Académique Documents
Professionnel Documents
Culture Documents
EDITOR IN CHIEF
Rudy J. M. Konings
European Commission, Joint Research Centre,
Institute for Transuranium Elements, Karlsruhe, Germany
SECTION EDITORS
Todd R. Allen
Department of Engineering Physics, University of Wisconsin, Madison, WI, USA
Roger E. Stoller
Materials Science and Technology Division, Oak Ridge National Laboratory, Oak Ridge, TN, USA
Shinsuke Yamanaka
Division of Sustainable Energy and Environmental Engineering, Graduate School of Engineering, Osaka University,
Osaka, Japan
Elsevier
Radarweg 29, PO Box 211, 1000 AE Amsterdam, The Netherlands
The Boulevard, Langford Lane, Kidlington, Oxford OX5 1GB, UK
225 Wyman Street, Waltham, MA 02451, USA
The following articles are US Government works in the public domain and not subject to copyright:
No part of this publication may be reproduced, stored in a retrieval system or transmitted in any form or by any means electronic,
mechanical, photocopying, recording or otherwise without the prior written permission of the publisher
Permissions may be sought directly from Elsevier’s Science & Technology Rights Department in Oxford, UK: phone (þ44) (0) 1865 843830;
fax (þ44) (0) 1865 853333; email: permissions@elsevier.com. Alternatively you can submit your request online by visiting the Elsevier web
site at http://elsevier.com/locate/permissions, and selecting Obtaining permission to use Elsevier material
Notice
No responsibility is assumed by the publisher for any injury and/or damage to persons or property as a matter of products liability,
negligence or otherwise, or from any use or operation of any methods, products, instructions or ideas contained in the material herein,
Because of rapid advances in the medical sciences, in particular, independent verification of diagnoses and drug dosages should be made
Cover image courtesy of Professor David Sedmidubský, The Institute of Chemical Technology, Prague
12 13 14 15 16 10 9 8 7 6 5 4 3 2 1
v
vi Editors Biographies
There are essentially three primary energy sources for the billions of people living on the earth’s surface: the
sun, radioactivity, and gravitation. The sun, an enormous nuclear fusion reactor, has transmitted energy to the
earth for billions of years, sustaining photosynthesis, which in turn produces wood and other combustible
resources (biomass), and the fossil fuels like coal, oil, and natural gas. The sun also provides the energy that
steers the climate, the atmospheric circulations, and thus ‘fuelling’ wind mills, and it is at the origin of
photovoltaic processes used to produce electricity. Radioactive decay of primarily uranium and thorium
heats the earth underneath us and is the origin of geothermal energy. Hot springs have been used as a source
of energy from the early days of humanity, although it took until the twentieth century for the potential of
radioactivity by fission to be discovered. Gravitation, a non-nuclear source, has been long used to generate
energy, primarily in hydropower and tidal power applications.
Although nuclear processes are thus omnipresent, nuclear technology is relatively young. But from the
moment scientists unraveled the secrets of the atom and its nucleus during the twentieth century, aided by
developments in quantum mechanics, and obtained a fundamental understanding of nuclear fission and fusion,
humanity has considered these nuclear processes as sources of almost unlimited (peaceful) energy. The first fission
reactor was designed and constructed by Enrico Fermi in 1942 in Chicago, the CP1, based on the fission of
uranium by neutron capture. After World War II, a rapid exploration of fission technology took place in the United
States and the Union of Soviet Socialist Republics, and after the Atoms for Peace speech by Eisenhower at the
United Nations Congress in 1954, also in Europe and Japan. A variety of nuclear fission reactors were explored for
electricity generation and with them the fuel cycle. Moreover, the possibility of controlled fusion reactions has
gained interest as a technology for producing energy from one of the most abundant elements on earth, hydrogen.
The environment to which materials in nuclear reactors are exposed is one of extremes with respect to
temperature and radiation. Fuel pins for nuclear reactors operate at temperatures above 1000 C in the center of
the pellets, in fast reactor oxide fuels even above 2000 C, whereas the effects of the radiation (neutrons, alpha
particles, recoil atoms, fission fragments) continuously damage the material. The cladding of the fuel and the
structural and functional materials in the fission reactor core also operate in a strong radiation field, often in a
dynamic corrosive environment of the coolant at elevated temperatures. Materials in fusion reactors are
exposed to the fusion plasma and the highly energetic particles escaping from it. Furthermore, in this
technology, the reactor core structures operate at high temperatures. Materials science for nuclear systems
has, therefore, been strongly focussed on the development of radiation tolerant materials that can operate in a
wide range of temperatures and in different chemical environments such as aqueous solutions, liquid metals,
molten salts, or gases.
The lifetime of the plant components is critical in many respects and thus strongly affects the safety as well as
the economics of the technologies. With the need for efficiency and competitiveness in modern society, there is
a strong incentive to improve reactor components or to deploy advanced materials that are continuously
developed for improved performance. There are many examples of excellent achievements in this respect. For
example, with the increase of the burnup of the fuel for fission reactors, motivated by improved economics and a
more efficient use of resources, the Zircaloy cladding (a Zr–Sn alloy) of the fuel pins showed increased
susceptibility to coolant corrosion, but within a relatively short period, a different zirconium-based alloy was
developed, tested, qualified, and employed, which allowed reliable operation in the high burnup range.
xxi
xxii Preface
Nuclear technologies also produce waste. It is the moral obligation of the generations consuming the energy
to implement an acceptable waste treatment and disposal strategy. The inherent complication of radioactivity,
the decay that can span hundreds of thousands of years, amplifies the importance of extreme time periods in the
issue of corrosion and radiation stability. The search for storage concepts that can guarantee the safe storage
and isolation of radioactive waste is, therefore, another challenging task for materials science, requiring a close
examination of natural (geological) materials and processes.
The more than 50 years of research and development of fission and fusion reactors have undoubtedly
demonstrated that the statement ‘technologies are enabled by materials’ is particularly true for nuclear
technology. Although the nuclear field is typically known for its incremental progress, the challenges posed
by the next generation of fission reactors (Generation IV) as well as the demonstration of fusion reactors will
need breakthroughs to achieve their ambitious goals. This is being accompanied by an important change in
materials science, with a shift of discovery through experiments to discovery through simulation. The progress
in numerical simulation of the material evolution on a scientific and engineering scale is growing rapidly.
Simulation techniques at the atomistic or meso scale (e.g., electronic structure calculations, molecular dynam-
ics, kinetic Monte Carlo) are increasingly helping to unravel the complex processes occurring in materials
under extreme conditions and to provide an insight into the causes and thus helping to design remedies.
In this context, Comprehensive Nuclear Materials aims to provide fundamental information on the vast variety of
materials employed in the broad field of nuclear technology. But to do justice to the comprehensiveness of the
work, fundamental issues are also addressed in detail, as well as the basics of the emerging numerical simulation
techniques.
R.J.M. Konings
European Commission, Joint Research Centre,
Institute for Transuranium Elements, Karlsruhe, Germany
T.R. Allen
Department of Engineering Physics,
Wisconsin University, Madison, WI, USA
R. Stoller
Materials Science and Technology Division,
Oak Ridge National Laboratory, Oak Ridge, TN, USA
S. Yamanaka
Division of Sustainable Energy and Environmental Engineering,
Graduate School of Engineering, Osaka University, Osaka, Japan
FOREWORD
‘Nuclear materials’ denotes a field of great breadth and depth, whose topics address applications and facilities
that depend upon nuclear reactions. The major topics within the field are devoted to the materials science and
engineering surrounding fission and fusion reactions in energy conversion reactors. Most of the rest of the field
is formed of the closely related materials science needed for the effects of energetic particles on the targets and
other radiation areas of charged particle accelerators and plasma devices. A more complete but also more
cumbersome descriptor thus would be ‘the science and engineering of materials for fission reactors, fusion
reactors, and closely related topics.’ In these areas, the very existence of such technologies turns upon our
capabilities to understand the physical behavior of materials. Performance of facilities and components to the
demanding limits required is dictated by the capabilities of materials to withstand unique and aggressive
environments. The unifying concept that runs through all aspects is the effect of radiation on materials. In this
way, the main feature is somewhat analogous to the unifying concept of elevated temperature in that part of
materials science and engineering termed ‘high-temperature materials.’
Nuclear materials came into existence in the 1950s and began to grow as an internationally recognized field
of endeavor late in that decade. The beginning in this field has been attributed to presentations and discussions
that occurred at the First and Second International Conferences on the Peaceful Uses of Atomic Energy, held in
Geneva in 1955 and 1958. Journal of Nuclear Materials, which is the home journal for this area of materials
science, was founded in 1959. The development of nuclear materials science and engineering took place in the
same rapid growth time period as the parent field of materials science and engineering. And similarly to the
parent field, nuclear materials draws together the formerly separate disciplines of metallurgy, solid-state
physics, ceramics, and materials chemistry that were early devoted to nuclear applications. The small priest-
hood of first researchers in half a dozen countries has now grown to a cohort of thousands, whose home
institutions are anchored in more than 40 nations.
The prodigious work, ‘Comprehensive Nuclear Materials,’ captures the essence and the extensive scope of the
field. It provides authoritative chapters that review the full range of endeavor. In the present day of glance and
click ‘reading’ of short snippets from the internet, this is an old-fashioned book in the best sense of the
word, which will be available in both electronic and printed form. All of the main segments of the field are
covered, as well as most of the specialized areas and subtopics. With well over 100 chapters, the reader finds
thorough coverage on topics ranging from fundamentals of atom movements after displacement by energetic
particles to testing and engineering analysis methods of large components. All the materials classes that have
main application in nuclear technologies are visited, and the most important of them are covered in
exhaustive fashion. Authors of the chapters are practitioners who are at the highest level of achievement
and knowledge in their respective areas. Many of these authors not only have lived through a substantial
part of the history sketched above, but they themselves are the architects. Without those represented here
in the author list, the field would certainly be a weaker reflection of itself. It is no small feat that so many
of my distinguished colleagues could have been persuaded to join this collective endeavor and to make the
real sacrifices entailed in such time-consuming work. I congratulate the Editor, Rudy Konings, and
xxiii
xxiv Foreword
the Associate Editors, Roger Stoller, Todd Allen, and Shinsuke Yamanaka. This book will be an important
asset to young researchers entering the field as well as a valuable resource to workers engaged in the
enterprise at present.
Dr. Louis K. Mansur
Oak Ridge, Tennessee, USA
Permission Acknowledgments
The following material is reproduced with kind permission of Cambridge University Press
Figure 15 of Oxide Dispersion Strengthened Steels
Figure 15 of Minerals and Natural Analogues
Table 10 of Spent Fuel as Waste Material
Figure 21b of Radiation-Induced Effects on Microstructure
www.cambridge.org
The following material is reproduced with kind permission of American Chemical Society
Figure 2 of Molten Salt Reactor Fuel and Coolant
Figure 22 of Molten Salt Reactor Fuel and Coolant
Table 9 of Molten Salt Reactor Fuel and Coolant
Figure 6 of Thermodynamic and Thermophysical Properties of the Actinide Nitrides
www.acs.org
i
ii Permission Acknowledgments
The following material is reproduced with kind permission of Taylor & Francis
Figure 9 of Radiation-Induced Segregation
Figure 6 of Radiation Effects in Zirconium Alloys
Figure 1 of Dislocation Dynamics
Figure 25 of Radiation Damage Using Ion Beams
Figure 26 of Radiation Damage Using Ion Beams
Figure 27 of Radiation Damage Using Ion Beams
Figure 4 of Radiation-Induced Effects on Material Properties of Ceramics (Mechanical and Dimensional)
Figure 7 of The Actinides Elements: Properties and Characteristics
Figure 20 of The Actinides Elements: Properties and Characteristics
Figure 18a of Primary Radiation Damage Formation
Figure 18b of Primary Radiation Damage Formation
Figure 18c of Primary Radiation Damage Formation
Figure 18d of Primary Radiation Damage Formation
Figure 18e of Primary Radiation Damage Formation
Figure 18f of Primary Radiation Damage Formation
Figure 1 of Radiation-Induced Effects on Microstructure
Figure 27 of Radiation-Induced Effects on Microstructure
Figure 5 of Performance of Aluminum in Research Reactors
Figure 2 of Atomic-Level Dislocation Dynamics in Irradiated Metals
Figure 3 of Atomic-Level Dislocation Dynamics in Irradiated Metals
Figure 5 of Atomic-Level Dislocation Dynamics in Irradiated Metals
Figure 10a of Atomic-Level Dislocation Dynamics in Irradiated Metals
Figure 10b of Atomic-Level Dislocation Dynamics in Irradiated Metals
Figure 10c of Atomic-Level Dislocation Dynamics in Irradiated Metals
iv Permission Acknowledgments
3.01.1 Introduction 2
3.01.2 Properties of Metal Fuel Alloys 4
3.01.2.1 Physical Properties 4
3.01.2.1.1 Density 4
3.01.2.1.2 Solidus and liquidus temperatures 5
3.01.2.1.3 Phase transition temperatures 6
3.01.2.1.4 Heat capacity 7
3.01.2.1.5 Thermal conductivity 7
3.01.2.1.6 Thermal expansion 8
3.01.2.2 Mechanical Properties 9
3.01.2.3 Diffusion Properties 12
3.01.2.4 Effects of MA Addition 13
3.01.3 Metal Fuel Fabrication 14
3.01.3.1 Fuel Slug Fabrication 15
3.01.3.1.1 Injection casting 15
3.01.3.1.2 Other methods 18
3.01.3.2 Fuel Pin Assembly 19
3.01.4 Steady-State Irradiation Behavior 19
3.01.4.1 Steady-State Irradiation Tests 19
3.01.4.2 Fuel Constituent Migration 20
3.01.4.3 Fission Gas Release and Gas Swelling 21
3.01.4.4 Restructuring and Deformation of the Fuel Slug 25
3.01.4.5 Fuel–Cladding Mechanical Interaction 27
3.01.4.6 Change in Fuel Slug Temperature 28
3.01.4.7 Fuel–Cladding Chemical Interaction 28
3.01.4.8 Behavior of Fission Products 29
3.01.4.9 Behavior of Breached Fuel Pins 30
3.01.4.10 Behavior of MA-Bearing Metal Fuel 31
3.01.4.11 Factors Controlling Fuel Lifetime 31
3.01.5 Transient Behavior 32
3.01.5.1 Transient Tests 32
3.01.5.2 Linear-Power-to-Melting 32
3.01.5.3 Liquefaction at the Fuel–Cladding Interface 33
3.01.5.4 Molten Fuel Motion 35
3.01.5.5 Fuel Pin Failure Mechanism 36
3.01.5.6 Failed Fuel Behavior 37
3.01.6 Summary and Future Development 37
References 37
1
2 Metal Fuel
EBR-I, II Experimental Breeder Reactor-I, II burnup limit of the Mk-II driver fuel to 8 at.%.
FBTA Fuel behavior test apparatus Another issue in metal fuel development at the
FCCI Fuel–cladding chemical interaction time was to explore appropriate compositions of
FCF Fuel cycle facility Pu-bearing fuel, which is essential in fuel cycle sys-
FCMI Fuel–cladding mechanical interaction tems for fast breeder reactors. The Mk-I and Mk-II
FFTF Fast Flux Test Facility driver fuels of EBR-II were the U–5 wt% Fs alloy,
Fs Fissium, a mixture of metals: 49.2Mo, where Fs stands for fissium, a mixture of metals:
39.2Ru, 5.6Rh, 3.8Pd, 2Zr, and 0.2Nb 2.46Mo, 1.96Ru, 0.28Rh, 0.19Pd, 0.1Zr, and 0.01Nb
(in wt%) (in wt%), which is the equilibrium composition of
Fz Fizzium, a mixture of metals: 27.5Mo, residual materials left in the melt-refining process.1
29.5Ru, 5Rh, 10Pd, and 28Zr (in wt%) Because the U–Pu–Fs alloys showed unsatisfactory
IFR Integral Fast Reactor compatibility with cladding materials, various other
INL Idaho National Laboratory U–Pu-based alloys were examined from the stand-
KAERI Korea Atomic Energy Research Institute point of physical properties, irradiation performance,
LOF Loss of flow and compatibility with cladding materials. As a result,
MA Minor actinides the ANL researchers considered that U–Pu–Zr alloys
RBCB Run-beyond-cladding breach would be the best because of their solidus temperature
RE Rare earths and compatibility with stainless steels. The above
SD Smear density history of metal fuel development until the 1980s is
TOP Transient overpower described in Stevenson,1 Walters et al.,2 Hofman and
TREAT Transient reactor test facility Walters,3 Hofman et al.,4 and Crawford et al.5
TRU Transuranium element The key features of metal fuel design – U–Pu–
UTOP Unprotected transient overpower 10 wt% Zr fuel slug and 75% smear density – were
WPF Whole-pin furnace embodied in the Integral Fast Reactor (IFR) pro-
gram6,7 initiated at ANL in 1984. A schematic view
of a metal fuel is shown in Figure 1. The cylindrical
fuel alloy rod is called a ‘fuel slug.’ Because sodium
3.01.1 Introduction does not react with U–Pu–Zr alloys, the annular gap
between the fuel slug and the cladding can be filled
Metal fuels are ideal for fast reactors because they with sodium (bond Na) to ensure thermal conduction
have higher densities of fissile and fertile materials from the fuel slug to the coolant. A relatively large
than any other fuel forms and provide higher reactor gas plenum, which is a space above the fuel slug, is
core performance such as higher breeding ratio and provided to mitigate the pressure of the fission gas
less fissile inventory. Early experimental fast reactors – accumulating in the course of irradiation. In the IFR
Experimental Breeder Reactor I (EBR-I), EBR-II, program, 2000 test pins of the U–10 wt% Zr binary
the Enrico Fermi Reactor, and the Dounreay Fast alloy fuel and 600 test pins of the U–Pu–10 wt% Zr
Reactor (DFR) – therefore utilized uranium alloys ternary fuel were irradiated in EBR-II and the Fast
as driver fuel. The burnup of metal fuel in those days Flux Test Facility (FFTF)8 until the program had to
was limited to a few atom percent (at.%) because of be terminated in 1994. Of these test pins, about 300
the increase in the fuel–cladding mechanical interac- U–Pu–Zr pins and 1500 U–Zr pins exceeded 10 at.
tion (FCMI) caused by gas swelling of fuel alloys. % burnup.8 The highest burnup achieved was more
Before the full potential of metal fuel was revealed, than 19 at.% for the U–19 wt% Pu–10 wt% Zr fuel
the global trend of fast reactor fuel development was pin,5,9 whereby the high burnup capability of the
directed toward oxide fuels. However, continuous metal fuel was demonstrated. All of the driver fuel of
efforts were made to raise the burnup limit of driver EBR-II was converted to Mk-III fuel (U–10 wt%
fuel of the EBR-II at Argonne National Laboratory Zr), and more than 10 000 U–10 wt% Zr fuel pins
(ANL) in the United States. It was found that reduc- were irradiated.8 A wide variety of irradiation tests,5
ing the fuel smear density to about 75% was effective in-pile transient tests,10 and out-of-pile heating
in promoting fission gas release before fuel–cladding tests11,12 in the IFR program revealed steady-state
contact and in suppressing FCMI at an early stage of irradiation behavior and transient performance of
irradiation. Here, ‘smear density (%)’ is defined as metal fuel.
the cross-sectional area ratio of the fuel slug to the An important factor in selecting a fuel form for
cladding inside. This finding increased the design fast reactors is ease of fuel recycling, that is,
Metal Fuel 3
U–(10-20) at.% Pu–Zr data trend for 500 ppm oxygen and carbon: ANL20
U–15 wt% Pu–Zr (as cast): Harbur et al.23
U–15 wt% Pu–Zr (extruded): Harbur et al.23
U–Zr: Rough18
U–(12.9,17.2) at.% Pu–22.5 at.% Zr (as cast): Boucher and Barthelemy19
U–Zr: estimation
U–30 at.% Pu–Zr: estimation
18.0
17.5
17.0
Density (g cm–3)
16.5
16.0
15.5
15.0
10 15 20 25 30
Zr concentration (at.%)
Figure 2 Density of U–Zr and U–Pu–Zr alloys.
Metal Fuel 5
Data in Ref. Eqn [1] Deviation Data in Ref. Eqn [2] Deviation
the solidus temperature Tsol (K) and liquidus tem- 1400 30 at.% Pu
perature Tliq (K) obtained from the optimized ter-
1300
nary phase diagram by the following relations.
1200 40 at.% Pu
Tsol ¼ A0 þ A1 CZr þ A2 CZr
2
þ A3 CZr
3
½1
1100
A0 ¼ 1408 1187CPu þ 967CPu
2 0 10 20 30 40 50 60 70 80
(a) Zr concentration (at.%)
A1 ¼ 572 732CPu þ 4960CPu
2
2000
A2 ¼ 740 þ 3305CPu 29182CPu
2
0 at.% Pu
1900 10 at.% Pu
A3 ¼ 624 3139CPu þ 36120CPu
2
1800 20 at.% Pu
Liquidus temperature (K)
B0 ¼ 1408 749CPu þ 2
93CPu 1600
1500
B1 ¼ 1313 þ 3869CPu þ 5072CPu
2 40 at.% Pu
1400
B2 ¼ 1052 6637CPu 44769CPu
2
1300
B3 ¼ 521 þ 1683CPu þ 66380CPu
2
1200
Here, CZr , CPu , and CU are the atomic fractions of Zr, 1100
Pu, and U, respectively. Correlations [1] and [2] are 0 10 20 30 40 50 60 70 80
applicable for CPu =CU < 1 and CZr < 0:8. In the case of (b) Zr concentration (at.%)
the U–Zr binary alloy, CPu ¼ 0. The values calculated Figure 3 Evaluated solidus and liquidus temperatures
by using these relations are shown in Figure 3 and also of U–Zr and U–Pu–Zr alloys.
6 Metal Fuel
in Table 1, which indicate that there are deviations for bcc e-plutonium and bcc b-zirconium; g1 and
from the reported data: <60 K for the solidus g2 are the uranium-rich and zirconium-rich mod-
and <130 K for the liquidus. ifications of g, respectively, that are formed by a
monotectoid reaction in the U–Zr binary system.
a: Orthorhombic allotropic modification of ura-
3.01.2.1.3 Phase transition temperatures
nium that dissolves up to 15 at.% of plutonium,
A U–Zr binary phase diagram was shown by
but has limited solubility for zirconium.
Massalski.37 Kurata et al.33 evaluated this alloy system
b: Tetragonal allotropic modification of uranium
on the basis of various published thermochemical data
that dissolves up to 20 at.% of plutonium, but has
and phase boundary data. O’Boyle et al.25 experimen-
limited solubility for zirconium.
tally determined the U–Pu–Zr ternary phase diagram
Z: A high-temperature intermediate phase in the
at several temperature cross sections. Kurata33 opti-
U–Pu binary system that is believed to be tetrago-
mized the U–Pu–Zr ternary phase diagram, as dis-
nal and has limited solubility for zirconium.
cussed in Section 3.01.2.1.2. Figure 4 illustrates the
z: A complex cubic U–Pu intermediate phase that
phase transition temperatures estimated from several
dissolves up to 5 at.% zirconium.
U–Pu–Zr isotherms by O’Boyle et al.25 The phases
d: A hexagonal intermediate phase in the U–Zr
shown in the figure are as follows25:
system that occurs approximately at the composi-
g: Body-centered cubic (bcc) allotropic modifica- tion UZr2 and has extensive solid solubility for
tion of uranium that has complete solid solubility plutonium.
1000
γ
990 992 K
980 γ
γ1+γ2 Tγ
970 γ
960 ∼958 K γ
γ1+γ2 Tγ γ
950 β+γ
γ1+γ2+β ∼943 K
940
γ1+γ2
γ+β Tγ
930
β+γ+ζ
∼923 K
Temperature (K)
920
Tγ
∼913 K
910 α+γ
900 γ+ζ
γ+α+ζ
890 890 K γ+ζ
Figure 4 Phase transition temperatures of U–Zr and U–Pu–Zr alloys estimated from O’Boyle and Dwight.25
Metal Fuel 7
In Figure 4, Ta is the temperature below which was measured by a comparative method at ANL.27
the g-phase disappears, and Tg is the temperature Takahashi et al.28 measured the U–Zr thermal diffu-
above which the g solid solution dominates. sivities by a laser-flash method, from which they
evaluated the thermal conductivities based on the
3.01.2.1.4 Heat capacity U–Zr heat capacities estimated from the elemen-
Heat capacity data for U–Pu–Zr alloys have not been tal heat capacities. These data are summarized in
reported to date. Takahashi et al.30 and Matsui et al.31 Figure 6. Matsui et al.31 evaluated the thermal con-
measured the heat capacities of U–Zr alloys, which ductivities of the U–20 at.% Zr alloy on the basis of
are presented in Figure 5. The curves in the figure its heat capacity that they measured by the direct
are the heat capacities that have been calculated on heating pulse calorimetry as well as the U–Zr thermal
the basis of a thermodynamic assessment of the U–Zr diffusivities measured by Takahashi et al.28 The eval-
binary system by Kurata et al.32 The calculated values uated values were consistent with the data reported
below 850 K are in good agreement with the data by in Touloukian et al.,24 Argonne National Laboratory
Matsui et al.,31 but the calculated values above 900 K report,27 and Takahashi et al.28 The data for U–Pu–Zr
are in good agreement with the data by Takahashi alloys, which were measured by a comparative method
et al.30 Because the heat capacity of plutonium is similar with reference to an Armco iron sample, are contained
to that of uranium, the heat capacity of U–Pu–xZr in an Argonne National Laboratory report.21 These
alloys may be similar to that of U–xZr alloys. data are listed in Figure 7, with the U–Pu data
reported in Kelman et al.22
3.01.2.1.5 Thermal conductivity Billone et al.38 showed the thermal conductivity
Touloukian et al.24 contains the thermal conductivity relation for U–Zr and U–Pu–Zr alloys as follows:
data on U–Zr alloys, which can also be found
in Rough.18 The data for U–25.1 at.% Zr alloy k0 ¼ A þ BT þ CT 2 ½3
50
45
40
Heat capacity (J K−1 mol−1)
35
30
U–14 at.% Zr: data in Takahashi et al.30
U–20 at.% Zr: data in Matsui et al.31
U–35 at.% Zr: data in Takahashi et al.30
U–72 at.% Zr: data in Takahashi et al.30
25
U–15 at.% Zr: calculation
U–20 at.% Zr: calculation
U–35 at.% Zr: calculation
U–70 at.% Zr: calculation
20
300 500 700 900 1100 1300 1500 1700 1900
Temperature (K)
Figure 5 Heat capacity of U–Zr alloys. ‘Calculations’ are based on Kurata et al.32
8 Metal Fuel
45
40
35
30
Thermal conductivity (W m−1 K−1)
25
20
U: Takahashi et al.28
0
300 500 700 900 1100 1300
Temperature (K)
Figure 6 Themal conductivity data of U–Zr alloys.
1 2:23WZ k0 ¼ 16:309 þ 0:02713T þ 46:279CZr
A ¼ 17:5 2:62WP
1 þ 1:61WZ þ 22:985CZr
2
53:545CPu ½4
2 1 þ 0:061WZ
B ¼ 1:54 10 0:90WP
1 þ 1:61WZ T < 1173K; CZr < 0:72; CPu < 0:16
C ¼ 9:38 106 ð1 2:70WP Þ where CZr and CPu are the atomic fractions of Zr
1 1 and Pu, respectively. For the U–Zr binary alloy,
where k0 is the thermal conductivity (W m K ), T
CPu ¼ 0. The values calculated for U–Pu–22 at.%
is the temperature (K), and WZ and WP are the
Zr alloys with relation [4] are shown in Figure 8.
weight fractions of the zirconium and plutonium,
respectively.
Relation [3] does not reflect the U–Zr data by 3.01.2.1.6 Thermal expansion
Takahashi et al.,28 which came after the publication The thermal expansion of U–Zr alloys is reported in
of Billone et al.38 Ogata36 proposed the following Rough,18 but these data are for the Zr-rich side. For
simpler relation for the U–Zr and U–Pu–Zr thermal U–Pu–Zr alloys, the data are contained in Boucher
conductivities, reflecting all of the available data and Barthelemy19 and Kelman et al.,22 as summarized
plotted in Figures 6 and 7. in Table 2.
Metal Fuel 9
45
40
35
30
Thermal conductivity (W m-1 K-1)
25
20
15
0
300 500 700 900 1100 1300
Temperature (K)
45
3.01.2.2 Mechanical Properties
The modulus of elasticity, yield strength, and ultimate
40
tensile strength of various compositions of U–Pu–Zr
35 0 at.% Pu alloys are given by Harbur et al.23 and Kittel et al.,26
Thermal conductivity (W m−1 K−1)
Table 3 Modulus of elasticity, yield strength, and ultimate tensile strength of U–Zr and U–Pu–Zr alloys
Ref. Composition (at.%) Temperature Modulus of elasticity Ultimate tensile Yield strength, 2% offset
(K) (GPa) strength (MPa) (MPa)
Continued
Metal Fuel 11
Table 3 Continued
Ref. Composition (at.%) Temperature Modulus of elasticity Ultimate tensile Yield strength, 2% offset
(K) (GPa) strength (MPa) (MPa)
600
Modulus of elasticity (GPa)
140
500
120
100 400
80 300
60
200
40
100
20
0
0 200 400 600 800 1000
200 400 600 800 1000
Temperature (K)
Temperature (K)
Figure 10 Ultimate tensile strength of U–Pu–Zr alloys.
Figure 9 Modulus of elasticity of U–Zr and U–Pu–Zr
alloys.
400
e_ ¼ 8:0 102 s3 expð14350=T Þ ½6
U–Pu–Zr: Kittel et al.26
350 U–Pu–Zr: Harbur et al.23 where e_ is the creep strain rate (s1), s is the stress
(MPa), and T is the temperature (K). Relation [6] for
300 the g solid solution is consistent with the data in
Kelman et al.22 and Kittel et al.,26 but eqn [5] gives
Yield strength (MPa)
250
the lower bound of the data in Harbur et al.23 For
200
U–Zr alloys, some data for the Zr-rich side appear
in Rough.18
150 Ogata et al.41 estimated the creep strain rate of
U–22.5 at.% Zr alloy from the relaxation behavior
100 of compressive stress applied to the sample above
1000 K. These data suggest that the creep strain rate
50 of U–Zr alloys is significantly lower than that of
U–Pu–Zr alloys. Robinson et al. have reported the
0
600 700 800 900 1000 creep strain rate data for uranium metal.42
Temperature (K)
Ref. Composition (at.%) Temperature (K) Stress (MPa) Creep strain rate (s1)
Continued
Metal Fuel 13
Table 5 Continued
Ref. Composition (at.%) Temperature (K) Stress (MPa) Creep strain rate (s1)
Interdiffusion (chemical) diffusion coefficients in For U–Pu–Zr ternary alloys, Petri and Dayananda48
the bcc solid solution (g-phase) of U–Zr binary alloys examined the interdiffusion coefficients at 1023 K,
were measured by Adda et al.43 in the temperature where the g-phase is dominant, by using various diffu-
range 1223–1348 K. Ogata et al.44 measured the diffu- sion couples consisting two alloys out of U, U–20Zr,
sion coefficients in the g-phase from 973 to 1223 K and U–22Pu–3Zr, and U–22Pu–20Zr (in at.%) alloys.
found a depression in the coefficients at a zirconium Thermodiffusion tests with U–Pu–Zr ternary alloys
content of about 0.3 atom fraction, corresponding to were performed by Harbur et al.,23 Kurata et al.,49 and
the miscibility gap at this zirconium content at 995 K. Sohn et al.,50 where the redistributions of U and Zr were
For the d-phase, Akabori et al.45 measured the interdif- observed. Analyses of these experimental data will
fusion coefficients at 823 and 853 K, and pointed out contribute to an understanding of the phenomenon of
that the diffusion coefficients in the d-phase are sig- fuel constituent migration in metal fuel.
nificantly smaller than those extrapolated from the
g-phase to the d-phase. These U–Zr interdiffusion
3.01.2.4 Effects of MA Addition
coefficients in the bcc solid solutions and the d-phase
are plotted in Figure 12. Adda et al.46 evaluated the The recent interest in MA transmutation, as discussed
intrinsic diffusion coefficients of U and Zr in the in Section 3.01.1, has led to the evaluation of the
g-phase at 1223, 1273, and 1313 K, and showed that properties of MA-bearing fuel alloys. Kurata et al.39
the U diffusion is much higher than the Zr diffusion in performed the dilatometric analysis of U–19Pu–
the U-rich side. The interdiffusion in U–Pu binary 10Zr, U–19Pu–10Zr–2MA–2RE, and U–19Pu–10Zr–
alloys at 1023 K was investigated by Petri et al.47 5MA–5RE (in wt%) rod samples. Dilatometry
14 Metal Fuel
1.0E−08
1.0E−09
Interdiffusion coefficient (cm2 s–1)
1.0E−10
1.0E−11
1348 K: Adda et al.43
1313 K: Adda et al.43
1273 K: Adda et al.43
1.0E−14
0 20 40 60 80 100
Zr concentration (at.%)
Figure 12 U–Zr interdiffusion coefficients in the bcc solid solutions and the d-phase.
16
conductivities of U–19Pu–10Zr and U–19Pu–10Zr–
5MA–5RE (in wt%) were measured by a comparative 14
method,39 as shown in Figure 13. The figure suggests
12
that the thermal conductivity of the U–Pu–Zr alloys
is not sensitive to MA and RE additions up to 5 wt%. 10
The elastic modulus, shear modulus, and Poisson’s
8 U–19Pu–10Zr–5MA–5RE (wt%)
ratio of the U–19Pu–10Zr–5MA–5RE alloys were
similar to those of the U–19Pu–10Zr alloy within 6 U–19Pu–10Zr (wt%)
experimental error,39 as indicated in Table 4. Other
4
property data for MA-bearing fuel alloys are now being 200 300 400 500 600 700
measured at INL. Temperature (⬚C)
Figure 13 Measured thermal conductivities of U–19Pu–
10Zr and U–19Pu–10Zr–3Np–2Am–0.2Y–1.2Ce–3.6Nd
3.01.3 Metal Fuel Fabrication alloys.
A practical process for nuclear fuel fabrication needs and has been applied to fuel slug fabrication for the
to be cost efficient (or simple), suitable for remote EBR-II driver and test fuel pins since the 1960s.1,51
operation, and capable of mass production while In the demonstration of metal fuel recycle at ANL
reducing the amount of radioactive waste. Injection in the 1960s, the fuel slugs were refabricated remotely
casting is one of the processes that meets these needs in the hot cell from partially decontaminated
Metal Fuel 15
radioactive uranium recovered from irradiated tube molds with the top ends closed are set above
U–5 wt% Fs fuel.1 The U–Zr and U–Pu–Zr fuel the crucible. The crucible’s interior is coated with
slugs for test subassemblies irradiated in EBR-II and yttria and the mold’s interior is coated by zirconia
FFTF were also fabricated by injection casting.51 for protection against reaction with molten uranium
More than 100 000 metal fuel pins including both alloy. The furnace is closed and filled with highly
U–5 wt% Fs and U–10 wt% Zr fuels were fabricated purified Ar gas. The crucible is inductively heated
by injection casting in the United States.51 The metal up to 1833 K, which is sufficiently higher than the
fuel for earlier fast reactors, EBR-I and the Enrico liquidus temperature of the fuel alloy (e.g., 1656 K for
Fermi Reactor, were made by various methods such as U–10 wt% Zr). In order to ensure the homogeneity
rolling and swaging, coextrusion, and centrifugal cast- of the melt, it is kept at the high temperature and
ing,2 but these were not better than injection casting. stirred electromagnetically by applying full power
This section describes fuel slug fabrication meth- to the crucible.51,53,54 After the vessel is evacuated,
ods, focusing on the injection casting process. The the molds are lowered and their bottom ends are
process of metal fuel pin assembly is also described. immersed in the melt. On again refilling the furnace
The development history and recent activities of metal with Ar gas, the pressure difference between the
fuel fabrication are summarized in Burkes et al.,51,52 mold’s interior (vacuum) and the furnace (Ar pres-
which is referred to in many parts of this section. sure) injects the melt into the molds. The injected
melt is quickly solidified from the top to the bottom.
After cooling, the fuel alloy castings are taken out of
3.01.3.1 Fuel Slug Fabrication
the molds. The mold must be broken when the cast-
3.01.3.1.1 Injection casting ing is taken out. Therefore, the mold is not reusable,
An outline of an injection casting process is illu- and the mold shards will be radioactive waste. How-
strated in Figure 14, based on Burkes et al.,51 the ever, the shards can be used as glass materials for
Argonne National Laboratory,53 and Ogata and waste forms such as glass-bonded sodalite, so that
Tsukada.54 The starting materials, that is, uranium these may not be considered as additional waste.55
and zirconium metals (and uranium–plutonium alloy, The fuel slugs are obtained by shearing off both ends
when U–Pu–Zr casting), are charged into the graph- of the castings. It is unnecessary to grind the fuel slug
ite crucible in the injection casting furnace, and silica surface unlike ceramic fuel pellets; the fuel slug
1. Charging starting metals 2. Closing vessel, melting 3. Lowering molds 4. Withdrawing molds
and vacuuming and injection and cooling
Mold bundle
drive shaft
Molds
Molten fuel Residual
alloy fuel alloy
Starting metals Heat shield (heel)
Induction coil
Vacuuming
Ar gas
Insulator
Crucible
Scraps
diameter is controlled by the inner diameter of the U–Zr slugs was satisfactory with respect to the pro-
mold. The casting parameters such as molten alloy visional specifications: average diameter precision
temperature, mold preheat temperature, pressuriza- 0.05 mm; local diameter precision 0.1 mm; den-
tion rate in injection, and cooling rate after injection sity 15.3–16.1 g cm3; zirconium content 101 wt%;
should be carefully determined according to the the total amount of impurities (O, C, N, and Si)
mold dimensions and fuel alloy composition. Inap- <2000 ppm. Typical distributions of the slug diame-
propriate parameters may cause casting defects such ter and density are presented in Figure 16. Figure 17
as shrinkage pipes, microshrinkage, and hot tears.51 shows the relationship between the slug average
Injection casting tests with the furnace shown in diameter and the mold inner diameter measured at
Figure 15 were conducted by CRIEPI,54 based the bottom-end opening. The solid line in the figure
on the experience in the United States. The maxi- denotes the slug diameters calculated by subtracting
mum metal charge of the furnace is 20 kg of the the thermal shrinkage of the U–Zr alloy and the
U–10 wt% Zr alloy per batch, which is close to that of zirconia coating thickness (estimated to be 0.01 mm)
commercial-scale equipment. The silica molds were from the mold inner diameter. In this calculation, it
6 mm in inner diameter and 500 mm in length. The was assumed that the alloy was cooled from the
graphite crucible is inductively heated at a frequency solidus temperature of 1566 K for the U–10 wt% Zr
of 3 kHz and a maximum power of 30 kW. The start- alloy (see Section 3.01.2.1.2) down to room temper-
ing metals were basically uranium metal blocks and ature, and the thermal expansion coefficient of the g
zirconium metal cut wire. In most of the casting
batches, slugs, heels, and scraps from the preceding
casting batch were also charged, simulating a practi- 30
Frequency (%)
cal fuel slug casting process. These metals were Provisional Local diameter
weighed and adjusted so that the composition of the 20 tolerance
alloy was U–10 wt% Zr. Complete melting and dis-
10
solution of the metals were ensured by maintaining
the metal temperature at 1780–1840 K for about 0
30 min. The argon gas pressurization rate in injection 40
was 0.2 MPa s1 and the terminal pressure was Provisional
Average diameter
Frequency (%)
30
0.2 MPa. Both ends of the castings were cut off tolerance
using a shearing device, and 400-mm-long U–Zr 20
slugs were obtained. Ten casting test batches resulted
10
in the production of more than 500 slugs of
U–10 wt% Zr alloy. The quality of the produced 0
5.60
5.65
5.70
5.75
5.80
5.85
5.90
5.95
6.00
(a) Slug diameter (mm)
30
Ar gas tank
Frequency (%)
20
10 Provisional
tolerance
0
15.3
15.5
16.3
15.7
15.9
16.1
14.3
14.9
15.1
14.5
14.7
5.90 Ca lc ula
Ca step. The calculation by ICAST showed that the coat-
5.85 ing inside the mold acts as a thermal insulator for the
mm molten alloy to be injected higher. This was verified
0.05
5.80 n− by an injection-casting test without the mold coating.
tio
lc ula KAERI has experience in injection casting of
5.75 Ca
U–10 wt% Zr–(2,4,6) wt% Ce ternary alloys,57
where Ce was a surrogate element for MA or rare
5.70
earth fission products. The test result showed that Ce
5.65 particles were dispersed in the U–Zr matrix.
Recent interest in MA-bearing fuel has resulted in
5.60 a reevaluation of injection casting. A major problem-
5.85 5.90 5.95 6.00 6.05 6.10
atic point in the injection casting of MA-bearing
Mold inner diameter at the bottom end (mm)
metal fuel slugs is that the crucible used in injection
Figure 17 Relationship between the slug average casting is not a closed system, where a relatively
diameter and the mold inner diameter measured at the
high vapor pressure of Am raises concerns about
bottom-end opening.
contamination of the furnace’s interior and loss of
Am from the process. Three full-length fuel slugs
solid solution of the U–Zr alloy was approximated by (4.3 340 mm) of the U–20Pu–10Zr–1.2Am–1.3Np
that of the U–Pu–Zr alloy, that is, 2.0 10–5 K1 (see (in wt%) alloy were fabricated by injection casting
Section 3.01.2.1.6). Figure 17 indicates that most of for the X501 irradiation test.58–61 Although no
the average slug diameters fall within the range of unusual macrosegregation of the major constituents
0.05 mm of the calculated value, so that the slug was observed, only 60% of the initial Am charge was
diameter can be controlled by the mold inner diameter. present in the as-cast fuel. Am loss was attributed to
Despite the repeated use of heel and scrap, the total volatile impurities (Ca and Mg) in the Am–Pu feed
amount of impurities (O, C, N, and Si) was still lower stock61 and evaporation at the casting temperature,
than the provisional limit. In the last test batch, 1.1Mo, 1465 C. Chemical analysis of sections from the top,
0.8Pd, 0.06Ce, and 0.1Nd (in wt%) were added to the center, and bottom of the fuel slug revealed that the
metal charge, simulating the fission product elements U, Pu, Zr, and Np levels were axially uniform, within
that may remain in the pyroprocess products. Precipi- experimental error, while the Am level was low
tations of these elements were not detected in the U–Zr (1.03 wt%) in the bottom section compared to those
slugs. Improvement in throughput can be achieved by in the top and central sections (1.33 and 1.32 wt%,
increasing the casting ratio (weight percentage of the respectively).59 Trybus62 performed an injection-
injected metal relative to the charged metal). Optimiz- casting test with U–7.5 wt% Zr–1.5 wt% Mn alloy,
ing the depth of the mold bottom end in the molten fuel which was the surrogate alloy for U–Pu–Zr–Am–Np
and the array pattern of the mold bundle resulted in a alloys. Mn has a vapor pressure similar to that of
reasonable casting ratio of 70–80%. Am at the casting temperature. In the surrogate cast-
The influence of some of the casting parameters, ing test, the alloying temperature and the vacuum
for example, molten alloy temperature, mold preheat just before injection were reduced to 1455 C
temperature, and pressurization rate in injection, and 13.3 kPa, respectively, from those in the X501
on the maximum casting length can be predicted fuel casting, which were 1495 C and 670 Pa, respec-
by calculating the temperature of the molten fuel tively. The casting was successfully completed, and
alloy during injection casting. For this purpose, an chemical analysis of the samples from the slug center
injection-casting simulation code, ICAST, was devel- indicated 1.42 wt% Mn, which is 90% of the initial
oped.56 The ICAST code calculates the temperature Mn charge. This means that minimal Mn (and Am)
of the mold and fuel alloy during each step of the loss is possible by changing the casting parameters.62
injection casting process: mold preheating, injection, According to the comprehensive discussion on Am
and cooling. Radiation heat transfer from the molten evaporation of Burkes et al.,52 the Am evaporation can
alloy surface and crucible wall is essential for be reduced by decreasing the fuel melt temperature,
18 Metal Fuel
increasing the cover gas pressure, and/or reducing the Optimizing the casting conditions is difficult when
Am concentration gradient in the cover gas. The fuel the fuel alloy has a large solidification range52,57
melt temperature can be decreased by adjusting the (temperature difference between the solidus and the
fuel alloy composition, for example, by reducing the Zr liquidus). A wide solidification range can lead to
content in the fuel alloy.63 From the standpoint of microshrinkage effects and loss of process control
increased cover gas pressure, injection casting may be during casting.52 Furthermore, pulling of the cast
disadvantageous because the furnace is evacuated must be properly aligned to avoid any asymmetric
before injection. The Am concentration gradient can variations in the rod diameter, thereby increasing the
be reduced by using a closed system for fuel alloy complexity of the unit for remote operation.52
melting. This is possible for the methods presented Finally, if continuous casting were to be used, the
below, other than injection casting. process would need to be highly automated to mini-
Nakamura et al.64 recently fabricated U–Pu–Zr mize the extent of human interaction required for
metal fuel slugs by injection casting for an irradia- casting a significant number of fuel slugs.52
tion test in the experimental fast reactor, Joyo.
In the fabrication process, a small amount of Am 3.01.3.1.2.3 Gravity casting
(0.3 wt%) accompanied the fuel alloy. Chemical Renewed interest is being taken in gravity casting,
analysis and g spectrometry of the samples from the where fuel melt is poured into molds by gravity with
graphite crucible and yttria coating indicated that or without the assistance of a pressure difference. In
Am selectively reacted with the graphite crucible gravity casting as well as centrifugal casting and con-
and yttria coating. This suggests that attention tinuous casting, the furnace containing the fuel melt
needs to be paid not only to Am volatility but also is not evacuated, unlike in injection casting. This is
to its chemical reactions with process materials. favorable for suppressing Am evaporation. The grav-
ity casting system is relatively simple.
3.01.3.1.2 Other methods Lee et al.57 fabricated U–10 wt% Zr rods by grav-
3.01.3.1.2.1 Centrifugal casting ity casting with a split graphite mold and a quartz tube
In a centrifugal casting process, the molten fuel alloy mold. A two-piece graphite mold was also used to
is poured vertically onto a rotating plate (distributor), facilitate the demolding operation after the casting.57
where the melt flow turns to the horizontal direction. Vacuum-assisted gravity casting was also tested by
The molds are aligned on the edge of the distributor KAERI, and U–10 wt% Zr and U–10 wt% Zr–6 wt%
and rotate with it. The melt is injected into the molds Ce alloys slugs were successfully fabricated.57
by the centrifugal force. This process was used to cast An advanced casting system (ACS) is being devel-
U–2 wt% Zr alloy fuel slugs for EBR-I, which were oped at INL to demonstrate minimal actinide fuel
significantly larger in diameter than for EBR-II loss by rapid melting and casting under careful atmo-
(9.8 mm compared to 3.3–4.4 mm).52 sphere control in a reusable crucible and molds.65
Although centrifugal casting could potentially be The first step of ACS development activity includes
used to fabricate fuel slugs with dimensions typical of design and construction of a bench-scale casting sys-
those in a commercial fast reactor, the process has tem (BCS), sized for 50–300 g castings, for use with
been considered somewhat complicated and time MA-bearing fuel alloys to demonstrate minimal
consuming.52 The number and type of manipulations transuranium element (TRU) loss.65 BCS is based
required to assemble and disassemble the furnace and on bottom-poured casting assisted by a pressure dif-
molds are significant, and there are concerns about ferential, and has the capability to be configured for
the relatively low throughput, compared with other injection casting.65
fabrication processes.52
3.01.3.1.2.4 Atomizing
3.01.3.1.2.2 Continuous casting The concept of He-bond particulate metal fuel63 was
Continuous casting is widely used in steel plants, and proposed as an advanced metal fuel, where a cladding
is also one of the candidates for MA-bearing metal tube is filled with fuel alloy particles and the spaces
fuel slugs. This process eliminates the need to use among the particles are filled with He gas, not
molds. KAERI produced a uranium rod with a sodium. A mixture of particles with two different
uniform diameter of 13.7 mm and a length of diameters can attain a fuel smear density (filling
2.3 m.57 The continuous casting of U–Zr alloy slugs fraction) of about 75%.63 The He-bond particulate
with a smaller diameter is under way. metal fuel has the following advantages: the He-bond
Metal Fuel 19
allows the gas plenum to be positioned below the fuel the fuel alloy leads to high compressibility of the
column section, so that the gas plenum temperature swollen fuel slug; and lanthanide fission products
is reduced and the fuel pin length can be shortened; agglomerate at the peripheral region of the fuel slug
nonuse of bond sodium will save the corresponding and react with the Fe-based cladding. These phe-
amount of oxidizing agent required in the electrore- nomena are closely related to each other.
fining process; and the fuel alloy particles can be This section describes such characteristic steady-
fabricated by gas atomization or centrifugal atomiza- state irradiation behavior of metal fuel, after review-
tion, neither of which needs molds and are expected ing the irradiation tests of U–Zr and U–Pu–Zr fuel
to have higher production throughput than injection pins. Recent MA-bearing metal fuel tests and their
casting. Furthermore, the furnace for atomizing can limited data are also explained. A large part of this
be a closed system for fuel alloy melting, which is section is based on comprehensive documents on
suitable for MA-bearing fuel fabrication. metal fuels.3–5,69–71
Spherical uranium alloy particles such as those of
U–Mo and U–Zr were successfully fabricated by
3.01.4.1 Steady-State Irradiation Tests
centrifugal atomization.57
The U–Pu–Zr alloys were first irradiation-tested in
3.01.3.2 Fuel Pin Assembly the CP-5 thermal reactor,67 where six U–15 wt%
Pu–12 wt% Zr slugs clad with 304SS, 316SS, and
A metal fuel pin assembling process is schematically
Hastelloy-X were irradiated at a maximum cladding
shown in Figure 18. This is based on the scheme used
temperature of 610 C up to 2.4 at.% burnup, and one
for the fabrication of the metal fuel test pins66 to be
U–18.5 wt% Pu–14.1 wt% Zr slug clad with V–20Ti
irradiated in the experimental fast reactor, Joyo. This
at a maximum cladding temperature of 655 C up to
scheme is similar to that for EBR-II driver fuel pins.1
12.5 at.%. The fuel slug length was scaled down by a
Fuel slugs are checked for dimensions and weight (or
factor of seven from that of the EBR-II driver, that is,
density). Bond sodium is extruded by using a bond
34.3 cm. Subsequently, 16 U–15 wt% Pu–10 wt% Zr
sodium extruder and shaped into rods. The weight of
(nominal composition) fuel slugs clad with 304LSS,
the bond sodium to be loaded into the cladding is
316SS, Hastelloy-X, and Hastelloy-X-280 were
determined from the measured or evaluated dimen-
irradiated in EBR-II at a maximum cladding temper-
sions of the cladding’s interior and the fuel slug so as
ature ranging from 600 to 652 C up to about 4.5 at.%
to meet the gas plenum volume specification. The
burnup without failure.68 These early irradiation
rod-shaped bond sodium is first inserted into the
tests in the 1960s revealed the main features of
cladding tube with the lower-end plug welded, fol-
irradiation phenomena such as fission gas release,
lowed by the fuel slug(s). One or two more slugs are
restructuring, fuel constituent migration, and clad-
inserted as required. After welding the upper-end
ding wastage by lanthanide fission products.
plug, the fuel pin is checked for leaks. Then, the fuel
The main body of metal fuel irradiation data was
pin is heated up to 500 C and oscillated vertically so
gained through irradiation tests conducted in the
that the annular gap between the cladding and fuel
IFR program.6 Beginning with three lead test assem-
slug is filled with the bond sodium. The gas plenum
blies, 600 U–Pu–Zr test pins and 8000 U–Zr
length is checked by an X-ray transmission method.
test pins were irradiated in EBR-II and FFTF.8 In
The US historical experience of metal fuel pin
the tests, fuel pins with a wide variety of specifications
assembling is described in detail in Burkes et al.51
were irradiated under a wide range of conditions, as
follows5: Pu contents 0–28 wt%; Zr content 2–14 wt%;
3.01.4 Steady-State Irradiation smear density 70–85%; cladding material: an austen-
Behavior itic stainless steel (316SS), a titanium-stabilized aus-
tenitic stainless steel (D9), and a ferritic/martensitic
In the course of neutron irradiation, metal fuel exhi- steel (HT9); peak burnup 19 at.%; and peak clad-
bits a characteristic behavior, as shown in Figure 19, ding temperature <660 C. The full lineup of test
which is different from that of ceramic fuel. For assemblies in the IFR program is summarized in
example, when compared with oxide fuel, a metal Crawford et al.5 Representative test assemblies are
fuel slug tends to hold more fission gas atoms, and listed in Table 6.
accordingly showing a higher rate of gas swelling in Recent metal fuel irradiation tests have focused on
the early stages of irradiation; a higher creep rate of MA-bearing metal fuel, as summarized in Table 7.
20 Metal Fuel
Composition
Mater. check Mater. check Mater. check Mater. check
adjustment
Unacceptable Judgment
Lower-end-plugged
Acceptable product cladding
Inspection Loading
Fuel slug
Weld
Insulator
Heat treatment
Disassembly
Decontamination
Unacceptable
Radioactivity check
Weld check
Leak check
Na bonding
Appearance check
X-ray inspect.
Radioactivity check
Judgment
Among them, the X50158–60 test assembly has been 3.01.4.2 Fuel Constituent Migration
completed, and postirradiation examinations for the
The as-cast metal fuel slug shows macroscopically
METAPHIX,72–76 AFC-1,77–79 AFC-2,80,81 and
uniform distribution of the fuel constituents. Its ini-
FUTURIX-FTA82,83 tests are ready or in progress.
tial microstructure consists essentially of a metastable
Some of the test results have been reported.
Metal Fuel 21
Test assembly no. X419, X420, X423 X425 X430 X441 X447 IFR-1 MFF-2
X421
Remark Lead tests Swelling behavior Lead tests Large diameter, Code benchmark, High temp. Full length Full length, U–Zr
high Pu high smear
density
Reactor for irradiation EBR-II EBR-II EBR-II EBR-II EBR-II EBR-II FFTF FFTF
Fuel alloy composition U–10Zr, U–xPu– U–10Zr, U–xPu– U–10Zr, U– U–10Zr, U–19Pu–yZr U–10Zr U–10Zr, U–xPu– U–10Zr
(wt%) 10Zr (x ¼ 8, 19) 10Zr (x ¼ 3, 8, xPu–10Zr U–xPu–10Zr (y ¼ 6, 10, 14) 10Zr (x ¼ 8, 19)
19, 22, 26) (x ¼ 8, 19) (x ¼ 19, 22, 26)
Cladding material D9 316SS HT9 HT9 HT9, D9 HT9 D9 HT9
Cladding outer 5.84 7.37 5.84 7.37 5.84 5.84 6.86 6.86
diameter (mm)
Cladding thickness 0.38 0.41 0.38 0.41 0.38 0.46 0.56 0.56
(mm)
Fuel slug outer 4.32 5.66 4.32 5.66 5.71 4.98 4.98
diameter (mm)
Fuel slug length (mm) 343 343 343 343 343 343 914 914
Fuel smear density (%) 72 75 72 75 70,75, 85 75 75 75
Plenum/fuel volume 1 1 1 1.4 1.1,2.1 1.4 1.2 1.3
ratio
Peak linear power rate 394 427 482 492 459 361 492 541
(W cm1)
Peak cladding temp. 590 522 590 540 600 660 615 618
( C)
Peak burnup 18.4 at.% 4.9 at.% 19.3 at.% 11.5 at.% 12.7 at.% 10 at.% 94 GWd t1 94 GWd t1
Table 7 Irradiation tests of MA-bearing metal fuel
Metal Fuel
23
24 Metal Fuel
Zr
Pu 1 mm
(a)
U
0.5 mm
Pu
Figure 20 Optical micrography and measured constituent
redistributions of U–19 wt% Pu–10 wt% Zr fuel at 1.9 at.%
burnup. Reproduced from Kim, Y. S.; Hofman, G. L.; Zr
Hayes, S. L.; Sohn, Y. H. J. Nucl. Mater. 2004, 327, 27.
60
intermediate z-phase region, where Zr is depleted,
appears to be dense in Figure 20, but contains
50 uniformly distributed fine spherical bubbles, as indi-
cated in Figure 24. When the central region temper-
40 ature is relatively low (<650–700 C), where the g
single phase does not form, a Zr-depleted region is
U–Fs
30 formed in the central part of the slug, and the cross
U–Pu–Zr
section shows a two-ring structure. This restructur-
20 U–Pu–Fs
ing (ring structure formation) precedes fuel constitu-
ent migration according to Porter et al.,84 but it may
10
be possible that the constituent migration affects the
0 restructuring.
0 20 40 60 80 100 120 140 Figures 25–27 present irradiation test data of fuel
Fuel volume increase (%) slug axial elongation.67,93,97 After rapid swelling in
Figure 22 Fractional fission gas release versus fuel the early stage (2 at.%), the elongation levels off
volume increase. at an asymptotic value of 4–10%, depending on the
100
90
Fractional fission gas release (%)
80
70
60
50
40 U–(0,8,19)Pu–10Zr, 72% SD: Pahl et al.93
U–19Pu–10Zr, 70–75 % SD: Tsai et al.9
30
U–19Pu–10Zr, 85% SD: Tsai et al.9
20 U–0Pu–10Zr, 75% SD: Pahl et al.93
0
0 2 4 6 8 10 12 14 16 18 20
Peak burnup (at.%)
Figure 23 Fractional fission gas release peak burnup.
26 Metal Fuel
(a)
0.5 mm
5 μm 50 μm 10 μm
Figure 24 Pore morphology of irradiated U–Pu–Zr fuel shown by optical metallography and SEM images: (a) optical
micrograph of axial cross section, SEM images of (b) intermediate zone, (c) central zone, (d) outer zone. Reproduced from
Kim, Y. S.; Hofman, G. L.; Yacout, A. M. J. Nucl. Mater. 2009, 392, 164.
U–10 wt% Zr U–8 wt% Pu–10 wt% Zr U–19 wt% Pu–10 wt% Zr U–10 wt% Zr U–19 wt% Pu–10 wt% Zr
10
8
8
6
6
4
4
2
2
0 2 20
0 4 6 8 10 12 14 16 18
0
Peak burnup (at.%) 0 1 2 3 4 5
Peak burnup (at.%)
Figure 25 Fuel slug axial elongation versus peak
burnup for test assemblies X419-421 (fuel slug diameter: Figure 26 Fuel slug axial elongation versus peak burnup
4.32 mm, see Table 6). Data from Pahl et al.93 for test assemblies X423 (fuel slug diameter: 5.66 mm,
see Table 6). Data from Hofman et al.67
6 1.5
4
1
2
0.5
0
0 2 4 6 8 10 12
Peak burnup (at.%) 0
6 8 10 12 14 16 18 20
Figure 27 Fuel slug axial elongation versus peak burnup Local burnup (at.%)
for test assemblies X430 (fuel slug diameter: 5.66 mm,
see Table 6). Data from Crawford et al.97 Figure 28 Cladding diametral strain of HT-9–clad
U–Pu–Zr fuel pins of smear density 75%.
2.5 1.2
Maximum diametral strain of cladding (%)
1 0.4
0.5 0.2
0
0 0.2 0.4 0.6 0.8 1 1.2 1.4
0
68 70 72 74 76 78 80 82 84 86 x / L0
Fuel smear density (%)
Figure 30 Cladding diametral strain profiles of HT-9–clad
Figure 29 Maximum cladding diametral strain versus fuel U–19 wt% Pu–10 wt% Zr fuel pins at 15.5 at.% (T473) and
smear density for HT-9–clad U–Pu–Zr fuel pins. 15.9 at.% (T470) peak burnups.
a peak burnup of 15–16 at.%. The cladding diameter conductivity from the radii of the molten region of
increase peaks at the middle of the fuel slug and is the irradiated fuel slugs subjected to the transient
lower at the both ends, similar to the local burnup overpower tests, as follows:
profile, that is, solid fission product swelling distribu- Before sodium infiltration
tion. FCMI may be at a lower level, particularly in ð3=2Þe
the high-temperature upper part of the slug, because keff ¼ k0 1 Pg ½7
there is less solid fission product swelling and cracks
and after sodium infiltration
and/or open pores can easily collapse at a lower
" #
compression stress due to its higher creep rate. The 3=2
1 kNa =k0 P
cladding diametral change profile may reflect this keff ¼ k0 1 3 Na 1 Pg ½8
2=e þ ð3 2=eÞðkNa =k0 Þ 1 Pg
FCMI behavior.
The FCMI behavior described above is for where keff is the effective thermal conductivity of the
swelling-resistant steel cladding. In the case of aus- irradiated fuel slug (W m1 K1), k0 is the thermal
tenitic steel cladding, lower FCMI is predicted conductivity of the unirradiated fuel slug (W m1
because irradiation swelling of the cladding increases K1), kNa is the thermal conductivity of sodium
the cladding diameter. (W m1 K1), Pg is the volume fraction of gas-filled
pores, PNa is the volume fraction of sodium-filled
pores, and e is a constant (e ¼ 1.72). According to
3.01.4.6 Change in Fuel Slug Temperature eqn [7], for example, the effective conductivity of the
The formation of gas bubbles and cavitation at the fuel slug attaining 40% gas swelling without any
beginning of irradiation degrades the effective ther- sodium infiltration decreases down to 40–50% of
mal conductivity of the fuel slug.3,4 When open pores the unirradiated slug.
are formed as a result of the interconnection among
the bubbles, fission gas inside of the bubbles comes
3.01.4.7 Fuel–Cladding Chemical
out through the open pores and the bond sodium
Interaction
goes into some of the open pores.3,4 Partial infiltra-
tion of the bond sodium recovers the degraded effec- Part of lanthanide fission products agglomerate at the
tive thermal conductivity of the slug.3,4 This change peripheral region of the fuel slug and react with the
in thermal conductivity is supported by the tempera- inside of the cladding.3,68,71,93 The reaction region of
ture measurements for the slugs in instrumented the cladding is so brittle3,70 that it should be regarded
fuel pins.101,102 Bauer and Holland103 evaluated the as wastage. Above a threshold temperature, a liquid
effect of bond sodium infiltration on slug thermal phase is formed at the interface between the fuel slug
Metal Fuel 29
1.00 0.25
Fuel stack zone Extruded
bond Na
0.60 0.15
0.40 0.10
137Cs
0.20 0.05
0.00 0.00
700 800 900 1000 1100 1200 1300 1400 1500
Axial position of metal fuel pin (mm)
137
Figure 32 Cs and 106Ru g autoradiography of 2.5 at.% peak burnup U–Pu–Zr fuel pin from the METAPHIX test.
remaining part may precipitate in the fuel slug as a Table 8 Fission yield of Zr, Mo, Tc, Ru, Rh, and Pd,
result of the very low solubility of these elements in compared with composition of Fizzium (Fz)
U and Pu. In fact, the g (from 137Cs and 106Ru) Element Fission yield (wt%)90,98 Composition of Fz
autoradiography of an irradiated metal fuel pin75 pre- (wt%)22
sented in Figure 32 indicates the transfer of Cs to the
Zr 19.7 28
bond sodium in the plenum region. According to Mo 23.9 27.5
the binary alloy phase diagrams37 (also see Chapter Tc 6.3 0
2.05, Phase Diagrams of Actinide Alloys), noble Ru 23.1 29.5
metals form intermetallic compounds with U and Rh 7 5
Pu, and Zr and Mo dissolve in U and Pu. Figure 32 Pd 20 10
also shows that the Ru distribution is similar to the
burnup profile, which means that Ru is immobile.
Based on the above consideration, Zr, Mo, Tc, Ru, Fz from the U–Pu–10 wt% Fz alloys, are calculated
Rh, and Pd of fission products will affect the solidus by using eqn [1], as shown in Table 9. The table
temperature of U–Zr and U–Pu–Zr fuel alloys. The indicates that 10 wt% Fz addition decreases the soli-
effect of the fission products on the solidus tempera- dus temperature of U–Pu alloys by 100 K. Assuming
ture can be estimated as follows90: When 1 wt% of that the decrease in the solidus is proportional to the
heavy metals (U and Pu) fission, the total production amount of Fz addition, 0.36 wt% Fz addition leads to a
of Zr, Mo, Tc, Ru, Rh, and Pd is 0.36 wt% of the initial decrease of 3.6 K in the solidus temperature of U–Pu
heavy metal. A breakdown of the production is shown alloys. Therefore, it can be estimated that the effect of
in Table 8. Studies on U–Pu–Fz alloys were per- burnup on the solidus temperature of the fuel slug is
formed in the United States,22,26 where Fz stands for 3.6 K/at.%.
fizzium, the alloy of Zr, Mo, Ru, Rh, and Pd having the
composition22 shown in Table 8. The Fz composition
3.01.4.9 Behavior of Breached Fuel Pins
is similar to the fission yield shown in Table 8.
Therefore, the effect of fission product accumulation As can be understood from the use of bond sodium,
on the solidus temperature of U–Pu alloys can be fuel alloys show good compatibility with sodium.
assumed to be the same as that of Fz addition. The Even when a metal fuel pin is breached, the fuel
solidus temperatures of U–Pu and U–Pu–10 wt% Fz slug does not react with the sodium reactor coolant
alloys have been reported, and are shown in Table 9.22 and the breach site is not enlarged. On the other
The solidus temperatures of the U–Pu alloys, the hand, oxide fuel pellets react with the sodium reactor
compositions of which are calculated by subtracting coolant and form low-density reaction products,
Metal Fuel 31
which enlarge the breach site. In order to confirm the product (Cs) behavior similar to those of MA-free
benign behavior of breached metal fuel pins, a num- U–Pu–Zr fuel pins. This means that the addition of a
ber of run-beyond-cladding-breach (RBCB) small amount of MA to the U–Pu–Zr matrix does
tests3,4,111 were conducted in EBR-II. In the tests, not affect the irradiation behavior of the fuel.
the claddings of irradiated fuel pins were thinned The MA-specific irradiation phenomena observed
mechanically to promote cladding breach, and were so far have been Am migration along with uranium
then reinserted into the reactor. The results of these and zirconium migration and helium gas release.
tests confirmed the following benign behavior of Am-rich phases were present only in the uranium-
breached metal fuel pins.3,4,111 depleted central and outer zones after irradiation,
At the time of cladding breach, bond sodium and as shown in Figure 21.59 Release of the helium
fission gas were expelled through the breach site. gas produced in MA transmutation was higher than
Delayed neutron (DN) precursors came out along the fission gas release in both the X50159 and
with the bond sodium, and a short DN signal was METAPHIX76 tests.
detected by a monitoring system. The cesium dissolved The AFC-1 test result showed that the fission gas
in the bond sodium was also expelled. The fission gas release behavior of the nonfertile or low-fertile fuel
released into the reactor coolant was monitored. After alloys (see Table 7) followed the same trend as that
the expulsion of bond sodium and fission gas, fission gas of the U–Pu–Zr-based fuel alloys when correlated
expulsion was detected intermittently. The crack at the with fission density.78
breach site remained small even over 54–233 days of One of the issues in MA-bearing fuel performance
RBCB operation, because the fuel pin was depressur- is FCCI because MA, particularly Am, is expected
ized and FCMI was kept at a low level. The weight loss to behave like lanthanide elements, which react with
of the breached fuel pin was accounted for by the Fe-based cladding materials. However, no FCCI data
expulsion of the bond sodium, cesium, and fission gas; are available at this point because the inner cladding
fuel loss was negligible. The HT-9 cladding fuel pin temperature in the X501 test was too low (<540 C)
(13.5 at.% burnup) exhibited crack extension and wid- to observe a sufficient level of FCCI. From META-
ening to some extent, but no fuel alloy was extruded PHIX-2 or METAPHIX-3, FCCI data will be
over 150 days of RBCB operation. obtained in the future.
As described in Section 3.01.4.7, the expul-
sion of bond sodium from the breach site would
3.01.4.11 Factors Controlling Fuel Lifetime
deteriorate the thermal conductivity of the fuel
slug and raise the fuel slug temperature. This may Important factors controlling the lifetime of metal
promote the supply of lanthanides and intensify fuel pins of low smear density (<75%) during
the wastage.94 In fact, cladding wastage by lanthanide steady-state irradiation are plenum pressure rise due
fission products after 150–200 days of RBCB operation to fission gas accumulation and cladding wastage by
was approximately twice that in unbreached fuel pins lanthanide fission products. The plenum gas pressure
at comparable burnup and cladding temperature.4 produces circumferential tensile stress in the clad-
ding. The decrease in the effective cladding thickness
due to the wastage raises the cladding stress. The
3.01.4.10 Behavior of MA-Bearing
cladding at the higher temperature upper part of the
Metal Fuel
fuel slug has lower strength and more wastage than
U–Pu–10 wt% Zr fuel pins containing less than any other part. Accordingly, the margin to cladding
5 wt% MA irradiated in the X50159 and META- failure is minimized at the upper part of the slug.
PHIX75,76 tests consistently showed microstructure, In the irradiation test of U–10 wt% Zr fuel pins
constituent migration, fission gas release, and fission clad with HT-9, two fuel pins were breached at
32 Metal Fuel
9.5 at.% burnup; both breach sites were at the upper power.10 Every cladding breach site was located
part of the slug.94 A maximum wastage of nearly near the top of the fuel slug.10 Examination of intact
100 mm in depth was found at the upper part of the fuel pins after the test confirmed that the melting
slug in the sibling pins.94 region extended over the upper half of the slug
FCMI may increase at a higher (>10 at.%) burnup, and >90% of the cross section near the top of the
particularly in the case of swelling-resistant ferritic fuel slug.4,10 The test conditions and the results are
steel cladding. However, no FCMI-induced cladding summarized in Table 10.4,5
failure has been found, even in the highest peak burnup Out-of-pile transient tests was conducted in the
of 19.4 at.% attained to date. This is because, as dis- whole-pin furnace (WPF) system12,104 in the Alpha–
cussed in Section 3.01.4.5, FCMI may be at a lower Gamma Hot Cell Facility (AGHCF) at ANL-Illinois
level at the upper part of the slug, where the margin to to evaluate fuel behavior in loss-of-flow (LOF)
cladding failure is minimized. Therefore, FCMI may events, where fuel pin power decreases to a decay
not be an important factor controlling the fuel lifetime. heat level and the radial temperature profile in the
fuel slug is flattened. In the test, an irradiated fuel pin
was put into a 316SS capsule and externally heated in
3.01.5 Transient Behavior the furnace.12 The cladding and plenum tempera-
tures were maintained at a constant value in the
Metal fuel also shows characteristic behavior in tran- range of 650–820 C in the six FM series test.112
sient conditions. For example, when the cladding In the other test, these temperatures were varied to
temperature exceeds a certain threshold value, the simulate an LOF event.12,112
interdiffusion between the fuel slug and cladding Out-of-pile tests in the fuel behavior test appara-
causes liquefaction of the diffusion zone and exten- tus (FBTA)12,104,105 in AGHCF were conducted to
sive wastage of the cladding inside. The molten fuel examine compatibility between the fuel slug and
slug expands axially because of rapid expansion of cladding at elevated temperatures by externally heat-
the fission gas bubbles without producing excessive ing irradiated fuel pin segments. The test data were
cladding stress. These phenomena during transient helpful in understanding FCCI behavior in metal
events are important in evaluating the margin to fuel fuel, particularly liquid-phase formation. Compati-
pin failure and failed fuel pin behavior. bility between the fuel slug and cladding has also
In this section, transient tests for metal fuel are been studied by conducting out-of-pile isothermal
reviewed briefly, the linear-power-to-melting is dis- annealing tests for diffusion couples consisting of
cussed, and then the transient behavior of metal fuel unirradiated fuel alloys and Fe-based alloys.112–120
is described.
3.01.5.2 Linear-Power-to-Melting
3.01.5.1 Transient Tests
When metal fuel transient behavior is discussed, it
A series of tests, called the M-series, was performed may be important to know the linear power required
at the Transient Reactor Test Facility (TREAT) in to cause fuel centerline melting, which is called
ANL-Idaho (currently INL) to study the cladding ‘linear-power-to-melting’122 for the metal fuel.
failure threshold and other safety-related behavior This subsection describes the method for a conserva-
of metal fuel during transient overpower (TOP) tive evaluation of the linear-power-to-melting.90,91
events.10 In the six M-series tests, a total of 15 pre- The linear-power-to-melting for metal fuel is
irradiated fuel pins of U–5 wt% Fs clad with 316SS, evaluated on the basis of calculations of the tempera-
U–19 wt% Pu–10 wt% Zr clad with D9, and ture distribution and solidus temperature of a fuel
U–10 wt% Zr clad with HT-9 were subjected to slug, which are influenced by fuel specifications, irra-
overpower conditions simulating TOP events.10 The diation conditions, and irradiation behavior, such as
sodium loop that carries two or three fuel pins for the fuel alloy composition, smear density, axial linear
each test was installed in TREAT.10 During an expo- power profile, inlet coolant temperature, gas swelling,
nential power rise in an 8 s period of the TREAT fuel constituent migration, bond sodium infiltration
reactor, the fuel pin power increased up to four times into the swollen fuel slug, and solid fission product
the nominal power, and the peak coolant temperature accumulation.
was about 1200 K.10 Cladding failure occurred for The radial temperature distribution at a certain
five fuel pins at nearly four times the nominal axial position of a fuel slug is calculated by using a
Metal Fuel 33
Test Fuel composition (wt%) Cladding Peak burnup(at.%) Peak overpower Cladding Maximum axial
(normalizeda) failure expansion (%)
Source: Hofman, G. L.; Walters, L. C.; Bauer, T. H. Prog. Nucl. Energy 1997, 31, 83; Crawford, D. C.; Porter, D. L.; Hayes, S. L. J. Nucl.
Mater. 2007, 371, 202.
a
Normalized by nominal conditions: peak linear power of 40 kW/m, 630 K inlet, and 150 K coolant temperature rise.
b
Test fuel pins were subjected to twice over power transients.
c
Expansion may have been caused by localized sodium bond boiling.
one-dimensional steady-state heat conduction model. content of the Zr-depleted zone is assumed to
The thermal diffusivity of metal fuel is small enough be 2.5 at.% based on the irradiation data,84,85,86 and
to use the steady-state model even for analysis of then the solidus temperature of this zone is calcu-
overpower reactor events such as the TREAT test. lated by using eqn [1]. During irradiation, the solidus
For obtaining a conservative estimate of the fuel slug temperature may be decreased because of the accu-
temperature, the influence of irradiation behavior is mulation of fission products. A solidus temperature
treated as follows: The effective thermal conductivity decrease rate of 3.6 K/at.% burnup is assumed, as
of the fuel slug is calculated from eqn [7], which does discussed in Section 3.01.4.8.
not incorporate thermal conductivity recovery due to Figure 33 shows the linear-power-to-melting
bond sodium infiltration. Porosity Pg in eqn [7] is evaluated in the case of a nominal peak linear
conservatively estimated from the smear density and power of 380 W cm1, a peak cladding inner temper-
the irradiation data of fuel slug axial elongation. The ature of 923 K, an inlet coolant temperature of 668 K,
porosity is assumed to be uniform and constant from and a fuel smear density of 75%. For a Pu content
the beginning of irradiation. As-fabricated uniform of <20 wt%, the linear-power-to-melting of U–Pu–
fuel composition is assumed without considering fuel 10 wt% Zr fuel is higher than 600 W cm1.
constituent migration, because it has only a minor This means that the U–Pu–10 wt% Zr fuel has
influence on fuel slug temperature, as stated in a sufficient margin to fuel slug melting. As also
Section 3.01.4.2. shown in Figure 33, the U–Pu–3 wt% Zr fuel has
A conservative estimate of the solidus temperature a linear-power-to-melting of >550 W cm1 for a
is obtained as follows: As a result of fuel constituent Pu content of <15 wt%. This indicates that the
migration, Zr is enriched in the high-temperature U–Pu–3 wt% Zr fuel will be feasible for a large-size
(>970 K) central region where the bcc g single reactor core from the standpoint of the margin to fuel
phase is formed. The solidus temperature in this slug melting.
region may be higher than that of the as-fabricated
fuel slug, but is evaluated on the basis of the
3.01.5.3 Liquefaction at the Fuel–Cladding
as-fabricated fuel composition for conservativeness.
Interface
On the other hand, a Zr-depleted zone is formed at
a temperature range of approximately 860–940 K FCCI during steady-state irradiation is dominated by
(in the case of U–20 wt% Pu–10 wt% Zr). The Zr the reaction of lanthanide fission products with the
34 Metal Fuel
800
U–Pu–10 wt% Zr
U–Pu–6 wt% Zr
700 U–Pu–3 wt% Zr
600
500
400
300
0 5 10 15 20 25 30
Pu content (wt%) (a) (b)
400
No liquefaction
U–Zr/Fe or Fe–Cr:
Ogata et al.117 and
760 Nakamura et al.118
U–Pu–Zr/Fe: Ogata et al.119
620 Ambiguous
test with the WPF system, the HT-9–clad U–19 wt% Cladding breach
Cladding
Pu–10 wt% Zr fuel pin of 3.0 at.% peak burnup was penetration
heated to 820 C and was breached 112 min later near
the top of the fuel slug.12 The cross section of this fuel
pin at the breach site is shown in Figure 38,12 indi-
cating that liquefaction covers the whole cross sec-
tion. In this test, an axial extrusion of 14 mm (4%
of the fuel slug length) was measured.12 This type of
liquefaction is not an instantaneous phenomenon
Resolidified
because times on the order of minutes are required fuel
for a sufficient amount of Fe to diffuse into the fuel
slug,12 whereas the extrusion by fuel slug centerline
1 mm
melting in a TOP event is instantaneous. Foamed fuel
Figure 38 Cross section near the cladding breach site
3.01.5.5 Fuel Pin Failure Mechanism of U–19 wt% Pu–10 wt% Zr fuel pin of 3.0 at.% peak
burnup after being held at 820 C for 112 min (FM-2 test).
Fuel pin failure in transient events is caused by a Reproduced from Liu, Y. Y.; Tsai, H.; Billone, M. C.;
combination of the plenum gas pressure rise due to Holland, J. W.; Kramer, J. M. J. Nucl. Mater. 1993, 204, 194.
coolant temperature increase and FCCI (reaction with
lanthanide fission products during steady-state irradi- system, three of four failed fuel pins showed cladding
ation plus liquid-phase penetration).10,12 FCMI is less breach near the top of the slug112; the other failed pin
important even when the fuel slug melts, as stated in had a breach site at the gas plenum region, the cladding
the previous section. This mechanism leads to cladding temperature of which was exceptionally higher than
breach near the top of fuel slug, where the cladding that of the fuel slug section.12
strength is lower and FCCI is more significant than in The dominating factor in the fuel pin failure mech-
the other cladding part. In fact, every cladding breach anism will depend on the fuel burnup and the event
in the TREAT M-series tests occurred near the top of type. A high-burnup fuel pin will burst as a result
the fuel slug.10 In the FM-series tests in the WPF of higher plenum gas pressure before liquid-phase
Metal Fuel 37
penetration develops. A low-burnup fuel pin will take operation in a hot cell. For MA-bearing metal fuel
more time before the cladding bursts, and liquid-phase fabrication, however, other methods such as gravity
penetration will develop during that time. In the case casting are now being examined from the standpoint
of a transient event where the cladding temperature of suppressing Am evaporation.
exceeds 1350 K, the accelerated penetration rate as Steady-state irradiation, transient, and out-of-pile
expressed by eqn [10] will cause cladding failure in tests extensively performed under the IFR program
a very short time. have produced a great amount of knowledge about
The HT-9–clad U–10 wt% Zr fuel pin of 2.9 at.% the irradiation behavior of metal fuel. A high-burnup
peak burnup exceptionally survived 4.8 times the capability was demonstrated, and a sufficient margin
nominal power transient in the TREAT M-7 test.10 to fuel failure in accidental events was confirmed.
A cross section of the fuel pin after the test showed Fuel behavior in severe accidents, such as in-pin
less cladding penetration than expected from eqn axial extrusion and rapid ejection–dispersion of mol-
[10]. One of the reasons for this lower level of pene- ten fuel alloy, was also characterized, which would
tration could be low interdiffusivity of the elements enhance the safety performance of the reactor core.
in the U–10 wt% Zr slugs,10 as suggested by its high Recent metal fuel development activities have been
solidus temperature (>1500 K). devoted to MA-bearing metal fuel. In metal fuel
performance, there seem to be no crucial issues so far.
For the future commercialization of metal fuel,
3.01.5.6 Failed Fuel Behavior
further tests are recommended, such as in-pile
In the failed fuel pins in the TREAT M-series tests, the TOP-simulating tests for high-burnup (>10 at.%)
molten fuel alloy was ejected rapidly through the U–Pu–Zr fuel pins, and in-pile LOF-simulating
breach site and dispersed upward with the coolant.10 tests, which have not been covered in the IFR pro-
The driving forces of fuel ejection could be expansion gram. Modeling and simulation, as well as irradiation
of trapped fission gas,10 bond sodium vapor pressure,10 data accumulation, is important for a better under-
and plenum gas flow toward the breach site. The post- standing of metal fuel irradiation behavior. Advanced
failure coolant flow rate was about two-thirds of the metal fuel concepts such as lined cladding are encour-
prefailure value, indicating only partial flow blockage.10 aged for the development of high-performance fast
Ejection and dispersion of molten fuel alloy were also reactors.
observed in the FM-series test in the WPF system.12
Acknowledgments
3.01.6 Summary and Future
Development The author expresses his respect for the extensive and
pioneering development of metal fuel at Argonne
Physical and mechanical property data on U–Zr and National Laboratory and Idaho National Laboratory
U–Pu–Zr alloys have been accumulated to date. in the United States from the 1960s to date. A signifi-
These data are useful for understanding the main cant part of this chapter consists of their fruitful results.
features of metal fuel irradiation behavior. However,
accumulation of property data is preferred for design-
ing commercialized fast reactor fuel. Irradiation References
behavior modeling also requires more data, particu-
larly on the diffusion properties for the constituent 1. Stevenson, C. E. The EBR-II Fuel Cycle Story; American
migration model and on the mechanical properties Nuclear Society: La Grange Park, IL, 1987.
2. Walters, L. C.; Seidel, B. R.; Kittel, J. H. Nucl. Technol.
for the fuel slug deformation model. The development 1984, 65, 179.
of high-burnup and MA-bearing fuels will require 3. Hofman, G. L.; Walters, L. C. In Nuclear Materials, Part 1;
thermodynamic assessment of multicomponent alloy Frost, B. R. T., Ed.; Material Science and Technology,
A Comprehensive Treatment; Cahn, R. W., Haasen, P.,
systems: U–Pu–Zr plus fission products and/or MA. Kramer, E. J., Eds.; VCH Verlagsgesellschaft: Weinheim,
Injection casting is a mature technology for fuel 1994; Vol. 10A, Chapter 1, pp 1–43.
slug fabrication, which is cost efficient (or simple), 4. Hofman, G. L.; Walters, L. C.; Bauer, T. H. Prog. Nucl.
Energy 1997, 31, 83.
suitable for remote operation, and capable of mass 5. Crawford, D. C.; Porter, D. L.; Hayes, S. L. J. Nucl. Mater.
production. The United States has significant experi- 2007, 371, 202.
ence in injection casting of fuel slugs, even by remote 6. Chang, Y. I. Nucl. Technol. 1989, 88, 129.
38 Metal Fuel
7. Till, C. E.; Chang, Y. I. Progress and status of the integral 33. Kurata, M. Calphad 1999, 23(3–4), 305–337.
fast reactor (IFR) fuel cycle development. In Proceedings 34. Brandes, E. A., Ed. Smithells Metals Reference Book,
of the International Conference on Fast Reactor and 6th ed.; Butterworth: London, 1983.
Related Fuel Cycles, Kyoto, Japan, 28 Oct–1 Nov 1991. 35. Maeda, A.; Suzuki, Y.; Ohmichi, T. J. Alloys Compd.
8. Leggett, R. D.; Walters, L. C. J. Nucl. Mater. 1993, 204, 1992, 179, L21–L24.
23. 36. Ogata, T. J. Nucl. Sci. Technol. 2002, (Suppl. 3), 675.
9. Tsai, H.; Cohen, A. B.; Billone, M. C.; Neimark, L. A. 37. Massalski, T. B., Ed. Binary Alloy Phase Diagrams, 2nd
In Proceedings of the 3rd JSME/ASME Joint International ed.; ASM International: Materials Park, OH, 1990.
Conference on Nuclear Engineering, Kyoto, Japan, 1995; 38. Billone, M. C.; Liu, Y. Y.; Gruber, E. E.; Hughes, T. H.;
Vol. 2, p 849. Kramer, J. M. In Proceedings of the International
10. Bauer, T. H.; Wright, A. E.; Robinson, W. R.; Holland, J. Conference on Reliable Fuels for Liquid Metal Reactors,
W.; Rhodes, E. A. Nucl. Technol. 1990, 92, 325. Tucson, AZ, 7–11 Sept; American Nuclear Society:
11. Cohen, A. B.; Tsai, H.; Neimark, L. A. J. Nucl. Mater. La Grange Park, IL, 1986.
1993, 204, 244. 39. Kurata, M.; et al. CRIEPI Report T92005; Central
12. Liu, Y. Y.; Tsai, H.; Billone, M. C.; Holland, J. W.; Research Institute of Electric Power Industry: Japan,
Kramer, J. M. J. Nucl. Mater. 1993, 204, 194. 1992.
13. Inoue, T.; Sakata, M.; Miyashiro, H.; Matsumura, T.; 40. Gruber, E. E.; Kramer, J. M. In Radiation-Induced
Sasahara, A.; Yoshiki, N. Nucl. Technol. 1991, 93, 206. Changes in Microstructure;13th International
14. Yokoo, T.; Sakata, M.; Inoue, T.; Kang, J.; Suzuki, A. Symposium (Part I); 1987; ASTM STP 955, p 432.
Nucl. Technol. 1996, 116, 173. 41. Ogata, T.; Akabori, M.; Ogawa, T. Measurement Test on
15. Carmack, W. J.; Porter, D. L.; Chang, Y. I.; et al. J. Nucl. Creep Strain Rate of Uranium–Zirconium Solid Solutions,
Mater. 2009, 392, 139. JAERI-Tech 96-052; Japan Atomic Energy Research
16. Ogata, T.; Nakamura, K.; Ohta, H.; Kurata, M.; Yokoo, T. Institute: Japan, 1996.
Research and development of fast reactor metal fuel by 42. Robinson, S. L.; Sherby, O. D.; Armstrong, P. E. J. Nucl.
CRIEPI. In International Conference on Fast Reactors and Mater. 1973, 46, 293.
Related Fuel Cycles (FR’09), Kyoto, Japan, 7–11 Dec 43. Adda, Y.; Philibert, J.; Faraggi, H. Rev. Metall. 1957, 54,
2009. 597.
17. Lee, C. B.; Lee, B. O.; Lee, C. T.; et al. Status of 44. Ogata, T.; Akabori, M.; Itoh, A.; Ogawa, T. J. Nucl. Mater.
metallic fuel development for sodium-cooled fast 1996, 232, 125.
reactor. In Global 2009, Paris, France, 6–11 Sept 2009. 45. Akabori, M.; Itoh, A.; Ogawa, T.; Ogata, T. J. Alloys
18. Rough, F. A. Battelle Memorial Institute Report, BMI- Compd. 1998, 271–273, 597.
1030, 1955. 46. Adda, Y.; Mairy, C.; Andreu, J. L. Rev. Metall. 1960, 57,
19. Boucher, R.; Barthelemy, P. CEA-2531 (1964), ANL- 550.
TRANS-138; Argonne National Laboratory: Argonne, IL, 47. Petri, M. C.; Hins, A. G.; Sanecki, J. E.; Dayananda, M. A.
1964; translated by B. Blumenthal. J. Nucl. Mater. 1994, 211, 1.
20. Argonne National Laboratory, Metallurgy Division. 48. Petri, C.; Dayananda, M. A. J. Nucl. Mater. 1997, 240,
Annual Progress Report for 1965, ANL-7155, 1965. 131.
21. Argonne National Laboratory, Reactor Development 49. Kurata, M.; Inoue, T.; Sari, C. J. Nucl. Mater. 1994, 208,
Progress Report, ANL-7230, 1966. 144.
22. Kelman, L. R.; Savage, H.; Walter, C. M.; Blumenthal, B.; 50. Sohn, Y. H.; Dayananda, M. A.; Hofman, G. L.; Strain, R.
Dunworth, R. J.; Rhude, H. V. In Proceedings of the 3rd V.; Hayes, S. L. J. Nucl. Mater. 2000, 279, 317.
International Conference on Plutonium, London, 22–26 51. Burkes, D. E.; Fielding, R. S.; Porter, D. L.;
Nov 1965; Barns and Noble: New York, 1967; p 458. Crawford, D. C.; Meyer, M. K. J. Nucl. Mater. 2009, 389,
23. Harbur, D. R.; Anderson, J. W.; Maraman, W. J. Studies 458.
on the U–Pu–Zr Alloy System for Fast Breeder Reactor 52. Burkes, D. E.; Fielding, R. S.; Porter, D. L. J. Nucl. Mater.
Applications LA-4512; Los Alamos Scientific Laboratory: 2009, 392, 158.
Los Alamos, NM, 1970. 53. Argonne National Laboratory, Unpublished document.
24. Touloukian, Y. S.; et al. Ed. Thermophysical Properties of 54. Ogata, T.; Tsukada, T. Engineering-scale development of
Matter, Thermal Conductivity, Metallic Elements and injection casting technology for metal fuel cycle. In
Alloys. IFI/Plenum: New York, 1970; Vol. I. Global 2007, Boise, ID, 9–13 Sept 2007.
25. O’Boyle, D. R.; Dwight, A. E. In Proceedings of the 4th 55. Sato, K.; Fujioka, T.; Nakabayashi, H.; Kitajima, S.;
International Conference on Plutonium and Other Yokoo, T.; Inoue, T. Conceptual design on an integrated
Actinides, Santa Fe, NM; Material Society: New York, metallic fuel recycle system. In Global 2003, New
1970; p 720. Orleans, LA, 16–20 Nov 2003.
26. Kittel, J. H.; Ayer, J. E.; Beck, W. N.; et al. Nucl. Eng. Des. 56. Nakagawa, T.; Ogata, T.; Tokiwai, M. Trans. Am. Nucl.
1971, 15, 373. Soc. 1989, 60, 315.
27. Argonne National Laboratory. Chemical Technology 57. Lee, C. T.; Oh, S. J.; Ryu, H. J.; et al. Casting technology
Division Annual Technical Report for 1986, ANL-87-19; development for SFR metallic fuel. In Global 2009, Paris,
1987. France, 6–11 Sept 2009.
28. Takahashi, Y.; Yamawaki, M.; Yamamoto, K. J. Nucl. 58. Meyer, M. K.; Hayes, S. L.; Carmack, W. J.; Tsai, H.
Mater. 1988, 154, 141. The EBR-II X501 Minor Actinide Burning Experiment,
29. Leibowitz, L.; Veleckis, E.; Blomquist, R. A.; Pelton, A. D. INL/EXT-08-13835; Idaho National Laboratory: Idaho
J. Nucl. Mater. 1988, 154, 145. Falls, ID, 2008.
30. Takahashi, Y.; Yamamoto, K.; Ohsato, T.; Shimada, H.; 59. Meyer, M. K.; Hayes, S. L.; Carmack, W. J.; Tsai, H.
Terai, T.; Yamawaki, M. J. Nucl. Mater. 1989, 167, 147. J. Nucl. Mater. 2009, 392, 176.
31. Matsui, T.; Natsume, T.; Naito, K. J. Nucl. Mater. 1989, 60. Kim, Y. S.; Hofman, G. L.; Yacout, A. M. J. Nucl. Mater.
167, 152. 2009, 392, 164.
32. Kurata, M.; Ogata, T.; Nakamura, K.; Ogawa, T. J. Alloys 61. Trybus, C. L.; Sanecki, J. E.; Henslee, S. P. J. Nucl.
Compd. 1998, 271–273, 636. Mater. 1993, 204, 50.
Metal Fuel 39
62. Trybus, C. L. J. Nucl. Mater. 1995, 224, 305. 82. Donnet, L.; Jorion, F.; Drin, N.; et al. The FUTURIX-FTA
63. Ogata, T.; Mizuno, T. Directions of metal fuel experiment in PHENIX: Status of fuel fabrication. In
development for fast reactors. In Global 2009, Paris, Global 2005, Tsukuba, Japan, 9–13 Oct 2005.
France, 6–11 Sept 2009. 83. Jaecki, P.; Pillon, S.; Warrin, D.; et al. Update on the
64. Nakamura, K.; Ogata, T.; Kato, T.; Nakajima, K.; FUTURIX-FTA experiment in PHENIX. In Global 2005,
Arai, Y. Fabrication of metal fuel slug for an irradiation Tsukuba, Japan, 9–13 Oct 2005.
test in JOYO. In Global 2009, Paris, France, 6–11 84. Porter, D. L.; Lahm, C. E.; Pahl, R. G. Metall. Trans. 1990,
Sept 2009. 21A, 1871.
65. Marsden, K. Report on Development of Concepts for 85. Hofman, G. L.; Hayes, S. L.; Petri, M. C. J. Nucl. Mater.
the Advanced Casting System in Support of the 1996, 227, 277.
Deployment of a Remotely Operable Research Scale Fuel 86. Kim, Y. S.; Hofman, G. L.; Hayes, S. L.; Sohn, Y. H.
Fabrication Facility for Metal Fuel, INL/EXT-07–12469; J. Nucl. Mater. 2004, 327, 27.
Idaho National Laboratory: Idaho Falls, ID, 2007. 87. Ogawa, T.; Iwai, T.; Kurata, M. J. Less Common Met.
66. Nakamura, K.; Kato, T.; Ogata, T.; Nakajima, K.; Iwai, T.; 1991, 175, 59.
Arai, Y. U–Pu–Zr metal fuel fabrication for irradiation test 88. Ishida, M.; Ogata, T.; Kinoshita, M. Nucl. Technol. 1993,
at JOYO. In International Conference on Fast Reactors 104, 37.
and Related Fuel Cycles (FR’09), Kyoto, Japan, 7–11 Dec 89. Kim, Y. S.; Hayes, S. L.; Hofman, G. L.; Yacout, A. M.
2009. J. Nucl. Mater. 2006, 359, 17.
67. Hofman, G. L.; Pahl, R. G.; Lahm, C. E.; Porter, D. L. 90. Ogata, T.; Mizuno, T. Estimation of Linear Power to
Metall. Trans. 1990, 21A, 517. Melting of Metal Fuel, CRIEPI Report L08001; Central
68. Pahl, R. G.; Porter, D. L.; Lahm, C. E.; Hofman, G. L. Research Institute of Electric Power Industry: Japan,
Metall. Trans. 1990, 21A, 1863. 2008.
69. Pahl, R. G.; Porter, D. L.; Crawford, D. C.; Walters, L. C. 91. Ogata, T.; Mizuno, T. Evaluation of linear-power-
J. Nucl. Mater. 1992, 188, 3. tomelting of fast reactor metal fuel. In Proceedings of the
70. Beck, W. N.; Fousek, R. J.; Kittel, J. H. The irradiation American Nuclear Society 2009 Annual Meeting, Atlanta,
behavior of high-burnup uranium-plutonium alloy GA, 14–18 Jun 2009.
prototype fuel elements; Argonne National Laboratory, 92. Barnes, R. S. J. Nucl. Mater. 1964, 11, 135.
Report ANL-7388, 1968. 93. Pahl, R. G.; Wisner, R. S.; Billone, M. C.; Hofman, G. L.
71. Murphy, W. F.; Beck, W. N.; Brown, F. L.; Koproeski, B. In Proceedings of the International Fast Reactor Safety
J.; Neimark, L. A. Postirradiation examination of U-Pu-Zr Meeting, Snowbird, UT, 12–16 Aug 1990; Vol. IV. p 129.
fuel elements irradiated in EBR-II to 4.5 atomic percent 94. Pahl, R. G.; Lahm, C. E.; Hayes, S. L. J. Nucl. Mater.
burnup; Argonne National Laboratory, Report ANL-7602, 1993, 204, 141.
1969. 95. Tsai, H.; Neimark, L. A. In Proceedings of the
72. Ohta, H.; Yokoo, T.; Ogata, T.; et al. Irradiation International Conference on Design and Safety of
experiment on fast reactor metal fuels containing minor Advanced Nuclear Power Plants, Kyoto, Japan, 25–29
actinides up to 7 at.% burnup. In Global 2007, Boise, ID, Oct 1992; p 28. 2–1.
9–13 Sept 2007. 96. Rest, J. J. Nucl. Mater. 1993, 207, 192.
73. Breton, L.; Garces, E.; Desjardins, S.; et al. METAPHIX-1 97. Crawford, D. C.; Hayes, S. L.; Pahl, R. G. Trans. Am.
non-destructive PIE in the irradiated elements cell of Nucl. Soc. 1994, 71, 178.
PHENIX. In Global 2007, Boise, ID, 9–13 Sept 2007. 98. Ogata, T.; Yokoo, T. Nucl. Technol. 1999, 128, 113.
74. Ogata, T.; Nakamura, K.; Ohta, H. Trans. ANS 2008, 99. Yacout, A. M.; Salvatores, S.; Orechwa, Y. Nucl. Technol.
98, 977. 1996, 113, 177.
75. Ohta, H.; Ogata, T.; Yokoo, T.; et al. Nucl. Technol. 2009, 100. Integral Fast Reactor Program Annual Progress
165, 96. Reports FY 1991; Argonne National Laboratory, Report
76. Ohta, H.; Ogata, T.; Yokoo, T.; et al. Post-irradiation ANL-IFR-169, 1992.
examinations on fast reactor metal fuels containing minor 101. Beck, W. N.; Fousek, R. J. Trans. Am. Nucl. Soc. 1969,
actinides – fission gas release and metallography 12, 78.
of 2.5 at.% burnup fuels. In Global 2009, Paris, France, 102. Betten, P. R. Trans. Am. Nucl. Soc. 1985, 50, 237.
6–11 Sept 2009. 103. Bauer, T. H.; Holland, J. W. Nucl. Technol. 1995, 110, 407.
77. Hilton, B. A.; Hayes, S. L.; Meyer, M. K.; Crawford, D. C.; 104. Tsai, H. In Proceedings of the International Fast Reactor
Chang, G. S.; Ambrosek, R. The AFC-1AE and AFC-1F Safety Meeting, Snowbird, UT, 12–16 Aug 1990; Vol. II,
irradiation tests of metallic and nitride fuels for actinide p 257.
transmutation. In Global 2003, New Orleans, LA, 16–20 105. Tsai, H.; Liu, Y. Y.; Wang, D.-Y.; Kramer, J. M.
Nov 2003. In Proceedings of the International Conference on
78. Hilton, B. A.; Porter, D. L.; Hayes, S. L. AFC-1 Fast Reactor and Related Fuel Cycles, Kyoto, Japan,
Transmutation Fuels Post-Irradiation Hot Cell Examination 28 Oct–1 Nov 1991; Atomic Energy Society of Japan:
4 to 8 at.% Final Report – Irradiation Experiments Japan, 1991.
AFC-1B, AFC-1F and AFC-1AE, INL/EXT-05–00785, 106. Inagaki, K.; Ogata, T. Trans. Am. Nucl. Soc. 2010, 102,
Rev. 1; Idaho National Laboratory: Idaho Falls, ID, 2006. 831.
79. Hilton, B. A.; Porter, D. L.; Hayes, S. L. Trans. ANS 2008, 107. Tokiwai, M.; Yuda, R.; Ohuchi, A.; Amaya, M. J. Nucl. Sci.
98, 773. Technol. 2002, 1(Suppl. 3), 913.
80. MacLean, H. J.; Hayes, S. L. Irradiation of metallic and 108. Keiser, D. D.; Cole, J. I. An evaluation of potential liner
oxide fuels for actinide transmutation in the ATR. materials for eliminating FCCI in irradiated metallic
In Global 2007, Boise, ID, 9–13 Sept 2007. nuclear fuel elements. In Global 2007, Boise, ID, 9–13
81. Hayes, S. L. Irradiation of Metallic Fuels with Rare Earth Sept 2007.
Additions for Actinide Transmutation in the Advanced 109. Ryu, H. J.; Lee, B. O.; Oh, S. J.; Kim, J. H.; Lee, C. B.
Test Reactor – Experiment Description for AFC-2A and J. Nucl. Mater. 2009, 392, 206.
AFC-2B, INL/EXT-06–11707, Rev. 2; Idaho National 110. Yang, S. W.; Ryu, H. J.; Kim, J. H.; Lee, B. O.; Lee, C. B.
Laboratory: Idaho Falls, ID, 2007. J. Nucl. Mater. 2009, 392, 206.
40 Metal Fuel
111. Batte, G. L.; Hofman, G. L. In Proceedings of the 118. Nakamura, K.; Ogata, T.; Kurata, M.; Itoh, A.; Akabori, M.
International Fast Reactor Safety Meeting, Snowbird, UT, J. Nucl. Mater. 1999, 275, 246.
Aug 12–16; 1990; Vol. IV, p 207. 119. Ogata, T.; Nakamura, K.; Kurata, M.; Mignanelli, M. A.
112. Integral Fast Reactor Program; Annual Progress Report J. Nucl. Sci. Technol. 2000, 37(3), 244.
FY 1993, ANL-IFR-244, 1994. 120. Nakamura, K.; Ogata, T.; Kurata, M.; Mignanelli, M. A.
113. Hofman, G. L.; Hins, A. G.; Porter, D. L.; Leibowitz, L.; J. Nucl. Sci. Technol. 2001, 38(2), 112.
Wood, E. L. Chemical interaction of metallic fuel with 121. Ogata, T.; Nakamura, K.; Itoh, A.; Akabori, M. Trans. Am.
austenitic and ferritic stainless steel cladding. Nucl. Soc. 2010, 102, 829.
In Proceedings of the International Conference on 122. Walter, A. E.; Reynolds, A. B. Fast Breeder Reactors;
Reliable Fuels for Liquid Metal Reactors, Tucson, AZ, Pergamon: New York, 1981.
7–11 Sept 1986; pp 4–121. 123. Bauer, T. H.; Fenske, G. R.; Kramer, J. M. Cladding
114. Keiser, D. D., Jr.; Dayananda, M. A. J. Nucl. Mater. 1993, failure margins for metallic fuel in the integral fast reactor.
200, 229. In Proceedings of the 9th International Conference on
115. Keiser, D. D., Jr.; Dayananda, M. A. Metall. Mater. Trans. Structural Mechanics in Reactor Technology, Lausanne,
1994, 25A, 1649. Switzerland, 17–21 Aug 1987; Vol. C, p 31.
116. Keiser, D. D., Jr.; Petri, M. C. J. Nucl. Mater. 1996, 240, 51. 124. Walter, C. M.; Kelman, L. R. J. Nucl. Mater. 1966, 20, 314.
117. Ogata, T.; Kurata, M.; Nakamura, K.; Akabori, M.; Itoh, A. 125. Betten, P. R.; Bottcher, J. H.; Seidel, B. R. Trans. Am.
J. Nucl. Mater. 1997, 250, 171. Nucl. Soc. 1983, 45, 300.
3.02 Nitride Fuel
Y. Arai
Nuclear Science and Engineering Directorate, Japan Atomic Energy Agency, Ibaraki, Japan
3.02.1 Introduction 41
3.02.2 Fabrication of Nitride Fuel 43
3.02.2.1 Actinide Nitride Compounds 43
3.02.2.2 Preparation from Metal or Hydride 44
3.02.2.3 Carbothermic Reduction 44
3.02.2.4 Other Nitride Formation Processes 45
3.02.2.5 Nitride Pellet Fabrication 46
3.02.2.6 Nitride Particle Fabrication 47
3.02.3 Irradiation Behavior of Nitride Fuel 47
3.02.3.1 Irradiation Experience 47
3.02.3.2 Fuel Design 47
3.02.3.3 Chemical Forms of FP 48
3.02.3.4 Restructuring 49
3.02.3.5 FP Gas Release 50
3.02.3.6 Swelling and FCMI 50
3.02.3.7 Fuel–Clad Chemical Interaction 51
3.02.4 Reprocessing of Nitride Fuel 51
3.02.5 Outlook of Nitride Fuel 52
References 53
41
42 Nitride Fuel
dedicated fuel for MA transmutation systems such as the high neutron capture cross section of 14N
an accelerator-driven system (ADS), U-free nitride (99.6% abundance in natural nitrogen) deteriorating
fuel, such as (Pu,MA)N diluted by ZrN, has been neutron economics. However, the fuel fabrication
studied mainly in Japan. process has improved since the late 1980s, and the
At the beginning of the nuclear era, the develop- breeding ratio and doubling time have not been
ment of fast reactor fuel cycles was centered on the center of the development of fast reactor fuel
breeding ratio and doubling time. The reason was cycles. Furthermore, it was found that nitride fuel is
that metallic fuel, the binary or certain ternary alloy less hygroscopic in nature than carbide fuel, which
of U and Pu, was adopted in the first generation of will be advantageous for technological development.
fast reactors. The metallic fuel, however, had disad- It was also found that nitride fuel dissolves well in
vantages for commercial use, such as anisotropic nitric acid without any formation of Pu oxalate,
crystal structure, low melting temperature with which will be compatible with hydrochemical repro-
phase transformations, and high fission product (FP) cessing technology such as the PUREX process. So
gas-induced swelling. So a solid solution of uranium since the late 1980s, the global interest has moved
dioxide (UO2) and plutonium dioxide (PuO2), from carbide fuel to nitride fuel.
namely (U,Pu)O2 (MOX), has been a reference fuel We can find two distinguished monographs about
for fast reactors and used in many test and prototype nitride fuel: one is written by Matzke1 published in
reactors all over the world, although the breeding 1986 and the other by Blank2 in 1994. These mono-
ratio is smaller and the doubling time is longer than graphs describe nitride fuel and carbide fuel as MX-
those of metallic fuel. type fuel (X ¼ N or C) for fast reactors in detail from
On the other hand, nitride fuel, as well as carbide scientific and technological viewpoints. It should also
fuel, has the advantages of both metallic fuel and be mentioned that (U,Pu)N fuel with high Pu con-
oxide fuel as shown in Table 1. It has a high thermal tent, in which the Pu/(U þ Pu) molar ratio is roughly
conductivity and high metal atom density like metal- 0.45–0.55, was studied in France as a fast reactor fuel
lic fuel, while it has a high melting temperature and for incineration of Pu in the 1990s. The good disso-
isotropic crystal structure like oxide fuel. These char- lution in nitric acid and stable crystal structure even
acteristics led to the motivation for developing at high Pu content led to the potential CAPRA core
nitride fuel for fast reactors because the high thermal with (U,Pu)N fuel for incineration of Pu.3 Although
conductivity and high melting temperature allow a not being described in this chapter, another interest-
high linear power operation; alternatively, the large- ing aspect of nitride fuel was pointed by Lyon et al.,4
diameter fuel pins can be used for a given linear who indicated the superior safety margin in case of
power. The high metal atom density allows a low hypothetical loss of flow (LOF) and transient over-
fissile material inventory with flexible core design power (TOP) events.
and good neutron economics, leading to an improved In a space reactor program called SP-100 in the
breeding ratio and doubling time. United States, UN with highly enriched 235U was
However, the development of nitride fuel has chosen as a reference fuel because it has the most
fallen behind that of carbide fuel, which has similar favorable properties and will show the best perfor-
physical and chemical properties. The reason in- mance for space reactor fuels.5 An extensive work
cludes an unexploited fuel fabrication process and was carried out in SP-100 program and Hayes et al.
Table 1 Comparison of typical properties between oxide, metallic, and nitride fuels for fast reactors
a
At 0.1 MPa N2 pressure.
Nitride Fuel 43
summarized the physical,6 mechanical,7 transport,8 Table 2 Crystal structures and lattice parameters of
and thermodynamic properties9 of UN, while FP gas nitrides of Th, U, Np, Pu, Am, and Cm
release and swelling of UN were summarized by Compounds Structure Lattice parameter
Storms10 and Ross et al.,11 respectively. Ross et al.12 (nm)
also compiled and analyzed the thermal conductivity
ThN NaCl-type fcc 0.5167
data of UN. On the other hand, the diffusional and
Th3N4 Th3P4-type a ¼ 0.3871
mechanical properties were reviewed by Routbort hexagonal
et al.13 previously. c ¼ 2.7385
Since the late 1990s, the partitioning and transmu- UN NaCl-type fcc 0.4889
tation of MA has attracted global interest. It may a-U2N3 þ x Mn2O3-type bcc 1.0685
b-U2N3 x La2O3-type a ¼ 0.3696
contribute to the decrease of toxicity of high-level
hexagonal
radioactive waste (HLW) and the mitigation of burden c ¼ 0.5840
for its final disposal. Several transmutation systems UN2 x CaF2-type fcc 0.531
and MA-containing fuels have been proposed so far. NpN NaCl-type fcc 0.4899
Among them, the Japan Atomic Energy Agency PuN NaCl-type fcc 0.4905
AmN NaCl-type fcc 0.4995
( JAEA) proposed a subcritical ADS as a transmuta-
CmN NaCl-type fcc 0.5027
tion system and MA nitride fuel as a dedicated fuel for
transmutation.14 Besides the thermal and neutronic
properties, the mutual solubility of actinide mononi-
trides in a wide range of composition and combination U – Pu – N
N
becomes an advantage of the fuel with high MA con- 1000 ⬚C
tent. Fabrication of MA nitride fuel and its property α-U2N3 + (U,Pu)N + N2
measurements have been carried out in JAEA.15–18 β + (U,Pu)N
In this chapter, fabrication of nitride fuel and its 1 atm
irradiation behavior are summarized in Sections
3.02.2 and 3.02.3, respectively. A brief description α-U2N3
α + β + (U,Pu)N
β-U2N3
about reprocessing of spent nitride fuel is given in
UN PuN
Section 3.02.4, because the reprocessing technologies α-U2N3
+ (U,Pu)N
are closely related with the specific issues of nitride fuel
as 14C formation from natural nitrogen and 15N enrich-
ment. On the other hand, properties of nitride fuel are
described in Chapter 2.03, Thermodynamic and
Thermophysical Properties of the Actinide Nitrides. Solid Liquid
In addition, an outlook of nitride fuel is briefly given U Pu
in Section 3.02.5. Figure 1 Ternary U–Pu–N phase diagram at 1273 K.
Reproduced from Matzke, Hj. Science of Advanced LMFBR
Fuels; North-Holland: Amsterdam, 1986.
3.02.2 Fabrication of Nitride Fuel
3.02.2.1 Actinide Nitride Compounds
Holleck,19 is shown in Figure 1. The system is char-
Although nitride fuel usually stands for a mononi- acterized by a complete solubility of UN and PuN.
tride or its solid solution, such as UN and (U,Pu)N, It is considered that (U,Pu)N phase has a narrow
higher nitrides other than mononitrides exist in the composition range of the N/(U þ Pu) molar ratio.
Th–N and U–N binary systems. Table 2 summarizes Although Pu2N3 does not exist in the Pu–N system,
the crystal structures and lattice parameters of acti- a sesquinitride phase was identified in the U–Pu–N
nide nitrides reported in the Th–N, U–N, Np–N, system at a Pu/(U þ Pu) molar ratio of 0.15.23 As seen
Pu–N, Am–N, and Cm–N binary systems. The in Table 2, actinide mononitrides have the same
binary U–N and Pu–N, and ternary U–Pu–N sys- crystal structure with similar lattice parameters ex-
tems were investigated and reviewed by Holleck,19,20 cept for ThN, which leads to the mutual solubility.
Tagawa,21 and Potter22 in detail. In a mononitride lattice with NaCl-type structure,
The ternary U–Pu–N phase diagram at 1273 K small nitrogen atoms are incorporated into a dense
in Matzke’s monograph,1 originally calculated by face-centered cubic packing of metal atoms.
44 Nitride Fuel
(3 1 1)
Nitride preparation methods from metal or hydride
were investigated mainly in the 1960s. They include
(2 2 2)
the nitridation of U or Pu metal in N2 or NH3 at
CmN
CmN
AmN
AmN
NpN
NpN
about 1073–1173 K, arc-melting of U or Pu metal
PuN
PuN
under N2 pressure, nitridation of fine grained U or
Pu powder formed by the decomposition of hydrides
with N2 or NH3 and direct reaction of UH3 or PuH2.7
with N2 or NH3. In the case of uranium nitrides, the
60 61 62 63 64 65 66 67
products were often U2N3, which was subsequently
2q (deg)
decomposed to UN and N2.
These reactions are exothermic and should be Figure 2 X-ray diffraction pattern of (Np,Pu,Am,Cm)N
carried out slowly by temperature cycling for better prepared by carbothermic reduction.81 Reprinted with
permission from OECD/NEA (2007), Actinide and Fission
control of the products. Furthermore, these methods Product Partitioning and Transmutation, Ninth Information
necessitate a high-purity inert gas atmosphere, since Exchange Meeting, Nı̂mes, France, Sept 25–29, 2006,
the fine-grained powders of metal, hydride, and p 119, www.nea.fr.
nitride are chemically active and likely to react with
moisture and oxygen in air even at room temperature.
So it is difficult to apply the metal or hydride route to
NH3 stream can be applied although a higher initial
a technological fuel production line and these meth-
C/MO2 mixing ratio is necessary than that for the two-
ods were restricted to a laboratory scale experiment.
step reaction. For the preparation of UN and (U,Pu)N,
the atmosphere is changed to Ar or He from N2 or
3.02.2.3 Carbothermic Reduction N2–H2 at a temperature lower than about 1673 K to
prevent the formation of higher nitrides.
Carbothermic reduction is the most widely used pro-
In the case of preparation of solid solution such
cess for preparing nitride fuel. The starting material
as (U,Pu)N, both the reduction of the mixture of
is a dioxide and carbon, and the general reaction is
respective dioxides and the solid solution formation
expressed as
of respective mononitrides can be applied. Figure 2
MO2 þ 2C þ 0:5N2 ¼ MN þ 2CO ½I shows the X-ray diffraction (XRD) pattern of (Np,Pu,
Am,Cm)N prepared by the carbothermic reduction of
where M represents an actinide element, such as the mixture of respective dioxides, from which the
U and Pu. The mixture of dioxide and carbon is formation of quaternary mononitride solid solution
heated in N2 gas stream, usually at 1773–1973 K. It was confirmed.
is considered that the carbothermic reduction could Mechanism and kinetics of carbothermic reduc-
be applied in a technological production line as well tion were investigated by several authors, such as
as in a laboratory scale experiment, in contrast to the Muromura et al.,27–30 Lindemer,31 Greenhalgh32 and
metal or hydride route.24 Furthermore, homogeneous Bardelle et al.,26 mainly by chemical and XRD ana-
products can be obtained by carbothermic reduction. lyses, and weight change measurement for UN, PuN,
However, high amounts of oxygen, up to several and (U,Pu)N. Muromura et al. investigated the mech-
thousand parts per million, are likely to remain in the anism of carbothermic reduction at 1693–2023 K for
products as impurity in case the initial carbon to diox- UN in detail. According to their results, the reaction
ide mixing molar ratio, C/MO2, is 2.0. Therefore, an is divided into four stages: (1) formation of UN1 xCx
excess amount of carbon is usually added to the mixture from UO2, (2) decarburization of UN1 xCx, (3) for-
to reduce the oxygen content and the residual carbon is mation of UN1 xCx with equilibrium composition,
removed from the products by heating in N2–H2 and (4) pure UN formation. They also claimed that
stream as CH(425) or HCN26 after carbothermic reduc- the carbothermic reduction followed the first-order
tion. The initial C/MO2 mixing ratio was chosen at rate reaction expressed as
2.2–2.5 for the preparation of UN and (U,Pu)N. Besides
lnð1 aÞ ¼ kt ½1
the two-step reaction constituted by the carbothermic
reduction in N2 stream and the following decarburiza- where a represents the reaction ratio, k the rate
tion in N2–H2 stream, a one-step reaction in N2–H2 or constant, and t the time, with an activation energy
Nitride Fuel 45
of 347 kJ mol1. This value is consistent with that preparation of Am-bearing nitrides by the two-step
reported by Greenhalgh,32 360 kJ mol1. On the reaction, the carbothermic reduction in N2 stream
other hand, Muromura et al. claimed that the decar- was carried out at 1573 K, which was lower than the
burization in N2–H2 or NH3 stream after carbother- cases for UN and (U,Pu)N by about 200 K.36,38 Then
mic reduction followed the phase boundary-type rate the temperature was raised to 1773 K for the decarbu-
reaction expressed as rization in N2–H2 stream. It is considered that the
intermediate product of AmCO is likely to vaporize
1 ð1 aÞ1=3 ¼ kt ½2 congruently during the carbothermic reduction. On
the other hand, the vaporization of Pu during car-
with activation energies of 285 kJ mol1 in 25% bothermic reduction can be neglected, which is differ-
N2–75% H2 stream and 175–185 kJ mol1 in NH3 ent from the preparation of Pu-bearing carbides by
stream, respectively. carbothermic reduction carried out in vacuum.
Kinetics was also investigated by thermogravime-
try for (U,Pu)N33 and (U,Np)N.34 The results almost
3.02.2.4 Other Nitride Formation
agreed with that for UN by Muromura et al.; the
Processes
carbothermic reduction in N2 stream followed
the first-order rate reaction with activation energies Four processes were reported for the preparation of
of 307 kJ mol1 for (U,Pu)N and 344–385 kJ mol1 nitride with regard to pyrochemical reprocessing
for (U,Np)N. Furthermore, the decarburization for of spent fuel. The first one is the direct dissolution
(U,Np)N in 92% N2–8% H2 stream followed the of spent nitride fuel in liquid Sn, followed by the
phase boundary-type rate equation with an apparent pressurization with N2. It was reported that UN
activation energy of 210 kJ mol1. However, it should powder with high density sank to the bottom and
be pointed out that the decarburization includes could be mechanically separated from the liquid
both the removal of free carbon resulting in a phase.39 The second and third processes concern
decrease in weight and the replacement of carbon the nitridation of actinides recovered in liquid Cd
by nitrogen in carbonitride resulting in an increase cathode by molten salt electrolysis. The second one is
in weight. the nitridation by N2 gas bubbling, in which N2 gas is
Typical impurities in nitride fuel prepared by passed into liquid Cd phase at 773–823 K. Kasai et al.
carbothermic reduction are oxygen and carbon. reported that they succeeded in preparing UN or
It was found that the level of impurities could be U2N3 granules by the N2 gas bubbling method.40 It
kept lower than 1000–2000 ppm for both oxygen and was found, however, that the method was not applica-
carbon by adjusting the initial C/MO2 mixing ratio. ble to the nitridation of Pu in liquid Cd because of
Carbonitrides such as UN1 xCx and PuN1 xCx are the thermodynamic stabilization of Pu in liquid Cd
characterized by complete solubility of the UN–UC phase.41 On the other hand, the third one is the
and PuN–PuC systems, while solubility limits of hypo- nitridation–distillation combined reaction, in which
thetical UO in UN and PuO in PuN were reported at the liquid Cd cathode-containing actinides are
7% and 14%, respectively.35 It was reported that the heated in N2 stream at 973 K. In this method, the
carbon impurity content in mononitride prepared by nitridation of actinides and distillation of Cd proceed
carbothermic reduction is related to the thermody- simultaneously. Preparation of (U,Pu)N, PuN, and
namic equilibrium composition of carbonitride with AmN has been reported so far by the nitridation–
free carbon under nitrogen atmosphere.17 When the distillation combined method.41,42 The fourth one is
same condition of carbothermic reduction was applied called LINEX process, in which actinides dissolved
for UN, NpN, and PuN, the carbon impurity content in the chloride molten salt are converted to nitride by
decreased with the increase of atomic number of the direct reaction with Li3N.43
actinides. Indeed, a rather high initial C/MO2 mix- In addition, a new process was reported by Yeamans
ing ratio was chosen for the preparation of AmN and et al.44 They successfully synthesized UN from UO2 by
(Pu,Cm)N,36,37 since the monocarbides of Am or making it react first with NH3(HF)2 at ambient tem-
Cm are thermodynamically unstable. perature to form (NH4)4UF8, and then with NH3 at
It is well known that Am-bearing species have high 1073 K to UN2, followed by the decomposition to UN
vapor pressures in comparison with the other acti- at 1373 K in Ar. This method has the advantage of a
nides. Vaporization of Am during fuel fabrication pro- low-temperature operation in comparison with the
cess should be kept as low as possible. In the case of carbothermic reduction of dioxides.
46 Nitride Fuel
3.02.2.5 Nitride Pellet Fabrication N2–H2 stream, resulted in lower density than sinter-
ing in low N2 partial pressure, such as in Ar or Ar–H2
Nitride fuel pellets are usually prepared by a classical
stream.45,47 This is an opposite tendency of the
powder metallurgical manner; the product of car-
self-diffusion coefficient of Pu in (U,Pu)N at differ-
bothermic reduction is ground to powder by use of
ent N2 partial pressures.48 The residual oxygen
a ball mill, pressed into green pellets and sintered in a
impurity contents might affect the density of pellets
furnace at 1923–2023 K. An organic binder is some-
sintered in different atmospheres. On the other hand,
times added to the ground powder to facilitate the
sintering in N2 or N2–H2 stream is indispensable
pressing. Finally, the diameter of sintered pellets is
for Am-bearing nitride pellets from the viewpoint
adjusted by use of a centerless grinder. As is men-
of mitigating loss of Am by evaporation. It was
tioned later, one of the characteristics of nitride fuel is
reported that the density higher than 85% TD was
that both He- and Na-bonded pins can be applied. In
attained for (Np,Am)N and (Pu,Am)N pellets by
general, an He-bonded fuel pin is characterized by
sintering in N2–H2 stream at temperatures lower
low-density pellets (i.e., 80–85% of theoretical den-
than 1953 K.49
sity (TD)) and a small gap width between pellets and
In addition to the classical powder metallurgical
cladding tube, whereas a Na-bonded fuel pin is char-
manner, a direct pressing (DP) method was proposed
acterized by high-density pellets (i.e., >90% TD)
by Richter et al.50 In this method, the nitride com-
and a large gap width.
pacts after carbothermic reduction were not ground
Actinide nitride powder has a low sinter-ability in
to powder but directly pressed into green pellets,
comparison with that of oxide or carbide powder,
followed by sintering under the conventional manner.
which is derived from a low diffusion rate of metal
The DP method has the advantage of avoiding dust
atoms in mononitrides. So a rather high sintering
production and shortening preparation period. The
temperature (i.e., T >1973 K) is necessary for pre-
(U,Pu)N pellets prepared by the DP method had a
paring dense UN or (U,Pu)N pellets higher than 90%
density of about 83% TD with levels of oxygen and
TD.45 Although a small amount of Ni powder is an
carbon impurities lower than 0.1 wt%.51 The open
effective sintering aid for carbide fuel, it is not appli-
porosity predominated in the pellets prepared by the
cable to nitride fuel. On the other hand, Bernard et al.,
DP method.
reported that oxygen impurities tend to promote the
An isostatic hot-pressing technique was applied to
sintering of (U,Pu)N pellets.24 However, the increase
fabrication of dense UN specimens for thermal and
of oxygen impurities in UN and (U,Pu)N up to
mechanical property measurements. Speidel et al.
1 wt% resulted in the decrease of density and grain
prepared UN pellets higher than 95% TD by conso-
size of sintered pellets.46 Microstructures of (U,Pu)N
lidating the powder sealed in a refractory metal con-
pellets with different oxygen impurity contents are
tainer under a pressure of 6.9 MPa at 1753–1813 K.52
shown in Figure 3.
Furthermore, a spark-plasma sintering (SPS) method
Sintering atmosphere also affects the sintered
for nitride fuel has been applied to preparation in a
density of nitride fuel pellets. It was reported that
laboratory scale experiment recently.53 The SPS
sintering in high N2 partial pressure, such as in N2 or
Oxide
20 μm 20 μm
Reactor Bonding Max. linear power (kW m1) Max. burnup (% FIMA) References
Temperature (⬚C)
bonding gas.59 1200
A Na-bonding concept is characterized by a large
gap width (i.e., >0.5 mm) between fuel pellets and ΔTfuel
cladding tube and a high density of fuel pellets (i.e.,
>90% TD). This concept has the advantage of 800
Tsurface
keeping the fuel temperature relatively low due to
good thermal conductivity of liquid Na. Further- A B C
more, the temperature of fuel pellets is considered Time scale
as quasiconstant. A shroud tube was sometimes used extented
400
in order to maintain the fuel fragments in their origi- 0 1 2 3 4 5 6
nal geometry. On the other hand, the disadvantage Burnup (at.%)
of a Na-bonding concept includes the difficulty in Figure 4 Schematic change in temperature for MX-type
fuel pin fabrication and spent fuel reprocessing. Fur- fuel pin. Reproduced from Blank, C. J. Less Common Met.
thermore, with regard to safety consideration, the 1986, 121, 583–603.
possibility of loss of Na in a breached pin has to be
evaluated.
At present, a He-bonding concept is considered Table 4 Chemical forms of typical FP in the irradiated
as the reference for (U,Pu)N fuel. A He-bonding (U,Pu)N fuel
concept is characterized by a small gap width (i.e.,
Elementa Chemical Element Chemical
<0.2 mm) and a low density of fuel pellets (i.e., forms forms
80–85% TD). The temperature of fuel pellets
becomes high in comparison with the fuel with Na Ba Ba3N2 Ce CeN
Cs Cs, CsI, CsTe I CsI
bonding, especially at an early stage of irradiation.
Kr Kr La LaN
However, the small gap is closed by free swelling of Mo Mo Nd NdN
fuel pellets at a burnup of 2–3% FIMA, which Pd (U,Pu)(Pd,Ru,Rh)3 Pm PmN
enhances the gap conductance and lowers the fuel Pr PrN Rb Rb, RbI
temperature. A schematic change in fuel tempera- Rh (U,Pu)(Pd,Ru,Rh)3 Ru (U,Pu)(Pd,Ru,Rh)3
Sm SmN Sr Sr3N2
ture for He-bonded (U,Pu)C fuel pin is illustrated
Tc Tc Te Te, CsTe
in Figure 4, which is also applicable to (U,Pu)N Xe Xe Y YN
fuel pin.69 The irradiation period A, as shown in Zr ZrN
Figure 4, corresponds to the first rise of power and
a
lasts for one to several days, the period B the resin- Elements with concentrations <0.08 at.% at burnup of 10% FIMA
are not shown.
tering of pellets center and closure of He gap, and
the period C the quasistate irradiation period in
which the fuel–clad mechanical interaction (FCMI)
3.02.3.3 Chemical Forms of FP
starts. In order to accommodate the swelling and
mitigate the strain on the cladding tube at burnup Chemical forms of FP in nitride fuel were evaluated
progressing, a rather low smear density (i.e., 75–80% by a thermodynamic equilibrium calculation and
TD) is adopted for He-bonded fuel pin. burnup-simulated experiments70,71 as well as postir-
Blank proposed the ‘cold fuel concept’ for MX- radiation examinations (PIE).72 These results agreed
type fuel, in which the maximum fuel temperature is with each other in general but it is difficult to identify
kept lower than one-half or one-third of the melting the phases other than mononitride by XRD or metal-
temperature in Kelvin.69 If this concept is realized, lographic analysis even at a burnup higher than
both low fission gas release and mild restructuring 10% FIMA.
and mild swelling characteristics can be compatible Table 4 shows the most probable chemical forms
in both Na-bonded and He-bonded fuels. of FP in the irradiated (U,Pu)N fuel. Among them,
Nitride Fuel 49
gaseous FP such as Xe and Kr exist as an elementary in Figure 5 were identified by Matzke.1 Zone I found
state. Semivolatile FP such as Cs, I, and Te are likely in the central of the fuel pellet was characterized by
to exist as an elementary state or compounds such as very porous structure. The pores were grown to
CsI and CsTe. Rare earth elements such as Nd, Ce, roughly the grain size and FP gas release was high.
Pr, and Y, and Zr and Nb are considered to be A small central hole was sometimes observed in
dissolved in (U,Pu)N and form a mononitride solid Zone I of the He-bonded (U,Pu)N fuel.59,60 The
solution. On the other hand, Mo and Tc are consid- mechanism of formation is, however, different from
ered to exist in an elementary state together, and that in MOX fuel; according to Coquerelle et al.72 it
Ba and Sr are considered to form lower nitrides results from the migration of lenticular pores up the
such as Ba3N2 and Sr3N2. Platinum group elements radial temperature gradient in MOX fuel, whereas it
such as Pd, Ru, and Rh are likely to form an interme- is apparently created by an in-pile resintering mecha-
tallic compound, (U,Pu)(Pd,Ru,Rh)3, in the irradiated nism in (U,Pu)N fuel. Zone II was characterized by
nitride fuel. pseudocolumnar grains observed in MOX fuel. How-
The change of N/(U + Pu) ratio in mononitride ever, it was not observed in (U,Pu)N but only in
phase was evaluated and Bradbury et al.71 reported (U,Pu)C irradiated at a linear power higher than
that it increased by 2.1% at a burnup of 10% FIMA. 100 kW m1. Zone III was characterized by the struc-
Furthermore, it should be mentioned that the chem- ture accompanied with grain growth, grain boundary
ical forms of FP are also influenced by the oxygen bubbles, and healing of cracks. FP gas release was
and carbon impurity contents, that is, rare earth ele- relatively high and swelling of (U,Pu)N fuel was
ments are likely to form oxide precipitates and Zr is mostly responsible in this zone. On the other hand,
likely to form ZrC dissolved in mononitride phase. Zone IV had the as-fabricated structure. In the case of
It was reported that the lattice parameter of the low-density pellets, slight densification occurred
mononitride phase of (U,Pu)N fuel did not signifi- because of in-pile resintering. Both FP gas release
cantly change with burnup progressing.70 This ten- and swelling were small in Zone IV.73
dency was explained by the compensation of the Temperature range of each zone found in (U,Pu)N
increase in lattice parameter due to the dissolution fuel irradiated in DFR to 4% FIMA was roughly
of rare earth elements and the decrease due to the evaluated by Matzke1 as follows. Zone I appeared at
dissolution of ZrN. a temperature higher than 1673 K, while Zone IV
predominated at a temperature lower than 1423 K.
The intermediate temperature ranging from 400 to
3.02.3.4 Restructuring
500 K corresponded to Zone III. On the other hand,
Because of relatively low fuel temperature and tem- under the ‘cold fuel concept’ proposed by Blank,69
perature gradient, the restructuring of (U,Pu)N fuel most part of fuel pellets should represent the as-
is mild in comparison with MOX fuel for fast reac- fabricated structure seen in Zone IV characterized
tors. However, in the He-bonded (U,Pu)N fuel irra- by low FP gas release and mild swelling.
diated at high linear power, a distinct restructuring Richter et al.74 observed the macro- and
was observed, in which three structural zones shown microstructures of unirradiated (U,Pu)N pellets
Zone IV III II I
Structure of Grain growth, grain Pseudocolumnar Very porous
as-fabricated fuel boundary bubbles grain zone, elongated central zone
grains and pores
Figure 5 Schematic presentation of structural zones observed in MX-type fuels. Reproduced from Matzke, Hj. Science of
Advanced LMFBR Fuels; North-Holland: Amsterdam, 1986.
50 Nitride Fuel
the increase of FP gas-induced swelling, in case FP temperature for the beginning of FP gas bubble pre-
gas bubbles are formed and retained in the pellets. cipitation was 1200–1300 K for (U,Pu)N fuel irra-
Volumetric swelling is also caused by the accumula- diated to 4.3% FIMA in JOYO.
tion of solid FP and crack formation in the pellets. In contrast to FP gas-induced swelling, the
The mechanism of swelling of MX-type fuel was swelling rate due to the accumulation of solid FP is
studied in detail in European Institute for Transura- considered almost independent of temperature and
nium Elements (ITU) and the results were summar- burnup. Its contribution to the volumetric swelling
ized in the monographs of Matzke1 and Blank.2 On was evaluated at 0.5%/% FIMA by a thermodynamic
the other hand, the creep rate of (U,Pu)N fuel is low equilibrium calculation.70
in comparison with MOX or metallic fuel at Ross et al.11 proposed an equation for representing
operating temperatures due to a slow diffusion rate the volumetric swelling of UN fuel as a function of
of metal atoms in nitride fuel. Therefore, the FCMI is average fuel temperature, burnup, and as-fabricated
considered the most important subject for the evalu- pellet density by statistically dealing with 75 reported
ation of (U,Pu)N fuel performance, especially at a data in the SP-100 program,
high burnup. Indeed, in the irradiation tests carried
DV =V ð%Þ ¼ 4:7 1011 Tav
3:12
Bu0:83 D0:5 ½4
out in the 1960s and 1970s, several He-bonded (U,Pu)
N fuel pins were possibly breached by FCMI.59,64 where Tav is the average fuel temperature (K), Bu the
However, most of He-bonded (U,Pu)N fuel pins burnup (% FIMA), and D the as-fabricated pellet
breached were irradiated at a linear power higher density (% TD).
than 100 kW m1 and had a smear density higher
than 85–90% TD. In these pin designs, the strain on
3.02.3.7 Fuel–Clad Chemical Interaction
the cladding tube became severe because of the
swelling of fuel pellets, although the integrity of Nitride fuel has a good enough chemical compatibility
fuel pins depended on the properties of cladding with stainless steels used for cladding tube in fast
tube itself. After the 1980s, the trend of He-bonded reactors. Although the formation of chromium nitride
(U,Pu)N fuel design has been changed to a mild is possible for hyperstoichiometric nitrides under
linear power condition (i.e., <80 kW m1) and a low a thermodynamic equilibrium condition, no results
smear density of fuel pins (i.e., 75–80% TD). have been reported for the deterioration of mechanical
In contrast, the large gap between the pellets and properties of the cladding tube. On the other hand, a
cladding tube will not be completely closed during the slight carburization of inner surface of cladding tube
irradiation period in Na-bonded (U,Pu)N fuel pins. In was found in Na-bonded (U,Pu)N fuel with high oxy-
this case, however, the relocation of the fragments of gen and carbon contents previously. Semivolatile FP-
(U,Pu)N pellets caused the ovalization of cladding tube induced intergranular oxidation of the cladding tube
leading to localized FCMI, which sometimes resulted seen in MOX fuel did not occur in (U,Pu)N fuel since
in breaching. In the irradiation test of (U,Pu)N fuel in the oxygen potential in the gap region is low enough.
EBR-II, a shroud tube was used inside the cladding A hypostoichiometric composition of (U,Pu)N fuel
tube and the effectiveness for maintaining the fuel should be avoided because of the reaction of free
fragments in the original position was confirmed.58 metal phase and cladding tube, leading to the forma-
The ovalization was also observed in the He-bonded tion of (U,Pu)Fe2 and (U,Pu)Ni5-type compounds.1
(U,Pu)N fuel pin irradiated in JOYO with a relatively In general, it is concluded that both He-bonded
large He-gap width.75 and Na-bonded nitride fuels have no problems of
The volumetric swelling rate of (U,Pu)N fuel irra- fuel–clad chemical interaction (FCCI) except for
diated to 9.3% FIMA was evaluated at 1.83%/% the fuel with high impurity contents or metal phase
FIMA without constraint of cladding tube and at precipitation.
1.44%/% FIMA under constraint of cladding tube
by Bauer et al.64 The similar values were also reported
by Lyon et al.76 and Tanaka et al.75 It should be noted 3.02.4 Reprocessing of Nitride Fuel
that FP gas-induced swelling strongly depends on
fuel temperature. Bauer et al.62 reported that the Reprocessing of spent nitride fuel has been investi-
swelling rate sharply rose at 1573–1673 K for (U,Pu)N gated only in a laboratory scale experiment so
fuel mainly irradiated in thermal test reactors. On the far. Both hydrochemical77,78 and pyrochemical43,79
other hand, Inoue et al.61 suggested that the threshold processes were proposed for the reprocessing
52 Nitride Fuel
technology. The disposal of long-lived 14C produced to apply the pyrochemical reprocessing technology
by 14N(n,p)14C reaction in case of nitride fuel with developed for metallic fuel in the United States. In
natural nitrogen or the recovery of expensive 15N in this case, spent nitride fuel is electrochemically dis-
the case of nitride fuel with 15N-enriched nitrogen solved at the anode with the evolution of N2 gas, and
becomes a debatable point in the reprocessing of U and U + Pu are recovered in solid and liquid cath-
nitride fuel. odes, respectively, by electrorefining in a LiCl–KCl
Two methods, the direct dissolution of spent ni- molten salt. For the moment, Cd has been considered
tride fuel in HNO3 and the voloxidation of spent as the most promising material for liquid cathode.
nitride fuel followed by the dissolution in HNO3, Then actinides are to be converted to nitride in molten
are considered as hydrochemical processes. The cadmium.41 In addition to the reference pyrochemical
product of hydrochemical reprocessing is the nitric process mentioned above, a concept called as LINEX
solution of U þ Pu to be converted to oxide, and then process was proposed by Ogawa et al.,43 in which spent
to nitride by carbothermic reduction. On the other nitride fuel is dissolved in the molten salt, followed by
hand, the pyrochemical reprocessing of spent nitride the nitridation of actinides by the reaction with Li3N.
fuel has many similarities with the reprocessing of Since 15N can be recovered as 15N2 gas in the off-gas
U–Pu–Zr metallic fuel developed in the United system of electrorefining process, the pyrochemical
States. In this case, the product of reprocessing is reprocessing technology is promising for nitride fuel
U þ Pu alloy recovered in cathode by molten salt with 15N-enriched nitrogen.
electrorefining, which is to be converted to nitride
again. In addition, a kind of partial reprocessing tech-
nique was proposed, in which spent nitride fuel was 3.02.5 Outlook of Nitride Fuel
pulverized by heating in N2 stream to sesquinitride,
accompanied by the release of FP gas and semivola- Research and development of nitride fuel as an
tile FP.80 A part of the product is subjected to nitride advanced fuel for Gen IV-type fast reactors and a
fuel fabrication without further reprocessing and the dedicated fuel for MA transmutation systems is still
other part is subjected to further reprocessing. By ongoing. Its superior characteristics as nuclear fuel
adopting the partial reprocessing, the material make it possible to design the core with high perfor-
throughput will be drastically decreased in compari- mance and enough safety margins. For the industri-
son with normal reprocessing. alization of nitride fuel, however, several subjects
Selection of two hydrochemical processes, direct should be solved by a lot of efforts.
dissolution or voloxidation, will depend on the The irradiation performance of nitride fuel should
enrichment of 15N in nitride fuel. In case natural be demonstrated at a high burnup. The highest
nitrogen is used in nitride fuel, the direct dissolu- burnup attained by irradiation tests is below 10%
tion method will be promising. The direct dissolution FIMA under fast neutron circumstances. FCMI of
method has the advantage of a simple flowsheet and nitride fuel will be a limited factor at burnup pro-
recovery of 14C as 14CO2 in the off-gas system. The gressing for the industrialization. In addition to the
dissolution of (U,Pu)N in HNO3 proceeds quasicon- modeling of fuel behavior under normal and tran-
gruently and any formation of Pu oxalate seen for sient conditions, the accomplishment of irradiation
(U,Pu)C fuel has not been observed. The conven- tests, in which the average burnup is 15% FIMA at
tional hydrochemical reprocessing technology devel- the lowest, is essential. The irradiation condition
oped for oxide fuel can be applied after the should be well defined and the fuel itself should be
dissolution process. well characterized.
In case 15N-enriched nitrogen is used in nitride From the viewpoint of fuel cycle, reprocessing is
fuel, the voloxidation method will be promising for important. Including the hydrochemical and pyrochem-
hydrochemical process. The voloxidation method has ical technologies, the process for reprocessing was
the advantage of recovery of 15N as 15NOx gas in the tested only in a beaker-scale experiment. The material
off-gas system, which prevents 15N from being con- balance of heavy metals should be evaluated through
taminated with 14N and isotropic exchange reaction the process and the data for designing the apparatus and
in the dissolution process. material handing are almost lacking.
On the other hand, similarities of electrical con- Of course, in case of using the nitride fuel with
15
ductivity and free energy change of formation for N-enriched nitrogen, the economical enrichment
chlorides in nitride and metallic fuels make it possible technology should be developed. The technology
Nitride Fuel 53
should also be easy for enlarging the scale and Their Optimization; IAEA: Vienna, 1988; pp 43–52,
IAEA-TECDOC-466.
friendly to the environments. Especially, the use of 25. Suzuki, Y.; Arai, Y.; Sasayama, T. J. Nucl. Mater. 1983,
15
N-enriched nitrogen is inevitable for the dedicated 115, 331–333.
fuel for MA transmutation systems. Recycling of 26. Bardelle, P.; Warin, D. J. Nucl. Mater. 1992, 188, 36–42.
15 27. Muromura, T.; Tagawa, H. J. Nucl. Mater. 1977,
N in fuel fabrication process will be necessary for 71, 65–72.
economic reason, while recovery of 15N in reproces- 28. Muromura, T.; Tagawa, H. J. Am. Ceram. Soc. 1978, 61,
sing will be an optional process in this case. 30–35.
29. Muromura, T.; Tagawa, H. J. Nucl. Mater. 1979, 80,
330–338.
30. Muromura, T. J. Nucl. Sci. Technol. 1982, 19, 638–645.
References 31. Lindemer, T. B. J. Am. Ceram. Soc. 1972, 55, 601–605.
32. Greenhalgh, W. O. J. Am. Ceram. Soc. 1973, 56,
553–557.
1. Matzke, Hj. Science of Advanced LMFBR Fuels; 33. Pautasso, G.; Richter, K.; Sari, C. J. Nucl. Mater. 1988,
North-Holland: Amsterdam, 1986. 158, 12–18.
2. Blank, H. In Materials Science and Technology, 34. Arai, Y.; Sari, C.; Spirlet, J. C. J. Nucl. Mater. 1991, 185,
A Comprehensive Treatment; Frost, B. R. T., Ed.; 159–166.
Nuclear Materials VCH: Weinheim, 1994; Vol. 10A, Part I, 35. Lorenzelli, R. In Plutonium 1970 and Other Actinides;
pp 191–363. Nuclear Metallurgy; Miner, W. N., Ed.; Metallurgical
3. Languille, A.; Millet, P.; Rouault, J.; Pillon, S. J. Alloys Society of the American Institute of Mining, Metallurgical,
Compd. 1998, 271–273, 517–521. and Petroleum Engineers, Inc. (AIME): New York, 1970;
4. Lyon, W. F.; Baker, R. B.; Leggett, R. D. In Proceedings of Vol. 17, pp 818–828.
LMR: A Decade of LMR Progress and Promise 36. Itoh, A.; Akabori, M.; Takano, M.; et al. J. Nucl. Sci.
Washington, DC, Nov 11–15, American Nuclear Society: Technol. 2002 (Suppl. 3), 737–740.
La Grange Park, IL, 1990. 37. Takano, M.; Itoh, A.; Akabori, M.; et al. J. Nucl. Mater.
5. Matthews, R. B.; Chidester, K. M.; Hoth, C. W.; Mason, R. 2001, 294, 24–27.
E.; Petty, R. L. J. Nucl. Mater. 1988, 151, 334–344. 38. Takano, M.; Itoh, A.; Akabori, M.; et al. In Proceedings of
6. Hayes, S. L.; Thomas, J. K.; Peddicord, K. L. J. Nucl. International Conference on Future Nuclear Systems
Mater. 1990, 171, 262–270. (Global’99), Jackson Hole, WY, Aug 29–Sept 3, 1999;
7. Hayes, S. L.; Thomas, J. K.; Peddicord, K. L. J. Nucl. American Nuclear Society: La Grange Park, IL, 1999;
Mater. 1990, 171, 271–288. CD-ROM.
8. Hayes, S. L.; Thomas, J. K.; Peddicord, K. L. J. Nucl. 39. Anderson, R. N.; Parlee, N. A. D.; Gallagher, J. M. Nucl.
Mater. 1990, 171, 289–299. Technol. 1972, 13, 29–35.
9. Hayes, S. L.; Thomas, J. K.; Peddicord, K. L. J. Nucl. 40. Kasai, Y.; Kakehi, I.; Moro, S.; et al. In Proceedings of
Mater. 1990, 171, 300–318. International Conference on Future Nuclear Systems
10. Storms, E. K. J. Nucl. Mater. 1988, 158, 119–129. (Global’99), Jackson Hole, WY, Aug 29–Sept 3, 1999;
11. Ross, S. B.; El-Genk, M. S.; Matthews, R. B. J. Nucl. American Nuclear Society: La Grange Park, IL, 1999;
Mater. 1990, 170, 169–177. CD-ROM.
12. Ross, S. B.; El-Genk, M. S.; Matthews, R. B. J. Nucl. 41. Arai, Y.; Iwai, T.; Akabori, M.; Minato, K. Nucl. Technol.
Mater. 1988, 151, 313–317. 2008, 162, 244–249.
13. Routbort, J. L.; Singh, R. N. J. Nucl. Mater. 1975, 58, 42. Hayashi, H.; Shibata, H.; Akabori, M.; Arai, Y.; Minato, K.
78–114. Electrochemistry 2009, 77, 673–676.
14. Mukaiyama, T.; Takizuka, T.; Mizumoto, M.; et al. Prog. 43. Ogawa, T.; Akabori, M.; Suzuki, Y.; et al. In Proceedings
Nucl. Energy 2001, 38, 107–134. of International Conference on Future Nuclear
15. Suzuki, Y.; Arai, Y. J. Alloys Compd. 1998, 271–273, Systems (Global ‘97), Yokohama, Japan, Oct 5–10, 1997;
577–582. Atomic Energy Society of Japan: Tokyo, 1997;
16. Minato, K.; Akabori, M.; Takano, M.; et al. J. Nucl. Mater. pp 812–815.
2003, 320, 18–24. 44. Yeamans, C. B.; Silva, G. W. C.; Cerefice, G. S.; et al.
17. Arai, Y.; Minato, K. J. Nucl. Mater. 2005, 344, J. Nucl. Mater. 2008, 374, 75–78.
180–185. 45. Arai, Y.; Fukushima, S.; Shiozawa, K.; Handa, M. J. Nucl.
18. Minato, K.; Takano, M.; Otobe, H.; et al. J. Nucl. Mater. Mater. 1989, 168, 280–289.
2009, 389, 23–28. 46. Arai, Y.; Morihira, M.; Ohmichi, T. J. Nucl. Mater. 1993,
19. Holleck, H. In Thermodynamics of Nuclear Materials 1974; 202, 70–78.
IAEA: Vienna, 1975; Vol. II, pp 213–264. 47. Tennery, V. J.; Godfrey, T. G.; Potter, R. A. J. Am. Ceram.
20. Holleck, H.; Kleykamp, H. In Gmelins Handbuch der Soc. 1971, 54, 327–331.
Anorganischen Chemie, Vol. 55, Uranium; Supplement 48. Bradbury, M. H.; Matzke, Hj. J. Nucl. Mater. 1978, 75,
C7, Nitrogen Compounds; Springer-Verlag: Berlin, 1981; 68–76.
Chapter 4.1. 49. Arai, Y.; Akabori, M.; Minato, K. In Proceedings of
21. Tagawa, H. J. Nucl. Mater. 1974, 51, 78–89. International Conference on Future Nuclear Systems
22. Potter, P. E. In Plutonium 1975 and Other Actinides; (Global 2007), Boise, ID, Sept 9–3, 2007; American
Blank, H., Lindner, R., Eds.; North-Holland: Amsterdam, Nuclear Society: La Grange Park, IL, 2007; pp 980–988.
1976, pp 211–232. 50. Richter, K.; Gueugnon, J.; Kramer, G.; Sari, C.; Werner, P.
23. Lorenzelli, R. Contribution a L’Etude du Systeme (U,Pu) Nucl. Technol. 1985, 70, 401–407.
C,N; CEA Report R-3536; Commissariat à l Energie 51. Richter, K.; Blank, H. In Advanced Fuel for Fast Breeder
Atomique (CEA): Fontenay-aux-Roses, France, 1968. Reactors: Fabrication and Properties and their
24. Bernard, H.; Bardelle, P.; Warin, D. In Advanced Fuel for Optimization; IAEA: Vienna, 1988; pp 61–70,
Fast Breeder Reactors: Fabrication and Properties and IAEA-TECDOC-466.
54 Nitride Fuel
52. Speidel, E. O.; Keller, D. L. Fabrication and Property of Oct 28–Nov 1, 1991; Atomic Energy Society of Japan:
Hot-Pressed Uranium Mononitride; Battelle Memorial Tokyo, 1991; Paper No. 2–6.
Institute: Columbus, OH, 1963; BMI-1633. 66. Jaecki, P.; Pillon, S.; Warin, D.; et al. In Proceedings of
53. Muta, H.; Kurosaki, K.; Uno, M.; Yamanaka, S. J. Nucl. International Conference on Nuclear Energy Systems for
Mater. 2009, 389, 186–190. Future Generation and Global Sustainability (Global 2005),
54. Ledergerber, G.; Kopajtic, Z.; Ingold, F.; Stratton, R. W. Tsukuba, Japan, Oct 9–13, 2005; Atomic Energy Society
J. Nucl. Mater. 1992, 188, 28–35. of Japan: Tokyo, 2005; Paper No. 448.
55. Ledergerber, G.; Ingold, F.; Stratton, R. W.; et al. Nucl. 67. Mayorshin, A. A.; Kisly, V. A.; Shishalov, O. V.; et al.
Technol. 1996, 114, 194–204. In Proceedings of International Conference on Atoms for
56. Ganguly, C. Metals Mater. Processes 1990, 1, 253–274. Prosperity: Updating Eisenhower’s Global Vision for
57. Shirasu, Y.; Yamagishi, S. Preparation of UN Microspheres Nuclear Energy (Global 2003), New Orleans, LA, Nov
by Internal Gelation Process; JAERI-Research 97–050; 16–20, 2003; American Nuclear Society: La Grange Park,
Japan Atomic Energy Research Institute, 1997; IL, 2003; p 1989.
In Japanese. 68. Iwai, T.; Nakajima, K.; Kikuchi, H.; et al. Post Irradiation
58. Bauer, A. A.; Cybulskis, P.; Green, J. L. In Proceedings of Examinations of Inert Matrix Nitride Fuel Irradiated in JMTR
Topical Meeting on Advanced LMFBR Fuels, Tucson, AZ, (01F-51A Capsule); JAEA-Research 2007-026; Japan
Oct 10–13, 1997; American Nuclear Society: La Grange Atomic Energy Agency, 2007.
Park, IL, 1977; pp 299–312. 69. Blank, C. J. Less Common Met. 1986, 121, 583–603.
59. Blank, H. In Proceedings of Topical Meeting on Advanced 70. Arai, Y.; Maeda, A.; Shiozawa, K.; Ohmichi, T. J. Nucl.
LMFBR Fuels, Tucson, AZ, Oct 10–13; American Nuclear Mater. 1994, 210, 161–166.
Society: La Grange Park, IL, 1977; pp 482–501. 71. Bradbury, M. H.; Matzke, Hj. J. Nucl. Mater. 1980, 91,
60. Fromont, M.; Lamontagne, J.; Asou, M.; Aubrun, I. 13–22.
In Proceedings of International Conference on Nuclear 72. Coquerelle, M.; Walker, C. T. Nucl. Technol. 1980, 48,
Energy Systems for Future Generation and Global 43–53.
Sustainability (Global 2005), Tsukuba, Japan, Oct 9–13, 73. Colin, M.; Coquerelle, M.; Ray, I. L. F.; Ronchi, C.;
2005; Atomic Energy Society of Japan: Tokyo, 2005; Walker, C. T. Nucl. Technol. 1983, 63, 442–460.
Paper No. 329. 74. Richter, K.; Sari, C. J. Nucl. Mater. 1991, 184, 167–176.
61. Inoue, M.; Takashi, I.; Arai, Y.; Asaga, T. In Proceedings of 75. Tanaka, K.; Maeda, K.; Katsuyama, K.; et al. J. Nucl.
International Conference on Atoms for Prosperity: Mater. 2004, 327, 77–87.
Updating Eisenhower’s Global Vision for Nuclear Energy 76. Lyon, W. F.; Baker, R. B.; Leggett, R. D.; Matthews, R. B.
(Global 2003), New Orleans, LA, Nov 16–20; American In Proceedings of International Conference on Fast
Nuclear Society: La Grange Park, IL, 2003; Reactors and Related Fuel Cycles (FR’91), Kyoto, Japan,
pp 1694–1703. Oct 28–Nov 1, 1991; Atomic Energy Society of Japan:
62. Bauer, A. A.; Brown, J. B.; Fromm, E. O.; Storhok, V. W. Tokyo, 1991; Paper No. 14–8.
In Proceedings of Topical Meeting on Fast Reactor Fuel 77. Olschewski, N. H.; Glatz, J. P.; Bokelund, H.; Leroy, M. J. F.
Element Technology, New Orleans, LA, Apr 13–15, 1971; J. Nucl. Mater. 1992, 188, 244–148.
American Nuclear Society: La Grange Park, IL, 1971; 78. Till, J. E.; Bomar, E. S.; Morse, L. E.; Tennery, V. J. Nucl.
pp 785–817. Technol. 1978, 37, 328–339.
63. ITU Annual Report 96. Investigation of High Burn-up 79. Rogozkin, B. D.; Stepennova, N. M.; Fedorov, Y. E.; et al.
Nitrides as Candidate Fuel for Incineration of Plutonium; Mononitride fuel for fast power reactors. In The 2nd
EUR 17296; European Commission: Luxembourg, 1996; JAERI-RDIPE Seminar on Lead Cooled Fast Reactor,
pp 84–85. Moscow, Oct 27, 1993.
64. Bauer, A. A.; Cybulskis, P.; Petty, R.; de Muth, N. S. 80. Fujioka, T.; Kosaka, Y.; Enokida, Y.; Kawakita, T.
In Proceedings of Topical Meeting on Fast Breeder In The 5th International Nuclear Conference on
Reactor Fuel Performance, Monterey, CA, Mar 5–8, 1979; Recycling, Conditioning and Disposal (RECOD98), Nice,
American Nuclear Society: La Grange Park, IL, 1979; France, Oct 25–28, 1998.
pp 827–841. 81. Arai, Y.; Akabori, M.; Minato, K. In Proceedings of Ninth
65. Kochetkov, L. A.; Poplavskii, V. M.; Rineiskii, A. A.; et al. Information Exchange Meeting on Actinide and Fission
In Proceedings of International Conference on Fast Product Partitioning and Transmutation, Nimes, France,
Reactors and Related Fuel Cycles (FR’91), Kyoto, Japan, Sept 25–29, 2006; OECD: Paris, 2008; pp 117–126.
3.03 Carbide Fuel
A. K. Sengupta, R. Agarwal, and H. S. Kamath
Bhabha Atomic Research Centre, Mumbai, India
3.03.1 Introduction 56
3.03.1.1 History of Carbide Fuel 56
3.03.1.2 Glimpses of Carbide Fuel 57
3.03.2 Physical Properties 59
3.03.2.1 Thermophysical Properties of Carbide Fuel 59
3.03.2.2 Thermochemistry of Carbide Fuels 60
3.03.3 Fabrication of Carbide Fuel 64
3.03.3.1 Melting Casting 64
3.03.3.2 Hydride/Hydrocarbon Route 64
3.03.3.3 Carbothermic Reduction Route 64
3.03.3.3.1 Direct pressing method 65
3.03.3.3.2 Sol–gel (wet) route 66
3.03.3.4 Quality Control 67
3.03.3.4.1 Chemical quality control method 68
3.03.3.4.2 Physical quality control 69
3.03.4 In-Pile Performance 70
3.03.4.1 Introduction 70
3.03.4.1.1 Burnup 70
3.03.4.1.2 Swelling 71
3.03.4.1.3 Performance of Na-bonded and He-bonded fuel pins 72
3.03.4.1.4 Irradiation creep 73
3.03.4.1.5 Experience on irradiation performance 75
3.03.4.1.6 Fuel–clad chemical interaction 78
3.03.4.2 Effects of Burnup on C/M Ratio and Chemical State of Fission Product 79
3.03.4.3 MA-Containing Fuel 80
3.03.5 Fuel Reprocessing and Waste Management 82
3.03.6 Summary 83
References 84
55
56 Carbide Fuel
IDMS Isotopic dilution mass spectrometry FBRs is about 60 times greater than that obtained
IGCAR Indira Gandhi Centre for Atomic from LWR, even after allowing for losses in the fuel
Research, Kalpakkam cycle.1 Hence, for effective utilization of the limited
ITU Institute of Transuranium Elements resources of U, the fast reactor concept came into exis-
LDP Large development plant tence. Consequently, efforts were in progress to develop
LHR Linear heat rating fast breeder reactors (FBRs) in several countries such
LMFBR Liquid metal fast breeder reactor as the United States, the United Kingdom, France,
LWR Light water reactor and the erstwhile USSR. In the 1950s, the early fast
MA Minor actinides reactor fuel developed was metallic (experimental
MKI & MKII Mark I and Mark II fuels of FBTR, breeder reactor I, EBR I) and it became the first fast
Kalpakkam, India reactor in the United States to generate electricity.
NIMPHE NItrure Mixte dans a Phenix This led to the development of the second generation
PFR Prototype fast reactor of larger fast reactors, which started operating in
PGM Platinum group metals the 1970s. With the inception of fast reactor technol-
PHWR Pressurized heavy water reactor ogy, the power density of the nuclear reactor core
PIE Postirradiation examination went up manyfold, thus enabling compact reactors
PUREX Plutonium and uranium recovery by producing larger power. These first few prototype
extraction reactors were BN350 (USSR), PHENIX (France),
SEM Scanning electron microscopy and PFR (United Kingdom). Large prototype fast
SFR Sodium-cooled fast reactor reactors of capacity 500 MWe were developed in
TD Theoretical density the USSR (BN600) in 1980 and another one of
TIG Tungsten inert gas 1200 MWe capacity was developed in France (Super
TIMS Thermal ionization mass Phenix). The first-generation large breeder reactor
spectrometry used uranium–plutonium-mixed oxide fuels in aus-
TREAT Transient-overpower test tenitic stainless steel clad with sodium as coolant.
TRU Transuranium element The technology for mixed oxide fuel fabrication and
U/M, Pu/M Uranium to (U þ Pu) ratio, plutonium its irradiation performance are quite well established
to (U þ Pu) ratio and a burnup of 10 at.% or more has been achieved in
UREX Uranium recovery by extraction several reactors: for example, PFR (United Kingdom),
VPPM Parts per million by volume Phenix, fast flux test facility (FFTF) (United States),
XGAR X-ray gamma autoradiography and JOYO (Japan). The European fast reactor (EFR)
XRD X-ray diffractrometry program has also considered mixed oxide fuel for
XRF X-ray fluorescence spectroscopy the first demonstration-type reactor. Although FBRs
based on mixed oxide fuel proved quite successful
in enhancing energy production, this option was not
a plausible futuristic solution particularly for develop-
3.03.1 Introduction ing countries which have limited uranium reserve
to meet their long-term sustainable energy demand.
3.03.1.1 History of Carbide Fuel
The anticipated shortage of fissile material in future
The world energy requirement from nuclear sources led to the search for a fuel having higher breeding
was mostly met by the use of uranium in pressurized ratio and lower compounded doubling time (CDT)
heavy water reactors (PHWR) or light water reactors (less than 15 years) than oxide (about 25 years).
(LWR). However, uranium reserve is limited and it The CDT is defined1 as the time needed to double
was felt necessary that the fuel cycle based on ura- the inventory of fissile material allowing for out-
nium must be a closed fuel cycle. In the vast majority of-reactor time, reprocessing, and losses during
of LWRs or HWRs, nuclear power is generated by out-of-reactor time in a system of breeder reactors
the fission of 235U by thermal neutrons. The where the excess fissile material is used to start up new
235
U content in natural uranium is only 0.7% and breeders as soon as the fuel from each cycle is repro-
the remaining is 238U which is nonfissionable. cessed. The high doubling time of oxide fuel is attrib-
238
U can, however, be converted to fissionable 239Pu, uted to its lower heavy-atom density. It also operates
which can be subsequently used as fuel for fast reactors. at a lower linear heat rating because of its low thermal
The energy generated from natural uranium using conductivity. These limitations of the oxide fuel led to
Carbide Fuel 57
the opening up of worldwide research for a suitable problem of 14N because of the (n.p.) reaction leading to
fuel with higher fissile atom density and thermal con- formation of 14C. Hence to get the full benefit of a
ductivity. Metallic fuel based on U–Pu–Zr or U–Zr nitride fuel, 14N should be replaced by 15N, which is
was presumed to be a possible solution, but it too had an expensive proposition. The (n.p.) reaction will have
limitations such as low melting point and high swelling a negative effect on the breeding ratio. However, a fuel
rates. Other possible fuel compositions with higher cycle based on carbide fuel could be the most feasible
fissile atom density were nonoxide ceramics, for exam- solution, and extensive research on different aspects of
ple, uranium–plutonium-mixed carbide or nitride. carbide fuel based on out-of-pile and in-pile experi-
These fuels have higher thermal conductivity (about ments has been carried out in countries such as the
5 times higher than oxide; Figure 1), high fissile United States, Germany, France, and India with the
heavy-atom density, and a reasonably high melting objective of developing their fast reactor program
point compared to metallic fuel.2 The higher thermal based on ‘carbide fuel.’
conductivity of carbide fuel results in more efficient The development of the carbide-fueled FBR was a
heat transfer from the fuel to the coolant compared to little sluggish during the 1970s. The practical diffi-
oxide and this, in combination with a high melting culty of fabrication of carbide fuel economically was
point, makes carbide fuel suitable for operation at probably the cause, since the high-purity inert cover
high specific power without causing any fuel central gas required for fuel fabrication was expensive and
melting. This also enables the use of large diameter maintenance of C/M ratio was difficult. Apart from
fuel pins, more fissile material per pin, and more this, the thermo-chemical in-pile behavior of carbide
power generation. Increase in the oxygen/metal ratio fuels, e.g., extent of loss of ductility of the cladding
with burnup in case of oxide fuel increases the proba- due to carburization was not fully understood. Behav-
bility of fuel–clad chemical interaction (FCCI), ior of the fuel under off-normal conditions, like loss
whereas for carbide and nitride fuel, the nonmetal to of sodium bonding during transient-overpower con-
metal ratio (C/M or N/M) remains constant or may ditions, was also not systematically investigated. In
even decrease depending upon the composition of the the absence of this information, utilization of carbide
fuel. High-specific-power operation permits fewer fuel was limited to test pin or capsule irradiation only.
pins, compared to oxide, and a more compact core. The Indian FBR program, however, started with the
This in totality reduces the plant cost significantly. introduction of plutonium-rich mixed uranium–
These findings were very encouraging and led to plutonium carbide as the driver fuel for 40 MWth
initiation of a large amount of development work on (13 MWe) loop-type reactor (fast breeder test reactor,
carbide and nitride fuels from 1960 to 1970, and more FBTR). The design of the Indian FBR was similar to
than 5000 advanced fuel pins have been fabricated that of Rapsodie (Rapid Sodium) Fortissimo version
and irradiated. However, nitride fuel has an inherent based on oxide fuel. The reactor became critical in
the year 1985 and it is the only reactor operative on a
full core of carbide fuel. Carbide fuel cannot be used
in LWR because of its incompatibility with the cool-
30
ant. However, it can be safely used with liquid metal
U–19Pu–10Zr (Na or lead) or gas cooled (CO2 or He) in Generation
25 MC
MN
Thermal conductivity λ (Wm-1 K-1)
Table 1 History of carbide fuels used in different countries and in different reactors
Fuel type Reactor Country/ Bond Density (% TD) Burnup (at.%) Clad References
organization
Table 2 Properties of mixed uranium–plutonium oxide, carbide, nitride, and metallic fuels for SFR
studies were brought together in the International Both carbide and nitride fuels offer the best poten-
symposium on ‘Carbide in Nuclear Energy,’ Vols I tial for LMFBR performance in the long run because
and II held at Harwell in 196311 at a time when carbide of their higher thermal conductivity, fissile-atom den-
was being seriously thought of as a reactor fuel. Subse- sity, and chemical compatibility with liquid sodium.
quently, following the oil crisis in 1974, a national Mixed oxide fuel has a higher melting point than
advanced liquid-metal-cooled fast breeder reactor carbide or nitride fuels but the higher thermal con-
(LMFBR) fuels development program was initiated in ductivity of the carbide fuels compensates for it. Some
the United States on the basis of the data available important parameters of these different types of fuel
in the exploratory years of basic development work are given in Table 2.
on carbide fuel, and a unified national approach was There are two concepts available for the carbide
pursued to look into all the aspects of carbide fuels. fuel pin depending upon the type of bond between
Carbide Fuel 59
the fuel pellet and the cladding material: He-bonded Table 3 Specifications of high- and low-plutonium-
and Na-bonded carbide fuels. The average operating containing fuels
fuel temperature of the He-bonded pin is high (Pu0.7U0.3)C (Pu0.2U0.8)C
because of low thermal conductivity of the He bond
compared to the Na bond. This design requires a Plutonium (wt%) 66 1 21.3 1
Pu/(U + Pu) 0.70 0.225
larger fuel–clad gap and low fuel density (85%) com-
Oxygen (ppm) 6000 360
pared to oxide fuel so that the higher swelling of the Oxygen + nitrogen (ppm) 7500 400
fuel due to high operating temperature can be accom- M2C3 (wt%) 5–20 12.5 1.4
modated. The fission gas release will also be higher Density (% TD) 90 1 80 1
compared to that from a sodium-bonded pin. In case Grain size 10–12 mm 12 mm
of the sodium-bonded pin, the fuel–clad gap is larger
and the fuel density higher so that there is enough
radial space for accommodating fuel swelling and the fuel specification for carbide fuels and Figure 2 shows
end of life of fuel is determined by the fuel–clad gap the design of the Pu-rich carbide fuel pin used in
closure. Sodium-bonded fuel does not undergo FBTR.
much of fuel restructuring at comparable linear
powers used for oxide-fueled fast reactors because
of the lower central temperature. The purity of the 3.03.2 Physical Properties
sodium bond is very important and is limited to a
3.03.2.1 Thermophysical Properties of
maximum of 10 ppm. High O in sodium reduces
Carbide Fuel
thermal conductivity, affecting thermal performance
of the fuel. The thermophysical properties that are of impor-
A hyperstoichiometric (C/M > 1, carbon to tance and affect the fuel performance are solidus/
metal ratio) fuel composition is chosen so that it liquidus temperature, thermal conductivity, coeffi-
contains some amount of sesquicarbide M2C3 phase cient of thermal expansion, elastic/fracture proper-
(M ¼ U þ Pu), which takes care of the decrease in ties, creep, and hardness at ambient and at high
(C/M) ratio with burnup. This decrease in (C/M) temperatures.
ratio with burnup may result in the formation of The solidus/liquidus temperatures along with ther-
actinide metal phase, which may form low-melting mal conductivity limit the fuel operating temperature
eutectic with the cladding, thereby limiting the life of in terms of linear heat rating (W cm1), and the fuel
the fuel pin. This is more so for fuels with the com- center which ‘sees’ the highest temperature does not
ponents of high plutonium content because, for Pu exceed the solidus temperature. The liquidus tempera-
fission by fast neutron, the fission yield curve shifts to ture gives some indication of the physical state of the
the right, generating more noble fission products. fuel in case of core meltdown under accidental condi-
Depending upon the temperature of operation (type tions. Thermal conductivity is an important factor that
of bond), they may cause bonding of the fuel with determines the rate of heat transfer from the fuel to the
cladding material resulting in fuel–clad mechanical clad. As mentioned above, these properties also put a
interaction (FCMI). O and N impurities also play limit on the fuel surface temperature. Thermal con-
important roles, as they act as ‘carbon equivalent’, ductivity, though an intrinsic property, varies with a
which decides the carbon potential of the fuel. number of parameters which are characteristic to the
O and N are impurities that are picked up during sample. Some of these parameters are density or poros-
fabrication and their contents are very important to ity (shape, size, and distribution), composition, pres-
make the fuel of the desired specification. Solubility ence of a second phase, grain size, etc. Coefficient of
of O is more in PuC than in UC-rich carbide fuel. thermal expansion is an important design parameter
Hence, plutonium-rich fuel has more O solubility for the fuel pin (both for fuel and cladding material). It
than UC-rich fuel. The details of all these aspects depends on the composition as well as the extent of
are discussed in the following section, which deals second phase present. The stresses generated by the
with the fuel properties, characterization, fabrica- fuel over the cladding material are partly due to the
tion, and postirradiation examination (PIE). The difference in the coefficient of thermal expansion
reprocessing and waste management of carbide fuel between the fuel and the cladding material. The elastic
has also been covered, to give a glimpse of the property and the fracture property of the fuel are
overall carbide fuel cycle. Table 3 shows the typical primarily responsible for the extent of FCMI. Hardness
60 Carbide Fuel
Angular position
of top plug with
respect to bottom plug
263⬚
View in direction S
Æ5.1 Æ5.1 S
Top plug
13 13 S1
Leak tight 8
weld (TIG) 4 Wire weld
Spring on the
top plug
106
Spring
support 4
3 Æ 0.76 mm wire
Insulation wound throughout
pellet the length
pitch = 90 mm
SS316 Clad
±1.5
320
515.3
505.5 531.5
Fuel pellet
531.5
Insulation
pellet 3
Disk
61.5
Pellet support
tube
1.8
Leak tight
weld (TIG) 3
Bottom plug 13 S2 13
Figure 2 Schematic diagram of the fuel pin design of a fast breeder test reactor.
of the fuel determines the extent of FCMI: a softer Thermophysical Properties of the Actinide Car-
fuel will exert lower stresses on the cladding, thus bides of this Comprehensive. A summary of the data
have lower FCMI. Both thermal-induced and irradia- for high (70%) and low (20%) plutonium-containing
tion-induced creep of the fuel also determine the fuel12 is given in Table 4.
extent of pellet–clad mechanical interaction. Creep
properties also depend on a number of variables such
3.03.2.2 Thermochemistry of Carbide Fuels
as composition, presence of a second phase as precipi-
tates, grain size, etc. Thermophysical properties of Uranium forms three compounds with carbon, that is,
actinide carbide have been extensively described and UC, U2C3, and UC2, out of which only U2C3 is a
discussed in Chapter 2.04, Thermodynamic and stoichiometric compound. UC is stable over a wide
Carbide Fuel 61
temperature and composition range and melts con- of metal and nonmetal atoms, the latter restricting
gruently at slightly substoichiometric composition, the slip and thereby hardening the crystal. In addi-
at 2780 K. UC2 is stable in two phases, a-UC2 and tion, the two p-states of carbon do not favor the
b-UC2. U2C3 decomposes into UC þ a-UC2 on heat- formation of octahedral ligands; hence many of
ing from 2096 to 2110 K and into UC þ C below the fission products are not soluble in monocarbide
1400 K. The Pu–C system has four compounds: lattice. During irradiation, carbide fuels are known
Pu3C2, PuC1–x, Pu2C3, and PuC2. The compound to swell more than the oxide fuel. Swelling is a com-
Pu3C2 decomposes into Pu þ PuC at 848 K. Though plicated phenomenon, controlled by many factors.
the crystal structures of carbides of U and Pu are However, the close-packed structure of monocarbides
very similar, phase diagrams of U–C and Pu–C is known to contribute to higher swelling in carbide
are very different. These differences are mainly due fuels, especially at low temperatures (T/Tm < 0.3), as
to (i) the presence of Pu3C2 compound, (ii) the low bulky fission product atoms of Xe/Kr cannot be
stability of PuC compared to UC, and (iii) the accommodated in the carbide lattice. The transport
high stability of Pu2C3 compared to U2C3 (Chapter properties of vacancies and interstitials in these struc-
2.04, Thermodynamic and Thermophysical Prop- tures also add to this problem. The presence of the
erties of the Actinide Carbides). more open type structure of the M2C3 phase in hyper-
UC and PuC are highly dense, face-centered stoichiometric mixed carbide fuel reduces the
cubic (fcc) packed metal atoms with octahedral swelling to some extent. The structures of UC and
holes occupied by carbon atoms. The brittleness of PuC are isomorphous with monocarbides of transi-
carbides is due to alternate close-packed planes tion metals and some of the actinide metals, and thus
these fission products dissolve in the monocarbide
phase to varying degrees.
Table 4 Thermophysical properties of high- and low-
plutonium-containing carbide fuels An increase in O and N impurities results in an
increase of carbon activity and CO pressure of the
Properties (U0.3Pu0.7)C (U0.8Pu0.2)C fuel (Figure 3). O impurities are also known to con-
Solidus temperature (K) 2148 3023 tribute significantly to the actinide redistribution in
Thermal conductivity 12.0 19.0 the carbide fuels and fuel restructuring during burnup.
(W m1 K) at 1273 K Much experimental data as well as assessed and criti-
Coefficient of thermal 13.8 106 10.9 106 cally reviewed works are available in the literature
expansion (300–1800 K)
investigating the impact of O þ N impurities on the
Hardness (MPa) at 1250 K 1200 1400
behavior of carbide fuels.13–17 In order to understand
6 ´ 10–16 0.011
T = 1000 K
4 ´ 10–16 pCO
0.010
aC
aC 0.009
Pu/M = 0.55
2 ´ 10–16
Carbon activity
pCO 0.008
pCO
1 ´ 10–16
0.007
8 ´ 10–17 aC of SS-clad at 1000 K
6 ´ 10–17
0.006
pCO
4 ´ 10–17
aC 0.005
Pu/M = 0.7
2 ´ 10–17 0.004
600 800 1000 1200 1400 1600 1800 2000 2200
[O] in ppm
Figure 3 Effect of O content and Pu/(Pu + U) on carbon potential and carbon monoxide partial pressure of carbide fuel.
62 Carbide Fuel
the effect of O and N on the behavior of carbide fuels, example, MC, MC2 (with high U), and M2C3 (with
it is important to understand their interactions high Pu). Solubility of nitrogen stabilizes UC2 and
with individual carbides. The structures of UC and PuC2 below their decomposition temperatures. Car-
PuC are isomorphous with UO/PuO and UN/PuN, bon activity of mixed carbide decreases with increase
and therefore these compounds show reasonable in Pu content because of negative Gibbs energy
solubility in the monocarbide phase. Though UO change for the reaction PuC þ U2C3 ! UC þ Pu2C3.
and PuO are not stable compounds, O replaces This also results in Pu enrichment of M2C3 phase.
C and gets stabilized in the monocarbide lattice. Opposite effects of Pu and N þ O contents also mean
PuC can accommodate more oxygen (65 mol% PuO) that Pu-rich fuel can accommodate higher nonmetal-
than UC (35 mol% UO). PuC–PuO and PuC–PuN lic impurities than U-rich fuel (Figure 4). The effect
follow a near-ideal-solution behavior, whereas the of Pu and O þ N content on carbon activity can
UC–UO system shows a negative deviation from also be seen in the phase diagram of carbide system
ideality and has limited solubility. UC–UN and (Figure 5). Carbon precipitation results in shrinking
PuC–PuN form solutions over the complete compo- of the MC þ M2C3 phase field with increase in O þ N
sition range. UC–UN solution shows a slight positive content and its expansion with increase in the Pu
deviation from ideality on the UC side. Because of the fraction.
increased ionic character of nitrogen and oxygen Mixed carbide shows actinide segregation, with
compared to carbon, addition of oxygen or nitrogen a higher plutonium content in the M2C3 phase
impurity in UC, PuC, or MC results in slight contrac- than in the MC phase,20,21 as shown in Figure 6.
tion in the lattice with a small positive deviation from The segregation of actinides in the two phases
Vegards law.18,19 As discussed in Chapter 2.04, Ther- MC and MC1.5 reduces with increase in the nonme-
modynamic and Thermophysical Properties of the tallic impurity contents O and N.12 This can be
Actinide Carbides, nonstoichiometry of carbides explained from the negative Gibbs energy changes
decreases by substitution of C by O or N. MC and of the reactions Pu2C3 þ UN ! PuN þ U2C3 and
M2C3 phases of Pu-rich mixed carbide fuel become Pu2C3 þ UO ! PuO þ U2C3. Hence, the presence
stoichiometric at high N þ O content (6000 ppm). of O þ N impurities stabilizes more U in M2C3 and
Pu-rich hyperstoichiometric carbide fuel is biphasic more Pu in MC. Segregation of actinides in two
MC þ M2C3 in the temperature range of operation of phases is more pronounced in plutonium-rich fuel
reactor. However, U-rich hyperstoichiometric fuel and the effect reduces with increase in temperature
may have three phases at high temperatures, for and also with increase in sesquicarbide fraction.
t%
0
w
+x %
wt
:8
−5 )C 1 +x
U2 C -U
:4 C
3
C
0.
8 )C 1 3
C3
2
U
,M
U 0.8 2
−10 u
2
0. , M t %
pm
. 2 w
(P u0 pm
(P :8
0p
p
C3
00
−15 6 00 , M2
60
Carbon potential (kJ mol-1)
]: m
+N pp
]:
−20 00
N
[O ]: 2
+
N
[O
+
−25 [O
x : 8 wt%
)C 1+ m, M 2C 3
−30 U0 . 8
+ N ]: 6000 pp
2 [O
(Pu 0
.
U )C 1+x
: (Pu 0.55 0.45
−35 MARK II
15 wt%
m, M 2C 3:
−40 N]: 6000 pp
u U )C 1+x [O +
MARK I: (P 0.7
0.3
−45 SS
31
−50 6
Pu2C -PuC
985 K
935 K
−55 3
PuC-Pu
−60
−108
UC-U
−110
600 800 1000 1200 1400 1600 1800 2000 2200
Temperature (K)
Figure 4 Comparison of carbon potentials of carbide system for different Pu/(Pu + U) and (O + N) values.
Carbide Fuel 63
C Pu/(Pu+U) = 0.2
0.0 1.0 Pu/(Pu+U) = 0.55
Pu/(Pu+U) = 0.7
1000 K
M
X
0.2 0.8
(U,
Pu
0.6
)C 1.5
(U,Pu)C1. 5 + 0.4
(U,Pu)CNO
+C
XC
(U,
Pu
pm
)CN
0.6 0.4
m
0p
O+
pp
00
C
00
pp
=2
m
=2
m
p
00
]
pm
0.8 0.2
] = 200 p
[O
pp
]
20
[O
0p
00
]=
]=
20
0
[O
=2
[O
(U,Pu) + (U,Pu)CNO
[O]
[O
1.0 0.0
(U,Pu) 0.0 0.2 0.4 0.6 0.8 1.0 (U,Pu)N
XMN
Figure 5 Effect of Pu/(Pu + U) and oxygen impurity on the phase diagram of (U,Pu)–C–N system.
0.9 0.9
0.8 0.8
0.7 0.7
0.6 0.6
xPuC1.5 in MC1.5
xPuCNO in MCNO
0.5 0.5
0.4 0.4
0.3 0.3
0.2 0.2
C/M = 1.005 C/M = 1.005 C/M = 1.005
[N] = 1000 ppm [N] = 1000 ppm [N] = 1000 ppm
0.1 [O] = 1000 ppm [O] = 5000 ppm [O] = 1000 ppm 0.1
T = 1000 K T = 1000 K T = 1500 K
0.0 0.0
UC1.5 UCNO UC1.5 UCNO UC1.5
UCNO
Figure 6 Effect of O and N contents and temperature on segregation of U/Pu in (U,Pu)C and (U,Pu)C1.5 phases.
In case of M þ MC system, when metal phase pre- the mixed carbide by oxidation and dissolution in
cipitates, the metal phase is richer in Pu, but this HNO3. Excess oxidation of the carbide leads to the
effect decreases with increase in temperature. In formation of M3O8 phase, which has limited Pu solu-
this phase field, increase in nonmetallic impurities bility. This results in excessive segregation of pluto-
results in a slight increase in segregation. Segregation nium in the MO2 þ x phase, giving problems during
behavior is also important for back-end processing of dissolution in HNO3.
64 Carbide Fuel
3.03.3 Fabrication of Carbide Fuel form of a button. This button is partially melted
many times for homogenization before finally melt-
Fabrication of carbide fuel on commercial scale is a ing and dropping into a mold. Pellets as large as
difficult task and needs additional care because of its 1.8 cm in diameter and 15 cm long could be made
pyrophorocity apart from high radio toxicity, and the by this technique. For larger pellets, the skull-melting
concern for criticality restricts the batch size. More- technique was followed, in which molten carbide
over, carbide powders formed during carbothermic forms a shell-type cast in a copper mold and acts as
reduction of oxides are prone to oxidation and hydro- its own containment. Melting casting route results in
lysis. This requires high-purity inert-gas cover large-grained materials compared to that obtained by
(nitrogen or argon) in the fabrication line consisting powder metallurgical methods.22 Melting casting
of glove boxes. The O and moisture content should method is, however, uneconomical due to the high
be less than 25 vppm (each) to minimize O pickup cost of metal fabrication.
during the fuel fabrication process and reduce the
possibility of any fire hazards due to pyrophorocity.
A mixed carbide fuel fabrication facility consisting of 3.03.3.2 Hydride/Hydrocarbon Route
a series of interconnected glove boxes, maintained
This method is followed for small-scale production of
under once-through inert (nitrogen) cover gas is
high-purity carbides, where the metal hydride reacts
shown in Figure 7. In 1960, when research on carbide
with graphite. Actinide carbides MC and M2C3 can
fuel was initiated, three different methods were fol-
be prepared from a mixture of hydride and graphite:
lowed, namely melting casting, metal hydriding–
dehydriding, and carbothermic reduction of oxide. PuH2 þ 0:85C ! PuC0:85 þ H2 ðgÞ ½I
The carbide produced by the latter two techniques
is processed further by powder metallurgy techni- The reaction occurs under vacuum at high tempera-
ques for the manufacture of fuel pellets. ture (1800–2600 C) and this is followed by sintering
at 1500 C for 4 h.23,24
For making UC, the reaction between uranium
3.03.3.1 Melting Casting
metal and a hydrocarbon gas is carried out at
Melting casting process, also known as ‘deep casting,’ 600–800 C, which yields a fine and easily sinterable
was followed during the early days of carbide fuel product. However, the reaction needs to be con-
development. This process was very reliable because trolled carefully, as continued flow of methane pro-
of some advantages over conventional powder metal- duces UC2. Hence, the methane flow must be judged
lurgy route; for example, the products were highly fairly accurately to produce only UC. In this process,
dense and very pure. In this method, UO2 or U metal a fine powder of UH3 is prepared by reacting bulk
chips with graphite are arc-melted and made into the uranium metal with hydrogen at 200–275 C, and
the powder formed is decomposed above 430 C to
produce fine uranium powder. This metal powder
is then reacted with methane (or propane) at
600–800 C25–27 to produce UC.
with a suitable organic binder. The pellets are heated account. Figure 8 shows the process flow sheet for
for 2 h at 1700 C in a vacuum induction furnace. the fabrication of high-plutonium (>30% PuC)
The reacted pellets are ground to a fine powder for mixed-carbide fuel pellet by the conventional pow-
compaction. der pellet route. In a recent development at BARC,
Uranium–plutonium-mixed carbide is also pre- India, for the fabrication of high-plutonium mixed-
pared by carbothermic reduction of UO2–PuO2. carbide pellet for the FBTR, the powder handling
The physical state and homogeneity of the mixture time during grinding/mixing of feed powder and ball
influences the reaction rate and the quality of the milling of carbide powder was considerably reduced
final product. With increase in specific surface area, by incorporation of a high-energy ball mill and
the reaction rate increases but reaction temperature attritor. This has considerably reduced the total
decreases, which reduce Pu loss by volatilization. time for production of a batch and also reduced the
Agglomeration of powders during blending should radiation dose for the working personnel. The pro-
be avoided to ensure good microhomogeneity of the cess control steps for the fabrication are indicated in
mixture. The reaction is controlled by monitoring blue in Figure 8.
the evolved CO gas during reaction using an IR Pu evaporation loss is undesirable, as it may lead
detector in the vacuum pump exhaust. Direct car- to criticality risk by accumulation in the colder
bothermic reduction of UO2 þ x þ PuO2 mixture regions of the furnace. This may also affect fuel
does not give the desired result due to formation economy. Pu loss can be minimized by increasing
of CO2 at low temperatures.28 The resultant reac- the CO partial pressure. Richter et al.30 found that
tion is increasing the CO pressure from 0.8 to 7 Pa reduced
Pu loss from 7.2% to 0.2%.
ð1 zÞUO2þx þ zPuO2 þ ð3 þ n þ 0:5 ð1 zÞÞC
A possible process simplification,31 ‘the one-step’
! ð1 2nÞðU1z Puz ÞC þ nðU1z Puz Þ2 C3 conversion of actinide oxalates precipitated from
½IV nitrate solution (PUREX process) into carbide, was
þ 2CO þ 0:5 x ð1 zÞCO2
studied in CEA, France, as a part of the carbide fuel
When the starting material is UO2 þ x, the reaction is development program for the GEN-IV reactor. This
complicated by the uneven distribution of plutonium work focused on carbothermic reduction of actinide
in the two carbide phases and the formation of (III) oxalate. As a developmental effort, neodymium
CO2 and CO. Therefore, for a better control on the was chosen as a surrogate of Pu, instead of cerium,
reaction mechanism, UO2 þ x could be first reduced Neodymium in oxidation state þIII, has characteris-
to UO2 by heating it with carbon in vacuum at 10–2 tic violet color, which allows easy visual check on
torr at 850 C for 30 min. A mixture of UO2 and homogeneity and precipitation.
PuO2 is then dry-blended with carbon. The reaction
is complete after heat treatment for 1–1½ h at a
3.03.3.3.1 Direct pressing method
temperature of 1400–1450 C for smaller pellets
The conventional powder metallurgy route of car-
and 1550–1600 C for bigger pellets. Carbothermic
bide fuel fabrication has certain disadvantages, for
reduction temperature and atmosphere (vacuum or
example, after the carbothermic reduction when the
inert) are important parameters, as they decide the
carbide clinkers are hammer-milled prior to compac-
completion of the reduction reaction and the extent
tion and sintering, fine carbide powder is generated
of residual O. It was observed at the Institute for
which picks up O impurity from the cover gas. Also,
Transuranium Elements (ITU), Karlsruhe,29 that
handling of the fine powder is risky because of its
carbothermic reduction under vacuum (1 Pa) yielded
pyrophorocity.
better results compared to that carried out under Ar
In the ITU, Richter et al.30 developed the ‘direct
atmosphere and the reduction temperature was
pressing’ method, in which carbide clinkers after
found to be about 100 C less. The nitrogen impurity,
carbothermic reduction in a predetermined shape
the source of which is the Ar cover gas, is also less in
(40–50% TD) are directly pressed to make pellets
the final product. Carbothermic reduction tempera-
of density 65–75% TD, which are subsequently sin-
ture also depends to a great extent on the Pu content
tered to obtain the final product.
of the material. While fixing the C content in the
The advantages of this method are as follows:
oxide–graphite mixture for carbothermic reduction,
the effect of the binder–lubricant, used during press-
Handling of fine carbide powder is reduced, which
ing of MC powder, should also be taken in to eliminates the chances of O pick-up.
66 Carbide Fuel
Analysis:
UO2 Pu, U, C, impurities
Carbothermic
PuO2 Weighing Grinding and mixing Tablet reduction (1485 ⬚C)
Carbon Analysis:
U, Pu, O, N, C,
impurities
Crushing
Phase content
MC clinkers
U, Pu, O, N, C, impurities,
total gas
Phase contents
Pu distribution
Dimension and density
Figure 8 Fabrication flow sheet for production of plutonium-rich mixed carbide fuel for the Indian fast breeder test reactor.
Self-ignition of fine powder in case of accidental The sphere-pac fuel pins prepared at Paul Scherrer
leakage of air is minimized. Institute (PSI)34 for the AC-3 irradiation test at FFTF
Production of dust is reduced. adopted the wet route for production of sintered fuels
No impurities are picked up from grinding in the of low density (80–85%) with high open porosity. In
ball mill during comminution. this method, an acid-free, concentrated feed solution of
The material flows easily for compaction. uranium and plutonium nitrate is mixed with an aque-
It takes less time for processing and hence is more ous solution of hexamethyletetramine (HMTA) urea
economical. and dispersed carbon black. The mixture is cooled and
dropped into a column of hot silicone oil (for making
The direct pressing method has been used for mixed
large particles of 800 mm) or injected into a stream
carbide fuel fabrication method.
of hot oil (for making smaller fuel particles of 80 mm).
The temperature rise in the droplets leads to the
3.03.3.3.2 Sol–gel (wet) route decomposition of HMTA and urea to form ammonia,
The internal gelation route (wet method)32,33 of parti- which, in turn, precipitates ammonium diuranate and
cle fuel fabrication is dust-free and provides ceramic plutonate within the droplet, forming a solid sphere.
fuel with perfect homogeneity of Pu and U on an The details of the fabrication procedure may be
atomic scale compared to the dry powder route. In obtained in Blank5 and Geithoff et al.,6 and is shown
addition, the wet route can be successfully used as a in Figure 9 along with the other methods. About 27
simple, economical, proliferation-resistant, radiologi- sphere-pac pins, which included spare and archive
cally and critically safe, and remotely controlled fuel pins, were fabricated at PSI for loading in the AC-3
production step by recycling the minor actinides (MA) cluster for irradiation in FFTF. The fuel column
from reprocessing step as part of closing the fuel cycle. density ranged from 78.8% to 80.3%.
Carbide Fuel 67
Carbide synthesis
Press Press
Compaction Compaction Form sphere
Blend: Add carbon Disks Disks (gelation)
Hammer mill
comminution Dry
Ballmill
preparation
Powder
Crush
and Calcine
Mill Pressing Repressing
Pelletiz recompact
Reaction sinter
Vacuum inert
Sintering
gas or nitrogen
Press slug (1600–1800 ⬚C)
Sieving
Granulate
Dimensions
Press pellets Pellet control Vibro-fill pin
Pellet fabrication
Density analysis
De-bond
Centerless grid
Load pin
Figure 9 Comparative flow sheet for conventional carbothermic reduction, direct pressing method, and wet method for
fabrication of carbide fuel pellet.
3.03.3.4.1 Chemical quality control method U(IV) form by adding excess of TiCl3. The excess
3.03.3.4.1.1 Uranium and plutonium analysis reductant is oxidized by HNO3. The HNO2 produced
TIMS and ICP/MS are used for isotopic determina- in situ oxidizes Pu(III) to Pu(IV), thereby eliminating
tion of U and Pu. Isotopic dilution mass spectrometry plutonium interference in uranium determination.
(IDMS), X-ray fluorescence spectrometry (XRF), Sulfamic acid is added to destroy any excess HNO2
gravimetry, redox methods with dichromate titration, and the acidity is brought down to 3 M by dilution with
ceric titration, or high-precision titration, and coulo- water. A solution of Fe(III) is added to oxidize U(IV) to
metry are some of the methods used for U and Pu U(VI), and the Fe(II) thus produced is titrated with the
determination. Gravimetry is a fast method for the K2Cr2O7 solution. The advantage of this method is that
approximate determination of U þ Pu of the carbide U can be determined in the presence of Pu and Fe. End
samples by studying the weight change due to com- point detection of the titration can be done by
plete oxidation of the sample and then reducing the the visual indicator method, amperometric method,
oxides to O/M ¼ 2.0 using the CO/CO2 equilibrium or potentiometric method. Biamperometry is more
method.36 IDMS and the redox method are two sensitive than potentiometry for end point detection;
important methods for chemical analysis of samples however, better automation is possible with the poten-
for U and Pu determination. IDMS method is free tiometric technique.
from chemical interference from other elements For Pu determination in the presence of uranium,
but the redox method is more precise. All redox an aliquot is taken in 1 M H2SO4 and an excess of AgO
methods are based on U(VI)/U(IV), Pu(III)/Pu(IV), is added to oxidize all the plutonium in the solution
and Pu(VI)/Pu(III), Pu(IV) systems using different to Pu(VI). The excess AgO is reduced by adding sul-
reducing/oxidizing agents. For chemical analysis, the famic acid, which does not reduce Pu(VI). A known
carbide sample is first converted into its oxide by excess quantity of standardized Fe(II) is added to
heating the samples in air at 500 C, followed by disso- reduce Pu(VI) to Pu(IV) and the unreacted Fe(II) is
lution in HNO3–HF and then removal of fluoride by titrated with standard K2Cr2O7 solution.41
evaporation. Direct treatment of carbide with HNO3 For combined determination of U þ Pu in mixed
does not give a clear solution easily. Moreover, some carbide or determination of either U or Pu in their
organic species such as oxalic acid and mellitic acid respective pure carbides, a common titrimetric method
(benzene hexacarboxylic acid) are reported to form used in nitric acid medium is carried out by reducing
during direct dissolution in HNO3. These species U/Pu with titanium(III) solution in the presence of
affect the electrochemical determination of U/Pu. sulfamic acid, and titrating with cerium(IV) solu-
Different methods have been tried for the dissolution tion42; however, Fe, V, Cu, and Mo, if present, will
of UC and (U,Pu)C sintered pellets for destructive show interference. Therefore, IDMS in combination
quantitative analysis. Chander et al.37 have reported with the redox method gives more reliable results.
that refluxing of carbide with an 18 M H2SO4–15 For the determination of U and Pu in powder and
M HNO3 (1:1) mixture results in complete dissolution pellet forms, a dry and quick method based on XRF
in 1 h. This process also excludes the need to convert is followed. First, standard powder samples of mixed
carbide into oxide, thus reducing the time for analysis plutonium–uranium carbides are prepared and chem-
and the risks involved in oxidation of pyrophoric car- ically analyzed. Then a calibration curve is obtained
bide at high temperature. Moreover, organic species by plotting the intensity ratios of Pu and U of these
such as oxalic and mellitic acids are either destroyed standards against the chemically determined pluto-
or do not interfere in the analysis process. nium content. This calibration curve is used for
Earlier, uranium determination in solution, in pres- subsequent determination of Pu/U ratio in unknown
ence of Pu, was done by the modified Davies and samples. The scatter in the results of determination of
Gray method,38,39 in which U is reduced to U(IV) in Pu and U by XRF were found to be within 0.5 wt%.
a 10 M H3PO4 medium with Fe(II) and then titrated
with K2Cr2O7 solution. However, during processing of 3.03.3.4.1.2 Analyses of nonmetallic elements
these Pu-containing analytical wastes, the strong com- Commonly, the techniques used for assay of carbon
plexing properties of phosphates posed problems. So involve oxidation of carbon to carbon dioxide by
another redox method was developed in which TiCl3 is igniting the sample in a stream of oxygen and then
used as reductant in H2SO4 medium (6 M), contain- flowing the gases generated, passing through the
ing millimoles of HNO3.40 All the U is brought to the Schutze oxidizing reagent (iodine pentoxide on silica
Carbide Fuel 69
gel) to ensure quantitative oxidation of any CO, col- 3.03.3.4.2 Physical quality control
lecting and weighing the CO2, and CO2 determination 3.03.3.4.2.1 Fuel pellet inspection
directly by manometry,43 IR spectrometry,44 con- Sintered fuel pellets are subjected to visual inspection
ductometry,45 or gas chromatography using a thermal for any surface defects. Each pellet is inspected visu-
conductivity detector,46 or indirectly after conversion ally and compared with a standard one for classifica-
of carbon dioxide to methane with subsequent detec- tion for acceptance and rejection. The type of pellet
tion by gas chromatography using Flame Ionization defects are end chipping, surface cracks, etc. Dimen-
Detector (FID).47,48 sional measurements such as length and diameter are
For the determination of nitrogen and oxygen also carried out to estimate linear mass (g cm1) and
impurities, the sample is fused in a graphite crucible to eliminate oversized and undersized pellets. An
at very high temperature along with a nickel flux. In automated pellet inspection system supported by a
this reducing atmosphere, the gases evolved are car- software may be employed for this purpose, which
bon monoxide and nitrogen. A gas chromatograph is reduces the time of inspection considerably.
used for separating the gases. First, carbon monoxide is Ceramographic analysis: Microstructural analysis of
oxidized to carbon dioxide with the Schutze reagent carbide pellets are carried out to measure grain size,
(measured by an IR detector), which is then removed cracks, pores, and inclusions. Typical microstructures
by ascarits and anhydrones. Then the pure nitrogen of mixed carbide pellets containing 70% and 20%
is measured by thermal conductivity.49 The uncer- PuC are shown in Figure 10. Quantitative metallog-
tainty of the methods is 5% for oxygen and nitro- raphy will reveal the extent of sesquicarbide phase
gen impurities and 2% for carbon. present in the sample and this data could be corre-
The metallic impurities are normally determined lated to X-ray diffractometry results. X-ray diffrac-
by ICP emission spectroscopy. A quantitative estima- tometry, however, cannot reveal the second phase
tion of MC and M2C3 contents of mixed carbide M2C3 if it is less than 5%, which, however, can be
samples can be made by the X-ray diffraction line detected by metallography. M2C3 appears as a bright
profile.50 The strongest reflections for MC (200) and and discontinuous phase mostly located at the triple
M2C3 (310) are chosen as analytical lines and the points of the grain structure.
K-values of these lines are calculated from the crystal Metallographic sample preparation of sintered
structure, Lorentz polarization, multiplicity, tempera- mixed carbide fuel pellets needs utmost control of
ture, and absorption factors. The K-values, I (measured atmosphere during etching. Mixed carbide pellets
integrated intensity), and V (volume fraction) of the are prone to oxidation because of their highly cor-
phases are related by the following equation, from rosive nature. The surface of the polished sample
which the volume fraction of the MC and M2C3 can gets tarnished before it can be etched because of the
be calculated (precision 5% of the absolute value): presence of O or moisture in the cover gas atmo-
sphere in the glove box. Oxidation of the polished
VMC KM C IMC carbide pellet surface is indicated by the coloration
¼ 2 3 ½1
VM2 C3 KMC IM2 C3 of the surface.
20 mm
Figure 10 Photomicrograph of mixed carbide fuel with (a) 70% PuC and (b) 20% PuC showing bright/white areas of
sesquicarbide phase. Reproduced from Matthews, R. B.; Herbst, R. J. Nucl. Technol. 1983, 63, 9–22.
70 Carbide Fuel
3.03.3.4.2.2 Fabrication of fuel pins parameters under different conditions. The in-pile
Fabrication of carbide fuel pins consists of loading the fuel behavior depends to a large extent on the fuel
components sequentially as per the fuel pin design. operating temperature, which in turn depends on the
The sequence of loading the cladding tube starts with type of pin or bond concept, namely, He-bonded fuel,
the lower reflector, followed by bottom insulator Na-bonded fuel, or He-bonded particle fuels.
pellets, fissile carbide fuel pellet stack, upper insula- A large number of mixed carbide fuels have
tor pellets, upper reflector, plenum spring, and, been irradiated in countries such as the United
finally, the plenum spacer tube. The closure of the States, France, the United Kingdom, Germany and
cladding tube is carried out at the top end cap using Switzerland, and at the ITU, Karlsruhe, Germany.
tungsten inert gas (TIG) welding in a helium-filled The summary of all such test pin irradiation results
welding chamber. Helium gas acts as an inert cover has been in given Matthews and Herbst,2 Ganguly
gas for the carbide fuel pellets and also as thermal et al.,50 Bart et al.,52 European Institute of Transura-
bond between the fuel pellet and the cladding tube nium Elements,53 and Kummerer.54 Test pin irradia-
ensuring heat energy transfer to the coolant. In case tion was also carried out in Japan at their research
of any faulty welding, end cap should be replaced and reactors JRR-2 and JMTR at JAERI and subsequently
rewelding done. The pins are inspected for dimen- in the experimental fast reactor (JOYO). In India, fast
sional tolerance and bowing. X-radiography, dye- reactor program started with the commissioning of the
penetrant test, and He leak testing are carried out FBTR, where the driver fuel of high plutonium (PuC–
to check the integrity of the fuel pin. 70% MKI) has seen a burnup of 16 at.%. USSR, too,
The technique of fabrication of sodium-bonded has test-irradiated carbide fuel pins in their BR-10
carbide fuel pin is similar to that of sodium-bonded reactor and had a full carbide core (burnup up to
metal fuel pin. The environment of the sodium filling 5 at.%), and subsequently experimental pins of both
box should be kept free of oxygen or moisture types, He-bonded and Na-bonded, were irradiated in
(<50 ppm O and moisture) to avoid any deleterious BOR60 up to 10 at.% burnup. In all these test irradia-
effect on fuel performance because of possible tion programs, mixed carbide powder was used to
O pickup by sodium and oxidation of sintered carbide make fuel pellets or vibrocompacted fuel, except for
fuel pellets. The detailed procedure of making sodium- the Swiss program where sphere-pac fuel pins were
bonded pin is given in Burkes et al.51 In brief, the used. The fuel was made at the Paul Sherrer Institute
procedure of making sodium-bonded pin consists in (Switzerland) and irradiated in the FFTF in the
loading the carbide pellets in the cladding tube filled United States. Thirty-one test pins have been irra-
with liquid sodium and allowing the pellets to settle diated as part of this joint venture.
down by displacing liquid sodium up the cladding tube. The performance analysis of the mixed carbide
Further heating after loading of fuel pellets is necessary fuel can be best understood on the basis of their
to remove trapped gases. Care should be taken to avoid burnup period, the structural changes occurring dur-
the formation of voids or pockets of trapped gases ing the burnup and subsequently the swelling of the
between the fuel and the cladding inner surface. The pin. The basis of such analysis applies to all types of
reason for the formation of such pockets of trapped pin concept, that is, helium-bonded pins, sodium-
gases is the nonuniform flow or flow of sodium in bonded pin, or the helium-bonded sphere-pac fuel.
preferred channels due to nonuniform wetting of the
pellet or the cladding inner surface by liquid sodium. 3.03.4.1.1 Burnup
The upper end plug is subsequently closed. The bond The burnup period could be best understood by
quality of the sodium-bonded pin is assessed by radi- dividing it into three stages based on the fuel surface
ography or eddy current test. temperature which changes with the change in fuel
burnup. The schematic diagram of the changes in
surface temperature and the fuel structural changes
3.03.4 In-Pile Performance are given in Figure 11.
Stage A: At the beginning of life when the virgin
3.03.4.1 Introduction
fuel is introduced into the reactor, it undergoes
Studies on in-pile fuel behavior of carbide fuel help cracking. This results in the movement of the frag-
in controlling the fuel operating parameters under mented fuel toward the cladding surface due to vibra-
normal or steady-state condition during the useful tion induced by the coolant flow. The fuel–clad gap
life of fuel pin and in setting the new operating partly reduces but does not close. The fuel surface
Carbide Fuel 71
on
ati
zones (Figure 12) as given below. These four radial
ific
Fuel surface temperature Ts
II Hig
zones were first conceived by Colin et al.,3 and were
ns
De
h ra
TS(A)
subsequently modified by Blank.55 These zones
IS
te o
ta
become well defined when the fuel sees a burnup of
bl
f
Lo
swe
e 1.3 at.%.
w
fu
el
ra
lling
te
Figure 11 Variation of fuel surface temperature with bonded fuel). However, for a fuel with high oxygen
burnup for a He-bonded mixed carbide fuel; A, B, and C content (>1000 ppm), this zone can exist even below
represent three burnup period. Reproduced from
1000 W cm1. Zone-III contributes to high micro-
Blank, H. Material Science and Technology, Nuclear
Materials, Pt I; Cahn, R. W., Haasen, P., Kramer, E. J., Eds.; scopic swelling. Based on linear heat rating, Zone-II
VCH Publisher: New York, 1994; Vol. 10A. may not exist at all and Zone-I may directly connect
to Zone-III (for stable fuel with linear heat ratings of
1000 W cm1). Zone-IV is the coolest zone of the
temperature reaches its maximum as indicated in fuel cross section, which is next to the fuel cladding.
Figure 11. The first stage may last for several days The as-fabricated (with porosity 10%) structure of
from the beginning of life. the fuel is more or less retained in this zone.
Stage B: In the second stage, unrestrained free
swelling, crack healing, or resintering starts. Resinter- 3.03.4.1.2 Swelling
ing occurs for thermally unstable fuels, which causes Swelling of fuel is a manifestation of the effect of
an initial increase in gap size resulting in the rise in irradiation. As the fuel undergoes irradiation, there is
the fuel surface temperature. This occurs in a very decrease in density and increase in fuel volume,
short period in the beginning of stage B. During this which can be assessed by measurement of dimension
stage, free grain boundary swelling occurs and crack of the fuel pellet and clad diameter. The factors that
healing starts, which results in the relocation of fuel contribute to swelling are (i) fission gases, for exam-
fragments further from the center of the fuel to the ple, Kr and Xe (mostly Xe) having high yield (25%)
periphery. This causes fuel–clad gap closure, increase and low solubility; and (ii) volatile fission products
in heat transfer coefficient, and decrease in fuel sur- such as Cs, Rb, I, and Br. The contribution to geo-
face temperature (Ts). The decrease in fuel–clad gap metric swelling can be assessed by quantitative met-
has a limiting value depending upon the fuel surface allography, SEM, or EPMA. Swelling performance of
roughness and the cladding inner surface roughness. dense fuels depends on the fuel operating tempera-
The decrease in cracks at the center transforms into ture, smear density, fuel–clad gap, and burnup. The
porosity and the formation of wedge type of cracks difference in the operating temperature range for
begins, and when the fuel–clad gap reduces to the helium-bonded and sodium-bonded fuel pin is that,
level of surface roughness of the fuel and clad inside for the former, the working range is at the beginning
surface, fuel–clad deforms elastically at first. Stage of the irradiation as indicated by ‘A’ (Figure 11)
B ends after a burnup of approximately 3 at.%, which, when the fuel clad is still open, and the range changes
however, depends on the type of fuel; for example, to ‘C’ when the gap is closed. The burnup range is
fuel with a high plutonium and oxygen content is less 3 < F < 12 at.%. Sodium-bonded fuel is a relatively
stable, as it densifies easily and swells more. cool fuel compared to helium-bonded fuel. It swells
Stage C: This stage starts after fuel–clad gap clo- very slowly and comes in contact with the cladding
sure when a steady minimum surface temperature is only after about 11 at.% burnup. However, it has
reached and it continues till the end of fuel life. For been found that, beyond a linear heat rating of
carbide fuel, this could be up to 15 at.%. At this stage, 1000 W cm1, the cool fuel concept is not valid. For
more homogeneous fuel–clad interaction starts. helium-bonded fuel, it starts with a short period of
72 Carbide Fuel
Cladding material
Fuel
Swelled fuel
No fuel–clad gap
T/Tm > 0.5, and after the gap closure the fuel temper- gap is more. So, under normal operating conditions
ature drops and major part of the fuel operates below for a linear heat rating <1100 W cm1, the structural
0.5. Only the inner part operates at a temperature >0.5. changes for the sodium-bonded pin are more or less
For helium-bonded pin, the thermal stability limit of constant and predictable. In the beginning (stage A),
the as-fabricated fuel can be matched to the initial fuel the virgin fuel fractures, reducing thermoelastic
operating temperature and some advantages of the hot stresses and drastically changing temperature gradi-
fuel can be achieved. For the purpose, the porosity in the ent of the fuel. The chips from fractured fuel pin may
central zone can be increased at the cost of fuel–clad gap, relocate causing FCMI. However, the concept of
thus enhancing the initial short period ‘A.’ The porosity using a shrouded tube to collect the small fractured
for gas release takes place at the later period of ‘C.’ The pieces to reduce FCMI is not advisable, as it may
outer zone in contact with the clad remains cool during cause unnecessary complications in fuel pin design.
the period ‘C’, and the swelling is partly accommodated There is no abrupt change in fuel temperature in this
into the porosity of the as-fabricated fuel. concept. The decrease in thermal conductivity with
burnup is compensated by a slow decrease in linear
3.03.4.1.3 Performance of Na-bonded and rating. Since the fuel density is high for Na bonding,
He-bonded fuel pins the gas release is less. FCMI can be avoided till end of
3.03.4.1.3.1 Sodium-bonded pin life by proper selection of initial fuel–clad gap.
For the sodium-bonded fuel pin, the fuel density is The sodium-bonded pin has some inherent dis-
higher than that of He-bonded fuel and the fuel–clad advantages; for example, the long residence time of
Carbide Fuel 73
fuel may result in loss of Na bond, which results in stresses generated on the clad due to swelling be
increase in temperature, more gas release, and further circumferential to avoid any failure. Clad carburiza-
reduction of bond quality. Also, Na bond acts like tion is a life-limiting factor, as it makes the clad
a medium for transfer of C from the fuel to clad, surface hard and brittle and cracks form and propa-
causing clad carburization. Fabrication of the Na- gate easily. FCMI can be independent of the fuel–
bonded fuel pin is more expensive than that of the clad gap size when the rate of free swelling is very
He-bonded pin. This will also affect the cost of high (3–10% per at.% burnup) for center tempera-
reprocessing because of the presence of sodium in ture as high as >2000 C. The high swelling cannot
the feed stock. However, with the large fuel–clad gap, be fully accommodated by the fuel–clad gap for a
which is filled with sodium, FCMI can be avoided high-burnup fuel. Increasing the cladding thickness
altogether in the case of a sodium-bonded pin. Also, can make the clad stronger to accommodate restraint
the low temperature of the sodium-bonded pin keeps swelling.
the swelling rate of the fuel low.
3.03.4.1.4 Irradiation creep
3.03.4.1.3.2 He-bonded pin Creep is a time- and temperature-dependent defor-
The performance of He-bonded fuel to a great extent mation mechanism under stress. The source of stres-
depends on the design parameters: namely, fuel–clad ses in the fuel under irradiation is the pressure
gap, smear density, type of clad, pin diameter, linear generated by fission gases produced within the
power, and burnup. He bonding is the most preferred fuel. The growth of fission gas bubbles within the fuel
bonding concept of carbide fuel, partly because Na results in swelling of the fuel by creep deformation.
bonding, apart from the cost, may deteriorate with The fuel–clad gap provided in the fuel pin design is
burnup and hence the bond quality. The in-pile per- utilized to accommodate this swelling known as ‘free
formance of the He-bonded pin depends to a large swelling.’ However, after the fuel–clad gap closure,
extent on the porosity of the as-fabricated fuel, which free swelling is restrained and a back stress is generated
undergoes structural changes when it passes through by the clad on the fuel. This results in restrained
two stages of burnup, ‘A’ and ‘B.’ The structural swelling of the fuel, which is accommodated within
changes at the end of stage B decide the safe burning the available pores of the fuel by creep deformation.
of the fuel in stage C. For He-bonded fuel, the initial Carbide has a close-packed fcc structure of the NaCl
temperature rise is much higher at the beginning of type. Unlike close-packed metals, deformation of car-
life because of the lower thermal conductivity of the bide requires much higher stresses because of the
He bond compared to Na bond. The lower thermal strong covalent bonding existing between the metal
conductivity of the He bond also warrants the reduc- atom and the carbon atom. It also needs sufficient
tion of the fuel–clad gap compared to the Na-bonded thermal excitation to leave the lattice site. Accommo-
fuel. As a consequence, the fuel density for He- dation of fuel swelling (restrained) within the porosity
bonded fuel is lower and a fabrication porosity of of the ceramic carbide fuel results in the extension of
about 15% is recommended (equivalent to 85% pel- the primary creep region and this causes an increase
let density). in the strain rate before it reduces and attains the
The fuel–clad gap closure in the early burnup steady-state creep. The tertiary creep region lies out-
period results in lowering of the fuel temperature, side the life span of the carbide fuel and hence is not
and free swelling changes into restrained swelling important.
under the contact pressure developed at the clad– Two types of creep deformation are operative in
fuel interface. Hence, the mechanical properties of the fuel: temperature-dependent thermal creep and
the fuel and clad (creep, fracture toughness) to a great radiation-induced or radiation-assisted creep. Accord-
extent predict the fuel behavior during the remaining ing to Seitzer et al.56 and Dienst,57 thermal creep starts
period of burnup. It has been observed that the He- at 1000 C and dominates over radiation-induced
bonded pin can be operated safely up to 15 at.% creep. However, it was also inferred that the relatively
burnup (peak burnup 20 at.%). The irradiation expe- flat and lower temperature profile in carbide due to
rience with various design parameters has been sum- higher thermal conductivity compared to mixed oxide
marized by Matzke.29 and the high neutron fluence provides sufficient means
Cladding breaches due to FCMI or FCCI can be for the reduction of stresses in the fuel by radiation-
due to loss of ductility of the clad, carburization of induced creep. The details of plasticity of carbide can
the clad, or fuel swelling. It is desirable that the hoop be obtained in the reviews by Routbort and Singh58 and
74 Carbide Fuel
MC1+x
1´10-2 (U0.45Pu0.55)C - MKII10
(U0.31Pu0.69)C0.93 63
104 (U0.3Pu0.7)C - MKI10
1´10-3
(U0.79Pu0.21)C63
de/dt (h-1)
1´10-4
UC1–x
1´10-5
Irradiation creep
s
1´10-6 103
TC TS
-7
1´10
5 6 7 8 9 10
104/T (K-1)
distributed in the MC phase, hindering dislocation induced by FCMI and fission gas pressure. The FC-1
motion and hence reducing creep deformation. It is FFTF experiment (a full-size, 91-rod FFTF assem-
recommended that the amount of M2C3 phase in the bly) attained goal burnup with breach. A peak fuel
fuel should be optimized to allow creep to occur. burnup of 20 at.% in 10 MC fuel rods clad in type 316
Though Pu-rich carbide fuel is harder than U-rich stainless steel was achieved in EBR II. Of those rods,
fuel, beyond a temperature of 1553 K (average volu- five had experienced a 15% transient-overpower test
metric temperature of fuel), the plutonium-rich fuel in EBR II after attaining 12 at.% burnup. Thirteen
behaves in the same way as the U-rich carbide fuel, other He-bonded rods and three Na-bonded rods
for which in-reactor performance indicated no failure attained 16 at.% burnup in EBR II without breach.
due to FCMI. Hence, from extrapolation it could be The FFTF AC-3 experiment results showed that,
presumed that the Pu-rich fuel will also behave in a for the relatively low-temperature conditions of the
similar manner. test, the pellet fuel and sphere-pac exhibited only
minor observed differences in behavior, and both
3.03.4.1.5 Experience on irradiation types of fuel performed in a manner consistent with
performance the rest of the MC fuel database. Carbide fuel failures
3.03.4.1.5.1 US experience typically result from FCMI, due to high fuel swelling,
The US experience of carbide fuel irradiation has which leads to early fuel–cladding gap closure. Also,
been summarized recently by Crawford et al.66 because it generally operates at relatively low tem-
A large number of He-bonded and sodium-bonded perature, fuel creep is not effective in relieving clad-
pins were irradiated in EBR II as an initial test for ding stress. For this reason, MC fuel pin design must
subsequent loading in FFTF to simulate conditions incorporate a large fuel–cladding gap and make use
for a large development plant (LDP). The irradiation of a low-density fuel to delay the onset of FCMI.
condition simulates the peak-cladding temperature No fuel failures have been attributed to the clad-
condition and the peak power condition of FFTF. ding carburization phenomenon. Ten transient-
The objective of this study was to see the effects of overpower tests of MC fuels were conducted in
pin diameter (7.89–9.40 mm), pellet density (81 TREAT using fuel irradiated in EBR II (burnups
and 87% TD), pellet–clad gap (0.13–0.78 mm), type up to 12 at.%) for the purposes of establishing that
of cladding alloy, and bond type (He or Na) on cladding breach would occur above 115 and 125%
in-reactor lifetime, fuel and cladding swelling, fission overpower. The results suggested FCMI-induced
product behavior, fuel–cladding mechanical interac- breaches, but most importantly indicated comfort-
tion, and fuel restructuring. About 470 MC fuel rods able margins to failure. The rods indicated only
with sodium or helium bonding were irradiated in small cladding strains and small amounts of liquid-
EBR II with different types of cladding: for example, phase penetration of the cladding. The conclusion of
SS316, PE-16, stainless steel D9, and D21. Over 200 these tests was that nothing in fuel transient-over-
MC fuel rods were irradiated in FFTF in two assem- power response would prevent or limit application of
blies: the ACN-1 experiments with rods fabricated MC fuels to fast reactors. The EBR II tests also
using 316SS and D9 cladding; and the FC-1 test, a included rods irradiated beyond goal burnup to
full-size, 91-rod FFTF assembly with 316SS and D9 breach, and one intentionally defected rod irradiated
cladding and ducts. The AC-3 test consisted of 91 for 100 days beyond cladding breach. The defected
full-size, D9-clad rods of which 25 rods contained the rod exhibited a reaction between the fuel and the
sphere-pac fuel made in PSI, Switzerland, and 66 coolant (O in the coolant), which resulted in a higher
rods contained pellet fuel irradiated to 9 at.% burnup specific-volume reaction product and caused expan-
without breach. The higher thermal conductivity of sion as well as widening of the defect with no release
carbide makes the average fuel operating tempera- of activity into the coolant. Other rods irradiated to
ture very low and it behaves like a brittle material natural breach in EBR II did not exhibit that phe-
showing cracks. However, the cracks resulted in fuel nomenon. MC fuels appear to operate benignly after
relocation and did not cause premature pin failure. In cladding breach. In another experiment, in EBR II,
the EBR II test 21, fuel breaches were observed the effect of sodium void was tested by irradiating an
before reaching the goal burnup. Out of this, 15 MC fuel rod with a purposely induced void, simulat-
were PE-16-cladded rods, and clad failure was attrib- ing void formation due to Na expulsion during irra-
uted to irradiation embrittlement of the cladding diation. The fuel rod exhibited microstructural
alloy, rendering it less capable of enduring the stress changes, reflecting a local high fuel temperature
76 Carbide Fuel
and no loss of cladding integrity. This experiment by direct pressing and irradiated at a linear power of
indicated that MC fuel would withstand an Na bond 730 W cm1 with a burnup of 5.8 at.% is being carried
expulsion of some magnitude. The US experience out and results are awaited.
with MC fuels was not very large; however, it was A short-term irradiation experiment at the begin-
sufficient to instill confidence that such fuels have ning of life in the High Flux Reactor Petten was carried
irradiation performance adequate for use in SFRs. In out for He-bonded mixed-carbide low in oxygen
particular, He-bonded rods clad in lower swelling (CARLO) and rich in oxygen (CARRO).53 The U:Pu
cladding alloys with around 80% fuel smeared den- ratio was 80:20, the O content was 290 and 1590 ppm,
sity appeared to show the best performance and the carbon content was 4.85 and 5.05 wt% for
potential. CARLO and CARRO, respectively. The pellet diame-
ter was 6.4–6.2 mm. The clad ovalization was 4
3.03.4.1.5.2 German fast breeder project and 6 mm before irradiation and 16 and 20 mm
Kummerer54 summarized the results of irradiation after a burnup of 0.5 at.% for CARLO and CARRO,
experiments under different conditions based on a respectively. The main objective of this test was to
reference concept. Altogether, 101 pins were irra- study the thermal stability of the fuel. Axial g-scanning
diated, out of which 83 pins were irradiated under of the pins indicated that for CARLO the 137Cs activity
steady-state condition, which includes 15 failed pins. was more at the extremities of the pin having a lower
Eighteen pins underwent steady-state and cycling, temperature. It was also observed for CARRO, but the
which reported one pin failure. The parameters, for concentration profile was less pronounced.
example, pin diameter, type of bond, fuel–clad gap, For CARLO fuel, wedge type crack was observed
smear density, cladding material and wall thickness, at the periphery, indicating strong difference in the
linear power, and burnup, were varied. Based on the swelling between the fuel center and periphery.
results of these irradiation, a reference pin concept CARRO did not show any such crack indicating
with cold-worked austenitic steel (1.4970) cladding, swelling at the periphery. There was no fuel–clad
pellet diameter 7.0 mm, pellet density 84% TD, gap for CARLO but for CARRO a gap existed, indi-
fuel–cladding gap of 400 mm, helium bond, smear den- cating in-pile densification at high temperature.
sity 75% TD, pin diameter 8.5 mm, and clad wall From the analysis of the pellet geometry after irradi-
thickness of 0.55 mm evolved for further irradiation ation, it was observed that in-pile densification in
for final performance test. He-bonded carbide fuel is not a general phenomenon
The ‘NItrure Mixte dans a Phenix (NIMPHE)’ but is restricted to unstable CARRO fuel which is
fuel irradiation experiments were designed to study exposed to high temperature at the beginning of life
the influence of the fabrication route on the fuel and that this condition was expected to exist for fuel
stability. The direct pressing route, as proposed by pins with large initial gap and/or higher linear heat
Richter,67 produces pellets with improved thermal rating. O plays a very vital role in stronger radial
stability; however, the starting materials, the reaction transportation of Pu and less axial Cs transportation
temperature, and the pressing conditions, all influence and increase in FCCI by carburization of the clad.
the thermal stability. Also, direct pressing of granules The excess O combines with Cs and forms a complex,
decreases the densification under irradiation. thus reducing the mobility of either O or Cs.
(a) (b)
(U0.8 Pu0.2)C1.0 (U0.8 Pu0.2)C1.1
(c) (d) the pellet (Figure 16) took place at the central part
of the pellets with the formation of large pores. Migra-
Figure 15 Photomicrographs of (U0.3Pu0.7)C1 + x fuel
tion of semivolatile Cs to plenum and other cooler
pin at different burnups: (a) 25 GWd t1, (b) 50 GWd t1,
(c) 100 GWd t1, and (d) 155 GWd t1. Reproduced from parts of the pin was detected. Clad carburization at the
Sengupta, A. K.; Basak, U.; Kumar, A.; Kamath, H. S.; grain boundary near the surface was also observed.
Banerjee, S. Experience on mixed carbide fuel with high The results of the other reported irradiation up to
plutonium content for Indian fast breeder reactor- an 3 at.% are awaited.
overview; Journal of nuclear materials 385; 2009;
pp 161–164.
3.03.4.1.5.6 French experience
Between the years 1960 and 1970, about 80 MC fuel
pins (80% sodium-bonded), including two NIMPHE
and clad. After 10 at.% burnup, the pellet–clad gaps
pins in collaboration with TUI as mentioned above,
were not observed at the center of the fuel column.
were irradiated in MTRs (OSIRIS, SOLOE) and then
The maximum increase in stack length was 2.61%.
in Rapsodie and PHENIX reactors.70 The sodium-
Radiography of fuel pins after 15.5 at.% burnup
bonded fuel had a high density (>90% TD) but low
showed pellet–pellet gap at the end of the column,
smear density of about 70–80% TD with a large fuel–
and the stack length increase varied from 2.7% to
clad gap. The fuel could reach a burnup of 14 at.%
3.7%. Maximum fission gas release was 16% and the
without a limited number of clad failures. The
corresponding internal pressure in the fuel pin was
swelling rate was 1.6%/at.% burnup below
2.1 MPa after 15.5 at.% burnup. The Xe/Kr ratio was
1000 C. They perceived some major design risks,
around 13. The cross-sections of the fuel at various
such as lack of sodium bond, locally resulting in over-
burnups up to 15.5 at.% are shown in Figure 15. It
heating of the fuel, excessive swelling, and hence a
indicates circumferential cracks due to thermal stres-
possible clad failure. Clad carburization in case of
ses, and discrete zone free from any porosity is
sodium-bonded pin was also presumed to be an issue
observed near the outer circumferential area. This
which might cause clad embrittlement. Also, trapping
was attributed to creep of the fuel due to FCMI. The
of small carbide chips in the large gap between the fuel
end of the fuel column reveals circumferential crack-
and the clad may also cause localized clad straining.
ing and fuel–clad gap closure.
For the He-bonded pins, it was observed that smear
density should be limited to 70–75% TD for high
3.03.4.1.5.5 Japanese experience
LHRs (800–1000 W cm1) and high burnup
The capsule irradiation tests of carbide fuels (C/M
operation.
ratio of 1.0 and 1.1) started in 1983 in JFR 2 and
In summarizing the international experience on
JMTR.69 Three capsules of carbide fuels containing
carbide fuel performance, the following conclusions
two pins in each were irradiated at about 1.5 and 3 at.%
can be drawn:
burnup. The pin diameter was 6.5 and 9.4 mm, and the
linear heat rate was 450 and 650 W cm1. After 1.5 at.%
Sodium-bonded fuel design has some intrinsically
burnup, characteristic cracking and restructuring of good performance characteristics, such as higher
78 Carbide Fuel
LHR compared to the He-bonded design; how- sodium-bonded fuel was more compared to the
ever, it has several drawbacks such as a more com- He-bonded pins.
plex process of sodium-bonding technique, their
Out-of-pile burnup simulation experiments showed
more stringent quality assurance, and a greater that neither individual fission products nor the
extent of clad carburization. global chemical effect of burnup was detrimental
Swelling of carbide fuel is an issue when compared was 2–3 times greater than that in helium-bonded
with oxide fuel. Both solid fission products as well elements. The carburized layer in sodium-bonded
as fission gases dissolved in matrix contribute to pins was as high as 150 mm. However, no fuel ele-
swelling. Swelling rate increases by growth and ment failures in both He- and Na-bonded pins had
merging of bubbles and it is a function of tem- been attributed to clad carburization.
perature. Beyond a critical temperature, swelling
increases drastically. Saibaba et al.17 have estimated the carbon potential
and Pco of hyperstoichiometric (Pu0.7U0.3)C1 þ x up to
2200 K as a function of O, N, and M2C3 contents, and
3.03.4.1.6 Fuel–clad chemical interaction
Pillai et al.72 have reported the carbon potential data
Carbon activity and partial pressure of CO are
of SS316 up to 1000 K. These calculations show that
important parameters responsible for clad carburiza-
tion, which make the clad brittle, resulting in clad
the carbon potential of hyperstoichiometric
breach. In sodium-bonded fuel, carbon transfer from (Pu0.7U0.3)C containing relatively high M2C3
the fuel to the clad takes place by dissolution of (20%) is lower than that of (U0.8Pu0.2)C with
carbon in sodium liquid. However, in He-bonded very low M2C3(2%);
fuel pins, carbon transfer takes place through CO.
the Pco of (Pu0.7U0.3)C1 þ x containing relatively
After 2–3 at.% burnup, the fuel may swell and come high oxygen (–1 wt%) is lower than that of its
in direct contact with the clad. But, transport of uranium-rich counterpart with very low oxygen
carbon down the temperature gradient within the (0.1 wt%);
fuel also takes place through CO.
carbon activity of the fuel increases with increase
Thermodynamic computations and experimental in temperature but that of SS316 decreases with
studies on the chemical compatibility between temperature.
mixed carbide, SS316, and sodium coolant have
been restricted to uranium-rich (U/(U þ Pu) > 0.7) In the MC þ M2C3 phase mixture, plutonium
compositions. The results of the out-of-pile and in- concentration is greater in the M2C3 phase. On
pile experiments as summarized by Ganguly and extending the same basis of calculations to hyperstoi-
Sengupta71 are given below: chiometric (Pu0.7U0.3)C1 þ x, the extent of plutonium
segregation in the M2C3 phase has been found to be
The chemical compatibility between mixed carbide more than 90%. The mixed sesquicarbide in this case
fuel and sodium was excellent. Out-of-pile test with could, therefore, be assumed to be pure Pu2C3. On
helium-bonded MC1 þ x containing up to 12 wt% the basis of thermodynamic data, the calculated car-
M2C3 at 873 K for 1000 h showed no clad carburiza- bon potentials of PuC, UC, U2C3, and Pu2C3 as a
tion, but at 1000 K carbide precipitation along function of temperature are shown in Figure 4. The
the grain boundaries and slip lines in a zone up to carbon potentials of PuC and Pu2C3 are lower than
50 mm in depth from the fuel–cladding interface, was that of SS316 up to 1300 and 1200 K, respectively.
observed. The hardening of the SS316 cladding The carbon potential of UC is lower than that of
up to 100 mm was observed. Similarly, out-of-pile PuC, Pu2C3, and SS316 at all temperatures but the
experiments with sodium-bonded MC1 þ x showed values for U2C3 are relatively high. The carbon
no carburization at 873 K, but at 1000 K an increase potential of hyperstoichiometric plutonium-rich
in carbide precipitation in grain boundaries mixed carbide has been found to be lower than
and along slip lines was noticed up to 90 mm its uranium-rich counterpart at all temperatures.
of the surface. The depth of carburization for Theoretical calculations, therefore, indicate that
Carbide Fuel 79
residence time of the fuel in the reactor, and with reasonable fission yield do not form stable mono-
dissolution in MC lattice (6 mol.%). Ln2C3 shows in the corrosive effects of iodine. However, the
limited solubility in isostructural U2C3 but consider- g-radiation field of the reactor dissociates cesium
able solubility in Pu2C3.74,75 It was observed that, in iodide molecules, which is followed by ionization of
high-burnup, irradiated, and simulated carbide fuels, iodine,78 thereby increasing its pressure by a factor
the solubility of lanthanides in carbide fuel is much of 107 in a FBR.79 Therefore, iodine is still an
lower than the sum total of solubilities of the individ- element of concern in carbide fuels, though its
ual lanthanide carbide. In Pu-rich carbide fuel, higher effect is significantly reduced due to its bonding
solubility of lanthanides in the sesquicarbide phase with Cs. However, fuel–clad chemical compatibility
compared to the monocarbide phase results in a will not be an issue, as the life-limiting factor in
decrease in carbon potential of the fuel with burnup carbide fuels will be FCMI at a stage when the fuel–
and this effect is more pronounced at high tempera- clad gap is closed.73
tures. With increase in burnup, the difference in car-
bon activity of the center and the surface of the fuel
3.03.4.3 MA-Containing Fuel
decreases, which should result in reduced mass trans-
fer of carbon from the center to the surface. In case of Advanced fast reactor fuels (carbide or nitride) such
U-rich fuel, limited solubility of Ln2C3 in U2C3 will as metal or oxide fuels are likely to contain recycled
result in a smaller decrease in carbon potential with MA, for example, Am, Np, and Cm. They are also
burnup. Depending upon the temperature of fuel, referred to as transmutation fuels. The goal of devel-
U-rich fuel with MC2 phase will stabilize LnC2 and opment of such a fuel is to develop and qualify a
reduce the carbon potential of the fuel. nuclear fuel system that performs all the functions
Kleykamp76 indicated that, depending upon the of a traditional fast-spectrum nuclear fuel while
amount of O present in the fuel, there could be destroying recycled actinides. The transmuted fuel
formation of oxides of lanthanides instead of carbides has a high burnup potential, resulting in long, unin-
and this would prevent decrease in the C/M ratio to terrupted burning of the fuel thereby reducing the
some extent. The sesquioxides of lanthanides are spent fuel volume and eventually the fuel cycle cost.
more stable than the oxides of uranium or plutonium. The potential of fully closed actinide recycle is well
In fact, both lanthanides and alkaline earths act as understood, as the separation of the MA from the
oxygen getters in the carbide fuels. Thus, with spent fuel and their recycling substantially reduce the
increase in burnup, the oxygen dissolved in the MC long-term radiotoxicity in the waste, for example,
phase reduces. The PIEs of carbide fuel reported in from 10 000 to 500 years. The Superfact experiments
the literature have also confirmed the presence of performed by CEA and ITU in the Phenix reactor
oxides of lanthanides and alkaline earths. Thus, the using oxide fuel are considered to be a milestone in
formation of stable sesquicarbide of lanthanides this respect. The irradiation performance database
decreases the carbon potential of the carbide fuel for MA-containing MC or MN fuels is limited and
with burnup, but oxygen in the fuel binds some of in the very initial stage of development compared to
the lanthanides, thereby diluting this effect on the mixed oxide or metal fuels.
carbon potential. Carbide fuels can be used in MA incinerator reac-
Another important chemical species detrimental tors because of their hard neutron flux. Very limited
to carbide fuel performance is iodine, which is very information is available on the carbides of MA. Np
corrosive for the SS cladding. Gotzman and Ohse77 forms stable monocarbide and sesquicarbide, whereas
carried out out-of-pile experiments on the chemical only sesquicarbide of Am is known so far. The Gibbs
compatibility of UC with Cs, Te, I, and Se energies of formation of these carbides are similar to
simulated fuel corresponding to 10 at.% burnup. those of uranium and plutonium carbides (DfG
They concluded from their experimental results (1000 K) in kcal mol1 ¼ 25.4 (UC), 19.3 (NpC),
that carburization was possible only when the virgin 12.7 (PuC), 66.5 (U2C3), 51.2 (Np2C3), 38.5
fuel (no simulated fission product) was in contact (Pu2C3), 18 (Am2C3)).80 As MA carbides and U/Pu
with stainless steel. However, with Cs, Te, I, and Se carbides are isostructural, carbides of MA are
addition to the fuel UC1 þ x, there was no incompat- expected to dissolve in fuel matrix to some degree.
ibility. This is due to the high thermodynamic sta- As the fission yield of MA is very low, they will not
bility of gaseous Cs2I2 and CsI; most of the iodine affect the fuel’s thermochemical behavior in a regular
available in the irradiated carbide fuel is present fuel. But in case of incinerator carbide fuels with a
as Cs compounds, resulting in a drastic reduction high fraction of MA, the stability of MA in different
Carbide Fuel 81
phases will have an appreciable effect on the carbon fuel swelling and is also an additional source for fuel
activity of the fuel. pin overpressurization. The data on in-pile perfor-
The procedure for the fabrication of MA- mance of MA-containing carbide fuel are limited,
containing carbide fuel is the same as that of conven- unlike oxide or metal fuel, and the scope of its use is
tional carbide fuel fabrication, that is, by carbothermic still wide open. Under the US fast reactor program,
reduction of the oxide feed materials. In this case, the reference and alternative 1000 MWth advanced
feed material should be (U,Pu,MA)O2. The reduction burner reactor (ABR) core concepts were developed
temperature is around 1500 C under low vacuum to using ternary metallic (U–TRU–Zr) and mixed
prevent volatilization loss of Pu and Am. An additional oxide fuel with UO2 þ TRU. However, better ther-
consideration for pellet-type fuel fabrication is the mal and neutronic properties of carbide and nitride
effect of radiation and decay heat on the binders used fuels prompted them81 to consider TRU carbide as
during pellet pressing. Traditional organic or metalor- an alternative fuel form. ABR core concepts with
ganic binders tend to decompose quickly upon incor- these fuels were developed based on the reference
poration into a system with high MA loading. High a 1000 MWth ABR core with the expectation that the
activity, high decay heat, high neutron emission rate, carbide fuel can mitigate the disadvantages of both
and high g activity, all require special consideration metallic and oxide fuels in the ABR: for example,
during the fabrication of MA-containing fuels. During favorable passive safety features in a severe accident
fabrication, one needs to take precautions such as use compared to the oxide core, a higher discharge
of biological shield from g and neutron emitters and burnup compared to the metallic core, and a poten-
containment to prevent the risk of incorporation of tial to increase thermal efficiency. For consistency,
toxic radioactive material into the body. The presence the potential design goals used in the reference
of MA in fuel necessitates the use of hot cells or ABR core concepts were also employed in this
hybrid laboratories, combining standard glove boxes study: a 1000 MWth power rating, medium TRU
with remote operation by telemanipulators. Lead (Pb) conversion ratio of 0.75, a compact core, 1-year
glasses provide shielding from g emitters, and a com- operational cycle length at least with a capacity
bination of water, lead, and cadmium (or boron) pro- factor of 90%, and sufficient shutdown margin. Gen-
vides shielding from neutrons emanating directly erally, it has been observed that the core performance
through radioactive decay (e.g., 244Cm) or by (a, n) parameters of the carbide and nitride fuels in the
reactions, which typically occur with light elements ABR core lie between the values of the metallic
(O, N, or C) in the fuel. Handling of MAs and fabrica- and oxide cores. The neutron spectrum is softer
tion of MA containing fuel is best undertaken in a dust- than that of the metallic core, but harder than that
free atmosphere, and hence processes that reduce dust of the oxide core. The fuel resident time of
production are not just desirable but essential. In this 60 months is longer than that of the metallic core,
respect, sodium-bonded fuel with large pellet–clad but shorter than that of the oxide core. The required
gaps allows more tolerances in the fuel pellet diameter TRU enrichment (25% for the carbide fuel) and
and elimination of the centerless grinding step which average discharge burnup of 100 GWd t1 are also
generates waste in the form of dust and contaminates higher than those of the metallic core, but lower than
the cell’s inside surface. The wet route with automated those of the oxide core. This indicates that the core
handling facility in a hot cell under nitrogen cover gas can be more compact or the operating temperature
is the best possible option. In the ITU, Karlsruhe,28 a increased further to reduce the plant cost. The replace-
qualified advanced process of fabrication of MA- ment of the bonding material by helium gas increases
containing fuel exists, which consists of glove boxes the fuel temperature by 200 C, but overall impacts
shielded by water (50 cm) and lead (5 cm) for handling on the kinetic parameters and reactivity feedback coef-
a limiting mass of 150 g of 241Am and 5 g 244Cm. In this ficients are negligibly small. As a result, the carbide
process, all operations are carried out by remote con- core satisfies all sufficient conditions for core passive
trol telemanipulators. safety that are not met by the oxide core. The good
The irradiation behavior of MA-containing fuels thermal conductivity and high melting temperature
is different from that of conventional fuels. lead to a significant decrease in the average fuel tem-
Increased fuel pin helium gas inventory is expected perature (comparable to metallic fuel), and hence pro-
due to capture and decay sequences associated with vide a large margin to fuel melt and favorable passive
241
Am and 242Cm, which decay by a emission (half- safety features without additional design fixes that
life of 169 days) to 238Pu. This could lead to higher were required in the oxide core concepts.
82 Carbide Fuel
3.03.5 Fuel Reprocessing and Waste in different laboratories across the world to find a
Management solution to this problem. It was found that the advanced
dissolution process with silver-catalyzed electrolysis
This aspect of the nuclear fuel cycle has been dis- could circumvent this problem. A study with the use
cussed in Chapter 5.14, Spent Fuel Dissolution and of ozone for the destruction of organic species has also
Reprocessing Processes; Chapter 5.16, Spent Fuel been successful.83 Most of the experiences in the pro-
as Waste Material; Chapter 5.18, Waste Glass; cessing of carbide fuel are limited to laboratory-scale
Chapter 5.19, Ceramic Waste Forms; Chapter experiments. A compact hot-cell facility has been com-
5.20, Metallic Waste Forms; Chapter 5.21, Graphite missioned in Kalpakkam, India,84 for reprocessing
and Chapter 5.22, Minerals and Natural Analogues FBTR spent fuel and to establish the process and
of this comprehensive and may be referred to for equipment for high-burnup, plutonium-rich fuels
further information. This section highlights carbide- (Figure 18). For this purpose an advanced Purex pro-
specific issues. Different nations have different cess has been developed. The fuel pin is chopped and
back-end strategies: for example, direct disposal dissolved in an electrolytic dissolver. For recycling,
(once-through fuel cycle), storage and postponed deci- Pu and U are separated in the pure form after centrifu-
sion (wait and watch), and reprocessing and recycling gation and three cycles of solvent extraction. The sol-
(closed fuel cycle). The reprocessing of the carbide vent extraction process is based on the tri-n-butyl
fuel can be carried out by the hydrometallurgy route phosphate (TBP) solvent. Detailed studies have been
or the pyrometallurgy route. Due to the extensive carried out on the various parameters that influence
work carried out on the oxide fuel, especially LWR the third phase formation: for example, effects of car-
fuel, the hydrometallurgy route via the PUREX pro- bon chain length of the diluents, acidity of the aqueous
cess is the most developed technology for the fuel phase, etc. TAP (tri-n-amyl phosphate)/n-dodecane
processing. The same process can be used for burnt has been found to have high capacity to load Pu(IV)
carbide fuel with some modifications since carbide fuel without third phase formation under normal extrac-
from an FR differs from the burnt fuel of LWR because tion conditions as compared to TBP/n-dodecane.
of the following issues: Long-chain monamides are also being investigated
as alternate extractants. The amides, for example,
and economically viable option, the fuel cycle should GEN-IV reactor concept eliminates the possibility of
be a closed cycle where the technology of fuel repro- any fire hazards due to sodium, of which there have
cessing and waste management also plays a vital role. been some reported instances. (See Chapter 2.04,
In such a case, the recycling of MA needs to be Thermodynamic and Thermophysical Properties
considered, as it will significantly reduce the waste of the Actinide Carbides and Chapter 5.14, Spent
load and also make the fuel proliferation-resistant. Fuel Dissolution and Reprocessing Processes).
The technology for reprocessing of carbide fuel is not
well established, although a small-scale facility has
been set up and demonstrated by IGCAR, Kalpakkam, Acknowledgments
India, for reprocessing of mixed carbide fuels based
on the PUREX process. The problem of enhanced The authors express their deep sense of gratitude to
formation of PGM, which form insoluble residues Dr. Srikumar Banerjee, Chairman, Atomic Energy
along with fissile elements, and the formation of Commission, Department of Atomic Energy, Gov-
some complex organic compounds during dissolu- ernment of India, for his guidance, support, and valu-
tion, which interfere during separation process by able suggestions in structuring this review.
forming a third phase, need to be understood and They also thank Dr. V. Venugopal, Director,
the process steps need to be modified suitably. The Radiochemistry and Isotope Group, BARC, India,
fabrication of MA-containing carbide fuels needs for his keen interest and encouragement during the
remote and automated operations in hot cells. course of this work.
A sodium-bonded fuel option having a large fuel– Thanks are also due to Mr. Kasi Viswanathan,
clad gap (having greater tolerance for fuel pellet Associate Director, GRIP Group, for providing the
diameter) will have some advantages over the irradiation examination data of MKI carbide fuel for
He-bonded fuel having low tolerance for fuel pellet FBTR.
diameter. The centerless grinding step of the fuel The authors also sincerely thank Ms. Mareena
pellet diameter, which generates dust, is not required Michael for her help in preparing the manuscript.
in this type of fuel. A remote and automated wet Contributions made by staff members of the
route (gelation method) of production of carbide Radiometallurgy Division and Fuel Chemistry Divi-
fuel will be an attractive option, as it would reduce sion, BARC, and Reprocessing Division, IGCAR,
the dust hazards of radioactive powder material. Kalpakkam, towards both the front end and back
Of late, there has been renewed interest86 in the end of the high-plutonium-containing carbide fuel
use of carbide fuel for ‘gas-cooled fast reactor (GFR).’ cycle are gratefully acknowledged.
as part of ‘The Generation IV Project.’ The GFR has
the advantage of a fast neutron spectrum as well as of
high temperature. The high outlet temperatures pro- References
vide improved economy in the power conversion unit
and permit process heat application. The design and 1. Matzke, H. J. Ceramics Int. 1991, 17, 315–323.
development of an innovative refractory fuel in an 2. Matthews, R. B.; Herbst, R. J. Nucl. Technol. 1983, 63,
9–22.
advanced ceramic cladding has been the aim of this 3. Colin, M.; Coquerelle, M.; Roy, I. L. F.; Rondei, C.;
concept. One of the concepts uses mixed carbide Walker, C. T.; Blank, H. Nucl. Technol. 1983, 63,
pellets (cylindrical pellet or disk with low L/D 442–460.
4. Majorshin, A. A.; Zabud’ko, L. M.; Bibilashiviliji, Y. U. K.;
ratio), and fiber-reinforced SiC–SiCf composite has Bogratur, S. M. Use of carbide-nitride fuel in the USSR.
been selected as cladding material. Mixed nitride and A review; IAEA TECDOC-466; IAEA: Vienna, 1988;
oxide has been considered as reserve options. The pp 53–59.
5. Blank, H. J. Less Common Metal 1986, 121, 583.
fuel configuration will be either a classical pellet in
6. Geithoff, D.; Muhling, G.; Richter, K. J. Nucl. Mater. 1992,
tube or an innovative plate type: that is, the fuel pellet 188, 43–48.
is disk shaped with a low length-to-diameter ratio in 7. Levine, J. P.; Nayak, N. P.; Boltax, A. In Transaction of the
7th International Conference on SMiRt, Cahigo; New York,
a honeycomb structure and the cladding is in the
Amsterdam, North Holland, 1981; Vol. C5/2, p 161.
form of a plate covering both sides. A tungsten layer 8. Alder, H. P.; Ledergerber, G.; Stratton, R. W. Advanced
is provided between the fuel and the clad to prevent fuel for fast breeder reactors produced by gelation;
IAEA-TECDOC-466; 1988, pp 81–96.
any interaction between them and also to contain the 9. Benard, H. J. Nucl. Mater. 1989, 166, 105–111.
fission products. Helium gas will be used as coolant 10. Sengupta, A. K.; Basak, U.; Kumar, A.; Kamath, H. S.;
instead of sodium. The use of He as a coolant in Banerjee, S. J. Nucl. Mater. 2009, 385, 161–164.
Carbide Fuel 85
11. Russell, L. E.; Bradbury, B. T.; Harrison, J. D. L.; Acid, NBL-252, New Brunswick Laboratory, U.S. Atomic
Hedger, H. J.; Mardon, P. C., Carbides in Nuclear Energy, Energy Commission, NJ, USA, 1970, 1–25.
Macmillan & Co. Ltd.: London; Vol. 1 & II,1964. 40. Xavier, M.; Nair, P. R.; Lohithakshan, K. V.; Marathe, S. G.;
12. Potter, P. E. Some phase relationships and equilibria for Jain, H. C. J. Radioanalyt Nucl. Chem. 1991, 148(2),
the uranium-plutonium monocarbides including the 251–256.
effects of oxygen on some equilibria; Plutonium 1975 and 41. Rao, C. L.; Nair, G. M.; Singh, N. P.; Ramaniah, M. V.;
Other Actinides, Blank and Lindemer, Baden Baden, Srinivasan, N. Zeitschrift fur Analytische Chemie 1971,
Sep 10–13, 1975. 254, 126–127.
13. Heiss, A. J. Nucl. Mater. 1975, 55, 207–223. 42. Corpel, J.; Regnaud, F. Anal. Chim. Acta 1962,
14. Potter, P. E. J. Nucl. Mater. 1972, 42, 1–22. 27, 36–39.
15. Agarwal, R.; Venugopal, V.; Sood, D. D. Thermodynamic 43. Crossley, D.; Philips, G. The determination of carbon in
analysis of U–Pu–C–N–O system; BARC/1992/e/022; plutonium metal, alloys, oxides and carbides, UKAEA
Bhabha Atomic Research Centre: Mumbai, India, 1992. Report, AERE-R-3790, 1961.
16. Srivastava, D.; Garg, S. P.; Goswami, G. L. J. Nucl. Mater. 44. Simmons, R. E.; Randolph, M. H. Anal. Chem. Acta 1962,
1989, 161, 44–56. 34, 1119–1122.
17. Saibaba, M.; Vana Varamban, S.; Mathews, C. K. J. Nucl. 45. Rizvi, G. H.; Rastogi, R. K.; Natarajan, P. R. BARC:I-891;
Mater. 1987, 144, 56–64. Bhabha Atomic Research Centre: Trombay, India, 1986.
18. Benedict, U.; Richter, K. J. Nucl. Mater. 1975, 55, 352–354. 46. Sankaran, P. S.; Balachander, N.; Sayi, Y. S.;
19. Benedict, U.; Dufour, C.; Scholten, O. J. Nucl. Mater. Venkataramana, P.; Natarajan, P. R. Zeitschrift fur
1978, 73, 208–212. Analytische Chemie 1983, 315, 496–498.
20. Tetenbaum, M. Advanced LMFBR fuels, Met. A 47. Dobbs, R. A.; Wise, R. H.; Dean, R. B. Anal. Chem. 1967,
7807–720200, 1977, pp 179–188. 39(11), 1255–1258.
21. Rosen, S.; Nevitt, M. V.; Mitchell, A. W. J. Nucl. Mater. 48. Jain, G. C.; Ganguly, C. J. Nucl. Mater. 1993, 207, 169–176.
1963, 9(2), 137–142. 49. Chetty, K. V.; Radhakrishna, J.; Sayi, Y. S.;
22. Frost, B. R. T. J. Nucl. Mater. 1963, 10(4), 265–300. Balachander, N.; Venkataramana, P.; Natarajan, P. R.
23. Pascard, R. Pulvermetallurgie in der Atomkerntechnik Radiochem. Radioanal. Lett. 1983, 58(3), 161–168.
Metallwerk Plansee AG, Reutte/Tirol, 4th Plansee 50. Ganguly, C.; Jain, G. C.; Ghosh, J. K.; Roy, P. R. J. Nucl.
Seminar, 1961, pp 387–419. Mater. 1988, 153, 178–188.
24. Anselin, F. Preparation et Etude des Nitrures et 51. Burkes, D. E.; Fielding, R. S.; Porter, D. L.; Crawford, D. C.;
Carbonitrures D ’uranium et de Plutonium (Preparation and Meyer, M. K. J. Nucl. Mater. 2009, 389(3), 458–469.
Study of the Nitrides and Mixed Carbidenitrides of 52. Bart, G.; Botta, F. B.; Hoth, C. W.; Ledergesber, G.;
Uranium and of Plutonium), Report CEA-R-2988; Mason, R. E.; Straton, R. W. J. Nucl. Mater. 2008, 376,
Commissariat a l’Energie Atomique, Fontenay-Aux-Roses 47–59.
(France), Centre d’Etudes Nucleaires, 1966. 53. Performance of He-bonded LMFBR carbide fuels at the
25. Rough, F. A.; Chubb, W. Progress on the Development beginning of life and consequences for their subsequent
of Uranium Carbide-Type Fuels; Battelle (USA) Report, behavior, Blank, H.; Campana, M.; Coquerelle, M.;
BMI-1488, 1960. Richter, K. Technical Committee Meeting on ‘advanced
26. Frost, B. R. T. J. Nucl. Mater. 1963, 10(4), 265–300. fuel for fast breeder reactors: fabrication and properties
27. Takada, Y.; Imoto, S.; Sano, T. J. Atomic Energy Soc. and their optimization’, Vienna (Austria) 3–5 Nov. 1987,
Japan 1961, 3, 513–518. IAEA 1988, IAEA-TECDOC-466, 71–79.
28. Fernandez, A.; McGinley, J.; Somers, J.; Walter, M. 54. Kummerer, K. R. J. Nucl. Mater. 1984, 124, 147–152.
J. Nucl. Mater. 2009, 392, 133–138. 55. Blank, H. J. Nucl. Mater. 1975, 58(2), 123–139.
29. Matzke, H. J. Science of Advanced LMFBR Fuels; 56. Seitzer, M. S.; Perrin, J. S.; Wilcox, B. A. Reactor Technol.
Elsevier Science Publishers: Amsterdam, 1971, 14(2), 99.
The Netherlands, 1986. 57. Dienst, W. J. Nucl. Mater. 1977, 65, 1–8.
30. Richter, K.; Coquerelle, M.; Gabolde, J.; Werner, P. In 58. Routbort, J. L.; Singh, R. N. J. Nucl. Mater. 1975, 58,
Proceedings of a Symposium on Fuel and Fuel Elements 78–114.
for Fast Reactors, Brussel, 1973; IAEA: Vienna, 1974; 59. Report AERE-R7326, Hall, RA. J. Am. Ceram. Soc. 1973,
Vol. 1, pp 71–84. 56, 173–177.
31. Handschuh, A.; Dubois, S. A.; Vaudez, S. A.; 60. Blank, H. Material Science and Technology, Nuclear
Grandjean, S. B.; Leturcq, G. B.; Abraham, F. J. Nucl. Materials, Pt I; Cahn, R. W., Haasen, P., Kramaer, E. J.,
Mater. 2009, 385, 186–188. Eds.; VCH Publisher: New York, 1994; Vol. 10A.
32. Pope, J. M. Trans. Am. Nucl. Soc. 1981, 38, 204–205. 61. Tokar, M.; Nutt, A. W.; Leary, J. A. Mechanical properties
33. Stratton, R. W. Trans. Am. Nucl. Soc. 1981, 39, 421–423. of carbide and nitride reactor fuel; LA-4452; LASL: Los
34. Ledergerber, G.; Herbst, R.; Zwicky, H. U.; Kutter, H.; Alamos, 1970.
Fischer, P. J. Nucl. Mater. 1988, 53, 189. 62. Tokar, M. High temperature compressive creep and
35. Chandrasekharan, K. N.; Pandey, V. D.; Khan, K. B.; hot hardness of uranium plutonium carbide, LA-4704,
Kumar, A. J. Nondestructive Test. Eval. 2008, 6(4), 12–16. 1971.
36. Bartscher, W. Zeitschrift fur Analytische Chemie 1972, 63. Tokar, M. Hot hardness of hyperstoichiometric mixed
258, 346–348. uranium plutonium carbide fuel, LA-DC-72–187, 1970.
37. Chander, K.; Patil, B. N.; Kamat, J. V.; Khedekar, N. B.; 64. Tokar, M. J. Am. Ceramic Soc. 1973, 56, 173.
Manolkar, R. B.; Marathe, S. G. Nucl. Technol. 1987, 65. Sengupta, A. K.; Basak, U.; Ganguly, C. J. Mater. Sci. Lett.
78(1), 69–74. 1987, 6, 20.
38. Davies, W.; Gray, W. A rapid and specific titrimetric 66. Crawford, D. C.; Porter, D. L.; Hayes, S. L. J. Nucl. Mater.
method for the precise determination of uranium using iron 2007, 371, 202–231.
(II) sulphate as reductant, Talanta 1964, 11(8),1203–1211. 67. Richter, K.; Gueugnon, J.; Kramer, G.; Sari, C.; Werner, P.
39. Eberle, A. R.; Lerner, M. W.; Goldbeck, C. G.; Rodden, C. J. Nucl. Technol. 1985, 70, 401.
Titrimetric Determination of Uranium in Product Fuel and 68. Kasiviswanathan, K. V.; Venugopal, V.; Jojo, J.;
Scrap Materials after Ferrous Ion Reduction in Phosphoric Muralidharan, N. G.; Johny, T.; Venkiteswaran, C. N.
86 Carbide Fuel
In Proceeding of a Symposium Research and Fuels, IAEAPL- 463/15; IAEA: Vienna, Aug 7–11, 1972;
Development of Uranium-Plutonium; IGCAR: Kalpakkam, pp 255–268.
India, 2008. 78. Cubicciotti, D.; Davies, J. H. Nucl. Sci. Eng. 1974, 60, 314.
69. Suzuki, Y.; Arm, Y.; Handa, M.; Shiba, K. Research and 79. Konashi, K.; Yato, T.; Kaneko, H. J. Nucl. Mater. 1983,
development of uranium plutonium mixed carbide and 114, 86.
nitride fuels at JAERI; IAEA TECDOC-466, Vienna, 1988; 80. Holley, C. E.; Rand, M. H.; Storms, E. K. The
pp 73–82. chemical thermodynamics of actinide elements and
70. Rouault, J. Conventional and innovative fuels for fast compounds, Part 6, The actinide carbides; IAEA: Vienna,
reactors: A French review, Consultancy Meeting on 1984.
Conventional and Innovative Fuels for Fast Reactor; IAEA: 81. Kim, T. K.; Grandy, C.; Hill, R. N. Carbide and nitride fuels
Bienna, Dec 18, 2007. for advanced burner reactor IAEA-CN-176/07–10;
71. Ganguly, C.; Sengupta, A. K. J. Nucl. Mater. 1988, 158, Argonne National Laboratory, 9700 S. Cass Av. Argonne,
159–165. IL 60439, FR09, Kyoto, Dec 7–11, 2009; p 274.
72. Rajendra Pillai, S.; Mathews, C. K. J. Nucl. Mater. 1987, 82. Bokelund, H.; Cadeci, M.; Ougier, M. Radiochim. Acta
150, 31–41. 1982, 30, 49–55.
73. Agarwal, R.; Venugopal, V. J. Nucl. Mater. 2006, 359, 83. Bokelund, H.; Apostolidis, C.; Glatz, J. P. J. Nucl. Mater.
122–131. 1989, 166, 181–188.
74. Holleck, H. In Thermodynamics of Nuclear Materials 1974; 84. Natarajan, R.; Raj, B. J. Nucl. Sci. Technol. 2007, 44(3),
IAEA: Vienna, 1975; Vol. I, p 213. 393–397.
75. Haines, H. R.; Potter, P. E. In Thermodynamics of Nuclear 85. Schapira, J. P. Nucl. Instrum. Methods Phys. Res. 1989,
Materials 1974; IAEA: Vienna, 1975; Vol. I, p 45. A280, 568–582.
76. Kleykamp, H. In Proceedings of a Topical Meeting on 86. Brunel, L.; Chauvin, N.; Mizuno, T.; Pouchon, M. A.;
Advance LMFBR Fuels; Leary, J., Kittle, H., Eds.; Somers, J. The Generation IV Project GFR fuel and other
American Nuclear Society: Arizona, Oct 10–13, 1977; core materials, GIF Symposium, Paris (France), Sept 9–10,
pp 166–167. 2009.
77. Gotzmann, O.; Ohse, R. W. In Proceedings of a Panel on
the Behavior and Chemical State of Irradiated Ceramic
3.04 Thorium Oxide Fuel
P. R. Hania and F. C. Klaassen
Nuclear Research and Consultancy Group, Petten, The Netherlands
3.04.1 Introduction 88
3.04.2 Incentives for Using Thorium 89
3.04.2.1 Thorium as an Abundantly Available Resource for Nuclear Fuel 89
3.04.2.2 Radiotoxicity Reduction with Thorium 90
3.04.2.3 Reduction of Excess Military Plutonium 90
3.04.3 Physical Properties of Thorium Oxide Fuel 90
3.04.3.1 Crystal Structure 91
3.04.3.2 Thermal Expansion 91
3.04.3.3 Thermal Conductivity 92
3.04.3.4 Thermophysical Properties 93
3.04.3.5 Oxygen Potential 93
3.04.4 Thorium Oxide Fuel Fabrication 94
3.04.4.1 Powder Compaction 94
3.04.4.1.1 ThO2 94
3.04.4.1.2 Mixed oxides 95
3.04.4.2 Powder–Plasticizer Methods 96
3.04.4.3 Sol–Gel Methods 96
3.04.5 Behavior of Thorium Oxide Fuel Under Irradiation 97
3.04.5.1 Neutronic Properties of Thorium-Based Fuel 97
3.04.5.2 In-Core Behavior of Thorium Oxide Fuel 99
3.04.5.2.1 Restructuring 100
3.04.5.2.2 Thermal conductivity 100
3.04.5.2.3 CANDU fuel 101
3.04.5.2.4 United States 101
3.04.5.2.5 India 102
3.04.5.2.6 Fission product behavior 102
3.04.5.2.7 (Th,Pu)O2 103
3.04.6 Reprocessing and Refabrication 104
3.04.6.1 The THOREX Process 104
3.04.6.2 Beyond Thorex 105
3.04.6.3 Radiation Issues in Reprocessing and Refabrication 106
3.04.7 Conclusions 106
References 106
87
88 Thorium Oxide Fuel
article by Rodriguez and Sundaram from 1981 con- Table 1 Total identified world thorium reserves
taining many references to the older literature.12 (in 106 kg Th), sorted by amounts per country
Here we give an introduction with more emphasis Country Identified Th resources
on the materials perspective, including the more (<80 USD per kg Th)
recent trend of replacing the HEU with LWR- or
Australia 452
WG-plutonium (WG-Pu), stemming from a virtual
United States 400
ban on HEU. Specifically, we treat the basic proper- Turkey NA
ties, fabrication and reprocessing aspects, and the India 319
irradiation characteristics of thorium oxide fuels. Brazil 302
Three recent complementary articles from the Venezuela 300
Norway 132
Encyclopedia of Materials should be mentioned, which Egypt 100
give up-to-date overviews of general fuel fabrication Russian Federation 75
issues, thorium fuel cycles, and reprocessing issues, Greenland 54
respectively.13–15 Canada 44
The chapter is organized as follows: in Section South Africa 18
Others 33
3.04.2, the rationale behind the investigation of tho-
Total 2229
rium as a fissile source material is highlighted. In
Section 3.04.3, the basic physical properties of tho- Data taken from Uranium Red Book; OECD 2009.
rium oxide fuels are discussed in comparison with the NA ¼ not available.
oxides of uranium and plutonium. The fabrication
aspects are discussed in Section 3.04.4. In Section
3.04.5, the behavior of thorium oxide fuel under amount of thorium resources is not very well known.
irradiation is treated. Reprocessing issues are dis- Estimates of thorium resources (identified resources,
cussed in Section 3.04.6. retrievable at a cost less than USD 80 per kg Th) are
The authors do not claim to have presented a given in Table 1, taken from the Uranium Red Book
complete picture of thorium in every detail. How- 2009.16 This overview estimates a total of known
ever, they do hope to have presented a thorough resources of 2.5 million tons Th (identified resources)
overview as well as numerous references that may and an additional prognosticated amount of 1.9
help the reader further. million tons.
Currently, there is little demand (and industrial
need) for thorium; it is obtained mostly as a by-
3.04.2 Incentives for Using Thorium product from the production of rare earth metals
or from the production of titanium-, zirconium-, or
As indicated in the introduction, there are three main tin-bearing minerals from monazite deposits.16 Little
incentives for the use of thorium. These are the use of systematic exploration has therefore been conducted
thorium as a resource alternative for uranium, the specifically for thorium, which explains the relative
potential of the thorium fuel cycle to reduce the uncertainty about the total world reserves.
radiotoxicity of spent fuel, and a third, more specific The large availability and relatively easy retrieva-
application of thorium as a matrix for the incinera- bility have been a large incentive to look at thorium
tion of excess WG-Pu. These three applications are as nuclear fuel.17 Specifically, this has been the case in
discussed in the following sections. countries with large thorium reserves, such as India,
Brazil, and, recently, Norway.
The largest advantages of 232Th over the com-
3.04.2.1 Thorium as an Abundantly
parable fertile isotope 238U are the amount of neu-
Available Resource for Nuclear Fuel
trons produced per fission and the fission-to-capture
Thorium is more abundant than uranium: The con- ratio, which are both higher than in 238U (Section
tent of thorium in the earth’s crust is estimated to be 3.04.5.1). This allows breeding with thorium in ther-
about three to four times larger than that of uranium. mal reactors as well, whereas for 238U, a fast spectrum
Its most common source is the rare earth mineral is mandatory for breeding. Thorium can thus be used
monazite. Monazite sand is found in large amounts directly in light or heavy water reactors or in HTR. It
in India and Brazil, and Australia has large deposits should be noted, though, that achieving breeding
as well. It should be noted, however, that the exact with thorium is not straightforward. The exact gain
90 Thorium Oxide Fuel
in resource efficiency, which can be achieved in a generally, the choice of three toppings, either HEU,
thorium cycle, as compared with the ‘standard’ ura- low enriched uranium (LEU), or plutonium (Pu). It is
nium cycle, depends highly on the chosen fuel cycle clear that the use of HEU offers the most benefits in
scenario, that is, the combination of chosen reactor terms of radiotoxicity reduction. But the use of HEU
type, fuel type, and fuel management scenario. The in a civil nuclear fuel cycle is not preferred due to
efficiency of breeding 233U together with the burnup reasons of proliferation risks. Furthermore, the effect
rate of the fissile topping, and the choice of whether of resource efficiency through the use of abundant
or not to reprocess the spent fuel, ultimately deter- thorium is counterbalanced by the amount of natural
mines the efficiency of a chosen thorium fuel cycle. uranium material needed to enrich uranium to
The Shippingport reactor constitutes a realization enrichments of 90%. The combination of thorium
of breeding in a light water-moderated reactor, but with HEU, although preferred from a radiotoxicity
this was a small (60 MWe) reactor with an optimized point of view, is therefore not a viable option for a
breeding fuel cycle. In heavy water reactors, neutron sustainable fuel cycle. The more the amount of ura-
absorption is smaller, which translates to a higher nium (238) or plutonium present in the initial fuel,
breeding potential. The graphite-moderated HTR the more the amount of transuranics produced, which
offers a similar but smaller advantage over light water. determines the lifetime of the spent fuel. Neverthe-
In the longer term, the use of thorium is foreseen in less, a significant reduction of the radiotoxicity can
molten salt reactors (MSR) as ThF4 (see Chapter 3.13, be achieved. Once an equilibrium thorium cycle is
Molten Salt Reactor Fuel and Coolant). achieved, the benefits increase, as then the fuel
is based on the combination of 232Th/233U only.
3.04.2.2 Radiotoxicity Reduction
with Thorium 3.04.2.3 Reduction of Excess Military
Plutonium
A second incentive to look at thorium is connected
to the potential of waste production. Thorium fuels A third, more specific potential application of tho-
produce far less transuranic elements (i.e., elements rium is connected to the incineration of excess
with atom number >92). As these transuranic ele- military plutonium. More than 250 tons of WG-Pu,
ments determine the long-term radiotoxicity of the containing around 93% 239Pu, has been produced in
spent fuel, the use of thorium greatly reduces this the world for military purposes, mostly by the United
radiotoxicity. A common definition of the lifetime of States and the Russian Federation. Part of these stock-
spent nuclear fuel is connected to the decay time, in piles has been declared as excess plutonium, and both
which the spent fuel reaches the radiotoxicity level of the United States and Russia have agreed to dispose of
the original uranium ore from which it was produced. 34 ton WG-Pu. In the disposition program, it has been
According to this definition, actinides (transuranics) well defined how the excess plutonium will be incin-
in the spent UO2 fuel have a lifetime of 130 000 years, erated, that is, through its use as (U,Pu)O2 fuel in
whereas fission products have a lifetime of only nuclear power plants in the United States and as fuel
270 years.18 The lifetime is thus dominated by the for fast reactors in Russia. Nevertheless, the mixture
transuranic elements plutonium, americium, and of thorium with WG-Pu into (Th,Pu)O2 fuel may
curium, with atom mass numbers of 238 and higher. provide a technical option to reduce the threat of
These are produced by subsequent neutron captures military plutonium. The application of (Th,Pu)O2
in 238U. The lower mass number of 232Th ensures fuel may similarly be used to more efficiently reduce
that these higher isotopes are produced in much stockpiles of separated civil plutonium.
less amounts.
The potential to reduce radiotoxicity of spent
nuclear fuel with thorium was investigated exten- 3.04.3 Physical Properties of
sively in Europe under the 4th Framework program Thorium Oxide Fuel
‘Thorium as a waste management option’4 and later
in the 5th Framework program ‘Thorium cycle.’108 The following section describes the important
The effect of reduction in radiotoxicity depends thermophysical properties of ThO2 in comparison
greatly on the choice of ‘topping,’ which is the fissile with UO2 and PuO2. The comparison with the more
start-up material. At the start-up of a thorium cycle, common fuel oxides UO2 and PuO2 (see Chapter 2.02,
that is, when no fissile 233U is available, one has, Thermodynamic and Thermophysical Properties
Thorium Oxide Fuel 91
of the Actinide Oxides) is made throughout to show changes linearly with the additive fraction (Vegard’s
that the three compounds are very similar and to simul- Law behavior). However, locally the individual heavy
taneously highlight the differences, which turn out metal–oxygen bond lengths in the mixed oxide
mostly favorably for ThO2. What sets ThO2 fuel apart stay somewhat closer to the values of the pure
is that the thorium ion (unlike uranium and plutonium) compounds.30,31
adopts in compounds a single oxidation state (4þ). This It follows from the Vegard’s Law behavior of
single valence implies that the oxide has very little the stoichiometric mixed oxide that their room-
nonstoichiometry, that is, a low number of oxygen temperature densities may simply be obtained by
vacancies and interstitials. The absence of nonstoichio- linear interpolation of the weights and cell volumes
metry is reflected in a relatively high thermal conduc- given in Table 2 and additionally that the solid solu-
tivity (Section 3.04.3.3), a well-defined oxygen tions thus formed show ideal behavior.20 Vapor pres-
potential (Section 3.04.3.5), high thermal stability, sure measurements on ThO2–UO2 solid solutions
low chemical reactivity, low matrix diffusivities, and have indicated that the same ideality also holds at
some difficulty in the sintering of pellets (Section higher temperatures.32,33 This allows one to reverse
3.04.4). For more thorough discussions of the existing the argument and claim that the lattice parameter
literature on thermophysical properties, we refer the also changes linearly with composition in the high-
reader to two exellent reviews20,21 and a paper by temperature region. Therefore, the thermal expan-
Sobolev et al. discussing some relationships between sion of the mixed actinide oxides can be obtained by
the different properties.123 linearly interpolating the thermal expansions of the
pure compounds.20
3.04.3.1 Crystal Structure
Close similarities exist between the three actinide 3.04.3.2 Thermal Expansion
oxides: In common with UO2 and PuO2, ThO2 pos-
Touloukian et al. recommend the following relation
sesses an fcc lattice, and all three materials can be
for the thermal expansion of pure ThO234 based on
heated to melt without undergoing phase transi-
a large set of measurements in excellent agreement:
tions.20 In addition, thermal conductivities and ther-
mal expansions are similar. The three oxides may DL=L0 ð293 KÞ ¼ 0:179 þ 5:097 104 T
furthermore be mixed in all proportions, forming a
single-phase material. þ 3:732 107 T 2
The associated lattice constant for this series of 7:594 1011 T 3 ð150 2000 KÞ
actinides decreases as Th < U < Pu, as shown in
Table 2, which compares some physical parameters The thermal expansion coefficient aL is obtained by
for the stoichiometric compounds. differentiation. Figure 1 compares this expansion
XRD measurements indicate that when changing for ThO2 with that observed for UO2, and PuO2,
the composition from pure ThO2 to pure PuO227 or which have all been described using a single set of
UO2,28,29 the lattice parameter of the cubic lattices relations25:
a
Value taken from a recent laser flash measurement,124 but note that PuO2 melts about 300 K below this point in older studies.25
The molecular weight of Pu, which has no stable isotopes, is conventionally fixed at 244. Lattice constants, cell volumes, and melting
temperatures are taken from IAEA Technical Document20 and thermophysical data from Bakker et al.19 and Cordfunke and Konings.23
92 Thorium Oxide Fuel
14 Bakker, 95% TD
ThO2
1.0
6
4
0.5
2
0.0 0
400 600 800 1000 1200 1400 1600 1800 2000 400 600 800 1000 1200 1400 1600
Temperature (K) Temperature (K)
Figure 1 Thermal expansion of ThO2 and UO2/PuO2. In Figure 2 Thermal conductivity of stoichiometric ThO2,
the range shown, both relations are valid. The expansion UO2, and PuO2.
curve for ThO2 has been adjusted here to a reference
temperature of 273 K.
In Figure 2, a comparison is also made with the
correlations given for UO225,38,39 and PuO2.37 The
DL=L0 ð273 KÞ ¼ 0:266 þ 9:802 104 T thermal conductivity of UO2 is obviously below that
2:705 108 T 2 of ThO2 in the temperature region shown in this
þ 4:391 1011 T 3 ð273 923 KÞ figure (up to 1600 K); however, an electronic contri-
bution to the thermal conductivity of UO2 kicks in at
DL=L0 ð273 KÞ ¼ 0:328 þ 1:179 103 T temperatures above 2000 K, whereas this contribu-
2:429 107 T 2 tion is absent for ThO2.40
For PuO2, Cozzo et al. argue that previous stud-
þ 1:219 1010 T 3 ð923 K Tm Þ
ies41,42 yielding values close to those for UO2 had
where Tm is the melting temperature. The compari- most probably been performed on samples of ill-
son shows that thermal expansion is slightly lower for defined stoichiometry and that stoichiometric PuO2
ThO2. For a detailed discussion of temperature- in fact has the largest thermal conductivity of the
dependent theoretical density, linear expansion, and three oxides.37 This argument indicates that due to
melting points of ThO2 and ThO2–UO2 mixtures, we the multivalence of the metal atoms, control over stoi-
refer to a recent assessment in the IAEA Technical chiometry is not trivial. Even when the stoichiometry is
Document.21 carefully controlled during fabrication, PuO2 and UO2
will become nonstoichiometric upon heating or when
in oxidative or reductive environments. This results,
3.04.3.3 Thermal Conductivity
for many practical conditions, in some loss of thermal
At moderate temperatures, where the electronic conductivity, which is difficult to control.
contribution can be neglected, empirical values for Mixing of the actinide oxides generally depresses
the temperature-dependent thermal conductivity of the thermal conductivity, mostly because the additive
ionic solids may be fitted with the general function heavy metal ions act as phonon-scattering centers.
l = 1/(A þ BT). Here, the constant A describes The scattering term A is therefore affected more
the effect of material defects that are present strongly than the phonon interaction term B. Gibby
independent of temperature and act as phonon scat- has observed this trend for the mixing of uranium and
tering centers, while the term BT represents the plutonium dioxide,41 and on ThO2 the depression is
temperature-dependent effect of phonon–phonon quite pronounced.43
interactions.35 Bakker et al.20 have analyzed a large Using a small dataset from Murabayashi44 and
set of experimental data and used a selection to obtain Springer and Lagedrost45 selected from the available
values of the parameters A (4.20104m K W1) and literature, Bakker et al.20 derive values of the parameters
B (2.25104m W1) for 95% dense ThO2. More A and B for 95% dense Th1yUyO2 and y up to 0.1
recent experimental data are available from Pillai36 (T ¼ 300–1800 K): A ¼ 4.195 104 þ 1.112y 4.499y2
and Cozzo.37 The three results are in reasonable (m K W1), B ¼ 2.248 104 9.170 104y þ
agreement (Figure 2). 4.164 10 y (m W1). These authors reject on
3 2
Thorium Oxide Fuel 93
theoretical grounds an earlier assessment by Berman The standard entropy, based on measurements
et al.46 from a partially overlapping dataset; however, of the low-temperature heat capacity, is
Berman gives correlations for higher uranium con- S 0(298.15) ¼ 65.23 J K1 mol1.50 Bakker et al.20 have
centrations. According to Bakker’s recommendation, fitted earlier measurements of H(T)H0(298 K)
the addition of 10% UO2 to ThO2 results in a reduc- under the constraint Cp0 (298 K) ¼ 61.76 J K1mol
tion of the conductivity by 40–50%. More recently, from the low-T heat capacity50 to obtain a function
Pillai et al. have found that the addition of as little as for the integrated high-temperature heat capacity:
2% UO2 decreases the thermal conductivity of thoria H 0 ðT Þ H 0 ð298Þ ¼ 55:962T þ 25:62895 103 T 2
by 10–30%.36 In the recent IAEA review,21 new data
are considered along with selected literature to 12:2674 106 T 3
obtain A and B values for uranium contents of 0, 4, 6, þ 2:30613 109 T 4
10, and 20%. Compared with Bakker’s analysis, a þ 5:740310 105 T 1
stronger temperature dependence is obtained in this
20581:7
work (the conductivities being equal at about 1400 K).
In addition, the effect of uranium concentration is The given temperature dependence shows an excess
smaller, that is, for a given uranium concentration, enthalpy at temperatures above 2500 K, similar to that
the conductivity depression is smaller. of other actinide oxides, which has been related to a
For the addition of PuO2, few sources are premelting l-transition at 3090 K.38 Similar disconti-
available, but Cozzo et al.37 report the correlation nuities in the slopes of the enthalpy–temperature
A ¼ 6.071 103 þ 0.572y 0.5937y2 (m K W1), curves at about 0.8 of the melting temperature were
4 1
B ¼ 2.4 10 (m W ). This correlation has a mini- found for a series of thoria–urania solid solutions.51
mum at around 45% Pu, at which point the thermal Recently, Dash et al. have performed extensive
conductivity has been reduced by more than a factor measurements on Th1yUyO2 for y 0.2 (127–
2 at 500 K. An equation obtained from a different 1698 K), making a comparison to older data.52 They
dataset is given in21: A ¼ 0.08388 þ 1.7378104y arrive at the following expression for the heat capac-
(m K W1), B ¼ 2.62524 þ 1.7405104y (m W1). ity as a function of T (298 T 2000 K) and
For a plutonium concentration of 5%, this result is y (0.019 y 0.9):
about equal to that of Cozzo, but the effect of Pu Cp ¼ ð66:26 þ 10:91y Þ þ ð0:00923 0:00065y ÞT
addition is found to be stronger. Basak reports similar
findings for the addition of 4% PuO2.47 7:70 105 þ 6:7 105 y T 2
The above comparison of the available correlations The correlation of Dash et al. for pure ThO2 corre-
for (Th,U)O2 and (Th,Pu)O2 fuels reveals significantly sponds reasonably well to that obtained by Bakker et al.
different results, which seems to indicate measurement With regard to the mixed oxides, it can be said that
uncertainties related to sample microstructure and both (Th,Pu)O2 and (Th,U)O2 experience a slight
stoichiometry. When using the Berman40 and Bakker20 melting point depression. For (Th,U)O2, the melting
correlations for (Th,U)O2 and the Cozzo37 correlation points correspond to that of ideal solid solutions;
for (Th,Pu)O2, the fissile concentrations at which for (Th,Pu)O2, not enough data exist to confirm this
the thermal conductivity of thoria-based fuel becomes ideality.20 The Cp(T) data for the (Th,U)O2 are quite
equal to that of UO2 is about 10%. However, the well reproduced by a weighted average of the Cp(T)
correlations given in a recent assessment in an IAEA values of ThO2 and UO2, with little deviation.20,52
Technical Document21 suggest that the thermal con- The IAEATechnical Document21 uses Bakker’s corre-
ductivity of UO2 is approached with the addition of lation for ThO220 and Fink’s correlation for UO239 to
about 20% uranium or only 6% Pu. construct Neumann-Kopp heat capacities.
–50
DG ðO2 Þ ¼ 0:029592T lnðx Þ þ 0:003436T
–100
y = 0.05
y = 0.1 lnð0:5619 0:1161x Þ 0:033028T
lnð0:5619y 1:1161x Þ 348
DG(O2) (kJ mol–1)
–150
y = 0.2 þ 0:1767T kJ mol1
–200
The correlations derived by Schram and Dash et al.
–250
Schram are shown in Figure 3; the data are in reasonable
–300
Dash agreement, considering that they are based on oxygen
potential measurements55–57 and heat capacity mea-
–350
0.000 0.005 0.010 0.015 0.020 surements,52 respectively.
x Finally, we note that phase separation occurs for
Figure 3 Oxygen potentials of Th1yUyO2+x at 1473 K off-stoichiometric conditions at high U and Pu con-
and different values for x and y, as determined by Schram53 centrations (Section 3.04.4.1). For instance, for high
based on a Lindemer–Bessman model for UO2+x (black), U concentrations phase separation occurs, and the
and as determined by Dash et al.52 from direct mixed fcc oxide exists in equilibrium with a separate
measurements on the mixed oxides (red).
U3O8 phase.54 Such a phase equilibrium would intro-
duce an oxygen potential plateau.
PuO2 and UO2 under realistic conditions. We can
combine this information with the ideality of the solid
solution and claim that thorium in mixed oxides can be 3.04.4 Thorium Oxide Fuel
regarded as an inert solvent (for PuO2 or UO2) that Fabrication
does not take part in the chemical equilibria describing
the oxygen potential.53 This reduces the oxygen chem- Fabrication of thorium-based oxide fuel is well devel-
istry of mixed thorium oxides to the chemistry of the oped. Three routes have been applied successfully to
fissile additive (U or Pu). create thorium-based oxide fuel: conventional bin-
On a microscopic level, hyperstoichiometry in derless powder pressing, spheroidization of powder–
ThO2–UO2 solid solutions manifests as interstitial plasticizer mixtures, and the sol–gel process. The
oxygen. Cohen shows that at 1200 C, the maximum latter two cases yield microspheres with diameters
amount of interstitial oxygen increases from 0 for in the range 50–1000 mm, which can be pressed into
pure ThO2 to 0.25 for Th0.1U0.9O2.25.54 This increase pellets, used directly in Sphere-Pac/Vipac arrange-
with U concentration follows from the fact that a ments (see Chapter 3.11, Sphere-Pac and VIPAC
valency change from 4þ to 5þ in two U atoms or Fuel), or coated with carbon and silicon carbide
(less likely) from 4þ to 6þ in one U atom is needed layers to create TRISO fuel for HTR (see Chapter
to compensate for the presence of the extra oxygen. 3.06, TRISO Fuel Production). Much of
The incorporation of PuO2 in the ThO2 matrix simi- the available information is from the Indian experi-
larly allows for the creation of oxygen vacancies at ence in powder pressing58 and from the German–
elevated temperatures under inert or reducing atmo- American developments in sol–gel methods.2,4,6–10
spheres. It may be clear that, because the additive
(UO2 or PuO2) will be fissioned away during irradia-
3.04.4.1 Powder Compaction
tion and the thorium ions cannot undergo a valency
change, the oxygen potential in thorium mixed 3.04.4.1.1 ThO2
oxides fuel should in fact rise faster with burnup The procedures for fabricating ThO2 pellets by pow-
compared to conventional (U,Pu)O2. der compaction are derived from the procedures
Schram53 has used the above reasoning to describe developed for UO2 and (U,Pu)O2 fuel.2,10 However,
(Th,U)O2+x with a Lindemer–Bessman type model as thorium is found only as a 4-valent cation, it is not
for the uranium, which describes the collected (lim- as important to control the oxygen potential during
ited) amount of oxygen potential data reasonably sintering as in the case of uranium or plutonium, and
well. On the other hand, Dash et al.52 have used sintering of thorium oxide may be performed in both
their heat capacity measurements on Th1yUyO2 to oxidizing and reducing conditions (air, argon, vac-
calculate the oxygen potential in a more direct way52 uum, or Ar/H2). On the other hand, the thermal and
(x 0.024, y 0.2): chemical stability of ThO2 discussed in the previous
Thorium Oxide Fuel 95
section somewhat decreases its sinterability, and high tend to be better sintering aids under a reducing
densities are more difficult to produce. atmosphere,58 while the niobate functions best in an
The thorium dioxide powder is usually obtained oxidizing atmosphere. We refer the reader to Kutty
by calcination of the oxalate, Th(C2O4)2, which pre- et al.58 for an analysis of the effects of dopants.
cipitates from a nitrate feed solution (nitric acid with a Finally, it is well known that water easily adsorbs
pH 0.8) upon dropwise addition of oxalic acid: to the ThO2 surface, chemically by forming a high
concentration of hydroxyl groups and physically via
ThðNO3 Þ4 ðaqÞ þ 2C2 O4 H2 ðaqÞ hydrogen bonds.61–63 Care should therefore be taken
! ThðC2 O4 Þ2 ðsÞ þ 4HNO3 ðaqÞ to store the resulting pellets in dry conditions.
phase was found to be present besides the ThO2– route, 1 wt% of Ca(NO3)2 may be added to the
PuO2 fcc structure.58 Similarly, XRD data for both heavy metal solution as a sintering aid, while 30 g
ThO2–30%UO2 and ThO3–50%UO2 sintered in air carbon black per mol heavy metals may be added
revealed the presence of small amounts of U3O8 to produce spherical pores during the calcination
(nearly hexagonal).10,64 step.6–8
Several sol–gel routes have been applied in
3.04.4.2 Powder–Plasticizer Methods the past, of which two have been most successful.
The KEMA internal gelation process developed
The pellet fabrication route as described above is
in the Netherlands for uranium and plutonium has
usually performed without the addition of binder mate-
been adapted for thorium in India and Germany,70,71
rials. India is developing a method to produce micro-
while the external gelation of thorium (EGT) or KFA
spheres for HTR TRISO particles (see Chapter 3.06,
process was developed in Germany (Jülich)6 and the
TRISO Fuel Production) or sphere-pac (see Chapter
United States (Oak Ridge).
3.11, Sphere-Pac and VIPAC Fuel) fuel from ThO2
In the internal gelation process, the nitrate feed
powder in which a binder material is added. In this so-
solution is mixed with a solution of hexamethylene
called CAP (coated agglomerate pelletization) process,
tetramine (HMTA, (CH2)6 N4) and urea (CO(NH2)2)
ThO2 powder is mixed with a plasticizer (e.g., a paraf-
of similar concentration at a temperature of around
finum/petrolatum mixture) at slightly elevated tem-
0 C 71–73; upon mixing at this low temperature, the
peratures, after which the plastic mixture is simply
heavy metal ions form complexes with urea. The
molded in an extruder and subsequently in a spheroi-
resulting mixture is dispersed as fine droplets by a
dizer to form small spheres.67–69 In a second step, the
hollow vibrating needle (frequency in the order
ThO2 spheres may be coated with a layer of the fissile
102–103 Hz). The dispersed droplets fall into a hot
material. This method should minimize dust formation
(50–90 C) bath of silicone oil, and the droplet temper-
as well as the number of steps to be performed under
ature rises quickly. This causes decomposition of the
shielded conditions (Section 3.04.6.3).
heavy metal–urea complexes as well as hydrolytic
decomposition of HMTA. The latter produces ammo-
3.04.4.3 Sol–Gel Methods
nium hydroxide. After hydroxylation of the heavy
The sol–gel process offers an alternative to the con- metal ions by NH4OH, the resulting heavy metal
ventional powder mixing technology, which may be hydroxides form agglomerates of microcrystallites.
automated and is dust-free, thereby offering a strong This induces the sol to gel transition.
advantage in radiation safety, which is especially Details of the reaction for thorium (and also for
important when handling irradiated thorium (Section (Th,U)O2 with up to 10% U) have been described by
3.04.6.3). The main disadvantage is that detailed con- Kumar71 and Pai et al.74 In brief, the concentrated Th
trol of the process is rather complex, which may give (NO3)4 solution is partially preneutralized by adding
problems when scaling up the process. formaldehyde or ammonia. The optimum Th4+ con-
As is the case with the powder compaction pro- centration after mixing is 1–1.4 M, and the ratio of both
cess, this technique starts from nitrate feed solutions HMTA and urea to Th4+ ions is 1.4:1 (Figure 4). The
of heavy metals (Th(NO3)4, Pu(NO3)4, and/or formation of opaque hard gel spheres upon gelation at
UO2(NO3)2), although the used concentrations are a temperature of 60–70 C is taken as an indication that
somewhat higher (2–3 M). But instead of adding crack-free spheres will be formed after drying and
oxalic acid to induce precipitation, droplets of the calcination.71–73 Following the gelation step, the gels
chosen heavy metal solution are exposed to ammo- are prewashed using CCl4 to remove the oil.
nia, which induces the formation of microcrystallites In the external gelation method, no ammonia-
and thereby gelation of the sol. The resulting gel is producing additive (HMTA) is used to hydrolyze or
washed and dried, producing microspheres. After polymerize the heavy metal sol.70 Instead, after being
calcination and sintering, the sol–gel microspheres released from the vibrating needle, the sol drops pass
(with diameters in the range 50–1200 mm) may be through gaseous ammonia, which quickly gels the sur-
crushed and pressed into pellets or alternatively face of the droplet. The partially gelled drops fall into a
used as is in a sphere-pac column. In the case of solution with a pH of 8 containing 1% ammonia and
HTR fuel, coatings are applied to the microspheres 4 M NH4NO3, which completes gelation of the inside
to produce the well-known TRISO particles, as of the droplets. As for the internal gelation process, the
is done for UO2 fuel. As in the powder–pellet best results were obtained when the heavy metal
Thorium Oxide Fuel 97
232
Th, the resulting fissile isotope is 233U; in 238U, it
1.5 239
is Pu. It is therefore useful to compare the neu-
Region A:
opaque soft gel
tronic properties of the two sets of nuclides and
1.4 (60–75 ⬚C) discuss their differences. The top and bottom panels
in Figure 5 compare the neutron capture cross-sections
of fertile nuclides and the fission cross-sections of
1.3
fissile nuclides, respectively. In Table 3, some neu-
tronic properties of the relevant nuclides are listed;
1.2 the table is an abstract from Kaye and Laby.75
Cracked during processing
Regarding the fertile materials, thermal capture is
Yielded defect-free microspheres
almost 3 times higher for 232Th than for 238U, but
1.1 Too soft for processing
resonance capture is more than 3 times higher for
238
U. In the fast region of the spectrum, the cross-
1.0 sections are similar. More important for thermal
0.9 1.0 1.1 1.2 1.3 1.4 1.5 breeders are the characteristics of the fissile nuclides.
Thorium (M) It can been seen in Figure 5 that the fission cross-
Figure 4 Gelation field diagram for ThO2 as published section of 233U is least dependent on neutron energy,
by Kumar.68 being relatively small in the thermal region and
relatively large in the epithermal and fast regions.
solution (90 C) was ‘preneutralized’ up to a pH,123 of Table 3 shows that the resonance integral for
3.25–3.5 and a viscosity of 0.03 Pa s before gelation.70,122 233
U fission is more than 2 times larger than that for
Treatment of the gel spheres is very similar in the 239
Pu. However, the most significant advantage of
233
two routes described. The spheres are ‘aged’ and U compared with 235U and 239Pu is the very high
washed in a 1% NH3 solution, to improve the internal fission-to-capture ratio (sf/sc), which in a thermal
structure and to remove organic material, respectively. spectrum is about 10 for 233U but only about 2.5 for
The particles are then dried at 100–400 C in humid 239
Pu.2 This produces a high neutron yield per
air to avoid crack formation, calcined at around 700 C absorption ¼ nsf/(sf þ sc) in a thermal spectrum
in air, and finally sintered at 1000–1200 C. The sol–gel and up to energies of about 100 keV, that is, it yields a
particles sinter very well due to the small crystallite relatively good neutron economy in a wide range of
size of the oxide formed from the agglomerate of spectra and especially in thermal and epithermal
hydroxide microcrystallites. The particles thus reach reactors. The value of also decreases less with
a density close to the theoretical density. temperature compared to 235U and 239Pu.
On the other hand, a significant drawback of the
3.04.5 Behavior of Thorium Oxide Th–U cycle is the larger time constant for b-decay of
Fuel Under Irradiation the intermediate species: 27 days for 233Pa compared
with 2.3 days for 239Np. This has consequences for
For in-depth treatment of the effects of irradiation on reactor physics and handling of spent fuel. Th-fueled
UO2 and MOX fuel, see Chapter 2.17, Thermal reactors with high neutron densities would contain
Properties of Irradiated UO2 and MOX; Chapter significant levels of 233Pa due to its slow decay, and
2.18, Radiation Effects in UO2 and Chapter 2.19, because it possesses a relatively high absorption
Fuel Performance of Light Water Reactors (Ura- cross-section, the protactinium will act as a ‘neutron
nium Oxide and MOX). poison.’ The thorium cycle therefore benefits from
98 Thorium Oxide Fuel
10 000
1000
100
10
s (barn)
0.1
0.01
232Th
238U
1E - 3
1E - 4
1E - 12 1E - 10 1E - 8 1E - 6 1E - 4 0.01 1 100
E (MeV)
1 00 000
10 000
1000
s (barn)
100
10
233U
235U
1 239Pu
0.1
1E - 12 1E - 10 1E - 8 1E - 6 1E - 4 0.01 1 100
E (MeV)
Figure 5 Capture (top) and fission (bottom) cross-sections of relevant nuclides.
sc sf n sc sf n sc sf n
Fissile materials
233
U 42.20 468.2 2.495 134.16 751.71 2.498 0.063 1.841 2.596
235
U 86.70 504.81 2.433 131.97 271.53 2.438 0.095 1.219 2.583
239
Pu 274.32 699.34 2.882 184.06 289.36 2.876 0.065 1.8 3.091
241
Pu 334.11 936.65 2.946 169.13 570.66 2.933 0.226 1.626 3.151
Fertile materials
232
Th 6.533 – – 84.97 – – 0.102 0.071 2.093
240
Pu 262.65 6.13 102 2.784 8448.7 3.74 2.785 0.095 1.349 3.013
238
U 2.414 1.05 105 2.489 277.7 2.16 103 2.49 0.07 0.3 2.598
Source: Data taken from Tables of Physical & Chemical Constants. 4.7.2 Neutron cross-sections. Kaye & Laby Online, version 1.0, 2005;
www.kayelaby.npl.co.uk.
c is the capture cross-section, f the fission cross-section, and is the average amount of neutrons produced per fission.
Cross-sections are given in barns (1024 cm2).
Thorium Oxide Fuel 99
systems with low neutron densities and from systems the Lingen BWR in Germany, the CANDU-prototype
with on-line removal of fission products. NPD and the NRX and NRU research reactors in
In a fast spectrum, the advantages of 233U over Canada, and the CIRUS reactor in India.76
235
U and 239Pu are in large part lost or countered In most of these irradiations and especially in the
by some disadvantages: First, 239Pu produces more earlier ones, the fissile component in the fuel has
neutrons than 233U in the fast part of the spectrum, been HEU. Some more recent work has focused on
the latter value staying relatively constant over the plutonium-containing fuels: two relatively recent
spectrum. Second, the ‘fertile’ material 238U is much projects on this topic (OMICO and Thorium Cycle)
more fissionable than 232Th in this region, upping the have been performed within the 5th Framework pro-
neutron yield. And finally, the capture cross-section gram of the European Commission,77 while several
advantage of 232Th over 238U is strongly reduced thorium–plutonium irradiations have been carried
(see Figure 5).2 out in India and Canada.8
One of the most important characteristics of tho- It is instructive to make the following points based
rium nuclear fuel from a radiological point of view is on data for the unirradiated materials (Section
the formation of 232U. 232U is generated in thorium fuel 3.04.3), before discussing the available data on actual
via several routes, but mostly from 233U through in-core behavior of thorium mixed oxide.
(n,2n)-reactions. It has a relatively short half-life of
69 years and a multistep decay chain ending at the The higher melting point and somewhat higher
stable 208Pb. One of the daughters, 208Tl, is radioactive, thermal conductivity of ThO2 fuels compared
with a hard g of 2.6 MeV. Therefore, the presence of with those of UO2-based fuels result in a larger
232
U, even in small amounts, in spent Th-fuel compli- temperature margin to fuel melting. This is accom-
cates handling during fabrication and reprocessing panied by enhanced physical stability, which adds
considerably. It is therefore an important consideration to safety under accident conditions.
in the fuel cycle approach to minimize the amount of The larger margin ‘to melt’ is also an indication
232
U in spent fuel or alternatively to opt for a once- that all thermally activated processes (creep, oxy-
through application of thorium, without reprocessing. gen diffusion, and migration of fission gases and
At the same time, the intense g-rays produced by the volatile fission products) proceed more slowly than
spent fuel enhance proliferation resistance. in UO2-based fuel.78 The very low concentration
We conclude that from the perspective of neu- of oxygen defects in ThO2, and the slightly
tron physics, the Th–U cycle generally compares reduced temperature gradient in the pellets add
well with the existing U–Pu cycle and that a signifi- to this effect. This slowdown of thermal processes
cant breeding advantage exists for thermal reactors. has a few notable consequences:
The current prevalence of uranium-based fuels in – a reduced chance of cladding corrosion by cor-
thermal reactors is therefore not due to neutronic rosive fission products79
properties of the fissile and fertile components but – lower fission gas release compared with UO2-
mostly for historic reasons and perhaps due to the based fuel. This reduces rod internal pressure
absence of a fissile component in naturally occur- (especially important under accident condi-
ring thorium. tions) as well as the loss of gap conductance
due to the heavy fission gases Xe and Kr.80 On
the downside, higher retention of fission gases
3.04.5.2 In-Core Behavior of Thorium
could results in increased fuel swelling and
Oxide Fuel
stronger fuel–cladding interaction.81
Much experience with thorium oxide fuel has been The high chemical stability of ThO2 suggests that
gained in the past. Several reactors have run at least the fuel is more resistant to corrosion in case of
partially on ThO2-based fuel: the AVR and the proto- coolant ingress.
type THTR in Germany, (both graphite-moderated The net result of uranium fissioning and oxidation
HTRs), the Elk River, Indian Point 1, Peach Bottom of some of the fission products is a gradual increase
HTGR and Shippingport LWBR reactors in the of the effective O/M ratio in the fuel. The inability
United States, and, most recently, the heavy water- of ThO2 to be further oxidized, together with the
cooled power stations Kakrapar-1and Kakrapar-2 in slow removal of oxygen dissolved in the matrix,
India. In addition, several irradiation tests have could result in a significantly higher oxygen poten-
been performed, a.o. in the Dragon HTR in England, tial in the fuel.81 As a result, it is likely that more
100 Thorium Oxide Fuel
fission products are present in the fuel in oxidized that were sintered in an oxidative atmosphere
states. For example, molybdenum, which has an showed more grain growth than the ones sintered
oxidation potential slightly above that of fresh in a reductive atmosphere. An irradiation of fuel
fuel, will start to oxidize even during the initial containing up to 50% uranium was carried out to
phases of the irradiation,81 whereas in the case of study grain growth in more detail.84 Columnar grain
UO2, this occurs only at higher burnups. growth appeared similar to that in UO2 fuel, but
proceeded at an estimated 350 K higher than for
The advantages of ThO2 over UO2 inferred above
UO2. In two other studies, it was estimated that
are expected to diminish with increasing uranium or
about 10–20% more power was needed to produce
plutonium content at varying rates. The fact that
a given structural change for thoria-based fuel,85 in
many quantitative uncertainties remain may be illu-
line with the idea that thoria offers more dimen-
strated by the case of postirradiation thermal con-
sional stability.
ductivity. In Section 3.04.3.3, a remark has been
made on the uncertainties in measurements on fresh
3.04.5.2.2 Thermal conductivity
fuel. Irradiation effects introduce additional uncer-
A low-burnup linear power ramp test at a uranium
tainty through the following effects:
concentration of 10%86 has indicated that the in-pile
A decrease in thermal conductivity during the early thermal conductivity of (Th,U)O2 follows that of pure
stages of irradiation due to the formation of lat- UO2 until about 1000 C, above which point UO2
tice defects by neutron and fission damage. Anneal- conductivity becomes independent of temperature,
ing of this damage in UO2 occurs at temperatures but (Th,U)O2 conductivity decreases further. This
above 500 C, so lattice defects are thought to have trend seems in line with the presence of an electronic
only a minor influence under normal operating contribution for UO2,40 but the observed crossover
conditions. Similarly, a loss of effective conductiv- temperature is much too low. A qualitative explana-
ity due to pellet cracking may be annealed at higher tion is that annealing of irradiation damage is much
temperatures. It is not immediately clear whether larger for UO2 (Section 3.04.3.3). Very limited data
these annealing mechanisms hold similarly for are available regarding the extent of short-term ther-
ThO2. mal conductivity loss in ThO2- based fuel, and the
A longer-term decrease in thermal conductivity related annealing temperature. In a low-temperature
due to the formation of fission products. For measurement of the in-pile thermal conductivity
UO2, the thermal conductivity decreases at a rate of (Th,U)O2 containing 1.3% UO2,87 the initial con-
of 6–8% per 10 GWd per t UO2.11 ductivity of 12.5 W mK1 at 60 C decreased rapidly
The modification of the porosity structure. The Xe with increasing exposure, to a 50% lower plateau
and Kr formed during irradiation may cause the value; postirradiation annealing led to a virtually
formation of large bubbles at higher burnups or complete recovery at 1000 C. On the other hand,
higher temperatures, lowering the overall thermal Jacobs compared the unirradiated and in-pile irra-
conductivity. diated conductivity of Th0.9U0.1O2 and concluded
that there was no statistical difference.11 Matolich
An overview of information from irradiation tests
and Storhok measured the postirradiation thermal
is given in Section 3.04.5.2.1–7. The discussion
conductivities of ThO2–UO2 pellets containing 3,
mostly relates to (Th,U)O2, with a shorter piece on
10, and 15% uranium, which had been irradiated
(Th,Pu)O2 at the end (see Chapter 2.17, Thermal
under varying conditions.11 No significant differ-
Properties of Irradiated UO2 and MOX).
ences were observed for the 3% sample, which was
attributed to the relatively high irradiation tempera-
3.04.5.2.1 Restructuring ture (900 C on average). The 10% sample showed
Short zero burnup ramp tests have been performed a higher postirradiation conductivity, which was
in the NRX reactor on (Th,U)O2 with uranium explained as due to the formation of columnar grains.
concentrations between 2.7% and 19% in order to The higher burnup 15% sample finally showed a loss
study the restructuring under fast reactor conditions of thermal conductivity of more than 50% even after
and establish the linear powers needed for central annealing at 1200 C. Taken together, these observa-
fuel melting.82,83 It was found as expected that the tions do not provide clear conclusions, but reveal
power ‘to melt’ decreased with uranium concentra- the many different parameters that influence the
tion in the range 5.4 –19%. After irradiation, pellets thermal conductivity.
Thorium Oxide Fuel 101
3.04.5.2.3 CANDU fuel gas release was observed for especially the 12.7% and
The overall conclusion from a summary of Canadian 25.6% samples. Postirradiation examination showed
experience from the 1960s88 was that thorium-based that significant columnar recrystallization was accom-
mixed oxide behaves similarly to UO2 when com- panied by oxygen release; the decrease in thermal
pared at the same fraction of the melting tempera- conductivity associated with the oxygen release
ture.88 North American experience from the early served to explain the high fission gas fractional release.
1970s has been summarized in contributions for the During the Thorium Utilization program, sol–gel
ANS winter meeting 1977.80,89,90 From a study on sphere-pac (Th,U)O2 fuel (4.5% U) was compared
thorium–uranium oxide fuel containing 1.6% ura- with pellet fuel, at high burnups under LWR condi-
nium in a CANDU-type reactor (linear power up tions. The behavior of the different fuel types was
to 650 W/cm, burnup 16.7 GWd per tHM), it was similar, and both types ‘had basic performance char-
concluded that ‘FGR is approximately 1/10 that acteristics for a power reactor fuel up to 10 at.%
of UO2 fuel operating at the same power.’89 Other burnup.’11 Fission gas fractional release was generally
conclusions from CANDU studies on thoria fuel con- less than 20%, and fuel volume changes were less than
taining 1–3% uranium are that power ramping after 0.5% per % burnup. In another high-burnup irradia-
prolonged irradiation at lower powers induces defects tion on pellet fuel of the same composition and a
consistent with UO2 fuel, but that ‘fission product burnup of 10%, fission gas fractional release was
release rates (following cladding failure) are between 14%, ‘comparable to other oxide fuels under similar
one and two orders of magnitude lower, and fuel conditions.’
deterioration is much less.89 A series of irradiations During the Indian Point PWR program, a full core
in the NRX (heavy water-moderated) reactor led to of (Th,U)O2 pellet fuel containing 6.5 and 9.1%
the conclusion that for higher uranium concentra- U operated at linear powers of 70–410 W cm1 to
tions of 5–10%, the (Th,U)O2 fuel behaved ‘essen- burnups of 3 at.% with little change except for fuel
tially the same as UO2 fuel,’ whereas even higher cracking. Fission gas fractional release was below 2%
concentrations (up to 19%) ‘resulted in poorer in all cases, and it was pointed out that a similar
performance.’90 amount was expected from UO2 fuel under similar
An extensive set of Canadian (CANDU) high- conditions.
burnup irradiations up to 1985 is summarized in The extensive experience from the Shipping-
Hastings et al.91 Nine irradiations are described of port light water breeder reactor program has
pinlets and multipin (Th,U)O2 fuel bundles contain- been summarized in Atherton et al.92 and Olson
ing pellets produced with different fabrication para- et al.93 The performance of the seed and blanket
meters. In these experiments, the maximum linear ThO2/233UO2 fuel rods was deemed ‘excellent,’
power was 570 W cm1 and maximum burnup around with no conclusive evidence of failure (the ‘seed’
35 GWd per t. No failures were observed. Fission gas fuel had a uranium content of 5–6 wt%, whereas
release was never above 5%. Intermediate power the blanket fuel contained only 1.5–3 wt%). Nota-
ramping up to 600 W/cm at a maximum burnup of ble conclusions from postirradiation examination
21 GWd per t in the BDL-421 experiment did not were as follows92:
cause any failure. It should be stressed here that
CANDU fuel contains little fissile material, and Fission gas fractional release at the end of the core
because HEU was used, the investigated uranium life was less than 0.2%
concentration was only 1–3%. The lack of significant grain growth confirmed that
the operating temperature was below 1415 C and
3.04.5.2.4 United States that the fuel was stable.
American experience from four extensive irradiation Fuel pellets remained cylindrical and essentially
programs is discussed in detail in Hart et al.,11 which intact. Limited axial and circumferential cracks
forms a very useful resource for data on thorium oxide were observed, but there were no fuel chips or
fuel irradiations. Here we summarize the most impor- dislocated pellet fragments.
tant findings, leaving out the experimental details. Fuel pores remained small up to burnups of
(Th,U)O2 fuel with U concentrations of 6.4%, 20 GWd per tHM; larger grain boundary fission gas
12.7%, and 25.6% were irradiated in the MTR and bubbles became evident only at higher burnups.
the engineering test reactor (ETR) at very high linear No oxide was observed on the fuel side of the
heat rates more typical for fast reactors. High fission zircaloy cladding.
102 Thorium Oxide Fuel
Essentially no iodine was detected in the fuel– per tHM, no release was observed.95 Finally, Shiba
cladding gap or on the cladding inner surface, but et al. found that xenon release from (Th,U)O2
low levels of 137Cs were present. powders increased with urania content and that the
release was more or less independent of preparation
From Shippingport irradiation tests on fuel with
conditions.96
a wider range of uranium concentrations (0–20%),
fission gas release measurements irradiated at peak
3.04.5.2.6 Fission product behavior
linear powers from 72–738 W/cm up to burnups in
(See Chapter 2.20, Fission Product Chemistry in
the range 0.9–56.5 GWd per tHM have yielded release
Oxide Fuels) Postirradiation fuel composition ana-
rates in the range 0.1–5.2% with an average below
lyses of Th0.8U0.2O2 by Berman97 and Th0.95O0.05O2
1%, ‘much less than typical for UO2 fuels.’80 The
by Padden et al.98 revealed the presence of metallic
stability of fuel with up to 10% U in the temperature
globules of Mo alloyed with Tc,Ru,Rh,Pd at the edges
range 600–1000 C was superior to that of UO2 at the
of columnar grains. These inclusions are also present
evaluated burnups (<1%),11 whereas thermal con-
in high-burnup UO2 and (Pu,U)O2 fuel. ‘Gray phase’
ductivity was unchanged from fresh fuel for a sample
perovskite inclusions of the BaZrO3 type, however,
with a burnup of 0.03%. Low-temperature creep
were not observed. Ugajin et al. studied the chemical
studies, however, indicated that fission-induced
form of solid fission products in some detail by repro-
creep was higher than in UO2 fuel. For samples con-
ducing the elemental composition of high-burnup
taining less uranium (2 instead of 10%), the creep was
Th0.81U0.19O299 and did find the expected ‘gray
again higher by 25%. The following general conclu-
phase’ inclusions in the form of (Ba,Sr)(Zr,Ce)O2.
sions from US experience are given in Hart et al.11:
Compared with (U,Pu)O2 fuel, the absence of uranate
1. Fission gas release for fuel with uranium content and plutonate and the large enrichment of Ce in the
up to 10% is comparable to that of UO2 under gray phase are notable; Cs was also absent from
typical LWR conditions. Within the investigated the gray phase. The solutes (rare earths, Ce, Zr)
range, increasing the UO2 content has little effect were found to induce a contraction of the unit cell of
on the release behavior. the fluorite matrix. The authors finally argue that the
2. Dimensional stability is better than for UO2, and thermochemistry of the solid fission products in pure
very little irradiation-induced swelling occurs. Up ThO2 would be similar.99 Momin et al. quantitatively
to 10% burnup, the volume change is less than 1% evaluated the unit cell contraction induced by the
per % burnup. dissolution of rare earths and confirmed that, unlike
3. Thermal conductivity remains unchanged under UO2, ThO2 and (Th0.8U0.2)O2 did not form single-
irradiation but decreases with increasing UO2 phase fcc solutions of Ba and Sr at all.100
content. Thermal expansion behavior is essentially Some experimental results on the diffusion of
the same as for UO2 fuel (note the disagreement volatile fission products in thorium oxide fuels
with perhaps more accurate out-of-pile measure- (I (Br), Cs (Rb), Te) reveal the same trends that
ments, Section 3.04.3.2). are indicated by fission gas release measurements
4. Structural changes under irradiation are generally (Xe and Kr): a drastic slowdown of diffusion is
less than for UO2 under similar circumstances. observed as compared with urania,101–105 to the
5. Due to the similarities in behavior, a fast route to point that bulk diffusion of I and Te was ‘too low to
the licensing of thoria-based fuels would be to be measurable’ below 1000 C for trace-irradiated
establish correlations with the existing database Th0.98U0.02O2.102 Initially, the activation energy of
on UO2 fuel under normal operating conditions. migration increases and the release decreases with
Normalization of the UO2 database for ThO2- burnup, but these changes level off at higher burn-
based fuel would require significantly less data- ups.104 (Th,U)O2 fuel containing 10% uranium has
points. However, off-normal and safety-related been irradiated in the the experimental boiling water
data will be necessary. reactor (EBWR) up to a uranium burnup of 3% while
monitoring the release of volatile species.106 Escape
3.04.5.2.5 India of radionuclides, notably iodine, was ‘much smaller
In India, ThO2 rods have been tested in the central than for UO2 fuel in PWRs,’ while only Xe and Kr
parts of some reactor cores as a means to flatten were significantly released. Similarly, tests with de-
the power profile.94 From a ThO2 fuel element of fected fuel in the EBWR showed no increase in gross
the Kakrapar power station with a burnup of 11 GWd activity of the reactor water.11
Thorium Oxide Fuel 103
(a) (b) 1 mm
Figure 6 (a) Axial ceramograph of an unirradiated (Th,Pu)O2 pellet containing 3% Pu. (b) Radial ceramograph taken
after irradiation in the KWO Obrigheim reactor. Reproduced from Somers, J.; Papaoiannou, D.; Sommer, D. Irradiation of
thorium–plutonium mixed oxide fuel to 37.7 GWd/t in the Obrigheim pressurised water reacter (KWO). In Proceedings
of GLOBAL 2009, Paris, France, 2009.
104 Thorium Oxide Fuel
of regular (U,Pu)O2 fuel, when taking into account phase, after which it is back-extracted (‘stripped’) into
the differences in burnup.111 an aqueous phase. The now roughly separated ura-
nium and plutonium streams are decontaminated
3.04.6 Reprocessing and by additional extraction–back-extraction cycles, in
Refabrication which the TBP concentration, the acid concentra-
tions in aqueous and organic phases, and the temper-
Reprocessing (see Chapter 5.14, Spent Fuel Disso- ature are varied to obtain desired separation levels.
lution and Reprocessing Processes and Chapter Although the technical feasibility of reprocessing
5.15, Degradation Issues in Aqueous Reprocessing thorium–uranium mixed oxide was demonstrated in
Systems) of spent UO2 and (U,Pu)O2 fuel for LWRs the 1950s in the United States,5,113–116 where rela-
(i.e., the PUREX process) is currently performed in tively large reprocessing plants based on the
France and Japan and consists of the following steps: THOREX reprocessing scheme have operated for
years,117,118 and later in Germany,119 thorium repro-
Head end: Mechanical processing (‘chopping’) and cessing has not fully matured. The main reasons are
dissolution of spent fuel. UO2 and (U,Pu)O2 (up to the ban on using the HEU needed to seed the tho-
plutonium contents of 35%112) may be dissolved in rium, and the industries’ choice for the uranium–
concentrated nitric acid, allowing for separation plutonium fuel cycle. Issues specific to thorium,
from the insoluble cladding. reviewed in references,117,120,121 are discussed in the
Separation of U, Pu, and fission products into next subsection.
different stream solutions, based on the partition-
ing of the different components over aqueous and
organic phases. 3.04.6.1 The THOREX Process
Purification of the different product and waste The THOREX process was developed to separate
streams. fissile uranium from the fertile thorium matrix, in
Back end: Vitrification of waste and conditioning of analogy with the PUREX process. The high stability
the reusable U (as UO2(NO3)2) and Pu (as PuO2). of the ThO2 matrix, which is an advantage when
A representation of the separation steps is given in considering in-pile behavior and waste disposal, ren-
Figure 7(a). In short, U and Pu are first extracted ders ThO2 and (Th,U)O2 practically insoluble in
from the nitric acid solution into an organic phase nitric acid. Spent thorium fuel is instead dissolved
containing 30% tri-n-butylphosphate (TBP), which using the ‘THOREX reagent’ (13 M HNO3 þ 0.03–
has a strong affinity for ions of valence þ4 and þ6 0.05 M F þ 0.1 M Al3þ).5,119 Here, the fluoride (as
and thus separates Pu(IV) and U(VI) from most of the HF) is added to the nitric acid solution in order to
fission products. The Pu(IV) is then reduced to catalyze dissolution. However, the fluoride also facil-
Pu(III) with a much lower solubility in the organic itates corrosion of the stainless steel reactor vessels.
PUREX
THOREX
FP (aq) Th (aq)
Th, U, FP (aq)
100
Partition coefficient
10
U(VI)
Th(IV)
Pu(IV)
1
Zr
Nb
0.1 Ru
Pu(III)
Acidity (M)
0.01
0 2 4 6 8 10 12 14
Figure 8 Partition coefficients (ratios of concentrations in the organic and aqueous phase, respectively) of Th, U, and Pu, as
well as for some more problematic fission products. The numbers are for extraction by 20% TBP in kerosene. Merz, E.
Wiederaufarbeitung im Thoriumbrennstoffkreislauf: ein Problemkatalog, KfA Jülich, 1984.
Aluminum ions are added (as Al(NO3)3) to reduce waste stream. The most widely used salting agent is
this problem by forming a complex with the fluoride. therefore HNO3. This case is referred to as the ‘acid
Dissolution of the fuel may be further enhanced THOREX’ flowsheet.
by starting from a lower sintered density compared To avoid the formation of a second organic phase,
with that of UO2-based fuel. In addition, a reduction the concentrations of thorium and HNO3 should be
in the dissolution time by a factor 2 may be obtained limited; the loading of thorium in particular is limited
by adding about 1.5% MgO to the fuel (which also to about 30% of the theoretical capacity. These con-
enhances sintering of the pellets;3 Section 3.04.4.1). centration limits lead, however, to high thorium
It is to be noted that the zircaloy cladding was found losses during the extraction step (Figure 8). As a
to be sufficiently resistant to corrosion by the ‘tho- countermeasure, concentrated HNO3 is added at
rium reagent.’116 the end of the extraction step, where thorium con-
Figure 7(b) gives an overview of the THOREX centrations are low.
separation phase. As for PUREX, the first step The subsequent separation of Th and U rests on
involves extracting the heavy metals from the aque- the difference in extractability between U(VI) and
ous phase by complexation with TBP: Th(IV). Here the absence of a reduction step is an
advantage with respect to the PUREX process, but
Th4þ þ 4NO
3 þ xTBP < > ThðNO3 Þ4 xTBP
the smaller difference in extractability between the
heavy metal ions forms a drawback. In India, an
alternative route is being developed in which U and
Figure 8 gives partition coefficients (ratios of con-
Th are separated during the first extraction step. In
centrations in the organic and aqueous phase) for the
this case, less (5%) TBP is added to the organic
relevant actinides. It may be clear from the figure that
phase, resulting in U-only extraction.
Th(IV) is less extractable than U(VI) or Pu(IV). This
necessitates the use of a salting agent119 to drive the
3.04.6.2 Beyond Thorex
complexation reaction to the right-hand side by
increasing the concentration of NO 3 counterions. The THOREX flowsheets were developed in the
Al(NO3)2 has been used for this, but it has the disad- 1950s and 1960s for fuels in which HEU contributes
vantage that the aluminum adds to the volume of the the fissile material. However, with the current ban on
106 Thorium Oxide Fuel
HEU and with the aim of reducing plutonium stocks, 3.04.7 Conclusions
the more likely fuel to be used in LWRs is thorium–
plutonium mixed oxide. After reaching the required From the above, the general conclusion can be drawn
burnup, this fuel contains significant amounts of Th, that (Th,U)O2 fuel significantly outperforms enriched
U, and Pu. No scheme currently exists to separate the UO2 for uranium concentrations up to about 5%,
three components, but there are some theoretical ideas while fission gas retention and dimensional stability
on what such a scheme would look like, and India has are still slightly better for concentrations up to about
concrete plans to breed 233U from Pu–Th mixed oxide 15%.90 Much confidence has been built for the use of
fuel in the currently developed advance heavy water thoria-based fuels in thermal reactors under accepted
reactor. Wilson discusses the difficulties of full threeway UO2 operating conditions, for U concentrations smal-
separation, which are related to interferences between ler than 10%; the more limited experience with
the requirements for PUREX and THOREX separa- (Th,Pu)O2 indicates very similar behavior. For U or
tion, and considers keeping the U and Pu streams Pu concentrations above 15%, the limited data sug-
unseparated to simplify the flowsheets.120 gest that performance falls below that of UO2 on
account of significantly reduced thermal conductivity
and melting point depression. As the main interest in
3.04.6.3 Radiation Issues in
ThO2-based fuel at the time of writing is in the
Reprocessing and Refabrication
burning of Pu in LWRs or HWRs, further irradiation
In addition to described chemical issues, the testing of (Th,Pu)O2 in the 6–10% Pu concentration
THOREX process presents some problems related range and to high plutonium burnups is warranted.
to radiation levels. First, the almost 100% fissile Fabrication procedures for thorium mixed oxides are
nature of the uranium stream raises criticality issues, straightforward and analogous to that of UO2 and
which may be solved by changing the shape and size of (U,Pu)O2. The main technical challenges therefore lie
the reaction vessels and diluting the uranium stream. in the reprocessing steps, and subsequent (shielded)
Furthermore, the 233Pa precursor to 233U (Section refabrication. Without closure of the thorium-
3.04.5) has a half-life of 27 days, which means that a uranium cycle, its added value with respect to ura-
9 month cooling time (about ten half-lives) is neces- nium resource savings will remain limited.
sary to avoid an increase in reactivity during reproces-
sing and refabrication. Cooling is in any case needed
due to the relatively large decay heat of thorium- References
based fuel and from the perspective of recovering as
1. Clayton, J. C. American Chemical Society Meeting;
much 233U as possible.117 Pittsburgh, PA, 1993, http://www.osti.gov/
Furthermore, it has been mentioned in the intro- energycitations/servlets/purl/10191380-PWWUeT/
duction that irradiated thorium-based fuel carries native/10191380.pdf.
2. Lung, M. Nuclear Science and Technology project report
some 232U, which has daughters that emit hard EUR 17771 for the European Commission, 1997.
g-rays. This complicates handling during fabrication 3. Anantharaman, K.; Shivakumar, V.; Saha, D. J. Nucl.
and reprocessing considerably through the require- Mater. 2008, 383, 119–121.
4. Gruppelaar, H.; Schapira, J. P. Nuclear Science and
ment for biological shielding. However, because the Technology project report EUR 19142 for the European
initial step (232U to 228Th, t1/2 ¼ 73.6 years) is rela- Commission, 2000.
tively slow, handling of freshly separated uranium 5. The use of Thorium in Nuclear Reactors, document
WASH-1097 of the US Atomic Energy Commission, 1969.
may be less problematic during the first year after 6. Thorium based nuclear fuel: Current status and
irradiation.2,118 Here we hit a trade-off with the perspectives, IAEA TECDOC 412, 1987.
decay time of 233Pa. 7. Thorium based fuel options for the generation of
electricity: developments in the 1990s, IAEA TECDOC
After reprocessing, the thorium stream itself con- 1155, 2000.
tains 228Th (t1/2 ¼ 1.9 years), which also leaves the 8. Thorium fuel utilization: Options and trends, IAEA
thorium output radioactive for decades. As a result, TECDOC 1319, 2002.
9. Potential of thorium based fuel cycles to constrain
it may be advantageous to blend the reprocessed plutonium and reduce long lived waste toxicity, IAEA
uranium oxide with fresh ThO2 and consider the TECDOC 1349, 2003.
spent thoria as waste. In fact, due to its (previously 10. Thorium fuel cycle – potential benefits and challenges,
IAEA TECDOC 1450, 2005.
discussed) high chemical stability, and in particular 11. Hart, M. E.; Griffith, C. W.; et al. ThO2-based pellet
its high corrosion resistance, ThO2 is an excellent fuels – their properties, methods of fabrication and
waste form.75,122,123 irradiation performance; DOE report PNL-3064; US, 1979.
Thorium Oxide Fuel 107
12. Rodriguez, P.; Sundaram, C. V. J. Nucl. Mater. 1981, 100, 41. Gibby, R. L. J. Nucl. Mater. 1971, 38, 163–177.
227–249. 42. Golovnin, I. S. Atom. Energy 2000, 89, 627–637.
13. Ganguly, C. Nuclear reactor fuel fabrication. 43. Springer, J. R.; Aldridge, E. A.; Goodyear, M. U.;
In Encyclopedia of Materials: Science and Technology, Wright, T. R.; Lagedrost, J. F. Fabrication,
2008; pp. 6324–6339. characterization and thermal property measurements of
14. Rodriguez, P. Thorium fuel cycles. In Encyclopedia of ThO2-UO2 fuel materials, BMI-X-10210; Oct 1967.
Materials: Science and Technology, 2008; pp. 1–8. 44. Murabayashi, M.; Nucl, J. Sci. Technol. 1970, 7, 559.
15. Ramanujam, A. PUREX and THOREX processes. 45. Springer, J. R.; Lagedrost, J. F. Thermal properties of
In Encyclopedia of Materials: Science and Technology, ThO2–UO2 fuel materials (LWBR development program),
2008; pp. 7918–7924. Batelle Memorial Institute report BMI-X-10231; 1968.
16. Uranium Red Book (Uranium 2009: Resources, 46. Berman, R. M.; Tully, T. S.; Belle, J.; Goldberg, I. DOE
Production and Demand’), NEA/IAEA joint publication, report WAPD-TM-908; 1972.
NEA No. 6891, 2009. 47. Basak, U.; Sengupta, A. K.; Ganguly, C. J. Mat. Sci. Lett.
17. Herring, J. S.; MacDonald, P. E.; Weaver, K. D.; 1989, 8, 449–450.
Kullberg, C. Nucl. Eng. Des. 2001, 203(2001), 65–68, 48. Cox, J. D.; Wagman, D. D.; Medvedev, V. A. CODATA
1–20. key values for thermodynamics, Hemisphere: New York,
18. Magill, J.; Galy, J. Radioactivity Radionuclides Radiation; 1989.
Springer Verlag: Germany, 2005. 49. Martin, D. G. J. Nucl. Mater. 1988, 152, 94.
19. Schram, R. P. C.; Kuijper, J. C.; Sommer, D.; et al. The 50. Osborne, D. W. J. Chem. Phys. 1953, 21, 1884.
thorium cycle project. In Proceedings of the International 51. Fischer, D. F.; Fink, J. K.; Leibowitz, L.; Belle, J. J. Nucl.
Workshop on P&T and ADS Development, 2003. Mater. 2008, 118(2–3), 342–348.
20. Bakker, K.; Cordfunke, E. H. P.; Konings, R. J. M.; 52. Dash, S.; Parida, S. C.; Singh, Z.; Sen, B. K.;
Schram, R. P. C. J. Nucl. Mater. 1997, 250, 1–12. Venugopal, V. J. Nucl. Mater. 2009, 393, 267–281.
21. Thermophysical properties database of materials for light 53. Schram, R. P. C. J. Nucl. Mater. 2005, 344, 223–229.
water reactors and heavy water reactors, IAEA Technical 54. Cohen, I.; Berman, R. M. J. Nucl. Mater. 1966, 18,
Document 1496, 2006 77–107.
22. Sobolev, V.; Lemehov, S. J. Nucl. Mater. 2006, 352, 55. Ugajin, M. J. Nucl. Mater. 1982, 110, 140.
300–308. 56. Ugajin, M.; Shiratori, T.; Shiba, K. J. Nucl. Mater. 1983,
23. De Bruycker, F.; Boboridis, K.; Rini, M.; Konings, R. J. M.; 116, 172.
Somers, J.; Manara, D. The melting behavior of 57. Ugajin, M. J. Nucl. Sci. Technol. 1983, 20, 228.
plutonium dioxide. In Proceedings of the EMRS Spring 58. Kutty, T. R. G.; Hegde, P. V.; Khan, K. B.; et al. J. Nucl.
Meeting 2010, Strasbourg, France, 2010. Mater. 2002, 312, 224.
24. De Bruycker, F.; Boboridis, K. ;; Manara, D.; Pöml, P.; 59. White, G. D.; Bray, L. A.; Hart, P. E. J. Nucl. Mater. 1981,
Rini, M.; Konings, R. J. M. Materials Today 2010, 13(11), 96(3), 305–313.
52–55. 60. Smit-Groen, V. M.; Schram, R. P. C.; Somers, J.;
25. Carbajo, J. J.; Yoder, G. L.; Popov, S. G.; Ivanov, V. K. Contrad, R. HFR Fabrication Report Thorium Cycle,
J. Nucl. Mater. 2001, 299, 181–198. document Th5-DOC-016 of the 5th framework
26. Cordfunke, E. H. P.; Konings, R. J. M. Eds. program Thorium Cycle of the European Commission
Thermochemical Data for Reactor Materials and Fission (2002).
Products; Elsevier, 1990. 61. Fuller, E. L., Jr.; Agron, P. A. J. Coll. Int. Sci. 1976, 57,
27. Freshly, M. D.; Mattys, H. M. General Electric Report 193–200.
HW-76559 1962, 11, 6. 62. Lords Smith, R. E. J. Nucl. Mater. 2004, 328, 215–224.
28. Lambertson, W. A.; Mueller, M. H.; Gunzel, F. H. J. Am. 63. Gammage, R. B.; Fuller, E. L., Jr.; Holmes, H. F.; Coll, J.
Ceram. Soc. 1953, 36, 397. J. Coll. Int. Sci. 1972, 38(1), 91–96.
29. Slowinski, E.; Elliott, N. Acta Crystall. 1952, 5, 768. 64. Kutty, T. R. G.; Khan, K. B.; Achuthan, P. V.; et al. J. Nucl.
30. Gmelins Handbuch der anorganischen Chemie, 8th Mater. 2009, 389, 351–358.
edition, Thorium, Springer Verlag, Berlin (vol. C1, C2). 65. Kurina, I. S.; Gudkov, L. S.; Rumyantsev, V. N. Atom.
31. Hubert, S.; Purans, J.; Heisbourg, G.; Moisy, P.; Energy 2002, 12, 461.
Dacheux, N. Inorg. Chem. 2006, 45, 3887–3894. 66. OECD/NEA report. Trends in the nuclear fuel cycle:
32. Aitken, E. A.; Edwards, J. A.; Joseph, R. A. J. Phys. Economic, environmental and social aspects, 2001.
Chem. 1966, 70, 1085. 67. Kutty, T. R. G.; Khan, K. B.; Somayajulu, P. S.; et al.
33. Alexander, C. A.; Ogden, J. S.; Cunningham, W. Batelle J. Nucl. Mater. 2008, 373, 299–308.
Memorial Institute Report BMI-1789; 1967. 68. Kutty, T. R. G.; Kulkarni, R. V.; Sengupta, P.; et al. J. Nucl.
34. Touloukian, Y. S.; Kirby, R. K.; Taylor, R. E.; Lee, T. Y. R. Mater. 2008, 373, 309–318.
Thermal Expansion Nonmetallic Solids; IFI/Plenum: New 69. Kutty, T. R. G.; Somayajulu, P. S.; Khan, K. B.; Kumar, A.;
York, 1970. Kamath, H. S. J. Nucl. Mater. 2009, 384,
35. Olander, D. R. Fundamental Aspects of Nuclear Reactor 303–310.
Fuel Elements, ERDA, 1976; Chapter 10. 70. Ganguly, C. Bull. Mat. Sci. 1993, 16, 509–522.
36. Pillai, C. G. S.; George, A. M. J. Nucl. Mater. 2000, 277, 71. Kumar, N. Nucl. Tech. 1991, 96, 169–177.
116–119. 72. Vaidya, V. N. Sol–gel process for ceramic nuclear fuels,
37. Cozzo, C.; Staicu, D.; Somers, J.; Fernandez, A.; BARC newsletter, http://www.barc.ernet.in/publications/
Naisse, F.; Konings, R. J. Nucl. Mater. 2011, 416, nl/2002/200203–01.pdf.
135–141; Proceedings of the 2010 EMRS Spring 73. Vaidya, V. N. J. Sol Gel Sci. Tech. 2008, 46,
Meeting. 369–381.
38. Ronchi, C.; Hiernaut, J. P. J. Alloys Comp. 1996, 240, 74. Pai, R. V.; Mukerjee, S. K.; Vaidya, V. N. J. Nucl. Mater.
179. 2004, 325, 159–168.
39. Fink, J. K.; Chasanov, M. G.; Leibowitz, L. Argonne 75. Tables of Physical & Chemical Constants. 4.7.2 Neutron
National Laboratory Report ANL-CEN-RSD-77–1; 1977. cross-sections. Kaye & Laby Online, version 1.0
40. Young, R. A. J. Nucl. Mater. 1979, 87(2–3), 283–296. (2005) www.kayelaby.npl.co.uk
108 Thorium Oxide Fuel
76. Note that the Peach Bottom and Fort st. Vrain reactors 102. Kaimal, K. N. G.; Kerkar, A. S.; Shirsat, A. N.; et al.
ran on thorium carbide fuel. J. Nucl. Mater. 2003, 317, 189–194.
77. Schram, R. P. C.; Klaassen, F. C. Prog. Nucl. Energy 103. Shiba, K.; Itoh, A.; Akabori, M. J. Nucl. Mater. 1984,
2007, 49, 617–622. 126(1), 18–24.
78. Fehrenbach, P. J.; Miller, A. I. In Nuclear Engineering 104. Akabori, M.; Fukuda, K. J. Nucl. Mater. 1990, 186(1),
Handbook; Kok, K. D., Ed.; CRC Press, 2009; 47–53.
Chapter 14. 105. Matzke, H.; Ray, I. L. F.; Verrall, R. A. Diffusion
79. Bailly, H. Menessier, D.; Prunier, C. Eds. The Nuclear and behaviour of iodine and rubidium in UO2 and ThO2,
Fuel of Pressurized Water Reactors and Fast Reactors – 183. In: Water reactor fuel behaviour and fission
Design and Behaviour; CEA (Lavoisier Publishing): product release in off-normal and accident conditions,
France, 1999. IAEA document IWGFPT/27, 1986.
80. Goldberg, I.; Spahr, G. L.; Giovengo, J. F.; Waldman, L. A. 106. Robertson, R. F. S. Tests on Defected thoria–urania
Trans. Am. Nucl. Soc. 1977, 27, 308–310. fuel specimens in EBWR, ANL report 6022; 1960.
81. Basu, M.; Mishra, R.; Bharadwaj, S. R.; Das, D. J. Nucl. 107. Ganguly, C. ThO2-based fuels for PHWRs – Fabrication,
Mater. 2010, 403, 2204–215. characterization and test irradiation. In 2nd International
82. Rao, S. V. K. Investigation of ThO2-UO2 as a nuclear fuel, Conference on CANDU Fuel, Canada, 1989, 398.
AECL report 1785; 1963. 108. Karam, M.; Dimayuga, F.C.; Montin, J. Fission gas
83. Rao, S. V. K. J. Nucl. Mater. 1964, 12(3), 323–329. release of (Th,Pu)O2 CANDU fuel.
84. Neimark, L. A. Trans. Am. Nucl. Soc. 1962, 5(1). 109. Verwerft, M.; Wéber, M.; Lemehov, S.; et al. Oxide fuels:
85. Jeffs, A. T. Thermal conductivity of ThO2–PuO2 under microstructure and composition variations (OMICO).
irradiation; AECL report 3294; 1969. In Proceedings of the FISA, 2003.
86. Kraerheim, G.; Rolstad, E. In-core study of fuel/clad 110. Somers, J.; Papaoiannou, D.; Sommer, D. Irradiation
interaction and fuel centre temperature, OECD Halden of thorium–plutonium mixed oxide fuel to 37.7 GWd/t
Reactor Project report HRP-107; 1969. in the Obrigheim pressurised water reacter (KWO).
87. MacEwan, J. R.; Stoute, R. L. Annealing of irradiation Proceedings of GLOBAL 2009, Paris, France, 2009.
induced thermal conductivity changes in ThO2–1.3 wt% 111. Hania, P. R. et al. unpublished results from the HFR
UO2. In 70th Annual Meeting of American Ceramic Thorium Cycle irradiation.
Society, Mar 1969; paper 11-N-68.. 112. Treatment and recycling of spent nuclear fuel,
88. Lewis, W. B.; Duret, M. F.; Craig, D. S.; Veeder, J. R.; M. Lecomte (editor), a CEA Nuclear Energy Division
Bain, A. S. Report AECL-3980; Atomic Energy of Canada monograph.
Limited, 1971. 113. Flag, J. F. Ed. Chemical Reprocessing of Reactor Fuels;
89. Jones, R. W.; Lee, H. R.; Hahn, H.; Walter, J. F.; Celli, A. Academic Press, 1961.
Trans. Am. Nucl. Soc. 1977, 27, 304–305. 114. Stoller, S. M.; Richards, R. B., Eds. Reactor Handbook,
90. Lane, A. D.; Jones, R. W.; Hamel, D. Am. Nucl. Soc. Vol. II: Fuel Reprocessing; Interscience: New York, 1961;
1977, 27, 305–306. 2nd edn.
91. Hastings, I. J.; Carter, T. J.; MacGillivray, G.; 115. Blanco, R. E.; Ferris, L. M.; Ferguson, D. E. Aqueous
MacDonald, R. D.; Judah, J. Canadian High burnup Processing of Thorium Fuels; ORNL-3219; 1962.
fuel experience. In Proceedings of 6th Annual 116. Nicholson, E. L.; et al. Recent developments in
Conference of the Canadian Nuclear Society, 1985. thorium fuel processing. In Second Symposium on
92. Atherton, R. Water cooled breeder program Thorium Fuel Cycle, Gatlinburg, TN, USA, 1966.
summary report; DOE R&D report WAPD-TM-1600; 117. Greneche, D.; Szymczak, W. J.; Buchheit, J. M.;
1987. Delpech, M.; Vasile, A.; Golfier, H. Rethinking the thorium
93. Olson, G. L.; McCardell, R. K.; Illum, D. B. Fuel fuel cycle: An industrial point of view. In Proceedings of
summary report: Shippingport Light Water Breeder the ICAPP Conference; 2007; Paper 7367.
Reactor; INEEL report EXT-98–00799; 2002. 118. Brooksbank, R. E.; McDuffee, W. T.; Rainey, R. H.
94. Anantharaman, K. Irradiation experience with thoria A review of thorium reprocessing experience, ORNL
based fuels, paragraph 3.2 in BARC report 1532; 1990. report for the DOE.
95. Viswathan, U. K.; Anantharaman, S.; Sahoo, K. C. 119. Zimmer, E.; Merz, E. Chemical processing of HTR fuels
Measurement of fission gas release from irradiated applying either THOREX or PUREX flowsheets.
nuclear fuel elements, Report BARC/2005/E/026 of the In Proceedings of the Specialists’ Meeting on
Bhabha atomic research centre, 2005. Gas-Cooled Reactor Fuel Development and Spent Fuel
96. Shiba, K.; Itoh, A.; Ugajin, M. J. Nucl. Mater. 1981, 96(3), Treatment, 1983, IAEA IWGGCR8.
255–260. 120. Wilson, P. D. Development requirements in reprocessing
97. Berman, R. M. WAPD-TM-1236, 1976. thorium fuels; In Thorium fuel utilization: options and
98. Padden, T. R.; Burton, R.; Campbell, C. E. WAPD-TM-644, trends; IAEA TECDOC 1319; 2002 [5].
1976. 121. Carter, W. L. Thorium Utilization Program: A Survey of
99. Ugajin, M.; Shiratori, T.; Shiba, K. J. Nucl. Mater. 1979, Processing Methods for Thorium Reactor Fuels; Report
84, 26–38. ORNL-TM-241; Oak Ridge, 1962.
100. Momin, A. C.; Mirza, E. B.; Mathews, M. D. J. Nucl. 122. Markowitz, J. M.; Clayton, J. C. Corrosion of oxide
Mater. 1991, 185, 308–310. nuclear fuels in high temperature water, WAPD-TM-909;
101. Das, D.; Kaimal, K. N. G. Release behavior of gaseous 1970.
and volatile fission products in irradiated thoria based 123. Thomas, L. E.; Einziger, R. E.; Buchanan, H. C. J. Nucl.
fuel materials – a laboratory scale study. In Conference Mater. 1993, 20(1), 310–319.
on Proceedings of High Burn-Up Issues in Nuclear Fuels, 124. Tel, H.; Eral, M.; Alta, Y. J. Nucl. Mater. 1998, 256,
Mumbai, India, 2005; pp. 137–142. 18–24.
3.05 Actinide-Bearing Fuels and Transmutation Targets
S. Pillon
Commissariat à l’Energie Atomique, St Paul Lez Durance, France
109
110 Actinide-Bearing Fuels and Transmutation Targets
1E + 09
Spent fuel
1E + 08
Plutonium
1E + 07 Minor
actinides
Radiotoxicity (SV/tHMi)
1E + 06 Uranium
Fission
1E + 05 products
Cesium
1E + 04
Technetium
1E + 03
Iodine
1E + 02
Activation
products
1E + 01
10 100 1000 10 000 100 000 1000 000
Time (years)
Figure 1 Evolution of the radiotoxicity of spent fuel (as unloaded from the reactor) versus time. Radiotoxicity is an
indicator of dangerousness. It is expressed according to the following formula: Radiotox (Sv) ¼ Activity (Bq) dose factor.
The dose factor is established by the ICPR (International Commission on Radiological Protection).
Actinide-Bearing Fuels and Transmutation Targets 111
radiotoxic and long-lived elements from spent fuel and manufacture the fuel, in a few hundreds and a few
transmute them into short-lived or nonactive ones in tens of thousands of years, respectively, the ‘light’
the nuclear reactors is being carried out on an inter- waste cleared from MA reaches the reference thresh-
national level. old in a few hundred years.
The research carried out in France until 2006, Transmutation of MAs cannot take place unless
under the scope of the Act dated 30 December 1991 they have been individually, or as a group, separated
concerning management of high activity and long- from the spent fuel. Separation and transmutation
lived radioactive waste1,2 shows on the one hand that are, therefore, the two inseparable stages on which
all types of nuclear reactors do not provide the intensive research has been carried out since 1991.
same transmutation potential and on the other hand This chapter evaluates transmutation of MAs by
that all radionuclides cannot be efficiently trans- describing the main research breakthroughs con-
muted.3 Therefore, the best transmutation perfor- cerning manufacturing of fuel and targets and their
mance is obtained in the fast neutron reactor (FR) behavior under irradiation.
and transmutation is only reasonably applicable for
MAs, chiefly americium, neptunium, and curium.
Regarding FPs, their contribution to the radiotoxicity 3.05.2 The Various Transmutation
inventory decreases after a few hundred years. This Strategies of MAs
factor, along with technical difficulties (mainly for
129 3.05.2.1 Which Reactors for
I, disadvantaged by its high volatility, which limits
Transmutation?
thermal stability of iodine compounds in the reactor
and for 135Cs, whose low content requires a prior Transmutation refers to the transformation, under a
isotopic separation level), greatly reduces the benefit neutron flux, of long-lived radionuclides into nonra-
of their transmutation. dioactive elements or elements with a much shorter
Figure 2 therefore shows the benefit that can be lifetime. This process may be achieved by fission
gained from transmutation of the radiotoxic footprint reaction or by neutron capture reaction. In the case
of high activity waste on geological storage. While of a fission disappearance, the nucleus is transformed
the radiotoxicity of spent fuel and of vitrified waste into FPs, mainly short-lived ones (<50 years). In the
containing FPs and MAs decreases to the arbitrary case of disappearance by capture, the nucleus is
referenced value, which constitutes the radiotoxicity transformed into another radionuclide, which does
of the natural uranium mineral that was used to not necessarily allow significant reduction of the
10 000
1000
Relative radiotoxicity
100
10
Isotope Thermal spectrum reactor Epithermal spectrum reactor Fast neutron reactor
(PWR – UOX) (PWR – MOX) (SFR)
Table 2 Neutron balance of the transmutation reaction PWR thermal spectrum, which, combined with the
per type of reactor fact that an FR flux is generally 10 times higher than
Isotope Thermal spectrum Fast that of a PWR, makes the FR a better candidate for
reactor spectrum transmutation. MAs notably contribute to the neutron
(PWR) (SFR) balance and to maintaining the chain reaction in FRs
237
Np þ1.10 0.55
whereas they poison the core of PWRs. Transmutation
241
Am þ1.07 0.54 of FPs is costly in the thermal spectrum as well as in the
243
Am þ0.32 0.67 fast spectrum as they are net consumers of neutrons.
244
Cm 0.66 1.47 Therefore, transmutation studies today are pre-
245
Cm 2.29 2.68 dominantly geared toward MAs in two major FR
Minor actinide (total) þ0.91 0.61
99
Tc þ1.01 þ1.02
families: the sodium-cooled or gas-cooled fast reac-
129
I þ1.01 þ1.02 tors (respectively SFRs and GFRs) and the subcriti-
135
Cs þ1.01 þ1.01 cal hybrid reactors driven by an accelerator (ADS
LLFP elements (total) þ2.3 þ2.2 for accelerator-driven system) cooled with lead or
LLFP isotopes (total) þ0.6 þ0.6 lead-bismuth.
The first type are fourth-generation reactors for
power production (electricity and heat) which are
long- or medium-term radiotoxicity. Thus, for effi- currently being developed by international partners
cient transmutation, fission reactions are of course under the scope of the GIF (Generation IV Interna-
privileged over capture reactions. tional Forum)4 to replace the PWR system beyond
Table 1 gives the average efficient cross-sections 2040. The second type is a reactor aimed at the
of capture sc and fission sf for each radionuclide, as incineration of nuclear waste from power production.
well as the a ratio of these cross-sections for three The subcritical nature of the reactor allows the core
characteristic neutron spectra: a PWR spectrum with to be heavily loaded with MAs while ensuring its
UOX fuel, a PWR spectrum with MOX fuel, and a control and safety. However, the complexity of this
FR spectrum. The a ratio indicates the probabilities system does not allow it to be considered as an
of disappearance by capture (section sc) rather than electrogenic reactor from an economical standpoint.
by fission (section sf) at the first neutron interaction. Their implementation in a ‘double-strata’ system is
Note that it is reduced by a factor of 5 at least for therefore considered: stratum No. 1 is made up of
most radionuclides when we move from the neutron electrogenic reactors (PWR and/or FR) with their
spectrum of a PWR (thermal or epithermal) to that associated cycle and stratum No. 2 is composed of
of the FR. incinerator reactors in which all the high activity
Table 2 shows the neutron balance of trans- nuclear waste of the nuclear fleet is burned with
mutation reaction for the main actinides per spec- their own fuel cycle. This type of reactor was studied
trum type. A positive balance corresponds to a net till March 2010 in the European integrated project
consumption of neutrons per transmutation reaction, EUROTRANS of the sixth framework program.5
whereas a negative balance corresponds to a produc- Today, the challenge of the research concerning
tion of excess neutrons. separation and transmutation is to show the feasibility
Both tables show that the FR spectrum offers on an industrial scale especially for the SFR system
better conditions for transmutation by fission than the and on a demonstration scale for ADS, while
Actinide-Bearing Fuels and Transmutation Targets 113
time in the SFR is incompatible with the lifetime of 3.05.2.3 Transmutation in ADS
structure materials to achieve the goal of a once-
In transmutation-dedicated reactors, MAs are the
through recycling. The transmutation process needs
fuel as they fundamentally contribute to core reactiv-
to be accelerated by moderating or softening the
ity. The ADS fuel is a very innovative concept with
neutron flux. Moderation allows the efficient fission
respect to the fuel of a critical core. The fuel is
cross-sections to be locally increased while preserv-
nonfertile, that is to say free of uranium, so as to
ing the high level of neutron flux in the SFR. This is
increase the transmutation performance by preventing
the objective of ‘moderate targets.’ Only hydrogen
the production of superior actinides through capture
moderators (CaHx, ZrHx, and YHx) allow fission
on the uranium. It has to contain a high concentration
rates higher than 90% to be obtained without
of actinides (of 45–70%) of the heavy atoms asso-
exceeding damage limit of 200 dpa for the cladding.
ciated with plutonium (for the complementary part)
Furthermore, as the fission rate and the mass
whose isotopic vector is very degraded (80–90% of
capacity of transmutation increase in reverse direc-
even isotopes Pu 238, 240, and 242 and 10–20%
tion versus MA content in the target, the 90% objec-
of uneven isotopes Pu 239 and 241) stemming from
tive of fission rate can only be attained by limiting the
multirecycling of plutonium in the first and second
americium content of target fuel pin to <15 vol%
strata (bottom box in Figure 3). Like targets, the acti-
(5% volume in the assembly). This is the limit of
nide compounds are spread in an inert matrix to better
physical composition. This is one of the main draw-
control their power and thermics while avoiding higher
backs of this type of recycling. The low capacity
actinide production.
would require involving a large number of reactors
There is therefore a certain composition analogy
in the fleet in the transmutation process.
between ADS fuel and SFR targets on inert supports.
Major research work needs to be carried out for both
3.05.2.2.3 Heterogeneous recycling in
to assess their in-pile behavior (see Section 3.05.4.2).
breeder blanket
This atypical composition moreover leads to a
Presently, one promising recycling option seems to
high g and neutron emission which will have to be
be the concept of MA incineration through a hetero-
taken into account in the fuel cycle to be implemen-
geneous method in the radial blanket of SFRs.7 In this
ted in order to ensure its reprocessing and its
concept, a relatively high quantity of MAs (which
manufacturing (see Section 3.05.5).
may reach 20% of heavy nuclei in mass) is spread
within the UO2 (U-depleted) breeder blanket to be
transmuted in the peripheral zone of the core. 3.05.3 Composition of
The blankets loaded with MAs are also to be Transmutation Fuel
manufactured in dedicated plants, but they may be
3.05.3.1 Actinide Compounds
reprocessed by dilution with the standard flux of fuel
assemblies in the current plants. This is a major asset 3.05.3.1.1 Oxides
with regard to the target. Moreover, the use of the Irrespective of the recycling method considered,
UO2 matrix as a support for MAs, the behavior of MAs are transmuted preferably in their dioxide
which is well known as it is currently the fuel of most form of cubic fluorite structure (see Table 3),
industrial reactors, should make the fuel design eas- which, based on the large experience available on
ier, even though its operation in the reactor under UOX and MOX fuel, is known to provide good sta-
very specific conditions raises very relevant questions bility under irradiation (also refer to Chapter 2.02,
(see Section 3.05.4.3). Finally, a correlative advan- Thermodynamic and Thermophysical Properties
tage in introducing MAs in the blankets, the major of the Actinide Oxides).
role of which is producing plutonium, is that of There are no major obstacles to forming a solid
limiting their proliferative nature. solution in a large composition range as the ionic
Table 3 Lattice parameter (a) and theoretical density (d) of actinide bioxides
radii of actinides are very close to each other and the nitride may have to be enriched in 15N and this
15
decrease with the atomic number. MAs are therefore N must be recycled. This results in two added tech-
generally introduced in the fuel in the form of a solid nical constraints which may be insurmountable for a
solution of (U,Np,Pu,Am,Cm)O2 x type, whenever large-scale application from an economical standpoint.
it involves homogeneous recycling, (Am,Cm)O2 x Finally, a partial dissociation of nitride at high temper-
type when it has to do with heterogeneous recycling ature was seen in pile with the appearance of pluto-
on inert support, or (U,Am,Cm)O2 x type whenever nium metal phase;8–10 this risk reduces the safety
it involves heterogeneous recycling in breeder blan- margins of the fuel with respect to accidental situa-
ket. Neptunium is generally not considered in het- tions. The relevance of carbide was therefore recently
erogeneous recycling even though there are no increased for research concerning GFR fuel.11
opposing contraindications to load it: during the sep- In comparison to nitride, carbide offers the
aration treatment, neptunium is easily extracted with advantage of greater thermal stability for slightly
uranium or with plutonium and therefore naturally less favorable physical properties. However, carbide
follows the uranium or plutonium cycle rather than fuel is more difficult to synthesize. The existence
that of americium and/or curium. domain of the monocarbide (U,Pu)C1 x is very
The compatibility of actinide oxides with the narrow (whereas the mononitride accepts a wide
European R&D facilities of course made it easier to stoichiometric deviation phase region) so that the syn-
develop this type of fuel. It has now attained an thesis conditions must be strictly checked to prevent
unequaled level of knowledge, establishing a certain formation of actinide metals in the hypostoichiometric
level of confidence as regards the feasibility of new phase region or formation of actinide sesquicarbide in
concepts proposed which has not yet been established the hyperstoichiometric phase region. The formation
with the alternate solutions described in Section of actinides in metallic form is to be forbidden as their
3.05.3.1.2 (also refer to Chapter 2.21, Fuel Perfor- low melting point would cause their fusion in the
mance of Fast Spectrum Oxide Fuel). reactor as well as the risk of low temperature of inter-
metallic compounds being formed with the compo-
3.05.3.1.2 Alternate compounds nents of the steel cladding. Actinide sesquicarbides
The properties of actinide oxides do not appear are not thermodynamically stable and dissociate, thus
to be necessarily the best ones in comparison with causing an increase of the carbon potential and a risk
those of other compounds such as actinide metal of embrittlement of the steel cladding by carburiza-
alloys or actinide carbide and nitrides, which already tion. Moreover, carbide requires more critical handling
have or had developments in the past as alternative than nitride, given its pyrophoric nature.
to the MOX fuel (see Table 4) (also refer to Finally, the greater thermodynamic stability of
Chapter 3.02, Nitride Fuel and Chapter 3.03, Car- americium nitride (AmN) as opposed to that of amer-
bide Fuel). icium carbide (Am2C3) leads to thinking that the
For example, nitride and carbide fuels were stud- americium-nitride–based solid solutions would be
ied because of their advantages in terms of thermal more stable than the carbide-based ones.
properties and heavy nuclei density. However, one Both of them require adapting manufacturing facil-
drawback of nitride is that it produces carbon 14 by ities for handling powders (inert atmosphere, water
reaction (n,p) on nitrogen 14 under neutron flux. and oxygen-free, carbothermic furnace, controller of
To reduce the 14C release from reprocessing plants, O and C impurities, etc.) as these compounds are
Table 4 Comparison of a few properties of the various types of SFR fuel (Pu/(U þ Pu) ¼ 20%)
a
500 C for UPuZr.
116 Actinide-Bearing Fuels and Transmutation Targets
hygroscopic and reactive to oxygen, especially in the properties, such as for instance thermal conductivity of
form of powder. oxides or temperature stability of nitrides.
Metal alloys of UPuZr type are the reference fuel It is very challenging to choose the matrix as it
of the US fast reactor and form the subject of thor- must comply with especially restrictive specifications:
ough studies abroad as an alternative to oxide fuel.
good thermal properties (high melting point, high
They have the advantage of very high thermal con-
thermal conductivity, and expansion property
ductivity and high thermal creep reducing the risk of
close to that of the actinide compounds); further, it
fuel–cladding mechanical interaction (also refer to
will be seen that these criteria must be reviewed
Chapter 3.01, Metal Fuel).
if operation at sufficiently high temperatures is
However, their low melting point and their poor
desired to promote release of fission gases or helium
chemical compatibility with the austenitic or ferrite-
to reduce macroscopic swelling of composites;
martensitic steel cladding require design accommo-
low activation under the effect of the neutron flux;
dations to maintain the fuel at low temperatures in
resistance to irradiation (low swelling, structural
all situations. One of these involves introducing a
and dimensional stability, isotropic behavior, pres-
sodium metal bond (liquid at operating temperature)
ervation of original properties, etc.);
between the fuel and cladding. They therefore belong
chemical compatibility with the actinide com-
to the family of ‘cold’ fuels and require a completely
pounds, the coolant, and the cladding;
different cycle of technology (pyrochemical pro-
reprocessable if it is multirecycled (dissolution/
cesses) from that of the refractory ceramic fuel called
separation in aqueous medium or molten salts,
‘hot’ fuel developed in Europe and reprocessed by
depending on the targeted recycling strategy) or
hydrometallurgy.12
on the contrary chemically stable if it must confine
Metal fuel has a complex metallurgy with multiple
FPs in case of direct storage after irradiation
phase transitions and uncertain mutual solubility of
(‘once-through’ recycling);
actinides. The volatility of metallic americium is high
integration by a simple and robust manufacturing
(4 times more than that of plutonium). Alloys devel-
process.
oped by the Americans for transmutation fuel therefore
contain much higher zirconium contents than those of Materials were preselected on the basis of these
standard fuel and are produced with ‘flash’ processes criteria14,15 and this is shown in Table 5. The white
reducing the high temperature heat treatment times boxes give the reference matrices for which signifi-
as much as possible to limit volatilization of MAs.13 cant out-of-pile and in-pile characterization work
Moreover, all these types of alternate fuel offer the was carried out, the gray boxes indicate high poten-
advantage of good compatibility with the sodium tial matrices although they do not offer additional
coolant in comparison to oxides which delays the decisive assets with respect to reference matrices,
extraction of a failed fuel assembly from the reactor and the black boxes are the matrices which have
and enables consideration of sodium fuel–cladding been ruled out for points considered as unsurmoun-
bond. This is advantageous in that it promotes the table. These points are also highlighted in black.
thermal exchange between the fuel and the cladding The inventory of high potential matrices was nar-
and keeps the fuel cold or increases its operating rowed down to a few matrices of which the most
power while maintaining acceptable safety margins studied were oxide composites: spinel (MgAl2O4),
with respect to the operating limits of temperature. magnesia (MgO), molybdenum (depleted in isotope
95 to limit its activation and in isotope 98 to avoid the
99
Tc formation), and tungsten; for nitride composites:
3.05.3.2 Inert Matrices
zirconium and titanium nitrides (also refer to
The design of SFR targets or ADS fuels involves Chapter 3.09, Inert Matrix Fuel).
dilution of actinide compounds in an inert matrix to Stabilized zirconia in its cubic form was not initi-
form either a composite (multiphased compound) or a ally considered as a choice matrix insofar as its very
solid solution (one-phase compound). Inert matrix low thermal conductivity does not predispose it as an
refers to a nonfertile or nonfissile material whose ideal candidate. Its excellent behavior under irradia-
neutron absorption is low and which does not lead to tion and the good thermochemical properties of the
formation of active or radioactive elements. Dilution complex compounds formed with actinide oxides,
makes it possible to control the power density released such as the zirconia-based disordered solid solutions,
during transmutation and improve the target thermal for example, (Am,Zr,Y)Ox or ordered solid solutions
Table 5 Selection of inert matrices for composite transmutation fuels and targets
Inert matrices Crystal structure Thermal properties Neutron AnO2 AnN Zry Steel Na H2O
activation
117
118 Actinide-Bearing Fuels and Transmutation Targets
such as pyrochlore Am2Zr2O7, however led to it being Finally, it is of course the standard fuel which
studied more specifically to improve the intrinsic determines the choice for homogeneous transmutation
properties of actinide oxides.16 These compounds fuel: oxide for development in France, Europe, and
are expected to offer on the one hand a wide phase Japan, or metal alloys for US developments. Mixed
region of the fluorine or pyrochlore-type cubic phase nitrides or carbides remain the alternative candidate
existence thus promoting incorporation of Cm, Pu but research has not yet really started in this respect.
elements, and FPs, and on the other hand a lesser
chemical activity limiting the risk of cladding oxida-
tion induced by the high oxygen potential of AmOx . 3.05.4 Behavior of Refractory
Ceramic Fuel Under Irradiation
3.05.3.3 Composition of Fuel and Targets 3.05.4.1 Homogeneous Transmutation Fuel
The typical composition of the various types of trans- Researches on homogeneous transmutation fuel involve
mutation fuel and targets selected is grouped in assessing the possible consequences of the presence of
Table 6. MAs on the physical–chemical properties and on the
Metal alloys and nitrides were developed only behavior of fuel under irradiation (redistribution of
under the scope of fuel for ADS.17,18 The excel- the fissile species within the fuel, helium production,
lent thermal properties of these compounds are a volume swelling, gas release, etc.). The challenge is
significant advantage for these fuels which are highly primarily in ensuring that the low MA contents do
enriched in MAs (6 g (Pu þ MA) per cm3 of fuel) not significantly modify the general behavior of the
and whose performance (linear power) is rapidly fuel in order to be able to benefit from important
limited by the thermal limits of the fuel (margin to feedback of SFR fuel (also refer to Chapter 2.21,
melting). Fuel Performance of Fast Spectrum Oxide Fuel).
With regard to transmutation targets and breeder
blankets containing MAs, the limitation of the linear 3.05.4.1.1 Beginning-of-life behavior
power and the relatively low content of the targets (burnup < 0.5 %FIMA)
(2 g (Pu þ MA) per cm3 of target) did not encour- The properties of MA oxides are not generally as
age exploration of nonoxide fuel as oxides fulfill the favorable as those of UOX or plutonium oxides
set thermal constraints. Their lifetime is especially (MOX) (this is also true for nitrides or metal alloys).
limited by the resistance of the cladding, which is In particular, the introduction of americium and
subject to very long irradiation periods and to high curium in the UOX or MOX fuel leads to a
irradiation doses. Targets are therefore composites decrease of the melting point and of the thermal
made up of actinide pyrochlore or oxide solid solu- conductivity and even more significantly as the MA
tion and ceramic or refractory metal as inert matrix content is high.19,20
and the blankets of mixed uranium and MA solid One of the major points of concern to which the
solutions. researches were directed is the impact of both
Actinide-Bearing Fuels and Transmutation Targets 119
degradations on the behavior of fuel at the beginning and 24 h of a few pellets of different types of fuel
of life when the fuel central temperature is the hot- containing from 3 to 5% of americium and a mixture
test, exceeding 2000 C (for a melting temperature of of 2% americium and 2% neptunium, at a linear power
about 2700 C). This high operation temperature is representing that of standard fuel (420 W cm1).
the conjunction of three factors: A restructuring after the first 10 min was noted, the
extent of which was comparable to that of standard
The poor thermal conductivity of the fuel (see
fuel jointly irradiated. The restructuring was almost
Table 3).
finished after 24 h with a migration of americium
The presence of a gaseous thermal bond between
toward the central hole as in the case of plutonium.
the fuel and the cladding, acting as a thermal
Neptunium remained immobile. No sign of fusion was
barrier. This bond is required to accommodate
detected further to this change in composition.
fuel swelling during irradiation.
The fuel restructuring. The fast diffusion of
3.05.4.1.2 Behavior at moderate combustion
oxygen from the pellet center to the periphery
rate (burnup < 6–7 %FIMA)
reduces the O/M ratio of the fuel in the center
At this stage of irradiation, the main concerns which
and locally affects the thermal conductivity. At the
result from the presence of MAs in the fuel pertain
same time, the oxide is enriched in plutonium at
to the consequences of a high production of helium
the edge of the central hole (preexisting or formed
on the gas release and on the fuel swelling leading to
very early during the irradiation) because uranium
thermal and mechanical fuel pin behavior change and
migrates to the periphery by an evaporation–
to any risk of premature corrosion of the cladding
condensation mechanism caused by the fact that
under the effect of a more oxidizing internal chemical
the vapor pressure of UO2 (the gaseous species
environment than that of a fuel without MAs.
UO3) is higher than that of PuO2 for typical fast
Regarding the first point, the specificity of fuel,
reactor conditions. A consequence of this enrich-
targets, and blankets as long as they contain MAs is
ment in plutonium is a slight decrease of the melt-
the production of a helium quantity which is all the
ing temperature at the center of the pellet.
more significant as the 241Am content is high. This
At this stage of irradiation, a check should be helium production results from a disintegration of
performed to ensure that the fuel-bearing MAs does curium 242, itself formed by capture and successive
not melt or it keeps enough margins with respect to b disintegrations of 241Am (see Figure 4). It con-
melting. tinues out of pile by natural decrease of 242Cm.
This aspect of beginning-of-life behavior recently As helium is much more mobile than xenon and
formed the subject of a thorough experimental study krypton, its total release at the rather high operating
by the JAEA in the Joyo fast reactor.21 The ‘Am1’ temperatures experienced by homogeneous transmu-
program concerned very fast irradiations of 10 min tation fuel can be expected. Helium possibly retained
242m
IRm Am 243 244
241
IT 141 year
Am IRg Am Am
432 year 242
Am
237 EC 16 h
Np
(17.3%)
in the fuel could however contribute to increase the Moreover, the cladding corrodes by chemical reac-
gaseous swelling of the fuel and precipitate the end of tion with certain FPs such as cesium or tellurium
life of the pin by an early mechanical interaction present in the JOG, which forms iron, chromium, or
between the pellet and the clad. nickel compounds with the steel elements in rela-
Regarding the second point, the high tempera- tively high conditions of oxygen potential that are
tures of the fuel promote radial and axial migration experienced from 5 to 6% of burnup. The depth of
of the volatile FPs such as cesium, rubidium, tellu- the cladding affected by the corrosion increases with
rium, selenium, iodine, or bromine. Like fission gases, irradiation and temperature and also depends on the
the volatile FPs leave the center of the restructured cladding deformation rates. The presence of ameri-
zone and are condensed in the colder zones: either at cium oxide whose oxygen potential is higher than
the periphery of the fuel or in the high or low part that of the other actinide oxides results in a change
of the fissile column. At an average combustion rate of the thermodynamic conditions of the fuel in its
(5–7 at.%), the gap between the pellet and the clad- cladding. This modification affects the migration of
ding which was closed because of the fuel swelling volatile FPs and their local accumulation in the JOG,
expands again over a thickness which may reach and a deleterious effect with respect to internal cor-
150 mm at 10 at.%. It is then filled with FP com- rosion of the cladding by chemical reactions can be
pounds of barium, palladium, cadmium, tellurium, expected.
and cesium molybdate type. It is called the Joint The Superfact program22–24 initiated by Institute
Oxyde Gaine ( JOG). Its composition changes with for Transuranium Elements (ITU) and Commissariat
the rate of combustion, given the increase of the fuel à l’Energie Atomique (CEA) in the 1980s provided
oxygen potential. It does not contribute to the dam- encouraging answers as to the behavior of fuel with
age of the cladding, but decreases the diameter of the low MA contents. Four pins representative of SFR
fuel pellets. At higher combustion rates (typically fuel containing either 2% of Np or 2% of Am were
>12 at.%), it is the cladding that suffers the most irradiated in the Phénix reactor up to 6.5 at.% of
profound transformations: it is deformed (by swelling burnup. Even if the operating powers (and therefore
and by irradiation creep, under the effect of internal temperatures) were slightly less than the usual powers
pressure from fission gases), corroded, and is subject of the SFR fuel, given a lower enrichment of the
to increasingly severe stresses. Each of these phe- tested transmutation fuel in Pu (24% against 28%),
nomena may limit the lifetime of the fuel element the comparative postirradiation examinations of the
depending on the grade selected for the cladding. transmutation fuel and the standard fuel performed
The best performance in terms of burnup reaches under optical microscope on a radial cross-section
20 at.%. Nowadays, with optimized grades (cold- in the maximum flux plane showed normal restructur-
worked improved 15–15 Ti austenitic steels (AIM1), ing with differences that can only be explained by
for instance), the permissible doses of the cladding the difference in operating power (Figure 5):
reaches 160 dpa. Nevertheless, it can be admitted
that the future improvements in cladding materials The central hole formed by migration of pores is
will shortly bring this limited dose to 200 dpa. smaller for the transmutation fuel.
1 mm 1 mm 1 mm
Figure 5 Superfact: radial cross-section at the maximum flux plane of the Phénix pins containing mixed uranium plutonium
oxide (MOX) fuel with 2% of neptunium (on the left), 2% of americium (on the right), and standard MOX (in the center)
(20 months of cooling).
Actinide-Bearing Fuels and Transmutation Targets 121
The columnar grains are visible at the edge of the is why SFR fuel rarely exceeds combustion rates
hole over a zone whose width is consistent with the of 20 at.%.
operating temperature. There is still no experiment today making it pos-
The transmutation fuel like the reference closed sible to state that the presence of MAs in the fuel does
the fuel–cladding gap during operation at high not accelerate the end of life of the fuel and the
temperature, as shown by the presence of circum- difficult access or even the unavailability of the fast
ferential cracks. The gap opened up again at room reactors in the world (mainly BOR-60, BN-600,
temperature. Joyo, and Monju) since Phénix shutdown has been
Despite the start of mechanical interaction, no preventing such a demonstration from being per-
significant damage of the cladding was noted. formed before 10 years.
The internal cladding remains sound with corro- It is within this context that the Idaho National
sion thicknesses not exceeding 50 mm. Laboratory (USA) is currently implementing an
additional program to Superfact and Am1 in the
The helium produced during irradiation (200
American thermal reactor (ATR) aimed at attaining
mm3 g1 for transmutation fuel and <50 mm3 g1
high combustion rates of 15–20 at.%.25 Irradiation is
for a standard fuel), which was supposed to be totally
designed to reach conditions of a rapid neutron
released, and helium produced during the 20 months
flux by introduction of a neutron shield able to cut
of cooling at room temperature (of which approxi-
the epithermal section of the flux, which is responsi-
mately half remained within the fuel) did not lead to
ble for the high increase of fuel power density and
any significant swelling.
the acceleration of the helium production kinetics
As regards the fission gases, the differences between
(see Table 1 comparing the efficient cross-section
the transmutation fuel and the standard fuel in the gas
of capture for 241Am in an epithermal spectrum
volumes released (from equal fuel mass) can be
and a fast spectrum). This experiment should par-
explained only by the difference in the combustion
tially provide the missing elements pertaining to the
rate between the fuel pins. The Superfact fuel does
resistance of the cladding to internal corrosion and
not release its fission gases in a manner different from
the behavior of transmutation fuel with high combus-
that of the standard fuel, as shown in Figure 6. Helium
tion rates.
produced by transmutation of americium did not
The international project GACID26 (Global
appear to be the driving effect for increased release
Actinide Cycle International Demonstration), per-
of fission gases.
formed in the scope of the GIF, which is aimed at
demonstrating homogeneous transmutation in the
3.05.4.1.3 End-of-life behavior (15–20 %FIMA) Japanese reactor Monju, at the scale of the assembly
The life of a fuel is limited by the chemical corrosion and for the 2025 horizon should be mentioned to
phenomena of the cladding which reduces its complete the panorama. This demonstration will be
mechanical resistance (see Section 3.05.4.1.2). This preceded by irradiation of one fuel pin in Monju
around 2017.
Finally, the sodium-cooled fast prototype
120 Astrid27 which is planned from 2020 should also
contribute to completing the missing demonstrative
Fission gas release rate (%)
100
elements.
80
Table 7 Characteristics of the various irradiation types for inert matrix fuels and targets
Inert matrices are subject to the intense fast neutron performed with matrix and actinide oxide mixtures
flux required for transmutation. These neutrons pro- in order to test the resistance of the matrix to damage
duce recoil nuclei which will interact with the matrices by actinide FPs. The role of irradiation tempera-
by elastic collisions. Furthermore, they are exposed to ture on possible restoration of irradiation defects
part of the fission fragments and a-particles produced by thermal annealing was also assessed with this
by transmutation of actinides. Table 7 shows the char- type of experiment. UOX, which is much easier
acteristics of the various irradiations undergone by to handle than americium oxide, was often used as
inert matrices during transmutation. actinide compound for these prospective studies.
Most of the energy transmitted to the inert matrix It was only in the last phase that irradiations involved
by irradiation is achieved by the electronic interac- americium oxide in order to study the damage by
tions caused by the fission fragments (close to a-particle. Even though these experiments are much
200 MeV per fission). This is therefore the main heavier and costly to be implemented, they however
cause of nonmetallic material damage when there is remain necessary for testing the effect of high helium
fission. However, it remains limited around the fissile production from 241Am.
atom with a zone of less than about 10 mm of radius This program generated a significant amount of
around this atom. Its effects are consequently largely irradiation experiments more often performed in
dependent on distribution of actinide atoms in the European or international frameworks in experimen-
inert matrix and microstructure of the composite. tal reactors in Europe or Russia.28–30 Table 8 sum-
The a-particles emitted by the a-emitter isotopes, marizes the main irradiations which yielded progress
the flux of which may greatly increase when there in understanding the damage mechanisms of matrices.
are several captures during transmutation and to a
lesser extent the recoil nuclei, may also cause elec- 3.05.4.2.2 Neutron damaging
tronic interactions in the material. The stopping 3.05.4.2.2.1 Effect on matrix swelling
power will however remain very limited (less than a The Matina1 series of experiment in Phénix31,32 pro-
few keV nm1). vides an interesting set of results on neutron irradia-
Therefore, after having selected the relevant inert tion of the relevant inert matrices up to fast fluence
matrices based on their physical–chemical and ther- (E > 0.1 MeV) of 5.9 1026 n m2 and for an esti-
modynamic properties, researches were especially mated operating temperature of around 650 C. It
focused on understanding their damage mechanisms shows that spinel and refractory metals have the
in order to choose the more resistant matrices and best dimensional stability under flux (Figure 7).
elaborate the main behavioral laws and models that This stability which has already been observed33–35
the designer needs to ensure performance and safety is interpreted by a high recombination capacity
of the fuel. It essentially involves macroscopic of the point irradiation defects and formation of
swelling, degradation with irradiation of thermal isolated loops inducing a low macroscopic swelling
properties, and release of gaseous FPs and helium (Figure 8). Moreover, spinel tends to restore these
which directly impact the thermics of the composites defects at relatively low temperatures between 700
and the mechanics of the clad. and 950 C.36
A methodological approach was implemented with In comparison, a segregation of loops was observed
initial study concerning the consequences of neutron for magnesia resulting in an interconnected network
irradiation on matrices alone. Then, experiments were which in the case of vacancy loops will be an efficient
Table 8 Irradiation program for transmutation fuels and targets developed in Europe since the 1980s
Transmutation Experiment name Test Fuel composition Irradiation objectives Experiment status
route reactor
Homogeneous Superfact Phénix (U,Pu,Np)O2x (2% Np), Behavior under irradiation up to 6–7 at.% Completed (1984–1992)
transmutation (U,Pu,Am)O2x (2% Am)
in SFR TRABANT 1 HFR (U,Pu,Np)O2x (5% Np) Behavior under irradiation up to 11–12 at.% Completed (1994–2000)
Am1 Joyo (U,Pu0.30,Am0.05)O1.95–1.98, Beginning-of-life behavior Completed (2001–2008)
(U,Pu0.30,Np0.02,Am0.02)O1.95–1.98
Am1 Joyo Behavior under irradiation up to 5 and 10 at.% Suspended
GACID Monju (U,Pua,Amb)O2x Irradiation at the full pin scale Design in progress
GACID Joyo (U,Pua,Amb,Cmc)O2x Behavior under irradiation up to 10 at.% or Suspended
more
GACID Monju (U,Pua,Amb,Cmc)O2x Irradiation at the full pin scale To be defined
METAPHIX 1 Phénix U–19Pu–10Zr Behavior under irradiation up to 2.4 at.% Completed (2003–2004)
METAPHIX 2 Phénix U–19Pu–1,2Np–0,6Am–0,2Cm– Behavior under irradiation up to 6.6 at.% Completed (2003–2006)
1,4Nd–0,2Y–0,2Ce–0,5Gd–10Zr
U–19Pu–3Np–1,6Am–0,4Cm–10Zr
Continued
123
124
Actinide-Bearing Fuels and Transmutation Targets
Table 8 Continued
Transmutation Experiment name Test Fuel composition Irradiation objectives Experiment status
route reactor
Mgo
1 mm
1 mm 200 mm
MgAl2O4
1 mm 200 mm
1 mm
Al2O3
1 mm 50 mm
Figure 8 Matina1A: macro- and microstructures of inert matrices after 180 EFPD of irradiation in Phénix at about
650 C (optical microscopy).
of microdispersed type was observed, has an excellent temperature close to 750 C up to a fission rate of
resistance in the same conditions of irradiation when 28 at.% experienced very significant swelling of 18%
it is macrodispersed. in volume and 24% for an average transmutation rate
of about 50 at.%. This swelling, which is higher than
3.05.4.2.4 Damage by a-particles that connected with amorphization (see Section
The EFTTRA-T4 and T4bis experiment in the high 3.05.4.2.3), can be partly explained by retention of
flux reactor (HFR) showed the decisive role of 80% of the helium formed (this is 29.4 cm3 (normal
helium on swelling of a composite target.41–43 temperature and pressure) for 1.6 cm3 of fuel) in the
A target made up of a microdispersion of MgAl2O4 target and formation of a network of large-sized
and AmO2 (0.42 g Am per cm3) irradiated at a bubbles which can be recognized in Figure 11
Actinide-Bearing Fuels and Transmutation Targets 127
showing scanning electron microscopy (SEM) and More recently, the Ecrix-H experiment44,45 deliv-
electron probe micro analysis (EPMA) micrographs ered complementary information on the impact of
of the EFTTRA-T4 target. The pore size increased the helium production on a composite made up
and the pore number decreased with increasing dis- of americium oxide microdispersed in magnesia, a
tance of the pore from the pellet surface. matrix resistant to amorphization. With respect to
T4bis, the content of americium was increased up
to 0.7 g cm3. Ecrix-H was irradiated in a moderated
environment (Figure 12) to accelerate transmutation
(environment representative of the ‘once-through’
recycling strategy6) and at a temperature slightly
higher than that of EFTTRA-T4 and T4bis
(890 C). It reached 25% of americium fission rate
and more than 90% of americium transmutation rate.
The hydrostatic measurements and metrologies
of the americium pellets demonstrate a moderate
1 mm de-densification after irradiation limited to 5–7 vol%
for a significant helium proportion retained in the
targets of 72% of the gas produced. This result attests
to a rather satisfactory behavior of this type of target.
Nevertheless, fracturing of the targets (some large
fragments per pellet) can be observed as well as rela-
tively abundant porosity. Large macropores or bubbles
are also seen within the large-sized americium oxide
particles located in the center of the pellets as opposed
to those much less large located at the pellet periphery
(Figure 13). The americium particles contain mainly
plutonium after transmutation.
The high helium production may therefore prove
to be a source of additional swelling. The designs
nowadays take into account this problem by trying
to optimize the microstructure to promote helium
evacuation as it is produced. This requires a micro-
structure with high open porosity, interconnected and
1 mm stable under irradiation.46 The use of a matrix with
lower thermal conductivity such as yttrium-stabilized
Figure 9 Microstructures after irradiation (1.3 at.% of
zirconia to promote thermal diffusion of helium by
burnup) of the microdispersed composite MgAl2O4 þ UO2
irradiated in Siloe at temperatures close to 600 C increasing the operation temperature of the target is
(Thermhet bottom and left) and in Phénix at 1400 C also another design of interest.47 The Helios experi-
(Matina1A on the right). ment in the HFR, completed in February 2010 and
Particle
Damaged matrix
Sound matrix
Figure 10 Damage of inert matrix based on the particle density and the size of the particle. On the left: the particle
density is identical in both representations. On the right: micrography of an irradiated composite showing the damaged zone
of the matrix around agglomerates and also of small particles spreading in the matrix.
128 Actinide-Bearing Fuels and Transmutation Targets
(a) 42 mm 42 mm 42 mm
Steel block
Sodium
Figure 12 Ecrix-H: irradiation device (left) of the experiment designed to accelerate the americium fission rate. The calcium
hydride ring (right) softens the neutron flux.
waiting for postirradiation examination, will provide of composite (i.e., 2 g Am per cm3) were irradiated in
results for both designs.48,49 the Phénix reactor. The irradiation test was com-
The demonstration that inert matrix fuel contain- pleted in March 2009 and is pending examination.
ing microdispersed americium up to 0.7 g cm3 is
able to operate without redhibitory damage up to 3.05.4.2.5 The nonoxide composites of ADS
25% of fission rate, provided that the inert matrix Since 2000, nuclear research institutes have been
selected is resistant to FP damage (as is the case with developing experimental programs to design innova-
MgO44), gives strong hope of reaching the transmu- tive U-free fuel for accelerator-driven transmutors
tation objective targeted for more heavily loaded fuel and perform common irradiation tests to qualify
in MAs: from 2 to 4.5 g transuranics per cm3, respec- their behavior under irradiation.51–53
tively representing targets and fuel of SFR and ADS. In Europe, oxide composites are preferred as
Additional information in that direction will be ADS fuels because of the large know-how on UOX
gained in a short term from the Futurix-FTA experi- and MOX fuel technology gained for 50 years and the
ment50 where pins containing (Pu,Am)O2 þ MgO recent development on transmutation targets for reac-
fuel loaded with up to 2.5 g of heavy metal per cm3 tors of existing technology, as described above.51,54,55
Actinide-Bearing Fuels and Transmutation Targets 129
500 mm
Am-target
Pellet–clad gap
Clad
500 mm
Am-target
Pellet–clad gap
Clad
(a)
20 mm
Surface
Pé é
Pore
(b) 20 mm
Figure 13 (a) Ecrix-H: macroscopic view (optical microscopy) of the pellets in their cladding after irradiation in Phénix.
(b) Ecrix-H: scanning electron microscopy observation of a fissile particle located near the periphery (left), and in the center
(right) of the pellet.
3.00
2.80 Radial flux profile
Reflecte
Core 2.60
2.40
ur
Radial profile
2.20
Pins A B C D E F G H I
2.00
Ring 1 2 3 4 5 6 7 8 9
1.80
1.60
1.40
1.20
1.00
1 2 3 4 5 6 7 8 9
Radius
10% MA
1600 Tc (°C)
1400
1200
1000
1:T_COMB_CENTER
800 2:T_COMB_CENTER
3:T_COMB_CENTER
4:T_COMB_CENTER
600 5:T_COMB_CENTER
6:T_COMB_CENTER
7:T_COMB_CENTER
400 8:T_COMB_CENTER
9:T_COMB_CENTER
10:T_COMB_CENTER
200
0
0 833 1667 2500 3333 4167
Equivalent full power days (EFPD)
Figure 14 Flux profile (upper right) within a blanket assembly depending on the radial position of the pins according to 9
zones (upper left) and evolution of the fuel center temperature versus irradiation time of the pins depending on their axial
position (bottom). The rise in temperature at middle corresponds to the turnover transient.
suffered by a fuel pin depending on its position in the reduces the lifetime of the pin or abnormal corrosion
assembly.7 Figure 14 shows the flux profile (upper significantly reduces the resistance of the cladding.
right) within a blanket assembly located at the core As the neutron flux is lower at the core periphery, the
periphery, depending on the radial position of the pins. MABB irradiation time is very long, 4000 effective
The assembly has been divided into 9 zones (upper full power days, which is close to 10 years of irradia-
left): zone 1 is the nearest from the core and zone 9, the tion in the reactor.
furthest one. The flux is roughly 3 times lower from Even though the behavior of UO2-based fuel is
zone 1 to zone 9. The assembly is consequently sched- well-known, given the significant feedback on PWRs
uled to be turned over at midlife, to bring the less and FRs, MABBs offer two specific features that
irradiated pins closer to the center of the core (in ring differentiate them from standard fuel or from a
9) and on the contrary move the more irradiated ones homogeneous transmutation fuel: the high produc-
further away (in ring 1) in order to homogenize the tion of helium connected with the high quantity of
transmutation rate within the pins. americium (20%) paired with operation at low
This irradiation history results in a significant vari- temperature (500–1500 C).
ation of temperature within a single pin (Figure 14) Little or no experimental data are available for
throughout the irradiation time, ranging between this range of operating temperature and increased
500 and 1500 C. The lifetime of MABB pins is swelling of the UOX can be feared, given probable
presumed limited by cladding damage which cannot retention of the helium and its precipitation in micro-
exceed more than 180–200 dpa for the best grades metric bubbles.
known to date, unless the consequences of strong The only feedback that can be given deals with
interaction between the blanket and the cladding the Superfact experiment in the Phénix reactor,22–24
Actinide-Bearing Fuels and Transmutation Targets 131
50 000
Std C2 et Ti
45 000
07 et
40 000
04 et
35 000
Linear power (W m-1)
30 000
25 000
05 et
20 000
06 et
15 000
10 000
5000
0
0 50 100 150 200 250 300 350 400
Irradiation time (EFPD)
Figure 15 Superfact: power history of the fuel pins containing low (pin 07: 2% Np; pin 04: 2% Am) and high
(pin 05: 45% Np; pin 06: 20% Am þ 20% Np) contents of minor actinides irradiated in Phénix.
annealing of samples from UOX and MOX (even if the concerning the behavior of MABB under irradiation
order of magnitude of the helium quantity produced at and perform an initial selection of the various micro-
the end of irradiation for MABB is approximately two structures. Subsequent irradiations are scheduled in
orders of magnitudes higher than the helium produc- order to gradually improve the representativeness of
tion of UOX or MOX fuel). PWR pellets irradiated at MABB irradiations in MTRs (particularly via use of
different combustion rates and cooled for 10–15 years neutron filters), so as to be able later to test their
release helium at an average temperature of behavior in SFR prototypic conditions.
1200 C.57 These results are consistent with those The goal will be to obtain stable, homogeneous,
obtained in the scope of Superfact irradiation. and well-controlled temperatures under irradiation.
At the lowest temperatures, recent studies on unir- Several irradiation temperatures are targeted: 600,
radiated polycrystalline UO258 reveal a slow and con- 800, 1000, and 1200 C. Two microstructures charac-
tinuous desorption of helium up to 800 C, which terized by different porosity rates and open/closed
becomes very fast above this temperature with a pref- porosity ratios will be tested.
erential release at the grain boundaries. Under an ion In the MTR, the helium production rate asso-
flux, irradiation defects created in the UO2 matrix ciated with transmutation of the 241Am is markedly
precipitate helium into bubbles of a few dozen to a accelerated. Therefore, the goal of the initial analyti-
few hundred nanometers slowing down the apparent cal irradiations will also be to check the biased effect
diffusion of helium atoms which may become mobile expected from this accelerated production on MABB
again at around 800–1000 C when the defects are pellets swelling. This check is required with a view to
restored, to increase or coalesce, resulting in macro- using envelope gaseous release/swelling values to
scopic swelling. Helium release is therefore highly optimize with certain robustness the predesign
dependent on the temperature–irradiation damage MABB assemblies.
pair and possibly also on the O/M ratio, for which Two irradiation tests are currently planned: Marios
significant research work remains to be accomplished and Diamino, respectively scheduled in the HFR and
to control the mechanisms, their kinetics, and the tem- OSIRIS reactors. Marios irradiation started in March
peratures for their appearance. Figure 17 gives a sim- 2011, for end of irradiation in 2012. The targeted
plified view of these release/swelling mechanisms. americium transmutation rate should be 55 at.% for
310 equivalent full power days of irradiation. Marios
3.05.4.3.2 Status of ongoing irradiation irradiation project is carried out in the frame of the
projects Fairfuels European project of the seventh framework
The first irradiations of MABB samples scheduled in program.
MTR are analytic-type experiments implementing a Diamino will be loaded in OSIRIS in May 2012,
suitable design so as to study gas release and fuel for an end of irradiation of the last fuel pins in 2013.
swelling at various temperature levels.59 This selec- Figure 18 shows the design principles of these
tion should make it possible to rapidly obtain data irradiations. Small MABB disks are placed in TZM
CCAM
W Mini-pin of about
TZM
ZrYO2 11 cm long
Detail of the
crucibles
Inconel clad
Figure 18 Marios: irradiation device of the experiment designed to analyze the minor actinide-bearing blanket swelling
at controlled temperature in the High Flux Reactor (Petten).
Isotope Half-lives Activity Neutron emission Thermal power Ingestion dose factor
(years) (1010 Bq g1) (n g1 s1) (W g1) (103 Sv g1)
237
Np 2.14 106 0.00261 0 0 0.028
238
Pu 87.7 63.4 2511 0.568 630
239
Pu 24 390 0.227 0.02 0.002 2.6
240
Pu 6 540 0.843 905.6 0.007 9.8
241
Pu 14.3 382 0 0.003 90
242
Pu 3.87 105 0.0141 1678 0 0.15
241
Am 432 12.7 1.2 0.115 150
242m
Am 152 36.0 150 0.004 410
243
Am 7 380 0.736 0 0.007 8.8
242
Cm 0.446 12 300 2.08 107 122 4300
243
Cm 28.5 170 0 1.685 1300
244
Cm 18.1 299 5892 2.825 1800
245
Cm 8 530 0.633 5448 0.006 7.6
difficulties even though the order of magnitudes are and curium at <80 C, the thickness of the bed must
less than that of curium, and then finally that not exceed 0.5 mm in natural convection. No indus-
of 241Am. trial experiments in the nuclear field to date exist to
In addition to appropriate radioprotection mea- manage such thermal constraints.
sures (concrete cells with thick walls of light In order to limit the technical–economical over-
materials: polyethylene or water to stop the neu- run of remote handling in shielded cells, the compac-
trons), involving remote handling in heavily shielded tion of the facilities, and the simplicity and sturdiness
cells, permanent cooling of tools, vessels, or diverse of the production processes of curium-based fuels
cells is required, from the conversion phase (repro- and targets are absolute necessities. It is fundamental
cessing end) to the assembly handling in the reac- to limit the volume and number of cells, make the
tor, including assembly mounting phase, interim maintenance of the equipment easier, reduce the
storage (also in shielded cells), and assembly volume of contaminated waste, reduce the produc-
transport. tion, control, and inspection time, and design an easy
Through hydrometallurgical treatment of the recycling process for scraps and waste.
spent fuel and at the end of the separation phases, The consequences may be unfavorable for the
the actinide solutions used for oxide conversion performance of the fuel or the target in the reactor.
at the precipitation stage traditionally require acti- The search for improved microstructures (macrodis-
nide concentration globally ranging between 5 and persion, e.g., see Section 3.05.4.2.3) to fulfill ambi-
50 g l1. By taking the example of a concentration of tious transmutation objectives leads to making the
25 g l1, this solution releases up to 2.5 W l1 if it manufacturing procedure more complex by introdu-
contains 241Am, 10 W l1 if it contains a mixture cing a granule formation stage (macromasses), the
of 90% Am þ 10% Cm, and 70 W l1 for pure Cm size, composition, and distribution of which in
(in this latter case, criticality constraints to this con- the inert matrix will have to be controlled. A good
centration are not included). For comparison, at La compromise between performance and easy man-
Hague plant, the hottest vessel containing the con- ufacturing will have to be found. A systematic study
centrates of FP releases 10 W l1. Moreover, if a of the impact of processes on microstructure and
sol–gel process is used at the conversion stage instead performance in pile seems essential in order to
of a precipitation process, the concentration of metals reach it. The cost saving connected with the best
in solution before gelation requires an even higher performance in reactor may compensate at least cost
order of magnitude. The problem is even more overrun of the manufacturing.
difficult in the solid state insofar as heat is less homo- The constraints are especially of radiological ori-
geneous than in the liquid phase and therefore gin for americium. Remote handling in metallic stain-
harder to release. It is calculated that to maintain a less steel or lead cells is enough. Furthermore, at
powder bed of the previous mixture of americium certain stages of the process (quality control on
Actinide-Bearing Fuels and Transmutation Targets 135
small volume sampling, for example), clamp boxes accumulate in the filters and on the walls of glove
can be used. The use of complex processes, in this boxes or cells which is a real radiological problem,
case, allowing fuel with improved microstructures to given the characteristics of handled isotopes and
be obtained, seems more accessible even though the the risk of neutron emission by reaction (a,n) of
technological breakthrough required by the use of americium on the light elements (of the inert matrix,
remote handling is still needed. However, 238Pu, for example).
which may be found in large quantities in the ADS The possibility of eliminating all these stages
fuel after several joint recycling stages of plutonium would be a major asset at the industrial scale.
and americium, may lead to constraints which may This is the goal of the coprecipitation (Section
eventually require reviewing the process as for 3.05.5.2.2) or sol–gel (Section 3.05.5.2.3) synthesis
curium fuel. processes which should obtain the mixture of con-
Finally, for neptunium, the conventional handling stituents directly at the required composition and
conditions of (U,Pu)O2 fuel (glove boxes) would be with a controlled quality of the grains or granulates
enough for handling recently purified neptunium. (size distribution and solidity), thus eliminating the
However, old neptunium requires reinforced biological milling and granulation stage. The development of
protection given the presence of protactinium, a the Vipac/Spherepac process consisting in the
radioactive neptunium decay product. As the powder direct filling of the fuel pin with granulates (Section
metallurgy processes are inevitably accompanied by 3.05.5.2.4) should push optimization of the man-
the retention of fine powders, which are difficult to ufacturing process even further by eliminating
eliminate and therefore destined to age within the pressing (also refer to Chapter 3.11, Sphere-Pac
systems, radioprotection very quickly will become and VIPAC Fuel). Each eliminated stage would also
restrictive. provide its set of reductions as regards the control
operations.
Another constraint connected with the powder
3.05.5.2 Manufacturing Processes
metallurgy relates to thermolysis and radiolysis of
3.05.5.2.1 Metallurgical processes the organic compounds such as binders or plasticizers
The powder metallurgy processes such as those used intended to increase the mechanical properties of
industrially to produce MOX fuel of PWRs, are green pellets, or pore formers used to check the size
well-adapted to production of a large variety of trans- and shape of porosities and adjust the fuel density.
mutation targets and fuel in laboratory. These are Even when used in very low quantities within the
flexible and universal processes which fulfill the qual- powders, the quick degradation of these products
ity requirements connected with performance in the submitted to thermal and radiological effects leads
reactor. They have proven their capacity to imple- to unacceptable volume of scraps because of low
ment fabricated microstructures (solid solutions, density, cracking, or even fracturing of the products
micro- or macrodispersed composites based on inert after sintering.
matrix). The composites (UO2 þ MgAl2O4, UO2 þ
MgO, AmO2 þ MgO, UO2 þ AmO2) elaborated for 3.05.5.2.2 Coprecipitation processes
the Matina1, Ecrix, Marios, and Diamino experi- The precipitation or coprecipitation process is the
ments (see Sections 3.05.4.2 and 3.05.5.2) are exam- standard method for obtaining a solid phase from
ples of this application. actinides in solution, as the latter traditionally stem
For MA-based nitrides and carbides, the volatility from extraction and purification operations during
of the americium nitride and carbide at temperatures processing of spent fuel (by presuming a treatment
less than the usual sintering treatment temperature, by hydrometallurgical process). Moreover, it should
however, poses a problem which has not yet been be noted that even in the case of fuel or target
resolved. manufacturing by dry process (powder metallurgy),
Nonetheless, the major drawback which compro- this stage of precipitation completed by a heat treat-
mises use of metallurgical processes for industrial ment would be necessary to stabilize and temporarily
production of americium targets and fuel is firstly store in solid form (conditioning) the various MAs
the difficulty in managing fine powders (dust) at all resulting from processing of spent fuel (also refer to
stages of the process (dosing, milling, mixing, granu- Chapter 5.14, Spent Fuel Dissolution and Repro-
lation, grinding, sieving, press filling, etc.). They cessing Processes).
136 Actinide-Bearing Fuels and Transmutation Targets
Even though it may seem simple to implement conversion of plutonium at the industrial stage is a
this precipitation phase (it is a mixture of the sure asset for later development of this process.
reagents), its control requires taking into account Moreover, these routes very generally result in pow-
all the physical–chemical and chemical engineering ders and may therefore require additional shaping
parameters. The main properties affected by these operations (agglomeration, pressing, etc.). The pur-
parameters are the chemical composition of the suit of R&D concerning interesting coprecipitation
precipitate and its physical characteristics (flowabil- processes would present the interest of broadening
ity, grain size distribution, filterability, microstruc- the shaping perspectives of the solid (particle size
ture, etc.), the reproducibility of obtaining these increase, flowability improvement, and agglomera-
characteristics, and the concentration of actinides tion by wet process) to reduce subsequent fuel
in the effluents. An approach aimed at attaining this manufacturing stages, however, within the limits
control has been under study for 10 years in the inherent to these routes leading to divided solid.
framework of oxalic conversion of plutonium and is
transposed at each stage of the progress to the copre- 3.05.5.2.3 Sol–gel processes
cipitation of several actinides and possibly an inert In the 1980s–1990s, several sol–gel processes were
element (Figure 20). An application of this co- tested for elaboration of actinide materials (mixed
conversion process is the synthesis of the mixed oxides, carbides, or nitrides). The main advantage of
(Pu, Am)O2 for the Futurix-FTA experiment in sol–gel processes involve the easy shaping during the
Phénix61 (see Section 3.05.4.2). This approach gelification stage, thanks to the fluidity of the initial
should help to obtain more systematically a solid solution. In particular, many works describe con-
compound with targeted and original characteristics trolled fabrication of dense or porous microspheres
(instead of nonchosen characteristics as it is often the depending on their subsequent use such as for
case today), especially as regards the grain size distri- instance Spherepac or pelleted fuel. The fuel from
bution or microstructure. This compound could be a the Superfact experiment was entirely produced by
direct precursor of the fuel thus simplifying its ITU with a process of this type.24
manufacturing process, for example, by eliminating Apart from providing the main advantage of
upstream stages of mixing and comilling oxide shaping the synthesized solid compound, the co-
powders of each actinide. gelification process also makes it possible to generally
However, the matter of managing high activity obtain homogeneous compounds through a perfect
effluents generated by certain coprecipitation routes initial mixture of elements in solution and therefore
(namely ammoniacal or carbonate precipitation) promotes the formation of solid solutions during
has still not been resolved. As regards oxalic co- a later heat treatment phase. Once optimized, this
conversion, the feedback stemming from oxalic technique allows spherical particles of homogeneous
Figure 20 Oxalic coconversion of An(IV)-An(III) solutions in oxide: coprecipitation is carried out in a vortex effect
reactor (left); different morphologies of powder obtained depending on the process parameters (right).
Actinide-Bearing Fuels and Transmutation Targets 137
Scraps
Dosing Pellets
primary blend inspection
Scraps Bead size
Scraps
selection
Mixing
Grinding
Primary blend
Scraps Pin
Dosing fabrication
final blend
Sintering
Scraps
Homogenization
Pin control
Final blend Scraps (Pu and density linear
Pressing distribution)
Conformed
pellets
(U,Pu)O2
Pin
fabrication
to confirm this good performance for actinide con- 3. Warin, D.; Zaetta, A.; Varaine, F.; Grouiller, J. P.; Pillon, S.
In Proceedings of PHYSOR 2006, Vancouver, BC,
tents and transmutation rates representative for the Sept 10–14, 2006.
heterogeneous concept (up to 1.5 g cm3 and 90% of 4. Bouchard, J. In Proceedings of the GIF Symposium, Paris,
fission for monorecycling, 2.5 g cm3 and 20–30 at.% France, Sept 9–10, 2009, pp 11–13.
5. Knebel, J. In OECD/NEA First Workshop on Technology
of fission for multirecycling, and more than 4 g cm3 and Components of ADS (TCADS-1), Karlsruhe, Germany,
and 20–30 at.% of fission for ADS). Mar 15–17, 2010, p 43.
As regards heterogeneous transmutation in breeder 6. De Saint Jean, C.; Tommasi, J.; Varaine, F. In Proceedings
of the International Conference Global 2001 on Back-End
blankets, the program is still at an early stage to of the Fuel Cycle: From Research to Solutions, Paris,
determine, with certainty, the feasibility of this new France, Sept 9–13, 2001; paper n 141.
concept. However, positive feedback concerning inert 7. Valentin, B.; Palancher, H.; Yver, C.; Garat, V.;
Massara, S. In Proceedings of the International
matrix targets operating in similar temperature ranges Conference Global 2009 on The Nuclear Fuel Cycle:
leads to the expectation that it should be possible to Sustainable Options & Industrial Perspectives, Paris,
control their swelling. France, Sept 6–11, 2009, pp 2095–2103.
8. Fromont, M.; Lamontagne, J.; Asou, M.; Aubrun, I. In
For nitride or metallic fuels of ADS, the feedback Proceedings of the International Global 2005 Conference
is still insufficient to reliably assess their performance. on Nuclear Energy System for Future Generation and
In general, the demonstration of technical feasi- Global Sustainability, Tsukuba, Japan, Oct 9–13, 2005;
paper n 329.
bility of MA targets and fuel is confronted at the 9. Blank, H. Euratom Report 13220 EN; 1991.
manufacturing level with handling difficulties for 10. Inoue, M.; Iwai, T.; Arai, Y. In Proceedings of the
isotopes, which are high neutron and thermal emit- International Global 2003 Conference on Atoms for
Prosperity: Updating Eisenhower’s Global Vision for
ters (244Cm and 238Pu). Today very few laboratories Nuclear Energy, New Orleans, LA, Nov 16–20, 2003,
handle this type of isotopes in significant quantities pp 1694–1703.
and research still has to be carried out to optimize 11. Chauvin, N.; Malo, J. Y.; Garnier, J. C.; et al. In
Proceedings of the International Conference Global 2007
the adapted manufacturing processes. Wet processes on Advanced Nuclear Fuel, Cycles and Systems, Boise, ID,
such as coprecipitation, sol–gel, and INRAM offer Sept 9–13, 2007, pp 423–433.
the interesting perspective of simplifying the fabri- 12. Carmack, W. J.; Porter, D. L.; Chang, Y. L.; et al. J. Nucl.
Mater. 2009, 392, 139–150.
cation and making it robust. They have been devel- 13. Meyer, M. K.; Hayes, S. L.; Carmack, W. J.; Tsai, H.
oped successfully for production of different J. Nucl. Mater. 2009, 392, 176–183.
americium-bearing fuels for the irradiation experi- 14. Cocuaud, N.; Picard, E.; Konings, R. J. M.; Conti, A.;
Matzke, H. J. In Proceedings of the Global 1997
ment needs, which makes it possible to consider International Conference on Challenge Towards Second
extending their application to curium-based fuel, Nuclear Era with Advanced Fuel Cycles, Yokohama,
however, not without significant research intended Japan, Oct 5–10, 1997, pp 1044–1049.
15. Kleykamp, H. J. Nucl. Mater. 1999, 275, 1–11.
to control chemical engineering and management of 16. Raison, P. E.; Haire, R. G. J. Nucl. Mater. 2003, 230,
waste and effluents. 31–35.
17. Hayes, S. L.; Hilton, B.; McLean, H.; Utterbeck, D.;
Carmack, J.; Pasamehmetoglu, K. In Proceedings of the
International Conference Global 2009 on The Nuclear Fuel
Acknowledgments Cycle: Sustainable Options & Industrial Perspectives,
Paris, France, Sept 6–11, 2009, pp 1997–2005.
The author would like to thank for their informed 18. Wallenius, J. Final Technical Report of the CONFIRM
Project; 2009.
proofreading Mrs Fabienne Delage, Mr Jean-Marc 19. Schmidt, H. E.; Sari, C.; Richter, K. J. Less Common Met.
Bonnerot, Mr Syriac Bejaoui, and Mr Stephane 1985, 121, 621–630.
Grandjean from the Commissariat à l’Energie Ato- 20. Morimoto, K.; Kato, M.; Ogasawara, M.; Kashimura, M.;
Abe, T. J. Alloys Compd. 2008, 452, 54.
mique et aux Energies Alternatives. 21. Maeda, K.; Sasaki, S.; Kato, M.; Kihara, Y. J. Nucl. Mater.
2009, 385, 413–418.
22. Walker, C. T.; Nicolaou, G. J. Nucl. Mater. 1995, 218,
References 129–138.
23. Prunier, C.; Boussard, F.; Koch, L.; Coquerelle, M. Nucl.
Technol. 1997, 119, 141–148.
1. Courtois, C.; et al. High Level Radioactive Waste/Research 24. Babelot, J. F.; Chauvin, N. Euratom Report; 1999.
and Results, Law of 30 December, 1991/Direction 1; CEA 25. Maclean, H. J.; Hayes, S. L. In Proceedings of the
Report DEN/DDIN/2005-568; 2005. International Conference Global 2007 on Advanced Nuclear
2. Warin, D. In Proceedings of the International Global 2005 Fuel, Cycles and Systems, Boise, ID, Sept 9–13, 2007,
Conference on Nuclear Energy System for Future pp 1341–1345.
Generation and Global Sustainability, Tsukuba, Japan, 26. Nakashima, F.; Tomoyasu, M.; Brunel, L.; Pillon, S.;
Oct 9–13, 2005; paper n 271. Pasamehmetoglu, K. O.; Carmack, J. In Proceedings of
Actinide-Bearing Fuels and Transmutation Targets 141
the GIF Symposium 2009, Paris, France, Sept 9–10, 2009, Cycle: Sustainable Options & Industrial Perspectives,
pp 239–246. Paris, France, Sept 6–11, 2009, pp 2006–2015.
27. Verwaedte, D.; Serpantié, J. P.; Rouault, J. In Proceedings 49. D’Agata, E.; Lapetite, J. M.; Klaasen, F.; et al. Prog. Nucl.
of the International Fast Reactors and Related Fuel Cycles Energy 2011, 53, 748–755.
FR’09 Conference, Kyoto, Japan, Dec 7–11, 2009; 50. Jaecki, P.; Pillon, S.; Warin, D.; et al. In Proceedings of the
IAEA-CN-176/INV-04; paper n FRP-02. International Global 2005 Conference on Nuclear Energy
28. Pillon, S.; Sudreau, F.; Gaillard-Groleas, G. Nucl. Technol. System for Future Generation and Global Sustainability,
2006, 153, 264–273. Tsukuba, Japan, Oct 9–13, 2005; paper n 448.
29. Warin, D.; Conrad, R.; Haas, D.; et al. In Proceedings of the 51. Delage, F.; Belin, R.; Chen, X.; et al. In OECD/NEA First
International Conference Global 2001 on Back-End of Workshop on Technology and Components of ADS
the Fuel Cycle: From Research to Solutions, Paris, France, (TCADS-1), Karlsruhe, Germany, Mar 15–17, 2010,
Sept 9–13, 2001; paper n 204. pp 69–79.
30. Pillon, S.; Zaboudko, L.; Kurina, I.; et al. In Proceedings of 52. Wallenius, J. In CONFIRM: Collaboration on Nitride Fuel
the International Conference Global 2001 on Back-End of Irradiation and Modelling, AccApp/ADTTA’01, 2001 ANS
the Fuel Cycle: From Research to Solutions, Paris, France, Winter Meeting, Reno, NV, Nov 11–15, 2001.
Sept 9–13, 2001; paper n 292. 53. Hilton, B. A.; Carmack, W. J.; Voit, S. L.; Hayes, S. L.;
31. Chauvin, N.; Albiol, T.; Mazoyer, R.; et al. J. Nucl. Mater. Pasamehmetoglu, K. O. In Proceedings of the
1999, 274, 91–97, pp 15–24. International Conference Global 2007 on Advanced
32. Bonnerot, J. M.; Warin, D.; Ferroud-Plattet, M. P.; Nuclear Fuel, Cycles and Systems, Boise, ID, Sept 9–13,
Gosmain, L.; Lamontagne, J. In Proceedings of the 2007, pp 1688–1691.
International Conference Global 2007 on Advanced Nuclear 54. Pillon, S.; Wallenius, J. Nucl. Sci. Eng. 2006, 153, 1–8.
Fuel, Cycles and Systems, Boise, ID, Sept 9–13, 2007. 55. Maschek, W.; Artioli, C.; Chen, X. N.; et al. In Proceedings
33. Garner, F. A.; Hollenberg, G. W.; Hobbs, F. D.; et al. of the Tenth Information Exchange Meeting on Actinide
J. Nucl. Mater. 1994, 212–215, 1087–1090. and Fission Product Partitioning and Transmutation, Mito,
34. Clinard, F. W.; Hurley, G. F.; Hobbs, L. W.; Rohr, D. L.; Japan, Oct 6–10, 2008.
Youngman, R. A. J. Nucl. Mater. 1984, 1–3, 1386–1392. 56. Arai, Y.; Minato, K. J. Nucl. Mater. 2005, 344, 180–185.
35. Konings, R. J. M.; Bakker, K.; Boshoven, J. G.; Conrad, R.; 57. Ronchi, C.; Hiernaut, J. P. J. Nucl. Mater. 2004, 325, 1–12.
Hein, H. J. Nucl. Mater. 1998, 254, 135–142. 58. Martin, G.; Desgardin, P.; Garcia, P.; Sauvage, T.;
36. Staicu, D.; Bonnerot, J. M.; Konings, R. J. M.; Gosmain, L. Carlot, G.; Barthe, M. F. In Proceedings of the Materials
In Proceedings of the International Conference Research Society Symposium on Scientific Basis for
Global 2009 on The Nuclear Fuel Cycle: Sustainable Nuclear Waste Management; Dunn, D., Poinssot, C.,
Options & Industrial Perspectives, Paris, France, Sept Begg, B., Eds.; Mater. Res. Soc. Symp. Proc.; Materials
6–11, 2009, pp 2016–2022. Research Society: Warrendale, PA, 2007; Vol. 985.
37. Yano, T.; Ichikawa, K.; Akiyoshi, M.; Tachi, Y. J. Nucl. 59. Bonnerot, J. M.; Pillon, S.; Bejaoui, S.; et al. In
Mater. 2000, 283–287, 947–951. Proceedings of the Eleventh Information Exchange
38. Noirot, J.; Desgranges, L.; Chauvin, N.; Georgenthum, V. Meeting on Actinide and Fission Product Partitioning and
J. Nucl. Mater. 2003, 320, 117–125. Transmutation, San Francisco, CA, Nov 1–5, 2010.
39. Neeft, E. A. C.; Bakker, K.; Schram, R. P. C.; Conrad, R.; 60. Pillon, S.; Somers, J.; Grandjean, S.; Lacquement, J.
Konings, R. J. M. J. Nucl. Mater. 2003, 320, 106–116. J. Nucl. Mater. 2003, 320, 36–43.
40. Dehaudt, P.; Mocellin, A.; Eminet, G.; et al. In IAEA 61. Grandjean, S.; Arab-Chapelet, B.; Robisson, A. C.; et al.
Technical Committee Meeting, Moscou, Russia, Oct 1–4, J. Nucl. Mater. 2009, 385, 204–207.
1996; IAEA-TECDOC-970. 62. Grandjean, S.; Arab-Chapelet, B.; Robisson, A. C.; et al.
41. Konings, R. J. M.; Conrad, R.; Dassel, G.; Pijlgroms, B. J.; In Proceedings of the International Conference Global
Somers, J.; Toscano, E. J. Nucl. Mater. 2000, 282, 2009 on The Nuclear Fuel Cycle: Sustainable Options &
159–170. Industrial Perspectives, Paris, France, Sept 6–11, 2009,
42. Wiss, T. A. G.; Konings, R. J. M.; Walker, C. T.; Thiele, H. pp 1309–1316.
J. Nucl. Mater. 2003, 320, 85–95. 63. Bychkov, A.; Kisly, A.; Shishalov, O.; et al. In Proceedings
43. Wiss, T. A. G.; Damen, P. M. G.; Hiernaut, C.; Ronchi, C. of the International GLOBAL 2005 Conference on Nuclear
J. Nucl. Mater. 2004, 334, 47–57. Energy System for Future Generation and Global
44. Bejaoui, S.; Lamontagne, J.; Bonnerot, J. M.; Brunon, E. In Sustainability, Tsukuba, Japan, Oct 9–13, 2005;
Proceedings of the International Conference Global 2009 paper n 583.
on The Nuclear Fuel Cycle: Sustainable Options & 64. Bart, G.; Bakker, K.; Hellwig, C.; et al. J. Nucl. Sci.
Industrial Perspectives, Paris, France, Sept 6–11, 2009, Technol. 2007, 44, pp 329–336; n 3.
pp 2104–2115. 65. Richter, K.; Fernandez, A.; Somers, J. J. Nucl. Mater.
45. Béjaoui, S.; Lamontagne, J.; Esbelin, E.; et al. In 1997, 249, 121–127.
Proceedings of the Eleventh Information Exchange
Meeting on Actinide and Fission Product Partitioning and
Transmutation, San Francisco, CA, Nov 1–5 2010.
46. Desmoulière, F.; Léorier, C.; Donnet, L.; Vaudez, S.; Relevant Websites
Bonnerot, J. M. In Proceedings of the Atalante 2008
International Conference, Montpellier, France, May 19–23,
2008. http://nuklear-server.fzk.de/eurotrans/ – European Integrated
47. Degueldre, C. J. Alloys Compd. 2007, 444–445, 36–41. Project EUROTRANS.
48. D’Agata, E.; Klaasen, F.; Sciolla, C.; Frenandez- http://www.neutron.kth.se/research/activities/ads-design/ –
Carretero, A.; Bonnerot, J. M. In Proceedings of the European Project CONFIRM.
International Conference Global 2009 on The Nuclear Fuel http://www.fp7-fairfuels.eu/ – European Project FAIRFUELS.
3.06 TRISO Fuel Production
K. Sawa
Japan Atomic Energy Agency, O-arai, Ibaraki, Japan
143
144 TRISO Fuel Production
External Internal
Buffer layer
NH4OH Aging
Drying H2O
Figure 1 TRISO-coated particle.
H2O, NH3
Calcining
CO2
H2
3.06.2 Fabrication Processes Reduction Gases
where Q is the flow rate of metal solution, D is the integrity of the coated fuel particle depends on
diameter of the droplets, and f is the frequency of the these different coating layers and the failure of
vibrating nozzles. these layers is directly connected to the safety of
Most of the degradation of kernel sphericity is the reactor. The coating technology of these layers
caused by the deformation at the stage of droplet involves a fluidization of the kernel microsphere bed
formation and wet-ADU particle. At the stage of and a well-known technique of chemical vapor depo-
formation, a process is applied to prevent the defor- sition (CVD) of coating materials.2
mation of droplets while they are landing on the Figure 34 shows the fabrication flow diagram of
ammonia water. In the process, droplets are solidified coated fuel particles. Coating layers are deposited on
while falling in ammonia gas blown against them. At the kernels in a CVD process using a fluidized bed
the stage of wet-ADU particles, a process is applied type of coater. The TRISO coating process is divided
in which aging, washing, and drying are carried out in into four coating processes for the porous PyC, IPyC,
the same conical dryer. It is possible to mitigate the SiC, and final PyC (outer PyC; OPyC) layers. A mix
impact of the particles by retaining very soft wet- of the gases of acetylene (C2H2) and argon (Ar) is used
ADU particles in the same conical dryer during the for the deposition of porous and low-density PyC for
operations, and thus prevent deformation. the first layer; propylene (C3H6) and Ar for the depo-
sition of dense PyC for the second and fourth layer;
and methyl-tri-chloro-silane (MTS) and hydrogen
3.06.2.2 PyC- and SiC-Coating Processes5,6
for the deposition of SiC for the third layer. After
The coating technologies for PyC- and SiC-coating the first and second layer coating, test specimens for
layers are the most crucial and delicate as the density measurement, etc. are generally sampled.4
Window
UO2 kernels
Gas outlet
Graphite heater
Cooling
water
Third layer coating
(Silicon carbide)
CH3SiCl3 + H2 Gas inlet nozzle Particle discharge can
Coater
Figure 3 PyC- and SiC-coating processes. Reproduced from Kato, S.; Yoshimuta, S.; Hasumi, T.; et al. Fabrication of HTTR
first loading fuel, IAEA-TECDOC-1210, 2001; pp 187–199.
146 TRISO Fuel Production
The amount of charged UO2 kernels corresponds to mechanical shocks to the particles given by violent
3 kg of uranium per coating batch in the case of the fuel particle fluidization in the coater and by the unload-
for the Japanese high-temperature-engineering test ing procedure of the particles. The coating process
reactor (HTTR). At a desired temperature, reactants was improved by optimizing the mode of the particle
are put into the coater to produce a coating layer on the fluidization and by developing the process without
particles fluidized in the coater. After achieving the unloading and loading of the particles at the interme-
desired thickness of the layer, the reactant gas supply diate coating process.
is replaced by argon. The coater and the coated parti-
cles are cooled down, and then the coated particles are
3.06.2.3 Manufacturing Process of Fuel
removed from the coater. All the coated particles
Element
are classified by means of a vibrating table to exclude
the odd-shaped particles in the case of the HTTR fuel. 3.06.2.3.1 The case of the block type
The as-manufactured quality of the fuel has been (example of the HTTR fuel)
improved by the modification of fabrication condi- Figure 44 shows the manufacturing process of
tions and processes.7 The coating failure during the fuel compacts. First, natural graphite powder,
coating process is caused mainly by the strong electro-graphite powder, and a binder are mixed,
Mixing
Overcoating
Preheating
Drying
Weighing
Pressing
Green compact
Numbering Pressing
Carobonization A1234
Green
Compact Discharge
Degassing
Fuel compact
Figure 4 Manufacturing process of fuel compacts. Kato, S.; Yoshimuta, S.; Hasumi, T.; et al. Fabrication of HTTR first
loading fuel, IAEA-TECDOC-1210, 2001; pp 187–199.
TRISO Fuel Production 147
and then the mixture becomes graphite matrix after a of a fuel sphere. Then, an additional 5-mm layer of
fine grinding process. Coated fuel particles are over- matrix graphite material is added to form a ‘nonfuel’
coated with the graphite matrix and warm-pressed to zone. The resulting fuel sphere achieves its final
make the annular cylinder of green compacts. Green diameter by a machining process and is then carbo-
compacts are preliminarily heat treated for carboni- nized and annealed at 2000 C. The finished fuel
zation at 800 oC under nitrogen atmosphere and then spheres are subject to numerous quality control
sintered at 1800 oC under vacuum to make fuel com- steps, among which the 100% X-ray controls the
pacts.8 Among these processes, the green compact centricity of the fuel core and the freeness of uranium
pressing is the most complicated, and involves weigh- of the shell. It is also very important to check the free
ing, preheating, loading and unloading of overcoated uranium in the fuel sphere, which would result from
particles, printing an identification number on a green SiC shell breakage during fabrication.
compact, and so on.
In order to decrease the failure fraction of coating
layers at the stage of a fuel compact, it is necessary to 3.06.3 Inspection
disperse coated fuel particles in a green compact as
3.06.3.1 The Inspected Parameters
uniformly as possible. From this point of view, the
punching speed and temperature during the warm- Inspections are carried out to determine whether the
pressing were optimized to fix the best timing of the fuel elements conform to specifications that certify
softening by heating and the plastic flow of matrix nuclear and thermal-hydraulic design, irradiation per-
graphite by pressing.7 formance, and so on. From the viewpoint of purpose,
the inspections are divided into three categories:
3.06.2.3.2 The pebble bed type9 (1) compulsory, (2) user’s requirement or optional, and
The fuel element of the pebble bed type HTGR is a (3) vendor’s quality control.10 The sampling rate is also
spherical fuel element shown in Figure 5. In the fuel determined by considering the uniformity of the in-
sphere production facility, the coated particles are spected data. The three categories are basically classi-
overcoated with a layer of matrix graphite powder fied as (a) small-scattering data, (b) medium-scattering
(MGP). Again quality assurance steps are applied as data, and (c) large-scattering data. One sample is mea-
with the kernel and coated particle. The overcoated sured from an inspection lot with small-scattering data.
particles are dosed into MGP and pressed to the core For the inspection lot with medium-scattering
data, three samples are measured, and all of them should
satisfy the required criterion. For the large-scattering
data, measured data should meet a statistically required
Graphite matrix
criterion with 95% confidence. The inspection item,
purpose, method, and sampling rate in the HTTR fuel
fabrication are summarized in Table 1.10
Coated particles
Table 1 The inspection item, purpose, method, and sampling rate in the HTTR fuel fabrication
Fuel kernel
235
U enrichment B Mass spectrometer analysis and g-ray 1 sample/enrichment
spectrometer analysis
Diameter B Optical particle size analysis 1 sample (100 particles)/fuel
kernel lot
Sphericity A Optical particle size analysis 3 samples (100 particles/sample)/
fuel kernel lot
Density B Mercury substitution 3 samples/fuel kernel lot
O/U ratio A Oxidation and weighing 1 sample/fuel kernel lot
Impurities A, B Emission spectrometer analysis 1 sample/enrichment
Coated fuel particle (CFP)
Layer thickness A Solvent substitution or sink–float 3 samples/CFP lot
Optical anisotropy A Polarization photometer 1 sample (5 particles/sample)/
factor enrichment
Diameter B Optical particle size analysis 1 sample (100 particles)/CFP lot
Appearance A Visual observation 1 sample (2000 particles)/CFP lot
Cross-section A Ceramography 1 sample (20 particles) /CFP lot
Sphericity A Selection by vibration table All CFPs
Strength A Point crushing 30 particles/enrichment
Fuel compact
235
U enrichment D Mass spectrometer analysis and gamma-ray 1 sample/enrichment
spectrometer analysis
U content B g-ray spectrometer analysis All fuel compacts
O/U ratio A Oxidation and weighing 1 sample/fuel compact lot
Graphite powder A Density, impurities, grain size, and water 1 sample/graphite powder lot
content
Binder A Contents, ash, melting point, and impurities 1 sample/binder lot
Exposed uranium A Deconsolidation and acid leaching 2 samples/fuel compact lot
fraction
SiC-failure fraction A Burn and acid leaching 3 samples/fuel compact lot
Packing fraction B Weighing and calculation 3 samples/fuel compact lot
Matrix density A Weighing and calculation 3 samples/fuel compact lot
Dimensions C Micrometer All fuel compacts
Appearance A Visual observation All fuel compacts
Marking D Visual observation All fuel compacts
Strength A Compression 3 samples/enrichment
Cross-section A Ceramography 1 sample/fuel compact lot
Impurities B Emission spectrometer analysis 1 sample/enrichment
References 6. Fukuda, K.; Ogawa, T.; Hayashi, K.; et al. J. Nucl. Sci.
Technol. 1991, 28, 570–581.
7. Ikawa, K.; Iwamoto, K. Density determination of the
1. Petti, D. A.; Buongiorno, J.; Maki, J. T.; Hobbins, R. R.; coating layers of fuel particles. JAERI-M-5674, 1974
Miller, G. K. Nucl. Eng. Des. 2003, 222, 281–297. (in Japanese).
2. Lee, Y. W.; Park, J. Y.; Kim, Y. K.; et al. Nucl. Eng. Des. 8. Kobayashi, F.; Shiozawa, S.; Hayashi, K.; et al. An
2008, 238, 2842–2853. inspection standard of fuel for the high temperature
3. Cordfunke, E. H. P. J. Inorg. Nucl. Chem. 1962, 24, engineering test reactor. JAERI-M-92–079, 1992
303–307. (in Japanese).
4. Kato, S.; Yoshimuta, S.; Hasumi, T.; et al. Fabrication 9. Fuel performance and fission product behavior in gas
of HTTR first loading fuel, IAEA-TECDOC-1210, 2001; cooled reactor. IAEA-TECDOC-978, 1997.
pp 187–199. 10. Sawa, K.; Suzuki, S.; Shiozawa, S. Nucl. Eng. Des. 2001,
5. Sawa, K.; et al. J. Nucl. Sci. Technol. 1999, 36, 683–690. 208, 305–313.
3.07 TRISO-Coated Particle Fuel Performance
D. A. Petti, P. A. Demkowicz, and J. T. Maki
Idaho National Laboratory, Idaho Falls, ID, USA
R. R. Hobbins
RRH Consulting, Wilson, WY, USA
151
152 TRISO-Coated Particle Fuel Performance
Pyrolytic carbon
Silicon carbide
Uranium dioxide or oxycarbide kernel
Prismatic
Matrix Kernel
Buffer layer 5 mm graphite layer
Figure 1 TRISO-coated particle fuel and compacts and fuel spheres used in high temperature gas reactors.
is a noninstrumented rig that can hold 10–13 graphite VXF positions and then moved to the higher flux
disks (15 mm in diameter and 2 mm thick), each of removable beryllium positions. Neither of these posi-
which can hold 50 particles. The rig can also hold tions can accommodate pebbles. A third facility, the
coated particles in axial holes, 1.2 mm in diameter, large VXF positions (six total), are farther out in
and a uniform volume of coated particles, 12–18 mm the reflector (and therefore have lower fluxes), but
in diameter and 20–255 mm high, in a graphite are 72 mm in diameter and also 500 mm long. As with
matrix. Another rig, termed a ‘CP hole,’ is 27 mm in the HFR, there is a large axial flux gradient that must
diameter and that can handle six to eight capsules. be considered in the design of any experiment in any
A third rig, identified as ASU-8, is a 60-mm hole that of these facilities.
can handle three compacts. The largest channel avail-
able is Vostok, which is 120 mm in diameter and 3.07.2.1.5 ATR
contains four cells. All of these rigs can irradiate The Advanced Test Reactor (ATR) at Idaho National
fuel at representative temperatures, burnups, and Laboratory (INL) is a light-water-cooled, beryllium-
fluences for HTGRs. There is a large degree of reflected reactor fuel in a four-leaf clover configura-
flexibility in the testing options at IVV-2M. Their tion to produce high neutron fluxes for materials
rigs can handle particles, compacts, and spheres. testing and isotope production.11 The clover leaf
configuration results in nine very high flux positions,
3.07.2.1.3 HFR Petten termed ‘flux traps.’ In addition, numerous other holes
The High Flux Reactor (HFR) in Petten, Netherlands, of varying size are available for testing. Several posi-
is a multipurpose research reactor with many irradi- tions can be used to irradiate coated-particle fuel.
ation locations for materials testing.9 The HFR has The 89-mm-diameter medium I position (16 total)
two different types of irradiation rigs/locations in the and the 100–125-mm-diameter flux traps can accom-
facility: one that can accommodate compacts and modate pebbles. Specifically, the use of a medium
another that can accommodate spheres. Rigs for I position early in the irradiation, required because
spheres are multicell capsules, 63–72 mm in diameter of the enrichment of the fuel, followed by transfer of
that can handle 4–5 spheres in up to 4 separate cells. the test train to the northeast flux trap can provide
For compacts rigs/locations are 32 mm in diameter irradiation conditions representative of a pebble bed
and 600 mm in useful length. They can handle three reactor. Approximately 10–12 pebbles in five or six
or four parallel channels of compacts. For the three- individually swept cells can be envisioned in the test
channel configuration, approximately 30 compacts train. The large B positions in ATR (four total) are
could be irradiated in the rig. There is a large axial 38 mm in diameter and 760 mm in length. They can
flux gradient across the useable length (40% spread accommodate six individually swept cells, with two
maximum to minimum) that must be considered in graphite bodies per cell, containing up to three 2-in.
the design of any experiment. long compacts per body. Thus, 36 full-size US com-
pacts can be irradiated in this location. Of special
3.07.2.1.4 HFIR note, here is the very flat burnup and fluence profile
The High Flux Isotope Reactor (HFIR) at Oak Ridge available axially in the ATR over the 760 mm length.
National Laboratory (ORNL) is a light-water-cooled, This allows for nearly identical irradiation of large
beryllium-reflected reactor that produces high neu- quantities of fuel.
tron fluxes for materials testing and isotope produc-
tion.10 Two specific materials irradiation facilities 3.07.2.1.6 SAFARI
locations are available for gas reactor fuel testing: The SAFARI Reactor in Pelindaba, Republic of
(a) the large RB positions (eight total) that are South Africa, is an isotope production and research
46 mm in diameter and 500 mm long, and can accom- reactor.12 The core lattice is an 8 9 array, consisting
modate capsules holding up to 24 compacts (three in of 28 fuel assemblies, 6 control rods, and a number
each graphite body, eight bodies axially) in a single of aluminum and beryllium reflector assemblies.
swept cell; and (b) the small vertical experimental The reactor is cooled and moderated by light water
facility (VXF) positions (16 total) that are 40 mm in and operates at a maximum power level of 20 MW.
diameter and 500 mm long, and can accommodate In-core irradiation positions include six high-flux
capsules holding up to 16 compacts (eight in each isotope production positions: two hydraulic, two
graphite body, two bodies axially) in a single swept pneumatic, and two fast transfer systems that are
cell. Capsules can be irradiated in the lower flux small accessible during operation. Several other irradiation
156 TRISO-Coated Particle Fuel Performance
positions can also be accessed when the reactor As discussed in Section 3.07.2.7, the irradiation
is shut down. A large poolside facility allows for a performance database suggests that modest levels of
variety of radiation applications. An intermediate acceleration (1.5–3) appear to be acceptable with-
storage pool and a transfer canal allow for easy and out jeopardizing fuel performance in the irradiation,
safe transport of activated materials to a hot cell. and should be a baseline requirement for future gas
reactor irradiations. This acceleration level can be
3.07.2.1.7 TRISO-coated particle fuel translated into a maximum power per fuel body or
irradiation testing power per particle that can be used by experimenters
The historical experience in irradiation testing of in the design of the irradiation capsule.
coated particle fuels suggests that multicell capsules Given the limitations of materials test reactors
wherein fuel can be tested in separate compartments around the world, the TRISO-coated particle irradi-
under different temperature, burnup, and fluence con- ation database contains results from tests conducted
ditions allow for tremendous flexibility and can actually under a range of accelerations. Successful German
save time and money in an overall fuel qualification TRISO-coated particle fuel irradiations in the
program. Although there are differences in details of European HFR-Petten reactor were conducted using
the test trains used in each of the reactors, they share a an acceleration of less than a factor of three. By com-
number of important similarities in the state of the art parison, other German irradiations in the Forschung-
with irradiation testing of this fuel form. In this section, zentrum Reaktor Juelich (FRJ) reactor at Jülich had
these important similarities are presented to highlight a neutron spectrum that was too thermalized. This
the technical considerations in executing this type of resulted in the fuel receiving too little fast fluence
testing. to be prototypic of a high-temperature gas reactor.
Because of the differences in neutron flux spec- Similarly, historic US irradiations in ORNL’s HFIR
trum between a gas reactor and a light-water materi- reactor had too high a thermal flux resulting in signifi-
als test reactor, simultaneous matching of both the cant burnup acceleration of the irradiation. On the basis
rate of burnup and the rate of accumulation of fast of these considerations, the large B positions (38 mm
neutron fluence is difficult to achieve. In addition, the diameter) in the ATR (see Figure 2) were chosen for
traditional 3-year fuel cycle of high-temperature gas the US Department of Energy’s (DOE) Advanced Gas
reactors makes real-time irradiation testing both time- Reactor (AGR) Program fuel irradiations because the
consuming and an expensive part of an overall fuel rate of fuel burnup and fast neutron fluence accumula-
development effort. To overcome these shortcomings, tion in these positions provide an acceleration factor
irradiations in material test reactors have historically of less than three times that expected in the high-
been accelerated relative to those in the actual reactor. temperature gas reactor.
Usually, the time acceleration is focused on achieving
the required burnup in a shorter time than would be 3.07.2.1.8 Thermal and physics analysis
accomplished in the actual reactor, with the value of considerations
the fast fluence left as a secondary variable that must Given the complexity of the capsules currently being
fall between a minimum and maximum value. designed, the extensive review by safety authorities of
The level of acceleration can also impact the the thermo-mechanical stresses, and the importance of
potential for fuel failure during irradiation. The each capsule in terms of irradiation data for fuel quali-
level of acceleration at a given test reactor power, fication, three-dimensional physics and thermal ana-
coupled with fuel loading in the experiment, results lyses are essential in irradiation capsule design. These
in a power density for the fuel specimen in the analyses are critical to ensure that the fuel reaches the
experiment. The power density peaks at the begin- intended burnup, fluence, and temperature conditions.
ning of the irradiation when the fissile content is To achieve high burnups with these fuels requires
highest and decreases as the fissile material is burned detailed physics calculations to determine the time to
out of the fuel. As the level of acceleration increases, reach full burnup. Given the concerns about severely
the temperatures in the fuel kernels increase above accelerated irradiations, it is not uncommon for such
that in the fuel matrix because of the thermal resis- irradiations to take approximately 2 years to reach full
tances associated with the coatings of the particle,13 burnup in LEU TRISO-coated particles. In addition,
and the potential for high temperature, thermally because thermocouples should not be attached directly
driven failure mechanisms to play a deleterious role to the fuel, thermal analysis is used to calculate the fuel
in fuel performance becomes more important. temperature during the irradiation.
TRISO-Coated Particle Fuel Performance 157
North
ON-1 ON-2
I-19
I-20 I-1 I-2
I-3
H positions
I-16 I-6
I-15 I-7
East large B
position location
I-14 I-8
for AGR-1
Control drum
OS-8 OS-9 OS-10 OS-11 OS-12
Examples of a test train for fuel compacts used in Direct temperature measurements of the coated
INL’s ATR and the pebbles used in HFR-Petten are particles are problematic because direct metal con-
shown in Figures 3 and 4 respectively. tact (e.g., thermocouple wires or sheaths) with the
These irradiation capsules have extensive instru- fuel element is not recommended as the metals can
mentation to measure temperature, burnup, and fast attack the TRISO fuel coatings. Thus, temperatures
fluence at multiple locations in the test train. Tradi- must be calculated on the basis of thermocouples
tional commercial thermocouples have been used located elsewhere in the capsule. Thermocouples
extensively in past irradiations, but thermocouples are generally located as close as possible to the
can suffer from drift and/or de-calibration in the fuel body to minimize the uncertainties on the
reactor. Redundancy in thermocouple measurements calculated fuel temperatures related to irradiation-
is another consideration in light of the low reliability induced dimensional change and thermal con-
of thermocouples at high temperatures and long ductivity changes of the materials in the capsule.
times in neutron fields typical of TRISO-coated par- Encapsulating the fuel element in a graphite sleeve
ticle fuel irradiations. Melt wires are inexpensive and or cup and inserting thermocouples into the graphite
have been used as a backup to thermocouples where has been used successfully in many designs. The
space was available in the capsule. However, melt high conductivity of graphite minimizes the effect
wires only indicate that a certain peak temperature of irradiation-induced dimensional changes on the
has been reached, and not the time of that peak. calculated fuel temperature.
158 TRISO-Coated Particle Fuel Performance
Gas line
Fuel stack
Thermocouple
Capsule
spacer nub
argon cannot be used in fuel experiments where experiment using ion chambers and sodium iodide
online fission product monitoring is used because detectors is an excellent means to capture any
the activated argon will reduce detectability of the dynamic failures of the coated particles associated
system. Therefore, helium and neon are used instead. with the instantaneous release upon failure. Grab
Computer-controlled mass flow controllers are typi- samples can provide excellent noble gas isotopic
cally used to automatically blend the gases (on the information. The temporal resolution and the number
basis of feedback from the thermocouples) to control of isotopes that can be measured depend on the fre-
temperature. The gas blending approach allows for a quency of the grab samples and the delay time
very broad range of control. Automatic gas verifica- between acquisition of the grab sample and offline
tion (e.g., by a thermal conductivity analyzer) has analysis. Weekly grab samples are typical in most
been implemented in some experiments to prevent irradiations, although daily or even hourly samples
the inadvertent connection of a wrong gas bottle. Gas are possible if failure has occurred, assuming opera-
purity is important and an impurity cleanup system tion and associated analysis costs are not too high.
should be implemented during each irradiation. Flow Typical isotopes to be measured include 85mKr, 87Kr,
88
rates and gas tubing should be sized to minimize Kr, 131mXe, 133Xe, and 135Xe. Measurement of very-
transit times between the mass-flow controllers and long-lived isotopes (e.g., 85Kr) would be useful in
the experiment, as well as between the experiment elucidating fission product release mechanisms from
and the fission product monitors. the kernel, but would also require waiting for the
decay of the shorter lived isotopes in the sample.
3.07.2.1.10 FPMS considerations Online gamma spectroscopy, although the most
In addition to thermal control, sweep gas is used to expensive in terms of hardware costs, can provide
transport any fission gases released from the fuel to a the most detailed real-time information with detailed
fission product monitoring system (FPMS). A num- isotopic spectrums as often as necessary subject to
ber of techniques have been used historically to data storage limitations of the system. An example of
quantify the release of fission gases from the fuel in the system used for the US AGR program is shown in
these irradiation capsules. Techniques include gross Figure 5. With such systems, transit times from the
gamma monitoring, online gamma spectroscopy, and experiment to the detector should be minimized to
offline gamma spectroscopy of grab samples. Online allow measurement of short- and medium-lived iso-
gross gamma monitoring of the effluent gas in the topes, but must remain long enough to allow decay of
Temperature control
gas mixing system
Vessel wall
He
Ne
Filter
H and V exhaust
Particulate
filters
6
5
Silver
4 Capsules zeolite
in-core
3
2 Fission product
monitoring system
1 Grab sample
Figure 5 Integrated fission product monitoring system used in US AGR program irradiations.
160 TRISO-Coated Particle Fuel Performance
any short-lived isotopes associated with the sweep Table 1 German particle fuel development sequence
gases (2–3 min). With this delay time, 89Kr, 90Kr,
135m Date of design Fuel form
Xe, 137Xe, 138Xe, and 139Xe should also be capa- consideration
ble of being measured online. Measurements of
xenon gas-release during reactor outages are recom- 1972 BISO coated (Th, U)O2
1977 Improved BISO coated (Th, U)O2
mended to provide information on iodine release TRISO-coated UCO fissile particles with
behavior from the decay of xenon precursors. Multiple ThO2 fertile particles
options for fission gas-release measurements should be TRISO-coated (Th, U)O2
considered for long irradiations where reliability of the 1981 LEU TRISO-coated UO2
overall fission gas measurement system can be a con-
cern. Redundancy is also recommended for online
systems so that failure of a spectrometer does not timely manner, tests conducted by German research-
jeopardize the entire experiment. ers were generally accelerated by factors of 2–3.
On the basis of the online concentration data, a The following sections present irradiation experi-
release-to-birth ratio (R/B), a key parameter used in ment summaries for fuels of ‘modern’ German design.1
reactor fuel behavior studies,14 can be calculated and For these experiments, this definition extends to
provide some insight into the nature of any particle high-enriched (Th, U)O2 TRISO-coated particles
failures. Because these instruments are online during fabricated since 1977, and low-enriched UO2 TRISO-
the entire irradiation, a complete time history of gas coated particles fabricated since 1981. Table 2 pro-
release is available. Gas release early in the irradia- vides the physical attributes of the fuel used in these
tion (i.e., from the start of the irradiation) is indicative tests. Mixed oxide fuel test summaries are presented
of initially failed particles or contamination outside first, followed by the UO2 tests.
of the SiC layer. Release later during the irradiation is
indicative of in situ particle failure. The timing of the 3.07.2.2.1 R2-K12 and R2-K13
failure data can then be correlated to temperature, The R2-K12 and R2-K13 cells were irradiated in the
burnup, and/or fluence that can be used when cou- R2 reactor at Studsvik, Sweden. The main objective
pled with PIE to determine the mechanisms respon- of the R2-K12 experiment was to test mixed oxide
sible for the fuel failure. (Th, U)O2 and fissile UC2/fertile ThO2 fuel ele-
ments, whereas for R2-K13, the main objective was
to test mixed oxide (Th, U)O2 fuel elements and
3.07.2.2 German Experience
supply fuel for subsequent safety tests.
Previously, particle fuel development was conducted In R2-K12, four full-size spherical fuel elements
by German researchers in support of various HTGR were irradiated in four independently gas-swept cells.
designs that employed a pebble bed core. These Two cells contained mixed oxide fuel spheres, while
reactors were intended to produce process heat or the other two contained fissile/fertile fuel spheres. As
electricity. The sequence of fuel development used the German researchers did not develop the two-
by German researchers followed improvement in particle fissile/fertile system further, only the mixed
particle quality and performance and was largely oxide results were reported. R2-K13 was a combined
independent of developments in reactor technology. experiment with the United States. Four indepen-
German fuel development can be categorized dently gas-swept cells were positioned vertically on
according to the sequence of fuels tested as provided top of one another. The top and bottom cells each
in Table 1. contained a full-size German fuel sphere. The middle
German irradiation test conditions generally two cells contained US fuel and will be discussed in
covered projected fuel operating conditions, where Section 3.07.2.3. Configuration and irradiation data
fuel was to reach full burnup at fast fluences from both experiments are given in Tables 3 and 4.
of 2.4 1025 n m2 and operate at temperatures Cold gas tests on each fuel sphere during PIE
up to 1095 C for process-heat applications and indicated that all the particles had remained intact
up to 830 C for electrical production applications. in both R2-K12 and R2-K13. These tests are con-
With the exception of irradiation duration, the vari- ducted after the fuel has been stored (for 14 days) at
ous experiments performed bounded expected nom- room temperature and a quasi-steady-state release of
inal conditions or were purposely varied to meet fission gas has been reached. The fuel is then swept
other test objectives. In order to obtain results in a with a carrier gas that is monitored for various fission
TRISO-Coated Particle Fuel Performance 161
Source: Gontard, R.; Nabielek, H. Performance Evaluation of Modern HTR TRISO Fuels; Tech. Rep. HTA-IB-05/90; Forschungszentrum
Jülich GmbH: Jülich, Germany, 1990.
Notes: The entries are one standard deviation. Entries in square brackets, [ ], are estimated values.
BAF is the Bacon anisotropy factor for the layer, where values closer to one are more isotropic.
Table 3 R2-K12 and R2-K13 configuration proportional to the gas source, and in a calibrated
R2-K12 R2-K13
system, indicates the number of failed particles.
The fuel sphere from R2-K12 Cell 1 was partially
Number of cells 2 2 deconsolidated and visual inspection revealed two
Number of fuel spheres 2 2 kernels ‘without coating.’ Segments from each of the
Spherical fuel element 59.9 mm 59.77 mm
diameter
two fuel spheres were also metallographically exam-
Fuel zone diameter 47 mm 47 mm ined; those examinations revealed a reaction zone on
Fuel type HEU (Th, U)O2 HEU (Th, U)O2 the inner side of the buffer layer, as well as tangential
LTI – TRISO LTI – TRISO cracks between the buffer and the inner pyrocarbon
Particle batch EO 1607 EO 1674 layer. Only one particle exhibited a radial crack in the
235
U enrichment 89.57% 89.01%
235
U per fuel element 1.002 g 1.02 g
buffer layer beyond the reaction zone. All of the SiC
232
Th per fuel element 4.961 g 10.125 g and PyC layers examined remained intact.
Heavy metal per fuel 6.076 g 11.27 g
element 3.07.2.2.2 BR2-P25
Number of particles per 10 960 19 780 The BR2-P25 capsule was irradiated in the BR2 reac-
spherical fuel element tor at Mol, Belgium. The primary objective of this
Defective SiC layersa <1 105 <5 106
experiment was to test (Th, U)O2 mixed oxide fuel.
(U/U-total)
One independently gas-swept cell contained 12 com-
a
Defective SiC layer fractions reported for German fuel are per pacts. Each compact was cylindrical in shape and
pebble with the exception of loose particle experiments that are contained a small fuel sphere. Configuration and irra-
per particle batch.
Source: Gontard, R.; Nabielek, H. Performance Evaluation of
diation data are given in Tables 5 and 6, respectively.
Modern HTR TRISO Fuels; Tech. Rep. HTA-IB-05/90; During PIE, Compacts 3 and 7 were electrolytically
Forschungszentrum Jülich GmbH: Jülich, Germany, 1990. deconsolidated with no particle failures being evident.
Ceramographic examination of cross-sections from
gases (usually 85mKr) and heated to 60 C. Sudden Compacts 4 and 8 revealed some radial cracks in
increases in the amount of fission gas detected indi- the buffer layers; however, no defective particles
cate failed particles. The amount of increase is were found.
162 TRISO-Coated Particle Fuel Performance
R2-K12 R2-K13
Source: Gontard, R.; Nabielek, H. Performance Evaluation of Modern HTR TRISO Fuels; Tech. Rep. HTA-IB-05/90; Forschungszentrum
Jülich GmbH: Jülich, Germany, 1990.
3.07.2.2.4 SL-P1
3.07.2.2.3 HFR-P4 The SL-P1 experiment was irradiated at the Siloë
The HFR-P4 capsule was irradiated at the HFR Reactor in Grenoble, France. The objective of the
in Petten. The main objective of this experiment experiment was to test reference LEU fuel up to the
was to compare the fuel performance of particles potential limits for burnup and fast fluence at 800 C.
TRISO-Coated Particle Fuel Performance 163
3.07.2.2.5 HFR-K3 and irradiation data are given in Tables 11 and 12,
The HFR-K3 capsule was irradiated at the HFR in respectively. Subsequent PIE reported no failures.
Petten. The primary objective of this experiment was
to determine the performance of reference LEU fuel 3.07.2.2.6 FRJ2-K13
from an accelerated test. Four full-size spherical fuel FRJ2-K13 cells were irradiated at the DIDO reactor in
elements were irradiated in three independently gas- Jülich, Germany. The main objective of this test was to
swept cells. The cells were vertically positioned on supply irradiated reference fuel for subsequent safety
top of one another, with the middle cell containing tests. Fuel performance was also to be examined under
two fuel spheres. To minimize flux gradient effects the controlled irradiation conditions of significant
on the test fuel, the entire test rig was rotated 90 burnup with negligible fast neutron fluence. Four full-
several times during the irradiation. Configuration size spherical fuel elements were irradiated in two
164 TRISO-Coated Particle Fuel Performance
independently gas-swept cells. The cells were vertically 11 h. The 85mKr R/B ratio from each capsule increased
positioned on top of each other, with the fuel spheres to a maximum of 108 at the start of the transient and
similarly positioned within the cells. Configuration and then dropped back to the pretransient levels after the
irradiation data are given in Tables 13 and 14, respec- temperature was returned to the nominal test condition.
tively. Subsequent PIE reported no failures.
3.07.2.2.8 FRJ2-P27
3.07.2.2.7 FRJ2-K15 FRJ2-P27 cells were irradiated at the DIDO reactor
FRJ2-K15 cells were irradiated at the DIDO reactor in Jülich, Germany. The main objectives of this test
in Jülich, Germany. The main objectives of this test were to investigate fission product release at various
were to demonstrate the high burnup potential of cyclic temperatures and to determine the effective-
reference fuel used in AVR reload 21-1 and to per- ness of thicker SiC layers on the retention of 110mAg.
form in-core temperature transient tests. Fuel perfor- Each of the three independently gas-swept cells
mance was also to be examined under the controlled contained three compacts and two coupons (trays).
irradiation conditions of significant burnup with neg- The compacts were cylindrical in shape and contained
ligible fast neutron fluence. Three full-size spherical an (unspecified) outer fuel-free zone. The coupons
fuel elements were irradiated in three independently were graphite disks with holes, annularly spaced, for
gas-swept cells. Configuration and irradiation data the insertion of 34 particles. Of the two coupons that
are given in Tables 15 and 16, respectively. contained the thicker SiC particles (51 mm vs. 36 mm),
Capsules 2 and 3 underwent a temperature transient one was placed in Cell 1, and the other in Cell 3.
test at a burnup of 10% FIMA. The temperature of Configuration and irradiation data are provided in
the sphere surfaces was raised to 1100 C and held for Tables 17 and 18, respectively.
HFR-K6 HFR-K5
high as 1400 C and reach full burnup (commensurate experiment summaries are listed in chronological
with 235U enrichment and kernel composition) at fast order and are not grouped by fuel type. Listed config-
fluences of 4 1025 n m2. With the exception of irra- uration and irradiation data are actual values, not
diation duration, the various experiments performed specification values or ranges. Interpretations of PIE
either bounded expected nominal conditions or were results are from the original sources and no overt
purposely varied to meet other test objectives. In order attempt has been made to reinterpret the results.
to obtain results in a timely manner, US tests were
accelerated by factors of 3–10. 3.07.2.3.1 F-30
The particle fuel irradiation experiments and PIE The F-30 experiment was irradiated in the Gen-
results summarized below consider only selected eral Electric Test Reactor (GETR) at Pleasanton,
tests of key US fuel types. These fuel types include California.17 The primary objective of this experi-
TRISO fissile/BISO fertile particles, weak acid resin ment was to demonstrate the irradiation performance
(WAR) TRISO fissile/BISO fertile particles, TRISO of Fort St. Vrain production fuel. Five independently
fissile/TRISO fertile particles, and TRISO-P fissile gas-swept cells contained the fuel. Cells 1, 3, and 4
particles (conventional TRISO-coated particles with contained only fuel compacts, Cell 2 contained only
an additional ‘protective’ pyrolytic carbon layer above loose particles, and Cell 5 contained both fuel com-
the outer pyrolytic carbon layer) as well as TRISO pacts and loose particles. Configuration and irradia-
fissile particles. General Atomics and Babcock & tion data are given in Tables 22 and 23, respectively.
Wilcox manufactured the majority of the kernel and Postirradiation metallographic examination of
coating batches. However, some of the batches were seven fuel compacts containing fissile and fertile
manufactured by ORNL. The following US particles was performed. In addition, five sets of
TRISO-Coated Particle Fuel Performance 167
Number of cells 5
Total number of fuel compacts 13
Cylindrical fuel compact diameter 12.45 mm
Cylindrical fuel compact lengths 18.54 and 49.28 mm
Fissile fuel type HEU (Th, U)C2 TRISO
Nominal Th/U ratio 4.25
235
U enrichment 93%
Fissile particle diameter 429–560 mm
Fertile fuel type ThC2 TRISO
Fertile particle diameter 648–771 mm
Number of fissile particle batches 7
Number of fertile particle batches 9 100 mm
Defective SiC layer <5 104–1.5 103
fraction – fissile particles Figure 7 A typical IPyC layer crack in a fissile F-30 fuel
Defective SiC layer 3 104–1.0 103 particle. Reproduced from Scott, C. B.; Harmon, D. P.
fraction – fertile particles Post Irradiation Examination of Capsule F-30; GA-A13208,
UC-77; General Atomics Report, 1975.
HRB-4 HRB-5
Number of cells 1 1
Number of fuel compacts 6 6
Cylindrical fuel compact 12.4 mm 12.4 mm
diameter
Cylindrical fuel compact 25.4 mm 25.4 mm
lengths
Fissile fuel type WAR UC2 WAR UC2
TRISO TRISO
Fertile fuel type ThO2 BISO ThO2 BISO
235
U enrichment 5.99% 5.99%
Fissile particle diameter 639 mm 639 mm
Fertile particle diameter 805 mm 805 mm
Fissile particle batch OR52A OR52A
Fertile particle batch T01424BIL T01424BIL
Total number of fissile 17 780 17 780
particles Figure 8 Typical HRB-4 fissile particle irradiated to 27.7%
Total number of fertile 4180 4180 FIMA and 10.5 1025 n m2 fast fluence. Reproduced
particles from Scott, C. B.; Harmon, D. P. Post Irradiation
Examination of Capsules HRB-4, HRB-5, and HRB-6;
GA-A13267, UC-77; General Atomics Report, 1975.
HRB-4 HRB-5
Number of cells 1
Number of fuel compacts 6
Cylindrical fuel compact diameter 12.4 mm
Cylindrical fuel compact length 25.4 mm
Fissile fuel type HEU (Th, U)C2 TRISO
Nominal Th/U ratio 4.25
235
U enrichment 93.15%
Fissile particle diameter 556 mm
Fertile fuel type ThO2 BISO
Fertile particle diameter 888 mm
Fissile particle batch CU6B-2427
Fertile particle batch T01451BIL-W
Defective SiC layer <5 104
fraction – fissile particles
Number of cells 2
Total number of fuel compacts 88
Cell 1 cylindrical fuel compact 15.75 mm diameter,
dimensions (16 compacts) 25.4 mm long
Cell 2 cylindrical fuel compact 15.75 mm OD,
dimensions (48 compacts) 3.30 mm ID,
12.70 mm long
Cell 2 cylindrical fuel compact 15.75 mm diameter,
dimensions (24 compacts) 50.8 mm long
Fissile fuel type WAR UCxOy TRISO
(Th, U)O2 TRISO
UC2 TRISO
235
U enrichment Not reported
Fissile particle diameter 600–753 mm
Fertile fuel type ThO2 BISO
Fertile particle diameter 806–889 mm
Figure 11 Typical HRB-6 fissile particle irradiated to Number of fissile particle batches 15
26.5% FIMA and 7.9 1025 n m2 fast fluence. Reproduced Number of fertile particle batches 16
from Scott, C. B.; Harmon, D. P. Post Irradiation
Examination of Capsules HRB-4, HRB-5, and HRB-6;
GA-A13267, UC-77; General Atomics Report, 1975.
Table 29 OF-2 irradiation data
Number of cells 1
Total number of fuel compacts 20
Cylindrical fuel compact diameter 12.50 mm
Cylindrical fuel compact length 9.52 mm
Fissile fuel type UCxOy TRISO
(Th, U)O2 TRISO
UO2 TRISO
235
U enrichment 19.18–19.66%
Fissile particle diameter 760–813 mm
Fertile fuel type ThO2 TRISO
Fertile particle diameter 786–882 mm
Number of fissile particle batches 5
Number of fertile particle batches 8
Defective SiC layer 7.0 107–
fraction – fissile particles 1.3 104
Metallographic examination was performed on 15 observed in 8% of the particles. UCO fuel particles
fuel compacts, and 8 of them contained fissile parti- that did not display localized chemical attack, had
cles. A few fissile particles were reported to have SiC uniform attack along the inner SiC layer (usually on
layer cracks but these cracks were attributed to one side of the particles). This uniform attack was
metallographic preparation. It should be noted that attributed to rare earth fission products. Figure 16
visual inspection of each compact during capsule displays typical uniform fission product attack in a
disassembly indicated that between 0% and 9% of UCO fuel particle. It should be noted that with
the visible particles (from compact surfaces and loose optimized UCO stoichiometry, the kernel retains rare
particles that had fallen off) had failed SiC layers. earth fission products and does not display kernel
However, this visual inspection did not distinguish migration as found here with non-optimized UCO
between fissile and fertile particles. kernels containing excess UC2 leading to rare earth
The metallographic examination of fissile parti- migration.
cles revealed that between 0% and 3% of the IPyC Metallographic examination of fertile particles
layers had failed (cracked) and that the IPyC layers indicated that between 0% and 2.4% of the particles
had debonded from the SiC in 0% to 7.7% of the in each compact had total coating failure, defined as
particles. Buffer layers did not crack in the UO2 or cracked OPyC and SiC layers. These failures were
(Th, U)O2 fuel but did crack in 10–71% of the UCO attributed to pressure vessel failure. Figure 17 dis-
fuel particles. Kernel extrusion was reported only in plays a typical failed fertile particle. Separate tallies
UCO fuel. Figure 14 displays typical kernel extru- of particles where only the SiC layer had failed were
sion, and Figure 15 presents a typical photomicro- not reported. Other fertile particle observations
graph of kernel migration. include the following:
In several particles of each fuel form, high con-
1.5–29.1% of the particles had failed OPyC layers
centrations of fission products were observed in
8–70% of the particles had failed IPyC layers
small, localized regions at the SiC–IPyC layer inter-
11–85% of the particles had IPyC layers debonded
face. In addition to fission product accumulation,
from the SiC
localized chemical attack was also observed in the
6–26% of the particles had cracked buffers
SiC layers of several (Th, U)O2 and UO2 fuel particles.
no kernel migration was observed
This localized attack, which had penetrated 2 mm
a few kernels had extruded into buffer cracks.
into the SiC, was attributed to palladium, and was
Number of cells 1
Total number of particle trays 184
Maximum number of loose 116
particles per tray
Particle tray outer diameter 22.3–23.6 mm
Particle tray inner diameter 11.1 mm
Fissile fuel type UCO TRISO and
silicon-BISO
(Th, U)O2 TRISO and
silicon-BISO
UC2 TRISO and
silicon-BISO
UO2 TRISO and
silicon-BISO
UO2* TRISO and
silicon-BISO
235
U enrichment 19.5%
Fissile particle diameter 742–951 mm
Fertile fuel type ThO2 TRISO, BISO
and silicon-BISO
Figure 16 Photomicrograph of a UCO particle (batch
Fertile particle diameter 773–836 mm
6157-08-020) from Compact 10 irradiated at 1040 C to
Number of fissile particle batches 19
27.8% FIMA and to a fast fluence (E > 0.18 MeV) of
Number of fertile particle batches 22
7.1 1025 n m2 displaying fission product attack of the
SiC layer. Reproduced from Young, C. A. Pre- and Post
Note: Two types of UO2* fuel were tested, one with ZrC dispersed
Irradiation Evaluation of Fuel Capsule HRB-14; GA-A15969,
in the buffer and the other with pure ZrC layer around the kernel.
UC-77; General Atomics Report, 1980.
Note: Two types of UO2* fuel were tested, one with ZrC dispersed
in the buffer and the other with pure ZrC layer around the kernel.
debonding was observed. It was also observed that demonstrate the irradiation performance of reference
between 10% and 30% of the particles with failed NP-MHTGR fuel at the upper bounds of nominal
IPyC layers also displayed cracked SiC layers. operating conditions. The same reference fuel was
also irradiated in the NPR-1 and NPR-2 tests. For
3.07.2.3.12 NPR-1A NPR-1A, 20 fuel compacts were placed vertically in a
The NPR-1A capsule was irradiated in the ATR at single, gas-swept cell. Originally, the test was sched-
the INL.27 The primary objective of the test was to uled for 104 days of irradiation, but was terminated
after 64 days because of indications of a significant
number of fuel particle failures. Configuration and
Table 44 NPR-1A configuration irradiation data are given in Tables 44 and 45.
Number of cells 1
Postirradiation metallographic examination of one
Number of fuel compacts 20 fuel compact was performed. This examination indi-
Cylindrical fuel compact diameter 12.37–12.50 mm cated that 1% of the SiC layers had failed. On the
Cylindrical fuel compact lengths 49.33 mm basis of the online gas measurements, it was estimated
Fuel type HEU UCO TRISO-P that approximately 48 particles had failed, which
235
U enrichment 93.15%
Fuel particle diameter 758 mm
correspond to 0.06% of the total particle population.
Fuel particle batch FM19-00001 composite
Total number of fuel particles 75 360 3.07.2.3.13 AGR-1
Defective SiC layer fraction 3 106 The AGR-1 experiment involves six separate cap-
sules, each containing approximately 50 000 particles
in the form of fuel compacts. It is an instrumented
Table 45 NPR-1A irradiation data
lead experiment, irradiated in an inert sweep-gas
atmosphere with individual online temperature mon-
Start date 2 October 1991 itoring and control of each capsule. A horizontal cap-
End date 3 January 1992 sule cross-section at the top of the test train is shown
Duration (full power days) 64
Peak burnup (% FIMA) 64
in Figure 20. A vertical section of the capsule was
Peak fast fluence 2.1 shown previously in Figure 3, and the experiment
(1025 n m2, E > 0.18 MeV) flow path was shown previously in Figure 5. The
Average temperature ( C) 977 sweep gas also has online fission product monitoring
Peak temperature ( C) 1220 on its effluent to track performance of the fuel in each
BOL 85mKr R/B 4 109
EOL 85mKr R/B 1.8 105
individual capsule during irradiation (see Figure 5).
The first of eight planned experiments, AGR-1, had
Graphite
Thermocouples
Insulating
Stack 1
Stack 3
Hf shroud
Property Specified range for Actual mean value population standard deviation
mean value
Baseline Variant 1 Variant 2 Variant 3
a
Specification does not apply to variants 1 and 2.
b
Value is an estimate of an attribute property, not the mean of a variable property.
been under irradiation at the INL ATR and was Table 47 AGR-1 configuration data
completed in November, 2009; PIE is scheduled to Number of cells 6
begin in April, 2010.28,29 Table 46 presents pertinent Number of compacts per cell 12
attributes of the fuel that is being irradiated in AGR- Cylindrical compact diameter 12.34–12.36 mm
1.30 Configuration data are presented in Table 47. Cylindrical compact height 25.0–25.3 mm
Irradiation of the experiment began on 24 December Fuel type LEU UCO TRISO
Particle batch – capsule 3 and 6 Baseline
2006, and will continue for approximately 2.5 years to Particle batch – capsule 1 and 4 Variant 3
reach a peak burnup of 19% FIMA for the fuel Particle batch – capsule 2 Variant 2
compacts. Particle batch – capsule 5 Variant 1
235
Table 48 provides a summary of the irradiation U enrichment 19.6%
status of the AGR-1 experiment. Detailed as-run Number of particles per compact 4 150
Number of particles per capsule 49 800
physics and thermal analyses are performed cycle Defective SiC layers <4 105
by cycle to track fuel burnup, fast neutron fluence
damage, and fuel temperatures during the irradiation.
Peak burnups ranged from 16 to 19% FIMA and
fast fluences were between 3.0 and 4.0 1025 n m2 time-averaged volume-averaged temperatures are
(E > 0.18 MeV). calculated as the irradiation progresses. After 514
On the basis of the fuel temperature distribut- effective full power days, the time-averaged peak
ions during each cycle, time-averaged peak and fuel temperatures ranged between 1120 and 1180 C
180 TRISO-Coated Particle Fuel Performance
Table 49 German and Chinese fuel attributes for EU1 and EU1bis
To evaluate the fuel performance of the HTR-10 13 500 loose fuel particles, were irradiated in the
fuel, two fuel elements (spheres) were chosen at Russian IVV-2M reactor. Selected physical attributes
random from each of the first two production batches. of the fuel used in this test are listed in Table 52.
These four fuel spheres, together with approximately The irradiation test rig consisted of five capsules.
Capsule 1, located at the top, contained the loose
particles, while Capsules 2 through 5, in descending
Gaslube connectors order, each contained a test fuel sphere. Irradiation
Thermocouple plug test data are summarized in Table 53.
After 3.9% FIMA, where 85mKr R/B values were
Penetration reactor top lid
in the 3 107 range, gas-release suddenly increased
Dynamic seating
penetration plug
in Capsule 4 with R/B values exceeding 103. This
result indicated a significant number of fuel particle
Second containment failures. Further measurements of gas-release from
gas out
this capsule were terminated at this point. PIE indi-
Biological shield cated that oxygen impurities in the helium flow gas
(such as from air ingress) had reacted with the fuel
Refa-170
element matrix material. This reaction turned the
Thermocouples
matrix into a graphite powder agglomerate and dam-
Second containment (Refa) aged the capsule thermocouple, the internal metal
+270 Purge gas out containment, and the fuel particles. No intact fuel
particle from this capsule has been found.
1
+190 From beginning to end of irradiation, the remaining
First containment three capsules displayed slight variations in gas-release
2
+120 (capsule) (other than for the in-pile heat-up tests). These varia-
3
tions were attributed to fluctuations in temperature
+50
control and implied that no complete (through all
0 Core
4 0.4 0.5 0.6 0.7 0.8 0.9 1.0
layers) particle failures had occurred during irradiation.
–20
The high R/B values were due to high U contamination
5 levels and initially defective particles. Quality was
–90
improved in subsequent fuel production batches.
–160 6 During irradiation, after 4.1% FIMA, the tempera-
ture in Capsule 3 was raised from a nominal 1000–
–230 7
1200 C for 200 h and then returned to the nominal
level. Likewise, after 6.1% FIMA, Capsule 3 tempera-
–310 ture was raised from a nominal 1000–1250 C for 200 h
First containment purge gas in and again returned to the nominal level. In both cases,
Second containment gas in the 85mKr R/B increased by approximately an order of
magnitude during the heat-up. When the tempera-
Figure 22 HFR-EU1bis capsule schematic. Reproduced
tures were returned to the nominal level, R/B values
from Fütterer, M. A.; et al. Nucl. Eng. Des. 2008, 238(11), also returned to those that existed prior to the heat-up.
2877–2885. This result indicated that no particles had failed
Number of cells 2 1
Number of spheres per cell 2 in cell 1 3 in cell 2 5
Spherical fuel element diameter 60 mm 60.4 mm 60.4 mm
Fuel type LEU UO2 LTI – TRISO LEU UO2 LTI – TRISO LEU UO2 LTI – TRISO
Particle batch INET V000802 AVR HT 384–393 AVR HT 354–383
Number of particles per sphere 8500 9560 9560
Number of particles per capsule 17 000 28 680 47 800
TRISO-Coated Particle Fuel Performance 183
a
After 12 irradiation cycles.
Table 52 Characteristics of Chinese fuel during the heat-up tests. Following irradiation, 2014
Capsule 3 particles, approximately one-fourth of the
Batch number 1 2
total number in this sphere were examined by an
Coated particle irradiated microsphere gamma analyzer (IMGA),
Kernel form UO2 UO2 which identified one failed particle (most likely an
235
U enrichment (wt%) 17 17 initially failed particle). Ceramographic examinations
Kernel diameter (mm) 500.9 497.0
Kernel density (g cm3) 10.81 10.90
of particles from this capsule did not reveal any PyC
Buffer thickness (mm) 91 96 or SiC layer cracks, fission product attack of the SiC,
IPyC thickness (mm) 48.1 44.7 or kernel migration (amoeba effect).
SiC thickness (mm) 35.0 36.6 IMGA examination of 2117 Capsule 2 particles,
OPyC thickness (mm) 41.6 42.1 approximately one-fourth of the total, identified
Buffer density (g cm3) 1.02 1.05
IPyC density (g cm3) 1.84 1.84
two failed particles, which again, were most likely
SiC density (g cm3) 3.20 3.20 initially failed particles. Ceramography of Capsule
OPyC density (g cm3) 1.90 1.84 2 particles were similar to Capsule 3 particles where
IPyC BAF 1.03 1.02 no irradiation induced defects were found.
OPyC BAF 1.03 1.03 Following the irradiation phase of the experiment, a
Spherical fuel element
Fuel element diameter (mm) 60 60
high-temperature, in-pile heat-up test was conducted
Matrix density (g cm3) 1.73 1.72 with Capsule 5. This heat-up test was performed
U per fuel element (g) 5.00 4.98 in-pile because a hot cell furnace was not available.
U contamination 1.1 106 7.6 107 Capsule 5 was moved to the center high flux region
U-free/U-total 1.4 104 2.3 104 of the test reactor while the remaining capsules were
transferred to low flux reactor regions where fuel
a
Cutoff energy for fast neutron fluence not specified.
b
The tabulated end of life (EOL) R/B for capsule 5 was measured at the end of the irradiation test phase and prior to the start of the
high-temperature heat-up test.
184 TRISO-Coated Particle Fuel Performance
temperatures did not exceed 1000 C. Temperature in these conditions, phenomena such as kernel migra-
Capsule 5 was increased step-wise to the intended tion, metallic fission product diffusion through the
1600 C, held for approximately 4 h, and then reduced SiC layer, and fission product chemical attack of
step-wise after completion of the test. The total the SiC layer become limiting, particularly at higher
duration of the heat-up test was 50 h. Since the burnup. To address these service conditions with LEU
thermocouple in Capsule 5 failed during transfer, SiC TRISO fuel, the Japanese particle design has
temperatures were calculated where further analyses been optimized for a relatively low maximum burnup
indicated that test temperatures had exceeded the (e.g., larger kernel, thinner buffer), with a maximum
intended 1600 C. Gas release measurements during burnup limit for service in HTTR of 3.6% FIMA.
the heat-up indicated many particle failures. Postirra- The test data for the HTTR fuel particle design
diation IMGA examination of 2189 particles identi- are provided in Table 54. The burnups in most of the
fied that 127, or 6% of the particles sampled, had irradiations were very low, between 0.5% and 4%
failed. Ceramography revealed that most particles had FIMA. The fast fluence values were not available in
remained intact. However, the failed particles displayed all cases, but on the basis of 13 and 15 Oarai Gas Loop
several types of defects including radial and tangential (OGL-1) irradiations, the fast fluence was probably
SiC cracks, IPyC cracks, and thorough failures in all less than 0.5 1025 n m2 for all the irradiations,
layers. None of the observed particles displayed fission approximately an order of magnitude below values
product attack of the SiC or kernel migration. for commercial HTGRs. Defect levels in the fuel
showed continual improvement with each subsequent
irradiation, and the R/B values were very low.
3.07.2.6 Japanese Experience
Table 55 summarizes the conditions and results of
The Japanese have conducted a number of irradiation testing of the Japanese ‘high burnup’ fuel
irradiations of high quality LEU TRISO fuel for the design, plus HTTR fuel irradiated to extended
High-Temperature Test Reactor (HTTR), including burnup. The most recent irradiation was conducted
the Japanese fuel design for first and second core load- in the HFIR reactor at ORNL, designated HRB-22.
ings of HTTR and a ‘high burnup’ particle design, The range of burnup experienced by the fuel was
developed for later introduction into the HTTR. The 4.1–6.7% FIMA, with an average of 5.3% FIMA.
gas reactor development program in Japan has been Four particle failures (out of a total of 32 200 particles
focused on high-temperature process-heat applica- in the capsules) were observed during the irradiation
tions from its inception. In addition, the Japanese phase. This would produce an estimated population
‘pin-in-block’ prismatic fuel design has higher calcu- failure fraction of 2.8 104 at 95% confidence at a
lated maximum fuel temperatures for a given reactor fairly low burnup. In the high burnup fuel irradiation
coolant temperature than traditional prismatic gas in 91F-1A, conducted in the HTTR, Japanese fuel
reactor designs. As a result, the temperatures in the was taken to burnups of between 7% and 9% FIMA
HTTR and projected for commercial gas reactor at temperatures between 1120 and 1340 C. In-pile
applications are elevated, with time averaged peak gas-release and postirradiation leaching indicated 4
fuel temperatures in the range of 1350 C. Under particle failures, two in the higher temperature upper
Tests OGL-1
Source: Minato, K.; et al. HRB-22 Capsule Irradiation Test for HTGR Fuel; JAERI-research 98-021; 2008; Sawa, K.; et al. JAERI-research
2001-043; Irradiation Test of High Burnup Coated Fuel Particles for High Temperature Gas-cooled Reactor (91F-1A Sweep Gas Capsule
Irradiation Test); 2001; Sawa, K.; et al. JAERI-research 2002-012; Acceleration Irradiation Test of First-loading Fuel of HTTR up to High
Burnup; 2002; Hayshi, K.; et al. JAERI-research 2000-001; Irradiation Experiments of the 13th–15th OGL-1 Fuel Assemblies; 2000.
capsule, and two in the lower capsule. Pressure vessel Figure 23 plots EOL online gas-release (R/B) as a
failure was the stated cause of the failures. The 95% function of temperature, burnup, and fluence sepa-
confidence value for particle failures during the rately, and no trend is evident. Releases observed in
irradiation was 4.6 104. In irradiation 94F-9A con- the experiments are a function of (a) the initial level
ducted in the Japan Material Test Reactor, TRISO of heavy metal contamination in the fuel, and
fuel was irradiated at 1200 C to burnups between (b) in-service failures.
5.5% and 7% FIMA. No failures occurred in one
capsule, and in the other capsule, only one particle 3.07.2.7.1 Heavy metal contamination
out of 50 000 failed. Another irradiation of ‘high BOL 85mKr R/B is often used as a metric of
burnup’ fuel was conducted in the OGL-1 loop to a as-manufactured fuel quality. Fuel developed by
burnup of 1% FIMA with very low fast neutron German researchers exhibited overall low initial
fluence with a failure fraction of 1 104 observed defect fractions, with measured R/B values ranging
during a temperature transient test to 1500 C. between 1010 and 106. These results also demon-
strated that fuel quality improved over time, as was
the case for fuel used in the AVR. It should be noted
3.07.2.7 Irradiation Performance Summary
that the lowest ever 85mKr R/B measured in an
Extensive TRISO-coated particle irradiations have in-reactor irradiation is from German fuel (1010).
been performed around the world over the past Both the Chinese and Japanese fuel exhibited slightly
40 years. The majority of the particles have been irra- higher levels of contamination, as indicated by the
diated in the United States and Europe/Germany, initial R/B ratios in their irradiations. By contrast, the
with smaller amounts of fuel irradiated by Chinese high gas release from the historic US database (all
and Japanese programs. A summary of salient features experiments except AGR-1) is a combination of
of all the irradiations in which R/B measurements high initial levels of heavy metal contamination in
were available is found in Table 56. The irradiation the fuel and in-service failures. During the 1970s and
experiments reviewed here, including many others 1980s, the lower fuel quality of the historic US fuel
not summarized above, contained particle fuels with resulted in R/B ratios that varied between approxi-
widely different coatings and kernels and were con- mately 2 108 and 104, indicating that the initial
ducted at a variety of burnups, temperatures, and contamination levels in US fuel displayed great varia-
fluences. bility and could be rather high. The level did not
186 TRISO-Coated Particle Fuel Performance
Figure 23 TRISO-coated fuel performance as a function of temperature, burnup, and fast fluence.
significantly change until the early 1990s, when seri- resulted in the observed high gas-release (R/B),
ous efforts at reduction of initial contamination were which is approximately a factor of 1000 times greater
undertaken in the fabrication campaigns. With changes than that of the German fuel under a broad range of
to fabrication processes, the most recent US irradiation, conditions (i.e., temperature, burnup, and fluence).37
AGR-1, is demonstrating very low levels of contami- The current US fuel in AGR-1 is the culmination of
nation with initial R/B ratios in the range of 108.31 a significant effort to overcome the limitations
observed in past irradiations. Fabrication processes
3.07.2.7.2 In-service failures were changed, and irradiations were not accelerated.
The behavior of German fuel under irradiation is The result is a fuel that has survived irradiation to
excellent. Measured EOL 85mKr R/B values ranged high burnup with no failures.
between 3 109 to 105 for the various German The smaller statistical databases associated with
irradiation experiments. In most cases, the maximum the fuel from China, Japan, and Europe are more
R/B was measured at EOL; however, in some irradia- mixed, with some irradiations indicating good per-
tions, the final portion of the experiment was con- formance, and some indicating fuel failures. In many
ducted at lower temperatures, which caused the R/B cases, the root cause of the failures is unclear because
to decrease. Nevertheless, these results indicate that of less extensive PIE.
no fuel particle had failed while under irradiation.
PIE results confirmed the more extensive and more 3.07.2.7.3 Failure mechanisms
reliable gas-release data. Out of approximately Review of the US irradiation database indicates that a
380 000 UO2 and 80 000 (Th, U)O2 particles tested, number of different failure mechanisms in the indi-
there were no in-pile failures and only a few ‘dam- vidual layers of the TRISO coating contributed
aged’ particles due to experimental anomalies. Gas to the less-than-satisfactory US fuel performance.
release was attributed to as-manufactured defects and Failures of the coating layers were attributed to
heavy metal contamination only. (a) pressure vessel failure, (b) kernel migration
The historic US fuel irradiations indicate that in- (amoeba effect), (c) fission product attack of the SiC
reactor fuel failures occurred at the percent level. layer, (d) irradiation-induced IPyC cracking and/or
PIEs of the historic US fuel revealed in most experi- debonding leading to cracking in the SiC layer,
ments up to percent level particle failures, with very and (e) matrix–OPyC interaction and irradiation-
high levels of individual layer failures. These failures induced OPyC failure.
TRISO-Coated Particle Fuel Performance 187
Fission product and impurity attack of the SiC and AGR-1 fuel taken to 19% FIMA demonstrates that
kernel migration are thermally driven phenomena the changes made to the process were correct and
strongly influenced by burnup, temperature, and the resulted in excellent in-pile fuel performance.
temperature gradient across the particle. The tem-
perature gradient is a strong function of the power 3.07.2.7.4 Acceleration effects
density in the fuel body, which is directly related to The rate of accumulation of burnup and fast fluence
the level of acceleration in the experiment. relative to that expected in the reactor (i.e., the
The PyC-related mechanisms are strongly related degree of acceleration) is an important factor in
to the anisotropy and porosity in the coatings. The the behavior observed in these experiments. For
anisotropy has a strong influence on the shrinkage most of these fuels, the time to reach goal burnup
and swelling behavior of the PyC layers under irra- and fast fluence is 1095 days (3 years), whereas the
diation. However, the anisotropy measurements, irradiation times varied greatly. For historic US
especially optical methods (OPTAF), are not reliable irradiations, acceleration varied between factors of
predictors of PyC failure under irradiation as 2–10. One Japanese irradiation (HRB-22) was also
indicated by the lack of correlation between the greatly accelerated. By contrast, the German, Chi-
measured BAF and PyC failure (e.g., OF-2, HRB-5, nese, and European irradiations, as well as the US
HRB-6), and the high level of PyC failure observed AGR-1 irradiation, were all generally accelerated by
in most historic US irradiations. The fabrication only factors of two to three. These low acceleration
processes in the United States have been changed, factors limited thermal gradients across the fuel,
with the introduction of more accurate methods to which could have jeopardized fuel integrity with
measure anisotropy to improve the survivability of enhanced fission product attack and kernel migra-
the PyC layers. The lack of fuel failure in the US tion (Table 56).
Germany
R2-K12/1 HEU (Th, U)O2 TRISO 308/3 1100 11.1 5.6 0.300
R2-K12/2 1280 12.4 6.9 0.200
R2-K13/1 HEU (Th, U)O2 TRISO 517/2 1170 10.2 8.5 0.070
R2-K13/4 980 9.8 6.8 0.050
BR2-P25 HEU (Th, U)O2 TRISO 350/3 1070 15.6 8.1 1.000
HFR-P4/1 LEU UO2 TRISO 351/3 940 14.7 8.0 0.080
HFR-P4/2 945 14.9 8.0 0.080
HFR-P4/3 1075 14.0 8.0 0.008
SL-P1 LEU UO2 TRISO 330/3 794 11.3 6.8 1.200
HFR-K3/1 LEU UO2 TRISO 359/3 1200 7.5 4.0 0.200
HFR-K3/2 920 10.0 5.8 0.100
HFR-K3/3 920 10.6 5.9 0.100
HFR-K3/4 1220 9.0 4.9 0.300
FRJ2-K13/1 LEU UO2 TRISO 396/2.75 1125 7.5 0.2 0.020
FRJ2-K13/2 1150 8.0 0.2 0.020
FRJ2-K13/3 1150 7.9 0.2 0.007
FRJ2-K13/4 1120 7.6 0.2 0.007
FRJ2-K15/1 LEU UO2 TRISO 533/2 970 13.2 0.2 0.010
FRJ2-K15/2 1150 14.6 0.2 0.005
FRJ2-K15/3 990 13.9 0.1 0.003
FRJ2-P27/1 LEU UO2 TRISO 232/4.7 1080 7.6 1.4 1.600
FRJ2-P27/1 1320 8.0 1.7 10.000
FRJ2-P27/1 1130 7.6 1.3 0.120
HFR-K5/1 LEU UO2 TRISO 563/2 Cycled 6.7 2.9 <0.3
HFR-K5/2 proof test 8.8 <4.3 <0.3
Continued
188 TRISO-Coated Particle Fuel Performance
Table 56 Continued
Note: US fluence is for E > 0.18 MeV, fluence is not specified by China, and fluence values of all others are for E > 0.10 MeV.
TRISO-Coated Particle Fuel Performance 189
3.07.3 Safety Testing and fuel fabrication processes (e.g., SiC coating rate)
on fission product release are investigated.
The release of fission products from TRISO-coated
irradiated fuels heated at elevated temperatures to
3.07.3.1 Facility Overview
simulate accident conditions is reviewed in this sec-
tion. For a small HTGR, the temperature evolution Several operational facilities currently exist for
during a loss of coolant accident with complete testing high-temperature fission product retention
depressurization is calculated to reach 1600 C performance of irradiated TRISO fuel. The three
(including a 100 C uncertainty margin) for approxi- primary facilities discussed here (KüFA, Fuel Acci-
mately 30 h, as shown in Figure 24. The preponder- dent Condition Simulator (FACS), and Core Con-
ance of the experimental data is from tests with fuels duction Cooldown Test Facility (CCCTF)) are all on
of German manufacture, but insights are also the basis of the basic concept of a refractory metal
provided from a few experiments using Japanese and (tantalum) furnace operating at maximum tempera-
US fuels. There is some evidence of particle failure tures of 1800–2000 C with a helium atmosphere
caused by internal pressure (pressure vessel failure), and a water-cooled cold finger intruding into the
but of most interest is degradation of the SiC layer in hot zone to collect condensable fission products on
the TRISO coating during accident performance test- an exchangeable plate connected to the end of the
ing at elevated temperatures, 1500–1800 C. Because cold finger.
releases of cesium are sensitive to the integrity of the
SiC layer, much attention is paid to this fission product. 3.07.3.1.1 KüFA at ITU
Releases of krypton are sensitive to pressure vessel Initial development on KüFA (from the German
failure, but otherwise trail releases of cesium due to Kühlfingerapparatur, ‘cold finger device’) began in
holdup by PyC layers. Releases of other fission pro- the early 1980s at Forschungszentrum Jülich.38 The
ducts, such as strontium, europium, and cerium are system was designed to simulate reactor accident
treated where data are available. Silver, although not a conditions by heating spherical fuel elements in
safety concern because of its low yield, has potential helium to temperatures of up to 1800 C while mea-
consequences for reactor maintenance and diffuses suring the release of condensable fission products, as
readily through the SiC layer, even at 1200 C. The well as noble fission gases. The system was used in the
influences of irradiation characteristics (e.g., fuel 1980s and 1990s to test high-temperature fission
burnup, neutron fluence, and irradiation temperature), product retention on various irradiated particle fuel
SiC microstructure (e.g., grain size and orientation), specimens.1,38,39 More recently, the system was
2000
Accident simulation tests
1800
1600
Temperature (⬚C)
1400
1200
MODUL, max. fuel element
temperature in hypothetical
1000 and design basis accident
*including 100 ºC margin
800
Reactor operation
600 Temperature
400
0 100 200 300 400 500
Time (h)
Figure 24 Temperature evolution during a depressurized reduction cooling of a small HTR, and in heating tests with
irradiated fuel elements. Reproduced from Schenk, W.; et al. J. Nucl. Mater. 1990, 171(1), 19–30.
190 TRISO-Coated Particle Fuel Performance
transferred from Jülich, upgraded, and installed in the efficiency of the first. The activity in the traps is
hot cells at the Institute for Transuranium Elements monitored continuously with NaI detectors during
(ITU) in Karlsruhe for continued accident testing.40 a heating test.
The system consists of a double-walled, water- Upgrades to this system since the original work
cooled metal jacketed furnace with tantalum electrical- at Jülich include an increased maximum temperature
resistance heating elements (maximum power of capability to 2000 C, updated temperature regulation
70 kW at 2400 A), tantalum hot zone components, system, and a redesigned cold finger to allow sufficiently
and a water-cooled cold finger that holds an low plate temperatures during operation of the furnace
exchangeable metal plate to condense fission pro- at up to 1800 C. The upgraded KüFA system has re-
ducts during operation (see Figures 25 and 26). cently been described in further detail elsewhere.40
The cooling system maintains the plate at a temper-
ature of less than 100 C during furnace operation. 3.07.3.1.2 INL’s FACS
The cold finger can be removed from the furnace The FACS furnace at the INL is designed to heat fuel
through a sliding valve and gas lock system that specimens at temperatures up to 2000 C in inert
maintains the purity of the gas in the furnace during atmospheres while monitoring the release of volatile
exchange operations. Plates can then be packaged fission metals and fission gases (see Figure 27). The
and removed from the facility for gamma spectro- furnace has tantalum hot zone components and is
scopic counting or for acid leaching and radiochemi- heated using a graphite resistance element. The fuel
cal analysis. The entire system is operated inside an specimen is heated inside an 85 mm diameter tanta-
alpha containment box which itself resides inside a lum flow tube and is supported in a tantalum holder
shielded hot cell. Fission gases in the sweep gas that sits atop a tantalum tube inserted from the
circuit are captured in two charcoal cryo-traps, with bottom of the furnace. The fuel specimen tempera-
the second trap serving to indicate the trapping ture is monitored with a Type C (W–Re) thermo-
couple housed within the tantalum tube that
supports the sample holder, with the junction located
12.5 mm below the specimen. The hot zone can
accommodate samples up to 75 mm in diameter.
Special sample holders have been designed to
accommodate the compacts or spherical pebbles,
although any specimen geometry can be accommo-
dated through design of a custom tantalum holder.
The furnace main chamber is double-walled,
water-cooled stainless steel. Both the top and bottom
flanges are water-cooled stainless steel and can be
removed from the main chamber for sample loading
or maintenance operations as needed. The water-
cooled cold finger is of aluminum construction and
holds a 35 mm diameter plate on the end by means of
a mechanical lock using a rotary actuator. The tem-
perature at the base of the cold finger, where the
condensation plate is attached, is monitored using a
Type J thermocouple. Figure 28 depicts a cutaway
view of the furnace system. Condensate plates are
introduced to the furnace through a separate trans-
fer chamber, purged with helium prior to opening a
gate valve to the main chamber. The entire conden-
sate plate exchange operation is performed automati-
cally by computer control. The operator must only
remove the used plate once it has been retracted into
Figure 25 The cold finger device KüFA installed in the hot the transfer chamber and replace it with a fresh plate
cells at ITU. © European Atomic Energy Community, 2011. to be placed on the cold finger during the next
See http://itu.jrc.ec.europa.eu. exchange cycle.
TRISO-Coated Particle Fuel Performance 191
Coldfinger jacket
(vacuum/He filled)
Valve
Water cooled
Removable
condensation plate
Heat element in Ta
Steel heat sink
Spherical fuel
element
W/RE–thermocouple
Vacuum pump
connection
Helium output Helium input
Power
connection
Figure 26 KüFA system at ITU, Karlsruhe. Reproduced from Freis, D.; et al. Post irradiation test of high temperature reactor
spherical fuel elements under accident conditions. In 4th International Topical Meeting on High Temperature Reactor
Technology, Washington, DC, USA, October 1–4, 2008.
The helium sweep gas is introduced to the furnace This system, located outside of the hot cell, consists
from the bottom and injected into the hot zone of dual liquid nitrogen-cooled charcoal traps to con-
beneath the specimen and flows toward the conden- tain the noble gases, each with a high purity germa-
sation plate at the end of the cold finger as shown in nium detector and adjustable collimator. A schematic
Figure 28. The helium is then swept from the fur- diagram of the cooling water and He gas circuits is
nace through particulate and moisture filters and shown in Figure 29.
carried to the fission gas monitoring system for anal- Condensation plates that are removed from the
ysis of 85Kr, 133Xe, or other radioactive noble gases. furnace during operation will be packaged in clean
192 TRISO-Coated Particle Fuel Performance
Transfer
chamber
Condensation
plate cradle
Heat shields
Sample holder
Figure 28 Three-dimensional cut away diagram of the FACS main chamber and interior components.
He to condensation
plate release
Cold finger
Top cover
Manifold
Building
Heat Chamber pump
cooling
exchanger Bottom cover
water Return
Electrodes
Exhaust
Fission gas
monitoring
Flow
system
meters Particulate
FM
filter
FM
Manifold
FM Bottom cover
FM Chamber
FM Electrodes
FM Exhaust
He to sample
flow tube
He to main
chamber
Cold finger
Spool piece
Gate valve
assembly
Sweep Sweep
gas out gas in
Deposition
cup Furnace
insulation
Tantalum
can
Graphite
heating
element Water-
cooled
furnace
Specimen jacket
assembly
(graphite)
Graphite
holder
Graphite
Dual composite Figure 31 Core Conduction Cooldown Test Facility.
type 'C' reflector
thermocouple
Ionization
chamber H O H2O Tritium
2 CO
Chart
recorder H2O bath
Furnace
1 Gas 2 Heated pipe
Figure 32 Schematic diagram of the KORA facility developed at Forschungzentrum Juelich in the 1990s. Reproduced from
IAEA. Fuel Performance and Fission product Behavior in Gas-cooled Reactors. IAEA TECDOC-978; IAEA: Vienna, Austria,
November 1997.
Table 57 Results of accident simulation tests with irradiated spherical fuel elements
Table 58 Results of accident simulation tests at 1600–1800 C with irradiated fuel compacts
100 10–2
Compact
2100 ⬚C Fuel element
10–1
10–3
10–2
10–4
fractional release
fractional release
10–3
14% FIMA
1800 ⬚C
1700 ⬚C
85Kr
85Kr
10–5
10–6
10–8 10–8
0 100 200 300 400 500 0 100 200 300
(a) Heating time (h) (b) Heating time (h)
10–2
1800 ⬚C
10–3
fractional release
1700 ⬚C
10–4
10–5
85Kr
1600 ⬚C
10–6
10–7
0 100 200 300
(c) Heating time (h)
Figure 33 Accumulated fractional release of 85Kr as a function of heating time at constant temperature; (a) spherical
elements at 1600–1800 C, (b) compacts of 8–14% FIMA at 1600 C, and (c) compacts in the range 10–12% FIMA at
1600–1800 C. Reproduced from Schenk, W.; et al. J. Nucl. Mater. 1990, 171(1), 19–30.
198 TRISO-Coated Particle Fuel Performance
with defective SiC, while, in both cases, the pyrocar- of heating as considerable silver was released from
bon layers remain intact. fuel particles to the matrix during irradiation at tem-
Integral fractional releases of silver, cesium, kryp- peratures in the range 1000–1200 C for 359 days.
ton, and strontium are shown as a function of time at The release of cesium is considerably greater than
1600 C for sphere HFR-K3/1 in Figure 35. The the release of krypton and strontium, because
release of silver is on the order of 1–2% at the outset krypton is held up by the PyC and strontium is
retained in the UO2 kernel and graphite matrix to a
greater extent than cesium. In Figure 35, the radial
profiles of silver and cesium in the graphite matrix
10–03
exhibit strong concentration gradients typical of
137Cs release AVR 88/33, 8.5
85Kr release AVR 82/20, 8.6 materials that are diffusing, whereas the strontium
AVR 71/22, 3.5
10–04 AVR 88/15, 8.7 profile is much more flat, indicating little diffusive
One particle, 100% release
release from the matrix. In this same figure, the
release of cesium is observed to climb strongly after
Release fraction
HFR-K3/1.75% FIMA
100 4.0 ⫻ 10 m (E > 16 fJ)
25 –2
100
1020–1200 ⬚C
10–1 10–1
Fractional of fission products in matrix
110mAg
10–2 90Sr
10–2
137Cs
Fractional release
After 500 h
10–4 10–4 at 1600 ⬚C
10–5 10–5
137Cs
10–6 10–6
85Kr
Fuel zone
10–7 10–7
90Sr
10–8 10–8
0 100 200 300 400 500 –30 –10 0 10 30
Heating time at 1600 ⬚C (h) Fuel element
radius (mm)
Figure 35 Fission product release and distribution in sphere HFR-K3/1 after irradiation at 1000–1200 C for 359 days and
1600 C heating. Reproduced from Schenk, W.; et al. J. Nucl. Mater. 1990, 171(1), 19–30.
TRISO-Coated Particle Fuel Performance 199
100
the range 1900–2500 C is thermal decomposition
1800 ⬚C of the SiC layer.50
100 mm 50 mm 20 mm
Figure 37 Ceramographic sections through particles heated at 1600 C (complete particle followed by enlarged views of
SiC layers) showing increasing degradation of the SiC layer with increasing burnup. Reproduced from Schenk, W.; et al.
J. Nucl. Mater. 1990, 171(1), 19–30.
these do not appear to have significantly affected the cycle to a value of 1 103. The release further
test results. No particle failures were detected during increased during the second 1800 C heating cycle,
this test. Fractional releases at the end of the test were ending with a final value of 0.03. The results of the
6 106 for 85Kr and 8 106 for 137Cs. heating test are shown in Figure 39. SEM character-
ization of the particles at the end of the test revealed
3.07.3.3.3 HFR K6/3 failed particles with cracked SiC layers.
This heating test was carried out in four stages: The results indicate good particle performance at
1600 C for 100 h, 1700 C for 100 h, 1800 C for up to 1700 C, with significant Cs release through
100 h, and 1800 C for 300 h. The temperature was intact coatings starting at 1800 C and multiple par-
returned to room temperature in between stages. 85Kr ticle failures starting with the second 1800 C cycle.
fractional release was low (<105) until the start of It is unclear if thermal cycling was responsible for the
the fourth heating cycle, when a drastic increase observed particle failures.
in 85Kr was observed. By the end of the fourth heating
cycle, the 85Kr fractional release was 4 104, indi- 3.07.3.3.4 HFR K6/2
cating several particle failures. The 137Cs fractional This heating test was performed in two stages: the first
release was low (106) after the first two heating was at 1600 C for 100h followed by a second stage at
cycles but increased dramatically during the third 1800 C for 200 h. The temperature was not cycled to
TRISO-Coated Particle Fuel Performance 201
10–1
SiC PyC Kernel PyC SiC
36 33 34 31 28
LTI Buffer Buffer LTI
24 layer layer
41 10 18 18 10–2
78 45
Ba
Parts of weight
41 74 48 21
10–3
36
78 Ru Ag Ru
50 78
77 26 13 10–4
81
26
78 47 34 30
10–5
0 10 20 30 40 50 60 70 80
39 (b) Measure points radial through coating
10–1
(a) SiC PyC Kernel PyC SiC PyC
PyC LTI Buffer Buffer LTI LTI
layer layer
Pd Cs Pd
10–1 PyC SiC PyC Kernel PyC SiC PyC 10–2
LTI LTI Buffer Buffer LTI LTI
Parts of weight
layer layer
Cs
Parts of weight
10–2 Cs 10–3
Pd Pd
Cs
10–3 10–4
10–4 10–5
0 10 20 30 40 50 60 70 80 0 10 20 30 40 50 60 70 80
(c) Measure points radial through coating (d) Measure points radial through coating
Figure 38 Microprobe profiles of fission product elements through coatings of particles from HFR-K3. (a) Arrangement of
sectioned particles (HFR-K3/3) for microprobe measurements. The numbers show the percentage loss of cesium from each
particle after heating at 1800 C. (b) Barium, ruthenium, silver profiles in a particle with 78% cesium loss after 1800 C test.
(c) Cesium, iodine, palladium profiles in a particle from HFR-K3/1 (0.01% cesium loss from sphere) after 1600 C test.
(d) Cesium, iodine, palladium profiles in a particle (78% cesium loss) from HFR-K3/3. Reproduced from Schenk, W.; et al.
J. Nucl. Mater. 1990, 171(1), 19–30.
Table 59 Averaged fission product distribution for spherical fuel elements after accident simulation tests
Fuel type Heating Time at Nuclide Fractional fission product content in: Fractional release
temperature temperature from fuel element
( C) (h) Kernel Coating Matrix
137
UO2 TRISO 1600 Up to 500 Cs 5 101 5 101 2 105–1 103 5 105
90
Sr 9.5 101 5 102 2 103–5 103 1 106
110m
Ag 8 101 2 101 9 104 1 103–3 102
137
UO2 TRISO 1800 Up to 200 Cs 2 102 6 101 1.5 101 5 102
90
Sr 7 101 8 102 2 101 3 103
110m
Ag 9 102 2 101 3 102 7 101
Source: Freis, D.; et al. Post irradiation test of high temperature reactor spherical fuel elements under accident conditions. In 4th
International Topical Meeting on High Temperature Reactor Technology, Washington, DC, USA, October 1–4, 2008.
202 TRISO-Coated Particle Fuel Performance
Table 60 Fuel description/irradiation conditions for spheres accident tested in KüFA at ITU
Fuel element AVR 73/21 AVR 74/18 HFR K6/3 HFR K6/2
Source: Freis, D.; et al. Post irradiation test of high temperature reactor spherical fuel elements under accident conditions. In 4th
International Topical Meeting on High Temperature Reactor Technology, Washington, DC, Oct 1–4, 2008.
room temperature at the end of the 1600 C test, but 2000 0.1
several interruptions of the experiment during the first 1800
stage occurred because of system malfunctions. The 0.01
1600
results of this heating test are shown in Figure 40. 1400
Fractional release
Temperature (⬚C)
1E-3
The 85Kr fractional release was below the detection 1200
limit of the system until 90 h into the 1800 C cycle, 1000 1E-4
when the release increased up to 107. Approxi-
800
mately 120 h into the 1800 C cycle, another continuous 1E-5
release was observed that reached approximately 105 600
Temperature (⬚C)
by the end of the test. 137Cs was released fairly contin- 400
85 1E-6
Kr
uously throughout the test up to a 2 103 fractional 200 137
Cs
release at the end of the test. 0 1E-7
0 100 200 300 400 500 600 700 800
In general, these initial tests demonstrated excel-
Time (h)
lent fuel performance at 1600 and 1700 C with zero
particle failures. Higher 137Cs and 85Kr releases were Figure 39 Results of heating test HFR K6/3 at ITU.
Reproduced from Freis, D.; et al. Post irradiation test of high
observed at 1800 C, with some particle failures
temperature reactor spherical fuel elements under accident
occurring for the sphere HFR K6/3. conditions. In 4th International Topical Meeting on High
Temperature Reactor Technology, Washington, DC, USA,
October 1–4, 2008.
3.07.3.4 US Experience and Future Plans
3.07.3.4.1 Past experience
Very little work has been done on accident testing of burnup was in the range of 21–25% FIMA and the
US fuels as most irradiations produced significant fast neutron fluence was in the range of 3.4 –
irradiation-induced particle failure fractions. How- 5.5 1025 n m2. The irradiation was accelerated
ever, postirradiation anneals of long durations at with a residence time of 169 effective full power
temperatures up to 1500 C were performed to accel- days. Only the fission product release data at
erate diffusive fission product releases from a variety 1500 C are discussed here, as cesium was not
of fuel types.52 UO2, UC2, UCO, UO2*(1), and UO2*(2) released at the lower temperatures. No fission prod-
fuel particles were irradiated in the HRB-15B cap- uct releases were measured at any temperature from
sule in HFIR.52 In the UO2*(1) fuel, the kernel was UO2*(1) fuel particles. Ten particles of each fuel type
coated with a ZrC layer, while in the UO2*(2) fuel, were annealed for 11 866 h at 1500 C. Integral
ZrC was dispersed in the buffer layer surrounding releases for each 10-particle batch were measured
the kernel. These changes were intended to control from individual particles by gamma counting each
free oxygen released during fission, which should particle before and after the test and by periodic
improve fission product retentiveness. The fuel gamma monitoring of fission product collectors
TRISO-Coated Particle Fuel Performance 203
during the anneal as a function of time. The agree- of the releases of cesium and silver from the UO2 fuel
ment of the integral releases from each 10-particle particles in Figures 41 and 42 indicate a diffusion
batch by these two methods was excellent. release mechanism through the SiC layer. However,
Cesium was released from only the UO2 and UC2 the release of cesium from the UC2 fuel batch is
fuel particles as is shown in Figure 41. These same sudden in Figure 41, and the release of silver shows
two fuel batches released the greatest fractions of a rapid increase at the same time as the sudden
silver as illustrated in Figure 42. The time signatures release of cesium, as pointed out in Figure 43.
The distribution of fission product releases among
particles within the fuel batches in Table 61 indicates
2000 0.01 that the release of cesium from the UO2 fuel particles
1800
is from two of the ten particles and from only one
1E-3 particle in the UC2 fuel batch. This same table shows
1600
that the release of silver was 100% from the UO2 fuel
1400
1E-4
Fractional release
batch, and 82% from the UC2, with 6 of the 10 UC2
Temperature (⬚C)
1200
particles releasing 100% of their silver inventories,
1000 1E-5
two particles releasing 85–95%, one particle releasing
800 50%, and one particle retaining 100%. Particle-to-
1E-6
600 particle variations in fission product release dominate
400
1E-7 the behavior implied by the data shown in Table 61.
200 The microstructures in Figure 44 show that the
0 1E-8 SiC layer in the UO2 batch exhibits large columnar
0 50 100 150 200 250 300 350 grains, whereas the UCO batch exhibits a strong
Time (h)
laminar grain structure in the SiC. The UC2 and
Temperature (⬚C)
85
UO2*(1) batches exhibit laminar structures in the
Kr
137
Cs
SiC that are somewhat weaker than in the UCO
batch. The results in Table 61 indicate that silver
Figure 40 Results from heating test HFR K6/2 at ITU. release at 1500 C is greatest (100%) in the case of
Reproduced from Freis, D.; et al. Post irradiation test of high
columnar SiC, least (3%) for strongly laminar SiC,
temperature reactor spherical fuel elements under accident
conditions. In 4th International Topical Meeting on High and intermediate (82%) for somewhat less strong
Temperature Reactor Technology, Washington, DC, USA, laminar SiC. Although Cs was released from only
October 1–4, 2008. three of the 50 particles annealed at 1500 C, two
102
101
Cesium release (%)
100
134Cs
137Cs
10−1
0 10 20 30 40 50
Annealing time (Ms)
Figure 41 Release of Cs from various types of TRISO-coated fuel particles at 1500 C. Reproduced from Bullock, R. E.
J. Nucl. Mater. 1984, 125(3), 304–319.
204 TRISO-Coated Particle Fuel Performance
102
UO2
UC2
UO2* (2)
UCO
100
10−1
0 10 20 30 40 50
Annealing time (Ms)
Figure 42 Release of 110Ag from various types of TRISO-coated fuel particles at 1500 C. Reproduced from Bullock, R. E.
J. Nucl. Mater. 1984, 125(3), 304–319.
102
Ag
101 Cs
Release (%)
100
10−1
0 10 20 30 40 50
Annealing time (Ms)
Figure 43 Abrupt 10% increase in 110Ag release from UC2 particles at 1500 C when one of the ten test particles released
its entire Cs inventory.
particles had columnar SiC and one had a somewhat containing only UO2 in the kernel (9–16%). As
weak laminar SiC. The sensitivity of cesium release shown in Table 61, cerium release is 45% in UC2,
to SiC grain structure was recognized in Myers53 only 1% in UCO, and nil in UO2 particles. These
where the diffusivity of cesium through columnar behaviors are related to the thermodynamics of rare-
SiC was given as an order of magnitude greater earth oxides and carbides, according to Homan et al.,54
than through laminar SiC. The diffusion equations where oxides formed in UO2 (such as Eu2O3 and
from Myers53 are accessible in Table A-3 of IAEA.3 Ce2O3) are less likely to escape from the kernel than
As shown in Figure 45, releases of europium are are the more mobile rare-earth carbides formed in
greatest (37–46%) for the fuel batches containing UC2. In UCO, europium forms a carbide and cerium
UC2 in the kernel, compared with fuel batches forms an oxide.54
TRISO-Coated Particle Fuel Performance 205
Table 61 Distribution of fission product release within particle batches during postirradiation annealing
Annealing TRISO particle typea Release breakdown from the ten particles within a test batch for:
temperature 134 110m 154 144
( C) Cs Ag Eu Ce
137
Cs
1500 UC2 with laminar SiC 9 ¼ 0% 1 ¼ 0% 15% < 5 < 25% 12% < 3 < 18%
1 ¼ 99% 1 ¼ 50% 45% < 3 < 55% 18% < 3 < 25%
10 ¼ 12% 85% < 2 < 95% 2 ¼ 100% 70% < 3 < 80%
6 ¼ 100% 10 ¼ 46% 1 ¼ 99%
10 ¼ 82% 10 ¼ 45%
1500 UO2 with 8 ¼ 0% 10 ¼ 100% Uniform release of 16% 10 ¼ 0%
columnar SiC 2 ¼ 99%
10 ¼ 24%
1500 UC0.4O1.6 with laminar SiC 10 ¼ 0% 7 ¼ 0% Uniform release of 37% 10 ¼ 1%b
10% < 3 < 20%
10 ¼ 3%
1500 UO2*(2) with laminar SiC 10 ¼ 0% 7 ¼ 0% 2 ¼ 0% 10 ¼ 0%
70% < 3 < 80% 0% < 5 < 10%
10 ¼ 27% 15% < 3 < 25%
1350 UC0.4O1.6 with laminar SiC 10 ¼ 0% 10 ¼ 0% Uniform release of 23% 10 ¼ 0%
1350 UO2*(2) with laminar SiC 10 ¼ 0% 7 ¼ 0% 10 ¼ 4%b 10 ¼ 0%
45% < 3 < 55%
10 ¼ 19%
1200 UC0.4O1.6 with laminar SiC 10 ¼ 0% 10 ¼ 0% Uniform release of 6% 10 ¼ 0%
1200 UO2*(2) with laminar SiC 10 ¼ 0% 10 ¼ 2%b 10 ¼ 0% 10 ¼ 0%
a
There was zero release within about 5% as determined from individual particle counting before and after annealing for all isotopes
from each of the ten particles in all test combinations not listed, that is, UO2*(1) at all temperatures, and UC2, UO2, and UC0.4O1.6 at
1350 and 1200 C. As no release on collectors was detected at the 0.01% level from the combined ten particles within each of these test
batches, it can be assumed that release from any one of these particles was certainly <0.01% and was probably not more than 0.001%.
b
These total releases from ten particles were too small and too uniformly distributed among particles to allow the determinations of
individual release values.
These results,42 with admittedly relatively few be performed on the AGR-1 fuel with maximum
particles, indicate that under a long annealing time temperatures between 1600 and 1800 C. High-
at 1500 C (a) silver and cesium releases are at a temperature margin tests up to 2000 C may also be
maximum in the case of SiC with a columnar grain performed if warranted, based on the results of the
structure; (b) europium releases are largest in UC2 lower temperature tests. The primary released radio-
fuels, but can be significant in UCO and, to a lesser nuclides of interest are 85Kr, 137Cs, 134Cs, 110mAg,
90
extent, in UO2 fuels; and (c) cerium release is signifi- Sr, and 154Eu. (131I and 133Xe are also of interest,
cant only in UC2 fuel. but will require reirradiation of the fuel in a test
reactor to regenerate these isotopes because of rapid
3.07.3.4.2 Future plans decay times. Work is underway to develop this capa-
The NGNP/AGR fuel program is currently bility at INL.)
planning an extensive campaign of accident testing Condensate plate analysis will consist of initial
of irradiated TRISO fuel compacts to measure fission measurement of sample activity using gamma spec-
product retention at elevated temperatures and pro- trometry. Following this step, the fission products
vide verification of fuel performance code predic- will be chemically stripped from the plates, and the
tions. The first accident tests will be performed on solutions analyzed for other fission products of inter-
irradiated fuel compacts from the AGR-1 experi- est. The primary fission product of interest will be
90
ment, using both the FACS and CCCTF test equip- Sr, a beta emitter. The strontium can be separated
ment described earlier. Individual fuel compacts from the rest of the analytes in an ion exchange resin,
in each capsule will be chosen for accident tests and the 90Sr activity measured using scintillation
based primarily on their irradiation history (i.e., techniques. The feasibility of chemical separation of
burnup, fast fluence, and irradiation temperature). iodine from the solution has also been demonstrated,
Both isothermal and transient temperature tests will and the separated iodine solution can be analyzed for
206 TRISO-Coated Particle Fuel Performance
102
UC2
UCO
UO2
100
10−1
0 10 20 30 40 50
Annealing time (Ms)
Figure 45 Release of 154Eu from various types of TRISO-coated fuel particles of 1500 C. Reproduced from Bullock, R. E.
J. Nucl. Mater. 1984, 125(3), 304–319.
100 100
110 mAg
110mAg 137Cs
10–1 10–1
Fractional release
Fractional release
125Sb
10–2 154Eu 10–2 85Kr
154Eu
137Cs
10–3 85Kr 10–3
10–4 10–4
10–5
0 50 100 150 200 250 300 10–5
0 50 100 150 200 250
Heating time (h) Heating time (h)
Figure 46 Time-dependent fractional releases of fission Figure 47 Time-dependent fractional releases of fission
products during the ACT3 heating test at 1700 C for 270 h, products during the Act 4 heating test at 1800 C for 222 h,
obtained by the online measurements of fission gas-release obtained by the online measurements of fission gas-release
and intermittent measurements of metallic fission product and intermittent measurements of metallic fission product
release. Reproduced from Minato, K.; et al. Nucl. Technol. release. Reproduced from Minato, K.; et al. Nucl. Technol.
2000, 131(1), 36–47. 2000, 131(1), 36–47.
product release from particle to particle. At 1700 C, X-ray microradiographs, ceramography, and electron
silver release varies from 10% to 100%, cesium microprobe.
from 0% to 20%, and europium from 5% to 30%. Accumulations of fission products in the buffer
At 1800 C, not including the failed particle, the show up as bright spots in X-ray microradiographs
silver release varied from 24% to 100%, cesium of fuel particles from which large fission product
from 0% to 55%, and europium from 0% to 69%. releases were measured, as shown in Figure 50
Individual particles were examined at JAERI by (1700 C) and Figure 51 (1800 C). In Figure 50,
208 TRISO-Coated Particle Fuel Performance
0.8
0.6
0.4 144Ce
95Zr
106Ru
0.2
125Sb
0
0 5 10 15 20 25
(a) Particle number
Inventory ratio of post-/preheating tests
0.8
0.6
0.4
110mAg
0.2 137Cs
154Eu
0
0 5 10 15 20 25
(b) Particle number
Figure 48 Inventory ratios of post- to preheating tests in individual particles in ACT3 measured with the IMGA system: (a) 95Zr,
106
Ru, 125Sb, and 144Ce; (b) 110mAg, 137Cs, and 154Eu. Reproduced from Minato, K.; et al. Nucl. Technol. 2000, 131(1), 36–47.
1.2 1.2
110mAg 137Cs 154Eu
Inventory ratio of post-/
1.0 1.0
preheating tests
preheating tests
0.8 0.8
0.6 0.6
0.4 0.4
0 0
0 5 10 15 20 25 0 5 10 15 20 25
(a) Particle number (b) Particle number
Figure 49 Inventory ratios of post- to preheating test in individual particles in ACT4 measured with the IMGA system: (a) 95Zr,
106
Ru, 125Sb, and 144Ce; (b) 110mAg, 137Cs, and 154Eu. Reproduced from Minato, K.; et al. Nucl. Technol. 2000, 131(1), 36–47.
particle ACT3–5 had relatively little fission prod- bright spot in the buffer. A similar trend is shown in
uct release and shows no evidence of fission product Figure 51 for a relatively non-releasing particle
accumulation in the buffer, whereas particle (ACT4–3) and two strongly releasing particles
ACT3–6 had relatively large releases and shows a (ACT 4–9 and ACT 4–13). These figures also show
TRISO-Coated Particle Fuel Performance 209
Figure 50 X-ray microradiographs and ceramographs of the particles after the ACT3 heating test: (a) and (b) show particle
ACT3-5; (c) and (d) show particle ACT3-6. Reproduced from Minato, K.; et al. Nucl. Technol. 2000, 131(1), 36–47.
that the buffers of the low-releasing particles are gas-cooled reactor, is achievable if the following con-
intact, whereas those of the heavily releasing particles ditions are met:
are severely cracked. In addition, the kernels of the
a. Disciplined and controlled fabrication of the
releasing particles in Figures 50 and 51 exhibit
coated particles under conditions that produce
larger pores than the kernels of the nonreleasing
the proper microstructure of the kernels and coat-
particles. The SiC layers of all the particles show
ings with robust quality control to demonstrate the
signs of degradation as seen in Figures 50–52. Accu-
statistical adequacy of the pedigree of the material.
mulations of fission products, especially palladium,
Much of the poor performance observed in the
were found in all particles at the IPyC/SiC interface,
past is associated with improper fabrication
as shown in Figure 53, and sometimes within the
conditions.
SiC layer.
b. Selection of normal operation conditions that will
The Japanese work corroborates results from
result in very low failures. For a pebble bed reactor
Germany in that wide variations are measured in
using 500-mm UO2 TRISO fuel, an envelope of
fission product release from particle to particle, palla-
1150 C, 10% FIMA, and 4 1025 n m2 was
dium buildup at the IPyC/SiC interface is observed
demonstrated by German fuel developers in the
in both releasing and nonreleasing particles, and fis-
1980s and remains the standard for current pebble
sion product releases increase with increasing test
bed designs. For prismatic gas-cooled reactors
temperature.
using 350-mm UCO TRISO fuel, an envelope
of 1250 C, 19% FIMA, and 4 1025 n m2
3.07.4 Conclusions appears achievable, on the basis of the results of
the US AGR-1 irradiation.
TRISO-coated particle fuel has been fabricated and c. Well-planned series of irradiations that limit the
tested around the world. A review of the fuel perfor- burnup and fluence acceleration to preclude arti-
mance database indicates that high-quality low- ficial failures that would not be expected in the
defect TRISO fuel, the heart of the high-temperature actual reactor application, yet contain enough fuel
210 TRISO-Coated Particle Fuel Performance
100 mm 100 mm
(a) (b)
100 mm 100 mm
(c) (d)
100 mm 100 mm
(e) (f)
Figure 51 X-ray micrographs and ceramographs of the particles after the ACT4 heating test: (a) and (b) show particle
ACT4-3; (c) and (d) show particle ACT4-9; and (e) and (f) show particle ACT4-13. Reproduced from Minato, K.; et al. Nucl.
Technol. 2000, 131(1), 36–47.
to draw statistically significant conclusions about for UCO TRISO, but work is underway to begin
fuel performance. To meet an in-service failure such testing in the near future. The current acci-
probability of 104 at 95% confidence, the mini- dent database at 1700 and 1800 C is inadequate to
mum number of particles to be tested is approxi- determine with statistical confidence if TRISO-
mately 300 000, allowing for some failures during coated UO2 or UCO can sustain these higher acci-
the experiment. dent temperatures. Future R&D programs will
d. Selection of an accident envelope that will result investigate the fuel performance at these elevated
in very low failures. For both pebble bed and temperatures.
prismatic designs, limiting accident temperatures e. Well-planned accident safety testing that tests
in depressurized conduction cooldown scenarios enough fuel to draw statistically significant con-
to 1600 C is acceptable. The large accident data- clusions (population on the order of 200 000 par-
base developed by German fuel developers in the ticles) and measures all relevant radionuclides
1980s remains the standard for gas reactors world- important to the ultimate safety and licensing of
wide. There has been very limited accident testing the high-temperature gas-cooled reactor.
TRISO-Coated Particle Fuel Performance 211
20 mm Acknowledgments
(a)
SEI Pd
(a) 10 mm (b)
Figure 53 Electron probe microanalysis of coating layers of particle ACT4-3 after the ACT4 heating test: (a) secondary
electron image and (b) X-ray image for palladium. Reproduced from Minato, K.; et al. Nucl. Technol. 2000, 131(1), 36–47.
212 TRISO-Coated Particle Fuel Performance
40. Kostecka, H.; Ejton, J.; de Weerd, W.; Toscano, E. Post 46. Kendall, J. K.; et al. PBMR fuel performance envelope and
irradiation testing of HTR-fuel elements under accident test program, Pebble Bed Modular Reactor Limited,
conditions: Part II. In 2nd International Topical Meeting on Document No. 039116; 2007.
High Temperature Reactor Technology, Beijing, China, 47. Proksch, E.; Strigl, A.; Proksch, E.; Nabielek, H. J. Nucl.
Sept 22–24, 2004. Mater. 1982, 107(2–3), 280–285.
41. Morris, R. N.; et al. The core conduction cooldown test 48. Verfondern, K.; et al. Nucl. Technol. 1990, 91, 235–246.
facility: Current status and issues; ORNL/NPR-91/7; 49. Schenk, W.; Nabielek, H. Nucl. Technol. 1991, 96, 323–336.
Jan 1992. 50. Nabielek, H.; et al. Nucl. Technol. 1989, 84(1), 62–81.
42. Minato, K.; et al. Nucl. Technol. 2000, 131(1), 36–47. 51. Freis, D.; et al. Post irradiation test of high temperature
43. Morris, R. N.; et al. HRB-17 and HRB-18 HEU TRISO UCO reactor spherical fuel elements under accident conditions.
unbonded irradiated core conduction cooldown tests; In 4th International Topical Meeting on High Temperature
ORNL/NPR-92/9; Jun 1992. Reactor Technology, Washington, DC, Oct 1–4, 2008.
44. Kellerbauer, A.; Schenk, W.; Pott, G.; Nabielek, H. Private 52. Bullock, R. E. J. Nucl. Mater. 1984, 125(3), 304–319.
communication; Institute for Transuranium Elements; Sept 53. Myers, B. F. Cesium diffusion in silicon carbide during post
2, 2009. irradiation anneals; Technical Note: KFA-HBK-TN-01183;
45. Schenk, W.; et al. J. Nucl. Mater. 1990, 171(1), Research Center Jülich; 1984.
19–30. 54. Homan, F. J.; et al. Nucl. Technol. 1977, 35(2), 428–441.
3.08 Advanced Concepts in TRISO Fuel
K. Minato and T. Ogawa
Japan Atomic Energy Agency, Tokai-mura, Ibaraki, Japan
215
216 Advanced Concepts in TRISO Fuel
ORR Oak Ridge Research Reactor in a porous structure due to the b-SiC to a-SiC
PyC Pyrolytic carbon transformation and the thermal dissociation of SiC,9
R/B Release-to-birth ratio which leads to an extensive release of fission products
VHTR Very-high-temperature reactor from the TRISO-coated fuel particles. The fuel tem-
peratures were limited to well below 1973 K during
the design-basis accidents in HTGR designs.10–12
Chemical interaction of the SiC coating layer with
3.08.1 Introduction fission products is one of the possible performance
limitations of the TRISO-coated fuel particles. The
The TRISO-coated fuel particle consists of a micro- fuel performance of TRISO is described in Chapter
spherical fuel kernel and coating layers of porous 3.07, TRISO-Coated Particle Fuel Performance.
pyrolytic carbon (PyC), inner dense PyC (IPyC), The fission product of palladium is known to react
silicon carbide (SiC), and outer dense PyC (OPyC). with the SiC layer. Corrosion of the SiC layer could
The chemical form of the fuel kernel can be oxide, lead to fracture of the coating layers or provide a
carbide, or a mixture of the two. The function of localized fast diffusion path, which degrades the
these coating layers is to retain fission products fission-product retention capability within the parti-
within the particle. The porous PyC coating layer, cle. Since the fission yield for palladium from 239Pu is
called the buffer layer, attenuates fission recoils and about tenfold that from 235U,13 a careful particle fuel
provides void volume for gaseous fission products design should be made in the actinide burning.
and carbon monoxide in the cases of oxide and oxy- The PyC layer develops gas permeability with
carbide fuels. The IPyC coating layer acts as a con- increasing fast neutron doses. Intactness of the IPyC
tainment to gases during irradiation and protects layer is crucial in keeping the integrity of the SiC
the fuel kernel from the reaction with the coating layer in the oxide fuel. When the IPyC layer fails, or
gases during the SiC coating process. The SiC coat- develops gas permeability, the SiC layer will react
ing layer provides mechanical strength for the parti- with CO gas to form volatile SiO.14 The PyC layer
cle and acts as a barrier to the diffusion of metallic will also develop anisotropy above 2173 K, which is
fission products, which diffuse easily through the deleterious to its irradiation behavior.
IPyC layer. The OPyC coating layer protects the To improve the high-temperature performance of
SiC coating layer mechanically. the TRISO-coated fuel particles, a new material other
The recent interest in the coated particle fuel than SiC is needed. Zirconium carbide is a candidate,
concept includes its application outside the past and ZrC-coated particles, where the SiC layer was
experience of the high-temperature gas-cooled reac- replaced by a ZrC layer, have been tested. New con-
tor (HTGR)1: figurations of the coating layers, with a layer contain-
ing SiC or ZrC added to the TRISO coating, have
1. Very-high-temperature reactor (VHTR) with a
been proposed and tested to improve the chemical
gas outlet temperature of 1273 K for supplying
stability of the TRISO-coated particles. For the appli-
both the electricity and the process heat for hydro-
cation to the fast reactor fuel, TiN coating layers have
gen production, as proposed in the Generation-IV
been proposed and tested instead of PyC layers.
International Forum.2
The following sections summarize the designs
2. Actinide burning in deep-burn modular helium
and the research and development of the advanced
reactor (DB-MHR) with fuel kernels consisting of
concepts in TRISO fuel: (1) ZrC-coated particle
high concentrations of transuranium elements.3,4
fuel, (2) ZrC-containing TRISO-coated particle fuel,
3. Advanced gas fast reactor (GFR) with nitride fuel,
(3) SiC-containing TRISO-coated particle fuel, and
which aims at a more improved performance com-
(4) TiN-coated particle fuel.
pared with the conventional liquid metal fast
breeder reactor (LMFBR) and/or the efficient
actinide burning.1,5
3.08.2 ZrC-Coated Particle Fuel
Although SiC has excellent properties, it gradually
3.08.2.1 Designs of ZrC-Coated Particle Fuel
loses its mechanical integrity at very high tempera-
tures, especially >1973 K.6–8 The annealing tempera- Zirconium carbide (ZrC) is known as a refractory
ture during fuel element fabrication is limited to and chemically stable compound, which melts eutec-
2073 K for 1 h. Higher temperatures should result tically with carbon at 3123 K. The properties of
Advanced Concepts in TRISO Fuel 217
Scrubber
increasing the linear coating rate of ZrC. Increases in
Quartz
the ratio of CH4 to ZrCl4 in the coating gas resulted
in a decreased metallic appearance of the coating and
an increase in the C/Zr in the deposit. Increases in
H2 inhibited these effects. Carbon
The ZrC coating layers were prepared using a gas wool
mixture of C3H6, H2, ZrCl4, and Ar with the same Graphite
coater and ZrCl4 powder feeder as described ear-
lier.23 In general, ZrC coating layers made with
CH4 and C3H6 were similar and were affected simi-
larly by variations in the hydrocarbon and hydrogen Induction
concentrations. The coating layers produced using coil
C3H6 were more sensitive to changes in hydrogen
concentration than those produced using CH4.
The coating processes based on the in situ genera-
tion of zirconium halide vapor are developed at Japan
Atomic Energy Research Institute (JAERI; now Japan
Atomic Energy Agency, JAEA) to avoid the handling Alumina
of highly hygroscopic halide powder.17 Several pro-
Zr Br4
cesses were studied: the chloride process,24 the meth-
Quartz
ylene dichloride process,25 the iodide process,26,27
and the bromide process.28–31 Among these processes,
Furnace
Zr
and reliable; in this process, ZrBr4 is produced by
reacting bromine with zirconium sponge inside the
coater. ZrBr4 is preferred to ZrCl4 since the reaction
of excess chlorine with hydrogen is a potential explo-
sion hazard.
Figure 2 shows an apparatus for ZrC deposition
by the bromide process.32 In this process, bromine,
which is liquid at room temperature, was carried by
Br2 + Ar
CH4 + H2
Ar
Ar
Ar
The lower one is for the reaction of bromine with in liquid tetrabromoethane (C2H2Br4) by the density
zirconium sponge and the upper one, for the chemical difference. The specific gravity of tetrabromoethane
vapor deposition of ZrC at 1873 K in the maximum. is 2.965 Mg m3.
The off-gas treatment equipment removes soot, hydro- The density of the SiC layers is measured by
gen bromide, and residual hydrogen. the sink-float technique or a liquid gradient column,
in which a liquid having the same density as the
sample is needed. The density of the SiC layers
3.08.2.3 Characterization Techniques for
is around 3.21 Mg m3, and a liquid mixture of
ZrC-Coated Particle Fuel
methylene iodide (CH2I2) having a density of
It is important to characterize the key basic proper- 3.325 Mg m3 and benzene (C6H6) having a density
ties of the coating layers that are critical to the fuel of 0.8785 Mg m3, for example, is used for the mea-
performance. Although most of the characterization surement of density. In the case of the ZrC density
techniques used for the ordinary TRISO-coated par- measurement, no suitable liquid is present since the
ticle fuels can be applied to the ZrC-coated particle density of the ZrC layers is around 6.6 Mg m3, and
fuels, some techniques have to be developed primar- so, other techniques are needed. Gas pycnometry,
ily for the ZrC-coated particle fuels. which required at least 100 mg of the samples,33 was
In the case of the ordinary TRISO-coated fuel developed for the ZrC density measurement. The
particles, the PyC layers are burnt off to recover the stoichiometry of the ZrC layer affects the thermal
SiC fragments for characterization, such as density, conductivity,36 fission product retention,18,37 etc. The
composition, and strength measurements. However, properties of ZrC are summarized in Chapter 2.13,
it is almost impossible to separate the ZrC from the Properties and Characteristics of ZrC. Analysis of
PyC layers by the same method since ZrC, in contrast the free carbon is important for controlling the quality
with SiC, does not form a protective oxide layer, of the ZrC coating. Plasma oxidation with emission
resulting in oxidation of ZrC to ZrO2 when exposed monitoring was also applied to the quantitative analysis
to air at high temperatures. To begin with, a method of the free carbon in ZrC powder.38 The emission was
of obtaining fragments of the ZrC coating layer from monitored with an optical color analyzer and was cali-
the ZrC-coated particles is needed. brated with standard samples of ZrO2 + C mixtures.
The plasma oxidation method was developed to The oxidation rates of the free and the combined
obtain the ZrC fragments from the coating layers con- carbons are so different that it is possible to estimate
taining the PyC.34 The difference in the oxidation rates the amount of the former from the emission. With
between PyC and ZrC is very large. In this method, the powdered ZrC of about 10 mg, free carbon of <1 wt%
samples were set in the plasma oxidation apparatus, could be easily determined. Without this method, the
where low-pressure oxygen was ionized by high fre- composition of the ZrC was estimated by burning the
quency induction coupling. Plasma reaction was moni- ZrC and PyC together, weighing ZrO2 and CO2, and
tored by a color analyzer and an optical power meter. then subtracting the contribution of PyC from the total
The color changed from the pale violet of pure oxygen amount of CO2.34
to pale blue during vigorous oxidation of free carbon, A new method to measure ZrC stoichiometry was
and again to pale violet when the PyC was completely developed using the infrared light absorption during
removed and a very thin oxide scale was formed on the combustion in oxygen and the inductively coupled
ZrC. The brightness also changed dramatically during plasma-atomic emission spectrometry (ICP-AES).
the reaction. The obtained ZrC fragments were exam- With this method, accuracy of the C/Zr atom ratio
ined by Raman spectroscopy and X-ray diffraction. was on the order of 0.01.33
It was confirmed that the bulk of the ZrC remained The defective SiC layer fraction is a very impor-
unaffected by the plasma oxidation.35 tant indicator to show the quality of the ordinary
The physical grinding technique was also devel- TRISO-coated particle fuels. This is measured by
oped to obtain the ZrC fragments.33 In this technique, the burn-leach method, where the fuel compact
quartz powder having the Mohs hardness of 7 was or the coated fuel particles are heated at 1173 K in
used since the Mohs hardness for ZrC is about 8–9 air to oxidize the graphite matrix of the compact and
and that for graphite is 6. The fragments of the the OPyC layers, followed by the acid leaching
combined layers of ZrC and PyC were ground with of the exposed uranium. The defective SiC coating
quartz powder. After grinding, the fragments of the layer exposes uranium during burning. However, this
ZrC layers without PyC were separated from the rest method cannot be applied to the ZrC-coated particle
220 Advanced Concepts in TRISO Fuel
fuels since the intact ZrC layer is oxidized and exposes appeared to be poor in comparison with the ordinary
the actinide oxides to be leached by nitric acid solution. TRISO-coated particles. However, there was no evi-
A new method of measuring the defective ZrC layer dence of palladium attack on any of the ZrC layers.42
fraction is needed. The ZrC-TRISO-coated UC2, UO2, and UCxOy
particles made at General Atomic Company (GAC;
now General Atomics, GA) were irradiated in
3.08.2.4 Performance of ZrC-Coated
HFIR.43–45 The irradiation conditions of the ZrC-
Particle Fuel
TRISO-coated fuel particles in the United States are
3.08.2.4.1 Irradiation performance listed in Table 1.42–45 The performance of the ZrC-
Although systematic irradiation experiments on the TRISO-coated particles was very favorable. But there
ZrC-coated fuel particles have not been completed, was evidence that the ZrC-TRISO-coated UC2 parti-
some promising data have been obtained; the ZrC cles in the HRB-7 and HRB-8 capsules might not be as
layer is less susceptible to chemical attack by fission effective in retaining fission products as the ordinary
products and fuel kernels, and the ZrC-coated fuel TRISO-coated particles; the electron probe micro-
particles perform better than the ordinary TRISO- analysis showed the rare-earth fission products on the
coated particles at high temperatures, especially above outside of the ZrC coating, whereas all cesium was
1873 K. Some early irradiation tests showed poor per- retained within the coating.43 The ZrC coating layers
formance of the ZrC-coated particles, which may be in the ZrC-TRISO-coated UO2 and UC2 particles in
attributed to the fact that the fabrication conditions the HRB-15A capsule suffered no fission product
had not been optimized. attack, while the ordinary TRISO-coated fuel particles
Irradiation tests on several coating designs of the showed some degree of SiC-fission product interac-
ZrC-coated particles were carried out in the High Flux tion. The ZrC-TRISO-coated UC2 particles, however,
Isotope Reactor (HFIR) and the Oak Ridge Research showed poor retention of silver and europium, with
Reactor (ORR).39–45 The irradiated ZrC-coated fuel great variety from particle to particle in this respect.44
particles, which were made at LASL, were (1) ZrC- The irradiation tests on the ZrC-coated UO2 par-
TRISO-coated particles, (2) ZrC-TRISO type coated ticles characterized by the zirconium-carballoy layer
particles without OPyC layer, (3) ZrC-coated particles without the OPyC made at JAERI were carried out in
with ZrC-doped OPyC layer, and (4) ZrC-coated the Japan Materials Testing Reactor (JMTR).17 The
particles with graded C–ZrC layer(s). Prior to the particles in the 73F-12A and 73F-13A capsules expe-
tests on the fuel-kerneled coated particles, carbon- rienced very high temperatures exceeding 1873 K,
kerneled inert coated particles were tested to deter- where the failure fractions of the ZrC-coated parti-
mine the stability of the ZrC at temperatures of 1173 cles were rather high though they were on par with
and 1473 K and fast neutron fluencies of 3.5 1025 to those of the ordinary TRISO-coated particles or
10.7 1025 m2.39,40 The irradiation results of the below. The ceramographic examination revealed no
inert coated particles were encouraging. In the test of interaction between the UO2 kernel and the zirco-
the fuel-kerneled particles, it was found that the graded nium-carballoy layer when they came into contact
coatings were cracked and it was postulated that the with each other due to the kernel migration, showing
cracking was associated with the low PyC deposition better performance of the zirconium-carballoy layer
rate and was not related to the ZrC.41 The ceramo- than the SiC layer against chemical attack.
graphic examination showed that the performance of The ZrC-TRISO-coated UO2 particles made at
the ZrC-coated particles in the HRB-12 capsule JAERI were irradiated in JMTR and the Japan
Capsule Fuel kernel Temperature (K) Burnup (% FIMA) Fast neutron fluence (m2, E > 29 fJ) Reference
Capsule Fuel kernel Temperature (K) Burnup (% FIMA) Fast neutron fluence (m2, E > 29 fJ) Reference
25
78F-4A UO2 1373 4.0 2.2 10 [46]
80F-4A UO2 1173 1.5 1.2 1025 [46]
ICF-26H UO2 1673–1773 1.8 1.0 1025 [46]
VOF-8H UO2 1643 (15 K mm1) 1.6 <1.0 1024 [46]
VOF-14H UO2 1873 (15 K mm1) 1.6 <1.0 1024 [46]
88F-3A UO2 1673–1923 4.5 – [47]
ZrC-TRISO-coated UO2 particle that survived the product retention of the ZrC-TRISO-coated UO2
postirradiation heating at 2673 K.46 The OPyC and particles. Three tests were carried out at 1873 K
ZrC layers expanded, while the IPyC layer did not. for 4500 h, at 2073 K for 3000 h, and at 2273 K for
There was a significant difference in the behavior 100 h in a cold-wall furnace with a graphite
of the ZrC coating layer compared with that of the heater.51,52 For each test, 100 particles were sampled
SiC at high temperatures. The ZrC can sustain a very from fuel compacts irradiated at 1173 K to 1.5%
large strain, whereas the SiC is brittle in nature. The FIMA. During all the heating tests, no through-
high plasticity is explained by the fact that resistance coating failure was detected by the 85Kr release
of ZrC crystal lattice to the dislocation motion monitoring. The X-ray microradiography on the
becomes very weak above 2473 K.58 In these heating coated particles after the heating tests revealed no
tests, the particles were heated in a loose condition OPyC failure, which confirmed the results of the
without mechanical support from the surrounding gas release monitoring.
graphite matrix of the fuel compact. The presence Typical polished cross-sections of the ZrC-
of the graphite matrix could offset the coating expan- TRISO-coated fuel particles after the heating tests
sion and would further reinforce the integrity of the are shown in Figure 7.51,52 No failure was observed
ZrC-TRISO-coated particles.46 in the ZrC and OPyC coating layers of the particles
Isothermal postirradiation heating tests were also heated at 1873 K for 4500 h, as shown in Figure 7(a)
performed to study the coating integrity and fission and 7(b). In some particles, the IPyC coating layers
100 µm (f) 20 µm
(e)
Figure 7 Ceramographs of the ZrC-TRISO-coated UO2 particles after the postirradiation heating tests; (a) and (b) at
1873 K for 4500 h, (c) and (d) at 2073 K for 3000 h, and (e) and (f) at 2273 K for 100 h. Adapted from Minato, K.; Ogawa, T.;
Fukuda, K.; et al. J. Nucl. Mater. 1995, 224, 85–92; Minato, K.; Ogawa, T.; Fukuda, K.; et al. J. Nucl. Mater. 1997, 249, 142–149.
Advanced Concepts in TRISO Fuel 225
were cracked in the radial direction. No palladium The diffusion coefficients for silver, barium, pro-
attack or thermal degradation of ZrC was observed.51 methium, and cerium in the ZrC coating layer were
On the polished cross-sections of the particles evaluated from annealing experiments.60 Based on the
heated at 2073 K for 3000 h, no failure of the ZrC distribution of the nuclides in the ZrC measured by
and OPyC coating layers was observed, as shown in removing the ZrC stepwise, the diffusion coefficient
Figure 7(c) and 7(d). However, some degradation of DBa(ZrC) was estimated to be 1.3 1017 m2 s1 and
the ZrC coating layer seemed to have occurred. The DCe(ZrC) was estimated to be 6.4 1018 m2 s1 at
inner and outer surfaces of the ZrC coating layers in 1773 K. The ZrC coating layer showed better reten-
most of the particles heated at 2073 K were not tion of these elements, though the characteristics of
smooth. In some particles, the IPyC coating layers the ZrC coating layer were not reported.
were cracked in the radial direction, where about a The postirradiation heating tests of the ZrC-
quarter of the thickness of the ZrC coating layers TRISO-coated UO2 particles were performed at
seemed to have been attacked along the grain 1873 K for 4500 h, at 2073 K for 3000 h, and at 2273 K
boundaries.52 for 100 h, to study the release behavior of the fission
The ceramography on the ZrC-TRISO-coated products.51,52 For each heating test, 100 of the ZrC-
fuel particles after heating at 2273 K for 100 h TRISO-coated particles, which had been irradiated at
revealed that the ZrC coating layers as well as the 1173 K to 1.5% FIMA, were used. The furnace in a hot
IPyC coating layers were damaged in most of the cell was composed of a graphite heater, a graphite
particles observed, as shown in Figure 7(e) and 7(f ). sample holder, graphite holder disks, and carbon insu-
The ZrC coating layers were damaged through the lators within a stainless steel vessel. The coated fuel
thickness. Based on the results of the ceramographic particles were placed individually in the holes of
examination, the electron probe microanalysis, and the graphite disks. Each heating test was divided into
the thermodynamic analysis, the observed deteriora- several time steps. At the end of each time step,
tion of the ZrC-TRISO-coated fuel particles was the graphite components and the carbon insulators
attributed to the reaction of ZrC with CO gas caused were replaced by new ones to measure the released
by the failure of the IPyC coating layer.52 metallic fission products by g-ray spectrometry. The
In the case of the ordinary TRISO-coated fuel fission gas release monitoring during the tests and
particles, fission gas release was observed in the post- the X-ray microradiography after the tests revealed
irradiation heating of fuel elements at 2073 K for 100 that no through-coating failure occurred in the tests.
or 200 h.53,54 Although the number of the ZrC- The measured fractional releases of 137Cs are
TRISO-coated particles tested was small compared shown in Figure 8 as a function of heating time.52
with that on the fuel element, it is probably safe to say The calculated fractional release of 137Cs from the
that the ZrC-TRISO-coated fuel particles have higher ordinary TRISO-coated particles at 1873 K is also
capability of fission gas retention than the ordinary presented in the figure for comparison. This curve
TRISO-coated fuel particles at high temperatures. was drawn based on the effective diffusion coefficient
of 137Cs in the SiC coating layer61 and the particle
3.08.2.4.4 Retention of fission products dimensions. The fractional release of 137Cs was found
Retention of the fission product cesium by the ZrC to be below 1 103 even after heating at 1873 K for
coating layer has been demonstrated to be better than 4500 h or at 2073 K for 3000 h, whereas it was more
that by the SiC coating layer though the data for the than 1 101 after heating at 2273 K for 100 h. The
other elements are limited compared with those for sudden increase in the fractional release at 2273 K was
the SiC coating layer. probably attributed to the degradation of the ZrC
The diffusion coefficients for strontium and coating layer observed in the ceramography. The
barium in the ZrC coating layer were obtained in high cesium retention of the ZrC-TRISO-coated
strontium soaking experiments and postactivation fuel particles was confirmed to 2073 K.
annealing experiments.59 The diffusion coefficient Based on a diffusion model, where a fuel kernel
for Sr in the ZrC coating layer, DSr(ZrC), was esti- with a single coating layer was assumed, the effective
mated to be 2 1018 m2 s1 at 1673 K and that for diffusion coefficients for 137Cs in the ZrC coating
Ba, DBa(ZrC), was estimated to be 2.9 1018 to layer, DCs(ZrC), were evaluated to be between
4.6 1018 m2 s1 at 1673 K. The retention of these 1 1018 and 5 1018 m2 s1 at 1873 K, and
elements by the ZrC coating layer was better than between 2 1018 and 1 1017 m2 s1 at 2073 K.52
that by the SiC coating layer. The present value for DCs(ZrC) at 1873 K was more
226 Advanced Concepts in TRISO Fuel
100 100
2273 K
10–1 10–1
Fractional release
Fractional release
10–2 10–2
2273 K
SiC -1873 K
10–3 10–3
2073 K 1873 K
2073 K
10–4 10–4
1873 K
10–5 10–5 0
100 101 102 103 104 105 10 101 102 103 104 105
Heating time (h) Heating time (h)
than one-tenth of the reported effective diffusion kind was also observed in the annealing tests of the
coefficients of 137Cs in the SiC coating layer, short-term irradiated ZrC-TRISO-coated UO2 par-
DCs(SiC),61,62 whereas at 2073 K, the value of DCs(ZrC) ticles.59 The fast diffusion of ruthenium would be
was less than one-hundredth of those for DCs(SiC). explained by the empirical rule that fast diffusion
The measured fractional releases of 106Ru are shown other than ordinary substitutional diffusion may occur
in Figure 9 as a function of heating time.52 The release when the solute has the atomic radius ratio to the solvent
behavior of 106Ru was evaluated by the same diffusion below 0.85; the radius ratio of Ru to Zr is 0.836.
model that was used for the evaluation of 137Cs release. Besides 137Cs, 134Cs, and 106Ru, the release of
For the best fit, as shown in Figure 9, the effective radionuclides of 144Ce, 154Eu, and 155Eu were detected
diffusion coefficients of DRu(ZrC) ¼ 3 1016 m2 s1 by g-ray spectrometry of the graphite components and
at 1873 K and DRu(ZrC) ¼ 5 1015 m2 s1 at 2073 K carbon insulators. The fractional releases of 154Eu
were obtained. The measured fractional release data at were larger than those of 137Cs at any temperature.
2273 K did not fit the calculated curve, which was Although the release data of 137Cs and 106Ru could be
drawn based on the extrapolated diffusion coefficient, treated quantitatively, the accuracy of the measured
DRu(ZrC) = 5 1014 m2 s1, from the data at 1873 and values for 134Cs, 144Ce, 154Eu, and 155Eu was not
2073 K. The sudden increase in the measured fractional enough to evaluate the effective diffusion coefficients
release and the large fractional release compared with since the counts by g-ray spectrometry were small.52
the calculated curve suggest the deterioration of the The radionuclide 110mAg, whose half life is
ZrC coating layer. 250.4 d, is known as one of the most releasable radio-
The release of 106Ru was not reported in the nuclides from the ordinary TRISO-coated fuel par-
postirradiation heating tests of the ordinary ticles. The release of 110mAg was not detected in this
TRISO-coated fuel particles at 1873 K for 500 h and particular experiment. However, it cannot be con-
at 2073 K for 200 h.53,54 The retention of 106Ru in the cluded that the ZrC coating layer has excellent
110m
ZrC-TRISO-coated fuel particles was not better than Ag retention capability. The present experiment
that in the ordinary TRISO-coated fuel particles. was carried out about 9 years after the end of irradia-
The present DRu(ZrC) obtained was almost the tion of the ZrC-TRISO-coated fuel particles and
same as DCs(SiC) in literature.61,62 Behavior of this the ratio of inventories of 110mAg to 106Ru was less
Advanced Concepts in TRISO Fuel 227
than 1 103 just before the heating tests, suggesting It was found that the cesium retention of the fuel
that the inventory of 110mAg was too small to be kernel was enhanced when the IPyC layer failed in
detected even if 110mAg was released.52 the ZrC-TRISO-coated UO2 particles. The mecha-
The retention of the fission product cesium by the nism for the enhancement of cesium retention was
ZrC-TRISO-coated fuel particles was further stud- not yet well understood, though the interaction of
ied. The particles after the postirradiation heating ZrC with CO gas results in reduction of the oxygen
tests at 1873 K for 4500 h and at 2073 K for 3000 h, potential of the fuel, which may influence chemical
together with the as-irradiated particles, were exam- forms of the fission products in the fuel.64 An addi-
ined individually with X-ray microradiography and tional ZrC coating layer on the UO2 kernel of the
g-ray spectrometry.63 coated fuel particle would enhance the cesium reten-
Fifteen particles each were crushed individually tion capabilities. The coated UO2 particles gettered
to recover the fuel kernel and the coating layers with ZrC, where ZrC was coated on the UO2 kernel
separately, and fission product inventories of the or dispersed throughout the buffer layer, were tested,
fuel kernel and coating layers of each particle were which is described in Section 3.08.3.
then measured with g-ray spectrometry. The inven-
3.08.2.4.5 Behavior under oxidizing
tory of 137Cs measured on the whole particle before
conditions
crushing it was in good agreement with the sum of
No experimental work on the performance of the
the inventories measured on the fuel kernel and the
ZrC-coated fuel particles under accidental oxidizing
coating layers for each particle. The fractional con-
tent of 137Cs in the fuel kernel for each as-irradiated conditions was found in the literature.
A thermodynamic analysis was performed on
particle was about 98% or about 2% of 137Cs was
the system of ZrC–C–(O2 or H2O)–He.65 The analy-
released from each kernel to the coating layers, as
sis showed that there were two kinds of oxidation
shown in Figure 10(a).63 The 137Cs retention behav-
behavior of ZrC, which was similar to that of SiC:
ior of each of the particles during the irradiation was
active oxidation and passive oxidation. While a loss
similar. No anomalies were found in the X-ray micro-
in mass occurs in the active oxidation, the passive
radiography. The release of 137Cs from the fuel kernel
could be attributed to the fission recoil. oxidation results in a net mass increase. The active-
The normalized 137Cs inventories in the whole to-passive transition of oxidation occurs depending
particles heated at 2073 K for 3000 h agreed with on temperature and initial O2 or H2O pressure.
Under most of the accidental air or water ingress
one another and the mean value agreed well with
conditions, the passive oxidation of ZrC was expected
that for the as-irradiated particles, which was consis-
to occur. According to the experiment on the passive
tent with the result of the fractional release measure-
oxidation of ZrC at temperatures of 1130–2160 K,66
ment. Figure 10(b) shows the fractional contents
the protective layer of ZrO2 would not be formed on
of 137Cs in the fuel kernel and the coating layers
ZrC, in contrast with the case of SiC. The preferen-
for each particle after postirradiation heating at
2073 K for 3000 h.63 It was found that part of the tial oxidation along grain boundaries occurred in
137 ZrC. Between 1130 and 1560 K, this preferential oxi-
Cs was released from the kernel to the coating
dation resulted in intercrystalline fracture due to
layers in each particle during the heating test and
grain boundary stresses. At higher temperatures, the
that the fractional content of 137Cs in the kernel was
different from particle to particle though 137Cs was stresses were apparently sufficiently released so that
not released from the particles practically.63 the samples remained intact.66 From this point of
view, ZrC may not be a retentive layer in air or
When the result of the X-ray microradiography
water ingress accidents.
was combined with that of the g-ray spectrometry, a
surprising relation was found between them. The first
seven ZrC-TRISO-coated fuel particles shown in 3.08.3 ZrC-Containing TRISO-
Figure 10(b), which showed relatively good retention Coated Particle Fuel
of 137Cs in the fuel kernels, had radially broken IPyC
3.08.3.1 Designs of ZrC-Containing
layers and deformed fuel kernels. The remaining eight
TRISO-Coated Particle Fuel
particles whose cesium retention of the fuel kernels
was relatively poor showed no anomalies in the X-ray Although the TRISO-coated UO2 particle fuel has
microradiography. The same tendency was observed performed very well, it has some disadvantages
in the particles heated at 1873 K for 4500 h.63 that stem from the oxide fuel. When the oxide fuel
228 Advanced Concepts in TRISO Fuel
Coatings
Kernel
0.8
0.6
Fraction
0.4
0.2
0
A2 A15 A3 A12 A9 A5 A4 A7 A14 A1 A8 A10 A11 A6 A13
Particle name
1
Coatings
Kernel
0.8
0.6
Fraction
0.4
0.2
0
C5 C1 C12 C7 C4 C10 C9 C11 C2 C14 C15 C8 C6 C13 C3
Particle name
Figure 10 Fractional contents of 137Cs in the fuel kernel and coating layers for each ZrC-TRISO-coated UO2 particle:
(a) as-irradiated at 1173 K to 1.5% FIMA and (b) after postirradiation heating at 2073 K for 3000 h. Reproduced from
Minato, K.; Ogawa, T.; Koya, H.; et al. J. Nucl. Mater. 2000, 279, 181–188.
is used, the fission-dependent buildup of gaseous large thermal gradient within the particle is estab-
products of CO/CO2 occurs from reactions between lished. In extreme cases, the kernel can traverse
released oxygen and carbon coatings. The resulting through the buffer and IPyC layers, which may
pressure within the particle must be accommo- interact with the SiC layer.
dated by increasing the thickness of the buffer The mixed UO2–UC2 fuel67–71 greatly reduces the
layer depending on target burnups to prevent the CO pressure via the chemical buffering or gettering
pressure-vessel failure. This penalizes the reactor action, and kernel migration rates are also reduced in
design because the core fissile atom density has to this gettered fuel. Similar results have been obtained
be lowered. Especially in the actinide burning, this by the addition of nonfissionable carbides, such as
penalty may be large. The accumulation of CO SiC,72 to oxide kernels, and additional benefits might
within the particle also leads to kernel migration be gained by placing getters of this type at the outer
toward the hotter side of the particle when the boundary of the UO2 kernels.19
Advanced Concepts in TRISO Fuel 229
Two configurations have been proposed and with porous and dense PyC layers, were tested76:
tested for the gettered TRISO-coated UO2 particle UC2/ZrC/buffer/PyC, (8Th,U)O2/ZrC/buffer/PyC,
fuel containing ZrC positioned outside the kernel.19 UC2/buffer/ZrC/PyC, and (8Th,U)O2/buffer/ZrC/
The purpose of the ZrC in the particle is to serve PyC. The irradiation experiments on these coated
as an oxygen getter for regulating the buildup of particles have certainly revealed that the ZrC layer
gas pressure caused by oxygen release during fission- has exceptional resistance to chemical attack by fission
ing of oxide fuel, though the work described in products and good mechanical stability under irradia-
Section 3.08.2 on the use of ZrC in coated fuel tion. Nonetheless, the following section focuses on the
particles has been directed toward the development two types of particles shown in Figure 11.
of a replacement for the SiC barrier layer.
Figure 11(a) and 11(b)19 shows ceramographs of
3.08.3.2 Performance of ZrC-Containing
the two configurations of the gettered TRISO-coated
TRISO-Coated Particle Fuel
UO2 particle fuel containing ZrC. The first one had
an additional solid layer of ZrC deposited over the 3.08.3.2.1 Irradiation performance
kernel, which was protected from coating gases used The two types of TRISO-coated UO2 particles get-
in the ZrC deposition process73 by the previous tered with ZrC were irradiated in the HFIR. Three
application of a thin PyC seal coat directly over the capsule irradiation experiments (HRB-15A, HRB-
kernel. This substrate, consisting of a UO2 kernel, a 15B, and HRB-16) were carried out, in which three
seal coat, and a ZrC layer, was then used as the core of kinds of sample forms were used: loose particles in
an ordinary TRISO particle. The other one had a the holes of graphite trays, particles bonded into
buffer layer in which ZrC was dispersed. The ZrC place in graphite trays with carbonaceous matrix, and
dispersed buffer layer was coated by a codeposition fuel rods were used. Table 3 summarizes the irradia-
process74,75 on the kernel with a PyC seal coat, to tion conditions.44,45,77 The irradiation temperatures
protect the kernel during the codeposition process. were about 1173 K (HRB-15B) and about 1473 K
The IPyC, SiC, and OPyC layers were then coated (HRB-15A, HRB-16), and the burnups were more
on the ZrC dispersed buffer layer. than 20% FIMA. For comparison, the ordinary
Besides the two types of TRISO-coated UO2 par- TRISO-coated particles having kernels consisting of
ticles gettered with ZrC shown in Figure 11, four UO2, UC2 and mixed UO2–UC2 were also included in
fuel particle designs, using ZrC layers in combination these capsule irradiation experiments.19
As made Irradiated
(a) (b)
Solid ZrC
150 µm 100 µm
(c) (d)
Dispersed ZrC
150 µm 140 µm
Figure 11 Ceramographs of two configurations of ZrC-gettered TRISO-coated UO2 particles before and after irradiation;
(a) and (b) with a solid layer of ZrC deposited over the kernel, and (c) and (d) with ZrC dispersed throughout the buffer
layer. Reproduced from Bullock, R. E.; Kaae, J. L. J. Nucl. Mater. 1983, 115, 69–83.
230 Advanced Concepts in TRISO Fuel
Capsule ZrC type Temperature (K) Burnup Fast neutron fluence Reference
(% FIMA) (m2, E > 29 fJ)
The irradiated particles were examined by X-ray buffer layers, and the kernels had extruded into
microradiography, ceramography, and electron probe these coating voids to produce the irregular shapes.19
microanalysis. Figure 11(b) and 11(d)19 shows cer- It is obvious why the expansion behaviors of the
amographs of the irradiated TRISO-coated UO2 two types of particles were so different. As long as the
particles gettered with ZrC. Both types of the ZrC- ZrC layer remained intact, the growth of gas bubbles
gettered TRISO-coated UO2 particles behaved very was restricted, and consequently these kernels swelled
well under irradiation. No detrimental irradiation by only about 1% in diameter. However, large gas
effects were ceramographically found, except for car- bubbles invariably formed when the ZrC layer failed,
bon that precipitated near the outer edge of the and this caused the larger kernel expansions. The
kernels, which may have resulted from the getter- compressive restraint due to the ZrC layer would be
ing action of the ZrC. Oxygen released from UO2 expected to restrict gas-bubble growth in the kernel.19
kernels during fissioning would react predominantly
with carbon from the seal coat to form CO, which 3.08.3.2.2 Retention of fission products
would then be gettered through the reaction: The fission product retention of the fuel particles
2CO þ ZrC ! ZrO2 þ 3C. was examined by postirradiation isothermal annealing
No failures of the three outer structural coatings tests. Ten particles each of the two types of TRISO-
(OPyC/SiC/IPyC) on these ZrC-gettered TRISO- coated UO2 particles gettered with ZrC, together with
coated UO2 particles were observed. The solid ZrC the ordinary TRISO-coated particles having kernels
coating over the kernel occasionally failed under consisting of UO2, UC2, and mixed UO2–UC2, from
irradiation, but an intact coating is not required for the irradiation experiment of HRB-15B in which the
effective gettering, and it was surprising that the irradiation temperature was about 1173 K, were used.
majority of the ZrC layers did remain intact.19 The inventories of fission products in the sample
Kernel migration was not observed in the two particles were measured before and after the heating
types of the TRISO-coated UO2 particles gettered experiment and the released ones collected on sleeves
with ZrC, while the UO2 fuel kernels migrated by as were measured at certain time intervals by g-ray spec-
much as 35 mm in the ordinary TRISO-coated UO2 trometer. The postirradiation heating tests were con-
particles. The absence of kernel migration in the ducted at temperatures of 1473, 1623, and 1773 K for
ZrC-gettered TRISO-coated UO2 particles was good more than 36 Ms (10 000 h).19,20
evidence that this potentially detrimental phenome- The TRISO-coated UO2 particles gettered with a
non in the oxide fuels can be controlled by the addi- solid ZrC overcoating on the kernel did not release
tion of oxygen getters.78 any measurable fission products during postirradia-
In addition to demonstrating the absence of kernel tion annealing at 1773 K for over 42 Ms (12 000 h).
migration, ceramographic examination of the UO2 There was significant release of silver and europium
particles gettered with a solid overcoating of ZrC from the TRISO-coated UO2 particles with ZrC
also yielded information on the reduced kernel dispersed in the buffer layer under the same condi-
expansions observed in X-ray microradiography. A tions, as well as from the ordinary TRISO-coated
one-to-one particle comparison between the X-rays UO2, UC2, and mixed UO2–UC2 fuels. The X-ray
and the ceramographic sections of 54 particles exam- microradiographs revealed that the solid ZrC layers
ined revealed that of the 42 particles with round on all the retentive test particles were apparently
kernels that expanded only 1% had intact ZrC intact and in place on perfectly round kernels before
and buffer layers. However, the other 12 particles and after annealing. An intact ZrC layer seems to be
with 10% kernel expansions had failed ZrC and necessary for explaining the outstanding retention
Advanced Concepts in TRISO Fuel 231
of these particles, as opposed to its mere presence as (O/U > 1.1) fuel kernels are used.68,69,81 Lanthanide
an oxygen getter, because UO2 particles with ZrC is kept as a stable oxide within the oxide and oxycar-
dispersed in the buffer layer released significant bide (O/U > 1.1) fuel kernels. This is a typical exam-
amounts of fission products.19 ple of the first method mentioned earlier.
While the retention mechanism was not yet well On the other hand, the compositions of the fuel
understood, it appeared that coated particle fuel with kernels of oxide, oxycarbide, and carbide showed little
a solid ZrC layer had the potential for retaining highly effect on the corrosion of the SiC layer by palladium:
diffusive silver and europium isotopes for long peri- the corrosion of the SiC layer by palladium has been
ods of time at temperatures as high as 1773 K.19 observed in the TRISO-coated particle fuel with
Both the rare-earth release and the kernel migra- almost all kinds of fuel compositions of UO2, UC2,
tion are largely controlled in UCO fuel in which UCxOy, PuO2x, 3ThO2–PuO2x, ThO2, and (Th,U)
proper proportions of UO2 and UC2 phases are O2.49,50,70,82–84 A new idea was needed to keep the
mixed in the kernel.68,69 Gas pressures in such parti- fission product palladium within the fuel kernel.
cles will certainly be reduced compared to those in Based on the out-of-reactor experiments,85 a con-
pure UO2 fuel, and UCO fuel kernels should retain cept of the TRISO-coated UO2 and UCO particles
all the rare earths, other than Eu, well. 68 Likewise, gettered with SiC dispersed in the kernels has been
the addition of the ZrC oxygen getter in the UO2 fuel suggested to prevent the corrosion of the SiC layer by
types accomplishes the same purpose, while also the fission product palladium.86 The idea is that the
allowing much higher retention of Eu in the pure formation of a compound UPd3Si3C5 with a high
oxide kernels. The TRISO-coated UCO particles melting temperature of >2225 K would keep palla-
were selected over the ZrC-gettered TRISO-coated dium in the kernel. Although the thermodynamic
UO2 particles as the reference fissile fuel for steam- calculations have been carried out, the fabrication
cycle HTGRs, based more on fabrication considera- and irradiation tests of particles of this kind have
tions and the availability of a wider database than upon not been performed yet.
the fission product release results described earlier.20 To prevent the corrosion of the SiC layer by fission
However, the excellent performance for the ZrC- product palladium based on the second method
gettered UO2 TRISO-coated particle has made this described above, three types of new combinations of
fuel a strong candidate for VHTR or DB-MHR.79 the coating layers have been proposed and tested.87 The
To improve the high-temperature stability and the idea is to add a layer that traps palladium by chemical
fission product retention, ZrC-TRISO-coated UO2 reaction inside the SiC layer of the TRISO coating.
particles, instead of the ordinary TRISO, gettered Two kinds of additional layers have been selected: an
with a solid ZrC overcoating on the kernel were SiC + PyC layer and an SiC layer. The SiC þ PyC
suggested as one of the possible design variations.80 layer is composed of SiC with free carbon.88 This kind
of layer has been studied to improve the capability of
fission product retention of the dense PyC layer.89
3.08.4 SiC-Containing TRISO-Coated Figure 12 shows ceramographs of three types of
Particle Fuel advanced coatings, together with that of the TRISO
coating for comparison.87 The type-A coating has an
3.08.4.1 Designs of SiC-Containing
additional layer of SiC þ PyC adjacent to the inside
TRISO-Coated Particle Fuel
of the SiC layer. As the SiC þ PyC layer has a better
The chemical interaction of the SiC layer of the capability of fission product retention,89 the thickness
TRISO-coated particle fuel with fission products of the IPyC layer can be reduced; the IPyC layer
occurs when the fission products are released from should act as a seal to prevent the chemical reaction
the fuel kernel and subsequently reach the SiC layer of the fuel kernel with coating gases during the coat-
through the IPyC layer. Generally, two methods may ing process.
be available to prevent the corrosion of the SiC layer: The type-B coating is similar to the type-A coat-
(1) to keep the fission products within the fuel kernel ing, but the dense PyC layer is present between the
and (2) to make a barrier to the diffusion of the fission SiC þ PyC and SiC layers. The expected role of
products to the SiC layer. the intermediate dense PyC layer is to interrupt the
The corrosion of the SiC layer by lanthanide seen radial extension of the corrosion zone from the
in the TRISO-coated UC2 and UCxOy (O/U 1.1) SiC þ PyC layer to the SiC layer. The corrosion
particles could be avoided, when oxide or oxycarbide zone would extend circumferentially in the SiC þ PyC
232 Advanced Concepts in TRISO Fuel
Den Den
se P se
yC PyC
SiC SiC
Den
SiC se
+ Py PyC
C SiC
Den +P
se P yC
yC Den
se
Por PyC
ous
PyC Por
ous
PyC
UO UO
2 20 µm (b) 20 µm
(a) 2
Den
se P Den
yC se P
yC
SiC SiC
Den
se P
yC Den
S se P
Den iC yC
se P
yC
Por Por
ous ous
PyC PyC
UO UO
2 2
(c) 20 µm (d) 20 µm
Figure 12 Ceramographs of three types of advanced coatings, together with that of the TRISO coating for comparison;
(a) type-A coating, (b) type-B coating, (c) type-C coating, and (d) TRISO coating. Reproduced from Minato, K.; Fukuda, K.;
Ishikawa, A.; et al. J. Nucl. Mater. 1997, 246, 215–222.
layer when the intermediate dense PyC layer is pres- 3.08.4.2 Fabrication of SiC-Containing
ent. The thickness of the IPyC layer can be reduced in TRISO-Coated Particle Fuel
the same manner as the type-A coating.
No new technology was needed and a conventional
In the type-C coating, SiC is used for an additional
coating apparatus without modification could be used
layer. The role of the inner SiC layer is to react with
to fabricate the SiC-containing TRISO-coated particle
fission products, and the intermediate dense PyC layer
fuel. This point is of great advantage for fuel fabrication.
is expected to interrupt the radial extension of the
Three types of advanced coatings, type-A, type-B,
corrosion from the inner to outer SiC layers.
and type-C, were prepared. The porous PyC layer was
The increase in the thickness of the SiC layer of
deposited at 1573 K with the pyrolysis of acetylene
the TRISO coating may be one of the solutions to the
(C2H2) in the flowing argon, and the dense PyC layer
corrosion of the SiC layer by fission products. How-
was deposited at 1638 Kwith the pyrolysis of propylene
ever, the thicker coating layers will result in a smaller
(C3H6) in the flowing argon. The SiC layer was chemi-
amount of the fuel material contained in a unit vol-
cally vapor deposited at 1873 K using methyltrichlor-
ume. With these advanced coatings, no corrosion
osilane (CH3SiCl3; MTS) and hydrogen. The SiC þ
of the SiC layer by fission products is expected to
PyC layer was deposited at 1873 K with MTS and
occur without increasing the total thickness of the
argon.88 The content of free carbon in the SiC þ PyC
coating layers.
Advanced Concepts in TRISO Fuel 233
In the out-of-reactor tests, palladium was designed to The nitrides of transuranium elements could be
diffuse into the coating layers from the outside of the fabricated with the prevention of the vaporization
particles. The heated coated fuel particles were polished loss of the transuranium elements by keeping the
and examined with an optical microscope.87 nitrogen partial pressure sufficiently high during
In the type-B coating, which has the intermediate the fabrication processes.91
dense PyC layer between the SiC + PyC and SiC layers, The coating has to be selected so as to be compatible
part of the SiC + PyC layer was completely attacked with the fast neutron environment, the nitride kernel,
through the coating thickness by palladium, while no and the matrix or structural materials. The use of SiC
corrosion was found in the SiC layer. This interruption as a coating material would have to be avoided as it
of the radial extension of the corrosion zone from reacts with transition metals, such as Ti, Fe, Cr, and
the SiC + PyC layer to the SiC layer was exactly the Ni,92 unless the matrix or the structural container is
expected role of the intermediate dense PyC layer. made of ceramics with good thermal properties, such
The interruption of the radial extension of the as TiN and TiSi2, which are compatible with SiC.1
corrosion zone from the outer to inner SiC layers Titanium nitride with enriched 15N may be a good
by the presence of the intermediate dense PyC layer choice because of its refractoriness (Tm = 3223 K),
was also observed in the type-C coating. The inter- low neutron cross-sections, and compatibility with
mediate dense PyC layer between two SiC layers steels. It is advantageous for the coating that TiN is
functioned effectively, as expected. little miscible with actinide nitrides.93
In the out-of-reactor experiment, the SiC + PyC Two configurations of the coated particles have
layer was proved to be effective as a barrier to the been proposed. One is the duplex coating, where the
diffusion of palladium to the SiC layer, and the inner layer is a low-density buffer layer and the outer
intermediate dense PyC layer was found to inter- layer is of high-density fission products container.
rupt the radial extension of the corrosion zone The other configuration is a porous kernel and a
from the barrier layer to the SiC layer. However, it thin but strong and dense outer layer to maximize
should be noted that the intermediate dense PyC the fraction of fuel in the particle while maintaining
layer could not interrupt the radial extension of the integrity of the miniature pressure vessel.1,5
the corrosion zone when the amount of palladium Ferritic steel may be a good choice for the metal
exceeded the capacity of the barrier layer to react matrix of a pebble-bed type core or the porous metal
with palladium. frit of a particle bed type core. TiN is completely
stable when in contact with the ferritic steels.1
behavior of LASL-made extruded fuel rods and 69. Tiegs, T. N.; Lindemer, T. B.; Henson, T. J. J. Nucl. Mater.
LASL-made coated particles, LA-6988-MS; Los Alamos 1981, 99, 222–234.
National Laboratory, 1977. 70. Benz, R.; Förthmann, R.; Grübmeier, H.; Naoumidis, A.
42. Homan, F.; Tiegs, T. N.; Kania, M. J.; et al. Irradiation In Thermodynamics of Nuclear Materials 1979; IAEA:
performance of HTGR fuel rods in HFIR experiments HRB-11 Vienna, 1980; pp 565–586, Vol. 1, IAEA-SM-236.
and-12, ORNL-5584; Oak Ridge National Laboratory, 1980. 71. Stinton, D. P.; Lackey, W. J.; Spence, R. D. J. Am. Ceram.
43. Valentine, K. H.; Homan, F. J.; Long, E. L.; et al. Irradiation Soc. 1982, 65, 321–324.
performance of HTGR fuel rods in HFIR experiments 72. Horsley, G. W.; Weldrick, G. J.; Turnbull, J. A.; Shipp, R.
HRB-7 and-8, ORNL-5228; Oak Ridge National J. Am. Ceram. Soc. 1976, 59, 1–4.
Laboratory, 1977. 73. Reynolds, G. H. J. Nucl. Mater. 1974, 50, 215–216.
44. Ketterer, J. W.; Kania, M. J.; Bullock, R. E.; Siman-Tov, I. I. 74. Reynolds, G. H.; Kaae, J. L. J. Nucl. Mater. 1975, 56,
Capsule HRB-15A postirradiation examination report, 239–242.
GA-A 16758; GA Technologies, 1984. 75. Stevens, D. W.; Sterling, S. A. Influence of ZrC and SiC
45. HTGR Fuel Technology Program. Semiannual report for additives on the thermal migration of BISO coated (Th,U)
the period ending March 31, GA-A 17110; GA O2 kernels, GA-A 15202; General Atomic Company, 1979.
Technologies, 1983. 76. Reynolds, G. H.; Janvier, G.; Kaae, J. L.; Morlevat, J. P.
46. Ogawa, T.; Fukuda, K.; Kashimura, S.; et al. J. Am. Ceram. J. Nucl. Mater. 1976, 62, 9–16.
Soc. 1992, 75, 2985–2990. 77. Ketterer, J. W.; Bullock, R. E. Capsule HRB-15B
47. Minato, K.; Ogawa, T.; Sawa, K.; et al. Nucl. Technol. postirradiation examination report, GA-A 15940; General
2000, 130, 272–281. Atomic Company, 1981.
48. Minato, K.; Ogawa, T.; Kashimura, S.; et al. J. Nucl. Mater. 78. HTGR Fuel Technology Program. Contract summary
1990, 172, 184–196. report for the period ending October 31, 1983, GA-A
49. Ogawa, T.; Ikawa, K. High Temp. Sci. 1986, 22, 179–193. 17369; GA Technologies, 1983.
50. Tiegs, T. N. Nucl. Technol. 1982, 57, 389–398. 79. McEachern, D.; Baxter, A.; Rodriguez, C. TRISO coated
51. Minato, K.; Ogawa, T.; Fukuda, K.; et al. J. Nucl. Mater. particles containing zirconium for use in deep-burn
1995, 224, 85–92. modular helium reactors. In Proceedings of Global 2003,
52. Minato, K.; Ogawa, T.; Fukuda, K.; Sekino, H.; New Orleans, LA, Nov 16–20, 2003, pp 523–524.
Kitagawa, I.; Mita, N. J. Nucl. Mater. 1997, 249, 142–149. 80. Kendall, J. M.; Bullock, R. E. Advanced coated particle
53. Schenk, W.; Pott, G.; Nabielek, H. J. Nucl. Mater. 1990, fuel options, In Proceedings of the Conference on High
171, 19–30. Temperature Reactors, HTR-2004, Paper no. B09, Beijing,
54. Schenk, W.; Nabielek, H. Nucl. Technol. 1991, 96, 323–336. China, Sept 22–24, 2004 IAEA: Vienna, 2004.
55. Ogawa, T.; Ikawa, K. J. Nucl. Mater. 1981, 99, 85–93. 81. Smith, C. L. J. Am. Ceram. Soc. 1979, 62, 600–606.
56. Storms, E. K.; Griffin, J. High Temp. Sci. 1973, 5, 291–310. 82. Montgomery, F. Fission-product SiC reaction in HTGR
57. Ogawa, T.; Minato, K.; Fukuda, K.; et al. Nucl. Technol. fuel, DOE/ET/35300-T3; General Atomic Company, 1981.
1991, 96, 314–322. 83. Saurwein, J. J.; Holzgraf, J. F.; Miller, C. M.; Myers, B. F.;
58. Trefilov, V. I.; Milman Yu, V.; Grigoriev, O. N. In Single Wallroth, C. F. Final report on the Peach Bottom test
Crystals of Refractory Compounds; Gurin, V. N, Ed.; element program, GA-A 15999; General Atomic Company,
Pergamon: Oxford, 1988; pp 225–277. 1982.
59. Ogawa, T.; Ikawa, K. J. Nucl. Mater. 1982, 105, 331–334. 84. Kleykamp, H. Mikrosondenuntersuchungen zum
60. Chernikov, A. S.; Khromov Yu, F.; Lyutikov, R. A. Verhalten der Spaltprodukte in hoch abgebrannten
In Proceedings of IAEA Specialists Meeting on Fission HTR-Brennstoffen, KfK 2213; Kernforschungszentrum
Product Release and Transport in Gas-Cooled Reactors, Karlsruhe, 1975.
IAEA IWGGCR/13, Berkeley, Oct 22–25, 1985; IAEA: 85. Lauf, R. J.; Lindemer, T. B.; Pearson, R. L. J. Nucl. Mater.
Vienna, 1986; pp 170–181. 1984, 120, 6–30.
61. Verfondern, K.; Martin, R. C.; Moormann, R. Methods and 86. Lindemer, T. B. Inert-matrix fuels in gas-cooled reactors.
data for HTGR fuel performance and radionuclide release In Invited presentation, Nuclear and Environmental
modeling during normal operation and accidents for safety Technology Division, American Ceramic Society, Materials
analyses, Jü1-2721; Forschungszentrum Jülich, 1993. Science and Technology 2008 Conference and Exhibition,
62. Minato, K.; Ogawa, T.; Fukuda, K.; et al. J. Nucl. Mater. Pittsburgh, PA, Oct 5–9, 2008.
1993, 202, 47–53. 87. Minato, K.; Fukuda, K.; Ishikawa, A.; Mita, N. J. Nucl.
63. Minato, K.; Ogawa, T.; Koya, H.; Sekino, H.; Tomita, T. Mater. 1997, 246, 215–222.
J. Nucl. Mater. 2000, 279, 181–188. 88. Minato, K.; Fukuda, K. J. Nucl. Mater. 1987, 149, 233–246.
64. Minato, K.; Fukuda, K.; Sekino, H.; Ishikawa, A.; Oeda, E. 89. Bullock, R. E. J. Nucl. Mater. 1983, 113, 81–100.
J. Nucl. Mater. 1998, 252, 13–21. 90. Lindemer, T. B. Thermochemical analysis of gas-cooled
65. Minato, K.; Fukuda, K. In Proceedings of a Technical reactor fuels containing Am and Pu oxides,
Committee Meeting on Response of Fuel, Fuel Elements ORNL/TM-2002/133; Oak Ridge National Laboratory,
and Gas Cooled Reactor Cores Under Accidental Air or 2002.
Water Ingress Conditions, IAEA-TECDOC-784, Beijing, 91. Takano, M.; Itoh, A.; Akabori, M.; Minato, K.; Numata, M.
China, Oct 25–27, 1993 IAEA: Vienna, 1995; pp 86–91. Study on the stability of AmN and (Am,Zr)N. In
66. Berkowitz-Mattuck, J. B. J. Electrochem. Soc. 1967, 114, Proceedings of Global 2003, New Orleans, LA , Nov
1030–1032. 16–20, 2003, pp 2285–2291.
67. Nickel, H. High Temp. High Press. 1976, 8, 123–141. 92. Minato, K.; Ogawa, T.; Kashimura, S.; et al. J. Nucl. Mater.
68. Homan, F. J.; Lindemer, T. B.; Long, E. L.; Tiegs, T. N.; 1990, 175, 14–19.
Beatty, R. L. Nucl. Technol. 1977, 35, 428–441. 93. Arai, Y.; Nakajima, K. J. Nucl. Mater. 2000, 281, 244–247.
3.09 Inert Matrix Fuel
P. Pöml, R. J. M. Konings, J. Somers, and T. Wiss
European Commission, Joint Research Centre, Institute for Transuranium Elements, Karlsruhe, Germany
G. J. L. M. de Haas and F. C. Klaassen
Nuclear Research and Consultancy Group, Petten, The Netherlands
GE General Electric
Abbreviations
GETR General Electric Test Reactor
AFCI Advanced Fuel Cycle Initiative
GWd Giga watt days
BNFL British Nuclear Fuels Limited
HM Heavy metal
BOL Beginning-of-life
HMTA Hexamethylenetetramine
Cercer Ceramic – ceramic
HFR High flux reactor
Cermet Ceramic – metal
HRP Halden Reactor Project
CSZ Calcia-stabilized zirconia
IFA Instrumented fuel assembly
DoE US Department of Energy
IFE Institute for Energiteknikk
EOL End-of-life
IMF Inert matrix fuel
EPMA Electron probe microanalysis
INL Idaho National Laboratory
ETR Engineering Test Reactor
ITU Institute for Transuranium Elements
FGR Fission gas release
JAEA Japan Atomic Energy Agency
FIFA Fissions per initial fissile atoms
JAERI Japan Atomic Energy Research Institute
FIMA Fissions per initial metal atoms
JRR Japan Research Reactor
237
238 Inert Matrix Fuel
In a strict sense the term inert matrix fuel (IMF) refers The development of zirconia-based nuclear fuel
to any nuclear fuel containing a low activation matrix started as early as in the 1960s with the successful
as carrier for the fissile material. Since the early days irradiation of UO2–ZrO2 fuel after some pioneering
of nuclear technology, this idea has been investigated, work on the fast-neutron irradiation of zirconia in
originally with the goal to improve fuel properties the 1950s.1,2
or to save uranium resources. However, currently, The first studies on IMF of interest today were
the term IMF is strongly associated with plutonium performed in the late 1960s by two indepen-
fuel that does not contain uranium to obtain the dent groups, both contracted by the United States
highest efficiency for destruction of excess plutonium Atomic Energy Commission (USAEC). One group
(separated civil plutonium or plutonium from dis- (GE-Hanford) investigated the irradiation perfor-
mantled weapons) in a single irradiation campaign. mance of ZrO2–PuO2 and MgO–PuO2 fuels in the
This chapter focuses on this application. The term Engineering Test Reactor (ETR) facility at Hanford,
IMF is also used in the context of uranium-free while the other group (Nuclear Materials and Equip-
fuels for transmutation of minor actinides (MA), ment Corporation, NUMEC) tested the preparation
although in many cases this is not appropriate as the and irradiation of PuO2–UO2–ZrO2 fuel in the
fissile content is too low for fuel purposes and it General Electric Test Reactor (GETR) facility at
is better to call this type of materials targets. Trans- Pleasanton.3–7 Early phase diagram investigations
mutation fuels and targets are described in detail in for the systems ZrO2–PuO2 and MgO–PuO2 were
3.05, Actinide Bearing Fuels and Transmutation also performed at Hanford.8,9 The programs at
Targets. Hanford focused on plutonium utilization and
A variety of materials has been proposed as inert ZrO2–PuO2 and MgO–PuO2 fuels were thought to
matrix for nuclear fuel, ceramics such as MgO, ZrO2, be of interest for potential Pu-containing ceramic
or CeO2, refractory materials such as graphite or SiC, spike enrichment applications.5 The NUMEC pro-
and metals such as stainless steel, zirconium, or gram, however, aimed for Pu recycling in thermal
Inert Matrix Fuel 239
reactors by substituting part of the UO2, using ZrO2 as incorporated in the YSZ phase forming a SS. This
an inert diluent. program included three irradiation tests on Pu- and
Interest in these programs and funding however U-ROX fuels in the Japan Research Reactor No. 3
ceased quickly and results of postirradiation exami- ( JRR-3). PIE was performed at the Real Fuel Exami-
nations (PIE) were sparse (Hanford), or not published nation Facility of JAERI.18 Fuel capsules of the ROX
at all (NUMEC). In the 1990s, IMF saw a renaissance, design were also irradiated in the OTTO experiment
because of two main issues related to the back-end of in the HFR in Petten (see previous paragraph).
the fuel cycle. Pu stockpiles worldwide were growing, In the United States, there was a recent resur-
as a result of their continuous production in nuclear gence of interest in IMF within the framework of
power plants and in particular the dismantling of the US Department of Energy’s (DoE) Advanced
nuclear weapons. In addition, it was recognized that Fuel Cycle Initiative (AFCI). Within this program,
there is a need for dealing with ‘MA’ (e.g., neptunium, two groups, the Idaho National Laboratory (INL)
americium, and curium) that are a component of and the University of Nevada, work on the devel-
spent nuclear fuel characterized by long-term radio- opment of magnesia–zirconia ceramics doped with
toxicity. Advanced concepts were developed to deal plutonia.19–21 Composite MgO–(Zr,Pu)O2 fuel was
with these issues, namely, the burning of Pu and the proposed for LWR use to enhance the thermal con-
transmutation of MA in nuclear reactors. ductivity (MgO) and to provide protection from the
With respect to the IMF for plutonium destruc- LWR coolant attack (ZrO2).
tion, a number of extensive studies were initiated. For clarity and readability, a short overview of
At the Paul Scherrer Institute (PSI) in Switzerland, the major programs is given in Table 1.
the concept of a homogeneous yttria-stabilized zirco-
nia (YSZ)-based plutonium fuel for light water
3.09.3 Neutronic Considerations
reactors (LWRs) was developed, designated for a
once-through irradiation.10 The test pins of this
There is a distinct difference between traditional
IMF, fabricated by two different routes, were irra-
uranium fuels and uranium-free IMFs, particularly
diated in the framework of the Organization for
with respect to power peaking and reactivity coeffi-
Economic Co-operation and Development/Nuclear
cients during irradiation. The power peaking for
Energy Agency (OECD/NEA) Halden Reactor Proj-
IMF is the result of the burnout of the fissile isotopes
ect (HRP) experiment IFA-651. The objective of
without the formation of new fissile material through
this program was to determine the general thermal
breeding from fertile material. It occurs at beginning-
behavior of YSZ-based IMF under irradiation condi-
of-life (BOL), when the fissile content is still high.
tions similar to those in current LWRs. The irradiation
Such power peaking could be reduced/controlled
test started in June 2000 and in May 2003; one IMF rod
by introducing a burnable poison in the fuel. For
was discharged for PIE and the results published.11–14
details on the use of burnable poison in nuclear fuel
Fuel capsules with similar compositions were also
the reader is referred to Chapter 2.16, Burnable
irradiated in the once-through-then-out (OTTO)
Poison-Doped Fuel. For that reason, PSI developed
experiment, an international, joint experiment of the
yttrium-stabilized zirconium-based IMF in which a
Japan Atomic Energy Research Institute ( JAERI), the
small amount of Er as burnable poison had been
Nuclear Research & Consultancy Group (NRG), and
introduced (Zr0.74Y0.14Er0.04Pu0.08)O2.25 In this way,
PSI. Fuel specimens designed by the participating
BOL behavior is almost identical to reference mixed
organizations were irradiated under the same condi-
oxide (MOX) fuel. JAERI suggested the introduction
tions in an in-core position of the High Flux Reactor
of fertile thorium into the fuel, to control the peak
(HFR) in Petten (Netherlands) under the same con-
power by 233U formation.26
ditions. The main properties to be studied were irra-
Extensive analysis of a full IMF reactor core was
diation stability and temperature, swelling, fission
performed at PSI.27,28 The effects of IMF assemblies
gas release (FGR), and chemical interactions in the
on the reactor core reactivity coefficients can be
fuel.15,16
summarized as follows:
At JAERI (now Japan Atomic Energy Agency,
JAEA) the concept of so-called rock-like oxide The Doppler coefficient for a full IMF core (1.2
(ROX) fuels was developed for the once-through pcm/K) at 1200 K is half of the value of the refer-
burning of Pu in LWRs.17 ROX fuel is a multi-phase ence UO2 core (2.4 pcm/K), whereas that of a
mixture of mineral-like compounds such as YSZ, spi- full MOX core (2.6 pcm/K) is more negative
nel (MgAl2O4), and corundum (Al2O3). Plutonium is than the reference.
Overview of the major irradiation programs with IMF
240
Table 1
Program Site/reactor Period Composition of Linear hear Fuel pellet ø (mm)/cladding T center (K) PIE References
fuels studied rate (LHR)
Table 2 The plutonium burnout and minor actinide (MA) production (in %) of a full MOX core and full IMF core after 1300
full power days for reactor grade and weapons grade plutonium
Source: Kasemeyer, U.; Paratte, J. M.; Grimm, P.; Chawla, R. Nucl. Technol. 1998, 122, 52–63.
The moderator temperature coefficient of the IMF In a UO2 fuel the fissile material is distributed
core is in between the reference UO2 core and the homogeneously and the fission will occur homo-
full MOX core. geneously over a fuel volume unit. In IMF, however,
The void coefficients of the full IMF core are especially in the particle-dispersed types, the
sufficiently negative, even in case of 100% voiding, fissile material is concentrated in the Pu-rich parti-
similar to the UO2 and full MOX cores. cles that are distributed in the inert matrix. There-
fore, the use of a volumetric burnup (e.g., GWd m3)
Burnout calculations by Kaesemeyer et al.27 showed
provides a better description of the burnup (and
that the plutonium consumption in (YSZ)-based plu-
hence also fission product damage) of IMF rather
tonium IMF is significant. For example, in pressurized
than a specific burnup related to the mass of HM
water reactors (PWRs) about 70% of weapons grade
(GWd/tHM).
plutonium and 60% of reactor grade plutonium can
A good overview on this topic can be found in
be destroyed in a single cycle (see Table 2).27 The Pu
Appendix 1.1 of Chapter 1 of the 2005 OECD report
left at the end-of-life (EOL) in the IMF fuel is signifi-
on Fuels and Materials for Transmutation.30
cantly less than in case of MOX fuel. As a conse-
quence, the minor actinide production is also
significantly lower in case of the IMF core. Compara-
3.09.4 Fabrication
ble results have been reported by Schneider et al.29 for
YSZ-based fuel and composite MgO–(Zr,Pu)O2
Most commonly, two processes are used for fuel
(40 wt% MgO) fuel. In addition, a significant reduc-
fabrication: powder blending and precipitation. The
tion of the decay heat can be achieved compared to
powder blending process is a dry method based on
direct disposal of equivalent UO2 fuel.
the mixing and milling of different powder feeds. It
has the big advantage that it is robust against criti-
3.09.3.1 Burnup Indicators of Inert
cality, but the obtained product is not always homo-
Matrix Fuels
geneous at the microscale. The precipitation process
Fuel burnup is measured in different quantities, such is based on the mixing of aqueous feed solutions
as fissions per initial metal atoms (FIMA), fissions per from which a solid powder is obtained by copreci-
initial fissile atoms (FIFA), GWd/tHM, etc. These pitation. This precipitate is generally very homoge-
quantities work well for standard UO2 fuels, but are neous, but because of the use of aqueous solutions
not very suitable for inert matrix fuels. This is related the process needs to be secured against criticality
to the fact that the heavy metal densities, and conse- risks, generally by using small(er) feed batches. The
quently the local specific burnup of standard fuels two processes can also be combined by producing
and IMF differ strongly (up to a factor of 20). microspheres of homogeneous Pu–YSZ that are used
242 Inert Matrix Fuel
as starting material to produce either a SS fuel or nozzle for drop formation (Figure 1). In this process,
composite fuels, as will be explained in the next monodisperse microspheres with a size larger than
sections. 250 mm are obtained. These microspheres cannot be
used directly to fabricate the fuel, but must be crushed,
3.09.4.1 Liquid Processes ground, and milled to obtain a processable powder.
However, the microspheres can be directly used to
Liquid processes are principally employed for the fab-
produce composite fuel by a powder blending process.
rication of SS fuels. Generally, they use nitrate solu-
tions as the starting materials. These solutions are
mixed and converted to produce either powder by 3.09.4.2 Powder Blending Processes
coprecipitation, or microspheres by external or inter-
Powder blending processes are commonly used
nal gelation processes.6,16,31 The resulting powder can
in LWR fuel fabrication for fabricating MOX fuel,
be processed directly into fuel pellets by ball milling,
and could be used for IMF fabrication in today’s
compacting, and sintering in a reducing atmosphere
plants, if required. This is a straightforward method,
(N2/H2 or Ar/H2). The pellets are then finished by
even though it has the disadvantage of producing
centerless grinding to meet the pellet dimension spe-
more dust than in the case of liquid processes.
cifications. If microspheres were produced, these are
To produce a solid solution IMF, powders of the
usually calcined at temperatures between 1070 and
starting materials such as UO2 or PuO2, ZrO2, and
1270 K in a reducing N2/H2 atmosphere. However,
Y2O3 are mixed in appropriate proportions. These
some production routes calcine also in air at 673 K.16
powders are then blended in a mill (ball or attrition
An advanced external gelation process was devel-
mill), pressed into pellets, and then sintered in reduc-
oped at the Institute for Transuranium Elements
ing atmospheres like N2/H2, or H2. Sintering tem-
(ITU), Karlsruhe, Germany,32,33 and tests with Ce and
peratures applied are between 1723 and 2023 K.4,34,35
Pu proved it to be very suitable for the fabrication of
The production of composite fuels is more com-
polydispersed (Zr,Y,Pu)O2x microspheres with a size
plex. Figure 2 shows, as an example, a flow diagram for
distribution in the 40–150 mm range. In this size range,
the production of fuel pellets of the particle-dispersed
the spheres are free-flowing and do not produce dust, a
and homogeneous mixture types for the ROX irradia-
known disadvantage of the traditional powder proces-
tion experiments. Basically, the starting materials for
sing. In this process, a rotating cup atomizer is used to
many composite fuels are microspheres loaded with Pu
generate droplets that are solidified in an ammonia
prepared by a liquid process (Figure 3). The particles
solution (Figure 1). The obtained powder can be
can be size selected (or crushed and sieved) permitting
processed, that is, compacted and sintered, directly.
In some studies, powder was produced by the stan-
dard internal gelation process that using a vibrating (U,Zr,Y)O2 Matrix materials UO2, ZrO2, Y2O3
calcined particles (MgAl2O4, Al2O3) powder
powder
Calcination Calcination
Sintered pellet Sintered pellet
Compaction Compaction (U,Zr,Y)O2 particles homogeneous mixture type
dispersed type
Sintering Sintering
External gelation Internal gelation Figure 2 Flow diagram for fabrication of the rock-like
oxide fuel pellets. Reproduced from Shiratori, T.;
Figure 1 Flow diagram for fabrication of (Pu,Zr,Y)O2 Yamashita, T.; Ohmichi, T.; Yasuda, A.; Watarumi,
particles with an external or internal sol–gel process. K. J. Nucl. Mater. 1999, 274, 40–46.
Inert Matrix Fuel 243
Figure 3 External appearance of (Y,Pu,U,Zr)O2x microspheres (left) and (Er,Y,Pu,Zr)O2x microspheres (right) for
the once-through-then-out (OTTO) experiment. Reproduced from Schram, R. P. C.; van der Laan, R. R.; Klaassen, F. C.;
Bakker, K.; Yamashita, T.; Ingold, F. J. Nucl. Mater. 2003, 319, 118–125.
Figure 4 External appearance (left), internal appearance (middle), and a-autoradiography of a pellet of the
once-through-then-out (OTTO) experiment, a macrodispersion of (Er,Y,Pu,Zr)O2x þ MgAl2O4 spinel, sintered at 1873 K.
Reproduced from Schram, R. P. C.; van der Laan, R. R.; Klaassen, F. C.; Bakker, K.; Yamashita, T.; Ingold, F.
J. Nucl. Mater. 2003, 319, 118–125.
size-tailored IMF concepts. In some cases, just powder small amounts of other oxides such as MgO, Y2O3,
blending methods are also used to obtain a microdis- CaO, Ce2O3, and also PuO2.
persed fuel (particle size 10–50 mm).31 The blending of For the application of IMFs, ZrO2 is usually stabi-
the Pu-bearing host with the matrix is generally lized by Y2O3 or CaO. Jin and Du36 assessed the phase
straightforward. Nevertheless, due caution must be diagrams for both systems using the CALPHAD tech-
exercised as segregation of powders can occur, because nique (Figures 5 and 6). They showed that Y2O3
of their differing densities. The final pellets are obtained stabilizes the cubic phase in a wide composition
by compaction and sintering in reducing atmospheres at range down to 796 K (for about 32 mol% YO1.5). The
temperatures between 1820 and 2020 K.15,31,34 Figure 4 stability domain of calcia-stabilization (ZrO2) is much
shows a pellet of a microdispersion fuel used in the narower and the minimum temperature is 1413 K.
OTTO irradiation experiment. The first phase diagram for the PuO2–ZrO2 sys-
tem was published in the 1960s by Carroll.8,9 This
early work suffered from the reduction of the helium
3.09.5 Fuel Properties atmosphere used, which caused the investigated
material to be not stoichiometric PuO2 but rather a
3.09.5.1 Phase Relations
reduced form (PuO1.61). In addition, the tetragonal
Pure zirconia (ZrO2) has a monoclinic crystal struc- form of ZrO2 was not considered. The system ZrO2–
ture at room temperature and tetragonal and PuO2 was assessed by Kinoshita et al.37 using the
cubic crystal structures at higher temperatures. CALPHAD technique (Figure 7). Their phase diagram
The transition from tetragonal to monoclinic during shows that the stability domain of the cubic phase
cooling involves a large volume change and concom- extends over the whole composition range, but the
itant cracking and thus needs to be avoided. ZrO2 can lower temperature boundary steadily increases with
be stabilized in the tetragonal/cubic form by adding increasing PuO2 content. In the region relevant to
244 Inert Matrix Fuel
3400 3000
Liquid Liquid
2700
HSS
2900 2983
CSS 2681 2712 2400
2642 2623
L + HSS 2676 Cubic
L 2100
Temperature (K)
2400
YSS
Temperature (K)
2681
1800
YSS 2676
Tetragonal
1900 YSS + HSS 1500
1649
TSS 1200 Monoclinic
1454 CSS + YSS
1400
900
3172 2500
Liquid CSS + liquid Liquid + MgO
3100
2983 CSS 2381, 0.55
(2373, 0.52±0.03)
2100
2700 2641
2642 CSS + MgO
Temperature (K)
L+CAO SS
2582 TSS
CSS +CSS
CSS+C-CAZRO3
2485 1700
2300 2232 C-CAZRO3+CAO SS 1679, 0.126
(1679, 0.132) TSS + MgO M-ZrO2 + MgO
CSS-O-CAZRO3 1300
1900 O-CAZRO3+CAO SS 0.00 0.20 0.40 0.60 0.80 1.00
ZrO2 Mole fraction MgO MgO
1626
TSS 1589 Figure 8 MgO–ZrO2 phase diagram. CSS, TSS ¼ cubic,
1500 1413
1454
1418 tetragonal ZrO2 solid solutions, respectively; M-ZrO2 ¼
1397 monoclinic ZrO2. Reproduced from Du, Z.; Jin, Y.
1284
TSS+MSS MSS+O-CAZRO3 CALPHAD 1991, 15, 59–68.
1100
0 20 40 60 80 100
ZrO2 Mole percent CaO CaO
The systems ZrO2–MgO and PuO2–MgO are of
Figure 6 ZrO2–CaO phase diagram. CSS, TSS,
relevance for composite fuels with MgO as matrix.
MSS ¼ cubic, tetragonal, monoclinic ZrO2 solid solutions,
respectively; CaO SS ¼ periclase (CaO) solid solutions; The system ZrO2–MgO was assessed by Du and Jin,38
L ¼ liquid. Reproduced from Jin, Z.; Du, Y. CALPHAD 1992, on the basis of a large number of experimental studies
16, 355–362. (Figure 8). Zhang et al.39 assessed the PuO2–MgO
phase diagram on the basis of the experimental data
IMF (5–20 mol% PuO2), the effect is limited and at from Hough and Marples40 (Figure 9). The system
the operating temperature of a reactor a two-phase was found to be a simple eutectic with limited solu-
mixture (cubic þ tetragonal) can be formed. bility of MgO in PuO2.
Inert Matrix Fuel 245
2800
3.09.5.2.2 Composite fuels
Halite The overall thermal conductivity of a composite fuel
2600
+ strongly depends not only on that of the individual
liquid fcc components but also on the microstructure. Hellwig
+
liquid et al.23 estimated the thermal conductivity of different
fcc
2400 fuel types (Figure 10). For a composite made of
fcc
+
spherical particles that are randomly distributed in
halite a matrix, the overall thermal conductivity is given by
2200 the following equation43:
0.0 0.2 0.4 0.6 0.8 1.0
MgO xPuO
2 PuO2 lD lC lM 1=3
ð1 CD Þ ¼ ½2
Figure 9 MgO–PuO2 phase diagram. Reproduced from lD lM l C
Zhang, H.; Huntelaar, M. E.; Konings, R. J. M.; Cordfunke, where lM, lD , and lC represent the thermal conduc-
E. H. P. J. Nucl. Mater. 1997, 249, 223–230.
tivity of matrix, dispersed phase, and composite,
6.5 CSZ-IMF
6.0 UO2, unirradiated
5.5 UO2 at 60 MWd kg−1
5.0
4.5
4.0
3.5
3.0
2.5
2.0
1.5
1.0 Correlations given for 95% dense fuel
0.5
0.0
600 800 1000 1200 1400 1600 1800 2000 2200
Fuel temperature (K)
Figure 10 Estimated thermal conductivity of different fuel types. Reproduced from Hellwig, C.; Streit, M.; Blair, P.; et al.
J. Nucl. Mater. 2006, 352, 291–299.
246 Inert Matrix Fuel
2 3
MgO accelerators and compared their results to the behavior
of natural analogs. They observed no amorphization of
10
MgAl2O4 ZrO2 under Xe irradiation even under extreme condi-
tions. Swelling rates reported were low, that is, 0.19% at
room temperature and 0.72% at 925 K during irradia-
Equation [1]
tion with low energy (60 keV) Xe ions up to a Xe
0
500 700 900 1100 1300 1500 fluence of 1.8 1016 cm2. Further, for the study on
T (K) natural analogs no amorphization was reported. The
Figure 11 Thermal conductivity of different matrix authors conclude that under extreme irradiation con-
materials. ditions with very high ion doses, zirconia remains
crystalline and does not undergo amorphization.
respectively, and CD (0 C 1) represents the con- MgO is known to be stable against neutron
centration of the dispersed phase. A somewhat more irradiation at high temperatures;47 at low temperatures,
practical equation for the effective thermal conduc- at which the irradiation experiments have generally
tivity (leff) of a composite has been proposed by been performed, the formation of unfaulted, elongated
Miller44 for cermet fuels with spherical inclusions interstitial loops and networks has been observed result-
in a wide range of concentrations, and for lM lD: ing in significant swelling.48,49 Annealing to about
1=3 1273 K leads to recovery of about 60% of the radia-
leff tion damage effects created at low temperature. The
leff ¼ lD þ ð1 VD ÞðlM lD Þ ½3
lM stability of MgO towards heavy ions is not so well
The thermal conductivity of the zirconia phase known. Friedland50 observed a mixed damage structure
can be approximated by eqn [1] assuming that it of random disordered regions and extended defects
will not vary significantly as a function of the Pu after Kr implantation in single-crystal MgO at room
concentration. Moreover, it is much lower than the temperature. Further, in this case it was shown that
conductivity of possible matrix phases considered for annealing at high temperature leads to recovery of the
composite fuels (Figure 11). radiation damage effects, which is almost complete at
In practice, however, the fissile particles can have 1273 K. Neeft et al.51 showed that the dimensional
a random distribution and random shapes. Therefore, changes in macrodispersed MgO þ UO2 composite
eqns [1] and [2] give the upper limit for the thermal samples of about 90% theoretical density irradiated to
conductivity with respect to fissile particle distribu- high burnup (thus subjected to the combined effects of
tion. For randomly distributed particles, the thermal neutron and fission product impact) at a temperature of
conductivity can be significantly lower, depending on 700–1000 K are limited (0.9 1.0 DV/V (%)).
the difference in thermal conductivity of the two
materials (i.e., matrix and fissile inclusions).45 For 3.09.5.4 Chemical Properties
nonspherical particles, the situation becomes more
complex and finite element modeling is required to ZrO2 is a very stable material. It has a high melting point
obtain a reliable estimate of the overall thermal (2983 K) and a total vapor pressure of 0.01 Pa at
conductivity.45 2273 K.52 It is chemically quite inert; it is not soluble in
nitric acid and does not react with steam. These are
important properties of an inert fuel material, for exam-
3.09.5.3 Radiation Stability
ple, in the case of a cladding failure and penetration of
IMF made from stabilized zirconia is a very robust, the cooling liquid into the fuel pin. In absence of an
stable material. It forms solid solutions with lantha- oxidizing atmosphere, no interaction of ZrO2 and zirco-
nides, transition elements, or other fission products. nium metal is to be expected. In addition, in view of
A very important positive factor is its suitability is nuclear waste disposal, the resistance against chemical
its stability against radiation. Clinard et al.46 and corrosion of ZrO2, and herewith the ZrO2-based IMF, is
Vance and Boland47 performed neutron and fission highly advantageous for the direct disposal and storage
fragment irradiation experiments on ZrO2 and of the spent fuel.
Inert Matrix Fuel 247
MgO has an even higher melting point (3100 K) capsules containing ZrO2–PuO2 sintered fuel pellets
than ZrO2 and a total vapor pressure of 100 Pa at were irradiated in the ETR. Four capsules contained
2000 K.52 Its thermal conductivity of 13 W m 1K1 ZrO2–1.93 wt% PuO2 and four contained ZrO2–
at 1273 K is excellent compared to that of ZrO2 9.76 wt% PuO2 pellets. Alpha autoradiography of
(2.2 W m 1K1 at 1273 K). MgO does not react the sintered, nonirradiated ZrO2–1.93 wt% PuO2
with liquid sodium, which would favor its use in pellets revealed nonuniform distribution of the
sodium-cooled reactors. Its most significant disad- PuO2. In the ZrO2–9.76 wt% PuO2 pellets, the
vantage is, however, the formation of hydroxides PuO2 was homogeneously distributed. Detailed
with water. This makes fuel based on MgO alone information on the irradiation conditions can be
not attractive for water-cooled reactors. found in Freshley5.
No external dimensional changes were observed
as a result of the irradiation. In-reactor plasticity was
3.09.6 Irradiation Behavior evidenced by unusual radial cracking. Fission gas
data revealed that between 40 and 70% of the
3.09.6.1 The 1960s US Programs
xenon and krypton was released in some samples.
3.09.6.1.1 NUMEC: PuO2–UO2–ZrO2 The postirradiation structures indicated that the
For the irradiation tests at GETR, three irradiation maximum fuel temperatures in some specimens
capsules were prepared for irradiation to 9, 30, and probably exceeded the melting point of 2973 K.
50 MWd kg1 (Pu þ U þ Zr). Six fuel compositions Unusual structural features, probably caused by
were tested, including 6 and 10 vol.% PuO2, and 25, inhomogeneous PuO2 distribution and low operating
50, and 75 vol.% ZrO2, with the balance being UO2.7 temperatures, were developed in the ZrO2–1.93 wt%
The 235U enrichment was different for each of the six PuO2 samples during irradiation. Short radial cracks
compositions, varying from 2.2% to 21.5%. Unfortu- were formed near the periphery of the irradiated sam-
nately, PIE results from this program are scarce. ples and no columnar grains or central voids were
During irradiation, two pins containing the 10 vol.% observed in the specimens with calculated linear
PuO2–40 vol.% UO2–50 vol.% ZrO2 and 6 vol.% heat rates as high as 49 kW m1. Autoradiographs
PuO2–44 vol.% UO2–50 vol.% ZrO2 fuel composi- indicated that little diffusion of PuO2 occurred
tions failed as a result of molten fuel contacting the at estimated fuel temperatures as high as 1673 K.
cladding surface in the gas plenum region. Four pins Recrystallization of ZrO2 occurred in the central
showed diametrical swelling ranging from 1 to 2% part of the specimens with consequent movement of
average. Local diametrical swelling of 3.6% occurred voids to the grain boundaries. Localized high-PuO2
at the midpoint of the gas plenum chamber on one pin. regions were characterized by spherical voids and
Multichannel gamma scans showed local and axial unidentified white spherical particles.
redistribution of fuel in the fuel pins. Gross redistri- Figure 12 shows a cross-section of an irradi-
bution of the fuel occurred in the two fuel elements ated ZrO2–9.76 wt% PuO2 sample. Columnar grain
having external cladding failure. Axial reorientation growth and the formation of a central void were
of other fuel pins also occurred. Fuel was deposited observed and Freshley5 identified five distinct zones.
1.27 cm above the as-loaded fuel height, and also The outermost ring was found to be structurally
deposited on the walls of the tubing and spring. unchanged fuel (no recrystallization or grain growth)
Preliminary examination of the fuel indicated that with the exception of radial cracking and auto-
central melting probably occurred in all of the fuel pins radiographs indicated no discernible evidence of fission
indicating that the thermal conductivity of the ternary product migration. The next inner ring appeared trans-
oxides was significantly lower than the 1.8 W m 1K1 lucent and even though recrystallization occurred,
value used to estimate maximum design centerline there was no apparent fission product migration in
temperatures of
2573 K. The overall thermal perfor- this region. Close examination revealed a uniform dis-
mance of the ternary PuO2–UO2–ZrO2 fuels was thus tribution of very small white second-phase particles
significantly poorer than predicted. throughout the two outermost zones. The third ring
inward consisted of large grains from which voids had
3.09.6.1.2 Hanford: ZrO2–PuO2 migrated to the boundaries. Further, a high concentra-
The irradiation behavior of ZrO2–PuO2 fuels inves- tion of spherically shaped voids and white second-phase
tigated at Hanford has been described to a limited particles were associated with a high concentration of
extent in the literature.5 Eight Zircaloy–clad fuel fission products as indicated by autoradiography.
248 Inert Matrix Fuel
Transverse cross-section
4.4⫻
85⫻
Autoradiograph (positive)
4.4⫻
85⫻
Figure 12 Transverse cross-section of an irradiated ZrO2–9.76 wt% PuO2 fuel specimen. Estimated irradiation conditions:
86 kW/ m1 (18.9 kW/m2); maximum center temperature, 3400 C; maximum exposure, 0.61 1020 fissions cm3.
Fuel pin outer diameter: 1.44 cm. Reproduced from Freshley, M. D. In Plutonium Handbook; Wick, O. J., Ed.; The American
Nuclear Society: La Grange Park, IL, 1980; Vol. 2, pp. 643–706. # The American Nuclear Society.
The fourth ring inward was described as a high-density dispersed throughout. An autoradiograph indicated a
ring free of voids and second-phase particles, and rela- uniform fission product distribution in the center,
tively free of fission products. The central region was which was thought to be the initial molten region.
composed of columnar and equiaxed grain structures One ZrO2–9.76 wt% PuO2 capsule failed during
and relatively small voids with second-phase particles irradiation, resulting in severe hydriding and
Inert Matrix Fuel 249
extensive cracking of the Zircaloy cladding. The fail- indicated a uniform fission product distribution in
ure was thought to have been caused by fuel expan- these samples. Discrete PuO2 particles were uni-
sion resulting from excessive temperatures; it was formly dispersed throughout the MgO matrix.
estimated that 20% of the fuel volume was molten. Irradiation of MgO–PuO2 specimens at higher
No core–coolant reaction or fuel washout occurred. linear power (165 kW m1) caused central void for-
mation and columnar grain growth (Figure 13).
3.09.6.1.3 Hanford: MgO–PuO2 Large translucent columnar grains formed at high
Eight Zircaloy–clad capsules containing sintered temperatures near the center of the specimens. The
MgO–PuO2 fuel pellets were irradiated in the large grains were surrounded by a circumferential
ETR.5 Four of the capsules contained MgO–2.71 wt% ring of smaller opaque columnar grains, which were
PuO2 pellets and four contained MgO–12.95 wt% in turn surrounded by a ring of unaffected material
PuO2 fuel pellets. Minor cracking of the fuel was near the periphery. Ceramographic, autoradiographic,
observed in samples that operated at relatively and chemical analyses showed extensive radial reloca-
lower heat ratings (95 kW m1) and autoradiographs tion of PuO2 and fission products. The PuO2 and
4⫻ 4⫻
Transverse Autoradiograph
cross-section (positive)
Outer
surface Center
Opaque 50⫻
Undisturbed
material columnar Translucent columnar grains
grains
High concentration
of PuO2 and fission
products
Figure 13 Transverse section of an irradiated MgO–2.71 wt% PuO2 specimen. Estimated irradiation conditions:
1650 W cm1 (362 W cm2); maximum burnup, 0.61 1020 fissions cm3. Fuel pin outer diameter: 1.44 cm. Reproduced from
Freshley, M. D. In Plutonium Handbook; Wick, O. J., Ed.; The American Nuclear Society: La Grange Park, IL, 1980; Vol. 2,
pp. 643–706. # The American Nuclear Society.
250 Inert Matrix Fuel
fission products were rejected from the large translu- disk-shaped fuel targets containing plutonium as the
cent columnar grains and concentrated in a ring at the fissile material were irradiated for 97 days to demon-
boundary between the large and small columnar strate the feasibility of the ROX fuel concept.22,54
grains. A ring of PuO2 particles coincides with the Three fuel pins of a blend of PuO2, YSZ, MgAl2O4,
high concentration fission product ring. The distribu- and Al2O3 were irradiated. In the second test, pellet-
tion of the insoluble PuO2 second phase was not type fuels (20% enriched uranium as fuel instead of
altered in the smaller columnar grain region even Pu) were employed to obtain data on the basic irradi-
though extensive recrystallization occurred. The plu- ation behavior such as swelling, FGR, phase stability,
tonium concentration varied by two orders of magni- and microstructure change.55 Five different types
tude in the three zones of this sample. of fuel were deployed: YSZ–UO2 SS (U–YSZ),
The fraction of fission gases released from MgO– U–YSZ particles dispersed in a spinel or corundum
PuO2 fuel increased with increasing linear power and matrix (particle size 250 mm), and U–YSZ blended
higher fuel temperatures. FGR ranged from less than with a spinel or corundum matrix (particle size
5% at 95 kW m1 to as high as 38% at 165 kW m1. 10–50 mm).56,57 The fuels were irradiated for 100 days.
X-ray diffraction (XRD) patterns of the irradi- The third irradiation test was performed to confirm the
ated MgO and PuO2 showed no line broadening or irradiation behavior of ROX fuel. Three fuels were
changes in lattice parameter in the burnup range of used, containing 20% enriched U as fissile material:
0.32 1020 to 1.02 1020 fissions cm3. XRD pat- a single-phase fuel of U–YSZ, and two particle-
terns of the MgO–2.71 wt% PuO2 specimen irra- dispersed fuels of U–YSZ particles dispersed in a spi-
diated at high temperatures (Figure 13) showed a nel or corundum matrix. The fuels were irradiated for
three-fold reduction in the intensity of MgO reflec- 275, 248, and 245 days, respectively.34
tions from the region of high PuO2 and fission prod- The blended PuO2–YSZ–MgAl2O4–Al2O3 fuels
uct concentrations. showed remarkable swelling, that is, an increase in
A full-size experimental MgO–PuO2 19-rod cluster diameter of 11%. XRD analysis showed that another
fuel element was irradiated in the plutonium recycle phase was present in the fuel before and after irradia-
test reactor (PRTR).53 The Zircaloy–clad rods were tion: hibonite (M2+Al12O19 and/or M3+M2+Al11O19).
fabricated by swage compacting incrementally loaded The fraction of hibonite in the sample increased
arc-fused MgO and calcined PuO2 fuel whose nominal during irradiation because of the fission product-
composition was 2.1 wt% PuO2. Coolant activity lanthanides and by the migration of plutonium to
records suggest that the element failed approximately the hibonite phase. The final volume also increased
3 h after the reactor reached 60-MW operation. The through the conversion of corundum to hibonite.
test was continued for approximately 8 days and the The particle-dispersed fuels showed lower swelling
element was discharged because of excessive coolant than the blended fuels. This confirmed that the
activity. PIE revealed a longitudinal split (0.64-cm swelling behavior was improved by using larger YSZ
wide and 3.8-cm long) in one rod. Approximately particles. Among the particle-dispersed and blended
23 cm of fuel had been lost to the coolant. A combina- U–YSZ fuels the compound blended in a spinel matrix
tion of related factors probably caused the failure. High showed the largest swelling. The swelling was attribu-
local temperatures were generated at the point of rod ted to spinel and it was suggested that it was caused
failure because of unexplained plutonium segregation. by fission fragment damage.18 The swelling of the
Water released from the MgO in combination with corundum-based fuels was substantial, but not so
fluoride contamination in the PuO2 may have caused large as for the spinel-based fuels. It was considered
rapid corrosive penetration of the cladding. Ingress of that the irradiation temperature in this case was high
water and the continued operation of the fuel rod enough to anneal the radiation damage. XRD analysis
caused fuel swelling and splitting of the cladding due of the irradiated fuels clearly showed the existence
to the MgO–water reaction, hydriding of the Zircaloy of a crystalline corundum phase, with little indica-
cladding, and fuel washout. tion of amorphization, at irradiation temperatures of
800–1000 K. The U–YSZ SS fuels showed low swelling
rates.
3.09.6.2 JAEA: ROX
Although the blended PuO2–YSZ–MgAl2O4–
Three irradiation tests (see Table 1) of ROX fuels Al2O3 disk fuels showed high FGR, comparison of
were carried out in JRR-3 at JAERI at a nominal their FGR with of the other fuel types was difficult
reactor power of 20 MW. In the first test, small because of that fragmentation of the disks caused
Inert Matrix Fuel 251
Z1 Z2 SB CB
0.5 mm 0.5 mm
0.5 mm
Figure 14 Scanning electron microscopy images of ROX fuels. Z1, Z2 ¼ (Y,U,Zr)O2x solid solution, second, third
irradiation, respectively. SB, CB ¼ (Y,U,Zr)O2x þ spinel or corundum, respectively, homogeneously blended, second
irradiation. SD1, SD2, CD1, CD2 ¼ particle-dispersed (Y,U,Zr)O2x þ spinel or corundum, second, third irradiation,
respectively. Reproduced from Nitani, N.; Kuramoto, K.; Nakano, Y.; et al. J. Nucl. Mater. 2008, 376, 88–97.
by very high power density. The FGR of the the second campaign possibly led to a two orders of
single-phase YSZ fuel of the third irradiation experi- magnitude faster annealing and thus to a lower FGR.
ment was about 5%, in spite of a central-hole forma- The cross-sectional appearances of the fuels of the
tion, high irradiation temperature, and substantial second and third irradiations are shown in Figure 14;
burnup. Although the linear power of the fuel irradi- Figure 15 shows radial EPMA line scan profiles of U,
ation was comparable to conventional UO2 fuel, its Zr, Al, and Mg for a U–YSZ fuel blended with spi-
irradiation temperature was higher, because of the nel.18,57 Restructuring and central-hole formation
lower ZrO2 thermal conductivity. This fuel showed were observed in the spinel-based fuels of the second
similar level of FGR as UO2 fuel at similar volumet- irradiation campaign. These fuel pellets comprised
ric fission density and temperature. roughly three regions; the center region consisted
The FGR of the single-phase YSZ fuel of the only of YSZ, the middle region of YSZ and corun-
second irradiation experiment (lower burnup) was dum, and the outer region of the original YSZ and
relatively higher, which could have been caused by spinel (Figure 15). MgO was detected on the inner
the lower density of the fresh pellet (
86% TD). surface of the cladding tube. Thus, the spinel-based
The composite fuels showed higher FGR than YSZ fuel irradiated at high centerline temperatures of
single-phase fuels. Contrary to the swelling behavior, about 2000 K showed decomposition and restructur-
the FGR of particle-dispersed fuels was larger than ing. The high irradiation temperature, the large tem-
that of the blended fuels. The FGR of the U–YSZ perature gradient, and the low oxygen potential
particle-dispersed corundum fuel (third irradiation) caused spinel decomposition and subsequent vapori-
was extremely high, about twice as high as the zation of MgO. By examining the pellet cross-section
U–YSZ particle-dispersed spinel fuel. As the YSZ and the temperature distribution, Nitani et al.18 sug-
particles for the particle-dispersed fuels were almost gested that the vaporization of MgO occurred at
identical, and the irradiation conditions were al- temperatures exceeding 1700 K.
most the same, the difference in behavior should be The restructuring of the fuel pellet was not
linked to the difference in the properties of the observed in the MgAl2O4-based particle-dispersed
matrices. In contradiction to theory, the FGR of fuel of the third campaign irradiated below 1400 K.
the corundum-based fuels of the second irradiation, This was achieved by reducing the total power peak-
which were irradiated at higher temperatures, was not ing factor through a modification of the assembly
as high as for the corundum-based fuel of the third design of ROX and by addition of Er as a burnable
irradiation. Nitani et al.18 attributed this behavior to a poison. Aggregation of the YSZ particles and restruc-
slower radiation damage annealing of corundum in the turing were rarely observed for the corundum-based
fuel of the third campaign that was irradiated at lower fuels. Regardless of irradiation temperatures, the fuel
temperatures. The higher irradiation temperatures of pellets showed uniform structure throughout the
252 Inert Matrix Fuel
References 27. Kasemeyer, U.; Paratte, J. M.; Grimm, P.; Chawla, R. Nucl.
Technol. 1998, 122, 52–63.
28. Paratte, J. M.; Akie, H.; Chawla, R.; et al. Nucl. Technol.
1. Rubin, B. F.; Berman, R. M.; Bleiberg, M. L. The 2000, 130, 159–176; cited by (since 1996): 4.
irradiation behavior of ZrO2–UO2 fuels, Technical 29. Schneider, E.; Deinert, M.; Herring, S.; Cady, K. J. Nucl.
Report WAPD-264; Bettis Atomic Power Lab: Mater. 2007, 361, 41–51.
Pittsburgh, PA, 1962. 30. Fuels and Materials for Transmutation, OECD NEA No.
2. Wittels, M. C.; Sherrill, F. A. J. Appl. Phys. 1956, 27, 5419, 2005.
643–644. 31. Shiratori, T.; Yamashita, T.; Ohmichi, T.; Yasuda, A.;
3. Freshley, M. D.; Carroll, D. F. Trans. Am. Nucl. Soc. Watarumi, K. J. Nucl. Mater. 1999, 274, 40–46.
1963, 6, 396. 32. Fernández, A.; Haas, D.; Konings, R. J. M.; Somers, J.
4. Freshley, M. D.; Carroll, D. F. Trans. Am. Nucl. Soc. J. Am. Ceram. Soc. 2002, 85, 694–696.
1963, 6, 348–349. 33. Croixmarie, Y.; Abonneau, E.; Fernández, A.; Konings,
5. Freshley, M. D. In Plutonium Handbook; Wick, O. J., Ed.; R. J. M.; Desmoulière, F.; Donnet, L. J. Nucl. Mater. 2003,
The American Nuclear Society: La Grange Park, IL, 1980; 320, 11–17.
Vol. 2, pp 643–706. 34. Nitani, N.; Kuramoto, K.; Yamashita, T.; Ichise, K.;
6. Caldwell, C. S.; Fisher, F. D.; Roth, J. Trans. Am. Nucl. Ono, K.; Nihei, Y. J. Nucl. Mater. 2006, 352,
Soc. 1966, 9, 544. 365–371.
7. Puechl, K. H.; Caldwell, C. S.; Roth, J. PuO2–UO2–ZrO2 35. Hellwig, C.; Kasemeyer, U.; Ledergerber, G.; Lee, B. H.;
Fuel Development Program, Progress Report NUMEC- Lee, Y. W.; Chawlaa, R. Ann. Nucl. Energy 2003, 30,
3505–33; Nuclear Materials and Equipment Corporation: 287–299.
Apollo, PA, 1968. 36. Jin, Z.; Du, Y. CALPHAD 1992, 16, 355–362.
8. Carroll, D. F. J. Am. Ceram. Soc. 1963, 46, 194. 37. Kinoshita, H.; Uno, M.; Yamanaka, S. J. Alloys Comp.
9. Carroll, D. F. J. Am. Ceram. Soc. 1964, 47, 650. 2003, 354, 129–137.
10. Degueldre, C.; Kasemeyer, U.; Botta, F.; Ledergerber, G. 38. Du, Z.; Jin, Y. CALPHAD 1991, 15, 59–68.
In Proceedings of Materials Research Society Symposium; 39. Zhang, H.; Huntelaar, M. E.; Konings, R. J. M.;
Materials Research Society: Warrendale, PA, 1996; Cordfunke, E. H. P. J. Nucl. Mater. 1997, 249,
Vol. 412. 223–230.
11. Streit, M.; Wiesenack, W.; Tverberg, T.; Hellwig, C.; 40. Hough, A.; Marples, J. A. C. J. Nucl. Mater. 1965, 15,
Oberländer, B. C. J. Nucl. Mater. 2006, 352, 298–309.
349–356. 41. Ronchi, C.; Ottaviani, J. P.; Degueldre, C.; Calabrese, R.
12. Oberländer, B. C.; Jenssen, H. K.; Hellwig, C.; J. Nucl. Mater. 2003, 320, 54–65.
Nordström, L. A. In Enlarged Halden Programme Group 42. Hellwig, C.; Kasemeyer, U. J. Nucl. Mater. 2003, 319,
Meeting on High Burn-up Fuel Performance, Safety and 87–94.
Reliability, Lillehammer (NO), Oct 16–21, 2005. 43. Schulz, B. High Temp. High Press. 1981, 13,
13. Degueldre, C.; Wiesenack, W. In Proceedings of Materials 649–657.
Research Society Symposium. Materials Research 44. Miller, J. V. Estimating Thermal Conductivity of Cermet
Society: Warrendale, PA, 2008; Vol. 1104. Fuel Materials for Nuclear Reactor Application;
14. Restani, R.; Martin, M.; Kivel, N.; Gavillet, D. J. Nucl. Technical Report NASA TN D-3898; NASA, Washington,
Mater. 2009, 385, 435–442. DC, 1967.
15. Schram, R. P. C.; Bakker, K.; Hein, H.; et al. Prog. Nucl. 45. Bakker, K.; Konings, R. J. Alloys Comp. 1998, 271–273,
Energy 2001, 38, 259–262. 632–635.
16. Schram, R. P. C.; van der Laan, R. R.; Klaassen, F. C.; 46. Clinard, F. W.; Rohr, D. L.; Ranken, W. A. J. Am. Ceram.
Bakker, K.; Yamashita, T.; Ingold, F. J. Nucl. Mater. 2003, Soc. 1977, 60, 287–288.
319, 118–125. 47. Vance, E. R.; Boland, J. N. Radiat. Effects 1978, 37,
17. Akie, H.; Takano, H.; Muromura, T.; Nitani, N. Prog. Nucl. 237–239.
Energy 1995, 29, 345–352. 48. Wilks, R. J. Nucl. Mater. 1968, 26, 137–173.
18. Nitani, N.; Kuramoto, K.; Nakano, Y.; et al. J. Nucl. Mater. 49. Clinard, F. Jr.; Hurley, G.; Hobbs, L. J. Nucl. Mater. 1982,
2008, 376, 88–97. 108–109, 655–670.
19. Medvedev, P.; Frank, S. M.; O’Holleran, T. P.; Meyer, M. K. 50. Friedland, E. Nucl. Instrum. Meth. Phys. Res. B Beam
J. Nucl. Mater. 2005, 342, 48–62. Interact. Mater. Atoms 1994, 85, 316–320.
20. Medvedev, P. G.; Jue, J. F.; Frank, S. M.; Meyer, M. K. 51. Neeft, E. A. C.; Bakker, K.; Schram, R. P. C.;
J. Nucl. Mater. 2006, 352, 318–323. Conrad, R.; Konings, R. J. M. J. Nucl. Mater. 2003, 320,
21. Holliday, K.; Hartmann, T.; Mulcahy, S. R.; Czerwinski, K. 106–116.
J. Nucl. Mater. 2010, 402, 81–86. 52. Kleykamp, H. J. Nucl. Mater. 1999, 275, 1–11.
22. Yamashita, T.; Kuramoto, K.; Nitani, N.; et al. J. Nucl. 53. Evans, E. A. Ceramics Research and Development
Mater. 2003, 320, 126–132. Operation Quarterly Report October–December, 1962;
23. Hellwig, C.; Streit, M.; Blair, P.; et al. J. Nucl. Mater. 2006, USAEC Report NUMEC-3505–33HW-76300; Hanford
352, 291–299. Atomic Products Operation, General Electric Co.: Rich
24. Degueldre, C.; Hellwig, C. J. Nucl. Mater. 2003, 320, Land, Washington, 1963.
96–105; In Proceedings of the 2nd Seminar on European 54. Yamashita, T.; Nitani, N.; Kanazawa, H.; et al. J. Nucl.
Research on Materials for Transmutation. Mater. 1999, 274, 98–104.
25. Ledergerber, G.; Degueldre, C.; Heimgartner, P.; 55. Kuramoto, K.; Yamashita, T.; Shiratori, T. Prog. Nucl.
Pouchon, M. A.; Kasemeyer, U. Prog. Nucl. Energy 2001, Energy 2001, 38, 423–426.
38, 301–308. 56. Yamashita, T.; Akie, H.; Nakano, Y.; Kuramoto, K.;
26. Yamashita, T.; Akie, H.; Kimura, H.; Takano, H.; Nitani, N.; Nakamura, T. Prog. Nucl. Energy 2001, 38,
Muromura, T. In IAEA/TCM Fuel Cycle Options for 327–330.
LWRs and HWRs, Victoria & Canada, Apr 28–May 1, 57. Nitani, N.; Kuramoto, K.; Yamashita, T.; Nihei, Y.;
1998. Kimura, Y. J. Nucl. Mater. 2003, 319, 102–107.
256 Inert Matrix Fuel
58. de Haas, G. J. L. M.; Klaassen, F.; Schram, R.; et al. Engineering (ICONE17), July 12–16, 2009, Brussels,
In HOTLAB European Hot Laboratories Research Belgium.
Capacities and Needs Plenary Meeting 2004, Halden, 60. Zabudko, L.; et al. In Transactions of 6th Russian
Norway, Sept 6–8, 2004. Conference on the Reactor Material Science, Sept 11–15,
59. Calabrese, R.; Vettraino, F.; Tverberg, T. In Proceedings 2000. SSC RF Research Institute of Atomic Reactors:
of the 17th International Conference on Nuclear Dimitrovgrad, Russia; pp 147–150.
3.10 Composite Fuel (cermet, cercer)*
M. K. Meyer
Idaho National Laboratory, Idaho Falls, ID, USA
257
258 Composite Fuel (cermet, cercer)
Cladding
Coating
(a)
Fuel
(b)
Figure 1 (a) Schematic representation of composite fuel. (b) Schematic of a composite fuel using a coated fuel particle
in which the coating acts as a buffer layer between the fuel particle and the matrix material.
microdispersions.9 The particle-size distribution in stress on the matrix due to gas-driven pressure.
macrodispersions is selected to retain the majority Dispersions using this concept have been proven to
of fission fragments within the fuel particles and work well in combination with ductile matrices such
allow the majority of the matrix material to remain as steel and niobium.10
undamaged. Macrodispersed fuels often use a fissile A buffer layer can be used to more effectively
particle size in the range of 50–200 mm. Microdisper- isolate the fuel particles from the matrix using a
sions have been considered for some concepts buffer, as depicted in Figure 1(b). The buffer may
because of convenience of fabrication or prevention be used to provide free volume for fission gas accu-
of matrix cracking in cercer fuels during irradiation. mulation or fuel particle swelling without exerting
The fissile particle size in microdispersions is typi- a stress on the matrix. The buffer layer may be a
cally on the order of a few micrometers. coating or a series of coatings applied to th parti-
Two types of fuel particle designs have been con- cle surface, or it may be composed of free space.
sidered for macrodispersed fuels. These are particles A highly developed variant of coated particle fuel
fabricated with and without a coating layer. The (TRISO) is used in high-temperature gas reactors
simplest dispersion fuel concept and the one with (HTGR) and is considered for use in water-cooled
the highest potential for fuel loading consists simply reactors.11 Incorporation of fuel particles of this type
of a fuel particle embedded in an inert matrix. This into an inert matrix has the potential to increase fuel
particle may incorporate a thin barrier coating to reliability and burnup performance, but at the
prevent the occurrence of fuel–matrix chemical expense of fuel heavy-metal density. Use of free
interaction during fabrication and irradiation. Such space between fuel particles in cercer fuels helps to
particles may include intentional porosity to provide prevent cracking in the ceramic matrix due to fuel
free volume for fission gas, thereby reducing the local particle swelling.12
260 Composite Fuel (cermet, cercer)
180.0
Vf = 0.1
160.0 Vf = 0.2
Vf = 0.3
100.0
80.0
60.0
40.0
20.0
0.0
0 100 200 300 400 500 600
Particle diameter, D (mm)
Figure 3 Plot of distance between particle surfaces as a function of fuel particle diameter for fuel particle volume
fractions (Vf) between 0.1 and 0.5.
150.0
140.0
130.0
120.0
7
110.0 14
21
Undamaged ligand thickness (mm)
100.0 29
36
90.0 43
50
80.0
70.0
60.0
50.0
40.0
30.0
20.0
10.0
0.0
0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.5 0.55
Fuel volume fraction
Figure 4 Plot of undamaged matrix ligand thickness as a function of fuel particle volume loading. The legend indicates
the ratio of particle diameter to fission fragment range (D/2lm). For example, for D/2lm ¼ 14, the particle diameter is 28 times
the fission fragment range.
262 Composite Fuel (cermet, cercer)
1.00
0.90
0.80
0.70
Fraction undamaged matrix, Vudm
0.60
0.50
0.40
0.30
Vf = 0.01
Vf = 0.05
0.20 Vf = 0.1
Vf = 0.2
Vf = 0.4
0.10 Vf = 0.5
Vf = 0.6
0.00
0 100 200 300 400 500
Particle diameter, D (mm)
Figure 5 Volume of undamaged matrix for a fission fragment range of 10 mm in an ideal dispersion. Vf is the fuel
particle volume loading.
of particles results in a larger fractional volume of temperature history, and the creep exponent and acti-
matrix damage. vation energy for creep of the matrix material. The
" 3 # calculated creep parameter was compared to experi-
Vf 1 mental data to determine a value of a that represented
Vudm ¼ 1 1þ 1 ½3
1 Vf D=2lm a failure threshold for a specific fuel loading and
particle size. Wier explicitly modeled dispersion fuel
Because cermet fuels depend on retention of matrix as a spherical shell of matrix material containing a
properties, deviations from the ideal microstructure fuel particle.15 Failure was assumed when internal
shown in Figure 1 result in larger matrix damage fission gas pressure caused the outer fibers of the
fractions. The behavior of cermet fuels also depends sphere to flow plastically. In this model, the effect
on the retention of matrix strength as a function of of fuel particle porosity in delaying pressurization of
temperature. the shell and the effect of fuel particle size and fuel
These principles have been applied to models that volume fraction on the thickness of the spherical
have attempted to quantify the failure threshold shell are accounted for explicitly. While both these
based on creep or plastic flow of the matrix due to models provided qualitative guidance to fuel designers,
fission gas pressurization. Keller et al. postulated a neither of them was able to analytically predict failure
mechanism based on the general proportionality because of the complex stress state of the matrix and
between fission gas release as the result of burnup irradiation-induced material property changes.
and creep in the matrix.14 A creep parameter (a) was More recently, finite element methods have been
derived as a function of burnup, irradiation time, applied to quantify the stress state in cermet
Composite Fuel (cermet, cercer) 263
dissolution. Liquid droplets are formed by using a closely spaced, resulting in an opportunity for the
rotating disk, cup, or spray orifice and quenched in overlap of fission fragment damage zones and a higher
an ammonia bath where they are converted to parti- probability of matrix cracking.
cles. The particles are then washed, dried, and cal- Development work thus focused on the achieve-
cined to form spherical beads. ment of an ideal dispersion, culminating in the late
1960s with the coating of fuel particles with matrix
3.10.4.1.3 Infiltration of porous microspheres material to achieve a uniform particle spacing, fol-
In the case of composite fuel concepts where low lowed by HIP (hot isostatic pressing) processing
heavy-metal density is preferable, as is often the to densify the fuel core.30 Fuel particle coating was
case for targets for plutonium or minor actinide incin- typically accomplished by coating the particles with a
eration, infiltration of a porous microsphere may be binder and then mixing with fine matrix powder.22
beneficial. Porous ceramic beads are subjected to These fabrication methods produced nearly ideal
repeated steps of infiltration with metal nitrate solu- microstructures, although few irradiation test data
tions followed by calcining to achieve heavy-metal are available to verify the fuel performance of these
mass fractions of 40 wt% or more.27,28 nearly perfect microstructures.
2.500
Upper bound
Lower bound
Thermal conductivity (W dt–1 ⬚C)
1.000
0.500
0.000
0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70 0.80 0.90 1.00
Volume fraction of fuel + reaction
Figure 6 Plot of measured thermal conductivity versus thermal conductivity calculated using eqn [11] (labeled
autocoherent conductivity). Also shown are upper and lower bounds from Hashin’s theory.
266 Composite Fuel (cermet, cercer)
fuels46; however, the range of applicability of eqn [12] fast-reactor fuels. Stainless steel–oxide dispersions
is not as broad as that of eqn [11]. were also developed for thermal reactor applications
such as the portable military (SM, PM) experimental
Kcomp 1=3
Kcomp ¼ Kf þ ð1 Vf ÞðKm Kf Þ ½12 units. Data regarding the performance of these fuels
Km are reviewed below. Early papers by Weber and cow-
The Maxwell–Euken equation [13] can be used in orkers4,53 and a later overview of early work in the
many cases to correct the porosity of the matrix or United Kingdom published in 196422 provide a useful
fuel phases47: summary of progress typical of developmental work
1P on stainless steel matrix cermet fuels.
KP ¼ K100 ½13 Early attempts at fabrication and testing produced
1 þ bP
dispersion fuels capable of burnups on the order of
where KP is the effective conductivity of the porous 2 at.% (total uranium). As irradiation testing and eval-
matrix, K100 is the conductivity of the fully dense uation progressed into the 1960s, it became clear
material, and P is the pore volume fraction. The that fuel performance was heavily dependent on
effect of pore shape on conductivity is taken into the microstructural characteristics of the dispersion.
account by the correction factor b. The value of b is Failure of early specimens was typically by matrix
1/2 for randomly dispersed spherical pores and >2/3 cracking, initiated at lamellar features in the micro-
for oblate spheroids aligned perpendicular to the structure (termed ‘stringers’) formed on fuel fabrica-
direction of heat flux. The measured thermal con- tion. As fuel particles swelled and induced stress on
ductivity of spinel–yttria-stabilized zirconia (YSZ) the matrix, the fuel plates would tend to unzip along
composite fuels has also been compared with the the oxide stringers. This was corrected by careful
value calculated by a form of the Maxwell–Euken attention to fabrication techniques (see Section
equation and found to agree within 5% for particle 3.10.4.2). In 1960, SM-2 series reactor fuel test speci-
volume fractions of 0.225 and 0.823, but not for an mens with a volume loading of 20% reached burnups
intermediate loading of 0.436.48 of 40–45 at.% in the material test reactor at surface
Finite element analysis methods can also be used temperatures of 370 C.29 No plate had a density
to explicitly model the contribution of fuel particles, decrease of more than 3% during this test. Burnup
matrix, and interface to thermal conductivity.49 limits of approximately 74 at.% at 538 C were later
BeO and SiC have been modeled as a second phase estimated for this fuel based on a series of irradia-
dispersed in UO2 matrix using this method. The tions to 84 at.%.54 Full-size 20 vol% plate-type ele-
model was and found to be in good agreement with ments fabricated using spherical uranium particles
the experimentally measured thermal conductivity reached 65 at.% burnup (thermal flux) at surface
BeO–UO2 composite fuel. temperatures of 315–427 C with no blistering or
cracking.55 Increasing the fuel loading to 27 vol%
3.10.6 Irradiation Behavior resulted in severe matrix cracking after irradia-
tion to 61 at.% burnup. An estimated burnup limit
3.10.6.1 Cermet Fuels of 27 at.% at 482 C was established for a thin clad
Most of the data available on the irradiation behavior 27 vol% fuel intended for Fermi Core B.56 It may be
of cermet fuels (other than aluminum-based research that the thin cladding caused the fuel to perform
reactor fuels) was generated in the 1950s and 1960s. poorly. Plate-type fuel developed at Harwell with
Some transient performance data have been gener- 30 vol% UO2 loading survived to 16.2 at.% burnup
ated on specific systems. Analysis has also been per- at approximately 620 C in a fast flux. This value was
formed to attempt to evaluate the severe accident achieved using a uniform dispersion of low-density,
behavior50,51 and neutronic implications of a high spherical oxide particles. A plate loaded to 50 vol%
core zirconium inventory.52 UO2 reached a burnup of 13.5 at.% without failure,
although matrix damage was evident on postirradia-
3.10.6.1.1 Stainless steel matrix fuels tion examination. The irradiation characteristics of
Most of the fuel performance data available for dis- nitride dispersions in stainless steel were reported
persion fuels outside of the research reactor operat- to be similar to those of oxides in the temperature
ing environment have been compiled on stainless range of 930–1090 C to 5 at.% burnup.57 UN irra-
steel–oxide dispersions. These fuels were initially diated in SM-2 tests failed in a manner similar to
developed in the late 1950s and early 1960s as oxides with the same U density.
Composite Fuel (cermet, cercer) 267
The general conclusion that can be drawn from volume loading. UKAEA (United Kingdom Atomic
these data is that higher fuel volume loading com- Energy Agency) fuel produced in the later stages of
bined with higher irradiation temperatures decrease development, for example, performed much better
the potential for high burnup. These factors are than predicted by this plot.
illustrated on a plot of fuel core fission density
versus fuel surface temperature in Figure 7. Fuel 3.10.6.1.2 Zirconium and silicon–aluminum
core fission density takes into account both the alloy cermet fuel
effects of fuel loading and fuel particle burnup Both steady-state and severe accident data have
(e.g., 25 vol% loading at 50 at.% burnup is equiva- been reported for UO2 cermet fuel with a silicon–
lent to 50 vol% loading at 25 at.% burnup), so that aluminum (silumin) alloy matrix.58 In-pile tests of
various combinations of fuel loading and burnup fuel specimens manufactured with defects showed
may be compared on a single plot. The solid line that minimal fission products release at normal pH
indicates estimates of temperature and burnup con- and a maximum fuel temperature of 617 K. Reactivity
ditions under which fuel is expected to fail from insertion accident testing to energy deposition levels
those where stable fuel performance is expected. of 837–962 J g1 of fuel did not result in fuel damage.
The two dashed lines chart the increase in fuel Irradiation tests to a burnup of 100 MWd kg1
burnup life as fabrication techniques are improved. U have been conducted without leaks or volume
The predictions of this plot appear to be conserva- changes for both silumin and zirconium alloy matrix
tive, possibly skewed to the left by early experience fuels.59
and by the thin clad Fermi Core B fuel performance Reactions of YSZ with zirconium, silicon, and
data (open squares). Microstructures that approach Zircaloy-4 were investigated.60 Little interactions of
that of a uniform dispersion of spheres allow for the metals with the ceramic were observed below
better performance at high temperatures and high 1273 K.
BMI-32
36
ORNL-46
Probably stable Probably unstable
BMI-33
32
BMI-GCRE
28 KAPL
Fission density (fissions cm–3)
NDA-1
PWAC
24 ORNL-21
(8)
( ) Number of
20
samples
(8) ORNL (1963)
16 (8) ORNL (1958)
BMI (1961)
(12) (8)
12 Open symbols indicate
(2) failure
Filled symbols
8 (8) indicate nonfailure
(8)
4 (8)
(8) (2) (2)
(8)
(2)
(2)
0
0 200 400 600 800 1000 1200 1400 1600 1800 2000 2200 2400
Surface temperature (⬚F)
Figure 7 Fuel performance data summary for stainless steel matrix UO2 composite fuels. Reproduced from
Thurber, W. C.; McQuilkin, F. R.; Long, E. L.; Osborne, M. F. Report ORNL-3709; Oak Ridge National Laboratory, 1964.
268 Composite Fuel (cermet, cercer)
amorphization of the spinel matrix that triggered a The range of neutron irradiation experiments
large volume swelling of approximately 15%. These with spinel using micro- and macrodispersions of
results indicate that fission fragment-driven swelling uranium, and dispersions of americium discussed
behavior in MgAl2O4 is markedly different at differ- above, along with irradiation with swift heavy
ent irradiation temperatures. ions80–86 and studies of self-aging87 indicate that the
The EFTTRA-T3 irradiation experiment68 in- large volume of helium generated88 as a result of
cluded uranium-bearing cermets in MgO, MgAl2O4, irradiation of 241Am coupled with amorphization
Y3Al5O12, Y2O3, and CeO2 matrices subjected to a of the spinel matrix at lower temperatures are
burnup of 17–20% fraction initial metal atoms at responsible for the observed high swelling rates.
700–1000 K. The volume loading of UO2 was nomi-
nally 2.5%. MgAl2O4 matrix specimens were prepared 3.10.6.2.2 Magnesia-based composite fuels
in both the macro- and microdispersed form. MgO, Magnesium oxide has been investigated as a compos-
Y2O3, and CeO2 matrices were found to perform well ite fuel matrix and as a second phase used to increase
in these tests. Experimental results indicated that the thermal conductivity of zirconium oxide and
Y3Al5O12 was not suitable as a matrix material due to pyrochlore phases that act as actinide hosts.
high swelling and high fission gas release. Microdispersions of UO2 in MgO were irradiated
A series of three irradiation tests performed in the to a uranium burnup of 1.3% over 61 days (fast flu-
JRR-3 reactor on rock-like oxide fuels (ROX) includ- ence 2 1026 n m2) at a calculated peak centerline
ing MgAl2O4 and MgAl2O4/Al2O3 (spinel/corun- temperature of approximately 1073 K in the fast neu-
dum) composite fuels provided similar data on the tron spectrum Phénix reactor.66 The MgO matrix
high swelling rate of MgAl2O4 microdispersions at an pellets exhibited a slight increase in porosity.
irradiation temperature of 980 K.69–73 Swelling was The melting behavior of materials in the Pu–Am–
lower in macrodispersed fuels, although fission gas Mg–O system has been investigated and summarized
release was higher due to matrix cracking. At fuel by Miwa et al.89 Eutectic melting was observed at
temperatures above 1700 K, severe restructuring of 2573 K.
the fuel was observed in both the spinel matrix The thermal conductivity of Nd2Zr2O7 pyro-
macro- and microdispersions. The irradiated pellet chlore was experimentally determined to be approxi-
structure consisted of a central hole, an adjacent region mately 1.3 W m1 K1, relatively independent of
consisting only of the YSZ fissile phase, a two-phase temperature in the range of 500–1600 K, resulting in
corundum/YSZ band, and another band consisting of unacceptably high pellet centerline temperatures at
the original YSZ, corundum, and spinel phases due to normal fuel pin operating power (250 W cm1).90
spinel decomposition and MgO migration away from Calculations showed that incorporation of magnesia
the center of the pellet to the periphery of the fuel pin. as a matrix for the pyrochlore host phase could be
MgAl2O4 has shown poor performance when used used to increase thermal conductivity by a factor of
as a matrix phase for americium. A microdispersion approximately 4 for a composite with 50 vol% MgO.
of 11 wt% americium in magnesium aluminate spinel Fabrication studies,91 heavy ion irradiation,92,93 mea-
was fabricated by infiltration and irradiated as part surement of thermal diffusivity,94,95 and systems anal-
of the EFTTRA-T4 experiment74–76 to 92% ameri- ysis of the use of MgO–pyrochlore composites for
cium transmutation at a fuel centerline temperature transmutation in boiling water reactors have been
near 900 K. The fuel experienced gross swelling of conducted.
18 vol%. The fabrication, thermal properties, and dissolu-
Both micro- and macrodispersions were irradiated tion behavior of dual-phase MgO/(ZrO2) ceramics
in the OTTO (once through then out) irradiation have been investigated as a composite material with
experiment to investigate the irradiation stability of the goal of improving the thermal conductivity of
zirconia-based solid solution fuel particles in a spinel zirconia, which appears to be an excellent actinide
matrix.77–79 The dispersions were irradiated at a peak host phase, but for its thermal conductivity.96–100
measured temperature between 1032 and 1345 K. In all cases, the matrix remains as pure MgO, with
During this test, the microdispersed fuel pin oper- the host phase stabilized with MgO. It was shown
ating at the lowest temperature failed due to high that significant increases in thermal conductivity
radial fuel swelling at a burnup of 136 kWd cm3. were obtained in as-fabricated material relative to
The macrodispersed fuels exhibited low swelling single-phase zirconia, with values in the range of
and fission gas release of 4.3–6.0%. 6–9 W 1 K1 at 1273 K.
270 Composite Fuel (cermet, cercer)
3.10.6.2.3 Transient testing of oxide from 1.6 to 5% as a fraction of initial metal atoms.
cercer fuels Fractional fission gas release was measured during
Pulse irradiation experiments were performed on low-burnup tests and found to be on the order of 106.
ROX fuels to investigate the fuel behavior under
reactivity insertion accident conditions.101,102 Macro- 3.10.6.2.5 Other carbides and nitrides as
and microdispersions of (Y,U,Zr)O2X in MgAl2O4 fuel matrix materials
(spinel) fuel were tested to peak fuel enthalpies of Zirconium carbide, titanium carbide, aluminum
2.2 kJ g1. The failure threshold for the ROX fuel was nitride, zirconium nitride, silicon nitride, and tita-
determined to be approximately 1.9 kJ g1, comparable nium nitride have been considered as inert matrices
to the failure threshold of unirradiated UO2 fuel. for specific high-temperature reactor concepts.109–112
Relatively little information is available regarding the
3.10.6.2.4 Carbide matrix composite fuels irradiation behavior of these materials.
SiC is an attractive composite fuel matrix material ZrC has been used as a coating in TRISO-type
due to good thermal conductivity and resistance to fuel and has therefore been of interest to fuel devel-
structural changes caused by neutron irradiation (see opers. A bibliography of ZrC irradiation testing
Chapter 4.07, Radiation Effects in SiC and SiC- results and recent data on the irradiation behavior
SiC). The thermal conductivity of SiC is known to of substoichiometric ZrC113 found swelling of <0.2%
decrease substantially with neutron irradiation103 and after irradiation to neutron fluences in the range of
heavy ion irradiation.104–107 Studies on fabrication, 1–10 1021 n cm2 at 908–1753 K. Small increases in
thermal conductivity, coolant and cladding compati- hardness, indentation fracture toughness, and elastic
bility, and heavy ion irradiation41 of SiC composite modulus were noted. Thermal conductivity was
fuels with oxide mass fractions less than 25% unchanged at irradiation temperatures of 1573–
concluded that (using realistic values for thermal 1753 K, but decreased slightly at 1273 K. Specimens
conductivity of neutron-irradiated SiC) calculated of slip cast/sintered and explosion-pressed ZrC were
fuel centerline temperatures would be approximately irradiated in the engineering test reactor in the late
1000 K lower than UO2 at the same heat generation 1960s at two different temperatures (130–355 C and
rate. No swelling was measured after irradiation with 1000–1100 C) to a fast neutron fluence of 5 1021
72 MeV iodine ions to a fluence of 1 1015 cm2 at n cm2.114 Volume swelling in the lower temperature
room temperature to 1473 K. range was 3%.
Figure 9 shows lateral and transverse cross- Few data are available on the irradiation behavior
sectional images of a fuel rod (1.27 cm diameter) of TiC. Specimens of slip cast and sintered and
consisting of a dispersion of pyrocarbon-coated par- explosion-pressed TiC were irradiated in the engi-
ticles in an SiC matrix formed by reaction bond- neering test reactor test, described above. The irradi-
ing.108 This fuel was fabricated and irradiation ation behavior under was similar to that of ZrC, with
tested as part of the high-temperature advanced gas swelling of material irradiated at low temperatures
reactor fuel development effort in Great Britain. in the range of 2–3 vol%, and swelling of material
Irradiation experiments were conducted at tempera- irradiated at the higher temperature in the range of
tures between 750 and 1200 C and rod powers of 0.25–0.5 vol%.
39 kW m1 in two experiments. Burnup values ranged Aluminum nitride neutron irradiation properties
have been studied at both high and low fluence in a
fast neutron spectrum. It was found that AlN exhib-
ited anisotropic swelling above 2.4 1020 n m2 at
743 K as would be expected from an HCP ceramic.
Specimens increased 0.15% in volume when irra-
diated to 5.3 1021 n cm2. No cracks or voids were
found on microstructural examination, and bending
strength remained relatively high at 350 MPa. At a
fluence of 9.1 1021 n cm2, however, bending
strength was reduced to less than 10 MPa due to
Figure 9 Silicon carbide matrix fuels containing
microcracking at grain boundaries caused by swelling
pyrocarbon-coated fuel particles after irradiation to 5% anisotropy. Thermal diffusivity decreased nearly
burnup (fraction initial metal atoms) at 1473 K. 20% on specimens irradiated to 8.3 1022 n cm2.
Composite Fuel (cermet, cercer) 271
Thermal diffusivity decrease with further irradiation 5. Keller, D. L. Nucleonics 1961, 19(6), 45–48.
6. Power Reactors and Experiments the World Around,
was minimal.115,116 Nucleonics 1960, 18(11), 148.
A TiN specimen irradiated in the Phénix reactor 7. Hofman, G. L.; Snelgrove, J. L. In Dispersion Fuels in
showed no detectable swelling on radiographic exam- Materials Science and Technology; Cahn, R. W.,
Haasen, P., Kramer, E. J., Eds.; VCH Verlagsgesellschaft
ination after a fast neutron fluence (E > 0.1 MeV) of mbH: Weinheim, Germany, 1994; Vol. 10A, Chapter 2.
1.95 1022 n cm2.66 8. Lambert, J. D. B. Proc. Br. Ceram. Soc. 1967, 7,
331–342.
9. Chauvin, N.; Konings, R. J. M.; Matzke, H. j. J. Nucl.
3.10.6.2.6 Graphite matrix fuels Mater. 1999, 274, 105–111.
In 1955, based on the requirements of the Rover Mars 10. Keller, D. L.; Cunningham, G. W.; Murr, W. E.; Fromm, E. O.;
Lozier, D. E. High-Temperature Irradiation Test of UO2
exploration program, the NERVA (Nuclear Engine Cermet Fuels; Report BMI-1608; Battelle Memorial
for Rocket Vehicle Application) nuclear rocket Institute, 1963.
engine concept was developed. The NERVA engine 11. Porta, J.; Pinto, P. Lo.; Bonnet, M.; et al. Prog. Nucl.
Energy 2001, 38, 407–410.
forced hydrogen through a solid core operating at 12. Noirot, J.; Desgranges, L.; Chauvin, N.; Georgenthum, V.
temperatures near 2500 C. Graphite, although not J. Nucl. Mater. 2003, 320, 117–125.
resistant to corrosion by hydrogen, was selected as 13. White, D. W.; Beard, A. P.; Willis, A. H. Irradiation
Behavior of Dispersion Fuels; Report KAPL-P-1849;
the matrix material for the first series of reactors Knolls Atomic Power Laboratory Report, 1957.
tested, the KIWI reactors.117 Early KIWI fuel designs 14. Keller, D. L.; Hulbert, L. E.; Dunnington, B. W. A Method
used a dispersion of fine (4 mm) UO2 particles in a of Correlating Irradiation Effects in Dispersion Fuels;
Report BMI-1408; Battelle Memorial Institute, 1960.
graphite matrix, which was converted to UC2 on 15. Wier, J. R. A Failure Analysis for the Low-Temperature
element fabrication at high temperature. Severe Performance of Dispersion Fuel Elements; Report
graphite corrosion in the KIWI-A test led to a series ORNL-2902; Oak Ridge National Laboratory Report, 1960.
16. Van Duyn, L. V Master’s Degree Thesis in Mechanical
of tests using fuel elements coated with niobium Engineering; Georgia Institute of Technology: Atlanda,
carbide. Successively better coating methods resulted GA, 2003.
in the KIWI-A3 test operating for approximately 17. Ding, S.; Jiang, X.; Huo, Y.; Li, L. J. Nucl. Mater. 2010,
374, 453–460.
5 min at 100 MW, with minimal damage to the fuel 18. Neeft, E. A. C.; Bakker, K.; Belvroy, R. L.; et al. J. Nucl.
elements. Beginning with KIWI-4E, the fuel particle Mater. 2003, 317, 217–225.
design was changed to a 50–150-mm-diameter UC2 19. Lutz, H. E.; Claussen, N. J. Eu. Ceram. Soc. 1991, 7,
209–218.
kernel that was coated with pyrolytic carbon to pre- 20. Mardon, P. G. Powder Metall. 1983, 26(1), 1–9.
vent oxidation of the UC2 during storage, which 21. Goslee, D. A. Nucleonics 1963, 21(7), 48–52.
caused loss of graphite when it was subsequently 22. Frost, B. R. T.; Cope, L. H.; Lambert, J. D. B.; et al.
In Proceedings of the 3rd Geneva Conference P/153,
reduced back to UC2 during reactor operation. 1964.
Later reactor designs used a ZrC coating to allow 23. Fernandez, A.; McGinley, J.; Somers, J.; Walter, M.
reactor operation for 60 min at 2400–2600 K, al- J. Nucl. Mater. 2009, 392, 133–138.
24. Idemitsu, K.; Arima, T.; Inagaki, Y.; Torikai, S.;
though spalling of the ZrC coating resulted due to Pouchon, M. A. J. Nucl. Mater. 2003, 319, 31–36.
differences in the coefficient of thermal expansion of 25. Fernandez, A.; Haas, D.; Konings, R. J. M.; Somers, J.
the coating and matrix. The final reactor test bed J. Am. Ceram. Soc. 2002, 85, 694.
26. Vettraino, F.; Magnani, G.; La Torretta, T.; et al. J. Nucl.
for the NERVA program was the NF-1 (Nuclear Mater. 1999, 274, 23–33.
Furnace-1),118,119 which showed that graphite-based 27. Fernandez, A.; Richter, K.; Somers, J. J. Alloys Comp.
fuel elements could withstand a temperature of 1999, 271–273, 616–619.
28. Richter, K.; Fernandez, A.; Somers, J. J. Nucl. Mater.
2400 K for 110 min with graphite loss of 1–2%. 1997, 249, 121–127.
29. Paprocki, S. J.; Keller, D. L.; Cunningham, G. W.;
Kizer, D. E. Report BMI-1487; Battelle Memorial Institute,
1960.
References 30. Weimar, P.; Thummler, F.; Bumm, H. J. Nucl. Mater.
1969, 31, 215–225.
31. Fernandez, A.; Koenings, R. J. M.; Somers, J. J. Nucl.
1. Holden, A. N. Dispersion Fuel Elements; Gordon and Mater. 2003, 319, 44–50.
Breach Science: New York, 1967. 32. Savchenko, A. M.; Vatulin, A. V.; Morozov, A. V.; et al.
2. Samoilov, A. G.; Kashtanov, A.; Volkov, V. Dispersion J. Nucl. Mater. 2006, 352, 372–377.
Fuel Nuclear Reactor Elements; Israel Program for 33. Savchenko, A. M.; Vatulin, A. V.; Morozov, A. V.; et al.
Scientific Translations: Jerusalem, 1968. J. Nucl. Mater. 2006, 352, 334–340.
3. Bhattacharyya, S. K. An Assessment of Fuels for Nuclear 34. Savchenko, A.; Konovalov, I.; Vatulin, A.; et al. J. Nucl.
Thermal Propulsion; Report ANL/TD/TM01-22; Argonne Mater. 2007, 362, 356–363.
National Laboratory, 2001. 35. Pouchon, M. A.; Nakamura, M.; Hellwig, C. h.; Ingold, F.;
4. Weber, C. E. Prog. Nucl. V Metall. Fuels 1959, 2, 295–362. Degueldre, C. J. Nucl. Mater. 2003, 319, 37–43.
272 Composite Fuel (cermet, cercer)
36. Shiratori, T.; Yamashita, T.; Ohmichi, T.; Yasuda, A.; 67. Noirot, J.; Desgranges, L.; Chauvin, N.; Georgenthum, V.
Watarumi, K. J. Nucl. Mater. 1999, 274, 40–46. J. Nucl. Mater. 2003, 320, 117–125.
37. Kleykamp, H. J. Nucl. Mater. 1999, 275, 1–11. 68. Neeft, E. A. C.; Bakker, K.; Schram, R. P. C.;
38. Matzke, H. j.; Rondinella, V. V.; Wiss, T. J. Nucl. Mater. Conrad, R.; Konings, R. J. M. J. Nucl. Mater. 2003, 320,
1999, 274, 47–53. 106–116.
39. Ronchi, C.; Ottaviani, J. P.; Degueldre, C.; Calabrese, R. 69. Nitani, N.; Kuramoto, K.; Nakano, Y.; Yamashita, T.;
J. Nucl. Mater. 2003, 320, 54–65. Kimura, Y. J. Nucl. Mater. 2008, 376, 88–97.
40. Williams, J. Report TID-7546; US Atomic Energy 70. Yamashita, T.; Nitani, N.; Kanazawa, H.; et al. J. Nucl.
Commission: 1957, 554–562. Mater. 1998, 274, 98–104.
41. Verrall, R. A.; Vlajic, M. D.; Krstic, V. D. J. Nucl. Mater. 71. Yamashita, T.; Kuramoto, K.; Nitani, N.; et al. J. Nucl.
1999, 274, 54–60. Mater. 2003, 320, 126–132.
42. Beecher, S. C.; Dinwiddie, R. B. Modeling the thermal 72. Nitani, N.; Kuramoto, K.; Yamashita, T.; Nihei, Y.;
conductivity of fiber-reinforced ceramic composites. Kimura, Y. J. Nucl. Mater. 2003, 319, 102–107.
In Proceedings of the 17th Annual Conference on 73. Kuramoto, K.; Yamashita, T.; Shiratori, T. Prog. Nucl.
Composites and Advanced Ceramics, Cocoa Beach, FL, Energy 2001, 38, 423–426.
1993. 74. Konings, R. J. M.; Conrad, R.; Dassel, G.; Pijlgroms, B.;
43. Hashin, Z.; Shtrikman, S. J. Appl. Phys. 1962, 33, Somers, J.; Toscano, E. J. Nucl. Mater. 2000, 282,
3125–3131. 159–170.
44. Marelle, V.; Huet, F.; Lemoine, P. Thermo-mechanical 75. Konings, R. J. M.; Conrad, R.; Dassel, G.; Pijlgroms, B.;
modeling of U–Mo fuels with MAIA. In Proceedings of the Somers, J.; Toscano, E. Report EUR 19138 EN;
8th International Topical Meeting on Research Reactor European Commission, 2000.
Fuel Management, Munchen, Germany, 2004. 76. Wiss, T.; Konings, R. J. M.; Walker, C. T.; Thiele, H.
45. Hayes, S. L.; Hofman, G. L.; Meyer, M. K.; Rest, J.; J. Nucl. Mater. 2003, 320, 85–95.
Snelgrove, J. L. Modeling of high-density U–Mo 77. Hellwig, C. h.; Streit, M.; Blair, P.; et al. J. Nucl. Mater.
dispersion fuel performance. In Proceedings of the 2002 2006, 352, 291–299.
International RERTR Meeting, San Carlos de Bariloche, 78. Schram, R. P. C.; Bakker, K.; Hein, H.; et al. Prog. Nucl.
Argentina, 2002. Energy 2001, 38, 259–262.
46. Miller, J. V. Report NASA-TN-D-3898; National 79. Schram, R. P. C.; van der Laan, R. R.; Klassen, F. C.;
Aeronautics and Space Administration, 1967. Bakker, K.; Yamashita, T.; Ingold, F. J. Nucl. Mater. 2003,
47. Marino, G. P. J. Nucl. Mater. 1971, 38, 178–190. 319, 118–125.
48. Nitani, N.; Yamashita, T.; Matsuda, T.; Kobayashi, S. I.; 80. Beauvy, M.; Dalmasso, C.; Thiriet-Dodane, C.;
Ohmichi, T. J. Nucl. Mater. 1999, 274, 15–22. Simeone, D.; Gosset, D. Nucl. Inst. Method Phys. Res. B
49. Latta, R.; Revankar, S.; Solomon, A. Heat Transfer Eng. 2006, 242, 557–561.
2008, 29, 357–365. 81. Gosset, D.; Simeone, D.; Dutheil, M.; Bouffard, S.;
50. Porta, J.; Aillaud, C.; Baldi, S. J. Nucl. Mater. 1999, 274, Beauvy, M. J. Eur. Ceram. Soc. 2005, 25, 2677–2681.
174–180. 82. Wiss, T.; Matzke, H. j. Rad. Meas. 1999, 31, 507–514.
51. Bonnet, M.; Baldi, S.; Porta, J. Prog. Nucl. Energy 2001, 83. Fromknecht, R.; Hiernaut, J. P.; Matzke, H. j.; Wiss, T.
38, 387–390. Nucl. Inst. Method Phys. Res. B 2000, 166–167,
52. Baldi, S.; Porta, J.; Peneleau, Y.; Pelloni, S.; Parette, J. M.; 263–269.
Chawla, R. Prog. Nucl. Energy 2001, 38, 351–354. 84. Wiss, T. A. G.; Damen, P. M. G.; Hiernaut, J. P.;
53. Weber, C. E.; Hirsch, H. H. Dispersion-type fuel elements. Ronchi, C. J. Nucl. Mater. 2004, 334, 47–57.
In Proceedings of the Conference on Powder Metallurgy in 85. Hojo, H.; Yamamoto, H.; Aihara, J.; et al. Nucl. Instrum.
Nuclear Engineering; Philadelphia, PA, 1955. Methods Phys. Res. B 2006, 250, 123–127.
54. Schaffer, L. D. Army Reactors Program Annual Progress 86. Damen, P. M. G.; van Veen, A.; Matzke, H. j.; et al.
Report; Report ORNL-3386; Oak Ridge National J. Nucl. Mater. 2002, 306, 180–189.
Laboratory, 1962. 87. Walter, M.; Somers, J.; Fernandez, A.; Haas, D.;
55. Goslee, D. E. Nucleonics 1963, 21, 48–52. Dardenne, K.; Denecke, M. A. J. Nucl. Mater. 2007, 362,
56. Thurber, W. C.; McQuilkin, F. R.; Long, E. L.; Osborne, M. F. 343–349.
Report ORNL-3709; Oak Ridge National Laboratory, 1964. 88. See, for example, Meyer, M. K.; Hayes, S. L.;
57. Gates, J. E.; Freas, D. G.; Saling, J. H.; Dickerson, R. F. Carmack, W. J.; Tsai, H. J. Nucl. Mater. 2009, 392,
Report BMI-1446; Batelle Memorial Institute, 1960. 176–183.
58. Troyanov, V.; Popov, V.; Barannaev, I. Prog. Nucl. Energy 89. Miwa, S.; Sato, I.; Tanaka, K.; Hirosawa, T.; Osaka, M.
2001, 38, 383–386. J. Nucl. Mater. 2010, 400, 32–36.
59. Savchenko, A. M.; Vatulin, A. V.; Glagovsky, E. M.; et al. 90. Lutique, S.; Konings, R. J. M.; Rondinella, V. V.;
J. Nucl. Mater. 2010, 396, 26–31. Somers, J.; Wiss, T. J. Alloys Comp. 2003, 352, 1–5.
60. Arima, T.; Tateyama, T.; Idemitsu, K.; Inagaki, Y. J. Nucl. 91. Yates, S. J.; Xu, P.; Wang, J.; Tulenko, J. S.; Nino, J. C.
Mater. 2003, 319, 24–30. J. Nucl. Mater. 2007, 362, 336–342.
61. See, for example, Report HW-76300; Hanford Atomic 92. Lutique, S.; Staicu, D.; Konings, R. J. M.; Rondinella, V. V.;
Products Operation, 1962. Somers, J.; Wiss, T. J. Nucl. Mater. 2003, 319, 59–64.
62. Report ANL-7236; Argonne National Laboratory, 1968. 93. Patel, M. K.; Vijayakumar, V.; Avasthi, D. K.; et al. Nucl.
63. Keller, D. L.; Cunningham, G. W.; Murr, W. E.; Fromm, E. O.; Instrum. Methods Phys. Res. B 2008, 266, 2898–2901.
Lozier, D. E. Report BMI-1608; Battelle Memorial 94. Yates, S. J.; McClellan, K. J.; Nino, J. C. J. Nucl. Mater.
Institute, 1963. 2009, 393, 203–211.
64. IAEA-TECDOC-970. Studies on Fuels with Low Fission 95. Shukla, P.; Watanabe, T.; Nino, J. C.; Tulenko, J. S.;
Gas Release; IAEA: Vienna, 1997. Phillpot, S. R. J. Nucl. Mater. 2008, 380, 1–7.
65. See, for example, J. Nucl. Mater., 1999, 274; J. Nucl. 96. Medvedev, P. G.; Jue, J. F.; Frank, S. M.; Meyer, M. K.
Mater. 2003, 319. J. Nucl. Mater. 2006, 352, 318–323.
66. Chauvin, N.; Albiol, T.; Mazoyer, R.; et al. J. Nucl. Mater. 97. Medvedev, P. G.; Frank, S. M.; O’Holleran, T. P.;
1999, 274, 91–97. Meyer, M. K. J. Nucl. Mater. 2005, 342, 48–62.
Composite Fuel (cermet, cercer) 273
98. Medvedev, P. G.; Lambregts, M. J.; Meyer, M. K. J. Nucl. 109. Davies, T. J.; Evans, P. E. J. Nucl. Mater. 1964, 13,
Mater. 2006, 349, 167–177. 152–168.
99. Holliday, K.; Hartmann, T.; Czerwinski, K. J. Nucl. Mater. 110. Streit, M.; Ingold, F.; Pouchon, M.; Gauckler, L. J.;
2009, 392, 487–493. Ottaviani, J. P. J. Nucl. Mater. 2003, 319, 51–58.
100. Holliday, K.; Hartmann, T.; Poineau, F.; Kennedy, J. R.; 111. Meyer, M. K.; Fielding, R.; Gan, J. J. Nucl. Mater. 2007,
Czerwinski, K. J. Nucl. Mater. 2009, 393, 224–229. 371, 281–287.
101. Nakamura, T.; Kusagaya, K.; Sasajima, H.; 112. Yamane, J.; Imai, M.; Yano, T. Prog. Nucl. Energy 2008,
Yamashita, T.; Uetsuka, H. J. Nucl. Sci. Technol. 2003, 50, 621–624.
40, 30. 113. Snead, L.; Katoh, Y.; Kondo, S. J. Nucl. Mater. 2010, 399,
102. Nakamura, T.; Sasajima, H.; Yamashita, T.; Uetsuka, H. 200–207.
J. Nucl. Mater. 2003, 319, 95–101. 114. Patriarca, P.; Rucker, D. J. Fuels and Materials
103. Price, R. J. J. Nucl. Mater. 1973, 46, 268–272. Development Program Quarterly Progress Report for
104. Cabrero, J.; Audubert, F.; Pailler, R.; Kusiak, A.; Period Ending, Sept. 30, 1969; Report ORNL-4480;
Battaglia, J. L.; Weisbecker, P. J. Nucl. Mater. 2010, 396, Oak Ridge National Laboratory, 1970.
202–207. 115. Yano, T.; Iseki, T. J. Nucl. Mater. 1993, 203, 249–254.
105. Kerbiriou, X.; Costantini, J. M.; Sauzay, S. M.; et al. 116. Yano, T.; Iseki, T. J. Nucl. Mater. 1991, 179–181, 387–390.
J. Appl. Phys. 2009, 105, 073513. 117. Taub, J. A Review of Fuel Element Development for
106. Jiang, W.; Wang, H.; Kim, I.; et al. Phys. Rev. B 2009, 80, Nuclear Rocket Engines; Report LA-5931; Los Alamos
161301(R). Laboratory, 1975.
107. Sosuke, K.; Tatsuya, H.; Akira, K. Mater. Trans. 2005, 46, 118. Davidson, K. V.; Martin, W. W.; Schell, D. H.; Taub, J. M.;
1388–1392. Taylor, J. W. Development of Carbide-Carbon
108. Shennan, J. V. Dispersed Ceramic Fuels for the Composite Fuel Elements for Rover Reactors; Report
Advanced Gas-Cooled Reactor. In Preparation of LA5005; Los Alamos National Laboratory, 1972.
Nuclear Fuels, Nuclear Engineering Part XVIII; 119. Lyon, L. L. Performance of (U,ZrC–Graphite (Composite)
Tevebaugh, A. D., Keller, D. L., Eds.; Chemical and of (U,Zr)C(Carbide) Fuel Elements in the Nuclear
Engineering Progress Symposium Series No. 80, 1967; Furnace-1 Test Reactor; Report LA-5398-MS;
Vol. 63, 96–110. Los Alamos Laboratory National laboratory, 1973.
3.11 Sphere-Pac and VIPAC Fuel
M. A. Pouchon
Paul Scherrer Institut, Villigen PSI, Switzerland
G. Ledergerber
Kernkraftwerk Leibstadt AG, Leibstadt, Switzerland
F. Ingold
Swiss Federal Office of Energy, Bern, Switzerland
K. Bakker
Nuclear Research and Consultancy Group, Petten, The Netherlands
275
276 Sphere-Pac and VIPAC Fuel
Figure 2 Production of Vipac fuel coming from an oxide electrowinning (dry reprocessing). The deposit formed at the
cathode during the oxide electrowinning process is scraped off and crushed. The product is then milled and sieved into
several size fractions. In order to reach a high filling density of the fuel pin, these fractions have to be fed parallel and in the
right ratio into the pin. Alternatively, a premixing can be performed. Additionally, often a metallic oxygen getter is added.
Russia, the so-called RIAR process.10 For a more gen- behind the Sphere-pac concept is the direct co-
eral overview of the process, see Bychkov et al.,11 conversion of these solutions to form the nuclear
Kirillovich et al.,12 Ivanov et al.,13 and Bychkov fuel particles for direct use in the pin. This final
et al.14 When coming from dry reprocessing, this particle formation through a fully aqueous method
concept promises to be very simple and cost-effective. promises excellent distribution of the fissile material
in the matrix and the formation of solid solution in
the ceramic. Furthermore, it implies less need for
3.11.1.2 Sphere-pac Fuel
mechanical devices, and a significant powder reduc-
Sphere-pac fuel is also composed of particles, which tion, and hence less potential for contamination, in
are directly filled into a fuel pin. There are, however, the process. Conventionally, the nitrate solutions are
several major differences from the Vipac fuel. The most precipitated as powders, which are then pressed to
important one is the defined spherical particle shape of form pellets. So in both cases, a solidification step
potentially several size classes, which allows a predict- is performed, conventionally before the shape for-
able arrangement in the fuel pin. Another difference is mation step, and in the Sphere-pac, it is part of
the production method and the most preferable source. the process. The nitrate solutions obtained in a wet
Originally, the concept was suggested at the Oak Ridge reprocessing route are converted into spherical fuel
National Laboratory (ORNL) in the United States. particles using a sol–gel technique. In contrast to the
Even some mixed forms of spheres and angular Vipac concept, where the solidification is part of the
shards are being suggested in Flipot and Papeleux.15 separation process (deposition on the anode), this
Today, the Sphere-pac production is the pre- formation of particles is a separate step, and there-
ferred succeeding step to a wet reprocessing route. fore offers several options, which are further
This can be a PUREX (plutonium and uranium described in Section 3.11.5.2.1. In order to reach
recovery by extraction) process, combined with acceptable fuel densities in the pin, several size frac-
one of its follow-up processes (e.g., selective actinide tions are produced and filled into the pin. When
extraction (SANEX), . . .). The dissolved and sepa- coming from an aqueous reprocessing route, this con-
rated uranium, plutonium, and minor actinides cept promises to be simple, safe, powderless, and
(MAs) will be present as nitrate solution. The idea therefore cost effective.
Sphere-Pac and VIPAC Fuel 279
3.11.2 Features of Particle Fuel After some time, it will be sintered to different
zones (for the case of Sphere-pac concept see
The introduction of particles into the fuel pin, as a Figure 10 in Chapter 3.25, Modeling of Sphere-
replacement of the classic fuel pellets, implies Pac Fuel). With increasing distance from the clad-
changes in some important properties. The following ding, the temperature will rise. Therefore, it will be
primary properties imply changes in the fuel highest in the fuel center. Because of pore migration,
comportment: a center-hole might form (Zone I). The ceramic fuel
material at its highest temperature (just beyond the
Thermal conductivity (tendency to lower conduc-
center-hole) will be sintered to a dense matrix, very
tivity, because of the hindering macrostructure, the
similar to the pellet structure (Zone II). Toward
particle or sphere arrangement)
the cladding and with decreasing temperature, one
– Temperature (tendency to higher centerline
will find a sintered ceramic, but with increasing
temperature)
porosity as a remaining feature of the original parti-
– Restructuring (extended restructuring as a
cle arrangement (Zone III). Finally, at its coolest
result of higher temperatures)
region near the cladding, the fuel will remain in its
– Fission gas release (tendency to higher release,
original macrostructure, where the contact points of
because of higher temperature and therefore
the particles undergo a necking, which is an increase
more important restructuring. Depending on
of their contact area by approaching their centers
reactor system and hence the general fuel tem-
and displacing material around the touching axis
perature and matrix, this effect can be negligi-
(Zone IV). The borders between these zones will
ble or important)
evolve during reactor operation, moving outward
Mechanical properties
with time and temperature.
– Fuel-cladding mechanical interaction (FCMI)
Because of this important structural change of the
(softer fuel, because of the possibility to expand
fuel, the physical properties are changing dynamically.
into the vacant space between the spheres and
The thermal conductivity, for example, importantly
because of higher temperatures, but conse-
increases already in the beginning, when the first
quently also higher thermal expansion)
necking is introduced. The particle fuel specific mod-
– Possibility to introduce a heterogeneous distri-
els treated in Chapter 3.25, Modeling of Sphere-Pac
bution of fissile material (with the option of
Fuel take into account these special features.
adjusting the thermal behavior by a cermet
The tendency for a higher fission gas release has
concept or coating the particles)
been observed for matrices with low thermal conduc-
Specific surface
tivity and therefore high centerline temperatures (e.g.,
– Fission gas release (tendency to higher fission
see Section 3.11.6.2.10). In other programs, where the
gas release, because of the larger surface)
matrix was better conducting and the center tempera-
– Reaction surface (in case of a cladding breach,
tures were therefore lower, this effect was not observed
the fuel could offer a higher reaction surface
(e.g., see Section 3.11.6.2.7). The grain structure of
area with the entering coolant)
the particles can in such cases play a larger role than
Structural integrity
the macrostructure.
– Cohesion of fuel parts (at the beginning of
The lower structural integrity, especially at begin-
irradiation, the spheres are only packed
ning of life (loose or weakly bond spheres), in con-
together without strong bonds; this raises safety
nection with a pin failure raises somewhat higher
concerns in case of a cladding failure).
safety concerns than for the initially compact pellet
type fuel. Also, the reaction surface is larger. This and
These properties are also the subject of the Sphere-
other aspects are treated in the next Section 3.11.3.
pac related modeling codes presented in Chapter
3.25, Modeling of Sphere-Pac Fuel (see also Botta
and Hellwig16 and Wallin et al.17). 3.11.3 Safety Aspects of
The macrostructure of the particle arrangement Particle Fuel
therefore importantly changes the fuel comportment.
This is especially true for the initial, fresh fuel. The initially loose fuel particles and the high specific
During reactor operation, the high temperatures fuel surface raise questions about the safety behavior
will cause an important restructuring of the fuel. in case of a very early cladding failure (e.g., during
280 Sphere-Pac and VIPAC Fuel
fuel loading or transportation). The specific charac- In contrast, cladding failure during the life of an
teristics of these two particle arrangements result in irradiation test must be anticipated, basically by the
two major issues: definition of a ‘test.’ These failures occur because
of internal loading by either fuel cladding mechan-
1. washing out of fuel particles into the coolant
ical interaction, which is lower for particle fuel, or
because of missing adhesion to the fuel package and
fission-gas pressure that drives the creep-rupture
2. enhanced chemical reaction with the coolant
mechanism in FBR qualified austenitic and ferritic
because of the larger surface area.
stainless steel claddings. Sodium will enter a breach
The question (1) can be partially answered and quan- after depressurization (fission-gas release), although
tified with the initial necking behavior of the fuel. In creep-rupture failures typically can reclose until
the models developed in Chapter 3.25, Modeling of repressurized. Sodium migrates via fuel surfaces and
Sphere-Pac Fuel, the initial necking, plus the fur- grain-boundaries and as vapor through voids. Particle
ther sintering behavior (see Figure 6 in Chapter fuel columns have 20% interconnected void vol-
3.25, Modeling of Sphere-Pac Fuel), is elaborated. ume in the outer radial unsintered regions where the
From the necking values, a mechanical stability fuel temperature is <1100 C. In this region, sodium
(adhesion force of single particles) can be calculated will migrate to the sodium saturation temperature
as a function of the local temperature and mechanical isotherm (900 C) zone where the reaction kinetics
load history. For a further quantification, models are the highest. In the analysis of over 100 failed
would have to be developed, taking into account MOX fuel pins, the rate of sodium entry had been
the flow characteristics, the force imposed by the sufficiently low such that liquid sodium migration in
coolant flow, the adhesion forces, and the transport the fuel region was never observed. In addition, the
of particles. An experimental approach was conducted reaction is affected by fuel O/M, morphology, and
at Paul Scherrer Institute (PSI) which addressed out of burnup. At beginning of life (BOL), the O/M in
pile loss of spheres, simulating a UO2 Sphere-pac the outer radial regions is <2.00 (typically 1.97–9,
in the Phenix reactor.18 but >1.96 minimum for reaction). The reaction prod-
The question (2) has been studied for the case of uct (Na3(U,Pu)O4) grows radially outward occupying
MOX particles in a sodium cooled fast reactor.19 The the voids until the excess oxygen is depleted. The
unanticipated cladding failure of an MOX particle diagram illustrates how the fuel-sodium-reaction-
fuel (Vipac and Sphere-pac) pin raises the issue of product (FSRP) develops in a few hours and in a
fuel behavior in flowing sodium. The assessment few days. Fuel particles in the peripheral regions
of the behavior must be on the basis of credible become encased in a matrix of FSRP when a failed
failure modes and the subsequent fuel-sodium reac- fuel pin remains in the core.
tion product formation. Relative to MOX pellet-fuel In creep-rupture failure scenarios, the early
pins, there is little particle fuel failure behavioral breach behavior will be basically stable until internal
data. However, the large failed pellet-fuel pin data- loading is reestablished by FCMI due to FSRP.
base20 provides the basis for predicting their behavior Relative to pellet fuels, FCMI is measurably lower
under credible cladding failure scenarios. Cladding because of the ease of its accommodation. As the
ruptures occur because of internal loading, either reaction continues, it tends to form an FSRP seal,
FCMI or fission-gas pressure that drives the creep- and in the event that the breach were to open to
rupture mechanism. Birth defects, due to a fabrica- flowing sodium, it would retard the fuel from being
tion error, must also be considered probable. These swept away. A birth defect at either the top or bottom
defects are of concern because during initial FBR weld zones will not be expected to grow because of
startup conditions breached cold MOX fuel is not the absence of a loading mechanism, even after the
readily detectable. The critical issue is whether the long term FSRP is accommodated by the void space
breach can develop to a size where flowing sodium in the 80% smear density fuel. Fuel column breach
can sweep over the fuel. growth to a size greater than the smallest fuel particle
A pin failure during reactor startup or very early conservatively will take several days; this scenario is
in life would be due to fabrication errors (birth only valid for the reactor condition where their gas
defects) and is a credible issue even though it may failures go undetected, which in itself is very unlikely.
be considered highly unlikely. These defects would In the extreme case, if fuel pin components fail by
normally be associated with the top or bottom end- an unacceptable failure mode, such as brittle failure,
cap weld zone, well away from the fuel column. or a catastrophic birth defect breach (neither is
Sphere-Pac and VIPAC Fuel 281
considered a credible failure mode in nuclear grade the sequential production of the particle fractions
materials or processes), fuel loss from any fuel form naturally delivers the size classes separately, and
will have to be assumed. the particles are spherical, offering comparatively
low friction resistance during the filling procedure.
3.11.4 Concepts of Particle Fuel These features offer the following possibilities:
1. reach tailored filling densities in the range of
The rough classification of particle fuel into Vipac
68–90% of the theoretical density (also designated
and Sphere-pac is already mentioned. Both fuel types
as smear density), as the number of sphere frac-
are preferentially used in closed fuel cycle concepts
tions and sizes can be fully controlled,
and in the area of solutions conversion with transura-
2. beside the economic, simultaneous filling of mul-
nic isotopes coming from reprocessing and being
tiple size fractions of spheres, it is also possible to
used in a fast reactor. The reason is their production
fill different size-fractions successively, starting
process simplicity and the consequently easy appli-
with the largest one, and infiltrating the smaller
cation in a remote process. The kind of actinide
ones into the spaces between the larger spheres
separation, whether dry or aqueous, will determine
(infiltration filling),
the kind of particle fuel. This section tries to elabo-
3. with the infiltration filling, an even better control
rate on extended options for each of these types.
is possible because of the geometrical arrangement
of each size fraction, avoiding any segregation
3.11.4.1 Vipac Fuel effects in the density distribution,
4. with the infiltration filling, it is also possible to
The most important option within Vipac fuel is the
evenly distribute a secondary material within the
addition of an oxygen getter. This will mostly be
primary one in order to act as an oxygen getter or
added as uranium metal and serves to reduce oxida-
improvement for thermal conductivity in a
tion exposure of the cladding material. The Vipac
ceramic metallic fuel concept (cermet),
fuel has mostly been developed for fast reactor utili-
5. the secondary material phase mentioned in the last
zation with uranium as matrices material as uranium,
point would be dominant near the cladding wall,
MOX, and MA-containing MOX.
as the main volume of the larger particles cannot
However, initially the Vipac fuel has also been
approach the even (with slight curvature, but even
investigated for thermal reactors. In particular, the
compared to particle size) wall (no close packing
early US programs3–5 and a program in Denmark
possible near an even border). Therefore, the sec-
(see Section 3.11.6.1.1) considered this option.
ondary phase could fulfill special cladding near
In Lee et al.,21,22 the Vipac concept is also studied as
requirements.
an alternative fuel form for the DUPIC (direct use of
spent pressurized water reactor (PWR) fuel in From these additional possibilities, many concepts
CANDU) concept. Today, the DUPIC fuel fabrication arise. These are specified in Sections 3.11.4.2.1–
process mostly concentrates on preparation of pellets. 3.11.4.2.4.
Cladding
Fuel
insulation
Ceramic
(a) (b)
Figure 3 Scheme demonstrating the filling techniques: (a) Infiltration filling: First the coarse fraction is filled, then it is fixed,
and the small fraction is infiltrated through a sieve-like holder. These actions are performed under slight vibration. (b) Parallel
filling: Both fractions are filled in parallel. Either vibration is applied during the filling process or afterward.
10.0 90
9.5
85
9.0
80
8.5
75
8.0
70
7.5
7.0 65
6.5 60
6.0 55
5.5 50
(a) (b) 5.0
Figure 4 Sphere-pac concept with two or three size (a)
fractions. 10.0 90
9.5
85
9.0
eight or more. Even when choosing a very large 80
8.5
sphere size of the coarse fraction, the smallest frac- 75
8.0
tion will be too fine for handling. However, in 70
7.5
McGeary,23 an example is shown, where three size 65
7.0
fractions of 2.8 mm (mesh 7) with 67%, 0.231 mm 60
6.5
(mesh 60) with 23%, and 0.038 mm (mesh 400) with
6.0 55
10% are filled into a 38 mm cylinder, achieving a Insulation fraction
5.5 50
packing density of 90%.
5.0
-25
0
25
50
75
100
125
150
175
200
225
250
275
300
resulting density is given. At the end of this section, spherical shapes is shown, where the shape factor K
earlier experiments are described. reduces to 0.
Systematic packing density studies have been per- A systematic review on the packing behavior of
formed in Ayer and Soppet,28 where spherical, angular urania has also been performed in Rossendorf,
shapes and rounded, angular shapes are compared. Germany.29 Besides the particle sizes and the ratio
These two are designated as ‘angular grit’ and ‘ball- of the size fractions, the applied frequencies (up to
milled grit.’ For the filling density and for a unitary 5000 Hz) and accelerations (up to 100 g) are
size fraction of particles, one comes to an equation addressed. Additionally, the compaction by external
of the following form: mechanical reduction of the pin diameter (hammer-
K CD
ing) and thereby achieved increased density are
A Be d1 ½1 described. Later, in Steinkopff et al.,7,8 a Vipac fabri-
d1
cation unit and procedure for BOR-60 fuel are
where A is the limiting density, B and C are constants described. Ternary, quaternary, and polydisperse size
defining the boundary conditions, and K is a shape fraction arrangements are experimented. The typical
factor. D is the containing cylinder diameter and d1 is vibration parameters are accelerations of 1.3–2 g and
the particle diameter. The shape factor depends on the frequencies of 38–100 Hz during filling, and 14–56 g
surface to volume ratio and has here its limits for a and 260–2000 Hz for the postcompaction.
tetrahedron on one side (angular shape), and a sphere
on the other side. For finding it experimentally, the
three particle types are experimented for large diame- 3.11.5.2 Sphere-pac Fuel
ter ratios of the container to the particles (D/d1 > 7) in The production of Sphere-pac has to comprise the
order to have a negligible effect from the boundary solidification step, as the product of an aqueous
condition. For spherical shapes K ¼ 0, and for the angu- separation process, such as in the PUREX process, is
lar grit K ¼ 0.032. For smaller ratios of D/d1, a depen- a nitrate solution. This offers a large flexibility in the
dency between B, C, and K is probable, but has not way the solidification is realized and what kind of
been further investigated. For angular shapes, which particles are to be formed. The particles in their
mostly compare with Vipac particles of a unitary size dry form are to be further processed to different
fraction, the following equation describes the density: materials such as MOXs, nitrides, and carbides –
0:032 coprocessing of different fissile materials and low
0:072e d1
0:207D
0:635 ½2 decontamination schemas with fission products (FPs)
d1
and inert, nonuranium matrices are possible. Special
In this study, a secondary fraction was introduced by attention has to be given to the concentration of
infiltration, which somehow is not today’s standard the fissile isotopes, which needs to be tailored in the
procedure for Vipac anymore. The derived packing case of the particles for the reactor operation or the
density for the binary size fraction package is found to final fuel form. This requires a collocation of repro-
be described by cessing unit and fuel fabrication plant, which makes
the fuel cycle more proliferation resistant.
0:017 0:098d1
0:037e d1 0:07e d2
0:207D
0:812
d1 3.11.5.2.1 Particle production
( 0:098d1 )
0:006e d2 0:207D þ
0:098d1 Many methods are known to form particles from
þ 0:014e d1 d2 ½3 solutions and suspensions. The large advantage of
d1
such a process is the absence of powder handling
where d2 is the size of the secondary fraction. Under and the initial chemical homogeneity of the metal
some conditions, when the size ratio between the cyl- solutions, which result in ceramic phases with the
inder and the large particles, and that between the large metals in solid solution.
and the small particles are large enough (D/d1 5 and Generally, sol–gel methods are used (see Sections
d1/d2 10), the last two terms (in curly brackets) are 59.5.2.1.1–59.5.2.1.4), as they prove to be reliable
negligible, reducing eqn [3] to its first line. and resistant to the radioactive environment. Three
For a ternary package, the limiting density for methods are shown in Figure 7. These are the inter-
angular shapes is 0.903 (for the unitary and for the nal gelation (Figure 7(a)), the external gelation
binary package it was 0.635 and 0.812, respectively). (Figure 7(b)), and the water extraction process
In Section 3.11.5.2.2, an equivalent approach for (Figure 7(c)). The internal gelation allows the
Sphere-Pac and VIPAC Fuel 285
NH3 H2O
HTMA
Preneutralized
Me(NO3)X+NH3 NH3 NH3 M-Sol
M-Sol
H2O H2O
Me(OH) Me(OH)NO3
NH3
H2O
(a) (b) (c)
Figure 7 The three sol–gel processes: (a) internal gelation where the gelation process is introduced by applying heat, (b) the
external gelation which is driven by a mass transfer of ammonia, and (c) where water is extracted into organic alcohols.
coprocessing of different metals in an adjusted nitrate external gelation process at an industrial scale with
solution; a wide range of sphere sizes can be pro- the production of about 400 kg UO2 kernels for AVR
duced and the process is capable for non-oxide reloads.32
ceramics such as nitrides and carbides; the external For a schematic of the external gelation process
gelation does not require a heat carrier, and water see Figure 7(b); it is on the basis of the preparation of
extraction does not introduce any organics in the a uranium feed solution of uranyl nitrate in which an
precipitated particles. In the following three chapters, organic polymer such as a polyvinyl alcohol (PVA)
the external and internal gelation processes, and the and a modifier agent such as tetrahydrofurfuryl alco-
water extraction method are described. The first two hol (THFA) have been added. This feed solution is
are on the basis of the precipitation of a metal nitrate then dispersed into droplets through a vibrating noz-
solution (e.g., uranyl or zirconyl nitrate) by ammonia zle. The spherical droplets are formed in air from the
into ammonium metal hydroxide (such as ammonia effect of surface tension, before passing through an
diuranate or zirconate). The fundamental difference ammonia gas layer, hardening the bead surface into
between the internal and the external gelation pro- the final shape. The droplets are then collected in an
cesses lies in the source of ammonia. In the external aqueous concentrated ammonium hydroxide bath,
gelation process, the ammonia is available from an where ammonia induces the precipitation of the ura-
external ammonium hydroxide solution, into which nyl nitrate as ammonium diuranate (ADU). The
the sol (dispersed as droplets) falls. Whereas, in the gelled droplets are aged for a suitable length of time
internal gelation process, an ammonia precursor added to ensure completion of the reaction. The excess
into the solution decomposes thermally to release ammonium nitrate produced during ADU precipita-
ammonia internally within the droplet of the sol. tion is removed from the gelled droplets by washing
in water or in a diluted ammonium hydroxide solu-
3.11.5.2.1.1 External gelation tion. A second washing stage is realized using isopro-
The external gelation processes (named by different panol to dehydrate the wet gel beads. After drying,
authors SNAM-,30 KFA-process,31 or the gel sup- the beads are calcined at 400–600 C in air, then
ported precipitation (GSP)) are the widest used pro- reduced, and sintered into UO2 kernels in Ar/H2
cesses in the sol–gel production of nuclear fuel. The atmosphere at 1600–1700 C.34
Tristructural-isotropic (TRISO) fuel-kernels of the
original AVR (German gas-cooled reactor called 3.11.5.2.1.2 Internal gelation process
‘‘Arbeitsgemeinschaft Versuchsreaktor’’) high temper- The internal gelation process (also called the HMTA
ature gas cooled reactor in Jülich were produced by (hexamethylenetetramine) process) is an aqueous
external gelation, as also the kernels for the Japanese production method for ceramic spheres, where
High Temperature Test Reactor (HTTR), the South because of a temperature-driven acidity shift hydrox-
African High Temperature Reactor (pebble bed modu- ide precipitation is initiated, which results in a thix-
lar reactor, PMBR), and the Chinese High Tempera- otrop solidification of mixed solution, which is also
ture Reactor HTR-10. The German experience32,33 called gelation. It was originally developed at KEMA
acquired during the 1980s–1990s has demonstrated in Arnhem, the Netherlands by van der Brugghen
high quality manufacturing of UO2 kernels using the et al.35 and Kanij et al.36 At the PSI in Switzerland, the
286 Sphere-Pac and VIPAC Fuel
internal gelation was further developed and different fraction is dropped directly into an oil bath. At the
research programs were conducted, including irradi- extremely high production rate of the small droplets,
ation tests. By introducing an advanced technology, the oil jet becomes necessary, in order to efficiently
the microwave heating of the spheres,37 some disad- transport them away and keep a distance between the
vantages of the present internal gelation process can spheres prior to solidification. This way, an agglom-
be avoided. This sophisticated approach was the sub- eration is avoided. The solid particles are then trans-
ject of some past work at PSI38,39 and at other places, ported to a belt-filter, where the silicon oil is removed
and it is the main topic of a new fuels program being by a small vacuum unit and fed back into the oil
initiated at PSI. circuit; additionally the spheres are rinsed with an
The precipitation of the internal gelation is driven organic solvent. The oil being recirculated has to be
by the acid (NO 3 ) catalyzed decomposition of filtered (especially during the production of small
HMTA and hence formation of ammonia. This pro- spheres), and the water picked up during the process
cess is strongly temperature sensitive, and can conse- has to be removed. The spheres then proceed to a
quently be controlled by applying heat. If working second belt filter, where the vacuum cleaning is
with uranium solutions, urea needs to be added in repeated to remove the potentially oil contaminated
order to prevent the formation of weaker complexes organic solvent, and the spheres are washed again
with HMTA, and thereby the preterm precipitation. with fresh solvent. The product is then collected in
For a detailed overview, the process description, or baskets and the remaining solvent is removed. The
the chemistry involved in the internal gelation pro- solvent is recycled (distilled) and used again. In the
cess for uranium, uranium and plutonium, zirconium, solvent washing step, the organic residues are removed.
and americium, see Ledergerber.40 The collected spheres undergo another washing
The process, as applied in PSI for the fabrication procedure using a diluted ammonia solution. In this
of the different fuel materials for the irradiation final washing step, the residual reaction products and
test, made use of a cooled (at roughly 270–272 K) excess of nitrates are removed. As in the case of the
feed solution which was dropped into a hot (roughly solvent, the ammonia is also recycled. Then, the
373–379 K) water-immiscible liquid such as silicon spheres are dried in hot air. For a detailed description
oil. Because of the temperature increase, the chemical and some production flowcharts, including the recy-
processes described before are initiated. By using cling steps, see Ledergerber40 and Pouchon et al.41
different nozzles and by optionally applying vibra- Figure 9 shows a scheme of the production unit,
tion, a large range of particle sizes has been produced. with the mixing of the feed solution, the drop gener-
The sizes varied typically between 40 mm and 1.5 mm. ation, the cleaning using the band filters, and the
Figure 8 shows the production of small and large ammonia washing. Additionally, the thermal treat-
spheres. In case of the small spheres, roughly 15 000 ment is schematically shown. This includes air dry-
drops are generated per second, and in case of the ing, calcination, and sintering. Finally, direct filling of
large ones, about 6 drops per second. As shown in the spheres into a fuel pin is shown, which is one of
Figure 8, the small fraction is vibrated into an oil jet, the possible applications of the product, the so-called
which pours into an oil bath, whereas the large Sphere-pac fuel.
The process was demonstrated for three fractions
of uranium oxide and uranium plutonium oxide for com-
parative irradiation tests in commercial reactors in
Switzerland.
For uranium plutonium carbide, the process flow
sheet was demonstrated with the preparation of
25 kg of mixed carbide fuel for a joint (US-Swiss)
irradiation test in the US Fast Flux Test Facility
(FFTF).42 The fuel met all the requirements of the
specifications concerning stoichiometry, chemical
composition, and structure. All the PSI processes
conformed to the stringent US specifications. Mix-
(a) (b) tures or ‘charges’ of sinter batches of each fraction
Figure 8 Production of a (a) small size fraction (40 mm) were made, enough for the filling of groups of five
and of a (b) large size fraction (1 mm). pins. Carbon, oxygen, and plutonium analyses were
Sphere-Pac and VIPAC Fuel 287
Ammonia washing
HTMA
Nitrates
urea
Feed solution
Pin filling
Sintering
Drying Calcination
Figure 9 Scheme of the internal gelation production unit used at Paul Scherrer Institute.
performed and a ceramographic specimen prepared X-ray diffraction for the chemical composition and
to qualify each batch before mixing the charges con- on the correlation of the crushing strength to structural
taining some 2 kg of large spheres or 750 g of fines. parameters. The following can be said in conclusion:
The charges were fully characterized at PSI with the
Carbon-containing spheres fabricated by the inter-
exception of the trace elements and nitrogen analyses
nal gelation process and carbothermically reduced
which were performed at Institute for Transuranium
with nitrogen addition result in either porous or
(ITU). Good reproducibility of the carbon and oxy-
dense uranium nitride microspheres. The demon-
gen contents, the two most important characteristics
strated process flow sheet reduces the number of
of a carbide fuel, was achieved over the whole
process steps remarkably.
campaign.
The addition of only 7% hydrogen to the purging
The fabrication capabilities for three size fractions
gas of the carbothermic reduction promotes the
of uranium oxide43 and for uranium–plutonium
growth of the uranium nitride grains. With a low
MOXs microspheres44,45 have been demonstrated in
specific gasflow, the carbon losses are only slightly
campaigns for irradiation experiments of segmented
increased. The pore and grain structure can be
fuel rods in lead test assemblies for commercial
adapted to fulfill requirements of a press feed for
LWRs. No problems have been encountered and
pellet fabrication. The low interlinked grain struc-
product specifications have been reached comparable
ture was associated with a crunchy crushing and a
to those of pelletized fuel.
low crushing strength.
Uranium nitride microspheres46 were fabricated by
internal gelation and carbothermic reduction in The nitride fabrication process has been further eval-
order to demonstrate the feasibility of a nitride irra- uated in Streit49 and Daumas,50,51 For the simulation
diation in the Phenix sodium-cooled fast reactor for of an americium–zirconium nitride target for an irra-
‘Nimphe 3.’47,48 The influence of the thermal treat- diation under fast neutron flux, uranium–zirconium
ment, the reaction atmosphere on the chemical com- nitride microspheres have been fabricated by using
position, and the structural parameters of the spheres the internal gelation process. The samples show a
were systematically investigated. High density (>95% good thermal conductivity and an acceptable density
TD) spheres of 800 mm diameter were obtained by of 84–86% TD, which is a reasonable value for a fast
reacting in argon–hydrogen followed by reaction in reactor application, where fission gas release must be
nitrogen–hydrogen. Porous spheres with a distinct attended to.
pore and grain structure and low crushing strength as As described in the introduction, this process
feed for pellets have been fabricated in nitrogen and offers many advantages compared to the classical
nitrogen–hydrogen atmosphere. Special emphasis was pellet route. Besides the already mentioned cleanness
put on a reliable determination of nitrogen and on and lack of important mechanical components, it is to
288 Sphere-Pac and VIPAC Fuel
be mentioned that the unit installed at PSI was run- the X-band, the applicators were upright mono-mode
ning extremely reliably over a very long period. cavities, where several E-field maxima were estab-
When changing the product composition, a readjust- lished. The required microwave power levels were
ment of the parameters was necessary, which some- 80–100 W, resulting in a temperature increase of
times was tedious. However, when the parameters 60–80 K. This was sufficient to achieve adequate gela-
were optimized, the production was reliable and tion. Using the S-Band, Yamagishi and coworkers54–56
especially also reproducible. A production example required 5000 W to achieve the same temperature
for three size fractions of MOX microspheres is given increase of 80 K. The decrease of energy transfer
in frame of the so-called FUJI program (fuel irradia- efficiency when decreasing the frequency is due to
tion test for JNC and PSI), which was performed in the lowering of the quantum energy and some scat-
collaboration between Japan Nuclear Cycle Develop- tering influence.58 Rosin et al.57 evaluated the cavity
ment Institute (JNC) (today JAEA) in Japan, PSI, and design and preheating of the droplets at the nozzle.
Nuclear Research and Consultancy Group (NRG) in When reviewing these studies, the use of the X-band,
the Netherlands.41 Pouchon et al.41 also have discussed with about one to two orders of magnitude higher
the characterization of the microspheres for their energy transfer efficiency,57 seems to be essential for
porosity, microstructure, plutonium distribution, a successful microwave gelation although the genera-
chemical composition, and density. tors and amplifiers are much more expensive for the
From a radioprotection point of view, the omission X-band, and the operation requires a license (because
of powders is a huge advantage, but on the other of interaction with some telecommunication frequen-
hand, the introduction of the silicon oil and the cies). When applied to an industrial scale, these factors
washing solvents require additional processing steps become much less grave, and the (energy) efficient
to process the waste being produced. The equipment operation of a plant becomes more important.
built into glove boxes at PSI included efficient units In this section, the technique being applied at
to recycle the oil and the washing fluids; however, in a PSI (formerly EIR, Eidgenössisches Institut für
continuous large scale fuel production, these recy- Reaktorforschung), initially described by Jungo and
cling units plus the still produced fluid waste would Ledergerber37 and Ledergerber,38,39 shall be further
represent a disadvantage. In the frame of one of the elaborated. The preparation of the feed solution and
collaboration programs of FUJI, one study shows that the chemical reactions leading to the gelification are
the installation costs of an internal gelation unit are the same as described in Section 3.11.5.2.1.2. The
higher, and the unit is more space requiring than the change is in the next step, where the heat is applied
pellet process.52 during the free falling passage of the droplets through
The disadvantages lead to further efforts to opti- a microwave cavity. This cavity is designed in the
mize the process. One of these efforts is described in X-band (8.2–12.4 GHz or a wavelength of 3.6–2.4 cm).
the next subsection. It is the attempt to change the To prevent a contamination of the cavity, the drops
heat induction into the droplets, from the heat carrier fall through a quartz tube, which vertically trans-
concept (silicon oil) to a more advanced approach by verses it. The gelled drops then directly fall into the
using microwaves. ammonia wash column. Therefore, in this process,
the recycling of the oil as well as the solvent is not
3.11.5.2.1.3 Microwave internal gelation needed as in the silicon oil process that has the silicon
The internal gelation process can to a large extent be oil column and the two belt-filters. Figure 10 shows a
controlled by temperature. The gelation does not comparative flowchart for the two techniques. On the
necessitate a chemical interaction with the environ- left, the simpler microwave gelation and on the right
ment anymore. The feed solution already contains all the internal gelation into silicon oil with the addi-
ingredients to conduct the solidification. Therefore, tional washing procedure are shown. For a more
it is irrelevant how, and from which medium the heat detailed flowchart, including the oil and organic sol-
is introduced into the spheres. The idea of using vent regeneration, see Ledergerber.39
microwave heating is not very new and has already An upright mono-mode cavity is used, where a
been practiced at some places.37–39,53–57 Knotik et al.53 pattern of standing waves is generated; see Figure 11
and Ledergerber37–39 were working in the 8–14 GHz for a scheme and a picture of the equipment. The
regime around the X-band, whereas Yamagishi and mode is transverse electric TE10, with its single loop
coworkers54–56 and Rosin et al.57 used the commer- across the largest guide lateral dimension. A detailed
cially widely used 2.45 GHz in the S-band. Around description of the microwave equipment is given in
Sphere-Pac and VIPAC Fuel 289
Feed solution
Drop
generator Cooling
liquid
Microwave
trap
Quartz tube
Cavity
Wave
guide
Microwave
trap
Washing
column
Ammonia
solution
Figure 11 Scheme and picture of the microwave unit. Reproduced from Ledergerber, G. ‘‘Internal gelation using
microwaves’’ IAEA-TECDOC, advanced fuel technology and performance. In Proceedings of an Advisory Group Meeting,
Würenlingen, Switzerland, Dec 4–6, 1984; 1985; Vol. 352, pp 165–174, with permission from IAEA, Vienna.
The given shape of spheres with the given size arrangement is reached, which is not orthorhombic.
fractions allows some prediction in the packing With a ratio of 1, the highest possible density can be
behavior of the arrangement. From solid geometry reached, which is 66.7%; then, it sharply decreases to
consideration, spheres of the same size can be 47% for a ratio of 2 and slowly increases with larger
arranged in four different schemes in space, these ratios to reach 56.4% for the ratio of 4.04 with its
are the simple cubic with a smear density of 53.36%, 12 spheres per layer. The smallest orthorhombic
the orthorhombic with 60.46%, the double nested arrangements are at ratios of 4.41, 6.41, 8.41, and
with 69.81%, and the closed packed with 74.05% of 10.41 with the corresponding number of spheres per
the theoretical density. However, with the boundary layer of 14, 31, 55, and 85 while the corresponding
condition of a cylindrical enclosure (as in the fuel smear densities are 55.4%, 58.1%, 59.9%, and 60.4%,
pin), the packing will be almost completely rhombic respectively. The number of spheres per layer (NL) can
(assuming a much larger cylinder diameter compared easily be calculated with the following equation23:
to the diameters of spheres). For very large container
NL ¼ 1=4ð13r 2 þ 29r Þ þ 4 ½4
to sphere diameter ratios, the measured packing den-
sities are around 62.5%,23 which would indicate about where r is the number of rings around the center four
one-fifth of double nested, and the rest orthorhombic spheres. This results in the three numbers given above
packing. (14, 31, 55, 85) are for 1,p
2, ffiffiffiffiffiffiffi
3, and 4 rings, respectively.
The container size has an important influence on With a layer distance of 3=4 times the sphere diam-
the packing arrangement. With an increasing ratio eter, the density can be calculated. Assuming pin inner
between the cylinder diameter and the sphere size, diameters between 0.5 and 1.0 cm and coarse size
the packing density generally increases to reach the fractions of around 1 mm, the ratios 6–10 seem to be
limit (at an infinite ratio) of the orthorhombic pack- technically relevant. However, this is a theoretical
ing (62.5%). However, below a ratio of about 4.04, consideration with just fitting arrangements, which
where 12 spheres can be placed in one layer, the ideal can practically not be reached.
Sphere-Pac and VIPAC Fuel 291
In the late 1960s, workers at the Argonne National the small size fraction is, the better it is able to fill the
Laboratory further investigated the packing theory interstices of the coarse material. The knee of the
using computer technique for curve-fitting and least curve in Figure 12 occurs at a size ratio of about 7.
squares analysis and came to the following equation Below this value, the packing efficiency decreases
for the packing density67: rapidly. This corresponds closely to the triangular
D pore path formed by any three pffifficlose
ffi packed spheres
0:635 0:216e0:313d1 ½5 1
of equal size, which is 2= 3 1 ¼ 6:46 (see
where D is the diameter of the cylindrical container right vertical dotted line in Figure 12). This ratio is
(cladding) and d1 is the diameter of the inserted larger (smaller pore and hence smaller, acceptable
spheres. The equation was derived for packages with fine-spherepdiameter)
ffiffiffi 1 than that for the square pore,
relative densities between 0.55 and 0.635 and a mini- which is 21 ¼ 2:41 (see left vertical dotted
mal ratio of the container diameter to the sphere line in Figure 12), or that for any intermediate pore.
diameter of 3. Therefore, for the infiltration filling, the triangular
Equation [5] has a reduced form of eqn [1] in pores govern the movement of the fine spheres
Section 3.11.5.1.2 with the shape factor K ¼ 0, throughout the stationary packing of coarse spheres.
where the limiting density A ¼ 0.635 and the bound- However, for an orthorhombic one-size packing,
ary condition factors B ¼ 0.216 and C ¼ 0.313. there is also a square pore passageway running the
Binary packages : With a second, smaller fraction, length of the packing separated by concentric ‘tubes’
the interstices between the large spheres are partially of spheres with triangular pore holes in them (see
filled, and a higher smear density (which is the geo- McGeary23). Full-length packing, although ineffi-
metrical density) can be reached. Figure 12 shows cient, could therefore occur even for a size ratio
experimental filling examples from McGeary,23 below 7. When filling the size fractions in parallel
where the finer fraction is introduced by infiltration (see Section 3.11.5.2.2.1), no pore size limitation
(see Section 3.11.5.2.2.2). With increasing size ratio applies. With decreasing ratio, the packing effi-
between the coarse and the fine sphere fractions, a ciency, however, still decreases.
clear increase in the filling density can be identified. According to Cumberland and Crawford,68 the
Qualitatively, this behavior is expected as the finer maximal smear density with two size fractions (large
ratio of the coarse to the fine fraction diameter) is 86%
where the small size fraction would be present with a
85
weight ratio of 27%.
The above eqn [5] for the unary packing density
can be extended for the additional secondary fraction,
80 where the succeeding smaller spheres occupy the
remaining space by the following fraction67:
% of theoretical density
d1
75 0:635 0:737e0:201d2 ½6
where d2 is the sphere diameter of the second, smaller
fraction.
70
Coarse fraction diameters: This leads to the total packing density of
3.15 mm (7 mesh)
D d1
0:867 0:079e0:313d1 0:269e0:201d2
1.55 mm (14 mesh)
65 1.09 mm (18 mesh)
0.91 mm (20 mesh) d1
½7
D
0.28 mm (60 mesh) 0:159e0:313d1 0:201d2
60
The equation is valid for packages with relative
0 5 10 15 20 densities between 0.72 and 0.867 and the minimal
Coarse to fine diameter ratio (-)
diameter ratio of 3 between the cladding and the
Figure 12 Experimental filling densities of two size coarse spheres, and 5 between the coarse and
fractions as a function of the coarse to fine fraction diameter the fine spheres.
ratio according to McGeary.23 The illustrations show the
triangular pore indicating a size ratio of 6.46 and the square
Ternary packages23: An experimental packing den-
pore indicating a size ratio of 2.41. Both pore types are sity of 90.0% is reported for a ternary package with
present in the orthorhombic packing. 67% of 3150 mm, 23% of 279 mm, and 10% of 43 mm
292 Sphere-Pac and VIPAC Fuel
(size ratios of 72.9:6.5:1) in a cylinder of 38.1 mm The fixation of the first fraction is necessary,
diameter. The particle sizes and the cylinder size because the small fractions introduced in the second
are, however, technically irrelevant for nuclear appli- step would cause the large spheres to float (penetra-
cations. A full range of maximum packing ratios for tion of the fine particles in the gaps below the large
definite size propositions is given in Lee,69 also indi- spheres and levitation of these).
cating the maximal smear density for infinite size A concrete example of a filling procedure can be
ratios between the three fractions, which is 92.6% found in the Section 3.11.6.2.11.
for the volume fractions 0.671:0.242:0.087. An exam-
ple of a ternary package of nuclear fuel can be
found in the program described in Section 3.11.6 Conducted Irradiation
3.11.6.2.8 reporting smear densities of 86.5% for Programs
sphere sizes of 39, 275, and 1155 mm.
3.11.6.1 Vipac Fuel
As in the case of the unary and binary packages, in
Ayer and Soppet,67 a filling density equation was 3.11.6.1.1 Studies at Risø in Denmark
developed for ternary ones: In the early 1970s, comparative studies between pellet
D
d1 d and Vipac fuel for thermal reactors were performed by
0:201d2
0:951 0:029e0:313d1 0:098 e0:201d2 þ e 3
Knudsen et al.70 in Denmark. Two UO2 test fuel pins in
stress-relieved Zircaloy-2 cladding (outer diameter
d1 d2
0:201 d2 þd3 14 mm and thickness 0.6 mm) containing pellets and
0:198e ½8
Vipac fuel were irradiated in the Halden boiling water
where d3 is the sphere diameter of the third fraction. reactor (HBWR) to a burnup of 20 600 MWd MT1
The equation is valid for packages with relative UO2 (assembly average) at assembly powers decreasing
densities between 0.84 and 0.951 and the minimal from 440 to 280 kW (82 cm length). The pellet had
diameter ratio of 3 between the cladding and the 95% theoretical density and a pellet-cladding gap of
coarse spheres, and 5 between the coarse and 0.23 mm. The Vipac contained granules that were
the middle as well as between the middle and the graded in the range of 45–4000 mm and had a com-
fine spheres. pacted density of 84%. After the base irradiation, vir-
tually no change in cladding diameter could be
detected. Prior to this base irradiation, overpower
3.11.5.2.2.1 Parallel filling
irradiation tests were performed in the DR3 reactor
Figure 3(b) shows a scheme of the parallel filling
at Risø (DIDO – heavy water reactor type called
process. The parallel filling is very similar to the
after ‘‘D2O’’ – class reactor). The pellet pin was first
technique described for the Vipac fuel (see Section
tested at the HBWR power level at 330 W cm1,
3.11.5.1.2). Both fractions are filled in parallel into
then at 470 W cm1 with no failure, and finally at
the fuel pin with two feeders. The ratio of the size
720 W cm1, where the pellet pins immediately
fractions, the feeding speed, a possible feeding delay,
showed failure indications. A similar test was per-
and/or extension of a single fraction, plus potential
formed with the Vipac fuel, where the power was first
vibration during feeding, have to be evaluated pre-
increased to 390 and then to 640 W cm1. This caused
cisely beforehand. A concrete example of a filling
failure indication after 7 min. The ramp rates were
procedure can be found in the Section 3.11.6.2.11.
50–60 W cm1 min1. postirradiation examination
(PIE) studies showed cladding hydriding in case of the
3.11.5.2.2.2 Infiltration filling Vipac fuel. As an overall conclusion, the performances
Figure 3(a) shows a scheme of the infiltration filling of the Vipac and the pellet fuel were seen to be similar.
process. First, the coarse fraction is filled into the pin; An objective of the study was the validation of the
in order to reach the highest possible packing density, modeling codes HOTCAKE (1D finite difference
this action is performed under slight vibration. Then, code) and WAFER-1 (3D pellet type code), and it
this arrangement of coarse spheres is fixed with a was found that the codes were in reasonable agree-
holder. This holder must keep the filled spheres in ment with the experiments conducted so far and that
place, but should at the same time allow the small the local cladding stresses and strains calculated
fraction to pass through. Then, the small spheres are could justify the observed cladding failure. It is also
filled into the remaining space (between the coarse stated that the codes could not distinguish between
spheres), under low energy vibration again. the pellet and the Vipac fuel.
Sphere-Pac and VIPAC Fuel 293
(a) (b)
Figure 13 Macroscopic and microscopic pictures of (U,Np)O2 (12.5% burnup) Vipac fuel after irradiation in BOR-60; the
cladding outer/inner diameter is 6.9/6.1 mm: (a) 90 mm (relative to the Midplane) and (b) 145 mm (relative to the Midplane).
Data from PSI.
294 Sphere-Pac and VIPAC Fuel
3.11.6.1.4 HFR Petten: FUJI program – the lower 200 mm is 76.9%. Figure 16 shows Vipac
Vipac fuel fuel after an initial sintering test (IST) and after
3.11.6.1.4.1 MOX matrix with 20% Pu restructuring test. The cladding tube had an inner
In the frame of the FUJI program,73 both types of (outer) diameter of 6.7 mm (7.5 mm) and a height of
particle fuels, the Vipac and the Sphere-pac concept, 410 mm with a fuel column length of 250 mm.
were compared to the Japanese reference pellet-type
fuel (for a description of the entire program see
3.11.6.2 Sphere-pac Fuel
Section 3.11.6.2.11). The production and filling pro-
cedures for Vipac fuel were conducted at PSI74 and as 3.11.6.2.1 US irradiation programs
no dry reprocessing unit was available to produce the conducted at/for the Oak Ridge National
electrode deposit as basis for the Vipac fuel, a simu- Laboratory
lating production process was applied. Randomly To assess the performance of Sphere-pac oxide fuel,
shaped particles were produced by crushing nonsin- 19 such rods, plus one pellet fuel rod, were irradiated
tered powder compacts (green pellets) into small in uninstrumented capsules in the ETR (Engineering
particles, and subsequently calcined at 800 C for
2 h and sintered at 1600 C for 20 h. Figure 14 10.0 90
shows typical Vipac particles after sintering. 9.5 85
The particles were classified into six size frac- 9.0
80
100
125
150
175
200
225
250
275
300
25
50
75
0
(a) (b)
Test Reactor at the Idaho National Laboratory).75 In another irradiation test, two instrumented cap-
For the production of the spheres, see Kitts et al.76 sules with four MOX rods each were experimented in
Two sphere sizes (360–600 mm plus <40 mm in diam- ETR at the Idaho National Laboratory.61 Each rod
eter) were used to achieve up to 84% smear density. contained nine thermocouples and pressure transdu-
Both UO2 and MOX were tested, where the MOX cers for fission gas release measurements. According
had Pu contents of 15% and 20%. Low burnup tests to Bradley et al.,61 the two Sphere-pac and the two
(0.6–1.4 at.%) were conducted with heating rates of pellet rods in capsule 121 have a Pu content of 20%.
345–1140 W cm1 resulting in a fission gas release of The burnup ranges from 6.7% to 9.8%. The density
4–27%. As restructuring of the fuel proceeds expo- of the pellet fuel is 81%, and the densities of the
nentially with temperature, most of the restructuring Sphere-pac fuel are 84% and 85.5%. At linear heat-
occurred early in life for maximal temperatures ing rates between 400 and 570 W m1, the peak inner
above 1650 C.77 Another series of tests addressed cladding temperature reaches 460–625 C. The
burnups from 5.5 to 6 at.%. This series contained corresponding fission gas release ranges from 75%
two Sphere-pac rods (MOX and UO2) and a UO2 to almost 100%. At the highest linear heat rate of
pellet type fuel. The corresponding linear powers 570 W m1 for the pellet rod, fuel melt was observed
were 655 and 720 W cm1 for the Sphere-pac fuel, at the center. Both pellet rods show chemical interac-
and 820 W cm1 for the pellet fuel, resulting in fis- tion with the cladding. The Sphere-pac fuel is
sion gas releases of 44% for both Sphere-pacs, and completely restructured for the higher linear power
47% for the pellet fuel. Finally, one MOX and several of 540 W m1, and radial cracks are formed, which,
UO2 rods were taken to high burnups of 8.7–11.3 at.% however, do not propagate to the cladding, but are
for MOX and 13.8 at.% for UO2 at heating rates of terminated at the outer edge of the restructured
290–330 W cm1 and 365 W cm1, respectively. For region. The fuel adheres to the cladding; however,
this fuel, the PIE still showed an unrestructured fuel no chemical interaction is detected.
annulus of 0.6 mm from the cladding. For the MOX,
the fission gas release ranged from 61% to 89%. For 3.11.6.2.2 German Sphere-pac program
the UO2, the fission gas release was 96%. From 1965 to 1968, the Kernforschungsanlage (KFA)
Another irradiation program addressed the in Jülich (today Forschungszentrum Jülich) con-
restructuring kinetics and the effective gap conduc- ducted a Sphere-pac program for the use in heavy
tance for Sphere-pac fuel.75 Here, six MOX fuel rods water moderated and cooled breeder reactors
were irradiated in three instrumented capsules (SG-1, (HWBRs).75,78 The fuel was (Th,U)O2 containing
SG-2, SG-3) in the ORR (Oak Ridge Research 2.15%78 and 4.45%75,78 uranium. The uranium was
Reactor).76 Five of them were Sphere-pacs, and one enriched to 93%78 and 90%75 235U, respectively. Two
was a pellet fuel rod. The Pu content was either 15% size fractions were used; the large spheres (75 wt%)
or 20%. In case of the pellet fuel, it was 20%. In were in the range of 630–1000 mm (630–800 mm in
all cases, the fuel density was 81–82%. The linear some cases) in diameter, and the small ones (25 wt%)
heat rate ranged from 395 to 525 W cm1. Eight between 33 and 1000 mm (33–80 mm in some cases);
thermocouples were located at the longitudinal reaching smear densities between 81% and 82%. The
center of each fuel rod to measure fuel and cladding cladding was Zircaloy-4. According to Fritz,78 the
temperature at three locations. Four additional ther- Sphere-pac for the prototype fuel rods (T1–T12 in
mocouples were placed in the capsule wall to mea- Table 1) was achieved by the infiltration technique.
sure the heat flow from the rod to the capsule wall. A vibration of 50 Hz with an acceleration of roughly
The Sphere-pac rods in SG-2 operated at central 10 g was applied. The irradiations were conducted at
temperatures of 1500 C; they showed only slight the FRJ-2 (Forschungreaktor Jülich 2, a research
restructuring in the central half of the fuel. SG-3 was reactor of the DIDO class) and at the MZFR (Mehrz-
operated at higher central temperatures of 2000 C, weckforschungsreaktor in Karlsruhe, also a heavy
which resulted in a complete restructuring of the water cooled and moderated research reactor). The
central half of the Sphere-pac fuel. The pellet fuel key properties of the irradiation programs are shown
which was placed in SG-3 also faced considerable in Table 1. After irradiation, none of the fuel rods
restructuring. The gap conductance between the described in Beatty et al.75 showed any indication for
outer fuel surface and the inner cladding surface was interaction of the fuel with the cladding. The maxi-
measured to be 7300 W m2 K1 for the pellet fuel, mal cladding diameter increases were 0.02–0.05 mm.
and 19 300 W m2 K1 for the Sphere-pac fuel. Neither axial gaps nor axial shrinkage of the fuel
296 Sphere-Pac and VIPAC Fuel
Designation reference RBE-2578 RBE-2678 T-1 to T-5 to RBE-3575 RBE-3675 RBE-3775 RBE-3875
T-478 T-1278
Uranium
Content 4% 4% 2.1% 2.1% 4% 4% 4% 4%
Enrichment 93% 93% 93% 93% 90% 90% 90% 90%
Smear density 81.68% 81.15% 82% 82% 81–82% 81–82% 81–82% 81–82%
Reactor FRJ-2 FRJ-2 MZFR MZFR FRJ-2 FRJ-2 FRJ-2 FRJ-2
Maximum linear power 360–595 570–635 492 525 560 570 580 600
(W cm1)
Burnup [MWd kg1 2.07 2.49 56 60 56 57 58 60
(Th þ U)]
column was detected. The fission gas release for the acquisition of the necessary know-how for designing
RBE-35, -36, -37, and -38 ranged from 41% to 51%. and manufacturing fuel elements to be used in a
Alpha-autoradiography revealed 233U enrichment in commercial fast power station.
the outer regions of the fuel.
In addition, eight MOX (Pu content 9.3–21.5%) 3.11.6.2.3.1 IFA-124
fuel rods with a steel cladding (alloy 1.4988, X 8 Cr The purpose of this study is a compatibility test of
Ni Mo V Nb 16 13 – a similar steel to ASTM SS 316) sol–gel Sphere-pac fuel with Zircaloy-2 cladding
were irradiated at the FR-2 (Forschungsreaktor 2, (0.7 mm thickness with a fuel diameter of 12.76 mm).
a heavy water cooled and moderated tank type The two fuel rods contained (U,Pu)O2 sol–gel
research reactor) in Karlsruhe.79 Two size fractions particles (1.8% Pu in metal) filled to 79.5% of theo-
of 800–1100 mm MOX spheres and 125–200 mm UO2 retical density. The irradiation was conducted for
were utilized to obtain smear densities from 52% 2 months until a fuel failure was suspected. The
up to 81%. The test pins were designated with burnup was at 500 MWd MT1, the cladding tem-
1G1, 1G2, 1G3, 1G4, 1G5, 1G9, 1G10, and 1G11. perature was 250 C, and the maximum linear heat
During a period of 2 min up to 10 h, a linear power rate was 520 W cm1. The PIE was conducted at the
of around 650 W cm1 was applied, resulting in Risø Center in Denmark. The cladding defect was
cladding temperatures of 280–392 C. The conclu- not confirmed; however, in the maximum power zone
sion from these tests was a good operating behavior a diametral strain of about 1.8% with a maximal
of the particle fuel. Low mechanical interaction and ovalization of 0.7% was measured. No significant
good thermal conductance are attributed to this increase in length was found. Gamma-scanning and
type of fuel. However, the test does not include a densitometry showed several discontinuities in the
direct comparison with pellet type fuel. According fuel bed, especially in the maximum power zone.
to Beatty et al.,75 the tests do address the needs of The axial power distribution is typical for the Halden
neither thermal reactors (steel cladding and too high reactor with its rods being inserted from top. The
cladding temperatures) nor fast reactors (cladding rods were also punctured for a gas analysis. High
temperatures too low). concentrations of CO and CO2 were found, whereas
the hydrogen concentration was very low. A possible
3.11.6.2.3 Italian Sphere-pac program explanation of the low hydrogen concentration is the
In the late 1960s, Italy conducted programs to hydrating of the Zircaloy cladding, which in this case
produce and test MOX Sphere-pac fuels.80 The acts as a hydrogen sink. This is confirmed by cross-
corresponding fuel assemblies were designated as sections taken from different axial pin positions, where
IFA-124/170, GIANO 1,2,3,4, VENCA G2, and the cladding in the high power region was found to be
SDK 1,2. The irradiation programs took place in the severely attacked by corrosion and hydride formation.
SORIN swimming-pool reactor Avogadro RS-1 at Also the formation of a crack (0.14 mm from the inner
Saluggia in Italy and in the HBWR Halden reactor surface) is documented for this region, and the clad-
in Norway. These studies were part of the Italian ding thickness decreased from 0.7 to 0.43 mm. In the
fast reactor program by the Comitato Nazionale per low power region, the hydride quantity was found to
l’Energia Nucleare (CNEN, today ENEA) for the have normal preirradiation values.
Sphere-Pac and VIPAC Fuel 297
From the fuel structure observed in postirradia- the better mechanical interaction of Sphere-pac fuel
tion examination, the maximum central temperature compared to pellet type of fuel, which was claimed by
has been estimated to be around 2100 C. The central Fitts and Miller,86 could indeed be observed and
void was 22% of the fuel radius, the columnar grains exploited. The need for better mechanical behavior
were 45%, and the sintering zone was 84%. was expected to arise from, among others, the need
to carry out a load-follow operation of NPPs. These
3.11.6.2.3.2 Gaino 1 studies included extensive destructive PIE in order to
Two Sphere-pac UO2 and (U,Pu)O2 (16% Pu in study amongst others the sintering behavior in
metal) fuel rods were irradiated in a central position the spheres and the behavior of fission products
of the RS-1 swimming pool reactor. The operation (such as I and Cs) that might influence the cladding
conditions were the same, both achieving a burnup of integrity. The irradiations shown in Table 2 are
15 000 MWd MT1. The cladding was an AISI 316 described in the sections hereafter.
SS with a thickness of 0.3 mm and a fuel diameter
of 7 mm. The linear heat rate was 410 W cm1. After 3.11.6.2.4.1 The R109 program in the PWR loop
irradiation, the rods were examined at the Dounreay in the HFR
alpha-gamma Hot Cell DERE. X-ray examination The R109 program has been performed in a PWR
showed intact rods, being centered along the capsule loop in the high flux reactor (HFR), in three bundles,
axis. Some common transversal fractures could be designated as R109-1, R109-2, and R109-3.87 The
seen in the fuel. Diameter, deflection, and length program compared the irradiation behavior of fuel
measurements showed no notable defects. Gas punc- rods consisting of (1) UO2 Sphere-pac fuel, (2) a mix
turing showed a fission gas release of 60.9% for the of small UO2 and larger (1 mm diameter) (U,Pu)O2
MOX and 52.3% for the urania fuel. spheres, and (3) UO2 pellets. The Sphere-pac fuel
was three-fraction type, with typically 60 wt%
3.11.6.2.4 Netherlands joint Sphere-pac of 1050 mm, 20 wt% of 100 mm, and 20 wt% of
research program <38 mm. In the mixed small UO2 and larger (U,Pu)O2
In the Netherlands, a joint Sphere-pac research pro- spheres fuel rods, the fissile atoms and thus the heat
gram has been performed in which KEMA, Energy production are mainly concentrated in the large
Research Centre of the Netherlands (ECN), and 1000 mm spheres, as these contained the plutonium.
Gemeenschappelijke Kernenergiecentrale Nederland The behavior of the inhomogeneous Sphere-pac fuel
(GKN) (the operator of the Dodewaard nuclear appeared to be almost identical to that of the homo-
power plant (NPP)) participated during the approxi- geneous UO2 based Sphere-pac fuel. Adding the plu-
mate period of 1969–1990 (Table 2). The nuclear tonium only in the large diameter spheres has an
sections of ECN and KEMA merged in 1998 into advantage from a fabrication perspective, compared
the NRG. This research was based on Sphere-pac to the scenario where for all three sphere diameters
fuel production experience gained at KEMA. This plutonium has to be handled. For comparison, the
Netherlands research was started in order to learn if pure UO2 Sphere-pac and UO2 pellet fuel rods
Table 2 The KEMA, ECN, GKN irradiation program of Sphere-pac fuel rods
Source: Data from ‘‘Van der Linde, A., Jun. 1990. Tabletten en bolleltjes: een vergelijkend splijtstofonderzoek. Energiespectrum 14(6),
165–172.’’
298 Sphere-Pac and VIPAC Fuel
were irradiated in parallel. A color etched ceramo- 27 MWd kg1 UO2). Maximum powers and maxi-
graph of the mixed small UO2 and larger (U,Pu)O2 mum burnup values are shown in Table 1. On the
spheres fuel rods after irradiation is shown in basis of the nondestructive PIE of all rods in the
Figure 17. Because of the difference in chemical Dodewaard pool, three rods of each assembly were
composition, the (U,Pu)O2 spheres etch to a brownish selected for transport to the hot laboratories of ECN
color, while the UO2 spheres have a yellowish color. in Petten. None of the investigated 70 Sphere-pac and
70 pellet rods failed during its duty in the Dodewaard
3.11.6.2.4.2 B202/B203 in the Dodewaard BWR core and none of the 12 rods showed indications of
Dodewaard was a 54 MWe, direct cycle, 6 6 square incipient cladding failure during PIE. PIE has been
lattice fuel grid, natural convection cooling boiling reported in large detail,85,88 including optical micros-
water reactor (BWR) which operated between 1968 copy, scanning electron microscopy (SEM), electron
and 1997. The low power and the natural convection probe microanalysis (EPMA) g-autoradiography, and
made the Dodewaard reactor an extraordinary reac- b/g-autoradiography. An example of the results is
tor which was used for various research purposes. It shown in Figures 18 and 19. Significant efforts have
operated on the basis of passive cooling concepts been made in studying the behavior of volatile fission
currently adopted in advanced BWR designs. products (cesium among others). It was concluded
A comparison test program was performed,85 com- for Sphere-pac fuel (compared with pellet fuel) that
paring four fuel assemblies: A010 (flat-ended pellets), the absence of radial cracks in the outer regions
B208 (double dished pellets), and B202 and B203 of the Sphere-pac column suppresses largely the
with Sphere-pac fuel. The Sphere-pac fuel consisted outward transport of volatile cesium compounds onto
of two fractions of UO2 spheres (a fraction of about the inner clad surface. This renders Sphere-pac rods
1000 mm diameter and another fraction <150 mm). (under normal power conditions) much less prone
The rod filling gas was pure helium of 0.1 MPa pres- to the occurrence of stress corrosion phenomena in
sure. The inner diameter of the Zircaloy-2 cladding the cladding. It was concluded for this irradiation that
was 11.71 mm. Rods in the B202 and B203 assemblies ‘‘The measured differences between Sphere-pac and
were identical with a fuel smear density of about
85.9% TD. Each assembly consisted of 35 UO2 fuel
rods. The elements were placed symmetrically about
the center of the core. Discharge of B202 and B208
occurred after four irradiation cycles in October 1975
with 1066 days of irradiation and an assembly burnup
of about 19 MWd kg1 UO2. Irradiation of B203
and A010 was continued until February 1977 (1530
days of irradiation, assembly average burnup about
Zircaloy
Bonding
layer
40 mm
Figure 19 Sphere to sphere and sphere to clad bonding in rod B203-A1 (burnup 36 MWd kg1 UO2, maximum power
410 W cm1). Courtesy of Klaas Bakker, NRG.
pellet UO2 rods were in effect unimportant, although the Sphere-pac rods were in general about 20 W cm1
permanent rod lengthening of Sphere-pac rods was larger than for the pellet rodlets. From Figure 20 and
about half that of pellet rods and fission gas release other related research, it was concluded in a later
was substantially larger in Sphere-pac rods, in partic- publication by van der Linde89 that for rapidly increas-
ular in low power rated rods where pellet stacks ing powers in the range 460–600 W cm1 Sphere-pac
release was <1%.’’ fuel is more likely to fail.
3.11.6.2.4.3 IFA-204 and IFA-416 in the Halden 3.11.6.2.4.4 Single sphere irradiated at 1900 K
reactor in HFR
The IFA-204 and IFA-416 were UO2 Sphere-pac pins In the framework of source term studies at KEMA-
with three size fractions: large 840–1000 mm, medium Arnhem in the Netherlands,90,91 a single UO2 sphere
88–149 mm, and small (<44 mm) leading to a smear was irradiated in the HFR.92 This sphere had an
density of 87–88% TD. Some of the rods in these as-fabricated diameter of 1.15 mm, an enrichment of
assemblies were equipped with fuel column extens- 20% 235U in U, and a density of 96–100% TD. The
ometers and pin length extensometers. The main aim sphere was irradiated during 544 h in a compacted bed
of these irradiations was to study the mechanical inter- of Al2O3 granules in an aluminum capsule in the D8
action between the Sphere-pac fuel and the cladding. core position of the HFR in Petten, The Netherlands.
The IFA-416 irradiation studied the length increase of The irradiated UO2 spheres were completely evapo-
Sphere-pac and pellet UO2 fuel pins, measured in-pile rated in a tungsten crucible shortly after irradiation.
during the irradiation. The IFA-416 bundle consisted The crucible content was monitored with a germanium
of three UO2 pellet rods and four UO2 Sphere-pac g-spectrometer and the particle bundle emerging from
rods. The maximum pin average power during irradia- the crucible was analyzed with a quadrupole mass spec-
tion in the Halden BWR was 380 W cm1 and the pin trometer. The behavior of various fission products dur-
average burnup was 33 MWd kg1 UO2. Four pins have ing accident conditions has been studied in this manner.
been instrumented with pin elongation detectors, the
results of which are shown in Figure 17. The fuel 3.11.6.2.4.5 Conclusion of the combined
cladding mechanical interaction threshold power (Pt) Netherlands research program
was substantially larger for the Sphere-pac rods than The conclusion of the combined Netherlands
for the pellet rods up to rod-average burnup of about research in the Sphere-pac field focusing on LWR-
12 MWd kg1 UO2. During further irradiation until a application was89 as follows:
burnup of 33 MWd kg1 UO2, the values of Pt varied The joint, 20-year KEMA, ECN, and GKN-
between 180 and 250 W cm1, where the Pt values for research on the behavior of UO2-Sphere-pac fuel,
300 Sphere-Pac and VIPAC Fuel
·
800 Measured Calculated
points curves Fuel-cladding
Thermal + mechanical
Sphere-pac : · · interaction
700 ·
Pellet : ·
Pt: Onset power for FCMI ·
600
·
Pin elongation (mm) ·
500 · · Fuel-cladding
Thermal interaction
400 · ·
Pt
· ·
· ·
300 · ·
· · ·
· ·
·· ·
· Pt
200 ·
·
·
·
4 8 12 16 20 24 28 32 36 40
Average pin power (kw m-1)
Figure 20 Elongation of fuel pins of the IFA-416 assembly during a continuous, normal power increase at a pin average
burnup of 23.8 MWd kg1 UO2. Reproduced from Fig. 2, p. 31 in van der Linde, A. Pin length response of Sphere-pac and
pellet UO2 fuel pins to power increase during irradiation up to 33 MWd/kg UO2 in the Halden Boiling Water Reactor; OECD
Halden Project Report HPR-305/10; 1983.
in comparison with the standard UO2-pellet fuel, various isotopic compositions has become available
showed that the Sphere-pac fuel under normal through ultra-centrifuge.93 Applying molybdenum
LWR-operation conditions has a significantly smaller with a reduced concentration of the molybdenum
risk of failure of the Zircaloy cladding compared to isotope 95Mo yields a strong reduction in the para-
pellet fuel. However, during rapid increases of the sitic neutron absorption in the molybdenum phase.
fuel rod powers to values above 600 W cm1 during This improves the neutronics properties of molybde-
accident conditions, Sphere-pac fails more rapidly num based cermet Sphere-pac fuel.
than pellet fuel rodlets. Future research is focused Within the F-Bridge program, research is ongoing
on the combination of the stability of UO2 spheres into the effects of molybdenum on both the thermal
during irradiation with the high thermal conductivity conductivity and neutron economy of the fuel.
of metals, making a cermet Sphere-pac fuel. Sphere-pac fuel is currently of renewed interest in
According to van der Linde,89 the cause of the the context of MA burning in sodium-cooled fast
easier failure of Sphere-pac fuel at high power is the reactor because of its improved gas release character-
lower thermal conductivity of the Sphere-pac fuel istics, and the potentially large advantage of dust-free
compared to the pellet fuel. Because of this, rapid fuel manufacturing. A first irradiation test comparing
power increases above 450 W cm1 cause so much FR-MOX Sphere-pac and pellet pins containing
more thermal expansion of the Sphere-pac fuel col- 2–5% 241Am is planned within the running European
umn than that of standard UO2 pellet column that the framework program FAIRFUELS; the irradiation is
Sphere-pac fuel fails more easily. Currently, the pro- scheduled to start in the summer of 2011 in the HFR.
blems of FCMI have been mitigated to a significant
extent by optimizing the fuel design (smaller rod 3.11.6.2.5 BR2 reactor: experiment
diameter, improved cladding design, improved creep MFBS 7 – uranium plutonium carbide
characteristics of the UO2 pellet), which has reduced The MFBS 7 experiment consisted of three Sphere-
justification for further studies in the Sphere-pac field. pac fuel pins with 9 mm diameter containing mixed
The use of a molybdenum sphere fraction in order carbide fuel. The purpose of the experiment was
to enhance the thermal conductivity in a cermet was to compare the irradiation behavior of pelleted and
discussed in Van Der Linde.24 The main advantage of Sphere-pac carbide fuels under simulated fast reactor
molybdenum is considered to be the high thermal conditions, that is, under epithermal conditions.
conductivity, while the neutron absorption in the MFBS 7 was a flowing sodium cooled loop located
molybdenum is considered a disadvantage under in a position near the core edge of the BR2 reactor at
LWR conditions. In recent years, molybdenum with Mol in Belgium. Three fuel pins have been clustered
Sphere-Pac and VIPAC Fuel 301
to a fuel bundle. Two pins were filled with Belgium- and 1977. The fuel of the one Sphere-pac fuel pin
made pelletized fuel and the third pin was filled with - consisted of a two-size fraction mixture of 700 and
vibro-compacted microspheres of (U0.855Pu0.145)C1þx , 60 mm nominal diameter microspheres. The heavy
of two size fractions of about 700 mm and about 60 mm metal composition was 16 at.% plutonium in a ura-
diameter. The uranium was enriched to 93 at.% 235U. nium matrix. The Uranium was enriched to 93 at.%
235
In the pin, a smear density of 76% for a 200 mm fuel U in order to achieve the needed high liner heat
stack was achieved and the fill gas was helium at 1 atm generation rate. The fuel was vibro-filled and vibro-
at 25 C. The cladding material was AISI type 316L in compacted using an infiltration method for the fine
the solution annealed condition, and the thickness was fraction. The total length of the fuel stack was
0.5 mm.94 500 mm with a smear density of 75% TD. The
The irradiation took place from 1974 to 1977 cladding was AISI type 316L stainless steel with
achieving a burnup of 4.7–5.5% fissions per initial 20% cold work. Wall thickness was 0.4 mm with an
metal atom (FIMA) at a linear heat generation of outside diameter of 6.61 mm. The coolant inlet tem-
720–870 W cm1. A brief summary of the profilome- perature in the plenum above the reactor core was
try and related results can be found in Table 3.95 The 230 5 C.96
fuel stack length of the Sphere-pac fuel pins had Maximum increases in diameter were found of the
increased in the end by 0.3%. The fuel stack length order of 1.0% for Sphere-pac fuel pin. The middle
of the pellet pins was not measured. Fuel ceramogra- section of the fuel pin, covering a cladding tempera-
phy was performed on fuel sections with 4.7–5.5% ture range of 400–580 C, showed diameter increases
FIMA burnup. The sections showed slightly sintered of 0.50–0.65%.96 Comparable pellet pins showed a
microspheres in the outer region close to the cladding.
The center part was completely transformed to porous
pellet-like fuel with a sharp transition from slightly
sintered to porous pellet-like fuel. The cross-sections
with 5.0% and more burnup showed a tendency to
form a center hole; compare Figure 21. In no case
could the formation of a gap be observed. Detailed
porosity estimations showed that the density within
the porous pellet-like region varied (more porosity in
the center, and less porosity close to the transition
to the slightly sintered microspheres), see Figure 22.
Nevertheless, the overall density of the porous
pellet-like region stayed at the as-fabricated density.
3.11.6.2.6 Dounreay Fast Reactor: Figure 21 Fuel and cladding cross-section. Burnup 5.5%
experiment DFR 527/1 – uranium plutonium FIMA, linear power 87 kW m1, and outer clad temperature
carbide 510 C. Figure 3(b) (p. 133) in Delbrassine, A., Smith, L., Jun.
1980. Pellet and sphere-pac (U,Pu)C fuel comparative
The Dounreay Fast Reactor (DFR) experiment was a irradiation tests. Nuclear Technology 49, 129–135.
comparative irradiation test of pelletized and Sphere- Copyright by the American Nuclear Society, La Grange
pac mixed carbide fuel pins in the DFR between 1975 Park, Illinois, USA.
Table 3 Brief summary of MFBS 7 irradiation conditions and average PIE results
significant higher diameter increase. The fuel column In the section with the 458 C cladding tempera-
height change was not measured explicitly. More ture and the highest linear power, the microspheres
detailed results are listed in Table 4.95 in the outer region close to the cladding are slightly
Fuel ceramography was performed on three fuel sintered while the inner region is completely trans-
sections (450 W cm1/5.3% burnup/610 C mean formed to porous pellet-like fuel with a sharp transi-
cladding temperature; 620 W cm1/7.3% burnup/ tion from slightly sintered to porous pellet-like fuel.
458 C mean cladding temperature; 490 W cm1/ An enrichment of pores in the center of the fuel can
5.7% burnup/320 C mean cladding temperature).96 be observed; see Figure 23.
The microstructure of the sections shows a clear The last section with the highest cladding temper-
dependence on the mean cladding temperature and ature but the lowest linear power shows also a
on the linear power. This reveals the importance of completely sintered center region. But compared to
temperature distribution for structure evolution. the section with the highest linear power, this region
In the section with the lowest cladding tempera- is smaller and no pore migration can be observed,
ture, the microsphere structure is still visible in the Figure 23.
center although the fine fraction seems to be In no case could the formation of a gap be
completely sintered. observed.
70
87 kW m-1
60 5.5% FIMA
Porosity area (%)
50
Cladding wall
40 Preirradiation
smear density = 76%
30
20
10
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5
Radius (mm)
Figure 22 Porosity profile observed by the burnup of 5.5% FIMA. PSI Report.
Table 4 Brief summary of DFR 527/1 irradiation conditions and average PIE results
Pellet pins 81.7 7.7 665 720 13 1.55 One pin cracked
Sphere- 75 7.7 650 650 14 0.9 Intact pin
pac pin
Sphere-Pac and VIPAC Fuel 303
3.11.6.2.7 FFTF: AC-3 experiment – uranium microstructures of these two fuel types, assuming
plutonium carbide97–99 closed porosity in the large spheres of the Sphere-
Helium bonded mixed carbide fuel pins with pellet pac fuel versus open porosity in the pellet fuel.
and Sphere-pac fuel, clad in D9 stainless steel, were A peculiarity concerning the pin dimensional
irradiated in FFTF to 8.5 at.% peak burnup and changes (axial profilometry, ovality, fuel column
14.71 022 n cm2 peak fast neutron fluence. The growth) and the fission gas release is the generally
time averaged linear power at the peak power plane higher scatter of the pellet pin results. This reflects
was around 660 W cm1, and the corresponding the basically different geometrical structure of the
cladding surface temperature about 440 C. The two pin concepts, that is, the higher mechanical
only planned and well defined difference between and thermodynamic homogeneity of a Sphere-pac
pellet and Sphere-pac fuel (apart from the basically bed as compared to a pellet stack, leading to more
different geometrical structure and the related pro- evenly distributed stresses, strains, and tempera-
duction procedure) concerned their range of the tures. A brief summary of the profilometry and
smear density: approximately 75.4–78.7% TD in pel- related results can be found in Table 5.42,100
let pins, and 79–80% TD in Sphere-pac pins (data for Fuel ceramography photographs show slightly sin-
PIE pins only, but representative for all pins). tered microspheres over the whole cross-section for
During 630 equivalent full power days of reactor linear powers of about 350 W cm1. Cross-sections
operation, none of the 66 pellet pins and 25 Sphere- of fuel irradiated with more than 600 W cm1 show
pac pins failed. PIE was carried out on seven pellet a tendency to form porous pellet-like fuel in the
and eight Sphere-pac pins. The results and conclu- center. Nevertheless, the transition from slightly
sions are summarized below: sintered to porous pellet-like fuel is very broad
and smooth. In no case could the formation of a
Fission gas release in general was relatively low
gap be observed; see Figure 24, respectively.97
when compared to EBR-II data. Significantly
higher values have been measured in pellet pins
(7.8%) than in Sphere-pac pins (4.8%). This dif-
ference is most likely related to basically different
(a) (b)
Table 5 Brief summary of AC-3 irradiation conditions and average NDE results
Smear Calculated Peak Time average Peak cladding Pin length Fuel column
density average cladding peak linear diametric change change height change
(% TD) burnup temperature power (% DD/D) (% DL/L) (% DL/L)
(% FIMA) ( C) (W cm1)
By and large, the irradiation behavior of both Table 6 Fuel and segment characteristics of the KWU-
pellet and Sphere-pac pins was very good besides PSI Gösgen experiment
some marked differences concerning detailed aspects Pellet Sphere-pac
as listed above. Stronger differences between the two
and slight advantages for the Sphere-pac pin concept Fuel type Standard, Vibro-compacted
dished
might be expected and found in comparative tests at Material/enrich. UO2/4.27 UO2/4.24
higher ratings and to a higher burnup with a further (% 235U)
developed Sphere-pac pin design. The improvements Pellet/part. size 9.11 O/D 11 39, 275, 1155
would consist in (1) placing the fissile material only (mm mm1)
in the large high density spheres and (2) replacing Fuel density (% TD) 95.4 99.63, 98.65,
98.75
the insulator pellets with an insulator sphere col- Stoichiometry (O/M) 2.002 2.003
umn of equal height. This would also result in sim- Mean grain size (mm) 11 20, 10, 7
plifications in the fabrication process not only for Mean pore size (mm) – 1.2
carbides but also for carbonitrides which have been Segments
fabricated in a similar manner.1,46 The conclusions Length (mm) 518 518
Fuel column length 394 394
were that the choice of Sphere-pac or pellet fuel for (mm)
reactor application is dependent on preferred dif- Smear density (% TD) 89.5 86.5
ferences in fabrication (e.g., economics and environ- Fuel/clad gap BOL 190 0.0
mental factors) and not on differences in irradiation (mm)
behavior.42 Cladding Zircaloy-4 Zircaloy-4
Inner diameter (mm) 9.3 9.3
The experimental results gained in this program Outer diameter (mm) 10.75 10.75
are compared to modeling using the PSI codes Fill gas/pressure (bar) He/22.5 He/22.5
SPACON and SPHERE. For these considerations
see Chapter 3.25, Modeling of Sphere-Pac Fuel.
3.11.6.2.8 Siemens-PSI Gösgen PWR: Table 7 Relative increase in diameter and length of the
uranium oxide fuel segments irradiated in Gösgen
In 1986, 12 segments containing Sphere-pac UO2 fuel Number of Relative increase in Relative increase in
produced by the internal gelation method at PSI and cycles diameter length
8 segments containing UO2 pellets produced by
Sphere- Pellet Sphere- Pellet
Siemens AG were loaded in the Gösgen PWR for a pac fuel fuel pac fuel fuel
comparative irradiation test. The segments were
connected to form four equivalent full-length fuel 1 0.06% 0.31% 0.132% 0.135%
2 0.03% 0.59% 0.329% 0.377%
rods. The Sphere-pac fuel consisted of three frac-
3 0.15% 0.70% 0.609% 0.570%
tions UO2. For fuel and segment characteristics see 4 0.17% 0.72% 0.925% 0.879%
Table 6. One rod was removed at each annual shut-
down for PIE at nominal burnup levels of 14, 30, 41,
and 54 MWd kgU1. The average linear power was
close to 300 W cm1 for the first two cycles and about
250 W cm1 for the last two cycles.43 3.11.6.2.9 HBWR Halden: ramp testing of
The PIE revealed no failure or center melting uranium oxide Sphere-pac fuel
of the pins. The changes in diameter and length of The KWU-PSI Halden test was performed on
the fuel segments are listed in Table 7.101 The two refabricated segments from the experiment
fuel ceramography of a pin irradiated during one KWU-PSI Gösgen (see Section 3.11.6.2.8). The
cycle showed in the center completely necked small test segments selected for this test were irradiated
fraction spheres and slightly necked medium and for one and three operation cycles, respectively KKG
large fraction spheres; Figure 6. No effect of macro- (Swiss nuclear reactor ‘‘Kernkraftwerk Goesgen’’).
scopic shrinkage was visible.101 A fission gas release The objective was to compare pellet fuel and
of 0.8% was observed (rod average), and the Sphere-pac fuel under ramping conditions with
corresponding calculation with the SPHRE-II code respect to fission gas release and failure threshold.
resulted in 1.36% (see Chapter 3.25, Modeling of The instrumented irradiation was performed in the
Sphere-Pac Fuel, Table 3). Halden Test Reactor, HBWR, and included ramp
Sphere-Pac and VIPAC Fuel 305
tests on two segments (one pellet, one Sphere-pac) in burnup of the studied segments was in the range
parallel under the same conditions. of 53–59 GWd tM1 after six irradiation cycles in
The test was carried out in 1994/95. The burnup Beznau-I reactor (PWR) at an average power level
was increased from 14 to 21.5 MWd kg1 UO2 at a of about 170 W cm1.
maximum linear heat rate of 480 W cm1. The diam- After the base irradiation, the rods were trans-
eter of the pellet segment increased from 0.31% ferred to PSI for examination. One Sphere-pac seg-
after the basic irradiation to 0.17%, related to the ment out of four segments were transferred to NRG
initial diameter. The Sphere-pac fuel segment showed for ramp testing at power rate of 100 W cm1 min1
no significant change in diameter. The visual inspec- until a power of 400–500 W cm1 was maintained
tion and the fuel ceramography revealed no failure during 24 h. These tests were successfully performed
and no center melting. The fuel structure showed for the pelletized segments but the Sphere-pac
an outer section with considerable sintered fine failed at a power of about 300 W cm1. The PIEs
spheres and slightly sintered medium and large revealed that numerous cracks followed the zirco-
spheres. The inner part of the fuel was completely nium hydride needles and indicated the failure
sintered. The transition zone between these two mechanism of delayed hydrogen clacking. The fission
zones was small. The sintered inner region showed gas release determined by puncture, the luminosity of
differences in grain appearance and porosity. The the b-g auto-radiographies, and the volatile fission
density in the center was lower than the smear den- products content found by EPMA indicated a high
sity while the fuel close to the transition to the release in the case of the Sphere-pac fuel. After base
slightly sintered region appeared to be denser as it irradiation, fission gas release of 5% was determined
was observed and described for MFBS 7 (see in the pelletized fuel and 29% for the Sphere-pac
Section 3.11.6.2.3). While the experimentally iden- segment. These amounts are comparable to those
tified fission gas release was 0.8% for the base irradi- found after ramp test in the pelletized segments.
ation in the Gösgen PWR, the ramp test in the The Sphere-pac segment presents a complete
Halden reactor resulted in a remarkable fission gas restructuring of its center after ramp. The rim effect
release of 45% (compared to the 36% determined seems to be spread over a larger distance in Sphere-
by modeling; see Chapter 3.25, Modeling of Sphere- pac fuel. The peripheral zone of fine pore density was
Pac Fuel, Table 3). The test and its evaluation are about 450 mm and the Pu build-up was about 320 pm.
described in Nordström and Botta.102
Figure 25 shows a ceramography with the differ- 3.11.6.2.11 HFR Petten: FUJI program
ent sintering stages in the same cut. In this project, four kinds of fuel, that is, MOX
A comparison to modeling can be found in Sphere-pac, MOX Vipac (see Section 3.11.6.1.4),
Chapter 3.25, Modeling of Sphere-Pac Fuel. MOX pellet, and Np-MOX Sphere-pac fuels were
simultaneously irradiated under identical conditions.
3.11.6.2.10 NOK Beznau/HFR Petten-M308 It was decided to test the two available types of
program: uranium plutonium oxide44,45 particle fuel, Sphere-pac and Vipac, as each of the
The Sphere-pac fuel was fabricated with the internal types has potential advantages and disadvantages. All
gelation process in PSI. The U–Pu homogeneity together, 16 fuel segments were fabricated at the PSI:
was ensured by coprecipitation of uranium and plu- 5 MOX pellet segments, 9 Sphere-pac segments with
tonium. The segments were filled with three batches MOX spheres, and 2 segments with MOX Vipac fuel
of different size spheres (around 1100, 300, and (see Section 3.11.6.1.4). The MOX contained 20%
50 mm). The average plutonium content of the fuel Pu in all segments. The nine Sphere-pac segments
stack was 4% of the metal content. with MOX spheres (two sphere size fractions) were
The two Sphere-pac segments lengths were in both produced by two different filling methods yielding a
cases 527 mm (nominal value) with a fuel stack length higher and a lower smear density (one segment with
of 420 mm. For the base irradiation in the Beznau lower smear density). The filling methods were infil-
Power Plant, the segment was assembled together tration (higher density) and parallel filling (lower
with segments with pelletized fuel in two rods which density).
had been built into a standard 14 14 fuel assembly. For the parallel filling each fraction was precisely
The test aimed to investigate the behavior of weighed beforehand and fed into the two remote-
pelletized and Sphere-pac MOX fuel after base irra- controlled feeders. The feeder rates were adjusted
diation and during power ramp test. The average to start and stop simultaneously. The 250 mm fuel
306 Sphere-Pac and VIPAC Fuel
200 mm
2 1 1
2
3
d: distance from rim-image center
x: geometric center dx: 4.65 mm
3
100 mm
4 4
d4: 2.9 mm 5 5
d5: 3.4 mm
6
6
7
7 X
d6: 3.9 mm
d7: 4.65 mm
500 mm
Figure 25 Fuel ceramography images of the S04 Sphere-pac fuel in polished condition, showing the state of sintering along
the fuel radius (one transverse cross-section from center position to cladding and several details). Reproduced from
Nordström, L. -Ã.; Nakamura, M. PIE and modelling results for the irradiation and transient test of Sphere-Pac UO2 fuel
(IFA-550.11). In EHPG-Meeting, Sandefjord, Norway, May 2004, OECD Halden Proceedings HPR-362, with permission from
Institutt for Energiteknikk, OECD Halden Reactor Project.
column length was filled within 25 s. During this tapped against the pin and caused it to jump. These
procedure, the pin was not vibrated. Then, the fuel jumps could easily be more than 1 mm in amplitude
column was axially fixed with a spring device, in and were adjusted to 0.5 mm.
order to hinder segregations of the two fractions. For the infiltration filling, the coarse fraction (710–
Then, the vibration was started. For the vibration, a 800 mm) was poured through a 2 mm aperture into
50-Hz vibrator was placed under the pin and the the pin, in order to achieve a coordinated filling. The
distance between the vibrator and the pin was pin was softly tapped by a vibrator during this filling
adjusted. The voltage of the vibrator could be procedure. This soft vibration was achieved by choos-
adjusted up to 220 V which corresponded to an ing a small gap between the vibrator and the pin.
amplitude of <1 mm. Because of the gap, the vibrator Then, a sieve-like device was placed, in order to fix
Sphere-Pac and VIPAC Fuel 307
the coarse spheres while allowing the fine spheres The use of the PSF allows the application of power
(63–75 mm) to pass through. The fine spheres were histories independent of the reactor operation. In all
poured onto the sieve and flowed down the vibrated irradiation tests, the power was decreased to zero very
pin through the spaces between the coarse spheres, rapidly after the irradiation was completed. The 25 cm
filling them up. For more details, see Hellwig et al.103 long active fuel columns were pro forma divided into
The fuel of two Sphere-pac segments with higher nine equidistant slices for the test evaluation and
smear density contained additional 5% Np. All these modeling. The axial power profile was very pro-
fuel segments were produced and delivered to NRG nounced, for example, covered in the IST test ranging
for four irradiation tests in total. Table 8 (see also from 230 to 620 W cm1 between midpoints of the
Nordström et al.104) gives an overview of the main two end slices. Power histories are shown together
characteristics of the three fuel segments described with the modeling results further below. The radial
in this chapter, one each from the first three irradia- temperature distribution was adjusted to fast reactor
tion tests. conditions by controlling the cladding temperature
The irradiation took place in the pool-side- and the fuel power, in order to obtain data concerning
facility (PSF) of the HFR in Petten, in an irradiation the sintering and restructuring behavior under irradi-
rig with two fuel channels, each of them holding a pin ation conditions similar to those in a fast reactor.
consisting of two fuel segments. The irradiation tests For more details on the fabrication of the different
can be distinguished as follows: fuel segments and the features of the FUJI experi-
ment see Pouchon et al.41 and Hellwig et al.105
IST: from zero to full power within 36 h (this reflects All four irradiation tests were completed success-
the start-up procedure of the JOYO fast reactor). fully without any rod failure.
Restructuring 1 test (RT1): from zero to full power The results of this study were compared to
within 36 h, then a holding time of 48 h at full modeling, see Chapter 3.25, Modeling of Sphere-
power. Pac Fuel.
Restructuring 2 test (RT2): from zero to full power
within 36 h, then a holding time of 96 h at full
power. 3.11.7 Conclusions
Power-to-melt test (PTM): first part comparable
to the RT1; subsequently the power was raised at Particle fuel is an innovative fuel form, which has
the rate of 5% per minute until central fuel melt- extensively been tested for light water and especially
ing could be observed in neutron radiographs. fast reactors. It has, however, not been used
Table 8 Fuel and segment characteristics, as fabricated data for the FUJI program
Source: Nordström, L. Å.; Hellwig, Ch.; Nakamura, M.; Bakker, K. PIE and modelling results for the FUJI irradiation test of Sphere-pac MOX
fuel. In EHPG-Meeting, Lillehammer (NO), Oct 2005; OECD Halden Proceedings HPR-364.
308 Sphere-Pac and VIPAC Fuel
commercially yet and can still be considered as an Additionally, there is still a large potential for
innovative fuel concept. future materials, for example, oxide dispersion
The basic idea is a simplification of the production strengthened (ODS) steels.
process, by forming fuel particles straight out of the The fuel pins are designed for fission gas
base solution and introducing them directly into the release. Therefore, the Sphere-pac concept, with
fuel pin. This not only promises a major saving on its potentially larger release, does not represent a
mechanical devices, such as grinding machines and disadvantage.
presses, but also offers an almost dustless operation. Besides the classical urania fuel base (UO2, MOX,
These advantages become relevant for highly active MA-MOX), other matrices are also of interest.
fuels, particularly if the source is a reprocessing of These are other oxides (e.g., YSZ as inert matrix
spent fuel. Depending on whether it is an aqueous fuel), carbides, nitrides, or carbo-nitrides. The pro-
reprocessing or it is based on pyroprocessing, the duction of a potential alternate matrix (e.g., by
particle fuel is Sphere-pac fuel or Vipac (also vibro- carbothermic reaction) is very easy as experienced
packed) fuel, respectively. in case of the Sphere-pac fuel.
In operation, the particle fuel behaves partially Particle beds can have advantageous FCMI, as the
different from the classical pellet fuel form, even network of interstice between the spheres provides
though the matrix is the same. At the BOL, the space for swelling of the fuel matrix. This is espe-
particle fuel has a different geometrical appearance, cially useful for materials with high swelling rates,
with a much higher specific surface, a potentially as for example carbide fuels.
lower smear density, and a connected network of inter-
A reservation in the Sphere-pac usage in any
stices. After reactor operation for some time, high
system has to be made because of the observation
temperatures will cause major restructuring resulting
described in Section 3.11.6.2.4.4; that is, van der
in a partial pellet-like appearance.
Linde reports a faster failure of Sphere-pac fuel
In order to reach high smear densities, several size
pins during accidental conditions (linear heat rates
fractions are filled into the pin. In case of the Vipac
above 450 W cm1) because of the low thermal con-
concept, the filling is always a parallel process,
ductivity of the near cladding, nonrestructured
whereas in case of the Sphere-pac fuel, the filling
Sphere-pac fraction, which causes a higher tempera-
can be a parallel or a sequential process. The sequen-
ture increase of the inner restructured ring, and with
tial process is commonly called infiltration filling.
thermal expansion a higher strain onto the cladding,
Clear disadvantages of particle fuels are the initial
compared to pellet fuel.
lower thermal conductivity, the high specific surface,
The application in LWRs, however, is not a pref-
and the large restructuring at BOL. On the other hand,
erable option.
even for the initial fuel there is no gap between the
The reasons for not using particle fuel in the
fuel and the cladding and the restructuring will lead to
present form in LWRs are as follows:
a pellet-like structure in a major part of the fuel.
According to the numerous experiments being The LWRs are not the preferred systems for burn-
performed (see Section 3.11.6.1 for Vipac and ing MAs; therefore, it does not make sense to
Section 3.11.6.2 for Sphere-pac fuel), the fast sodium introduce MA containing fuel. For nonremote
cooled reactor is the ideal system for particle fuel. operation, the present pellet production processes
This has several reasons. are highly cost effective and well experienced. It is
Reasons to use particle fuel for fast reactors: very unlikely that an alternate production process
can compete with such an established concept.
The fast spectrum is adequate for actinide and MA The thermal spectrum limits the materials selection
burning. Low decontaminated fuel from reproces- for the cladding to typically zirconium based alloys.
sing is highly radioactive. It must therefore be Highly sophisticated steel materials are excluded
produced remotely with a preferably simple pro- because of their high cross-sections.
cess, such as the internal gelation and the scrap On one hand, the LWR fuel pins are designed to a
extraction from reprocessing. certain fission gas release. A higher release can
Fast reactors provide larger flexibility in the choice have a negative influence on the corrosion behav-
of cladding material. The steels being commonly ior and potentially enhance corrosion stress crack-
used for the past and today’s fast reactors showed ing. On the other hand, the higher initial specific
very good compatibility with the particle fuel. fuel surface of particle fuel, plus the bigger
Sphere-Pac and VIPAC Fuel 309
restructuring will enhance fission gas release, com- 4. Hauth, J. J. In Perspectives in Powder Metallurgy, Vol. 2,
Vibratory Compacting Principles and Methods;
pared to the pellet. Especially for some power Hausner, H. H., Ed.; Plenum Press: New York, 1967;
ramp tests, this was confirmed experimentally pp 253–276.
(see Sections 59.6.2.8, 59.6.2.9, and 59.6.2.10). 5. Hauth, J. J. Nucleonics 1962, 20(9), 50–54.
6. Herbig, R.; Rudolph, K.; Lindau, B.; Skiba, O.;
Similar to the Vipac concept, where metallic ura- Maershin, A. J. Nucl. Mater. 1993, 204, 93–101.
7. Steinkopff, H.; Krompaß, R.; Schumann, K.; et al.
nium is added as an oxygen getter, one could also Kernenergie 1987, 30(11/12), 458–465.
envisage having a heterogeneous concept in case of the 8. Steinkopff, H.; Krompaß, R.; Skiba, O. V. J. Nucl. Mater.
Sphere-pac fuel. Here, a secondary phase could even 1991, 178(2–3), 163–170.
9. Tzykanov, V. A.; Skiba, O. V.; Tzarenko, A. F.;
form an interconnected second phase throughout the Porodnov, P. T.; Mayorshin, A. A.; Makarov, V. A.
fuel, by introducing it as a smaller infiltrated fraction. In In European Nuclear Conference (ENC’90), Lyon, France,
this case, it could, for example, enhance the thermal Sept 23–28, 1990; ENS/ANS-Foratom Conference
Transactions; Verlag TÜV: Rheinland, 1990; Vol. IV,
conductivity. As for the Vipac, it could also regulate pp 2075–2077.
the oxygen potential, or bind fission gases. Both con- 10. Vavilov, S.; Kobayashi, T.; Myochin, M. J. Nucl. Sci.
cepts therefore still have large development potential. Technol. 2004, 41(10), 1018–1025, http://dx.doi.org/
10.3327/jnst.41.1018.
Because of the wall effects, a secondary fine 11. Bychkov, A. V.; Skiba, O. V.; Porodnov, P. T.;
fraction of a Sphere-pac is also present in a higher Kormilitzin, M. V.; Babikov, L. G. Development of the
concentration near the cladding. This enables the pyroelectrochemical process for demonstration fuel
cycle of an actinide burner reactor. In International
possibility to introduce a secondary fraction with Conference on Evaluation of Emerging Nuclear Fuel
near cladding requirements, for example, a different Cycle Systems GLOBAL’95, Versailles, France,
fissile concentration. Sept 11–14, 1995; pp 516–523.
12. Kirillovich, A. P.; Bychkov, A. V.; Skiba, O. V.; Babikov, L. G.;
It is important to emphasize that particle fuel is an Lavrinovich, Y.; Loukinykh, A. N. Safety analysis of fuel
innovative nuclear fuel form, and that both the Vipac and cycle processes based on ‘‘dry’’ pyrochemical fuel
the Sphere-pac concepts were extensively tested, and reprocessing and vibropac technology. In Global’97,
International Conference on Future Nuclear Systems,
proved to have very good irradiation behavior for fast Yokohama, Japan, Oct 5–10, 1997; Vol. 2, pp 900–905.
systems. Particle fuel is therefore not an exotic system, 13. Ivanov, V. B.; Skiba, O. V.; Mayershin, A. A.; Bychkov, A. V.;
but can realistically be envisaged for future usage. Demidova, L. S.; Porodnov, P. T. Experimental, economical
and ecological substantiation of fuel cycle based on
pyroelectrochemical reprocessing and vibropac
technology. In Global’97, International Conference on
Future Nuclear Systems, Yokohama, Japan, Oct 5–10,
Acknowledgements 1997; Vol. 2, pp 906–9011.
14. Bychkov, A. V.; Vavilov, S. K.; Skiba, O. V.; Porodnov, P. T.;
The compilation of this overview chapter was Paravin, A. K.; Popkov, G. P. Pyroelectrochemical
reprocessing of irradiated FBR fuel. III. Experiment
partially financed by the CCEM project PINE. of high burn-up fuel of the BOR-60 reactor.
The IAEA, the JAEA, the American Nuclear Society In Global’97, International Conference on Future
and the OECD Halden Reactor project are thanked Nuclear Systems, Yokohama, Japan, Oct 5–10, 1997;
Vol. 2, pp 912–917.
for their courtesy to reproduce figures from their 15. Flipot, A. J.; Papeleux, P. Energ. Nucl. 1970, 17(1),
reports. 46–52.
16. Botta, F.; Hellwig, Ch. Nucl. Sci. Eng. 2000, 135, 1–12.
17. Wallin, F. H.; Nordström, L. Å.; Hellwig, Ch. The
Sphere-pac fuel code SPHERE-3. In Proceedings of the
References IAEA TCM on Nuclear Fuel Behaviour Modelling at High
Burnup, Windermere, Jun 2000; CD-ROM.
18. Botta, F. Out of pilse loss of sphere tests on
1. Ledergerber, G.; Ingold, F.; Stratton, R. W.; Alder, H. P. UO2-Sphere-pac pins with cladding cracks,
Nucl. Technol. 1996, 114, 194–204. TM-43-91-03; Technical Report; Paul Scherrer
2. Nakamura, M.; Sekine, N.; Pouchon, M. A.; Miyamoto, H.; Institut: Villigen, Switzerland, 1991.
Nakajima, Y. 2002. Development of vibro-packed fuel 19. Bottcher, J. H. FBR MOX particle-fuel behavior in failure
design code ( ); scenarios. In Proceedings of the 2000 Fall Meeting of the
Technical Report 5; http://jolisfukyu.tokai-sc.jaea.go.jp/ Atomic Energy Society of Japan, Aomori-daigaku,
fukyu/gihou/pdf2/n15-05.pdf. Japan, 2000; Vol. 3, p 808.
3. Hauth, J. J. Vibration-Compacted Ceramic Fuel 20. Bottcher, J. H.; Inoue, M.; Nomura, S. Operational
Elements, HW-60346; Technical Report, General Electric reliability testing of MOX fuels for future FBRs.
Company; Hanford Atomic Products Operation: In Proceedings of the Global 1997, Yokohama, Japan,
Richland, WA, 1959; Hauth, J. J.; Anicetti, R. J.Vibrational Oct 5–10, 1997; pp 1086–1092.
Compaction of Ceramic Powders for Nuclear Fuel 21. Lee, J. S.; Yang, M. S.; Park, H. S.; Boczar, P.;
Elements, HW-SA-1735; Technical Report; United States Sullivan, J.; Gadsby, R. D. In Nuclear Fuel Cycle and
Atomic Energy Commission: Washington, DC, 1959. Reactor Strategies: Adjusting to New Realities,
310 Sphere-Pac and VIPAC Fuel
International Symposium Held in Vienna, Jun 3–6, 1997; Fuel Performance, Avignon, France, Apr 21–24, 1991;
International Atomic Energy Agency: Vienna, Austria, Vol. 1, pp 174–183.
1997; pp 153–157, IAEA TECDOC-990. 44. De Limelette, C. NOK – M308 international programme:
22. Lee, J. S.; Choi, J. W.; Ko, W. I.; Jung, I. H.; Yang, M. S.; A MIMAS and SPHERE-PAC experimental data base for
Park, H. S. Cost benefits analysis of vibratory high burnup MOX fuel – final report, M308 2001/19
compaction technology for dupic fuel fabrication. Rev. A – BN Ref. 0104015/221; Technical Report;
In World Nuclear Congress. Transactions Vol. III: Poster Belgonucleaire: Bruxelles, 2001.
Papers, ENC 98 World Nuclear Congress, Nice, France, 45. Markgraf, J. W. F. NOK – M308 international programme:
Oct 25–28, 1998; pp 517–520. A MIMAS and SPHERE-PAC experimental data base for
23. McGeary, R. K. J. Am. Ceram. Soc. 1961, 44, 513, high burnup MOX fuel – power ramp testing of fuel rods
DOI: 10.1111/j.1151-2916.1961.tb13716.x. SM36, SB05, SM41 and SP58, M308 99/06 Rev. A – BN
24. Van Der Linde, A. Cer-Met Sphere-pac Fuel Potential; Ref. 9907798/221; Technical Report; Belgonucleaire:
Netherlands Energy Research Foundation, Petten, 1990; Bruxelles, 1999.
ECN-RX-90-031. 46. Ledergerber, G.; Kopajtic, Z.; Ingold, F.; Stratton, R.
25. Pouchon, M. A.; Nakamura, M.; Hellwig, Ch. J. Nucl. Sci. J. Nucl. Mater. 1992, 188, 28–35, http://dx.doi.org/
Technol. 2002, (Suppl. 3), 830–833. 10.1016/0022-3115(92)90450-Y.
26. Pouchon, M. A.; Nakamura, M.; Hellwig, Ch.; Ingold, F.; 47. Prunier, C.; Bardelle, P.; Pages, J. P.; Richter, K.;
Degueldre, C. J. Nucl. Mater. 2003, 319, 37–43. Stratton, R. W.; Ledergerber, G. In International
27. Kofuji, H.; Sato, F.; Myochin, M.; et al. J. Nucl. Sci. Conference on Fast Reactors and Related Fuel
Technol. 2007, 44(3), 349–353, http://dx.doi.org/ Cycles, Kyoto, Japan, Oct 28–Nov 1, 1991; Atomic
10.3327/jnst.44.349. Energy Society of Japan: Japan, 1991; Vol. 2,
28. Ayer, J. E.; Soppet, F. E. J. Am. Ceram. Soc. 1966, 49(4), pp 15.9/1–15.9/12.
207–210, DOI: 10.1111/j.1151-2916.1966.tb13235.x. 48. Faugere, J. L.; Pelletier, M.; Rousseau, J.; Guerin, Y.;
29. Steinkopff, H.; Schurig, C. Kernenergie 1967, 10, Richter, K.; Ledergerber, G. The CEA helium-bonded
357–363. nitride fuel design. In International Conference on
30. Brambilla, G.; Gerontopouos, P.; Neri, D. Energ. Nucl. Design and Safety of Advanced Nuclear Power Plants,
1970, 17(4), 217–224. Tokyo, Japan, Oct 25–29, 1992; Atomic Energy Society
31. Ganguly, C.; Langen, H.; Zimmer, E.; Merz, E. R. Nucl. of Japan: Japan, 1992; Vol. 3, pp 28.1/1–28.1/6.
Technol. 1986, 73(1), 84–95. 49. Streit, M. Ph.D. Thesis, Eidgenössische Technische
32. Nickel, H.; Nabielek, H.; Pott, G.; Mehner, A. W. Nucl. Hochschule, ETH Zürich, Switzerland, 2004;
Eng. Des. 2002, 217(1–2), 141–151, http://dx.doi.org/ http://e-collection.ethbib.ethz.ch/view/eth:27028?
10.1016/S0029-5493(02)00128-0. q=15403.
33. Heit, W.; Huschka, H.; Rind, W.; Kaiser, G. G. Nucl. 50. Daumas, S. Ph.D. Thesis, Université d’Aix-Marseille 1,
Technol. 1985, 69, 44–54. Marseille, France, 1997.
34. Charollais, F.; Fonquernie, S.; Perrais, C.; et al. Nucl. Eng. 51. Daumas, S.; Ledergerber, G.; Ingold, F.; Bauer, M.;
Des. 2006, 236(5–6), 534–542, http://dx.doi.org/10.1016/ Prunier, C. Nitride targets elaborated by sol–gel
j.nucengdes.2005.11.022. processing for actinides incineration. In Proceedings of
35. van der Brugghen, F. W.; Noothout, A. J.; Hermans, M. E. A.; the Global 1995, Versailles, France, Sept 11–14, 1995.
Kanji, J. B. W.; Votocek, O. A U(VI)-process for 52. Tanaka, K.; Kawaguchi, K.; Koike, K. In Proceedings of
microspheres production. In Symposium on Sol–Gel the Global 2003, Advanced Fuel Fabrication System
Processes and Reactor Fuel Cycles, Gatlinburg, TN, Concepts Using Advanced Aqueous Processing
1970; CONF-700502, pp 253–263. Product, New Orleans, LA; 2003; pp 1354–1365.
36. Kanij, J. B. W.; Noothout, A. J.; Votocek, O. The KEMA U(IV) 53. Knotik, K.; Leichter, P.; Bonek, E. Atomenergie
process for production of UO2 microsphere. In Sol–Gel Kerntechnik 1981, 39, 191–194.
Process for Fuel Fabrication, IAEA-161; 1974; 54. Yamagishi, S.; Hasegawa, A.; Ogawa, T. Development of
pp 185–195. Rapid Gelation Apparatus with Microwave Heating;
37. Jungo, Ch.; Ledergerber, G. Verfahren und Vorrichtung JAERI-Tech 94-010; 1994.
zur Herstellung von Mikrokugeln durch interne Gelierung 55. Yamagishi, S.; Hasegawa, A. Method and apparatus for
von Mischfeedtropfen. European Patent EP 0 048 360 producing gel particles. U.S. Patent 5,581,589, 1996.
B1, 1982; Process and apparatus for producing 56. Yamagishi, S. J. Nucl. Mater. 1998, 254, 14–21.
microspheres. U.S. Patent 4,431,164, 1984. 57. Rosin, A.; Schmidt, A.; Gerdes, T.; et al. Microwave
38. Ledergerber, G. Trans. Am. Nucl. Soc. 1982, 40, 55–56. assisted internal gelation of droplets – a case study.
39. Ledergerber, G. ‘‘Internal gelation using microwaves’’ In Proceedings of the 10th International Conference on
IAEA-TECDOC, advanced fuel technology and Microwave and High Frequency Heating, Modena, Italy,
performance. In Proceedings of an Advisory Group 2005; pp 346–349, http://users.wpi.edu/qwug/files/
Meeting, Würenlingen, Switzerland, Dec 4–6, 1984; Rosin-Schmidt-et-al_2005.pdf.
International Atomic Energy Agency, 1985; Vol. 352, 58. Morrison, A.; Cross, M. J. Bell System Tech. J. 1974,
pp 165–174. 53(6), 955–1019.
40. Ledergerber, G. Internal Gelation for Oxide and Nitride 3.11. Pouchon, M. A.; Ingold, F. Internal gelation at PSI: The
Particles; JAERI-Review 96-009; 1996. past and the future. In Proceedings of the 2009
41. Pouchon, M. A.; Ingold, F.; Kopajtic, Z.; Tomita, Y.; International Congress on Advances in Nuclear Power
Kono, S. Fabrication and characterization of MOX Plants, ICAPP 2009, Shinjuku, Tokyo, May 10–14, 2009;
microspheres for the FUJI project. In Proceedings of the No. 9494.
Global 2003, New Orleans, LA, 2003; pp 653–657. 60. Wymer, R. G. Laboratory and engineering studies of
42. Stratton, R. W.; Ledergerber, G.; Ingold, F.; Latimer, T. W.; sol–gel processes at ORNL. In Proceedings of a Panel on
Chidester, K. M. J. Nucl. Mater. 1993, 204, 39–49. ‘‘Sol–Gel Processes for Ceramic Nuclear Fuels’’; IAEA:
43. Stratton, R. W.; Botta, F.; Hofer, R.; et al. A comparative Vienna, 1968; pp 131–172.
irradiation test of UO2 Sphere-pac and pellet fuel in the 61. Bradley, R. A.; Conlin, J. A.; Cox, C. M.; et al. Fast
Gösgen PWR. In International Topical Meeting on LWR Breeder Reactor Oxide Fuels Development – Final
Sphere-Pac and VIPAC Fuel 311
Report, ORNL-4901 UC-79b; Technical Report; 79. Lahr, H. W. H. Nucl. Technol. 1976, 31, 183–190.
Oak Ridge National Laboratory: Oak Ridge, TN, 1973. 80. Cervellati, A.; Falcinelli, G.; Lepscky, C.; Cogliati, G.;
62. McBride, J. P.; McCorkle, K. H.; Pattison, W. L.; Guermani, P.; Testa, G. Irradiation tests of sol–gel products.
Finney, B. C. Nucl. Technol. 1972, 13, 148–158. In Symposium on Sol–Gel Processes and Fuel Cycles,
63. McBride, J. P.; McCorkel, K. H. In Laboratory Studies of Gatlinburg, TN, May 4–7, 1970; pp 374–413,
Sol–Gel Processes at the Oak Ridge National Laboratory, CONF-700502.
ORNL-TM-1980; McBride, J. P., Ed.; Oak Ridge National 81. van der Linde, A.; Lucas Luyckx, H. J. B. Performance of
Laboratory: Oak Ridge, TN, 1967; http://www.osti.gov/ Sphere-pac UO2 fuel pins irradiated in a loop; ECN
energycitations/product.biblio.jsp?osti_id=4197859. Report ECN-38; Mar 1978.
64. Fu, X.; Liang, T.; Tang, Y.; Xu, Z.; Tang, C. J. Nucl. Sci. 82. van der Linde, A.; Lucas Luyckx, H. J. B.;
Technol. 2004, 41(9), 943–948, http://dx.doi.org/ Verheugen, J. H. N. Fabrication, irradiation and post-
10.3327/jnst.41.943. irradiation examinations of MO2 and UO2 Sphere-pac
65. Streit, M.; Ingold, F.; Pouchon, M. A.; Gauckler, L. J.; and pellet fuel pins, irradiated in a PWR loop; Report
Ottaviani, J. P. J. Nucl. Mater. 2003, 319, 51–58, ECN-91; Apr 1981.
http://dx.doi.org/10.1016/S0022-3115(03)00133-8. 83. van der Linde, A. Irradiation performance and post-
66. Baader, F. H.; Graule, T. J.; Gauckler, L. J. Ind. Ceram. irradiation examinations of the instrumented Sphere-pac
1996, 16(1), 36–40. UO2 assembly IFA-204, irradiated up to 1.7% FIMA in the
67. Ayer, J. E.; Soppet, F. E. J. Am. Ceram. Soc. 1965, 48(4), Halden Boiling Water Reactor: ECN; Report ECN-127;
180–183, DOI: 10.1111/j.1151-2916.1965.tb14708.x. Dec 1982 and OECD Halden Project Report HPR-290;
68. Cumberland, D. J.; Crawford, R. J. The Packing of Dec 1982.
Particles, Handbook of Powder Technology Elsevier: 84. van der Linde, A. Comparative testing of instrumented
Amsterdam, 1987, Vol. 6, p 45. Sphere-pac and pellet UO2 fuel rods up to rod average
69. Lee, D. I. J. Paint Technol. 1970, 42(550), 579–587. burnups of 44 MWd/kg UO2 in the Halden Boiling Water
70. Knudsen, P.; Bagger, C.; Kjaer-Pedersen, N. Analysis of Reactor; Report ECN-208; Jun 1988.
overpower performance of high-burnup pellet and Vipac 85. van der Linde, A. Sphere-pac versus pellet UO2 fuel in
UO2–Zr fuel pins. In Transactions of the American the Dodewaard BWR; ECN Report ECN-213; Apr 1989;
Nuclear Society; American Nuclear Society: La Grange Non-destructive post-irradiation examinations of
Park, IL, 1976; Vol. 24, p 172. Sphere-pac and pellet UO2 fuel rods irradiated in the
71. Mayorshin, A. A.; Bychkov, A. V.; Kormilitsyn, M. V.; et al. Dodewaard BER. In Proceedings of EHPG Meeting on
Experimental transmutation of neptunium in the BOR-60 Fuel Performance Experiments and Analysis; OECD
reactor in the form of vibropac UNpO2 fuel. In Seventh Halden Project Report HPR-330/18; 1986; Comparative
Information Exchange Meeting on Actinide and Fission testing of UO2 Sphere-pac and pellet fuel in the
Product Partitioning and Transmutation, Session III: Dodewaard BWR. In Proceedings of the European
Progress in Fuels and Targets, Jeju, Republic of Korea, Nuclear Conference, Lyon, France, Sept 23–28, 1990;
Oct 14–16, 2002, OECD/NEA, OECD Nuclear Energy ENC’90, pp 1900–1903.
Agency: France, 2002; http://www.nea.fr/html/pt/docs/ 86. Fitts, R. B.; Miller, F. L. Nucl. Technol. 1974, 21,
iem/jeju02/session3/SessionIII-11.pdf. 304–310.
72. Gratchyov, A. F.; Skiba, O. V.; Tsykanov, V. A.; et al. 87. Van der Linde, A.; Verheugen, J. H. N. Nucl. Technol.
J. Nucl. Sci. Technol. 2007, 44(3), 504–510, http://dx.doi. 1982, 59, 70–77.
org/10.3327/jnst.44.504. 88. van der Linde, A. Microscopic examinations of a
73. Morihira, M.; Nakamura, M.; Ozawa, T.; et al. Final report Sphere-pac and pellet UO2 fuel rod irradiated during
of the FUJI project concerning the research and 1530 days in the Dodewaard BWR; Report ECN-199;
development of advanced Sphere-pac fuel among JNC, 1987.
PSI and NRG, JNC TY8400 2005-006; Technical Report; 89. van der Linde, A. Energiespectrum Juni 1990; p 165,
Japan Nuclear Cycle Development Institute (JNC): [in Dutch].
Tokai-mura, Ibaraki, Japan, 2005. 90. Andriesse, C. D. Melting Nuclear Fuel, KEMA Scientific
74. Shigetome, Y.; Kono, S.; Hellwig, C.; Heimgartner, P. and Technical Reports; 1985, Vol. 3(7).
Vipac fuel fabrication for irradiation tests of the FUJI 91. Tanke, R. H. J. J. Nucl. Mater. 1992, 188, 262–272.
project. In Proceedings of the Global 2003, New Orleans, 92. van der Linde, A. Behaviour of a single UO2 sphere,
LA, Nov 16–20, 2003. irradiated at approximately 1900 K in the HFR at Petten;
75. Beatty, R. L.; Norman, R. E.; Notz, K. J. Gel-Sphere-pac Report ECN-191; 1986.
Fuel for Thermal Reactors: Assessment of Fabrication 93. Bakker, K.; Klaassen, F. C.; Schram, R. P. C.; et al. Nucl.
Technology and Irradiation Performance, ORNL-5469; Technol. 2004, 146, 325–331.
Technical Report; Oak Ridge National Laboratory: Oak 94. Smith, L.; Aerne, E. T.; Bürgisser, B.; Flückiger, U.;
Ridge, TN, 1979. Hofer, R.; Petrik, F. The post irradiation examination of a
76. Kitts, F. G.; Fitts, R. B.; Olsen, A. R. In International Sphere-pac (UPu)C fuel pin irradiated in the BR-2 reactor
Symposium Plutonium Fuels Technology, Scottsdale, (MFBS 7 experiment); EIR-Report Nr. 376; Eidg. Inst. f.
AZ, 1967; Horton, K. E., Macherey, R. E., Allio, R. J., Eds.; Reaktorforschung: Würenlingen, 1979.
Nucl. Met. 13; American Institute of Mining, 95. Delbrassine, A.; Smith, L. Nucl. Technol. 1980, 49,
Metallurgical, and Petroleum Engineers: New York, 1968; 129–135.
pp 195–210. 96. Bischoff, K.; Smith, L.; Stratton, R. W. Performance of a
77. Olsen, A. R.; Cox, C. M.; Fitts, R. B. Trans. Am. Nucl. Soc. Sphere-pac mixed carbide fuel pin irradiated in the
1969, 12(2), 605–606. Dounreay Fast Reactor (DFR 527/1 Experiment);
78. Fritz, R. Die Entwicklung und Auslegung eines Prototyp EIR-Report Nr. 415; Eidg. Inst. f. Reaktorforschung:
Brennstofftestelements für einen Würenlingen, 1980.
schwerwassermoderierten und schwerwassergekühlten 97. Bart, G.; Botta, F. B.; Hoth, C. W.; Ledergerber, G. O.;
Thorium Brutreaktor mit vibrationsverdichteten Mason, R. E.; Stratton, R. W. Final Report of a
Brennstoffpartikeln, Jül-654-RG; Technical Report; Comparative Irradiation Test of Mixed Carbide Fuel in the
Kernforschungsanlage: Jülich, Germany, 1970. US Fast Flux Test Facility (Joint USA-Swiss Experiment
312 Sphere-Pac and VIPAC Fuel
AC-3); PSI Bericht Nr. 06-08; Paul Scherrer Institut: In EHPG-Meeting, Loen (NO), May, 1996; OECD
Würenlingen, Jun 2006. Halden Proceedings HPR-347.
98. Bart, G.; Botta, F. B.; Hoth, C. W.; Ledergerber, G.; 103. Hellwig, Ch.; Heimgartner, P.; Tomita, Y.; Kono, S.
Mason, R. E.; Stratton, R. W. J. Nucl. Mater. 2008, 376(1), Sphere-pac filling for irradiation tests of the FUJI
47–59, http://dx.doi.org/DOI:10.1016/j. project. In Proceedings of the Global 2003, New Orleans,
jnucmat.2008.01.022. LA, 2003; pp 1348–1353.
99. Ledergerber, G.; Herbst, R.; Zwicky, H. U.; Kutter, H.; 104. Nordström, L. Å.; Hellwig, Ch.; Nakamura, M.; Bakker, K.
Fischer, P. J. Nucl. Mater. 1988, 153, 189–204, http://dx. PIE and modelling results for the FUJI irradiation test of
doi.org/10.1016/0022-3115(88)90211-5. Sphere-pac MOX fuel. In EHPG-Meeting, Lillehammer
100. Mason, R. E.; Hoth, C. W.; Stratton, R. W.; Botta, F. (NO), Oct 2005; OECD Halden Proceedings HPR-364.
Trans. Am. Nucl. Soc. 1992, 66, 215–217. 105. Hellwig, Ch.; Bakker, K.; Nakamura, M.; Ingold, F.;
101. Reindl, J.; Schleuniger, P.; Zumsteg, R.; Zwicky, H. U. Nordström, L. Å.; Kihara, Y. Nucl. Sci. Eng. 2006, 153,
Zerstörungsfreie Nachbe-strahlungsuntersuchungen an 233–244.
den segmentierten Brennstäben 1104, 1105 und 1106, 106. van der Linde, A. Pin length response of Sphere-pac and
Zwischenbericht Nr. 2; TM-43-92-16; Paul Scherrer pellet UO2 fuel pins to power increase during irradiation
Institut: Villigen, 1992. up to 33 MWd/kg UO2 in the Halden Boiling Water
102. Nordström, L. Å.; Botta, F. Comparative FGR ramp Reactor; OECD Halden Project Report HPR-305/10;
tests with pellet and Sphere-pac fuel, IFA-550.9. 1983.
3.12 Uranium–Zirconium Hydride Fuel
D. Olander
University of California, Berkeley, CA, USA
K. Konashi
Tohoku University, O-arai, Ibaraki, Japan
M. Yamawaki
University of Tokyo, Tokyo, Japan
313
314 Uranium–Zirconium Hydride Fuel
3.12.5 Comparison of LM-Bonded Hydride Fuel and Oxide Fuel in LWRs 345
3.12.6 Hydride Fuel Fabrication 347
3.12.7 Gap Closure During Operation 349
3.12.7.1 Fuel Swelling 349
3.12.7.1.1 Thermal expansion 350
3.12.7.1.2 Hydrogen expansion 351
3.12.7.1.3 Solid fission-product swelling 351
3.12.7.1.4 Total fuel-surface strain 351
3.12.7.2 Cladding Creepdown 351
3.12.7.2.1 Secondary thermal creep 351
3.12.7.2.2 Primary thermal creep 351
3.12.7.2.3 Irradiation creep 351
3.12.7.2.4 Total cladding creep strain 352
3.12.7.3 Gap Closure 352
3.12.7.3.1 Dimension changes 352
3.12.7.3.2 Parameters of typical hydride fuel element 352
3.12.7.3.3 Combined effect 352
3.12.8 Conclusions 353
Appendix A Permeation of Hydrogen Through Stainless Steel 354
Appendix B Effect of Fission Products on H/Zr Ratio of Fuel 354
Reaction of Impurity Oxygen 354
Subsequent Change in H/Zr with Burnup 355
References 355
a
Fissions per initial metal atom; also called ‘metal-atom fraction fissioned.’
types: the SNAP reactor, the LWR-hydride reactor,5 part the reason for the choice of Hastelloy cladding
and the LWR-oxide reactor (a standard boiling-water for SNAP reactor fuel.
reactor, BWR). LWRs cannot tolerate Hastelloy as cladding
Noteworthy in Table 1 is the wide range of the because of its high nickel content and the conse-
uranium contents of the fuels. Hydride fuels operate quent negative effect on neutron economy. Also,
most reliably with low U concentrations, which is the LWR hydride fuel cannot operate for a long period
reason for the choice of 10 wt% U, corresponding to a at >800 C because of the assured catastrophic reac-
uranium density of 0.6 g U cm3, in the SNAP reactors. tion of the Zircaloy cladding with the gaseous hydro-
To produce acceptable amounts of nuclear energy with gen inside the fuel element. Fortunately, the hydride
so little uranium per unit volume, the enrichment was fuel peak temperature at the LWR power density is
93%. In order to attain the same 235U density of the only about 550 C.6
fresh fuel as in an oxide-fueled core, the enrichment of Given the high 235U content of SNAP reactor fuel,
the LWR-hydride fuel would have to be scaled high burnup was easily attainable. The maximum
according to the total uranium density, or burnup of the oxide fuel (60 MWd kgU1) is 1/5
that of the SNAP hydride fuel.
eðLWR-hydrideÞ ¼ eðLWR-oxideÞ ð9:3=3:7Þ
is either aluminum or stainless steel (SS). All pro- b-Zry metal and the ceramic d-ZrHx) are important
posed or constructed hydride-fueled reactors have for hydride fuels: hydrogen diffusion in the control of
adopted the basic TRIGA fuel. the kinetics of hydriding during fabrication and
The exception to the usual low power of these hydrogen redistribution during operation.
research reactors is the Romanian TRIGA, which The hydrogen diffusivities in these two phases14,15,16
utilizes standard-size LWR fuel pellets made of are shown in Figures 2 and 3. While the agreement
(U,Zr)H1.6 instead of UO2. This hydride fuel is among the measured diffusivities in b-Zr is fair (maxi-
reported to have operated at a linear heat rate mum of a factor of 10 discrepancy), the data in Figure 3
(LHR) of 80 kW m1 at a fuel centerline tempera- exhibit differences of a factor of 50 between the lowest
ture of 820 C with a forced-convection coolant tem- and highest lines. Table 2 provides additional details
perature of 60 C.7,8 The design limits set for the on hydrogen diffusion in the d-hydride. The two early
high-power TRIGA core are7,8 fuel temperatures of measurements were obtained using the classical
750 C at steady state and 1050 C during transients. method of absorption of hydrogen from H2 into a
disk or cylinder specimen of ZrHx. In this method,
3.12.1.4 Advanced Reactors diffusivities are determined from curves of weight
gain versus time or progression of the delta phase
Geologic disposal of reprocessed nuclear waste
into the metal. The latest measurement by Majer
requires reduction of both the radioactivity and the
et al.15 utilized nuclear magnetic resonance (NMR),
heat release due to the minor actinides. Hydride fuel
which detects the mean residence time (t) of a proton
containing 237Np, 241Am, and 243Am is a candidate
on an interstitial site in the lattice. This measurement
transmutation target for this purpose.9 U–Th–Zr
yields the activation energy for diffusion (Ed, last
hydride has been employed as a surrogate of fuel con-
column in Table 2). Related techniques yield the
taining the minor actinides.10 Section 3.12.2 reviews
mean jump distance (L), which corresponds to the
some properties of this hydride, and its irradiation
separation of tetrahedral interstitial sites in the face-
behavior is described in Section 3.12.3.
centered cubic (fcc) structure of the Zr atoms in
the crystal. The diagram of the structure of the
3.12.2 (U, Zr) Hydride Properties d-hydride crystal in Figure 4 shows that there are
eight tetrahedral interstitial sites in the unit cell,
3.12.2.1 Hydrogen (Ordinary) Diffusion
which consists of four zirconium atoms (eight cor-
Unlike oxide fuel,11–13 where diffusion of corrosion- ner atoms shared by eight contiguous unit cells and
product hydrogen in a-Zry (zircaloy) is the key trans- six face-centered atoms shared between two adjoin-
port property, the other two major phases (in the ing unit cells).
3e−4
2e−4
D (cm2 s-1)
1e−4
3e−5
0.6 0.7 0.8 0.9 1.0
103/ T (K-1)
Figure 2 Diffusivity of H in b-Zr.
318 Uranium–Zirconium Hydride Fuel
1e−4
1e−5
3e−7
1e−7
Majer et al.15
3e−8
1e−8
0.8 1.0 1.2 1.4 1.6 1.8
103/T (K-1)
Figure 3 Diffusivity of H in d-hydride.
Authors Year Methoda H/Zr range T range ( C) D0 (cm2 s1) ED/R (K)
Albrecht and Goode14 1960 Diff 1.56–1.86 500–750 6.0 102 17 600
Paetz and Lücke16 1971 Diff 1.50–1.70 650–800 2.5 101 8960
Majer et al.15 1.5 103
b
1994 NMR 1.58–1.86 330–700 7100b
a
Diff, rate of absorption of hydrogen by solid; NMR, nuclear magnetic resonance.
DH ¼ fL2 =6t
dependence of the preexponential factor closely fol- The solution to eqn [10] is18,19
lows the theoretical expectation:
x ¼ Ae TQ =T ½11
DH ¼ ½fL ð1 x=2ÞvexpðED =RT Þ
2
½9
The constant of integration A is determined from the
The quantity in the square brackets is the preexpo- specified average H/Zr ratio of the fuel:
nential factor D0. In the range of the experiments of ð
Majer et al.15 shown in Table 2, the activation energy 2 R
xaverage ¼ 2 rxðr Þdr ½12
for diffusion (divided by the gas constant) is a weak R 0
function of H/Zr ratio, increasing from 7100 K at
where R is the radius of the fuel pellet.
H/Zr ¼ 1.58 to 7800 K at H/Zr ¼ 1.86. This feature
Figure 5 shows the extent of hydrogen redistribu-
may be due to incursions of the e-hydride in the
tion in ZrH1.6 in a parabolic temperature distribution.
sample.
The solid curve is for a helium-bonded fuel element
and the dashed curve is for a fuel element with a
liquid metal (LM) in the fuel-cladding gap. Redistri-
3.12.2.2 Hydrogen Thermal Diffusivity
bution is extensive; the original H/Zr ratio of 1.6
The other transport property of ZrHx that affects fuel has diminished to 1.46–1.49 at the center of the
performance is the thermal diffusivity (or heat of pellet and grown to 1.74–1.78 at the periphery. This
transport). This property causes hydrogen in the redistribution has only a minor effect on the neu-
Zr matrix to move toward cold regions, even in tronic characteristics of hydride fuel.
the absence of a concentration gradient. The process
reaches steady state when the gradient due to con-
centration diffusion just balances that due to thermal 3.12.2.3 Thermodynamic Properties
diffusion. In cylindrical coordinates, this condition of The most important thermodynamic properties of
zero net flux is given by the Zr–H system are the phase diagram and the
dx TQ dT equilibrium H2 pressure as a function of H/Zr and
þ x ¼0 ½10 temperature. These two properties are contained
dr T 2 dr
in Figure 6, which shows isobars of the equilibrium
where T is the temperature (in K), x the H/Zr ratio, hydrogen pressure superimposed on the binary phase
r the radial position in the fuel pellet, and TQ is the diagram.20,21 At temperatures exceeding 800 C, the
heat of transport of H in ZrHx divided by the gas equilibrium H2 pressure over the hydride is suffi-
constant. The only known measure of this quantity is ciently large that accumulation of this gas in the
that due to Sommer and Dennison,17 who report plenum or hydriding of the cladding would occur,
TQ ¼ 640 K. accompanied by a reduction of the H/Zr ratio of
1.80
p = 9.6 ⫻ 10-3
1.75 LHR = 375 W cm-1
p = 4.7 ⫻ 10-3
T0 = 555 ⬚C (LM); 680 ⬚C (He)
1.70 Ts = 385 ⬚C (LM); 510 ⬚C (He)
1.65
H/Zr ratio
1.60
1.55
He
1.50
LM
1.45
1.40
0.0 0.2 0.4 0.6 0.8 1.0
Fractional radius
Figure 5 Hydrogen redistribution in ZrH1.6 – with helium or liquid metal in the fuel-cladding gap; T0 and TS are the
centerline and surface temperatures, respectively.
320 Uranium–Zirconium Hydride Fuel
1200
1100
K=1
1000
900 K=0
K = −1
800
Temperature (⬚C)
K = −2 δ
700
β β+δ
α+β K = −3
600
K = −4
α
500 K = −5
K = −6 ε
400 K = −7
K = −8
K = −9
300 K = −10
K = −11
K = −12
200
α+δ δ+ε
100
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
H/Zr atom ratio
Figure 6 H–Zr phase diagram with isobars of equilibrium H2 pressures superimposed. Adapted from Zuzek, E.;
Abriata, J. Bull. Alloy Phase Diagrams 1990, 11, 385; Wang, W.-E.; Olander, D. J. Am. Ceram. Soc. 1995, 78, 3323.
the fuel. If all released H2 migrated to the plenum, the hydride fuel on startup.19 In the parabolic tem-
the neutron-moderating capability of the fuel would perature distribution engendered by fission, thermal
be affected. expansion results in azimuthal tensile stresses at
the pellet periphery that exceed the fracture stress
(barely or quite a bit, depending on which value of
3.12.2.4 Thermal and Mechanical
the fracture stress in Table 3 is accurate). The com-
Properties
bination of hydrogen redistribution toward the cool
Table 3 summarizes the principal thermal and periphery of the pellet and expansion with increasing
mechanical properties of the delta hydride. H/Zr ratio results in significant compressive stresses,
As the uranium metal in the hydride has a con- called ‘hydrogen stresses,’ in the outer portion of
ductivity of 28 W m1 K1, the two-phase thermal the pellet. This phenomenon was first noted (but
conductivity is 20 W m1 K1. This is about 7 times not analyzed) in Gylfe et al.,1 where it was termed
that of UO2. ‘migration stress.’
To reduce the fuel centerline temperature in The total stress is the sum of the thermal and
hydride fuel, a low-melting, inert liquid metal (LM, hydrogen stresses. These two axial components
Pb–Bi–Sn) replaces helium as the gap filler. This and their sum are shown in Figure 7 for typical
replacement lowers the centerline temperature by LM-bonded fuel-pellet operating conditions. Axial
well over 100 C.26 stresses are shown because this component is larger
The combination of the density variation with than the total azimuthal stress. According to Table 3,
H/Zr ratio (second row of Table 3), thermal expan- the fracture stress lies somewhere between the
sion (first row of Table 3), and hydrogen redistribu- dotted lines in the figure. The thermal stress acting
tion due to the in-reactor temperature distribution alone is capable of cracking the pellet at the surface.
(Figure 5) results in peculiar mechanical behavior of However, the hydrogen stress is highly compressive
Uranium–Zirconium Hydride Fuel 321
Heat capacity (J mol1 K1) (25 þ 4.7x) þ (0.31 þ 2.01x)T/100 þ (1.9 þ 6.4x)/T2 105 25
a
a
300<T(K)<1000; x ¼ H/Zr.
Source: GA Technologies, Inc. Uranium–zirconium hydride TRIGA LEU fuel. In Research Reactor Core-Conversion Guidebook, Vol. 4,
Fuels, IAEA-TECDOC-643, Appendix I-7 (19).
600
400
200
0
Axial stress (MPa)
−200
−400 Hydrogen
Total
−600
Thermal
−800 Fracture stress
−1000
−1200
−1400
0.0 0.2 0.4 0.6 0.8 1.0
Fractional radius
Figure 7 Axial stresses in an liquid-metal-bonded hydride fuel rod. Reproduced from Olander, D.; Ng, M. J. Nucl. Mater.
2005, 346, 98.
at large radii and more than overcomes the tensile Figure 7 must take into account the possible dif-
thermal stress. The total axial stress is tensile out to a ferences in the time constants of the thermal- and
fractional radius of 0.8 and exceeds the range of the hydrogen-stress components.
fracture stresses over most of this interval.
However, the thermal and hydrogen stress com-
3.12.2.5 Chemical Properties
ponents have different time responses. The thermal
stress develops as soon as the steady-state tempera- Chemical interaction between zirconium hydride
ture distribution in the pellet is established. This and Zircaloy cladding is thermodynamically favored,
occurs in a couple of days over which reactor but the kinetics are unknown. Hence, designation
startup takes place. On the other hand, the kinetics of Zircaloy as the cladding metal for hydride fuel
of hydrogen redistribution, which is controlled by rods must be approached with great caution. First,
the diffusivity of hydrogen in the hydride, is slower. consider the worst case.
The transient hydrogen redistribution calculations The surface temperature of a helium-bonded
in Huang et al.18 show a strong dependence upon hydride fuel pellet operating at a peak LHR of
the LHR: at 30 kW m1, steady state is achieved 37 kW m1 with a 35-mm gap thickness is 500 C
in 75 h, but at 15 kW m1, 10 days are required. (Table 1 of Olander and Ng19). Taking into account
Interpretation of curves such as those shown in the redistribution shown in Figure 5 and the phase
322 Uranium–Zirconium Hydride Fuel
diagram of Figure 6, the fuel surface generates an 2. Avoidance of waterside intergranular stress-
equilibrium H2 pressure of 5 103 atm. This partial corrosion cracking (IGSCC) of the SS.
pressure is not sufficiently large to move hydrogen
The primary disadvantages of this approach include
along the gap to the plenum, but it can drive hydrogen
to the cladding inner surface where metallic zirconium 1. The absence of the soft Zr liner (in BWRs) for
essentially reduces p(H2) to zero. The rate of cladding alleviating cracking by pellet–cladding interaction
hydriding depends on the series resistances of (PCI).
H diffusion in the fuel and the cladding and the 2. The SS liner may also be susceptible to internal
kinetics of H2 formation at the fuel surface and H2 stress corrosion by fission products (FPs).
decomposition at the cladding inside diameter (ID). 3. Differential expansion between Zr and SS may
Means of avoiding hydride of the cladding by this cause debonding of the two.
process are discussed below.
Replacing helium in the gap with an LM prevents
hydrogen from reaching the cladding inner surface 3.12.2.5.2 SiC internal coating or sleeve
as long as the gap is open. However, closure of Silicon carbide is very resistant to chemical attack by
the fuel-cladding gap generates PCMI. Hard con- or permeation of hydrogen; neutronically, it is practi-
tact of the two solids may provide a mechanism cally inert and is stable against radiation damage to
to overcome the sluggishness of the hydrogen reac- very high neutron fluences. It is used very successfully
tions at the two surfaces. The conservative design in high-temperature gas-cooled reactor (HTGR) fuel
response to this situation is to avoid PCMI entirely as a hydrogen (tritium) permeation barrier.28
by making the initial gap large enough that the
combination of fuel swelling and cladding creep- 3.12.2.5.3 Glass-enamel coating
down does not close the gap in the lifetime of the A glass-enamel coating metal cladding, about
fuel. An initial cold, radial gap of 150 mm (com- 0.08 mm thick, has been successfully bonded to the
pared with 80 mm in conventional oxide fuel ele- inner surface of the Hastelloy cladding in SNAP
ments) is sufficient to prevent PCMI in BWR reactors and survived operation at temperatures up
hydride fuel to a burnup of 60 MWd kgU1. With to 700 C.29
LM-bonding, no fuel-temperature increase results
from the larger gap size. 3.12.2.5.4 Zirconia coatings
Other methods have been suggested to separate The solubility of H2 in ZrO2 is very low (several ppm
the fuel and Zircaloy cladding. Most of them involve atomic); consequently, the oxide should be very resis-
adding a hydrogen permeation barrier to the clad- tant to hydrogen permeation. From the point of view
ding’s inner surface. These include of fabrication, oxidizing the inner surface of Zircaloy
cladding is the simplest and certainly the least expen-
sive of all coating methods. However, its effectiveness
3.12.2.5.1 Stainless-steel liner depends on its resistance to cracking, which afflicts
The transition metals do not form stable hydrides, the waterside corrosion scale on the cladding outer
and the solubility of hydrogen in stainless steel (SS) is surface.
low (but not zero). To exploit this property, duplex Oxidizing the outer surface of the hydride fuel
cladding consisting of a Zircaloy outer cladding and pellets has been demonstrated (and patented30) and
an SS inner liner has been considered as a means of appears to provide a satisfactory hydrogen perme-
reducing hydrogen loss from the fuel. Analysis of the ation barrier up to 800 C. In addition, when the fuel-
hydrogen permeation process is given in Appendix cladding gap closes, the oxide layer between the
A. Using the permeability of hydrogen in SS reported hydride and Zircaloy prevents direct contact between
in Sommer and Dennison17 and LeClaire,27 years are the latter two solids.
required to reduce the average H/Zr ratio of the fuel
from 1.6 to 1.55. From a hydrogen-retention point of
view, the SS liner is very effective. 3.12.2.6 (U, Th, Zr) Hydride Properties
The advantages of the liner over cladding made
Figure 8 shows that the (U–Th–Zr) hydride consists
entirely of SS are
of three phases: U-metal, ThZr2Hx, and ZrHx.
1. Reduced parasitic neutron capture by the ele- As shown in Figure 9, the U–Th–Zr hydride
ments in the alloy can hold more hydrogen at temperatures above
Uranium–Zirconium Hydride Fuel 323
1173 K than that of the U–Zr hydride.31 This is due over the temperature range 300–950 K.33 The results
to the high thermodynamic stability of ThZr2Hx are shown in Figure 11. The thermal conductivity,
phase formed in the U–Th–Zr hydride. This finding k, was calculated from the relation between the
led to the concept of hydride fuel containing 237Np, measured thermal diffusivity, a, and the specific heat:
241
Am, and 243Am as a transmutation target.32 k ¼ a Cp r, with
Figure 10 shows the linear expansion of four
a ¼ ð1:11x 21:2Þ=T þ 2:29 102 þ
hydrided U–Th–Zr alloys measured by Konashi and
Yamawaki.10 The thermal expansion coefficient ð3:18 106 ðH=MÞ þ 7:59 105 ÞT ðcm2 s1 Þ
increases with the hydrogen-to-metal ratio.
The thermal diffusivities of the UTh4Zr10H18–27 r ¼ 8:4 2:99 102 ðH=MÞðg cm3 Þ
hydrides were measured using the laser-flash method
Cp ¼ 0:110 þ 6:87 104 T þ 6:36 103 ðH=MÞð Jg1 KÞ
where H/M is the hydrogen-to-metal ratio.
2 2
U:Th:Zr = 1:4:10
U:Th:Zr = 1:2:5
U:Th:Zr = 1:1:4
U:Th:Zr = 2:1:5
Pure Zr
1.5 1.5
H/(Th + Zr)
1 1
1173 K 105 Pa
0.5 0.5
8E + 03 6E + 04 1E + 05 1173 1073 973 873 773
H2 pressure (Pa) T (K)
Figure 9 Equilibrium hydrogen concentration in U–Th–Zr alloys under various hydrogen pressures at 1173 K (left figure)
and that at various temperatures under 105 Pa (right figure). Those of pure Zr are shown as lines without symbols being
attached. Reproduced from Yamamoto, T.; Kayano, T.; Suwarno, H.; Yamawaki, M. Sci. Rep. RITU 1995, A40, 13.
324 Uranium–Zirconium Hydride Fuel
0.9
Pure U
0.8 Pure Th
UTh4Zr10H27
0.7
U2ThZr6H13.3
Thermal expansion (%)
0.6 UTh2Zr6H15.2
UO2
0.5
UThZr49.5
0.4 Pure Zr
0.3
0.2
0.1
−0.1
250 350 450 550 650 750 850
Temperature (K)
Figure 10 Thermal expansion of U–Th–Zr–H. Reproduced from Konashi, K.; Yamawaki, M. In Characterization and Quality
Control of Nuclear Fuels; Ganguly, C., Jayaraj, R. N., Eds.; Allied Publishers: New Delhi, 2004; pp 92–106.
0.060
UTh4Zr10H24 UTh4Zr10H18
0.045
0.040
0.035
0.030
0.025
0.020
300 400 500 600 700 800 900
Temperature (K)
Figure 11 Thermal diffusivity of UTh4Zr10H18–27 (open symbols: increasing temperature; solid symbols: decreasing
temperature).
In this section, the properties of actinide hydrides Flotow et al.,39 and Bartscher.40 We have mainly
are discussed from the viewpoint of the development used the data given in these reviews and also some
of nuclear materials; phase-diagram, equilibrium other data recently reported.
hydrogen pressure, structure, and thermodynamic
properties. Several comprehensive reviews of the 3.12.2.7.1 Th–H system
properties of the actinide metal hydrides have been Pressure–composition–temperature (P–C–T) diagrams
published by Libowitz,36 Ward,37 Mueller et al.,38 for the Th–H system are shown in Figure 12 for
Uranium–Zirconium Hydride Fuel 325
1
673 623
1073 K
573
0.1
973
P(H2)(105Pa)
0.01
973
473
823
0.001
773
373
723 298
0.0001
0 0.5 1 1.5 2 2.5 3 3.5
H/Th
Figure 12 Pressure-composition isotherms for the Th–H system.
equation40: 700
logðCðat:%ÞÞ ¼ 2:966 1732=T ½13
Phase II includes two types of nonstoichiometric 600
dihydrides, that is, ThH2x and ThH2þx, which
have a body-centered tetragonal (bct) structure.37 The 500
a parameter decreases slightly with increasing hypos-
toichiometry, whereas the c parameter increases.40 400
0 0.5 1 1.5 2 2.5 3 3.5
Phase III is a higher hydride of composition Th4H15
(ThH3.75), with the structure of a complex body- H/Th
centered cubic (bcc) array with space group I-43d.37 Figure 13 Phase diagram of the Th–H system.
Figure 14 shows the crystal structures of ThH2 and
Th4H15. The lattice parameters for these phases are lnðpðH2 ÞÞ ¼ AðT ðKÞÞ1 þ B ½15
listed in Table 4.
The P–C–T diagrams can be interpreted to refer yielding the enthalpy of reaction (RA, where R
to the general equilibrium: is the universal gas constant) and entropy of
reaction (RB).
2 2
ThHs þ H2 ¼ ThH2x ½14 The thermodynamic properties of the compounds
2s x 2s x ThH2 and Th4H5 have been studied experimen-
where ThHs denotes hydrogen-saturated uranium tally using calorimetric techniques up to 350 K. From
metal. The equilibrium pressures of this reaction these measurements, the heat capacity (Cp0 ) and the
can be represented by a linear equation: absolute entropy (S0) were derived.39 The heat capacity
326 Uranium–Zirconium Hydride Fuel
a
Phase IV in Figure 17.
b
Phase III in Figure 17.
ThH2 36.71 0.07 50.73 0.10 145.09 1.00 8.3850 þ 0.17509(T (K)) 8.9285 105(T (K))2
þ 7.358 104(T (K))2
ThH3.75 51.32 0.10 54.42 0.10 215.80 4.40
UH3 49.29 0.10 63.67 0.13 126.99 0.12 15.4897 þ 0.2728(T (K)) 2.0120 104(T (K))2
þ 5.2770 108(T (K))3
NpH2 47.28 66.11
PuH2 43.04 72.84 0.60 170.8 2.9 47.3070.1763(T (K)) þ 7.231 104(T (K))2
5.845 107(T (K))3
Uranium–Zirconium Hydride Fuel 327
100 100
10 795 K
10
741
701
1
643
609 1
741 K
P(H2)(105 Pa)
0.1 574
701
P(H2)(105 Pa)
513
0.01 0.1
643
473
609
0.001
423 0.01 574
0.0001
373 513
0.001
10-5 473
0 0.5 1 1.5 2 2.5 3
H/Pa 423
III
UD3 from 250 to 350 K. The enthalpy of formation of
600
UH3 recommended by Flotow et al. is based on the
direct reaction calorimetric measurement by Abraham
500
and Flotow.47 Uncertainties of the values in Table 8
are at the 95% confidence limit.
400
(a) (b)
Figure 19 Crystal structure of (a) a-UH3 and (b) b-UH3. Reproduced from Taylor et al.41 Acta Mater. 2010, 58(3), 1045–1055.
200 H/U
0.25 1.0 3.0
860 ⬚C
100 830 800 1200
UH3
770 1120
740
1000 (γ-U)
50 700
P(H2)(105 Pa)
680
30 800 770
T(⬚C)
10 400
570
200
5
0
3 0 10 20 30 40 50 60 70
0 0.5 1 1.5 2 2.5 3 at.% H
U
H/U
Figure 21 Phase diagram of the U–H system.
Figure 20 Pressure–composition isotherms for the U–H
system.
100
310
10
255
1
350
252
(0.2945) 628 ⬚C
P(H2)(105 Pa)
249
(0.00124) 476 ⬚C 148
310
0.01
(0.00034) 425 ⬚C
0.001
(0.00053) 364 ⬚C
0.0001
0 1 2 3
H/Np
Figure 22 Pressure–composition isotherms for the Np–H system. (Open circles denote hydriding; filled circles denote
dehydriding. Squares connected by dashed lines are temperature ramp experiments.)
400 (x ~
= 0.15)
PuH2þy (Figure 26) with an extended solid solution is
formed, finally leading to the hexagonal phase at com-
300 280 Cubic position approaching PuH3. A rather complex phase
+hex. relationship was obtained by the rapid reaction at high
Np(α) + NpH2+x
temperature and for high hydrogen pressure. Thevalues
200
Hex. of DH0f of PuH2þy were obtained by the solution calori-
metric study (Table 10).40 The composition depen-
dence of the DH0f can be described by the equation:
100
0 0.5 1 1.5 2 2.5 3 DHf0 ðPuH2þy ; 298Þðk J mol1 Þ
H/Np
Figure 23 Phase diagram of the Np–H system.
¼ 78:80 45:98H=Pu ½20
This yields for the enthalpy of formation of ‘PuH2’:
(phase I).49,50 The two-phase region, phase II consists
DHf0 ¼ 170:8 k J mol1 ½21
of the saturated metal and PuH2x. Hydrogen dis-
sociation pressures for two-phase regions up to Further thermodynamic studies are very limited.
H/Pu ¼ 2 are shown in Figure 25.51 A fcc dihydride, Oetting et al.52 measured the heat capacity of 242PuH1.9
Uranium–Zirconium Hydride Fuel 331
from 8 to 600 K. The results show a complex behavior 3.12.2.7.6 Am–H system
with numerous anomalies. A l-type transition at The P–T–C data for the Am–H system are plotted
36 K is associated with antiferromagnetic ordering. in Figure 27.40 The phase diagram is shown in
In addition, a significant excess heat capacity associ- Figure 28.40 Hydrogen dissolves in a-americium
ated with the crystal field splitting of the electronic and b-americium. By adding more hydrogen, the
ground state of the Pu3þ ions. The resulting thermody- fcc-AmH2x is formed, which coexists with the
namic data for ‘PuH2’ are hydrogen-saturated metal. The lower composition
limit of the subsequent americium dihydride phase
S 0 ð298:15KÞ ¼ ð72:84 0:60Þ J K1 mol1 ½22
is very close to H/Am ¼ 2. Hydrogen dissolves in the
dihydride phase until the composition of about
AmH2.7 is reached. When more hydrogen is added,
Cp ðT Þ ¼ 47:307 0:1763ðT ðKÞÞ þ 7:231 104 ðT ðKÞÞ2
the hexagonal AmH3 appears.
5:845 107 ðT ðKÞÞ3 ½23 The value of DH0f (298 K) of AmH2.0 was calculated
from the P–C–T data to be 183.7 14.6 kJ mol1
800 and that of DS0f (298 K) was 48.1 3.8 J K1 mol1.40
L
600
Curium dihydrides have been prepared and character-
ized by Bansal and Damien,53 and later by Gibson and
T (⬚C)
PuH3
200
sial reports regarding the stability of trivalent curium
hydride. Bansal and Damien53 have not detected any
0 sign of the hexagonal CmH3 phase, while Gibson and
0 20 40 60 80 100 Haire54 have reported observing hexagonal CmH3 in
at.% their experiments. The thermodynamic stability of
Figure 24 Phase diagram of the Pu–H system. CmH3 is investigated by using the density functional
1000
1
800 ⬚C
750
100
700
0.1
P(H2)(102 Pa)
650
P(H2)(105 Pa)
10 600
0.01
1
0.001 500
0.2
0.0002
0 0.4 0.8 1.2 1.6 2.0
H/Pu
Figure 25 Pressure–composition isotherms for the Pu–H system.
332 Uranium–Zirconium Hydride Fuel
theory approach.55 It appears that CmH3 is unstable of CmH2.0x was calculated from the P–C–T data to
with respect to metallic curium and gas-phase hydro- be 187 14 kJ mol1 and that of DS0f (770 K) was
gen. This conclusion is consistent with the earlier 142 10 J K1 mol1.
experimental result where attempts to synthesize
CmH3 were unsuccessful. The value of DH0f (770 K) 3.12.2.7.8 Th–Zr–H system
The ThZr2–H system has been characterized by equi-
librium pressure measurements shown in Figure 29
and X-ray diffraction analysis.56 At temperatures
above 1180 K, hydrogen dissolves in the bcc metal
up to H/M ¼ 2 (M ¼ Th þ Zr). A two-phase region
observed for 0.2 < H/M < 0.6 is followed by the
formation of a cubic Laves-type (Fd3m, C15) hy-
dride across the composition range 0.6 < H/
M < 2.2. The enthalpy and entropy of hydriding
ThZr2 to form ThZr2H6 at 1250 K are 413 25 kJ
mol1 and 334 20 J K1 mol1, respectively.
The heats of formation of ThZr2Hx were evalu-
ated by using two models.57 The first correlation
is the well-known Miedema’s empirical model. The
Figure 26 Crystal structure of PuH2.
second one is the Griessen–Driessen’s model, which
is a semiempirical band structure model, in which
each metal is characterized by the difference between
Table 10 Enthalpy and entropy of formation of PuH2þy the Fermi energy and the energy of the lowest
H/Pu DH0f (298) (kJ mol1) conduction band of the host metal. The evaluated
results were compared with the experimental data
1.99 170.8 2.9 in Figure 30. It is shown that the Griessen–Driessen
2.21 179.6 2.1
model predicts the heat of formation more precisely
2.61 199.6 2.9
than the Miedema model does.
1
879 ⬚C
827 ⬚C
0.1
775 ⬚C
375 ⬚C
730 ⬚C
574 ⬚C
475 ⬚C
0.01
678 ⬚C
P(H2)(105 Pa)
275 ⬚C
125 ⬚C
630 ⬚C
0.001
574 ⬚C
0.1
528 ⬚C
0.01 475 ⬚C
3.12.2.7.9 Np–Zr–H, Am–Zr–H, and Pu–Zr–H entropy change into consideration, the phase diagram
systems of the Np–Zr–H system was drawn (Figure 32).
No experimental data on these ternary hydrides Extensive investigations have been performed on
are available. The values of the heats of formation hydrides of actinide elements with nuclear interest.
were calculated by the method described in Section The studies of the actinide hydrides were mainly
3.12.2.1.58 The results for various intermetallic forms focused on U hydrides and Th hydrides. Experimen-
are listed in Table 11, that is, A5BHx, A2BHx, ABHx, tal data are limited for other transuranium hydrides
AB2Hx, AB3Hx, and AB5Hx. Figure 31 shows the due to the difficulty of handling radioactive materials.
enthalpies per hydrogen atom. These values need to Especially, only a few papers were published on Am
be multiplied by the number of x for conversion to the hydrides and Cm hydrides.
enthalpies in the unit of kJ mol1. By taking the Another reason for less accumulation of experi-
mental data is a lack of material of minor actinides.
First principles calculation is a useful tool to under-
stand and predict the behavior of actinide hydrides.
1200
Advances in density functional theory are making it
possible to treat relativistic effects of heavy elements
AmHs(β) AmH3 and strongly correlated electron systems.
1100
AmHs(β) + AmH2-x
1000 3.12.3 Irradiation Effects
AmH2+x
T(K)
900 AmH2.7 In common with all nuclear fuels, the key irradiation
AmHs(α) + properties are fuel swelling, fission gas release, and
AmH3
fission-product chemistry. These are functions of
800 temperature (or LHR) and burnup.
AmHs(α) + AmH2-x
700
3.12.3.1 Burnup Units
600
The literature on hydride fuels expresses burnup either
0 1 2 3 as F, the number of fissions per unit volume, or FIMA,
H/Am ‘fissions per initial metal atom’ (%FIMA is equivalent
Figure 28 Phase diagram of the Am–H system. to the burnup unit ‘metal atom %’ that is commonly
1
1321 K
1300 K
1253 K
0.1
1073 K
P(H2)(105 Pa)
1183 K
973 K
0.01
873 K
The values of enthalpy are for x ¼ 1. When x ¼ 1.8, 4, or 6, the values need to be multiplied by the number of x and the unit in kJ mol1.
-20
A5BHx
A2BHx
-30
ABHx
Enthalpy of formation (kJ mol-1 H)
AB2Hx
-40
AB3Hx
Np-Zr-Hx
AB5Hx
-50
-60 U-Zr-Hx
-70 Pu-Zr-Hx
Am-Zr-Hx
-80
Th-Zr-Hx
-90
Hydrides
Figure 31 Heats of formation for hydrides of common intermetallic-type actinides.
Uranium–Zirconium Hydride Fuel 335
H
1.0
0.9
0.8
0.7
CaF2 _C1 + bcc_A1
Mole_fraction H
+FCT_L⬘12
0.6
0.5
CaF2 _C1 + bcc_A
0.4 +fcc_A1
0.3
0.2
CaF2 C1
0.1
bcc_A2#(β-N p)
0 0.2 0.4 0.6 0.8 1
Np
CaF2 _C1 + bcc_A2#
M ole_fraction ZR Zr
+bcc_A2
Conversion from FIMA to BU is accomplished with Thus, 1% FIMA in this particular hydride fuel is
3:7 10 19 equivalent (in terms of energy production per unit
BU ¼ NZr ð1 þ yÞ FIMA volume) to a burnup of 15 MWd kgU1 in UO2.
rU
1 wU
¼ 2:4 10 3
ð1 þ yÞ FIMA ½26 3.12.3.2 Fuel Swelling due to Void
wU
Formation
As shown in Table 1, the uranium density of the
hydride fuel is substantially lower than that of oxide Very early in life, oxide fuel shrinks (densifies)
fuel. If oxide and hydride fuels are to be compared on because of destruction of fabrication porosity by
the basis of energy output per unit fuel volume (which fission fragments. The early-time volume change of
is proportional to F ), the uranium density should be hydride fuel is either nonexistent (T < 700 C) or
removed from eqn [25]. A more appropriate definition exhibits large increases (T > 700 C) termed ‘offset
for this purpose is the ‘equivalent oxide’ burnup: swelling.’4 Following these initial swelling/densifica-
tion stages, both fuels continue to swell linearly with
r0U
BUequivalent oxide ¼ BU ¼ 4:0 1020 F ½27 burnup as a result of introduction of FPs into the
9:3
matrix.
where 9.3 is the density of uranium in UO2. Fuels with In the SNAP program,4 swelling of hydride fuel
the same equivalent-oxide burnup have generated the was measured both by in-reactor and by quite sophis-
same amount of energy per unit volume. ticated irradiation tests. In normal irradiation tests,
Because FIMA is commonly used for hydride fuels, the effects of temperature and fission rate cannot be
conversion to the equivalent-oxide burnup is required. separated because the latter controls the former. The
This is accomplished with the aid of eqn [26]: SNAP program utilized a test rig in which cooling of
0 the in-reactor specimens was accomplished by a gas
r0U r 1 wU
BUequivalent oxide ¼BU ¼ 2:4 103 U flowing along the exterior of the specimen holder.
9:3 9:3 wU
Without changing the fission rate, the specimen tem-
ð1 þ yÞ FIMA ½28a perature was controlled by using different mixtures
For the hydride with 45 wt% U considered here, of helium and xenon in the coolant gas, thereby
eqn [28] reduces to enabling examination of the effect of each variable
separately. Swelling was determined by density mea-
BUequivalent oxide ¼ 1:5 103 FIMA ½28b surements of the irradiated specimens.
336 Uranium–Zirconium Hydride Fuel
Microscopic examination of swollen hydride fuel it is 5%, but it vanishes at 650 C. In these plots,
revealed large cavities adjacent to uranium particles. the offset swelling saturates at a burnup of 103
These cavities are believed to be true voids, not FIMA; however, no experimental evidence was
fission-gas bubbles. Voids are generated in the hydride offered to support this supposition.
matrix close to the uranium particle by the radia- The top graph in Figure 34 shows unusual
tion damage created by escaping fission fragments. changes in offset swelling with the fission rate or
Figure 33 is a transmission electron micrograph of the LHR. The decrease in offset swelling with this
these voids.4 The uranium particles are not imaged in variable is attributed to behavior of voids as the
this transmission electron microscope (TEM) but primary source of the volume change.
were reported as 5 mm in diameter and 40 mm The data underlying the plots in Figure 34 are
apart. The temperature of the fuel at the location of shown in Figure 35. Unfortunately, the burnup rates
the TEM was estimated at 785 C and the burnup for each data point are not given in Lillie et al.4 The
was 2 103 FIMA, which is about twice the lines shown in the graph were estimated from offset
saturation burnup for offset swelling. The void-size swelling values taken from Figure 34. They fit the
distribution was roughly bell-shaped, ranging from equations:
the microscope’s resolution limit of 10 nm up to a logð% DV =V Þvoid ¼ A Bð103 =T Þ ½29
maximum diameter of 120 nm. The average diameter
(not defined) was about 60 nm. The number density of The coefficients A and B depend on the burnup rate
these voids, averaged over an area comprising numer- according to
ous uranium particles, was 5 107 nm3. From A ¼ 9:2 þ 2:2 103 ðFIMA per yearÞ
these figures, the overall volume swelling should be
3
B ¼ 8:7 þ 2:4 103 ðFIMA per yearÞ ½30
DV pD p 603
¼ Nvoid ¼ 5 107 ¼ 0:06 Comparison of the void production in irradiated
V void 6 6
hydride reactor fuel with the well-developed theories
The (DV/V) values reported in Lillie et al.4 of void swelling in metals provides some insight into
were 10% in the immediate vicinity of the uranium the behavior shown in Figure 34. In particular, the
particles and 4% averaged over a large cross-section features for comparison are the dependence of void
of the fuel. The swelling calculated by the above size/swelling on temperature and burnup rate and
equation lies between these two observed values. the conditions that lead to saturation of void growth.
Figure 34 shows that offset swelling can be quite
large and highly temperature-sensitive: at 760 C, 3.12.3.2.1 Temperature dependence
Figure 34 shows a sharply increasing offset swelling
with increasing temperature. The typical bell-shaped
1m curves of void-growth rate versus temperature in
metals (e.g., Figure 40.36 of Olander59) exhibit similar
behavior as the onset of offset swelling in hydride
fuel. However, the temperature effect observed in
hydride fuel is on the magnitude of offset swelling,
while the temperature effect seen in metals refers to
the rate of void swelling.
7
Burnup rate, 10−3 FIMA per year
6
0.9
5
2
760 ⬚C
0
0.0 0.1 0.2 0.3 0.4 0.5 0.6
Burnup (100 ⫻ FIMA)
7
6 760 ⬚C
5
Burnup rate = 0.9 ⫻ 10− 3 FIMA per year
Swelling (%)
3
705 ⬚C
650 ⬚C
1
0
0.0 0.1 0.2 0.3 0.4 0.5 0.6
Burnup (100 ⫻ FIMA)
Figure 34 Swelling behavior of (U,Zr)Hx Space Nuclear Auxiliary Power fuel.
20
SNAP reactor
Irradiation test
10
0.9 ⫻ 10−3 FIMA per year
1.8 ⫻ 10−3 ’’
0.5
0.2
0.92 0.94 0.96 0.98 1.00 1.02 1.04 1.06 1.08
103/T(K-1)
Figure 35 Space Nuclear Auxiliary Power fuel-offset void-swelling data. From Lillie, A.; et al. Zirconium Hydride Fuel
Element Performance Characteristics; Atomics International/USAEC Report AI-AEC-13084; 1973.
by fission in the solid lattice and bubble destruction by into account in Huang et al.61 Below 750 C, these
fission-fragment re-solution. Although this model is volatile elements are retained in the fuel (see Section
advanced for cavities containing fission gas, there 3.12.3.4). The rare gases may exist as nanobubbles that
is no reason that it could not be applied to voids were not observed in the SNAP studies.
in hydride fuel. (Lillie et al.4 assume that the cavities
observed in irradiated hydride fuel were true voids, 3.12.3.4 Fission-Gas Release
but the possibility that the cavities contained fission
A major difference between oxide and hydride fuels
gas was not investigated.)
involves the mechanisms responsible for the release
of fission gas during irradiation. This phenomenon
3.12.3.3 Fuel Swelling due to FPs has been exhaustively studied for uranium dioxide
fuel, and there is enough empirical evidence to sup-
For UO2, fuel dimensional changes during irradiation port the predictions of analytical and computational
are due principally to the ingrowth of solid FPs, with models. However, the data on fission-gas release from
smaller contributions due to fission-gas bubbles and hydride fuel lack robustness and careful validation.
early-on, to removal of as-fabricated porosity (densi- What data are available suggest that recoil is the only
fication). The solid fission-product swelling rate is mechanism of fission-gas release up to temperatures
about 0.07% per MWd kgU1. in the neighborhood of 700 C.3 The release fraction
In hydride fuel, the fission-product-swelling rate due to recoil is of the order of 104, supporting the
is represented by the slope of the lines in Figure 34 assumption that fission gas is not released from
after cessation of offset swelling. Following this period, hydride fuel. The justification for this statement is
hydride fuel swells at a constant rate of 3% per presented below.
%FIMA. Converting from FIMA to equivalent-oxide Many ways are available for quantitative measure-
burnup gives a swelling rate of 3/1.5 103 ¼ 0.002, or ment of fission-gas release from irradiated fuel.
DV =V ¼ 0:2% per BUequivalent oxide In-reactor tests include measurement of
On this basis, the fission-product swelling of hydride 1. The buildup of pressure within an instrumented
fuel is 3 times larger than that of oxide fuel. It is also test fuel rod resulting from the release of the stable
twice as large as the value computed for the contribu- Xe isotopes (132Xe, 134Xe, and 136Xe) and 85Kr.
tion of solid FPs by Huang et al.61 A good part of the The released gas accumulates in the plenum and
reason for this significant difference is the presence of the fuel-cladding gap. The fraction released
Xe þ Kr, Te, and I in the fuel, which were not taken increases with time.
Uranium–Zirconium Hydride Fuel 339
2. The rate of release of a short-half-life isotope same model and its parameters can be used to accu-
(85mKr or 131Xe) by passing a carrier gas through rately predict release under power reactor condi-
a test fuel rod or fuel specimen. The released tions, an expectation that is rarely met.
radioisotopes are trapped on refrigerated charcoal All fission-gas release theories, from the Booth
and then measured by g-spectroscopy. Steady- model on, were developed for application to UO2
state release is achieved and the ratio of the release fuel. Because the microstructure of U–Zr hydride in
rate (R) to the birth rate (B) (i.e., the production no way resembles that of UO2, with one exception, the
rate in the fuel) is determined. This R/B ratio is models are useless for understanding release from the
often (erroneously) called the fraction released. former fuel. The exception is the mechanism by which
The postirradiation test involves fission fragments born within 10 mm of an exposed
surface escape from the solid. This so-called recoil
3. Annealing the specimen in a furnace through mechanism produces a release rate-to-birth rate ratio
which a carrier gas flows. Typically, 133Xe g-activity equal to 1/4(S/V )m, where S/V is the surface-to-
is used to measure the quantity released. For such volume ratio of the fuel, m is the range of fission
an experiment, the fraction released is the ratio of fragments in the fuel, and is the fraction of the escap-
the amount of nuclide released in a certain time ing fission fragments that are not reimplanted in the
divided by the quantity of the isotope contained in opposite surface. Typical values of recoil R/B are 104.
the specimen following prior irradiation. There are numerous reports of FP release from
Interpretation of these experiments ranges from U–Zr hydride (TRIGA) fuel.3,62,63 Unfortunately,
the very simplest Booth model to partially mechanis- none of them are the original sources of the data
tic models included in large fuel-behavior codes. The they report. A compendium of these second-hand
Booth model is based on diffusion from the interior of reports is shown in Figure 36. This plot appears in
the grains of the fuel to the grain boundaries, whence all reports of fission-gas release from hydride fuel,
release occurs (see Section 15.4 of Olander59). The with the latest experimental results added to it. Most
basic properties that govern the interpretation of all of the results come from postirradiation anneal
three of the tests described above are the diffusivity experiments, with a few experiments of the in-reactor
of the isotope in the grains (D) and the radius of the type 2 described above. None of the data originated
grains (a), which appear as the combination a2/D. from in-reactor fuel rods (method 1 above), the claim
This parameter is the only one needed in the Booth being that no fission gas isotopes could be detected in
model and appears in different forms in the quantita- postirradiation examination. How seriously to take
tive interpretation of the tests. Whatever model is this claim is uncertain, since the temperature of the
used to analyze test data, the expectation is that the fuel was poorly known or unknown.
1e − 1
Postirradiation anneal-fraction release (Xe)
In-pile-release-to-birth rate (85mKr)
Diffusional release
1e − 2 Recoil release
Release fraction
1e − 3
1e − 4
1e − 5
0 200 400 600 800 1000 1200
Irradiation temperature (⬚C)
Figure 36 Fission-gas release from U–Zr hydride fuel. Reproduced from Olander, D.; Ng, M. J. Nucl. Mater. 2005, 346, 98.
340 Uranium–Zirconium Hydride Fuel
The dashed horizontal line in Figure 36 is attrib- 3.12.3.5 Chemical and Physical States of
uted (by the cited references) to recoil release, which, the FPs
given the magnitude of the fraction release and its
The chemical and physical states of the FPs in irra-
temperature independence up to 600–800 C, is a rea-
diated U/Zr hydride fuel has been studied by Huang
sonable interpretation. The curve through the remain-
et al.61 using the thermochemical code Thermo-Calc.
ing data points is not model based. The only claim is
Mixtures of the elements U, Zr, H, O, and a variety
that it is due to some undefined ‘diffusion’ process.
of FPs were analyzed for the equilibrium composi-
Two problems are immediately evident from
tion for element ratios simulating burnup in reactor.
Figure 36.
In addition to the chemical states, Thermo-Calc also
First, release fractions measured by postirradia-
predicts the phases present at equilibrium. Oxygen
tion anneal tests are reported as a function of tem-
was assumed to be present as an impurity (in the form
perature only. However, the release fraction from
of ZrO2) at a concentration of 1000 wt ppm. Unfortu-
this type of test is a function of time as well as
nately, the U/Zr and H/Zr ratios were not specified,
temperature. (Justification for this omission is based
but they are taken here to be 0.31 (45 wt%) and 1.6,
on the claim that the postirradiation anneal data in
respectively.
Figure 33 apply to fuel that has been irradiated long
Table 12 summarizes the states of FPs in irradiated
enough for all fission-product activity to achieve
UO2 and (U0.31Zr)H1.6O0.01. The former are taken
equilibrium, when (it is claimed), the release fraction
from Olander59 and the latter from Huang et al.61
is a maximum.3 This does not obviate the need to
report the length of time of the anneal.)
Second, the low-temperature data points in
3.12.3.6 Irradiation Behavior of (U, Th, Zr)
Figure 36 were reported to have been determined by
Hydride
the postirradiation technique. However, recoil release
can only be detected while fission is occurring. Irradiation tests of the UTh4Zr10H20 fuels were con-
Despite the rudimentary state of understanding of ducted in Japan Materials-Testing Reactor (JMTR).64
the mechanism(s) of fission-gas release from hydride The test apparatus and irradiation conditions are
fuel, the available data strongly suggest that if fuel summarized in Table 13 and a section of the capsule
temperatures are kept below 650 C during opera- is shown in Figure 37. After irradiation, nondestruc-
tion, fission-gas release is negligible. In Shuffler et al.,6 tive examinations were performed by visual inspec-
the design limit for the centerline temperature at tion, profilemetry, X-ray photography, and axial
steady state is 750 C. In view of the observed swelling g-scanning. Visual inspection and X-ray radiography
behavior, this value may be somewhat aggressive. It is did not reveal any irregularities. g-Scanning showed
essential that irradiation tests of hydride fuel be con- FPs in the fuel pellet region. No released fission gas
ducted in order to verify the behavior shown in was detected by cladding puncturing the capsules.
Figure 34. This may be due to the low burnup (equivalent
Table 12 Probable chemical and physical states of fission products in oxide and hydride fuels
a
The element in parentheses is closely related chemically and physically to the first element.
b
Dissolved in UO2.
c
RE ¼ rare earths (La, Ce, Pr, Nd, Sm, Pm, Eu, Gd).
d
NM ¼ noble metals (Pd, Ru, Rh, Tc, Pt).
Uranium–Zirconium Hydride Fuel 341
a
FIMA ¼ F/(NTh þ NU), F, number of fissions per unit volume; NTh and NU, atomic densities of Th and U in fuel.
12.0
50
110
166 (mm)
Figure 37 Irradiation capsule used in Japan Materials-Testing Reactor tests. Reproduced from Konashi, K.; et al.
Irradiation of hydride fuels in JMTR. In Proceedings of the International Congress on Advances in Nuclear Power Plants
(ICAPP 2007), Nice, France, May 13–18, 2007.
oxide burnup 1 MWd kgU1 for 98F–10U and 5 the X-ray diffraction response (Figure 40). The
MWd kgU1 for 99F–11U). Two distinct regions X-ray diffraction pattern of the outer region of
in the cross-sections of UTh4Zr10H20 pellets were the irradiated pellets is clearly flattened compared
observed: an outer dark region and an inner bright with the inner region and the unirradiated pellets.
region (Figure 38). Cracks were also observed in the Both of the latter show sharp diffraction pattern,
pellets, although these might have occurred during which suggests that the outer layer at the pellet has
sample preparation for postirradiation examination. been amorphized. The microhardness of the unir-
Alternatively, the cracks might have been initiated radiated pellet is constant along the pellet radius,
from microcracks in the as-fabricated pellet. but after irradiation, the microhardness increases in
During irradiation, a radial temperature gradient the inner region of the pellet and decreases in the
is built up in the fuel pellet. As a result, hydrogen outer region.
migrates to the regions of lower temperature, causing The thermal diffusivities of the irradiated pellets
a considerable increase of hydrogen content near (by the laser-flash method) are compared with that
the pellet periphery. In destructive examination of of the unirradiated pellet in Figure 41. There does
irradiated pellet, hydrogen migration was found to not appear to be any effect of irradiation on this
have affected the microhardness (Figure 39) and property.
342 Uranium–Zirconium Hydride Fuel
f
2RE þ 32 O2 ¼ ðREÞ2 O3 DGðREÞ ¼ 1674kJ mol1 ½31b
2 O3
3.12.4 In-Reactor Chemical Behavior Subtracting eqn [31b] from twice eqn [31a] gives
reaction [III], so the standard free energy change of
3.12.4.1 Impurity Oxygen
the latter is
The small quantity of impurity oxygen in the fresh
hydride fuel is present as ZrO2. The oxides of the
0
DGðIIIÞ ¼ 2DGðREÞH
f
2
DGðREÞ
f
2 O3
¼ 1456 kJ mol1
rare-earth elements and of Ba, Sr are more stable
and the law of mass action gives the oxygen pressure
than ZrO2, so as the FPs grow in, the following
reaction takes place: h i2=3
pO2 ¼ expð1456 103 =8:3145 800Þ pH 2
2
2RE þ 1:5ZrO2 þ 1:5xH ! RE2 O3 þ 1:5ZrHx ½I
¼ 2 1067 atm
2ðBa; SrÞ þ ZrO2 þ xH ! 2ðBa; SrÞO þ ZrHx ½II
For a typical hydrogen pressure generated by the
The ‘H’ in these reactions is taken from the hydride fuel (3.5 103 atm), the oxygen potential is 1025
in order to combine with the ‘liberated’ Zr, thus kcal mol1. Zirconium does not oxidize at this
reducing the value of x. The details of this process oxygen pressure.
Uranium–Zirconium Hydride Fuel 343
ThZr2Hx
(333)
e-ZrH1.74 No irra.
(220) Center
Inner
α-U ε-ZrH1.74 Outer
(022)
δ-ZrH1.6
ThHx
Intensity
(220) ThZr2Hx
ThHx ThZr2Hx α-U
Th02 ThZr2Hx
(422) (440)
Tho2 (311) Th02 (531)
(220) (222)
Zr
(102)
0.04
Thermal Diffusivity (cm−2 s−1)
0.03
Non-irrad.T:↑
Non-irrad. T:↓
Irrad. T:↑
Irrad. T:↓
0.02
0 100 200 300 400 500 600
Temperature (⬚C)
Figure 41 Thermal diffusivities of pellets (98F–10U). Reproduced from Konashi, K.; et al. Irradiation of hydride fuels in
JMTR. In Proceedings of the International Congress on Advances in Nuclear Power Plants (ICAPP 2007), Nice, France,
May 13–18, 2007.
3.12.4.3 Reduction of the H/Zr Ratio for 90% consumption of the original 235U, which
During Irradiation occurs at 22 months of operation.
This process may not be as serious as it first
After conversion of impurity ZrO2 to the FP oxides,
appears. The hydrogen removed from the ZrHx
all subsequent RE and Ba, Sr produced by fission
remains in the fuel as FP hydrides, so the total
extracts hydrogen from ZrHx to form the hydrides
hydrogen concentration is unchanged. Since the fuel
(RE)H2 and (Ba, Sr)H2. As shown in Appendix B, the
does not lose hydrogen, the neutronics of the system
result is a continual reduction in the H/Zr ratio of
are not affected by the switch of the metal atoms
the hydride at a rate of D(H/Zr)/D(FIMA) 0.1, or,
to which the hydrogen is attached. Moreover,
equivalently, a decrease of H/Zr from 1.586 to 1.584
344 Uranium–Zirconium Hydride Fuel
the fission-product hydrides are more stable thermo- mechanistic point of view. One of two steps can
chemically than ZrHx, so at a given temperature, the control the kinetics of this reaction. The first is diffu-
equilibrium H2 pressure over a mixture of ZrHx and sion of hydrogen (as interstitial H atoms) in the
FPH2 is lower than that over pure ZrHxþy, where y zirconium hydride matrix to the free surface. The
depends on the burnup, as indicated above. The second is recombination of H atoms at the surface
lower H2 pressure reduces the risk of overpressuriza- to form H2, which then escapes to the gas phase. For
tion of the rod in the event of a transient that sub- either rate-limiting step, the rate should have
stantially increases fuel temperature. decreased with time, with a consequent upward cur-
vature of this plot. The constant dehydriding rate
implied by the straight line in the top graph implies
3.12.4.4 Hydrogen Loss to the Gas Phase
zero order for the above reaction (i.e., the rate is
During normal operation, loss of hydrogen from independent of H concentration in the hydride).
ZrH1.6 to the gas phase in the fuel element is not These experiments need to be repeated in a typical
significant, owing to the low equilibrium H2 pressure in-reactor atmosphere (20 atm He) in order to ver-
at fuel temperature. However, in accident conditions ify this strange behavior.
accompanied by high temperature excursions and Figure 43 is a conventional Arrhenius plot of the
cladding failure, rapid hydrogen release may consti- dehydriding rates. The line is represented by
tute a unique hazard. Release from the fuel occurs
34 000
when the equilibrium hydrogen pressure of the fuel Rate ¼ 8 1011 exp mmol H per cm2 s1
T
(from Figure 6, modified for dilution by soluble
fission-product hydrides) exceeds that in the gap or The number in the exponential term corresponds to
plenum gas. The dehydriding reaction is an activation energy of 280 60 kJ mol1.
The significance of these results can be appre-
ZrHx ! ZrHy þ 1=2ðx yÞH2 ½IV
ciated from the following example in a fuel rod with
In order to determine the kinetics of this reaction, a breached cladding. Suppose a 1-cm diameter, 1-cm
specimen of ZrHx was suspended from a microbal- long pellet of U0.31ZrH1.6 is at a uniform temperature
ance in a vacuum furnace and the mass loss recorded of 700 C. The 80 vol% hydride in this fuel has a
as a function of time.65 Figure 42 shows a typical density of 5.64 g cm3. The pellet contains 0.0475 mol
result. The most significant aspect of this plot is of Zr and 0.076 mol of H. From Figure 43, the H flux
its linearity. This is wholly unexpected from a at this temperature is 3 107 mol H per cm2 s1,
1.75
T = 692 ⬚C
1.70
1.65
H/Zr ratio
1.60
1.55
1.50
1.45
−2
Pressure experiments–Zirconium
−3 Thermogravimetry–Zirconium
Thermogravimetry–Zircaloy
−4
−6
−7
−8
−9
−10
−11
0.80 0.85 0.90 0.95 1.00 1.05 1.10 1.15
1000/ T(K-1)
Figure 43 Dehydriding kinetics of ZrHx; Arrhenius plot of rate data. Reproduced from Terrani, K.; et al. J. Nucl. Mater.,
submitted.
so the rate of H loss from the sides of the pellet 3.12.5 Comparison of LM-Bonded
exposed to the open gap is 106 mol H per second. Hydride Fuel and Oxide Fuel in LWRs
In 1 h, the pellet loses 3.6 103 mol H, which cor-
responds to a reduction in the H/Zr ratio from 1.6 to Estimated temperatures are shown in Table 14 for
1.52. If continued, this rate of H loss may lead to a three pressurized-water reactor (PWR) fuel ele-
significant buildup of H2 in the ex-core environment ments: two hydride fuels, one with helium in the
and to increased reactivity of the H-depleted fuel with fuel-cladding gap and the other with a LM bond.
steam. In intact LWR fuel rods, on the other hand, For comparison, the comparable temperatures in a
the small plenum volume allows the H2 pressure to conventional oxide-fueled system are also given.
build up to its equilibrium value before significant This table illustrates some of the favorable fea-
H depletion of the fuel occurs (see Section 3.12.5). tures of hydride fuel relative to oxide fuel. The first is
the much lower fuel temperatures for the same LHR.
This is due to the large thermal conductivity of the
3.12.4.5 Stability of Hydride Fuel in Water hydride fuel, which is approximately equal to that of
The compatibility of hydride fuel with coolant water is Zircaloy (Zry). However, the important consideration
a prime consideration in the use of this material in is not the temperatures per se; it is how the lower
LWRs. Although long-term tests of the fuel in high- temperatures in the hydride affect fuel performance.
pressure water at 300 C have not been performed, At the maximum temperature of the helium-bonded
quench tests of TRIGA fuel (identical in composition hydride fuel, FP release is due to recoil only (see
to the proposed power-reactor fuel) have been Section 3.12.3.4), whereas the UO2 fuel is hot
reported.66 Specimens about the size of a BWR oxide enough for release of fission gases by diffusion to be
fuel pellet were heated up to 1200 C (atmosphere not significant. The approximately fourfold lower DTfuel
specified) for a matter of minutes and dropped into in the hydride fuel compared with the oxide version
water. The authors report the retrieved specimens to considerably reduces the thermal stresses in the
have suffered ‘minor cracking,’ but otherwise they former, probably enough to avoid pellet cracking
were in excellent condition. The average weight loss entirely. For the conditions given in Table 14, the
for eight tests was 2%. This was attributed to hydrogen stored energy in the helium-bonded hydride fuel is
loss, partially counterbalanced by surface oxidation. one-quarter of that in the oxide fuel at the same
346 Uranium–Zirconium Hydride Fuel
LHR. The factor is even larger if the hydride fuel has Hydride fuel-element designs call for replace-
a LM in the fuel-cladding gap. ment of helium in the fuel-cladding gap by an LM.
The choice of the H/Zr atom ratio of 1.6 is a The alloy of choice is the ternary Pb–Sn–Bi eutectic,
compromise between incorporating the largest possi- which melts at 120 C and is chemically benign.26
ble hydrogen concentration for neutron moderation It serves two important purposes. The first is to pro-
without generating too large a hydrogen overpressure. tect the cladding inner wall from noxious FPs. The
At 700 C, for example, Figure 6 gives p(H2) ¼ 0.1 atm second is to reduce the fuel temperatures by 100 C or
for ZrH1.6. Hydrogen gas filling a 15 cm3 plenum at this more compared with helium-bonded fuel elements.
pressure constitutes <0.1% of the hydrogen in a stan- Demonstration of thermal benefits of LM-bonded
dard LWR-size pellet of (U0.31Zr)H1.6, so there is no fuel requires a test-reactor irradiation. Until this is
risk of hydrogen depletion of the fuel by this mecha- done, the following simple transient test served the
nism. The choice of an H/Zr ratio of 1.6 places the fuel same purpose. Helium- and LM-bonded, UO2-
squarely in the center of the d-Zr–H phase, which fueled, Zircaloy-clad rodlets were quenched from
eliminates the danger of modest composition variations 600 C by immersion in a large body of water at
resulting in the precipitation of another phase. How- 25 C. Each rodlet had a thermocouple in a hole drilled
ever, the H2 pressure does not limit the maximum into the centerline of the middle pellet in order to
operating temperature during normal operation. measure the transient response of the fuel centerline
temperature. The external heat-transfer coefficient
Table 14 Operating temperatures of PWR fuel rods was determined by performing the same test on a
with hydride or oxide fuel solid copper rod of the same diameter as the miniature
fuel elements. Figure 44 shows the experimental
Temperature U0.31ZrH1.6 (He) U0.31ZrH1.6 (LM) UO2 (He)
results along with the theoretical predictions obtained
Fuel 680 555 1450a by solving the heat conduction equation in fuel with
centerline the boundary condition accounting for the series resis-
DTfuel 170 170 665
DTgap 125b 1b 125b
tances due to the fuel-cladding bond material, the
DTclad 46 46 46 cladding, and the external fluid film. The only differ-
DTfluid 39 39 39 ence between the two rodlets was the bond material.
Two conclusions can be drawn from this test. The
The fuel OD is 12 mm and the LHR is 37.5 kW m1. The coolant first is that the LM is a much more efficient heat
temperature is 300 C.
a
UO2 thermal conductivity ¼ (22.7 þ 7 103T (K))1. transfer medium than that of helium, which is not
b
For a gap thickness of 35 mm. surprising considering the difference in the thermal
600
Experimental
500
Calculated
Centerline temperature (⬚C)
400
300 He
200
100
LM
0
0 10 20 30 40 50
Time (s)
Figure 44 Transient responses of fuel centerline thermocouples in three-pellet rodlets with helium or a liquid metal in the
fuel-cladding gap.
Uranium–Zirconium Hydride Fuel 347
Vacuum
for clearing To time
the pump switch
N2 (75 atm)
N2 (75 atm)
Membrane
Gauge Pressure
release
Hydrogen Gas purge
H2
Injection
UF6
Oscilloscope from
measurement of
reactor pressure
H2 (20 atm)
Pressure
Collection of
Cylinder of UF4 powder
UF6
Heater
Jack for
holding
apparatus
Figure 46 Apparatus for reducing UF6(g) to UF4(s). Reproduced from Sauteron, J. Les Combustibles Nucleaire, Herman,
1965.
3. pickling with a solution of HNO3 and HF add hydrogen at 100 Torr above the equilibrium
4. washing in distilled water, alcohol pressure of the b-Zr/d two-phase region, which is
130 Torr at this temperature. The objective is to
Addition of hydrogen to the alloy must be done
bring the H/Zr ratio to about unity, which is the
very carefully to avoid cracking the hydride because
terminal solubility of hydrogen in b-Zr. As long as
of the volume change mentioned above. The pre-
the hydride phase is not formed, there is no risk of
ferred method is to heat the metal to 800 C and
Uranium–Zirconium Hydride Fuel 349
Probe
Valve
Detachable valve
Diffusion pump
(secondary vacuum)
U
Ingot Water
Coil
Crucible
Support Valve
Molds
Figure 47 Induction-melting furnace for production of U–Zr ingots. Reproduced from Sauteron, J. Les Combustibles
Nucleaire, Herman, 1965.
cracking the metal. The hydrogen pressure is then 3.12.7.1 Fuel Swelling
increased to 1 atm, and the gas is admitted in small
The variables affecting this calculation are the initial
pulses to avoid excessively rapid absorption in the
fuel radius R0 and the fuel centerline and surface
metal, which would risk cracking of the hydride
temperatures, T0 and TS. These are taken from
being formed. Additional details of the fabrication
Table 14.
process are provided in Huffine68 and in many of the
The increase in the fuel radius, DR, is related to
patents for TRIGA fuel elements.30,69
the azimuthal strain at the fuel surface, etot
y , by
DR ¼ etot
y R0 ½32
3.12.7 Gap Closure During Operation
where
The objective of this section is to determine the thick-
y ¼ ey þ ey þ ey
etot th H FP
½33
ness of the fuel-cladding gap as a function of burnup.
The fuel is U0.31ZrH1.6 and the gap is filled with an The three terms on the right are due to the following
LM. Because of the low fuel temperature, there is no expansion mechanisms: temperature distribution,
fission-product release (Figure 36) and the only hydrogen redistribution, and fission-product swelling.
fission-related swelling is due to solid FPs and fission The first two terms are calculated using the method in
gas in atomic form in the fuel (Section 3.12.3.3). Olander and Ng.19
350 Uranium–Zirconium Hydride Fuel
unfortunately mixes units. The fast flux in eqn [55] is Thermomechanical state:
in n m2 s1, while the fast fluence in eqn [51] is in Dp ¼ pressure difference between coolant and rod
n cm2. The two are related by F ¼ ft. Strain rates interior ¼ 3 MPa. The hoop stress (compressive) in
are expressed in h1 but the flux utilizes seconds): the cladding is
eirr ¼ 3:56 1024 f0:85 sy t ½55 1=2dO 6
sy ¼ Dp ¼3 ¼ 23MPa
dC 0:78
3.12.7.2.4 Total cladding creep strain
The total cladding creep strain as a function of time T ¼mean cladding temperature. For the thermal
(in h) is given by conditions in Table 14, T ¼ 635 K.
E = Young’s modulus for SRA-Zry-4. At 635 K, it
C ¼ ep þ ess þ eirr
etot ½56 is 7.7 104 MPa70
The creep strains in this equation are considered to Irradiation conditions:
be positive quantities even though they apply to f ¼ 1018 nm2 s1
creepdown of the cladding tube. This is taken into
account in eqn [58].
The fast fluence is
to thermal and hydrogen stresses. The difference in LHR, the enrichment varies as the square of the fuel
the initial fuel radii for the two cold gap thicknesses is pellet diameter. In order to maintain the same LHR
due to the requirement that the cladding geometry for the two fuel pellet dimensions, the enrichment of
remains constant. The increase thereafter is due the small pellet must be 9% larger than for the large
solely to fission-product swelling. With an 80 mm pellet. If the fuel in the rod with the 80 mm gap has an
initial cold gap, PCMI occurs at an equivalent-oxide enrichment of 12%, this figure needs to be increased
burnup of 10 MWd kgU1. With a 150 mm initial to 13% if the gap thickness is increased to 300 mm.
gap, closure is delayed until 27 MWd kgU1 The reason for the necessity of a large initial gap
equivalent-oxide burnup. The goal of avoiding gap to avoid PCMI at 60 MWd kgU1 is the sizeable
closure until 60 MWd kgU1 cannot be achieved fission-product swelling rate of the hydride fuel.
using the LM in the gap unless the initial gap According to the only data available, this quantity is
is 300 mm. According to eqn [58], a gap of this size 3 times larger in hydride fuel than in oxide fuel.
contains fuel of 4.92 mm radius, compared with
5.17 mm for the 80 mm cold radial gap. For a fixed
3.12.8 Conclusions
10
sq = 23 MPa
The principal properties of (U,Zr)Hx, notably its high
f = 1 ⫻ 1014 cm− 2 s−1
.
thermal conductivity, provide this material with sev-
8
F = 3 ⫻ 1013 cm−3 s−1 eral attractive features as a nuclear fuel. The low fuel
ep
6 temperature reduces fission-product release to negli-
4
gible amounts and decreases stored energy in the
ess fuel. With a low-melting LM in the fuel-cladding
2 gap, PCMI can be delayed or even avoided by
0
increasing the thickness of the cold gap. Initial fuel
0 2000 4000 6000 8000 cracking due to thermal stress may be avoided
Time (h) because of the countervailing effect of hydrogen
Figure 49 Components of cladding creepdown. redistribution.
5.22
Cladding
5.20
Radial position (mm)
dg0 = 80 µm
5.18
5.10
0 1000 2000 3000 4000 5000 6000 7000 8000 9000
Time (h)
Figure 50 Gap closure – fuel and cladding inner radii as functions of burnup.
354 Uranium–Zirconium Hydride Fuel
A cautionary note is that the appearance of voids with H2 pressure pH2 in the fuel-cladding gap and
produced by fission-fragment damage the hydride zero pressure at the Zry/SS interface is
lattice adjacent to the uranium particles. The swelling pffiffiffiffiffiffiffi
CH pH2
(DV/V) due to the voids appears early in irradiation J H ¼ DH ¼ ðDH KH Þ ½A2
(burnup 103 FIMA) and is highly temperature- dSS dSS
sensitive: it is absent at 650 C, reaches 1% at 700 C, The product DHKH is the permeability of hydrogen in
and 5% at 750 C. LM bonding of the fuel-cladding the metal. For stainless steel, LeClaire gives the
gap is necessary to avoid temperatures where severe equation27:
initial swelling occurs.
Although many of the essential radiation proper- DH KH ¼ 3:3 107 expð7900=T Þmol m1 s1 Pa0:5 ½A3
ties of hydride fuel have been determined, a well-
To assess the utility of this protection scheme in the
instrumented irradiation test under power-reactor
present design, consider a 0.4 mm-thick SS liner at
conditions is needed, particularly in order to assess
340 C exposed at its inner surface to the equilib-
(1) the effect of released fission gas (if any) on the
rium hydrogen pressure generated by the hydride
thermal effectiveness of the LM bond in the fuel-
fuel (5 103 atm, Figure 6). The above two equa-
cladding gap and (2) whether the initial void swelling
tions give a permeation rate JH ¼ 5 1012 mol H per
occurs during power-reactor conditions. Additional
cm2 s1. The fractional change rate of the H/Zr ratio
laboratory tests include measurement of the compat-
(x) of the fuel is
ibility of the hydride fuel with PWR coolant water at
300 C and with steam at T > 1000 C representative dðx=x0 Þ J
¼ H ½A4
of accident conditions. dt NH0 dI
A significant unknown is the cost of large-scale
where x0 is the initial H/Zr ratio (1.6), NH0 ¼ 0.08 mol
fabrication of hydride fuel elements, which, when
H per cm3 is the molar density of hydrogen in ZrH1.6,
added to the increased enrichment cost, may com-
and dI is the ID of the cladding (1 cm).
promise its competitiveness vis-à-vis oxide fuel.
nYZr 1 + nYZr+ 12NO FP Fission This step, shown in Figure B1(b), is based on 1 mol
n moles
fission
moles mole Zr of ZrH1.59 at the burnup FIMA* of oxygen removal
from ZrO2. Following n0 additional fissions per mole
Zr
FP
oxide of initial Zr, 1 þ n0 YZr moles of Zr are present. Since
the active fission products form dihydrides, the quan-
(a)
tity of hydrogen is reduced to 1.59–2n0 (YRE þ YBa, Sr).
1 mol The H/Zr ratio is
ZrH1.59
n⬘YRE H 1:59 2n0 ðYRE þ YBa; Sr Þ
Zr
moles ¼ ½B4
n⬘YZr
1 + n⬘YZr
H RE Zr 1 þ n0 YZr
moles
moles Zr
Zr The number of moles of uranium fissions is given by
n⬘YBa, Sr
eqn [24]:
1.59-2n⬘(YRE+YBa, Sr) moles
moles H Ba, Sr n0 ¼ ðF F
Þ=NAv ¼ rZr ð1 þ yÞðFIMA FIMA
Þ
1 mol per 91:2 g Zr ¼ 0:050 mol Zrper cm3 ½B3 F_ t ¼ NU0 ð1 þ y 1 Þ FIMA ½B8
The fission density required for this process is For the uranium density given in Table 1 (3.7 g
Meeting on Reduced Enrichment for Research & Test on Future Nuclear Systems (Global’97), Yokohama, Japan,
Reactors, Petten, The Netherlands, Oct 1985. Oct, 1997.
4. Lillie, A.; et al. Zirconium Hydride Fuel Element 33. Tsuchiya, B.; Konashi, K.; Yamawaki, M.; Nakajima, Y.
Performance Characteristics; Atomics International/ J. Nucl. Sci. Technol. 2002, (Suppl. 3), 855.
USAEC Report AI-AEC-13084; 1973. 34. Konashi, K.; Tsuchiya, B.; Yamawaki, M.; Fujimura, K.;
5. Shuffler, C.; Malen, J.; Diller, P.; et al. Nucl. Eng. Des. Sanda, T. Development of actinide–hydride target for
2009, 239, 1505–1530. transmutation of nuclear waste. In Proceedings of the
6. Shuffler, C.; Malen, J.; Trant, J.; Todreas, N. Nucl. Eng. International Conference on Back-End of the Fuel Cycle:
Des. 2009, 239, 1442–1460. From Research to Solutions (GLOBAL 2001), Paris,
7. Iorgulis, C.; et al. Neutronic calculations regarding the new France, Sept 9–13, 2001.
LEU 66 fuel bundle for the 14 MW TRIGA. In International 35. Konashi, K.; Yamawaki, M. Adv. Sci. Technol. 2010, 73,
Meeting on Reduced Enrichment for Research Reactors, 51–58.
Sao Paulo, Brazil, 1998. 36. Libowitz, G. G. The Solid-State Chemistry of Binary Metal
8. Toma, C.; et al. Characterization of TRIGA LEU Fuel Hydrides; W.A. Benjamin: New York, Amsterdam, 1965.
Behavior in 14 MW Core; Un-numbered Report; Institute 37. Ward, J. W. In Handbook on the Physics and Chemistry of
for Nuclear Research: Romania, 2002. the Actinides; Freeman, A. J., Keller, C., Eds.; Elsevier:
9. Konashi, K.; Tsuchiya, B.; Yamawaki, M.; Fujimura, K.; Amsterdam, 1985; pp 1–74.
Sanda, T. Development of actinide–hydride target for 38. Mueller, W. M.; Blackedge, J. P.; Libowitz, G. G., Eds.
transmutation of nuclear waste. In Proceedings of the Metal Hydrides; Academic Press: New York, London,
International Conference on Back-End of the Fuel Cycle: 1968.
From Research to Solutions (GLOBAL 2001), Paris, 39. Flotow, H. E.; et al. In The Chemical Thermodynamics of
France, Sept 2001. Actinide Elements and Compounds; Flotow, H. E.,
10. Konashi, K.; Yamawaki, M. In Characterization and Quality Haschke, J. M., Yamauchi, S., Eds.; Part 9 of the IAEA
Control of Nuclear Fuels; Ganguly, C., Jayaraj, R. N., Eds.; Series ‘‘The Actinide Hydrides’’; International Atomic
Allied Publishers: New Delhi, 2004; pp 92–106. Energy Agency: Vienna, 1984; STI/PUB/424/9.
11. Parfenov, V.; et al. Corrosion of Zirconium and Its Alloys; 40. Bartcher, W. In Hydrogen Metal System I; Lewis, F. A.,
Atomizdat: Moscow, 1967; p 179. Aladjem, A., Eds.; Solid State Phenomena; 1996;
12. Gelezunas, V.; et al. J. Electrochem. Soc. 1963, 110, 799. vol. 49–50, pp 159–238.
13. V’yugov, P.; Ryabchikov, L. Izv. Akad. Nauk. SSR Met. 41. Taylor, C. D.; Lookman, T.; Lillard, S. R. Acta Mater. 2010,
(Russian Metallurgy) 1979, 3, 214. 58(3), 1045–1055.
14. Albrecht, W.; Goode, W. Diffusion of Hydrogen in 42. Chiotti, P. In Binary Alloy Phase Diagrams, 2nd ed.;
Zirconium Hydride; USAEC Report BMI-1426; 1960. Massalski, T. B., Ed.; 1990, Vol. 2, Vol. 2, pp 2070–2071.
15. Majer, G.; et al. J. Phys. Condens. Matter 1994, 6, 2935. 43. Manchester, F. D.; San Martin, A. J. Phase Equilib. 1995,
16. Paetz, V.; Lücke, K. Z. Metallkd. 1971, 62, 657. 16, 263–275.
17. Sommer, A.; Dennison, W. Thermal Diffusion of Hydrogen 44. Nagasaki, S.; Hirabayashi, M. Phase Diagrams of
in Nonstoichiometric Zirconium Hydride; USAEC Report Binary Alloy; Agne Gijutsu Center: Tokyo, Japan, 2001;
NAA-SR-5066; 1960. p 308.
18. Huang, J.; Tsuchiya, B.; Konashi, K.; Yamawaki, M. 45. Flotow, H. E.; Osborne, D. W. Phys. Rev. 1967, 164, 755.
J. Nucl. Sci. Technol. 2000, 37, 887. 46. Flotow, H. E.; Lohn, H. R.; Abraham, B. M.; Osborne, D. W.
19. Olander, D.; Ng, M. J. Nucl. Mater. 2005, 346, 98. J. Am. Chem. Soc. 1959, 81, 3529.
20. Zuzek, E.; Abriata, J. Bull. Alloy Phase Diagrams 1990, 47. Abraham, B. M.; Flotow, H. E. J. Am. Chem. Soc. 1955,
11, 385. 77, 1446.
21. Wang, W.-E.; Olander, D. J. Am. Ceram. Soc. 1995, 78, 48. Ward, J. W.; Bartscher, W.; Rebizant, J. J. Less Common
3323. Met. 1987, 130, 431–439.
22. Beck, R.; Mueller, W. In Metal Hydrides; Mueller, W., 49. Okamoto, H. In Binary Alloy Phase Diagrams, 2nd ed.;
Blackledge, J., Libowitz, G., Eds.; Academic Press: Massalski, T. B., Ed.; 1990; Vol. 2, 2054–2055; Vol. 2,
New York, 1968; Section 7.3. pp 2054–2055.
23. Yamanaka, S.; et al. J. Alloys Compd. 1999, 50. Mulford, R. N. P.; Sturdy, G. E. J. Am. Chem. Soc. 1956,
293–295, 23. 78, 3897–3901.
24. Merten, U.; et al. Preparation and properties of U–Zr 51. Mulford, R. N. P.; Sturdy, G. E. J. Am. Chem. Soc. 1955,
hydrogen alloys. In Proceedings of the Second 77, 3449–3452.
Conference on Peaceful Uses of Atomic Energy, Geneva, 52. Oetting, F. L.; Hodges, A. E.; Haschke, J. M.; et al.
1958; p 789. J. Chem. Thermodyn. 1984, 16(11), 1089–1102.
25. Yamanaka, S.; et al. J. Nucl. Mater. 2001, 294, 94. 53. Bansal, B. M.; Damien, D. Inorg. Nucl. Chem. Lett. 1970, 6,
26. Wongsawaeng, D.; Olander, D. Nucl. Technol. 2007, 159, 603–606.
279–287. 54. Gibson, J. K.; Haire, R. G. J. Phys. Chem. 1990, 94,
27. LeClaire, A. J. Nucl. Mater. 1984, 122–123, 1558. 935–939.
28. Greenspan, E. Personal communication from Massoud 55. Milman, V.; Winkler, B.; Pickard, C. J. J. Nucl. Mater. 2003,
Simnad, 1998. 322, 165–179.
29. Hesketh, K. W.; Stanbridge, J. R. Technology objectives 56. Bartscher, W.; Rebizant, J. J. Less Common Met. 1988,
for nuclear fuel in 2020. In Proceedings of GLOBAL-93, 136, 385–394.
Seattle, WA, Sept, 1993. 57. Konashi, K.; Pudjanto, B. A.; Terai, T.; Yamawaki, M.
30. Eggers, G. Method of Making ZrH Fuel Element, U.S. J. Phys. Chem. Solids 2005, 66, 625–628.
Patent 4,071,587, 1978. 58. Pudjanto, B. A.; Konashi, K.; Terai, T.; Yamawaki, M.
31. Yamamoto, T.; Kayano, T.; Suwarno, H.; Yamawaki, M. J. Phys. Chem. Solids 2005, 66, 665–670.
Sci. Rep. RITU 1995, A40, 13. 59. Olander, D. R. Nuclear Reactor Fuel Elements; National
32. Yamawaki, M.; Suwarno, H.; Yamamoto, T. Development Technical Information Services, Document No. 26711;
of MA-containing hydride fuel as MA burning target 1976.
material. In Proceedings of the International Conference
Uranium–Zirconium Hydride Fuel 357
60. Olander, D. R.; Wongsawaeng, D. J. Nucl. Mater. 2006, 66. Lindgren, J.; Simnad, M. Trans. Am. Nucl. Soc. 1979.
354, 94. 67. Van Houten, R. Nucl. Eng. Des. 1974, 31, 434.
61. Huang, J.; Tsuchiya, B.; Konashi, K.; Yamawaki, M. 68. Huffine, C. In Metal Hydrides; Mueller, W., Blackledge, J.,
J. Nucl. Mater. 2001, 294, 154. Libowitz, G., Eds.; Academic Press: New York, 1968;
62. GA Technologies, Inc. Uranium–zirconium hydride TRIGA Chapter 13.
LEU fuel. In Research Reactor Core-Conversion Guidebook, 69. Simnad, M.; et al. Method of Hydriding. U.S. Patent
Vol. 4, Fuels, IAEA-TECDOC-643, Appendix I-7 (19) 3,135,697, 1964.
63. West, G.; Simnad, M.; Copeland, G. Final Results 70. Limback, M.; Andersson, T. A model for analysis of the
from TRIGA LEU Fuel Postirradiation Examination effect of final annealing on the in- and out-of-reactor creep
and Evaluation Following Long-Term Irradiation behavior of Zircaloy cladding. In Zirconium in the Nuclear
Testing in the ORR; General Atomics Report GA-A18641; Industry: Eleventh Symposium, 1996; ASTM STP 1295,
1986. pp 448–468.
64. Konashi, K.; et al. Irradiation of hydride fuels in JMTR. 71. Lyon, W. F.; Jahingir, N.; Montgomery, R. O.; Yagnik, S.
In Proceedings of the International Congress on Advances Capabilities of the FALCON steady state and transient fuel
in Nuclear Power Plants (ICAPP 2007), Nice, France, May performance code. In Proceedings of the 2004
13–18, 2007. International Meeting on LWR Fuel Performance, Orlando,
65. Terrani, K.; et al. J. Nucl. Mater., submitted. FL, Sept 19–22, 2004.
3.13 Molten Salt Reactor Fuel and Coolant
O. Beneš and R. J. M. Konings
European Commission, Joint Research Centre, Institute for Transuranium Elements, Karlsruhe, Germany
359
360 Molten Salt Reactor Fuel and Coolant
MSR
Molten salt reactor
Control
rods
Purified
salt
Turbine
Freeze
plug Heat Precooler Heat sink
Pump sink
Intercooler
Compressor
Figure 1 Schematic drawing of the molten salt reactor. Reproduced from US DOE Nuclear Energy Research Advisory
Committee and the Generation IV International Forum, A Technology Roadmap for Generation IV Nuclear Energy Systems,
http://www.ne.doe.gov/genIV/documents/gen_iv_roadmap.pdf © Generation IV International Forum.
Table 1 Typical fuel salt inlet and outlet temperatures of (as discussed in Section 3.13.7). Therefore, issues
some MSR concepts such as swelling or crack formation that appear in
MSR concept Tinlet Toutlet References the case of ceramic fabricated fuels are avoided.
Compatibility with nickel-based alloys (Ni–Mo– fertile (ThF4) and fissile elements (UF4). As second-
Cr–Fe) that can be used as structural materials. ary coolant, a NaF–NaBF4 (8–92 mol%) mixture was
foreseen, particularly because the tritium retention
In the second half of the 1950s, the molten salt of this salt is much better than FLIBE.
technology was transferred to the civilian nuclear In the 1990s, there was a renewed interest
program of the United States. At the time, many in molten salt technology, which originated from
reactor concepts were being studied and the interest programs that were looking into the possibilities
in breeder reactors was growing. It was recognized of transmutation of actinides. When addressing
that the MSR would be ideal for thermal breeding of transmutation of minor actinides, the absence of
uranium from thorium,7 and the Molten Salt Reactor complicated fuel and fuel pin fabrication and the
Experiment (MSRE) was started at ORNL to dem- compatibility with pyrochemical processing in the
onstrate the operability of MSRs. Because of the molten salt fuel cycle were recognized as important
breeding aspect, the neutron economy in the reactor advantages, in comparison with conventional pellet
was considered to be of key importance, and 7LiF– fuel types. Also, the interest in the use of thorium as a
BeF2 (FLIBE), with 5% ZrF4 as oxygen getter, was nuclear fuel kept up the interest in MSRs. As a result,
selected as fuel carrier because of the very low neu- the MSR is now one of the six reactor concepts
tron capture cross-sections of 7Li (sthermal ¼ 0.045 selected for the Generation IV initiative, which is
barn) and Be (sthermal ¼ 0.0088 barn). Natural lith- looking at next generation nuclear reactors. Current
ium cannot be used as part of the nuclear fuel as MSR designs, however, move away from thermal
it contains about 7.6% of 6Li (the remaining 92.4% graphite-moderated concepts, and favor nonmoder-
is 7Li), which has a very high parasitic neutron ated concepts that have a fast(er) neutron spectrum.
capture cross-section (sthermal ¼ 940 barn). There- Fuel selection for the nonmoderated reactor con-
fore, enrichment of 7Li is required before it can be cepts is more flexible, and elements other than 7Li
used as a fuel matrix. The MSRE was a graphite- can be considered. One reason is that the neutron
moderated reactor of 8 MWth (megawatt thermal) capture cross-section of the alkali halides and alkali-
and operated from 1965 to 1969. Two different fissile earth halides is generally lower in the ‘fast’ spectrum
sources were used: initially, 235UF4 was used with than in the thermal spectrum; also, the neutron econ-
33% enrichment and later, 233UF4 was added to the omy is not as sensitive in the ‘fast’ spectrum as in
carrier salt, making the MSRE the world’s first reac- the thermal one. Therefore compounds like NaF,
tor to be fueled with this fissile material.10 FLIBE was KF, RbF, or CaF2 can be considered as part of
used as coolant in the secondary circuit. The results the fuel matrix. Moreover, there are some ‘fast’
of MSRE, which have been reported in great detail,10 MSR concepts, for example, the REBUS-3700 con-
revealed that all the selected materials (fuel, struc- cept,11 which are based on the chloride matrix (35Cl:
turals) behaved well and that the equipment behaved sfast ¼ 0.0011 barn, whereas sthermal ¼ 43.63 barn).
as predicted. In this respect, it was very successful.
After the MSRE, a design for a prototype molten
salt breeder reactor (MSBR) was made by ORNL in 3.13.3 Fuel Concepts of MSR
the early 1970s.3 The program was stopped in 1976
in favor of the liquid metal cooled fast reactor7: The fuel in the MSR must fulfill several require-
although the technology was considered promising, ments with respect to its physicochemical properties
there were technological problems that had to be (as will be discussed in Section 3.13.4). These
solved. The MSBR design was a 2250 MWth reactor, requirements are very well met by the various sys-
optimized to breed 233U from 232Th in a single fluid tems containing alkali metal and alkali-earth fluor-
system. Online pyrochemical cleanup was planned to ides; hence the fluoride systems are the most
clean the fuel solvent from the neutron-absorbing recognized candidates for MSR fuels.
fission products. Nevertheless, interruption of reactor In the previous section, the MSBR has been men-
operation was planned every 4 years to replace the tioned as a graphite-moderated reactor that is based
graphite moderator, as experiments had revealed sig- on the 7LiF–BeF2–232ThF4–UF4 system.3 232ThF4 is
nificant swelling of graphite due to radiation damage. a fertile material that is used to produce fissile 233UF4
Because of the (semi)continuous online clean up of by a neutron capture and two consecutive b-decays
the fuel, the addition of zirconium to the fuel was not of 233Th and 233Pa. This fuel composition based on
necessary, and FLIBE could be used as carrier of the the FLIBE matrix still remains an ideal candidate
Molten Salt Reactor Fuel and Coolant 363
when the MSR is designed as a thermal breeder One of the very recent MSR designs is the
reactor (moderated reactor). In this case, neutron REBUS-3700 concept, which is based on a chloride
economy is very critical and only isotopes with very salt as a fuel. It is a fast breeder reactor proposed by
low neutron capture cross-section in the thermal Mourogov and Bokov11 and it is based on a 238U/239Pu
spectrum can be part of the fuel matrix. Thus, 7LiF cycle, where 238U serves as a fertile material bred
and BeF2 are the prime compounds for consideration. to fissile 239Pu by neutron capture and two consecu-
One of the current MSR concepts that uses fuel tive b-decays of 239U and 239Np. Both uranium and
technology similar to that of the MSBR is the MSR plutonium are present in the form of trichlorides
FUJI concept.4 Originally proposed by Furukawa, it dissolved in a matrix of liquid NaCl. In general, the
is a rather small graphite-moderated concept with an chlorides have higher vapor pressures and lower
installed thermal capacity of 450 MW. thermodynamic stability at high temperatures com-
Nowadays the nonmoderated reactors are attract- pared to fluorides, but, on the other hand, their
ing interest because they offer the possibility of trans- melting points are lower. Therefore, more fissile
muting the long-lived actinides produced mostly in material can be dissolved in the matrix, which is
light water reactors. The transmutation is most effec- essential for fast breeder reactor designs. However,
tive in the fast neutron spectrum; however, due to the the chlorides can be used only in fast reactors and not
presence of the fluorine atom in the fuel, partial in thermal ones due to the relatively high parasitic
moderation is maintained, and the neutron spectrum neutron capture cross-section of the chlorine atom,
of the MSR is, rather, shifted to the epithermal range. as already discussed in Section 3.13.2.
Nevertheless, at this energy, all the minor actinides A summary of various applications of molten salts
are fissionable, and the fission-to-capture ratio for in future nuclear reactor designs is given in Table 2.
these nuclides is still much higher than in the thermal As the primary choices for the MSR fuels or coolants
spectrum.12 Furthermore, the nonmoderated reactor are based on the fluoride systems, the chloride sys-
does not require graphite blocks (moderator in the tems are not discussed further.
thermal MSR) in the reactor core: they are very
susceptible to radiation damage and must be periodi-
cally replaced.
3.13.4 Properties of the MSR Fuels
At the moment, there are two main directions
and Coolants
for the nonmoderated MSR concepts. The first is
an actinide burner design based on the Russian
In this section, the physicochemical properties of the
MOSART (Molten Salt Actinide Recycler and Trans-
primary MSR fuel and coolant choices from Table 2
muter) concept,6 for which the 7LiF–(NaF)–BeF2–
are discussed, with the emphasis on the melting
AnF3 system is proposed as a fuel salt. The startup
behavior, actinide solubility in the fuel matrix, den-
and feed material scenarios can include plutonium
sity, viscosity, heat capacity, thermal conductivity, and
and minor actinides from pressurized water reactor
vapor pressure. All these quantities are highly rele-
(PWR) spent fuel. Depending upon the feed material,
vant for the reactor design calculations and a sum-
the fuel salt at equilibrium contains 0.7–1.3 mol% of
mary of these properties for typical coolant and fuel
actinide and lanthanide trifluorides. The second one
compositions is given in Tables 3 and 4 respectively.
is an innovative concept called MSFR (molten salt fast
Optimized phase diagrams of the relevant fluoride
reactor), which has been developed by Centre
systems used as MSR fuels, coolants, or heat transfer
National de la Recherche Scientifique (CNRS) in
salts are also shown in this section.
France.5,13–16 The fuel in this concept is based on
the 7LiF–232ThF4 matrix, with the addition of acti-
nide fluorides as a fissile material. There are two
3.13.4.1 Structural Aspects of Molten Salts
initial fissile choices in the MSFR concept: (1) the
233
U-started MSFR and (2) the transuranic-started Molten fluoride salts are essentially ionic liquids in
MSFR with a mix of 87.5% of Pu (238Pu 2.7%, 239Pu which cations and anions form a loose network. Some
45.9%, 240Pu 21.5%, 241Pu 10.7%, and 242Pu 6.7%), cations occur in their simplest form, such as Liþ and
6.3% Np, 5.3% of Am, and 0.9% of Cm in the form of Naþ, but some form molecular species like BeF2,
fluorides, corresponding to the transuranic element which is a structural analogue to SiO2, known to be
composition of a UO2 fuel after one use in a PWR and highly associated and forming a network structure
5 years of storage.17 that exhibits a glass transition characteristic. In a
364 Molten Salt Reactor Fuel and Coolant
a
Advanced high-temperature reactor, graphite-moderated, thermal reactor.
b
Very high-temperature reactor, graphite-moderated, gas cooled reactor.
c
Molten salt cooled fast reactor, the solid fuel fast reactor with MS as a coolant.
d
Sodium cooled fast reactor.
e
Heat transfer salt is a medium that will be used to deliver heat from the reactor to the hydrogen production plant.
f
To separate sodium and the steam circuits.
a
Value for T ¼ 1023 K.
recent study by Salanne et al.,18 a molecular dynamic so forth, resulting in a polymer of several BeF2 4
study was performed on the LiF–BeF2 system in order units. This polymerization is also a reason why the
to understand the structure of the (Li,Be)F2x melt. viscosity of pure BeF2 is much higher compared to
Figure 2 shows the distribution of various species that of other fluorides discussed in this chapter.
observed in the solution as a function of BeF2 BeF2
4 species were also experimentally observed
composition. At low concentrations of BeF2 in LiF, by spectroscopic studies, as reported by Toth and
the mixture behaves as a well-dissociated ionic melt Gilpatrick.19 Lanthanide fluorides, ThF4 or PuF3
consisting of Liþ, BeF2
4 , and F species. As BeF2 also form molecular species in their liquid form,
concentration increases, the BeF2 4 units start to but in comparison to BeF2, they do not exhibit poly-
bond together sharing a common F ion, first creat- merization. Dracopolous et al.20,21 investigated
ing Be2 F3 7
7 species, followed by Be3 F10 species, and the structure of molten KF–YF3 and KF–LnF3
Molten Salt Reactor Fuel and Coolant 365
100
80
F-
BeF42-
60
Be2F73-
%F
4-
Be3F10
40 5-
Be4F13
‘Polymer’
20
0
0 20 40 60 80
mol% BeF2
Figure 2 Percentage of F atoms involved in various species observed in the LiF–BeF2 system as a function of
composition; ‘polymer’ means a cluster with a Be nuclearity >4, whereas F implies that the ion is coordinated only to Liþ.
Reproduced from Salanne, M.; Simon, C.; Turq, P. J. Phys. Chem. B 2007, 111, 4678–4684.
(Ln ¼ La, Ce, Nd, Sm, Dy, Yb) systems using Raman and Konings,25 the latter version being preferred
spectroscopy and found that at x(LnF3)0.25, LnF3 6 as the authors considered not only the equilib-
are the predominant species surrounded by Kþ rium points measured,26–28 but also the mixing
cations. At higher concentrations of LnF3, the enthalpies of the (Li,Be)Fx liquid solution
lanthanides are forced to share common fluorides measured by Holm and Kleppa.29 The LiF–BeF2
and start to create loose structures of bridged phase diagram is shown in Figure 3; it is charac-
octahedra. On the basis of these two studies, the terized by two eutectic invariant equilibria found at
authors concluded that lanthanide melts have simi- T ¼ 636 K and xðBeF2 Þ ¼ 0:517, and T ¼ 729 K and
lar structural behavior. In case of thorium, a tetrava- xðBeF2 Þ ¼ 0:328 in the calculation. Two interme-
lent ion is the only known species in molten diate phases, Li2BeF4 and LiBeF3, are present in
fluorides. As reported by Barton,22 ThF4 forms the system as well, the first melting congruently
mainly anionic complexes of the general formula at T ¼ 729 K, whereas the latter decomposes below
ThFm 23
4þm , and the existence of ThF5 is claimed. In the solidus at T ¼ 557 K. A miscibility gap appears
case of uranium, tri- or tetravalent ions are stable in in the BeF2-rich side, with the monotectic tem-
the molten fluoride salt. It has been demonstrated19 perature found at T ¼ 772 K, while the critical
that UF4 dissolves in the fluoride melts, forming temperature was found at Tc ¼ 812 K and
complexes of coordination numbers 7 or 8. It has x(BeF2) ¼ 0.826.
been shown that in fluoride-rich systems, the UF4 8
species predominates, while with the reduction of
fluoride ions, the UF3 3.13.4.2.2 LiF–PuF3
7 species is produced according
The thermodynamic assessment of the LiF–PuF3
to UF4 8 Ð UF7 þ F . Furthermore, the same
3
authors confirmed that approximately equal amounts system was made in a study by van der Meer et al.30
of UF4 3 and later by Beneš and Konings,31 using a different
8 and UF7 occur in the LiF–BeF2 melt of
intermediate composition. thermodynamic model based on the equilibrium data
measured by Barton and Strehlow.32 The calculated
phase diagram as obtained from the data of Beneš
3.13.4.2 Phase Diagrams
and Konings is shown in Figure 4, indicating very
3.13.4.2.1 LiF–BeF2 good agreement with the experimental data. The
The LiF–BeF2 phase diagram has been assessed by system is characterized by a single eutectic at
van der Meer et al.24 and more recently by Beneš T ¼ 1018 K and x(PuF3) ¼ 0.212.
366 Molten Salt Reactor Fuel and Coolant
1300
1100
900
T (K)
700
500
300
0 0.2 0.4 0.6 0.8 1
x (BeF2)
Figure 3 Calculated LiF–BeF2 phase diagram from Beneš and Konings25: ◊ experimental data by Roy et al.26; □ data by
Thoma et al.27; and △ data by Romberger et al.28 Reproduced from Beneš, O.; Konings, R. J. M. J. Chem. Thermodyn.
2009, 41, 1086–1095.
1800
1600
1400
T ( K)
1200
1000
800
0.0 0.2 0.4 0.6 0.8 1.0
x (PuF3)
Figure 4 The calculated LiF–PuF3 phase diagram based on the thermodynamic data taken from Beneš and Konings31:
○ experimental data measured by Barton and Strehlow.32 Reproduced from Beneš, O.; Konings, R. J. M. J. Nucl.
Mater. 2008, 377(3), 449–457.
1800
1600
1400
T ( K)
1200
1000
800
0.0 0.2 0.4 0.6 0.8 1.0
x (PuF3)
Figure 5 The calculated NaF–PuF3 phase diagram based on the thermodynamic data taken from Beneš and Konings31:
○ experimental data measured by Barton et al.33 Reproduced from Beneš, O.; Konings, R. J. M. J. Nucl. Mater. 2008,
377(3), 449–457.
1800
1500
1200
T ( K)
900
600
300
0.0 0.2 0.4 0.6 0.8 1.0
x (PuF3)
Figure 6 The estimated BeF2–PuF3 phase diagram. Reproduced from Beneš, O.; Konings, R. J. M. J. Chem.
Thermodyn. 2009, 41, 1086–1095.
system, assuming an ideal behavior of the liquid by Thoma et al.34 The calculated phase diagram is
phase. The estimated BeF2–PuF3 phase diagram is shown in Figure 7. It is a simple eutectic system with
shown in Figure 6, consisting of a single eutectic the eutectic at T ¼ 800 K and x(ThF4) ¼ 0.019.
point at T ¼ 783 K and x(PuF3) ¼ 0.031.
3.13.4.2.6 LiF–AnF4
3.13.4.2.5 BeF2–ThF4 The LiF–ThF4 system is a reference salt for the MSFR
The BeF2–ThF4 system was assessed by van der concept. The equilibrium diagram of the LiF–ThF4
Meer et al.24 using the equilibrium data measured system was reported by Thoma et al.35 on the
368 Molten Salt Reactor Fuel and Coolant
1400
1300
1200
T ( K) 1100
1000
900
800
700
600
0 0.2 0.4 0.6 0.8 1
x (ThF4)
Figure 7 The calculated BeF2–ThF4 phase diagram. Reproduced from van der Meer, J.; Konings, R. J. M.; Jacobs, M. H. G.;
Oonk, H. A. J. J. Nucl. Mater. 2005, 344, 94–99.
1600
1000
900
1400 800
T ( K)
700
600
1200
500
0.0 0.0 0.1 0.1 0.2 0.2
T ( K)
x (UF4)
1000
800
600
0.0 0.2 0.4 0.6 0.8 1.0
x (ThF4)
Figure 8 The equilibrium diagram of the LiF–ThF4 system assessed in Beneš et al.49: ○ thermal analysis data obtained
by Thoma et al.35; □ supercooled data; invariant equilibria as reported in Thoma et al.35 Inset: calculated ThF4–UF4
pseudobinary system with constant amount of LiF at 78 mol%. Reproduced from Beneš, O.; Beilmann, M.; Konings, R. J. M.
J. Nucl. Mater. 2010, 405, 186–198.
basis of thermal analysis and thermal quenching. latter study,24 is shown in Figure 8. The LiF–ThF4 phase
Based on their data, the phase diagram was thermody- diagram consists of four mixed compounds: Li3ThF7,
namically assessed by van der Meer et al.24 and more which melts congruently and Li7Th6F31, LiTh2F9, and
recently by Beneš et al.49 The phase diagram from the LiTh4F17, all melting peritectically. Two eutectic points
Molten Salt Reactor Fuel and Coolant 369
were found at xeut1 ¼ ð22:4 1Þmol% ThF4 with Thoma et al.34 It contains a single eutectic at
Teut1 ¼ ð841 1ÞK, and xeut2 ¼ ð28:3 1Þmol% 1.5 mol% ThF4 and Teut ¼ ð629 3ÞK; no ternary
ThF4 with Teut2 ¼ ð838 1ÞK, the first selected as a compounds were found. van der Meer et al.36 calcu-
fuel composition of the MSFR concept. lated the ternary from the assessed binaries and found
In this notation, AnF4 is represented mainly good agreement with the experimental diagram. The
by ThF4, which serves as a fertile material, and by calculated phase diagram of the LiF–BeF2–ThF4
UF4, which is the fissile material, normally presented system is shown in Figure 9, as a projection of the
with a concentration of up to 4 mol%. As UF4 and liquidus surface.
ThF4 form close-to-ideal solid and liquid solutions, In the MSBR concept, the proposed fuel
the melting point of the fuel is only slightly affected composition in the LiF–BeF2–AnF4 system was 71.7–
by the UF4/ThF4 substitution. The effect of UF4 addi- 16.0–12.3, where the AnF4 fraction was made up of
tion is demonstrated in the inset graph of Figure 8, 12.0 mol% ThF4 and 0.3 mol% UF4. In this section,
which shows the calculated liquidus line (the very AnF4 is represented by pure ThF4, which is possible for
upper line) of the ThF4–UF4 pseudobinary system the same reasons as discussed in Section 3.13.3.1.1.
with the amount of LiF constant at 78 mol%. The If we then assume that the concentration of ThF4
left axis of the graph corresponds to the proposed must be 12.3 mol%, it is possible, according to ther-
LiF–ThF4 (78–22 mol%) fuel composition (eutectic1 modynamic data, to determine the lowest melting
of the LiF–ThF4 system) and the right axis corre- temperature of such a system and its exact composition.
sponds to the LiF–UF4 (78–22 mol%) composition; It has been found at T ¼ 786 K and LiF–BeF2–ThF4
thus, in this case, all ThF4 is substituted by UF4. (67.1–20.6–12.3 mol%) (Composition 1), thus reason-
As can be seen from the figure, the liquidus line along ably close to the data of the MSBR fuel (T ¼ 771 K and
this section is nearly constant, with a total drop of LiF–BeF2–AnF4 (71.7–16.0–12.3 mol%) (Composition
only 18 K. 2)). This means that, keeping the safety margin of 50 K,
the inlet temperature of the reactor must be a mini-
3.13.4.2.7 LiF–BeF2–AnF4 mum of 836 K. This is a promising result because it
The LiF–BeF2–ThF4 system is the reference salt for is lower than the inlet temperature in MSBR, which
a MSR designed as a thermal breeder. The equilib- was found to be 839 K, as discussed in Section 3.13.1.
rium diagram of this system was measured by According to the modeled phase diagram (Figure 9),
ThF4
13
00
12
00
11
00
10
00
90
0
90
10 0
00
11
00
LiF BeF2
Figure 9 Calculated liquid surface of the LiF–BeF2–ThF4 phase diagram. Isotherms are labeled in K with interval of 25 K.
Reproduced from van der Meer, J.; Konings, R. J. M.; Oonk, H. A. J. J. Nucl. Mater. 2006, 357, 48–57.
370 Molten Salt Reactor Fuel and Coolant
the calculated liquidus temperature of the MSBR com- the MOSART reactor.41 Note here that, in order
position (Composition 2) is 795 K. to simplify the study, all actinides were represented
As the melting temperatures of Compositions 1 and by plutonium. This was possible as plutonium is the
2 are very close, we focus (see Table 4) the discussion major constituent of all actinides considered in
only on the preferred composition of the MSBR con- the MOSART fuel. A pseudoternary phase diagram
cept (LiF–BeF2–ThF4 (71.7–16.0–12.3 mol%) (Com- of the LiF–NaF–BeF2–(PuF3 ¼ 1.3 mol%) system is
position 2)). This salt has also been more extensively shown in Figure 10. The melting temperature of
studied, and thus more of its properties are known. the lowest eutectic composition is calculated at
775 K, which is much lower than the designed inlet
3.13.4.2.8 LiF–NaF–BeF2–AnF3 temperature of the MOSART concept6 and therefore
The LiF–NaF–BeF2–PuF3 system is a reference salt in acceptable for reactor purposes.
the MOSART concept. The full thermodynamic The optimized fuel composition as found in
description of this quaternary system has been assessed Beneš and Konings25 varies slightly from that of
in a recent study by Beneš and Konings,25 using the the MOSART concept (LiF–NaF–BeF2–PuF3 (14.8–
solubility data of PuF3 measured by Barton,37 Mailen 57.4–26.5–1.3)). Because the authors of the MOSART
et al.,38 and Ignatiev et al.39,40 for the optimization of the concept did not have a full thermodynamic description
PuF3-containing ternary subsystems. Based on this of the whole LiF–NaF–BeF2–PuF3 system, they took
work,25 the optimized fuel composition is LiF–NaF– the eutectic of the LiF–NaF–BeF2 system with the
BeF2–PuF3 (20.3–57.2–21.2–1.3), which is exactly lowest BeF2 content, as reported in Thoma,42 and
the point that corresponds to the lowest eutectic directly dissolved 1.3 mol% of AnF3 in it. Hence, they
in the LiF–NaF–BeF2–PuF3 system, with a fixed con- did not consider the shift of the eutectic composition
centration of PuF3 at 1.3 mol% as an equilibrium while adding AnF3, which was demonstrated in Beneš
concentration of AnF3 after 10 years of operation of and Konings.25
BeF2
(805)
80 Miscibility gap
0
90
0
10
00
1015
1018
3)
10 (94
00
8)
(85
6)
(89
2) (83
847
10 785
00 858 775
11
00
(1114) (1263)
LiF (917) NaF
Figure 10 Calculated pseudoternary phase diagrams of the LiF–NaF–BeF2 system with constant amount of PuF3 ¼ 1.3 mol%.
Reproduced from Beneš, O.; Konings, R. J. M. J. Fluor. Chem. 2009, 130, 22–29.
Molten Salt Reactor Fuel and Coolant 371
1400
1200
1000
T ( K)
NaF + L
800
600
NaF + b-NaBF4
NaF + a-NaBF4
400
0 50 100
NaBF4 (mol%)
Figure 11 The equilibrium diagram of the NaF–NaBF4 system. Reproduced from Beneš, O.; Konings, R. J. M. J. Fluor.
Chem. 2009, 130, 22–29.
NaF
(1269)
12
00
11
00
A
(991) 1000 1)
10
00 (92
B
11 C
00 726
(1131) (1120)
KF (763) LiF
Figure 12 Calculated liquid surface of the LiF–NaF–KF phase diagram. Isotherms are labeled in K with interval of 25 K.
Primary phase fields: (A) (Li,Na,K)F; (B) (Na,K)F; (C) (Li,Na)F. Reproduced from Beneš, O.; Konings, R. J. M. J. Fluor. Chem.
2009, 130, 22–29.
LiF
0.1
0.9
C
0.8
0.2
A
0.7
0.3
B
Liquid
0.4
0.6
0.5
0.5
0.6
0.4
0.3
0.7
0.2
0.8
0.1
0.9
D
ThF4 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 BeF2
Figure 13 Isothermal plot of the LiF–BeF2–ThF4 phase diagram at T¼ 839 K. Reproduced from Beneš, O.;
Konings, R. J. M. J. Fluor. Chem. 2009, 130, 22–29.
of MSBR), keeping a constant ratio of LiF/BeF2 ¼ at T ¼ 839 K. The straight bold line represents the
0.818/0.182. This ratio corresponds to the fuel com- LiF/BeF2 ratio at 0.818/0.182 within the whole field
position proposed in MSBR. Figure 13 shows the of the diagram, while the ThF4 concentration varies
ternary phase diagram of the LiF–BeF2–ThF4 system from 0 to 100 mol% as it moves from point ‘C’
Molten Salt Reactor Fuel and Coolant 373
towards ‘D.’ The solubility of ThF4 in the LiF–BeF2 solubility in the LiF–BeF2 (67–33) composition has
matrix thus derived is between 9.2 and 20.8 mol%. been calculated for the temperature range of
The interval of the solubility is represented by the ‘A’ 780–930 K, giving:
and ‘B’ signs, respectively, which correspond to the
log10 Q ðmol%Þ ¼ 4:0975 þ 4:32 103 T ðKÞ ½2
intersection of the ‘CD’ line with the surface of the
liquid field.
3.13.4.3.5 PuF3 in molten LiF–NaF–BeF2
3.13.4.3.3 UF4 in molten LiF–ThF4 According to the thermodynamic model of the LiF–
To our best knowledge, there are no experimental data NaF–BeF2–PuF3 system published in Beneš and
of the UF4 solubility in the LiF–ThF4 binary matrix. Konings,25 the solubility of PuF3 in the recom-
However, based on the thermodynamic assessment mended fuel matrix composition (LiF–NaF–BeF2
of the LiF–NaF–ThF4–UF4 system,49 the solubility (20.6–57.9–21.5)) was calculated for the temperature
of UF4 in the LiF–ThF4 (78–22) composition (primary range of 823–973 K and fitted with the polynomial
fuel choice of the MSFR concept) has been calculated equation below:
for a temperature range of 840–880 K giving:
log10 Q ðmol%Þ ¼ 5:3526 þ 9:7386 103 T ðKÞ
log10 Q ðmol%Þ ¼ 42:7475 0:1052T ðKÞ
3:4105 106 T 2 ðKÞ ½3
þ 6:6086 105 T 2 ðKÞ ½1
Based on this equation, the total PuF3 solubility
3.13.4.3.4 PuF3 in molten LiF–BeF2 in the LiF–NaF–BeF2 (20.6–57.9–21.5) melt at the
The solubility of PuF3 in the LiF–BeF2 binary melt inlet temperature of the MOSART reactor concept
has been measured by Barton22 and Mailen et al.38 (T ¼ 873 K) is 3.55 mol%. This value is slightly
Barton measured the PuF3 solubility in LiF–BeF2 higher than the measured value in the MOSART
(71.3–28.7) and LiF–BeF2 (63–37) compositions matrix composition (LiF–NaF–BeF2 (15–58–27)),
for the temperature range of 736–927 K, whereas which was determined to be 3.08 mol%.6 Higher
Mailen et al. measured the PuF3 solubility in LiF– solubility was achieved in the former case because
BeF2 (67–33) composition for the temperature range of the lower content of BeF2, which is the main fuel
of 59–657 K. Furthermore, Barton measured the component responsible for low AnF3 solubility, as
PuF3 solubility at T ¼ 838 K in the LiF–BeF2 matrix discussed in Beneš and Konings.25
as a function of composition from x(LiF) ¼ 0.52 to
0.72. Beneš and Konings25 recently evaluated the 3.13.4.3.6 PuF3 in molten LiF–BeF2–ThF4
LiF–NaF–BeF2–PuF3 system thermodynamically The solubility of PuF3 in various compositions of
and found very good agreement with all experimen- LiF–ThF4 and LiF–BeF2–ThF4 melts were measured
tally determined solubility data by Barton and Mailen by Sood et al.,50 between 783 and 1060 K. Results
et al. On the basis of their assessment, the PuF3 of their measurements are reported in Table 5,
showing the derived coefficients for the general The agreement between the studies is excellent, as
equation: shown in Figure 14 in an isothermal section at
873 K. From the results, we interpolate for the 66–34
log10 Q ðmol%Þ ¼ A þ B=T ðKÞ ½4
composition:
ðmPa sÞ ¼ 0:116expð3755=T ðKÞÞ ½6
3.13.4.4 Density and Viscosity
3.13.4.4.1 LiF–BeF2
3.13.4.4.2 LiF–AnF4
The density of liquid LiF–BeF2 has been measured
The density of LiF–ThF4 mixtures was measured by
by Blanke et al.51 from 0 to 55 mol% BeF2, by Cantor
Porter and Meaker57 and Hill et al.58 The data are in
et al.52 for 50.2, 74.9, and 89.2 mol% BeF2, and by
good agreement and clearly indicate a linear depen-
Cantor53 for the 34 mol% BeF2 composition. As dis-
dence of the molar volume on composition as shown
cussed by van der Meer et al.,36 the molar volumes
in Figure 15, confirming ideal behavior. The density
derived from the measured density data indicate
of the 78–22 composition as measured by Porter and
ideal behavior, suggesting that the density can be
Meaker,57 is given by:
interpolated from the molar volume data for the
pure components. However, the density and molar rðkg m3 Þ ¼ 5543 1:25T ðKÞ ½7
volume of liquid BeF2 are known only at a single
temperature (T ¼ 1073 K), and not at all as a function The density of liquid LiF–UF4 mixtures were
of temperature. Therefore, we have selected the measured by Blanke et al.51 and Porter and Meaker.57
results of the 66:34 composition from Cantor53: The results are in excellent agreement, as shown in
Figure 15. The results indicate a linear dependence
rðkg m3 Þ ¼ 2146:3 0:4884T ðKÞ ½5 of the molar volume on composition, confirming
The viscosity of liquid LiF–BeF2 has been measured ideal behavior.
by Cohen and Jones54 and Abe et al.55 for the compo- The viscosity of LiF–ThF4 mixtures was measured
sitions 31and 32.8 mol% BeF2, respectively, as well as by Chervinskij et al.59 from 0 to 100 mol% ThF4. The
by Blanke et al.,51 Cantor et al.,52 and Desyatnik et al.56 results reveal a strong deviation from ideal behavior
for a wide(r) range of compositions and temperatures. around the eutectic composition. An isothermal sec-
tion in Figure 15 shows a steady increase from LiF
to ThF4. The viscosity of the 78–22 composition
interpolated from the results is given by:
5
ðmPa sÞ ¼ 0:365expð2735=T ðKÞÞ ½8
The viscosity of the LiF–UF4 system measured by the
3 same group60 shows a less strong increase with the AnF4
content compared to ThF4 (Figure 15). As a result, the
above equation probably overestimates the viscosity
log h (Pa s)
3.13.4.4.3 LiF–BeF2–ThF4
The densities of the three compositions of the
-1 LiF–BeF2–ThF4 system, with almost constant
LiF concentration, were measured by Cantor.53
B Unfortunately, the density of the LiF–BeF2–ThF4
(71.7–16.0–12.3) composition has not been measured;
-3 however, a very close composition (LiF–BeF2–ThF4
0.00 0.20 0.40 0.60 0.80 1.00
x (BeF2)
(70.06–17.96–11.98)) has been determined and the
corresponding density function is given below:
Figure 14 The viscosity of liquid LiF–BeF2 at 873 K:
▽ Cohen and Jones54; △ Blanke et al.51; □ (curve A), rðkg m3 Þ ¼ 4043:9 0:8064T ðKÞ ½9
Cantor et al.52; ○ (curve B), Desyatnik et al.56;
◊ Abe et al.55 Reproduced from Beneš, O.; Konings, R. J. M. It has been shown by van der Meer and Konings61 that
J. Fluor. Chem. 2009, 130, 22–29. the molar volumes and thus the densities of all three
Molten Salt Reactor Fuel and Coolant 375
50 25
An = Th
20
40
Vm (cm3 mol−1)
An = U
15
h (mPa s)
An = Th
30
10
An = U
20
5
10 0
0.00 0.20 0.40 0.60 0.80 1.00 0.00 0.20 0.40 0.60 0.80 1.00
x (AnF4) x (AnF4)
Figure 15 The molar volume (left) and viscosity (right) of liquid LiF–ThF4 and LiF–UF4 at 1273 K. Right figure: ○ data by
Chervinskij et al.59; □ data by Desyatnik et al.60; Left figure: ▲ data by Hill et al.58; ▼ data by Porter and Meaker57; □ data
by Blanke et al.51; ○ data by Porter and Meaker.57 Reproduced from Beneš, O.; Konings, R. J. M. J. Fluor. Chem. 2009,
130, 22–29.
LiF–BeF2–ThF4 compositions measured in Cantor53 reported density at the same temperature taken from the
behave almost ideally. Based on this triplet of data and study by Briant and Weinberg63 is 3250 kg m3: values
with the assumption of the ideality, it is possible to that are in close agreement.
estimate the density function of temperature of pure The viscosity of liquid LiF–BeF2–ThF4 of two
BeF2, which has not been measured yet. The density of compositions was measured by Cantor.53 The viscos-
liquid BeF2 was measured by Mackenzie,62 but only ity of the quaternary LiF–BeF2–ThF4–UF4 (71–16–
at 1073 K, obtaining the value of 1947 10 kg m3. 12–1) composition, which is nearly identical to
Cantor et al.52 also measured the density, but, due to our reference selection (LiF–BeF2–ThF4 (71.7–16–
the experimental difficulties, they derived only an 12.3)), has been reported in Powers et al.64 for the
approximate value: 1960 kg m3 at 1123 K. The value temperature range of 873–1073 K, giving:
of MacKenzie is recommended and taken as a constraint
in our estimation. The obtained density for liquid BeF2 ðmPasÞ ¼ 0:062expð4636=T ðKÞÞ ½12
as a function of temperature is shown below:
1.0 2100
2050
0.9
Density 2000
0.8
1950
0.7
log10 h (mPa s)
1900
r (kg m–3)
0.6 1850
1800
0.5
Viscosity
1750
0.4
1700
0.3
1650
0.2 1600
750 800 850 900 950 1000 1050 1100 1150 1200
T ( K)
Figure 17 Viscosity and density functions of temperature reported by Chrenková et al.71 (– – –) and Powers et al.64 (——).
For comparison, the ideal density behavior is represented by a dotted line. Reproduced from Beneš, O.; Konings, R. J. M.
J. Fluor. Chem. 2009, 130, 22–29.
slope, which is consistent with our observations that temperature. Some time later, Cooke et al.74 reported
most of these fluoride systems are ideal. For this reason, more detailed results, indicating that the thermal con-
we recommend the data by Chrenková et al.: ductivity increases slightly, from l ¼ 1.0 W m1 K1
at 923 K, to about 1.2 W m1 K1 between 1023 and
rðkg m3 Þ ¼ 2579:3 0:6240T ðKÞ ½19 1133 K. Kato et al.75 measured the thermal diffusivity
The viscosity of the eutectic melt of the LiF–NaF–KF of the compositions 66–34 mol% and 53–47 mol%.
system has been measured by Chrenková et al.71 for From their results, we calculate 1.1 W m1 K1 for
the temperature range of 773–973 K and Powers et al.64 the 66–34 mol% composition, which is in good agree-
for the temperature range of 873–1073 K. The com- ment with Cooke’s results, and we recommend l
parison between the data by Chrenková et al. and (LiF–BeF2 (66–34)) ¼ 1.1 W m1 K1.
by Powers et al. is shown in Figure 17. The data by
Chrenková et al. have been selected: 3.13.4.5.2 LiF–AnF4
To our best knowledge, heat capacity or thermal
log10 ðmPa sÞ ¼ 1:6044 þ 1944=T ðKÞ ½20 conductivity have not been measured for the LiF–
ThF4 system. We have estimated the heat capacity of
the LiF–ThF4 (78–22 mol%) composition on the
3.13.4.5 Heat Capacity and Thermal
basis of the comparison between the ideal heat capac-
Conductivity
ity and the measured data from other fluoride sys-
3.13.4.5.1 LiF–BeF2 tems taken from Powers et al.64 The average positive
The heat capacity of liquid LiF–BeF2 (66–34 mol%) deviation from the ideal behavior has been found to
has been measured by Hoffman and Cooke (as cited be 11%. If we combine this difference with the ideal
in Cantor et al.72), and Douglas and Payne,73 who heat capacity of the LiF–ThF4 (78–22 mol%) compo-
obtained 2.41 J K1 g1 (unknown temperature range) sition, we obtain our suggested value:
and 2.37 J K1 g1 (773–873 K), respectively. The Cp ¼ 1.0 J g1 K1.
value Cp(LiF–BeF2 (66–34 mol%)) ¼ 2.39 J K1 g1 There are not enough data to accurately esti-
has been selected. mate the thermal conductivity of the LiF–ThF4
The thermal conductivityof LiF–BeF2 (66–34 mol%) (78–22 mol%) composition; however, we suggest
has been measured by Cooke (as reported in Cantor that the value be slightly higher than the value of the
et al.72) to be 1.0 W m1 K1, independent of the LiF–BeF2 (66–34 mol%) composition and close to
378 Molten Salt Reactor Fuel and Coolant
the value for LiF–BeF2–ThF4 (71.7–16–12.3 mol%) sources74,76 indicate that in the measured composition
composition, which was derived for T ¼ 1023 K range, the thermal conductivity decreases with increas-
(see Section 3.13.4.5.3). Our suggested value ing (BeF2 þ MF4) content as indicated in Figure 18.
for LiF–ThF4 (78–22 mol%) composition is The LiF–BeF2–ThF4 (71.7–16.0–12.3 mol%) compo-
l ¼ 1.5 W m1 K1. sition is just outside this range (x(BeF2 þ MF4) ¼
28.3 mol%), and linear extrapolation would suggest
3.13.4.5.3 LiF–BeF2–ThF4 l ¼ 1.51 W m1 K1 at T ¼ 1023 K (solid line in
Araki and Kato76 measured the thermal diffusivity of Figure 18). However, such linear extrapolation
liquid LiF–BeF2–ThF4 (64–18–18 mol%), from would suggest a relatively high thermal conductivity
which they derived the thermal conductivity using of LiF–ThF4 (78–22 mol%). Alternatively, one could
their heat capacity data and an estimated density. The extrapolate the results in a nonlinear way (dashed line
results indicate an almost constant value in the tem- in Figure 18). This would suggest l ¼ 1.49 W m1
perature range of 850–1000 K: 0.95–0.98 W m1 K1. K1 at T ¼ 1023 K, which is very close to previously
The recommended heat capacity according to established value. In this case, the thermal conductivity
Araki and Kato is Cp ¼ 1.23 J g1 K1. Both data, of LiF–ThF4 (78–22 mol%) is 1.6 W m1 K1, which is
heat capacity and thermal conductivity, are measured more realistic. For LiF–BeF2–ThF4 (71.7–16.0–12.3)
for a LiF–BeF2–ThF4 composition that is slightly composition we recommend:
different from the one considered in this work l ¼ 1:5 Wm1 K1 ½21
(71.7–16.0–12.3 mol%). Cooke et al.74 reported (in
graphical form only) the thermal conductivity of The heat capacity of the quaternary LiF–BeF2–ThF4–
liquid LiF–BeF2–ThF4–UF4 (67.5–20–12–0.5 mol%) UF4 (71–16–12–1 mol%) composition, which is
for the temperature range of 800–1150 K. The data nearly identical to our reference selection (LiF–BeF2–
scatter around l ¼ 1.2–1.4 W m1 K1, with a sug- ThF4 (71.7–16–12.3 mol%)), has been reported in
gested maximum at 973 K. This result is somewhat Briant and Weinberg,63 giving Cp ¼ 1550 J kg1 K1.
different from that of Araki and Kato.76 As the results This value is also fairly close to the estimated value,
for liquid LiF–BeF2 from both groups are in good based on the approach published by Khokhlov et al.65
agreement, the variation probably arises from differ- (discussed in the following section), which gives
ences in BeF2 and MF4 content (where M ¼ Th, U, Cp ¼ 1.506 J g1 K1. We select the measured value,
and Zr). The results from the above-mentioned Cp ¼ 1.550 J g1 K1.
1.6
1.4
l ( W m-1 K-1)
1.2
1.0 T = 1023 K
0.8
0.28 0.30 0.32 0.34 0.36
x (BeF2 + MF4)
Figure 18 Extrapolation of the thermal conductivity of the LiF–BeF2–ThF4 (71.7–16.0–12.3 mol%) composition at
T ¼ 1023 K. (——) linear fit; (– – –) polynomial fit. (▪) Experimental data from Cooke et al.74 and Araki and Kato.76
Reproduced from Beneš, O.; Konings, R. J. M. J. Fluor. Chem. 2009, 130, 22–29.
Molten Salt Reactor Fuel and Coolant 379
LiF–ThF4 (78–22 mol%) composition has been cal- where the total vapor pressure is highlighted by a
culated for the temperature range between 839 and bold curve, whereas the most volatile species are
1473 K. The result is given in the equation below: reported by thin lines. The graph does not include
Pu containing species because even the most volatile
log10 pðPaÞ ¼ 11:90212 989=T ðKÞ ½28
among these, PuF4, has a much lower pressure than
The vapor pressure of the LiF–ThF4–UF4 (78–18– the species reported, and therefore they have been
4 mol%) composition is slightly lower compared to a excluded from the figure. The total vapor pressure is
system with no UF4 content. The calculated boiling represented by the following equation:
temperature of the LiF–ThF4 (78–22 mol%) compo- log10 pðPaÞ ¼ 11:6509 12 827=T ðKÞ ½30
sition is T ¼ 1874 K.
which gives p ¼ 0.001 Pa and p ¼ 0.046 Pa at the
3.13.4.6.3 LiF–BeF2–ThF4 designed inlet temperature (Tinlet ¼ 873 K) and the
According to the thermodynamic data by van der outlet temperature (Toutlet ¼ 988 K) of the MOSART
Meer et al.,36 the vapor pressure of the LiF–BeF2– reactor,6 respectively. Both values are very low, and
ThF4 (71.7–16.0–12.3 mol%) composition has been hence the composition shift of the fuel as a consequence
calculated for the temperature range of 823–1473 K of the incongruent vaporization can be neglected.
and the obtained result is shown in the following The calculated boiling temperature is T ¼ 1973 K.
equation:
3.13.4.6.5 NaF–NaBF4
log10 pðPaÞ ¼ 11:158 10 790:5=T ðKÞ ½29
The vapor pressure of BF3 in the NaF–NaBF4 system
The calculated boiling temperature of the LiF– has been measured by Cantor et al.78 They measured
BeF2–ThF4 (71.7–16.0–12.3 mol%) composition is the equilibrium of the BF3 gaseous species over the
T ¼ 1744 K. melt for the composition range of 5–100 mol%
NaBF4 and the temperature range of 698–1473 K.
However, in their report they ‘only’ show the results
3.13.4.6.4 LiF–NaF–BeF2–AnF3
for 900, 1000, and 1100 K. Based on this triplet of
In the study by Beneš and Konings,25 the vapor pres-
data, the vapor pressure equation of NaF–NaBF4
sure of the potential fuel composition (LiF–NaF–
(8–92 mol%) has been determined, giving:
BeF2–PuF3 (20.3–57.1–21.2–1.3 mol%)) has been
calculated, and the results are reported in Figure 19, log10 pðPaÞ ¼ 11:6386550:6=T ðKÞ ½31
1E - 4
tal
To
1E - 5
F2
Na 2
pvapor (atm)
1E - 6 F2
Li 2
F 2
Be
F3
1E - 7 Li 3
eF
3
F LiB
Li
1E - 8 F
Na
1E - 9
900 1000 1100 1200 1300
T ( K)
Figure 19 Calculated vapor pressure of the x(LiF) ¼ 0.203, x(NaF) ¼ 0.571, x(BeF2) ¼ 0.212, x(PuF3) ¼ 0.013 potential
fuel composition. Reproduced from Beneš, O.; Konings, R. J. M. J. Chem. Thermodyn. 2009, 41, 1086–1095.
Molten Salt Reactor Fuel and Coolant 381
log10 pðPaÞ ¼ 10:74810 789=T ðKÞ ½32 Whether Pa2O5 will precipitate or not depends on
three factors: oxide and protactinium concentrations,
and the oxidation state of the fuel, which, in the MSR,
3.13.5 Role of Oxygen Impurities is controlled by the UF4/UF3 ratio, as discussed in
Section 3.13.8. As reported in Rosenthal et al.,80 with
In the previous section, the physicochemical pro- 100 ppm Pa and 30 ppm oxide present, the UF4/UF3
perties of pure fluoride salts have been discussed. ratio must be at least 105 in order to start the Pa2O5
However, the behavior of these systems can be sig- precipitation. Nevertheless, such oxidizing conditions
nificantly affected by the presence of the oxide are easily avoided, as the typical UF4/UF3 ratio in the
ion that might be resulting from contamination of MSR is set to 100 (see Section 3.13.8).
the salt system; for example, the presence of reactive Even stronger oxidizing conditions (UF4/UF3 > 108)
oxides such as H2O can result in precipitation of are required to precipitate PuO2, and hence this species
the UO2 phase.80 Therefore, the effect of added is avoided in the MSR fuel as well.
oxide on the fuel mixture containing LiF, BeF2, Although the Pa2O5 and PuO2 species will
ThF4, UF4, and PaF4 has been investigated in sev- not be formed in the fuel salt, the other actinide
eral studies,81–88 as reported in Rosenthal et al.80 who dioxides UO2, ThO2, and PaO2 can be formed
give a summary of the main conclusions from these under the redox conditions of the MSR and, due
works is given. It has been found that the solubility to the very low solubilities of these species in the
of the actinide dioxides in the MSBR fuel salt is low fluoride matrix (as given by eqns [33]–[36]), they can
and it decreases in the order, ThO2, PaO2, UO2, and easily precipitate in the solid form. Therefore, it is
PuO2. The temperature functions of the solubilities important to keep the fuel salt free from any oxide
of these oxides were estimated in the same study as contamination to avoid this inadvertent event. This
follows: will certainly require some care but, as mentioned in
Rosenthal et al.,80 the results of the MSRE project
log10 QThO2 ¼ 2:86 3280=T ðKÞ ½33
have shown that the oxide content can be maintained
at an adequately low level in order to achieve suc-
log10 QPaO2 ¼ 2:86 4920=T ðKÞ ½34 cessful long-term operation of the MSR.
log10 QPuO2 ¼ 2:86 7100=T ðKÞ ½36 Surprisingly, very little experimental work has been
done on the electrochemical properties of the main
where ions in molten fluoride salts. For the LiF–BeF2 system,
QMO2 ¼ xM4þ xO2 2 ½37 some direct measurements of the standard potentials
have been made. The standard potentials of the main
The ThF4 concentration in the MSBR concept is ions in the liquid LiF–BeF2 (67–33) melt have been
equal to x ¼ 0.12, and it has been shown80 that at reported by Baes.89–91 He has made an extensive anal-
such concentrations of thorium, the ThO2 precipita- ysis of the available literature, which is essentially
tion at T ¼ 773 K will start for xO2
8104. based on a comparative scale as only the Be2þ/Be0
Protactinium is produced in thorium-containing couple has been measured electrochemically92:
breeder fuel by neutron capture, and both tetravalent
BeðcrÞ þ 2HFðgÞ ¼ BeF2 ðslnÞ þ H2 ðgÞ ½I
and pentavalent species of protactinium are stable.
Thus, in addition to PaO2, Pa2O5 can precipitate Using equilibrium constants, Gibbs energies of the
in the oxide form. As reported in Rosenthal et al.,80 solutes, and activity coefficients, Baes derived the
382 Molten Salt Reactor Fuel and Coolant
Table 6 Standard potential in LiF–BeF2 (66–34) relative Table 7 Standard potential in LiF–CaF2 (77–23) relative to
to the HF(g)/H2 couple, E/V ¼ a þ bT (K) the F2/F pair measured by Chamelot et al.93 at T ¼ 1100 K
F2 buildup in the reactor is eliminated.95 Because the in the Ni-based alloys, see Section 3.13.9) from
MSR operates at high temperatures, the recovery the Hastelloy-N,83 is inhibited. As reported by
process is rapid and radiation damage to the salt is Rosenthal et al.,80 the UF4/UF3 ratio in the MSRE
very small. This has been confirmed in separate was 100. It is shown in Figure 21 that at this
experiments, using accelerators, and in in-pile tests concentration the ratio of dissolved chromium in
for the ARE and MSRE projects. None of these experi- the form of CrF2 and its metal form is <105.
ments have revealed indications that the fluoride salts Taking into account that the UF4/UF3 ratio is set
are unstable in radiation fields.8,95 It is believed that in such way not to form chromium fluoride, one
this radiation stability is responsible for the demand can assume that fluorides that have more negative
that only very stable salts must be considered in the free energy of formation DGf0 according to the
reactor in order to keep the construction alloys ther- general reaction:
modynamically stable with respect to the salt.
xMmetal þ yF2gas ! xMF2y DGf0 ½II
will dissolve in the fuel, whereas the ones that have
3.13.8 Fission Product Behavior higher DGf0 of the above given reaction will precipi-
tate in the form of insoluble metals.
The fission products that are formed during the During the operation of the MSR, free fluorine is
operation of the MSR can be divided into three formed from the fission processes. This fluorine pref-
main groups based on their solubilities in the carry- erably reacts with UF3, increasing the UF4/UF3 ratio
ing matrix: noble gases, stable salt-soluble fluorides, and thus changing the redox potential of the
and noble metals that are very difficult to dissolve in salt. This will certainly increase the corrosion rate
the fluoride matrix. Whether the fission product will of the structural material; therefore, the UF3 concen-
or will not be dissolved by the salt is determined by tration must be readjusted. This is achieved by adding
the redox potential of the salt. As demonstrated in small amounts of pure metals, for example, beryllium,
the MSRE project, the redox potential of the salt is which absorb fluorine. In the MSRE, a beryllium
controlled by the UF4/UF3 ratio in such way that rod was kept immersed in the salt until the UF3
the corrosion of the structural material, for example, concentration reached the correct value. On the
leaching of chromium (the least stable element other hand, as discussed in Section 3.13.9, too high
100
10–1
10–2
e
/F
MFx / M mole ratio
10–3
2
F
i
/N
Fe
r
/C
2
F
Ni
2
10–4
F
Cr
10–5
Mo
10–6
F3 /
Mo
10–7
10–8
100 101 102 103 104 105 106 107
UF4 / UF3 mole ratio
Figure 21 Variation of equilibrium concentration of structural metal fluorides as a function of the UF4/UF3 ratio in a molten
salt reactor fuel. Reproduced from Rosenthal, M. W.; Haubenreich, P. N.; Briggs, R. B. Tech. Rep. ORNL-4812; 1972.
384 Molten Salt Reactor Fuel and Coolant
a content of UF3 is not allowed in the MSR as it can and Xe in NaF–ZrF4 (53–47 mol%) and NaF–ZrF4–
corrode graphite or other structural components. UF4 (50–46–4 mol%); Blander et al.97 measured the
Hence, one must be able to control the redox poten- solubility of He, Ne, and Ar in LiF–NaF–KF eutectic;
tial of the salt in the other direction also to achieve and Watson et al.98 measured the solubility of He, Ne,
more oxidizing conditions, and thus increasing Ar, and Xe in LiF–BeF2 (64–36 mol%).
the UF4/UF3 ratio. In the MSRE, this was done by These studies all show that the solubilities
adding NiF2 into the fuel. At the redox conditions increase linearly with gas pressure and temperature
of the MSR, NiF2 decomposes to metallic nickel and decrease with increasing atomic weight of the gas
and fluorine, which then reacts with UF3 forming atom. All experimental data confirm that the solubil-
more UF4 (see Chapter 5.10, Material Perfor- ity of these noble gases follows Henry’s Law, which
mance in Molten Salts). says that in very dilute solutions, the pressure is
proportional to the mole fraction:
3.13.8.1 Noble Gases pi ¼ kH xi ½40
The solubility of inert gases in molten fluoride salts The proportionality factors kH derived from the
has been measured by a limited number of authors. experiments for LiF–BeF2 (64–36 mol%) are plotted
Grimes et al.96 measured the solubility of He, Ne, Ar, in Figure 22 as a function of the reciprocal tempera-
ture and listed in Table 9.
1E − 06
As is obvious from the table and the figure, the
noble gases are only slightly soluble in molten salts.
They can be removed from the fuel by sparging with
helium to an off-gas system. As reported in Engel
et al.,99 it was demonstrated during the operation of
the MSRE that about 80% of 135Xe, the highest noble
kH (mol cm− 3 bar− 1)
1E − 07 He
gas neutron poison, was removed using this method.
It was observed in Rosenthal et al.80 that no com-
Ne
pounds are formed with noble metals under the
MSR conditions, which prevents the noble metals
1E − 08 from being chemically bonded to the fuel salt.
Ar
3.13.8.2 Salt-Soluble Fission Products
Xe
Alkali metals (mainly Rb, Cs), alkali-earth (mainly
1E − 09 Sr, Ba), the lanthanides, and Y with Zr, all form
8 10 12 14 16
104/T (K)
stable fluorides and are soluble in the fuel salt.
Hence, they are expected to be found completely
Figure 22 The Henry’s constant kH of noble gases as a dissolved in the fuel, except for the ones that have
function of the reciprocal temperature in molten LiF–BeF2
(64–36). Reproduced from Watson, G. M.; Evans, R. B.;
noble gas precursors with relatively long half-lives
Grimes, W. R.; Smith, N. V. J. Chem. Eng. Data 1962, 7, and these are removed prior to their decay by the
285–287. off-gas system. It was demonstrated experimentally
Table 9 The Henry’s constant kH of noble gases in molten LiF–BeF2 (64–36 mol%) as a function of the temperature
He Ne Ar Xe
Source: Watson, G. M.; Evans, R. B.; Grimes, W. R.; Smith, N. V. J. Chem. Eng. Data 1962, 7, 285–287.
Molten Salt Reactor Fuel and Coolant 385
that these isotopes with noble gas precursors (89Sr 3.13.8.4 Iodine
and 137Cs) show ratios to the calculated inventory
A separate section is devoted here to iodine, as it does
that are significantly lower than those without, which
not fit into either of the above-mentioned categories.
generally scatter around 1.0.80 Although an online
At the redox conditions of the MSR, iodine is kept
cleanup will be made for the MSR fuel, some of
dissolved in the fuel salt in the form of the I anion.
the fission products from this group will remain dis-
From the thermodynamic point of view, it would be
solved in the salt during the lifetime of the reactor.
necessary to have a UF4/UF3 ratio of at least 104 in
However, as reported, for example, in the MSR FUJI
order to strip 0.1% of I as I2 in the off-gas system.100
concept,4 for a thermal spectrum reactor, the accu-
Such conditions are highly oxidizing and are not
mulation of these fission products will be relatively
allowed as the rate of chromium leaching from the
small, within the tenths of wt%, mostly represented
Hastelloy-N would be too high. Based on the MSRE
by Zr and lanthanides, followed by cesium. Such low
observations, the experimentally obtained inventories
concentrations will only negligibly affect the fuel
of iodine found in the fuel salt mostly ranged some-
properties, as demonstrated in a recent study by
where between 30% and 60% of the calculated ones.
Beneš and Konings.31
Why such relatively small amounts were observed
is not clear, but one of the possible explanations
3.13.8.3 Insoluble Fission Products is that the 131Te precursor of iodine with a half-life
of 25 min had been stripped from the fuel before it
This group of fission products is mostly represented
decayed to 131I or deposited on the surfaces of the
by noble metals that are more thermodynamically
reactor vessels.100
stable to oxidation than chromium in the structural
alloy, at the redox conditions maintained in the
MSR. Therefore, they are expected to be found in
the fuel in the metal form. Noble metals that will
3.13.9 The Effect of Corrosion
accumulate in the MSR in significant amounts are
Reactions on the Fuel Behavior
isotopes of Nb, Mo, Tc, Ru, Ag, Sb, and Pd. As
Corrosion of the structural materials of the reactor by
confirmed by the MSRE,9,80,100 selenium and tellu-
the molten salt may affect the chemistry and redox
rium are also expected to be present in the reactor
conditions of the fuel. The corrosion process to be
circuit in metal form; however, as discussed in
considered in this context is the oxidation of metals in
Section 3.13.9, the reduction of tellurium in the
the alloy phase and their dissolution in the salt phase.
salt to telluride can be achieved by increasing the
From the thermodynamic point of view, Cr is the least
UF3/UF4 ratio to about 0.05.101
stable element in relevant Ni-based alloys such as
As the noble metals are insoluble in the fluoride
Hastelloy-N, as can be seen in Figure 23. It is not useful
salt, it is important to understand the state of
to apply a protective coating of, for example, an oxide
these fission products within the fuel circuit. Based
layer on the structural material for the two following
on the MSRE samples,101 some metallic particles
reasons: although some oxides are relatively insoluble
were found in the helium sparge gas, some were
in the fluoride melt, most are readily dissolved, and all
deposited on the metallic surfaces of the primary
rapidly recrystallize.101 Therefore, the clean metal must
circuit, and a smaller fraction was found deposited
withstand the corrosive attack of the salt.
on the graphite specimens. The aim is to avoid the
As discussed by Briggs,102 three principal corro-
precipitation of these metallic fission products on
sion processes can be distinguished:
the structural materials as the deposited material
contributes to the heat generation from its decay Reactions due to oxides on the metal: Oxide films
even after the reactor shutdown, decreasing the on the surface of the alloy can be attacked by the
overall safety of the MSR. Furthermore, the depos- fluoride melt:
its on the graphite specimens in the case of ther-
Cr2 O3 þ3BeF2 ðsaltÞ ¼ 2CrF3 ðsaltÞþ3BeO ½III
mal reactors absorb neutrons during the operation
and lead to lower efficiency of the reactor. One
FeOþBeF2 ðsaltÞ ¼ FeF2 ðsaltÞþBeO ½IV
way of avoiding this detrimental precipitation is
to remove the insoluble fission products by helium
bubbling before they interact with the structural The formed oxides are of little consequence
materials. as long as they do not contain fissile elements, but
386 Molten Salt Reactor Fuel and Coolant
CrF2
to 85 ppm.104 This is lower than the approximate equi-
librium concentration. Also, the observed rates of cor-
-800 rosion in the MSRE have been significantly lower than
UF4
predicted from thermodynamic data and diffusion the-
BeF2 ory. It has been postulated that one of the principal
ThF4
reasons for the unexpectedly low values observed is
-1000 PuF3 LiF
that the metal surfaces of the fuel circuit have been
covered with a film of the noble-metal fission products
Nb, Mo, Tc, and Ru about 10 Å thick.104 Such low Cr
-1200 concentrations in the fuel will not have a large impact
300 600 900 1200 1500 on the properties of the fuel salt.
Temperature (K)
A special paragraph must be devoted to the influ-
Figure 23 The stability of selected metal fluorides. ence of the fission product tellurium on the corrosion
aspects of the fuel salt. When it is present in the metallic
the formed fluorides will act as oxidants of Cr in form in the fuel, it corrodes Ni-based alloys and
the alloy. embrittles its surface grain boundaries. This embrittle-
ment is a vital issue to consider because, in the long
Reactions with dissolved impurities: In addition to term, it can result in cracking of the structural materials
impurities generated by reactions with oxides, (see Chapter 5.10, Material Performance in Molten
impurities may result from incomplete removal of Salts). As reported in Keiser,105 an investigation using
HF or of easily reducible fluorides during the salt chromium telluride as tellurium source was carried out
purification, which is done by successive reactions in order to understand the mechanism of this type of
with HF–H2 and H2. In this case, the following corrosion. It was concluded that the intergranular
reactions should be considered: embrittlement produced in Hastelloy-N can be signif-
CrðalloyÞ þ 2HFðsaltÞ ¼ CrF2 ðsaltÞ þ H2 ðgÞ ½V icantly reduced by adding 1–2 wt% of niobium into the
Hastelloy-N. Another way to suppress the corrosion
CrðalloyÞ þ FeF2 ðsaltÞ ¼ CrF2 ðsaltÞ þ Fe ½VI rate is by increasing the reducing conditions of the fuel,
which is done by the increase of the UF3/UF4 ratio
Reactions with necessary constituents of the melt: (done in the MSRE by adding Be metal into the fuel).
At such conditions, tellurium would be present as Te2
The oxidation reaction of alloy components with
anion rather than in an ‘oxidized’ metallic form, and
UF4 has been suggested to play an important role
this is less aggressive against the nickel-based alloys.
in the corrosion103:
On the other hand, one must be cautious because
CrðalloyÞ þ 2UF4 ðsaltÞ setting the UF3/UF4 ratio too high will result in some
parasitic reactions of the UF3 with graphite (only in
¼ CrF2 ðsaltÞ þ 2UF3 ðsaltÞ ½VII
case of thermal-moderated reactor) and possibly with
other materials within the primary reactor circuit.101
Equation [VII] will define/buffer the electrochemi-
cal potential of the salt via the UF4/UF3 ratio:
3.13.10 Summary and Future Work
aCrF2 aUF
2
ðgxÞCrF2 ðgxÞ2UF3
K47 ¼ 3
¼ ½41
aCr aUF
2
4
ðgxÞCr ðgxÞ2UF4 The MSR is a very promising concept for the fu-
ture needs of nuclear energy. It can be designed as
where a is the activity and g is the activity coefficient. a thermal (graphite-moderated) or nonmoderated
Briggs102 showed that, based on reasonable estima- breeder reactor that produces more fuel from
232
tions of the activity coefficients for the above reac- Th than it consumes, or as an actinide burner
tion, the equilibrium concentration of CrF2 is of the to manage the plutonium and the long-lived minor
Molten Salt Reactor Fuel and Coolant 387
actinides (Np, Am, Cm) contained in significant bromides, and iodides, more measurements are
amounts in the nuclear waste coming from current needed in order to justify his approach.
nuclear power plants. For a thermal spectrum reac-
tor, the ideal candidate for the fuel matrix is a Attention must be also given to the corrosion of the
mixture of 7LiF and BeF2 because of its very low fuel salt against the structural material. It has been
neutron capture cross-section; for the nonmoder- reported in Sections 3.13.8 and 3.13.9 that the cor-
ated spectrum, other fluorides, for example, NaF, rosion rate in MSRE was maintained at a very low
can be considered. level during the operation of the reactor, mainly by
A critical review of physicochemical properties controlling the redox potential of the salt via the UF4/
of the primary fuel choices of various MSR designs UF3 ratio and also by frequent fission product removal;
(see Table 2) has been done in Section 3.13.4. From however, the outlet temperature of the MSRE was
the evaluated data, we can conclude the following: 927 K, which is rather low for current interests. The
demand for a higher temperature regime leads to a
The thermodynamic description of the fuel sys- higher corrosion rate of the Ni-based alloys (e.g.,
tems is available and, based on these results, the Hastelloy-N used in the MSRE); hence, new structural
liquidus temperatures, vapor pressures, and the materials must be investigated that would withstand
solubility of actinide fluorides in the fuel solvent the attack of the salt at elevated temperatures.
can be adequately derived. However, the addition
of other matrix components must be investigated
in order to decrease the melting temperature of the Acknowledgments
nonmoderated MSR fuel.
The density and the derived molar volume of most The authors want to acknowledge Dr. V. Ignatiev and
binary liquid systems can be described in terms of Dr. V. Khokhlov for their fruitful discussions on the
ideal mixtures, which means that the density can topic of the molten salt reactor.
be calculated by linear interpolation of the molar
volumes of the end members.
The viscosity of fluoride systems shows significant References
nonideal mixing behavior. Notably, the binary
LiF–ThF4 system shows a strong positive devia- 1. US DOE Nuclear Energy Research Advisory Committee
and the Generation IV International Forum, A Technology
tion from ideal behavior over a wide composition Roadmap for Generation IV Nuclear Energy Systems,
range. The same has been found in the LiF–BeF2 http://www.ne.doe.gov/genIV/documents/
system. Based on these observations, it is not pos- gen_iv_roadmap.pdf.
2. Robertson, R. C. Technical Report ORNL-TM-728; 1965.
sible to accurately estimate the viscosity of the 3. Robertson, R. C. Technical Report ORNL-4541; 1971.
fluoride salts, and therefore more measurements 4. Status of small reactor designs without on-site refuelling;
are required. Technical Report; IAEA-TECDOC-1536; 2007.
5. Delpech, S.; Merle-Lucotte, E.; Heuer, D.; et al. J. Fluor.
Data on experimental heat capacity of molten Chem. 2009, 130, 11–17.
fluoride systems containing actinide fluorides are 6. Zherebtsov, A. L.; Ignatiev, V. V. Experimental mock-up
generally lacking. Khokhlov et al.65 derived an esti- of accelerator-based facility for transmutation of
radioactive waste and conversion of military plutonium;
mation equation that is based only on the reverse Technical Report; Nr. 1606, Annual Report; 2006.
molar mass of the salt mixture. However, this 7. MacPherson, H. G. Nucl. Sci. Eng. 1985, 90, 374.
method must be considered only as an approxima- 8. Grimes, W. R.; Cuneo, D. R. Reactor Handbook;
Materials, Interscience: New York, 1960; Vol. I,
tion as it sometimes gives relatively large devia- Chapter 17, p 425.
tions to the experimentally determined heat 9. Grimes, W. R. Nucl. Appl. Technol. 1970, 8, 37.
capacity (e.g., pure LiF or NaF). Thus, more mea- 10. Haubenreich, P. N.; Engel, J. R. Nucl. Appl. Technol.
1970, 8, 118–133.
surements are required in order to describe the 11. Mourogov, A.; Bokov, P. M. Energ. Convers. Manag.
heat capacity behavior of the liquid fluoride solu- 2006, 47, 2761.
tions more precisely. 12. Gruppelaar, H.; Kloosterman, J. L.; Konings, R. J. M.
Advanced Technologies for the Reduction of Nuclear
Experimental data for the thermal conductivity of Waste, Netherlands Energy Research Foundation
molten fluoride systems are generally lacking. ECN; 1998.
Although Khokhlov et al.65 recently derived an 13. Mathieu, L. Ph. D. Thesis, France (in French), 2005.
14. Mathieu, L.; et al. Proposition for a very simple thorium
estimation method for this quantity on the basis molten salt reactor. In Global Conference, Tsukuba,
of the available data for the molten chlorides, Japan, 2005.
388 Molten Salt Reactor Fuel and Coolant
15. Mathieu, L.; et al. Progr. Nucl. Energ. 2006, 48, 48. Beneš, O.; Konings, R. J. M. Comput. Coupl. Phase
664–679. Diagrams Thermochem. 2008, 32, 121–128.
16. Merle-Lucotte, E.; et al. Fast thorium molten salt reactors 49. Beneš, O.; Beilmann, M.; Konings, R. J. M. J. Nucl.
started with plutonium. In Proceedings of the Mater. 2010, 405, 186–198.
International Congress on Advances in Nuclear Power 50. Sood, D. D.; Iyer, P. N.; Prasad, R.; et al. Nucl. Technol.
Plants (ICAPP), Reno, NV, 2006. 1975, 27, 411–416.
17. Merle-Lucotte, E.; Heuer, D.; Brun, C. L.; Allibert, M.; 51. Blanke, B. C.; Bousquet, E. N.; Curtis, M. L.; Murphy, E. L.
Ghetta, V. The TMSR as Actinide Burner and Thorium Technical Report USAEC MLM-1086; 1956.
Breeder, Note LPSC 07-37; 2007. 52. Cantor, S.; Ward, W. T.; Moynihan, C. T. J. Chem. Phys.
18. Salanne, M.; Simon, C.; Turq, P. J. Phys. Chem. B 2007, 1969, 50, 2874.
111, 4678–4684. 53. Cantor, S. Technical Report ORNL-TM-4308; 1971.
19. Toth, L. M.; Gilpatrick, L. O. Technical Report 54. Cohen, S. I.; Jones, T. N. Technical Report ORNL-2278;
ORNL-TM-4056; 1972. 1957.
20. Dracopolous, V.; Gilbert, B.; Børrensen, B.; Photiadis, G. M.; 55. Abe, Y.; Kosugiyama, O.; Nagashima, A. J. Nucl. Mater.
Papatheodorou, G. N. J. Chem. Soc. Faraday Trans. 1997, 1981, 99, 173–183.
93, 3081. 56. Desyatnik, V. N.; Nechayev, A. I.; Chervinskii, Y. F. Zh.
21. Dracopolous, V.; Gilbert, B.; Papatheodorou, G. N. Prikl. Khim. 1981, 54, 2310–2312.
J. Chem. Soc. Faraday Trans. 1998, 94, 2601. 57. Porter, B.; Meaker, R. E. Technical Report BMI RI-6836;
22. Barton, C. J. J. Phys. Chem. 1960, 64, 306. 1966.
23. Rosenthal, M. W.; Briggs, R. B.; Kasten, P. R. Technical 58. Hill, D. G.; Cantor, S.; Ward, W. T. J. Inorg. Nucl. Chem.
Report ORNL-4449; 1970. 1967, 29, 241–243.
24. van der Meer, J.; Konings, R. J. M.; Jacobs, M. H. G.; 59. Chervinskii, Y. F.; Desyatnik, V. N.; Nechaev, A. I. Zh. Fiz.
Oonk, H. A. J. J. Nucl. Mater. 2005, 344, 94–99. Khim. 1982, 56, 1946–1949.
25. Beneš, O.; Konings, R. J. M. J. Chem. Thermodyn. 2009, 60. Desyatnik, V. N.; Nechaev, A. I.; Chervinskii, Y. F. Russ.
41, 1086–1095. J. Phys. Chem. 1979, 53, 986–988.
26. Roy, D. M.; Roy, R.; Osborn, E. F.; Am, J. Ceram. Soc. 61. van der Meer, J.; Konings, R. J. M. J. Nucl. Mater. 2007,
1954, 37, 300. 360, 16–24.
27. Thoma, R. E.; Insley, H.; Friedman, H. A.; Hebert, G. M. 62. Mackenzie, J. D. J. Phys. Chem. 1960, 64, 306–309.
J. Nucl. Mater. 1968, 27, 166. 63. Briant, R. C.; Weinberg, A. M. Nucl. Sci. Eng. 1957, 2,
28. Romberger, K. A.; Braunstein, J.; Thoma, R. E. J. Phys. 797–803.
Chem. 1972, 76, 1154. 64. Powers, W. D.; Cohen, S. I.; Greene, N. D. Nucl. Sci. Eng.
29. Holm, J. L.; Kleppa, O. J. Inorg. Chem. 1969, 8, 207. 1963, 71, 200–211.
30. van der Meer, J. P. M.; Konings, R. J. M.; Hack, K.; 65. Khokhlov, V.; Ignatiev, V.; Afonichkin, V. J. Fluor. Chem.
Oonk, H. A. J. Chem. Mater. 2006, 18, 510–517. 2009, 130, 30–37.
31. Beneš, O.; Konings, R. J. M. J. Nucl. Mater. 2008, 377(3), 66. Janz, G. J. J. Phys. Chem. Ref. Data 1988, 17(Suppl. 2),
449–457. 21–309.
32. Barton, C. J.; Strehlow, R. A. J. Inorg. Nucl. Chem. 1961, 67. Kirshenbaum, A. D.; Cahill, J. A. J. Chem. Eng. Data
18, 143–149. 1962, 7, 98–99.
33. Barton, C. J.; Redman, J. D.; Strehlow, R. A. J. Inorg. 68. Smirnov, M. V.; Khokhlov, V. A.; Antonov, A. A. Viscosity
Nucl. Chem. 1961, 20, 45–49. of Molten Alkali Halides and Their Binary Mixtures;
34. Thoma, R. E.; Insley, H.; Friedman, H. A.; Weaver, C. F. Nauka: Moscow, 1979; p 102.
J. Phys. Chem. 1960, 64, 865. 69. Ignatiev, V.; Merzliakov, A.; Afonichkin, V.; et al.
35. Thoma, R. E.; Insley, H.; Landau, B. S.; Friedman, H. A.; In Proceedings of the 7th Exchange Meeting on Actinide
Grimes, W. R. J. Phys. Chem. 1959, 63, 1266. and Fission Product Partitioning Trans-Mutation, Jeju,
36. van der Meer, J.; Konings, R. J. M.; Oonk, H. A. J. Korea, Sept 9–14, 2002; pp 581–590.
J. Nucl. Mater. 2006, 357, 48–57. 70. Ignatiev, V.; Grebenkine, K.; Subbotin, V.; et al.
37. Barton, C. J. J. Phys. Chem. 1960, 64, 306–309. In Proceedings of the International Symposium on Ionic
38. Mailen, J. C.; Smith, F. J.; Ferris, L. M. J. Chem. Eng. Liquids, Carry le Rouet, France, Jun 26–28, 2003;
Data 1971, 16, 68–74. pp 203–310.
39. International Scientific Technical Centre, Moscow; 71. Chrenková, M.; Daněk, V.; Vasiljev, R.; Silný, A.;
Technical Report; ISTC Project #1606 Final Report; Kremetsky, V.; Polyakov, E. J. Mol. Liq. 2003, 102,
July 2004. 213–226.
40. Ignatiev, V.; Merzlyakov, A.; Subbotin, V.; Panov, A.; 72. Cantor, S.; Cooke, J. W.; Dworkin, A. S.; Robbins, G. D.;
Golovatov, Y. Atomnaya energiya 2006, 101, Thoma, R. E.; Watson, G. M. Technical Report
364–372. ORNL-TM-2316; Aug 1968.
41. Ignatiev, V.; Feynberg, O.; Smirnov, V.; Tataurov, A.; 73. Douglas, T. B.; Payne, W. H. J. Res. NBS 1969, 73A, 479.
Vanukova, G.; Zakirov, R. In Proceedings of 74. Cooke, J. W.; Hoffman, H. W.; Keyes, J. J. Technical
ICENES-2005, Brussels, Belgium, Aug 21–26, 2005. Report ORNL-TM-4396; 1969.
42. Thoma, R. E. Advances in Molten Salt Chemistry; Plenum: 75. Kato, Y.; Furukawa, K.; Araki, N.; Kato, Y. High Temp.
New York, NY, 1975; Vol. 3, Chapter 6, p 275. High Pres. 1983, 15, 191–198.
43. Selivanov, V. G.; Stender, V. V. Russ. J. Inorg. Chem. 76. Araki, N.; Kato, Y. Research on Thorium Fuel, Ministry of
1958, 2, 279–282. Education; Science and Culture: Tokyo, Japan, 1987;
44. Barton, C. J.; Gilpatrick, L. O.; Bornmann, J. A.; Stone, H. H.; pp 83–86.
McVay, T. N.; Insley, H. J. Inorg. Nucl. Chem. 1970, 33, 77. Smirnov, M. V.; Khokhlov, V. A.; Filatov, E. F.
337–344. Electrochim. Acta 1987, 32, 1019–1026.
45. Bergmann, A. G.; Dergunov, E. P. Dokl. Ac. Sc. URSS 78. Cantor, S.; Roberts, C. E.; McDuffie, H. F. Technical
1941, 31, 753. Report ORNL-4229; Dec 1967.
46. Sangster, J.; Pelton, A. D. J. Phase Equil. 1991, 12, 511. 79. Beneš, O.; Konings, R. J. M. J. Fluor. Chem. 2009, 130,
47. Chartrand, P.; Pelton, A. D. Metall. Trans. 2001, 32A, 1385. 22–29.
Molten Salt Reactor Fuel and Coolant 389
80. Rosenthal, M. W.; Haubenreich, P. N.; Briggs, R. B. 94. Hammel, C.; Chamelot, P.; Laplace, A.; Walle, E.;
Technical Report ORNL-4812; 1972. Dugne, O.; Taxil, P. Electrochim. Acta 2007, 52,
81. Bamberger, C. E.; Baes, C. F., Jr. J. Nucl. Mater. 1970, 3995–4003.
35, 117. 95. Blankenship, F. F. Technical Report ORNL-3708;
82. Bamberger, C. E.; Ross, R. G.; Baes, C. F., Jr. J. Inorg. 1964.
Nucl. Chem. 1971, 33, 767. 96. Grimes, W. R.; Smith, N. V.; Watson, G. M. J. Phys.
83. Baes, C. F., Jr. Nucl. Metall. 1969, 15, 617. Chem. 1958, 62, 862–866.
84. Rosenthal, M. W.; Briggs, R. B.; Haubenreich, P. N. 97. Blander, M.; Grimes, W. R.; Smith, N. V.; Watson, G. M.
Technical Report ORNL-4782; 1972. J. Phys. Chem. 1959, 63, 1164–1167.
85. Rosenthal, M. W.; Briggs, R. B.; Haubenreich, P. N. 98. Watson, G. M.; Evans, R. B.; Grimes, W. R.; Smith, N. V.
Technical Report ORNL-4728; 1971. J. Chem. Eng. Data 1962, 7, 285–287.
86. Rosenthal, M. W.; Briggs, R. B.; Haubenreich, P. N. 99. Engel, J. R.; Bauman, H. F.; Dearing, J. F.;
Technical Report ORNL-4676; 1971. Grimes, W. R.; McCoy, J. H. E. Technical Report
87. Rosenthal, M. W.; Briggs, R. B.; Haubenreich, P. N. ORNL-TM-6415; 1979.
Technical Report ORNL-4622; 1971. 100. Compere, E. L.; Kirslis, S. S.; Bohlmann, E. G.;
88. Rosenthal, M. W.; Briggs, R. B.; Kasten, P. R. Technical Blankenship, F. F.; Grimes, W. R. Technical Report
Report ORNL-4548; 1970. ORNL-4865; 1975.
89. Baes, C. F., Jr. In Proceedings of the Symposium on 101. Engel, J. R.; Bauman, H. F.; Dearing, J. F.; Grimes, W. R.;
Thermodynamics, Vienna, 1966; pp 409–433. McCoy, H. E.; Rhoades, W. A. Technical Report
90. Baes, C. F., Jr. In Symposium on Reprocessing of ORNL-TM-7207; 1980.
Nuclear Fuels, Iowa State University, Ames, IA, 1969; 102. Briggs, R. B. Technical Report ORNL-3708; 1964.
CONF-690801, pp 617–654. 103. Koger, J. W. Technical Report ORNL-TM-4286;
91. Baes, C. F., Jr. J. Nucl. Mater. 1974, 51, 149–162. 1972.
92. Hitch, B. F.; Baes, C. F., Jr. Inorg. Chem. 1969, 104. Rosenthal, M. W.; Briggs, R. B.; Kasten, P. R. Technical
8, 201. Report ORNL-4254; 1968.
93. Chamelot, P.; Massot, L.; Hamel, C.; Nourry, C.; Taxil, P. 105. Keiser, J. R. Technical Report ORNL-TM-6002; 1977.
J. Nucl. Mater. 2007, 360, 64–74.
3.14 Uranium Intermetallic Fuels (U–Al, U–Si, U–Mo)
Yeon Soo Kim
Argonne National Laboratory, Argonne, IL, USA
391
392 Uranium Intermetallic Fuels (U–Al, U–Si, U–Mo)
IL Interaction layer (reaction layer) adequate mechanical, physical, thermal, and chemical
INL Idaho National Laboratory (Idaho Falls, ID) properties for cladding material. Aluminum alloy as
LEU Low-enrichment uranium (<20 wt%235U) cladding material also has good corrosion resistance to
MEU Medium-enrichment uranium (35–45 any slightly acidic coolant. Another useful attribute of
wt%235U) aluminum is its relative compatibility to reprocessing.1
MTR Material testing reactor (at INL)
OM Optical microscopy 3.14.1.2 Historical Evolution of
ORNL Oak Ridge National Laboratory (Oak U Intermetallic Fuels
Ridge, TN)
ORR Oak Ridge research reactor (at ORNL)
The U intermetallic fuels for use in research and test
PIE Postirradiation examination
reactors are, in the order of earliest to latest, U–Al,
RERTR Reduced enrichment for research and
U–Si, and U–Mo. The basic driving force for the
test reactors (program)
development of a new fuel is to obtain a higher
SEM Scanning electron microscopy
uranium density in the fuel phase. Uranium metal
SEM SEM backscattered electron image
has the highest uranium density, but it is not usable
BEI
because of poor irradiation stability. The uranium
SEM SEM secondary electron image
intermetallics were introduced to achieve stable irra-
SEI
diation performance of uranium metal. The uranium
TEM Transmission electron microscopy
densities for the candidate fuels are given in Table 1.
Because of its structural similarity to matrix alu-
minum, the first uranium intermetallic fuel chosen
for research and test reactor purposes was U–Al alloy.
3.14.1 Introduction U–Al alloy has a well-established performance his-
tory as the fuel for the materials testing reactor
3.14.1.1 Background (MTR) and engineering test reactor (ETR).
Uranium intermetallic fuels such as U–Al, U–Si, and Fabrication of U–Al alloys with high uranium con-
U–Mo are chiefly meant for research and test reac- tents poses difficulties during the rolling process, and
tors in which neutron production, instead of power uranium inhomogeneity increases proportionally with
generation, is the main purpose. The operation tem- uranium content. The typical picture-frame method of
peratures of these fuels are lower than those of ura- fuel assembly and the related rolling fabrication method
nium ceramic fuels used for power generation such as are illustrated in Figure 2. When alloys of greater than
UO2. In general, the U intermetallic fuels can achieve 25 wt% uranium contents are needed, dispersions of
much higher fission densities than can the oxide fuel. UO2 instead are used.2 (Full use of UO2–Al dispersion
Currently available research reactor fuels are pre- has been limited in the United States, chiefly because of
dominantly in a dispersion form that is composed of swelling problems due to a reaction between UO2 and
fuel particles dispersed in an inert matrix. Figure 1 Al encountered early in its development, although this
illustrates the cross section of a dispersion fuel plate. is not the case for Russian-built reactors. Instead, a little
The fueled zone in a dispersion fuel plate, that is, less dense U3O8–Al dispersion has been used.) Table 1
the fuel particles–matrix mixture zone, is frequently gives the basic properties of fuels currently used (or
called the ‘fuel meat’ or ‘fuel core’ and is metallurgi- candidate fuels) for research and test reactors.
cally bonded to the cladding. Throughout this chap- The application of monolithic U–Al alloy in higher
ter, ‘fueled zone’ is used instead of fuel meat. power reactors such as the advanced test reactor
Aluminum is the most popular choice for the (ATR) at INL and high-flux isotope reactor (HFIR)
matrix because of its low neutron absorption cross at ORNL was deemed limited because of the fabrica-
section, low cost, and good fabricability. It also has tion limitations inherent for a high U-density fuel,
Figure 1 Schematic of the cross section of a dispersion fuel plate frequently adopted in research and test reactors.
The fueled zone (or fuel meat) is composed of fuel particles (darker phase) dispersed in an Al matrix (brighter phase).
Uranium Intermetallic Fuels (U–Al, U–Si, U–Mo) 393
Table 1 Basic properties of uranium intermetallic fuels tested.6,7 When handled in air, however, UAl2 is
more pyrophoric than UAl3, and this leads to com-
Fuel Melting point Physical Uranium
( C) density loading plications in fabrication and the potential for oxygen
(g cm3) (g cm3) impurities in the fuel.
The US Department of Energy (US DOE) initiated
U 1133 19.1 19.1
the RERTR (Reduced Enrichment for Research and
U–7Mo 1145 18.4 17.1
U–10Mo 1150 18.2 16.4 Test Reactor) program in 1978 to convert the world’s
U6Mn 726 17.8 17.1 research and test reactors using high-enrichment ura-
U6Fe 815 17.7 17.0 nium (HEU) to those using low-enrichment uranium
U3Sia 930b 15.6 15.0 (LEU). An enrichment in 235U of 20 at.% is the thresh-
U3Si2a 1665 12.2 11.3
old between HEU and LEU. To use a fuel with reduced
USi 1580 10.96 9.8
UAl2a 1590 8.1 6.6 enrichment, keeping the fuel phase volume the same in
UAl3a 1350b 6.8 5.0 the fueled zone (see Figure 1), requires using a fuel
UAl4 731b 6.1 4.2 having a higher uranium density to compensate for the
U0.9Al4a 641b 5.7 3.7 reduced fissile fraction in LEU.
UAlxc NA 6.4 4.5
In the RERTR program, the fuel form developed
UC 2500 13.6 13.0
UN 2630 14.3 13.5 to accomplish this is U3Si2, which allows the highest
UO2a 2875 10.96 9.7 possible uranium loading among the qualified fuel
U3O8a b
8.4 7.1 types. This fuel showed excellent stability during
Ald 660 2.7 0 irradiation. Fission gas bubble swelling is of no con-
a cern for fuel dimension expansion at typical research
Currently used Al dispersion fuels.
b
Decomposes. and test reactor applications. This fuel enabled LEU
c
d
This case is for the mixture of 60 wt% UAl3 + 40 wt% U0.9Al4. core conversion of 60% of the research and test
Al is included for reference.
reactors worldwide. Another U–Si intermetallic fuel,
U3Si, can achieve even a higher U loading than U3Si2,
high fuel swelling, and the need for adding B-10 but in plate-type designs, it shows unstable growth of
burnable absorber. The use of U3O8–Al dispersion, fission gas bubbles at high temperatures and burn-
that is, U3O8 particles dispersed in an Al matrix in ups.8 In addition, the interaction layer (IL) between
high-power reactors, was first considered and aban- U3Si and Al grows faster than that of U3Si2–Al. In
doned due to concern about the exothermic reaction a rod-type design, U3Si–Al dispersion, however, is
between U3O8 and Al3 and the interdiffusional reac- known to have acceptable performance due to the
tion growth between U3O8 and Al, although this fuel ability of the pin to constrain fission gas bubble
form is still used in some other reactors. swelling in a more stable manner.9
The fuel form of U–Al alloy with a U density high Failure to convert high-power research reactors
enough to satisfy the need for high-power rectors using HEU to LEU U3Si2 called for fuels of even
is a mixture of UAl2, UAl3, and UAl4, known as higher uranium density. Given the unstable irradiation
UAlx. It develops when the U weight fraction is behavior of the high-uranium-content compounds
pushed beyond 62 wt%. The exact fractions of the (e.g., U3Si and U6Fe), the fuel development effort has
compounds included in UAlx depend on the fabrica- shifted to uranium–molybdenum alloys with Mo con-
tion process. Whitacre4 reported that typical powder tent ranging 6–10 wt%, in both monolithic and disper-
lots used in the ATR contained phase fractions of sion fuel forms. Since 1997, the U–Mo alloys have
7.6 wt% UAl2, 78.6 wt% UAl3, and 13.8 wt% UAl4. been irradiation-tested under the auspices of the U.S.
UAlx has several positive features that enable its RERTR program and other programs in Argentina,
superior performance in high-power reactors. Fuel Canada, France, South Korea, and Russia. These tests
swelling can be reduced by accommodating fission have shown that U–Mo alloy has stable irradiation
product swelling in the powder dispersions, which behavior.
include pores left during fabrication. It also has ex- A major complication in U–Mo alloy dispersion
ceptional resistance to fission gas bubble formation. in Al is the reaction between U–Mo and Al. Under
In addition, fabrication with a uniform distribution certain irradiation conditions, fission gas bubble
of burnable absorbers is possible.5 growth in this reaction product is sufficient to cause
To achieve a higher U-density fuel, UAlx com- fuel plate failure. A small amount of silicon added
posed of dominantly UAl2 instead of UAl3 was to the matrix aluminum has been found to be a
394 Uranium Intermetallic Fuels (U–Al, U–Si, U–Mo)
Cover
plate
Fuel bearing
alloy or meat
compact
Frame
Cover
plate
(a)
Al powder
Al-alloy frame and
Fuel compact
cover for cladding
Fuel powder
Assemble
and weld
Shear excess
material to desired
plate dimensions
(b)
Figure 2 Illustration of plate fabrication. (a) Exploded view of the dispersion fuel compact assembly by the
picture-frame method. (b) Flow diagram of the hot-rolling fabrication method.
of the material increases when it becomes amor- Fission gas and any gas included during fabrication
phous. Amorphization of a crystalline material to remain in the fueled zone; in particular, fission gases
metallic glass is usually accompanied by an increase are contained in pores or fission gas bubbles. Gas
in volume – a quantity called ‘free volume’ – which pressure in large pores and fission gas bubbles, which
facilitates atomic mobility, enhancing diffusion.10 may be insufficient to cause detrimental creep or yield-
Fission gas mobility is also high in amorphous mate- ing of fuel, could instead result in blistering of a fuel
rial and the fuel material is more readily deformed plate when the plate is heated to a certain temperature.
by the growing gas bubbles. Hence, overall, fission Two types of mechanisms can be considered for blis-
gas bubble growth in an amorphous material is faster. tering. One is pore (or void) connection, and the other
The three U–Al intermetallics undergo amorphi- is pressure rupture of fission gas bubbles. Figure 3
zation depending on the fission rate and tempera- shows the images of a typical blister-tested plate. In
ture. The lower the irradiation temperature and the the typical blister test, the sample plate is held at a
higher the fission rate, the more readily the fuel specified temperature for 30–60 min during each
becomes amorphous. Among the three, UAl4 amor- annealing step.
phizes most readily and UAl2 the least. The reaction The temperature at which blisters form is termed
products between the fuel and matrix are also ura- the ‘blister temperature.’ The higher the blister tem-
nium aluminides and undergo amorphization. The perature, the more resistant the fuel is to blistering.
U–Si intermetallics (U3Si and U3Si2) also become As a design requirement, the minimum blister tem-
amorphous during irradiation, and the reaction prod- peratures, also called ‘blister threshold temperatures,’
uct between the fuel and matrix, U(Si, Al)3, also are typically tested with irradiated plates for the
undergoes amorphization. The U–Mo alloy is not anticipated power excursions.
amorphized during irradiation, but the reaction prod-
uct between the fuel and the matrix, (U, Mo)Alx, 3.14.2 U–Al
becomes amorphous.
In the following sections on U–Al, U–Si, and U–Mo 3.14.2.1 U–Al Fuel Properties
fuels, the areas of (1) physical properties, (2) fabrication The U–Al phase diagram is shown in Figure 4. There
methods, and (3) irradiation performance are dis- are three intermetallic compounds in the U–Al sys-
cussed. Each of the three review areas are described tem: UAl2, UAl3, and UAl4. UAl2 forms directly from
in detail. The physical properties section discusses the the liquid, but UAl3 and UAl4 form by peritectoid
phase diagram, lattice structure of important composi- reactions with aluminum as follows:
tions, and density. The section on fabrication methods
discusses relevant fuel particle fabrication processes. UAl2 þ Al ! UAl3 ½I
The section on irradiation performance includes fuel
swelling, IL growth between fuel particles and matrix UAl3 þ Al ! UAl4 ½II
aluminum, and blister threshold temperature.
Fuel particle swelling and IL growth are two major
fuel performance topics in research reactor fuel plates.
Both work to increase the fueled zone volume. In plate
geometry, fueled zone volume expansion is transferred
directly to plate thickness increase because the
restraint in this direction is the weakest. Monitoring
plate thickness is an effective method of tracking the
fueled zone swelling. An excessive plate thickness
increase is the indicator for potential fuel plate failure. (a)
A unique measure for sound fuel performance
considered in research and test reactors is ‘blister
threshold temperature’ testing with irradiated plates.
In typical research and test reactor fuel designs, (b)
because there is neither a gap between fuel and
Figure 3 Images of a U3Si2–Al dispersion fuel plate after a
cladding nor a plenum, no fission gas release and postirradiation blister test at 450 C. (a) Blistered plate
collection is possible outside of the fueled zone. surface morphology. (b) Cross section of a blistered plate in
This is another advantage of using dispersion fuel. the plate thickness direction.
396 Uranium Intermetallic Fuels (U–Al, U–Si, U–Mo)
Weight percent Al
5 10 20 30 40 60 80
1800
1620
L
1600
L + UAl2
1400 1350
UAl2 + L
Temperature (⬚C)
1200
1135 1105
γ
1000 UAl3 + L
UAl2 + g UAl2
+
UAl3
800 758
776 731
UAl2 + b UAl4 + L
660
UAl4
668 641
600
U0.9Al4
UAl2
UAl3
UAI2 + a UAl4
400 +
Al
U 20 40 60 80 AI
Atomic percent Al
Figure 4 U–Al phase diagram. Redrawn from Okamoto, H. In Binary Alloy Phase Diagrams; Massalski, T. B., Ed.; ASM
International: Materials Park, OH, 1990.
UAl2 has a face-centered cubic structure with a ¼ b ¼ where kT is the U–Al alloy thermal conductivity in
c ¼ 0.776 nm (the MgCu2-type crystal structure). UAl3 W m1 K1 and CU is the uranium content in the
has a simple cubic structure with a ¼ b ¼ c ¼ 0.426 nm alloy in wt%.
(the Cu3Au-type structure).11 UAl4 has a body- Thermal conductivity of UAl2 is not available,
centered orthorhombic structure with a ¼ 0.441, though it can be reasonably estimated as a value
b ¼ 0.627, c ¼ 1.371 nm.12 UAl4 is found as U-lattice lower than that of UAl3. In general, the higher the Al
deficient, and therefore U0.9Al4 is frequently used to content in a U–Al intermetallic, the higher the ther-
designate this compound. This compound is, however, mal conductivity because Al is more thermally con-
expressed stoichiometrically as UAl4.9. ducting than is U. Among the U–Al compounds, UAl4
The densities of the compounds are 8.14 g cm3 has the lowest thermal conductivity due to its defec-
for UAl2, 6.80 g cm3 for UAl3, and 6.06 g cm3 for tive structure. Because of peritectoid reactions given in
UAl4. However, it is reasonable to assume the density eqns [I] and [II], during heating a UAlx–Al dispersion
of UAl4 as 5.7 g cm3 considering its U-deficiency the UAl4 fraction increases while that of UAl3 remains
structure. nearly unchanged and that of UAl2 decreases. Hence,
the overall thermal conductivity of a UAlx–Al disper-
3.14.2.2 Thermal Conductivity of U–Al sion decreases during heating, but the thermal con-
Alloy and UAlx–Al Dispersions ductivity of UAlx–Al is determined chiefly by the Al
The thermal conductivity of U–Al alloy depends on matrix. For a UAlx (60 wt% UAl3 and 40 wt% UAl4)
the uranium composition in the alloy and on the volume fraction of 35% in the fueled zone, the ther-
temperature. At 65 C, the thermal conductivities of mal conductivity of the fueled zone is 62 W m1 K1,
the as-cast U–Al alloys are a linearly decreasing considering a porosity of 6% in the fueled zone,
function of the uranium concentration13: which is typical for this fuel. This is similar to that of
U3O8–Al but slightly lower than that of U3Si2–Al with
kT ðU Al alloyÞ ¼ 225 2:9CU ½1 the same fuel volume fraction and porosity.
Uranium Intermetallic Fuels (U–Al, U–Si, U–Mo) 397
The recommended data for the thermal con- Since U aluminides are brittle, the typical method
ductivities of UAlx–Al dispersions obtained from for powder fabrication is mechanical pulverization by
Hofman and Snelgrove14 are used to fit an equation the use of jaw crushers and hammer mills. U–Al
as follows: intermetallics are pyrophoric. In particular, UAl2 is
highly pyrophoric; hence fabrication of this com-
kT ðUAlx AlÞ ¼ 225 1:61vUAlx 0:121vUAl
2
x pound is more difficult, although it is the highest in
þ 1:45103 vUAl
3
x ½2 U density among the three U–Al intermetallics. The
desired particle size is controlled by the use of metal-
where kT is in W m1 K1 and vUAlx is the UAlx vol% lic sieves. The undersized or oversized particles are
(for the 60 wt% UAl3 and 40 wt% UAl4 mixture) recycled. The comminuted particles are irregular.
including typical fabrication porosity in the fueled They have sharp corners, cracks, and high surface-to-
zone. volume ratio. The fuel particles are also brittle, which
causes higher porosity in the fueled zone after plate
fabrication. Since the comminution is performed in air,
3.14.2.3 U–Al Fabrication
oxygen is absorbed into the fuel particles. The oxygen
This section deals with the fabrication methods used inclusion has an effect on fuel performance (to be
for the fuel phase made of U–Al intermetallics. The discussed later).
U–Al alloy is in the form of a single slab, whereas It is difficult to fabricate pure U aluminide com-
UAlx is in the form of a particle powder. The fueled pounds. Typically the product powder is a mixture of
zone of the fuel plate containing U–Al alloy is there- UAl2, UAl3, and UAl4. If necessary, pure compounds
fore monolithic in that no distinct aluminum matrix can be prepared by the use of uranium hydride,15 but
is involved, whereas that of UAlx is composed of fuel this method is not a commercially viable option
particles in a matrix of aluminum. In the overall fuel because of high cost, the hazardous process involved,
plate fabrication, both fuels have the same procedure and difficulty of reprocessing the scrap materials.
shown in Figure 2, except for the difference that lies Porosity is common in as-fabricated UAlx–Al and
in whether the fueled zone is a monolithic alloy or a U3Si2–Al dispersions, with the amount depending
dispersion compact. on the fabrication method. For all fabrication methods,
however, it typically increases with the fuel phase vol-
3.14.2.3.1 U–Al alloy ume fraction in the fueled zone. For example, in UAlx
U–Al alloy is produced directly by melting and cast- composed of 60 wt% UAl3 and 40 wt% U0.9Al4, the
ing proper amounts of U and Al metal together, density is rUAlx ¼ 0:6ð6:83Þ þ 0:4ð5:7Þ ¼ 6:4 gcm3
which determines the U density in the alloy. For a and the average stoichiometry x ¼ 0.6(3) þ 0.4(4.9) ¼
plate-type fuel, the alloy melt is poured into graphite 3.8. For this case, as-fabricated porosity as a function
molds to produce cast alloy slabs. Each monolithic of UAlx volume fraction can be estimated using infor-
slab of U–Al alloy is then sandwiched between Al mation given by Beeston et al.,5 as shown in Figure 5
alloy cladding slabs and hot-rolled into dimension to along with the data for U3Si2–Al.
form a plate-type fuel element.
This method can be used for fabrication of U–Al
3.14.2.4 U–Al Irradiation Performance
alloys with up to 25 wt% uranium. Up to 40 wt%
uranium, U–Al alloy can be fabricated in the same 3.14.2.4.1 Fuel swelling by fission products
way with a small amount of UAl4 precipitates in the Fuel swelling by fission products is conveniently
alloy. Alloys with such content ratios are liable to divided into two distinct parts. One is fuel swelling
contain small fractions of metastable UAl3. by solid and liquid fission products due to the differ-
ence between the volume of a uranium atom and solid
3.14.2.3.2 UAlx fission products, and the other is by gaseous fission
UAlx designates a mixture of UAl2, UAl3, and UAl4: products that form bubbles. The former takes place in
The exact composition varies depending on the pow- the matrix of fuel particles and is due to the atomic
der fabricator. Fabrication of UAlx is determined by volume difference between the uranium atom and
its composition. solid fission product atoms. (Fuel matrix here stands
The first step of the powder fabrication of U–Al for the solid part in the fuel particles, excluding the
compounds with the high U weight fraction is fission gas bubbles.) This also includes liquid fission
arc-melting of the mixture of U and Al metals. product atoms. Most fission gas atoms remain in the
398 Uranium Intermetallic Fuels (U–Al, U–Si, U–Mo)
Figure 5 Fabrication porosity in a UAlx–Al dispersion 3.14.2.4.1.2 Fuel swelling by gaseous fission
plate as a function of the volume of UAlx. Information
products
on U3Si2–Al dispersion was obtained from Matos and
Snelgrove,16 and the fabrication data for UAlx–Al were Fuel swelling by fission gas bubbles is more difficult
from Beeston et al.5 to quantify. The fission gas bubbles that form in UAlx
are so small that they were hardly observable by
matrix of fuel particles particularly at low burnup. authors in the literature. Francis20 observed fission gas
This also contributes to fuel matrix swelling, with bubbles in UAl3 with burnup of 60%, or 6.51027
the rate proportional only to burnup or fission den- fissions m3 (in fuel particles). He also reported that
sity, independent of the fabrication method, fuel type, no gas bubbles were observed in reaction layers,
and fuel temperatures. Hence, it is commonly appli- which were UAl4. He suggested that this was due to
cable to other U intermetallic fuels also. The latter the lower uranium density in UAl4 and the lower fission
is due to the fission gas bubble growth. As the gas density of 4.61027 fissions m3 than in UAl3 and
bubble swelling increases at higher burnup, however, also suggested that the defect structure in UAl4 might
more gas atoms move into and collect in gas bubbles. accommodate extra fission gas, delaying bubble forma-
The resulting solid swelling rate is therefore propor- tion. Miller and Beeston21 found no visible fission gas
tionately lower. Modeling this effect in detail has not bubbles in UAl2 up to 4.6 1027 fissions m3. Hofman,22
been tried due to the lack of accuracy in relevant too, found no fission gas bubbles in UAlx up to 7 1027
measurements. fissions m3. Miller and Beeston21 and Hofman22 com-
monly observed fission gas bubbles only in uranium
3.14.2.4.1.1 Fuel swelling by solid fission oxide clusters, but none in UAl2 or UAlx. The source of
products oxygen in the clusters is the air included in the fueled
Fuel swelling by solid (including liquid fission pro- zone from both the fuel powder and the later plate
ducts also) fission products is a result of difference in fabrication performed in an air environment. Figure 6
atomic volumes between uranium atoms destroyed by shows a scanning electron microscopy (SEM) image
fission and solid fission products and how the fission of the fracture surface of UAlx irradiated at 130 C
products stay in the alloy. Hofman and Walters17 esti- to 56% burnup of 93% enriched fuel (or 7 1027 fis-
mated burnup-dependent fuel swelling by solid fis- sions m3). It is unclear whether the oxide clusters acted
sion products for U–Zr alloy. Their estimation is as reservoirs absorbing fission gas, or whether UAl4
1.2% per at.% burnup in U–10Zr, which can be trans- in UAlx helps retard bubble formation. More recent
formed in terms of fission density of U–10Mo as high-resolution transmission electron microscopy
follows: (HR-TEM) investigation of irradiated uranium inter-
metallic fuels has revealed the presence of small fission
DV
¼ 3:5fd ½3 gas bubbles in the IL.23,24 Thus, it may be possible that
V0 s
very small micro bubbles are present in irradiated UAlx,
where fd is the fission density in 1027 fissions m3 . but were undetected due to limits on the resolution of
However, this estimation is applicable only for the microscopy techniques at the time.
situation in which alkali and alkaline-earth fission Recent TEM work by Gan et al.23 shows that small
products release and dissolve in the liquid sodium bubbles, visible only in TEM, are in fact formed in
Uranium Intermetallic Fuels (U–Al, U–Si, U–Mo) 399
Al
U–7 wt%Mo
Interaction
zone
1 μm
is lower than in any other fuel dispersions. This Al as thermodynamic data indicate (see Table 2).
advantage can be attributed to several factors, includ- The diffusion layers for both works by Kiessling and
ing higher as-fabricated porosity, the defect structure Castleman showed a considerable amount of pores.
of UAl4, and the high resistance of UAlx to large Out-of-pile measurements by Nazare et al.28
bubble formation. yielded activation energies of 220 kJ mol1 for the
reaction between UAl3 and Al, and 180 kJ mol1
3.14.2.4.2 Interaction between U–Al and Al
for the reaction between UAl2 and Al. The slightly
UAlx and Al react during irradiation even at low tem-
higher value for UAl3–Al dispersion will result in a
peratures due to irradiation-enhanced interdiffusion.
slower reaction rate than UAl2–Al dispersion.
The U–Al phase diagram shown in Figure 4 shows
Measured reaction data of UAlx–Al from in-pile
that UAl2 and UAl3 react with matrix Al according to
tests are scarce because reactions between the fuel
the peritectoid reactions given in eqns [I] and [II]. and matrix are not an irradiation performance issue
Because there is no higher compound in terms of Al
with this fuel. Relevant changes in volume fractions
content than UAl4, only UAl4 stays stable with Al.
due to fuel–matrix reactions are also generally small.
Reaction products on U aluminides are discernable
All three uranium aluminides undergo amorphiza-
in optical micrographs due to color difference, as
tion, depending on the fission rate and temperature.
shown in Figure 9. The volume fraction analyses of
Therefore, the reaction products of these fuels with Al
UAlx–Al before and after irradiation showed that the
are also subject to amorphization. The lower the irra-
volume fraction of UAl4 increases while that of UAl3 is
diation temperature and the higher the fission rate,
nearly unchanged and that of UAl2 decreases. the more the tendency of amorphization. Among the
The U–Al diffusion-couple tests in the tempera-
three, UAl4 is the easiest and UAl2 is the hardest for
ture range of 100–600 C showed that UAl3 is the
amorphization. Crystalline material is more stable
dominant phase created. UAl4 is less prevalent, and
during irradiation than amorphous material.
UAl2 is present the least.12 At lower temperatures,
there was a greater tendency to form UAl2.
3.14.2.4.3 U–Al blister threshold temperature
In U–Al diffusion-couple tests at 400–600 C, how-
No blister threshold temperature data were found
ever, Castleman25 did not observe UAl4 and rarely
for U–Al alloy fuel, which is probably because this
observed UAl2. This is because UAl2 is unstable around fuel is more prone to failures by breakaway swelling.
UAlx–Al dispersion fuels have high blister thresh-
old temperatures due to their high resistance against
formation of fission gas bubbles. Beeston et al.5 pro-
posed that the blister threshold temperature could be
measured as a function of fueled zone fission density,
and Nazare29 fitted Beeston’s data, adding more data
to give the following correlation:
For a fuel plate without B4C,
C
TB ¼ 921 59:2fd ½6
B
For a fuel plate containing B4C,
A
TB ¼ 907 54fd 224B ½7
Figure 9 Scanning electron microscopy image of a 93% UAl2 95 4209.738T lnT + 76.91T 98.8
235
U-enriched UAlx–Al dispersion after a 60% 235U burnup. UAl3 110 0705.464T lnT + 44.38T 114.0
A is UAl2, B is UAl3, C is the interaction product (UAl4), and UAl4 134 15031.48T lnT + 247.3T 130.0
D is U oxide. Adapted from Hofman, G. L. Nucl. Technol.
a
1987, 77, 110–115; Ryu, H. J.; Kim, Y. S.; Hofman, G. L. Chiotti and Kateley.26
b
J. Nucl. Mater. 2009, 385, 623. Kubaschewski and Alcock.27
Uranium Intermetallic Fuels (U–Al, U–Si, U–Mo) 401
1600 L L
1580 1510
1540
1414
1400
1315
Temperature (⬚C)
1200
1135
985
1000
γ + U3Si
USi3
930
800 776
β + U3Si
USi1.88
U3Si5
USi
668
600
U3Si2
α + U3Si
U3Si
450
400
USi2
U 20 40 60 80 Si
Atomic percent silicon
Figure 10 U–Si phase diagram. Redrawn from Okamoto, H. In Binary Alloy Phase Diagrams; Massalski, T. B., Ed.;
ASM International: Materials Park, OH, 1990.
discussed in Section 3.14.3.3. The thermal conduc- of U solid solution and U3Si, typically 7.5 wt% Si
tivities of U3Si–Al and U3Si2–Al dispersions are is added to the U–Si alloy. A prolonged heat treat-
fitted on the basis of the data given in Matos and ment at 925 C has shown to remove the presence of
Snelgrove16: U solid solution.32 The secondary phases typically
reside inhomogeneously in a fuel particle, which
kT ðU3 Si2 AlÞ ¼ 225 3:22vU3 Si2
causes inhomogeneous size distributions of fission
4:48 102 vU2 3 Si2 gas bubbles inside the fuel particles. In some cases,
U3Si occupies whole fuel particles in a nominal U3Si2
þ 6:76 104 vU3 3 Si2 ½8 fuel. As a result, postirradiation images show anomaly
where vU3 Si2 is the U3Si2 or U3Si volume fraction in in fission gas bubble size between fuel particles.
% including typical fueled zone porosity. Alloy ingots of U–Si are made by mixing and
melting of uranium and silicon with a desired Si/U
ratio. The ingots are sometimes annealed in an inert
3.14.3.3 U–Si Fabrication
atmosphere to complete compound formation. These
U3Si2 is used for plate-type fuels, whereas U3Si is ingots are then broken into smaller particles by a
currently applied only for rod-type fuels. In practice, powder fabrication process.
however, it is almost impossible to fabricate the exact Two types of powder fabrication methods for
stoichiometric form of one of these compounds. This U-silicide fuels are most commonly used. The rela-
is why fuel manufacturers instead start with slightly tively brittle compounds of U3Si2 and USi are
Si-rich alloys, which lead to final products more obtained by comminution of the alloy in a glove box
abundant in higher Si content compounds. For in a nitrogen atmosphere with a hardened steel mor-
example, the stoichiometric U3Si2 requires an Si tar and pestle. U3Si powder fabrication, because U3Si
composition of 7.3 wt%. To suppress the formation has greater toughness than U3Si2 and USi, is typically
Uranium Intermetallic Fuels (U–Al, U–Si, U–Mo) 403
140
120
U3Si
100
(ΔV/V0)f (%)
80
60
40 U3Si2
20
100% LEU BU
(ΔV/V0)s
0
0 2 4 6 8
Fission density (1027 fissions m–3)
50 μm
Figure 12 Fuel particle swelling of U–Si intermetallic
fuels. (a)
the fuel – the fission gas diffusivity – and the resulting The bubble morphology from higher temperature
swelling behavior. Although amorphization is a pre- tests is available in the literature.38,39 Figure 15
requisite for low-temperature high-swelling behavior, shows the fission gas bubbles at different tempera-
it needs to be accompanied by an increase in free tures. Compared to Figure 14(a), Figure 15(c) sug-
volume. gests that bubble growth in U3Si2 can be enhanced
to the level of U3Si if the temperature is increased
by 60 C. U3Si2 appears to experience a bubble
growth phenomenon at high temperatures similar to
that of U3Si at low temperatures; the low bubble
growth advantage of U3Si2 provided by the high
Si/U ratio is negated if the temperature is increased.
A mechanistic rate-theory model demonstrates that
the bubble coarsening process in irradiated amorphous
materials such as U3Si and U3Si2 depends on their
(a)
50 μm
(a)
(b)
50 μm
(c)
50 μm
viscosity.40,41 Estimated irradiation-induced viscosity Al/Si ratio in the IL is the highest for U3Si–Al, lower
values were obtained by comparing the calculated for U3Si2–Al, and the lowest for USi–Al. The composi-
bubble size distribution with the observed average bub- tions of the compounds lie on the tielines between the
ble size and density as a function of fission rate and uranium silicides and Al, as shown in the isothermal
burnup. This model predicts the viscosity values section of the ternary phase diagram (see Figure 17).
determined from the bubble size distribution analysis. For all cases, the reaction products U(AlSi)3 have a
The calculated temperature dependence of the density of 7.1 g cm3, and approximately equal
viscosity depends on the assumption that the rate of volumes of uranium silicide and Al are used. Only a
change of the calculated formation enthalpy with small volume change, 4%, is involved in the reaction.
respect to temperature is symmetric with respect to The compositions of the IL from in-pile tests
the uranium concentration. In addition, the tempera- of U3Si2–Al are also shown in Figure 17. The com-
ture independence of certain material properties positions deviate from the exact stoichiometry, that
(such as thermal expansion coefficient) has also been is, (Al þ Si)/U ¼ 3. This indicates that the reaction
assumed. Thus, only the trend of the calculations products become amorphous during irradiation, as
should be meaningful. It is important to note that, as has been observed in in-pile tests reported previ-
U3Si2 is irradiated, the Si/U ratio shifts to the right. In ously.45,46 Since the IL is amorphous, U, Al, and Si
any event, the calculations show that a 30 K increase atoms exist in a mixture without crystalline restric-
in temperature results in a viscosity for U3Si2 that is tion of stoichiometry.
similar to that of U3Si irradiated at the lower temper- When matrix Al exists, the IL of U3Si2–Al is rich
ature (see Figure 16). In addition, the calculated vis- in Al with the Al/Si ratio 3.5 and the (Al þ Si)/U
cosity of U3Si2 is much more sensitive to temperature ratio is 5.3.47 Leenaers et al.45,44 reported a larger
than that of U3Si. (5.0) Al/Si ratio and a smaller (4.6) (Al þ Si)/U
ratio than those by Kim. Kim’s (Al þ Si)/U ratio is
3.14.3.4.2 Interaction between U–Si and Al larger than Leenaers’ due to higher burnup. Using
U3Si, U3Si2, and USi react with Al to form a single the IL physical density of 7.1 g cm3 and assumption
intermetallic compound, U(AlSi)3. The solubilities of of a linear time-dependent burnup profile in the IL,
Al in U3Si, U3Si2, and USi are very small (1 at.%).11 the (Al þ Si)/U ratio increases to 4.5 at a fission
U(AlSi)3 has a composition intermediate between UAl3 density of 1.43 1028 fissions m3. This suggests that
and USi3, both of which are mutually soluble. The the (Al þ Si)/U ratios from both the present test
and Leenaers’ test are higher than the theoretical
assessment.
1012
The (Al þ Si)/U ratio was seen to decrease to 3.3
1011
with Al/Si 0.29 with depletion of the matrix Al
1010
109
around the U3Si2 particles during irradiation.47 A two-
108 phase mixture of USi2 and U(AlSi)2 was also observed.
Viscosity (Poise)
Si
From out-of-pile tests
[p,D,R]
From in-pile tests [p]
USi3 80 Reaction product composition
USi2 from in-pile tests [L]
From in-pile tests after Al depletion
[p]
Si 60 at.% From in-pile tests after Al depletion
%
Mi
USi
At
sc
ibi
lity
40
U3Si2
ga
p
U3Si
20
U(AlSi) 3
U 20 40 60 80 Al
UAI2 UAI3 UAI4
Atom % Al
Figure 17 U–Si–Al ternary phase diagram showing reaction products between U, Si, and Al. p – Hofman et al.,42
D – Dwight,11 R – Rhee et al.,43 L – Leenaers et al.44
40 mm 20 μm
(a) (a)
U3Si2
IL
40 mm
5 μm
(b)
(b)
Figure 19 Scanning electron microscopy (SEM) images
Figure 18 Optical microscopy images of U3Si–Al and
of high-burnup high-enrichment uranium fuel showing
U3Si2–Al dispersion fuels (low-enrichment uranium)
gas bubble formation in fuel and interaction layer (IL).
irradiated at 100 C in Oak Ridge research reactor (ORR).
(a) SEM backscattered electron image of U3Si–Al
(a) U3Si–Al irradiated for 319 days to a fission density of
dispersion fuel irradiated for 23 days to a fission density
5.3 1027 fissions m3 in ORR (A105). The dark phase
of 2.4 1028 fissions m3 (fuel particle) at 110 C in the
shows the U3Si fuel particles, the bright phase is the Al
high-flux isotope reactor (H3-2) showing large fission
matrix, and the gray phase in between is the interaction
gas bubbles in the IL. (b) SEM secondary electron image
phase. The fission gas bubbles are shown as black
of U3Si2–Al dispersion fuel irradiated for 130 days to a
circles in fuel particles. (b) U3Si2–Al irradiated for 300 days
fission density of 9.4 1027 fissions m3 (fuel particle) at
to a fission density of 4.7 1027 fissions m3 in ORR (A34).
100 C in Oak Ridge research reactor showing small
The dark phase shows the U3Si2 fuel particles, the bright
bubbles in the IL. Notice different scales in the micrographs.
phase is the Al matrix, and the gray phase in between is
the interaction phase.
of local thermal homogenization will take place in a temperature dependent than in-pile diffusion.
shorter time.49 Consequently, the contribution to dif- In-pile diffusion relies on vacancy migration that
fusion enhancement by thermal activation of a fission occurs at lower energies than the atom activation in
event is less important than that from the generation out-of-pile diffusion.
and annihilation of defects that directly influence Since the higher U/Si ratio is the cause for a
interdiffusion. faster IL growth rate for U3Si–Al than U3Si2–Al
The defect generation rate is proportional to the (see Figure 20), the effect of U/Si ratio on IL growth
fission rate. The reaction rate coefficient given by eqn with burnup, particularly at high burnup, is worth
[11] for out-of-pile data can be modified by including further discussion. The U/Si ratio in fuel is the
the fission rate: same in the IL as in the fuel particle. Therefore,
the higher the U/Si atom ratio in the IL, the higher
Qirr _ p
k ¼ Airr exp f ½12 the growth rate. In this context, the growth rate ought
RT
to decrease as the U/Si ratio of the IL decreases with
where f_ is the fission rate, p is the exponent of the burnup. The IL U/Si atom ratios are calculated for
fission rate dependence, Airr is a constant, and Qirr the U3Si–Al, U3Si2–Al, and USi–Al and plotted versus
effective activation energy for irradiation tests, which the fission density in Figure 21.
is related to the mobility of vacancies. The U/Si ratio of IL in U3Si–Al decreases to
To determine p, the measured IL thicknesses were that of U3Si2–Al at a fission density of 1.8 1028
normalized with the fission rate (p ¼ 1) and its square fissions m3 and subsequently to that of USi at
root (p ¼ 0.5), respectively. The result is compared in 2.5 1028 fissions m3. For LEU fuels, however,
Figure 20. When normalized with the fission rate with these fission densities are unreachable. In Figure 22,
a power p ¼ 1 or 0.5, Y2 must be linear with respect to only the H3-3 sample having HEU has a fission
time. However, the data given in Figure 20(a) with density of 2.4 1028 fissions m3. However, the IL
p ¼ 1.0 deviate from a straight line. For U3Si2–Al, the growth rate does not appear to slow in this sample,
data increase somewhat in a parabolic manner, as shown in Figure 22. Samples A115 and A116 have
whereas the data given in Figure 20(b) with p ¼ 0.5 relatively high fission densities 1.6 1028 and
follow straight lines for all fuels. This indicates that 1.2 1028 fissions m3, respectively, but the fitted
p ¼ 0.5 is a better fit. This result suggests, in analogy curves in Figure 22 rule out a decrease in the IL
with the crystalline irradiation-enhanced diffusion growth rates at high burnups. Instead, a closer exam-
theory of Sizmann,50 that recombination of fission- ination reveals that U3Si–Al tends to have faster than
fragment-induced defects may be the underlying the average IL growth at high burnups.
mechanism. Hofman et al.42 developed a correlation The U/Si ratio of U3Si2 drops to that of USi at a
with p ¼ 1 with acceptable accuracies because of its fission density of 9.5 1027 fissions m3. HEU samples
composite modeling. They applied a thermal activa- A121 and A122 were irradiated to fission densities
tion term and an irradiation term separately, which of 9.4 1027 fissions m3 and 1.4 1028 fissions m3,
fortuitously reduced the significance of the linear respectively. Neither shows any sign of a decrease in
dependence of the fission rate. IL growth with burnup. Instead, they experienced
The data-fitting for U3Si–Al results in above-average growth, consistent with the observation
Airr ¼ 4:3 106 and Qirr ¼ 56 kJ mol1, where the for U3Si–Al, which is the opposite of observations
fission rate is in fission cm3 s1 and Y in mm. For regarding the effect of U/Si ratio on IL growth.
U3Si2–Al, following the same procedure, Possible reasons for this contradiction are discussed
Airr ¼ 2:2 108 and Qirr ¼ 41 kJ mol1 are found below.
for U3Si2–Al. Fissions in the IL and in the fuel particles irre-
The activation energy for U3Si–Al obtained from versibly decrease the U/Si ratio by U depletion, but
irradiation test data is larger than that for U3Si2–Al. they increase fission product concentrations in the
This is the reverse of the out-of-pile data, so it may IL and the unreacted fuel. The transition-metal
be concluded that the mechanism for irradiation- elements including rare-earth elements are pro-
enhanced diffusion is different from the thermally duced at a rate of 1.3 atoms per fission. These
activated diffusion. For both U3Si–Al and U3Si2–Al, are known to take U atom sites in the U–Al IL and
the activation energies for the in-pile data are about react with Al, and they have low chemical affinities
an order of magnitude smaller than those for the out- to U. Consequently, these fission products contrib-
of-pile data because out-of-pile diffusion is more ute to IL growth, compensating for the reduction in
410 Uranium Intermetallic Fuels (U–Al, U–Si, U–Mo)
16
14
12 U3Si–Al (HEU)
Y 2/f (10-22 cm5 s) U3Si–Al (MEU)
10 U3Si–Al (LEU)
8 U3Si2–Al (HEU)
U3Si2–Al (MEU)
U3Si2–Al (LEU)
.
6
USi–Al (MEU)
4 USi–Al (LEU)
0
0 100 200 300 400 500
(a) Time (day)
25
20
U3Si–Al (HEU)
Y2/fr0.5 (10-15 cm3.5 s0.5)
U3Si–Al (MEU)
U3Si–Al (LEU)
15
U3Si2–Al (HEU)
U3Si2–Al (MEU)
U3Si2–Al (LEU)
10
USi–Al (MEU)
USi–Al (LEU)
5
0
0 100 200 300 400 500
(b) Time (day)
pffiffi
_ (b) Y2 normalized with
Figure 20 Y2 versus time plots to show the effect of the fission rate. (a) Y2 normalized with f, f_.
The interaction layer thickness data included in the plots are at 100 C, but H3-3 (high-enrichment uranium) at 225 C
is normalized to 100 C.
the U atom concentration by burnup. The 1.3 yield The full-scale plates were in the range 550–575 C.
ratio explains the higher than average IL growth These data show that the blister threshold tempera-
rates for the high fission density samples (A115 tures for U–Si compound plates are similar to those
and A116 for U3Si–Al and A121 and A122 for for UAlx plates.51 From these data, the blister thresh-
U3Si2–Al), which is apparently contradictory to the old temperatures are typically set in the range
observation that the lower the U/Si ratio, the lower 525–550 C.14
is the IL growth. When boron is added, the blister threshold tem-
perature decreases by about 100 C, which is
3.14.3.4.3 U–Si blister threshold temperature observed in a similar pattern as with UAlx–Al
For typical fuel particle loadings, miniature scale dispersion fuels (see Section 3.14.2.4.3). The blister
plates of U3Si2 and U3Si were blistered in the threshold temperature for U–Si intermetallic disper-
range of 515–530 C. When the fuel loading was sion fuels is less sensitive both to burnup and to
very highly increased, they blistered at 450–475 C. fuel volume loading than UAlx–Al dispersion fuels.52
Uranium Intermetallic Fuels (U–Al, U–Si, U–Mo) 411
U3Si
2.0 irregular in shape.
1.5
U3Si and U3Si2 are amorphized under irradiation
due to damage induced by fission. In the amorphous
1.0
U3Si2
fuel, fuel swelling depends on the viscosity of fuel.
The viscosity of an amorphized material is lower than
0.5
USi when it is crystalline: in other words, the fluidity of
~9.5 ~18.5
0.0 the material increases when it becomes amorphous.
0 5 10 15 20 25 30
Fission density (1027 fissions m–3)
Hence, fission gas bubbles grow faster when fuel is
amorphized. Although both are amorphous during
Figure 21 U/Si mole ratio changes in U3Si, U3Si2, and USi
irradiation, U3Si and U3Si2 have inherent differences
versus fission density.
in fission gas bubble growth in that U3Si has unstable
swelling, whereas U3Si2 has stable swelling. The
cause for the difference is attributed to the lower
U3Si–Al (HEU) U3Si2–Al (HEU)
60 U3Si–Al (MEU) U3Si2–Al (MEU)
free volume in U3Si2. Apparently, the additional Si
U3Si–Al (LEU) U3Si2–Al (LEU) A115 bonds in U3Si2 have a large effect on the amount of
USi–Al (MEU) USi–Al (LEU) free volume in the glassy state.
40
In out-of-pile tests, U3Si, U3Si2, and USi react
Y 2 (μm2)
fuel swelling. It was recognized early on in the devel- alloy remains in the g-phase indefinitely at room
opment of fast reactor fuels that molybdenum is one temperature.
of the strongest g-stabilizers of the transition-metal Another advantage of the g-phase uranium over
elements, even stronger than Zr, and that it enables the a-phase lies in its slower reaction with matrix
alloys with U to have relatively high U density (see aluminum than a-phase uranium (see e.g., Park et al.53
Table 1).8 A disadvantage of Mo as an alloying ele- and references therein).
ment is that it has higher neutron absorption cross The radiation stability of U–Mo alloy depends to
sections than Si and Al. Even though it has not been a considerable extent upon its ability to retain the
dealt with here, physics designs showed that the g-phase during irradiation. During irradiations at tem-
penalty caused by the parasitic neutron absorption peratures below the g- and a þ g0 -phase boundary,
was insignificant. which is 565 C for U–10 wt% Mo, the alloy tends
As can be seen in the U–Mo phase diagram shown to transform to the a- and g0 -phases because these are
in Figure 23, the solubility of Mo extends to 22 wt% the thermodynamically stable phases at these lower
(or 41 at.%) in the high-temperature body-centered temperatures. The effect of fission-induced displace-
cubic g-phase, but it is limited to a few percents in ments and thermal spikes is to oppose this tendency,
the a- and b-phases. The g-phase undergoes a eutec- the spikes tending to disorder the g0 -phase and to
toidal decomposition at 565 C, transforming to the produce a homogeneous composition of uranium
dual-phase mixture of the orthorhombic a-phase and and molybdenum in the g-phase. The critical fission
the ordered tetragonal g0 -phase which has the nomi- rate is the rate at which the minimum number of
nal stoichiometry of U2Mo. This transformation is displacements and thermal spikes that maintain the
slow when the molybdenum content is more than g-phase are in balance, with the thermodynamic ten-
about 6 wt%, so a g-phase metastable U–Mo alloy dency to transform to the a þ g0 -phases. Willard and
typically with 6–12 wt% Mo can be obtained by Schmitt54 reported a critical fission rate–temperature
quenching the alloy melt into the g-phase. This correlation above which the g-phase U–Mo is stable
and so fuel swelling is low. Their results are given
in Table 3.
Weight percent Mo The typical Mo content is in a range of 6–12 wt%,
0 10 20 30 and 10 wt% Mo in the U–Mo alloy is optimal for
1400 low fuel swelling. Fission gas bubble swelling appears
B A to be minimum at this content. Coincidentally, the
1300 L
10 wt% Mo in the U–Mo alloy is close to the eutectic
composition.
1200
The variation of the lattice parameter of U–Mo
1100 alloys with the Mo concentration has been provided
as follows55:
1000
Temperature (⬚C)
where xMo is the Mo content in mole fraction. If the chemical grinding process of the alloy ingot, and the
unit cell information is used, the density of U–Mo atomization method typically includes pouring the
can also be obtained by U–Mo alloy melt onto a rotating disk in an inert
atmosphere or using a rotating consumable electrode.
n½ð1 xMo ÞMU þ xMo MMo
rUMo ¼ ½15 The U–Mo alloy is ductile, which poses difficulty
NA ð3:4808 0:3314xMo Þ3 for comminution similar to U3Si. To overcome the
where n is the number of atoms in the unit cell (which toughness of the fuel alloy, slight oxidization of the
is 2 for the body-centered cubic g-phase), xMo is the fuel alloy is allowed during the comminution process.
Mo concentration in mole fraction, MU and MMo are The comminuted powders have more equiaxially
the molecular weights of U and Mo, respectively, and shaped grains and a more homogeneous distribution
NA is the Avogardro number. of grains than the atomized powder fuel because
At room temperature, the U–Mo alloy density is there is no thermal process involved during fabrica-
given by an empirical equation56: tion. In addition, they are heavily cold-worked and
contain a high concentration of dislocations. During
rUMo ¼ 19:1 9:11xMo ½16 hot-rolling and subsequent irradiation of the fuel
plates, this dislocation structure will undergo poly-
where xMo is the Mo content in mole fraction. gonization. The final subgrain structure is similar to
Equation [15] predicts a density larger by 1% than that of the atomized powder, providing nucleation
by eqn [16], and also larger by 0.7% than by eqn [14]. sites for gas bubbles. Comminuted powders are made
The lower predictions by eqn [16] are probably due from well-homogenized cast alloy rods, and therefore
to the measurement data that include small pores do not contain the ‘cored’ cellular structure typical of
in the fabricated samples. Equation [16] is recom- the rapidly solidified atomized powders.
mended for practical use, however, because it is more The microstructure of the atomized powder con-
likely to fit the real alloys. sists of a ‘cellular’ solidification structure which is
commonly found in rapidly cooled alloys that have a
pronounced solidus–liquidus gap. When the U–10Mo
3.14.4.2 Thermal Conductivity of
melt (or U0.78Mo0.22) in the U–Mo phase diagram
(U–Mo Alloy)–Al Dispersions
shown in Figure 23 is cooled, the melt follows the
The thermal conductivity of U–Mo alloy is given right arrow (line A). When it meets the liquidus line,
approximately as follows56: U0.64Mo0.36 solidifies as solid islands. As the cooling
progresses, the solid phase volume increases, while,
kTUMo ¼ 2:2 þ 0:032T ½17
simultaneously, the Mo content in the solid phase
where k is the thermal conductivity of U–Mo alloy decreases. In the atomization process, however, the
in W m1 K1 and T is the temperature in K in the cooling does not follow this equilibrium process.
range of 298 < T < 773 K. The thermal conductivity Instead, once it meets the solidus line, the remaining
of (U–10Mo)–Al dispersions, where 10 is for 10 wt% liquid phase, which consists of a lower Mo content
Mo, is given by fitting data for 40% fuel volume than the solid islands, solidifies abruptly, leaving an
fraction in the dispersion57: interconnected network with a lower Mo content.
ðU10MoÞAl This structure, when etched, may appear as a thick
kT ¼ 18:75 þ 0:70T 1:24 103 T 2 boundary because of its low Mo content and obscures
þ 7:53 107 T 3 ½18 the thinner grain boundaries. The boundaries between
the cells most likely form low-angle grain boundaries.
where k is in W m1 K1 and T is in K in the range Because the swelling rate of U–Mo alloy increases
298 < T < 773 K. as the Mo content decreases starting from 10 wt%,
one can argue that the lower Mo content region near
the grain boundaries has a more favorable environ-
3.14.4.3 U–Mo Fabrication
ment for bubble nucleation. A similar process for
3.14.4.3.1 U–Mo alloy powder fabrication a lower Mo content alloy, such as U–7Mo, is shown
Like U–Si intermetallic fuels, the powder fabrication by the arrow B in Figure 23. The gap between the
methods for U–Mo alloys widely adopted currently solidus and liquidus lines on arrow B is slightly
are the comminution and atomization. The commi- smaller than that on arrow A. Thus, U–7Mo will
nution powder method uses a mechanical and/or have slightly smaller solid islands, that is, smaller
414 Uranium Intermetallic Fuels (U–Al, U–Si, U–Mo)
CL CL
22
CMo (at.%)
5
C
17
0
0 b L 0 b L/2
(a) Distance from GB center (b) Distance from GB center
Figure 24 Model of the initial condition of the grain boundary and Mo content. Schematic of (a) a grain boundary, and
corresponding Mo content and (b) Mo content versus distance from grain boundary center.
1 ð
L=2
@Cðx; t Þ @ 2 Cðx; t Þ 4X x
¼D ½19 þ Cðx; 0Þcos 2np dx
@t @x 2 L n¼1 L
0
x t
with the boundary and initial conditions, cos 2np exp 4Dn2 p2 2 ½21
L L
@C
¼ 0 at x ¼ 0; L=2
@x Truncating the series in eqn [21] to n ¼ 10, we
Cðx; 0Þ ¼ 0 0
x
b; Cðx; 0Þ ¼ 5 b
x
L=2 ½20 obtain the change in Mo concentration as a function
of time at the grain boundary and grain interior, as
The solution is similar to that obtained by Bleiberg,59 shown in Figure 25. As seen in Figure 25, the Mo
which is as follows: concentrations at the grain center and at the grain
Uranium Intermetallic Fuels (U–Al, U–Si, U–Mo) 415
boundary converge with time. Complete homogeni- However, during the heat processes, the reaction
zation, however, takes 2000 days. This is far beyond between the fuel particles and matrix aluminum
the irradiation time of a typical research reactor. The increases at the fuel surface, which is in the a-phase.
concentration difference between the grain center The reaction rate of the a-phase with the matrix is
and the grain boundary is still 2.5 at.% at end of life considerably higher than that of the g-phase.
(EOL) (or 48 effective full power days, EFPD). For
this analysis, we used a grain size of 3 mm and grain-
3.14.4.4 U–Mo Irradiation Performance
boundary ‘width’ of 0.5 mm. This analysis leads to the
conclusion that, for the given irradiation condition of 3.14.4.4.1 Fuel swelling by fission products
typical plates, the grain boundaries remain at a lower Fuel swelling by solid fission products discussed
Mo concentration to the end of the test. in Section 3.14.2.4.1 is also applicable for U–Mo.
The swelling rate given by eqn [4] can be used.
However, fuel swelling kinetics by fission gases, that
3.14.4.3.2 U–Mo dispersion plate fabrication
is, fission gas bubble growth, is different and dis-
The U–Mo fuel uses the same plate fabrication method
cussed more in depth below.
as other U intermetallic fuels, which is hot-rolling
As discussed in Section 3.14.2.4, the solid fission
(see Figure 2 for details of the fabrication method).
product swelling is a linear function of burnup and
However, unlike other fuels, the hot-rolling procedure
given by
(including the follow-on blister test) has significant
effects on the performance of this fuel because of the DV
heat processes involved. The heat processes change ¼ 4:0fd ½4
V0 s
particle characteristics and enhance the reactions
between the fuel particles and matrix aluminum. where the swelling rate is in % and fd is the fission
When the as-fabricated U–Mo particles in the density in 1027 fissions m3.
metastable g-phase are heated during fabrication at U–Mo swelling, specifically, swelling by gas bub-
485 C, the fuel is forced to undergo a a þ g0 trans- ble growth, is known to have two distinct rates: slow
formation. The time to start this transformation is at low burnup and much faster at high burnup. The
governed by the temperature and is typically given phenomenon underlying the transition is grain refine-
by a TTT (time–temperature–transformation) dia- ment or ‘recrystallization’ of the g-phase U–Mo. After
gram. The time taken for the typical hot-rolling this transition, gas bubble agglomeration accelerates,
fabrication and follow-on blister test is enough to pro- resulting in faster swelling.60
duce partial transformation reactions. Times to trans- In Figure 27, evolution of fuel microstructure by
formation can be assessed for typical heat processes fission gas bubble formation and growth is shown
as shown in Figure 26. with three different burnups. Fission gas bubbles
As Bleiberg58,59 has shown, irradiation quickly first appear with an SEM resolution of 0.1 mm on
reverses the partly transformed phases a þ g0 to g. grain boundaries, heterogeneously over the fuel
cross section, as shown in Figure 26(a). There are
100 virtually no bubbles in the interior of the grains.
Time to start γ --> α + γ⬘ transformation (h)
5 mm
(a)
20 nm
5 mm
(b) For fd
3 1027 fissions m3,
DV
ð%Þ ¼ 1:0fd ½22
V0 g
For fd > 3 1027 fissions m3,
DV
ð%Þ ¼ 3:0 þ 2:3ðfd 3Þ þ 0:33ðfd 3Þ2
V0 g ½23
where fd is the fission density in 1027 fissions m3.
The measured fuel swelling data and prediction
correlation composed of eqns [4], [22], and [23] are
5 mm
(c) compared Figure 29.
80
Table 4 Differences of IL growth during irradiation from
out-of-pile tests
Er, Ge, Hf, Ti, Zr, Sn, and Si effectively stabilize the left-hand-side plates and the right-hand-side plates, in
UAl3 phase. Er, Ge, Hf, Ti, and Zr form a (U, X)Al3- that the right-hand-side ones having more Si show
type phase, whereas Sn and Si form a U(Al, X)3-type drastically reduced IL thicknesses.
phase. Chakraborty et al.70 repeated a similar study on The ILs of Si-added plates are commonly uneven
the effect of ternary additions to suppress UAl4 forma- and the thicknesses are different among neighboring
tion. Their results were generally consistent with those particles. This is due to the local difference of the IL
given by Thurber and Beaver, and they concluded that growth rate caused by the nonuniform Si diffusion
the additions of Si, Ge, or Zr were effective. and consequently nonuniform Si concentration in the
In the U.S. RERTR program, the effect of Si addi- IL. Furthermore, for fuel particles in densely popu-
tion to the aluminum matrix to suppress IL growth lated regions, the available amount of Si is effectively
and suppress pore formation in ILs was tested in-pile. restricted by the region enclosed by the fuel particles.
The most obvious comparison can be made with the As the IL grows thicker, the effectiveness of Si
postirradiation metallographs of the side-by-side plate addition weakens and, simultaneously, uniform ILs
pairs. As shown in Figure 31, the IL thickness and are formed, as is typically observed in plates having
microstructure are remarkably different between the a negligible amount of Si in the matrix. This suggests
U–7Mo/Al–0.2Si U–7Mo/Al–2Si
R5R020 (C5) R2R010 (C3)
BU = 60% BU=57%
Time = 135 EFPD Time=135 EFPD
RERTR–6 RERTR-6
13 μm 50 μm 4 μm
50 μm
U–7Mo/Al–0.1Si U–7Mo/Al–2Si
R0R010 (B3) R2R040 (B2)
BU = 86% BU = 86%
Time = 90 EFPD Time = 90 EFPD
RERTR–7 RERTR-7
50 μm
9 μm 3 μm
50 μm
U–7Mo/Al–2Si U–7Mo/Al–4.8Si
R2R078 (C3) R3R108 (C2)
BU = 78% BU = 78%
Time = 98 EFPD Time = 98 EFPD
RERTR–9A RERTR-9A
50 μm 6 μm 50 μm 4 μm
U–7Mo/Al–2Si U–7Mo/Al–3.5Si
R2R088 (B6) R6R018 (B7)
BU = 119% BU = 119%
Time = 115 EFPD Time = 115 EFPD
RERTR–9B RERTR-9B
50 μm 9 μm 50 μm 8 μm
Figure 31 Postirradiation optical microscopy images of cross sections of side-by-side test plates. The plate positions
are given in the parentheses. The numbers in front of Si denote the at-fabrication Si content in Al in wt%. Fuel for all plates
is U–7 wt% Mo alloy. Burnup is low-enrichment uranium equivalent BU for the spot where the micrograph was taken.
Uranium Intermetallic Fuels (U–Al, U–Si, U–Mo) 419
that RERTR-9B plates, with extremely high burnup, are available. Based on these data, the blister thresh-
reach the point where the IL reduction power of Si old temperatures are approximately in the range of
is depleted. 450–525 C, which is slightly lower than that of U–Si
In case of the RERTR-6, -7A, and -9A tests, it was dispersion fuels. However, these data were obtained for
commonly observed that U–Mo/Al with at least samples having extremely high burnups (beyond 100%
2 wt% Si in the matrix is free of pore formation in LEU equivalent BU) and fuel loading of 8.5 gU cm3.
the ILs. Instead, highly burned fuel particles show the On comparison with U–Si dispersion fuels in similar
formation of fission gas bubbles interconnected in their conditions in terms of burnup and fuel volume loading,
peripheries. This kind of porosity has generally rough it is reasonable to predict that both fuels are likely to
surfaces, different from the much smoother ones in the have similar blister threshold temperatures.
ILs of plates with a pure Al matrix. High-burnup plates
from the RERTR-9B show the latter kind of pores in
3.14.4.5 Summary for U–Mo
ILs. As shown in the left-hand side of the third row of
Figure 31, R2R078 from the RERTR-9A shows both U–Mo alloys provide high U density for the LEU
kinds of pores. conversion of high-power research and test reactors
The observation of no porosity in the ILs in plates in the RERTR program. They have good thermal and
with at least 2 wt% Si is noteworthy considering high chemical compatibility with aluminum cladding. For
burnup close to 90% and is attributed to an increase the typical g-stabilized U–Mo alloys with the Mo
in IL stability in the presence of Si, as theorized by content ranging 6–10 wt%, the alloys have very sta-
Ryu et al.64 They showed that the enthalpy of formation ble swelling properties, which pose no performance
of IL decreases (or becomes a larger negative number), issue. Fuel swelling has two distinct rates: slow at
when Si is added to the reaction product between low burnup and much faster at high burnup. The
U–Mo and Al. The average bond strength of a material phenomenon underlying in the transition is grain
typically increases with decreasing enthalpy of forma- refinement or ‘recrystallization’ of the g-phase
tion. Therefore, by analogy, the viscosity of the amor- U–Mo. After this transition, gas bubble agglomera-
phous IL also increases when Si is present. ILs are tion accelerates, resulting in faster swelling.
known to be amorphous during irradiation.46,61 Con- The IL growth is correlated in the same functional
sequently, increasing the viscosity of an amorphous form as for the U–Si with different constants. The
material is a key to resistance against its flow. main issue in the dispersion fuel development is the
As the IL thickness increases, however, the Si formation of a U–Mo–Al IL growing between the fuel
content in the IL decreases and the Si deficiency particles and Al matrix. The IL growth reduces the
reaches a threshold below which the Si effect cannot thermal conductivity of the dispersion fuel, and facil-
suppress IL growth and prevent pore formation. itates formation of pores in the ILs. The effect of the
R2R088 and R6R018 from the RERTR-9B showed former is not significant, but the latter provides a
that this threshold was reached. potential for fuel failure. The ILs are amorphous,
The IL thickness data of the plates with the Si-added whereas the U–Mo fuel is crystalline during irradia-
Al matrix were available from RERTR-6 and -7A. The tion. As discussed in the U–Si section, the fission gas
IL growth kinetics of plates with 2 wt% or more Si in filled pores in the ILs can grow rapidly because of the
the Al matrix was drastically lower than for the pure Al high fluidity of the amorphous ILs. To reduce the IL
matrix plates. The IL growth correlation with the growth and, more importantly, to prevent the forma-
Si-added matrix has the same form as in eqn [24], but tion of the pores in the ILs, a small amount of silicon
with different fitting constants of A ¼ 7.5 1018 and (2–5%) is added in the matrix aluminum. Irradiation
q ¼ 4026 K. The applicable range of this correlation is test results show drastic reduction in IL growth and
limited to the temperatures (lower than 170 C) and improved irradiation stability of the ILs. However,
fuel loading for which the measured data were addition of silicon raises concerns about the effects of
obtained and to the situation in which the effect of Si even small amounts of Si on the standard reprocessing
addition is in its maximum strength. of the spent fuel.
The blister threshold temperature data for U–Mo
3.14.4.4.3 U–Mo alloy blister threshold dispersion fuels are scarce. Only a few data points are
temperature available, which show slightly lower temperatures
The blister threshold temperature data for U–Mo dis- than of other U–Si dispersion fuels. However, these
persion fuels are scarce. Only a few data points data are obtained for much higher burnups and fuel
420 Uranium Intermetallic Fuels (U–Al, U–Si, U–Mo)
volume loadings compared to U–Si dispersion fuels. the strength of the bond so that it can survive during
Therefore, a reasonable prediction is that U–Mo irradiation.
dispersion fuels have the same blister threshold tem- Although they have long histories of use, U inter-
perature as U–Si dispersion fuels. metallic fuels in research and test reactors have been
the subject of much less theoretical studies than have
UO2 in power reactors. There is also need for more
3.14.5 Summary and Outlook data for better performance modeling. Behavior of
fission products in dispersion fuels is not necessarily
U–Al intermetallic fuels have chiefly been used for less complicated than in UO2 pellets. Fission product
low-temperature but, typically, high-burnup research behavior in the ILs is important to model fuel swelling.
and test reactors. U–Al fuel is stable with respect to Nonetheless, these are still largely unexplored. In this
fuel swelling, and its reaction with matrix aluminum respect, the recent TEM studies, for example, by Van
is slow. The use of HEU also makes these fuels of low den Berghe et al.,61 will advance the understanding of
fuel volume fractions in dispersion-type fuel geome- the behavior of fission products in these fuels.
try, which keeps the fuel temperature low even at U–Mo is also an excellent fuel under a fast neu-
high power and isolates irradiation damage in the tron spectrum that operates at much higher tempera-
fuel particles. In dispersion fuel, fission gas migration tures such as 600 C. It is recently gaining more
is also limited mostly to inside fuel particles. attention because of its benign properties toward
However, the recent impetuses to use LEU have constituent migration, which is characteristic for the
changed the design perspective significantly. The use typical U–Zr fuel.38,39 With negligible constituent
of LEU increases the fuel volume fraction in disper- migration, U–Mo is theoretically predicted to have
sion fuels, so the buffer zones provided by the inert much less fission product migration than in U–Zr.
matrix are reduced. The increase in fuel volume frac- The current understanding is that the fuel swelling of
tion due to the LEU conversion by a simple calcula- U–Mo at the high-temperature regime is slightly
tion is 93/19.9 ¼ 4.7 times for a fuel. The development better than that in U–Zr, but it is anticipated to be
of U–Si intermetallics has been important to LEU controllable by using strong cladding materials such
conversion. In particular, U3Si2 is stable with respect as oxide dispersion strengthened steel (ODS) or
to fuel swelling, and its reaction with matrix aluminum HT9. Use of this fuel will require much more thor-
is so low that it is not a performance issue. However, it ough studies.
is inapplicable for high-power reactors.
Apparently, U–Mo has had the potential to over- Acknowledgments
come the limitation that U–Si intermetallics have,
because its uranium density is much higher than The author is grateful to Dr. G. L. Hofman of Argonne
in U–Si intermetallics. The performance issues with National Laboratory for valuable data and discussion,
U–Mo are reactions between fuel particles and matrix and also acknowledges the U.S. RERTR program for
aluminum as well as formation of large fission gas making available the PIE data used in this work.
pores. These issues have been dealt with by adding
small amounts of alloying elements such as silicon in
the matrix to retard the reaction between the fuel References
particles and matrix and to increase the resistance
1. Peacock, H. B.; Frontroth, R. L. Properties of Aluminum–
of the reaction products against pore formation. Uranium Alloys; WSRC-RP-89-489; Westinghouse
Irradiation tests consistently have shown promising Savannah River Co: Aiken, SC, 1989.
results. In order to increase uranium loading even 2. Cunningham, J. E.; Boyle, E. J. Proc. Int. Conf. Peaceful
Uses Atomic Energy 1955, 9, 203–207.
higher, the use of a monolithic form, that is, a foil, 3. Fleming, J. D.; Johnson, J. W. Nucleonics 1963, 21, 84.
of U–Mo is proposed and under development. There 4. Whitacre, R. F. The UAlx Fuel Dispersion System; EG &
are some successes for this type of U–Mo fuel also. G Idaho: Idaho Falls, ID, 1990; EGG-PRP-8783, Rev. 2.
5. Beeston, J. M.; Hobbins, R. R.; Gibson, G. W.; Francis, W. C.
The main performance issue regarding this type of Nucl. Technol. 1980, 49, 136.
fuel is the separation of the fuel foil from initially 6. Miller, L.; Brown, K. R.; Beeston, J. M.; McGinty, D. M.
metallurgically bonded cladding during irradiation. Extended Life Aluminide Fuel for University Research
Reactors; EGG-SE-6464; Idaho National Engineering
Fission gas bubbles preferentially develop at the Lab Idaho Falls, ID; 1983.
bond, allowing thermal stresses to pull the materials 7. Thummler, F.; Lilienthal, H. E. Powder Metall. 1969, 12,
apart. It appears that the best solution is to increase 1–22.
Uranium Intermetallic Fuels (U–Al, U–Si, U–Mo) 421
8. Hofman, G. L.; Kim, Y. S. Nucl. Eng. Technol. 2005, 34. Birtcher, R. C.; Richardson, J. W.; Mueller, M. H. J. Nucl.
37, 299. Mater. 1996, 230, 158.
9. Kim, C. K.; Kim, K. H.; Jang, S. J.; Jo, H. D.; Kuk, I. H. 35. Birtcher, R. C.; Richardson, J. W.; Mueller, M. H. J. Nucl.
In Proceedings of the 1992 International Meeting Reduced Mater. 1997, 244, 251.
Enrichment for Research and Test Reactors (RERTR), 36. Grest, G. S.; Cohen, M. H. In Advances in Chemical
ANL/RERTR/TM-19, CONF-9209266, Sept 27–Oct 1, Physics; Prigogine, I., Price, S. A., Eds.; Wiley: New York,
1992; Argonne National Laboratory: Argonne, IL 1992. 1981; p 469.
10. Hofman, G. L.; Finlay, M. R.; Kim, Y. S. In Proceedings of 37. Komatsu, T.; Matusita, K.; Yokota, R. J. Non Cryst. Solids
the 26th International Meeting on Reduced Enrichment for 1985, 69, 347.
Research and Test Reactors, Vienna, Austria, Nov 7–12, 38. Kim, Y. S.; Hofman, G. L.; Rest, J.; Robinson, A. B. J. Nucl.
2004; Available from: www.rertr.anl.gov. Mater. 2009, 389, 443.
11. Dwight, A. E. A Study of the Uranium–Aluminum–Silicon 39. Kim, Y. S.; Hofman, G. L.; Yacout, A. M. J. Nucl. Mater.
System; ANL-82-14; Argonne National Laboratory: 2009, 392, 164.
Argonne, IL, 1982. 40. Rest, J. J. Nucl. Mater. 2004, 325, 107.
12. Kiessling, R. In Proceedings of the International Conference 41. Rest, J. Comput. Mater. Sci. 2008, 44, 207.
Peaceful Uses Atomic Energy, Geneva, Switzerland, 1956; 42. Hofman, G. L.; Rest, J.; Snelgrove, J. L.; Wiencek, T. C.
Vol. 9, p 69. In Proceedings of the International Meeting on
13. Jones, T. I.; Street, K. N.; Scoberg, J. A.; Baird, J. Can. Reduced Enrichment for Research and Test Reactor,
Metall. Q. 1963, 2(No.1), 53. Seoul, Korea, 1996; Available from:
14. Hofman, G. L.; Snelgrove, J. L. In Materials Science and http://www.rertr.anl.gov.
Technology; Cahn, R. W., Haasen, P., Kramer, E. J., Eds.; 43. Rhee, C. K.; Pyun, S. I.; Kuk, I. H. J. Nucl. Mater. 1991,
VCH: Weinheim, 1994; Vol. 10A, Chapter 2, pp 45–108. 184, 161.
15. Eding, H. J.; Carr, E. M. High Purity Uranium Compounds- 44. Leenaers, A.; Van den Berghe, S.; Koonen, E.; Jaquet, P.;
Final Report; ANL-6339; Argonne National Laboratory: et al. J. Nucl. Mater. 2004, 327, 121.
Argonne, IL, 1961. 45. Leenaers, A.; Koonen, E.; Parthoens, Y.; Lemoine, P.; Van
16. Matos, J. E.; Snelgrove, J. L. In Research Reactor Core den Berghe, S. J. Nucl. Mater. 2008, 375, 243.
Conversion Guidebook, Vol. 4, Fuels; IAEA: Vienna, 46. Ryu, H. J.; Kim, Y. S.; Hofman, G. L. J. Nucl. Mater. 2009,
Austria, 1992; p. 13, Appendix 1, IAEA-TECDOC-643. 385, 623.
17. Hofman, G. L.; Walters, L. In Materials Science and 47. Kim, Y. S.; Hofman, G. L. J. Nucl. Mater. 2010, 410, 1.
Technology; Cahn, R. W., Haasen, P., Kramer, E. J., Eds.; 48. Nazare, S. G. J. Nucl. Mater. 1984, 124, 14.
VCH: Weinheim, 1994; Vol. 10A, Chapter 1, pp 1–44. 49. Turnbull, J. A. J. Nucl. Mater. 1982, 107, 168.
18. Hofman, G. L. In Proceedings of the 2002 International 50. Sizmann, R. J. Nucl. Mater. 1978, 69, 386.
Meeting on Reduced Enrichment for Research and Test 51. Copeland, G. L.; Hobbins, R. W.; Hofman, G. L.;
Reactors, Bariloche, Argentina, Nov 3–8, 2002; Available Snelgrove, J. L. ANL/RERTR/TM-10; Argonne National
from: http://www.rertr.anl.gov. Laboratory: Argonne, IL, 1987.
19. Kim, Y. S.; Hofman, G. L.; Medvedev, P.; Shevlyakov, G.; 52. Snelgrove, J. L. Safety Evaluation Report: Related to the
Robinson, A. B.; Ryu, H. J. In Proceedings of the Evaluation of Low-Enriched Uranium Silicide-Aluminum
2007 International Meeting on Reduced Enrichment Dispersion Fuel for Use in Non-Power Reactors;
for Research and Test Reactors, Prague, Czech NUREG-1313; U.S. Nuclear Regulatory Commission,
Republic, Sept 23–27, 2007; Available from: http://www. Office of Nuclear Reactor Regulation: Washington, DC,
rertr.anl.gov. 1988.
20. Francis, W. C. Annual Progress Report Reactor Fuels and 53. Park, J. M.; Ryu, H. J.; Oh, S. J.; et al. J. Nucl. Mater. 2008,
Materials Development for FY 1965; IDO-17154; US AEC; 374, 422.
1966. 54. Willard, R. M.; Schmitt, A. R. Irradiation Swelling, Phase
21. Miller, L. G.; Beeston, J. M. Extended Life Aluminide Fuel Reversion, and Intergranular Cracking of U–10 wt% Mo
Final Report; EGG-2441; Idaho National Engineering Fuel Alloy; NAA-SR-8956; 1965.
Laboratory, 1986. 55. Dwight, A. E. J. Nucl. Mater. 1960, 2(1), 81.
22. Hofman, G. L. Nucl. Technol. 1987, 77, 110–115. 56. Rest, J.; Kim, Y. S.; Hofman, G. L.; et al. ANL-09/31;
23. Gan, J.; Keiser, D. D.; Wachs, D. M.; Robinson, A. B.; Argonne National Laboratory: Argonne, IL, 2009.
Miller, B. D.; Allen, T. R. J. Nucl. Mater. 2010, 396, 234. 57. Lee, S. H.; Park, J. M.; Kim, C. K. Int. J. Thermophys. 2007,
24. Jue, J. F. Private communication; Idaho National 28, 1578–1594.
Laboratory: Idaho Falls, ID; 2009. 58. Bleiberg, M. L.; Jones, L. J.; Lustman, B. J. Appl. Phys.
25. Castleman, L. S. J. Nucl. Mater. 1960, 1, 1–15. 1956, 27(11), 1270.
26. Chiotti, P.; Kateley, J. A. J. Nucl. Mater. 1969, 32, 59. Bleiberg, M. L. J. Nucl. Mater. 1959, 2, 182.
135–145. 60. Kim, Y. S.; Hofman, G. L.; Rest, J.; Shevlyakov, G. ANL-08/11;
27. Kubaschewski, O.; Alcock, C. B. Metallurgical Argonne National Laboratory: Argonne, IL, 2008.
Thermochemistry, 5th ed.; Pergamon: New York, 1979. 61. Van den Berghe, S.; Renterghen, W.; Leenaers, A. J. Nucl.
28. Nazare, S. G.; Onsracek, G.; Thummler, F. J. Nucl. Mater. Mater. 2008, 375, 340.
1975, 56, 251–259. 62. Ryu, H. J.; Han, Y. S.; Park, J. M.; Park, S. D.; Kim, C. K.
29. Nazare, S. G. Research Reactor Core Conversion Guidebook; J. Nucl. Mater. 2004, 321, 210.
IAEA: Vienna, Austria, 1992; Vol. 4, p 205. 63. Hamy, J. M.; Lemoine, P.; Huet, F.; Goigon, B.;
30. Kimmel, G.; Sharon, B.; Rosen, M. Acta Crystallogr. 1980, Jarousse, C.; Emin, J. In Transactions of the 8th
B36, 2386. International Topical Meeting on Research Reactor Fuel
31. Zachariasen, W. H. Acta Crystallogr. 1949, 2, 94. Management, Mar 21–24, 2004; ENS: Munchen,
32. Snelgrove, J. L.; Domagala, R. F.; Hofman, G. L.; Germany, 2004.
Wiencek, T. C. Report ANL/RERTR/TM-11, Argonne 64. Ryu, H. J.; Kim, Y. S.; Hofman, G. L. J. Nucl. Mater. 2006,
National Laboratory: Argonne, IL, 1987. 358, 52.
33. Wiencek, T. ANL/RERTR/TM-15; Argonne National 65. Thurber, W. C.; Beaver, R. J. ORNL-2602; Oak Ridge
Laboratory: Argonne, IL, 1995. National Laboratory: Oak Ridge, TN, 1959.
422 Uranium Intermetallic Fuels (U–Al, U–Si, U–Mo)
66. DeLuca, I. S.; Sumsion, H. T. KAPL-1747; Knolls Atomic 69. Exner, H. E.; Petzow, G. Metall 1969, 23, 220.
Power Laboratory: Schenectody, NY; 1957. 70. Chakraborty, A. K.; Crouse, R. S.; Martin, W. R.
67. Boucher, R. J. Nucl. Mater. 1959, 1, 13. ORNL-TM-2800; Oak Ridge National Laboratory:
68. Daniel, N. E.; Foster, E. L.; Dickerson, R. F. BMI-1388; Oak Ridge, TN, 1970.
Battelle Memorial Institute: Columbus, OH; 1959.
3.15 Metal Fuel-Cladding Interaction
D. D. Keiser, Jr.
Idaho National Laboratory, Scoville, ID, USA
3.15.1 Introduction
Abbreviations
AES Auger electron spectroscopy
As metallic fuels undergo irradiation, they swell until at
DOE Department of Energy
1–2 at.% burnup, contact is made with the cladding.1
EBR-II Experimental Breeder Reactor-II
At locations along the fuel–cladding interface where
EDS Energy-dispersive spectroscopy
there is good contact, interdiffusion can occur between
EPMA Electron probe microanalysis
fuel, fission product, and cladding constituents. This
FBTA Fuel behavior test apparatus
fuel–cladding chemical interaction (FCCI) can result
FCCI Fuel–cladding chemical interaction
in the development of interaction zones along the inner
FCMI Fuel–cladding mechanical interaction
surface of the cladding, which can impact the perfor-
ID Inner diameter
mance of a fuel element. Interaction zones can be
IFR Integral fast reactor
brittle, causing cracks that can affect the mechanical
OM Optical metallography
integrity of the cladding to be initiated, or they can
RED Radiation-enhanced diffusion
contain relatively low-melting phases. As a result, these
SEM Scanning electron microscopy
interaction zones are typically considered to be wastage
SS Stainless steel
and, as such, are not credited with any contribution to
TREAT Transient Reactor Test facility
the overall strength of the cladding. Therefore, how
WDS Wavelength-dispersive spectroscopy
metallic fuels and different types of cladding chemi-
WPF Whole Pin Furnace
cally interact needs to be better understood if one
wants to identify the life times of fuel elements in
423
424 Metal Fuel-Cladding Interaction
reactors, to establish the safety thresholds for reactor cladding during fabrication to allow for fuel swelling,
operation, or to model fuel performance. development of interconnected porosity in the fuel, and
Over the last 60 years, both in-reactor and out-of- eventual release of gaseous fission products to a plenum.
reactor experiments have been performed to improve At around 2 at.% burnup, the fuel finally contacts the
the understanding of FCCI in metallic fuel elements. cladding, but because of its highly porous nature, not
They looked at FCCI for different metallic fuel much stress is applied to the cladding. Many driving
alloys, including U, Pu–Al, Pu–Fe, U–Fs, U–Cr, forces exist at the fuel–cladding interface that result in
U–Zr, U–Pu–Zr, U–Mo, and U–Pu–Mo, along with the interdiffusion of fuel, fission product, and cladding
different claddings, including Type 304SS, Type constituents. Some examples are the gradients in tem-
316SS, HT-9, and V-Ti.2 (Fissium (Fs) is a term cre- perature, chemical potentials, and/or neutron flux that
ated to describe those fission products that remained can exist in the fuel, at the fuel–cladding interface, or
with U after melt refining, which was an early repro- in the cladding, along with the overall effect of
cessing method for metallic fuels. For a 5 wt% Fs alloy, radiation-enhanced diffusion (RED) where the fission
2.46Mo–1.96Ru–0.28Rh–0.19Pd–0.1Zr–0.01Nb would process results in the creation of a variety of defects (e.g.,
be present.) On the basis of early investigations, par- vacancies and interstitials) that, at intermediate tem-
ticular focus was given to Zr-containing metallic fuels peratures, can result in an increase in the kinetics
due to their superior compatibility with stainless of the overall process. As the various constituents
steel claddings.3 Because of this improved perfor- interdiffuse, a variety of phases can develop that typi-
mance, the bulk of the available FCCI data is available cally consist of many components. One factor that can
for the U–Zr and U–Pu–Zr alloys. actually impede the interaction of fuel and cladding
This chapter briefly describes the FCCI process is the presence of impurity-enriched layers at the inter-
and how phase diagrams and results from out-of-pile face.4 As a result, premature failure of the cladding
diffusion experiments can be used to gain insight into because of FCMI is typically not observed , and the
the process. Next, the observed behavior in actual fuels can be run up to higher burnups. With longer
irradiated fuel is discussed. The results of high- times in the reactor, FCCI zones have more time to
temperature heating tests that have been conducted develop, and in local areas along the inner surface of
using irradiated fuels to study FCCI during transient the cladding, the zones can become large enough to
conditions are then described, along with a brief affect fuel performance, particularly if cracks are
mention of tests conducted using the Transient Reac- initiated in the brittle interaction zones and penetrate
tor Test (TREAT) reactor. Finally, work looking at the cladding wall.
potential barrier materials to eliminate FCCI is also
discussed, as elimination of FCCI in an irradiated
3.15.2.1 Using Phase Diagrams to Gain
metallic fuel element would be beneficial.
Insight into FCCI
To gain initial insight into the phases that may
3.15.2 The Fuel–Cladding Interaction develop at the fuel–cladding interface, one can look
Process at available equilibrium phase diagrams, even though
the irradiation process deviates from equilibrium.
A variety of metallic fuels have been irradiated in fast The main phase diagrams to be considered are the
reactors.2 U–Mo alloys have been irradiated in binary combinations of the major fuel (e.g., U, Pu, and
Dounreay and Fermi-I, U–Fs alloys in EBR-I and Zr) and cladding (e.g., Fe, Ni, and Cr) constituents.5
EBR-II reactors, and U–Zr and U–Pu–Zr alloys in Additional discussion of the major phase diagrams
EBR-II. In the early fuel elements, the metallic fuel of interest can be found in Chapter 2.05, Phase
was in intimate contact with the cladding after fabri- Diagrams of Actinide Alloys. Things to be consid-
cation.2 As a result, when the fuel swelled during ered when looking at phase diagrams are the inter-
irradiation, large stresses were generated on the clad- metallic phases present in the phase diagrams, the
ding, which was called fuel–cladding mechanical inter- solid-solubility ranges in different phase regions, and
action (FCMI), and this would cause fuel element the temperatures and compositions at which liquid
failure at relatively low burnups. As a result, there was phases develop. The intermetallic phases that are
not much time for FCCI to have a large effect on fuel present in phase diagrams can form in the interaction
performance, as FCCI really becomes a factor only at zones during the interdiffusion process; wide ranges
higher burnups. However, as the development of metallic of solubility in terminal alloys that make intimate
fuels matured, space was left between the fuel and the contact can be an indication that interdiffusion rates
Metal Fuel-Cladding Interaction 425
may be high; and knowing the compositions of the how it combines with Fe (see the Fe–Zr phase diagram
relatively low-melting phases is beneficial as it is in Figure 3), Fe3Zr, Fe2Zr, FeZr2, and FeZr3 are
favorable for relatively low-melting phases to form phases to be considered at the fuel–cladding interface.
at the fuel–cladding interface due to FCCI in an Ternary and quaternary phase diagrams comprised
irradiated metallic nuclear fuel. Looking at the U–Fe of the major fuel and cladding constituents should
and Pu–Fe phase diagrams (see Figures 1 and 2), also be considered. Diagrams have been investigated
which combine either U or Pu as a major fuel constit- for U–Fe–Zr6–9 and U–Pu–Fe–Zr.10–12 Continuous
uent with the major cladding constituent Fe, one can solid solutions have been reported between Pu6Fe
see that as UFe2 and U6Fe phases are present in the and U6Fe and between PuFe2 and UFe22, and so
U–Fe phase diagram and PuFe2 and Pu6Fe phase are (U,Pu)6Fe and (U, Pu)Fe2 should be phases to look
present in the Pu–Fe system, there is a good chance for at the fuel–cladding interface in irradiated fuels
that these phases could be present in any interaction containing U–Pu–Zr alloys. Due to the fact that
zone that develops in an irradiated fuel. Looking at fission products are generated in ever-increasing
another significant fuel constituent, namely, Zr, and concentrations during the irradiation of a metallic
nuclear fuel and can be present at the fuel–cladding
interface, the phase diagrams that combine fission
2000
products with the major cladding constituents can
1800 also be of interest. Lanthanide fission products have
UFe2 been observed in relatively high concentrations
1600
at the fuel–cladding interface13 and are typically
1400 identified as playing a large role in the FCCI process.
U6Fe
In terms of applying phase diagrams to transient
T (K)
1200
fuel behavior, Nakamura et al. have applied data on
1000 the U–Pu–Zr–Fe system to FCCI that could occur
during a transient.14
800
2000 2400
1800 2200
PuFe2 Fe3Zr Fe2Zr FeZr3
2000
1600
1800
1400
1600 FeZr2
T (K)
T (K)
1200 1400
Pu6Fe
1000 1200
1000
800
800
600
600
400 400
0 0.2 0.4 0.6 0.8 1.0 0 0.2 0.4 0.6 0.8 1.0
Fe Pu Fe Zr
x (Pu) x (Zr)
Figure 2 The Pu–Fe phase diagram. Reproduced from Figure 3 The Fe–Zr phase diagram. Reproduced from
Kurata, M.; Nakamura, K.; Ogata, T. J. Nucl. Mater. 2001, Kurata, M.; Nakamura, K.; Ogata, T. J. Nucl. Mater. 2001,
294, 123–129. 294, 123–129.
426 Metal Fuel-Cladding Interaction
and then performing isothermal heat treatments to fuel constituents with different cladding constituents.16
determine the interdiffusion behavior of the various Later tests performed in the 1960s used the fuel alloys
constituents at a temperature of interest.15 This gives U–Fs, U–Pu–Fs, U–Zr, and U–Pu–Zr and different
information about thermal diffusion with no contribu- cladding alloys, including 304SS, 316SS, Hasteloy-X,
tion from any radiation enhancement, and for polished Haynes 56, Timken 16-15-6, Timken 16-25-6,
samples, which should produce the maximum kinet- Incoloy 800, V–20 wt% Ti, and V–10 wt% Ti.17–21
ics, as the presence of surface layers on the tested Diffusion experiments have also been performed
samples will only impede the interdiffusion process. using a systematic approach, in which simple systems
To study FCCI, individual fuel or fission product were first studied, followed by more complex systems.
constituents or fuel or fission product alloys are Binary U–10 wt% Zr fuel alloys have been mated
brought into intimate contact with cladding constitu- with the major cladding component Fe to construct
ents or alloys to form diffusion couples. Typically, a diffusion couples for study,22–24 and then more com-
jig, which can be comprised of two disks that can be plex diffusion couples were constructed to allow
tightened together using threaded rods, is first con- for the study of the more complicated U–Zr–Fe–Ni,
structed out of a low-expansion material (e.g., Kovar) U–Zr–Fe–Cr, and U–Zr–Fe–Ni–Cr systems. Another
in which well-polished samples of interest are higher order system that has been investigated using
inserted. It is then tightened to guarantee good con- diffusion couples includes U–Pu–Zr–Fe.25 Systems
tact between the samples and to ensure the presence with added fission products have also been studied
of compressive stress at the various sample interfaces. (e.g., U–Zr–Ce–Fe).26 More complex systems that can
The tightened jig can then be placed in a capsule (e.g., be investigated combine U–Zr and U–Pu–Zr fuel
quartz) sealed under vacuum or a reducing cover gas alloys with cladding steels. U–Zr alloys have
(e.g., 5% H2/95% Ar), which is placed in a furnace for been tested with D9, HT-9, and Type 316SS.27,28
heat treatment, or placed straight into a furnace that A U–Pu–Zr alloy has been tested with HT-9 and
has an inert gas environment for heat treatment. After D-9 stainless steels.29 Fission products have been
the jig is heated for a specific period of time at a added to U–Pu–Zr alloys and then mated with differ-
specific temperature, the jig is removed and rapidly ent cladding steels for diffusion couple experiments.30
cooled using a quenching medium. The stack of sam- The most complex systems that have been tested for
ples is removed from the capsule, sliced longitudi- FCCI employ U–Pu–Zr alloys with added fission
nally to expose the direction of diffusion, mounted, products and minor actinides that are mated to clad-
polished, and characterized using scanning electron ding steels.31 The diffusion behavior of lanthanide
microscopy (SEM) with energy-dispersive (EDS) or fission products is of interest due to their large role
wavelength-dispersive spectroscopy (WDS) or using in FCCI. Diffusion of Ce with Fe-based alloys has
electron probe microanalysis (EPMA). Images are been investigated,32 along with the interdiffusion of
taken to determine the thickness of the overall inter- Ce and Nd with each other.33
diffusion zone and the morphology of the various Rather than test interdiffusion behavior only at
phases. Point-to-point and linescan compositional temperatures at which all the phases that develop
analysis are employed to determine the composition are in the solid state, tests have been performed at
of the various phases, and composition profiles can be temperatures at which liquid phases can develop.
generated for simpler systems so that interdiffusion These tests have been used to try and determine
data can be calculated using a Boltzmann-Matano the temperature at which liquid phases may form
type of analysis.15 X-ray diffraction analysis of a diffu- in the diffusion zone during interactions of the fuel
sion couple cross-section can be employed for further and the cladding. In Zegler et al.34 the results of early
identification of phases present in the interdiffusion testing are discussed. In this report, the results from
zone. microprobe analysis of annealed U–5Fs/Type 304SS
A variety of diffusion couple experiments have couples annealed at 550, 600, 650, 700, 710, 720, and
been performed to investigate FCCI, looking at dif- 730 C have been discussed. The first signs of the
fusion behavior in relatively simple systems that development of a liquid phase were observed in the
relate to FCCI in fuel elements with metallic fuel 700 C couple. It was suggested that the liquid phase
and stainless steel cladding. During the earliest days formation was the result of the interaction of the
of sodium fast reactor (SFR) fuel development, diffu- following three phases: (1) a phase based on the inter-
sion couple experiments were conducted to gain a mediate phase UFe2, (2) a phase based on the binary
better understanding of the compatibility of metallic phase U6Fe with some Ni, and (3) a gamma uranium
Metal Fuel-Cladding Interaction 427
solid solution enriched with molybdenum. Due to the (e.g., C, O, or N), then Zr-rich layers that are stabi-
increased presence of the liquid phase in couples lized by impurities can develop at the interface and
annealed at 710 and 720 C, a notable increase in impede the interdiffusion process.4,35 Work has been
fuel–cladding interaction was observed over what completed using U–Zr alloys to investigate the devel-
was seen at 700 C. For the couple annealed at opment of nitride layers.36–38 This work involved
730 C, gross macroscopic melting was observed. creating nitrogen-rich layers on the surface of the
The development of interdiffusion zones in couples U–Zr, using nitrogen cover gas. The phases that
with 316SS was very similar to that observed for developed were then characterized using SEM and
304SS couples. The appearance of incipient melting Auger electron spectroscopy (AES).
in the diffusion zone occurred at slightly higher The data from out-of-pile diffusion experiments
temperatures for the couples with 316SS, namely, provides a good baseline set of information that
Tm ¼ 715 C versus Tm ¼ 705 C for the 304SS cou- can be employed to better understand FCCI in irra-
ples. Basically, a eutectic phase can form at a specific diated fuel elements. This information identifies the
composition in FCCI zones at temperatures near primary fuel and cladding constituents that dictate
700 C, and this phase is similar to the one present the phase formation in the interdiffusion zone the
in the U–Fe phase diagram at 725 C (see Figure 1). interaction of fuel and cladding at elevated tempera-
In cases where U–Pu–Zr alloys are mated with stain- tures. A schematic diagram of the main features of
less steels, there is the potential for a liquid phase to an interdiffusion zone that forms when U–Pu–Zr
develop below 700 C. It was reported that a liquid alloy fuel interacts with a stainless steel is presented
phase developed at 650 C in a diffusion couple between in Figure 4. The exact position of the original fuel–
a U–Pu–Zr alloy and Fe25 as also in a diffusion couple cladding interface shown in this figure is determined
between a U–Pu–Zr alloy with added fission products by performing a mass balance of the interdiffusing
mated to a stainless steel and heated at 650 C.35 species using the composition profiles produced by
Impurities in the fuel or cladding alloys have SEM/EDS analysis. Interdiffusion experiments have
been suggested to play a role in determining the shown that for U–Pu–Zr fuels, U and Pu interact with
FCCI behavior in annealed diffusion couples.4 Fe and Ni to form a phase layer that contains about
If enough impurities are present in these materials two times more cladding constituents nearest the
(Fe,Ni,Cr)-enriched layer
Cr enrichment that contains high levels of
possible Zr with some Pu and U
Figure 4 Schematic diagram of the main features observed in an interdiffusion zone that develops in a diffusion couple
between a U–Pu–Zr fuel alloy and stainless steel cladding annealed at a relatively high temperature.
428 Metal Fuel-Cladding Interaction
unreacted cladding in FCCI zones. Fe and Ni com- Table 1 Nominal compositions for stainless steel
bine with Zr to form Zr-enriched phases closer to the cladding materials (wt%)a
unreacted fuel alloy. The Cr that is present in stainless HT9 D9 316
steel becomes enriched in some diffusion couples near
the interface with the unreacted cladding. When fis- Ni 0.5 15.5 10–14
Cr 12.0 13.5 16–18
sion products are incorporated into fuel alloys for out- Mn 0.2 2.0 2 (max)
of-pile experiments, the lanthanides can penetrate Mo 1.0 2.0 2–3
relatively far into the cladding. It has been reported Si 0.25 0.75 1 (max)
that nearest the unreacted cladding in fuel/austenitic Ti – 0.25 –
cladding diffusion couples, the Ni depletion that C 0.2 0.04 0.08 (max)
W 0.5 – –
occurs when D9 is in contact with metallic fuels V 0.5 – –
can result in the formation of a ferritic layer in the S – – 0.03 (max)
cladding, which is similar to a ferritic layer in HT-9 P – – 0.045 (max)
that results from decarburization of the original
a
martensite during a diffusion couple anneal.4 How- Balance Fe.
a
Temperatures calculated for beginning of life (BOL) of fuel elements using the computer code LIFEMETAL.
b
DP-70 and DP-75 were fuel elements run in the reactor beyond breach of the cladding.
c
Not determined.
d
The label for this fuel element was not found.
Source: Keiser, D. D., Jr. In Nuclear Reactors, Nuclear Fusion, and Fusion Engineering; Olsson, A. A. A. P., Ed.; Nova Science:
Hauppauge, NY, 2009.
(a) (b)
amounts. Based on the X-ray maps shown in Figure 7,
Cladding
Fe diffuses into the fuel, Ce and Nd diffuse into the
Fuel
cladding, and Pd enriches in localized areas of the fuel
near the fuel–cladding interface.
Figure 8 shows an EPMA elemental trace taken
at the fuel–cladding interface in DP-16. The approx-
imate original cladding boundary corresponds to
where the Cr concentration drops to approximately
50 µm
zero. The elemental trace shows the same behavior
(c) (d) for Fe, Cr, U, Pu, Zr, Ce, and Nd as that observed with
X-ray maps. The lanthanides penetrated the farthest
into the cladding (40 mm) of any components ori-
ginating in the fuel.
With respect to the SEM analysis that was per-
formed on a sample taken from DP-16 at the axial
elevation x/L ¼ 0.84 that had a local burnup of
9.7 at.% and was exposed to a peak inner cladding
temperature of 542 C, a variety of fuel, fission pro-
(e) (f)
ducts, and cladding constituents were observed in
the different layers that constituted the interaction
layer. Tables 3 and 4, reproduced from Raghavan,12
show the layer compositions that were measured at
two different locations along the inner surface of
the cladding sample that was analyzed. High con-
centrations of lanthanide fission products and fission
product Pd were observed in the different layers.
(g) (h)
U
Zr
Pu
Fe
(i) (j) Cr
Ce
Nd
Table 3 Approximate compositions of spot readings taken in different regions of the phase layers at the first location in
the DP-16 SEM sample (in at.%)a
1 2 2 0 24 16 34 19 1 –
1 1 1 5 35 19 33 5 1 –
1 – – 5 38 13 26 7 2 9La
1 – – – 20 10 16 31 – 6La, 17Pb
2 – 34 6 – 11 23 – – 3Mo, 6La, 8U, 9Sn
2 2 28 7 – 16 28 – – La, 7U, 1Ag, 10Sn
3 28 20 5 15 10 12 – 11 –
4 63 10 4 – 7 13 – 0.3 1La
4 66 15 6 – 5 7 – 0.2 1Ag
4 67 13 6 – 5 7 – 0.2 1Ag
4 69 16 4 – 5 5 – 0.3 1Ag
a
Nominal unreacted cladding had a composition, in wt%, of 84Fe–12Cr–2Si–1Mn–0.4Ni for the components listed in Table 3 (determined
using SEM/EDS).
b
It is difficult to resolve lanthanides using SEM/EDS due to peak overlap. As a result, the numbers given for Ce and Nd should be taken as
qualitative numbers, and in some cases, La may be present with the Ce and Nd (based on EPMA X-ray maps).
Source: Keiser, D. D., Jr. In Nuclear Reactors, Nuclear Fusion, and Fusion Engineering; Olsson, A. A. A. P., Ed.; Nova Science:
Hauppauge, NY, 2009.
Table 4 Approximate compositions of spot readings of phase layers at the second location in the DP-16 SEM sample
(in at.%)
Layer Fe Cr Mo Si Pu U Ce Nd La Other
1 41 – – 2 15 – 10 21 4 7Pd
1 66 2 – 7 22 – 2 2 2 1Ni, 1Ba
1 55 14 0.4 2 3 1 7 15 2 0.4Ni
2 – 28 1 – 5 4 15 39 6 –
3 37 39 2 2 8 2 5 3 1 0.5Mn
4 69 17 – 1 0.2 – 5 6 – 1Mn, 0.4Ni
Source: Keiser, D. D., Jr. In Nuclear Reactors, Nuclear Fusion, and Fusion Engineering; Olsson, A. A. A. P., Ed.; Nova Science:
Hauppauge, NY, 2009.
Overall, the layers were very complex and contained observed for each specimen is correlated with the
a variety of fuel, cladding, and fission product axial position (x/L), the local burnup, and the clad-
constituents. ding ID temperature. Typically, the largest interac-
tion zones in this element were observed where a
3.15.3.2.2 Fuel element with U–Pu–Zr and D9 wedge-shaped crack had formed in the fuel adjacent
cladding to the inner surface of the cladding, and a large fission
Results from the characterization of seven fuel ele- product deposit (primarily Nd, Ce, and La) had
ments with ternary U–Pu–Zr fuel clad with D9 that formed (see Figure 5).
were irradiated as part of the Integral Fast Reactor For a sample taken at x/L ¼ 0.78, three different
(IFR) program have been reported.13 As shown in areas were analyzed at the fuel–cladding interface.
Table 2, the burnups varied from 3.0 to 17.0 at.% The first area that was analyzed was near a wedge-
burnup. A study of FCCI in fuel element A-850, shaped crack in the fuel that had been filled in with
which consisted of U–16 at.% Pu–23 at.% Zr fuel fuel. The second area contained a thin band (about
and D9 cladding and was irradiated to 11.3 at.% 40 mm thick) of lanthanides compressed between the
burnup, sheds light on how FCCI zones develop in fuel and the cladding. The last location was near a fuel
ternary fuel elements clad with D9. A total of seven crack, similar to that shown in Figure 5. The bulk of
samples were analyzed from this as-irradiated pin, the Fe diffusion was observed in the first area analyzed.
and in Table 5, the maximum cladding wastage At the wedge-shaped crack location, Fe penetrated into
Metal Fuel-Cladding Interaction 433
Axial position Local burnup Linear power Cladding ID Maximum interaction zone thickness
(x/L) (at.%) (kW /m) temperature ( C) (wastage) (mm)
Source: Keiser, D. D., Jr. In Nuclear Reactors, Nuclear Fusion, and Fusion Engineering; Olsson, A. A. A. P., Ed.; Nova Science:
Hauppauge, NY, 2009.
the fuel to a maximum depth of about 175 mm, and respectively, from the interface. The x/L position of
the concentration of Fe in local spots in the fuel was 0.95 was the highest temperature region of the pin,
as high as 25 wt%. Penetration of Fe into the fuel and this is where the light and dark phases seemed to
was not observed in the other two areas analyzed. be closest to the cladding. Therefore, the local tem-
Nickel interdiffusion was observed in all the three perature in a fuel element appears to control the
analyzed locations (180–200 mm into the fuel) leav- observed proximity of the light and dark phases in
ing an Ni-depleted layer in the cladding; the Ni relation to the fuel–cladding interface.
concentrations in the fuel varied between 6 and
12 wt%. Zirconium and Ni appeared to colocate in 3.15.3.2.3 Fuel element with U–Zr and
the fuel, especially when adjacent to the cladding. HT-9 cladding
With regard to Pu behavior, the high concentrations Of the fuel elements with U–23 at.% Zr alloy fuel
of Ni corresponded to local depletions of Pu in the that were irradiated in EBR-II and characterized to
fuel, while in the cladding, elevated levels of Pu were investigate FCCI, the most interesting fuel element
observed in the Ni-depleted region. The highest with HT-9 cladding was DP-11 (see Table 2). This
amount of Pu interdiffusion was observed in the first fuel element was irradiated to about 10 at.% burnup
area analyzed. This area had the highest levels of Fe at elevated cladding temperatures (630–660 C
depletion from the cladding and the most elevated beginning-of-life temperature). A specimen was taken
concentration levels of Pu in the cladding. The lan- from an axial position 37.47 cm from the bottom of the
thanide interdiffusion behavior appeared to mirror pin and analyzed using EPMA. Figure 9 shows an
that of the Pu, and the lanthanides were found to elemental trace through a region of the fuel–cladding
interdiffuse to similar depths in the cladding. Cerium interaction zone. The vertical scale in this trace is
and Nd were the most abundant lanthanides at all the uncorrected X-ray intensity, and the x-scale is position
three locations, and La was the only other fission (the cladding is to the left), with the cladding inner
product observed in a measurable concentration. diameter location being taken as zero. At this particular
Uranium and Zr did not penetrate into the cladding location, there is a 50-mm-wide zone of the cladding
at any of the locations analyzed. The maximum lan- that is high in Cr and low in Fe and Ni. Ni is also
thanide penetration was seen where lanthanide phases depleted from the cladding another 40 mm into the
appeared to be in direct contact with the cladding. cladding over what is observed for Fe. In the fuel,
As was the case for fuel element DP-16 (an HT-9 Fe has interdiffused around 250 mm and Ni, around
clad element), two differing contrast phases were 20 mm. In the 50-mm-wide band in the cladding con-
observed in the vicinity of the fuel–cladding interface taining high Cr and low Fe and Ni, lanthanide concen-
in element A-850. These phases contained differing trations are elevated. Ce and Nd are the principal
amounts of lanthanides; the lightest phase also elements in this layer, and lower levels of La, Pr, and
contained Pd and Ru. For the sample sliced at an Sm are also observed. On the basis of X-ray maps, the
x/L position of 0.6, the boundary between the light possible presence of Pd in the fuel–cladding interac-
and dark phases was seen around 200 mm from the tion zone is observed. Figure 10 shows typical X-ray
fuel–cladding interface. For x/L positions of 0.7 and maps taken to identify the locations of the various
0.95, the boundary was observed 170 and 150 mm, components at the fuel–cladding interface.
434 Metal Fuel-Cladding Interaction
Wt%
0 Cr
12
0
Ni
0.6
0
Cs
Ba
La
Ce
Pr
Nd
Sm
3.15.3.2.4 Fuel element with U–Zr and D9 performed on a transverse cross-section from this pin.
cladding Two regions near the fuel–cladding interface were
Results from the characterization of three fuel ele- analyzed using EPMA. In both regions, complex mul-
ments with binary U–23 at.% Zr fuel clad with D9 tiphase diffusion structures were noted. By moving
that were irradiated as part of the Integral Fast Reac- through the interaction zone, starting with the
tor program have been reported (Table 2). Using fuel unreacted cladding and moving toward the fuel, the
element C-709 (U–23Zr fuel, D9 cladding, 9.3 at.% first phase layer that was observed contained Fe, Cr,
burnup) as an example of the FCCI behavior in fuel and some lanthanides (no Ni). The next layer con-
elements with this fuel–cladding combination, one sisted of Fe, Cr, lanthanides, and some Ni, and was
can look at the results of an EPMA analysis that was followed by two lanthanide layers with and without
Metal Fuel-Cladding Interaction 435
Pd and Sn (the layer without Pd and Sn was toward Zr-rich layer with Ni and smaller amounts of Fe; (4)
the cladding), and a layer containing Zr, Ni, and a phase with Nd, Ce, La, and Pd; (5) a U layer with
Fe (and some Si). Closest to the unreacted fuel was a Fe and Zr precipitates (the layer was devoid of
multiphase layer, in which the matrix phase was lanthanides); and (6) a layer with a U-matrix and
enriched with U. A precipitate phase containing Zr, lanthanide phases, Zr and Ni phases, and Fe and Zr
Ni, and Fe was present on the cladding side of the phases.
layer, and a precipitate phase containing only Zr
and Ni was present closer to the center of the fuel 3.15.3.2.5 General comments about
element. Lanthanides were not present in high interdiffusion of fuel, cladding, and fission
amounts on the clad side of this zone but were products at the fuel–cladding interface in an
observed toward the fuel side. irradiated fuel element
The second region of the sample that was ana- Based on the detailed reports of irradiated fuel ele-
lyzed had interdiffusion zones similar to those of the ments in the context of FCCI, a few general com-
first. Going from the unreacted cladding toward the ments can be made about the interdiffusion behavior
fuel center, the layers observed included (1) a layer of certain fuel, cladding, and fission product consti-
containing Fe, Cr, and negligible Ni; (2) a layer with tuents. For the cladding constituents, based on the
Fe, Nd, Ce, La, and negligible amounts of Ni; (3) a results of the analyses of fuel elements with U–Zr
Fuel
Cladding
(a) 50 mm (b)
(c) (d)
(e) (f)
Figure 10 (Continued)
436 Metal Fuel-Cladding Interaction
(g) (h)
(i) (j)
(k)
Figure 10 (a) Electron microprobe specimen current image and X-ray maps for (b) U, (c) Zr, (d) Fe, (e) Cr, (f) Ba, (g) La,
(h) Ce, (i) Pr, (j) Nd, and (k) Sm. Fuel is to the left and cladding to the right. Reproduced from Keiser, D. D., Jr. In Nuclear
Reactors, Nuclear Fusion, and Fusion Engineering; Olsson, A. A. A. P., Ed.; Nova Science: Hauppauge, NY, 2009.
and U–Pu–Zr alloy fuels and D9, Type 316, and the farthest into the cladding (Ce and Nd appear to
HT-9 stainless steel cladding that were irradiated in penetrate the farthest) in both HT-9- and D9-clad
EBR-II, a variety of phases have been observed fuel elements. This penetration results in the devel-
to form locally where swelled fuel has contacted opment of phases containing Ce, Nd, and La nearest
the cladding. At the interface between the fuel the unreacted cladding. Fe and Ni depletion zones are
and the cladding, the observed phases on both sides observed in all the fuel elements, with the depletion of
combine the cladding elements Fe, Ni, and Cr with the Ni extending further into the cladding than the
the fuel components U, Zr, and Pu and the lanthanide depletion of the Fe. Where there is Ni depletion, a
fission products. Some of the phases near the fuel– ferritic layer has been observed4 along with a uniform
cladding interface contain Pd and Ru. front of lanthanide penetration into the ferritic layer.
Looking at the cladding side of the original For HT-9, a ferritic layer closest to the unreacted
fuel–cladding interface, the developed phases are cladding can also develop. This layer results from
enriched with cladding components and they contain decarburization of the martensite. Lanthanides have
differing amounts of the primary fuel components been observed to penetrate along grain boundaries in
(U, Pu, and Zr) and fission product components. The this layer.4 Of the nonlanthanide fission products, Pd
lanthanide fission products are observed to penetrate is found to interdiffuse the farthest into the cladding.
Metal Fuel-Cladding Interaction 437
On the fuel side of the fuel–cladding interaction phases to appear in a fuel element at a high tempera-
zone, some phases are observed that are enriched with ture could be at the fuel–cladding interface. This is not
the fuel constituents U, Pu, or Zr and contain Fe or especially important when considering the normal
Ni as the major nonfuel or components (with some operating temperature of a fuel element, except that
lanthanides). Other phases are observed that primar- it may possibly limit the allowable coolant temperature
ily contain lanthanides. When the low-Ni stainless that can be used for the reactor, but it can become more
steel HT-9 is employed as cladding in a fuel element, important when one considers transient conditions and
Fe is observed to penetrate the deepest into the fuel various accident scenarios. When considering accident
of any cladding constituent. As a result, the phases scenarios where the cladding could reach 700–800 C,
found on the fuel side of the interface in these fuel the exposure would be only for minutes. Yet, for rela-
elements contain more Fe than any other cladding tively lower temperature events (e.g., a loss of coolant
component. The fuel component that appears to com- combined with a disabled heat rejection system), the
bine with the Fe is Zr. For fuel elements clad with the exposure time could run into hours,4 and so it is
austenitic stainless steel D9, Ni diffuses much further important to develop a better understanding of how
into the fuel, and it is found to combine with Zr. Iron liquid phases develop at the fuel–cladding interface at
also diffuses into the fuel and forms phases. relatively high temperatures.
In general, interaction zone formation is localized. To try and identify the temperature at which liquid
In many of the analyzed fuel elements, fuel–cladding phases may first develop, annealing tests have been
interaction can be linked to the integrity of the Zr-rind. conducted using actual irradiated fuel from EBR-II.
The Zr-rind appears to be an impurity-stabilized layer Tests have been performed on segments of irradiated
that does not interact much with cladding or fuel. fuel39 and on whole-irradiated fuel pins.40 To simu-
In general, it remains stable during irradiation, but it late extended transient conditions, such as loss of flow
can develop cracks. Where the Zr-rind fails, some of events, a device called the fuel behavior test apparatus
the thickest interaction zones can be observed. The (FBTA) has been used to heat segments of irradiated
Zr-rind is most likely to remain intact in fuel elements U–Zr and U–Pu–Zr fuel clad with stainless steel.
exposed to lower temperatures and lower power. Tsai41 reported the results of 50 experiments that
As mentioned earlier, temperature and power var- were conducted using U–Pu–Zr and U–Zr alloys.
iations along the length of a fuel element appear to HT-9, Type 316, and D9 stainless steels were the
impact the amount of fuel–cladding interaction that cladding materials used. Most of the FBTA tests that
occurs, the maximum fuel–cladding interaction being were reported were run for 1 h. It was observed that at
observed in the combined high-temperature and temperatures of 725 C and above, there was liquid
high-power region of a fuel element. In this region phase formation that accompanied penetration of the
of a fuel element, the interdiffusion kinetics are at a cladding, the characterization of which was a goal of
maximum and the supply of fission products at the the testing. However, the amount of liquid phase was
fuel–cladding interface is relatively large. The result much greater in the fuel, especially at higher tem-
can be some of the largest observed interaction zones. peratures. This point is illustrated in Figures 11 and
In addition, the burnup of a fuel pin impacts the size 12, which show the results of 1-h testing at 750 and
of the fuel–cladding interaction zone. Higher burnup 800 C, respectively, using irradiated (3 at.% burnup)
fuel pins have larger inventories of fission products at U–10 wt% Zr fuel–HT-9 cladding segments. This
the fuel–cladding interface that can interact with the illustrates the effect of the relatively low quantities
cladding, and because of the relatively long times at of Fe required to cause melting on the uranium-rich
higher temperatures, these fuel pins have more time side of the U–Fe phase diagram.
to interact with the cladding. The cladding penetration manifested itself as a
grain boundary penetration or matrix dissolution by
the liquid phase. In cases where lanthanide fission
3.15.4 High-Temperature Irradiated products were particularly enriched at the fuel–
Fuel Annealing Tests to Investigate cladding interface, enhanced penetration into the
Liquefaction in FCCI Zones cladding was observed, perhaps as a result of prior
solid-state FCCI, or the potential low eutectic tem-
One potentially significant impact that the presence of peratures for alloys of lanthanide fission products
FCCI zones on the inner surface of the cladding has on and cladding components. Overall, no noticeable
fuel performance is the possibility that the first liquid interaction was observed at temperatures below
438 Metal Fuel-Cladding Interaction
now Idaho National Laboratory for generating very conditions that were bounded by the available data.
large bursts of thermal neutron flux over a large sample With respect to the way FCCI was handled in LIFE-
volume. The fast reactor safety program provided the METAL, the interaction zone thickness that was mod-
impetus for construction of the facility so that core eled to develop under specific reactor conditions on
meltdown problems could be studied in-pile. The reac- the inner surface of the cladding was treated as ‘clad-
tor was designed so that mockups of fuel elements, ding wastage’ (i.e., had no strength). Only the thickness
under controlled conditions, could be heated to of the unreacted cladding was used to provide the
destruction without harming the test reactor. The fuel necessary strength to keep the fuel element from fail-
used for this reactor was uranium dispersed in graphite. ing. For modeling metallic fuels under transient con-
The graphite acted as a large heat sink and as a moder- ditions, a computer code called FPIN2 was constructed
ator with a sizable negative temperature coefficient and validated using experimental data.45 A recent effort
of reactivity. When heated, the graphite released to model the irradiation behavior of metallic fuel
the energy of the thermal neutrons and, as a result, has included steady-state and transient FCCI models
increased the probability of their leakage from the core. that are based on precipitation kinetics.46 In the future,
The first meltdown experiments began in FCCI will most likely be modeled using the multiscale
September 1959. For the stainless steel clad, U–5 wt% techniques described elsewhere in this handbook.
Fs alloy EBR-II Mark I fuel that was tested using
TREAT, it was determined that the fuel elements did
3.15.7 Development of Diffusion
not suffer any damage, with outer cladding tempera-
Barriers
tures recorded up to 960 C.42 The temperature
ramp and the time for temperature elevation were
By using a liner material on the inner surface of a
sufficiently short and FCCI kinetics sufficiently slow
cladding, it may be possible to eliminate FCCI in
to ensure a minimum of FCCI. Local boiling and
metallic nuclear fuels during irradiation. The Zr-rind
expulsion of the sodium bond were observed in tests
mentioned earlier has shown a propensity to limit
up to this temperature. Above 960 C and up to
fuel–cladding interaction in irradiated fuel elements,
1000 C, cladding penetration may or may not occur,
but because of its tendency to develop cracks during
depending on the presence of sufficient pressure con-
irradiation, it would not be a reliable barrier. There-
tact points between contiguous surfaces of fuel and
fore, for a material to be a good liner, it would need
cladding. Above 1000 C and up to 1015 C, failure of
better in-pile mechanical properties than those of
the fuel element was assured. The alloy was observed
the Zr-rind, it should interact sluggishly with fuel,
to dissolve its way through the cladding and was then
cladding, and fission product constituents during
forced out by the pressure of vaporizing sodium. U–Zr
irradiation, and it should not produce any low-
and U–Pu–Zr alloys clad in HT-9, D9, and Type 316
melting phases because of any liner/fuel or liner/
were also tested using TREAT43 and these alloys
cladding interactions that may occur. It has been
behaved in a manner similar to U–5 wt% Fs. In gen-
demonstrated that metallic fuels, in general, react
eral, the fuel breaches that occurred during transient
sluggishly with refractory metals like Zr, V, Ti, Ta,
testing were observed at the top of the fuel column.
and Mo15,25,47–49 or with nonmetallic materials like
They were attributed to cladding rupture caused by
ZrN.50 Actual irradiation tests have been performed
the stresses from the pressurization of the plenum,
using a Zr sheath,51 and V-lined cladding tubes were
which were enhanced by the cladding thinning caused
fabricated for reactor testing, but actual in-reactor
by a eutectic-like formation of the molten fuel–
tests were never performed.50 For lined claddings to
cladding phase that penetrated the cladding wall.44
be employed in-reactor, rigorous in-reactor testing
will have to be conducted. Refractory metals tend to
get impurities from the fuel or cladding and can
3.15.6 Modeling FCCI
become brittle. So, it must be demonstrated that
these materials can survive in-reactor to high burnup.
During the IFR program, a computer code called
LIFEMETAL was developed to model the irradiation
performance of U–Zr and U–Pu–Zr alloy fuel.4 (See 3.15.8 Outlook
Chapter 3.23, Metal Fuel Performance Modeling
and Simulation.) This code was empirically based The outlook for improvements in the understanding
and was used to model fuel behavior within the reactor of FCCI in metallic nuclear fuels in the near future is
440 Metal Fuel-Cladding Interaction
favorable. Active research is ongoing in the areas of 15. Philibert, J. Atom Movements: Diffusion and Mass
Transport in Solids, Les Editions; Les Ulis Cedex A:
out-of-pile diffusion studies, phase diagram con- France, 1991.
struction, reactor testing using metallic nuclear 16. Kittel, J. H. Layer Formation by Interdiffusion Between
fuels clad with different stainless steels, continued Some Reactor Construction Metals, ANL-4937; June 17;
Argonne National Laboratory: Argonne, IL, 1949.
analysis of irradiation data from old EBR-II reactor 17. Rothman, S. J. Diffusion in Uranium, Its Alloys, and
tests, and testing using ion beams to simulate in- Compounds, ANL-5700, May 1961; Argonne National
reactor behaviors. In addition, there is much focus Laboratory: Argonne, IL, 1961.
18. Walter, C. M. Interdiffusion Between Uranium-5 w/o
on using multiscale computer modeling to investigate Fissium Alloy and Type 304 Stainless Steel, ANL-6816;
FCCI in irradiated metallic nuclear fuels. It is hoped Mar; Argonne National Laboratory: Argonne, IL, 1964.
that ultimately the result of this research will be the 19. Walter, C. M.; Lahti, J. A. Nucl. Appl. 1966, 2, 308–319.
20. Zegler, S. T.; Walter, C. M. Nucl. Metall. 1967, 13, 335–344.
development of new cladding materials or barriers on 21. Walter, C. M.; Golden, G. H.; Olson, N. J. U–Pu–Zr
claddings that can be employed to lessen or eliminate Metal Alloy: A Potential Fuel for LMFBRs, ANL–76–28,
the impact of FCCI on the irradiation performance of Nov 1975; Argonne National Laboratory: Argonne, IL,
1975.
a fuel element. 22. Keiser, D. D.; Dayananda, M. A. J. Nucl. Mater. 1993, 200,
229–243.
23. Ogata, T.; Kurata, M.; Nakamura, K.; Itoh, A.; Akabori, M.
Acknowledgments J. Nucl. Mater. 1997, 250, 171–175.
24. Nakamura, K.; Ogata, T.; Kurata, M.; Itoh, A.; Akabori, M.
J. Nucl. Mater. 1999, 275, 246–254.
This work was supported by the US Department of 25. Ogata, T.; Nakamura, K.; Kurata, M. J. Nucl. Sci. Technol.
Energy, Office of Nuclear Energy, under D.O.E. Idaho 2000, 37(3), 244–252.
26. Ryu, H. J.; Lee, B. O.; Oh, S. J.; Kim, J. H.; Lee, C. B.
Operations Contract DE-AC07-05ID14517. Accord- J. Nucl. Mater. 2009, 392, 206–212.
ingly, the US Government retains a nonexclusive, 27. Keiser, D. D., Jr; Dayananda, M. A. Metall. Mater. Trans.
royalty-free license to publish or reproduce the pub- 1994, 25A, 1–5.
28. Tak Lee, C.; Kim, H.; Kim, T. K.; Lee, C. B. J. Nucl. Mater.
lished form of this contribution, or allow others to do 2009, 395, 140–144.
so, for US Government purposes. The contributions 29. Keiser, D. D., Jr.; Peteri, M. C. J. Nucl. Mater. 1996, 240,
of Adam Cohen and Doug Porter toward investigation 51–61.
30. Sari, C.; Walker, C. T.; Kurata, M.; Inoue, T. J. Nucl. Mater.
of FCCI in irradiated fuels are acknowledged. 1994, 208, 201–210.
31. Cole, J. I.; Kennedy, J. R.; Keiser, D. D., Jr. An
investigation of fuel–cladding chemical interaction in
References metallic transmutation fuels. In The ANS Embedded
Topical Meeting on Nuclear Fuels and Structural Materials
for the Next Generation Nuclear Reactors, Anaheim, CA,
1. Hofman, G. L.; Walters, L. C.; Bauer, T. H. Prog. Nucl. Jun 8–12, 2008.
Energy 1997, 31(1/2), 83–110. 32. Tortorici, P. C.; Dayananda, M. A. J. Nucl. Mater. 1993,
2. Kittel, J. H.; Frost, B. R. T.; Mustelier, J. P.; et al. J. Nucl. 204, 165–172.
Mater. 1993, 204, 1–13. 33. Keiser, D. D. Scripta Mater. 1995, 33, 959–965.
3. Walters, L. C. J. Nucl. Mater. 1999, 270, 39–48. 34. Zegler, S. T.; Rhude, H. V., Jr.; Lahti, J. A. Compatibility of
4. Hofman, G. L.; Walters, L. C. In Metallic Fast Reactor Fuels Uranium-5 w/o Fissium Alloy with Types 304L and 316
Cahn, R. W., Haasen, P., Kramer, E. J., Eds.; VCH Stainless Steel, Report ANL-7596; Sept; Argonne National
Publishers: New York, 1994; Vol. 10, 1–43. Laboratory: Argonne, IL, 1969.
5. Kurata, M.; Nakamura, K.; Ogata, T. J. Nucl. Mater. 2001, 35. Keiser, D. D., Jr; Cole, J. I. TMS Lett. 2005, 2(3),
294, 123–129. 79–80.
6. Pelton, A. D.; Talley, P. K.; Leibowitz, L.; Blomquist, R. A. 36. Akabori, M.; Itoh, A.; Ogawa, T. J. Nucl. Mater. 2001, 289,
J. Nucl. Mater. 1994, 210, 324–332. 342–345.
7. Kurata, M.; Ogata, T.; Nakamura, K.; Ogawa, T. J. Alloys 37. Akabori, M.; Itoh, A.; Ogawa, T.; Ugajin, M. J. Alloys
Compd. 1998, 271–273, 636–640. Compd. 1994, 213/214, 366–368.
8. Nakamura, K.; Kurata, M.; Ogata, T.; Itoh, A.; Akabori, M. 38. Ogawa, T.; Akabori, M. J. Alloys Compd. 1994, 213/214,
J. Nucl. Mater. 1999, 275, 151–157. 173–177.
9. Raghavan, V. J. Phase Equilib. 2003, 24(4), 364–366. 39. Cohen, A. B.; Tsai, H.; Neimark, L. A. J. Nucl. Mater. 1993,
10. Kurata, M.; Nakamura, K.; Ogata, T. J. Nucl. Mater. 2001, 204, 244–251.
294(1–2), 123–129. 40. Liu, Y. Y.; Tsai, H.; Billone, M. C.; Holland, J. W.;
11. Nakamura, K.; Ogata, T.; Kurata, M.; Yokoo, T.; Kramer, J. M. J. Nucl. Mater. 1993, 204, 194–202.
Mignanelli, M. J. Nucl. Mater. 2002, 304(1), 63–72. 41. Tsai, H. In The 1990 International Fast Reactor Safety
12. Raghavan, V. J. Phase Equilib. 2003, 24(4), 367–370. Meeting, Aug 12–16, 1990; American Nuclear Society:
13. Keiser, D. D., Jr. In Nuclear Reactors, Nuclear Fusion, and Snowbird, UT, 1990.
Fusion Engineering; Olsson, A. A. A. P., Ed.; Nova 42. Dickerman, C. E.; Sowa, E. S.; Okrent, D.; Monaweek, J.;
Science: Hauppauge, NY, 2009. Miller, L. B. Studies of Fast Reactor Fuel Element
14. Nakamura, K.; Ogata, T.; Kurata, M. J. Phys. Chem. Solids Behavior Under Transient Heating to Failure 1. Initial
2005, 66, 643–646. Experiments on Metallic Samples in the Absence of
Metal Fuel-Cladding Interaction 441
Coolant, ANL-6334, August 1961; Argonne National 48. Keiser, D. D., Jr.; Cole, J. I. An evaluation of potential
Laboratory: Argonne, IL, 1961. liner materials for eliminating FCCI in irradiated
43. Carmack, W. J.; Porter, D. L.; Change, Y. I.; et al. J. Nucl. metallic nuclear fuel elements. In The Global 2007
Mater. 2009, 392, 139–150. International Conference on Advanced Nuclear
44. Crawford, D.; Porter, D. L.; Hayes, S. L. J. Nucl. Mater. Energy and Fuel Cycle Systems, Boise, ID, Sept 23–27,
2007, 371, 202–231. 2007.
45. Kramer, J. M.; Liu, Y. Y.; Billone, M. C.; Tsai, H. C. J. Nucl. 49. Kim, J. H.; Ryu, J.; Baek, J. H.; et al. J. Nucl. Mater. 2009,
Mater. 1993, 204, 203–211. 394, 144–150.
46. Karahan, A.; Buongiorno, J. J. Nucl. Mater. 2010, 396, 50. Tokiwai, M.; Horie, M.; Kako, K.; Fujiwara, M. J. Nucl.
283–293. Mater. 1993, 204, 56–64.
47. Kittel, J. H.; Ayer, J. E.; Beck, W. N.; et al. Nucl. Eng. Des. 51. Crawford, D. C.; Lahm, C. E.; Tsai, H.; Pahl, R. G. J. Nucl.
1971, 15, 373–440. Mater. 1993, 204, 157–164.
3.16 Ceramic Fuel–Cladding Interaction
K. Maeda
Japan Atomic Energy Agency, O-arai, Ibaraki, Japan
443
444 Ceramic Fuel–Cladding Interaction
EBR-II Experimental breeder reactor-II Fuel pins for FRs are generally designed with Type
EPMA Electron probe microanalysis 316 stainless steel cladding to operate with a peak
FAE Fuel adjacency effect cladding hot-spot temperature of 700 C.
FCCI Fuel and cladding chemical interaction Results and analyses of irradiation experiments
FCMI Fuel–cladding mechanical interaction related to FCCI with oxide fuels have been reported
FFTF Fast flux test facility in various technical society conferences and topical
FP Fission product and periodic reports since FCCI was first reported in
FPLME Fission product-induced liquid-metal the late 1960s. FCCI is nowadays recognized as one
embrittlement of the major factors determining integrity and life-
FR Fast reactor time of oxide fuel pins as demonstrated in numerous
GE General Electric Company in-pile and out-of-pile tests. Some mechanisms of
HEDL Hanford Engineering Development cladding attack have been proposed from the results
Laboratory of the many postirradiation investigations and thermo-
JOG Joint oxide-gaine (French) dynamic analyses of the postulated chemical reactions.
LWR Light water reactor Cladding and oxide fuels do not violently react,
MOX Mixed oxide even under a high oxygen potential condition; they
MX Nonoxide, where M stands for U þ Pu and only form a protective layer on the inner wall of
X stands for C or N the cladding. But when fuel pins are irradiated in a
NMA Nuclear microprobe analysis reactor, the additional effect of the generated FPs
N/M Nitrogen-to-metal induces cladding attack. A number of experiments
O/M Oxygen-to-metal have shown that both stoichiometric and hypostoi-
PFR Prototype fast reactor chiometric oxide fuels react with stainless steel clad-
PIE Postirradiation examination ding when irradiated in typical FRs. On the other
PNC Power Reactor and Nuclear Fuel hand, out-of-pile tests between (U,Pu)O2 or (U,Pu)
Development Corporation, currently O2x and several stainless steels have shown that no
Japan Atomic Energy Agency detectable reaction took place within the times and
RIFF Réaction à l’Interface Fissile Fertile (French) temperatures of interest for FRs. When hyperstoichio-
SIMS Secondary ion mass spectrometer metric fuel, (U,Pu)O2 þ x , was tested, cladding attack
SNR Schneller Natriumgekühlter Reactor was detected and the difference in reaction behavior
(German) was ascribed to the excess oxygen provided in the
hyperstoichiometric fuel. Inability to reasonably
extrapolate the out-of-pile results to the in-pile results
is of concern for the design of oxide fuel pins, and it
3.16.1 Introduction and Overview of indicates that the prediction of lifetime is complicated.
Ceramic Fuel–Cladding Interaction The thermochemistry of the fuel–cladding gap is
complex as well and difficult to predict because it
Ceramic fuels used for fast reactors (FRs) are oxide depends not only on concentrations of corrosive FPs,
fuels and nonoxide ceramic (MX-type, where M stands but also on major parameters such as fuel–cladding
for U þ Pu and X stands for C and N) fuels such as gap width, fuel oxygen-to-metal (O/M) ratio, cladding
carbide and nitride fuels. Ceramic fuel–cladding inter- temperature, fuel temperature, and radial temperature
actions (FCCI, fuel and cladding chemical interaction) gradient. FCCI is further regarded as sensitive to
are mainly divided according to oxide and MX-type linear heat rating and likely to change with fuel
fuels. FCCI is more complicated in oxide fuels than burnup. When the swelling of the cladding is high
in MX-type fuels because oxide fuels lead to oxidation and the fuel–cladding temperature gap is large, the
of cladding materials and formation of various oxides probability of attack is enhanced. Thus, cladding attack
of fission products (FPs) by irradiation. tends to be unpredictable, and it may be locally worse
Chemical interactions between uranium and plu- compared to the overall condition. The possible con-
tonium mixed oxide (MOX) fuels and/or FPs and sequence is complete penetration of the cladding by a
cladding materials are considered as one of the major chemical mechanism alone. In addition to this, it may
factors limiting the lifetime of fuel pins in FRs. This be considered that there is some form of stress corro-
limitation is especially important for long-term irra- sion cracking in the cladding. Actual creep strain on
diation of (U,Pu)O2 fuel pins clad in stainless steel. the cladding is from fuel and cladding mechanical
Ceramic Fuel–Cladding Interaction 445
interaction (FCMI) and/or internal FP gas pressure. nitriding of cladding, and also the formation of inter-
FCCI has been identified as a contributing factor in metallic compounds of fuel and cladding, have been
the breaches of oxide fuel pins. Observations at an investigated as a major FCCI of MX-type fuels.
axial location of a breach that was located at the The carbide fuels (U,Pu)C, which are designed to
approximate original top of the fuel column have be slightly hyperstoichiometric, will therefore be
shown extensive FCCI. The breach was a consequence in the two-phase region (U,Pu)C þ (U,Pu)2C3. The
of FCMI and internal FP gas pressure. presence of higher carbide phases carburizes the
As explanations of the observed effects of FCCI claddings. Hyperstoichiometric carbides can embrit-
have been speculative, fuel pin design could rest tle the cladding by forming grain boundary carbides
only on empirical equations rather than on fundamen- which can lead to intragranular failure of the steel
tal models. Cladding wastage equations by FCCI have after a moderate burnup. The creep and swelling
been developed for fuel pin designs. Most observations properties of stainless steels are sensitive to carbur-
of FCCI showed it to be the result of simple oxidation ization and precipitation of M23C6. As sodium can act
of the inner surface of the cladding. Three principal as a transfer agent, carbon transport rates through the
types of cladding attack in stainless steel can be distin- gap in sodium-bonded fuel greatly increase. Hypos-
guished. The first is a general oxidation of the inner toichiometric mixed carbides contain (U,Pu) metal as
surface of the cladding. The second is intergranular a second phase which may form low melting-point
attack and is the most important. The third is advanced eutectics with iron or nickel base cladding alloys.
attack which appears to be a transport of the cladding Hyperstoichiometric MN1 þ x-containing sesqui-
constituents into the fuel. It is typically seen as wastage nitride phase can cause nitrogen penetration and
of the cladding thickness in some local areas by mechan- form a reaction layer at the cladding inner surface,
ical or liquid-phase transport of cladding constituents which results in the clad nitriding. The nitriding
into the outermost oxide layer on the fuel pellets. of cladding generally decreases the ductility and
FCCI with oxide fuels has been recognized as an increases the mechanical strength. Hypostoichio-
important factor in the ability to achieve peak burn- metric MN1 x contains free metal leading to a eutec-
ups in the range of 10 at.% in FRs while maintaining tic melting reaction between the free (U,Pu) metal and
high coolant bulk outlet temperatures. However, in the cladding, which results in formation of (U,Pu)Fe2-
addition to cladding thickness losses due to FCCI, and (U,Pu)Ni5-type intermetallic compounds.
oxide fuels and FPs have the potential for reducing At present, it is clear that the knowledge base for
cladding load-bearing capabilities by mechanisms such MX-type fuels is much smaller and less detailed than
as liquid-metal embrittlement (FPLME, FP-induced that for oxide fuels, and addition to the base is a work
liquid-metal embrittlement). in progress. However, MX-type fuels merit much less
The other type of FCCI occurs for nonoxide concern regarding cladding–fuel compatibility than
ceramic (MX-type) fuels such as carbide and nitride oxide fuels.
fuels and the cladding. MX-type fuels are chemically In Sections 3.16.2–3.16.6, the causes of FCCI
unreactive to sodium coolant, so sodium may also pos- with oxide fuels are reviewed, considering the depen-
sibly be used as a medium for bonding between fuel and dence on irradiation conditions and fuel parameters
cladding instead of helium gas. MX-type fuels are as well as types of cladding material. Furthermore, the
generally irradiated at lower temperatures and lower role of corrosive FPs in the FCCI, the mechanism of
radial temperature gradients than oxide fuels, although FCCI, and the FCCI enhancement by oxygen poten-
at high linear heat rating, which results in low FPs tial are summarized. The MX-type fuel–cladding
release rate. The volatile FPs (Br, I, Cs, and Rb) do interaction is briefly described in Section 3.16.7.
not form carbides or nitrides. In particular, MX-type
fuel pins are kept with low oxygen potential at the inner
cladding surface; therefore, severe oxidative FCCI of 3.16.2 Cladding Compatibility with
the FPs is not expected. A number of irradiation experi- Oxide Fuels and FPs
ments have been performed with MX-type fuels to
3.16.2.1 Formation of Protective Oxides on
study FCCI. The compatibility with cladding materials
Cladding Materials
has been investigated in out-of-pile examinations and
thermodynamic analyses. As a consequence, unlike the Out-of-pile tests between (U,Pu)O2 or (U,Pu)O2 x
case of oxide fuels, FPs from MX-type fuels do not play and several stainless steels have shown that no
a major role in FCCI. Instead, the carburizing and detectable reaction took place for exposure times
446 Ceramic Fuel–Cladding Interaction
and temperatures of a typical FR. However, in hyper- the sensitization of stainless steel cladding occurs,
stoichiometric fuel, (U,Pu)O2 þ x , cladding attack was 500 C.2 This suggests that the corrosion resistance
detected. Excess oxygen was provided by the hyper- of the stainless steel cladding might become degraded
stoichiometric fuel, which was considered to cause because of chromium being held in carbide particles
the difference in the reaction behavior. Therefore, if in the cladding.
the fuel surface O/M ratio can be maintained just
below exact stoichiometry, oxidation of the cladding
3.16.2.2 Chemical Interaction Among
cannot take place.
Oxide Fuels, FPs, and Cladding
Of the three major constituents of austenitic stain-
less steel cladding, Fe, Cr, and Ni, chromium has As long as the protective layer stays intact, the stainless
the greatest affinity for oxygen and forms the most steel cladding is protected from further corrosion.
stable oxide. Initially, chromium begins to get oxi- However, a number of irradiation experiments have
dized when the oxygen partial pressure satisfies the shown that both stoichiometric and hypostoichiometric
equilibrium condition of the reaction fuels reacted with stainless steel cladding. Unlike irra-
diated fuel, fresh fuel does not corrode stainless steel
4=3CrðcladdingÞ þ O2 ðgÞ ¼ 2=3Cr2 O3 ðsÞ;
cladding to the same extent. It was suggested that
where the oxygen potential DG O2 ð¼ RT lnpO2 Þ of the irradiation damage might reduce the effectiveness of
fuel surface reaches 554 kJ mol1 at 727 C.1 How- this protective layer. But it was found that the extent
ever, fuel and cladding do not severely react, even of oxidation did not sufficiently increase while
when the oxygen potential is high; they only form a irradiation damage by fission fragments increased.3
protective layer on the inner wall of the cladding. The The thermodynamic tendency of oxide fuels is to
stable protective Cr2O3 thin layer prevents the fuel oxidize the cladding, and not to violently attack it
and cladding reaction from becoming thermochemi- in the absence of FPs because of the protection
cally equilibrated. provided by the oxide film formed on the surface of
In the initial stage of irradiation, the oxygen the steel.4 However, the protective layer is impaired
potential of the fuel surface rises because of oxygen by a chemical reaction of reactive FPs and oxygen
redistribution. Excess oxygen, after uranium and plu- with chromic oxide. Such evidence suggests that one
tonium fission in the fuel, leads to an increase in fuel or more of the FPs are responsible for accelerating
O/M ratio with burnup. Radial redistribution of oxy- the chemical reactions between fuel and cladding in
gen along the fuel radial temperature gradient irradiated fuel pins.3
enhances the increase of O/M ratio at the fuel sur- FCCI is the FP-accelerated oxidative attack of the
face. It appears unlikely that the oxygen potential at cladding that is frequently observed in FR fuel pins
the entire fuel–cladding interface can be kept low involving reactive FPs such as Cs, Te, and I.4 Specifi-
enough to prevent cladding oxidation throughout cally, cesium and tellurium are thought to contribute
the entire lifetime of the fuel element. Thus, a thin to the most aggressive intergranular attack modes.4,5
protective layer of oxide, mainly Cr2O3, soon forms The FCCI phenomenon is generally recognized to be
on the inside surface of the cladding, thereby physi- the result of the oxidation of chromium in the stain-
cally separating the substrate metal from the oxidiz- less steel cladding under the influence of FPs; clad-
ing medium. Further growth of this layer requires ding attack by Cs2Te has not been considered as an
that chromium ions diffuse from the substrate metal oxidation mechanism of the cladding materials.
to the outer surface of the coating or that oxygen ions In irradiation experiments, however, the protective
migrate in the opposite direction. The rates of both oxide layer is breached in some places and cladding
these processes are very slow at 727 C because of attack takes place, usually in a few isolated patches
the low values of the diffusion coefficients of the ions rather than uniformly. Whether a chemical reaction
in the oxide layer. If the thermochemically stable between components of the irradiated fuel and consti-
uniform layer is breached by mechanical forces or is tuents of the cladding can occur at all is determined by
dissolved by a component of the oxidizing environ- the thermodynamics of the reactions involved. Local
ment, the substrate metal is exposed to rapid attack. breakdown of the protective layer and subsequent
The integrity of the cladding relies on the kinetics corrosion appear to depend on the local accumulation
of the chemical attack in an environment where of observed major FPs, such as Cs and Te or I, which
oxidation is thermodynamically possible. In addition, are considered important corrosive elements.
the inner wall temperature of the cladding in The generated volatile FPs are released and accu-
an FR-MOX fuel pin reaches the range at which mulate at the fuel–cladding gap with increasing
Ceramic Fuel–Cladding Interaction 447
burnup. When fuel surface oxygen potential exceeds considered as intergranular corrosion accelerated by
the threshold necessary for oxygen transport to the sensitization which is seen as the loss of Cr by Cr23C6
cladding inner surface, excess oxygen and corrosive precipitation in the grain boundary.9 As this was the
FPs can interact with the cladding inner surface leading most aggressive form of FCCI observed along grain
to FCCI. Internal wastage of the stainless steel cladding boundaries deep in the cladding in the case of initial
is related to the complex phenomenon of corrosion O/M ratios above 1.98, this type of attack has been
established by the presence of FPs (Cs, I, and Te) and largely eliminated by using fuel with O/M ratios of 1.98
oxygen at the fuel–cladding interface. The threshold and below. By utilizing an appropriate lower O/M fuel
temperature for cladding attack is around 500 C. associated with longer irradiation for excess oxygen
in the fuel pin, a more uniform matrix interaction
tends to take place.
3.16.3 Morphology of Cladding A combined interaction form, consisting of matrix
Attack in Oxide Fuel Pins FCCI proceeded by intergranular FCCI, occurs in
fuel with moderate O/M ratio and high burnup.2
3.16.3.1 Observations of Cladding Attack
3.16.3.1.1 Deep localized cladding attack 3.16.3.1.2 FCCI at the top of the fuel column
Regions of chemical reactions between the fuel and The top of the fuel column at a cladding temperature
cladding have been generally observed, especially the near the maximum corresponds to the boundary of
hotter cladding temperature regions. Examination of fissile–fertile fuel pellets. At this location, axially
metallography samples showed occurrence of non- migrated and accumulated volatile FPs react with
uniform and deep localized cladding attack in irra- the cladding material. Axial isotopic gamma scans
diated fuel pins.6,7 Cladding attack usually occurred for high burnup pins have shown that there are larger
in an irregular manner over the inner surface of the amounts of cesium in the area of the upper insulator
cladding and in the case of intergranular attack, its pellets than in the area of the fuel.10 Because of the
depth of penetration varied from site to site. In addi- migration of cesium to the cold region in the irra-
tion, the observed deep localized interaction was diated fuel pins, cesium peaks are generally found at
usually of a different type than in the rest of the both ends of the fuel column.
sample. When access to the substrate metal was estab- These accumulations were generally related to the
lished, cladding attack by FPs occurred, either uni- formation of a phase consisting of U–Cs–O (Cs2UO4)
formly or only locally, but in some cases it penetrated at the UO2 blanket or insulator pellets, which caused
more than 100 mm into the cladding. This would be a localized inelastic deformations of the cladding (up to
significant reduction of the effective thickness of the 30 mm) at the fuel–blanket interfaces by a volumetric
cladding. Despite the reduction in cladding thickness, change.10,11 But Kleykamp12 confirmed that instead
actual fuel pin failure has rarely been observed. of Cs2UO4, a cesium uranoplutonate Cs2(U,Pu)4O12
The occurrence of a deep localized interaction of layer was formed on the grain boundaries of the
more than 100 mm in depth was observed in a sample UO2 blanket pellets in the irradiation experiment.
which had an initial O/M ratio larger than 1.98, and Furthermore, formation of compounds at the UO2
was irradiated to less than 5 at.% burnup at cladding blanket or insulator pellets led to a severe intergran-
temperature higher than 650 C.6 That suggests that ular attack of the cladding (up to 100 mm) in this
this type of interaction occurs primarily in high- region.10,11 Figure 1 shows ceramographs for a longi-
temperature regions with relatively low burnup. tudinal section removed from a fuel pin (maximum
This interaction is called deep localized FCCI burnup 14.5 at.%, 695 C cladding inner surface
and is an intergranular type of cladding attack, char- temperature, and initial O/M ratio 1.984).6 Both
acterized by a highly localized reaction product. the depth and character of the FCCI had changed
Because cladding attack tends to be random, it becomes significantly at the fissile–fertile transition zone.
locally worse compared to the overall condition. For a The maximum depth of cladding attack at the fertile
fuel with an initial O/M ratio of 1.99–2.00, there is and fissile fuel pellets was approximately 90 and
evidence that intergranular attack of sensitized stainless 135 mm, respectively. A similar localized form of clad-
steel cladding occurs in the matrix around the carbide ding attack occurred at higher temperatures of
particles in the grain boundaries.8 Microprobe exam- >600 C at the fissile–fertile interface.13 This fissile–
inations have shown this area to be depleted in chro- fertile interface reaction, termed RIFF (Réaction à
mium and manganese, with significant quantities of the l’Interface Fissile Fertile, in French), is associated
FP cesium present in the reaction product. This was with migration of volatile FPs to the end of the fuel
448 Ceramic Fuel–Cladding Interaction
column. There is no evidence of RIFF in PE-16 (high gradient in the region of the UO2 blanket or insulator
Ni alloy) and EM-12 (ferittic–martensitic steel alloy). pellets are significantly different from those at the
The occurrence of RIFF appears to depend on the region of the fissile fuel column. The predominantly
choice of cladding materials.13 radial heat transfer in the upper region of the fuel
The change in character of the cladding attack at the column and the absence of heat-generating material
top of the fuel column suggests a change in the mecha- in the UO2 blanket or insulator pellet suggest little or
nism of chemical interactions at locations of fissile and no thermal gradient across the UO2-cladding gap.
fertile fuel pellets. The reaction of fuel and cesium This effect has been taken into account in the design
suggests the presence of a high oxygen potential in of other irradiation experiments by reducing the volume
the fuel.14 Large cesium pressures, which are generally of the blanket pellets.15 The depth of chemical interac-
expected in hypostoichiometric fuel, lead to the forma- tion between cladding and fuel outside the blanket–fuel
tion of cesium uranate in the UO2 blanket or insulator interface has always been lower than 60 mm.
pellets. The FP inventory and the radial temperature
3.16.3.2 Types and Characteristics of
Cladding Attack
From the metallographic examination of stainless
steel–clad fuel pins irradiated to various burnup levels,
it is generally possible to observe the character of the
evolving cladding attack along the cladding tempera-
ture distribution in the fuel pin. The cladding attack
is classified into three types: (1) matrix, (2) intergran-
ular, and (3) combined matrix and intergranular (also
called ‘advanced’ or ‘evolved’).16 Figure 2(a) and 2(b)
show typical photomicrographs of matrix and inter-
granular types of attack in a fuel pin with Type 316
stainless steel cladding (burnup: 50 MWd kg1 M),
respectively. In addition, Figure 2(c) shows a severe
(U,Pu)O2 UO2
combined intergranular and matrix attack observed
in the fuel pin with Type 347 stainless steel clad-
ding (burnup: 140 MWd kg1 M).
The first type of cladding attack is a general oxi-
dation and is confined mainly to the shallow inner
surface of the cladding. The entire body of the inner
wall of the cladding is converted to a reaction zone
Figure 1 Cladding attack in the vicinity of the fissile–fertile containing the oxides of Fe, Cr, and Ni. In the regions
fuel interface. Reproduced from Lawrence, L. A. Nucl. of matrix attack, EPMA (electron probe microanaly-
Technol. 1984, 64, 139–153, with permission from ANS. sis) results show a depletion of iron and nickel and
an enhancement of chromium and cesium. Trace temperature, and temperature gradient. It is essential
amounts of iodine and tellurium are also observed in to develop correlations between the loss of cladding
the region of matrix attack.16 When this type of clad- strength and the various parameter groups such as
ding attack evolves, there is a definite segregation of irradiation conditions and fuel specifications. The
the cladding constituents in the reaction product layer. qualitative characteristics of FCCI and the observed
The reaction product of the matrix attack is a mixture effects of various fuel and irradiation parameters on
of metal particles and nonmetallic compounds in the FCCI are described next.
fuel–cladding gap. In addition to the three major con-
stituents of the cladding, the reaction zone contains the 3.16.4.1.1 Fuel parameters
FPs (Cs, Mo, and lesser amounts of I, Te, and Pd). The Severity and frequency of internal cladding attack
reaction zone does not appear to contain the heavy appear to be independent of both fuel form and fuel
metals (U and Pu), and neither does the cladding. density. In the case of vibrocompacted fuel, reduction
The attack on the grains is uniform with no strong in fuel density might contribute to increased FP
preference for attack along the grain boundaries. release and to eased radial migration via pores in
The second type of cladding attack is penetrating the fuel, which has a minor influence on the severity
the stainless steel cladding along grain boundaries of the cladding attack (see Chapter 2.02, Thermo-
and it is the most relevant for fuel pin failure. Inter- dynamic and Thermophysical Properties of the
granular attack occurs where the steel is sensitized. Actinide Oxides).
Such attack on the area of chromium depletion from Annular pellet fuel having a theoretical intrinsic
a steel layer adjacent to the grain boundaries by density of approximately 96%, which corresponds to
precipitation of carbides at the grain boundaries is a smear density of 80%, showed no clear difference in
in accord with metallographic observations. Opening cladding attack in comparison with vibrocompacted
of the grain boundaries from the cladding inner sur- fuel of the same smear density.20 From the results, a
face indicates that attack has occurred along them. general increasing trend of the depth of cladding
In addition, metallic and nonmetallic reaction attack at cladding temperatures above 500 C at
products are also detected in the fuel–cladding gap. approximately 3 and 5 at.% burnup was indicated in
This indicates that the grains have been chemically both pellet and vibrocompacted fuels. On the other
attacked, as evidenced by the roughened surface. hand, Batey and Bagley21 showed that the cladding
The third type of cladding attack is the combined attack in vibrocompacted fuel was significantly more
matrix and intergranular attack that is characteristi- severe in comparison with pellet fuels. Also, vibro-
cally observed in local areas, and is often accompanied compacted fuels pins that were irradiated at higher
by wastage of the cladding thickness caused by power levels (57–79 kW m1) showed two to eight
mechanical interaction or liquid-phase transport of times the depth of cladding attack expected from
cladding constituents in the outermost oxide layer fuel irradiations having the same inner surface clad-
adjacent to the fuel. The dissolution of iron, chro- ding temperatures.22,23 In contrast, sphere-packed
mium, and nickel in a medium of liquid cesium and fuel pins which are loaded with spheres of mixed
tellurium present in the fuel–cladding gap is known as UO2–PuO2 and UO2 have exhibited decreased
cladding component chemical transport (CCCT). It is depths of cladding attack in comparison with pellet
interesting to note that the constituents of the cladding fuels with a similar initial O/M ratio and irradiation
are not uniformly distributed in the reaction zone.17–19 history.24–30 Thus, there is no clear explanation of the
evidence for different cladding attack behavior
caused by different fuel forms.
3.16.4 Occurrence of Interaction From the results of metallographic observations,
Between Oxide Fuels and Cladding the influence of the initial O/M ratio on the severity
of cladding attack was emphasized in addition to the
3.16.4.1 Key Parameters in FCCI
influence of the cladding temperature,14 and it
Development
appeared that the type of the attack was controlled
The thermochemistry in the fuel–cladding gap is by the initial O/M ratio. The initial O/M ratio has a
complex and is difficult to predict because it depends significant influence on the depth of cladding
not only on concentrations of corrosive FPs, but also attack.31–34 Figure 3 shows the influence of initial
on major parameters such as the fuel–cladding gap O/M ratio on the depth of cladding attack.31,32
width, fuel O/M ratio, cladding temperature, fuel The effects of initial fuel stoichiometry on the
450 Ceramic Fuel–Cladding Interaction
1.0 25
O/M 1.96
O/M 1.94
0 0
900 1000 1100 1200 1300
(a) Maximum cladding temperature (⬚F)
(210)
Rapsodie I
£ O/M 1.96
Depth of cladding penetration (mm)
100
³1.98
50
£1.96
0
450 500 550 600 650 700
(b) Cladding inner surface temperature (⬚C)
Figure 3 (a) Maximum depth of cladding attack as a function of cladding inner surface temperature. The data represent
burnup ranging from 7 to 13 at.%. Reproduced from Weber, J. W.; Jensen, E. D. Trans. Am. Nucl. Soc. 1971, 14, 175–176,
with permission from ANS. (b) O/M influence on cladding attack as a function of cladding inner surface temperature.
Reproduced from Götzmann, O.; Dünner, Ph. In Technical Committee Meeting on Fuel and Cladding Interaction, Proceedings
of the International Working Group on Fast Reactors, IWGFR/16, Tokyo, Japan, Feb 21–25, 1977; IAEA: Austria, 1977;
pp 43–48, with permission from IAEA.
characteristics of the FCCI have been examined and or through reinforcement of FP attack as catalysts.
the results suggested that hypostoichiometric fuel has Therefore, the changes in characteristics and the cor-
advantages with respect to cladding attack. Further- relation of depth of FCCI have been determined as a
more, effects of initial O/M ratio on the maximum function of the initial O/M ratio.
depth of cladding attack with increasing burnup have
been confirmed.35 3.16.4.1.2 Effect of temperature
The effects of other parameters could not be clearly The changes in characteristics and the correlation of
evidenced. This suggests that variations of the other depths of FCCI have also been determined as a
parameters are not so critical. However, fuel impurities function of the cladding inner surface temperature.
such as C, Si, Ni, or halogens might aggravate internal The experimental data for this and the results show a
cladding corrosion either by independent interactions remarkable increase in the depth of cladding attack
Ceramic Fuel–Cladding Interaction 451
Rapsodie I
RAPS.-MON.
100
s
X
X O/M ³1.98
50
X O/M £1.96
0
500 550 600 650 700 750
Cladding inner surface temperature (⬚C)
Figure 4 Measured depth of cladding attack and the results of statistical analysis as a function of cladding inner
surface temperature. Reproduced from Götzmann, O.; Dünner, Ph. In Technical Committee Meeting on Fuel and Cladding
Interaction, Proceedings of the International Working Group on Fast Reactors, IWGFR/16, Tokyo, Japan, Feb 21–25,
1977; IAEA: Austria, 1977; pp 43–48, with permission from IAEA.
with temperature above a threshold of about 500 C. swelling of FR-MOX fuel increases with burnup and
Measured depth of cladding attack in specimens of results in fuel–cladding gap closure.18,54 At further
various irradiation experiments and the results of burnup, swelling of the cladding begins to occur
statistical analysis as a function of cladding inner sur- depending on the swelling properties of each material,
face temperature are shown in Figure 4.32 The depth such as length of the incubation period. A large fuel–
of cladding attack increases with higher initial O/M cladding gap forms again, and the FPs are released into
ratio. The threshold temperature of the cladding the gap in accordance with the formation of the so-
attack is higher with lower initial O/M ratio. There called JOG (joint oxide-gaine in French).18,19,54,55 As FPs
is general agreement that the temperature threshold in the gap gradually migrate to a colder region of the
for cladding attack is an inner surface temperature of fissile column, the gap conductance should be
500 C.4,17,21,32,36–52 From out-of-pile tests, the tem- degraded.56,57 Therefore, the maximum temperature
perature threshold of cladding attack was identified increase occurred across the fuel–cladding gap at near
as between 450 and 500 C, which is consistent with a relative axial position of 0.7 in the Phenix fuel pin as
in-pile tests. shown in Figure 5. This large temperature difference
Random nonuniform cladding attack has been across the gap would lead to a thermodynamic driving
observed at all cladding temperatures down to force for cladding attack.
500 C in FR fuel pins. Although high temperature
appeared to promote more widespread attack, the 3.16.4.1.3 Effect of burnup
depth of penetration showed no consistent variation With an increase in fuel burnup, there is significant
with temperature in the range normally employed in generation and migration of FPs such as Cs, Te, I, and
FRs. Certainly, the cladding attack generally increases Mo. In addition, the oxygen potential increases with
with temperature only in the interval of 500–600 C; irradiation in the fuel pin and fuel periphery. This
however, saturation and a decrease of the cladding determines the thermochemically stable chemical
attack were observed above 600 C.37 Figure 5 shows reactions which occur between FPs and cladding
the temperature dependence of depth of cladding constituents. A quasilinear relation exists between
attack and of neutron-induced swelling in stainless the maximum cladding depth of FCCI and burnup.
steel cladding used for the fuel pins of Phenix.53 The However, the influence of burnup on the depth of FP
maximum swelling occurred near the relative axial penetration into the cladding is not clear. Table 1
position of 0.7 in the Phenix fuel pins. Generally, summarizes the characterization of FCCI as a
452 Ceramic Fuel–Cladding Interaction
Swelling
0
700 800 900 1000
Cladding inner surface temperature (K)
Figure 5 Comparison of distribution of maximum depth of cladding attack with the profile of neutron-induced
cladding swelling of fuel pin. Reproduced from Fee, D. C.; Johnson, C. E. J. Nucl. Mater. 1981, 96, 80–104, with permission
from Elsevier.
Table 1 Characterization of cladding attack as a function of O/M ratio, burnup, and cladding inner surface temperature
O/M Burnup
a
Cladding inner surface temperature.
b
Number of fuel pins and samples examined with irradiated O/M and burnups.
c
Includes 16 pins/30 samples from HEDL P-15 test.
Source: Lawrence, L. A. Nucl. Technol. 1984, 64, 139–153.
function of O/M ratio, burnup, and cladding inner the rate-determining step of the cladding attack, that
surface temperature.6 On the basis of the results of is, the corrosion rate depends on the square root of
postirradiation examinations,32,46 a slight burnup the burnup. McCarthy and Craig58 have correlated
dependency on the severity of cladding attack was the cladding corrosion depths with cladding inner
obtained from the beginning until approximately wall temperatures assuming that the cladding corro-
10 at.% (Figure 6). sion is proportional to the burnup.
Sections from irradiated oxide fuel pins showed
no consistent variation in maximum cladding pene- 3.16.4.1.4 Effect of temperature difference
tration with burnup in the range up to 10.0 at.%, as between fuel and cladding
shown in Figure 7.41 With an increase in fuel burnup, the fuel pellet is
Perry et al.16 and Batey and Bagley21 investigated swollen, and the distance between the fuel and the
cladding attack over a wide range of burnups; how- cladding is generally decreased. However, when the
ever, a clear dependency was not obtained. Götzmann cladding diameter increase is larger at high burnup, the
et al.40 have evaluated the in-pile data from the view- distance between the fuel and the cladding becomes
point that the mass transport in the reaction layer was wider again. Thus, thermodynamic conditions might
Ceramic Fuel–Cladding Interaction 453
Microns Mils
Temperature of cladding inside surface (⬚F)
800 1000 1200 1400
30.5 1.2
Measured Predicted Avg. burnup
25.4 1.0 at.%
1.0
Depth of penetration
2.1
20.3 0.8 4.6
15.2 0.6
10.2 0.4
5.1 0.2
0.0
427 538 649 760
(a) Temperature of cladding inside surface (⬚C)
100
50
(420)
0
450 500 550 600 650 700
(b) Clad inner surface temperature (⬚C)
Figure 6 (a) Measured depth of cladding attack on fuel pins of type 316 (20 % CW) cladding at different local burnup and
predicted depth of cladding attack using HEDL correlation.32 Reproduced from Roake, W. E.; Hilbert, R. F.; Adamson, M. G.;
et al. In Technical Committee Meeting on Fuel and Cladding Interaction, Proceedings of the International Working Group
on Fast Reactors, IWGFR/16, Tokyo, Japan, Feb 21–25, 1977; IAEA: Austria, 1977; pp 137–158, with permission from IAEA
and (b) measured depth of cladding attack as a function of cladding inner surface temperature at different local burnup.46
Reproduced from Götzmann, O.; Dünner, Ph. In Technical Committee Meeting on Fuel and Cladding Interaction, Proceedings
of the International Working Group on Fast Reactors, IWGFR/16, Tokyo, Japan, Feb 21–25, 1977; IAEA: Austria, 1977;
pp 43–48, with permission from IAEA.
be rearranged with a widened fuel–cladding gap attack under normal conditions would be controlled
because of the notably marked influence of cladding by rather slow corrosion kinetics. On the other hand,
deformation, in addition to the axial migrations of at cladding strain (DD/D) beyond 1% cladding strain,
FPs and oxygen. cladding attack exhibits much faster kinetics because
The effect of cladding strain on the fuel–cladding of the abundantly supplied reactive agents.
gap widening has been investigated.54,55 Figure 8(a) As illustrated schematically in Figure 8(b),59 the
shows the effect of cladding strain on the depth of interface region is considered as fuel (a) and cladding
cladding attack with increasing burnup.55 The oxy- (b) separated by a gas-filled or reaction product-filled
gen potential continues to increase with irradiation at gap which supports the bulk of the temperature
cladding strain (DD/D) less than 1%. The cladding difference. When the swelling of cladding is high
454 Ceramic Fuel–Cladding Interaction
50
10 20 30 40 50 60 70 80 90 100
Burnup (MWD kg-1 M)
Figure 7 Depth of cladding attack as a function of local burnup. Reproduced from Koizumi, M.; Nagai, S.; Furuya, H.;
Muto, T. In Technical Committee Meeting on Fuel and Cladding Interaction, Proceedings of the International Working Group
on Fast Reactors, IWGFR/16, Tokyo, Japan, Feb 21–25, 1977; IAEA: Austria, 1977; pp 64–73, with permission from IAEA.
180
900
Fuel pins with
Corrosion depth (mm)
60
Fuel pins with
unstrained cladding
A ~30 mm B
20 DD/D < 1%
0 2 4 6 8 10 12 14
(a) Average burnup (at.%) (b)
Figure 8 (a) Effect of cladding strain on depth of cladding attack with increasing burnup. Reproduced from Bailly, H.;
Menessier, D.; Prunier, C. The Nuclear Fuel of Pressurized Water Reactors and Fast Neutron Reactors: Design and Behavior;
Intercept Ltd.: UK, 1999; p 475, ISBN: CEA 2-7272-0198-2, with permission from CEA and (b) schematic illustration of
temperature gradient in fuel–cladding gap interface. Reproduced from Adamson, M. G.; Aitken, E. A.; Lindemer, T. B. J. Nucl.
Mater. 1985, 130, 375–392, with permission from Elsevier.
and the fuel–cladding temperature gap is large, clad- increasing burnup, this gap is either partly or fully
ding attack might be more enhanced by different filled with reaction products.18,55–61 The reaction pro-
thermochemical conditions. Thus, the temperature ducts in the gap are also seen to have extruded down
gradient (or lack thereof) between fuel and cladding into radial cracks at the fuel periphery, and in some
is a key parameter determining the driving force and cases islands of a phase having bright metallic ingots
occurrence of cladding attack. The majority of irra- are contained in the corrosion product. The thermal,
diated fuel pins retain a significant fuel–cladding gap mechanical, and physicochemical properties of the
over the greater part of the fuel column. But with oxide fuels vary in a continuous manner owing to
Ceramic Fuel–Cladding Interaction 455
the presence of the reaction products, which thus inner surface to a maximum depth of 4 mil (102 mm)
directly influence fuel temperatures, cladding attack, because of fuel pellets in the top half of the fuel
and fuel–cladding interaction. The physical and column which each contained 4800 ppm of nitrogen.
chemical states of the different FP compounds deter- Cracks in the cladding were occasionally observed in
mine the volume occupied by these species and there- or near the nitrided layer, and they were caused by
fore swelling, that is, the volume increase of fuel pellets the differential expansion of this layer. However, pre-
resulting from fission. cipitates were observed along the cracks between
the nitrided layer and the remainder of the cladding,
3.16.4.1.5 Effects of cladding materials which were suggested to be chromium nitrides.
Corrosion resistance of three major types of stainless Although nitriding of the cladding materials would
steel claddings, M316 (austenite), FV548 (ferrite), and appear to complicate the situation of cladding attack,
Nimonic PE16 (high Ni), have been investigated in an the corrosion resistance of steels should be decreased
irradiation experiment.21 The incidence and severity of by the precipitation of chromium.63
cladding attack were almost identical or indistinguish-
able in oxide fuel pins irradiated under similar condi-
3.16.4.2 FCCI Model and Wastage Equation
tions irrespective of cladding type. Furthermore,
corrosion of cladding materials in the fully solution Several mechanism models have been proposed to
treated or 20% cold-worked conditions were investi- explain FCCI in MOX fuels based either on oxidation
gated for M316 and M316L stainless steel cladding or materials transport processes. However, the fuel pin
pins. The influence of cladding heat treatment on corro- design only varies with respect to empirical equations
sion sensitivity was not detected (see Chapter 2.08, rather than fundamental models because explanations
Nickel Alloys: Properties and Characteristics; of the observed effects of FCCI are speculative. There-
Chapter 2.09, Properties of Austenitic Steels for fore, FCCI wastage equations for design have been
Nuclear Reactor Applications; Chapter 4.02, Radi- based on collected in-pile test data. Wastage correla-
ation Damage in Austenitic Steels; Chapter 4.03, tions for fuel pin design were developed by relating the
Ferritic Steels and Advanced Ferritic–Martensitic loss of effective cladding thickness to irradiation and
Steels; Chapter 4.08, Oxide Dispersion Strength- fabrication parameters of the fuel pins. The following
ened Steels; and Chapter 4.04, Radiation Effects in wastage correlations, expressed using penetration
Nickel-Based Alloys). depth, have been proposed.
The corrosion susceptibility of M316 stainless The ANL (Argonne National Laboratory) correla-
steel (carbon content 0.05%) and M316L stainless tion, which was made by fitting the HEDL (Hanford
steel (carbon content 0.02%) was also investigated.21 Engineering Development Laboratory) data, was
Although carbide precipitation and associated chro- developed in order to evaluate in detail the relative
mium depletion at grain boundaries should facilitate importance of the temperature difference across the
intergranular attack, it was indicated that the initial fuel–cladding gap. The average depth of cladding
carbon content was not dominant among the factors attack was obtained by the following correlation:
determining corrosion resistance. Although carburiz-
Ave:WDepth ðANL; mmÞ ¼ 3 105
ing or nitriding of the cladding by transfer of carbon
and nitrogen from the fuel occurred, neither promo- B½4:4 104 exp 31200=TðfuelÞ
tion nor prevention of FCCI was shown.21 2:7 103 exp 24600=TðcladdingÞ ½1
On the contrary, out-of-pile results showed that
carbon may influence the nature of FP-induced clad- where B ¼ local burnup (at.%), T(fuel) ¼ temperature
ding attack by saturating alloy grains near the of fuel outer surface (K), and T(cladding) ¼ temperature
cladding surface in the form of intergranular carbide of cladding inner surface (K).
precipitates.62 Heavily carburized regions of the clad- Another wastage correlation developed at ANL64 is
ding material were chemically attacked by Cs–Te shown in eqn [2] and is based on EBR-II (experimental
mixtures; the result was an attack zone with a more breeder reactor-II) and FFTF (Fast Flux Test Facility)
uniform appearance than the deep intergranular type data. The EBR-II data included 305 data sets from
observed with alloys of large grain size. 104 fuel pins with Type 316 stainless steel cladding
The effects of nitriding of cladding materials on and 24 data sets from ten fuel pins with D9 stainless
the attack were identified in an irradiation experi- steel cladding, irradiated to approximately 17 at.%
ment.8 The cladding was irregularly nitrided on the burnup at T values up to 730 C. FFTF data included
456 Ceramic Fuel–Cladding Interaction
78 data sets from a mixed total of 24 pins irradiated cases that high initial O/M ≧ 1.98 and high burnup ≧
to about 15 at.% burnup at T values up to 685 C. 5 at.% are involved:
WDepth ðANL; mmÞ ¼ 0:5070 ðO=M 1:935Þ Max:WDepth ðGEðCCCTÞ; mmÞ ¼ 4:14 105 B 0:5
B ðT 705Þ ½2 expð9776=T Þ þ 0:86ðT 873Þ0:5 ðB 4Þ ½7
Here, O/M > 1.935, B > 0, and T > 705; B ¼ local where B ¼ local burnup (at.%) and T ¼ cladding
burnup (at.%) and T ¼cladding inner surface temper- inner surface temperature (K). In this equation, the
ature (K). The standard deviation of the correlation first term represents the conventional (oxidative)
was 12.5 mm. FCCI and the second term represents the contribu-
A third correlation was obtained from an tion from CCCT.
experimental relationship for a low-burnup fuel pin The SNR (Schneller Natriumgekühlter Reactor) cor-
(1.2 at.%), designated P-23A EBR-II. The average relation52 which was based on DFR (Dounreay FR)
depth of cladding attack yielded eqn [3]65: and Rapsodie (French experimental FR) data is
shown in eqn [8]. The correlation fits the average
Ave:WDepth ðHEDL; mmÞ ¼ 3:36 105 expð10063=T Þ ½3
corrosion depth and emphasizes the influence of the
where T ¼ cladding inner surface temperature (K). initial O/M ratio on the corrosion depth (i.e., a strong
Nonlinear regression analysis of the data from all increase for O/M between 1.96 and 1.98)
three fuel pins of two interim examinations of P-23A
Max:WDepth ðSNR; mmÞ ¼ 96:97½1 3:013 105
HEDL fuel pins at 2.4 and 5.0 at.% yielded eqn [4]66:
Ave:WDepth ðHEDL; mmÞ ¼ 2:43 106 B 0:517 ð2 O=MÞ4 exp½76:920=ðT 769Þ ½8
expð9806=T Þ ½4 for 1.96 O/M 2.00 and 769 T 923 K; T ¼
where B ¼ local burnup (at.%) and T ¼cladding cladding inner surface temperature (K) and O/M ¼
inner surface temperature (K). initial O/M ratio.
A later wastage correlation67 based on EBR-II data The PNC (Power Reactor and Nuclear Fuel
is shown in eqn [5]: Development Corporation) correlation which sug-
gests the maximum cladding corrosion depths for
Max:WDepth ðHEDL; mmÞ the fuel pin design is shown in eqn [9]41:
¼ 0:4521ðB þ CÞ ðO=M 1:942ÞðT 728Þ ½5
Max:WDepth ðPNC; mmÞ ¼ 4:64 104
for B 0, O/M > 1.942 and T > 728; for conditions ½9
ðO=M 1:93Þ ðT 360Þ B
outside these ranges, depth ¼ 0. Here, B ¼ local
burnup (at.%), C ¼ 12.23, O/M ¼ initial O/M ratio, where T ¼ temperature( C), O/M ¼ initial O/M
and T ¼ local time-averaged temperature of cladding ratio, and B ¼ burnup (MWd t1); but T ¼ 620
inner surface (K). (if cladding temperature 620), B ¼ 35 000 (if
One GE (General Electric Company) correlation burnup 35 000), and O/M ¼ 1.98 (if initial O/M
of the cladding attack based on 68 irradiated fuel pin ratio 1.98).
sections was developed and included in fabrication A wastage correlation developed at PFR (proto-
and operating parameters.4,66,67 This correlation is type FR),68 which was based on data (up to 10 at.%
given as eqn [6]: burnup), is shown in eqn [10].
Max:WDepth ðGE; mmÞ ¼ 2:36 108 Max:WDepth ðPFR; mmÞ
t ðO=M þ 0:001B 1:96ÞQ 2:33
ð1 pÞ ¼ 10 Maximum burnup ðat:%Þ ½10
expð14:185=1:987T Þ ½6
Relatively conservative wastage correlations [2], [5],
where T ¼cladding inside surface temperature (K), and [9] that employed major parameters such as clad-
t ¼ equivalent full power days of irradiation, O/M ¼ ding temperature, burnup, and O/M ratio are like the
initial O/M ratio, B ¼ local burnup (at.%), p ¼ smear envelopes of measured data points. Those parameters
density (a fraction of the theoretical density), and are conservatively considered relating to threshold
Q ¼ peak linear heating rate (W mm1). cladding temperature, incubation period (burnup),
Another GE model involves CCCT as shown in and threshold oxygen potential (O/M ratio). The
eqn [7]. This GE (CCCT) model applies only to the wastage correlation (10) was developed on the basis
Ceramic Fuel–Cladding Interaction 457
of experimental data, but it conservatively includes a 3.16.5 Mechanism of Oxide Fuel and
contribution from fabrication tolerances and external Cladding Interaction
corrosion (as well as FCCI).
3.16.5.1 Oxygen Potential of Irradiated Fuel
Wastage correlations [1], [3], [4], and [6–8]
obtained by fitting experimental data with major para- The occurrence of FCCI depends on the excess oxy-
meters such as cladding temperature, burnup, and gen available for interaction and on the cladding inner
O/M ratio were developed on the basis of the conven- surface temperature. The available oxygen is related
tional (oxidative) FCCI. The form of the correlation to the initial fuel O/M ratio and burnup. Thus, the
equations were mainly expressed as an exponential increase of the oxygen potential ðDG O2 ¼ RT lnpO2 Þ
temperature relationship. Thermally activated pro- due to burnup enhances the rate of the cladding
cesses assumed in cladding attack exhibit exponential attack.50,53 The chemical state of the FPs depends on
temperature dependence. In addition, the wastage cor- the oxygen potential of the fuel, a potential which in
relation [7] involves the term of CCCT contribution turn varies during irradiation, depending on the
expressed by threshold values (cladding temperature amount of oxygen consumed by the FPs and therefore
and burnup). The wastage correlation [6] includes a on the nature of the compounds that they form. FCCI
parameter of the smear density that has a large impact is thermochemically controlled mainly by the oxygen
on the fuel behavior, because of its large influence on potential (see Chapter 2.02, Thermodynamic and
the thermal conductivity of the fuel. Although much Thermophysical Properties of the Actinide Oxi-
data is collected for FCCI, the range of parameters and des; and Chapter 2.21, Fuel Performance of Fast
the uncertainty of the cladding temperature constitute Spectrum Oxide Fuel).
serious obstacles to developing consistent wastage In comparison with the fission of uranium, pluto-
correlations. The form of correlations, except the nium fission makes more oxygen available because of
correlation [7], was not chosen to reflect current higher yields of noble metals and lower yields of
knowledge of probable FCCI mechanism. The corre- oxide formers.39,49 The different fission yield curves
lations have not successfully predicted and consisted for plutonium and uranium result in greater increases
of actual corrosion behavior. Cladding wastage equa- in oxygen potential with burnup for plutonium-
tions are often used to predict the overall condition of containing fuels.69 Soon after reactor startup, before
cladding attack in fuel pins at each burnup level. The significant amounts of FPs have been generated, the
current model used in cladding wastage equations oxygen in the fuel redistributes rapidly to make the
cannot sufficiently account for localized deep attack. peripheral fuel nearly stoichiometric. Different tech-
Cladding attack depends on various corrosion niques for direct measurement of the radial oxygen
mechanisms, reaction potentials, FP transport (release, profiles have been employed in postirradiation exam-
radial and axial migration), and condensation within inations.67–76 Figure 9(a) compares the oxygen profile
the fuel–cladding gap, which is always changing during of irradiated samples obtained by an EMF method72
the course of an irradiation; thus severity of cladding with similar profiles, which were obtained by using
attack is not directly dependent on the above major different techniques (EPMA analysis of Mo and
parameters in the correlations. The kinetics of FCCI MoO2 content in metallic and ceramic inclusions74
in fuel pins during irradiation is determined through and X-ray measurements of lattice parameters75). In
consecutive processes: (a) generation and release of the figure, three different zones may be postulated to
corrosive FPs and supply of free oxygen, (b) transport exist.72 Zone a is near the central hole, in which rapid
of FPs and oxygen to the cladding, and (c) reaction depletion of oxygen occurs; this zone coincides
between the elements and cladding constituent ele- approximately with the columnar grain region. Zone
ments. The process (c) is strongly dependent on clad- b is an intermediate flat zone (or plateau). Zone c is
ding temperature. The process (b) is related to not near the periphery of the oxide pellet in which the
only fuel parameters and irradiation parameters but O/M approaches the stoichiometric value.
also nonuniform temperature within the fuel–cladding A slight amount of carbon in MOX fuel is included
gap. In addition to change in cladding temperature as an impurity during the fabrication process. It was
during irradiation, the nonuniform temperature asso- postulated that the oxygen potential at any point in a
ciated with fuel behavior produces a change in the fuel pin during use is controlled by a local equilibrium
character of FCCI. The amount of cladding wastage with a CO/CO2 mixture of constant composition,77
will be different at each burnup level because of rela- from which it follows that redistribution of oxygen
tive shifts in the processes. should occur to give a nearly stoichiometric fuel at
458 Ceramic Fuel–Cladding Interaction
2.00
Central
void
e
in zon
1.98
Oxygen to metal ratio (–)
nar gra
Temperature (⬚C)
Colum
2
1.94 Conte Kleykamp 82
emf. X-ray Mo/MoO2
-500 Burnup, at.% -120
Edge
Pu/(U + Pu) 0.2 0.3 Cente } 3.8
( ) r
1.92 Initial O/M 1.94 1.976 1.98 } 7.0
( ) -100
BU, at.% 3.4 Low 7 -400 11.2
Final (O/M) 1.99 1.976 1.994
-80
1.90 -300
0.2 0.4 0.6 0.8 1.0 900 1000 1100 1200 1300 1400
(a) Position (r/r0)2 (–) (b) Temperature (K)
Figure 9 (a) Comparison of present profile obtained with EMF measurements with experimental results of Kleykamp and
Conte et al. using the Mo/MoO2 method and X-ray techniques, respectively. Reproduced from Ewart, F.; Lassmann, K.;
Matzke, Hj.; Manes, L.; Saunders, A. J. Nucl. Mater. 1984, 124, 44–55, with permission from Elsevier. (b) Oxygen potential
measurements with the EMF cell on irradiated Phenix fuel of initial composition (U0.8Pu0.2)O1.98. Reproduced from Matzke,
Hj.; Ottaviani, J.; Pellottiero, D.; Rouault, J. J. Nucl. Mater. 1988, 160, 142–146, with permission from Elsevier.
the fuel periphery. Therefore, the oxygen potential with increasing americium content.79 In the Superfact
of the fuel periphery in contact with the cladding has experiment, four oxide targets containing high and low
the same stoichiometry in irradiated oxide fuels, concentrations of 237Np and 241Am were irradiated in
despite the different initial O/M ratios, after an equil- the Phenix reactor.80,81 The maximum depth of corro-
ibration period of oxygen redistribution. As a conse- sion was 50 mm in the SF13 (U0.74Pu0.24Np0.02O2 x)
quence, the influence of the initial stoichiometry on and SF16 (U0.74Pu0.24Am0.02O2 x) fuel pins.
cladding attack is not strong. From the data of Rand The depth of the cladding attack did not exceed the
and Markin,77 this stoichiometry at the fuel periphery limit set for standard MOX fuel pins with the same
corresponds to an oxygen potential DG O2 in the fuel– burnup.
cladding gap of approximately 418 kJ mol1 at
700 C, which is more than sufficient for growth of a
3.16.5.2 Characteristics of Major
Cr2O3-type protective oxide on the inner cladding
Corrosive FPs
surface (the oxygen potential DG O2 in equilibrium
with chromium and Cr2O3 at 700 C is 576 kJ mol1). The chemical state of the corrosive FPs depends on
The oxygen potential of irradiated Phenix fuel with an the oxygen potential of the fuel. Oxygen potential
initial composition Pu0.2U0.8O1.982 was measured78 and varies during irradiation, depending on the amount of
the results are shown in Figure 9(b). The DG O2 value oxygen consumed by the FPs and on the nature of the
increased with increasing burnup from 3.8 to 11.2 at.%, compounds that they form.
and it also increased linearly with temperature. As for Cesium, tellurium, and iodine are often called
oxygen stoichiometry of these irradiated fuels, the volatile FPs. These elements and part of the com-
DG O2 data measured for the pellet periphery pounds they form are gases at the fuel temperatures.
was regarded as giving the oxygen potential of the - Therefore, they will undergo considerable radial and
near-stoichiometric composition.72,78 axial migration to colder temperature regions. They
In addition, for MOX fuel containing small generally accumulate in the gap; in some cases, this
amounts of americium, the oxygen potential increases results in contact between fuel and cladding.
Ceramic Fuel–Cladding Interaction 459
The cesium is obtained in high yield. The large support for the presence of CsI, rather than elemental
excess of cesium over I and Te guarantees that not iodine, in the fuel pin. As other iodides, for instance, of
only almost all the iodine and tellurium are bound as Zr, Mo, or Ru are less stable,88 it can be assumed that
CsI and Cs2Te, but also that there is still a large CsI is the dominant iodide species in the fuel pin, and
amount of cesium left in the fuel pins. In FRs, the that it will escape as such when lack of cooling causes
Cs/I and Cs/Te ratios are approximately 10 and 4, the fuel temperature to rise. Finally, as the atomic
respectively.82 The other reactive FPs have very low ratio Cs/I is approximately 10, it can be expected
yields, and hence it is considered that cesium/oxygen that all, or almost all iodine is present as CsI.
reactions are dominant in the pins. Hofmann and Götzmann83 performed out-of-pile
As shown in many postirradiation investigations, reaction tests of cladding material with UO2 and UO2.08
FP-enhanced oxidation is the main cause of cladding in the presence of CsI or iodine. High-purity CsI did
attack in oxide fuel pins. Various out-of-pile experi- not react with the cladding steels even at 800 C for
ments have shown that the FPs (I, Te, Sb, Cd, In, and 1000 h. Nevertheless, reactions took place with the
Sn) directly attack stainless steel.83 Understanding cladding when the oxygen potential was above a critical
the dependence of those reactions on the cladding value. The maximum depth attained by the reaction
temperature and the oxygen potential in the system is with stoichiometric UO2 was about 10 mm at some
important to obtain a better understanding of the points. No reaction between CsI (simulated burnup
reaction possibilities of the various FPs. 20 at.%) and stoichiometric UO2 at 800 C for 1000 h
was detected; however, marked reactions took place
3.16.5.2.1 Iodine with UO2.08 (reaction depth, about 20 mm). On the
Investigations with iodine revealed that reactions other hand, reactions with UO2.08 were much reduced
with stainless steels take place at 400 C. Iodine reacts in the absence of CsI. The addition to UO2.08 of free
predominantly with Cr. Preferential reactions along iodine caused reactions up to 50 mm in depth in the
the grain boundaries of the cladding material yield a cladding material after 1000 h at 800 C. This indicated
type of reaction similar to that of pitting corrosion.83 that the oxidation of the cladding by hyperstoichio-
However, iodine is not present in elemental form, but metric fuel was considerably accelerated in the pres-
as CsI molecules which are thermodynamically stable ence of CsI and I.
under the conditions in the fuel pins. The equilib-
rium pressures of I2 and I in the fuels for various 3.16.5.2.2 Cesium
oxygen potentials are very low and much lower than A large excess of cesium over I and Te is generated by
the CsI pressure (5107 atm at 727 C).84 fission. Although almost all iodine and tellurium are
In addition, dissociation of CsI in the strong bound as CsI and Cs2Te, there is still a large amount
G-radiation field of the fuels has been postulated. of cesium left in the fuel pin. Elemental cesium has a
Cubicciotti and Davies85 have shown that free iodine high volatility, but formation of cesium uranate or
was released from thermochemically stable solid cesium molybdate lowers and determines the vapor
iodides by G-radiation. The chemical activity of pressure of cesium.
iodine released into the fuel–cladding gap was asso- Hofmann and Götzmann83 conducted out-of-pile
ciated with the amount of cesium which probably tests of cladding material with UO2 and UO2.08 in
could control the activity. the presence of Cs. Elemental cesium was compatible
Iodine in the elementary state causes severe attack with stainless steels up to 3000 h at 1000 C. Even minor
on stainless steel cladding. However, when bonded impurities of cesium with oxygen caused a reaction with
to Cs, it is not corrosive to the cladding materials. the cladding. The depth of cesium reactions showed a
Nevertheless, the state of iodine in the fuel has been dependence on the O/M ratio of the fuel. The cladding
questioned for a long time in view of the dissimilar attack by hyperstoichiometric fuel was considerably
transport of cesium and iodine in fuels,13 and their accelerated in the presence of cesium. While after
different release rates.86 Investigations of 129I and 1000 h, reaction zones of less than 5 mm were observed
137
Cs radial profiles in light water reactor (LWR) in contact with UO2.08 at 800 C, 100 mm was attained
pin sections indicated that iodine migrates slightly after the addition of Cs (simulated burnup 10 at.%). If
faster than Cs. This would contradict the possibility molybdenum was added to the mixture of UO2.08 þ Cs,
for excessive CsI formation in the fuels. However, the reactions with the cladding became weaker.
experimental evidence from release characteristics The Cs compound Cs2CrO4 reacted very vio-
and postirradiation investigations87 has provided lently with stainless steels at 800 C and 1000 h and
460 Ceramic Fuel–Cladding Interaction
grain boundary reactions took place, penetrating Te/TeO2 system is much higher than that in
more than 1000 mm. Cs2Cr2O7 also reacted with the the fuel. Formation energy of stable Cs2Te appears
cladding. The chemical interactions covered some to be approximately 376 kJ mol1, which is just
50 mm at 800 C and 1000 h.83 enough for Cs2Te to be stable at the conditions in
Formation of various cesium compounds is deter- the fuel. Cs2Te showed an appreciable volatility
mined by the oxygen potential. This indicates that at the conditions in a fuel.92 The vapor pressure
reactions between cesium compounds and cladding of Cs2Te is about 10–4 atm at 727 C, which means
occur only above a certain oxygen potential. The that an appreciable amount of tellurium is already
change in chemical constitution within the gap with present as Cs2Te in the gap at normal operating
variation of oxygen potential of a fuel (U0.7Pu0.3O2
x) conditions of the fuel pins. If gas-phase transport
in the presence of FPs (simulated burnups 2, 10, is the main path for FPs to reach the fuel surface, as
and 20 at.%) has been examined by thermochemical has been stated,93 it is important to know the vapor
calculations of thermodynamic equilibria.89 From pressure of Cs2Te.
thermodynamic studies, formation of the major cesium Pure tellurium showed marked penetration in
compounds was determined at certain oxygen poten- stainless steel up to depths comparable with those
tials of a fuel. The reactions which can buffer the observed in irradiated fuel pins at temperatures
oxygen potential within the gap are given in Table 2. below 800 C. Furthermore, out-of-pile experiments
Formation of cesium uranate and molybdate showed that the reaction of tellurium does not
was thermodynamically suggested.83 Postirradiation depend on the oxygen potential unlike in the case
examinations of highly irradiated fuel showed of cesium.94 The reactions with tellurium are clearly
Cs2MoO4 to be present on the fuel side of the fuel– functions of irradiation period and temperature.
cladding gap. By contrast, the formation of Cs2U4O12 Götzmann and Hofmann95 determined the maxi-
instead of Cs2UO4 or Cs2UO3.5 was observed in irra- mum penetration depths of tellurium in stainless
diated fuel.13 The formation of this uranate will cause steels as a function of the temperature and annealing
considerable swelling of the fuel90,91 and decrease the period. At 500 and 700 C, attack occurred along the
fuel–cladding gap width. grain boundaries of the cladding. Below 500 C a
uniform attack was observed, and tellurium migrated
3.16.5.2.3 Tellurium into the cladding by interfacial reactions. Above
Tellurium is a precursor of iodine. TeO2 cannot 700 C, the penetration depths of tellurium were
be formed because the oxygen potential of the dependent on the grain size of the cladding material.
Table 2 The reactions which can buffer oxygen potential within the fuel–cladding gap of fast reactor fuel pins
Source: Ball, R. G. J.; Burns, W. G.; Henshaw, J.; Mignanelli, M. A.; Potter, P. E. J. Nucl. Mater. 1989, 167, 191–204.
Ceramic Fuel–Cladding Interaction 461
The maximum penetration depth of tellurium was oxidize the Cr in the cladding material. Significant
greater in coarse grains than in fine grains. cladding attack is not expected until FPs (Cs, Te,
When the oxygen potential DG O2 is approximately and I) are available to participate in an enhanced
376 kJ mol1, Cs2Te is stable in the fuel. When the corrosion process. The quantities required are small,
oxygen potential reaches a sufficiently high level however, and sufficient quantities are generated
(>418 kJ mol1), Cs2Te becomes less stable than by approximately 0.4 at.% burnup to initiate the
cesium molybdate or urinate.96 Cs2Te can attack reactions. Therefore, it is expected that even at burn-
the cladding by reaction with the cladding constitu- ups as low as 1 at.%, significant reaction can be
ents, forming cesium chromate, iron telluride, and observed.55 An internal cladding corrosion phenome-
nickel telluride. Because of the dissolution of the non due to Te and I has been observed,7 which is linked
cladding materials into Te, the cladding constituents to the oxide thermal behavior at the very beginning of
diffuse into the fuel even at temperatures as low as life. This intergranular type of corrosion is very deep
400 C. In contrast, tellurium diffuses into the clad- and it is characteristic of pins having been operated at
ding material and is found at the reaction front high linear power at the very beginning of irradiation.
together with Cr.96 The fission yield for cesium is generally higher
However, the highly localized concentration of tel- than is needed to react with Te and I. Potentials
lurium required for cladding attack in FR fuel pins of tellurium and iodine are greatly reduced unless
appears to be inconsistent with the postirradiation cesium is consumed by other reactions.
investigations of numerous cladding attack regions in Figure 10 shows that tellurium-induced grain
which tellurium has not been detected.37,39,40,46 The boundary cladding attack in short-term irradiation
role of tellurium in cladding attack is largely based on tests (burnup 0.1 at.%) of annular fuel occurred in a
results from out-of-pile tests38,40,61,94,97–103 in which low density fuel pellet of 86% T.D. (theoretical den-
the average tellurium content per unit area of cladding sity) but not in a higher density fuel.7 The maximum
inner surface was much greater than that encountered penetration depth of this type of cladding attack was
in irradiation tests. 60 mm after 11 days of irradiation. Similar localized
intergranular cladding attacks are likely to appear at
the very beginning of life (from loading to an irradia-
3.16.5.3 Various Corrosion Reaction
tion of 10 days) and show very deep penetrations. This
Mechanisms
very beginning of life corrosion, termed ‘corrosion-de-
3.16.5.3.1 Corrosion early in life jeunesse’, in French, was in the form of spots of limited
Generally, postirradiation examinations of specimens area in fuel pins irradiated at high linear heating rate.55
irradiated at low power (20 kW m1) and low This cladding attack is caused by corrosive FPs
burnup (2 at.%) have shown the absence of fuel– (I and Te) accumulating in contact with the cladding.
cladding chemical attack.47,104 At low burnup, a radial Iodine and tellurium formed by fission generate Cs
redistribution of oxygen occurs early in life. Oxygen and Rb through radioactive decay after roughly
migration down the temperature gradient provides 10 days. Tellurium could rapidly form compounds
oxygen activity at the cladding surface sufficient to such as Cs2Te. However, because of the time required
(a) (b)
Figure 10 Tellurium-induced early-in-life cladding attack (burnup 0.1 at.%): (a) chromium oxide scale on cladding wall
and (b) intergranular attack with grain boundaries (60 mm). Reproduced from Götzmann, O. In Fast Reactor Core and Fuel
Structural Behaviour, Proceedings of the International Conference, Inverness, June 4–6, 1990; BNES: London, UK, 1990;
pp 1–8, with permission from BNES.
462 Ceramic Fuel–Cladding Interaction
for the formation of rubidium and cesium, the iodine evaluated for gaseous CsI based on kinetic theory.108
and tellurium elements are in excess at the very The calculations were carried out for an oxygen
beginning of irradiation. Under the influence of potential DG O2 ¼ 418kJmol1 at the cladding
high linear heat rating, these FPs migrate radially inner surface temperature of 600 C. The partial
and axially to the colder temperature regions. If, in pressure of iodine under the radiation condition was
these regions, the oxygen potential and the cladding approximately 102 Pa (107 atm), which was
temperature are high enough, significant intergranu- higher than that under the nonradiation condition
lar attack of cladding materials develops. approximately 109 Pa (1014 atm). Radiation
effect on the partial pressures of FeI2, CrI2, and
3.16.5.3.2 Iodine transport of cladding NiI2 as a function of cladding temperature is shown
constituents in Figure 11(b).108 These calculations indicated that
Free iodine reacts with cladding materials. But vapor transport of iron and chromium was possible
iodine is easily bound to Cs during irradiation, so within a FR fuel pin. And the pressure of NiI2 (under
only insignificant quantities of I are available to react the radiation condition) was so low that significant
with cladding materials. A reaction of the cladding transport may not occur.
materials with I needs a sufficiently high partial pres- Aubert et al.107 proposed a vapor transport mecha-
sure of I which is not possible from a thermodynamical nism by iodine which is similar to the Van Arkel–de
viewpoint. However, metallic inclusions of pure iron Boer process. According to the vapor transport
from the cladding by iodine transport were observed mechanism, iodine reacts with stainless steel at the
within the irradiated fuel in addition to intergranular cold region of the fuel–cladding gap to form metal
and matrix attack of cladding.16,105–107 Figure 11(a) diiodides and they diffuse through the gap into the
shows the appearance of metallic precipitates along fuel. The radiation-induced decomposition of metal
fuel cracks observed in irradiated MOX fuel.105 diiodides occurs and metal iodides are formed at the
The radiation effect on the partial pressure of I, fuel surface which corresponds to the hot regions.
involving the effect of recombination, has been This leads to transportation and precipitation of
0
PFel2 (rad)
-5
PCrl2 (rad)
-10
PNil2 (rad)
log P (Pa)
300
PCrI2 (non-rad)
800 mm
100x
-20
500
600
700
-25
PNil (non-rad)
2
elements in the stainless steel and the release of greater, and it was characterized by a definite segre-
iodine available for further reaction with the steel. gation of the cladding constituents of Fe, Cr, and Ni,
According to the Van Arkel–de Boer process, with the iron and nickel appearing in the fuel–
iodine reacts with cladding materials by the following cladding gap as a metallic layer separated from the
reversible reaction: reacted cladding by a nonmetallic layer containing
the chromium.
MðsÞ þ I2 ðgÞ Ð MI2 ðgÞ
CCCT, as its name implies, is a chemical transport
where M ¼ Fe, Cr, or Ni and MI2 is the metal iodide process in which Fe, Ni, and Cr are selectively
of M. Metal iodides decompose in the region of high removed from the cladding materials and transported
temperature. The free I2 molecules then diffuse back to the fuel surface by a melt of FPs consisting princi-
to the cladding to react with it. As these reversible pally of Cs and Te species. Tellurium of the transport
reactions on both sides of the fuel and cladding are medium is recycled in the melt and its activity in the
cyclically continued, the cladding constituents are melt is related to the ratio of chemically associated Cs
transported to the fuel surface and inside cracks by and Te (Cs:Te).
small quantities of iodine. But occurrence of this Out-of-pile experiments have provided strong
transport is possible in only relatively high oxygen evidence that the Cs:Te ratio is a critical parameter
potentials that are required to form a stable cesium that determines both severity and types of FCCI.62
uranate. The roles of Cs–Te mixtures in promoting both
In the high radiation field of FR fuel pins during FCCI and FPLME are essentially catalytic.4,62,110 In
use, radiation decomposition of CsI would release addition, under certain local conditions of Cs:Te ratio,
iodine for participation in the iron transport pro- stress, and temperature, FPLME may take place.111
cess.85,104 Some other reaction must remove cesium CCCT will only occur when Cs:Te <2:1.62 At this
from the gas phase in the fuel–cladding interface for sufficiently high tellurium activity, iron and nickel
iodine to be freed. That only iron inclusions are tellurides could be formed; however, formation of
found within the fuel is reasonably understood in chromium telluride depends on the activity of oxy-
consideration of the stability and volatility of possible gen which determines the formation of other chro-
metal iodides. The free energy of formation of NiI2 is mium oxide compounds such as Cr2O3 and Cs3CrO4.
too high for it to be formed at the prevailing iodine The dominant feature of this form of interaction is
partial pressure to transport the metal in significant the segregation of the chromium from the Fe and Ni
quantities. The transport of chromium would also be in the reaction product layer.
limited by the low gas-phase pressure because stable The tellurium activity in the gap is determined by
CrI2 is nearly three orders of magnitude less volatile an equilibrium reaction between condensed Cs2Te
than the other iodides. The mechanism by vapor and oxide fuels at its outer surface. This tellurium
transport of FeI2 in consideration of its stability and activity-controlling equilibrium reaction is very sen-
volatility is consistent with the observation of only sitive to oxygen potential or fuel stoichiometry.
iron inclusion in the fuel. It requires a sufficiently high oxygen potential to
Sufficiently stable cesium–metal–oxygen com- allow the decomposition of cesium telluride. This
pounds such as Cs–U–O and Cs–Mo–O decompose corrosion process is determined by the molar ratio
CsI and release free iodine, which depends on oxygen of cesium to tellurium (Cs:Te) in a melt of the mix-
potential84 and supports the hypothesis of iron trans- tures. The thermodynamic activity of tellurium and
port by the iodide process.109 the availability of tellurium and cesium for reactions,
are usually expressed as the Cs:Te ratio. The decom-
3.16.5.3.3 Cladding corrosion by Cs–Te position of cesium telluride occurs in the presence of
mixture the oxygen supplied by the fuel as per the following
When the temperature and oxygen potential are high reaction.
enough in the fuel–cladding gap, liquid Cs–Te dis-
Cs2 Te þ Cr þ 2O2 Ð Cs2 CrO4 þ Te
solves constituents of the cladding; separate metallic
phases are then formed in the fuel cracks and on the Because the above decomposition requires an oxygen
fuel surface. This nonoxidative mode of FCCI is potential high enough to destabilize Cs2Te, a suffi-
called CCCT.5,17–19 CCCT was seen in fuel pins ciently high tellurium activity exists for severe
with high O/M ratios (i.e., >1.97), at cladding tem- cladding attack. Because Cs2Te is less stable with
peratures >600 C, and at burnups of 5 at.% or increasing oxygen potential than Cs–Mo–O and
464 Ceramic Fuel–Cladding Interaction
Cs–U–O, the free tellurium in the gap can lead fuel–cladding gap.59 A purposely contaminated
to cladding attack by formation of tellurides of iron, Cs2Te specimen started melting at approximately
nickel, and chromium according to the following 700 C, and the melting point of Cs2Te
reactions. (810
10 C) was found to be extremely sensitive to
the presence of impurity oxygen. The changes in
Fe þ 0:9 Te Ð FeTe0:9 types of cladding attack behavior occurred over rela-
tively constant tellurium activity where the Cs:Te
Ni þ ð1 Y ÞTe Ð NiTeð1Y Þ withð1 Y Þ 0:6 ratio was varied from 4:1 to 2:1. At Cs:Te ¼ 2:1,
where the change in tellurium activity showed the
2Cr þ 3Te Ð Cr2 Te3 most significant and steepest change, the cladding
attack behavior was a type of deep localized inter-
From the results of out-of-pile experiments62,94,97,110 granular attack. The ratio of Cs:Te < 4:1 was suffi-
with various Cs:Te mixtures, a diagram of the tellu- ciently low for promotion of FCCI, and FPLME may
rium potential and Cs–Te binary solidus/liquidus occur under conditions of Cs:Te < 2:1 and an oxygen
boundaries versus Cs:Te ratio (or composition) is potential corresponding to stoichiometric or very
drawn in Figure 12(a) at 677 C (950 K) which slightly hyperstoichiometric fuel.4,62,110 The Cs:Te
was taken as a representative temperature for the ratios in the gap of irradiated pins was low (2:1)
FPLME/FAE
Cs: Te < 2:1
0 Bu ³ 0.2 at.%
10-1 é ³ 3⫻10-3 s-1
Stress or fracture strain
Oxygen potential DGTe (kJ mol-1)
Temperature (⬚C)
600 Liquid
Liquid 600
(Cs-rich)
+ FCCI
400 Cs2Te(s) Liquid
+
Cs2Te Cs Te Cs Te Te(s)
500
5 4 2 3
FPLME/FAE
Cs3Te2 CsTe
Cs(s) + Cs2Te(s)
0 400
Cs 60 40 20 Te 0 2 4 6 8 10
(a) Composition (at.%) (b) Burnup (at.%)
Figure 12 (a) Diagram illustrating the relationships between tellurium potential (DGTe), condensed phase Cs: Te ratio in gap,
and the solidus–liquidus phase boundaries of Cs–Te binary mixtures. Reproduced from Adamson, M. G.; Aitken, E. A.;
Lindemer, T. B. J. Nucl. Mater. 1985, 130, 375–392, with permission from Elsevier and (b) schematic illustrations of the
primary environmental domains for the various fuel–cladding (AISI 316) interaction mechanisms involving Cs and Te fission
products. Reproduced from Adamson, M. G.; Aitken, E. A. J. Nucl. Mater. 1985, 132, 160–166, with permission from Elsevier.
Ceramic Fuel–Cladding Interaction 465
relative to the fission yield ratio (6:1). It was pro- shows relationships between the temperature and
posed that the tellurium activity in the gap was deter- fuel average burnup.114
mined by an equilibrium reaction between condensed It has been postulated that contact of a crack tip
Cs2Te and oxide fuels at their outer surface. This and an agent would cause reduction of the energy
tellurium activity-controlling equilibrium is very sen- required for propagation of a crack.115 Out-of-pile
sitive to oxygen potential or fuel stoichiometry, in tests have been used to try to explain this type of
addition to the temperature. cladding embrittlement which is termed FAE (fuel
The cesium telluride reaction with the cladding adjacency effect). A chemical state of Cs:Te ratios <1
constituents, which is a reaction that forms cesium for embrittlement might be required for FAE.116
chromate, iron telluride, and nickel telluride, requires As for the dependence of oxygen potential, the
a high oxygen potential ðDG O2 > 418 kJ mol1 Þ.55 corrosion of cladding materials PE16 and M316
Cladding attack proceeds until Cs2Te reacts with the stainless steels by Cs:Te mixtures (1:1, 2:1, and 4:1)
FPs and forms Cs2Mo2O7 and PdTe at slightly higher was examined in sealed capsules under partial pres-
oxygen potential. sures of O2 set by metal/metal oxide couples at
Slotted ring tests in Cs–Te mixtures showed that 675 C for 168 h.117 Results of the nature of the cor-
stress had no significant influence on the grain rosion of alloys are shown in Table 3. Detailed fea-
boundary corrosion rate.112 In contrast, certain Cs– tures of cladding attack were classified into three
Te mixtures have been found to strongly embrittle types as follows. Type 1: deep intergranular penetra-
stainless steels.111 Corroded cladding materials were tion, embrittlement, and dissolution of the metal;
found to have considerably lower properties of Type 2: uniform attack of the alloy matrix causing
strength and ductility after contact with fuel com- alternate layering; Type 3: intergranular penetration,
pared with intact cladding materials irradiated to the embrittlement, and no dissolution of the metal. As for
same neutron dose.113 Schematic illustrations of the the dependence of oxygen potential on cladding
primary environmental domains for the various fuel– attack, the 2:1 and 4:1 mixtures produced a layering
cladding (AISI 316 stainless steel) interaction matrix type of cladding attack and the depth of attack
mechanisms involving Cs and Te are shown in increased with increasing oxygen potential. On the
Figure 12(b). The upper graph shows relationships other hand, cladding materials experienced severe
between stress level (or strain) and temperature intergranular attack for the 1:1 mixtures and the
for FPLME, FCCI, and CCCT, and the lower one attack seemed independent of oxygen potential.
Table 3 The nature of the corrosion of cladding alloys at 948 K of temperature by mixtures of Cs, Te, and oxygen
Alloy Cs:Te ¼ 1:1 liquid Cs:Te ¼ 2:1 solid Cs:Te = 4:1a liquid Cs:Te = 4:1b liquid
PE16 Deep intergrain Matrix layer; layers of Cs, Te, Cr Matrix layer; outer layer Deep intergrain + matrix
between Ni, Fe of Fe, Ni; inner layer of layer; Cs, Cr in gb
Cs, Te, Cr, Ti
Dissolution Corrosion by Cs + O2
Cs, Te, Ni in gb
150 mm (Cr/Cr2O3) 0 mm (Cr/Cr2O3)
150 mm (Mo/MoO2) 12
5 mm (Mo/MoO2)
150 mm (Ni/NiO) 40 mm (Ni/NiO) 15–20 mm 90 mm
Type 1 Type 2 Type 2 Type 2 þ 3
M316 Deep intergrain Matrix layer; outer layers of Fe, Ni; Matrix layer; layer of Cs, Deep intergrain;
inner layer of Cs, Te, Cr Te, Cr Cs, Cr in gb
Dissolution Between Fe, Ni Corrosion by Cs þ O2
Cs, Te, Ni in gb
200 mm (Mo/MoO2) 30
5 mm (Mo/MoO2) 48
5 mm 200 mm
120 mm (Ni/NiO)
Type 1 Type 2 Type 2 Type 3
a
With Mo/MoO2 buffer only.
b
With Ni/NiO buffer only.
gb, grain boundary.
Source: Pulham, R. J.; Richards, M. W. J. Nucl. Mater. 1990, 172, 206–219.
466 Ceramic Fuel–Cladding Interaction
3.16.6 Inhibition Methods for Oxide coexistent phases all have the potential to prevent
Fuel and Cladding Interaction oxidation of alloyed chromium: Ti/TiO, TiO/Ti2O3,
V/VO, Nb/NbO, NbO/NbO2, and Cr/Cr2O3.
The principal function of buffer–getter materials in Figure 13 shows the oxygen potential of these candi-
fuel pins is as an inhibitor acting as an oxygen sink, date inhibitors as a function of temperature in addi-
thereby reducing oxidation of the cladding. When tion to those of the oxide fuels with various O/M
oxygen is transported from the fuel to the buffer– ratios.62 Cesium reacts with the oxides of Nb, Ti, Cr,
getter material, oxidation of the buffer–getter occurs and V, and affinity of cesium is in the following order:
instead of cladding materials. Candidate materials Nb > Ti > Cr ¼ V.118
can be separated into buffers and getters by consider- Experimental verification of inhibitor effective-
ing their relative affinity for oxygen. Selection and ness was done to confirm other pertinent physical
placement of buffer–getter materials in fuel pins are and chemical properties, neutronics, and oxidation
basically evaluated relative to the chemical and neu- kinetics. Fuel pellets coated with getter materials
tronic characteristics of the materials. would be reasonable as they efficiently inhibit FCCI.
Buffers from elements such as V, Nb, and Cr are Meanwhile in addition to the kinetics of oxygen
considered, which will maintain a fixed fuel O/M trapping, the behavior of fuel and cladding contacting
ratio under given conditions. On the other hand, with getters has to be technologically considered.
getters are also considered from elements such as When the oxygen potentials increase, formation of
Zr, U, Ti, and certain rare earth metals, which cesium uranate causes a deformation of the cladding
will reduce the fuel O/M ratio from the initial near the fissile–fertile pellet transition zone.8,119,120
value. Thermal properties of the fuel are generally The cladding attack near the fissile and fertile
degraded; as a consequence, getters lower the O/M pellet transition zone has been reported to be signifi-
ratio of the oxide. At typical temperature ranges of cant.40,95,121,122 For these reasons, the oxygen poten-
the cladding in FR fuel pins, a buffer–getter additive tial should be kept at a sufficiently low value.
must maintain the fuel O/M ratio of the periphery Therefore, placing oxygen getters in the form of
below the initial value in order to protect Cr in pellets at the end of the fissile column could be a
the cladding from oxidation. The following pairs of possible method to control the oxygen potential.
1.96
2.000 1.95
1.94
1.93
4 2
−100 1.
99 1.99 −418
9 96
1.
K)
O5 96 (U
/Nb 2 1.9
O2
Nb O3 1.92
Cr 2
Cr/
−150 /Nb
O2 −628
NbO O
N b
Nb/
Fuel center line
O temperature
V/V
O3
/Ti 2
TiO
−200 iO −837
Ti/T
Furthermore, niobium foils were successfully applied formation of intermetallic compounds. Although
to reduce or suppress oxidizing grain boundary attack mixed carbides are more impure and inherently less
of the cladding, when they were placed between stable chemically than MOX, carbide fuel presents
the fuel and cladding in out-of-pile experiments fewer compatibility problems than oxide fuels (see
performed at 700 and 800 C for 1000 h.40,121 Porous Chapter 2.04, Thermodynamic and Thermophy-
niobium and titanium metal pellets were used in sical Properties of the Actinide Carbides; Chapter
feasibility irradiation tests as oxygen absorbers at 3.03, Carbide Fuel).
the ends of the fuel column.118 However, they were
ineffective in gettering excess oxygen of the fuel 3.16.7.1.1 Chemical reactions with FPs
pin or in reducing the average O/M ratio of the In carbide fuel pins, the oxygen potential is too low
fuel column. for the formation of complex oxides. In addition,
The effectiveness of niobium metal coatings on an increase in carbon potential during burnup is
the inner cladding surface to act as an oxygen getter not to be expected. As the FPs chemically react
was examined in an in-pile test.123 The niobium with carbon, the total carbon-to-metal (C/M) ratio
coating on the inner cladding surface worked well decreases with increasing burnup. FP attack on the
in reducing the oxygen potential. However, it was cladding of a carbide fuel pin cannot cause severe
found that niobium reacted with noble metal FPs interaction.
and formed relatively stable intermetallic phases. Out-of-pile compatibility tests using the burnup
simulated hyperstoichiometric uranium carbide
(UC1 þ x) fuel containing UC2 as the second phase
3.16.7 Nonoxide Ceramic Fuels and were performed with Type 1.4988 steel cladding.139
Cladding Interaction The annealings were carried out up to 5000 h at
temperatures from 500 to 900 C. Significant FP-
Nonoxide ceramic (MX-type) fuels are generally cladding material interactions were not observed in
irradiated at lower temperatures and lower radial the simulated specimen of 10 at.% burnup. Tellurium
temperature gradients than oxide fuels, although at caused some intergranular attack at a high concen-
high linear heat rating, which results in low FP tration of 20 wt% in the fuel, and it reduced carbide
release rates. The volatile FPs (Br, I, Cs, and Rb) precipitation in the cladding.
do not form carbides or nitrides. In particular, Reactions with the cladding in the presence of the
MX-type fuel pins are kept with a low oxygen poten- carbide fuel occurred only with free iodine. Most of
tial at the inner cladding surface during irradiation; the iodine available in the irradiated fuel is present as
therefore, severe oxidative chemical interaction of cesium compounds such as Cs2I2 and CsI.140 Because
cladding with the FPs is not expected. A number of the formation of cesium and iodine compounds,
of irradiation experiments to study FCCI have the corrosive effects of iodine will be reduced drasti-
been done with MX-type fuels (carbide124–133 and cally because gaseous cesium iodide molecules are
nitride134,135). The compatibility with cladding mate- thermodynamically stable. Larger additions of other
rials has been investigated in out-of-pile examina- FP elements did not cause severe reactions with
tions136,137 and thermodynamic analyses of fuel and cladding at temperatures up to 800 C.
cladding have been performed.138,139 Although there Thin reaction layers (10 mm) containing nickel,
are some exceptions, FPs from MX-type fuels have iron, uranium, plutonium, and rare earths were found
not been shown experimentally to have an important at high specific linear heating rates.140–142 Severe
role in FCCI. As a consequence, a little contribution interactions were thus not expected nor were they
of FPs to FCCI of MX-type fuels has been consid- observed. The phases observed were in agreement
ered. Unlike the case of oxide fuels, the carburizing with experiments simulating different stoichiome-
and nitriding of cladding, and also the formation of tries and burnups.143
intermetallic compounds of fuel and cladding, have
been investigated as major FCCI of MX-type fuels. 3.16.7.1.2 Formation of intermetallic
compounds
The fuel surface temperature of carbide fuels is
3.16.7.1 FCCI of Carbide Fuel
generally below 1000 C. Hypostoichiometric carbide
FCCI in fuel pins containing carbide fuel consists UC1 x forms uranium carbide and free U metal on
of clad carburization, slight reaction with FPs, and the basis of the U–C phase diagram.144–147 Plutonium
468 Ceramic Fuel–Cladding Interaction
carbide is usually hypostoichiometric PuC1 x below Table 4 Reaction between UC and metals and alloys
1000 C as determined from the Pu–C phase dia-
Al UC reacts with Al to produce UAl3 and
gram.145,146,148–151 The z carbide apparently is a near- UAl4 in 24 h at 620 C
stoichiometric compound that decomposes into e-Pu Be Pseudobinary between UC and UBe13,
and PuC at about 575 C.151 Formation of intermetallic UC–Be2C. UC dissolves 20 mol%
compounds with fuel and cladding constituents has to Be2C at 1700 C
Bi No reaction at 950 C
be considered for compatibility of fuel and cladding.
Co Possible pseudobinary, little solubility
Hypostoichiometric mixed carbides contain (U,Pu) Cr UC–Cr pseudobinary, eutectic at
metal as a second phase which may form low 1100 C. No mutual solubility
melting-point eutectics with iron or nickel base clad- between UC and Cr
ding alloys. An intermetallic phase has been identi- Cu UC–Cu pseudobinary
Fe Pseudobinary UC–Fe, eutectic at
fied in (U,Pu)C compatibility tests with Hastelloy 1100 C
X (48 wt% Ni) and Inconel 625 (61 wt% Ni) as Hf Complete solubility of HfC-UC;
(U,Pu)Ni5.132 The presence of a small quantity of free probable increase in m.p. (HfC
uranium metal has a detrimental effect on the irradi- m.p. 3900 C)
ation stability of uranium monocarbide.152,153 Mo UC þ Mo ! Mo2C at 1000 C
Na No reaction at 871 C
Investigation of compatibility of uranium carbide
Nb Complete solubility of NbC-UC; NbC
with 3d transition metals154 or with refractory melts 3485 C. Nb-UC compatible
metals136,137 showed formation of UNi5 phases. Reac- Ni UC-U6Ni pseudobinary
tions at 1000 C were observed between UC and not Pb No reaction
only Fe, Ni, and Cr, but also Nb, Th, Ti, and Zr. The Pu Large range of solubility of Pu in UC and
possibly of Pu in U2C3
free metal formed intermetallic compounds such as
Re Simple eutectic at 1850 C
(U,Pu)Fe2 or (U,Pu)Ni5 in the fuel and cladding Si UC þ 3Si ! USi3 þ C at 1000 C
interface. Sn No reaction
Out-of-pile tests were conducted for 1000 h at 600 Ta Continuous TaC–UC solid solution;
and 725 C to determine the compatibility of helium- Ta þ UC ! TaC þ U
Th Continuous ThC–UC solid solution
bonded (U,Pu)C þ 9–12 vol% (U,Pu)2C3 with aus-
Ti UC dissolves < 2 mold % TiC
tenitic steels Type 330, A286, M813, PE-16, Inconel V UC dissolves 4–9 mol% VC; VC
706, and Inconel 718 compared to Type 316 stainless dissolves no UC
steel as a reference.154 At 725 C, the alloys contain- W UC dissolves 10 mol% W;
ing >41 wt% nickel such as PE-16, Inconel 706, and W dissolves no UC
Zn UC wetted-condition not states
Inconel 718 formed a small amount of an intermetal-
Zr Continuous solid solution between UC
lic phase, containing mainly uranium, plutonium, and and ZrC, rising to ZrC m.p. of 3500 C
nickel and 3–8 mm in depth. Solid-state reactions Stainless steel No reaction at 800 C for 6 months.
between irradiated (U,Pu)C and stainless steel clad- (18Cr, 8–12Ni) 0.10 mm penetration in 6 days at
dings proceed very slowly at temperatures below 1100 C. Liquid phase at 1200 C
Zircaloy-2 Reacts in 1 h at 1200 C but not at
800 C. The reaction rate is significantly higher in 800 C – possible ZrC diffusion barrier
PE16 cladding. At temperatures above 950 C, the Nb – 40 at.% Ti Molten phase at 1200 C
solid-state reaction leads to liquid-phase formation in Inconel 0.008 mm corrosion in 100 h at 800 C
both stainless steels and PE16 and rapid cladding
dissolution ensues by the following reaction: Source: Frost, B. R. T. J. Nucl. Mater. 1963, 10, 265–300.
and mechanical strength because of cladding carbur- Hyperstoichiometric mixed carbides contain (U,
ization can determine the fuel lifetime as well as Pu)2C3 as a second phase from which carbon would
excessive cladding swelling. be transferred to the cladding. Out-of-pile examina-
tion results verify that 100% (U,Pu)2C3 fuel in
3.16.7.1.4 Key parameters of clad sodium at 800 C will carburize Type 304 stainless
carburization steel.187 Small quantities of sesquicarbide (<10%)
The carburization depth increases approximately may be acceptable because a slight amount of carbur-
linearly with irradiation time or with burnup. ization will probably have a negligible effect on clad-
The amount of carburization is dependent on carbon ding mechanical properties.188
activity (fuel stoichiometry and oxygen content), Carburization can be reduced to an acceptable
temperature (cladding temperature and fuel temper- degree in fuels with M2C3 levels below 5%.189 This
ature gradient), and on the bonding materials (helium was confirmed up to 5.7 at.%, 100 kW m1, and
gas and liquid sodium) between fuels and cladding. 650 C cladding temperature. From the investigations
of maximum depth of carburization in a group of
3.16.7.1.4.1 Carbon activity seven fuel pins that attained a burnup of approxi-
Carbon activity in austenitic stainless steels depends mately 5 at.%, the amount of (U,Pu)2C3 in the fuel
on the temperature and heat treatment history.181 appeared to have a significant effect on the depth of
The solubility of carbon in sodium increases signifi- clad carburization.
cantly on increasing the temperature.182 Hence, the Carburization effects in the Type 316 stainless
carbon activity decreases on increasing the tempera- steel cladding used for the sphere-pac mixed carbide
ture at constant carbon concentration. fuel containing up to 40% M2C3 in small spheres
The carbon activity in MC pins depends on the were studied.189 Carbon concentration profiles are
C/M ratio, but not on the presence of FPs.183 Clad shown in Figure 15 as a function of cladding
carburization with sodium- or helium-bonded fuels wall thickness. Sphere-pac mixed carbide fuel contain-
increases with increasing x in MC1 þ x . Moreover, ing >5% sesquicarbide caused severe carburization
the effective carbon activity is increased by radial and cracking of Type 316 stainless steel cladding
carbon transport in the thermal gradient and reduced at temperatures >600 C combined with linear heat-
strongly by the effect of oxygen impurities.184 The ing rates >80 kW m1. Depth of significant carbur-
oxygen content of the sodium can affect the rate of ization was up to 0.120 mm at 630 C and 0.200 mm
carbon transfer.185,186 at 670 C. Clad cracking was only associated with
Cracked
40% M2C3, cladding
temp. 610 ⬚C, 91 kW m-1
>5% M2C3, cladding
2
MAX temp. 650 ⬚C, 80 kW m-1
Carbon (wt%)
Intact
<5% M2C3, cladding
MAX temp. 650 ⬚C, 80 kW m-1
0
0 10 20 30 40 50 60 70 80 90 100
Fuel Clad wall thickness (%) Sodium
side side
Figure 15 Cladding carbon profiles for sphere-pac fuels with various sesquicarbide contents. Reproduced from
Stratton, R.W.; Smith, L. In Advanced LMFBR Fuels, Proceedings of Intl. Meeting on Advanced LMFBR Fuels, Tucson, AZ,
Oct 10–13, 1977; Leary, J.; Kittle, H., Eds.; ERDA-4455, Amer. Nucl. Soc.: La Grange Park, IL, 1977, 348–350, with permission
from ANS.
Ceramic Fuel–Cladding Interaction 471
temperatures >600 C and heavy carburization >20% Migration of carbon from the fuel to the cladding
penetration. Grains and grain boundaries were heavily would occur if the carbon activity in the cladding is
decorated with M23C6 and M7C3 but there was no less than in the fuel.164 The carbon activity in the
large carbide-rich layer. Clad cracking occurred. In cladding depends strongly on its structure and chro-
contrast, sphere-pac mixed carbide fuel containing mium content; it equals approximately 103 for AISI
<5% sesquicarbide caused only light carburization 316 stainless steel and 102 for AISI 410 stainless
without failure in a burnup to 5.7 at.% with peak clad steel.197,198 The carbon activity in UC is approximately
temperatures of 650 C and a mid-plane peak linear 102.199 In an exposure experiment of UC containing
heating rate of 100 kW m1. Depth of light carburiza- U2C3 phase to sodium for 2000 h at 700 C, carbon
tion ranged from 0.050 mm deep at 600 C up to transfer did not occur.196 This was explained by com-
0.240 mm deep at 655 C. The carbon was mainly in parison of carbon activity between fuel and cladding.
the grain boundary zones as a fine precipitate of M23C6. The carbon activity in the carbide in equilibrium with
Some fine M23C6 precipitation was also observed U was calculated to be 3 106 at 727 C, and the
within the grains near the fuel–cladding interface. carbon activity in (Cr,Fe)23C6 in equilibrium with aus-
In the hyperstoichiometric carbide fuel con- tenitic stainless steel is 2 102 at 727 C.
taining impurity oxygen, transport of carbon and The carbon activity in the fuel is influenced by
oxygen takes place to cooler fuel regions as gase- oxygen and nitrogen impurities in the carbide.
ous carbon monoxide along the temperature An ‘equivalent’ carbon content is evaluated by the
gradient.190 Carbon monoxide causes further carbur- following equation by taking into account nitrogen
ization of the inner cladding surface.176 Whereas ura- and oxygen content:
nium monocarbide containing oxygen (O2 ¼ 0.1%)
C ¼ wt%C þ 0:75wt%O þ 0:86wt%N
has not been observed to carburize the inner cladding
surface, distinct effects have been found with hyper- where only the molecular weights are considered.
stoichiometric carbides containing oxygen (O2 > 0.3%). Stoichiometric fuel has an equivalent carbon content
Out-of-pile chemical compatibility experiments of 4.8 wt%. The ‘equivalent’ carbon content suitable
between plutonium-rich MC fuels and Type 316 for FRs was evaluated as between 4.9 and 5.2 wt% of
stainless steel (20% cold-worked) cladding were hyperstoichiometric carbide, and the oxygen content
carried out at 700 C for 1000 h by using a capsule of carbide for helium-bonded fuel was limited to
containing four fuel pellets sandwiched between 0.4 wt%.156 The amount of oxygen is specified for
disks of the cladding. The pellets containing up to sodium-bonded fuel in accordance with the ‘equiva-
0.7% oxygen and 20% M2C3 caused insignificant lent’ carbon content upper limit of 5 wt% for phase
carburization of the cladding to less than 12 mm stability reasons, which reduces the risk of unaccept-
in depth. However, the pellets containing relatively able clad carburization.200 A preliminary model for
high oxygen (1%) and high M2C3 (60%) caused the clad carburization by sodium-containing carbon
carburization of the cladding to depths of up to has been proposed.197 It was assumed that the carbon
90 mm.191 The depth of clad carburization in an activity would establish the equilibrium and that car-
identical test carried out with uranium-rich MC con- bon diffusion into the cladding along the gradient of
taining 1% M2C3 and around 0.2% oxygen was the activity would control the growth of carburized
150 mm without breaching in the Type 316 stainless phase. Chromium-rich carbide would be precipitated
steel cladding.192,193 so as to achieve equilibrium of dissolved chromium
In addition, dissolved oxygen in liquid sodium has in the matrix. This model also assumed the carbide
caused some surface oxidation of the fuel and mass formation of three types Cr23C6, Cr7C3, and Cr3C
transfer of carbon, which lead to the decomposition according to chromium and carbon activities.
of UC2 platelets and to carburization of the clad-
ding.194 Hyperstoichiometric (U,Pu)C1 þ x generally 3.16.7.1.4.2 Cladding temperature
includes (U,Pu)2C3 instead of (U,Pu)C2 as a second The cladding temperature plays the most important
phase. The U2C3 phase is more compatible with role in clad carburization. The temperature gradient
sodium than the UC2 phase, because the carbon in the fuel causes the excess carbon to migrate to the
activity of U2C3 is lower than that of UC2.195 colder zones of the fuel.201
In addition, Pu2C3 phase has carbon potential lower Carburization depth in various stainless steels at
than that of U2C3 at all temperatures and for Type medium burnup of sodium-bonded carbide fuel pins
316 stainless steel up to 705 C.196 as a function of temperature is shown in Figure 16.
472 Ceramic Fuel–Cladding Interaction
Carburization depth generally increases depending least below 630 C.202–205 However, carburization
on temperature increase. Cold-worked Type 316 in cold-worked stainless steels shows a steep increase
stainless steel shows less carburization in comparison above about 630 C.
to solution-annealed Type 316 stainless steel at Carbon penetration into the cladding as measured
by EPMA is shown in Figure 17. Carbon penetration
400 seems to follow a simple diffusion process with a
Type 316 stainless steel (solution annealed) + constant surface concentration.202 Clad carburization
8–12 at.% corresponds to conditions of chemical diffusion.
Type 316 stainless steel (cold-worked)
Because a part of the carbon precipitates as
+ 3–5 at.%204,205
(Fe,Cr)23C6 during carbon penetration, the effective
+ 5 at.%204,205
300 + 10–12 at.%203 diffusion coefficients are lower than the value of
DIN 1.4970, cold worked carbon tracer diffusion.206 The absolute carbon con-
Carburization depth (mm)
P1 = 870 W cm-1
Burnup = 7.4 at.%
Tclad = 810 K
%M2C3 = 3.8%
Carbon (%)
0
0 100 200 300 400
Depth (µm)
Figure 17 Typical carbon concentration profile in the Type 316 stainless steel cladding. Cladding was carburized by
mixed carbide containing 3.8% sesquicarbide at 550 C of cladding temperature and burnup of 7.4 at.%. Reproduced from
Ronchi, C.; Coquerelle, M.; Blank, H.; Rouault, J. Nucl. Technol. 1984, 67, 73–91, with permission from ANS.
Ceramic Fuel–Cladding Interaction 473
0.8
0.7
0.6
AERE Harwell
0.4
Carbon concentration
Eir
0.3
0.2
0.1
0
0 10 20 30 40 50 60 70 80 90 100
Cladding thickness (%) Standard austenitic M23C3 MC3
Sodium Fuel microstructure (0.04 w/o carbon)
side side Sodium side Fuel side
(a) (b)
Figure 18 (a) Carbon concentration profile in the M 316 stainless steel cladding after irradiation with mixed carbide fuel
(EIR: measured by SIMS, AERE Harwell measured by nuclear microprobe). Reproduced from Bart, G.; Aerne, E.T.; Burri, M.;
Zwicky, H.-U. Nucl. Instrrum. and Meth B, 1986, 17, 360–362, with permission from Elsevier. (b) Appearance of carburized
stainless steel cladding specimen. The specimen was carburized by mixed carbide fuel (burnup 42 GWd t1). Reproduced
from Bart, G.; Aerne, E.T.; Burri, M.; Zwicky, H.-U. Nucl. Instrrum. and Meth B, 1986, 17, 360–362, with permission from
Elsevier.
(oxygen ¼ 1000 ppm) showed that the excess carbon The extensive data available for helium- and
of the M2C3 phases migrated toward the cladding, sodium-bonded fuels210–220 consistently show clad
so the M2C3 content increased in the cold outer carburization. The influence of the carbon content
zone.176 Clad carburization was severe compared to of the fuel has been identified both in helium- and
the case with carbide fuel with small oxygen content. sodium-bonded pins. Clad carburization occurs less
Therefore, radial redistribution of M2C3 by migra- in helium-bonded fuels, but more in sodium-bonded
tion of excess carbon toward the pellet periphery has fuels where excessive carburization can occur with
to be considered. Carburization of solution-annealed hyperstoichiometric (U,Pu)C1 þ x .221,222 The depth of
Type 316 stainless steel was measured across the carburization of Type 316 stainless steel and Incoloy
cladding thickness of an irradiated sodium-bonded 800 in sodium-bonded carbide fuel is two to three
fuel pin with a low initial M2C3 content 2%.209 The times greater than in comparable helium-bonded
thickness of the carburized zone was 0.120 mm at fuels.13,216
850 C for a burnup of 27 MWd kg1, which depends Behavior of carbon redistribution in helium- and
on cladding temperature. The mean carbon concen- sodium-bonded pins is different. The rate of carbon
tration in the cladding was about 2000 ppm. A carbon transport toward the cladding is determined by
diffusion coefficient in the fuel calculated using the the carbon diffusion with sodium-bonding or with
above results was D 3.1011 cm2 s1, which sug- gas-bonding.181 As sodium can act as a transfer agent,
gested that diffusion mechanisms may be enhanced carbon transport rates through the gap in sodium-
with irradiation. bonded fuel greatly increase. In contrast, carbon trans-
fer is slow in helium-bonded fuel with a gas-filled
3.16.7.1.4.3 Bonding medium gap. In helium-bonded pins, carburization seemed to
MX-type fuels are chemically unreactive to sodium become detrimental only after the establishment of
coolant, so sodium may also possibly be used as fuel and cladding contact. The hyperstoichiometric
a medium for bonding between fuel and cladding carbide fuel containing high oxygen content can
instead of helium gas. The clad carburization can be cause carbon redistribution by carbon monoxide and
developed in direct contact with fuel, or through the increase M2C3 content in the pellet periphery which
bonding mediums between the fuel and cladding gap results in severe clad carburization. Carburization
such as helium gas or liquid sodium. rates increase with increasing M2C3 content of the
474 Ceramic Fuel–Cladding Interaction
helium-bonded carbide and gas-bonded oxide pins 400 800 1200 1400
and appears to saturate at approximately 0.4 wt%.156 Temperature ( ⬚C)
This level of carburization should not affect Figure 19 The carbon potential of various cladding
mechanical properties significantly. components and UC and PuC as a function of temperature.
Reproduced from Matzke, Hj. Science of Advanced LMFBR
Fuels; Elsevier, UK, North Holland, Amsterdam,
3.16.7.1.4.4 Buffer addition
Netherlands, 1986, with permission from Elsevier.
An Ellingham-type diagram of some cladding con-
stituent elements and MX fuel systems is shown
in Figure 19.226 The free energies of formation of in the system would determine whether the move-
(U,Pu)C, (U,Pu)2C3, and carbides of cladding con- ment of carbon occurred in the sodium.
stituent elements show that chromium forms the To have good compatibility and good irradia-
most stable carbides. Therefore, carbon transport tion behavior, it is thus desirable to stabilize the
from the fuel to the cladding should occur at a rate monocarbide. One way is to retain the NaCl struc-
determined by the kinetics of the mechanism of ture by alloying with a material such as TiC or ZrC,
carbon transfer. The driving force for a chemical or forming carbonitrides or carbosulfides. However,
reaction is the tendency to decrease the chemical sodium- and vacuum-bonded compatibility tests at
potential. Carburization of cladding steels is a diffu- 600 and 700 C between stainless steels and carbosul-
sion controlled process in which the difference fide fuels of differing sulfur contents and stoichiome-
between the thermodynamic stability of carbides tries revealed that the amount of clad carburization is
in the carbon source and carbides produced in the a linear function of only the total nonmetal content
cladding is a driving force. (C þ O þ N þ S) of the fuel, and does not depend on
In an immersion test for about 336 h at 650 C, the actual sulfur content of the fuel.227 No apparent
carbon was transferred from carburized Type 316L differences in carburization of Type 316 stainless steel
stainless steel to V–15 wt% Cr–5 wt% Ti alloy.176 for a range of carbonitrides with C/(C þ N) ratios
The free energy of formation of titanium carbide varying between 0.85 and 0.20 were identified.228
(TiC) is about –172 kJ mol1 as compared with Another way to stabilize the monocarbide is to make
about 71 kJ mol1 for chromium carbide (Cr23C6) a ternary metallic addition to obtain a thermodynami-
formation. This indicated that carbon was transported cally stable three-phase field in which UC is a stable
in static sodium from a carbon-bearing material phase. From a study on carbonitride, it was concluded
to another material in order to form a more stable that it was not easier to prepare single-phase carboni-
carbide. Thus, the thermodynamic stability of tride than pure carbide, at least up to 15 mol% UN.229
Cr23C6-type carbide relative to the other carbides The principle of the buffering action of several metals
Ceramic Fuel–Cladding Interaction 475
or carbides has been described by Chubb.166 Among in Cr-modified fuels (1300–1350 C) compared
other materials studied, Fe, Cr, V, Re, and W were with those in Fe-modified fuels (1000–1050 C).
found to be effective as buffering agents. Fe-modified mixed carbide fuel may contain the
Attempts to stabilize the MC phase by buffer ternary (U,Pu)Fe2–Fe-(U,Pu)C eutectic which
addition were made using a buffer acting as a carbon melts at 1045 C.
acceptor from M2C3 and as a carbon donor to free Investigations into the constitution of these sys-
metal. The compounds used as buffers for UC fuel tems and systems using W and V additions have
were W–UWC2,166 Mo–UMoC2,230 V–UVC2,231 been reported.166,237 Out-of-pile compatibility tests
Cr–Cr23C6,231 Fe–UFe2,233 and Ti–TiC.230,234 At- of hypostoichiometric UC, Cr23C6-modified UC, and
tempts for buffering (U,Pu)C by using W, Fe, Cr, Fe-modified UC with stainless steels in sodium
Mo, and Ti were also made.230,234 Molybdenum was showed that fuel–cladding reactions were eliminated
reported to stabilize C23C6-type carbides in austen- by the additions.238 Similarly, modified (U,Pu)C has
itic stainless steels.235 also been shown to be stable in sodium.239 In these
Although sodium-bonded cladding materials such latter tests, modified (U,Pu)C fuels in sodium at
as AISI 316 and AISI 410 stainless steels were carbur- 800 C showed no metallographic evidence of car-
ized by uranium carbide after 2–16 weeks at 800 C, burization of Type 304 stainless steel after approxi-
both Cr and V additions to uranium carbide are mately 1400 h of testing.
effective to improve the compatibility of the carbide A number of researchers have suggested the use
fuel with the stainless steel claddings. The carbon of buffering agents or stabilizers to minimize the for-
activity in uranium monocarbide has been buffered mation of the potentially deleterious second
with Cr or V to avoid carbon transfer from carbide phases.164,189,238,240,241 Chromium, vanadium, and iron
fuels to the cladding.164 seem to be the three elements referred to most consis-
AISI 316 stainless steel, in contact with three- tently. They promote the formation of a polyphase fuel
phase alloys UC–Cr–Cr23C6, did not show any sign structure without the compatibility problems of either
of carburization after 20 weeks at 800 C. But, when free metal or higher carbide phases. Van Lierde et al.242
insufficient Cr was present to keep the three-phase suggested that very fine precipitates of vanadium car-
composition, formation of higher chromium carbides bide would work well as a buffering agent.
(Cr7C3) and carbon transfer to the steel occurred.164
Compatibility tests between vanadium stabilized
3.16.7.2 FCCI of Nitride Fuels
UC and AISI 410 or 316 stainless steels were per-
formed and the compatibility was better depending FCCI of mixed nitride fuels is similar to carbide fuels
on increased vanadium content. However, a good except regarding clad carburization. FCCI in nitride
compatibility between uranium carbide and vana- fuel pins consists of clad nitriding, slight reactions
dium alloys could be observed only if the reaction with FPs, and formation of intermetallic compounds.
rate was very slow.236 Clad nitriding by nitride fuels is less concerned with
It was also reported that chromium addition to FCCI than clad carburization by carbide fuels, and
modified monocarbide fuels was effective to improve there have been few problems reported on compati-
the compatibility of the carbide fuel with stainless bility with nitride fuels and cladding (see Chapter
steel claddings.236 2.03, Thermodynamic and Thermophysical Prop-
The burnup of (U,Pu) atoms in-pile tends to erties of the Actinide Nitrides and Chapter 3.02,
shift the fuel toward the hyperstoichiometric struc- Nitride Fuel).
ture. Therefore, a modified fuel was proposed to
buffer the carbon activity against irradiation-induced 3.16.7.2.1 Chemical reactions with FPs
changes and minimize the formation of potentially In nitride fuel, oxidation of the fuel is excluded.
deleterious second phases. Two systems employ- The information on the behavior of major volatile
ing Cr- and Fe-based additions were selected for FPs such as Ba, Cs, and Te is limited.243–246 Chemical
examination. The Cr-based modification, which forms in the irradiated fuel are indicated as Cs(1),
yields fuel containing stoichiometric (U,Pu)C with Ba3N2, BaTe, and Cs2Te by the thermodynamic cal-
Cr and chromium carbides as grain boundary sec- culation.247 The halogens (Br and I) and alkali metals
ond phases, is considered the better system of the (Rb and Cs) do not form nitrides but tend to form
two studied. This preference is strongly influenced compounds with the other volatile elements, such as
by the minimum melting points of second phases CsI. Only the chalcogenides (Te, Se) form nitrides
476 Ceramic Fuel–Cladding Interaction
(SeNx and Te3N4), but these probably decompose 727 C). With U2N3, only iron, nickel, and molybde-
even at low temperatures. Therefore, in irradiated num are stable on the basis of the potential for nitride
nitride fuels, Se and Te are expected to be present formation.
either in elemental form or in compounds with the
other volatile elements such as Cs2Te or CsI. 3.16.7.2.3 Clad nitriding
Chemical reactions with FPs were not observed in In the U–Pu–N system, the mononitrides show
a capsule of sodium-bonded fuel and Type 304 stain- complete miscibility. Plutonium sesquinitrides do not
less steel cladding at high cladding temperature exist in the Pu–N system. However, after Lorenzelli256
(795 C) and high burnup (15 at.%).248 The cladding reported a sesquinitride phase with a Pu/(U þ Pu)
was sensitized after irradiation; however, electron- ratio of 0.15, Potter257 suggested the (U,Pu)N phase
microprobe examinations have revealed no trace of diagram with either very small Pu solubility or with
either fuel or FPs in the cladding. up to 15 mol% ‘PuN1.5’ solubility in UN1.5 existed.
It should be noted that the most frequent composition
3.16.7.2.2 Formation of intermetallic of advanced fuels will be at about 20 mol% PuN.
compounds Also, MN fuels are usually specified to be single-
The binary systems U–N249–253 and Pu–N226 are less phased mononitrides (in contrast to MC fuels which
complicated than the systems of carbides. The main are frequently specified to contain 10–15 wt%
compatibility problems of MN with cladding are also M2C3). As the binary systems U–N and Pu–N show
because of its narrow homogeneity range. The two very narrow single-phase fields for the mononitrides,
main uranium nitrides are the mononitride UN and it seems reasonable to assume that the mononitride
the sesquinitride U2N3. Precipitations of U within solid solutions (U,Pu)N also exist over a relatively
grains or at grain boundaries are seen in the U-rich narrow composition range.
UN phase. High-purity stoichiometric MN fuel is not likely
Hypostoichiometric MN1 x-containing free to have any chemical interaction with Type 316
U or Pu should be avoided as a eutectic melting stainless steel cladding. Single-phased MN fuel also
reaction with the cladding can occur. Hypostoichio- shows excellent compatibility with stainless steel.
metric MN1 x contains free metal leading to a Plutonium sesquinitrides are not usually formed,
eutectic melting reaction between the free (U,Pu) but a sesquinitride phase can be formed with a
metal and the cladding, which results in formation Pu/(U þ Pu) ratio of 0.15 in the U–Pu–N system.256
of (U,Pu)Fe2 and (U,Pu)Ni5-type intermetallic com- Uranium-rich MN1 þ x containing a second phase of
pounds.254 An intermetallic reaction phase (probably mixed sesquinitride (M2N3) caused nitriding of Type
(U,Pu)(Ni,Fe)3) has been observed in the fuel–cladding 316 stainless steel cladding and consequent loss of its
gap.216 A low melting eutectic composed of free ductility.134 The sesquinitride in (U,Pu)N1 þ x is
metal of the fuel and the constituents of the cladding essentially U2N3 (with some Pu in solution) which
can cause melting of cladding at 1000 C in the hypos- decomposes at about 1350 C.
toichiometric sodium-bonded fuel.255 At 1000 C, the Hyperstoichiometric MN1 þ x-containing sesqui-
mixed nitride fuel appears to get oxygen from a variety nitride phase can cause nitrogen penetration and
of cladding materials; as a result, the activity of nitro- form a reaction layer at the cladding inner surface,
gen is increased by oxidizing the fuel surface. which results in the clad nitriding. Diffusion of nitro-
The prediction of a stable reaction with cladding gen into the cladding and nitride precipitation can
constituents considering the displacement of nitro- occur according to the following reaction.134,258
gen from the fuel to form a nitride was made by
M2 N3 þ 2Cr ! Cr2 N þ 2MN
thermodynamic calculations.255 Pure UN and PuN
are stable with respect to most major cladding con- As the N/M (nitrogen-to-metal) ratio of (U,Pu)N
stituents such as Nb, Ta, V, Mn, Cr, Fe, and Mo. increases with burnup, the amount of sesquinitride
Zirconium and titanium possibly form nitrides. Alu- (essentially U2N3) would increase with burnup. How-
minum is unstable with respect to UN, and its addi- ever, excessive nitriding of cladding by nitride fuels
tions must be restricted. Nickel is stable with respect irradiated to high burnups have not been reported.
to both UN and PuN on the basis of nitride forma- Irradiation tests confirmed the excellent compatibility
tion except for the formation of UNi5. However, of nitrides with stainless steel.176,248,259–271
(U0.8Pu0.2)N is unstable with respect to pure nickel Oxygen impurity in MN fuel plays a vital role
and Inconel-625 at low temperatures (527 and in controlling nitriding of Type 316 stainless steel
Ceramic Fuel–Cladding Interaction 477
cladding. The nitrogen activity is decreased by oxy- formation of a higher nitride phase. No reactions in
gen impurities in the fuel.134 Some 2–3 wt% of oxide the fuel were noted that could be associated with the
in MN stabilizes the U2N3 phase.270 Oxide stabilizes cladding. Mixed nitride fuel and Type 304 stainless
the M2N3 phase and decreases the nitrogen activity, steel cladding, sodium-bonded to one another, were
thus reducing clad nitriding. completely compatible at temperatures of 700–
On the other hand, if oxygen is present in MN as 800 C for at least 10 000 h.255 As expected, helium
solid solution or exists as impurity in the filling gas or bonding did not lead to significant nitriding either.
in the sodium bond, or originates from the cladding, it Hyperstoichiometric nitride fuels containing ses-
causes surface oxidation of MN by the following reac- quinitride may possibly nitride the cladding, based on
tion, leading to release of N2 and clad nitriding.134 thermodynamic considerations. The formation of
Cr2N and the compatibility of fuel and iron were
MN þ 2O2 ! MO2 þ 1=2O2
confirmed by a thermodynamic analysis of UN and
The compatibility test of (Pu0.3U0.7)N pellets contain- cladding materials.138 Chromium nitride precipitates
ing a large amount of oxygen (0.5 wt%) with Type were observed in the cladding, but no important det-
316 stainless steel cladding showed a discontinuous rimental effect was identified. The main impurities
clad nitriding around 35 mm.192 Similar effect of oxygen likely to be introduced during the nitride fabrication
content was observed in uranium-rich mixed nitride.258 process are carbon and oxygen. The influence of such
The helium-bonded mixed nitride fuel pin irra- impurities on the fuel properties of UN and (U0.8Pu0.2)
diated in EBR-II achieved a fuel burnup of 32.1 MWd N has been investigated by Arai et al.271
kg1 M at a peak linear power of 81.4 kW m1.248 The nitrided cladding generally decreases the duc-
Excellent compatibility was exhibited between Type tility and increases the mechanical strength. Compati-
304 stainless steel cladding and single-phase mixed bility of mononitride fuel with stainless steel was
nitride fuel. Cladding in contact with fuel containing confirmed by out-of-pile and in-pile tests with gas-
the U2N3 phase exhibited a reaction layer indicative and sodium-bonded fuel pins.226 A thin layer of the
of nitrogen penetration. A typical section of an irra- cladding inner surface was nitrided in gas-bonded fuel
diated stainless steel cladding specimen is shown in pins but no deterioration of the mechanical properties
Figure 20.248 The darkening of the cladding near the of the cladding was shown.176,259
fuel indicated both reaction with the cladding and The microhardness measurements in the compat-
ibility test of (Pu0.3U0.7)N pellets containing higher
oxygen (0.5 wt%) with Type 316 steel cladding
showed Vickers Hardness Numbers of 265 and 210
at locations in the reaction zone and away from it,
respectively.192
3.16.8 Outlook
oxide fuels. Further FCCI data of MX-type fuels of Conference, New Orleans, Luisiana, Apr 13–15, 1971;
Farmakes, R., Ed.; American Nuclear Society: Hinsdale,
irradiated at higher linear heating rate will be needed IL, 1971; pp 411–429.
to examine the compatibility of fuel and cladding at 17. Adamson, M. G.; Aitken, E. A. In Fast Breeder Reactor
high temperatures. The restructuring of MX fuel Fuel Performance, Proceedings of International
Conference on Fast Breeder Reactor Fuel Performance,
might provide indications of the mechanisms asso- Monterey, CA, Mar 5–8, 1979; Norman, E. C.,
ciated with FCCI. Adamson, M. G., Boltax, A., et al. Eds.; American Nuclear
In the case of oxide fuels, the observed regimes of Society: Hinsdale, IL, 1979; pp 536–550.
18. Maeda, K.; Asaga, T. J. Nucl. Mater. 2004, 327, 1–10.
complex intergranular and matrix attack in oxide 19. Maeda, K.; Tanaka, K.; Asaga, T.; Furuya, H. J. Nucl.
fuels cannot be explained by simple mechanisms. Mater. 2005, 344, 274–280.
Generally, the design approach with respect to 20. Lambert, J. D. B.; Neimark, L. A.; Murphy, W. F.;
Dickerson, C. E. In Fast Reactor Fuel Element
FCCI includes consideration of cladding wastage at Technology, Proceedings of Conference, New Orleans,
the 2s confidence level in order to reserve assurance Luisiana, Apr 13–15, 1971; Farmakes, R., Ed.; American
of achieving the burnup goal without cladding fail- Nuclear Society: Hinsdale, IL, 1971; pp 517–554.
21. Batey, W.; Bagley, K. Q. J. Brit. Nucl. Eng. Soc 1974, 13,
ure. Therefore, intergranular attack must be pre- 49–61.
vented or minimized. The actual embrittlement of 22. Neimark, L. A.; Brown, F. L. Report ANL-8046; Argonne
cladding by intergranular attack is difficult to exam- National Laboratory, 1973.
23. Swanson, K. M.; Sloss, W.; Batey, W. In Advanced
ine in compatibility tests. Experiments designed to LMFBR Fuels, Proceedings of International Meeting on
clarify the phenomenon should be encouraged. Advanced LMFBR Fuels, ERDA-4455, Tucson, AZ, Oct
10–13, 1977; Leary, J., Kittle, H., Eds.; American Nuclear
Society: La Grange Park, IL, 1977; pp 95–109.
24. Olsen, A. R. Trans. Am. Nucl. Soc. 1971, 14, 596–598.
References 25. Neimark, L. A.; Mushy, W. F.; Lambert, J. D. B.;
Michels, L. C. Report ANL-7833; Argonne National
Laboratory, 1971.
1. Olander, D. R. Fundamental Aspects of Nuclear Reactor 26. Cox, C. M.; Olsen, A. R.; Fitts, R. B.; Long, E. L. Trans.
Fuel Elements. USERDA, TID-26711-P1, 1976. Am. Nucl. Soc. 1971, 14, 173–175.
2. Lambert, J. D. B.; Strain, R. V. Trans. Am. Nucl. Soc. 27. Lotts, A. L. Report ORNL-4901; Oak Ridge National
1973, 17, 192. Laboratory, 1973.
3. Bennett, M. J.; Chaffey, G. H.; Antill, J. E. J. Nucl. Mater. 28. Neimark, L. A.; Renfro, C. W. Report ANL-RDP-41;
1969, 32, 299–309. Argonne National Laboratory, 1975.
4. Adamson, M. G. In Technical Committee Meeting on Fuel 29. Lloyd, M. H.; Haire, R. G. Nucl. Appl 1968, 5, 114–122.
and Cladding Interaction, Proceedings of the 30. Haas, P. A. Nucl. Technol. 1971, 10, 283–292.
International Working Group on Fast Reactors, IWGFR/ 31. Weber, J. W.; Jensen, E. D. Trans. Am. Nucl. Soc. 1971,
16, Tokyo, Japan, Feb 21–25, 1977; IAEA: Austria, 1977; 14, 175–176.
pp 170–189. 32. Götzmann, O.; Dünner, Ph. In Technical Committee
5. Adamson, M. G.; Aitken, E. A.; Caputi, R. W.; Potter, P. E.; Meeting on Fuel and Cladding Interaction, Proceedings
Mignanelli, M. A. In Thermodynamics of Nuclear of the International Working Group on Fast Reactors,
Materials, Proceedings of an International Symposium, IWGFR/16, Tokyo, Japan, Feb 21–25, 1977; IAEA:
Jülich, Jan 29 to Feb 2, 1979; IAEA: Vienna, Austria, Austria, 1977; pp 43–48.
1980; Vol. I, pp 503–538. 33. Stelnmetz, B.; Haas, D.; Verbeek, P.; Dunner, Ph.
6. Lawrence, L. A. Nucl. Technol 1984, 64, 139–153. In Technical Committee Meeting on Fuel and Cladding
7. Götzmann, O. In Fast Reactor Core and Fuel Structural Interaction, Proceedings of the International Working
Behaviour, Proceedings of the International Conference, Group on Fast Reactors, IWGFR/16, Tokyo, Japan, Feb
Inverness, June 4–6, 1990; BNES: London, UK, 1990; 21–25, 1977; IAEA: Austria, 1977; pp 48–51.
pp 1–8. 34. Lawrence, L. A.; Hata, D. C.; Weber, J. W. Nucl. Technol
8. Neimark, L. A.; Lambert, J. D. B.; Murphy, W. F.; 1979, 42, 195–206.
Renfro, C. W. Nucl. Technol 1972, 16, 75–88. 35. Kashihara, H.; Shikakura, S.; Yokouchi, Y.; Shibahara, I.;
9. Gellings, P. J.; Dejongh, M. A. Corros. Sci. 1967, 7, Matushima, H.; Yamamoto, K. In Fast Reactor Core and
413–421. Fuel Structural Behavior, Proceedings of the International
10. Johnson, I.; Johnson, C. E. In Thermodynamics of Conference, Inverness, June 4–6 1990; BNES: London,
Nuclear Materials; Proceedings of a Symposium, Vienna, 1990; pp 243–248.
Austria, Oct 21–25, 1974; IAEA: Vienna, Austria, 1975; 36. Batey, W.; Findlay, J. R. In Technical Committee Meeting
Vol. I, pp 99–107. on Fuel and Cladding Interaction, Proceedings of the
11. Venker, H.; Ehrlich, K. J. Nucl. Mater. 1975, 56, 115–120. International Working Group on Fast Reactors,
12. Kleykamp, H. J. Nucl. Mater. 1985, 131, 221–246. IWGFR/16, Tokyo, Japan, Feb 21–25, 1977; IAEA:
13. Martin, P.; Pelletier, M.; Every, D.; Buckthorpe, D. Nucl. Austria, 1977; pp 97–103.
Technol. 2008, 161, 35–44. 37. Conte, M.; Marcon, J. P. In Technical Committee
14. Haas, D.; van de Velde, J.; Gaube, M.; Ketels, J.; van Meeting on Fuel and Cladding Interaction, Proceedings
Loon, C. Nucl. Technol. 1977, 34, 75–88. of the International Working Group on Fast Reactors,
15. de Heering, D.; Pay, A.; Heyne, C. Trans. Am. Nucl. Soc. IWGFR/16, Tokyo, Japan, Feb 21–25, 1977; IAEA:
1976, 24, 139–141. Austria, 1977; pp 27–36.
16. Perry, K. J.; Melde, G. F.; McCarthy, W. H.; Duncan, R. N. 38. Delbrassine, A.; Flipot, A. J. In Technical Committee
In Fast Reactor Fuel Element Technology, Proceedings Meeting on Fuel and Cladding Interaction, Proceedings
Ceramic Fuel–Cladding Interaction 479
of the International Working Group on Fast Reactors, 57. Melis, J.-C.; Piron, J.-P.; Roche, L. J. Nucl. Mater. 1993,
IWGFR/16, Tokyo, Japan, Feb 21–25, 1977; IAEA: 204, 188–193.
Austria, 1977; pp 9–23. 58. McCarthy, W. H.; Craig, C. N. Trans. Am. Nucl. Soc.
39. Fee, D. C.; Johnson, I.; Johnson, C. E. Report ANL-7553; 1973, 17, 192.
Argonne National Laboratory, 1975. 59. Adamson, M. G.; Aitken, E. A.; Lindemer, T. B. J. Nucl.
40. Götzmann, O.; Hofmann, P.; Thümmler, F. J. Nucl. Mater. Mater. 1985, 130, 375–392.
1974, 52, 33–50. 60. Mignanelli, M. A.; Thetford, R. J. Nucl. Mater. 1993, 204,
41. Koizumi, M.; Nagai, S.; Furuya, H.; Muto, T. In Technical 173–179.
Committee Meeting on Fuel and Cladding Interaction, 61. Inoue, M.; Maeda, K.; Katsuyama, K.; Tanaka, K.;
Proceedings of the International Working Group on Fast Mondo, K.; Hisada, M. J. Nucl. Mater. 2004, 326, 59–73.
Reactors, IWGFR/16, Tokyo, Japan, Feb 21–25, 1977; 62. Adamson, M. G. In Technical Committee Meeting on Fuel
IAEA: Austria, 1977; pp 64–73. and Cladding Interaction, Proceedings of the
42. Coquerelle, M.; Götzman, O.; Pikering, S. Trans. Am. International Working Group on Fast Reactors, IWGFR/
Nucl. Soc. 1975, 20, 287. 16, Tokyo, Japan, Feb 21–25, 1977; IAEA: Austria, 1977;
43. Weber, J. E.; Jensen, E. D. Trans. Am. Nucl. Soc. 1971, pp 108–136.
17, 175. 63. Comonbier, L.; Hochmann, J. Stainless and Heat
44. Pascard, R. In Technical Committee Meeting on Fuel Resisting Steels; Edward Arnold, Publishers Ltd: London,
and Cladding Interaction, Proceedings of the UK, 1965; Chapter 3.
International Working Group on Fast Reactors, 64. Lawrence, L. A.; Bard, F. E.; Cannon, N. S. In LMR:
IWGFR/16, Tokyo, Japan, Feb 21–25, 1977; IAEA: A Decade of LMR Progress and Promise, Proceedings of
Austria, 1977; pp 23–26. ANS Winter Meeting 1990 – Special Cluster Sessions,
45. Uematsu, K.; Ishida, Y.; Koizumi, K.; Komatsu, J. Washington, DC, Nov 11–15; ANS: La Grange Park, IL,
In Technical Committee Meeting on Fuel and 1990; pp 170–175.
Cladding Interaction, Proceedings of the International 65. Weber, J. W. Report HEDL-TME-73–69; Hanford
Working Group on Fast Reactors, IWGFR/16, Engineering Development Laboratory, 1973.
Tokyo, Japan, Feb 21–25, 1977; IAEA: Austria, 1977; 66. Adamson, M. G.; Aitken, E. A. Report GEFR-00258;
pp 79–89. General Electric Company, 1977.
46. Roake, W. E.; Hilbert, R. F.; Adamson, M. G.; et al. 67. Adamson, M. G.; Aitken, E. A.; Evans, S. K.;
In Technical Committee Meeting on Fuel and Cladding McCarthy, W. H. GEAP-14075; General Electric
Interaction, Proceedings of the International Working Company Report, 1976.
Group on Fast Reactors, IWGFR/16, Tokyo, Japan, Feb 68. Swanson, K. M.; Linekar, G. A. B; Bagley, K. Q.
21–25, 1977; IAEA: Austria, 1977; pp 137–158. In Reliable Fuels for Liquid Metal Reactors, Proceedings
47. Lawrence, L. A.; Weber, J. W.; Devary, J. L. In Fast of ANS International Conference, Tucson, AZ, Sept 7–11,
Breeder Reactor Fuel Performance, Proceedings of 1986; ANS: La Grange Park, IL, 1986; pp 2–38.
International Conference on Fast Breeder Reactor Fuel 69. Johnson, C. E.; Johnson, I.; Blackburn, P.; Battles, I. E.;
Performance, Monterey, CA, Mar 5–8, 1979; Norman, E. C., Crouthamel, C. E. Proceedings of a Panel on the
Adamson, M. G., Boltax, A., et al. Eds.; American Nuclear Behaviour and Ceramic State of Irradiated Ceramic
Society: Hinsdale, IL, 1979; pp 432–444. Fuels; Aug 7–11, 1972; IAEA: Vienna, Austria, 1974;
48. Bradbury, M. H.; Pickering, S.; Whitlow, W. H. pp 1–30.
In Technical Committee Meeting on Fuel and Cladding 70. Sari, C.; Schumacher, B. J. Nucl. Mater. 1976, 61,
Interaction, Proceedings of the International Working 192–202.
Group on Fast Reactors, IWGFR/16, Tokyo, Japan, Feb 71. Gibby, R. L.; Woodley, R. E.; Adamson, M. G.;
21–25, 1977; IAEA: Austria, 1977; pp 51–64. Johnson, C. E. In Fast Breeder Reactor Fuel
49. Johnson, C. E.; Johnson, I.; Blackburn, P. E.; Performance, Proceedings of International Conference
Crouthamel, C. E. Reactor Technol 1972/1973, 15, on Fast Breeder Reactor Fuel Performance, Monterey,
303–338. CA, Mar 5–8, 1979; Norman, E. C., Adamson, M. G.,
50. Ohse, R. W.; Schlechter, M. In The Behaviour and Boltax, A., et al. Eds.; American Nuclear Society:
Ceramic State of Irradiated Ceramic Fuels, Proceedings Hinsdale, IL, 1979; pp 188–201.
of a Panel, Vienna, Aug 7–11, 1972; IAEA: Vienna, 72. Ewart, F.; Lassmann, K.; Matzke, Hj.; Manes, L.;
Austria, 1974; pp 299–314. Saunders, A. J. Nucl. Mater. 1984, 124, 44–55.
51. Bradbury, M. H.; Pickering, S.; Whitlow, W. H. J. Nucl. 73. Ewart, F. T.; Manes, L.; Matzke, Hj.; Mari, C. M.; Toci, F.;
Mater. 1978, 78, 272–280. Schreiber, R. J. Nucl. Mater. 1979, 81, 185–190.
52. Biermann, W. K.; Haas, D.; Heuvel, H. J. In Fast 74. Kleykamp, H. J. Nucl. Mater. 1977, 66, 292–300.
Breeder Reactor Fuel Performance, Proceedings of 75. Conte, M.; Gatesoupe, J. P.; Trotabas, M.; Boivineau, J. C.;
International Conference on Fast Breeder Reactor Cosoli, G. In Fast Breeder Reactor Fuel Performance,
Fuel Performance, Monterey, CA, Mar 5–8, 1979; Proceedings of International Conference on Fast Breeder
Norman, E. C., Adamson, M. G., Boltax, A., et al. Eds.; Reactor Fuel Performance, Monterey, CA, Mar 5–8, 1979;
American Nuclear Society: Hinsdale, IL, 1979; Norman, E. C., Adamson, M. G., Boltax, A., et al. Eds.;
pp 569–579. American Nuclear Society: Hinsdale, IL, 1979;
53. Fee, D. C.; Johnson, C. E. J. Nucl. Mater. 1981, 96, pp 301–319.
80–104. 76. Johnson, I.; Johnson, C. E.; Crouthamel, C. E.; Seils, C. A.
54. Tourasse, M.; Boidron, M.; Pasquet, B. J. Nucl. Mater. J. Nucl. Mater. 1973, 48, 21–34.
1992, 188, 49–57. 77. Rand, M. H.; Markin, J. L. In Thermodynamics of Nuclear
55. Bailly, H.; Menessier, D.; Prunier, C. The Nuclear Fuel of Materials, Proceedings of A Symposium, Vienna, Austria,
Pressurized Water Reactors and Fast Neutron Reactors: Sept 4–8, 1967; IAEA: Vienna, Austria, 1968; pp 637–650.
Design and Behavior; Intercept Ltd.: UK, 1999; p 475, 78. Matzke, Hj.; Ottaviani, J.; Pellottiero, D.; Rouault, J.
ISBN: CEA 2–7272–0198–2. J. Nucl. Mater. 1988, 160, 142–146.
56. Melis, J.-C.; Roche, L.; Piron, J. P.; Truffert, J. J. Nucl. 79. Osaka, M.; Kurosaki, K.; Yamanaka, S. J. Nucl. Mater.
Mater. 1992, 188, 303–307. 2006, 357, 69–76.
480 Ceramic Fuel–Cladding Interaction
80. Walker, C. T.; Nicolaou, G. J. Nucl. Mater. 1995, 218, 107. Aubert, M.; Calais, D.; Le Beuze, R. J. Nucl. Mater. 1975,
129–138. 58, 257–277.
81. Prunier, C.; Boussard, F.; Koch, L.; Coquerelle, M. Nucl. 108. Konashi, K.; Yano, T.; Kaneko, H. J. Nucl. Mater. 1983,
Technol. 1997, 119, 141–148. 116, 86–93.
82. Findlay, J. R. In The Behaviour and Ceramic State of 109. De Keroulas, F.; Le Beuze, R.; Calais, D.; van
Irradiated Ceramic Fuels, Proceedings of a Panel, Craeynest, A.; Conte, M. J. Nucl. Mater. 1972, 43,
Vienna, Aug 7–11, 1972; IAEA: Vienna, Austria, 1974; 313–320.
pp 211–220. 110. Adamson, M. G.; Reineking, W. H.; Lauritzen, T.;
83. Hofmann, P.; Götzmann, O. In The Behaviour and Vaidyanathan, S. In Liquid and Solid Metal-Induced
Ceramic State of Irradiated Ceramic Fuels, Proceedings Embrittlement of Metals; Kamdar, M. H. Ed.; TMS-AIME:
of a Panel, Vienna, Aug 7–11, 1972; IAEA: Vienna, New York, 1984.
Austria, 1974; pp 237–254. 111. Adamson, M. G.; Aitken, E. A.; Vaidyanathan, S. Nature
84. Cordfunke, E. H. P.; Konings, R. J. J. Nucl. Mater. 1988, 1982, 295, 49–51.
152, 301–309. 112. Maiya, P. S.; Busch, D. E. Met. Trans. A 1975, 6A,
85. Cubicciotti, D.; Davies, J. H. Nucl. Sci. Eng. 1976, 60, 409–415.
314–319. 113. Hamilton, M. L.; Johnson, G. D.; Hunter, C. W.;
86. Peehs, M.; Kaspar, G.; Neeb, K. H. J. Nucl. Mater. 1983, Duncan, D. R. In Dimensional Stability and Mechanical
119, 284–290. Behavior of Irradiated Metals and Alloys, Proceedings of
87. Campbell, D. O.; Malinauskas, A. P.; Stratton, W. R. Nucl. International Conference; British Nuclear Society:
Technol. 1981, 53, 111–119. London, 1983; p 211.
88. Besmann, T. M.; Lindemer, T. B. Nucl. Technol. 1978, 40, 114. Old, C. F. J. Nucl. Mater. 1980, 92, 2–25.
297–305. 115. Adamson, M. G.; Aitken, E. A. J. Nucl. Mater. 1985, 132,
89. Ball, R. G. J.; Burns, W. G.; Henshaw, J.; Mignanelli, 160–166.
M. A.; Potter, P. E. J. Nucl. Mater. 1989, 167, 191–204. 116. Cannon, N. S.; Hunter, C. W.; Kear, K. L.; Wood, M. H.
90. Cordfunke, E. H. P. In Thermodynamics of Nuclear J. Nucl. Mater. 1986, 139, 60–69.
Materials, Proceedings of a Symposium, Vienna, Austria, 117. Pulham, R. J.; Richards, M. W. J. Nucl. Mater. 1990, 172,
Oct 21–25, 1974; IAEA: Vienna, Austria, 1975; Vol. II, 206–219.
pp 637–650. 118. Weber, E. T.; Lawrence, L. A.; Wilson, C. N.; Gibby, R. L.;
91. Fee, D. C.; Johnson, C. E. J. Inorg. Nucl. Chem. 1978, 40, Adamson, M. G. In Fast Breeder Reactor Fuel
1375–1381. Performance, Proceedings of International Conference
92. Cordfunke, E. H. P.; Kleverlaan, F.; Ouweltjes, W. on Fast Breeder Reactor Fuel Performance, Monterey,
Thermochim. Acta 1986, 102, 387–390. CA, Mar 5–8, 1979; Norman, E. C., Adamson, M. G.,
93. Benedict, U. In Thermodynamics of Nuclear Materials, Boltax, A., et al., Eds.; American Nuclear Society:
Proceedings of an International Symposium, Jülich, Hinsdale, IL, 1979; pp 445–459.
Jan 29 to Feb 2, 1979; IAEA: Vienna, Austria, 1980; Vol. I, 119. Brucklacher, D.; Dienst, W. In Fuel and Fuel Elements for
pp 453–470. Fast Reactors, Proceedings of a Symposium on Fuel and
94. Aitken, E. A.; Evans, S. K.; Rubin, B. F. In The Behaviour Fuel Elements for Fast Reactors, Brussels, July 2–6,
and Ceramic State of Irradiated Ceramic Fuels, 1973; IAEA: Vienna, Austria, 1974; Vol. I, pp 147–161.
Proceedings of a Panel, Vienna, Aug 7–11, 1972; IAEA: 120. Johnson, C. E. Report ANL-RDP-18; Argonne National
Vienna, Austria, 1974; pp 269–285. Laboratory, 1973.
95. Götzmann, O.; Hofmann, P. Report KFK 1619; 121. Hofmann, P. Report KFK-1831; Kernforschungszentrum
Kernforschungszentrum Karlsruhe, 1972. Karlsruhe, 1974.
96. Götzmann, O.; Hofmann, P. Report KFK 1271/4; 122. Hofmann, P.; Götzmann, O. Report KFK-1272/4;
Kernforschungszentrum Karlsruhe, 1972. Kernforschungszentrum Karlsruhe, 1972.
97. Aitken, E. A.; Adamson, M. G.; Dutina, D.; Evans, S. K. In 123. Götzmann, O.; Kleykamp, H. J. Nucl. Mater. 1980, 89,
Thermodynamics of Nuclear Materials, Proceedings of a 71–79.
Symposium, Vienna, Austria, Oct 21–25, 1974; IAEA: 124. Elkins, P. E. Report NAA-SR-7502; Atomic International,
Vienna, Austria, 1975; Vol. I, 187–205. 1964.
98. Aitken, E. A.; Adamson, M. G.; Dutina, D.; Evans, S. K. 125. Alexander, C. A.; Ward, J. J.; Ogden, J. S.;
Report GEAP-12538; General Electric Company, 1974. Cunningham, G. W. In Carbides in Nuclear Energy;
99. Aitken, E. A.; Adamson, M. G.; Evans, S. K. Report Russell, L. E., Bradbury, B. T., Harrison, J. D. L.,
GEAP-12489; General Electric Company, 1974. Hedger, H. J., Mardon, P. G., Eds.; MacMillan: London,
100. Adamson, M. G.; Aitken, E. A. Report GEAP-12511; 1964; Vol. 1, p 192.
General Electric Company, 1974. 126. Chubb, W.; Keller, D. L. In Carbides in Nuclear Energy;
101. Adamson, M. G.; Aitken, E. A. Report GEAP-14136; Russell, L. E., Bradbury, B. T., Harrison, J. D. L.,
General Electric Company, 1976. Hedger, H. J., Mardon, P. G., Eds.; MacMillan: London,
102. Antill, J. E.; Warburton, J. B. J. Nucl. Mater. 1977, 71, 1964; Vol. 1, pp 208–230.
134–139. 127. Briggs, G.; Dutta, S. K.; White, J. In Carbides in Nuclear
103. Aitken, E. A.; Evans, S. K.; Rubin, B. F.; Ludlow, T. E. Energy; Russell, L. E., Bradbury, B. T., Harrison, J. D. L.,
Report GEAP-12268; General Electric Company, 1972. Hedger, H. J., Mardon, P. G., Eds.; MacMillan: London,
104. Hales, J. W.; Hoth, C. W. Trans. Am. Nucl. Soc. 1977, 27, 1964; Vol. 1, p 231.
237–238. 128. Nichols, J. L.; Marples, J. A. C. In Carbides in Nuclear
105. Fitts, R. B.; Long, E. L., Jr.; Leitnaker, J. M. In Fast Energy; Russell, L. E., Bradbury, B. T., Harrison, J. D. L.,
Reactor Fuel Element Technology, Proceedings of Hedger, H. J., Mardon, P. G., Eds.; MacMillan: London,
Conference, New Orleans, LA, Apr 13–15, 1971; 1964; Vol. 1, pp 246–260.
Farmakes, R., Ed.; American Nuclear Society: Hinsdale, 129. Götzmann, O.; Ohse, R. W. In The Behaviour and
IL, 1971; pp 431–458. Ceramic State of Irradiated Ceramic Fuels, Proceedings
106. Johnson, C. E.; Crouthamel, C. E. J. Nucl. Mater. 1970, of a Panel, Vienna, Aug 7–11, 1972; IAEA: Vienna,
34, 101–104. Austria, 1974; pp 255–268.
Ceramic Fuel–Cladding Interaction 481
130. Stumpf, W. E.; Götzmann, O. Report KFK-1384; of International Meeting on Advanced LMFBR Fuels,
Kernforschungszentrum Karlsruhe, 1971. ERDA-4455, Tucson, AZ, Oct 10–13, 1977; Leary, J.,
131. Götzmann, O.; Hofmann, P. Report KFK-1546; Kittle, H., Eds.; American Nuclear Society: La Grange
Kernforschungszentrum Karlsruhe, 1972. Park, IL, 1977; pp 315–325.
132. Latimer, T. W. Report ANL-7827; Argonne National 157. Frost, B. R. T. J. Nucl. Mater. 1963, 10, 265–300.
Laboratory, 1971. 158. Gmelins Handbuch der anorganischen Chemie, Vol. 73,
133. Beahm, E. C.; Culpepper, C. A. Trans. Am. Ceram. Soc. Transuranium Elements; Part C, Vol. 4, Compounds;
1979, 33, 262–264. Springer: Berlin, 1972.
134. Hebeisen, J. C.; Moak, D. P.; Pfeifer, W. H. Report 159. Blank, H. Report KFK-1111, Kummerer, K.,
BMI-1895; Battelle Memorial Institute, 1971. Thümmler, F., Eds.; Kernforschungszentrum Karlsruhe,
135. Hebeisen, J. C.; Moak, D. P.; Saling, J. H. Trans. Am. 1969; Chapter II.
Nucl. Soc. 1969, 12, 595–597. 160. Holleck, H. In Thermodynamics of Nuclear Materials,
136. Katz, S. J. Nucl. Mater. 1962, 6, 172–181. Proceedings of a Symposium, Vienna, Austria,
137. Weinberg, A. F.; Yang, L. Report GA-2548; General Oct 21–25, 1974; IAEA: Vienna, Austria, 1975; Vol. II,
Dynamics Corp., 1961. pp 213–263.
138. Imoto, S.; Namba, S. J. Nucl. Mater. 1974, 51, 106–111. 161. Reavis, J. G.; Johnson, K. A.; Leary, J. A. In Plutonium
139. Götzmann, O.; Hofmann, P. In Fuel and Fuel Elements 1970 and Other Actinides, Proceedings of the 4th
for Fast Reactors, Proceedings of a Symposium on International Conference, Santa Fé, NM, Oct 5–9, 1970,
Fuel and Fuel Elements for Fast Reactors, Brussels, Miner, W. N., Ed.; Nuclear Metallurgy, Metallurgical
July 2–6, 1973; IAEA: Vienna, Austria, 1974; Vol. I, Society of AIME 1970; Vol. 17, pp 791–798.
pp 233–245. 162. Dalton, J. T.; Harper, E. A.; Hedger, H. J.; Stannard, R.W.
140. Agarwal, R.; Venugopal, V. J. Nucl. Mater. 2006, 359, In Plutonium 1970 and Other Actinides, Proceedings of
122–131. the 4th International Conference, Santa Fé, NM, Oct 5–9,
141. Kleykamp, H. Reaktortagung Hannover; German 1970, Miner, W. N., Ed.; Nuclear Metallurgy, Metallurgical
Atomforum: Bonn, 1978; p 660. Society of AIME 1970; Vol. 17, pp 829–841.
142. Latimer, T. W.; Bost, D. S.; Kerrisk, J. F.; Barner, J. O.; 163. Mardon, P. G.; Potter, P. E. In Plutonium 1970 and Other
Green, J. L. Report LA-6077-MS; Los Alamos Scientific Actinides, Proceedings of the 4th International
Laboratory, 1975. Conference, Santa Fé, NM, Oct 5–9, 1970, Miner, W. N.,
143. Holleck, H.; Smailos, E.; Thümmler, F. Reaktortagung, Ed.; Nuclear Metallurgy, Metallurgical Society of AIME
Proc. Symp.; Berlin, 1974; p 384. 1970; Vol. 17, pp 842–858.
144. Benz, R.; Hoffman, C. G.; Rupert, G. N. High Temp. Sci. 164. Huet, J.-.; Casteels, F.; Cools, A.; et al. In Ceramic
1969, 1, 342–359. Nuclear Fuels, Proceedings of the International
145. Storms, E. K. The Refractory Carbides, Monograph Symposium on Ceramic Nuclear Fuels, Washington, DC,
Series on Refractory Materials; Academic Press: May 1969; Kruger, O. L., Kaznoff, A. I., Eds.; American
New York and London, 1967; Vol. 2. Ceramic Society: Columbus, OH, 1969; pp 138–146.
146. Halleck, H. J. Nucl. Mater. 1984, 124, 129–146. 165. Washburn, T. N.; Acott, J. L. In Fast Reactor Fuel Element
147. Laugier, J.; Blum, P. L. J. Nucl. Mater. 1971, 39, 245–252. Technology, Proceedings of Conference, New Orleans,
148. Storms, E. K.; Ackermann, R. J.; Rand, M. H. In Luisiana, Apr 13–15, 1971; Farmakes, R., Ed.; American
Thermodynamic Properties of the U-C, Pu-C and Nuclear Society: Hinsdale, IL, 1971; pp 741–752.
(U,Pu)-C Systems, Proceedings of Panel, Grenoble, 166. Chubb, W. Nucl. Sci. Eng. 1967, 29, 176–179.
France, May 6–8, 1974; IAEA: Vienna, Austria. 167. Browning, P.; Phillips, B. A.; Potter, P. E.; Rand, M. H. In
149. Mulford, R. N. R.; Ellinger, F. H.; Hendrix, G. S.; Plutonium and Other Actinides, Proceedings of the 5th
Albrecht, E. D. In Plutonium Metallurgy, Proceedings International Conference, Baden Baden, Sept 10–13,
of the Second International Conference, Grenoble, 1975; Blank, H., Lindner, R., Eds.; Elsevier: North-
France, Apr 19–22; Grison, E., Lord, W. B. H., Holland, 1976; pp 257–265.
Fowler, R. D., Eds.; Cleaver-Hume Press, Ltd.: London, 168. Batly, W.; et al. In Carbides in Nuclear Energy; Russel, J.,
1961; pp 301–311. Ed.; MacMillan and Co.: London, 1964.
150. Reavis, J. G.; Leary, J. A. In Plutonium 1970 and Other 169. Gamba, O. O. Report NAA-SR-10929; Atomic
Actinides, Proceedings of the 4th International International, 1965.
Conference, Santa Fé, NM, Oct 5–9, 1970, Miner, W. N., 170. Elkins, P. E. Report NAA-SR-7502; Atomic International,
Ed.; Nuclear Metallurgy, 1970; Vol. 17, pp 809–817. 1963.
151. Leary, J. A. In Ceramic Nuclear Fuels; Proceedings of the 171. Webb, B. A. Report NAA-SR-6246; Atomic International,
International Symposium on Ceramic Nuclear Fuels, 1963.
Washington, DC, May 1969; Kruger, O. L., Kaznoff, A. I., 172. Watanabe, H.; Kurasawa, T.; Kikuchi, T.; Furukawa, K.;
Eds.; American Ceramic Society: Columbus, OH, 1969; Nihei, I. J. Nucl. Mater. 1971, 40, 213–220.
pp 38–50. 173. Deighton, M. J. Iron Steel Inst. 1970, 208, 1012–1014.
152. Crane, J.; Gordon, E. In Carbides in Nuclear Energy; 174. Hall, E. L.; Briant, C. L. Met. Trans. A 1984, 15A, 793–811.
Russel, J., Ed.; MacMillan and Co.: London, 1964. 175. Pillai, S. R.; Mathews, C. K. J. Nucl. Mater. 1987, 150,
153. Frank, J. E.; Forrester, R. E.; Buck, J. S. NAA-SR-11440, 31–41.
1966. 176. Litton, F. B.; Morris, A. E. J. Less-Common metals 1970,
154. Beahm, E. C.; Culpepper, C. A. Nucl. Technol. 1977, 35, 22, 71–82.
173–177. 177. Blank, H. In Advanced LMFBR Fuels, Proceedings of
155. Latimer, T. W.; Andrew, J. F. In Advanced LMFBR Fuels, International Meeting on Advanced LMFBR Fuels,
Proceedings of International Meeting on Advanced ERDA-4455, Tucson, AZ, Oct 10–13, 1977; Leary, J.,
LMFBR Fuels, ERDA-4455, Tucson, AZ, Oct 10–13, Kittle, H., Eds.; American Nuclear Society: La Grange
1977; Leary, J., Kittle, H., Eds.; American Nuclear Park, IL, 1977; pp 482–501.
Society: La Grange Park, IL, 1977; pp 205–217. 178. Anselin, F.; Carteret, Y.; Bataller, S.; Clottes, G.;
156. Bagley, K. Q.; Batey, W.; Paris, R.; Sloss, W. M.; Mercier, C.; Dupouy, J. M. In Irradiation Behaviour of Fuel
Snape, G. P. In Advanced LMFBR Fuels, Proceedings Cladding and Core Component Materials, ; Proceedings
482 Ceramic Fuel–Cladding Interaction
of European Conference, Karlsruhe, F.R. Germany, 201. Wallace, T. C. Report LA-DC-8840; Los Alamos Scientific
Dec 3–5, 1994; KTG/BNES: London, 1975; pp 57–60. Laboratory, 1968.
179. Hofman, G. L.; Truffert, J.; DuPouy, J. M. J. Nucl. Mater. 202. Ronchi, C.; Coquerelle, M.; Blank, H.; Rouault, J. Nucl.
1977, 65, 200–203. Technol. 1984, 67, 73–91.
180. Latimer, T. W.; Andrew, J. F. In Advanced LMFBR Fuels, 203. Kerrisk, J. F.; DeMuth, N. S.; Petty, R. L.; Latimer, T. W.;
Proceedings of International Meeting, ERDA-4455, Vitti, J. A.; Jones, L. J. In Fast Breeder Reactor Fuel
Tucson, AZ, Oct 10–13, 1977; Leary, J.; Kittle, H., Eds.; Performance, Proceedings of International Conference
American Nuclear Society: La Grange Park, IL, 1977; on Fast Breeder Reactor Fuel Performance, Monterey,
pp 205–217. CA, Mar 5–8, 1979; Norman, E. C., Adamson, M. G.,
181. Stumpf, W. E.; Götzmann, O. Report KFK-1384; Boltax, A., et al., Eds.; American Nuclear Society:
Kernforschungszentrum Karlsruhe, 1971. Hinsdale, IL, 1979; p 804.
182. Snyder, R. B.; Natesan, K.; Kassner, T. F. J. Nucl. Mater. 204. Paris, R.; Jones, D.; Demant, J. T. Report AERE-R-7953,
1974, 50, 259–274. Atomic Energy Research Establishment (Harwell,
183. Thompson, R. In Carbon in Sodium, Proceedings of the England), 1975.
Specialists’ Meeting, Harwell, UK, Nov 27–30, 1979; 205. Paris, R.; Jones, D.; Demant, J. T.; Yates, A. Report
IAEA: Vienna, 1979; p 6. AERE-R-9788, Atomic Energy Research Establishment
184. Lorenzelli, N.; Marcon, J. P. J. Nucl. Mater. 1972, 44, (Harwell, England), 1980.
57–63. 206. Agarwala, R. P.; Naik, M. C.; Anand, M. S.; Paul, A. R.
185. Tetenbaum, M.; Hunt, P. D. Int. Coll. French Natl. Center J. Nucl. Mater. 1970, 36, 41–47.
Sci. Res., Odeillo, France, Sept 27–30, 1971, Vol. 205. 207. Bart, G.; Aerne, E. T.; Burri, M.; Zwicky, H.-U. Nucl.
186. Lyashenko, V. S.; Nevsorov, B. A. In The Corrosion of Instrum. Meth. B 1986, 17, 360–362.
Reactor Materials, Proceedings of a Conference, AEC-TR- 208. Pierce, T. B.; McMillan, J. W.; Peck, P. F.; Jones, I. G.
5408 (Translated from Paper No. CN-13/41), Salzburg, Nucl. Instr. Meth. 1974, 118, 115–124.
Austria, June 4–9, 1962; IAEA: Vienna, Austria, 1962. 209. Combette, P. In Ceramic Nuclear Fuels, Proceedings of
187. Nevsorov, B. A. In The Corrosion of Reactor Materials, the International Symposium on Ceramic Nuclear Fuels,
Proceedings of a Conference, AEC-TR-5412 (Translated Washington, DC, May 1969; Kruger, O. L., Kaznoff, A. I.,
from Paper No. CN-13/40), Salzburg, Austria, June 4–9, Eds.; American Ceramic Society: Columbus, OH, 1969;
1962; IAEA: Vienna, Austria, 1962. pp 343–347.
188. Powers, R. M.; Strasser, A. A. Trans. Am. Nucl. Soc. 210. Ronchi, C.; Coquerelle, M.; Blank, H.; Rouault, J. Nucl.
1967, 10, 471. Technol. 1984, 67, 73–91.
189. Levine, P. J.; Harbourne, B. L.; Biancheria, A.; Boltax, A. 211. Strasser, A.; Montgomery, M.; Powers, R. Trans. Am.
In Ceramic Nuclear Fuels, Proceedings of the Nucl. Soc. 1971, 14, 40–41.
International Symposium on Ceramic Nuclear Fuels, 212. Mikailoff, M.; Ratier, J.; Lallement, R. Trans. Am. Nucl.
Washington, DC, May 1969; Kruger, O. L., Kaznoff, A. I., Soc. 1974, 19, 84.
Eds.; American Ceramic Society: Columbus, OH, 1969; 213. Kerrisk, J. F.; DeMuth, N. S.; Petty, R. L.; Latimer, T. W.;
pp 156–165. Vitti, J. A.; Jones, L. J. In Fast Breeder Reactor Fuel
190. Stratton, R. W.; Smith, L. In Advanced LMFBR Fuels, Performance, Proceedings of International Conference
Proceedings of International Meeting on Advanced on Fast Breeder Reactor Fuel Performance, Monterey,
LMFBR Fuels, ERDA-4455, Tucson, AZ, Oct 10–13, CA, Mar 5–8, 1979; Norman, E. C., Adamson, M. G.,
1977; Leary, J., Kittle, H., Eds.; American Nuclear Boltax, A., et al., Eds.; American Nuclear Society:
Society: La Grange Park, IL, 1977; pp 348–350. Hinsdale, IL, 1979; p 804.
191. Adamson, M. G. Report AERE-R-6897, Atomic Energy 214. Clottes, G.; Dieumegard, M.; Dufayet, J. P.; et al., Fuel
Research Establishment (Harwell, England), 1972. and Fuel Elements for Fast Reactors, Proceedings of a
192. Ganguly, C.; Hegde, P. V.; Sengupta, A. K. In Advanced Symposium on Fuel and Fuel Elements for Fast Reactors,
Fuel for Fast Breeder Reactors: Fabrication and Brussels, July 2–6, 1973; IAEA: Vienna, Austria, 1974;
Properties and Their Optimization, ; Proceedings of Vol. II, pp 113–127.
Technical Committee Meeting, IAEA-TECDOC-466, 215. Kleykamp, H. Mikrochimica Acta 1983, (Suppl. 10),
Vienna, Austria, Nov 3–5, 1987; IAEA: Vienna, Austria, 217–229.
1988; pp 7–23. 216. Latimer, T. W.; Barner, J. O.; Kerrisk, J. F.; Green, J. L.
193. Latimer, T. W.; Petty, R. L. Report LA 8688, Los Alamos Report LA-6249; Los Alamos Scientific Laboratory, 1976.
Scientific Lab., New Mexico, 1981. 217. Latimer, T. W.; Petty, R. L.; Kerrisk, J. F.; DeMuth, N. S.;
194. Latimer, T. W.; Petty, R. L. Report LA 9482, Los Alamos Levine, P. J.; Boltax, A. In Fast Breeder Reactor Fuel
Scientific Lab., New Mexico, 1983. Performance, Proceedings of International Conference
195. Frost, B. R. T. J. Nucl. Mater. 1963, 10, 265–300. on Fast Breeder Reactor Fuel Performance, Monterey,
196. Watanabe, H.; Kurasawa, T.; Kikuchi, T.; Furukawa, K.; CA, Mar 5–8, 1979; Norman, E. C., Adamson, M. G.,
Nihei, I. J. Nucl. Mater. 1971, 40, 213–220. Boltax, A., et al., Eds.; American Nuclear Society:
197. Campbell, C. S.; Tyzack, C. In Alkali Metal Coolants – Hinsdale, IL, 1979; p 316.
Corrosion Studies and System Operating Experience, 218. Kittel, J. H.; Neimark, L. A.; Walters, L. G.; Einziger, R. E.;
Proceedings of the International Symposium, Vienna, Mahagani, D. E. Report ANL-80–40; Argonne National
Nov 28 to Dec 2, 1966; IAEA: Vienna, 1967; pp 159–170. Laboratory, 1980.
198. Kubaschewski, O.; Heyaer, G. Acta Metall. 1960, 8, 219. Coquerelle, M. In Jahrestagung Kerntechnik; Die
416–423. Kerntechnische Gesellschaft (KTG): Bonn, 1981; p 385.
199. Hoch, M.; Juenke, E. F.; Sjodahl, L. H. In 220. Bagley, K. Q.; Edmonds, E.; Powell, H. J.; Bellamy, R. G.;
Thermodynamics of Nuclear Materials; Proceedings of Paris, R.; Horspool, J. M. In Fuel and Fuel Elements for
A Symposium, Vienna, Austria, Sept 4–8, 1967; IAEA: Fast Reactors, Proceedings of a Symposium on Fuel and
Vienna, Austria, 1968; pp 497–508. Fuel Elements for Fast Reactors, Brussels, July 2–6,
200. Potter, P. E. In Thermochemie, Proceedings Colloques 1973; IAEA: Vienna, Austria, 1974; Vol. I, pp 87–100.
Internationaux du Centre National de la Recherche 221. Batey, W.; Donaldson, D. M.; Yates, G. In Plutonium,
Scientlflque No. 201; CNRS: Marseille, 1971; p. 253. Proceedings of the Third International Symposium, Inst.
Ceramic Fuel–Cladding Interaction 483
Metals, London, UK, 1965; Kay, A. E., Waldron, M. B., 247. Arai, Y.; Maeda, A.; Shiozawa, K.; Ohmichi, T. J. Nucl.
Eds.; Chapman and Hall: London, 1967; pp 888–908. Mater. 1994, 210, 161–166.
222. French, P. M.; Hodkin, D. J. In Plutonium, Proceedings of 248. Bauer, A. A.; Brown, J. B.; Fromm, E. O.; Storhok, V. W.
the Third International Symposium, Inst. Metals, London, In Fast Reactor Fuel Element Technology, Proceedings
UK, 1965; Kay, A. E., Waldron, M. B., Eds.; Chapman and of Conference, New Orleans, LA, Apr 13–15, 1971;
Hall: London, 1967; pp 697–720. Farmakes, R., Ed.; American Nuclear Society: Hinsdale,
223. Holden, R. B. Ceramic Fuel Elements; Gordon and IL, 1971; pp 785–817.
Breach Science Publishers: New York, 1966. 249. Holleck, H.; Kleykamp, H. Gmelins Handbuch der
224. Luner, C.; Cosgarea, A., Jr.; Fader, H. M. In Alkali Metal anorganischen Chemie, Vol. 55, Uranium; Supplement C7,
Coolants – Corrosion Studies and System Operating Nitrogen Compounds; Springer: Berlin, 1981;
Experience, Proceedings of the International Chapter 4.1.
Symposium, Vienna, Nov 28 to Dec 2, 1966; IAEA: 250. Tagawa, J. H. J. Nucl. Mater. 1974, 51, 78–89.
Vienna, 1967; pp 171–179. 251. Benz, R.; Bowman, M. G. J. Am. Ceram. Soc. 1966, 88,
225. Longson, B.; Thorley, A. W. J. Appl. Chem. 1970, 20, 264–268.
372–379. 252. Benz, R.; Hutchinson, W. B. J. Nucl. Mater. 1970, 36,
226. Matzke, Hj. Science of Advanced LMFBR Fuels; Elsevier: 135–146.
North Holland, 1986. 253. Hoenig, C. L. J. Am. Ceram. Soc. 1971, 54, 391–398.
227. Stumpf, W. E.; Venter, S. J. Nucl. Mater. 1976, 60, 254. Henshaw, J.; Mignanelli, M. A. Report AEAT-0075, AEA
195–202. Technology, 1996.
228. Pascard, R. Nucl. Met. Met. Soc. AIME 1968, 13; 255. Pardue, W. M.; Bauer, A. A.; Keller, D. L. In Ceramic
345–368. Nuclear Fuels, Proceedings of the International
229. Vangeel, J.; Huet, J.-J. In New Nuclear Materials Symposium on Ceramic Nuclear Fuels, Washington, DC,
Including Non Metallic Fuels, Proceedings of IAEA May 1969; Kruger, O. L., Kaznoff, A. I., Eds.; American
Symposium, Prague, July 1–5, 1963; IAEA: Vienna, Ceramic Society: Columbus, OH, 1969; pp 166–178.
Austria, 1963; Vol. 1, pp 385–400. 256. Lorenzelli, R. Report CEA R-3536, Commissariat a
230. Mouchnino, M. Report CEA-3353, Commissariat a l’Energie Atomique, 1968.
I’Energie Atomique, 1967. 257. Potter, P. E. J. Nucl. Mater. 1973, 47, 7–16.
231. Delbrassine, A.; Coquerelle, M. In Fuel and Fuel 258. Allbutt, M.; Junkison, A. R. Report AERE-R-5466,
Elements for Fast Reactors, Proceedings of a Atomic Energy Research Establishment
Symposium on Fuel and Fuel Elements for Fast Reactors, (Harwell, England), 1967.
Brussels, July 2–6, 1973; IAEA: Vienna, Austria, 1974; 259. Wullaert, R. A.; Lagedrost, J. F.; Bugl, J.; Gates, J. E.
Vol. I, pp 247–260. Report BMI-1638, Battelle Memorial Institute, 1963.
232. Milet, C. A. Report CEA-R-3656; Commissariat a 260. Bauer, A. A.; Cybulskis, P.; Green, J. L. In Advanced
l’Energie Atomique, 1968. LMFBR Fuels, Proceedings of International
233. Jacoby, W. R.; Palm, R. G.; Ray, W. E. Report Meeting on Advanced LMFBR Fuels, ERDA-4455,
WCAP-3791–3, Argonne National Laboratory, 1967. Tucson, AZ, Oct 10–13, 1977; Leary, J.; Kittle, H., Eds.;
234. Andrews, K. W.; Hughes, H. Precipitation Processes in American Nuclear Society: La Grange Park, IL, 1977;
Steels, Proceedings of Iron and Steel Institute, 1959; The pp 299–312.
Iron and Steel Institute: London, 1959, 43–50. 261. Bauer, A. A.; Cybulskis, P.; Petty, R. L.; DeMuth, N. S.
235. Fries, R. J. The Chemical Diffusion of Carbon in the In Fast Breeder Reactor Fuel Performance, Proceedings of
Group VI-B Metal Carbides, LA-3795-MS, 1968. International Conference on Fast Breeder Reactor Fuel
236. Herbst, R. J. Trans. Am. Nucl. Soc. 1966, 9, 17–18. Performance, Monterey, CA, Mar 5–8, 1979; Norman, E. C.,
237. Gorle, F.; Timmermans, W.; Casteels, F.; Vangeel, J.; Adamson, M. G., Boltax, A., et al., Eds.; American Nuclear
Brabers, M. In Thermodynamics of Nuclear Materials, Society: Hinsdale, IL, 1979; pp 827–832.
Proceedings of a Symposium, Vienna, Austria, Sept 4–8, 262. Bauer, A. A.; Saling, J. H.; Storhok, V. W. Trans. Am.
1967; IAEA: Vienna, Austria, 1968; pp 481–496. Nucl. Soc. 1969, 12, 598–599.
238. Jordan, K. R.; Herbst, R. J.; Palm, R. A. ANL-7120, Argonne 263. Keller, D. L. Report BMI-1925, Battelle Memorial
National Laboratory: Argonne, IL, 1965; pp 301–312. Institute, 1972.
239. Latimer, T. W.; Jacoby, W. R. Nuclear Met. 1967, 13, 264. Mikailoff, H.; Clottes, G. Bull. Inf. Sci. Technol. 1974, 196,
322; Plutonium Fuels Technology, Proceedings of 41–48.
the International Symposium; Latimer, T. W., 265. Bauer, A. A. Reactor Technol. 1972, 15, 87–104.
Jacoby, W. R., Eds. 266. Richter, K.; Coquerelle, M.; Gabolde, J.; Werner, P. In
240. Harbourne, B. L.; Levine, P. J.; Latimar, T. W.; Fuel and Fuel Elements for Fast Reactors, Proceedings of
Neimark, L. A. In Fast Reactor Fuel Element Technology, a Symposium on Fuel and Fuel Elements for Fast
Proceedings of Conference, New Orleans, LA, Apr Reactors, Brussels, July 2–6, 1973; IAEA: Vienna,
13–15, 1971; Farmakes, R., Ed., American Nuclear Austria, 1974; Vol. I, pp 71–84.
Society: Hinsdale, IL, 1971; pp 869–893. 267. Prunier, C.; Bardelle, P.; Pages, J. P.; Richter, K.;
241. Kelly, J. C. R., Jr.; Biancheria, A.; Anderson, C. A., Jr.; Stratton, R. W.; Ledergerber, G. In Fast Reactors and
Hobson, R. G.; Beckett, E. F. ANL-7520, Pt. 2, 1968, Related Fuel Cycles, Proceedings of International
pp 388–418. Conference on Fast Reactors and Related Fuel
242. Lierde, W. V.; Delbrassine, A.; Coheur, L.; Gorlé, F.; Cycles (FR0 91), Kyoto, Japan, Oct 1991; Atomic
Verstappen, G. Trans. Am. Nucl. Soc. 1974, 19, 99–100. Energy Society of Japan: Tokyo, Japan, 1991; Vol. 2,
243. Smailos, E. Report KfK-1953, Kernforschungszentrum pp 15.9-1–15.9-12.
Karlsruhe, 1974. 268. Tanaka, K.; Maeda, K.; Katsuyama, K.; Inoue, M.;
244. Fee, D. C.; Johnson, C. E. Report ANL-AFP-10, Argonne Iwai, T.; Arai, Y. J. Nucl. Mater. 2004, 327, 77–87.
National Laboratory, 1975. 269. Kummerer, K. R. J. Nucl. Mater. 1984, 124, 147–152.
245. Giacchetti, G.; Sari, C.; Walker, C. T. Nucl. Technol. 270. Leitnaker, J. M.; Beatty, R. L.; Spear, K. E. Trans. Am.
1976, 28, 216–225. Nucl. Soc. 1969, 12, 86–87.
246. Bradbury, M. H.; Matzke, Hj. J. Nucl. Mater. 1980, 91, 271. Arai, Y.; Morihira, M.; Ohmichi, T. J. Nucl. Mater. 1993,
13–22. 202, 70–78.
3.17 Thermal Spectrum Control Rod Materials
R. M. Horn, B. D. Frew, and P. Van Diemen
GE Hitachi Nuclear Energy, LLC, San Jose, CA, USA
485
486 Thermal Spectrum Control Rod Materials
IASCC Irradiation assisted stress corrosion with earlier GEH (then GE nuclear energy) designs.
cracking The CRB has four wings, each of which contains
IGSCC Intergranular stress corrosion absorber tubes or containers for the neutron poison
cracking material. The wings are attached to a central structure,
KWU Kraftwerk Union which itself is connected to the coupling structure
LOCA Loss of cooling accident below the CR and to a handle structure above that
LWR Light water reactors allows servicing during a refueling outage. Within the
NWC Normal water chemistry four wings, this design employs tubes, typically 18–21,
PWR Pressurized water reactors which are filled with the neutron poison material,
RCCA Rod Cluster Control Assemblies which is primarily B4C powder. The earliest designs
REO Rare earth oxides employed a sheath to cover the tubes and retain them
RIS Radiation induced segregation within the structure of the CRB. All of the structural
SCC Stress corrosion cracking materials used in the CRB have always been austenitic
SSE Safe shutdown earthquake stainless steel. The BWR utilizes CRs for both power
SSRT Slow strain rate test shaping and core reactivity control, although typically
STEM Scanning transmission electron <20% of the CRs are used at any given time for hot
microscopy operation. The CRs used during operation see very
VNC GE Vallecitos Nuclear Center high duty in terms of neutron poison material deple-
tion and/or transmutation, as well as irradiation effects
on the austenitic stainless steel structural materials.
A later section will discuss the current understanding
3.17.1 Overview of Light Water of the effects of neutron fluence on standard composi-
Reactor Control Rod Designs tion austenitic stainless steel materials as well as on the
specialized compositions developed for the BWR CR
There are major differences between the boiling water application. The control function requires that the
reactor (BWR) and pressurized water reactor (PWR) CRBs are available for both reactor scram (shutdown
control rod (CR) systems and designs. Strasser and function), and reactivity regulation to compensate for
Yario1 in their comprehensive review of the properties changes in core reactivity due to burnable absorber
of the burnable poisons, also covered the basic depletion, flow changes, CRB poison depletion effects,
designs and differences in the CRs, many of which fission product buildup, etc. Because many BWRs
are still in use today. Given that this chapter is generally are now operating on 2 year refueling cycles,
directed at structural materials used in the CR com- the CRs must have adequate mechanical and nuclear
ponents, the overview will start by focusing on the life to be able to complete a multiple number of
general design and material characteristics for each operating cycles. The evolution of the current design,
type of reactor. A more in depth discussion of the detailed in the later sections, has been driven by the
specific detailed design differences between vendors need to incorporate changes in the structural design to
and the relationship to the poison materials will be improve the design lifetime of the CRB. It should be
covered in later sections. acknowledged that it takes time to replace a BWR CR
during a refueling outage and then requires a tempo-
rary storage location until it can be disposed of in
3.17.1.1 BWR CR Design Summary
accordance with the appropriate regulations for radio-
The current BWR vendors include General Electric active waste material. Given the high costs associated
Hitachi (GEH) and Hitachi (HGNE), Toshiba/ with disposal, the utility is always looking for a longer
Westinghouse (formerly Asea Atom (ASEA) and life CR to minimize the volume of radioactive waste.
Asea Brown Boveri (ABB)), and Areva (which The general trend has been to improve the designs so
includes Siemens/Kraftwerk Union (KWU)). In the that they are not mechanically limited and then to
BWR system, the CRs enter from the bottom of the optimize the neutron poison material type, quantity
core. All current BWR plants use cruciform shaped and configuration, to obtain the maximum nuclear life.
control rod blades (CRBs) that fit between the fuel As stated earlier, the BWR neutron poison materials
assemblies. The specific design of the CRB varies have been primarily B4C, although recently, hafnium
from vendor to vendor, as is discussed later. Figure 1 (Hf ) has been used in key locations of the CRB to
displays a typical BWR design,1 which is consistent extend the design life.
Thermal Spectrum Control Rod Materials 487
Typical upper handle other CRs in the system. The CRBs are ready to
support both reactor scram (shutdown function),
and reactivity regulation to compensate for CRB
depletion effects, fission product buildup, etc.
The NRC (Nuclear Regulatory Commission)
accepted criterion for nuclear lifetime has been a
10% reduction in reactivity worth from its ‘new’
condition. Within this criterion, CRs are allowed to
be treated as ‘new’ when performing core physics
calculations. In a typical BWR, there may be as
many as 200 CRs, each with a unique absorber deple-
Neutron absorber rods tion history.
Absorber length
purposes, the reactivity worth is calculated in a con-
servative manner, that is, a best estimate reactivity
157.0
Absorber
worth of all rods is computed. Generally, a 10% con-
servation allowance is made for calculational uncer-
tainties. Also, the assumption that the most reactive
rod cluster is stuck out of the core is applied.
165.0
1.25
dia.
its high thermal neutron cross-section to perform the as the mechanical property changes. As the designs
poison function. With the capture of neutrons, the have advanced, this requirement has changed. As
transmutation of boron leads to the formation of generally described next, the life requirements are
helium and lithium resulting in swelling. As discussed driven by the ability to eliminate any risk of stress
in the following sections, hafnium has also been used, corrosion cracking (SCC) over the projected life. The
as the desire for long life has been driven by the CRB also must maintain adequate mechanical
utility commercial needs for long fuel cycles and strength and fracture resistance to maintain design
long life components. Hafnium has the advantages margin against the loads attributed to scram and
of continuing to absorb neutrons even after transmu- rapid insertion.
tation takes place and to remain in the metallic form
without any measurable swelling effects. All of the 3.17.2.2.1 Life requirements
BWR vendors, having started as licensees of General The fundamental requirement for any CRB is to be
Electric Nuclear Energy, now GEH, use similar able to perform its safety function during the speci-
approaches to these CR poisons. Some test specimens fied lifetime. The safety function is the capability to
using rare earth oxides (REO) have also been inves- shut down the reactor and maintain it in a safe shut-
tigated, given their high neutron absorbing capability. down condition. This applies equally to both BWR
These REO poisons have the potential for very long and PWR technology. For the BWR, the CRB per-
nuclear lifetime, but are expensive and result in forms the dual role of controlling the power level and
highly activated waste products creating storage and providing rapid shut down when a reactor trip signal
disposal concerns, and therefore have not been pur- is initiated. For control of the BWR power level, a
sued. Strasser and Yario1 list initial details of several small fraction (typically <20%) of the CRBs are
different designs and the typical design parameters. partially inserted into the reactor core. The high
Table 1 presents typical key materials used in the neutron and g-flux that these inserted CRBs are
different components of early and more current CRB exposed to result in neutron poison material deple-
designs. While the materials listed are those currently tion as well as structural material property changes.
used, it is informative to discuss the specific design The change in neutron poison effectiveness and
differences as well as the materials issues that have material properties is gradual with no known thresh-
led to the material approaches used in CRs of today. old effects. The only exception would be potential
It should be recognized that many plants are starting breach of the poison material containment structure,
to reach the 20þ years of life of standard duty CRBs, which could lead to escape of the poison material and
and only now are the designs that were developed in a more rapid change in neutron absorption effective-
the late 1980s and early 1990s reaching their end of ness. Since many of the CRB designs use compart-
service life. mented containment of the neutron poison material,
there is no common mode failure.
There are two requirements on the CRB lifetime,
3.17.2.2 Duty Requirements for CRBs
first is to define how much neutron fluence it will
The original neutron fluence requirements were take to reach a 10% reduction in reactivity worth
established by both the structural material’s ability (a measure of neutron absorption effectiveness) in
to withstand the stress from absorber swelling as well the cold shut down condition, and second, provide
CRB designation Capsule/tube material/ Structural material Assembly method Poison materials
poison container
Original GE Type 304 Type 304 Spot welded sheath and tubes B4C
Original Siemens Type 321/347 Type 321/347 Sheath and tubes B4C
GE D230 Type 316L/Type 304S Type 316L Sheath and tubes B4C
GE Marathon Type 316L/Type 304S Type 316L Laser welded tubes B4C and Hf
Hitachi Type 316L Type 316L Spot welded sheath and plates B4C and Hf
Westinghouse/ Type 316L Type 316L Drilled plate B4C and Hf
ABB
490 Thermal Spectrum Control Rod Materials
an accurate neutron fluence tracking method to earthquake (SSE). In this event the reactor fuel
determine where a CRB is in its lifetime. Computer undergoes horizontal deflection from the SSE, and
simulation programs can be used to predict the for the BWR, the CRB is rapidly inserted from the
amount of expected CRB lifetime, which will be bottom of the core and must conform the deflection
expended in any given fuel cycle. The utility owner of the much heavier fuel assemblies. In the PWR, the
can then select the CRB model which provides the control cluster drops in from the top of the core and
necessary number of whole fuel cycles. It would be must insert into the fuel assembly guide tubes while it
very undesirable to have a CRB reach its end of undergoes horizontal deflection from the SSE. The
life before the completion of a complete fuel cycle. magnitude of the fuel horizontal deflection is depen-
The reactor operator would have to withdraw the dent upon the magnitude of the SSE at the specific
end-of-life CRB and use an alternate CRB insertion plant site and the fuel assembly design characteristics,
pattern that may not have the same optimized fuel and is usually unique to each fuel design.
cycle operating strategy. During normal scram insertion (without the SSE)
In order to determine the nuclear lifetime of a for the BWR, the CRBs are rapidly accelerated and
CRB, a depletion analysis is required. The neutron decelerated to meet the plant technical specification
absorption rate in each individual absorber compart- insertion times. In order to minimize the deceleration
ment (e.g., absorber tube, drilled hole, etc.) is consid- loads at the end of the insertion, the drive mechanism
ered based on its location in the CRB. Neutron has a buffer system. This can be either hydraulic or
absorber located near the end of a wing of a cruci- mechanical. A failed buffer is also considered in the
form shape CRB receives a much higher neutron structural evaluation of the CRB, whereby there are
fluence than that located near the center. Neutron very high impact loads on the CRB at the end of the
poison materials such as boron carbide have a high scram insertion. This will result in high local stress on
neutron self-shielding effect. Although the thermal the coupling between the CRB and drive system.
neutrons are absorbed mainly at the surface of the Absorber material or absorber containment material
boron carbide, higher energy neutrons are absorbed that is free to move within the CRB outer structure
uniformly throughout the absorber cross-section. will also receive high impact loads at the end of the
The net result is an exponential decrease in the scram insertion. Actual scram loads are specific to a
CRB reactivity work as a function of neutron fluence. drive design and vary by vendor.
The first 10% reduction in reactivity worth, however, Other mechanical loads, such as internal pressure
can be approximated with a linear fit. This makes the buildup from helium generation or absorber swelling
tracking of the CRB lifetime much easier during under irradiation are also considered in the mechan-
power operation. ical design.
The PWR control elements are normally only
used for shut down of the reactor and are not used
3.17.2.3 GE CR Designs and Development
for power level control, although such usage is
increasing. These control elements then do not nor- The types of General Electric’s BWR CRs that are
mally see the same level of neutron and g-flux as the in service are the Duralife and Marathon series
BWR CRBs and have the potential for longer service (Figure 3).2 Both CR series utilize high purity stain-
life. The tips of the PWR control elements that are less steel (HPSS) to mitigate irradiation assisted
nearest to the reactor fuel will be exposed to some stress corrosion cracking (IASCC) and low cobalt
level of neutron and g-flux that needs to be tracked. materials for radiation reduction at the time of dis-
The objective for any supplier of CRBs is to spec- posal. The Duralife and Marathon CRs have absorber
ify a nuclear lifetime based on allowable neutron configurations with strategic placement of hafnium
fluence and design the absorber configuration and metal and additional amounts of boron carbide (B4C)
containment structure to be less limiting than the powder for increased lifetime. To improve the CRB
nuclear lifetime. This makes tracking the CRB life- lifetime, the primary developments have been
time simpler for the reactor operator. focused on the use of HPSS with excellent irradiated
properties to meet the exposure requirements. These
3.17.2.2.2 Upset load requirements developments started with the discovery of absorber
The worst case loads applied to the CRB are the tube cracking in 1978, with SCC identified as the
scram loads in combination with a design basis earth- primary cracking mode. This discovery led to an
quake, also referred to as the safe shutdown extensive program to find a material capable of
Thermal Spectrum Control Rod Materials 491
impurities and the maximum allowable concentration of another tube to protect them from the BWR envi-
of these impurities. Additionally, it identified composi- ronment. None of the capsules failed, with local
10
tions additions that increased the resistance against B depletion in excess of 90%. The capsules were
SCC processes. To verify the strain capability of stain- located in the fuel to maximize burnup in the shortest
less steel out of the BWR environment, irradiated period of time. This led to the development of the
samples of CPSS absorber tubing were tested in the HPSS that had significantly higher strain capability
GE Vallecitos Nuclear Center (VNC) hot cell, using in the irradiated state.3 Confirmation tests to demon-
an expanding mandrel test. The expanding mandrel strate the capability of HPSS tubing were performed.
test was designed to simulate the swelling character- Specifically, B4C absorber rods with HPSS tubing were
istics of B4C in the actual CRB. The tubing samples irradiated in both BWRs and PWRs. The absorber rods
ranged from 20 to 80% local depletion. The results of attained 3% strain in the BWR environment and the
the test demonstrated that all the samples had strain PWR environment without failure, as shown respec-
capability in excess of 5% strain without failure. The tively in Figures 4 and 5.4 In the same reactors, CPSS
results confirmed that stainless steel has high strain were found to fail at <1.5% strain (Figures 4 and 5).
capability when not subjected to the BWR environ- Constant extension rate tests (CERTs) were also
ment. Based on hot cell examination of CPSS absorber performed on irradiated samples of HPSS tubing.
rod tubing with cracks, the failures were occurring at The results confirmed that the HPSS tubing is resistant
<1% strain. This low strain value is less than the to SCC at a fast fluence of 3 1021 cm2, in sharp
demonstrated strain capability of irradiated CPSS contrast to the IGSCC susceptibility of CPSS
(i.e., 5% strain) outside the BWR environment. shown in Figure 6. Given this definite improvement,
Swelling capsules made of CPSS and filled with GE’s HPSS was continually evaluated in operating
B4C (70% of theoretical density) were irradiated in BWRs and established that the CRBs were able to
an operating BWR. The swelling capsules were inside meet the recommended end-of-life burnup. Hot cell
3.5
1 2 3 Cycles
%
GE* Intact
3.0
Defective
GE* results from absorber
tube measurements
2.5 KWU* results from in-core
stress corrosion capsules
KWU* KWU*
Diameter increase
2.0
1.5
1.0
0.5
0
CPSS HPSS CPSS HPSS CPSS 316
304 348
Figure 4 Strain to failure after exposure to boiling water reactor core conditions. Reproduced from Brayman, K. W.;
Van Diemen, P. Kerntechnik: vereinigt mit Kernenergie 1992, 57(2). © Carl Hanser Verlag Muenchen 1992.
Thermal Spectrum Control Rod Materials 493
3.5
1 2 Cycles
Intact
3.0 Defective
CP HP
( ) Diameter increase
estimated
2.5
Diameter increase (%)
2.0
1.5 ( (( (( ( ( ( ( ( ( (( (( ( ( (( (
( (
1.0 ( ( (
( (( (( ( ( ( ( (( ( ( ( ( (
( (
( ( ( ( ( (
0.5
0
a b d a b a b a b c a b
DIN 1.4541 AISI AISI Inconel Inconel Inconel
321 348 316 625 X-750 718
Figure 5 Strain to failure after exposure to pressurized water reactors core conditions. Reproduced from Brayman, K. W.;
Van Diemen, P. Kerntechnik: vereinigt mit Kernenergie 1992, 57(2). © Carl Hanser Verlag Muenchen 1992.
no IASCC
14
3.17.2.3.2 Development of crevice-free
12
High purity stainless steel design CRBs
10 Concurrent with the efforts to employ the HPSS
structural materials, it became clear that the sheath
8
design used in the original design Duralife CRBs led
6 to risks of spot weld induced crevices with associated
Threshold strain for local weld residual stresses. With the maturity of the
4 IASCC for commercial
design, cracks were found in the edge tubes and a
purity stainless steel
2 smaller number of cracks were found in the adjacent
tubes. No cracks were found in tube positions that
0 experienced the highest fast fluences levels up to
0 10 20 30 40 50 60 70 80 100
10
3.5 1021 cm2. The cracks in tube positions 1 and
Local B burnup (%)
2 were oriented longitudinally. The cracks were lon-
Figure 6 Comparison of failure strain for commercial gitudinal and were not distributed uniformly around
and high purity stainless steel alloys in boiling water the circumference of the tube as had been observed in
reactor environment after irradiation. Reproduced from
Brayman, K. W.; Van Diemen, P. Kerntechnik: vereinigt mit
the earlier investigations of the strain limited CPSS
Kernenergie 1992, 57(2). © Carl Hanser Verlag Muenchen absorber tubing material subject to IASCC. These
1992. cracks were consistent with the creviced design in
494 Thermal Spectrum Control Rod Materials
that location. These crevice positions promoted an In 1987 GE initiated a program to limit the cobalt
acceleration in the initiation of cracking due to the concentration, by specification, in all the CR compo-
aggressive environment associated with the crevice. nents to <500 ppm (typically 100–300 ppm). From
By eliminating the crevice, this incidence of cracking 1988 to 1990 each of the various CR components
could be mitigated. Likewise, the use of hafnium were changed to low cobalt materials. In the early
absorber rods in the tube edge positions addressed 1990s, this program was successful in producing CRs
the cracking problem. This concern, addressed in the with all low cobalt materials.
unirradiated structural components in the BWRs in
the late 1970s, is also addressed in all versions of the 3.17.2.3.5 CRB handle cracking
Marathon CR, which does not contain any wetted Following the efforts to remove the cobalt sources in
crevices in the design. the handle pin and roller assembly, cracking was
found in a number of CRBs after varying times of
3.17.2.3.3 Inclusion of hafnium materials operation. An example of the detected cracking is
In the GE designs, hafnium containing CRs have shown in Figure 7.7 The cracking emanated from a
been used successfully. The current design Marathon drilled hole in the handle into which was inserted the
CRs use hafnium rods inside the stainless steel tubes roller pin. The cracking experiences were limited to a
in order to provide longer life compared to B4C. The few BWR plants. The root causes were assessed to be
hafnium rods have been provided with enough clear- related to contamination of the machined hole, sig-
ance inside the tube to allow for 100% hafnium nificant cold work, and residual stresses from the
hydride formation and associated swelling without drilling and plug welding process, thereby being sim-
exceeding the allowable cladding strain limit. The ilar in cause to standard IGSCC cracking processes.
nominal and worst case clad strain, at 100% hafnium This led to the development of an interim captured
hydride, was found to be <0.5%, well below the cap- pin design for the pin and roller to preclude the
sule’s capability. Note that hafnium is not expected to development of any stress to initiate and propagate
hydride in the BWR environment but was conserva- cracking. The longer term solution was to implement
tively addressed in the design. a roller-less handle design, which employed only a
raised spacer pad or plain handle. This design was
3.17.2.3.4 Cobalt elimination efforts improved in several ways: it (1) eliminated the crevice
CRs have also been identified as a major in-core and the potential for trapped contamination, (2)
contributor of cobalt-60 to the reactor system.5 CRs eliminated cold work hardening from the drilling
are fabricated from stainless steel, which contains operation, and (3) eliminated any residual stress
cobalt-59 as an impurity. The original equipment from the plug weld.
Duralife CR design, in use since the 1960s, had pins
and rollers made of Haynes Alloy 25 and Stellite 3,
respectively. These materials were selected for their
superior wear resistance, but they contain 50%
cobalt. A joint study by GE and the Electric Power
Research Institute was initiated in 1980 to develop
alternate low cobalt pin and roller materials.6 Several
pin and roller materials were tested in the laboratory,
and subsequently the pin and roller materials with
the best wear resistance were placed in operating
reactors. GE selected replacement pin and roller
materials based on the results of this program. The
rollers were changed from Stellite 3 to Inconel X-750
and the pins were changed from Haynes Alloy 25 to
PH13-8 Mo stainless steel. Since 1983, low cobalt
pins and rollers have become standard on all replace-
ment CR, as well as being installed as retrofits on Figure 7 Cracking was found in a number of control rod
blades after varying times of operation in the pin and roller
original design CRBs. This change, eliminating the location. Reproduced from Department of Nuclear
use of cobalt-base alloys, resulted in the single largest Regulation Atomic Energy Council, R.O.C AEC/NRC
reduction in cobalt activity from CRs. Bilateral Technical Meeting, May 31–Jun 2, 2005.
Thermal Spectrum Control Rod Materials 495
3.17.2.3.6 Summary of the current GEH allow adequate clearance such that the outer absorber
Marathon CRB design tube under the worse case dimensional and swelling
The Marathon CR design has evolved to incorporate conditions of 100% local depletion will not be sub-
features to preclude IASCC during operation to its jected to any imposed strains. Finally, as stated earlier,
design end-of-life. All of the features are directed at the design of the roller region has been changed to
mitigating the factors that led to SCC: improved mitigate the potential for cracking there as well.
materials and the use of design features to prevent
swelling stresses and manufacturing stresses. Specifi-
3.17.2.4 Toshiba/Westinghouse (Formerly
cally, these include a B4C containment system, which
ABB) BWR CRB Designs and Development
provides a clearance between the outer diameter of
the containment system and the inner diameter of the In what is now Toshiba/Westinghouse, the CR design
absorber tube, thereby allowing the B4C to swell for a and development activities were conducted under ABB.
significant period of time without straining the tube Similar to GE, this CRB design has been provided
wall. Figure 8, applicable to the original Marathon extensively for BWRs. The basic design concept is
design, depicts the strain characteristics of the tubing different from GE in that it employs solid stainless
as a function of local burnup, with and without the steel blades into which are drilled holes that contain
B4C containment system. The B4C containment sys- the neutron absorber material. Figure 9 displays the
tem along with the HPSS tubing material act together various configurations in comparison to the GE designs
to reduce the susceptibility to IASCC even further.2 (Figure 3). Similar to GE, the original approach was to
Recent efforts have also led to tighter clearance use entirely B4C powder in all poison containing holes
requirements as well as modified structure shapes designated the CR 70 design. However, during the
based on finite element analysis (FEA) to reduce any early 1980s, the hole locations exhibited significant
predicted strains at high burnup. The latest simplified swelling, which limited the life of the CRBs. This
or ‘round’ absorber tube shape incorporates the most swelling occurred in the top segments of the blades as
conservative strain criteria on the outer absorber tube. expected because this is the segment of the blade that
Whereas the original Marathon design delays the receives the highest neutron fluence. Consistent with
onset of strain on the outer absorber tube (Figure 8), the GE CRBs, the Westinghouse components also
the newest Marathon design applies a zero strain cri- incorporated metallic hafnium to achieve longer
teria. This design, employed in advanced boiling water nuclear life in their CR82 and CR85 CRBs that are
reactor (ABWR) designs, sizes the inner capsule to also detailed in Figure 9.8 Because Hf does not swell as
it transmutes, the risk of SCC was reduced. This was
confirmed by exposures in a Swedish BWR over several
cycles. While these Hf tips are very efficient, the
remaining B4C regions can still experience cracking
as swelling progresses and the stainless steel hardens
Without containment with irradiation. Improvements were developed to
optimize both the hole design and to also employ
improved stainless steel similar to the GE approach.
Strain
CR 70 CR 82
Hafnium pins
Boron carbide
Boron carbide
Disconnectable shaft
Hf-tip design
All B4C design
(a) (b)
CR 85
Experimental CR
(Ultima)
Boron carbide
pellets
Hafnium pins
Smaller hole
diameter
Hafnium plugs
Closer hole to
Boron carbide hole separation
New steel
Hafnium bar
Ultima design
(c) (d)
Figure 9 (a–d) The various configurations in Westinghouse control rod blades in comparison to the GE designs
(shown in Figure 3). Reproduced from Jonson, A. Kerntechnik: vereinigt mit Kernenergie 1992, 57(2). © Carl Hanser Verlag
Muenchen 1992.
Building on the understanding of the degradation changed from Type 304 to Type 316. This switch
process, modifications were introduced into the basic was based on a large program that was undertaken
ABB BWR CR concept similar to the approaches to evaluate the IASCC resistance of the two types of
taken by GE. In order to improve strain capability alloys. The program tested specimens that were irra-
of the structural material, the type of steel was diated in the Barsebäck 1 reactor core to fast neutron
Thermal Spectrum Control Rod Materials 497
fluences in the range 3 1021 to 3.5 1021 cm2 (E > 1 distribution designed to minimize the stresses at the
MeV). Slow strain rate testing (SSRT) was per- surface facing the aggressive BWR core environment.
formed on the specimens in an autoclave connected Recent efforts have been made to optimize the design
to the water clean-up system of the reactor upon and fabrication processes to preclude the effects of
completion of the irradiation. Tests were performed irradiation and B4C swelling on the CRB stainless
in normal water chemistry (NWC) as well as a hydro- steel cracking.9 The primary design evolution for
gen peroxide doped environment, which was to their CR99 was to use hot isostatic pressed (HIP)
simulate the aggressive oxidizing nature of the envi- solid boron carbide pins as the absorber material.
ronment in the core. Type 316 stainless steel (SS) was The diameter of these pins is sized such that they
found to be superior to Type 304 SS: cracks were are smaller than the drilled holes, allowing the pins to
observed in 95% of the investigated specimens of swell/expand some within the holes. The swelling
Type 304, but only in 15% of the Type 316 speci- rate of these compacted powder pins may be different
mens (neutron fluence range 1.2–3.3 1021 cm2). The than the swelling rate for solid B4C. Similar to the GE
ABB study, in contrast to the GE study, did not show experience, the life of the CRB is being managed by
any benefits of low P (0.008%) or S (0.001%) impu- minimizing the effects of strain through design and to
rities in Type 304-SS. Additionally, the program also a lesser extent, through material selection.
confirmed the benefits of hydrogen water chemistry
(HWC) to mitigate IASCC at the fluence levels
3.17.2.5 Other Vendor Designs and
tested. All of these data relied on small specimens,
Development
which could reproduce the failure behavior. The
ABB redesign also addressed the role of the swelling There have been several patented CRB designs that
strain on the CRB life. Therefore, the design employed utilize only hafnium as the neutron poison material.
a new pattern of holes that moved them closer together These designs have been utilized primarily in Japanese
and reduced the hole diameter. This design change BWRs. The benefits of an all hafnium absorber
reduced the predicted stress at the outer fiber at design are the longer nuclear life of hafnium when
given internal pressure, that is, 10B depletion by a factor compared to boron, the stability of the material,
of 2–3, as shown in Figure 10. whereby it does not generate a gas, and it does not
swell under normal oxidizing environment. The dis-
advantages are the higher cost of hafnium and the
3.17.2.4.1 Summary of the current Toshiba/
weight of the CRB. For an equal volume of 70%
Westinghouse BWR CR99 CRB design
density boron carbide and hafnium, the weight of
In summary, the Westinghouse basic BWR CRB con-
the hafnium is nearly 7.5 times greater than the
cept has largely proven to be successful. In order to
boron carbide. The CRB drive systems were qualified
prevent the occurrence of cracking in the tip of CRs
for a specific weight CRB, and therefore, the hafnium
operating mainly in shut-down positions, a Hf tip
quantity must be reduced to preserve the drive integ-
design was introduced in the mid 1980s. For CRs to
rity. The nuclear lifetime of the CRB is reduced
be used, deeply inserted in the core, a more IASCC
proportionally to the reduction in hafnium quantity.
resistant steel was introduced with an absorber
Several patented hafnium absorber configurations
have been used where the hafnium is designed with
an inner coolant path to expose more surface area
120
to thermal neutrons (e.g., neutron flux trap). This
100
increases the initial reactivity worth with less hafnium,
Stress (MPa)
in conjunction with the residual stresses from spot limiting factor in CRB life. As can be seen in the
welds, produced sufficient stress to initiate the SCC schematic of the SCC process, the radiation acceler-
that was observed. Some cracking even led to loss ated process leading to material changes is the one
of sheath material from the blade (Figure 11).10 major parameter that could be improved within the
It should be noted that these CRB designs are no spectrum of environmental cracking (Figure 12).12–14
longer in service. As with the other BWR CRB designs, The dependence of IASCC on neutron fluence for
any source of stress has been sufficient to produce austenitic stainless steel is shown in Figure 13 for
cracking in stainless steels at the high fluences asso- BWR control blade sheaths,15,16 and in Figures 14
ciated with the end of life conditions for the CRBs.11 and 15 for laboratory tests on BWR-irradiated mate-
rial.17–19 Figure 14 not only shows that laboratory
tests mimic the field experience of irradiated com-
3.17.2.6 Overview of Limiting Structural
ponents with increasing fluence, but the smaller
Material Properties and IASCC Mechanism
inset within Figure 14 also shows the very strong
As discussed in the earlier sections, the unique and beneficial effect of reducing the corrosion potential.
complex form of cracking, IASCC, is the primary Above a fluence of 2 1020 n cm2 (E > 1 MeV),
Handle
(SUS316L)
About 25 cm wide
51
47
43
39
36
31
27
23
19
15
11
07
03
02 06 10 14 18 22 26 30 34 38 42 46 50
Positions of Hafnium blade type control rods in the reactor
: Control rod with cracks identified on its sheath
Sheath of about 3.7 m long
Hafnium
blade
Crack on
Tie-rod the tie-rod
(SUS316L)
Sheath
(SUS316L)
Missing part
Figure 11 Cracking observed in Japanese boiling water reactor in Hf-type control rod blade.
Thermal Spectrum Control Rod Materials 499
% IGSCC
at crack-tip N-Fluence 0.02
GB denudation 40 –0.05
Segregation
100
tests at 3.7 107 s1 on preirradiated Type 304 stainless
BWR creviced control blade sheath steel in 288 C water. The inset shows the effect of
80 corrosion potential via changes in dissolved oxygen
at a fluence of 2 1021 n cm2. GB, Grain boundary.
60 Reproduced from Angeliu, T. M.; Andresen, P. L.; Sutliff,
Threshold fluence for IGSCC J. A.; Horn, R. M. In Proceedings of the 9th International
≈ 5 ⫻ 1020 n cm–2 Symposium on Environmental Degradation of Materials in
40 Nuclear Power Systems – Water Reactors; AIME, 1999.
20
1022
Theoretical V observed relationships,
double ligament specimens
5 304 STST, 32 ppm O2 water, 288 ⬚C
3 7
⫻25 18
30 43 48 50–70 KSI
Fluence (n cm–2) (E > 1 MeV)
46 42 44 40 30–50 KSI
1021 62 60 61
59
71
69
70
1020
1019 1
10 102 103 104 105 106
Time to failure (h)
Figure 15 Comparison between observed and predicted time-to-failure for the effect of fast neutron fluence on
preirradiated Type 304 stainless steel tested at constant load in the laboratory in oxygen saturated, 288 C water.
is most important in oxidizing environments, while modeling also establishes that effective reduction in
radiation hardening may also be important in oxi- corrosion potential through HWC, akin to PWR
dizing environments and the dominant factor in environments, will retard the IASCC processes.
reducing environments. The recent research also This is best confirmed using crack growth behavior
suggests an important role on Si segregation to the as shown in Figure 18.23 The introduction of HWC
grain boundaries and as such could play a comple- shows dramatic reduction in rates. While this data is
mentary role to Cr in IASCC susceptibility. very promising, there are other data that suggest that
While there is significant attention paid to the the effectiveness is reduced at the higher fluences
grain boundary characteristics and their role on that are experienced by CRBs at the end of life
cracking behavior, the mechanical properties are fluences. Overall, the clearest path to improved life
also greatly affected. In particular, the yield strength may be to eliminate the stresses associated with
is increased by more than a factor of three and the B4C swelling.
uniform elongation reduced to 1–3% once the flu- In order to increase the resistance to IASCC, there
ence exceeds 1–2 1021 n cm2. This added harden- may be a need to design the compositions to either
ing is known to have great effects on both SCC address grain boundary susceptibility or to change
initiation and crack growth rates in unirradiated the deformation characteristics of the steels. Was et al.
materials. Figure 17 shows the clear effect of yield has shown that the stacking fault energy associated
strength on growth rates in austenitic materials. The with Ni content and other constituents could play an
strength levels for the CRB materials in their end-of- important role.24 In summary, all of these factors are
life condition exceed these strength levels.22 consistent with the general improvement of high
There have been efforts by Andresen and Ford to purity alloys for CRB materials, but they may not
model the crack growth rate dependencies on the be sufficient without efforts to reduce the corrosion
important degradation mechanisms in irradiated potential as well as reduce the strain through design.
stainless steels: RIS and irradiation hardening. Their
modeling predict a saturation in rates at 3 dpa.12–14
3.17.2.7 Summary of BWR CR Structural
The continuing depletion of Cr at the grain bound-
Problems
aries imply that the entire category of standard alloys,
304, 316, and 347 will very likely have similar As discussed with each of the designs, the CRBs
responses following irradiation. However, the employed in the BWR design have extreme duty
Thermal Spectrum Control Rod Materials 501
Cr (at.%)
Cr (wt%)
Fe (at.%)
Fe (wt%)
69 69
the CRB life when the B4C will exert a strain into
68 68 the stainless steel structure. The recent changes for all
67 of the designs have been focused on reducing or
67
eliminating the strains. Any strain could lead to the
66
risk of IASCC initiation in the reactor core environ-
ment. Great care is being taken in material tolerances,
15 Ni 14 Ni (at.%) the fitup stresses or the pellet-tube tolerances, to
Ni (wt%)
1.E – 06
Unsensitized 304, 304L and 316L SS and A600 Very high
288 ⬚C high purity water, 2000 ppb O2 martensite
CT tests at 27.5–30 MPa√m
Crack growth rate (mm s–1) Circles = high martensite SS
Triangles = alloy 600 Predicted
response
Two-sensitized points
1.E – 07 for comparison
Very low or no
martensite
Figure 17 The role of cold work equivalent to irradiation hardening on crack growth rates in austenitic structural materials.
1.E – 05
Sensitized 304 stainless steel
← 200 ppb O2
30 MPa√m, 288 ⬚C water
← 200 ppb O2
← 500 ppb O2
0.06–0.4 μS cm–1, 0–25 ppb SO4
Cold worked
Crack growth rate (mm s–1)
42.5
28.3
Cold worked
0.06 μ S cm–1
1.E – 09
–0.6 –0.5 –0.4 –0.3 –0.2 –0.1 0.0 0.1 0.2 0.3 0.4
Corrosion potential (VSHE)
Figure 18 The effect of hydrogen water chemistry (low corrosion potential) on crack growth rates in sensitized and
cold worked stainless steel materials. Reproduced from Andresen, P. L.; Angeliu, T. M.; Young, L. M. In Proceedings of
the Staehle Symposium on Chemistry and Electrochemistry of Corrosion and SCC; TMS: Warrendale, PA, Feb 2001.
Thermal Spectrum Control Rod Materials 503
which can vary from vendor to vendor. The major subsequently made gas-tight using welded end plugs.
poison as stated earlier is Ag–In–Cd for which there The use of the stabilized austenitic stainless steel is
is a low rate of reactivity loss with exposure. This factor consistent with the more common use of these alloys
along with the more benign mechanical requirements in Europe as the preferred approach for avoiding
(compared to BWR requirements) are the reasons for a sensitization during welding or thermal treatment
long design life. The Ag–In–Cd does present similar operations. The actual absorber material is an 80%
swelling concerns to that of the B4C poison material, Ag–15% In–5% Cd alloy. The remaining volume in
which is used by some vendors as well. The actual the CR is filled with helium at atmospheric pressure.
approach to incorporating the poison will be discussed The life-time of PWR clusters is limited on the
in the context of the specific designs. one hand by absorber-clad interaction, which is due
to absorber swelling and creep down of the cladding
surrounding the absorber, and, on the other hand by
3.17.3.2 Design Overview for PWR CRs
wall thinning due to wear. The nuclear life-time, that
The mechanical design of CRs must accommodate is, the time up to a reduction in RCCA worth of 10%,
the stresses introduced by the external pressure of is not considered here since it is considerably higher
reactor coolant, dynamic stresses produced during than 30 years.22 The swelling is attributed to the
scrams and during seismic events, loads during loss neutron capture, which then leads to the conversion
of cooling accident (LOCA), and by the mechanical of In into Sn and of Ag into Cd. While the reactor
interaction between clad and absorber materials. pressure acts as an external overpressure on the clad-
These are the same as discussed in detail for the ding, creep of the stainless steel cladding closes the
BWR CRs. The allowable stresses are based on values initial gap during irradiation and then the absorber
from design codes, such as the ASME Boiler and swelling will produce tensile stress in the clad. Simi-
Pressure Vessel Code even though they are only a lar to the BWR CRBs, the RCCA lifetime is limited
guide. The key features of the assemblies of some by the ductility of the cladding material at that point
vendors are given in Table 2. Example designs will when there has been significant accumulation of
be discussed in the next sections with an emphasis on fluence.
the materials issues that have manifested themselves. The biggest concern, however, is the wear of CR
cladding. This wear, potentially, can occur with axial
movement of the RCCAs within the fuel bundle and
3.17.3.3 Areva/Siemens Designs
with vibration during plant operation. Specifically,
The standard Areva/Siemens design of the RCCA vibration of either the CRs or the mating structures
features a spider assembly with compression spring, can produce local wear, which in turn will thin the
bolt and spring cup, and CRs (shown in Figure 19) RCCA cladding wall. This will naturally result in an
that are fastened onto the spider assembly.22 The spider increase in cladding stresses.
assembly arms are manufactured from a forging, using Given the changes in properties of the stainless
milling and electrical discharge machining processes. steel cladding, the PWR designs are also needed to
These fabrication methods avoid the need for welded assess the stainless steel material behavior as a func-
or brazed connections and also lead to better mechani- tion of fluence. Starting in 1979, Siemens conducted
cal strength. The individual CRs are fastened onto the a test program to study the swelling behavior of
spider as an assembly using a welded lock nut design. various absorber materials under irradiation. Their
Each rod comprises an absorber, which is enclosed in a program irradiated test rods up to a maximum
stabilized austenitic stainless steel cladding tube and fluence of 7 1022 cm2 (E > 0.6 MeV). Density,
Table 2 Design features and materials used in example PWR control rods
diameter, and length measurements were performed temperature changes, low cobalt and boron content,
with the reported results shown in Figure 20.22 The and application processing parameters that would not
measured data were consistent with other measure- affect the microstructure of the cladding. The pro-
ments.23 It is apparent from Figure 20 that absorber gram identified chromium carbide (Cr3C2) in a NiCr
material swelling is essentially isotropic and a linear binder as the optimum coating. It could be applied
function of fluence. In the material test program, by the detonation gun process,25 which kept the work
the diameter of the steel cladding enclosing the piece (cladding) temperature at 150 C during coat-
absorber specimens were also inspected and evalu- ing application. The coating thickness was maintained
ated after each irradiation cycle. These data were used at a level that changes from both creepdown and
to develop a design curve to predict the in-reactor straining during reactor insertion would not result in
‘creepdown’ of steel cladding tubes. With an under- either cracking or spalling of the coating. Siemens
standing of the changes as a function of neutron flu- RCCAs with Cr3C2 coated CRs have been fabricated
ence, the amount of clad thinning could be calculated and inserted since 1979. The operating experience was
with operation and then used to address any wear confirmed to be excellent, as discussed in Heins et al.22
and still meet the design loads.22 At the same time,
it was recognized that the coating of CRs was the
3.17.3.4 Toshiba/Westinghouse
best approach to improve wear resistance. Siemens
(ABB and CE) Designs
concurrently set up a test program to assess the
use of coated CRs. There were several requirements The original Westinghouse RCCA design was shown
for the coating. It would have to have adequate in the first section in Figure 1, which had 53 full rod
corrosion resistance with exposure to the reactor and 8 partial rod clusters with 24 rods per cluster.26
coolant, improved clad wear resistance without trans- The designs employed cold worked Type 304 stain-
ferring wear to the mating components, no detrimen- less steel cladding and the Ag–In–Cd poison pellets
tal effects from neutron irradiation, no spalling from inside the cladding with a nominal gap. In Europe,
Toshiba/Westinghouse (formerly ABB) developed a
hybrid CR assembly that, after modifications, could
also be employed in other PWRs. The assembly was
∅ 9.3 different and consisted of 24 rods with Alloy 625
Thread ⎯ 3720 420
cladding attached to the same spider assembly. This
Figure 19 Siemens design for single control rod typical design used Ag–In–Cd as the absorber over the
of the assembly shown in Figure 2. Reproduced from
Heins, L.; Dambietz, W.; Fuchs, H. P. Kerntechnik: vereinigt
100 cm in the lower part, and B4C pellets were used
mit Kernenergie 1992, 57(2). © Carl Hanser Verlag in the remainder of the rod. This hybrid rod design is
Muenchen 1992. shown in Figure 21.
Density measurements
Diameter measurements
Length measurements
Conservative
Best estimate
0 10 20 30 40 50 60 70 80
Neutron fluence ⫻1021(cm–2) (E > 0.6 Mv)
Figure 20 Siemens data showing swelling of Ag–In–Cd poison material to be linear with fluence. Reproduced from Heins, L.;
Dambietz, W.; Fuchs, H. P. Kerntechnik: vereinigt mit Kernenergie 1992, 57(2). © Carl Hanser Verlag Muenchen 1992.
Thermal Spectrum Control Rod Materials 505
of compositionally modified candidate stainless have shown the role of stacking fault energy on
steels. However, with the continuing use of austen- IASCC susceptibility.24 Any future alloy develop-
itic stainless steels, these improvements have been ment may need to move significantly away from the
more incremental in nature as opposed to producing 18 Cr-8 Ni region of composition to reap the benefits
major changes in irradiated SCC resistance. The of crystal structure in reducing the impact of irradia-
recent emphasis has been directed at design improve- tion damage. Additionally, the structural alloy devel-
ments to significantly reduce or eliminate any strains opment needed for next generation BWR core
in the structural tubes of the CRB structure. This internals will very likely include the use of much
requires strict attention to the design tolerances higher Cr levels. With the desire to increase the
and has led to limitations on the amount of poison license period for all LWRs, there may be a renewed
material present as well as the cumulative allowable interest in modified alloys as well as modified designs.
fluence at the end of design lifetime. The difficulties The use of Hf as a poison may become more impor-
of developing better alloys remain, given the need to tant even though there may be design challenges
irradiate any candidate materials for a significant time associated with the increased weight.
to confirm the improved properties. Irradiation cap-
sules are expensive and the volume of material pro-
duced for laboratory evaluations are small, leading 3.17.5 Summary
to the use of testing techniques that are compromised
by this small size. Additionally, the development CRB technology has developed over the history of
time needs to be followed by insertion of in-plant LWR reactor operation. The original designs and
lead CRBs or CRB components made from any material approaches have been constrained by the
improved materials. This will also require the accu- reactor design limitations: the ability to interface
mulation of significant in-reactor duty followed with the fuel design, the limitations on the weight
by post irradiation evaluations. The recent issues in order to meet the structural requirements, and the
with crevice corrosion can also be accommodated limitations on the available neutron poison materials.
through design or manufacturing changes although However, given the differences in operation, there are
there are possibilities that small alloy changes will differences in the designs that have evolved. Many of
help mitigate the problems. As to the wear problems these changes have been driven by the need to resolve
experienced in PWR CRs, the choice of the alloy SCC or wear issues or the need to incorporate added
seems to lead to the required improvements. Given service life to the CR.
the early identification of solutions, the only issue In the BWRs, the presence of the coolant and core
may be material issues that surface at higher fluence environment along with the naturally occurring
associated with license renewal. swelling of the B4C absorber material have led to
Given the extended time to assess the perfor- extensive efforts to develop austenitic stainless steel
mance of CRBs, there is a reduced likelihood that materials with higher IASCC resistance as well as
future development will be conducted at the level increased elongation characteristics after high neu-
that was done in the 1980s and 1990s. For BWRs, tron fluence exposure. This chapter has reviewed the
there is an opportunity, given there still exists a need evolution of the designs based on vendor experiences.
to develop improved SCC and IASCC resistant stain- This evolution has included the development of
less steels for core internals in new plants that could improved materials, as well as design changes to
be used in CRBs. Specifically, there is a current eliminate crevices and weld residual stresses. The
interest in controlled grain boundary alloys which recent focus has been directed at modification of
attempts to eliminate most of the grain boundaries the designs to eliminate any absorber-imposed strains
with susceptible orientations for SCC. Given the at the end of life to add margin against any risk of
small size of the components, this technology could SCC. Recent field problems have also revealed the
have an application in CRBs. Hitachi has pursued risk of crevice corrosion in the vicinity of Hf absorber
these alloys as they have reported recently. The pri- material that can also produce cracking prior to the
mary experimentation in the past has been on the achievement of the component lifetime goal.
category of Type 304 and Type 316L alloys. As dis- The CR concept used in PWRs is quite different
cussed this work has focused on high purity or stabi- as is the poison materials of choice. The experience
lizing additions to attempt to reduce IASCC has been good with the exception of wear issues
susceptibility. Recent work by Was, Busby, and others experienced in the 1980s. These problems have
Thermal Spectrum Control Rod Materials 507
been overcome with material changes to reduce the control rod in BWR plants. In 14th International
Conference on Environmental Degradation of Materials in
surface effect. Nuclear Power Systems, Virginia Beach, VA, Aug 23–27,
Overall, the designs have matured in the four 2009.
decades of LWR plant operation, but the nature of 12. Andresen, P. L. In Stress Corrosion Cracking: Materials
Performance and Evaluation; Jones, R. H., Ed.; ASM:
the core environment coupled with the desire for Materials Park, OH, 1992; pp 181–210.
increased life continues to play a role in the ability 13. Andresen, P. L.; Ford, F. P.; Murphy, S. M.; Perks, J. M.
of the newest designs to meet performance goals. The In Proceedings of the 4th International Conference on
Environmental Degradation of Materials in Nuclear Power
time frame required to both assess material changes Systems – Water Reactors; NACE: Houston, TX, 1990;
and the benefits is very long. The major changes are pp 181–210.
focused on design improvement to remove the dri- 14. Ford, F. P.; Andresen, P. L.; Jacobs, A. J. Life prediction of
irradiated components subject to environmentally
vers for material degradation. However, with the assisted cracking. In International Symposium on Plant
desire to develop new plants as well as increase the Aging and Life Prediction of Corrodible Structures
operating life of LWRs to 80 years, there may be an Sapporo, Japan, May 15–18, 1995.
15. Gordon, G. M.; Brown, K. S. In Proceedings of the 4th
opportunity to once again address the fundamental International Conference on Environmental Degradation of
problems of the CRB structural materials properties Materials in Nuclear Power Systems – Water Reactors;
at high neutron fluences: the issues of material NACE: Houston, TX, 1990; pp 14-46–14-62.
16. Brown, K. S.; Gordon, G. M. In Proceedings of the
changes when exposed to core irradiation increase Environmental Degradation of Materials in Nuclear
environmental susceptibility and significantly decrease Power Systems – Water Reactors; AIME: Warrendale, PA,
material ductility. 1987; pp 243–248.
17. Jacobs, A. J.; Hale, D. A.; Siegler, M. Unpublished data,
GE Nuclear Energy, San Jose, CA, Jan 1986.
18. Angeliu, T. M.; Andresen, P. L.; Sutliff, J. A.; Horn, R. M.
References In Proceedings of the 9th International Symposium on
Environmental Degradation of Materials in Nuclear
Power Systems – Water Reactors; AIME: Warrendale,
1. Strasser, A.; Yario, W. Control rod materials and burnable PA, 1999.
poisons: An evaluation of the state of the art and needs for 19. Jacobs, A. J.; Clausing, R. E.; Miller, M. K.; Shepherd, C.
technology development, EPRI NP-1974; July 1980. In Proceedings of the 4th International Conference on
2. Brayman, K. W.; Van Diemen, P. Kerntechnik: vereinigt mit Environmental Degradation of Materials in Nuclear Power
Kernenergie 1992, 57(2), pp 107–112. Systems – Water Reactors; NACE: Houston, TX, 1990;
3. Austenitic Stainless Steel Alloy, US Patent 4,863,682, pp 14-21–14-45.
awarded to D.J. Coates, G.M. Gordon, A.J. Jacobs, and 20. Andresen, P. L.; Gott, K.; Nelson, J. L. In Proceedings of
D.W. Sandusky, Sept 5, 1989. the 9th International Symposium on Environmental
4. Garzarolli, F.; Alter, D.; Dewes, P.; Nelson, J. L. Degradation of Materials in Nuclear Power Systems –
In Conference Proceedings of the Third Symposium on Water Reactors; AIME: Warrendale, PA, 1999.
Environmental Degradation of Materials in Nuclear Power 21. Andresen, P. L.; Angeliu, T. M.; Young, L. M.
Systems – Water Reactors, Traverse City, MI; The In Proceedings of the Staehle Symposium on Chemistry
Metallurgical Society: Warrendale, PA, 1988. and Electrochemistry of Corrosion and SCC; TMS:
5. EPRI NP-2263, BWR Cobalt Source Identification, Warrendale, PA, Feb 2001.
Palo Alto, CA, Feb 1982. 22. Heins, L.; Dambietz, W.; Fuchs, H. P. Kerntechnik:
6. EPRI NP-2329, BWR Control Rod Cobalt Replacement, vereinigt mit Kernenergie 1992, 57(2), pp 84–89.
Palo Alto, CA, Mar 1982. 23. Sipush, P. J.; Woodcock, J.; Chickering, R. W.
7. Department of Nuclear Regulation Atomic Energy Council, Lifetime of PWR Silver–Indium–Cadmium Control
R.O.C AEC/NRC Bilateral Technical Meeting, Rods, EPRI NP-4512, Project 1628-4; Final Report,
May 31–Jun 2, 2005. Mar 1986.
8. Jonson, A. Kerntechnik: vereinigt mit Kernenergie 1992, 24. Was, G. S.; Andresen, P. L. Stress Corrosion Cracking
57(2), pp 113–115. Behavior of Alloys in Aggressive Nuclear Reactor
9. Control rod and a control rod blade for a boiling water Core Environments, CORROSION–Vol. 63, No. 1,
reactor, US Patent 7469029, awarded to A. Lunden, pp. 19–45, 2007; NACE International.
Dec 23, 2008. 25. Schwartz, E. Detonation Gun Coatings for Nuclear and
10. NISA/METI Communication; Actions taken against the Related Industries. In International Colloquium on
cracks on hafnium blade-type control rods in BWRs, Hardfacing Materials in Nuclear Power Plants, Avignon,
Mar 3, 2006. Sept 1980.
11. Ooki, S.; Miyazaki, A.; Tezuka, H.; Taguchi, M.; 26. Vedsterlund, G.; Hallstadius, L.; Hoffmann, H.;
Suzuki, S. Investigation of damages of irradiated Corsetti, L. Kerntechnik: vereinigt mit Kernenergie
316L stainless steel used for hafnium plate type of 1992, 57(2), pp 102–106.
3.18 Fast Spectrum Control Rod Materials
T. Donomae and K. Maeda
Japan Atomic Energy Agency, O-arai, Ibaraki, Japan
3.18.1 Introduction
Abbreviations
ACCI Absorber–cladding chemical
Control rods have three primary functions: reactivity
interaction
compensation during the operating cycle, start-and-
ACMI Absorber–cladding mechanical
stop control for normal operation, and rapid shut-
interaction
down during abnormal conditions.
BU Burnup
As an absorber material, boron carbide (B4C) is
EFPD Effective full power days
mainly used in fast reactor systems, such as the
FBR Fast breeder reactor
Japanese experimental reactor JOYO. The current core
MK-II Mark-II (core type of JOYO,
configuration of JOYO is Mark-III (MK-III). Figure 1
1982–2000)
shows the control rod locations in the JOYO MK-III
MK-III Mark-III (core type of experimental
core; six control rod assemblies are loaded into it.
fast reactor JOYO, 2003–)
The control rods are inserted automatically when
Na bonding Sodium bonding
abnormal conditions occur. In both the MK-II and
NMR Nuclear magnetic resonance
MK-III cores, all control rods have the same struc-
SEM Scanning electron microscopy
ture and functions. Every control rod assembly has
TD Theoretical density
seven control rod elements, and B4C pellets are used
TEM Transmission electron microscopy
as absorbers. The boron in B4C includes 10B and 11B.
509
510 Fast Spectrum Control Rod Materials
Shielding subassembly
Outer reflector
Neutron source Inner reflector
Outer core fuel subassembly
Inner core fuel subassembly
5
6
1
Control rod subassembly
4
3
2
Figure 1 JOYO MK-III core. Reproduced from Aoyama, T.; Sekine, T.; Maeda, S.; et al. Nucl. Eng. Design 2007, 237,
353–368.
Source: Aoyama, T.; Sekine, T.; Maeda, S.; et al. Nucl. Eng.
Design 2007, 237, 353–368; Aoyama, T.; Sekine, T.; Tabuchi, S.
Nucl. Eng. Design 2004, 228, 21–34.
Control rod Absorber pin The B4C stack area is 650 mm high, which is
greater than the driver fuel length (500 mm). Figure 2
B4C pellet (10B: 90 wt%) shows the control rods of JOYO.1,2 The main para-
meters of the JOYO MK-II and MK-III cores are
Figure 2 Control rod of JOYO MK-III. Reproduced from
Aoyama, T.; Sekine, T.; Maeda, S.; et al. Nucl. Eng. Design
shown in Table 1.1,2 Control rods and drive mechan-
2007, 237, 353–368; Aoyama, T.; Sekine, T.; Tabuchi, S. isms are shown in Table 2. Reactor types are classi-
Nucl. Eng. Design 2004, 228, 21–34. fied into three groups.
Fast Spectrum Control Rod Materials 511
Continued
512 Fast Spectrum Control Rod Materials
Table 2 Continued
a
The ultimate shutdown system (USS) injects B4C balls.
b
Diverse shutdown rods þ control and shutdown rods.
c
Group 1 and 2 rods.
HSR, hydraulically suspended rod; GEM, gas expansion module; MRR, movable ring reflector (Ni).
Safety (shutdown) rods: Number of safety (shutdown) rods.
Regulating rods: Number of regulating rods (or combined regulating and safety rods).
Rapid shutdown rods: Number of rods contributing to rapid shutdown within the first and second shutdown systems.
Additional shutdown rods: Number of additional, diverse, shutdown rods, or devices.
Continued
Fast Spectrum Control Rod Materials 513
Table 2 Continued
d
Fuel pins.
e
Control and shutdown rods/diverse shutdown rods.
f
Diverse shutdown rods.
Continued
514 Fast Spectrum Control Rod Materials
Table 2 Continued
g
MK-III: (17.6 in MK-I).
h
MK-III: (17.8 and 18.5 in MK-I and MK-II, respectively).
i
The reactor was controlled by movement of fuel pins in triangular clusters.
j
Diverse shutdown rods.
k
Control and shutdown rods/diverse shutdown rods.
Continued
Fast Spectrum Control Rod Materials 515
Table 2 Continued
l
Movable ring reflector (Ni).
m
Control and shutdown rods/diverse shutdown rods.
[If B4C is used it is abbreviated below as BCx where x is the enrichment (%10B) and if boron powder or sintered powder is used it is
abbreviated as Bx]
Continued
516 Fast Spectrum Control Rod Materials
Table 2 Continued
BREST-OD-300 (Russian 0.72 0.70 (per rod) 0.15 (per GEM) 0.26 (per 3.6
Federation) HSR)
Commercial size reactors
Super-Phénix 1 (France) 10.0 8.5 0.4 – 10.0
Super-Phénix 2 (France) 12.0 – – – –
SNR 2 (Germany) 2.9 8.5 0.4 – –
DFBR (Japan) 8.9 + 1.6 – – – –
CDFR (United Kingdom) 4.0 – – 5.0 –
BN-1600 (Russian Federation) 2.8 0.4 0.2 6.7 –
BN-800 (Russian Federation) 4.1 0.4 0.2 6.1 9.0
EFRp 10.3 8.1 0.34 – 10.3
ALMR (United States) 6.8 – – – –
SVBR-75/100 (Russian 1.37 0.48 0.27 6.0 6.48
Federation)
BN-1800 (Russian Federation) To be
determined
BREST-1200 (Russian To be
Federation) determined
JSFR-1500 (Japan) – – – – –
Breeding core 2.3 – – – 6.8q
Break even core To be determined
n
With boron follower (0.38 Bef – without boron follower).
o
Group 1 and 2 with one rod stuck.
p
Diverse shutdown.
q
Group 1 and 2 with one rod stuck.
Continued
Fast Spectrum Control Rod Materials 517
Table 2 Continued
r
MK-III: [900, 700 (fine and coarse), respectively, MK-I].
s
By B4C ball injection system.
t
Control and shutdown rods/diverse shutdown rods.
Continued
518 Fast Spectrum Control Rod Materials
Table 2 Continued
u
Control and shutdown rods/diverse shutdown rods.
HSR, hydraulically suspended rod.
Safety
Continued
Fast Spectrum Control Rod Materials 519
Table 2 Continued
Safety
DFBR (Japan) Screw drive gear with magnetic holdup and SASSx
CDFR (United Kingdom) Screw drive gear with magnetic holdup
BN-1600 (Russian Federation) Rack drive gear with magnetic holdup and accelerating spring
BN-800 (Russian Federation) Rack drive gear with magnetic holdup and accelerating spring
EFR (1) Electromagnet in gas and (2) electromagnet in sodium
ALMR (United States) B4C balls released into open thimble at core center
SVBR-75/100 (Russian Federation) Rack drive gear with electromagnetic holdup, accelerating spring, and visible lock
BN-1800 (Russian Federation) To be determined
BREST-1200 (Russian Federation) Electromechanical
JSFR-1500 (Japan) Screw drive gear with magnetic holdup and SASSx
v
Motor drive/spring acceleration in MK-I.
w
Diverse shutdown rods.
x
SSAS, a passive shutdown system (Curie point magnets).
SASS, self-actuated shutdown system.
Coarse Fine
Continued
520 Fast Spectrum Control Rod Materials
Table 2 Continued
Coarse Fine
y
MK-III. Motor drive/spring acceleration in MK-I.
Control rods function as safety (shutdown) rods, Table 3 Required performance of absorbers
regulating rods, rapid shutdown rods, additional shut-
Item Performance
down rods, absorber pins, worth of controlled rods, requirement
vertical travel of control rods, rod-drop time, and
features of the drive mechanism. Neutron cross section Large
Thermal conductivity High
Melting point High
Compatibility (with cladding materials) Good
Weight Light
3.18.2 Absorber Materials Induced activity Small
3.18.2.1 Performance Requirements of
Neutron Absorber Materials
Table 3 lists the performance requirements of neu-
tron absorber materials. The most important is the absorption ability, low cost, ease of fabrica-
neutron cross section. A large cross section enables tion, and low radioactivity after irradiation.
the control of the reactor with just a few control rods. Natural boron contains about 20% 10B, and the
High thermal conductivity prevents the absorber pins remainder 11B.
from reaching excessive temperature. A high melting Figures 3 and 43,4 show the neutron absorp-
point for the absorber material means that the mate- tion cross sections of 10B and 11B. Under irradiation,
rial can be used at high temperature and will remain the main capture reaction is 10B(n,a)6Li. During
stable. Good compatibility with cladding materials this reaction, helium gas is produced. Some of the
reduces the chemical interaction between the absorber helium gas accumulates in the B4C pellets, and
material and the cladding tube. the rest is released. This phenomenon induces pellet
swelling and/or increases the inner pressure in the
cladding tube.
There are other candidate materials for absorber
3.18.2.2 Variety of Absorber Materials
materials, for example, tantalum and europium.
A conventional absorber material is B4C; it They have advantages with respect to gas release
has been selected because of its high neutron and reduced swelling compared to B4C.
Fast Spectrum Control Rod Materials 521
B-10
104
Total
103 Elastic
(n,a)
102 Capture
Cross-section (barns)
101
100
(n,a)
10−1
10−2
10−3
10−4
10−5
10−1 100 101 102 103 104 105 106 107
Neutron energy (eV)
Figure 3 Neutron absorption cross-section of 10B. Reproduced from Young, P. G.; Stewart, L. Prog. Nucl. Energy 1984,
13(2/3), 193–233; Shibata, K.; Kawano, T.; Nakagawa, T.; et al. J. Nucl. Sci. Technol. 2002, 39(11), 1125–1136.
B-11
103
Total
102 Elastic
Inelastic
101 Capture
Cross-section (barns)
100
10−1
10−2
10−3
10−4
10−5
sinterable. Magnesium oxide is removed by washing 92% theoretical density (TD) are brittle. Structur-
in a hydrochloric acid bath. ally, the absorber pins are contained within a clad-
2. Carbothermic method5 ding tube. Therefore, the B4C may be in the form
This endothermic reaction produces a higher grade of a powder or pellets. At less than 800 C,
product, but in the form of an agglomerate which B4C retains about most of the tritium produced
must be crushed to obtain a sinterable powder. during reactor use. Tritium is generated in B4C by
the reactions such as 10B(n,2a)T, 11B(n,9Be)T, and
3.18.3.2 Structure
10
B(n,a)7Li(n,n0 a)T, in general. The cladding tube
is not capable of blocking tritium gas that escapes
1. Phase diagram from the B4C matrix.
Figure 5 shows the phase diagram of the boron– The thermal conductivity of B4C is an important
carbon system.6,7 Above 20 at.% carbon, a eutectic characteristic during operation.
reaction with a corresponding carbon content of The (n,a) reaction releases an energy of 2.78 MeV
31 wt% occurs at around 2375 C. Industrially per event. Most of this energy is stored within the
manufactured B4C leads to a product closer to B4C matrix, so the thermal gradient in the absorber
the phase limit of 20% carbon atoms, and the material is a direct function of thermal conductivity.
product contains up to 1% free carbon mass.8 Figure 79 shows the temperature dependence of the
2. Crystal structure5–8 thermal conductivity. In this figure, symbols show
The unit cell of B4C is rhombohedral,5–8 but it is the burnup level of the specimen as capture per
often described in the hexagonal system. The para- cubic meter. For unirradiated materials, the thermal
meters for the B4C structure are shown in Table 5.8 conductivity steadily decreases with increasing tem-
Figure 6(a)5 and 6(b)7 are typical descriptions of perature. For irradiated material (lower curves), the
the crystal structure of B4C. Both of them are rhom- thermal conductivity is reduced to less than the unir-
bohedral unit cells and the lattice belongs to the radiated values and remains nearly constant over a
space group of D3d/R3m. The unit cell contains 15 wide range of temperatures.
atoms as B12C3. The central position of C–C–C is
partially occupied by boron, as shown by nuclear 3.18.3.3.2 Chemical interactions
magnetic resonance (NMR) studies,5 so the C–C–C The chemical interactions of B4C with stainless
chain is easily changed into C–B–C as B13C2. steel cladding have been studied in a series of
At each vertex of the unit cell, an icosahedron is out-of-pile investigations.10 A compatibility test
formed with boron atoms. Each triangular face of the has been performed in sodium, to provide reliable
icosahedron has a polar site.6 The icosahedra also have data for the design of vented absorber pins with
an equatorial site. These positions are directly linked Na bonding. The possibility for reducing or elim-
to atoms in the icosahedra that lie next to each other. inating absorber–cladding chemical interaction
(ACCI) has been studied for chromium coating of
the surface of the stainless steel cladding. Chemical
3.18.3.3 Physical and Chemical Properties
interaction between B4C and the cladding material
3.18.3.3.1 Physical properties was studied with B4C powder in an inert atmo-
B4C is a light (2.5 g cm3) and hard (Hv ¼ 4.1 104 sphere. ACCI was negligible after 8600 h of testing
MPa) ceramic material. Hot-pressed pellets of at 600 C.
Fast Spectrum Control Rod Materials 523
4500
Gas
3500 Liquid
Temperature (K)
Liquid + graphite
3000
0.164–2731
2657
2500
2371
Graphite
B4+dC
1500
0.1 0.3 0.5 0.7 0.9
B Mole fraction (C) C
(a)
2800
2600
USED:
[61 EII] Eutect. R.
2400
[61 EII] Solvus
[61 EII] Liquidus
Temperature (K)
1400
0 0.05 0.10 0.15 0.20 0.25 0.30 0.35
(b) B Mole fraction (C)
Figure 5 Phase diagram of B4C. Reproduced from He, J. L.; Shen, Z. Q.; Wu, E.; et al. J. Alloys Compd 2007, 437,
238–246; Seifert, H. J.; Aldinger, F. Struct. Bonding 2002, 101, 1–58.
524 Fast Spectrum Control Rod Materials
Table 5 Parameters of the B4C composition were measured, and the rate of mass transfer from
Rhombohedral Hexagonal
B4C to sodium and/or cladding followed a parabolic
time law.
a 5.162 Å 5.599 Å Metallographic cuts of the uncoated cladding
c – 12.075 Å material revealed chemical interaction on the inner
a 66 680 –
surface of a cladding tube. This interaction is shown
Source: Bailly, H.; Menesssier, D.; Prunier, C. The Nuclear Fuel of in Figure 8(a).10 The region of interaction is char-
Pressurized Water Reactors and Fast Neutron Reactors: Design acterized by two or three uniform reaction layers
and Behaviour 1999, 531–565. with considerably higher microhardness than the
matrix, separated from the cladding material by a
dark region. Based on the experimental results in
the pure B4C/Na/stainless steel system, a correla-
tion between the growth of the interaction region
B
B and temperature has been found, which is shown
C in Figure 8(b).10 From the test results, it was con-
cluded that in a sodium-bonded vented absorber pin
C
with B4C pellets, ACCI occurred and the depth of
interaction was 48 mm within 1 year at 600 C. Con-
sidering this ACCI behavior, the lifetime of an
absorber pin is at least 420 effective full power
days (EFPD). Coating the inner surface with nio-
bium or chromium is another appropriate measure
to reduce ACCI, as shown in material tests per-
(a)
formed outside the reactor.10
Polar site
Equatorial site
3.18.3.4 Irradiation Behavior
Chain atom - B
Chain atom - C The behavior of B4C as an absorber material in a fast
reactor is well understood.
Under irradiation, B4C pellets are fractured early
in the cycle. The main cause is the difference in
temperature between the surface and the center of
the B4C pellets.
The size of the fractured pieces depends on the
burnup. Figure 9 shows some fractured pieces of
B4C.9 For low burnup and low irradiation tempera-
ture (30 1026 cap m3 at 650 C), B4C pieces were
about 5–7 mm in size (Figure 9(a)) and some pellets
were not broken. A part of the pellets broke into
powder when irradiated to around 130 1026 cap m3
(b)
at 1140 C, as shown in Figure 9(b).
Figure 6 (a) Crystal structure of B4C. Reproduced
from Thevenot, F. J. Eur. Ceram. Soc. 1990, 6, 205–225.
(b) Crystal structure of B4C. Reproduced from He, J. L.; 3.18.3.4.1 Swelling
Shen, Z. Q.; Wu, E.; et al. J. Alloys Compd. 2007, 437, The dependence of swelling on burnup of B4C pellets
238–246. is shown in Figure 10.9
Swelling of B4C pellets tends to linearly increase
with burnup. Up to about 80 1026 cap m3, the aver-
By adding sodium to the B4C–cladding system, age swelling rate was 3.5% per 100 1026 cap m3
chemical interaction is considerably enhanced. below 1000 C. The swelling rate for irradiation
During heat treatment, the B4C pellets experienced temperatures above 1000 C was about 2.5% per
no damage, but small diameter and weight losses 100 1026 cap m3.
Fast Spectrum Control Rod Materials 525
30
Burnup (1026 cap m–3)
4
Unirradiated 14
Thermal conductivity (M m–2 K–1)
36
35
20
10
0
0 200 400 600 800 1000
Measurement temperature (⬚C)
Figure 7 Temperature dependence of thermal conductivity. Reproduced from Maruyama, T.; Onose, S.; Kaito, T.;
Horiuchi, H. J. Nucl. Sci. Technol. 1997, 34(10), 1006–1014.
To study the temperature dependence of the (17-9) in Table 6,11 during annealing. Figure 1611
swelling rate, the swelling rate of B4C pellets can correlates the helium release rate and the thermal
be plotted against the irradiation temperature. expansion curve of B4C pellets. The first peak in the
This correlation between temperature and swelling helium release correlates with a sharp increase in
has to be known accurately. Figure 119 shows that the thermal expansion. It was found that for the
swelling rate linearly decreases with increasing B4C pellet, which exhibits a second helium release
temperature. peak, the second peak does not correlate with ther-
When B4C was irradiated at around 600 C, the mal expansion, as the thermal expansion curve does
swelling rate was around 5% per 100 1026 cap m3, not have a second peak around second helium
but when it was irradiated at around 1000–1200 C, the release peak.
swelling rate was about 2.5% per 100 1026 cap m3. Figure 1711 presents the thermal expansion
Since the primary contributing factor for swelling results of irradiated and unirradiated B4C. Up to
is considered to be helium accumulation in the 400 C, the thermal expansion of the irradiated
pellets,9–11 the relation between swelling and helium specimen is larger than that of unirradiated speci-
retention in B4C pellets has been studied. If the mens. When these specimens were annealed above
B4C pellet contains a large amount of helium, it swells the temperature of the plateau region and cooled to
to a great extent as shown in Figure 12.9 This result room temperature, the thermal expansion curve
supports the belief that the main factor driving became almost the same as that of the unirradiated
B4C pellet swelling is the accumulation of 10 helium material.
atoms produced by the 10B(n,a)6Li reaction.
The dependence of helium retention on burnup 3.18.3.4.2 Lattice parameters
in B4C was calculated from the helium release Figures 18 and 1911 present the temperature depen-
is shown in Figure 13.9 Generally, the amount of dence of the lattice parameters and unit cell volume.
helium retained in the B4C linearly increases with The lattice parameter c increases dramatically from
increasing burnup. Results from another report on 500 to 900 C (Figure 18), and then suddenly
the release of helium at low temperature are shown decreases to the original value of the unirradiated
in Figure 14.10 specimen. The unit cell volume shows a slight
Figure 1511 shows the helium release rate of decrease around 500 C and has a peak around
two B4C pellets, identified as samples (3-2) and 750 C (Figure 19).
526 Fast Spectrum Control Rod Materials
200 mm 50 mm 30 mm
(a)
(⬚C)
500 600 700 800 900
10-4
1.4970/Na/B4C
1.4970/Na-Li/B4C
Growth of interaction zone (cm s-1/2)
1.4970/Na/Na/B4C
1.4981/Na/B4C
1.4981/Cr/Na/B4C
10-5
k = 0.522exp( 97200 )
RT
10-6
10-7
13 12 11 10 9 8
(b) 104
T (K-1)
Figure 8 (a) Metallographic observation after heat treatment of stainless steel cladding with B4C pellets and sodium
at 600 C for 8640 h. Reproduced from Dunner, P. H.; Heuvel, H.-J.; Horle, M. J. Nucl. Mater. 1984, 124, 185–194.
(b) Temperature dependence of the interaction layer on stainless steel with B4C pellets and sodium. Reproduced from
Dunner, P. H.; Heuvel, H.-J.; Horle, M. J. Nucl. Mater. 1984, 124, 185–194.
1 cm
(a)
1 cm
(b)
intergranular and transgranular porosity. The broke into small pieces, which is a behavior asso-
average grain size is about 20 mm. The micro- ciated with microcracks.
structure of an irradiated specimen is seen in At low burnup, observation of the B4C micro-
Figure 21(b).11 There are many irregular-shaped structure by transmission electron microscopy
microcracks caused by irradiation. Some cracks (TEM) indicates that the dominant defect structures
were along the grain boundaries and some formed were small, high-pressure, plate-like bubbles with len-
transgranularly. ticular strain fields.14 These defects were also observed
Figure 22(a)11 shows a polished surface of a in B4C at high burnup. The observed strain fields were
B4C pellet that was irradiated and then annealed for quite complex because of neighboring bubbles. Some
1 h at 1200 C. The cracks along the grain boundaries of these bubbles contained helium gas.13,14
grew to interconnect with one another during the After irradiation and annealing above 0.6Tm, there
annealing. Figure 22(b)11 shows the etched surface were small irregular cavities with strong lattice
of the material in Figure 22(a). Some of the grains strains and denuded zones at the grain boundary.
528 Fast Spectrum Control Rod Materials
5.0
4.0
3.5 [(Δ D/D%) / (100 ´ 1026 cap m–3)]
Pellet swelling (Δ D/D%)
3.0
2.0
Irradiation
temperature (⬚C)
<600
1.0
600–800
800–1000
1000–1200
0.0
0 50 100 150 200
26 –3
Burnup (10 cap m )
Figure 10 Dependence of pellet swelling on burnup. D is the pellet diameter. Reproduced from Maruyama, T.; Onose, S.;
Kaito, T.; Horiuchi, H. J. Nucl. Sci. Technol. 1997, 34(10), 1006–1014.
8.0
Burnup
(1026 cap m–3)
Swelling rate (DD/D%/100 ⫻ 1026 cap m–3)
£50
6.0 50–100
³100
4.0
2.0
0.0
600 800 1000 1200
Irradiation temperature (⬚C)
Figure 11 Dependence of pellet swelling on temperature. D is the pellet diameter. Reproduced from Maruyama, T.;
Onose, S.; Kaito, T.; Horiuchi, H. J. Nucl. Sci. Technol. 1997, 34(10), 1006–1014.
Here, Tm is melting point of B4C, which is with denuding near the grain boundary, for an irradi-
about 2450 C. ation temperature around 0.3Tm.
Annealing reduces the number density of cavities, On the other hand, the growth of these cavities on
as shown in Figure 23.12 The microstructure suggests annealing at higher temperatures suggests that gas
that radiation-induced voids were generated, along bubbles were present.15–17
Fast Spectrum Control Rod Materials 529
5.0
Burnup
(1026 cap m–3)
4.0 £ 50
Pellet swelling (DD/D%) 50–100
³ 100
3.0
2.0
1.0
0.0
0 100 200 300 400 500
Helium retention (m3 STP per m3 B4C)
Figure 12 Relation between helium release and pellet swelling. Reproduced from Maruyama, T.; Onose, S.; Kaito, T.;
Horiuchi, H. J. Nucl. Sci. Technol. 1997, 34(10), 1006–1014.
300
Helium retention (m3 STP per m3 B4C)
250
Full release
200
150
100 Irradiation
temperature (⬚C)
<600
50 600–800
800–1000
1000–1200
0
0 50 100 150 200
Burnup (1026 cap m–3)
Figure 13 Dependence of helium release on burnup. Reproduced from Maruyama, T.; Onose, S.; Kaito, T.; Horiuchi, H.
J. Nucl. Sci. Technol. 1997, 34(10), 1006–1014.
60
50
I
40
30
20 II
10
0
0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0 9.0
Burnup (×1026 cap m–3)
Curve I : Integral envelope
Curve II : Envelope for T < 600 ⬚C
Figure 14 Integrated volume of helium released from B4C. Reproduced from Dunner, P. H.; Heuvel, H.-J.; Horle, M.
J. Nucl. Mater. 1984, 124, 185–194.
Sample 10-2 10-6 10-8 3-2 17-9 Europium (Eu) has a high fast spectrum absorption
cross section. Natural Eu consists of 47.8% 151Eu and
10
B burnup (%) 1.0 1.2 1.8 1.0 3.0 52.2% 153Eu. 151Eu has a large cross section of about
1020(n,a) (cm3) 3.4 4.1 6.1 3.4 10.2
5300 and 2800 barns for generating 152Eu (half-life;
Reproduced from Suzuki, H.; Maruyama, T.; Wakasa, T. J. Nucl. 13.3 years) and 152mEu (half-life; 9.3 h), respectively.
Sci. Technol. 1979, 16(8), 588–595. As the result of the disintegration of Eu, the daughter
Fast Spectrum Control Rod Materials 531
3 3 Specimen (17-9) 3
Specimen (3-2)
Gas release
1 1 1
2
Thermal expansion
0 0 0 0
0 200 400 600 800 1000 1200 0 200 400 600 800 1000 1200
(a) Temperature (ºC) (b) Temperature (ºC)
Figure 16 Dependence of helium release rate and thermal expansion of B4C pellets on temperature. Reproduced from
Suzuki, H.; Maruyama, T.; Wakasa, T. J. Nucl. Sci. Technol. 1979, 16(8), 588–595.
products also have a large cross section. But Eu has B4C produces helium through the 10B(n,a)7Li
some disadvantages: it is a rare-earth element, reaction, which causes an increase in the gas
its thermal conductivity is low, and it has a high pressure in the absorber pins. The retained
level of radioactivity from its decay chain. helium in a pellet causes pellet swelling, which
532 Fast Spectrum Control Rod Materials
12.12
5.64
12.10
Lattice parameter a (Å) 5.62
12.06
5.58
12.04
5.56 … a-axis
… c-axis
12.02
5.54
12.00
5.52
0 500 1000
Temperature (⬚C)
The symbols O and D denote lattice
parameters of unirradiated specimen
Figure 18 Temperature dependence of lattice parameters. Reproduced from Suzuki, H.; Maruyama, T.; Wakasa, T. J. Nucl.
Sci. Technol. 1979, 16(8), 588–595.
0.4
Volume change (%)
0.2
0
0 500 1000
Temperature (⬚C)
Figure 19 Temperature dependence of unit cell volume. Reproduced from Suzuki, H.; Maruyama, T.; Wakasa, T. J. Nucl.
Sci. Technol. 1979, 16(8), 588–595.
in the fast reactor JOYO. In the case of a vented-type 3.18.5.2 B4C Pellet–Cladding Tube
pin, helium gas is released through the hole in the Interaction
cladding tube; there are no limits due to built-up
Currently, the lifetime of control rods in JOYO is
inner pressure.
limited by ACMI due to swelling of B4C pellets,
which is accelerated by relocation of pellet fragments.
Pins with a wider gap between the B4C pellet and
cladding tend to break at lower burnup. In the
cracked pins, extensive ACMI due to relocation of
the B4C pellet takes place, together with accumula-
tion of helium in the cladding material to cause
helium embrittlement. It is inferred that these phe-
nomena were caused by pin rupture when they were
in the reactor core.
In this case, the crack growth rate is extremely
slow and the ACMI-induced cracking in the absorber
elements does not influence either the reactor or
1.0 mm plant operations. It is on this basis that a strict condi-
(a) 34 ⫻ 1026 cap m–3, 670 ⬚C
tion to avoid cladding cracking is not necessary.
Accordingly, it is suggested that a realistic design
standard should consider the ACMI phenomena
and that the burnup limit should be based on the
nominal base calculation for average plastic strain
using depressed ductility data.
3.18.5.3 Outlook
A sodium-bonded type control rod has been devel-
oped that improves the thermal conductivity by
means of charging sodium into the gap between the
B4C and cladding by utilizing a shroud that wraps the
1.0 mm
pellet fragments in a thin tube. This new design will
(b) 126 ⫻ 1026 cap m–3, 1120 ⬚C be able to enlarge the gap between the B4C and
Figure 20 SEM images of irradiated B4C pellets. cladding without heating the B4C or causing frag-
Reproduced from Maruyama, T.; Onose, S.; Kaito, T.; ment relocation, thereby extending the lifetime of the
Horiuchi, H. J. Nucl. Sci. Technol. 1997, 34(10), 1006–1014. control rods.
20 mm 20 mm
Figure 21 Observation results of unirradiated and irradiated B4C pellets (optical micrographs). Reproduced from Suzuki, H.;
Maruyama, T.; Wakasa, T. J. Nucl. Sci. Technol. 1979, 16(8), 588–595.
534 Fast Spectrum Control Rod Materials
20 mm 20 mm
40
Measurement temperature: 25 ⬚C
20
10
0
0 10 20 30 40
a
b Burnup (1026 cap m–3)
535
536 Oxide Fuel Performance Modeling and Simulations
FDM Finite difference method used for R&D purposes, for the design of fuel rods,
FEM Finite element method new products, or modified fuel cycles, and for sup-
FGR Fission gas release porting loading of fuel into a power reactor, that is,
HBS High-burnup structure to verify compliance with safety criteria in safety case
HCP Hexagonal closed packed submissions. A list of commonly used fuel perfor-
HM Heavy Metals mance codes is provided in Table 1.
HZP Hot zero power
IAEA International Atomic Energy Agency 3.19.1.2 Geometrical Idealization and
IFA Instrumented fuel assembly Size of the Problem
LEFM Linear elastic fracture mechanics
LHR Linear heating rate In principle, our spatial problem is three-dimensional
LOCA Loss-of-coolant accident (3D). However, the geometry of a cylindrical fuel rod (a
LWR Light water reactor very long, very thin rod) suggests that any section of a
MC Monte Carlo fuel rod may be considered as part of an infinite body:
MD Molecular dynamics that is, neglecting axial variations. By further assuming
MIMAS Micronized master blend axially symmetric conditions because of the cylindrical
MOX Mixed oxide geometry, the original 3D problem is reduced to a 1D
NEA Nuclear Energy Agency one. Analyzing the fuel rod at several axial sections with
O/M Oxygen-to-metal ratio a (radially) 1D description is sometimes referred to as
OECD Organisation for Economic Cooperation quasi-2D or-11=2D. Most fuel rod performance codes
and Development fall into this category. Real 2D codes such as, for
PAS Positron annihilation spectroscopy instance, the FALCON code,1 offer the possibility to
PCI Pellet–cladding interaction analyze r–z problems (no azimuthal variation) and r–’
PCMI Pellet–cladding mechanical interaction problems (no variation in axial direction). An example of
PWR Pressurized water reactor a 3D code is TOUTATIS2 and DRACCAR3 and is dealt
RIA Reactivity-initiated accident with in Chapter 3.22, Modeling of Pellet–Cladding
S/V Surface-to-volume ratio Interaction. DRACCAR is addressed later in Section
SANS Small-angle neutron scattering 3.19.3.1.2. Generally, 2D or 3D codes are used for the
SCC Stress corrosion cracking analysis of local effects, whereas the other codes have
SM Shell model the capability to analyze the whole fuel rod during a
TAD Temperature-accelerated dynamics complicated, long power history.
TEM Transmission electron microscopy In order to estimate the ‘size’ of the problem at
hand, the number of time steps must also be specified.
For a normal irradiation under base load operation, that
is, under no-load follow operation, 100–500 time steps
3.19.1 Introduction are sufficient. However, for an irradiation in a research
reactor, such as the heavy-water boiling water reactor
3.19.1.1 Importance of Fuel Performance
(BWR) of the Organisation for Economic Cooperation
Modeling
and Development (OECD), Halden, many more varia-
In order to ensure the safe and economical operation tions of the linear rating with time are recorded. In such
of fuel rods, it is necessary to be able to predict their a situation, one must either simplify the complicated
behavior and lifetime. The accurate description of power history or increase the number of time steps to
the fuel rod’s behavior, however, involves various the order of several thousands. The simplest geometrical
disciplines ranging from the chemistry, nuclear and idealization needs 20 radial and 20 axial nodes; a 2D
solid-state physics, metallurgy, ceramics, and applied representation of a single pellet would approximately
mechanics. The strong interrelationship between need several hundred nodes. Therefore, local models,
these disciplines, as well as the nonlinearity of many which are in almost all cases nonlinear, must be very
processes involved, calls for the development of com- carefully constructed, since even for the simplest geo-
puter codes describing the general fuel behavior. Fuel metrical idealization the number of calls may easily
designers and safety authorities rely heavily on these reach the order of millions:
types of codes since they involve minimal costs in
15 radial 15 axial nodes 5000 time steps 3 iterations
comparison with the costs of an experiment or an
unexpected fuel rod failure. The codes are being ¼ 3:4 106 calls
Oxide Fuel Performance Modeling and Simulations 537
COMETHE Hoppe, N.; Billaux, M.; van Vliet, J.; Shihab, S. COMETHE version 4D release 021 (4.4-021), Vol. 1,
general description; Belgonucleaire Report, BN-9409844/220 A; Apr 1995
COPERNICa Bonnaud, E.; Bernard, C.; Van Schel, E. Trans. Am. Nucl. Soc. 1997, 77
ENIGMA Kilgour, W. J.; Turnbull, J. A.; White, R. J.; Bull, A. J.; Jackson, P. A.; Palmer, I. D. Capabilities and
validation of the ENIGMA fuel performance code. In Proceedings of the ENS Meeting on LWR Fuel
Performance, Avignon, France, 1992
FALCON, FREY Rashid, J.; Montgomery, R.; Yagnik, S.; Yang, R. Behavioral modeling of LWR fuel as represented in the
FALCON code. In Proceedings of the Workshop on Materials Modelling and Simulations for Nuclear
Fuel, New Orleans, LA, Nov 2003
FEMAXI Suzuki, M.; Saitou, H. Light Water Reactor Fuel Analysis Code FEMAXI-6 (Ver. 1); JAEA-Data/Code
2005-003, Feb 2006
FRAPCON Berna, G. A; Beyer, C. E.; Davis, K. L.; Lanning, D. D. FRAPCON-3: A computer code for the calculation of
steady-state, thermal-mechanical behaviour of oxide fuel rods for high burnup; NUREG/CR-6534,
PNNL-11513; Dec 1997
METEORa Struzik, C.; Moyen, M.; Piron, J. High burnup modelling of UO2 and MOX fuel with METEOR/
TRANSURANUS Version 1.5. In Proceedings of the International Topical Meeting on LWR Fuel
Performance, Portland, OR, Mar 1997
PIN-micro Pazdera, F.; Strijov, P.; Valach, M.; et al. User’s guides for the computer code PIN-micro; UJV 9512-T,
Rez; Nov 1991
START Bibilashvili, Y. K.; Medvedev, A. V.; Khostov, G. A.; Bogatyr, S. M.; Korystine, L. V. Development of
the fission gas behaviour model in the START-3 code and its experimental support. In Proceedings of the
International Seminar on Fission Gas Behaviour in Water Reactor Fuels, Cadarache, France, Sept 2000
TRANSURANUS Lassmann, K. The TRANSURANUS code – past, present and future; Review article, ITU Activity Report
2001 – EUR 20252, ISBN 92-894-3639-5; 2001
a
Based on TRANSURANUS.
of such problems, 2D or even 3D models are 3.19.2.1.1 Axial heat transfer in the coolant
indispensable. In general, three regimes must be covered in a light
One of the most important consequences of all water reactor (LWR):
uncertainties is that one must implement models of
1. The subcooled regime, where only surface boiling
‘adequate’ complexity.
occurs. This regime is typical for pressurized
water reactors (PWRs) under normal operating
conditions.
3.19.2 Basic Equations and 2. The saturated, two-phase regime. This regime
State of the Art is typical for BWRs under normal operating
3.19.2.1 Heat Transfer conditions.
3. The saturated or overheated regime. This regime
The objective of this section is to describe how the may be reached in all off-normal situations. A typical
temperature distribution in a nuclear fuel rod is calcu- example is a LOCA.
lated in a fuel rod performance code. The scope is
limited to a description of the important physical The fuel rod performance codes use 1D (axial) fluid
phenomena, along with the basic equations and the dynamic equations that can only cope with the first
main assumptions. Detailed numerical aspects as two regimes. For simulating the third type of regime,
well as mathematical derivations are provided in the whole reactor coolant system needs to be ana-
some reference works.4–6 lyzed by means of thermohydraulic system codes
The temperature distribution in a fuel rod is such as RELAP, TRACE, or ATHLET in order to
of primary importance for several reasons. First provide adequate boundary conditions to the fuel rod
of all, the commercial oxide fuels have poor performance code.
thermal conductivities, resulting in high tempera- The temperature calculation in the coolant serves
tures even at modest power ratings. Second, the two purposes. First of all, the axial coolant tempera-
codes are used for safety cases where one has ture in the basic (fictional) channel provides the
to show that no fuel melting will occur, or that (Dirichlet) boundary condition for the radial tem-
the internal pressure in the rod will remain below perature distribution in the fuel rod. It results from
a certain limit. Finally, many other properties the combined solution of the mass, momentum, and
and mechanisms are exponentially dependent on energy balance equations. The simplified equation
temperature. used in fuel performance codes reads
The most important quantity is of course the local @T @T 00 2prcl;o
power density q 000, which is the produced energy per cr þ crw ¼ qcl;c þ qc000 ½2
@t @z A
unit volume and time. It is usually assumed that q 000
depends only on the radius and on time. The linear where c represents the heat capacity, r the density,
00
rating is then simply given by w the velocity, T the temperature, qcl;c the heat
ð rcl;o flux from the cladding to the coolant, A the channel
q0 ¼ q 000 ðr Þ2pr dr cross-sectional area, rcl,o the cladding outer radius,
rf ;i and qc000 the power density in the coolant. In general,
ð rf ;o ð rcl;o 00
the heat flux from cladding to coolant qcl;c should be
000
¼ qf f ðr Þ2pr dr þ q 000
cl 2pr dr ½1 computed by means of a thermohydraulic code.
rf ;i rcl;i
Mathematically, the boundary condition is of the
where rf ;i /rcl;i is the inner fuel/cladding radius, convective type:
rf ;o /rcl;o is the outer fuel/cladding radius, q000
f and
@T ðr ; t Þ
q000 are the average power density in the fuel and 00
qcl;c ¼ l ¼ afT ðr ¼ rcl;o ÞTc g
@r rcl;o
cl
cladding, respectively, and f ðr Þ is a radial distribu-
tion (form) function (see below). Generally, the where a is the heat transfer coefficient between the
linear rating is a prescribed quantity and is a cladding and the coolant and Tc ¼ Tc ðz; t Þ is the
function of the axial coordinate z and the time t. (bulk) coolant temperature. Only for a steady-state
For some phenomena (e.g., cladding creep), the fast condition
neutron flux is also needed. It can be prescribed as
dT q 0
well but may also be calculated from the local crw ¼ þ qc000 ½3
power density q 000. dz A
Oxide Fuel Performance Modeling and Simulations 539
the heat flux from the cladding to the coolant is where q 00 is the heat flux in watts per unit area and
known and is given by hgap is the heat transfer coefficient between the fuel
q0 and the cladding (gap conductance). Heat transfer by
00
qcl;c ¼ convection can be neglected. In general, the heat
2prcl;o
transfer coefficient hgap depends on
Under normal operational conditions, the mass flow
1. gap width or contact pressure between fuel and
rate m_ ¼ Arw, and the coolant inlet temperature and
cladding
pressure are prescribed. In an off-normal or acciden-
2. gas pressure and composition
tal situation, the normal operational condition is the
3. Surface characteristics of cladding and fuel
initial condition, but the boundary conditions must
be provided by the thermohydraulic system codes. In fact, there are three parallel conduction routes:
The second objective of the heat flow calculation
hgap ¼ hrad þ hcon þ hgas
in the coolant is the derivation of the radial tem-
perature drop between the coolant and the cladding The contribution of the radiative component is
Tcl Tc, resulting from convection: given by
4
qc000 Cs Tf Tcl4
q00 ¼ afilm ðTcl Tc Þ ¼ hrad ¼
2prcl;o 1=ef þ 1=ecl 1 Tf Tcl
The heat transfer coefficient in the film depends where Cs is the Stefan–Boltzmann constant, e is the
on the type of convection (forced or natural) and emissivity, and T is the temperature. The radiative
the type of coolant (gas, liquid, liquid metal). In the component is <1% during normal operating condi-
subcooled regime of a PWR, the Dittus–Boelter cor- tions because of the limited difference between the
relation is largely applied, whereas in the saturated two surface temperatures.
regime of a BWR, the Jens–Lottes correlation is The component hcon reproduces the improvement
applied (see separate lecture on thermohydraulics). in heat transfer due to contact pressure:
3.19.2.1.2 Heat transport through the P b
hcon ¼ ald 2
cladding dH
The heat transport through the cladding occurs
where l and d are the mean values of the thermal
through conduction:
conductivity and the arithmetic mean roughness,
1@ @T respectively, P is the contact pressure, H is the
r lc þ qcl000 ¼ 0
r @r @r Meyer hardness of the softer material, and a and b
where lc is the cladding conductivity (20 W mK1 for are model parameters.
Zircaloy), and the heat generation in the cladding is The heat transfer through conduction in the gas is
generally neglected (the g-heating as well as the exo- often based on the model of Ross and Stoute9:
thermic clad oxidation process are generally disre- lgas
hgas ¼
garded). In order to account for the presence of an d þ s þ gf þ gcl
outside oxide layer with a thermal conductivity on the
where the thermal conductivity of a multicomponent
order of 2 W mK1 for ZrO2 (thickness <100 mm), the
gas is only composition-dependent and calculated by
total equivalent cladding conductivity can be obtained
means of
by applying the formula for serial thermal resistances. 2 3
In a similar manner, the appearance of crud on the
6 7
outer cladding surface is sometimes accounted for 6 7
6 7
through an additional heat transfer coefficient.
X n 66 7
l j 7
lgas ¼ 6 0 1 7
3.19.2.1.3 Heat transport from cladding to 6 7
j ¼1 6 7
6B 7
6 B 1 þ P w ck C
the fuel pellet n
C 7
The temperature difference in the pellet–cladding 4@ jk cj A5
k¼1
gap, DTgap , is calculated as follows7,8: j 6¼k
Fuel Cladding
1000
Data
He-filled rod
500
0
0 2.5 5.0 7.5
Radial position (mm)
Figure 3 Radial temperature distribution in a boiling water reactor rod at the beginning of life. Comparison between the
range of experimental results and predictions of the TRANSURANUS code for two different fill gases (He, Xe). The data refers
to a thermocouple measurement in the central hole of the fuel pellet, indicated by the dashed line. From van Uffelen, P.;
Konings, R. J. M.; Vitanza, C.; Tulenko, J. In: Handbook of Nuclear Engineering; Cacuci, D.G., Ed., Springer: Germany, 2010;
pp 1520–1627, Chapter 13, with permission from Springer ScienceþBusiness Media B.V.
542 Oxide Fuel Performance Modeling and Simulations
4. Elastic constants are often isotropic and constant 3.19.2.2.2 Calculation of strains
within a cylindrical ring, but can be anisotropic in 3.19.2.2.2.1 Elastic strain
some codes (e.g., Suzuki and Saitou14). The elastic strains for an isotropic material are
5. The total strain can be written as the sum of elastic reversible and given by
and nonelastic components. 1
eelastic ¼ ½sr nðst þ sa Þ
The first two assumptions reduce the problem to one r
E
dimension. The third assumption indicates that the 1
eelastic ¼ ½st nðsr þ sa Þ
stresses are related through a local equilibrium con-
t
E
dition for the radial force in the following form: 1
eelastic ¼ ½sa nðsr þ st Þ
a
E
dsr st sr
¼
dR R 3.19.2.2.2.2 Nonelastic strain
where sr and st represent the normal radial and 3.19.2.2.2.2.1 Thermal strain The nonelastic
tangential stresses, respectively, and R corresponds strains consist of various contributions. First of all,
to the radius of the deformed geometry. there is the thermal strain resulting from temperature
Since the fuel stack and cladding are treated as a differences, which is assumed to be isotropic and
continuous, uncracked medium, no discontinuities reversible:
are allowed in their displacements. This is translated
by the compatibility relations for the strains: eti ¼ aðT T0 Þ i 2 fr; t; ag
0.01
0.009 UO2 Zircaloy
0.008
0.007
Elongation (–)
0.006
0.005
0.004
0.003
0.002
0.001
0
300 500 700 900 1100
Temperature (⬚C)
Figure 4 Elongation of UO2 fuel and Zircaloy cladding due to thermal expansion () as a function of the temperature ( C).
Oxide Fuel Performance Modeling and Simulations 543
1
seff ¼ pffiffiffi½ðsr st Þ2 þ ðsr sa Þ2 þ ðst sa Þ2 1=2 3.19.2.2.2.2.4 Pellet cracking The fourth and last
2 nonelastic strain component stems from the pellet
so that yielding only occurs when the effective or cracking. Pellet cracking already occurs at startup
equivalent stress (seff) exceeds the yield stress deter- with fresh fuels due to the differential thermal expan-
mined from a uniaxial tensile test. Others have intro- sion since the hot pellet center expands more than
duced the anisotropic factors according to Hill’s the cold periphery. In order to assess the linear heat
methodology.6 Finally, a multidimensional yield sur- generating rate at which cracking in cylindrical pellet
face20,21 has also been proposed. In order to account occurs, the maximum thermal stress (¼st,max ¼ sa,max
for work-hardening, one generally assumes that the at pellet periphery) in an uncracked pellet submitted
yield stress changes with the total permanent defor- to a parabolic temperature gradient
mation. The plastic strain is therefore computed aEq 0
incrementally. st;max ¼
8pð1 uÞl
In order to answer the second question, each
increment of effective plastic strain is related must be compared with the (uniaxial) fracture stress,
to the individual plastic strain components by a which is 130 MPa. When using E ¼ 200 GPa,
flow rule: u ¼ 0.31, the thermal diffusivity a ¼ 105 K1, and
an average thermal conductivity of l ¼ 3 W mK1,
@seff radial cracks are predicted to be initiated in the pellet
Dei ¼ Deeff i 2 fr; t; ag
@si periphery at a linear heat rate q0 of the order of
5 kW m1. The number of cracks (Ncr) is dependent
When using the above-mentioned definition of the
on the linear heat rate. Oguma23 proposed a linear
equivalent stress, one obtains the Prandtl–Reuss
model for the number of radial cracks, which is illu-
flow rule22:
strated in Figure 6. In addition to radial cracks, also
p
p 3Deeff axial and (especially under ramping conditions) cir-
Dei ¼ Si i 2 fr; t; ag cumferential cracks are formed.
2seff
The consequences of cracking are very impor-
indicating that the plastic strain increment is tant in fuel performance modeling. Owing to the
proportional to the deviatoric stress Si ¼ si sh, larger thermal expansion of the fuel fragments in
where sh ¼ (sr þ st þ sa)/3 is the hydrostatic stress. comparison with that of a monolithic cylinder, and
For the time-dependent creep, one needs strain to vibration-induced motion, they move outward.
rate equations although the total creep strain is also This is called pellet ‘relocation’ and has a strong
computed incrementally by multiplying the strain impact on the thermal behavior, as shown in Figure 7.
rate with the time step length. For primary creep, It reduces the pellet–cladding gap size, thereby
typically an empirical expression is applied: reducing the temperature levels in the fuel at BOL.
This constitutes the largest contribution to the gap
e_ eff ¼ K sneff t m
closure (30–50%, depending on q0 ) but is also the
where K, n, m are constants. one which is subject to the largest uncertainty
For the secondary or steady-state creep, there are because of the stochastic nature of the cracking
three parallel processes. The vacancy diffusion or
Nabarro–Herring creep and the dislocation climb
dominate at high temperature and high stresses, Crack pattern
respectively:
Number of fragments
Bq 000 seff Ed
ð_eeff
c
Þ¼ exp vacancy diffusion
R2gr kT 16
0
E 8
ð_eeff
c
Þ ¼ B 0 q 000 seff
4:5
exp d dislocation climb
kT 0
0 10 20 30 40
The third process is irradiation-induced creep, dom- Rod power (kW m–1)
inating at low temperatures and assumed to be pro-
Figure 6 Calculated crack pattern from thermoelastic
portional to the effective stress and the local fission
stress. Reproduced from Oguma, M. Nucl. Eng. Des. 1983,
rate density or q 000. 76, 35–45.
Oxide Fuel Performance Modeling and Simulations 545
process. This also raises questions about the useful- When the pellets swell large enough so that they
ness of applying 3D stress calculations. come into contact with the cladding, which creeps
The contribution from relocation is generally down under influence of the pressure difference
accounted for in the tangential strain component as a between the coolant pressure and the fill gas pressure,
(linear) function of the linear heat rate: et ¼ u/r, where then relocation may be (partly) reversed. If both
u ¼ sdg, s being the initial radial gap size and dg the pellet swelling and cladding creep-down continue,
fraction of the gap size closing as a result of relocation. the gap is closed and the pellet fragments are pushed
An example based on the relocation model in the inward, so that the relocation approaches total
FRAPCON3 code24 is illustrated in Figure 8.25 elimination.
Fuel Cladding
Without relocation
1500 Data Xe-filled rod
Temperature (⬚C)
With relocation
1000
Data
He-filled rod
500
0
0 2.5 5.0 7.5
Radial position (mm)
Figure 7 Radial temperature distribution in a boiling water reactor rod (instrumented fuel assembly-505) at beginning of
life calculated by the TRANSURANUS code. Comparison between the analysis of a Xe-filled rod with (full line) and without
(dashed line) taking relocation into account. van Uffelen, P.; Konings, R. J. M.; Vitanza, C.; Tulenko, J. In: Handbook of Nuclear
Engineering; Cacuci, D.G., Ed., Springer: Germany, 2010; pp 1520–1627, Chapter 13, with permission from Springer
ScienceþBusiness Media B.V.
FRAPCON-3 modified
≥40 kW m–1
50
35 kW m–1
30 kW m–1
Relocation (%)
25 kW m–1
40
≤20 kW m–1
30
0 10 20 30 40 50
Slice averaged burnup (MWd kg–1 HM)
Figure 8 Fraction of gap closure due to pellet fragment relocation (dg), derived from the relocation model in the FRAPCON3
code. van Uffelen, P.; Konings, R. J. M.; Vitanza, C.; Tulenko, J. In: Handbook of Nuclear Engineering; Cacuci, D.G., Ed.,
Springer: Germany, 2010; pp 1520–1627, Chapter 13, with permission from Springer ScienceþBusiness Media B.V.
546 Oxide Fuel Performance Modeling and Simulations
The effect of relocation on the mechanical behav- 3.19.2.2.3.1 Radial boundary conditions
ior is also of primary importance since it reduces the In general, continuity of the radial stress and dis-
overall thermal stress in the pellets and may even placement at each radial zone is imposed and the
change the sign of the stress in pellet centers from radial stress at the outer cladding surface is deter-
compression (in a cylinder) to traction (in frag- mined by the coolant pressure: sr(rcl,o) ¼ pcool.
ments).26 The exact location and size of every crack The boundary condition in the rod depends on
would be required to accounting for the effects of the the configuration. When pellet–cladding mechanical
cracks exactly and to solve a 3D stress–strain problem interaction (PCMI) is not established, the radial stress
in each block. Instead, one simply modifies either the at the pellet periphery is determined by the fill gas
material constants14,25 or modifies the constitutive pressure in the fuel rod (pgas): sr(rf,o) ¼ pgas. For the
equations. An example of the former approach is boundary condition in the pellet center, two possibi-
that of Jankus and Weeks (Figure 9),27 where a lities exist. In hollow pellets, the radial stress at the
reduction of the elastic constants is proposed: pellet center is equal to fill gas pressure: sr(rf,i) ¼ pgas,
Ncr whereas in the case of full cylindrical pellets, the radial
2
E0 ¼ E and tangential stresses are equal at the pellet center.
3 When the fuel and cladding are in contact, a fuel
Ncr
0 1 pellet interfacial pressure exists (pfc), which deter-
n ¼ n
2 mines the boundary condition at the pellet surface:
sr(rf,o) ¼ sr(rcl,i) ¼ pfc. The other radial boundary
which means that an equivalent continuous and
conditions remain unchanged.
homogeneous solid body with directionally dependent
(anisotropic) properties is considered. As the pellet–
3.19.2.2.3.2 Axial boundary conditions
clad gap closes during irradiation, the contact pressure
The plane strain assumption entails that the axial strain
can press the fragments inward, thereby reducing the
is constant in the plane perpendicular to the axis. The
relocated radius to a minimum value. Some codes
axial strain is therefore determined by an axial force
also account for the restoration of the elastic con-
balance equation including the fill gas pressure, the
stants as the relocation is reversed (partially).14
plenum spring pressure, the fuel column weight, the
In order to modify the constitutive equations, a
friction forces, and the coolant pressure imposed on
plane stress condition has been proposed20: that is,
both end plugs of the cladding. The friction forces
the tangential stress is set equal to the fill gas pressure
depend on the fuel–cladding interaction and can only
once the radial crack appears. Both types of ap-
be taken into account iteratively. Indeed, when one of
proaches, however, do not account for crack healing.
the axial sections i is analyzed, it is not known whether
3.19.2.2.3 Boundary conditions the frictional forces between fuel and cladding origi-
In order to solve the main equations summarized in nating from a section above/below i need to be consid-
Section 3.19.2.2.1, boundary conditions are required. ered in the axial balance of forces. This is schematically
1.2
E⬘/E
1
Elastic constant ratio
V⬘/V
0.8
0.6
0.4
0.2
0
0 2 4 6 8 10 12
Number of cracks
Figure 9 Effect of the number of cracks on the elastic modulus and the Poission’s ratio according to Jankus and Weeks.27
Oxide Fuel Performance Modeling and Simulations 547
No radial contact
the important analytical targets in predicting fuel
‘slip’ behavior, since it can be a cause of fuel failure.
Cladding The mechanical interaction of the fuel with the
Case cladding is mainly caused by the different thermal
Fuel 1
expansion rates of the two components (see Section
3.19.2.2.2). However, the poor thermal conductivity
of the fuel results in a strong temperature gradient
across the pellet. This leads to an increase of the gap
Static friction
‘no slip’
closure rate at both ends of a pellet because even
Cladding
in the early stages of irradiation, the differential
thermal expansion in the pellet causes its so-called
Fuel 2 hour-glass shape (Figure 11). The resulting internal
thermal stresses exceed the fracture stress of UO2
(around 100 MPa) causing pellet fracture. Once the
pellet has fractured, it is able to deform much more
Static friction
Trapped
‘no slip’
‘Slip’ readily under the effects of the temperature field,
causing the pellet ends to bow outward leading to
Cladding
the hour-glass shape.
Fuel 3 PCMI in the early stage of irradiation is therefore
caused by the differential thermal expansion of the
pellets, leading to a bamboo or ridging deformation
of the cladding. The ridges in the cladding, the height
Trapped Sliding friction ‘Slip’ of which can reach 20 mm at high powers, coincide
‘slip’ with the pellet ends and can cause hoop stresses of
Cladding
around 400 MPa, which is close to the clad yield
Fuel
stress. When aggressive fission products such as
4
iodine are in their vicinity, these local concentrations
of stress and strain lead to the so-called stress corro-
sion cracking (SCC) in LWR fuel systems.
Figure 10 Four possible modes of an interaction between The onset of PCMI is affected by a number of
fuel and cladding. van Uffelen, P.; Konings, R. J. M.; Vitanza, design and fabrication parameters. First of all, the
C.; Tulenko, J. In: Handbook of Nuclear Engineering; pellet geometry is adapted to extend the onset of
Cacuci, D.G., Ed., Springer: Germany, 2010; pp 1520–1627, and to mitigate PCMI. In particular, the pellet height
Chapter 13, with permission from Springer Scienceþ
Business Media B.V.
has been reduced with respect to its diameter, while
chamfers were applied for delaying PCMI and reduc-
shown in Figure 10. In the case of a radial contact ing the probability of fuel chipping. Second, a rod
between the fuel and cladding, both bodies may stick to geometry is important. More precisely, the width of
each other, but some sliding may be possible in specific the original clearance between the pellets and the
conditions (sticking or static vs. sliding friction). Part of cladding should be large enough. Nevertheless, the
the fuel rod may be ‘trapped,’ which means that rather gap size is subject to many uncertainties, the largest
high axial forces may act on cladding and fuel. being the pellet fragment relocation. In addition, it is
One advantage of 2D and 3D finite element mod- generally assumed in fuel performance modeling that
els is that such effects are automatically included in the pellets are located concentrically within the clad-
the analysis through the use of specific gap elements, ding, although this is seldom the case. Any eccentric-
as explained in more detail in Chapter 3.22, Model- ity in the stacking arrangement, resulting from
ing of Pellet–Cladding Interaction. fabrication or fuel rod handling, is likely to lead to
premature onset of PCMI even though its effect will
3.19.2.2.4 Pellet–cladding interaction diminish as gap closure occurs.
3.19.2.2.4.1 Pellet–cladding mechanical Once PCMI has started, both the pellet geometry
interaction and the material properties of the interacting compo-
As dealt with in detail in Chapter 3.22, Modeling of nents influence the maximum stresses and strains
Pellet–Cladding Interaction, PCMI is one of as well as their evolution. Flat-ended pellet stacks
548 Oxide Fuel Performance Modeling and Simulations
will generate larger axial expansion in comparison elements of the pellet and cladding are summed up
with dished stacks, especially in fresh fuel. This is independently to determine the diameter change,
due to the fact that in flat-ended pellets, the hot granting that FDM is actually modified to some extent
central part determines the maximum length change, to cope with mechanical interactions among the ring
whereas for dished pellets there is no contact between elements. There have been attempts to develop 3D
pellets along the central axis (when the power is not too fuel performance codes,29 as explained in Chapter
high) and hence the axial expansion will be controlled 3.22, Modeling of Pellet–Cladding Interaction.
by the outer (cooler) regions of the pellet. In practice,
the ratio of axial to tangential strain can vary between 3.19.2.2.4.2 Irradiation-induced SCC
0.2 for large dishes and 2 for undished pellets.28 Nev- Failures during power variations are not only attrib-
ertheless, the axial expansion from flat-ended pellets uted to stress. Stresses increase at the intersection of
can diminish with burnup because of in-pile dishing, radial crack planes, the interpellet planes, and the
for instance caused by densification and creep in the inner surface of the cladding. The brittle nature of
hot central parts of the pellets during PCMI. the failure site, however, has led to the general consen-
In addition to the difficulties to reproduce 2D sus that, although stress is the primary initiator, propa-
(local) pellet deformations during PCMI by means gation of the crack is chemically assisted, and thus the
of 1D fuel performance codes, there are other chal- process is termed SCC. The chemical assistance for
lenges to be dealt with. Apart from pellet and cladding brittle cracking in all probability arises from the envi-
eccentricity and slanting, which mostly affect the onset ronment established by the release of fission products
of PCMI, there are uncertainties related to the assess- into the fuel–clad interspace, with isotopes of iodine as
ment of the stress in cracked pellets, to pinching by the main corroding species. There is some discussion as
assembly grids, as well as to the (static and dynamic) to whether the corrodent must be freshly released from
friction coefficient between pellets and cladding. In the fuel or whether it is sufficient for it to have accu-
particular, the slipping is severely restricted by the mulated in the fuel–clad gap throughout the irradia-
interaction layer that establishes between both com- tion period prior to the overpower transient.
ponents after gap closure at higher burnups. Since its discovery in the Canada deuterium
Therefore, for these PCMI evaluations, the (local uranium (CANDU) reactor and the BWR in the
2D) finite element method (FEM) is in principle a 1970s, the incidence of pellet–cladding interaction
more advantageous method than the finite difference (PCI)-SCC failure is thought to be affected by
method (FDM). FEM puts the reaction forces exert- four factors, as reviewed by Cox30: stress, time, mate-
ing among all the ring elements of the pellet and rial, and chemical environment. Understanding the
cladding into one entire matrix to obtain a numerical important variables for PCI-SCC will help fuel
solution,14 while in FDM displacements of each ring, designers to propose solutions. These include adapt-
ing the appropriate fuel geometry (reducing the
Pellet hourglassing
height to diameter ratio, introducing dishes and
chamfers), imposing restrictions on the power varia-
tions, modifying the fuel assembly design, and apply-
ing coatings on the inner cladding surface in order to
reduce the friction coefficient and to accommodate in
a ductile manner the biting of pellet ends. More
Ridges
precisely, in BWRs, Cu and Zr barriers were applied
in the form of thin metallic layers as an integral part
of the tube fabrication process. Pure zirconium (of
either crystal bar or sponge origin)31,32 has been
adopted as the standard barrier because irradiation
As fabricated Prior to PCMI After PCMI
experience showed that copper offered less protec-
Figure 11 Schematic presentation of pellet–cladding tion after high irradiation doses. However, pure zir-
mechanical interaction. P. van Uffelen, R.J.M. Konings, conium oxidizes more rapidly in comparison with
C. Vitanza and J. Tulenko, Analysis of Reactor Fuel Rod Zircaloy, while the terminal solubility of hydrogen
Behavior. In: Handbook of Nuclear Engineering
(D.G. Cacuci, Ed.), Chapter 13, p. 1520-1627, Springer
is lower in the liner.33 Accordingly, when a primary
Germany, 2010, with permission from Springer Scienceþ defect occurs in the cladding, for instance, due to
Business Media B.V. debris fretting, water ingress will oxidize both the
Oxide Fuel Performance Modeling and Simulations 549
fuel and the liner. This occurs typically in the lower develop into an incipient crack, starting and propa-
(cooler) part of the fuel rod. The hydride formation gating along the grain boundary.
near the cladding inner diameter can lead to a sun- For the simulation of crack propagation, most
burst hydride and would, due to the volume increase, authors apply linear elastic fracture mechanics
set up a tangential stress in the Zircaloy part of the (LEFM), in line with Kreyns et al.45 By fitting the
cladding, which can promote crack formation. The experimental data of Wood,46 they advocated that the
local hydrogen absorption can cause more severe crack velocities could be related to the fourth power
hydriding and therefore secondary defects in the of the stress intensity factor KI:
form of long axial splits and circumferential cracks.
da
Secondary cladding defects in LWR fuel can cause ¼ CKI4
large releases of uranium and fission products to the dt
primary coolant,34 and seem to be correlated with PCI where a is the crack length and C is a constant. The
caused, for instance, by control blade movements in crack intensity factor was shown to be controlled by
BWRs. Mitigation of these secondary failures has been the elastic stress field at a flaw tip as described by KI,
achieved by increasing the number of rods per assem- rather than the nominal applied stress:
bly35 (e.g., from 6 6 to 10 10 in BWR assemblies) pffiffiffi
in order to reduce the linear heat generation rate per KI ¼ s a Y
rod and by alloying the liner with either Fe36 or Sn.37 where s represents the nominal hoop stress and the
A more recent component of PCI-resistant fuel factor Y incorporates a correction factor to account
designs is the use of softer fuel pellets obtained by for the finite width of a defect-bearing component.
means of large-grained doped UO2 for both BWR Nevertheless, Kreyns et al.45 pointed out that small-
and PWR applications. Since the early 1990s, AREVA scale plasticity will occur near the crack tip, in a cone
has developed an optimized chromia-doped UO2 fuel with radius rp:
that exhibits significantly higher performance com-
pared to standard UO2.38 The grains are on the order 1 KI 2
of 60 mm compared to 10 mm in standard UO2. As a rp ¼ pffiffiffiffiffi
6p sy
result, the fuel releases less fission gas and is less prone
to PCI failures during ramp tests thanks to a larger where sy corresponds to the yield stress and
pffiffiffiffiffiffiffi
creep rate.39 Postirradiation examinations after ramp KI ¼ s aeff Y , which results in an effective crack
tests revealed a larger number of radial cracks on the length:
pellet periphery. Recent 2D finite element simulations
aeff ¼ a þ rp
of PCI have shown that this can be attributed to the
paffiffiffiffi
ffi
larger friction due to fuel–clad bonding in high- and the effective correction factor Yeff ¼ Y eff
a
burnup fuel and the reduced fracture stress of the becomes
doped fuel.40 The simulations also indicate that, unlike
Y
the hoop stress, the shear stress distribution in the Yeff ¼ rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
2 ffi
cladding is smoother thanks to the reduced fracture 1 ssy Y6p
2
surface. Park et al.44 applied an expression for F within would be below a certain threshold (KI KIA) and the
the range of 5 R/t 20, 2c/a 12, and a/t 0.8: stress intensity factor would decrease. However, quan-
titative information cannot be directly inferred from
F ¼ 1:12 þ 0:053x þ 0:0055x2 þ
their analysis, since the critical stress corresponding to
R 2 KIA should depend on the material (stress relieved vs.
2 20 t
ð1 þ 0:02x þ 0:0191x Þ recrystallized), the irradiation, as well as the loading
1400
aa . history. Furthermore, the LEFM should not be applied
where x ¼
t 2c without corrections for the viscoplastic behavior and,
The KISCC value provided by Park et al. was finally, because of the local inhomogeneities of the
3.3 and 4.8 MPa m1=2 for stress-relieved and recrys- material at the scale of the crack, one should apply
tallized Zircaloy-4, respectively. local rather than global criteria such as KI KIA.
Zhou et al.42 replaced the constant C in the equation With the advent of improved hardware and soft-
for the crack velocity by a function of the iodine ware on one hand and more detailed experimental
concentration and an Arrhenius-type temperature data on the other, more detailed models are being
dependency. More importantly, however, they also developed at various scales. At the electronic and
involved the pellet–cladding contact pressure in the atomic scale, first-principle computations should
estimation of the stress intensity factor. The local effect enable analyzing the individual effect of impurities
of the frictional shear forces is accounted for by adopt- such as iodine on the binding energies, in much the
ing the Coulomb friction model, according to which same way as Xin et al.52 have studied the properties of
the friction force is proportional to the contact pres- point defects and their interactions with Nb in Zr, or
sure. The extension by Zhou et al. is in line with recent Kaji and Tsuru53 analyzed the clustering of Ni in Fe.
findings of Michel et al.48 Based on 2D and 3D finite By means of finite element computations at the level
element computations, they showed that the tangential of crystallographic grains, Rousselier et al. suggest
stress concentration in the cladding is proportional avoiding the limitations associated with LEFM for
to the shear loading transmitted at the pellet–clad crack initiation and propagation. Because of the exces-
interface. As a result, the peak hoop stress at the inner sive computation time and the lack of precise data of
surface of the cladding depends on the interfacial shear some model parameters, their model is not applicable
stress and the uniform loading in the hoop direction. in fuel performance codes but should enable analyzing
Nevertheless, the simulations did not account either the effect of corrosive fission products on the inter-
for cladding anisotropy or for stress relaxation. granular damage by coupling the mechanical problem
Stress relaxation was accounted for in an empirical with a diffusion problem. A similar tool has already
manner by Mattas et al.49 They assumed the chemical been developed by Musienko and Cailletaud,54 albeit
(intergranular) crack growth rate to have an initial the corrosive environmental parameters are accounted
value and to decrease exponentially as the crack for by a phenomenological approach: that is, via an
depth increases. Chemical crack growth was postu- effective diffusion coefficient. At the mesoscopic scale,
lated to continue until a critical stress intensity for Kaji et al. developed a 2D model for SCC growth, in
cleavage and fluting was achieved, at which point order to analyze qualitatively the effect of load (normal
intragranular cleavage initiated until failure. vs. shear stress) and grain boundary corrosion on the
As pointed out by Rousselier et al.,50 stress relaxa- branching aspect of crack growth. The macroscopic
tion and the ensuing crack arrest are necessary to models are mostly based on finite element simula-
explain a so-called discontinuity observed during tions.40,48 These tools provide deeper insight and qual-
ramp experiments: depending on the maximum itative information about the various parameters
power of the fuel rod, either a through crack is obtained affecting PCI as explained above. Nevertheless, these
within maximum 10 min or the SCC damage is limited models are not yet capable of replacing the simplified
to a few micrometers, even after hours of operation at 1D models implemented in fuel performance codes,
the maximum power. This discontinuous behavior was despite the improvements of hardware and software.
observed above the SCC initiation threshold of about Marchal et al.40 are trying to develop analytical
300 MPa in Zircaloy-4.51 Rousselier et al. attributed this ‘enrichments’ for 1D models in fuel performance
to the stress relaxation and to the fact that the inter- codes based on the 2D models. The stochastic nature
granular crack could leave the stress concentration of cracking and the complex evolution of material
zone at the crack tip. Crack arrest was postulated to properties and boundary conditions during irradiation
occur when at the same time the stress intensity factor remain the most important difficulties to be tackled.
Oxide Fuel Performance Modeling and Simulations 551
slower moving species are rate-controlling for most with F_ the fission rate density and T the absolute
diffusion-related processes in UO2.) temperature.
The lattice diffusion coefficient is influenced by Unperturbed (intrinsic) diffusion of single inert
the temperature, deviations from stoichiometry and gas atoms (Xe, Kr) can be observed at low damage
additives (e.g., Cr, Nb), phase changes, and, therefore, and gas concentration (105 at.%). At higher gas and
also indirectly by the burnup. Also, the fission frag- damage concentrations, other effects should be taken
ments are assumed to contribute to the diffusion into account.
process, which is referred to as ‘irradiation-enhanced
diffusion.’ This is due to the interaction of the fission 3.19.2.3.1.3 Trapping
fragments and the associated irradiation damage In nuclear fuels, either natural (e.g., impurities, dislo-
cascades with the fission gas atoms in the lattice, cation lines, closed pores, etc.) or irradiation-produced
resulting in a displacement of the gas atoms. This imperfections in the solid (e.g., vacancy clusters in
effect dominates the diffusion process at temperatures fission tracks, fission gas bubbles, solid fission product
below 1000 C and is temperature independent. For precipitates, etc.) depress the amount of fission pro-
temperatures between 1000 and 1400 C, vacancies ducts available for diffusion by temporarily or perma-
necessary for the gas atom diffusion are assumed to nently trapping the migrating atoms. The experiments
be created both thermally and by the damage cascades show that, for the burnup characteristic of power reac-
related to fission fragments. Above 1400 C, a purely tors, gas atom trapping due to (intragranular) fission
thermally activated diffusion coefficient is applied: gas bubbles in the grains is predominant. The trapping
that is, thermally created vacancies for diffusion are rate depends on the number density and size of the
predominant. These three temperature regimes are intragranular bubbles, and hence on temperature, fis-
reflected in the three components of the single gas sion rate, and burnup. A second important effect of
atom diffusion coefficient (m2 s1) often applied in trapping occurs at the grain boundaries. It deals with
the fuel performance codes59: the delay for the onset of thermal FGR, via the bubble
interconnection mechanism (see Section 3.19.2.3.1.8).
D ¼ D1 þ D2 þ D3
where 3.19.2.3.1.4 Irradiation-induced resolution
A fraction of the gas atoms trapped in bubbles can be
6 35000 redissolved in the surrounding matrix through the
D1 ¼ 7:6 10 exp
T interaction of a fission fragment with the bubble.
pffiffiffi Two different types of mechanisms are proposed to
25 _ 13800
D2 ¼ 5:6 10 F exp explain the experimental observations. On one hand,
T
microscopic models consider the resolution of one
D3 ¼ 8 1040 F_ gas atom at a time when interacting with a fission
Oxide Fuel Performance Modeling and Simulations 553
fragment or an energetic atom from the collision fission product release from nuclear fuels. Migration
cascade. Macroscopic models on the other hand con- of bubbles in the oxide fuels has two other important
sider the complete bubble destruction, but there is consequences, namely the columnar grain growth
still discussion about the detailed mechanisms. For with the concomitant central void formation (ob-
(larger) grain boundary bubbles, resolution is sup- served in fast breeder reactor fuel) and the coales-
posed to be less effective. cence of the bubbles which gives rise to fuel swelling.
Under normal operating conditions, however, intra-
3.19.2.3.1.5 Grain boundary diffusion granular fission gas bubbles remain small (typically
Grain boundary diffusion is the most commonly below 20 nm) due to resolution, and show a small
observed route for solute migration in polycrystalline mobility at least up to 1800 C.61 This is partly
materials. It is generally accepted that diffusion in explained by the pinning by dislocations and other
crystalline solids proceeds more rapidly along grain crystal defects.
boundaries than through the lattice. This is due to the
atomic jump frequency in these planar defects, which is 3.19.2.3.1.8 Grain boundary (intergranular)
about a million times greater than the jump frequency bubble interconnection
of regular lattice atoms in stoichiometric materials at Fission gas bubbles appear along grain boundaries
0.6 times the absolute melting temperature. Neverthe- after a certain burnup, depending on the temperature
less, there is a switch from release assisted by grain history. When bubbles interconnect, they form a so-
boundary diffusion in trace-irradiated UO2 to trapping called tunnel network through which the gas can be
and eventual interlinkage of the intergranular bubbles released. The bubble interconnection is a reversible
(see Section 3.19.2.3.1.8). This switch occurs early in process, for the tunnel network can close again under
life, so that grain boundary diffusion is only considered the influence of the surface tension when the outgo-
to contribute to the precipitation of fission gas atoms in ing flux of gas atoms outweighs their supply.
grain boundary bubbles, rather than to the long-range The bubble interconnection has two important
transport along grain boundaries to the free surface of consequences. First of all, it determines the onset of
the pellets.60 release as the release remains small (due to athermal
release) before grain boundary bubbles interconnect
3.19.2.3.1.6 Grain boundary sweeping or with open grain edge tunnels. This incubation period
grain growth is reflected in the Vitanza threshold for FGR, which is
In LWR fuel under normal operating conditions, shown in Figure 13. The ensuing release corresponds
only normal grain growth is observed, that is, large to a seepage process. Second, when grain face bub-
grains grow at the expense of smaller ones. It affects bles interconnect and form snake-like tunnels, there
the FGR in two ways. First of all, grain boundary will be a sudden release of the gas accumulated in
sweeping provides another mechanism for the collec- these bubbles, referred to as ‘burst release.’ This can
tion of gas at these internal surfaces from which also be interpreted as a sudden interconnection or
release can occur. The collection results from the opening of grain face bubbles due to microcracking
low solubility of the fission gas, and hence the sweep- along grain boundaries during abrupt power varia-
ing grain boundary does not redeposit any gas in the tions. Cracking results in a sudden opening of a frac-
newly formed crystal behind it. The moving grain tion of the grain boundaries with the instantaneous
boundary acts as a fission gas collecting filter. At the venting of the corresponding fraction of the accumu-
same time, grain boundary bubbles hinder grain lated gas atoms.
growth to some extent. Interconnection of gas-filled bubbles takes place
Second, the average diffusion distance for the in general where diffusion-controlled precipitation
fission products created in the grain increases. Unlike occurs at the grain boundaries: that is, when both the
the first consequence, this tends to reduce the release temperature and the burnup are high enough. The
rate. Grain boundary sweeping occurs at tempera- conditions correspond roughly to the Vitanza62,63
tures above roughly 1600 C. threshold:
9800
3.19.2.3.1.7 Intragranular bubble migration Tc ð CÞ ¼ BU
ln 0:005
The migration of intragranular fission gas bubbles pro-
vides an alternative to the sequence ‘bubble formation– where Tc represents the central temperature in ( C)
resolution–gas atom diffusion’ in order to describe and BU the burnup in (MWd kg1) UO2.
554 Oxide Fuel Performance Modeling and Simulations
1800
1700 Original 1% data
Siemens 2% data
Figure 13 Original Halden (or Vitanza) criterion for the onset of fission gas release and supporting data. Data from
Vitanza, C.; Graziani, U.; Fordestrommen, N. T.; Vilpponen, K. O. Fission Gas Release from In-Pile Measurements;
HPR-221.10; OECD Halden Reactor Project; 1978; Vitanza, C.; Kolstad, E.; Graziani, C. Fission gas release from UO2 pellet at
high burnup. In Topical Meeting on Light Water Reactor Fuel Performance, Portland, OR, May 1979; American Nuclear
Society: Portland, OR, 1979.
3.19.2.3.2 Modeling the fission gas behavior created by fission in the fuel matrix. They then dif-
There are various approaches in FGR and swelling fuse in the grains toward grain boundaries by thermal
modeling. They can be classified into two categories. and irradiation-enhanced diffusion. Small intragra-
On one hand, there are purely empirical models, nular bubbles with a diameter of 1–2 nm are observed
including those based on soft computing techniques in irradiated fuel. They are created in the wake of
such as neural networks. These models are inexpensive fission spikes and then grow by diffusion (trapping).
to use and provide an efficient tool for the design of fuel They are continuously destroyed by fission spikes
rods within a limited range of application. However, (resolution). There is no bubble migration except at
they are not suitable for gaining knowledge about the temperatures above roughly 1800 C. The bubbles act
underlying mechanisms, nor do they enable us to extend as sinks for gas atoms, thereby reducing the amount
their range of application to higher discharge burnup of gas available for release.
values as required by the industry. On the other hand, This scenario leads to solving a diffusion equation
there are mechanistic models which aim at the physical in a sphere with a source term proportional to
description of the underlying phenomena as explained the local fission rate density (S ¼ Yfp F_ ), which
in Chapter 3.20, Modeling of Fission-Gas-Induced is based on the pioneering model of Booth.65 He
Swelling of Nuclear Fuels. Despite their great data proposed the equivalent sphere model. This theory
needs, such models provide an excellent basis both for considers a polycrystalline sinter as a collection of
the analysis of the mechanisms and for the extension uniform spheres with an equivalent radius in order
of the models beyond their range of calibration. to simplify the mathematical problem. The hypo-
Fuel performance codes nowadays tend to imple- thetical sphere radius (RB) is defined so that the
ment more and more mechanistic models, based on effective surface-to-volume ratio of the fuel (S/V) is
very detailed but stand-alone models (i.e., models that preserved:
are not implemented in a fuel performance code, such
V
as Van Uffelen57 and Noirot64). They all consider RB ¼ 3
S t
FGR to be a two-step process. The first step deals
with the gas behavior in the grains (intragranular part), where (S/V)t accounts for the sum of the geometric
whereas the second step deals with the gas behavior surface of the pellets as well as the surface due to
along the grain boundaries (intergranular part). open porosity. As irradiation proceeds, fission gases
are generated within the Booth sphere and migrate to
3.19.2.3.2.1 Intragranular behavior the surface, where the concentration is taken to be
For the behavior in the fuel grains, the following zero. He proposed that the fractions of stable gas
scenario is generally adopted. The gas atoms are release can be approximated by
Oxide Fuel Performance Modeling and Simulations 555
sffiffiffiffiffiffiffiffiffi
Dt Dt resolution, fission gas behavior models generally intro-
fann ðt Þ ffi 6 3 2 duce a simple resolution rate coefficient that is pro-
pRgr 2 Rgr
portional to the local fission rate density and depends
for so-called annealing conditions (i.e., without on the bubble size:
source term, but with an initial nonzero concentra- b ¼ 2pðRbl þ dÞ2 mff F_
tion), and
sffiffiffiffiffiffiffiffiffi where it is assumed that a bubble can be destroyed if
Dt 3 Dt its center lies within a distance d from the fission
firr ðt Þ ffi 4
pR2gr 2R2gr fragment track of length mff ¼ 7–10 mm. The condition
for applying Deff is that the traps are saturated. Experi-
for irradiation conditions (nonzero source term, but no
ments show that small intragranular bubbles stabilize
initial concentration). In a second model, he proposed
rapidly both in size and in diameter. Intragranular
the approximation for the release-to-birth ratio for
bubbles can therefore be considered saturated for irra-
unstable gas release under steady-state conditions66:
diation times of practical interest (beyond 0.5 MWd
rffiffiffiffi" rffiffiffiffi! rffiffiffiffi#
R 3 D l 1 D kgU1). Nevertheless, the difference between D and
¼ coth Rgrain Deff is only important for temperatures above
B Rgrain l D Rgrain l
1100 C. It should be underlined, however, that dur-
where l represents the decay constant of the species ing a power ramp the application of Deff provides an
under consideration. It should be underlined that the overestimation of the trapping effect.69
diffusion coefficient to be used is subject to an order of Over time, several models have been proposed
magnitude uncertainty. The expression in Section wherein the Booth sphere radius was taken to be
3.19.2.3.1 is often being used with a multiplication or equal to the average grain radius of the fuel, in order
reduction factor of about 5. to be able to account for grain growth. However, it must
Regardless of the uncertainty on the diffusion first be pointed out that there is no consensus about
coefficient, the Booth models themselves suffer which grain growth model should be applied, even
from several limitations: though the Ainscough et al.70 model is often applied:
1. They consider a constant temperature and fission dRgr 1 1
¼k
rate density. dt Rgr Rmax
2. They do not account for resolution and trapping at where k is a temperature-dependent rate constant and
intragranular bubbles. Rmax ¼ Rmax(BU) is the grain size at which growth stops.
3. They do not account for grain boundary sweeping. This burnup-dependent quantity is introduced in
4. They cannot reproduce an incubation period order to account for the retarding effect of fission
(see Vitanza curve). products on grain growth as burnup proceeds.
5. They do not account for resolution at grain Most FGR models only account for the increase of
boundary bubbles. the average diffusion distance when normal grain
All these limitations have been alleviated over time. growth occurs. Some other models only take into
First, several numerical techniques have been pro- consideration the sweeping effect, assuming either
posed to cope with time-varying conditions, which that the fractional release is proportional to the
have been compared in Lassmann and Benk.67 In grain boundary velocity, or that the gas in the total
order to deal with trapping and resolution, Speight68 fraction of grain volume swept by the grain bound-
found that, instead of solving one diffusion equation aries is released. So they all fail to properly incorpo-
coupled with an equation for the gas balance in the rate boundary motion into the intragranular diffusion
traps, one could solve a single diffusion equation for equation and artificially separate the two aspects of
the sum of the concentration in the matrix and in the grain growth on FGR. Only some stand-alone models
traps with an effective diffusion coefficient (Deff): have been proposed so far that account for both
simultaneously by solving the diffusion equation in
b
Deff ¼ D a sphere with a moving boundary (e.g., Suzuki and
bþg Saitou14 and Forsberg and Massih71).
where g ¼ 4pRbubble D corresponds to the trapping For alleviating the fourth and fifth limitations of
rate coefficient and b corresponds to the resolution rate the Booth models, an intergranular module has to be
coefficient. Whatever model is being considered for introduced.
556 Oxide Fuel Performance Modeling and Simulations
(RIA)79 and LOCA,80 but also provides some novel 3.19.3 Design Basis Accident
results in the description of power ramps.81 Modeling
3.19.2.3.2.3 Coupling intra- and intergranular Fuel designers and safety authorities rely heavily on
behavior fuel performance codes since they require minimal
In general, the intra- and intergranular modules of costs in comparison with the costs of an experiment
a fission gas behavior model are coupled in two or an unexpected fuel rod failure. Nevertheless, two
directions. On one hand, the intragranular module types of fuel performance codes are generally being
provides the source term for the intergranular module. applied, corresponding to the normal operating and
On the other hand, the intergranular module provides the design basis accident conditions (see Chapter
the boundary condition for the diffusion equation 2.22, Transient Response of LWR Fuels (RIA)
in the spherical grains and/or the supplementary and Chapter 2.23, Behavior of LWR Fuel During
source term near the grain boundary. In fact, most Loss-of-Coolant Accidents, respectively). In order
models make use of the Booth approximations; that is, to simplify the code management by limiting the
they assume zero boundary conditions, meaning that number of programs and to take advantage of the
the grain boundary is considered to be a perfect sink. hardware improvements, there is a tendency to gen-
Some models consider a finite segregation factor. In erate a single fuel performance code that can cope
order to account for the resolution effect on grain with the different conditions, which requires differ-
boundary bubbles, three different approaches are ent types of developments.
being utilized. The first group considers a correction On one hand, extending the application range of
factor for the Booth flux, accounting for the fact a fuel performance code originally developed for
that the resolution opposes the gaseous diffusion out steady-state conditions to accident conditions requires
of the grains. The second group of models applies a modifications to the basic equations in the thermo-
time-varying boundary condition that makes use of mechanical description of the fuel rod behavior,6 stable
a time-dependent flux balance. In the third and last numerical algorithms, and a proper time-step control,
group, a supplementary source term is defined in a fine in addition to the implementation of specific models
layer adjacent to the grain face, either as a uniform dealing with the high-temperature behavior of clad-
source in a fine layer or as a Dirac distribution at a ding such as observed under LOCA conditions.85 For
given distance from the grain boundary. Mainly the dealing with RIA events, one should check carefully
first two approaches have been implemented in fuel the thermal expansion model because of the edge-
performance codes. peaked power distribution, as well as the other models
affecting the effective cold gap width,86 and the model
3.19.2.3.2.4 Swelling for thermal heat transfer in the plenum. On the other
Theoretically speaking, the FGR and the fission gas hand, for fuel performance codes developed to simu-
swelling models should be closely related. In most late various aspects of the nuclear fuel behavior under
codes, however, semiempirical relations are being accident conditions, such as TESPA,87 MFPR,88
used for the gaseous swelling as a function of the FRAPTRAN,89 SCANAIR,90 FALCON,91 or
temperature and burnup (e.g., MATPRO correla- RANNS92 code, the thermomechanical behavior of
tion82,83), or as an empirical function of the released the fuel must be incorporated and/or the extension
fraction, for instance Billaux84: of models to normal operating conditions is necessary
to consider burnup-dependent phenomena such as
DV
¼ Að1 aFG FGR aCs CSRÞBU thermal conductivity degradation, FGR, gap closure
V gaseous FP
and swelling, as well as cladding corrosion. Such a
where A is a constant for solid swelling, FGR is posteriori modifications of the fuel performance code
the local fraction FGR from the grain, CSR is the may entail difficulties in terms of convergence and
fractional release of the volatile fission products calculation time.
(Cs, I) from the grain, BU is the local fuel burnup,
aFG ¼ 0.37, and aCs ¼ 0.45. The empirical nature 3.19.3.1 Loss-of-Coolant Accident
reflects the uncertainty pertaining to both the release
and swelling models (in particular during power 3.19.3.1.1 Specific LOCA features
ramps at high burnup), as can also be inferred from As explained in Chapter 2.23, Behavior of LWR
the large variety of models presented above. Fuel During Loss-of-Coolant Accidents, a LOCA
558 Oxide Fuel Performance Modeling and Simulations
in a water-cooled reactor consists of a large break of appreciably larger than in the depressurized reactor
the coolant primary system and the consequent loss vessel – it starts deforming and ballooning in the
of core cooling capacity. It is a design basis accident: space between the fuel assembly grid spacers. If
that is, one that the plant design must account and the ballooning exceeds a certain size, the fuel rod
accommodate for in terms of ensuring that a core cladding will burst open, breaking the first barrier
coolable configuration is maintained. Post-LOCA to radiation containment. In addition, the Zircaloy
coolability is achieved by an immediate reactor cladding undergoes a crystallographic transition from
scram, that is, a rapid insertion of control rods in the a-phase to a b-phase at temperatures exceeding
the core which ends the fission power production in 900 C (see Section 3.19.3.1.2.1), although the transi-
the nuclear fuel, and by the activation of emergency tion temperature is affected by the amount of hydro-
core cooling systems (ECCS) which enters into func- gen contained in the metal cladding as well as the
tion upon depressurization of the primary system, heating rate.
within a relatively short time after the start of the In the third phase, the cladding temperature
transient. continues to increase but gradually due to the
As schematically shown in Figure 15, a typical increasing radiant and convective heat transfer,
LOCA sequence in a PWR consists of three main exhibiting a relatively flat temperature trend for
phases. During the first phase, depressurization of the some period normally lasting for few minutes.
primary system occurs, which is accompanied by a Under the ECCS effect, the cladding tem-
reactor scram. In the course of this phase, the fuel perature trend is then reversed and, after a period
temperature may oscillate depending on position, of gradual temperature decrease, the cladding is
timing of the power reduction, and timing of the finally quenched by the ECCS water rising in
depressurization. The cladding temperature may the core.
even decrease since the water still has enough cooling
capacity and because the water temperature is
3.19.3.1.2 Specific LOCA modeling
decreasing as pressure decreases.
requirements
In the second phase, the core heats up because of
The major concern from the fuel safety criteria view-
the remaining heat generation in the fuel (stored and
point is the rapid cladding oxidation taking place
decay heat) and the loss of water. Normally, the fuel
when the cladding is heated up to high temperature
cladding temperature increases at moderate rate, ap-
and exposed to steam environment, as it occurs in a
proximately some few degrees per second in the ear-
LOCA. In fact, the safety criteria for LOCA are stated
lier phase of the temperature escalation. When the
as five requirements (although small variations exist
temperature exceeds 600 C, the cladding becomes
in each country) dealing with the calculated perfor-
more and more prone to plastic deformation and,
mance of the cooling system under the most severe
under the effect of the rod inner pressure – which is
LOCA conditions:
1. In order to preserve cladding ductility, its oxida-
tion shall be limited to 17% of the metallic wall
ECR thickness (i.e., equivalent cladding reacted, ECR),
Fuel PCT where the oxidation level is computed by means of
1200 Clad the Baker–Just correlation (see Section 3.19.3.1.2.2)
Temperature (⬚C)
Cooling
assuming conservatively that the inner and outer
Quenching
800
Oxidation surfaces of cladding are oxidized concurrently, that
Burst is, two-side oxidation.
2. In order to prevent rapid cladding embrittlement
Rupture as well as runaway oxidation, the cladding peak
400 Ballooning
temperature shall remain below 1204 C.
3. The total amount of hydrogen generation shall
50 100 150 not exceed 1% of the hypothetical amount gener-
Time (s) ated by the reaction of all the metal in the
Figure 15 Schematic evolution of the cladding
surrounding fuel.
temperature in a fuel rod during a loss-of-coolant 4. The calculated changes in core geometry shall
accident. leave the core amenable to cooling.
Oxide Fuel Performance Modeling and Simulations 559
5. After any operation of the ECCS, the core temper- by Forgeron et al.,95 the effect of the temperature
ature shall be maintained at an acceptable low variation rates on the phase transitions are directly
value and decay the heat removed for the extended taken into account by means of a differential
period required by long-lived radioactivity. equation:
The first two criteria can be assessed by means of fuel d’
¼ K ðT Þ’ð1 ’Þ
performance codes containing some specific models dt
such as high-temperature oxidation and deformation, where K(T) is an appropriate function of the temper-
as well as the crystallographic phase transition and ature in the form
the rupture criterion. These specific models are trea-
K ¼
T f 1 ð’Þexp C1 þ C2 T f 1 ð’Þ
ted in the next four sections.
with t as the time, and C1 and C2 as empirical con-
3.19.3.1.2.1 Phase transition model stants. During temperature variations, the fraction of
At high temperatures, the cladding may undergo a the b-phase can therefore be calculated through a
phase transition from the hexagonal closed packed numerical algorithm:
(HCP) a-phase to the face-centered cubic (fcc)
’i ¼ ’i1 þ K ðTi ; ’i1 Þ’i1 ð1 ’i1 ÞDt
b-phase. The a ! b crystallographic phase transition
of zirconium takes place principally in the tempera- where the indexes i and i 1 denote the actual
ture range of 800–1000 C, which is typical in a and the previous time steps, respectively. As an
LOCA. The phase transformation is not only temper- example, Figure 16 also represents a hysteresis loop
ature dependent but is also influenced by the com- of the phase transition calculated for Zircaloy-4 by
position and the impurities of the alloy, the corrosion means of the dynamic model at
10 C s1 heating/
level (oxygen and hydrogen contents) of the cladding, cooling rates.
and the temperature increase rate.93 Quasiequili-
brium conditions are generally assumed in the static 3.19.3.1.2.2 High-temperature oxidation
model, where the fraction of the b-phase (’) is a The cladding–steam reaction model is often based on
simple function of the temperature: ’ ¼ f (T ). Specific parabolic kinetic correlations for both the oxygen
functions are available for each alloy. Figure 16 mass gain and/or the ZrO2 layer thickness growth.
illustrates these functions for Zircaloy-4 and for The actual reaction rate constant is defined as a
Zr1%Nb,94 indicating that due to the niobium con- function of the temperature through an Arrhenius
tent the phase transition of zirconium is completed relation:
at temperatures about 100 C lower. In the case
of a dynamic model, as proposed, for example, Km ¼ Am eQm =T
where Km is the oxygen mass gain rate in mg cm2 s1/2,
Am is the pre-exponential factor, Qm is the activation
1.0
energy divided by the perfect gas constant, and T
Zr1%Nb is the cladding temperature in K. The kinetics
Zircaloy
of high-temperature oxidation of Zircaloy were
0.8
described by Baker and Just in 196296 by the follow-
Fraction of β-phase
ing equation:
0.6
w ¼ 2029t 1=2 e11450=T
0.4
where w is the weight of metal zirconium reacted per
+10 ⬚C s–1 unit surface in mg cm2, T is the temperature in K,
0.2
−10 ⬚C s–1 and t is the time (at the constant temperature T)
in seconds. Although a more precise correlation was
0 derived subsequently by Cathcart and Powel (C–P),97
700 800 900 1000 1100
Temperature (°C) the Baker–Just (B–J) correlation is quoted here because
it is the reference correlation for expressing the clad-
Figure 16 Fraction of the b-phase as a function of
temperature calculated by the equilibrium model for
ding zero ductility criterion for LOCA accidents. An
Zircaloy-4 and Zr1%Nb and by the dynamic model for improved correlation was developed by Cathcart and
Zircaloy-4 at
10 C s1 temperature rates. Powel97:
560 Oxide Fuel Performance Modeling and Simulations
1000
dt RT
where e_ corresponds to the effective strain rate, s is
100 the effective stress, R is the universal gas constant, T
is the temperature, A is the strength coefficient, Q is
10
the activation energy for the plastic deformation, n is
the stress exponent, and x represents the oxygen
1
5 6 7 8 9 10 weight concentration of the cladding.
10 000/T (1 K–1) In order to define the effective strain rate of the
Figure 17 Comparison of the correlations for high- cladding in the metallurgical phase transition range,
temperature corrosion of Zircaloy proposed by Baker and there are two different approaches in the literature.
Just (B–J)96 and Cathcart and Pawel (C–P).97 According to the first approach, the Norton para-
meters (A, Q , and n) are interpolated in the transition
w ¼ 601:8t 1=2 e10050=T range using the fractions of the a- and b-phases as
weighting factors:
The above two correlations are compared in Figure 17, lnðAab Þ ¼ ð1 ’ÞlnðAa Þ þ ’ lnðAb Þ
in terms of the temperature and time needed to achieve Qab ¼ ð1 ’ÞQa þ ’Qb
the same level of oxidation (17% of metal) for a two- nab ¼ ð1 ’Þna þ ’nb
side oxidation of a PWR 17 17 cladding. The figure
reveals that the B–J correlation is somewhat more The second approach is based on micromechanical
conservative in that in the high temperature range the considerations for composite materials assuming a
given oxidation level (17%) is predicted for somewhat statistically homogeneous distribution of a-phase
lower temperatures and/or lower times as compared to inclusions embedded in the b-phase matrix; that is,
the C–P correlation. The difference is 50 C in the each representative volume element has the same a
temperature interval 1100–1200 C, whereas it is prac- to b volume fraction corresponding to the macro-
tically negligible in the 800–1000 C range. scopic average value. According to the basic relation
Typical correlations for the Zr1%Nb-specific of self-consistency, the overall average strain, and con-
alloy were published by Solyany et al.98: sequently the macroscopic strain rate of the two-phase
cladding, is expressed as the weighted average of the
Km ¼ 920e10410=T strain rates of each phase.
e_ ab ¼ ð1 ’Þe_ a þ ’e_ b
and Gyó´ri et al.99
200 400 600 800 1000 1200 multirod calculation, thereby consuming large com-
1000 puting resources. However, only with this type of
T (⬚C)
numerical analysis can we simulate the actual thermal
Zircaloy
Zr1%Nb
and mechanical interactions taking place between the
nearest, second-nearest, or third-nearest neighbor
100 rods.3,103–105 In the assembly, the cladding circumfer-
stB (MPa)
five orders of magnitude larger than the rate during a 3.19.3.2.2.1 Thermal and mechanical calculations
normal power ramp of about 10 kW m1 min1. The for fast transients
clad load is multiaxial: the axial to tangential strain To cope with the speed of the temperature increase
ranges between 0 and 1. and the large radial gradients, an accident analysis
During the first phase of the RIA, the cladding code needs an appropriate number of radial ring
temperature remains moderate, typically below elements and numerical stability. In addition, the
600 C. Therefore, the cladding can fail due to temperature gradient generates a strong thermal
PCMI, depending on the cladding conditions (irradi- stress particularly within a pellet. Therefore, in gen-
ation damage, oxide thickness, hydride concentration eral, FEM is the most versatile and reliable method
and orientation, etc.). used in these situations.111,112
During the second phase of the RIA, the cladding The prediction of DNB and recovery to rewetting
temperature continues to rise because of the energy after the PCMI phase is hard to model because of the
generated by the fuel, and can lead to departure from limited amount of experimental data.113
nucleate boiling (DNB). Under these conditions, the
cladding temperature could remain above 1000– 3.19.3.2.2.2 FGR during RIA
1200 K for up to 10–15 s. During the temperature The internal gas pressure after a RIA is remarkably
increase, the clad stiffness is strongly impaired, while enhanced by the burst release of fission gas following
the inner gas pressure of the rod increases due to gas the grain separation,114 and is generally correlated
release and the temperature increase. A local balloon- with the peak fuel enthalpy.115 The fission gas origi-
ing deformation may occur and, if it becomes severe, nates mainly from the intergranular gas bubbles, even
clad rupture may result. though some contribution comes from gas pores
During the third stage of the RIA, the cladding is within the grain and the high-burnup structure
submitted to a rewetting or quenching, which may (HBS).115,116 Although a few models consider the
cause a fragile rupture, especially when clad oxida- thermal diffusion process of fission gas atoms from
tion is severe. inside the grains (e.g., Khvostov et al.79), it is generally
During the fourth and last stage of an RIA, fuel considered that it can be substantially neglected.115
dispersal can occur. Indeed, when the energy depo- This is mostly explained by the short duration of the
sition is high enough, fuel melting accompanied by high-temperature period of the pellet (e.g., a few or
excessive fuel swelling could cause clad failure by several tens of seconds) to allow for the thermal
overheating or overstraining. When dispersed into diffusion to contribute significantly to the FGR
the coolant upon clad failure, the hot fuel may increase, even if the fuel pin runs into a DNB phase.
cause a pressure pulse in the water due to vaporiza- In any case, it is necessary to evaluate the preac-
tion, potentially damaging neighboring assemblies or cident gas inventory within the pellet, particularly
other core components. in the intergranular gas bubbles. Obviously, this
Papin et al.110 summarize the experimental obser- evaluation depends on the performance of all the
vations and classify the typical sequences of rod models that are responsible for fission gas behavior
behavior during pulse irradiation on the basis of the during the base-irradiation period.
results obtained from the CABRI-REP-Na tests, and Most research is directed toward the PCMI fail-
Fuketa et al.108 from the NSRR tests. ure type because of the tendency to increase the
average discharge burnup levels for economic rea-
3.19.3.2.2 Specific RIA modeling sons, and because this is considered to be the most
requirements severe criterion for high-burnup rods. The predic-
Among the four failure modes described above, tion of fuel rod failure during a RIA requires the use
there are two dominant types of fuel rod failures of a computer code that accounts for the various
during a RIA. The first deals with a typical PCMI aspects of the fuel behavior. In order to guarantee
failure and is predominant for fuel rods irradiated to acceptable calculation times, however, only simpli-
an average burnup beyond 30–40 MWd kg1 HM fied models for the clad rupture can be afforded
because of gap closure and clad ductility reduction. rather than complex mechanistic models. On the
The second type of failure is more typical for fuel basis of the experimental results and main charac-
rods with a lower burnup level and is caused by teristics, several RIA rupture criteria have been
gas pressure at high temperature during the post- proposed and adopted in fuel performance codes.
DNB phase. The models are either based on the rupture strain,
Oxide Fuel Performance Modeling and Simulations 563
8
in terms of a rupture strain energy density (SED), >
< 1:0 e_ y < 8:37 104 s1
or in terms of an enthalpy limit. f1 ð_ey Þ ¼ 0:046 0:31 log10 e_ y 8:37 104 s1 e_ y 1s1
>
:
0:046 e_ y > 1s1
3.19.3.2.2.3 Rupture strain
The models based on rupture strain postulate clad f2 ðcHex ; T ; e_ y Þ ¼ 0:01 þ 0:99egðT ;_ey ÞcHex
failure when
gðT ; e_ y Þ ¼ 6:52 104 þ 2:21 103 ð6 þ log10 e_ y Þ
ey ey ðz; t Þ
p
p T 298
e_ y ðz; t Þ > 0 1 tanh
8:5
p p
where ey and e_ y represent the radially averaged clad
f3 ðfÞ ¼ 0:05 þ 0:95e2:8910 f
24
hoop strain and strain rate, respectively. The uniform
hoop elongation is often selected for two reasons.
The empirical correlation for the rupture hoop strain
First, the largest strains due to PCMI are in the
is valid for temperatures between 20 and 400 C,
tangential direction. Second, the true failure strain
deformation rates between 5 105 and 1 s1, hydro-
is somewhere between uniform elongation and total
gen concentrations in the range of 0–1941 ppm, and
elongation (includes both uniform and localized
fluences between 0 and 12 1025 n m2. Jernkvist118
strain) at failure. Geelhood et al.117 have implemented
implemented this rupture threshold in the
such a uniform strain-based failure model in the
SCANAIR-3.2 code, which was originally developed
FRAPTRAN code. The empirical correlation was
at the IRSN in France. It was verified on the basis of
derived from biaxial burst tests in addition to uniaxial
integral tests in the CABRI and NSRR pulse reactors.
tension tests on irradiated Zircaloy, and is a function
The comparison with the experimental data indi-
of the temperature and excess hydrogen in the clad-
cates that the strain-based criterion is conservative,
ding (i.e., the quantity of hydrogen above the solu-
and that the standard deviation of the relative differ-
bility limit). Therefore, they have had to apply a
ences between calculated and measured failure strain
correction factor because of the nonrepresentative
obtained with the modified SCANAIR-3.2 code is
stress conditions. They also implemented a new
0.57. A part of the scatter has been attributed to the
yield stress correlation, whereby the plastic strain
deviation of the loading conditions of the out-of-pile
hardening component is dependent on both the tem-
tests from those of the in-pile tests (tangential and
perature and the fast neutron fluence (representing
axial loading nearly equal), which required the ap-
radiation damage) and has been fitted on the basis of
plication of a correction factor to the measured hoop
the same set of experiments. As a result, their failure
strains.
predictions account for the temperature, the irradia-
Desquines et al.119 adopted a more physical-based
tion damage, the strain rate, and the hydrogen con-
modeling for defining the rupture strain. More pre-
tent. In a similar manner, Jernkvist118 proposed more
p cisely, they applied the concept of elastoplastic fail-
recently an empirical correlation for ey for recrystal-
ure mechanics, in line with Kuroda et al.,120 and
lized Zry-2 and stress-relieved Zry-4 on the basis of
combined this with a probabilistic model for the defect
more than 200 out-of-pile mechanical property tests:
size and the oxide layer thickness (or average hydro-
3e0y Sf1 f2 f3 gen content) in order to determine the failure proba-
efy ¼ bility as a function of the hoop strain at failure, the
f1 f2 þ f1 f3 þ f2 f3
oxide layer thickness, the temperature, and the height
where e0y is the plastic strain to failure of as-fabricated of the fuel rod.121 Although they lacked experimental
cladding at low strain rate, S is a ductility reduction data at temperatures above 400 C, they applied the
factor for clad tubes with spalled oxide layer respon- rupture criterion in the SCANAIR code at IRSN.
sible for detrimental effects of a nonuniform hydride
3.19.3.2.2.4 Strain energy density
distribution, and f1, f2, and f3 are ductility reduction
The strain-based failure criterion is conceptually
factors accounting for elevated strain rate, hydrogen-
simple, and the correlation can be determined in a
induced embrittlement, and radiation damage,
rather straightforward manner from total elongation
respectively. The functions have been fitted to the
as well as burst strain data. However, these data are
experimental data and are as follows:
not true material properties, and attention must be
e0y ðT Þ ¼ 2:82 102 þ 1:22 104 T paid to the impact of the specimen geometry and load
564 Oxide Fuel Performance Modeling and Simulations
biaxiality. Furthermore, the strain-based failure cri- and propagates in the reduced ductility region to the
terion depends on the mechanical loading and tem- inner part, and at a certain depth onset of through-
perature history prior to failure. As an alternative, a wall fracture occurs. A direct implication of these
critical strain energy density (CSED) has been pro- observations is that the plastic strain around the
posed by Rashid et al.122 and implemented in the crack tip is mainly responsible for the onset of
FALCON code. The measure for the clad to resist through-wall fracture: that is, final fracture. In other
failure is formulated as the total mechanical energy words, fracture toughness is associated with the small
(elastic þ plastic) per unit volume that can be ab- region around the crack tip and the cladding inner
sorbed by the material before it can fail. The threshold region, which is heated to high temperature, would
value for SED is derived from the experimental have a negligible contribution to the fracture tough-
stress–strain curves and is expressed as a function of ness. From this point of view, further study is justified
the fast neutron fluence, temperature, hydrogen con- to design an alternative failure criterion127 to the
centration, and material type, or, alternatively, the macroscopically derived quantities such as CSED or
oxide thickness. A correction factor was applied to plastic strain to failure.
account for deviations from the biaxiality of stress
during the tests, and it should be underlined that the
tests were limited in terms of strain rate (7 105 s1 3.19.4 Advanced Issues and
instead of 1–5 s1 for RIA). A similar approach was Future Needs
adopted by Bernaudat and Pupier123 and Leclercq
3.19.4.1 Deterministic Versus Probabilistic
et al.124 for application in the SCANAIR code applied
Analyses
by Électricité de France (EdF), although the CSED
limit is expressed as a function of the oxide thickness In the introduction, we have enumerated the various
or the average burnup: sources of uncertainties, and in the second section we
have provided more details about the limitations of
CSED ¼ 95:17 4:7653OX þ 8:759OX2
the models. In order to assess the technological effect
5:56 104 OX3 OX < 54mm of all sources of uncertainties, there are various tech-
niques that may be considered.
CSED ¼ 5:3 otherwise
First of all, there are various so-called sensitivity
where OX is the oxide thickness in micrometers methodologies, ranging from multiple runs with
and CSED is expressed in megapascals. The conver- input data or model parameters being varied to a
sion of the oxide thickness in average fuel burnup rigorous mathematical treatment based on perturba-
depends on the material under consideration, although tion theory. Apart from the numerical noise tech-
Bernaudat et al. concluded that the correlation derived nique which is included in the TRANSURANUS
for Zircaloy-4 could be (conservatively) applied to code, the latter has never been applied to fuel rod
ZIRLO and M5. modeling, probably because the mathematical effort
The SED is not as easily measured in mechanical would be too important to formulate all the nonlinear
property tests as the plastic strain to failure, and phenomena in the form of basic differential equa-
direct measurements of this quantity are seldom tions. The numerical noise analysis allows the esti-
made. In addition, the derivation of the SED from mation of the standard deviation, for instance, of the
the (nonrepresentative) experiments relies on a con- centerline temperature within just one run. For this
stitutive relation for the material, and is also sensitive purpose, the most important parameters of the fuel
to the load and temperature history. There is thus no rod (e.g., q0 and l) are varied slightly on a random
obvious benefit in using the critical SED instead of a basis in an interval, which is similar to introducing
critical strain in clad failure criteria for RIA.118 numerical noise. These changes of model parameters
Furthermore, the plastic strain of cladding at fail- result in modifications of the resulting quantities,
ure is a few to several ten of percent in calculation, which enables the analysis of the sensitivity of the
although the actual plastic strain is largely localized code with respect to its parameters. The computational
around the fracture site. Metallographic cross- effort is therefore identical to that of a deterministic
sectional observation of PCMI-failed cladding indi- run, but there is no information available about the
cates that some incipient cracks are generated in the effect between the various uncertain parameters.
outer oxide layer or in the dense hydride (embrittled) A second category of probabilistic approaches is
region under the oxide.125,126 The crack then grows the response surface technique. This is based on a
Oxide Fuel Performance Modeling and Simulations 565
careful combination of parameters called the experi- numerical noise analysis. The comparison shows that
mental design, such as the Latin hypercube sampling the latter method, even though computationally
or the Taguchi design. Peck128 tried this with the equivalent to a single deterministic run, provides a
FRAP code. The main problem is that one has to similar approximation for the standard error of the
decide in advance which parameter should be centerline temperature as the MC technique in the
selected and how fine the variation of each parameter case under consideration.
should be. Also, this approach requires a definition of In Figure 20, a potential problem of an MC
the interactions between the perturbed parameters. calculation is illustrated. More precisely, when the
This is not always easy to specify. standard deviations of the input variables and their
Finally, the Monte Carlo (MC) method is based distributions were chosen too high, the resulting
on random sampling of all variables that are consid- temperatures cover a range from 850 to 1260 C. It
ered. It seems most attractive in view of its simplicity, is not trivial to neglect results below a specific
but it entails high computational costs (factor 100– frequency.
500); the knowledge about the distributions of the Despite the restrictions, applying a limited varia-
variables is rather limited, and it is unclear how much tion of some fabrication variable parameters in fuel
each parameter uncertainty contributes to the varia- rod performance calculations is being considered in
tion of the outcome. In order to reduce the computa- Germany for two reasons. First of all, it is argued129
tional costs, quasi-MC methods may be applied. that probabilistic calculations can replace a deter-
Instead of random sampling, quasirandom sampling ministic calculation with superimposed unfavorable
sequences are recommenced, which fill the space tolerance limits on some fabrication parameters
more uniformly than uncorrelated random numbers. (worst case dataset) in fuel design calculations. The
They may be considered as a combination of the varia- results obtained with the deterministic worst case
tions of parameters in response surface techniques calculation are said to be very conservative, with a
with its experimental design and the standard MC degree of conservatism that would be difficult to
technique. quantify. Probabilistic calculations based on distribu-
In Figure 19, a comparison is made between a tions on the other hand would allow the replacement
deterministic and the two statistic analyses in the of the worst case dataset by a dataset leading to
TRANSURANUS code. The points with the error results with a known, defined conservatism. The
bars indicate the mean value and the standard devia- industry, of course, would like to reduce the margins
tion of the centerline temperature at four distinct in this way. Second, it has been observed that in some
times according to a MC analysis, whereas the dashed cases the deterministic conservative approach did not
line indicates the standard deviation according to the in fact predict the worst fuel performance as in a
1200
Centerline temperature (⬚C)
900
600
300
TRANSURANUS v1m1j03
Monte Carlo analysis (v1m1j03)
TRANSURANUS v1m1j03
Numerical noise analysis
0
0 6000 12 000 18 000
Time (h)
Figure 19 Comparison between a deterministic and two statistical analyses implemented in the TRANSURANUS code.
566 Oxide Fuel Performance Modeling and Simulations
0.05
0
600 800 1000 1200 1400
Centerline temperature (⬚C)
Figure 20 Fractional frequency of the centerline temperature predictions when varying only three input parameters
(q0 , hgap, l).
probabilistic approach. This, of course, depends on referred to as the ‘rim-effect’ because of its appear-
the proper definition of the worst case on one hand ance in the pellet rim (several 100 mm at a pellet
and on the limits of variation of the variable para- burnup of 55 MWd kg1 HM). The HBS is charac-
meters in the MC approach on the other. terized by micrometer-sized pores with a swelling
For these reasons, a semistatistical design approach contribution up to 15%, surrounded by small grains
has been proposed in Germany.130,131 The method on the order of 100–300 nm with a very low but
consists of computing Spearman’s rank coefficients constant concentration of Xe of around 0.25 wt%. It
following a limited number of MC simulations. The starts in the pellet periphery when the local burnup
rank coefficients are then converted into quantiles of exceeds a value of about 60 MWd kg1 HM, and the
the statistical distributions in order to produce a set local temperature remains below 1000 C. It then
of parameter combinations that lead to the fuel re- progressively grows toward the pellet center as long
sponse that covers with a predefined probability the as these two conditions are maintained.
whole spectrum of results when applying all possible The mechanisms responsible for the development
parameter combinations. This set of parameters is of the HBS, as well as a description of its consequences
then applied to deterministic fuel rod computations. on fuel behavior, are detailed in Chapter 2.19, Fuel
In a similar manner, a nonparametric order statis- Performance of Light Water Reactors (Uranium
tical method is being applied in the United States,132 Oxide and MOX). In the present chapter, we limit
offering a more realistic estimation of the operational ourselves to describing briefly how it is accounted for
margin through the statistical evaluation of numbers in fuel performance codes.
of fuel rods coming close to the design limit by In fuel performance codes, the effect of the HBS
assessing the statistical certainty for the actual occur- on the heat transfer is generally included in a corre-
rence of extreme cases. lation describing the thermal conductivity degrada-
tion (see Chapter 2.17, Thermal Properties of
3.19.4.2 The High Burnup Structure Irradiated UO2 and MOX)
1
The resonance absorption of epithermal neutrons by l¼
238
U leads to the buildup of 239Pu in the pellet A0 þ A1 BU þ BT
periphery after two successive b-decays. The high and the porosity buildup in the HBS as well as in an
local burnup in combination with the low tempera- improved gap conductance in order to account for
ture leads to the formation of the HBS, sometimes the bonding layer between the ceramic pellets and
Oxide Fuel Performance Modeling and Simulations 567
the metallic cladding. To this end, for example, the performance codes rely on empirically fitted func-
cladding and pellet roughness is empirically reduced tions reflecting the resonance capture, for instance in
above a certain burnup to account for the filling up RAPID136 or TUBRNP.137 The latter solves the local
by mainly zirconium oxide, which has a higher ther- differential equations for the uranium and plutonium
mal conductivity than that of the mixture of He and isotopes:
Xe in the residual gap. dN235 ðr Þ
Thermal conductivity degradation of the fuel is ¼ sa;235 N235 ðr ÞA
dBU
sometimes considered to saturate:
pffiffiffiffiffi dN238 ðr Þ
l0 pffiffiffiffiffiffiffipffiffiffiffiffi ¼ sa;238 N 238 f ðr ÞA
l ¼ pffiffiffiffiffiffiffi arctan D BU l0 þ lel dBU
D BU
dN239 ðr Þ
¼ sa;239 N239 ðr ÞA þ sc;238 N 238 f ðr ÞA
with l0 ¼ 1/(A þ BT ) and D a constant.133 It can be dBU
shown that in the low burnup limit, this expression
dNj ðr Þ
tends to the classical form of l0. Occasionally, partial ¼ sa;j Nj ðr ÞA þ sc;j 1 Nj 1 ðr ÞA
thermal conductivity restoration is even considered dBU
when the HBS is created. In this way, account is made The local concentration of 238U, N238(r), is written as
for the ‘cleaning’ of the matrix from point defects and N 238 f ðr Þ, where f (r) is a radial shape function
Ð with a
rout
fission products, which relies on an empirical formu- 2 f ðr Þr dr
lation of the lattice parameter variation.134 The latter normalization factor defined by rin
rout
2 r 2 ¼ 1,
in
approach seems to be in line with recent experimen- where rin and rout are the inner and outer fuel radii.
tal data.135 The function f (r), called the radial shape function, is
The effect of the long irradiation time in the f ðr Þ ¼ 1 þ p1 expðp2 ðrout r Þp3 Þ. The constants p1,
mechanical analysis is mostly accounted for in p2, and p3 are fitted to electron microprobe analysis
the cladding properties. The process of outer clad- (EMPA) data for each type of reactor.
ding corrosion liberates hydrogen from the water, and The impact of the HBS on FGR is incorporated in
reduces the thickness of the cladding metallic wall. various ways. The simplest way is based on an empir-
A fraction of that hydrogen, called the ‘hydrogen ical threshold depending on the local burnup, and/or
pickup fraction,’ is absorbed by the Zircaloy. This temperature, and/or initial grain size. Others include
fraction depends on the composition and heat treat- the burnup as a parameter in the diffusion coefficient
ment of the cladding (and is usually introduced as a or the S/V value (for the Booth sphere radius), or in
fitting constant). The hydrogen concentrates in the the grain boundary saturation value. Quantitative
cold part of the cladding. As soon as the solubility details of most models are not available in the open
limit is exceeded, it precipitates as brittle hydrides. literature. In the START3 code,138 the grain size
When these hydrides are oriented normal to the clad- reduction observed is modeled in an empirical manner
ding surface, the cladding strength is further reduced. as a function of temperature and burnup. As the
Reduction of the microhardness (Figure 21) and average distance for intragranular diffusion thereby
Young’s modulus (see Figure 22) observed in high- reduces, an increased release fraction is predicted.
burnup fuel pellets is not yet included in the codes. Nevertheless, there is evidence that direct release
This also holds for the bond between the pellet and from the HBS is small under normal operating condi-
the cladding. On one hand, it improves the PCI tions.139 Apart from this general statement that the
resistance since it is much softer than UO2. On the release is small, there is no consensus about the quanti-
other hand, the pellet–clad bonding would make tative contribution from the HBS to the overall release
the cladding subject to a biaxial stress state due to under normal operating conditions. This is very well
pellet stack elongation. This biaxial stress state is a reflected in the large spread of FGR predictions at very
mechanically tougher condition for cladding.92 high burnup for the codes involved in the FUMEX-II
The HBS is associated to some extent with the benchmark exercise organized by the International
radial power profile in the pellets. The resonance Atomic Energy Agency (IAEA) (Figure 23).
capture of epithermal neutrons in the peripheral Under design basis accident conditions, the
regions of the pellets renders the application of release models disregard the kinetics. Release is
neutron diffusion theory obsolete. In order to repro- assumed to be instantaneous (provided that the gap
duce the stronger gradient in the pellet rim, fuel is open) and comes from grain boundary cracking.
568 Oxide Fuel Performance Modeling and Simulations
40 400
Porosity (SEM-1000 X)
Porosity (OM-500 X)
Min. sq. fit. all porosity data
Local burnup (APOLLO 2 calc.) I
10 100
0 0
0.0 0.2 0.4 0.6 0.8 1.0
1000
9
Hardness (kg mm–2)
8
800
Hardness (GPa)
7
600 6
400 4
0.0 0.2 0.4 0.6 0.8 1.0
Pellet radial position, r/ro
Figure 21 Experimental values for microhardness of irradiated UO2 fuel with a burnup of 98 GWd t1 HM. Reproduced from
Spino, J.; Stalios, A. D.; Santa Cruz, H.; Baron, D. J. Nucl. Mater. 2006, 354, 66–84, with permission from Elsevier.
The precise modeling of this relies on the modeling Consequently, the fuel rod behavior codes do not
of the local conditions of temperature and stress, require fundamental modifications. The thermal
which is very difficult and hardly discussed in properties such as melting temperature, thermal
publications. conductivity, and expansion must be modified.140
Finally, the swelling in the HBS is mostly modeled Also, mechanical properties such as thermal creep
separately with a very simple linear or quadratic and theoretical density require appropriate
function of burnup (Figure 24). Also, the width of modifications.
the HBS structure is sometimes modeled empirically In addition to material properties, a limited num-
as a function of the burnup. ber of models must be adapted as well. The first
obvious modification pertains to the neutronic
model for computing the radial power profile and
3.19.4.3 Mixed Oxide Fuels
isotope production. In particular, the resonance cap-
As the plutonium concentration of mixed oxide ture in 240Pu must be dealt with in a similar way as for
238
(MOX) fuel in LWRs is low, its behavior in a reactor U, for instance with an empirical radial depen-
is not very different from that of UO2 fuel (more dence, and dedicated effective cross-sections for
so because at EOL roughly 80% of the heat pro- MOX configurations in PWRs and BWRs must be
duced in UO2 fuel comes from plutonium fissions). used.141 As such, one will be able to reproduce the
Oxide Fuel Performance Modeling and Simulations 569
240.00 UO2
HBRP UO2
HBRP rod 1
220.00 HBRP rod 2
HBRP rod 3
HBRP rod 4
HBRP rod 6
Young modulus (GPa)
200.00 HBRP UO2+5% Gd
HBRP rod 1 Gd
HBRP rod 2 Gd
180.00 HBRP rod 3 Gd
HBRP rod 4 Gd
HBRP rod 6 Gd
160.00 N118 initial
N118_68 GWJ tM–1
140.00
120.00
100.00
0 50 100 150 200
Fuel burnup (GWd tM–1)
Figure 22 Experimental Young’s modulus of irradiated light water reactor fuels as a function of burnup. Data from Baron
et al. Evolution of the nuclear fuel mechanical properties with burnup. An extensive European experimental program.
In International Topical Meeting on Light Water Reactor Fuel Performance, Kyoto, Japan, Oct 2–6, 2005; ANS: Kyoto, Japan, 2005.
40
START-3
35 TU (Bul)
BACO
INFRA
30 FRAPCON
FEMAXI
25 FEMAXI
FGR (%)
TRANSURANUS
PSI
20
15
10
0
0 10 20 30 40 50 60 70 80 90 100
Burnup (MWd kgU–1)
Figure 23 Cumulative fractional fission gas release predicted for a constant linear heat rate of 15 kW m1 by various
fuel performance codes involved in the FUMEX-II round robin exercise of the International Atomic Energy Agency.
Data from Killeen, J.; Turnbull, J. A.; Sartori, E. In International LWR Fuel Performance Meeting, San Francisco, CA, Sept 30 to
Oct 3, 2007; American Nuclear Society: San Francisco, CA, 2007, pp 261–273, with permission from John Killeen.
larger flux depression in MOX at BOL, whereas at in the Halden BWR. According to these experiments,
EOL the higher residual plutonium content at the the Vitanza threshold (see Section 3.19.2.3.1) would
center will lead to higher local powers compared to apply to UO2 and MOX fuel alike. Some experiments,
those in UO2 fuel. however, indicate that the microstructure of some
Higher FGR at EOL is the only apparent problem MOX fuels such as micronized master blend
mentioned for MOX fuel. It is mainly attributed to (MIMAS)-MOX could also contribute to the higher
the higher linear powers at EOL, in combination with release fractions. This has been ascribed to the hetero-
a lower thermal conductivity. This is corroborated geneity at the microscopic scale, in particular to the
by the similar threshold for thermal release measured Pu-enriched agglomerates characterized by a Pu
570 Oxide Fuel Performance Modeling and Simulations
20
Vennix 500x
Vennix 800x
Vennix 816x
Vennix 1600x
15 Van de Laar 1600x
Estimation from
density measurements
TRANSURANUS
Porosity (%)
10
0
0 0.5⫻105 1.0⫻105 1.5⫻105 2.0⫻105
Local burnup (MWd tU–1)
Figure 24 Porosity as a function of the local burnup for highly irradiated fuel and the corresponding correlation adopted in
the TRANSURANUS code. Reproduced from Lassmann, K; Schubert, A.; van de Laar, J.; Vennix, C. W. H. M. Recent
developments of the TRANSURANUS code with emphasis on high burnup phenomena. In IAEA Technical Committee
Meeting on Nuclear Fuel Behaviour Modelling at High Burnup, Lake Windermere, UK, June 19–23, 2000; pp 387–406.
content of around 25% and a size in the order of generated in fuel matrices by a-decay of transura-
50 mm or larger. The burnup and fission product con- nium nuclides, ternary fission, and the (n,a) reaction
centrations are therefore roughly three times higher in on 16O. The He production is 4–10 times higher in
these agglomerates. Nevertheless, plutonium breeding MOX in comparison to UO2 fuel. Studies on He
in the UO2 matrix surrounding the agglomerates behavior showed that its diffusion coefficient in UO2
together with fission product recoil tends to reduce is several orders of magnitude larger than that of noble
the heterogeneity, in particular for agglomerates smal- gases. Nevertheless, its release from the fuel occurs
ler than 30 mm, which can even ‘dissolve’ in the UO2 only if its content is above the solubility limit. The
matrix with progress of burnup. latter strongly depends on temperature and He fill gas
In view of the particularities of FGR in MOX sum- pressure,146 but is still being analyzed for irradiated
marized above, three different modeling approaches MOX. The solubility in fresh MOX fuel could explain
are adopted: some codes do not modify anything why less release is observed in PWR rods (He fill gas
because the use of existing uranium fuel FGR models pressure >20 bar), as shown in Figure 25, whereas
for MOX fuel applications have been reported to give significant release is observed in low prepressurized
satisfactory results.142 Some codes multiply the fis- BWR MOX rods. This is why some codes do not
sion rate with a ‘heterogeneity factor’ to account for consider the He release at all (in PWRs), while others
the higher athermal diffusion coefficient, or simply use a simple (linear) correlation between the He release
multiply the diffusion coefficient with a constant factor and FGR observed BWR MOX fuel (Figure 26).
(1.75). Some others have multiplied the FGR fraction
based on the model for UO2 with a so-called heteroge-
3.19.4.4 Multiscale Modeling
neity factor (1.3). Finally, there are stand-alone
mechanistic models for FGR in MOX fuel wherein In the previous sections, the main models included in
the evolution of the Pu distribution in agglomerates the fuel performance codes have been outlined.
and U-matrix is computed, the recoil from agglomer- Every code contains semiempirical correlations that
ates is accounted for, and/or the size distribution of Pu are very valuable for technological applications.
agglomerates is taken into account.143–145 However, they are valid only within the confines of
Another issue closely related to the FGR in the parameters and irradiation conditions covered
MOX is the He production and release. Helium is in the database on which they are developed.
Oxide Fuel Performance Modeling and Simulations 571
550
1 cycle 2 cycles 3 cycles 4 cycles
500
Helium volume (cm3)
450
400
300
Figure 25 Helium balance in mixed oxide pressurized water reactor fuel rods. After Blanpain, P.; Thibault, X.;
Pages, J. Recent results from the in reactor MOX fuel performance in France and improvement program. In International
Topical Meeting on Light Water Reactor Fuel Performance, Portland, OR, Mar 2–6, 1997; ANS: Portland, OR, 1997; pp 39–45.
Cecades
testing Finite
Mechanical element
macroscopic
property systems
experiments Fracture
mechanics
S-year TEM, in situ TEM Continuum material
failure
mechanics
constitutive
properties
en e
3D dislocation
rim tur
ms-s
ts
dynamics
pe uc
Timescale
spatial evolution
ex rostr
Rate/field of dislocation-
theory defect ensembles
c
4D evolution Thermo-
Mi
dynamics,
equations
els
kinetics
ns-ms
od
Kinetic M
Molecular monte carlo
dynamics- 3D spatial
semi-empirical evolution
potential TEM
Ab initio
ps-ns
electronic
structure SANS
methods PAS
Atomic-nm nm-mm mm-mm mm-m
Length scale
Figure 27 Comparison between He gas and fission gas release in boiling water reactor mixed oxide fuel. Reproduced
from Mitsugi, T.; Kushida, N.; Kikuchi, K. Behaviour of MOX fuel irradiated in a thermal reactor. In International Topical
Meeting on Light Water Reactor Fuel Performance, Portland, OR, Mar 2–6, 1997; ANS: Portland, OR, 1997; pp 54–61.
performance code including the various techniques designers, and nuclear fuel researchers. The various
described below is quite unlikely despite the hardware benchmarks for fuel rod performance codes organized
evolution over the last decades. Some of the reasons are by the IAEA155 have revealed that, in general and
related to the limitations of the various techniques under normal operating conditions,
involved and dealt with in detail in Chapter 1.08, Ab
Temperatures can be predicted with a relative
Initio Electronic Structure Calculations for Nuclear
error of the order to 10%.
Materials; Chapter 1.10, Interatomic Potential
The clad elongation as well as the cladding dia-
Development; Chapter 1.11, Primary Radiation
metrical deformation are predicted with a relative
Damage Formation; Chapter 1.12, Atomic-Level
error of about 30% (although based on less data).
Level Dislocation Dynamics in Irradiated Metals;
The ratio of predicted to measured values for FGR
Chapter 1.13, Radiation Damage Theory; Chapter
is generally within a factor of 2.
1.14, Kinetic Monte Carlo Simulations of Irradia-
tion Effects; Chapter 1.15, Phase Field Methods; and Most difficulties stem from
Chapter 1.16, Dislocation Dynamics.
Uncertainties in input values (fabrication and irra-
diation parameters).
3.19.5 Summary and Conclusions
The stochastic nature of pellet cracking, and its
consequences on the thermal (uncertainty on the
The modeling of nuclear fuel behavior involves many gap size, especially at BOL) as well as on the
phenomena and disciplines ranging from chemistry, mechanical analysis (validity of compatibility and
nuclear and solid-state physics, metallurgy, ceramics, constitutive equations). In addition, crack healing is
and applied mechanics. The strong interrelationship difficult to model properly.
between these disciplines, as well as the nonlinearity
A lack of direct measurements (e.g., stresses or
of many of the processes involved, calls for the devel- parameters such as diffusion coefficients and reso-
opment of computer codes describing the general lution rate coefficients) or large scatter in the
fuel behavior. experimental data.
After years of development, the fuel behavior codes
A limited set of experimental data available in the
have become standard tools for safety authorities, fuel open literature for new materials that are being
Oxide Fuel Performance Modeling and Simulations 573
introduced for reaching higher discharge burnups Modeling the behavior of MOX fuel for LWRs has
(e.g., new cladding material properties). also become a routine task in some specific countries
as it became a mature industrial product. Generally,
It should be underlined that some specific fuel per- the FGR fraction at EOL is higher in comparison with
formance codes (typically developed by fuel vendors) that of UO2 fuel. The difference in power history and
have been fine-tuned with a larger database but for a the lower thermal conductivity of MOX have been
particular type of fuel and therefore provide more identified to be the main causes, although the role of
accurate predictions for this type of fuel. the heterogeneous microstructure in MIMAS-type
It should also be pointed out that the same diffi- MOX continues to be assessed through research.
culties apply to multidimensional fuel performance The mechanistic models for FGR for MOX
codes. Although these are complementary to the available in the open literature do not account for
standard codes and offer an improved description of the evolved microstructure, which apart from the
specific phenomena such as PCI or cladding balloon- Pu-rich agglomerates and U-rich matrix may also
ing, and despite superior computational resources contain a Pu-rich coating layer at grain boundaries.
available nowadays, it is unlikely that they can over- With respect to helium, the fractional release
come all these difficulties. observed in MOX fuel rods loaded in BWRs is gen-
In view of the tendency to increase the discharge erally taken to be proportional with the FGR frac-
burnup of fuel rods, modifications have been made or tion. In contrast, the absence of He release observed
still need to be implemented. In particular, for the in PWR MOX has been ascribed to the higher He
thermal conductivity, recent measurements revealed solubility due to the elevated fill gas pressure. With
that the HBS leads to a recovery of thermal conduc- the long storage times of spent fuel, the issue of He
tivity (as point defects and fission products are release will gain importance in the codes. Neverthe-
removed from the fuel matrix).135 Nevertheless, this less, more experimental data are needed in the open
recovery cannot always compensate the impedance literature for MOX fuel.
increase to the heat flow resulting from the concomi- In order to complement the time-consuming and
tant porosity buildup, depending on the extent of the costly experiments for developing and optimizing
porosity buildup. advanced fuels for the next generation of reactors,
In parallel, the bonding layer between the fuel and multi-timescale modeling is being launched across
the cladding at these burnup levels not only elimi- the globe. Yet, despite promising results for the
nates progressively the effect of surface roughness on behavior of fission products in UO2, and a better
the heat transfer in the gap but also decreases the understanding of the basic phenomena, it is still too
temperature jump induced by gap gas. ambitious to conceive a fuel performance code based
With respect to the mechanical properties, various solely on fundamental computations. There are limita-
experimental techniques indicate a softening of the tions to each individual technique in terms of temper-
fuel when the HBS is established.139,156 ature range as well as space and time scale, in addition
As far as the fission gas behavior is concerned, to the stochastic nature of several phenomena (e.g.,
there is a consensus that most fission gas remains in fuel cracking) or the strong relationship between vari-
the HBS (90%), whereas earlier it was believed to ous phenomena entailing high nonlinearities.
be responsible for the observed increase in FGR Until recently, fuel performance codes were either
beyond average burnup levels of 50 MWd kg1 HM. developed for simulating the fuel rod behavior under
Despite the relatively benign effects of the HBS normal operating conditions and mild transients or for
on the thermomechanical behavior of the fuel rod simulating accident behavior. At present, there is a
under normal operating conditions, the HBS still tendency to widen the scope of both types of codes so
requires attention for modeling the transient behavior as to simulate the fuel rod behavior under all conditions
of nuclear fuel. In the event of an RIA, for example, it with a single code in a more consistent manner, and a
is important to know how much gas resides in porosity few codes have already successfully demonstrated the
in terms of a potential FGR. The quantification of the applicability for this extended range of conditions.
gas distribution at high burnup still requires more Finally, it remains important that all fuel perfor-
experimental efforts. However, the contribution of mance codes be further benchmarked with peer-
the HBS porosity gas to the rapid mechanical loading reviewed experimental data, such as those proposed
of the cladding has not been clearly observed and can in the frame of international benchmarks organized
be considered insignificant.107,116 by the OECD/NEA and the IAEA.
574 Oxide Fuel Performance Modeling and Simulations
Acknowledgments 20. Garcia, P.; Struzik, C.; Agard, M.; Louche, V. Nucl. Eng.
Des. 2002, 216, 183–201.
21. Massih, A. R.; Jernkvist, L. O. Mechanics and PCMI. In
The authors would like to express their gratitude for Lecture Notes of the OECD – HRP Summer School on
the constructive remarks from Dr M. Zimmermann Fuel Behaviour Modelling, Halden, Norway, Aug 31 to
Sept 3, 2004; Chapter 6.
(PSI) on the manuscript. 22. Owen, D. R. J.; Hinton, E. Finite Elements in Plasticity:
Theory and Practise; Pineridge: Swansea, UK, 1980.
23. Oguma, M. Nucl. Eng. Des. 1983, 76, 35–45.
References 24. Berna, G. A.; Beyer, C. E.; Davis, K. L.; Lanning, D. D.
FRAPCON-3: A Computer Code for the Calculation of
Steady-State, Thermal-Mechanical Behaviour of Oxide
1. Yagnik, S. K. In Overview of EPRI Fuel Performance Fuel Rods for High Burnup; NUREG/CR-6534,
Code FALCON, Seminar Proceedings; Cadarache, PNNL-11513; 1997.
France, Mar 3–6, 1998; Nuclear Science: Cadarache, 25. Lassmann, K.; Schubert, A.; Van Uffelen, P.;
France, 1998; pp 267–272. van de Laar, J. TRANSURANUS Handbook; 2005.
2. Bentejac, F.; Hourdequin, N. TOUTATIS, an application 26. Van Uffelen, P.; Sheindlin, M.; Rondinella, V.; Ronchi, C.
of the CAST3M finite element code for PCI In Pellet–Clad Interaction in Water Reactor Fuels
three-dimensional modelling. In International Seminar (PCI-2004); CEA Cadarache/DEN/DEC:
on Pellet–Clad Interaction in Water Reactor Fuels, Aix-en-Provence, France, 2004.
Aix-en-Provence, France, Mar 9–11, 2004. 27. Jankus, V. Z.; Weeks, R. W. Nucl. Eng. Des. 1972, 18,
3. Georges, R.; François, J.; François, B.; François, L.; 83–96.
Jean-Marc, R. DRACCAR, a new 3D-thermal mechanical 28. Hoppe, N. Nucl. Eng. Des. 1980, 56, 123–133.
computer code to simulate LOCA transient on Nuclear 29. Newman, C.; Hansen, G.; Gaston, D. J. Nucl. Mater.
Power Plants – Status of the development and the 2009, 293, 6–15.
validation. In ICAPP’09, Tokyo, Japan, 2009. 30. Cox, B. J. Nucl. Mater. 1990, 172, 249–292.
4. Olander, D. R. Fundamental Aspects of Nuclear Reactor 31. Schrire, D.; et al. Secondary defect behaviour in ABB
Fuel Elements; Technical Information Center – Energy BWR fuelIn Topical Meeting on LWR Fuel Performance,
Research and Development Administration, University of West Palm Beach, FL, Apr 17–21, 1994; pp 398–409.
California: Berkeley, CA, 1976. 32. Harbottle, J. E.; Kennard, M. W.; Sunderland, D. J.;
5. Bailly, H.; Ménessier, D.; Prunier, C. The Nuclear Fuel of Strasser, A. A. The behaviour of defective BWR barrier
Pressurized Water Reactors and Fast Reactors; and non-barrier fuelIn Topical Meeting on LWR Fuel
Lavoisier: Paris, France, 1999. Performance, West Palm Beach, FL, Apr 17–21, 1994;
6. Lassmann, K.; Van Uffelen, P. The Structure of Fuel Rod pp 391–397.
Codes; EUR 21400 EN; European Commission, 2004. 33. Van Swam, L. F.; Strasser, A. A.; Cook, J. D.; Burger, J. M.
7. Lassmann, K.; Hohlefeld, F. Überarbeitung des URGAP Behaviour of Zircaloy-4 and zirconium liner Zircaloy-4
Modells zur Berechnung des Wärmeübergangs cladding at high burnup. In Proceedings of the
zwischen Brennstoff und Hülle; EUR-10756 DE; International Topical Meeting on LWR Fuel Performance,
European Commission, 1986. Portland, OR, Mar 2–5, 1997; pp 421–431.
8. Lassmann, K.; Hohlefeld, F. Nucl. Eng. Des. 1987, 103, 34. Karlsson, J.; Lysell, G.; Pettersson, H.; Rönnberg, G. In-pile
215–221. testing of liner cladding an pellet performance in failed fuel
9. Ross, A. M.; Stoute, R. L. Heat Transfer Coefficient rods. In International Meeting on LWR Fuel Performance,
Between UO2 and Zircaloy-2; AECL-1552; Atomic Orlando, FL, Sept 19–22, 2004; Paper 1046.
Energy of Canada Limited (AECL): Mississauga, ON, 35. Garner, N. L.; Rentmeister, T.; Lippert, H.-J.; Mollard, P.
1962. Upgraded fuel assemblies for BWRs. In International
10. MacEwan, J. R.; Stoute, R. L.; Notley, M. J. F. J. Nucl. LWR Fuel Performance Meeting, San Francisco, CA,
Mater. 1967, 24, 109–112. Sept 30 to Oct 3, 2007; Paper 1091.
11. Lassmann, K. Nucl. Eng. Des. 1980, 57(1), 17–39. 36. Hoffmann, P. B.; Dewes, P. Post-irradiation examination
12. Lassmann, K. Nucl. Eng. Des. 1987, 103, 211–214. and ramp testing of fuel rods with Fe-enhanced Zr liner
13. Lassmann, K. J. Nucl. Mater. 1992, 188, 295–302. cladding at high burnup. In International Meeting on LWR
14. Suzuki, M.; Saitou, H. Light Water Reactor Fuel Analysis Fuel Performance, Orlando, FL, Sept 19–22, 2004;
Code FEMAXI-6 (Ver. 1); JAEA-Data/Code 2005-003 Paper 1059.
JAEA: Japan, 2006. 37. Sihver, L.; Hallstadius, L.; Wikmark, G. Recent ABB BWR
15. Assmann, H.; Stehle, H. Nucl. Eng. Des. 1978, 48, failure experienceIn International Topical Meeting on
49–67. LWR Fuel Performance, Portland, OR, Mar 2–5, 1997;
16. Suk, H. C.; Wang, W.; Kim, B. G.; Kim, K. S.; Heo, Y. H. pp 356–364.
Improvement of ELESIM-CANDU fuel performance 38. Delafoy, C.; Dewes, P.; Miles, T. AREVA NP Cr2O3-
analysis codeIn Technical Committee Meeting on Fission doped fuel development for BWRs. In International LWR
Gas Release and Fuel Chemistry Related to Extended Fuel Performance Meeting, San Francisco, CA, Sept 30
Burnup, Pembroke, ON, Apr 1992; pp 193–201. to Oct 3, 2007; Paper 1071.
17. Jackson, P. A.; Turnbull, J. A.; White, R. J. In A 39. Nonon, C.; Menard, J.; Lansiart, S.; et al. PCI behaviour
Description of the ENIGMA Fuel Performance Code; of chromium oxide doped fuel. In International Seminar
IAEA-TECDOC-697; IAEA: Vienna, Austria, 1989; pp 28–38. on Pellet–Clad Interaction in Water Reactor Fuels,
18. Jackson, P. A.; Turnbull, J. A.; White, R. J. Nucl. Energy Aix-en-Provence, France, Mar 9–11, 2004.
1990, 29(2), 107–114. 40. Marchal, N.; Campos, C.; Garnier, C. Comput. Mater.
19. Cunningham, M. E.; Beyer, C. E.; Medvedev, P. G.; Sci. 2009, 45, 821–826.
Berna, G. A.; Scott, H. FRAPTAN: A computer code 41. Misfeldt, I. A stress corrosion failure model. In IAEA
for the transient analysis of oxide fuel rods; Specialists Meeting on Fuel Element Performance
NUREG/CR-6739, Vol. 1, PNNL-13576; 2001. Computer Modelling, Blackpool, UK, 1978.
Oxide Fuel Performance Modeling and Simulations 575
42. Zhou, G.; Lindbäck, J. E.; Schutte, H. C.; Jernkvist, L. O.; 67. Lassmann, K.; Benk, H. J. Nucl. Mater. 2000, 280,
Massih, A. R. In Pellet–Clad Interactions in Water Reactor 127–135.
Fuels, Aix-en-Provence, France, Mar 9–11, 2004; OECD: 68. Speight, M. V. Nucl. Sci. Eng. 1969, 37, 180–185.
Aix-en Provence, France, 2004. 69. Lösönen, P. J. Nucl. Mater. 2000, 280, 56–72.
43. Miller, A. K.; Ocken, H.; Tasooji, A. J. Nucl. Mater. 1981, 70. Ainscough, J. B.; Oldfield, B. W.; Ware, J. O. J. Nucl.
99, 254–268. Mater. 1973, 49, 117–128.
44. Park, S. Y.; Kim, J. H.; Lee, M. H.; Jeon, Y. H. J. Nucl. 71. Forsberg, K.; Massih, A. R. Theory of fission gas release
Mater. 2008, 372, 293–303. during grain growth. In Structural Mechanics in Reactor
45. Kreyns, P. H.; Spahr, G. L.; McCauley, J. E. J. Nucl. Technology (SMiRT-16), Washington, DC, Aug 12–17,
Mater. 1976, 61, 203–212. 2001; Paper 1931.
46. Wood, J. C. J. Nucl. Mater. 1972, 45(2), 105–122. 72. White, R. J.; Tucker, M. O. J. Nucl. Mater. 1983, 118,
47. Anderson, T. L. Fracture Mechanics – Fundamentals and 1–38.
Applications, 2nd ed.; CRC Press: Boca Raton, FL, 1995. 73. Koo, Y. H.; Sohn, D. S.; Yoon, Y. K. J. Nucl. Mater. 1994,
48. Michel, B.; Sercombe, J.; Thouvenin, G. Nucl. Eng. Des. 209, 62–78.
2008, 238, 1612–1628. 74. Kogai, T. J. Nucl. Mater. 1997, 244, 131–140.
49. Mattas, R. F.; Yaggee, F. L.; Neimark, L. A. Iodine stress- 75. White, R. J. J. Nucl. Mater. 2004, 325, 61–77.
corrosion cracking in irradiated Zircaloy cladding. In 76. Rest, J.; Zawadski, S. A. FASTGRASS: A Mechanistic
Topical Meeting LWR Fuel, Portland, OR, Apr 30 to Model for the Prediction of Xe, I, Cs, Te, Ba and Sr
May 2, 1979. Release from Nuclear Fuel Under Normal and Sever-
50. Rousselier, G.; Leclercq, S.; Diard, O. Scenario for Accident Conditions; NUREG/CR-5840, ANL-92/3, 1992.
the damage of PWR fuel cladding in situations of 77. Olander, D. R.; Mubayi, V. J. Nucl. Mater. 1999, 270(1–2),
pellet–cladding interaction. In 17th International 1–10.
Conference on Structural Mechanics in Reactor 78. Khvostov, G. A dynamic model for fission gas release
Technology (SMiRT-17), Prague, Czech Republic, and gaseous swelling integrated into the FALCON fuel
Aug 17–22, 2003; Paper c03-3. analysis and licensing code. In Top Fuel 2009, Paris,
51. Schuster, I.; Lemaignan, C. J. Nucl. Mater. 1992, 189, France, Sept 6–10, 2009.
157–166. 79. Khvostov, G.; Zimmermann, M. A.; Ledergerber, G.
52. Xin, X. K.; Lai, W. S.; Liu, B. X. J. Nucl. Mater. 2009, 393, Parametric study of fuel rod behaviour during the RIA
197–202. using the modified FALCON code. In OECD/NEA
53. Kaji, Y.; Tsuru, T. Investigation of model for stress Workshop on Nuclear Fuel Behaviour During Reactivity
corrosion cracking of cladding materials. In 7th Initiated Accidents, Paris, France, Sept 9–11, 2009.
International Workshop Materials Modelling and 80. Khvostov, G.; Zimmermann, M. A.; Sugiyama, T.;
Simulations or Nuclear Fuel (MMSNF), Karlsruhe, Fuketa, T. On the use of the FALCON code for modeling
Germany, Sept 29–30, 2008. the behaviour of high burn-up BWR fuel during the LS-1
54. Musienko, A.; Cailletaud, G. Acta Mater. 2009, 57, pulse-irradiation test. In International Conference on the
3840–3855. Physics of Reactors, PHYSOR’08, Interlaken,
55. Suzuki, M.; Uetsuka, H.; Saitou, H. Nucl. Eng. Des. 2004, Switzerland, Sept 14–19, 2008.
229, 1. 81. Herranz, L. E.; Vallejo, I.; Khvostov, G.; Sercombe, J.;
56. Shimada, S.; Etoh, E.; Hayashi, H.; Tukuta, Y. J. Nucl. Zhou, G. Insights into fuel rod performance codes during
Mater. 2004, 327, 97–113. ramps: Results of a code benchmark based on the SCIP
57. Van Uffelen, P. Ph.D. Thesis, University of Liege: Liege, project. In Top Fuel 2009, Paris, France, Sept 6–11, 2009.
Belgium, 2002. 82. MATPRO library, in Report NUREG/CR-5273,
58. Grimes, R. W. In Fundamental Aspects of Inert Gases in EGG-2555; 1990.
Solids; NATO Advanced Science Institute, Series B; 83. MATPRO-Version 11 (Revision 2), A handbook of
Plenum: New York, 1991; Vol. 279, pp 415–429. material properties for use in the analysis of light water
59. Turnbull, J. A.; Friskney, C. A.; Findlay, J. R.; Johnson, F. A.; reactor fuel rod behaviour; NUREG/CR-0497,
Walter, A. J. J. Nucl. Mater. 1982, 107, 168–184. TREE-1280, Rev. 2; 1981.
60. Olander, D. R.; Van Uffelen, P. J. Nucl. Mater. 2001, 288, 84. Billaux, M. R. Fuels performance, limits, operational and
137–147. safety issues. In Lecture Notes at Frederic Joliot–Otto
61. Matzke, H. Radiat. Eff. 1980, 53, 219–242. Hahn 2005 Summer School, Karlsruhe, Germany,
62. Vitanza, C.; Graziani, U.; Fordestrommen, N. T.; Aug 24 to Sept 2, 2005.
Vilpponen, K. O. Fission Gas Release from In-Pile 85. Van Uffelen, P. J. Nucl. Mater. 2008, 383, 137–143.
Measurements; HPR-221.10; OECD Halden Reactor 86. Meyer, R. O. The U.S. Nuclear Regulatory Commission’s
Project, 1978. research on fuel behaviour under accident conditions. In
63. Vitanza, C.; Kolstad, E.; Graziani, C. Fission gas release International Meeting on LWR Fuel Performance,
from UO2 pellet at high burnup. In American Nuclear ‘‘Nuclear Fuel: Addressing the Future’’, Topfuel,
Society, Topical Meeting on Light Water Reactor Fuel Salamanca, Spain, Oct 22–27, 2006.
Performance, Portland, OR, May 1979. 87. Sonnenburg, H. G. Study of fuel rod criteria for LOCA
64. Noirot, L. In International Topical Meeting on Light Water condition in the light of recent experimental data. In
Reactor Fuel Performance, Kyoto, Japan, Oct 2–5, 2005; International Meeting on LWR Fuel Performance,
ANS: Kyoto, Japan, 2005. ‘‘Nuclear Fuel: Addressing the Future’’, Topfuel,
65. Booth, A. H. A Method of Calculating Fission Gas Salamanca, Spain, Oct 22–27, 2006.
Diffusion from UO2 Fuel and Its Application to the X-2-f 88. Veshchunov, M. S.; Ozrin, V. D.; Shestak, V. E.;
Loop Test; CRDC-721; AECL: Chalk River, ON, 1957. Tarasov, V. I.; Dubourg, R.; Nicaise, G. Nucl. Eng. Des.
66. Booth, A. H. A Suggested Method for Calculating the 2006, 236(2), 179–200.
Diffusion of Radioactive Rare Gas Fission Products from 89. Berna, G. A.; Scott, H. FRAPTRAN: A Computer Code
UO2 Elements and a Discussion of Proposed In-Reactor for the Transient Analysis of Oxide Fuel Rods;
Experiments that May Be Used to Test Its Validity; NUREG/CR-6739; US Nuclear Regulatory Commission:
DCI-27; AECL: Chalk River, ON, 1957. Washington, DC, 2001.
576 Oxide Fuel Performance Modeling and Simulations
90. Federici, E.; Lamare, F.; Bessiron, V.; Papin, J. In IAEA 112. Suzuki, M.; Fuketa, T. Nucl. Technol. 2006, 155, 282.
Technical Committee Meeting on Fuel Behaviour Under 113. Bessiron, V.; Sugiyama, T.; Fuketa, T. J. Nucl. Sci.
Transient and LOCA Conditions; Halden, Norway, 2001; Technol. 2007, 44, 723.
IAEA: Halden, Norway, 2001; pp 88–101. 114. Lemoine, F. J. Nucl. Mater. 1997, 248, 238.
91. Rashid, J.; Montgomery, R.; Yagnik, S.; Yang, R. 115. Jernkvist, L. O.; Massih, A. R.; Fuketa, T.; et al. Nuclear
Behavioral modeling of LWR fuel as represented in the Fuel Behaviour Under Reactivity-Initiated Accident (RIA)
FALCON code. In Workshop on Materials Modelling and Conditions; State-of-the-art report; NEA/CSNI/R (2010)1;
Simulations for Nuclear Fuel (MMSNF), New Orleans, LA, OECD-NEA: Paris, France, 2010.
Nov 2003. 116. Sasajima, H.; Sugiyama, T.; Chuto, T.; Nagase, F.;
92. Suzuki, M.; Fuketa, T. Nucl. Technol. 2006, 155, Nakamura, T.; Fuketa, T. J. Nucl. Sci. Technol. 2010, 47,
282–292. 202.
93. Massih, A. R. J. Nucl. Mater. 2009, 384, 330–335. 117. Geelhood, K. J.; Beyer, C. E.; Cunningham, M. E.
94. Gyó´ri, Cs.; Van Uffelen, P.; Schubert, A.; van de Laar, J.; Modifications to FRAPTRAN to predict fuel rod failures
Hózer, Z. Implementing experimental data on the due to PCMI during RIA-type accidentsIn Proceedings of
accidental behaviour of the WWER cladding obtained at the 2004 International Meeting on LWR Fuel
the AEKI in the TRANSURANUS fuel performance code. Performance, Orlando, FL, Sept 2004; pp 585–595.
In 6th International Conference on WWER Fuel 118. Jernkvist, L. O. J. Nucl. Sci. Technol. 2006, 43(5), 546–561.
Performance, Modelling and Experimental Support, 119. Desquines, J.; Cazalis, B.; Bernaudat, C.; Poussard, C.;
Albena, Bulgaria, Sept 19–23, 2005. Averty, X.; Yvon, P. In Proceedings of the 14th
95. Forgeron, T.; Brachet, J. C.; Barcelo, F.; et al. Experiment International Symposium on Zirconium in the Nuclear
and modelling of advanced fuel rod cladding behaviour Industry; Stockholm, Sweden, June 13–17, 2004; ASTM:
under LOCA conditions: Alpha-beta phase Stockholm, Sweden, 2004; pp 851–872.
transformation kinetics and EDGAR methodology. In 120. Kuroda, M.; Yamanaka, S.; Nagase, F.; Uetsuka, H.
12th International Symposium: Zirconium in the Nuclear Nucl. Eng. Des. 2001, 203, 185–194.
Industry, West Conshohocken, PA, 2000. 121. Le Saux, M. Comportement et rupture de gaines en
96. Baker, L.; Just, L. C. Studies of Metal Water Reactions at Zircaloy-4 détendu vierges, hydrurées ou irradiées en
High Temperature III; ANL-6548; Argonne National situation accidentelle de type RIA; Ecole National
Laboratory: Chicago, IL, 1962. Superieure des Mines (MINES ParisTech): Paris, 2008.
97. Cathcart, J. V.; Powel, R. E. Zirconium Metal-Water 122. Rashid, J. Y. R.; Montgomery, R. O.; Lyon, W. F.;
Oxidation Kinetics, IV. Reaction Rate Studies; Yang, R. A cladding failure model for fuel rods subjected
ORNLINUREG-17; ORNL: Oak Ridge, TN, 1977. to operational transients and accident transientsIn
98. Solyany, V. I.; Bibilashvili, Y. K.; Tonkov, V. Y. Nuclear Fuel Behaviour Modelling at High Burnup and Its
In OECD-NEA-CSNI/IAEA Specialists Meeting on Water Experimental Support. Proceedings of a Technical
Reactor Fuel Safety and Fission Product Release in Committee Meeting, Windermere, UK, June 19–23,
Off-Normal and Accident Conditions, Risø, Denmark, 2000; pp 187–199.
May 16–20, 1983; IAEA: Denmark, 1983; p 163. 123. Bernaudat, C.; Pupier, P. A new analytical approach to
99. Gyó´ri, Cs.; Hózer, Z.; Lassmann, K.; et al. Extension of study the rod ejection accident in PWRsIn Proceedings
TRANSURANUS code applicability with niobium of the International Topical Meeting on Light Water
containing cladding models (EXTRA)In EU Research in Reactor Fuel Performance, Kyoto, Japan, 2005;
Reactor Safety, Conclusion Symposium on Shared-Cost pp 602–614.
and Concerted Actions (FISA-2003), Luxembourg, 124. Leclercq, S.; Parrot, A.; Leroy, M. Nucl. Eng. Des. 2008,
Nov 10–13, 2003; Proceedings – EUR 21026, 238(9), 2206–2218.
Luxembourg, 2003; pp 584–589. 125. Tomiyasu, K.; Sugiyama, T.; Fuketa, T. J. Nucl. Sci.
100. Urbanic, V. F.; Heidrick, T. R. J. Nucl. Mater. 1978, 75(2), Technol. 2007, 43, 733.
251–261. 126. Suzuki, M.; Sugiyama, T.; Udagawa, Y.; Nagase, F.;
101. Pawel, R.; Cathcart, J.; McKee, R. J. Electrochem. Soc. Fuketa, T. Comparative analysis on behavior of high
1979, 126, 1105–1111. burnup PWR fuels pulse-irradiated in reactivity-initiated
102. Uchida, M.; Otsubo, N. Models of Multi-Rod Code accident conditions. In Water Reactor Fuel Performance
FRETA-B for Transient Fuel Behavior Analysis (Final Meeting, Paris, France, 2009.
Version); JAERI-1293; JAERI: Japan, 1984. 127. Sartoris, C.; Taisne, A.; Petit, M.; Barré, F.; Marchand, O.
103. Uchida, M.; Nakamura, J.; Otsubo, N. In Water Reactor In Workshop on Nuclear Fuel Behaviour During Reactivity
Fuel Element Performance Computer Modelling; Initiated Accidents; OECD-NEA: Paris, France, 2009.
Gittus, J. H., Ed.; Applied Science: London, 1982; pp 128. Peck, S. O. Automated uncertainty analysis methods in
631–647. the FRAP computer codes. In IAEA Specialist’s Meeting
104. Uchida, M.; Nakamura, J. Nucl. Eng. Des. 1981, 65, on Fuel Element Performance Computer Modelling,
63–69. Blackpool, UK, 1980.
105. Uchida, M. Nucl. Eng. Des. 1984, 77, 37–47. 129. Heins, L.; Groß, H.; Nissen, K.; Wunderlich, F. J. Nucl.
106. Meyer, R.; McCardell, R.; Chung, H.; Diamond, H.; Mater. 1991, 178, 287.
Scott, H. Nucl. Saf. 1996, 37, 271–288. 130. Sauer, G. Kerntechnik 2005, 70(3), 153–157.
107. Fuketa, T.; Sasajima, H.; Mori, Y.; Ishijima, K. J. Nucl. 131. Sauer, G.; Besenböck, W. Reliable statistical fuel rod
Mater. 1997, 248, 249–256. analysis. In Water Reactor Fuel Performance Meeting,
108. Fuketa, T.; Sasajima, H.; Sugiyama, T. Nucl. Technol. Seoul, Korea, Oct 19–23, 2008; Paper 8031.
2001, 133, 50–62. 132. Arimescu, V. I. Best-estimate BWR fuel thermal-
109. Nakamura, T.; Kusagaya, K.; Fuketa, T.; Uetsuka, H. mechanical analysis based on non-parametric order
Nucl. Technol. 2002, 138, 246. statistics. In Water Reactor Fuel Performance Meeting,
110. Papin, J.; Balourdet, M.; Lemoine, F.; Lamare, F.; Seoul, Korea, Oct 19–23, 2008; Paper 8094.
Frizonnet, J.; Schmitz, F. Nucl. Saf. 1996, 37, 289. 133. Sontheimer, F.; Landskron, H.; Billaux, M. R. A fuel
111. Suzuki, M.; Sugiyama, T.; Fuketa, T. J. Nucl. Sci. thermal conductivity correlation based on the latest
Technol. 2008, 45, 1155–1164. experimental resultsIn Proceedings Seminar on Thermal
Oxide Fuel Performance Modeling and Simulations 577
Performance of High Burn-up LWR Fuel, Cadarache, 143. Billaux, M.; van Vliet, J. Res Mech. 1986, 17, 41–57.
France, Mar 3–6, 1998; p 119. 144. Ishida, M.; Korei, Y. J. Nucl. Mater. 1994, 210,
134. Baron, D. About the modelling of fuel thermal 203–215.
conductivity degradation at high-burnup accounting for 145. Koo, Y. H.; Lee, B. H.; Cheon, J. S.; Sohn, D. S. Ann.
recovering process with temperatureIn Proceedings of Nucl. Energ. 2002, 29, 271–286.
the Seminar on Thermal Performance of High Burn-up 146. Blanpain, P.; Lippens, M.; Schut, H.; Federov, A. V.;
LWR Fuel, Cadarache, France, Mar 3–6, 1998; Bakker, K. Helium solubility in UO2: The HARLEM
pp 129–143. project. In Proceedings of the Workshop Materials
135. Walker, C. T.; Staicu, D.; Sheindlin, M.; Papaioannou, D.; Models and Simulations for Nuclear Fuels, Nice, France,
Goll, W.; Sontheimer, F. J. Nucl. Mater. 2006, 350, 19–39. June 1–2, 2006.
136. Lee, C. B.; Kim, D. H.; Song, J. S.; Bang, J. G.; Jung, Y. H. 147. Stan, M. Materials Models and Simulations in Support of
J. Nucl. Mater. 2000, 282(2–3), 196–204. Nuclear Fuels Development; LA-UR-05-5652; Los
137. O’Carroll, C.; Walker, C. T.; van de Laar, J. In Technical Alamos National Laboratory, 2005.
Committee Meeting on Water Reactor Fuel Element 148. Crocombette, J. J. Nucl. Mater. 2002, 305, 29–36.
Modelling at High Burnup and Experimental Support; 149. Crocombette, J.; Jollet, F.; Nga, L.; Petit, T. Phys. Rev. B
International Atomic Energy Agency: Windermere, 1994. 2001, 64, 104–107.
138. Bibilashvili, Y. K.; Medvedev, A. V.; Khostov, G. A.; 150. Wirth, B. D.; Odette, G. R.; Marian, J.; Ventelon, L.;
Bogatyr, S. M.; Korystine, L. V. Development of the Young-Vandersall, J. A.; Zepeda-Ruiz, L. A. J. Nucl.
fission gas behaviour model in the START-3 code and its Mater. 2004, 329–333, 103–111.
experimental support. In Proceedings of the International 151. Malerba, L. J. Nucl. Mater. 2006, 351(1–3), 28–38.
Seminar on Fission Gas Behaviour in Water Reactor 152. Domain, C.; Becquart, C.; Malerba, L. J. Nucl. Mater.
Fuels, Cadarache, France, Sept 2000. 2004, 335, 121–145.
139. Spino, J.; Stalios, A. D.; Santa Cruz, H.; Baron, D. J. Nucl. 153. Kotomin, E. A.; Mastrikov, Y. A.; Rashkeev, S. N.;
Mater. 2006, 354, 66–84. Van Uffelen, P. J. Nucl. Mater. 2009, 393, 292–299.
140. Beauvy, M. In Frederic Joliot–Otto Hahn Summer School 154. Watanabe, T.; Srivilliputhur, S. G.; Schelling, P. K.;
on Nuclear Reactors ‘‘Physics, Fuels and Systems’’, Tulenko, J. S.; Sinnott, S. B.; Phillpot, S. R. J. Am. Ceram.
Cadarache, France, Aug 23 to Sept 1, 2006; CEA-FZK: Soc. 2009, 92(4), 850–856.
Cadarache, France, 2006. 155. Killeen, J.; Turnbull, J. A.; Sartori, E. In International LWR
141. Schubert, A.; Van Uffelen, P.; van de Laar, J.; Fuel Performance Meeting, San Francisco, CA,
Sheindlin, M.; Ott, L. Present status of the MOX version Sept 30 to Oct 3, 2007; American Nuclear Society:
of the TRANSURANUS code (F2.5). In Enlarged Halden San Francisco, CA, 2007; pp 261–273.
Programme Group Meeting on High Burn-up Fuel 156. Baron, D.; Masson, R.; Gatt, J.; Spino, J.; Laux, D.
Performance, Safety and Reliability, Sandefjord, Norway, Evolution of the nuclear fuel mechanical properties with
May 9–14, 2004. burn-up. An extensive European experimental program.
142. Status and Advances in MOX Fuel Technology; IAEA STI/ In International Topical Meeting on Light Water Reactor
DOC/010/415; 2003. Fuel Performance, Kyoto, Japan, 2005.
3.20 Modeling of Fission-Gas-Induced Swelling of
Nuclear Fuels*
J. Rest
Argonne National Laboratory, Argonne, IL, USA
579
580 Modeling of Fission-Gas-Induced Swelling of Nuclear Fuels
RERTR Reduced Enrichment for Research and motion. All aspects of this synergistic process need to
Test Reactors be accounted for and modeled correctly in order to
SEM Scanning electron microscope obtain a model that can accurately predict fission-gas
TEM Transmission electron microscope release.
On a different level, below temperatures at which
defect annealing occurs, at relatively high doses, fuel
materials such as UO2 and uranium alloys such as
3.20.1 Introduction U–10Mo undergo irradiation-induced recrystalliza-
tion wherein the as-fabricated micron-size polycrys-
This chapter addresses various aspects of modeling talline grains are transformed to submicron-sized
fission-gas-induced swelling in both oxide and metal grains. As a result of this transformation, fission
fuels. The underlying theme underscores the simila- gases are moved from within the grain to the grain
rities and differences in gas behavior between these boundaries, transferring the materials response to gas-
two classes of nuclear materials. The discussion focuses driven swelling from intragranular to intergranular. In
more on a description of key mechanisms than on a addition, gas-bubble/precipitate complexes can act as
comparison of existing models. Three interrelated crit- pinning sites that immobilize potential recrystalliza-
ical phenomena that dominate fission-gas behavior tion nuclei, and thus affect the dose at which recrys-
are discussed: the role of intra- and intergranular tallization is initiated. The synergy between these
gas-bubble nucleation, irradiation-induced re-solution, different forces needs to be realistically captured in
and irradiation-induced recrystallization on gas-driven order to accurately model the phenomena.
swelling in these materials. The results of calculations Given the current uncertainties in materials prop-
are compared to experimental observations. erties, critical parameters, and proposed behavioral
A clarifying comparison of existing models is mechanisms, a key issue in modeling of fission-gas
clouded by the fact that many of the models employ behavior in nuclear fuels is realistic validation. In
different values for critical parameters and materials general, most of the model validation is accomplished
properties. This condition is fueled by the difficulty by adjusting/predicting these properties and para-
in measuring these quantities in a multivariate irra- meters to achieve agreement with measured gas
diation environment. Examples of such properties release and swelling, and with mean values of the
are gas-atom and bubble diffusion coefficients, bub- bubble-size distribution. However, the uncertainties
ble nucleation rates, re-solution rate, surface energy, in these properties and parameters generate an inher-
defect formation and migration enthalpies, creep ent uncertainty in the validity of the underlying
rates, and so on. physics and the physical reality of proposed behav-
The behavior of fission gases in a nuclear fuel is ioral mechanisms. This inherent uncertainty clouds
intimately tied to the chemical and microstructural the predictive aspects of any mechanistic approach to
evolution of the material. The complexity of the describe the phenomena. Thus, more detailed data
phenomena escalates when one considers the possi- are required to help clarify these issues.
bility that microstructure is dependent on the fuel The shape of bubble-size distribution data con-
chemistry. Some of the key behavioral mechanisms, tains information on the nature of the behavioral
such as gas-bubble nucleation, are affected by fuel mechanisms underlying the observed phenomena
microstructure. Likewise, mechanisms such as the that are not present in the mean or average values
diffusion of gas atoms and irradiation-produced of the distribution. This is due to information
defects are affected by fuel chemistry. Thus, a realis- contained in the first and second derivates of the
tic description of the phenomena entails an accurate bubble density with respect to bubble size. Literature
representation of the evolving fuel chemistry and descriptions of measured intragranular bubble-size
microstructure. A simple example of this is the distributions are few and far between, and measured
dependence of fission-gas release on the grain size: intergranular bubble distributions are all but non-
the larger the grains, the lower the fractional release existent. In Sections 3.20.2 and 3.20.3, recently
at a given dose. On the other hand, grain growth measured intra- and intergranular bubble-size distri-
occurs as a result of time at temperature as well as butions obtained from U–Mo alloy fuel are used
by irradiation effects and fuel chemistry (e.g., stoichi- for model validation, and the robustness of this
ometry). As the grain boundaries move, they encoun- technique in reducing uncertainties in proposed
ter fission products and gas bubbles that impede their mechanisms and materials properties as compared
Modeling of Fission-Gas-Induced Swelling of Nuclear Fuels 581
to employing average values is underscored. In this recrystallization story is clarified. Calculations are
regard, it will be shown in Section 3.20.3 that a compared to data for the dislocation density and
substantial increase in validation leverage is secured change in lattice displacement in UO2 as a function
with the use of bubble-size distributions compared of burnup. In addition, calculations are compared to
with the use of mean values. The results of a series of available data for the recrystallized grain-size distri-
calculations made with paired values of critical para- bution in UO2 and in U–10Mo.
meters, chosen such that the calculation of average Models such as those described in this chapter are
quantities remains unchanged, demonstrate that the stories that remain just stories until validated by
calculated distribution undergoes significant changes experiment. Fission-gas behavior in nuclear fuels
in shape as well as position and height of the peak. As has been studied since the early 1950s, and although
such, a capacity to calculate bubble-size distributions almost 60 years have elapsed, a definitive picture of
along with the availability of measured distributions these phenomena is still unavailable today. The rea-
goes a long way in validating not only values of key son, as stated above, is the difficulty in obtaining
materials properties and model parameters, but also reliable single-effects data in a multivariate irradia-
proposed fuel behavioral mechanisms. tion environment, coupled with the highly synergis-
Sections 3.20.2 and 3.20.3 contain discussions of tic nature of the beast.
gas-bubble nucleation in the high-temperature
equilibrium g-phase, and in the low-temperature
irradiation-stabilized g-phase of uranium alloy fuel, 3.20.2 Intragranular Bubble
respectively. The connection between these regimes Nucleation: Uranium-Alloy Fuel in
is that while intergranular multiatom nucleation the High-Temperature
appears to dominate at low temperature, intragranu- Equilibrium g-Phase
lar multiatom nucleation is the dominant nucleation
3.20.2.1 Introduction
mechanism at high temperature. Although the discus-
sion on gas-bubble nucleation focuses on uranium Figure 1 shows scanning electron microscope
alloys (because of the availability of measured bub- (SEM) micrograph of g-U–Zr–Pu alloy fuel.1 The
ble-size distributions), there is no reason to believe microstructure shown in Figure 1 is typical of most
that they would not be applicable to oxide fuel as well.
Section 3.20.4 presents an analysis of irradiation-
induced re-solution. Specifically, the analysis pre-
sents a rationale for why gas-atom re-solution from
grain-boundary bubbles is a relatively weak effect as
compared to that for intragranular bubbles. One of
the arguments is that intergranular bubble nucleation
results in bubble densities that are far smaller than
observed in the bulk material. For example, an inter-
granular bubble density of 1 1013 m2 is equivalent
to a bubble density of 2 1018 m3 for a grain size of
5 106 m. This is to be compared to observed intra-
granular bubble densities that are on the order of
1023 m3. In addition, typical intergranular bubble
sizes of tenths of a micron are to be compared to
nanometer-sized intragranular bubbles. This consid-
eration is supported not only by the experimental
results presented in Section 3.20.3, but also by the
results of the multiatom nucleation theory that form
the basis of the analysis.
Finally, in Section 3.20.5, models for the initia-
tion and progression of irradiation-induced recrys- 10.0 mm
tallization are reviewed, and a theory for the size of Figure 1 Scanning electron microscope micrograph of
the recrystallized grains is discussed. The role of g-U–Zr–Pu alloy fuel. Reproduced by permission of the
bubble nucleation and gas-atom re-solution in the experimentor, G. L. Hofman.
582 Modeling of Fission-Gas-Induced Swelling of Nuclear Fuels
uranium metal alloys irradiated in the equilibrium respectively, the change in the density of gas in intra-
g-phase. Swelling of this material is predominantly granular bubbles because of bubble nucleation, gas-
due to the growth of fission-gas bubbles. Its fission- atom diffusion to bubbles of radius, rb , and the loss of
gas behavior is characterized by high mobility at gas atoms from bubbles because of irradiation-
relatively high temperatures at which it exists at the induced re-solution.
equilibrium g-U–Zr–Pu phase. As seen in Figure 1, An implicit assumption in eqn [1] is that once a
the bubbles in this material comprise a relatively two-atom nucleus forms, it grows instantaneously to
broad size range. Some of the larger bubbles have a an m-atom bubble. Values of fn ranging from 107 to
sinuous plastic-like appearance, indicative of high 102 have been proposed, which makes the nucle-
mobility. A number of coalescence events are appar- ation factor little more than an adjustable parameter.5
ent, and some of the larger bubbles appear to be A substantial contribution to the spread of reported
growing into the smaller neighboring bubbles. values for fn is that most models describe the time
Most attempts at describing intragranular evolution of mean values of cb and rb which are
gas-bubble nucleation in nuclear fuels at higher compared to the respective mean values of the mea-
temperatures have relied on a homogeneous2 or sured quantities (comparing model predictions with
heterogeneous3 two-atom mechanism. In general, it average quantities is by far the dominant validation
is assumed that two atoms that come together in the technique reported in the literature). In this regard,
presence of vacancies or vacancy clusters become a as will be demonstrated in the following section,
stable nucleus. At lower temperatures, because of the the use of bubble-size distributions goes a long way
relatively strong effect of irradiation-induced re- toward the reduction of such uncertainties.6
solution, the number of nucleated bubbles increases As an approach to circumventing the deficiencies
due to the increase in the effective gas generation thus described, in what follows a multiatom bubble
rate.4 In theory, the number of nuclei will increase nucleation mechanism is proposed and implemented
until newly created gas atoms are more likely to be into a mechanistic calculation of the intragranular
captured by an existing nucleus than to meet other fission-gas bubble-size distribution. The results of
gas atoms and form new nuclei.2 In practice, because the calculations are compared to a measured bubble-
of the coarsening of the bubble-size distribution, the size distribution in U–10Mo irradiated at relatively
two-atom nucleation process continues throughout high temperature to 4% U-atom burnup. The multi-
the irradiation. atom nucleation model is compared to the two-atom
If both bubble motion and coalescence are model within the context of the data, and the impli-
neglected, the rate equation describing the time evo- cations of each mechanism for the observable quan-
lution of the density of gas in intragranular bubbles is tities are discussed.
given by In the next section, a multiatom nucleation mech-
anism is formulated. Section 3.20.3 presents an out-
d½mb ðt Þcb ðt Þ
¼ 16pfn Dg rg cg ðt Þcg ðt Þ line for a calculation of the time evolution of the
dt bubble-size distribution. In Section 3.20.4, a discus-
þ 4prb ðt ÞDg cg ðt Þcb ðt Þ bmb ðt Þcb ðt Þ ½1 sion is presented of processes that lead to coarsening
of the as-nucleated bubble distribution. In Section
where cg , cb are the densities of gas atoms and bub- 3.20.5, model calculations are used to interpret a
bles, respectively, mb is the average number of gas measured distribution in U–8Mo uranium alloy fuel
atoms per bubble, Dg is the gas-atom diffusion coeffi- irradiated to 4% U burnup at 850 K. In addition, in
cient, b is the gas-atom re-solution rate from bubbles, this section, a comparison between the multiatom
and fn , the so-called nucleation factor, is the proba- and two-atom nucleation mechanisms is attempted.
bility that two gas atoms that come together stick Finally, conclusions are presented in Section 3.20.6.
long enough to form a stable bubble nucleus. Often,
fn is interpreted as the probability that there are
3.20.2.2 A Multiatom Nucleation
sufficient vacancies or vacancy clusters in the vicinity
Mechanism
of the two-atom to form a stable nucleus. For exam-
ple, for heterogeneous bubble nucleation along fis- Fission gases Xe and Kr are generated in a nuclear
sion tracks in UO2, fn is approximately the average fuel at a rate of about 0.25–0.30 atoms per fission as a
volume fraction of fission tracks 104. The three result of decay of the primary fission products. About
terms on the right-hand-side of eqn [1] represent, seven times more Xe is produced than Kr. These gas
Modeling of Fission-Gas-Induced Swelling of Nuclear Fuels 583
atoms are very insoluble in the fuel in that they do where cgcrit is the concentration of gas at the solubility
not react chemically with any other species. Thus, limit, bv is the volume per atom (van der Waals con-
left in the interstices, because of their relatively large stant), and cb ðr 0 Þ is the concentration of bubble nuclei
size, they produce a strain in the material. In order to at the unrelaxed radius r 0, that is, the initial stage of
lower the energy of the system and to minimize the bubble nucleation is a volume-conserving process.
strain, the gas atoms tend to relocate in areas of Subsequently, in order to lower the free energy of the
decreased density, such as in vacancies and/or system, the overpressurized nuclei relax by absorbing
vacancy clusters. For example, in UO2, gas atoms vacancies until the bubbles reach equilibrium. At equi-
have been calculated to sit in neutral trivacancy librium, the bubble radius is r and, in the absence of
sites consisting of two oxygen ions and one uranium significant external stress, the pressure in the bubble is
ion.7 Given enough energy via thermal fluctuations, given by
and/or via irradiation, the gas atoms can hop ran- 2g
domly from one site to another and thus diffuse Pe ¼ ½3
r
through the material. The gas atom/vacancy com-
plexes can combine forming clusters of gas atoms where g is the surface energy per unit area. If it is
and vacancies. If enough gas atoms come together, assumed that the average gas bubble size r 0 is a function
they become transformed into a gas bubble which, of the equilibrium bubble size, then differentiating eqn
under equilibrium conditions, sits in a strain-free [2] with respect to the equilibrium radius r and rear-
environment. This process of forming gas bubbles is ranging terms yields
termed gas-bubble nucleation. 1 dcb ðr 0 Þ 1 dmðr 0 Þ 3 dr 0
¼ 0 ½4
According to phase transition theory, at relatively cb ðr Þ dr
0 mðr 0 Þ dr r dr
large supersaturations, a system transforms not by
atom-to-atom growth, but simultaneously as a Let us assume that during the relaxation phase
whole. In other words, the system is unstable against there is no interaction between the nucleated bubbles,
transformations into a low free energy state, and the that is,
new phase will have a certain radius defined by the r 0 ! r ; mðr 0 Þ ! mðr Þ ¼ mðr 0 Þ; cb ðr 0 Þ ! cb ðr Þ ¼ cb ðr 0 Þ
supersaturation. Solubility of rare-gas atoms in ura-
½5
nium alloys or ceramics is so low that it has not been
measured. In perfect crystals, the order of magnitude The nucleation problem thus consists of determining
of the solubility has been estimated to be 1010.8 This the two terms on the RHS of eqn [4]. The first term on
figure may be increased up to 105 in the vicinity of the RHS of eqn [4] can be determined from the equa-
dislocations. In addition, there may be a substantial tion of state (EOS), the capillarity relation, and the
effect from gas in dynamic solution, that is, as a result conditions expressed in eqn [5]. Using the van der
of irradiation-induced re-solution. Thus, in regions Waals EOS,
of nuclear fuels that are near irradiation-produced
defects and/or various microstructural irregularities, PðV mbv Þ ¼ mkT ½6
the solubility of the gas can be substantially higher where V ¼ 4/3pr 0 3 is the bubble volume. Recognizing
than in the bulk material. The gas concentration in that at nucleation the bubble size is small such that
these regions will increase until the solubility limit is 2g/r 0 s , where s is the external stress, and differ-
reached, whereupon the gas will precipitate into entiating eqn [6] with respect to the equilibrium
bubbles. Subsequently, nucleation is limited because radius r one obtains
of the gas concentration in solution falling below
the solubility limit. The trapping of the gas by the 1 dmðr 0 Þ 3 rkT
¼ 1 ½7
nucleated bubble distribution damps the increase mðr 0 Þ dr r 3ðrkT þ 2gbv Þ
in gas concentration. Eventually, the gas in solution
The remaining term on the RHS of eqn [4] can
may reach the solubility limit at which time the nucle-
be determined by invoking energy minimization as
ation event repeats. Thus, assuming that all the gas
the driving force for bubble equilibration. The change
precipitates into bubbles of equal size r 0, the concen-
in the Gibbs free energy due to bubble expansion is
tration of gas in the bubble at nucleation is given by
given by
bv cgcrit 4
mðr 0 Þ ¼ ½2 G ¼ pr 03 Gv þ 4pr 02 g ½8
4=3pr 03 cb ðr 0 Þ 3
584 Modeling of Fission-Gas-Induced Swelling of Nuclear Fuels
where Gv is the free energy driving bubble equilibra- Subsequent to the nucleation event, the as-
tion, which, in analogy with the treatment of the nucle- nucleated bubble-size distribution evolves under the
ation of liquid droplets in a vapor,9 can be expressed as driving forces of gas diffusion to bubbles, gas-atom
kT re-solution from bubbles, and bubble coalescence
Gv ¼ lnðPðr Þe =Pðr 0 ÞÞ ½9 due to bubble–bubble interaction via bubble motion
O
and geometrical contact. As stated earlier, additional
where O is the atomic volume. The critical bubble nucleation events are delayed because of the gas in
radius at equilibrium is given by the condition solution remaining below the solubility limit, as the
@G 2g gas generated by continuing fission events is trapped
¼ 0 ! r ¼ rcrit ¼ ½10 within the existing bubble-size distribution. This last
@r Gv
point is facilitated by the relatively high gas-atom dif-
Inserting the expressions for Pe and P from eqns [3] and
fusivities at the temperatures of interest (i.e., those
[6], respectively, into eqn [10], differentiating with
under which the equilibrium g-phase of the alloy
respect to the bubble radius r, and applying a little
exists). Eventually, the gas in solution may again
algebra results in
reach the solubility limit at which time the nucle-
dr 0 1 O kT ation event repeats.
4pr 02 ¼ þ ½11
dr X r 2g
where 3.20.2.3 Calculation of the Fission-Gas
Bubble-Size Distribution
kTr
X¼ ½12
4 03 The model consists of a set of coupled nonlinear dif-
2g pr mbv ferential equations for the intragranular concentration
3
of fission product atoms and gas bubbles of the form10
Making use of eqn [2] in eqn [11] results in
dCi
3 dr 0 m2 c b 2gO=rkT rkT dm kT O bv dm ¼ ai Ci Ci bi Ci þ ci ði ¼ 1; . . . ; N Þ ½15
¼ e þ þ dt
r 0 dr bv c g 2gm dr 2g r m dr
where Ci is the number of bubbles in the ith size class
½13 per unit volume; and the coefficients ai , bi , and ci obey
Finally, substituting eqn [5], eqn [7], and eqn [13] into functional relationships of the form
eqn [4] yields ai ¼ ai ðCi Þ
1 dcb 1 dm mcb
¼ þ bi ¼ bi ðC1 ; . . . ; Ci1 ; Ciþ1 ; . . . ; CN Þ
cb dr m dr bv cg
The variables in eqn [15] are defined in Table 1.
rkT dm mkT mO dm
e2gO=rkT þ þ bv ½14 ai represents the rate at which bubbles are lost from
2gm dr 2g r dr
(grow out of) the ith size class because of coalescence
The as-nucleated bubble-size distribution is then with bubbles in that class; bi represents the rate at
obtained by the simultaneous solution of eqns [7] which bubbles are lost from the ith size class because
and [14]. of coalescence with bubbles in other size classes and
dCi
Table 1 Definition of variables in eqn [15], ¼ ai Ci Ci bi Ci þ ci ði ¼ 1; . . . ; NÞ
dt
i Ci aiCiCi biCi ci
1 Concentration of Rate of gas atom loss Rate of gas atom loss Rate of gas atom gain due to
intragranular gas atoms due to gas-bubble due to diffusion into atom re-solution and fission
nucleation gas bubbles of uranium nuclei
2,. . .,N Concentration of Rate of gas bubble loss Rate of gas bubble Rate of gas bubble gain due
intragranular gas bubbles due to bubble loss due to to bubble nucleation and
coalescence with coalescence with coalescence, and diffusion
bubbles within the bubbles in other of gas atoms into bubbles
same size class size classes
Thus, the probability that a coalescence between an Equations [25] and [26] are proportional to the prob-
i-bubble and a j-bubble will result in a k bubble is abilities that any particular i-bubble will become an
586 Modeling of Fission-Gas-Induced Swelling of Nuclear Fuels
i þ 1 or an i 1 bubble, respectively; the ratio of and s is the width of the distribution that charac-
the probabilities is equal to the ratio of the rates. terizes divergences from spherical bubbles and the
The aforementioned definition of Nik and Nik 0 is uniform distribution assumption. In principle, s is a
consistent with the conservation of the total number measurable parameter.
of gas atoms. The ai and bi coefficients in eqn [28] now have an
additional term given by
3.20.2.4 Bubble Coalescence gc
ai ¼ 4pDvol Ri Pii
The bubbles are assumed to diffuse randomly gc
X
bi ¼ 4pDvol Ri Cj Pij ½32
through the solid alloy by a volume diffusion mecha-
j 6¼i
nism. The bubble diffusion coefficient Di of a bubble
having radius Ri is given by where Pij is given by eqn [30].
In what follows, it is assumed that DXe ¼ Dvol .
3a03 Dvol
Di ¼ ½27
4pR3i 3.20.2.5 Analysis of U–10Mo
where a0 is the lattice constant and Dvol is the volume High-Temperature Irradiation Data
self-diffusion coefficient of the most mobile species
Figure 2 shows the as-nucleated bubble-size distribu-
in the alloy. The coefficients ai and bi (e.g., the first
tion made with the simultaneous solution of eqns [7]
and second terms on the RHS of eqn [15]) are repre-
and [14] for a gas solubility of 107 at a fuel temperature
sented, respectively, by
X of 850 K. At a fission rate of 1 1020 fissions m3 s1,
ai ¼ 16pRi Di ; bi ¼ ðRi þ Rj ÞðDi þ Dj ÞCj ½28 the solubility limit is reached in 140 s. Subsequently,
j 6¼i nucleation is limited as a result of the gas concentration
in solution falling below the solubility limit. The trap-
The interaction cross-section represented in eqn [28]
ping of gas in solution by the nucleated gas bubbles
is based on an analysis of colloidal suspensions within
damps the rate at which the generated gas increases the
the framework of the continuum theory.11 Fission-gas
gas concentration in dynamic solution. It is important to
bubbles can also interact due to mobility from biased
point out that here the solubility limit is an unknown
motion within a temperature gradient. This aspect of
parameter. If the solubility limit was 106 or 105, the
the problem is handled in an analogous manner and
initial bubble nucleation event would occur after 1400
will not be considered here.
or 14 000 s of irradiation, respectively.
As the bubbles grow and interact, the average
Figure 3 shows m versus r obtained from the
spacing between bubbles shrinks. In addition, as
solution of eqn [7] for T ¼ 850 K and g ¼ 0:5 Jm2 .
seen from eqn [27] for the volume diffusion mecha-
As expected from the form of eqn [7], the number
nism, bubble mobility falls off as the inverse of the
of gas atoms grows exponentially with bubble size.
radius cubed such that, for all practical purposes,
Figure 4 shows the amount of gas in bubbles as a
relatively large bubbles are immobile. As the larger
function of bubble size corresponding to Figures 2
bubbles grow because of accumulation of the contin-
and 3. As is evident from Figure 3, although the
ual production of gas due to fission, the bubbles
bubble-size distribution shown in Figure 2 is rela-
intercept other bubbles and coalesce. This process is
tively broad, the majority of the gas generated prior
here termed geometrical coalescence. For spherical
to the nucleation event (i.e., within the first 140 s of
bubbles that are all the same size and that are uni-
irradiation) exists in bubbles having radii <1 nm. As
formly distributed, contact is reached when
discussed earlier, subsequent to the multiatom bubble
2Rb ð2Cb =3Þ1=3 ¼ 1 ½29 nucleation event, the concentration of gas in solution
stays below the solubility limit due to the trapping
In analogy with percolation theory, the probability of
effect of the nucleated gas bubbles such that addi-
an i-bubble contacting a j-bubble is given by
h pffiffiffiffiffiffi i tional multiatom nucleation events are delayed.
gc
Pij ¼ 0:5 1 erf 1 Rij Cij
1=3
0:5=s ½30 Thus, until the solubility limit is again exceeded, for
the situation shown in Figures 2–4, for irradiation
where times >140 s, the bubble distribution follows from the
1=3 evolution of the as-nucleated distribution shown in
1=3 2 Figure 2 because of bubble–bubble coalescence and
Rij ¼ Ri þ Rj ; Cij ¼ ðCi þ Cj Þ ½31
3 diffusion of generated gas to the existing bubble
Modeling of Fission-Gas-Induced Swelling of Nuclear Fuels 587
1E + 21
1E + 20
1E + 19
1E + 18
1E + 17
Cb (m−3)
1E + 16
1E + 15
1E + 14
1E + 13
1E + 12
1E + 11
0 2 4 6 8 10 12
r (nm)
Figure 2 As-nucleated bubble-size distribution made with the simultaneous solution of eqns [7] and [14] for a gas
solubility of 107. Reproduced from Rest, J. J. Nucl. Mater. 2010, 402(2–3), 179–185.
50 000
40 000
30 000
m
20 000
10 000
0 2 4 6 8 10 12
r (nm)
Figure 3 Number of gas atoms in a freshly nucleated bubble versus bubble radius corresponding to Figure 1.
Reproduced from Rest, J. J. Nucl. Mater. 2010, 402(2–3), 179–185.
population. When the solubility limit is again U-atom burnup using eqn [15] and the multiatom
exceeded, additional nucleation events occur within nucleation model described in Section 3.20.2 for
the evolving bubble population, and this complex of three values of the rare-gas solubility. The calcula-
bubbles again evolves under the driving forces of tions shown in Figure 5 were made using a gas-atom
bubble coalescence, gas-atom diffusion to, and gas- diffusivity, and re-solution rate given by
atom re-solution from bubbles.
Dvol ¼ 2 104 e33000=kT cm2 s1
Figure 5 shows the calculated bubble-size distri-
b ¼ 1 1018 f_ ½33
bution for an irradiation in U–8Mo at 850 K to 4%
588 Modeling of Fission-Gas-Induced Swelling of Nuclear Fuels
0.12
0.12
0.10 0.10
0.06 0.04
0.02
0.04 0.00
0.02 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0
0.00
0 2 4 6 8 10 12
r (nm)
Figure 4 Fraction of generated gas in bubbles versus bubble radius corresponding to Figures 2 and 3.
Reproduced from Rest, J. J. Nucl. Mater. 2010, 402(2–3), 179–185.
1E + 20
U–8Mo data
1E + 19 Multiatom nucleation
solubility limit = 2.5 ⫻ 10−9
1E + 18
Multiatom nucleation
solubility limit = 2.5 ⫻ 10−8
1E + 17
Multiatom nucleation
1E + 16 solubility limit = 2.5 ⫻ 10−7
Cb (r) (m−3)
1E + 15
1E + 14
1E + 13
1E + 12
1E + 11
1E + 10
0 2 4 6 8 10 12 14
Bubble radius (µm)
Figure 5 Calculated bubble-size distributions for an irradiation in U–8Mo at 850 K to 4% U-atom burnup using eqn [15]
and the multiatom nucleation model described in Section 72.2 for three values of the rare-gas solubility compared with
irradiation data. Reproduced from Rest, J. J. Nucl. Mater. 2010, 402(2–3), 179–185.
where f_ is the fission rate. The value for Dvol given in U–10Mo is about an order of magnitude less than
eqn [33] is about a factor of 10 less than the out-of-pile the U self-diffusion coefficient.13 In addition, it is
measured U self-diffusion coefficient in U–10Mo.12 not at all clear how these diffusion couple measure-
On the other hand, it is not clear what diffusion ments extrapolate to lower temperatures (lowest dif-
mechanism dominates gas behavior in these alloys. fusion couple temperature was 1073 K) and to an
For example, the Mo self-diffusion coefficient in irradiation environment. The value for b is
Modeling of Fission-Gas-Induced Swelling of Nuclear Fuels 589
approximately an order of magnitude less than esti- nucleation as expressed by the first term on the
mated for UO2.14 This value is consistent with esti- RHS of eqn [1]. Figure 7 shows the calculated bub-
mated irradiation-enhanced creep rates in U–10Mo, ble-size distributions for an irradiation in U–8Mo at
which are approximately an order of magnitude less 850 K to 4% U-atom burnup using eqn [15] and the
than for UO2.15 These effects can be traced to a higher two-atom nucleation model for three values of the
thermal conductivity in the metal alloy as compared nucleation factor compared with irradiation data.
to the metal oxide. Also shown are results for two different values of
The irradiation data shown in Figure 5 were the volume diffusion coefficient for fn ¼ 103. It is
converted to a volume density from the measured clear from Figure 7 that the two-atom nucleation
areal density16 using the Saltykov method.17 The model does not satisfactorily interpret the measured
error bars associated with the solid circle data points bubble-size distribution over a 6 orders of magnitude
are unknown, but they are most certainly substantial range in fn and 2 orders of magnitude range in Dvol .
for the smaller bubble sizes where undercounting Thus, comparing Figures 5 and 7, the multiatom
errors are typical. In addition, the fuel experienced nucleation model provides a better interpretation of
an end-of-life constraint of 10 mp (the effect of the data than the two-atom model. This becomes a
hydrostatic constraint on bubble size is included stronger statement when the relative insensitivity of
in the calculations). Given these uncertainties, the the calculated tail of the distribution to the value
bubble-size distribution is relatively flat for bubbles of the nucleation factor and the volume diffusion
having radii from 5 to 12 mm. As shown in Figure 4, coefficient for the two-atom model are compared
a solubility of 2.5 108 provides a plausible inter- to the ‘bracketing’ of the data by commensurate
pretation of the data. changes in solubility and diffusion coefficient for the
Figure 6 shows the dependence of the calculated multiatom model.
bubble-size distribution on the value of Dvol for a gas A more definitive differentiation between these
solubility of 2.5 108 compared with the measured two models requires data at a much lower burnup
quantities. As seen from Figure 6, not surprisingly, where the effects of bubble diffusion and coalescence
the value of Dvol has a reasonably strong effect on the are minimal. Unfortunately, such data are currently
calculated distribution. unavailable. Figure 8 shows a comparison of multi-
It is of interest to compare the multiatom atom and two-atom nucleation mechanisms for an
nucleation model with conventional two-atom irradiation to 0.04% U burnup of U–8Mo fuel at
1E + 20
U–8Mo data
1E + 19
Dvol = 6.5 ⫻ 10−20 m2 s−1
1E + 18 Dvol = 6.5 ⫻ 10−19 m2 s−1
Dvol = 6.5 ⫻ 10−18 m2 s−1
1E + 17
1E + 16
Cb (r) (m−3)
1E + 15
1E + 14
1E + 13
1E + 12
1E + 11
1E + 10
0 2 4 6 8 10 12 14
Bubble radius (µm)
Figure 6 Dependence of the calculated bubble-size distribution on the value of Dvol for a gas solubility of 2.5 108
compared with the measured quantities. Reproduced from Rest, J. J. Nucl. Mater. 2010, 402(2–3), 179–185.
590 Modeling of Fission-Gas-Induced Swelling of Nuclear Fuels
1E + 20
U–8Mo data
1E + 19 fn = 10-6
fn = 10-3
1E + 18
fn = 1
1E + 17 fn = 10-3
Dvol = 6.5 ⫻ 10-18 m2 s-1
Cb (r) (m-3)
1E + 16 fn = 10-3
Dvol = 6.5 ⫻ 10-20 m2 s-1
1E + 15
1E + 14
1E + 13
1E + 12
1E + 11
0 2 4 6 8 10 12 14
Bubble radius (mm)
Figure 7 Calculated bubble-size distributions for an irradiation in U–8Mo at 850 K to 4% U-atom burnup using eqn [15] and
the two-atom nucleation model for three values of the nucleation factor compared with irradiation data. Also shown are
results for two different values of the volume diffusion coefficient for fn ¼ 103 . Reproduced from Rest, J. J. Nucl. Mater.
2010, 402(2–3), 179–185.
1E + 21
1E + 20 Multiatom nucleation
Solubility limit = 2.5 ⫻ 10-8
1E + 19
Diatom nucleation
1E + 18 fn = 10-3
1E + 17
1E + 16
1E + 15
Cb (r) (m-3)
1E + 14
1E + 13
1E + 12
1E + 11
1E + 10
1E + 9
1E + 8
1E + 7
1E + 6
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4
Bubble radius (mm)
Figure 8 Comparison of multiatom and two-atom nucleation mechanism for an irradiation to 0.04% U burnup in U–8Mo
fuel at 850 K. Reproduced from Rest, J. J. Nucl. Mater. 2010, 402(2–3), 179–185.
850 K. As shown in Figure 8, the two-atom nucle- nucleation models. It is anticipated that low burnup
ation model leads to a substantially broader distribu- bubble distribution data will become available in the
tion than the multiatom model. This feature is carried relatively near future.18 Once this data become avail-
on to high burnup and, on comparing Figures 5 and 7, able, a more definitive differentiation between these
is one of the key differences between these two two models can be undertaken.
Modeling of Fission-Gas-Induced Swelling of Nuclear Fuels 591
The rate equation describing the time evolution of for swelling. If obcb was not included in eqn [35],
the density of gas in intragranular bubbles is given by then the density of bubbles could never decrease as
a result of irradiation. Likewise, if ð1 oÞbmb was
d½mb ðt Þcb ðt Þ
¼ 16pfn Dg rg cg ðt Þcg ðt Þ not included in eqn [36], the number of atoms in a
dt bubble could never decrease. However, the partition
þ 4pr b ðt ÞDg cg ðt Þc b ðt Þ bmb ðt Þcb ðt Þ ½34 of gas-atom re-solution between these two mechan-
isms, where o is the partitioning fraction, is an
The three terms on the RHS of eqn [34] represent,
assumption that remains to be tested experimentally.
respectively, the change in the density of gas in intra-
In what follows, equal partition is also assumed, that
granular bubbles due to bubble nucleation, the gas-atom
is, o ¼ 1=2.
diffusion to bubbles of radius r b , and the loss of gas
Because of the strong effect of irradiation-induced
atoms from bubbles because of irradiation-induced re-
gas-atom re-solution, in the absence of geometric con-
solution. Equation [34] can also be represented as the
tact, the bubbles stay in the nanometer size range. The
sum of two equations denoting, respectively, the time
density of bubbles increases rapidly early in the irra-
evolution of the fission-gas bubble density cb and of the
diation. At longer times, the increase in bubble con-
gas content in bubbles mb as follows:
centration occurs at a much-reduced rate. On the basis
dc b ðt Þ 16pfn Dg rg cg ðt Þcg ðt Þ of the above considerations, a quasi steady-state solu-
¼ obc b ðt Þ ½35
dt mb ðt Þ tion for the average bubble density cb and the average
number of gas atoms per bubble mb as a function of
dmb
¼ 4pr b ðt ÞDg cg ðt Þ ð1 oÞbmb ðt Þ ½36 the density of gas in solution cg and the gas-atom
dt radius rg is given by Spino et al.22
In eqn [35], fn is the bubble nucleation factor, and cg and 16pfn rg Dg cg2
rg are the gas-atom concentration and radius, respec- cb ¼ ½37
tively. In general, the value of fn is less than 1 reflecting bmb ðt Þ
the premise that gas-bubble nucleation within the fuel 1=2
matrix requires the presence of vacancies/vacancy 3bv 4pDg cg ðt Þ 3=2
mb ðt Þ ¼ ½38
clusters in order to become viable. The value of fn is 4p b
estimated on the basis of the hypothesis that gas-atom In eqn [37], fn is the bubble nucleation factor, and in
diffusion occurs by a vacancy mechanism and that a eqn [38], bv is the van der Waals constant. In general,
three gas-atom cluster is a stable nucleus. In this case, the value of fn is less than 1 reflecting the premise that
fn is approximately the bulk vacancy concentration gas-bubble nucleation within the fuel matrix requires
(i.e., 104). the presence of vacancies/vacancy clusters in order to
The first term on the RHS of eqn [35] can be become viable. The average bubble radius rb is related
interpreted to represent the generation rate of ‘aver- to mb through the gas law and the capillarity relation.
age’ size bubbles of radius r b . For every two-atom Imposing gas-atom conservation, that is, requiring that
bubble that is nucleated, 2=mb of a bubble of radius r b the sum of the gas in solution, in intragranular bubbles,
appears. In other words, nucleation of mb two-atom and on the grain boundary is equal to the amount of
clusters leads to the gain of one bubble of radius rb . gas generated (there is no gas released from the U–Mo
This ‘average size’ bubble is in the peak region of the fuel), the term cg ðt Þ is determined as
bubble-size distribution.
h i1=2
Both ‘whole’ bubble destruction and gas-atom 1 þ 1 þ 64pð1 fs Þfn rg Dg f_ bt =b
‘chipping’ from bubbles are included (last terms on cg ðt Þ ¼ ½39
RHS) in eqns [35] and [36] in order to capture the 32pfn rg Dg =b
behavior of an average size bubble (that characterizes where b is the number of gas atoms produced per
the full bubble-size distribution). Within the full fission event and fs is the fraction of gas released to
bubble-size distribution, there are bubbles that are the grain boundaries of grains of diameter dg, where,
destroyed by one fission fragment collision (e.g., bub- following Speight23
bles smaller than a critical size) and others that are
1=2
only partially damaged (e.g., bubbles larger than a 8 b 6 b
critical size). Including b in both eqns [35] and [36] fs pffiffiffi Dg t 2 Dg t ½40
pdg bþg dg b þ g
is an attempt to depict these processes using a sim-
plified formulation that enables an analytical solution where g ¼ 4pDg r b c b .
Modeling of Fission-Gas-Induced Swelling of Nuclear Fuels 593
3.20.3.3 Calculation of Evolution of gas diffusing from the grain interior removes bubbles
Average Intergranular Bubble-Size and from this size range, but these are replaced by the
Density simultaneous growth of smaller bubbles. The distri-
bution of intragranular gas consists primarily of fis-
Following the work of Wood and Kear,24 grain-
sion-gas atoms because of the strong effect of
boundary bubble nuclei of radius Rb are produced
irradiation-induced gas-atom re-solution. Bubbles
until such time that a gas atom is more likely to be
appear on the grain boundaries due to the reduced
captured by an existing nucleus than to meet another
effect of re-solution, ascribed to the strong sink-like
gas atom and form a new nucleus. An approximate
property of the boundary, as well as to the altered
result for the grain-boundary bubble concentration is
properties of bubble nucleation. In addition, nðr Þdr is
given by
affected by bubble–bubble coalescence. A differential
1=2
8zaK growth rate between bubbles of different sizes leads
Cb ¼ ½41 to a net rate of increase in the concentration of
121=3 p2 xDg d
bubbles in the size range r to r þ dr . This behavior
where a is the lattice constant, z is the number of sites is expressed by
explored per gas-atom jump, d is the width of the
boundary, x is a grain-boundary diffusion enhance- dnðr Þ d dr d dr
ment factor, and K is the flux of gas atoms per unit dr ¼ nðr Þ dr nðr Þ dr
dt dr dt d dr dt c
area of grain boundary. ½43
The intergranular bubble nucleation and growth
formulation incorporated here is on the basis of the where the subscripts d and c refer to growth by gas-
assumption that, although the effect of radiation- atom diffusion and bubble coalescence, respectively.
induced re-solution on intergranular bubble behavior The growth rate (dr =dt ) of a particular bubble is
is not negligible, a reasonable approximation can be related to the rate (dm=dt ) at which it absorbs gas
obtained by neglecting such effect in the governing from the boundary, either by diffusion of single gas
eqn. [25]. Under the above considerations, the flux K atoms, or by coalescence with another bubble. The
of atoms at the grain boundary is given by rate of growth due to gas-atom precipitation is con-
dg _ dð fs t Þ trolled by the grain-boundary gas-atom diffusion
K ¼ bf ½42
3 dt coefficient xDg and the average concentration Cg of
In general, in an irradiation environment where bub- fission gas retained by the boundary.
ble nucleation, gas-atom diffusion to bubbles, and Studies on the evolution of helium bubbles in
irradiation-induced re-solution are operative, a differ- aluminum during heavy-ion irradiation at room tem-
ential growth rate between bubbles of different size perature have shown that bubble coarsening can take
results in a peaked monomodal size distribution.25 The place by radiation-induced coalescence without bub-
position of the peak in the bubble-size distribution that ble motion.21 This coalescence is the result of the net
occurs under these conditions is defined by the bal- displacement of Al atoms out of the volume between
ance between diffusion of gas-atoms to bubbles and bubbles initially in close proximity. The resulting
irradiation-induced re-solution of atoms from bubbles. nonequilibrium-shaped bubble evolves toward a
As more gas is added to the lattice (e.g., as a result of more energetically favorable spherical shape whose
continued fission), the gas-atom diffusion flux to bub- final size is determined by the equilibrium bubble
bles increases and the peak shifts to larger bubble sizes pressure.
and decreases in amplitude, resulting in an increased Bubble coalescence without bubble motion (sput-
level of bubble swelling with increased burnup. The tering coalescence) can be understood on the basis of
model presented in this section describes the average the difference in the probability for an atom to be
behavior of this peak as a function of burnup. knocked out of the volume between a pair of bubbles
and the probability of an atom to be injected into this
interbubble volume. If the bubbles contained the
3.20.3.4 Calculation of Intergranular
same atoms as that comprising the interbubble vol-
Bubble-Size Distribution
ume, the net flux of atoms out of the interbubble
Let nðr Þdr be the number of bubbles per unit volume volume would be zero. However, as the gas bubbles
on the grain boundaries with radii in the range r to contain fission gas and not matrix atoms, the flux of
r þ dr . Growth by gas-atom collection from fission atoms into the interbubble volume is reduced by the
594 Modeling of Fission-Gas-Induced Swelling of Nuclear Fuels
bubble volume fraction, that is, the net flux out of Inserting eqns [44]–[46] into the second term
volume is equal to lV lðV VB Þ, where l is the on the RHS of eqn [43] and differentiating with
atom knock-on distance and VB is the intergranular respect to r,
bubble volume fraction. In this case, the growth rate
dnðr Þ 6 2 dnðr Þ
(dr =dt ) of a bubble being formed by the coalescence dr ¼ lds f_ pr 2 nðr Þdr þ lds f_ pr 3 dr
dt c d g d g dr
of two adjacent bubbles (and the commensurate
effective shrinkage rate of the adjacent bubbles) is ½49
related to the rate (dms =dt ) at which the interbubble Subsequent to intergranular bubble nucleation, gas
material is being sputtered away, where arriving at the boundary will be adsorbed by the
dms existing bubble population. The rate at which a
¼ bs ms ½44 grain-boundary bubble adsorbs gas is approximately
dt
given by
Using the van der Waals EOS,
dm=dt ¼ 12pr xDg Cg =dg ½50
2
dr 3ðrkT þ 2gbv Þ dm
¼ ½45 As mentioned in Section 3.20.3.3, re-solution of
dt 16pgðkTr 3 þ 3gbv r 2 Þ dt grain-boundary bubbles is not explicitly considered,
for example, in eqn [50]. The rationale for this is that
In continuation, the sputtering rate bs can be
because of the very strong sink-like nature of the
expressed as
grain boundary, gas-atoms ejected from a gas bubble
6 located on the boundary that land within the steep
bs ¼ lds f_ pr 2 ½46
dg portion of the concentration gradient are ‘sucked
back’ into the boundary and quickly reenter the
where the effective interbubble volume is assumed bubble such that the ‘effective’ re-solution rate is
to be disk-shaped with volume = ds pr 2 , and where relatively small.26
ds is the thickness of the material undergoing sput- Combining eqns [9] and [14]
tering. For a lenticular bubble with radius of curva-
ture r, the equivalent radius of a spherical bubble is 9r xDg Cg ðrkT þ 2gbv Þ2 3bv xDg Cg
dr =dt ¼ ½51
given by 4gdg ðkTr 3 þ 3gbv r 2 Þ dg r
1=3
3 1 Using the approximation on the RHS of eqn [51], the
r ¼ r 1 cos y þ cos y 2
½47
2 2 first term on the RHS of eqn [43] becomes
where
dnðr Þ 3bv xDg Cg 3bv xDg Cg dnðr Þ
ggb dr ¼ nðr Þ
cosðyÞ ¼ ½48 dt d dg r 2 dg r dr
2g ½52
and ggb is the grain-boundary energy.
It is assumed that bubble coalescence is The overall net rate of change of the concentration of
approached by the gradual erosion of the material bubbles in a given size range is given by the sum of
between the bubbles. This bubble coarsening process eqns [49] and [52]
can be visualized as lenticular intergranular bubbles
dnðr Þ 3bv xDg Cg 3bv xDg Cg dnðr Þ
separated by a distribution of solid disks. As these dr ¼ nðr Þ
dt dg r 2 dg r dr
disks are sputtered because of fission damage, the
majority of the sputtered atoms are injected into the 6 2 dnðr Þ
lds f_ pr 2 nðr Þdr lds f_ pr 3 dr ½53
adjacent bubbles, with the commensurate drawing dg dg dr
together of the bubbles until the joining process has
been completed. In order for this process to be viable, The equilibrium population of bubbles is obtained by
the gas atom knock-on distance should be sufficiently setting eqn [53] to zero
large such that the majority of atoms sputtered from 3bv xDg Cg 3bv xDg Cg dnðr Þ
the solid disk can enter the adjacent bubbles. Because nðr Þ
dg r 2 dg r dr
of the geometry of the lenticular gas bubbles and
6
solid disks, this distance will be substantially less lds f_ pr 2 nðr Þdr ¼ 0 ½54
than the interbubble spacing. dg
Modeling of Fission-Gas-Induced Swelling of Nuclear Fuels 595
where the last term in eqn [53] has been omitted will be adsorbed by the existing bubble population.
3bv xDg Cg 2 Combining eqns [45] and [58]
because lds f_ pr 3 for the conditions
dg r dg
3Cg bv ðrkT þ 2gbv Þ2
explored in this paper. ðdr =dt Þr ¼ r0 ¼
16pgð4tb Cb pr03 =3ÞðkTr 3 þ 3gbv r 2 Þ
Equation [54] must be solved subject to the rele-
vant boundary condition. In general, this boundary ½60
condition concerns the rate at which bubbles are The solution of eqn [54] subject to the boundary con-
formed at their nucleation size r0 . From a consider- dition expressed by eqns [55] and [60] is
ation of freshly nucleated bubbles25
64gCb2 p2 r 3 ðkTr 3 þ 3gbv r 2 Þexp½kðr 4 r04 Þ
Cb nðr Þ ¼
nðr0 Þdr ¼ dr =ðdr =dt Þr ¼r0 ½55 3bv Cg dg ðrkT þ 2gbv Þ2
tb ½61
The rate of bubble nucleation is provided by the where
Wood–Kear nucleation mechanism24 where on the
grain boundary the average time tb for a gas atom p f_ lds
k¼ ½62
to diffuse to an existing bubble (as discussed above 2bv xDg Cg
this is the time at which bubble nucleation would
essentially cease) is given by 3.20.3.5 Comparison Between Model
1 Calculations and Intragranular Data
tb ¼ ½56
pxDg Cb One of the major challenges in the field of fission-gas
behavior in nuclear fuels is the quantification of
Thus, from eqn [20], it follows that the bubble nucle-
critical materials properties. There is a direct corre-
ation rate is given by
lation between the accuracy of the values of critical
dCb Cb properties and the confidence level that the proposed
¼ ½57
dt tb underlying physics is realistic.
where is a proportionality constant that is deter- The values of the key parameters used in the
mined by imposing the conservation of gas atoms. model are given in Table 2. Many of them are
The observed grain-boundary bubbles are a combi- known or estimated from the literature27; the values
nation of lenticular-shaped objects whose size is sub-
stantially larger than the estimated thickness of the
Table 2 Values of parameters used in the calculations
grain boundary.20 In general, the solubility of gas on
the grain boundary is substantially higher than in the Parameter Value References
bulk material. In analogy with the treatment of intra- b 0.25 27
granular bubble nucleation in the high-temperature x 125 Annealed28
equilibrium g-phase discussed in Section 3.20.2.2, 850 Nonannealed6,28
the gas concentration on the boundary will increase _
b0 ðb ¼ b0 fÞ UO2: 2 1023 m3 22
until the solubility limit is reached (approximately U–10Mo: 2 1024 m3 28
_
D0 ðDg ¼ D0 fÞ UO2: 1039 m3 s1 29
given by tb ), whereupon the gas will precipitate into U–10Mo: 1040 m3 s1
bubbles. Thus, the rate at which a grain-boundary rg 0.216 nm 27
bubble adsorbs gas is approximately given by g 0.6 J m2 Annealed28
0.5 J m2 Nonannealed6,28
ðdm=dt Þr ¼ r0 ¼ bv Cg =ð4tb Cb pr03 =3Þ ½58 cosðyÞ 0.2 30
bv 8.5 1023 m3 per 27
where Cg is the gas concentration on the boundary atom
given by fn 1 28
ds 1 109 m 6
dg _ l 1.8 108 m 6
Cg ¼ fs bf t ½59
3 z 4 24
0 2.5745 108 m 6
As described by eqn [50], subsequent to bubble nucle- i 1 109 m 6
ation gas solubility on the boundary will drop to a
relatively low value and gas arriving at the boundary Source: Rest, J. J. Nucl. Mater. 2010, 407, 55–58.
596 Modeling of Fission-Gas-Induced Swelling of Nuclear Fuels
of the others (e.g., x) result from a comparison of the bubbles with diameters of 1–2 nm were observed to
present theory with measured data for bubble popu- form a superlattice in the U–Mo with a relatively
lations. As an example of estimated parameters, the close spacing (6–7 nm) and having an approximate
values of Dg and b used for U–Mo are assumed to be monomodal-like distribution.32 For this reason, as
an order of magnitude less than those for UO2. On listed in Table 1, the gas-bubble nucleation factor
the basis of irradiation-enhanced creep rates was taken to be equal to unity. In any event, the
measured in UO2, UN, and UC,31 the irradiation- physics presented in this section is not compatible
enhanced gas-atom diffusivity Dg is expected to be with the formation of a bubble superlattice.
lower in U–Mo than in UO2. In addition, as a result of
the higher thermal conductivity of the alloy as com- 3.20.3.6 Comparison Between Model
pared to the oxide, b is also expected to be lower in Calculations and Intergranular Data
U–Mo than in UO2. This argument is on the basis of
the expected larger interaction cross-section in the The calculated distributions are obtained by integrat-
metallic alloy with conduction electrons. However, ing eqn [61] over the bin sizes i , that is, the bubble
because of the (assumed) linear dependence of both density N ði Þ in units of m3 is given by
Dg and b on f_ , and because it is the ratio Dg =b that Z0 þi
appears in eqns [37]–[39], it is reasonable to assume N ði Þ ¼ nðr Þdr ½63
that this ratio of critical properties is the same for
0 þði1Þ
both materials.
The calculated intragranular bubble-size distribu- where 0 is the minimum bubble size. The inter-
tion for Z03 (fully annealed) is contrasted with data32 granular bubble size depends on the value of x (see
for the average bubble size and density in irradiated eqn [41] and Table 2), which is a grain-boundary
U–10Mo fuel (ground and atomized) as shown in gas-atom diffusion enhancement factor that reflects
Figure 9. Values for Dg and b obtained from data the fact that grain-boundary diffusion is decidedly
and analyses of UO2 are listed in Table 2. The faster than grain lattice diffusion.33,34 The effect of x
calculated results shown in Figure 3 are in reason- on the intergranular bubble nucleation is visible in
able accord with the observed estimates of the aver- eqn [41]. By increasing x the intergranular bubble
age bubble density and size. However, it should density is reduced with a commensurate increase in
be noted that highly over pressurized solid gas bubble size. The larger value used for x for the
1E + 19
Theory
Estimated data
1E + 18
Bubble density (cm−3)
1E + 17
1E + 16
1E + 15
0 2 4 6 8 10
Bubble diameter (nm)
Figure 9 Calculated intragranular bubble-size distribution for Z03 (fully annealed) contrasted with data32 for the average
bubble size and density in irradiated U–10Mo fuel (ground and atomized). The calculated distribution is not consistent with the
observed bubble superlattice.
Modeling of Fission-Gas-Induced Swelling of Nuclear Fuels 597
nonannealed miniplates reflects the increase in diffu- These miniplates were fully annealed and as such
sivity with decreased molybdenum content. have a uniform distribution of molybdenum across
The experimental database consists of both as- the fuel region. Z03 was fabricated by atomization,
atomized and g-phase annealed specimens. The range whereas Y01 was made from a ground powder. The
of burnup is from 5.8 to 9.2 at.% U, with fission rate calculated distribution is in very good agreement
from 2.3 to 6.8 1014 f cm3 s1, temperature from with the measured quantities.
66 to 191
C, and Mo content from 6 to 10 wt%.20 Figure 11 shows calculated and measured inter-
Table 2 shows the value of the key physical para- granular bubble-size distribution for U–10Mo as-
meters used in the model. The remaining critical atomized plates. As is evident from the comparisons
parameter x was determined by best overall interpre- in Figure 11, in general, the model calculations are in
tation of the measured intergranular bubble-size remarkable agreement with the data. Figure 12 shows
distributions for the g-phase annealed and for the calculated and measured intergranular bubble-size
as-atomized specimens, respectively. In addition, distribution for U–6Mo and U–7Mo as-atomized
the reduced value for the grain-boundary energy g plates, respectively. The deviation between calculated
for the nonannealed material reflects lower angle and measured results shown in Figure 12 is most
boundaries as compared to the annealed specimens.20 likely due to the lower Mo content and, therefore,
Figure 10 shows calculated results compared requires different (larger) values for Dg and x.
with RERTR-3 miniplates Z03 and Y01 data. The results of calculations shown in Figure 13
demonstrate the increased validation leverage
secured with the use of bubble-size distributions
compared with the use of mean values (i.e., average
4E + 8
Theory
quantities such as bubble density and diameter). Com-
Data Z03 paring model predictions with average quantities
3E + 8 is by far the dominant validation technique reported
in the literature. The graph on the left hand side
Bubble density (cm-2)
2.5E + 8 2E + 8
Theory 2E + 8 Theory
2.0E + 8 Data V03 Data V07
2E + 8
1E + 8
Bubble density (cm )
1.5E + 8
1E + 8
1E + 8
1.0E + 8
8E + 7
6E + 7
5.0E + 7
4E + 7
0.0 2E + 7
0.0
0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.05 0.10 0.15 0.20 0.25 0.30 0.35
Bubble diameter (μm) Bubble diameter (μm)
2.5E + 8 1.8E + 8
Theory 1.6E + 8 Theory
2.0E + 8 Data V8005B Data V002
1.4E + 8
Bubble density (cm )
1.2E + 8
1.5E + 8
1.0E + 8
1.0E + 8 8.0E + 7
6.0E + 7
5.0E + 7
4.0E + 7
2.0E + 7
0.0
0.0
0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.05 0.10 0.15 0.20 0.25 0.30 0.35
Bubble diameter (μm) Bubble diameter (μm)
2.5E + 8
Theory
2.0E + 8 Data V6019G
Bubble density (cm−2)
1.5E + 8
1.0E + 8
5.0E + 7
0.0
Figure 11 Calculated and measured intergranular bubble-size distribution for U–10Mo as-atomized plates V03, V07, V002,
V8005B, and V6019G. Reproduced from Rest, J.; Hofman, G. L.; Kim, Y. S. J. Nucl. Mater. 2009, 385(3), 563–571.
of burnup for U–8Mo fuel irradiated at various tem- at 33% swelling. However, in general, the calculation
peratures. The calculated swelling is a strong func- of the swelling at which the bubbles interconnect
tion of the irradiation temperature as well as the fuel is complicated by a relatively broad distribution of
burnup. It should be noted that the temperature nonspherical bubbles, nonuniformly distributed within
dependence of fuel that is under restraint (e.g., by the fuel regions (e.g., such as that in Figure 14).
cladding) is much softer than exhibited in Figure 14. The maximum gas release in these high swelling fuels
The curves in Figure 14 do not reflect any gas approaches 80%. There are many small bubbles
release that may occur. Empirically, gas release begins between the larger interconnected bubbles that con-
to occur when the swelling reaches 25–30%. If all the tinue to drive the swelling even at high gas release
bubbles are spherical, of the same size, and randomly values. However, even so, the calculated swelling
distributed, then interconnection will be initiated curves in Figure 8 are typical of those that have been
Modeling of Fission-Gas-Induced Swelling of Nuclear Fuels 599
1.6E + 8 5E + 8
Theory Data
1.4E + 8 Data S03 l = 1.26 × 10-8
4E + 8 l = 1.80 × 10-8
1.2E + 8 l = 2.34 × 10-8
1.0E + 8 3E + 8
8.0E + 7
2E + 8
6.0E + 7
4.0E + 7 1E + 8
2.0E + 7
0
0
0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.04 0.06 0.08 0.10 0.12 0.14 0.16
Bubble diameter (mm) Bubble diameter (μm)
7E + 8
2.5E + 8
Data
Theory 6E + 8 x = 21; d = 0.70 × 10-9
Data R6007F
2.0E + 8 x = 15; d = 1.00 × 10-9
5E + 8 x = 10; d = 1.43 × 10-9
Bubble density (cm-2)
3E + 8
1.0E + 8
2E + 8
5.0E + 7
1E + 8
0.0 0
0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.04 0.06 0.08 0.10 0.12 0.14 0.16
Bubble diameter (mm) Bubble diameter (μm)
Figure 12 Calculated and measured intergranular Figure 13 Sensitivity of the calculated distributions to the
bubble-size distribution for as-atomized plates S03 (U–6Mo) value of the gas-atom knock-on distance (left-hand graph).
and R6007F (U–7Mo). Reproduced from Rest, J.; Hofman, Results of a series of calculations made with paired values
G. L.; Kim, Y. S. J. Nucl. Mater. 2009, 385(3), 563–571. for the grain-boundary-diffusion enhancement factor and
the thickness of the grain boundary, chosen such that the
calculation of average quantities remains unchanged
measured. The key here is that Figure 14 shows unre- (right-hand graph). Reproduced from Rest, J.; Hofman,
strained swelling. If the fuel is given enough room, it G. L.; Kim, Y. S. J. Nucl Mater. 2009, 385(3), 563–571.
will keep on deforming.
If it is arbitrarily assumed that the maximum
allowable fuel swelling is 50%, then fuel safety mar- obtained for the rate equations, thereby providing
gins can be calculated using the results of Figure 14. for increased transparency and ease of use. The results
As an example of this type of calculation, Figure 15 support a multiatom gas-bubble nucleation mecha-
shows the calculated boundary between stable and nism on grain boundaries that have substantially
unstable unrestrained fuel swelling as a function of higher gas solubility than that in the grain interior.
fission density and fuel temperature. The solid line in The gas-atom diffusion enhancement factor on the
Figure 15 is the 50% unrestrained swelling threshold grain boundaries was determined to be 125–850 in
obtained from Figure 14. Also shown in Figure 15 order to obtain agreement with the measured distri-
is the fission density and fuel temperature for butions. The enhancement factor is about 8 times
RERTR-9. As shown in Figure 15, the calculated higher for as-fabricated powder plates than for the
safety margin for RERTR-9 is 150 K. annealed plates because of the lower Mo content on
the boundaries. This range of values for the enhance-
ment factor is consistent with values obtained in the
3.20.3.8 Conclusions
literature.30 The largest deviation between calculated
Calculations of intergranular bubble-size distribution and measured results (Figure 12) is most likely due to
made with a mechanistic model of grain-boundary several fuel plates that have a lower Mo content (6 and
bubble formation kinetics are consistent with the 7 wt% vs. 10 wt%) and, thus, require different (larger)
measured distributions. Analytical solutions are values for Dg and x.
600 Modeling of Fission-Gas-Induced Swelling of Nuclear Fuels
500
684 K
700 K
716 K
400 732 K
748 K
764 K
300 780 K
Swelling (%)
200
100
1E + 22
50% unrestrained swelling
threshold
147 K
8E + 21 RERTR-9
Fission density (cm−3)
6E + 21
Safety margin
4E + 21
2E + 21
The agreement between the model and the intergranular bubble-size distribution using a model
measured distributions for the 10 wt% Mo fuel sup- based on the growth of bubbles in a regular array35
ports the validity of a sputtering coalescence (bubble have not been successful.
coalescence without bubble motion) coarsening A number of the critical parameters listed in
mechanism on the grain boundaries. In this regard, Table 2 are assumed to be a factor of 10 less than
attempts by this author to reproduce the shape of the those listed in the literature for UO2. However, it is
Modeling of Fission-Gas-Induced Swelling of Nuclear Fuels 601
the ratio of these parameters (b=Dg , x=l) that appear a grain boundary provides a relatively short recap-
in the model solution; thus, the validity of their use ture distance for gas that has been knocked out of a
for U–Mo reduces to the ratios being approximately bubble, and as such neutralizes the ‘shrinking’ effect
the same for both materials. This assumption is sup- of the re-solution process. These grain-boundary
ported by the observed similarity (albeit remarkable) bubbles grow at an enhanced rate as compared
in bubble behavior and microstructure evolution to those in the bulk material. The importance of
between the two materials.36 understanding the physics underlying intergranular
The results demonstrate the increased validation bubble growth is underscored by the rim region of
leverage secured with the use of bubble-size dis- high-burnup fuels which are characterized by an
tributions compared with the use of mean values exponential growth of intergranular porosity toward
(i.e., average quantities such as bubble density and the pellet edge: a narrow band of fully recrystallized
diameter). Model predictions are sensitive to various porous material exists at the pellet periphery, and
materials and model parameters. Improved prediction a rather wide adjacent transition zone with partially
capability requires an accurate quantification of these recrystallized porous areas appears dispersed within
critical materials properties and measurement data. the original matrix structure.46 In particular, the
The results of this analysis enable the calculation understanding of the dynamics of irradiation-induced
of safety margins for unrestrained fuel swelling. recrystallization and subsequent gas-bubble swelling
These safety margins contain an uncertainty primar- requires a quantitative assessment of the nucleation
ily tied to uncertainties in the values of the volume and growth of grain-boundary bubbles.45,46
and Xe diffusion coefficients. A mechanistic model is described, for the growth of
grain-boundary bubbles during irradiation at relatively
low temperatures (i.e., where gas-atom diffusion is
3.20.4 Irradiation-Induced athermal) in order to quantify the effect of gas-atom
Re-solution re-solution on their growth. A variational method is
used to calculate diffusion from a spherical fuel grain.
3.20.4.1 Introduction
The junction position of two trial functions is set equal
After a short period of irradiation, the intragranular to the bubble gas-atom knock-out distance. The effect
structure of UO2 is populated with a high-density of grain size, gas-atom re-solution rate and diffusivity,
(1023 m3 ) of small (r 109 m) bubbles,19 sepa- gas-atom knock-out distance, and grain-boundary
rated by 5–10 bubble diameters. In general, observa- bubble density on the growth of intergranular bub-
tions that bubbles confined to the bulk (lattice) material bles is studied, and the conditions under which inter-
of irradiated nuclear fuels do not grow to appreciable granular bubble growth occurs are elucidated.
sizes at low temperatures (fuel temperatures where the
gas-atom diffusivity is irradiation enhanced, i.e., <0.5
melting temperature) are ascribed to the effect of 3.20.4.2 Flux Algorithm
irradiation-induced re-solution (see Chapter 2.18, The flux of gas atoms diffusing to the grain bound-
Radiation Effects in UO2).3,37 Gas-atom re-solution aries in a concentration gradient is obtained by solv-
is a dynamic bubble-shrinkage mechanism wherein ing the concentration of gas atoms Cg within a
fission fragments either directly or indirectly cause gas spherical grain that satisfies the equation
atoms to be lost from a bubble. Only when sinks, such as
grain boundaries, are present in the material can bub- @Cg 1 @ 2 @Cg
bles grow to sizes observable with a SEM.38 Most cal- ¼ 2 Dg r þ a f þ f ðlÞboundary ½64
@t r @r @r
culations on intergranular gas behavior found in the
literature have focused on the condition for grain-face where Dg is the gas-atom diffusion coefficient, f_ is
saturation and have not addressed the specific mechan- the fission rate, a is the average number of rare-gas
ics of intergranular bubble growth in the presence atoms produced per fission, r is the radial distance
of irradiation-induced re-solution.39–42 Calculations from the grain center, and f ðlÞboundary is the flux of
of grain-boundary bubble growth have been performed gas atoms from the boundary bubbles because
under the assumption that the effective gas-atom of irradiation-induced re-solution. This back flux of
re-solution rate from grain-boundary bubbles is gas can be thought of as an additional matrix gas-
negligible.43–45 This assumption has relied on heu- atom generation mechanism and is assumed to be
ristic arguments23 that the strong sink-like nature of distributed uniformly within a spherical annulus of
602 Modeling of Fission-Gas-Induced Swelling of Nuclear Fuels
thickness l, where l is the gas-atom knock-out dis- with a minimum of computer storage and running
tance. In eqn [64], intragranular bubble trapping of time by splitting the spherical grain into two concen-
fission gas has been neglected. However, this effect tric regions of approximately equal volume. In each
can be modeled by using an effective diffusion coef- region, the gas concentration was represented by a
ficient given by Turnbull3 quadratic function. In the inner region, the concen-
b tration function was constrained to have dCg =dr ¼ 0
Dgeff ¼ Dg ½65 at r ¼ 0. In the outer region, the concentration func-
bþg
tion was constrained to a value of Cg ¼ 0 at r ¼ dg/2.
where b is the gas-atom re-solution rate and g is the The two functions were also constrained to be con-
probability per second of a gas atom in solution being tinuous at the common boundary of the two regions.
captured by an intragranular bubble. Observed con- This left three free parameters: the concentrations
centrations 1023 m3 of intragranular bubbles of g g g
C1 , C2 , and C3 , respectively, for the radius ratio
1 nm radius3 with b ¼ 2 104 s1 yields a value r1 ¼ 0:2, r2 ¼ 0:4, and r3 ¼ 0:45, where r ¼ r =dg .
for g ¼ 2:5 104 s1 and Dgeff ¼ 0:44Dg . These positions are the midpoint radii of the inner
In general, eqn [64] is solved with the boundary region, the boundary between the regions, and the
conditions midpoint radius of the outer region, respectively.
However, this method is too crude if one is inter-
Cg ¼ 0 at t ¼ 0 for 0 r dg =2 ½66a ested in an accurate representation of the concentra-
tion gradient in the presence of irradiation-induced
Cg ¼ 0 at r ¼ dg =2 for t0 t t0 þ dt ½66b re-solution from grain-boundary bubbles where the
gas atom knock-out distance l is on the order of
@Cg 100 Å.5 In this case, it is necessary to formulate the
¼ 0 at r ¼ 0 for t0 t t0 þ dt ½66c
@r problem in terms of l. The radius ratio at the inter-
where dt is an increment of time and dg is the grain face of the two regions is now expressed as
diameter. For an increment of time dt the concentra-
rl ¼ 1=2 l=dg ½69
tion of gas atoms in a spherical grain described in eqn
[64] is The trial functions are as follows:
For the inner region,
1 d 2 dCg Cg Cg0
D g r þ þ af þ f ðlÞboundary ¼ 0 g
g
r dr
2 dr dt dt Cg ðrÞ ¼ 4 C1 r2l r2 þ C2 r2 r2l =4 =3r3l ½70
½67
For the outer region;
Euler’s theorem may now be used to obtain a varia-
g
tional principle equivalent to eqn [3]: C2
Cg ðrÞ ¼ ½8r2 2ð2rl þ 3Þr þ 2rl þ 1
ð g " 2
1=2d ð2rl 1Þ2
Dg dCg Cg2
d 4p þ 8C3
g
2 dr 2dt þ ½ð2rl þ 1Þr 2r2 rl ½71
0
! ð2rl 1Þ2
Cg0 ½68 Equations [69] and [70] are substituted for Cg in eqn
þ af þ f ðlÞboundary Cg r 2 dr ¼ 0
dt [68] and an extremum is found by differentiating
g g g
with respect to C1 , C2 , and C3 . This results in a set
which assumes that Dirichlet boundary conditions
of three coupled algebraic equations that can be
are to be applied. An approximate solution to the g g
directly solved to obtain the concentrations C1 , C2 ,
problem may now be obtained by choosing a trial g
and C3 , as follows:
function that satisfies the boundary conditions and
g
minimizes the integral in eqn [68] in terms of free g X1 F2 C2
C1 ¼ ½72
parameters in the function. Many types of trial func- F1
tion could be chosen, but it is easier to work with
F2 F4
piecewise functions than global functions. Quadratic X1 þ X3 X2
g F F5
functions are attractive because they allow an exact C2 ¼ 1 ½73
F2 F4
representation of eqn [64] for long times. Matthews F2 þ F4 F3
and Wood47 obtained a realistic level of accuracy F1 F5
Modeling of Fission-Gas-Induced Swelling of Nuclear Fuels 603
1000
dg = 50 μm
dg = 5 μm
800 dg = 0.5 μm
Bubble diameter (nm)
Bubble interlinkage
600
400
200
0
0.0 0.2 0.4 0.6 0.8 1.0 1.2
Fission density (1027 m−3)
Figure 16 Calculated bubble diameter versus fission density for three values of grain size. The arrow shows the point at
which the grain-boundary bubbles interlink. Reproduced from Rest, J. J. Nucl. Mater. 2003, 321(2–3), 305–312.
generation mechanism (last term on the RHS of eqn Table 3 Nominal values of key parameters used in the
[64]) and is given by calculation
atoms from the boundary to the matrix on intergran- re-solution result in a smaller net flux of gas atoms
ular bubble growth is dependent on the ratio l=dg : to the boundary (see eqn [80]) and, hence, smaller
smaller values of l=dg lead to larger values for grain-boundary boundary bubble sizes. The arrow in
the flux J. As shown in Figure 16, the calculated Figure 19 shows the point at which the grain-bound-
intergranular bubble sizes are larger for larger grain ary bubbles interlink. The effect of the back flux of gas
diameters (i.e., smaller values of l=dg ). The arrow atoms can also be seen in Figure 20, which shows the
in Figure 16 shows the point at which the grain- calculated fractional gas release to grain boundary
boundary bubbles interlink as given by eqn [86]. versus fission density for three values of gas-atom
The effect of calculated grain-boundary bubble re-solution rate. For re-solution rates in Figure 20
diameter on the value of l=dg for a fixed grain size of 2 104 and 2 103 s1, the fractional gas release
and variable l is demonstrated in Figure 17 where decreases after an initial increase. This behavior is due
calculated bubble diameter versus fission density to the growing strength of the back flux of gas atoms as
is shown for three values of gas-atom knock-out the number of gas atoms in boundary bubbles
distance. Decreasing the value of l (i.e., smaller increases (e.g., see last term in eqn [80]).
values of l=dg ) leads to larger fluxes and a commen- Figure 21 shows the calculated gas-atom concen-
surate increase in the calculated bubble size. tration versus fractional grain diameter across the
Figure 18 shows calculated bubble diameter ver- outer annulus of width l at a fission density of
sus fission density for three values of gas-atom diffu- 1 1027 m3 for three values of gas-atom re-solution
sivity. Higher gas-atom diffusion rates result in a rate. Higher values of gas-atom re-solution rate result
greater flux of gas atoms to the boundary (see eqns in a stronger back flux of gas atoms into the lattice,
[77] and [78]) and, thus, larger bubble sizes (see also and thus a larger concentration of gas within the
eqns [79] and [80]). adjacent annulus of width l. Although Figure 21
Figure 19 shows calculated bubble diameter ver- indicates that the flux of gas to the boundary
sus fission density for three values of gas-atom re- increases for larger values of the gas-atom re-solution
solution rate.
The re-solution rate that is being varied rate, the net flux decreases due to the larger back flux
here is b0 f , that is, the term in eqn [82] that is of gas into the lattice. This is clear from Figure 22
independent of bubble size. The strength of the which shows the calculated net flux of gas atoms to
gas-atom re-solution rate works counter to that of the grain boundary versus fission density for the
the gas-atom diffusivity; larger values of the gas-atom three values of the gas-atom re-solution rate.
500
l = 2.5 × 10−9 m
l = 5 × 10−9 m
400 l = 1 × 10−8 m
Bubble diameter (nm)
300
200
100
0
0.0 0.2 0.4 0.6 0.8 1.0 1.2
Fission density (1027 m−3)
Figure 17 Calculated bubble diameter versus fission density for three values of gas-atom knock out distance.
Reproduced from Rest, J. J. Nucl. Mater. 2003, 321(2–3), 305–312.
606 Modeling of Fission-Gas-Induced Swelling of Nuclear Fuels
500
Dg = 2 × 10−18 m2 s−1
Dg = 2 × 10−19 m2 s−1
400 Dg = 2 × 10−20 m2 s−1
Bubble diameter (nm)
300
200
100
0
0.0 0.2 0.4 0.6 0.8 1.0 1.2
Fission density (1027 m−3)
Figure 18 Calculated bubble diameter versus fission density for three values of gas-atom diffusivity. Reproduced from
Rest, J. J. Nucl. Mater. 2003, 321(2–3), 305–312.
1000
b = 2 × 10−5 s−1
b = 2 × 10−4 s−1
800 b = 2 × 10−3 s−1
Bubble interlinkage
Bubble diameter (nm)
600
400
200
0
0.0 0.2 0.4 0.6 0.8 1.0 1.2
Fission density (1027 m−3)
Figure 19 Calculated bubble diameter versus fission density for three values of gas-atom re-solution rate. The arrow
shows the point at which the grain-boundary bubbles interlink. Reproduced from Rest, J. J. Nucl. Mater. 2003, 321(2–3),
305–312.
Figure 23 shows the calculated bubble diameter estimated value of the effective intragranular diffusion
versus fission density for three values of grain- coefficient given by eqn [65] that accounts for gas-atom
boundary diffusion enhancement factor. As expected, trapping by intragranular bubbles is well within the
larger values of grain boundary-gas-atom diffusivity range of diffusivities explored in Figure 23.
lead to larger boundary bubble sizes because of the Figure 24 shows the calculated bubble diameter
increased flux of gas atoms into the bubbles. The versus fission density for three values of grain-boundary
Modeling of Fission-Gas-Induced Swelling of Nuclear Fuels 607
0.6
b = 2 × 10−5 s−1
b = 2 × 10−4 s−1
b = 2 × 10−3 s−1
Fractional gas release 0.5
0.4
0.3
0.2
0.0 0.2 0.4 0.6 0.8 1.0 1.2
Fission density (1027 m−3)
Figure 20 Calculated fractional gas release to grain boundary versus fission density for three values of gas-atom
re-solution rate. Reproduced from Rest, J. J. Nucl. Mater. 2003, 321(2–3), 305–312.
2.5
b = 2 × 10−5 s−1
b = 2 × 10−4 s−1
2.0
b = 2 × 10−3 s−1
Gas atom concentration (1024 m−3)
1.5
1.0
0.5
0.0
bubble density. The calculated bubble size increases for fully in the athermal, irradiation-enhanced regime.
smaller values of the grain-boundary bubble density An increase in the temperature (e.g., to 746 K charac-
because of the larger number of gas atoms per bubble. teristic of the rim region in UO2 fuel) will, accord-
The temperature used in Figures 16–24 is 373 K. ing to eqn [79], produce a proportionate increase
This low temperature was selected in order to be (e.g., doubling) of the bubble diameter.
608 Modeling of Fission-Gas-Induced Swelling of Nuclear Fuels
0.6
b = 2 × 10−5 s−1
b = 2 × 10−4 s−1
0.5 b = 2 × 10−3 s−1
Net flux (1014 m−2 s−1)
0.4
0.3
0.2
0.1
0.0 0.2 0.4 0.6 0.8 1.0 1.2
Fission density (1027 m−3)
Figure 22 Calculated net flux of gas atoms to the grain boundary versus fission density for three values of the gas-atom
re-solution rate. Reproduced from Rest, J. J. Nucl. Mater. 2003, 321(2–3), 305–312.
400
z = 100
z = 10
z=1
300
Bubble diameter (nm)
200
100
0
0.0 0.2 0.4 0.6 0.8 1.0 1.2
Fission density (1027 m−3)
Figure 23 Calculated bubble diameter versus fission density for three values of grain-boundary diffusion enhancement
factor. Reproduced from Rest, J. J. Nucl. Mater. 2003, 321(2–3), 305–312.
500
Nb = 1 × 1011 m−2
Nb = 1 × 1012 m−2
400
Nb = 1 × 1013 m−2
Bubble diameter (nm)
300
200
100
0
0.0 0.2 0.4 0.6 0.8 1.0 1.2
Fission density (1027 m−3)
Figure 24 Calculated bubble diameter versus fission density for three values of grain-boundary bubble density.
Reproduced from Rest, J. J. Nucl. Mater. 2003, 321(2–3), 305–312.
(i.e., high concentrations of gas and enhanced diffu- leading to an initial rapid nucleation of gas bub-
sion) that is much stronger than the bubble growth bles.37 These evolving gas bubbles act as sinks for
mechanism in the lattice. Second, given the rela- the newly arriving gas and deplete the boundary of
tively small size of the gas-atom knock-out single gas atoms, thus reducing bubble nucleation.
distance l, gas atoms that are ejected from bound- When the projected areal coverage of the boundary
ary bubbles find their way back (via the chemical by bubbles becomes significant, the direct getting of
potential gradient) to the boundary in relatively gas atoms arriving from the lattice by the boundary
short times. Third, as clearly delineated in Section bubbles augments this process. Conversely, within
3.20.3, bubble nucleation and growth on grain the matrix, although the bubble nucleation rate is
boundaries produce bubble densities that are far initially smaller than on the boundary, the nucle-
smaller, and bubble sizes much larger than observed ation continues for a longer time leading to a sub-
in the bulk material. For example, an intergranular stantially larger, albeit significantly smaller sized,
bubble density of 1 1013 m2 is equivalent to population of bubbles.53
a bubble density of 2 1018 m3 for a grain size of The analysis presented here lacks an integrated
5 106 m. This is to be compared to observed intra- calculation for the evolution of the intergranular
granular bubble densities that are on the order of bubble-size distribution. Instead, the intergranular
1023 m3. Additionally, the effects of intergranular bubble density is an input parameter (see Table 3). In
bubble nucleation and growth are augmented by order to calculate the intergranular bubble density, a
linkup, venting, and reformation processes.38,52 suitable model for bubble nucleation on grain bound-
As discussed in Section 3.20.3, the reason for the aries is required (such as that presented in Section
smaller density of bubbles on the grain boundary 3.20.3, or developed by Wood and Kear24). In Wood
as compared to the bulk is due to a multiatom and Kear24, the intergranular bubble density rises to a
nucleation mechanism as compared to a two-atom steady-state value relatively early in the irradiation.
mechanism operative in the bulk material. In addi- Although such a calculation is valuable, it was felt that
tion, there is a contribution from the relatively its incorporation was outside the scope of this chapter,
rapid arrival rate of gas atoms to a grain boundary and would not affect the general conclusions iterated
that has a substantially smaller volume than the bulk above. However, given an estimate of the intergranular
grain. Thus, early in the irradiation, the gas-atom bubble density (e.g., at low burnup), the condition
interaction rate on the boundary is relatively high, for grain-face saturation (i.e., the burnup at which
610 Modeling of Fission-Gas-Induced Swelling of Nuclear Fuels
saturation occurs; e.g., see Figures 14 and 17) can be In Section 3.20.5.4, calculations are compared to
calculated and compared directly with observation. data for the dislocation density and change in lattice
displacement in UO2 as a function of burnup.
In Section 3.20.5.5, calculations are compared to
3.20.5 Irradiation-Induced available data for the recrystallized grain-size distri-
Recrystallization bution in UO2 and in U–10Mo. Finally, conclusions
are presented in Section 3.20.5.6.
3.20.5.1 Introduction
As discussed in Section 3.20.1, fission-gas behavior
is intimately tied to the evolving fuel microstructure 3.20.5.2 Model for Initiation of
(see Chapter 2.18, Radiation Effects in UO2). Irradi- Irradiation-Induced Recrystallization
ation-induced recrystallization appears to be a general
The recrystallization nuclei are taken as the triple
phenomenon in that it has been observed to occur in a
points of an evolving cellular dislocation network.
variety of nuclear fuel types,36 for example, U–xMo,
The relevant kinetics are those of interstitial-loop for-
UO2, and U3O8 (see Chapter 1.03, Radiation-
mation and agglomeration that lead to the formation of
Induced Effects on Microstructure). The recrystalli-
a dislocation network in an environment of precipitate
zation process results in the transformation of the
pinning.54,58 The trigger point for irradiation-induced
original micron-size grains to submicron-size grains
recrystallization is defined as the point where the
that accelerate fission-gas swelling as a result of the
kinetically derived concentration of nuclei becomes
combination of short diffusion distances, increased
equal to the equilibrium number of nuclei determined
grain-boundary area per unit volume, and greater
from thermodynamic considerations.
intergranular bubble growth rates when compared
The fission density Fd dependent concentration of
with that in the grain interior.26 The initiation of
viable recrystallization nuclei Crx has been deter-
recrystallization is to be distinguished from the
mined as a function of the dislocation density rd
subsequent progression and eventual consumption of
on the basis of the concept of node pinning by
the original fuel grain. An expression has been derived
irradiation-induced precipitates associated with
for the fission density at which irradiation-induced
fission-gas bubbles54 as
recrystallization is initiated that is athermal and weakly
dependent on fission rate,54 and for the progression sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
9ð f ðnÞrd Þ7=2 ap 2 2l
of recrystallization55 independent of temperature Crx ¼ 7 5=2 ’g
½87
8p ðCA Cr Þ Fd
6 3pb v B0 b
and fission rate, linearly dependent on the fission
density, and inversely proportional to the cube of the where ’g=ap is a factor composed of terms related to
as-fabricated grain size. the production of precipitates and subgrain growth in
The driving force for recrystallization is the pro- the presence of precipitates, bv is the van der Waals
duction of interstitial loops due to irradiation. The constant, l is the atom knock-on distance,
continued generation of interstitial loops induces f ðnÞ ¼ ð1 n=2Þ=ð1 nÞ, n is Poisson’s ratio,
internal stresses in the material that leads to strain CA ¼ 3 for cubic cells, and Cr is within a factor of
in the form of lattice displacement. The initiation of unity, while b is the number of gas atoms produced
recrystallization in U–10Mo has been observed to per fission, and at the relatively low temperatures at
occur predominately along the preexisting grain which irradiation-induced recrystallization occurs,
boundaries.56 Subsequently, the recrystallization the gas-atom diffusivity is athermal and can be
front moves toward the grain center eventually con- expressed as Dg ¼ D0 f_ , where D0 is a constant of
suming the entire grain. Thus, the volume fraction of proportionality (1039 m5) and f_ is the fission rate.
recrystallized material is a function of irradiation The temperature dependence of Crx in eqn [87] is
time as well as the initial grain size. As gas-bubble contained in the interstitial and vacancy diffusivities.
swelling is higher in the recrystallized material than In general, these diffusivities are expressed as
in the un-recrystallized fuel, the swelling due to fission Di ¼ Di0 expðei =kT Þ and Dv ¼ Dv0 expðev =kT Þ,
gas is a function of the recrystallization kinetics. where ei and ev are the interstitial and vacancy migra-
In Section 3.20.5.2, models for the initiation tion enthalpies, respectively.
and progression of irradiation-induced recrystalli- The trigger point for irradiation-induced recrystal-
zation are reviewed. Section 3.20.5.3 presents a lization is defined as the point where the kinetically
theory for the size of the recrystallized grains.57 derived concentration of nuclei given by eqn [87]
Modeling of Fission-Gas-Induced Swelling of Nuclear Fuels 611
becomes equal to the equilibrium number of nuclei when the volumetric strain energy exceeds that
ni determined from thermodynamic considerations, necessary to create the new surface; and (3) the
that is, rate at which the defect front moves through the
ni ¼ n0i expðG =kT Þ ½88 newly created defect-free annulus is proportional
to the strain rate, which, in analogy with fission-
where G is the critical standard free energy that a induced creep, is proportional to stress and fission
node must acquire to recrystallize and, as the basic rate. The microscopic stress is a function of the
unit out of which the cellular dislocation network is lattice displacement, which is related to the genera-
composed is the interstitial loop, n0i is taken to be the tion rate of interstitial loops.
athermal component in the expression for the inter- The original grain boundaries act as nucleation sites
stitial-loop density. for the recrystallization transformation. Upon the ini-
tiation of recrystallization, given by eqn [90], and the
rl r3=2
ni ¼
0
¼ pffiffiffiffiffiffiffiffiffi creation of the defect-free annulus adjacent to the ring
pdl Athermal p3=2 CA Cr f ðnÞ of newly recrystallized material, the defects interior to
exp½þðev =2 ei Þ=2kT ½89 the annulus are considered to be in a ‘superheated’
condition. These defects must travel through the
Equating eqns [87] and [88], using eqn [89], and solv- denuded annulus in order to find appropriate nucle-
ing for Fd results in an expression Fdx for the critical ation sites required for the recrystallization transfor-
fission density at which recrystallization will occur, mation. The defects in the region interior to the defect-
1=5 free annulus consist of a cellular dislocation network.
ap rd 4=5 2l
Fdx ¼ When dislocation loops are large enough relative
’g 3bv D0 b to the interatomic distances but small relative to the
f ðnÞ6=5 exp½4ðev =2 ei Þ=15kT crystal dimensions, they produce a measurable lattice
½90 distortion that can be expressed as59
p9=5 ðCA Cr Þ12=5
aðt Þ=a0 ¼ pbv nl ðt Þdl2 ðt Þ=12 ½92
The fission density at which recrystallization is pre-
dicted to initiate as given by eqn [4] is athermal and The increase in the lattice parameter is the driving
very weakly dependent on fission rate. As such, Fdx is force for irradiation-induced recrystallization.
independent of ev and ei and depends primarily on the Recrystallization occurs when the strain energy den-
collision related parameters l, Dg , and ap =’g. Substitut- sity U in an annulus of width d is greater than the
ing nominal values of the parameters54 in eqn [90] leads the energy required to create a new surface, that is,
to the simplified expressions for Fdx (m3):
4pr 2 dU ¼ 4pr 2 ggb ½93
Fdx ¼ 4 10 ð f_ Þ2=15
24
½91a where ggb is the energy of the subgrain boundary and
1 a 2
Fdx ¼ 6 1024 ð f_ Þ2=15 ½91b U ¼ E ½94
2 a0
where eqn [91a] corresponds to UO2 and eqn [91b] to
U–10Mo. Any compositional dependence of eqn [90] is where E is the bulk modulus of the material.
most likely reflected in the value of the irradiation Thus,
enhanced diffusion coefficient. 2gs y
d¼ ½95
Eða=a0 Þ2
3.20.5.3 Model for Progression of where gs is the surface energy and y is the boundary
Irradiation-Induced Recrystallization dihedral angle which is given by
h pffiffiffiffiffiffiffiffiffiffii pffiffiffiffiffiffiffiffiffiffi
A model for the progression of recrystallization as y ¼ 2tan1 bv rd =2 2bv rd =2 ½96
a function of burnup has been formulated55 on the
basis of the following assumptions: (1) recrystalliza- The progression of recrystallization is assumed to
tion initiates at preexisting grain boundaries; (2) occur in the following sequence of events: (1) sub-
annuli located initially adjacent to the original sequent to the formation of the new surface the
grain boundary transform to defect-free regions annulus is cleared of defects; (2) the defect front
via the creation of the new recrystallized surface interior to the annulus is driven through the cleared
612 Modeling of Fission-Gas-Induced Swelling of Nuclear Fuels
annulus by stress generated by defect-induced lattice Using eqns [95] and [102],
displacement; and (3) the strain rate is proportional pffiffiffiffiffiffiffiffiffiffi
2gs y a _ 2gs 2bv rd =2 _
to stress and fission rate (analogous to fission-induced ndf ¼ B2 E f ¼ B2 f ½107
radiation creep). Eða=a0 Þ2 a0 ða=a0 Þ
The time for the recrystallization front to move a and using eqn [103]
distance dg =2 r ðt Þ from the surface of a spherical pffiffiffiffiffiffiffiffiffiffi rffiffiffiffiffiffiffiffiffi
grain of diameter dg is given by 2gs 2bv rd =2 48g B f ðnÞ _
B f_ ¼
2
ndf ¼ rffiffiffiffiffiffiffiffiffi s
f ½108
bv C A C r p 1=2 2 CA Cr 2p
½dg =2 r ðt Þ rd
tr ¼ ½97 12 f ðnÞ
ndf
The volume fraction of recrystallized fuel as a function
where ndf is the speed of the defect front through the
of fission density is then given by
annulus. The volume fraction of recrystallized fuel as
a function of time can be expressed as 2ndf t 3
Vr ¼ 1 1 ½109
dg
2r ðt Þ 3
Vr ¼ 1 ½98
dg or after substituting eqn [104] for ndf
" rffiffiffiffiffiffiffiffiffi #3
and using eqn [97] 96gs B2 ðFd Fdx Þ f ðuÞ
Vr ¼ 1 1 ½110
3 dg CA Cr 2p
2ndf t
Vr ¼ 1 1 ½99
dg Equation [110] is the major result of this section. An
approximate solution for Vr is given by
It is here assumed that ndf is proportional to the strain rffiffiffiffiffiffiffiffiffi
rate, that is, 288gs B2 ðFd Fdx Þ f ðuÞ
Vr ; Fd
Fdmax ½111
dg C A C r 2p
ndf ¼ d_e ½100
where
where the strain rate is given by
Fd ¼ f_ t : Fdx ¼ f_ tx ½112
e_ ¼ B2 f_ s ½101
and tx is the time at which recrystallization is initiated.
and the stress is a function of the lattice displacement Thus, from eqn [111], for Fd
F max the volume
fraction of recrystallized fuel is proportional to
a the fission density and inversely proportional to the
s¼E ½102
a0 grain size. In addition, Vr ðt Þ is independent of fuel
temperature.
The progression of recrystallization occurs after the An estimate of the parameter B2 is required in
cellular dislocation network has formed. Thus, from order to perform quantitative calculations. U–10Mo
eqn [92] fuel with initiation of recrystallization at 40%
rffiffiffiffiffiffiffiffiffi
bv C A C r p 1=2 burnup appears to reach completion by 70%
a=a0 ¼ r ½103 burnup. Consistency with these ‘rough’ observations
12 f ðnÞ d
requires B2 1029. Estimates of this microscopic
where parameter on the basis of macroscopic in-pile creep
rd ¼ pnl dl ½104 data of UO2 (see Chapter 1.04, Effect of Radiation
on Strength and Ductility of Metals and Alloys and
and the lowest energy configuration is a cellular Chapter 1.05, Radiation-Induced Effects on Mate-
dislocation network with cell size given by Rest and rial Properties of Ceramics (Mechanical and
Hofman,45 and Hanson and Kuhlmann-Wilsdorf60 Dimensional))27 yield B2 ¼ 1031 (i.e., using eqn
rffiffiffiffiffiffiffiffiffiffiffiffiffiffi [101] with e_ ðUO2 Þ ¼ 2:8 1010 s1 , s ¼ 2:4 108
p
dl ¼ C A C r
rd f ðnÞ
½105 dynes cm2, and f_ ¼ 1:2 1013 cm3 s1 ).
Figure 25 shows the volume fraction of recrystal-
Using eqns [100] and [101] and solving for ndf lized fuel in U–10Mo calculated with eqn [110] as
a function of burnup compared with the estimated
ndf ¼ d_e ¼ dB2 f_ s ½106 data. The calculation shown in Figure 25 utilized
Modeling of Fission-Gas-Induced Swelling of Nuclear Fuels 613
1.2
Theory
Data
1.0
0.6
0.4
0.2
0.0
40 50 60 70 80
Burnup (%)
Figure 25 Volume fraction of recrystallized fuel in U–10Mo calculated with eqn [106] using B2 = 2 1029 compared
with estimated data. Reproduced from Rest, J. J. Nucl. Mater. 2005, 346(2–3), 226–232.
0.6
Theory
Data
0.5
Volume fraction of recrystallized fuel
0.4
0.3
0.2
0.1
0.0
0 10 20 30 40 50 60 70
Grain diameter (μm)
Figure 26 Calculated grain-size dependence of the volume fraction of recrystallized fuel in UO2 fuel calculated with eqn [106]
compared with data. Reproduced from Nogita, K.; Une, K.; Hirai, M.; Ito, K.; Shirai, Y. J. Nucl. Mater. 1997, 248, 196;
Rest, J. J. Nucl. Mater. 2005, 346(2–3), 226–232.
B2 ¼ 2 1029. Equation [91b] was used to calculate the burnup dependence exhibited by eqn [110].
the initiation of recrystallization at 45% burnup as Figure 26 shows the calculated grain-size dependence
shown in Figure 25. Although the data shown in of the volume fraction of recrystallized fuel in UO2
Figure 25 allow an estimate of the value of the fuel calculated with eqn [110] compared with data.61
parameter B2, the data are not sufficient to validate The grain-size dependence of eqn [110], that is, 1=dg
614 Modeling of Fission-Gas-Induced Swelling of Nuclear Fuels
1E + 16
1E + 15
1E + 14
1E + 13
r (m−2)
1E + 12
1E + 11
1E + 10 Theory
Fit to data
1E + 9
1E + 8
0 20 40 60 80 100
Burnup (GWd tM–1)
Figure 27 Calculated dislocation density for T ¼ 953 K as a function of burnup. Also shown is a fit to measured dislocation
density from Nogita and Une. Reproduced from Rest, J. J. Nucl. Mater. 2006, 349(1–2), 150–159.65
Table 4 Values of various parameters used in the cal- temperature for various values of the volumetric
culation for UO2 configurational entropy change S upon recrystalli-
Parameter Value References zation. As demonstrated in Figure 30, the calculated
2 1
recrystallized grain size decreases as the fuel temper-
D0i 5m s 60
ature increases. For fuel temperatures <850 K and for
D0v 75 m2 s1 60
ei 0.6 eV 60
relatively large values of T S that are on the order
ev 2.4 eV 63 of U (104–103 J m3 as shown in Figure 31),
riv 82.5 Å 60 the second term in the denominator of eqn [116] will
Ziv 5 104 60 offset the first term (which decreases as the tempera-
a p ffiffiffi Å (UO2)
5.47 2 ture decreases) such that the recrystallized grain size
bv 2a=4 2
E 2 1011(1–1.09154 104T ) 64 calculated with eqn [116] becomes relatively flat.
N m2, (UO2, 0.95 TD) The effect of including the entropy term in eqn
gs 1 J m2 2 [116] is also demonstrated in Figure 31 which shows
B2 2 1034 m5 N1 55 the calculated volumetric strain energy U in UO2
S 2 106 J m3 K1 57 for f_ ¼ 2 1019 m3 s1 as a function of the fuel
Source: Rest, J. J. Nucl. Mater. 2006, 349(1–2), 150–159.
temperature compared with several values for the
volumetric configurational entropy change S upon
recrystallization. As shown in Figure 31, U becomes
3.20.5.6 Calculation of Recrystallized
comparable to T S at fuel temperatures <850 K.
Grain Size
Comparing Figures 29 and 30, it is clearly seen
Using eqns [116], [118], and [124]–[127], and the that the effect of temperature on recrystallized grain
properties listed in Table 4, the average recrystal- size is much stronger than that of fission rate. Although
lized grain size can now be calculated. Figure 29 this result is at odds with some reported results,67,68 it
shows the calculated recrystallized grain size in is consistent with new experimental evidence that the
UO2 at T ¼ 923 K as a function of the average fission recrystallized grain-size distribution is temperature
rate f_ . As is evident from Figure 29, the calculated dependent.69 Again, care must be taken in comparing
recrystallized grain size increases as the average fis- the results of the calculations for recrystallized grain
sion rate in the fuel increases. Figure 30 shows size at the moment of recrystallization with data on
the calculated recrystallized grain size in UO2 for grain-size distributions measured at local fuel burnups
f_ ¼ 2 1019 m3 s1 as a function of the fuel far beyond the recrystallization burnup.
Modeling of Fission-Gas-Induced Swelling of Nuclear Fuels 617
12
0.24
0.22 Theory
Recrystallized grain size (μm)
0.20
0.18
0.16
0.14
0.12
0.10
0.0 2.0E + 14 4.0E + 14 6.0E + 14 8.0E + 14 1.0E + 15 1.2E + 15
Fission rate (⫻1020 m–3 s–1)
Figure 29 Calculated recrystallized grain size in UO2 at T ¼ 923 K as a function of the average fission rate f._
Reproduced from Rest, J. J. Nucl. Mater. 2006, 349(1–2), 150–159.
Figure 32 shows the calculated recrystallized grain grain size in the recrystallized region of the fuel from
size in UO2 as a function of average linear power. The the work of Spino et al.71 The theoretical results shown
fuel temperature used to calculate the recrystallized in Figure 32 indicate that there is a relatively small
grain size in Figure 32 was estimated on the basis of change in the grain size for average linear power >18
the relationship between linear power and fuel tem- kW m2. Between 14 and 18 kW m2, the calculated
perature developed by Bagger et al. (e.g., see eqn [12] in grain size increases from 0.15 to 1 mm before flatten-
Bagger et al.70). Also shown in Figure 32 is the average ing out for linear power <14 kW m2.
618 Modeling of Fission-Gas-Induced Swelling of Nuclear Fuels
12
ΔS = 2 ⫻ 10–6 J m–3 K–1
10 ΔS = 2 ⫻ 10–7 J m–3 K–1
ΔS= 0
Recrystallized grain size (μm)
8
1E + 1
ΔU
1E + 0 ΔS = 2 ⫻ 10–6 J m–3 K–1
ΔS = 2 ⫻ 10–7 J m–3 K–1
1E – 1
Energy (J m–3)
1E – 2
1E – 3
1E – 4
1E – 5
1E – 6
700 750 800 850 900 950 1000 1050
Temperature (K)
Figure 31 Calculated volumetric strain energy in UO2 for f_ ¼ 2 1019 m3 s1 as a function of the fuel temperature
compared with the various values for the assumed volumetric configurational entropy change upon recrystallization.
Reproduced from Rest, J. J. Nucl. Mater. 2006, 349(1–2), 150–159.
Also shown in Figure 32 are the results of calcu- grain size on the linear power for values of the linear
lations made without the effect of including the power <15 kW m2.
entropy term (i.e., S ¼ 0) in eqn [113]. As is evident Unfortunately, for U–xMo there is a much greater
from Figure 32, not including this mechanism results uncertainty in the properties (e.g., those in Table 4)
in a relatively strong dependence of the recrystallized than for UO2. Figure 33 shows the calculated average
Modeling of Fission-Gas-Induced Swelling of Nuclear Fuels 619
3.0
2.5
1.0
0.5
0.0
12 14 16 18 20 22 24 26
Linear power (kW m–2)
Figure 32 Calculated average recrystallized grain size in UO2 as a function of linear power compared with measured
values. Reproduced from Hofman, G. L. Private Communication, Argonne National Laboratory, Jun 2005. Reproduced from
Rest, J. J. Nucl. Mater. 2006, 349(1–2), 150–159.
1.0
Theory (fission rate = 4.3 ⫻ 1020 m–3)
0.4
0.2
0.0
recrystallized grain size in U–10Mo made using the Table 4 are listed in Table 5. The results shown in
properties listed in Table 5 as a function of fuel Figure 33 indicate that the theoretical calculation for
temperature compared with the trend of measured the recrystallized grain size in U–10Mo is consistent
values.73 Only the properties that differ from those in with the trend of the data.
620 Modeling of Fission-Gas-Induced Swelling of Nuclear Fuels
Table 6 Description of fuel used in the analysis: also listed are calculated values for Fdx and dgx
Plate ID Burnup Fission rate Fuel Fission density Fdx : eqn [88b] Recrystallized/
(at.% U) (1020 f m3 s1) temperature F (1027 f m3) (1027 f m3) dxg : eqn [97]
(
C)
3.5E + 8
V07: 5.8% U
3.0E + 8 V03: 7.4% U
623AD: 15.6% U
Theory: 15.6% U
2.5E + 8
1.5E + 8
1.0E + 8
5.0E + 7
0.0
V V V In eqn [137], it has been assumed that in the recrys-
¼ ð1 Vr Þ þ Vr ½135 tallized region of the fuel the majority of the generated
V T V g V gx
V
gas is on the grain boundaries. This assumption is
V
V g is given by eqn [128], Vr by eqn [110],
where consistent with the fractional gas release calculated
and V gx is given by using eqn [40] for 4p2 Dg t =dg2 > 1.
! Figure 35 shows data for U–xMo total fuel
V Cb Cbx 1 swelling behavior versus burnup74 compared to the
¼ 4pRbx
3
þ þ 3 ½136
V gx dg dgx 3dgx theory (i.e., eqns [110], [128], [135], and [136]) for
two values of the nominal recrystallized grain size.
In eqn [136] it has been assumed that upon recrystal- The nominal recrystallized grain size dgx represents
lization the preexisting density of bubbles on the the average grain size of the distribution of recrystal-
boundaries Cb remains relatively unchanged. A new lized grain sizes.71 A calculation of dgx is the subject of
population of bubbles Cbx is formed and is given a paper currently in process by this author. A solid
by eqn [41] with the as-fabricated grain size dg fission product swelling contribution given by
replaced with the recrystallized grain size dgx . The Hofman and Walters75 has been added to eqn [39]
term 1=dgx3
represents the density of triple points per V
¼ 0:014BU ½138
unit volume. The triple point nodes are considered V S
very efficient sinks and as such it is assumed that a gas
where BU is the percent burnup with respect to all
bubble will form at each node. It has also been
the metal atoms.
assumed in eqn [136] that the radius Rbx of the pre-
From Figure 35, it is clear that the theory follows
existing grain-boundary bubbles, the newly formed
the trends of the swelling data as a function of burnup
bubbles, and the triple point bubbles are approxi-
across the transition from no recrystallization to full
mately the same. Rbx is given by eqn [130] with Nb
recrystallization. The three micrographs located above
replaced by Nbx where
the graph in Figure 35 show, respectively, the fuel
Vr bf_ t prior to recrystallization, at the initiation of recrys-
Nbx ¼ ! ½137 tallization, and when 100% of the fuel volume has
Cb Cbx 1
3 þ þ 3 been recrystallized. As seen in the second micrograph
dg dgx 3dgx of Figure 35, the initiation of recrystallization occurs
5 mm
70
60
Philips
40 Leeser
Shaudy
30 WAPD Theory: dgx = 0.2 μm
Kramer
Theory: dgx = 0.5 μm
20 RERTR-4
IRIS-1
Recrystalization RERTR-5
10 Future
0
0 20 40 60 80
Burnup (at.% 235U)
Figure 35 U–xMo fuel swelling behavior versus burnup compared with theory for two values of the nominal recrystallized
grain size. Reproduced from Rest, J. J. Nucl. Mater. 2005, 346(2–3), 226–232.
Modeling of Fission-Gas-Induced Swelling of Nuclear Fuels 623
primarily along the original grain boundaries. The to the initiation of this irradiation-induced transfor-
recrystallized grain size measured from the micro- mation. From Figure 36, it is also clear that the
graphs in Figure 35 is 0.2–0.5 mm. effect of the progression of recrystallization (as com-
Figure 36 shows U–xMo fuel swelling behavior pared to simultaneous recrystallization of the whole
versus burnup for two values of the nominal recrys- grain) is greater when the as-fabricated grain size
tallized grain size. The effect of the progression of is larger.
recrystallization, shown in Figure 36, is a marked Figure 37 shows UO2 fuel swelling behavior on
suppression in the swelling (solid line as compared the basis of measured relative immersion density22
to the dashed line, and dash-dotted-dotted line as versus burnup compared with theory (dgx ¼ 0:2 mm)
compared to the dotted line in Figure 36) subsequent for two values of the microscopic strain-rate parame-
40
Theory with progression
35 Theory with no progression
(ΔV/V0)f (%)
30
dg = 9 μm
25
dg = 4 μm
20
15
45 50 55 60 65 70
Burnup (at.% 235U)
Figure 36 U–xMo fuel swelling behavior versus burnup with and without the effects of the progression of
recrystallization (dgx ¼ 0:4 mm) for two values of the as-fabricated grain size. Reproduced from Rest, J. J. Nucl. Mater.
2005, 346(2–3), 226–232.
0.98
0.96
0.94
Relative density
0.92
0.90
ρ 2
= 0.983 − 1.7 ⫻ 10–4 BU3/2 ± 0.005
ρth
0.88
B2 = 2 ⫻ 10–29
B2 = 10–29
0.86
0 20 40 60 80 100 120
Pellet average burnup (GWd tM–1)
Figure 37 UO2 fuel swelling behavior based on measured relative immersion density versus burnup compared with
theory (dg ¼ 7 mm; dgx ¼ 0:2 mm) for two values of the microscopic strain-rate parameter. Reproduced from
Spino, J.; Rest, J.; Goll, W.; Walker, C. T. J. Nucl. Mater. 2005, 346, 131; Rest, J. J. Nucl. Mater. 2005, 346(2–3), 226–232.
624 Modeling of Fission-Gas-Induced Swelling of Nuclear Fuels
ter. The dotted lines in Figure 37 indicate the spread The entropy term in eqn [11] is responsible for
in measured density values. A solid fission swelling flattening out the calculated recrystallized UO2 grain
rate of 0.32%/10 GWd tM1(54) was added to eqn size versus temperature curve for fuel temperatures
[135] in order to calculate the total fuel swelling. <800 K (Figure 5). It is interesting to note that this
It is interesting to note that the UO2 solid fission term has a negligible effect on the calculated recrys-
product swelling is 3 times larger than that used tallized grain size for the U–10Mo irradiation condi-
for U–xMo (eqn [138]). For burnups greater than tions shown in Figure 7. On the other hand, the
80 GWd tM1, a value of B2 equal to that used for calculated dislocation density given by eqn [19] con-
U–xMo results in UO2 swelling rates that are larger tinues to decrease as the temperature decreases. This
than those measured. The results shown in Figure 37 result is in apparent contradiction to some observa-
suggest that the value of B2 in UO2 is about a factor of tions that the dislocation density in UO2 is relatively
2 smaller than that used for the U–xMo alloy (e.g., see constant across the fuel radius.80 However, in a
Figure 35). later publication, Ray et al.67 found about 1 order of
magnitude difference in the dislocation density and
about 4 orders of magnitude difference between
3.20.5.9 Discussion and Conclusions
dislocation loop densities measured in the rim and
The basic premise of the theory proposed in this in the center of the pellet irradiated to 74 GWd tM1.
work is that irradiation-induced recrystallization is One possibility that might account for the apparent
driven by the behavior of the interstitial loops in the low-temperature behavior (saturation) of the net-
material. Interstitial loops interact and lead to the work dislocation density is to include an additional
formation of dislocations and a cellular dislocation term in the expression for the loop climb controlled
structure. Evolving subgrains are characterized by glide velocity given by eqn [102], that is,
relatively low angles and remain thus until recrystal- 2Ziv
lization is initiated, which leads to the formation of nl ¼ Di ci þ zG ½139
bv
nanometer size grains having large angle boundaries.
However, Ray et al.76 report that the grain-boundary The last term in eqn [139] represents the motion of
angles are exclusively low (<5
) in the rim region. loops due to stress-induced glide where G is the shear
This observation supports the microstructural evo- modulus and z is a parameter that sets the magnitude
lution through polygonization and is inconsistent of this stress (e.g., Pierls stress). This term is only
with recrystallization even though the local burnup significant at relatively low temperatures (e.g.,
of the samples exceeded 200 GWd tM1. Similarly, T =TM < 0:3) where diffusion rates are low and,
Sonoda et al.77 report a low-angle subgrain struc- thus, the first term in eqn [139] is small.
ture in 90 GWd mT1 samples and, thus, support The theory presented in this work is novel in
polygonization as the prime mechanism for grain the sense that a complete picture consisting of the
refinement. initiation and progression of recrystallization and
On the other hand, Thomas et al.78 and Nogita and the determination of the recrystallized grain size is
Une79 report a mixture of low and high angle submi- provided by analytically derived expressions. In addi-
cron grains in the rim region at 100 GWd mT1. tion, analytical expressions are provided for the time-
These observations support the further evolution of dependent network dislocation density and change in
low-angle subgrains into recrystallized grains. lattice parameter.
The theory described here supports the latter view- It has been demonstrated that these phenomena
point that not only polygonization, but also recrystalli- can be simulated in both UO2 and in U–xMo with the
zation is occurring in these materials. Recrystallization same theory, albeit with various property differences.
provides a mechanism for the reduction in the lattice That this is possible reflects the likelihood that the
displacement as shown in Figure 2. In addition, recrys- defect behavior in both materials in the temperature
tallization provides a mechanism for the lowering of regime where recrystallization is observed to occur is
the free energy, consisting of decreasing volumetric characterized by the uranium interstitial, by the asso-
strain energy and increasing entropy, in exchange for ciated formation of interstitial loops, and by athermal
the creation of new surfaces associated with the newly diffusion of gas atoms. Results of the calculations are
formed recrystallized grains. Indeed, this scenario is consistent with the trends of the available data.
consistent with observations of relatively defect-free Calculation of intergranular bubble-size distribu-
interiors of the recrystallized grains in UO2.65 tions in U–10Mo alloy fuel made with a mechanistic
Modeling of Fission-Gas-Induced Swelling of Nuclear Fuels 625
model of grain-boundary bubble formation kinetics is the atomistic calculations are well suited (given real-
consistent with the measured distributions pre- and istic interatomic potentials) to calculate critical
postirradiation-induced recrystallization. Analytical model parameters and materials properties that can
solutions are obtained to the rate equations thus be incorporated into the mesoscale models. In addi-
providing for increased transparency and ease of tion, the atomistic models can be used to perform
use. The bubble distribution subsequent to recrystal- single-effects computer experiments that can facili-
lization evolves with the same physical kinetics as tate the validation of proposed mesoscale behavioral
that dominating the bubble evolution prior to recrys- mechanisms.
tallization. The results support the general feature that First principle modeling can provide the observa-
any differences in bubble evolution kinetics and mor- tions that are impossible to obtain from in-pile experi-
phology between pre- and postirradiation-induced ments. This resource can be useful as the development
recrystallization regimes are due primarily to gas con- of improved experimental methods have yielded
tent and initial and/or boundary conditions. results that motivate further theoretical investigation.
For example, it has been observed that bubble nucle-
ation mechanisms on the grain boundary do not appear
3.20.6 Final Thoughts to proceed according to established theory, and that
other processes may be involved. Similarly, many ques-
Analytical models are characterized by their flexibil- tions still abound regarding the exact mechanism for
ity and portability. The aim of these models is to gas-atom diffusion within the fuel. Even in the rela-
approximate the behavior of a fuel under one or a tively short run time of the proposed atomistic models,
variety of different conditions (e.g., see Rest,10 Rest processes such as the two just mentioned may be read-
and Zawadzki,81 Wood and Matthews,82 Matthews ily observable.
and Wood,83 Noirot84). Because the focus is on fuel The efficiencies of analytical, physically based
behavior, it is not only possible to describe the models provide compelling reasons to expand this
observed phenomenon from these models, but also kind of research. Mechanistic models that provide
to predict potential issues that have not yet occurred. accurate representations of fuel behavior are valuable
This is possible, even under varying conditions. Most tools that can be used to assess and evaluate a range
importantly, because many fuels share similar com- of fuels in a variety of operating environments.
position and material characteristics, analytical mod- (See also Chapter 1.01, Fundamental Properties of
els can be translated with minimal modifications to Defects in Metals; Chapter 1.02, Fundamental Point
model a range of different fuels. Defect Properties in Ceramics; Chapter 1.07, Radi-
The major impediment to accurate portability of ation Damage Using Ion Beams; Chapter 1.12,
analytical models is the relative shortage of fuel Atomic-Level Level Dislocation Dynamics in Irra-
property values. The property values, like some of diated Metals; Chapter 2.19, Fuel Performance of
the examples given above, form the qualitative dif- Light Water Reactors (Uranium Oxide and MOX);
ferences in nuclear fuels and often drive the dominat- Chapter 3.01, Metal Fuel; Chapter 3.19, Oxide
ing mechanisms that are manifested. Much can be Fuel Performance Modeling and Simulations and
learned from first principle computer simulations Chapter 3.23, Metal Fuel Performance Modeling
and there seems to be great promise in the develop- and Simulation).
ment of atomistic and molecular models. These
models may be able to provide accurate gas-atom
and vacancy, and interstitial diffusion coefficients,
References
bubble nucleation and re-solution rates, and values
for the thermal conductivity of the fuel, as well as
1. Hofman, G.; Snelgrove, J. In Materials Science and
many other properties that are presently unavailable. Technology; Frost, B. R. T., Ed.; VCH: Weinheim, 1994;
A promising area is the coupling between lower Vol. 10A.
length scale atomistic and higher length mesoscale 2. Greenwood, G. W.; Foreman, A. J. E.; Rimmer, D. E.
J. Nucl. Mater. 1959, 4, 305.
calculations (e.g., see Insepov et al.85). The mesoscale 3. Turnbull, J. A. J. Nucl. Mater. 1971, 38, 203.
(e.g., rate-theory approach) is well suited to a multi- 4. Whapham, A. D. On the formation of gas bubbles in
variate, mechanistic analysis of irradiation behavior fissile material during reactor irradiation, AERE-R 6595;
1970.
where the calculated quantities can be directly com- 5. Olander, D. R.; Wongsawaeng, D. J. Nucl. Mater. 2006,
pared to measured PIE data. On the other hand, 354, 94.
626 Modeling of Fission-Gas-Induced Swelling of Nuclear Fuels
6. Rest, J.; Hofman, G. L.; Kim, Y. S. J. Nucl. Mater. 2009, 41. Brailsford, A. D. AERE Report No. R 7446, 1973.
385, 563. 42. Manley, A. J. TRG Report No. 1681, 1968.
7. Jackson, R. A.; Catlow, C. R. A. J. Nucl. Mater. 1985, 127, 43. Rest, J. J. Nucl. Mater. 1984, 120, 195.
161. 44. Rest, J. J. Nucl. Mater. 1993, 207, 192.
8. Blin, J. CEA Report 1444; 1960. 45. Rest, J.; Hofman, G. L. J. Nucl. Mater. 2000, 277, 231.
9. Turnbull, D. Solid State Phys. 1956, III, 225–306. 46. Spino, J.; Baron, D.; Coquerelle, M.; Stalios, A. D. J. Nucl.
10. Rest, J. Grass-SST: A comprehensive, mechanistic model Mater. 1998, 256, 189.
for the prediction of fission-gas behavior in UO2-base fuels 47. Matthews, J. R.; Wood, M. H. Nucl. Eng. Des. 1980, 56,
during steady-state and transient conditions, NUREG/ 439.
CR-0202, ANL-78-53; 1978. 48. Hedger, H. J. AERE Report No. M 3151, 1980.
11. Gruber, E. E. J. Appl. Phys. 1967, 38, 243. 49. Reynolds, G. L.; Beere, W.; Sawbridge, P. T. J. Nucl.
12. Fedorov, G. B.; Smirnov, E. A. Diffusion in Reactor Mater. 1971, 41, 112.
Materials; Oxonian Press: New Delhi, 1984. 50. Marlowe, M. O.; Kaznoff, A. I. In Proceedings of the
13. Adda, Y.; Philibert, J. Diffusion of uranium with some International Conference on Nuclear Fuel Performance,
transition metals. In Proceedings of the 2nd United States British Nuclear Society, London, 1973.
International Conference on the Peaceful Uses of Atomic 51. Spino, J.; Vennix, K.; Coquerelle, M. J. Nucl. Mater. 1996,
Energy, Geneva, Switzerland, 1958. 231, 179.
14. Rest, J. Transanctions on RRFM 2009, Marrakech, 52. Tucker, M. O. Radiat. Eff. 1980, 53, 251.
Morocco, Mar 21–25, 2010; ISBN 978-92-95064-10-2. 53. Rest, J.; Birtcher, R. C. J. Nucl. Mater. 1989, 168, 312.
15. Hofman, G. L.; Kim, Y. S.; Robinson, A. B. Fission induced 54. Rest, J. J. Nucl. Mater. 2004, 326, 175.
swelling and creep of uranium–molybdenum alloy fuel. 55. Rest, J. J. Nucl. Mater. 2005, 346, 226.
In Transactions on 13th International Topical Meeting on 56. Rest, J.; Hofman, G. L. In Proceedings of the Materials
Research Reactor Fuel Management, Vienna 2009; ISBN Research Society Meeting, Symposium R, Boston, MA,
978-92-95064-07-2. Nov 2000.
16. Kryger, B. Contribution A L’etude Du Degagement Des 57. Rest, J. J. Nucl. Mater. 2006, 349, 150.
Gaz De Fission Dans Les Combustibles Nucleaires 58. Rest, J.; Hofman, G. L. J. Nucl. Mater. 1994, 210, 187.
Metalliques, CEA-R-3888; 1969. 59. Krivoglaz, M. A. X-Ray and Neutron Diffraction in Nonideal
17. DeHoff, R. T.; Rhines, F. N. Quantitative Microscopy; Crystals; Springer: Berlin, 1995; Chapters 3 and 4.
McGraw-Hill: New York, 1968; p 162. 60. Hanson, N.; Kuhlmann-Wilsdorf, D. Mater. Sci. Eng. 1986,
18. Hofman, G. L. Private Communication, Argonne National 81, 141.
Laboratory, Oct 2009. 61. Nogita, K.; Une, K.; Hirai, M.; Ito, K.; Shirai, Y. J. Nucl.
19. Baker, C. J. Nucl. Mater. 1977, 66, 283. Mater. 1997, 248, 196.
20. Kim, Y. S.; Hofman, G. L.; Rest, J.; Shevlyakov, G. V. 62. Garner, F. A.; Wolfer, W. G. In Effects of Radiation on
Characterization of Intergranular Fission Gas Bubbles in Materials: Eleventh Conference, ASTM STP 782;
U–Mo Fuel; Argonne National Laboratory Report, Brager, H. R., Perrin, J. S., Eds.; American Society for
ANL-08/11, Feb 2008. Testing and Materials: Philadelphia, PA, 1982;
21. Birtcher, R. C.; Donnelly, S. E.; Templier, C. Phys. Rev. B pp 1073–1087.
1994, 50, 764. 63. Matzke, H. Adv. Ceram. 1986, 17, 1.
22. Spino, J.; Rest, J.; Goll, W.; Walker, C. T. J. Nucl. Mater. 64. Marlowe, M. O. J. Nucl. Mater. 1968, 33, 242.
2005, 346, 131. 65. Nogita, K.; Une, K. Nucl. Instrum. Meth. 1994, B91, 301.
23. Speight, M. V. Nucl. Sci. Eng. 1969, 37, 180. 66. Spino, J.; Papaioannou, D. J. Nucl. Mater. 2000, 281,
24. Wood, M. H.; Kear, K. L. J. Nucl. Mater. 1983, 118, 320. 146.
25. Speight, M. V. J. Nucl. Mater. 1971, 38, 236. 67. Ray, I. L. F.; Matzke, Hj.; Thiele, H. A.; Kinoshita, M.
26. Rest, J. J. Nucl. Mater. 2003, 321, 305. J. Nucl. Mater. 1997, 245, 115.
27. Olander, D. R. Fundamental Aspects of Nuclear Reactor 68. Kinoshita, M.; Kameyama, T.; Kitajima, S.; Matzke, H. J.
Fuel Elements; Technical Information Center: ERDA: J. Nucl. Mater. 1998, 252, 71.
Springfield, VA, 1976. 69. Spino, J.; Stalios, A. S.; Santa Cruz, H.; Baron, D. J. Nucl.
28. Rest, J. J. Nucl. Mater 2010, 407, 55–58. Mater. 2006, 354, 66–84.
29. Matzke, H. J. Radiat. Eff. 1980, 53, 219. 70. Bagger, C.; Mogensen, M.; Walker, C. T. J. Nucl. Mater.
30. Hondros, E. D. In Precipitation Processes in Solids; 1994, 211, 11.
Russell, K. C., Aaronson, H. I., Eds.; The Metallurgical 71. Spino, J.; Vennix, K.; Coquerelle, M. J. Nucl. Mater. 1996,
Society of AIME: Warrendale, PA, 1976; p 1. 231, 179.
31. Cundy, M. R.; Von Der Hardt, P.; Loelgen, R. H., Eds. 72. Meyer, M. K.; Hofman, G. L. Properties of U–Mo alloy
J. Nucl. Mater. 1977, 65, 1–335. dispersion fuel. In IAEA Technical Committee on the
32. Van den Berghe, S.; Van Renterghem, W.; Leenaers, A. Development of High Density Uranium–Molybdenum
J. Nucl. Mater. 2008, 375, 340. Dispersion Fuels, Vienna, Austria, Jun 23–25, 2003.
33. Gjostein, N. A. In Diffusion; American Society for Metals: 73. Hofman, G. L. Private Communication, Argonne National
Metals Park, OH, 1973; Chapter 9, p 272. Laboratory, Jun 2005.
34. Fisher, J. C. J. Appl. Phys. 1951, 22, 74; Matzke, Hj. 74. Hofman, G. L.; Finlay, M. R.; Kim, Y. S. Post-irradiation
Radiat. Eff. 1980, 53, 219. analysis of low enriched U–Mo/Al dispersion fuel miniplate
35. White, R. J. J. Nucl. Mater. 2004, 325, 61. tests, RERTR 4&5. In Proceedings of the 2004 RERTR
36. Hofman, G. L.; Kim, Y. S. Nucl. Eng. Technol. 2005, 37, Meeting, Vienna, Austria, Oct 1–5, 2004.
299. 75. Hofman, G. L.; Walters, L. C. In Nuclear Materials;
37. Wood, M. H. J. Nucl. Mater. 1979, 82, 264. Brian Frost, R. T., Ed.; In Materials Science and
38. Turnbull, J. A.; Friskney, C. A. J. Nucl. Mater. 1978, 71, Technology: A Comprehensive Treatment; Cahn, R. W.,
238. Haasen, P., Kramer, E. J., Eds.; Weinheim, New York,
39. White, R. J.; Tucker, M. O. J. Nucl. Mater. 1983, 118, 1. 1994; Vol. 10A, p 1.
40. Dowling, D. M.; White, R. J.; Tucker, M. O. J. Nucl. Mater. 76. Ray, I. L. F.; Matzke, H. J.; Thiele, H. A.; Kinoshita, M.
1982, 110, 37. J. Nucl. Mater. 1997, 245, 115.
Modeling of Fission-Gas-Induced Swelling of Nuclear Fuels 627
77. Sonoda, T.; Kinoshita, M.; Ray, I. L. F.; et al. Nucl. Instrum. 82. Wood, M. H.; Matthews, J. R. J. Nucl. Mater. 1980,
Meth. 2002, B191, 622. 91, 35.
78. Thomas, L. E.; Beyer, C. E.; Charlot, L. A. J. Nucl. Mater. 83. Matthews, J. R.; Wood, M. H. A Simple Operational
1992, 188, 80. Gas Release and Swelling Model – 11. Grain
79. Nogita, K.; Une, K. J. Nucl. Mater. 1995, 226, 302. Boundary Gas; AERE Harwell, Didcot, Oxfordshire,
80. Ray, I. L. F.; Theile, H.; Matzke, H. J. J. Nucl. Mater. 1992, March 1986.
188, 90. 84. Noirot, L. J. Nucl. Sci. Technol. 2006, 43(226), 1149.
81. Rest, J.; Zawadzki, S. A. FASTGRASS: A mechanistic 85. Insepov, Z.; Rest, J.; Yacout, A.; et al. Derivation of
model for the prediction of Xe, I, Cs, Te, Ba, and Sr kinetic coefficients by atomistic methods for studying
release from nuclear fuel under normal and defect behavior in Mo. J. Nucl. Mater. 2011, doi: 10.1016/j.
severe-accident conditions, NUREG/CR-5840; 1992. jnucmat.2011.08.019.
3.21 Matter Transport in Fast Reactor Fuels
M. J. Welland
European Commission, Joint Research Centre, Institute for Transuranium Elements, Karlsruhe, Germany
629
630 Matter Transport in Fast Reactor Fuels
vk Velocity of species k
Symbols
vlattice Velocity of the solid lattice with respect
AAitken Constant of integration from the Aitken
to the nondiffusing ends/laboratory
equation
frame of reference
Dk Self-diffusion coefficient of species k
VU Valence of uranium
Dm,surf Surface diffusion coefficient
VPu Valence of plutonium
Dv,vol Volume diffusion coefficient
x Stoichiometry deviation
Dk Intrinsic diffusion coefficient of
DHvap,k Enthalpy of vaporization of species k
species k
G Jump frequency
D~AB Chemical interdiffusion coefficient
rT Gradient taken at constant temperature
fk Thermodynamic factor of species k
Vk Specific volume of species k
f0 Diffusion correlation factor
gk Activity coefficient of species k
Fx Conjugate thermodynamic force for
k Boltzmann constant
flow of x
l Jump distance
G Gibbs energy
mk Chemical potential of species k
hk Partial specific enthalpy of species k
s Surface energy
hfv Enthalpy of formation of vacancies
sstress Isostatic stress field in the solid
Hf Energy required to make space for the
Subscript
atom at the final site
k Index a particular chemical species
Hi Energy required to remove atom from
n Vacancy defect species
the initial site
i Interstitial defect species
Hm Migration energy
x Index of system state variable
Hs Energy required to prepare the saddle
point for a migrating atom
Jk0 Mass flux of species k measured with
respect to the nondiffusing ends 3.21.1 Introduction
Jk Mass flux of species k
Jq Reduced heat flux The fuel of fast neutron spectrum nuclear reactors
Ju Energy flux may be subject to extreme operating conditions.
Jx Generalized flow of x In particular, centerline temperatures may regularly
K Thermal conductivity exceed 2000 K and incur temperature gradients
lbw Characteristic depth of bow wave in in excess of thousands of degrees per centimeter.
front of migrating pore Such large temperatures and temperature gradients
LAB Kinetic parameter, either diagonal or produce a variety of mass transport phenomena that
cross-term related to mobility affect the chemical and physical state of the fuel.
nk Concentration of species k The physical evolution of the fuel affects the genera-
nek Equilibrium concentration of species k tion and diffusion of heat, in addition to mechanical
Nk Mole fraction of species k behavior and fission product release.
pk Partial pressure of species k The large temperatures and temperature gradi-
Pfill Pressure of the fill gas ents can directly or indirectly produce driving forces
q Atomic ratio of Pu/(U þ Pu) acting on the constituents of the fuel, causing their
Qk;vap Effective heat of transport through migration. Such driving forces act on all the consti-
carrier-gas mechanism tuents of the fuel, be they fission products or the
Q
k Effective heat of transport of species k primary constituents of the solid matrix itself. For
Q
AB Process heat of transport for discussion purposes, mass transport can be consid-
interdiffusion of species A and B ered as resulting in redistribution of the chemical spe-
r Radius cies, which leaves the general structure of the matrix
Rv Source of vacancies intact, and restructuring of the solid matrix itself.
s_ Entropy production rate Restructuring of the solid matrix occurs when, for
T Temperature example, voids migrate through the fuel. In a radial
TC Centerline temperature temperature gradient, voids migrate toward higher
Uk Energy of transport of species k temperatures and coalesce in the center, resulting
Matter Transport in Fast Reactor Fuels 631
in movement of solid toward the outer periphery of penalty of the maximum linear heat rating by
the fuel and forming the central void, an example as much as 8.2 kW m1, 15% for a liquid metal
of which is shown for a mixed oxide (MOX) fuel fast breeder reactor MOX fuel element compared
pellet in Figure 1. With regards to the thermal per- to the restructured case without plutonium
formance of the fuel, this relocation of fuel away from redistribution.1,2
the hot center toward the cooler periphery means Redistribution of chemical species can also degrade
that heat is generated closer to the heat sink, resulting the structural integrity of the fuel cladding by moving
in the lowering of the maximum temperature. Mean- reactive species toward the clad where they can attack
while, the removal of voids from the fuel matrix it. In particular, volatile fission products such as Ce,
changes the heat transport characteristics, and so Cs, Te, Se, Rb, and I, may be transported down the
restructuring can be seen to play a major role in the temperature gradient through cracks and, depending
overall heat transport process. on the local oxygen concentration, they may attack
Redistribution results in demixing of solutions and the cladding.3,4
migration of chemical species such as fission pro- In this chapter, matter transport is discussed as
ducts. For example, in MOX fuels, this phenomenon it occurs in a variety of reactor fuels. Due to the
results in the migration of oxygen through the sample higher operating temperatures of some fast reactor
and demixing of U and Pu in addition to other species fuels, these effects are more pronounced and relevant
present in the fuel. Several material properties are compared to conventional fuels. However, the key
dependent on the chemical state of the fuel including parameter incurring matter transport are usually the
thermal conductivity, heat capacity, density, and the temperature and temperature gradient and therefore
incipient melting temperature. Redistribution may thermal spectrum reactor fuels also exhibit these
also lead to safety concerns as, for example, redistri- effects if operated at high enough temperatures for
bution of Pu up the temperature gradient increases long enough.
the heat generation at the innermost region. As This chapter is divided into sections describing
avoidance of melting is a key concern for solid fuel the mass transport processes on different scales of
operation, this phenomenon may culminate in a space and time. Restructuring and redistribution are
discussed from a common standpoint although the
distinction is maintained for discussion purposes.
The mechanisms described are common to all mate-
rials; however, the characteristics of each fuel affect
the magnitude of each effect, and thus their relative
importance.
First, fundamental mass transport processes are
described in terms of the atomic transport mechanisms
and their thermodynamic driving force. Gradients in
temperature, chemical potential, and stress may all
contribute to matter transport and so are placed on a
common footing, the requirement of entropy produc-
tion. By comprehensively treating all the constituents
of the system, including vacancies, phenomena such as
the Kirkendall effect and Darken’s equation for inter-
diffusion are derived. We, therefore, seek to unite
several theories on a common footing based on irre-
versible thermodynamics, in Section 3.21.2.
The collective behavior of particles under the
action of these fundamental mechanisms leads to
the emergence of microscopic structures, which play
a major role in fuel restructuring and redistribution.
Figure 1 Microstructure of a fast reactor pelletized
In particular, the behavior of grains and voids sig-
mixed oxide fuel showing the effects of restructuring due
to mass transport. A discussion on the visible zones in nificantly affects fuel performance. Different charac-
this sample is presented in Section 3.21.5.1. © European teristics of voids permit their categorization into
Atomic Energy Community, reproduced with permission. bubbles and pores depending on, among other
632 Matter Transport in Fast Reactor Fuels
factors, the relative importance of the fundamental fuel is subject to extreme conditions in both magni-
mechanisms, which depend on the void’s size, shape, tude and gradient of state variables, for example,
and gas content. Grain growth and void behavior are both very high temperatures and very large tempera-
described in Section 3.21.3. ture gradients. Usually, a heat flux is expressed in
Evolution of the fuel through fundamental terms of the temperature gradient only, and mass
mechanisms and microscopic structures leads to mac- fluxes only function in terms of chemical potential
roscopically observable effects of redistribution and (mole fraction). In the extreme conditions produced
restructuring. The evolution of the fuel porosity and during fuel operation, coupling between the trans-
redistribution by phase changes and interconnected port of heat and mass becomes an important
porosity are discussed. The redistribution of chemical consideration in correctly predicting fuel behavior.
constituents is shown to result from the balance of The coupling between mass flux and temperature
driving forces in the solid matrix, through voids or gradient is known as thermodiffusion, or the Soret
as a result of phase change. These large-scale phe- effect, and is critical to the understanding of fuel
nomena are discussed in Section 3.21.4. behavior.
Finally, different fuel types are discussed using the The many interrelated phenomena applicable
developed theoretical framework. Most of the sections to restructuring and mass transport are placed on a
are dedicated to MOX fuels, as they are the most common base through the theory of irreversible pro-
commonly investigated fuels; they may be used in cesses (TIP). Within the TIP, all processes that pro-
pelletized, sphere-pac, and vipac forms. Carbide and duce entropy, and therefore occur spontaneously,
nitride fuels are also discussed, in addition to U–Pu–Zr are accounted for in such a way that coupling
fuels and finally tristructural isotropic (TRISO) par- between phenomena is derived naturally. Mathemat-
ticles. These fuels are discussed in Section 3.21.5. ical relationships between fluxes and thermodynamic
A few features related to restructuring and redistri- driving forces are derived comprehensively and con-
bution are not addressed in this work, but are worthy sistently, and naturally include such concepts as
of mention. High-burnup fuels, both in fast and ther- thermodiffusion.9
mal spectra reactors, have been shown to undergo
other changes in structure and phase. Volatile fission
products may migrate down the temperature gradient 3.21.2.1 The Theory of Irreversible
toward the fuel–cladding gap where they can form Processes
agglomerations, known as the ‘Joint Oxide Gain’ The TIP relates the evolution of a system, as
(JOG).5 At extended burnups, there is also a reaction described by its state variables x, to the change in its
layer observed between the cladding and fuel surface.6 thermodynamic state with respect to change in the
Also evolved is the so-called rim structure, which is state variables. The change in a state variable is
characterized by <1-mm-diameter grains in a porous described by a generalized flow, Jx, where the flow is
recrystallized configuration.7,8 Finally, it is postulated driven by its conjugate driving force, Fx. The driving
that the rim structure may in fact sinter again and form force is given as the gradient of a potential function
a ‘white layer,’ which is nearly free of voids.7 While related to state variable x, for example, a mass flux
these phenomena are undoubtedly relevant to fuel (flow) is driven by a chemical potential gradient. The
operation at longer burnups, this work focuses on fundamental principle of the TIP is that entropy is
behavior at lower burnups. produced when this flow proceeds along its driving
By approaching this topic by beginning with the force:
fundamentals and advancing through spatial scales, X
the connection between the different restructuring s_ T ¼ Jx Fx 0 ½1
x
and redistribution observations in a variety of fuels is
elucidated, and the common principles leading to where s_ is the rate of entropy generation and T is
mass transport are highlighted. the absolute temperature. Flow-driving force pairs
include vectorial fluxes such as heat and mass
flux driven by gradients in temperature and con-
3.21.2 Transport Phenomena centration, respectively, and also scalar flows such
as chemical reactions and phase changes driven
This Section describes some of the fundamental by minimization of free energy (maximization of
mechanisms that drive material transport. Nuclear universal entropy through entropy production).
Matter Transport in Fast Reactor Fuels 633
Note that the expression of the matter flux is in role and may be included as a separate diffusing
line with the Nernst–Einstein equation for matter species, which, unlike the chemical constituents of
flux driven by a force (potential gradient): the system, are not conserved but rather created and
Dk destroyed locally. Defects are included in eqn [5b] as
Jk ¼ nk v ¼ nk Fk ½6 the K þ 1 species.
kT
If we consider a system with a constant density of
where nk and Dk are the local concentration and normal lattice sites, and that diffusion occurs primarily
diffusion coefficient of species k, respectively.12 through defect mechanisms, then not all the K þ 1 mass
Through an argument of time-reversal symmetry, fluxes are independent. The assumption of a constant
it is possible to demonstrate that in the absence of atomic volume is clearly not always correct and may be
rotation or a magnetic field, the L matrix is symmetric: lifted by introducing the partial molar volumes.12 For
Ljj 0 ¼ Lj 0 j ½7 vacancies diffusing on the same crystal
P lattice, the flux
of vacancies must satisfy Jv ¼ Kk¼1 Jk. For inter-
This relationship is known as the Onsager reciprocal stitials, which may occupy a distinct lattice, the sign is
P
relation, and helps to reduce the number of coeffi- reversed and Ji ¼ Kk¼1 Jk . These relations can be
cients required by eqn [5]. exploited to eliminate the defect flux, but then
The nature of the coefficients may be intuited by require a modification to the thermodynamic driving
considering the role they play in linking a flux with the forces as given in eqn [11] to include the chemical
respective driving force. We may expect the diagonal potential of the defects. The choice to eliminate the
components Lqq and Lkk to be related to the thermal defect flux is entirely arbitrary and in some cases,
conductivity and diffusion coefficient of species k, such as the diffusion of oxygen in MOX, it is more
respectively. The off-diagonal terms define cross- convenient to eliminate a material flux instead.13 For
effects between fluxes such as Lkq relating to mass the sake of simplicity, only vacancies are considered
transport driven by a temperature gradient.16 The for the remainder of the derivation.
Onsager relation equates Lqk ¼ Lkq, where Lqk is related The creation and annihilation of vacancies consti-
to a heat flux being driven by a chemical potential tutes a flow in the sense of eqn [1], where the driving
gradient, a phenomenon known as the Dufour force is expressible as the deviation of the vacancy
effect.17 The Soret and Dufour effects are, therefore, chemical potential to its equilibrium value. To describe
considered reciprocal relations stemming from the this scalar flow (source), the source of vacancies
same source. Regarding the mass fluxes, the Lkk0 will be denoted as Rv. The linear phenomenological
coefficients are related to correlation coefficients theory then implies a source/sink of vacancies where
between mass fluxes of different species and are dis- creation or annihilation occurs due to local sub- or
cussed in more detail below.12 Logically, Lkk0 ¼ Lk0 k supersaturation:
as one may expect that the correlation effect of
one particle flux on the other applies equally so in @G
Rv / ½8
vice versa. @nv
We have now used the TIP to obtain expressions
for heat and mass fluxes, which may occur in nuclear nev nv
Rv ¼ Lv mv ¼ Lv kT ½9
fuel. In the next section, this framework is applied to nev
specific cases of matter transport in the solid and where Lv is a phenomenological kinetic coefficient
vapor phases. akin to the other L coefficients, and nev and nv are
equilibrium and actual defect concentration, respec-
3.21.2.2 Solid State Mass Transport tively. Generally, the Lv coefficient depends on the
concentration of vacancy sources and sinks. The
This section discusses the mass transport that occurs chemical potential of the vacancies, mv may be posi-
in the solid state, beginning with an examination tive or negative, but by definition, must be zero if
of an isothermal system before proceeding to the the vacancies are at their equilibrium concentration.
nonisothermal case. This does not imply that the vacancy concentration is
constant everywhere, but merely that it is equal to the
3.21.2.2.1 Mass transport equilibrium concentration value. The equilibrium
In considering mass transport of the K species in an value may vary with chemical composition and tem-
isothermal system, point defects play an important perature. Thermal equilibrium of the vacancies will
Matter Transport in Fast Reactor Fuels 635
The lattice, with a site concentration, n, moves with The Darken equation is a convenient approxima-
respect to the laboratory at a speed v lattice, which is tion for the chemical interdiffusion coefficient cal-
related to the vacancy flux: culated from the diffusion of radioactive tracers.
The intrinsic diffusion coefficients from eqn [16]
JA þ JB ¼ Jv ¼ nv lattice ½17
are taken and combined with the self-diffusion coef-
which, using eqn [15] becomes: ficient expressions in terms of the radioactive tracer
experiments from eqn [23]. If the correlation factor
v lattice ¼ DA DB rNA ½18
and the vacancy wind terms are assumed to cancel,
We can now perform a coordinate transformation then the intrinsic diffusion coefficients are expressed
to express the mass fluxes in the laboratory frame of simply in terms of the radioactive tracer diffusion:
reference in which they are typically measured using DA ¼ DA f ½24
JA0 ¼ JA þ nA v12:
an expression, that is particularly useful in eqn [20]
JA0 ¼ D~AB rnA ½19a because, as previously mentioned, f is equal for both
diffusing species in a binary solution:
JB0 ¼ D~AB rnB ½19b
D~AB ¼ ðNA DB þ NB DA Þf ½25
where the chemical interdiffusion coefficient
This equation is commonly used in the analysis of the
between species A and B, D~ has been defined:
experimental data.21,22
D~AB ¼ NA DB þ NB DA ½20 The approximation of neglecting the correlation
coefficients in the derivation of Darken’s equation
The matter fluxes in the laboratory frame of reference,
was lifted by Manning who determined other rela-
JA0 and JB0 , are what is commonly predicted in mass
tions relating the intrinsic diffusion coefficient to the
transport calculation and typically observed in experi-
tracer diffusion including the presence of the corre-
mentation. As one would expect in a system in which
lation factor, in addition to the vacancy wind13,20:
the local density does not vary in time, JA0 ¼ JB0 . If
the velocity of the local lattice is measured with inert g
D
AB ¼ ðNA DB þ NB DA Þf
markers, the intrinsic diffusion coefficients may be ½26
1 f0 NA NB ðDA DB Þ2
calculated using eqn [18]. 1þ
In the limiting case of infinite dilution of A in B, f0 ðNA DA þ NB DB ÞðNA DB þ NB DA Þ
it is clear that nA ! 0 and f ! 1. Furthermore, it can Similarly, there is a factor of f 1 on the Kirkendall
be shown that while LAA =nA approaches a finite velocity given in eqn [18]. Allnatt and Lidiard
value, LBA =nB approaches zero. Thus, have noted that the correction for the vacancy wind,
LAA although significant, will never be very large and
lim DA ¼ kT ¼ DA ½21 provide an example of a 30% increase over the
nA !0 nA
Kirkendall velocity calculated with eqn [18].13
where DA is the self-diffusion coefficient of A. The A subject of particular interest to the nuclear
same cannot be said about the DB in the infinite community is that of radiation-induced diffusion.
dilution of A in B.12 This is often described as an athermal contribution
A special case of the above theory is related to to the diffusion coefficient of slow-moving species
the diffusion of a radioactive tracer, A* in a matrix through the solid matrix in which the diffusion
of A. In this case, the mobility of the diffusing species coefficient may be described by23:
is equal and they mix ideally, implying that the
Dkrad / fission rate ½27
intrinsic diffusion coefficients are equal. From eqns
[21] and [16], we have: and may lead to significant enhancement by as much as
1010, especially in the cold regions where thermally
LAA nA
D A ¼ D A ¼ D A ¼ D A 1 ½22 activated self-diffusion is slow.23 As described by
LAA nA
Matzke, the reason for this is the damage to the lattice
structure and the nuclear collision process, as well as
DA ¼ DA fo ½23
local thermal and pressure effects due to the slowing-
where f0 is the correlation factor, related to the prob- down process. These considerations, therefore, con-
ability of a diffusing tracer atom interacting with the tribute to an enhancement of kinetic processes or
vacancy.12 produce a system out of thermodynamic equilibrium.23
Matter Transport in Fast Reactor Fuels 637
The basic equations of mass transport in a binary Jq ¼ LqA rT ðmA mv Þ LqB rT ðmB mv Þ
system have thus been derived, and the relationships
1
between different diffusion coefficients formally Lqq rT ½29e
established. The equations in [19] reconcile with a T
basic understanding of Fickian mass flux, and experi- The definition of the reduced heat flux presented in
mental observations, such as the movement of inert eqn [2] must now include the partial specific
markers in interdiffusion experiments, have also enthalpy of the vacancies:
been justified within the same framework. The pre- X
K þ1
cise meaning of the different diffusion coefficients Jq ¼ Ju hk J k
such as D~ has been rigorously defined and methods k
of determining this value from related experiment
¼ Ju ½ðhA hv ÞJA þ ðhB hv ÞJB ½30
such as radioactive tracers has been explained. This
is critical in determining the behavior of chemical System [29] may be expressed in a simplified form by
redistribution in nuclear fuel elements.24,25 two considerations. Firstly, the gradient of the vacancy
chemical potential may be expressed in terms of the
3.21.2.2.2 Mass and heat transport enthalpy of formation as described in eqn [10].
Having discussed mass transport in Section 3.21.2.2.1, Secondly, we can eliminate the cross-terms between
we now lift the isothermal condition and deduce the heat and mass fluxes by defining the reduced heat of
ramifications of considering a temperature gradient. transport of species k, Q k , as:
We also see that the coupling appears between heat X
K
transport and mass transport and introduces new Lkq ¼ Lki Q i ½31
considerations in the treatment of each. i¼1
Returning to the description of heat transfer The reduced heat of transport is related to the (unre-
described in eqn [5a], if we consider a homogeneous duced) heat of transport via the partial specific
system with no diffusion, then the heat flux is simply enthalpy in a manner similar to eqn [30].13
Fourier heat flux: The system in eqn [29] then becomes:
1 JA ¼ LAA rT mA LAB rT mB
Jq ¼ Lqq rT ¼ K rT ½28
T
1
½LAA ðQA hvf Þ þ LAB ðQB hvf Þ rT ½32a
and so we conclude that for a nondiffusing system, T
Lqq ¼ TK.
We now apply eqn [5] in its entirety for the case of JB ¼ LBA rT mA LBB rT mB
diffusion in a nonisothermal system. Once again, 1
diffusion is considered to occur by the vacancy mech- ½LBA ðQA hvf Þ þ LBB ðQB hvf Þ rT ½32b
T
anism, and so we arrive at a set of equations resem-
bling eqn [11] into which the temperature gradient is Jv ¼ JA JB ½32c
introduced. In the frame of reference moving with
the lattice we have: Rv ¼ Lv mv ½32d
JA ¼ LAA rT ðmA mv Þ LAB rT ðmB mv Þ 1
Jq ¼ Q A JA þ Q B JB þ ðLqq LAq LBq Þ rT ½32e
1 T
LAq rT ½29a
T Expression [32e] reveals the nature of the heat of
transport because, for an isothermal system, this
JB ¼ LBA rT ðmA mv Þ LBB rT ðmB mv Þ equation shows Q k as relating the heat flux with the
1 flux of species k, in excess of the enthalpy associated
LBq rT ½29b with the transported mass. This notion also reconciles
T
with the description of Wirtz for the heat of transport,
as discussed below. Allnatt and Lidiard comment that
Jv ¼ JA JB ½29c
in terms of determining the heat of transport, while
eqn [32a] lends itself to experimental determination,
Rv ¼ Lv mv ½29d [32e] may be used with rT ¼ 0 in computational
638 Matter Transport in Fast Reactor Fuels
techniques such as molecular dynamics, in which consider interdiffusion of this metal species (M) and
isothermal systems are more tractable.13 the fission product (F). This is tantamount to consid-
We may continue to simplify eqn [32] by the ering a dilute solution of A in B, in which case eqn
introduction of the effective heat of transport: [36] may be simplified greatly and expressed in terms
of tracer diffusion coefficients as:
Q
k ¼ Q k hv
f
½33
N
JFo ¼ nDF rNF F2 Q rT ½39
Now, we see that if the effective heat of transport is kT FM
positive, the species will preferentially migrate down
where Q
FM 1=DF DM Q M Q F . This is applied to
the temperature gradient. If it is negative, the species
Cs in Section 3.21.5.1.
will migrate up the gradient.
It can be seen that eqns [32 a–d] are similar to the
isothermal binary interdiffusion in eqn [11], and may 3.21.2.2.2.1 Single component
be treated as before in order to calculate the intrinsic vacancy/interstitial diffusion
diffusion coefficients. The mass fluxes equivalent to There is a special and useful case remaining, which
eqn [15] are then: differs from the above treatments by a fundamental
assumption. Previously, diffusion has been consid-
rT ered to occur via a vacancy mechanism where the
JA ¼ DA rnA ½LAA QA þ LAB QB ½34
T vacancies are present in local equilibrium in small
numbers. In some cases, it is convenient to have
rT
JB ¼ DB rnB ½LBA QA þ LBB QB ½35 a considerable vacancy population and to treat them
T as a significant interdiffusing species. Such a case is
These fluxes are defined relative to the lattice frame of relevant in considering thermodiffusion of oxygen
reference, moving at a speed calculated by eqn [17]. excess or deficiency in hyper- and hypostoichio-
Again, we arrive at a cumbersome equation due to metric fuels, respectively, in which case the defects
the presence of the cross-terms. If the cross-terms present are due to a nonstoichiometry rather than
(and correlation factor) are dropped as was done being intrinsically generated. This is discussed in
in defining Darken’s equation, then we arrive at an detail in Section 3.21.5.1. Note also that this diffu-
equation that is useful in experimental analysis: sion occurs on a different lattice than that hosting the
nNA NB h i1 cation diffusion.
JAo ¼ D~nAB rNA DA QA DB QB rT ½36 We no longer consider the creation and destruc-
kTf T
tion of crystal planes associated with the previous
with JB0 ¼ JA0 . Bober and Shumacher simplify eqn treatment of vacancy flows. In this case, the noni-
[36] by defining a process heat of transport, which com- sothermal diffusion system, analogous to eqn [29] is:
bines the reduced heats of transport in a form more
1
convenient for evaluation from measured concentra- JA ¼ LAA rT ðmA mv Þ LAq rT ½40a
tions distributions26: T
¼ 1
Jv ¼ JA ½40b
Q AB DB Q
B DA Q A ½37
~
DAB
1
In which case, the flux of A measured in the labora- Jq ¼ LqA rT ðmA mv Þ Lqq rT ½40c
tory frame of reference would be given by: T
The use of eqn [10] is no longer possible to express
Q
JAo ¼ nD~AB rNA þ NA NB AB rT ½38 the gradient of mv in terms of hvf because, as men-
kT 2
tioned, the vacancies are not at thermodynamic
If tracer diffusion coefficients are available, D~, DA , and equilibrium. Rather, we note that NA þ Nv ¼ 1 and,
DB may be replaced using eqns [24] and [25]. This through eqns [13] and [31], we can determine an
equation is useful in determining the redistribution expression for Jv:
of the dominant chemical species, for example, ura-
1 1 Q v
nium and plutonium in MOX as discussed in Section Jv ¼ Lvv kT þ f rNv þ 2 rT ½41
3.21.5.1. In order to consider thermodiffusion of fis- NA Nv kT
sion products, which are present generally in dilute
NA Nv Q v
solution at low burnup, we can treat the uranium or ¼ nDv rNv þ rT ½42
plutonium ions as a single metal species and then f kT 2
Matter Transport in Fast Reactor Fuels 639
the temperature gradient, and results in the steady by eqn [31] and an electric force acting on the diffus-
state equation: ing ions.36 Electrons (or holes) act as heat carriers
Hf H i Q by absorbing energy at hot regions and diffusing to
d lnðN Þ ¼ dT ¼ 2 dT ½51 cooler regions where the energy is released. Less
kT 2 kT
energetic charge carriers must flow up the tempera-
which is identical to eqn [48] for dilute solutions, ture gradient in order to maintain charge neutrality.
with the heat of transport identified as: There are two processes in which the charge car-
Q ¼ Hi Hf ½52 riers may affect migration. The first is through direct
interaction between energetic carriers and migrating
the energy at the saddle point having been found to ions, biasing their jump according to the effective
cancel. The above theory also allows us to derive a charge of the ion.36 The effective charge need not
relation between the activation energy for diffusion be of the same magnitude or sign as the chemical
and the heat of transport: valence as it contains contributions from the proba-
bility of interaction with electron and hole, the latter
jQ j Hm ½53
of which possesses a negative sign.
Doubt has, however, been cast on this conclusion by The second interaction is indirectly through the
the work of Jones et al. whose theoretical work pre- generation of an electrostatic potential across the tem-
dicts the heat of transport to be larger than the perature gradient, a phenomenon known as the
migration enthalpy in solid argon.30,31 Nonetheless, thermoelectric effect and characterized by the thermoelec-
eqn [53] implies that Q* Hm, as is sometimes used tric power, also known as the Seebeck coefficient of the
as an approximation in calculations considering this material. Migrating charge carriers are scattered by
phenomenon.9,32,33 impurities, phonons, etc. If this scattering is energy
Another conclusion of eqn [53] is that mea- dependent, an electrostatic potential may be estab-
surements of the heat of transport may provide infor- lished in the material. The electrostatic potential
mation on how the activation energy for diffusion is may then influence the diffusion of the electrically
divided between the initial, intermediate, and final charged ions via their valence.
positions of the diffusing particle. If the lattice requires The contribution of the electrical process to the
a relatively large amount of energy to create a path for heat of transport will not be written explicitly. How-
the particle, then the heat of transport must be small. ever, the notion must be borne in mind when inter-
If, on the other hand, the lattice is relatively open preting out-of-pile experimental data, as mentioned
along the migration path, then the migration enthalpy by Bober and Schumacher with regards to experi-
must originate in the binding energies of the atoms ments involving a gas-tight rhenium capsule.40 As
in initial and final sites, in which case the potential discussed in detail in Section 3.21.5.1 this effect
difference in energies may produce a more signifi- may be included within the TIP by considering the
cant Q*.34–36 flux of charge carriers, and their creation and
Alternate explanations suggest that the value of Q k destruction as a separate diffusing species.
might also be explained by the lattice energy asso-
3.21.2.2.3 Other thermodynamic driving
ciated with the movement of the species in the tem-
forces
perature gradient. This may be calculated directly for
Thus far, the thermodynamic driving forces of the
Q k ,27 or proposed to calculate Q k based on thermo-
type described in eqn [5] were the temperature
dynamic functions by assuming that U* in eqn [50]
gradient and the gradient in chemical potential as
is small in comparison to hk.37 In such a case,
resulting from a concentration gradient. While they
Q* hk.38,39 The partial molar enthalpy of k, hk is
are arguably the most important driving forces for
then calculated from thermodynamic information
nuclear fuel, there are other considerations that also
on the molar heat of solution of species k in the
produce observable effects.
solid, or may be related to the formation enthalpy
of a Frenkel pair.38,39 3.21.2.2.3.1 Stress-induced diffusion
Another explanation for the heat of transport sug- Atoms may experience a force due to a gradient in
gests that rather than being based on lattice energies, the stress field. This may lead to a driving force for
it is influenced by the electric processes in the fuel.26 atomic movement and, similarly, for void migration.41
The observed heat of transport would then be in fact The flux force experienced by atoms is given by the
a superposition of the true heat of transport, defined spatial derivative of the stress gradient.13 In addition,
Matter Transport in Fast Reactor Fuels 641
stress-induced diffusion can lead to the macroscopic chemical potential of the species in the solid should
phenomenon of creep. equal the chemical potential in the gas phase which,
for a perfect gas, is given by:
3.21.2.2.3.2 Phase change
Coexistence/contact between two phases may also pk ! mk;vap ¼ m0k;vap þ kT lnðpk Þ ½55
lead to mass redistribution if the phase change is
where pk is the partial pressure of species k.
noncongruent. The cause for this is the equilibrium
Alternatively, solid constituents may react with
between chemical potentials for species in either
other species present in the gas phase resulting in
phase. This is the principal of Gibbs energy minimi-
movement via a carrier-gas mechanism, as discussed
zation, which is an expression of entropy production
further below. Once in the vapor phase, these species
and, therefore, falls in line with the TIP. This effect is
are free to diffuse in the gas phase through the void
manifest in fuels, which exhibit multiple phases, such
space and may reach points where the physical state of
as metallic U–Pu–Zr, where three phases are under-
the solid is different enough for the vapor to condense.
stood to exist at different points in the temperature
Depending on the vapor pressure and the extent of the
gradient as discussed in Section 3.21.5.3. Redistribu-
void space, this phenomenon can produce a faster
tion by this mechanism also manifests in solid–liquid
matter transport mechanism than solid state diffusion.
transitions for partially molten fuels, as discussed in
The evaporation–condensation mechanism will
Section 3.21.4.3.
produce restructuring as material is physically trans-
3.21.2.2.3.3 Curvature ported to other areas. In addition, the collection of
Atoms at the interface of a phase possess an excess vapor species evolved does not necessarily have the
surface energy, s, and may also have a modified chemi- same overall chemical composition as the solid. Thus,
cal potential due to the curvature of that surface. This this process may separate the constituents of the solid
effect originates in the slightly different number of by favorably transporting species with higher partial
nearest neighbors surrounding atoms on a curve. The pressures. This is especially notable for MOX, as the
difference in chemical potential may be described by vapor pressure of the U-bearing species is approxi-
Herring’s formula, which, in a simplified form, results mately a factor of 10 greater than Pu-bearing species.
in the Gibbs–Thompson formula12,42: The redistribution of Pu can be seen as a result of
pore migration (Section 3.21.3.3) and macroscopic
1 1 gas-phase diffusion.
mk mk;r1 ¼r2 ¼1 ¼ Ok s þ ½54
r1 r2 This Section describes some of the physical con-
where r1 and r2 are the radii of curvature at the point ditions responsible for the evaporation–condensation
on the surface, taken as positive for a convex surface, mechanism, and briefly considers the associated
and mk;r1 ¼r2 ¼1 is the chemical potential of species k redistribution effect.
for a flat interface, Ok is the specific volume of species
k, and s is the surface energy. 3.21.2.3.1 Factors affecting equilibrium
The effect of this is motion in the solid state either vapor pressure
by surface diffusion or vacancy diffusion through the A number of factors affect the equilibrium vapor
body of the fuel in such a way that the curvature is pressure and so contribute to the evaporation–
reduced. The produced gradient in chemical poten- condensation mass transport mechanism. Local
tial produces a force referred to as the sintering stress, variation of temperature, composition, and surface
which is important for the ‘sintering stress’ in sphere- curvature, which vary in space, also contribute.
pac fuels (Section 3.21.3.1).
3.21.2.3.1.1 Temperature
A gradient in temperature is one of the clearest
3.21.2.3 Vapor Phase Transport
sources of different vapor pressures. This is due to
Apart from solid state matter transport, mass may the exponential dependence of the vapor pressure
also be transported via a gas-filled void space through with temperature as given by42:
the so-called evaporation–condensation mechanism.
DHvap;k
Whenever solid fuel is adjacent to a void, it spon- pk ¼ p0 exp ½56
kT
taneously vaporizes parts of its constituents as it
attempts to establish its equilibrium vapor pressure for vapor species k. The indication of vapor species is
according to the local conditions. In equilibrium, the critical in this equation. As Breitung and Reil note in
642 Matter Transport in Fast Reactor Fuels
their recommendation for enthalpy of vaporization 1 nm, the thermodynamic formalism begins to break
of UO2 that because UO2 vaporizes into a variety of down due to the small numbers of participating
species, their recommendation is in fact an effective molecules.42
enthalpy of vaporization for the ensemble of species.43
Due to the large temperature gradients present, 3.21.2.3.2 Temperature-driven vapor phase
this effect results in a potentially fast transport mech- transport
anism for material from the inner hot regions of the Matter transport in the gas phase may be driven by a
fuel to the outer cooler regions. The impact of this temperature gradient in one of two mechanisms. The
effect is discussed further with respect to MOX fuels first is simply due to a partial pressure gradient in a
and the carbides and nitrides. species, caused by eqn [56]; however, this mechanism
generally requires high temperatures in order to be
efficient. Alternatively, at low temperatures, compo-
3.21.2.3.1.2 Composition of the condensed phase
nents of the solid may be transported via a carrier gas.
The relative concentrations of vapors in equilibrium
with the solid may change with the chemical compo- 3.21.2.3.2.1 Gradient in vapor pressure
sition of the solid phase. If the overall fraction of At sufficiently high temperatures, evaporation of
one species in the vapor differs from the solid, then matrix material may occur. The partial pressure gra-
vaporization will alter the composition in the remain- dient would then drive transport of vapor species
ing solid. This may lead to the eventual establishment down the temperature gradient, where they would
of a congruent vaporization composition in which the vapor condense in regions of cooler temperature. This
has the same composition as the solid and thus will not mechanism will generally result in restructuring as
shift the composition. Such is the case for UO2
x and well as redistribution as the matrix itself may be
multicomponent compounds such as (U, Pu)O2
x transported, and vaporization may not be congruent.
where the congruent vaporization of U/Pu may also Figure 2 shows the vapor pressures above a sample of
depend on the state of oxidation, as discussed in 2%Am-MOX at two different temperatures, showing
Section 3.21.5.1.44 the change of dominant vapor species with O/M and
temperature.46 For example, this figure indicates that
3.21.2.3.1.3 Surface curvature vaporization from a hyperstoichiometric sample will
A final physical state is the curvature of the solid result in an enrichment of Pu at higher temperatures
surface. As described in Section 3.21.2.2, the curva- and U migration down the temperature gradient by
ture of the surface affects the chemical potential of UO3 diffusion. Likewise, O would be transported in
the condensed species. The equilibrium partial pres- the same way.47 Similarly, hypostoichiometric MOX
sure will also necessarily be affected in the same will shift the Am concentration accordingly. This
manner via eqn [55]. Thus, we arrive at the Young– mechanism is thought to be important when consid-
Laplace equation42,45: ering the migration of pores, which is discussed later.
p Ok s 1 1 3.21.2.3.2.2 Carrier-gas assisted
ln ¼ þ ½57
pk;r1 ¼r2 ¼1 kT r1 r2 Rand and Markin suggested that it might be possible
The second effect concerns the equilibrium vapor that contamination by other elements may hasten the
pressure above curved surfaces and has two notable vapor transport of an element by acting as a carrier
implications. gas.48 This is thought to be the case with H and
Arguments similar to those considered above lead C contamination of MOX, and O contamination of
to the determination of the gas inside a bubble, which carbide fuels, discussed in Sections 3.21.5.1 and
is in equilibrium with the isostatic stress in the solid, 3.21.5.4, respectively. The gas mixtures adjacent
sstress: to the solids would then be H2O and H2, and CO2
and CO. For example, even contaminations as low
2s as 3 ppm carbon in UO2 have a substantial effect.49
P¼ þ sstress ½58
r The gases are free to react with the solid and diffuse
which is useful in considering the pressure inside along, for example, the length of the crack. The
bubbles. ratio of carrier-gas species is determined by the chem-
Lupis notes that these effects become important ical potential of oxygen (carbon) in the fuel, which
only when the radii of curvature are approximately varies with conditions such as temperature and oxy-
0.1 mm. He also cautions that for radii on the order of gen (carbon) composition.
Matter Transport in Fast Reactor Fuels 643
-2 -2
-3 -3
-4 -4 Total UO3
UO3
-5 Total -5
UO2
-6 -6 PuO2
Log p (atm)
Log p (atm)
UO2
-7 -7 AmO2
PuO2
-8 -8 Pu
AmO2 AmO
UO
-9 -9 U
AmO PuO
-10 -10
-11 Pu -11
PuO
UO
-12 -12
1.80 1.85 1.90 1.95 2.00 2.05 1.80 1.85 1.90 1.95 2.00 2.05
(a) O/M ratio (-) (b) O/M ratio (-)
Figure 2 Vapor pressures of actinide-bearing species above Am containing mixed oxide at (a) 2073 K and (b) 2273 K.
Reproduced from Maeda, K.; Sasaki, S.; Kato, M.; Kihara, Y. J. Nucl. Mater. 2009, 389. Copyright (2009), with permission
from Elsevier.
Table 1 Enthalpies of formation of carrier-gas species 3.21.3.1 Sintering and Grain Growth
CO2 CO H2O H2 As discussed in Section 3.21.2.2, the curvature of an
0 1
h (kJ mol ) –393 –109 –234 0 interface and the minimization of interfacial area may
also act as a driving force for matter transport. The
Source: Aitken, E. A. J. Nucl. Mater. 1969, 30. interface may be the surface between the solid and
a void, or the boundary between two solid crystals
of different atomic structure/orientation. Both pro-
similar form but with the standard chemical poten- cesses may be simulated using either Monte Carlo
tials in place of specific molar enthalpies. The origin codes or multiphase transport models such as the
of this confusion may be the comparison of eqn [61] phase-field model.50–53
with [49]; however, as noted in the derivation of the
latter, this equation implies that Q* is independent of
T. If we take only the temperature-independent com- 3.21.3.1.1 Sintering
ponent of m0, then we obtain h0, resulting in eqn [63]: The minimization of the surface area at an interface
It should be emphasized that QO;vap is not the heat between solid and void leads to the sintering effect
of transport of oxygen in the solid or the same in the common in manufacturing processes. In this case,
vapor. matter is driven to move from regions of low curva-
Although it relates the transport of oxygen to the ture toward areas of high curvature, in addition to
applied temperature gradient, the transport mecha- reducing the overall surface area.
nism is indirect via chemical reaction and diffusion As a model, consider two spheres in contact.
in the vapor phase and it is the gradient in There is a sharp point near the contact area, toward
this reaction potential that results in the transport. which matter is driven both in the solid phase
Therefore, this is not a thermodiffusion-type flux and and through evaporation–condensation as described
should not be confused with the heat of transport in Sections 3.21.2.2 and 3.21.2.3, respectively. The
defined in Section 3.21.2.2. result is that the two particles weld together, forming
In reality, vapor transport would be accompanied a neck.12 If an external pressure is applied such that
by solid state thermodiffusion, which Aitken suggests the spheres are pushed together, creep will also con-
would result in a nonconstant ratio of vapor species in tribute to this process. Eventually, the process of
eqn [61]. He continues to theorize that this would curvature reduction results in the formation of spher-
culminate in a modification to eqn [62], which for the ical voids. Voids may now continue to shrink by the
dilute limit is38: emission of vacancies either into the bulk solid or
! through diffusion along grain boundaries toward a
0 1 0 0 1 Ls vacancy sink.12
hZðgÞ hO þ h r ¼ 1 þ krlnðxÞ ½64
2 2 ðsÞ ZOðgÞ T Lg This phenomenon may play an important role in
the restructuring of particle fuels as discussed for
where the mobility rates of the gaseous and solid MOX fuels in Section 3.21.5.1.54 A more detailed
migration, in the sense of eqn [5], are given by Lg discussion is also contained in Chapter 3.25, Mod-
and Ls, respectively. While eqn [64] recovers [62] for eling of Sphere-Pac Fuel.
Lg Ls, it does not recover eqn [48] for Lg Ls, a fact
that Aitken attributes to the lack of considering equil- 3.21.3.1.2 Grain growth
ibration of the solid with a gas phase when the later The fuel manufacturing process produces grains of
equations were discussed. varying crystal orientation in the solid. When the
temperature is favorable, that is, during sintering
3.21.3 Microscopic Structures and reactor operation, atoms along the curved inter-
face between grains will experience a driving force
Section 3.21.2 discussed some of the fundamental due to the difference in curvature. Grains will there-
mechanisms that drive matter transport. Attention fore spontaneously grow in such a way as to reduce
is now turned to how these phenomena lead to the the total interfacial area, and so big grains will grow at
development of microscopic structures and deter- the expense of smaller ones.
mine their evolution. In this section, we consider The process is complicated by the presence of
sintering and grain growth before taking up a detailed inclusions in the solid, be they solid or gas-filled
discussion of void migration. voids. Inclusions prefer to reside on grain boundaries
Matter Transport in Fast Reactor Fuels 645
due to the lower energy there than in the bulk solid. migration through the solid, and the associated
The lower-energy configuration of the inclusions on restructuring and redistribution of the solid matrix
a grain boundary results in a pinning effect known as and its constituents.
Zener pinning. In the case that the inclusion is Voids change shape or migrate as atoms move
immobile, it will effectively exert a force retarding around the surface or by the emission/absorption
the movement of the boundary, in which case there of vacancies from the surrounding solid. The phe-
must be a sufficient driving force applied to the nomenon of void migration has been a subject of
boundary in order to break free of the inclusion. study by several authors.9,35,56–62 Three atomic trans-
Alternately, the inclusion may be mobile, as may be port mechanisms are identified: surface diffusion,
the case for voids. The boundary movement rate is volume diffusion, and vapor phase transport. Vacancies
then coupled to the migration rate of the inclusions.55 and atoms on the surface are able to migrate by surface
Redistribution may also result as the gases contained diffusion as this offers a potentially faster migration
in small bubbles or other inclusions moved along by path than they experience in the solid. Alternately,
the boundary. vacancies may be emitted into the solid, diffuse
through the solid volume, and be absorbed at a dif-
ferent point resulting in a displacement of the void.
3.21.3.2 Void Migration
Thus far, we have only been considering vacancies;
In reality, cracks, pores and other vapor-filled void however, interstitials may also lead to the matter
spaces offer alternative modes of transport either transport. Due to the relatively small concentration
along the solid surface or through vapor phase of interstitials compared to vacancies, the influence
transport through the void. Both mechanisms can of this point defect may be neglected.9 Finally, as
result in the restructuring and redistribution of the the void is filled with vapor evolved from the local
constituents and may change the shape of the void surface, difference in the vapor pressure of matrix
itself. We now distinguish between microscopic voids material may incur vapor phase transport. All these
enclosed in solid matrix such as pores and bubbles mechanisms participate in void migration, but their
and macroscopic voids such as cracks, interconnected relative magnitudes vary for reasons of void geometry
porosity, and the space surrounding the fuel. This and material properties. These transport mechanisms
section, therefore, discusses microscopic voids, their are depicted schematically in Figure 3.
Void movement
Evaporation
Vacancy recaptured
Figure 3 Schematic description of void transport mechanisms showing vacancy surface diffusion, volume diffusion, and
vapor phase transport of matrix material.
646 Matter Transport in Fast Reactor Fuels
the same result.32 Kelly’s more sophisticated deriva- a temperature difference, for example, the concentra-
tion of the jump distance differs from that of Nichols tion gradient of matrix species will incur diffusion
as it considers that the void does not move at the same through the gas filling the void, leading to matter
jump distance as the vacancies in the solid, but rather transport and void migration. The magnitude of this
considers spontaneous and continuous emission of effect depends on the magnitude of the equilibrium
vacancies at one point and subsequent random vapor pressure and its gradient within the void, for
diffusion through the solid before they are recaptured example, the temperature and the temperature gra-
at a different position.66 The characteristic distance dient. The concentrations of the individual vapor
involved in this process is then the root-mean-square species vary with temperature and the composition
distance between the point of emission and that of of the condensed phase and will not necessarily exist
recapture, as illustrated in Figure 3. For the limiting with the same composition as the condensed phase.
case of rvoid am, this distance is (corrected for a In effect, preferential evaporation of one species over
factor of 2 as noted by Nichols)60: the others may shift the composition of the remaining
solid, in addition to acting as a redistribution mecha-
l2v;col ¼ 2rvoid am ½69
nism for the chemical species, which is independent,
the square root of which is then used in eqn [66] and potentially contrary to solid state diffusion.
for the vacancy jump distance. The result coincides The most accurate method to predict void migra-
with that of Nichols, but is placed on a sounder tion from vapor phase transport is by solving the
physical basis. diffusion of the vapor species across the void. This
Om implies solving the mass balance equation, which for
lvoid ¼ lv;vol ½70 a steady state concentration profile is:
3
4prvoid =3
r Jvap ¼ 0 ½73
The void jump frequency was evaluated by Olan-
der by considering the void to be a perfectly absorb- The mass flux may be calculated through the fill gas
ing sphere surrounded by a concentration of using the TIP. The case is simple as we have only one
vacancies in the solid.9 The resulting formula for a vapor component diffusing through a fill gas:
void in equilibrium is: 1
Jvap ¼ Lvap;vap rT mvap Lvap;q rT ½74a
3
4prvoid =3 T
Gvoid;vol ¼ Gv:vol ½71
Om 1
Jq ¼ Lq;vap rT mvap Lqq rT ½74b
which is the same as proposed by Nichols using the T
logic leading to eqn [67] but again with a firmer physi- Using similar arguments as used to develop solid state
cal basis. The resulting diffusion coefficient is then: mass flux in a temperature gradient in Section 3.21.2.2
3 but with eqn [55] to relate the chemical potential of
Dvoid;vol ¼ Dv;vol Om 3 ½72 the vapor to its partial pressure, pvap, we obtain:
4prvoid
Dvap 1
Jvap ¼ rpvap þ Q vap rT ½75
Void migration by volume diffusion must be trea- kT T
ted cautiously because it involves vacancy diffusion Through an analysis of kinetic theory, Sens asserts
through the solid. Similar to the treatment of the that Q vap ¼ 0 with a caveat that this is only valid for
Kirkendall effect in Section 3.21.2.2, the surround- diffusion without restriction and, therefore, is not
ing solid matrix itself is moving with respect to the necessarily valid in the presence of a barrier, such as
solid far from the void. An example of this treatment a porous membrane.61 Therefore, the mass flux of
is given below when calculating the void velocity in a vapor is simply:
thermal gradient in Section 3.21.3.2.
Dvap
Jvap ¼ rpvap ½76
3.21.3.2.1.3 Void migration by vapor phase kT
transport and the balance equation for the steady state is:
A third mechanism of void migration is through
Dvap
vapor phase transport. The solid surrounding the r rpvap ¼ 0 ½77
void spontaneously evolves into vapor in an attempt kT
to establish its equilibrium vapor pressure. If the The boundary condition of eqn [77] is the
equilibrium pressure varies along the surface, due to partial pressure of the vapor species at the surface.
648 Matter Transport in Fast Reactor Fuels
The solution of this equation may become compli- 3.21.3.2.4.1 Migration driven by thermodiffusion
cated if we consider that the vapor pressure need not When considering thermodiffusion, the applicable
be at the thermodynamically equilibrium value. In temperature gradient must be considered as the
addition, we can consider that the surface of the void presence of the void perturbs the macroscopic
moves at a rate of JvapOm. Finally, if thermodiffusion gradient, rT . The temperature gradient in the void,
is considered to be the driving force, then a complete ðrT Þvoid , is applicable for surface diffusion and vapor
treatment must include solving for the temperature diffusion, whereas the gradient in the solid immedi-
profile surrounding the void. This is discussed for a ately surrounding the void, ðrT Þsur , is applicable for
simple case in the next section and in more detail volume diffusion. The temperature gradient in and
when considering pores in Section 3.21.3.3. around the void may be derived in terms of the
It is, however, possible to treat the problem in a thermal conductivity of the void and the surrounding
simplified manner as a spherical superparticle. While solid, Kvoid and Ksur, respectively, and may be derived
this simplification neglects several important phenom- from a classical solution to yield60:
ena, it is useful for ready comparison between void 3Ksur
migration mechanisms. In this case, the equations are ðrT Þvoid ¼ rT ½80a
2Kvoid þ Ksur
similar to volume diffusion but include the density
difference between solid and gas. The jump distance
1 2ðKsur Kvoid Þ
is derived as was eqn [66]. The jump frequency is ðrT Þsur ¼ rT ½80b
similar to his derivation of eqn [71] but includes the 2Kvoid þ Ksur
concentration of vapor calculated from the equilib- The void velocity in a macroscopic thermal gradient,
rium partial pressure of the rate-controlling species rT is therefore:
and the departure from that equilibrium, a 1.
2a 1 3Ksur Q surf
Nichol’s diffusion coefficient is then32: vvoid;surf ¼ Dv;surf O rT ½81a
rvoid kT 2Kvoid þ Ksur T
3Dvap O2 apv
Dvoid;vap ¼ ½78
3
4prvoid kT vvoid;vap ¼ Dv;vap Oap
1 3Ksur DH
rT ½81b
kT 2Kvoid þ Ksur kT
The diffusion coefficient of vapor through the gas in
the void may be calculated from kinetic gas theory.9,61 Shewmon warns that the heats of transport for surface
Typically, the magnitude of the diffusion coefficient and volume migration need not be equal, and there-
depends on the total gas pressure in the void. For a fore any value calculated from experiments must
void in equilibrium with the surrounding solid, the then be attributed to either surface or volume
pressure varies with the radius as described in eqn [58]. according to the dominant mechanism.
Thus, the dependence of eqn [78], and the migration As pointed out by Biebsack and Diez, the velocity
speed of the void is dependent on the behavior of the of void migration due to vacancy diffusion in the solid
coefficient of vapor diffusion through the void fill gas, must be treated as a special case since it involves
which may be linearly dependent on the radius of the vacancy diffusion in the solid.57 As was the case in
bubble or independent of the radius, if D is only considering the Kirkendall velocity in eqn [18], we
proportional to T. must, therefore, account for the movement of the
surrounding lattice also. The surrounding lattice
3.21.3.2.4 Void migration forces moves as the vacancies diffuse and are annihilated/
Having discussed the mechanisms of void migration, created in the macroscopic temperature gradient:
we can now consider how driving forces of stress and Q v
temperature may influence void migration. Equation vlattice ¼ Dv rT ½82
kT 2
[6] is the basic form of this equation, but the driving
force to which the void superparticle is subject, Fvoid, which implies that the void velocity with respect to
is related to the molecular driving force, Fm by noting the bulk material is:
that the energy of moving the void forward by a 1 Q v 2ðKsur Kvoid Þ
distance must be equivalent to moving an equal vvoid;vol ¼ Dv;vol rT ½83
kT T 2Kvoid þ Ksur
amount of atoms the same distance:
An interesting feature in eqn [83] is that, unlike
4pr 3 migration due to surface diffusion or vapor transport,
Fvoid ¼ nm void Fm ½79
3 the direction of void migration depends on the relative
Matter Transport in Fast Reactor Fuels 649
conductivities of the void and surrounds. Although, 102 mm in radius, were implanted into copper foil
typically, the void conductivity is much less than that and a temperature gradient applied, which drove the
of the surroundings, this may not be the case in the bubbles to migrate due to thermodiffusion.64 When
treatment of migrating solid inclusions, for which the migration pattern was examined, it was deter-
the treatment of surface and volume diffusion applies. mined that the rate of migration was proportional
Therefore, even the sign of the heat of transport in to the inverse of the bubble radius, in accordance
eqn [83] must be considered carefully. with the predictions of Shewmon and Nichols. The
The impact of the stress field surrounding a void conclusion of size dependence is also supported by
has also been addressed.9,32,33,41 The analysis consid- molecular dynamics simulations and a phase-field
ers the work done by compressing the gas inside the model considering vacancy diffusion through the
void, changing the surface area and the change in solid bulk.33,65,68
energy of the surrounding solid. This driving force The three mechanisms need not operate indepen-
may be compared to that of thermal diffusion and dently, but can occur cooperatively, as is the postu-
ranges, depending on the chosen parameters, on the lated explanation of void migration in (U, Pu)C.
order of 14%.9,33 A more sophisticated approach In this fuel, the equilibrium vapor contains very little
investigated the effect of stress-driven migration and carbon, and so it is thought that while U and Pu
geometry evolution for voids located on grain bound- transport across the void via vapor transport, rapid
aries.41 It was suggested that stress-driven migration solid state carbon diffusion moves carbon to balance
may become comparable to thermodiffusion along the molecule.69 The migration rate of the void is then
the surface in the case of oblate pores with the polar determined by the smaller of the two rates which,
axis perpendicular to the grain boundary.33,41 in this case, is the vapor transport of the metal. This is
discussed more thoroughly in Section 3.21.5.2.
3.21.3.2.4.2 Pinning Nichols reminds us that while the rate of a mass
As mentioned in the discussion of grain growth, when transport process such as void migration is limited
a small void migrates to a grain boundary, the reduc- by the speed of diffusion, there are other factors
tion in surface energy can pin it. The typical migra- that may be relevant in reducing the speed of the
tion length of small voids may, therefore, be limited process.32 Such a rate-controlling step may be the
by the size of the grain.55,67 They do not, therefore, transfer of material across the void–solid interface.
provide a large contribution to fuel restructuring, As Nichols discusses, local equilibrium at the sur-
although they do significantly affect the macroscopic face may be considered quite generally except for
behavior of the fuel. If sufficient bubbles exist at interfaces, which are atomically smooth and crystal-
the grain boundaries, they can interlink, creating lographically low indexed.70 At such a surface, depo-
macroscopic voids, which span large distances in sition of diffusing atoms may involve the nucleation
the fuel solid. These voids enable the evaporation– of new atomic layers, a difficult and, therefore,
condensation on the macroscopic scale. The pinning slow process. This limitation may, however, be over-
of bubbles to grain boundaries and dislocations is come by a strong force driving the interface move-
the subject of several papers, notably a statistics- ment such as that of a thermal gradient in excess of
based grain/bubble growth algorithm50; however, 1000 K cm1, in which case the situation returns to
the details of this process are not elaborated on here. diffusion control.71 Therefore, it may be that inter-
face control may only be important in the case of
3.21.3.2.5 Comparison of void migration little or no driving force. Shewmon also considers the
behavior and experimental applications interface reaction control and concludes that for a
By comparing the dependence of the void mobility spherical void, the migration speed should increase
n
to the void radius, it is possible to gain insight into with rpore , an important conclusion as it ensures that
the particular surface, volume, or vapor diffusion the deduction of the surface diffusion mechanism
mechanisms that are dominant. Through examina- from the inverse radius behavior remains a valid
tion of eqns [81a], [81b], and [83] it is apparent that conclusion. The surface reaction rate will initially
only the surface diffusion mechanism is dependent determine the migration behavior until a radius is
inversely on the void radius. This conclusion was reached where the faster of surface or volume diffu-
first proposed by Shewmon,35 who used this theory sion will determine the speed of migration. In sup-
to analyze the results of Barnes and Mazey’s experi- port of this supposition, Shewmon cites the work of
ments in which He-filled bubbles, approximately Wernick72 who conducted experiments with molten
650 Matter Transport in Fast Reactor Fuels
aluminum wires in germanium, under a temperature 0.65 and 12.9 Å s1 with no apparent relation between
gradient. The migration rate displayed exponential the migration speed and the void size, indicating that
behavior for small wire radii before becoming radius the dominant diffusion mechanism was most likely
independent, which may be explained by interface not surface diffusion but vapor phase transport.74
control leading to volume diffusion. They also credit a large spread in migration velocities
A distinction may now be made between two types to the effect of dislocations and grain boundaries.
of voids, bubbles and pores, based on their migration Speight summarizes the comparison of the vapor
behavior as determined by size, shape, and gas and surface diffusion mechanisms for UO2 and, for a
content. Bubbles are considerably smaller, typically rough calculation with several assumed values, calcu-
contain fission gases at high pressure, and are in lates that the critical radius for the change between
mechanical equilibrium with the surrounding solid, surface and vapor diffusion mechanisms at 2000 K is
implying a spherical shape. Pores are larger, contain approximately 1 mm, a conclusion later supported
gas at low pressure, and are typically not in equilib- by extrapolation of molecular dynamics simulation
rium with the surrounding solid matrix and so are not results.33,56
spherical. Therefore, much of the preceding theory The treatments of surface and vacancy diffusion
may be directly applied to bubble migration, whereas above are in fact not restricted to gas-filled voids, but
pore migration must be dealt with in greater detail in also apply to solid inclusions in the fuel matrix.
the following section. A summary of these differences Michels et al. also examined the migration of solid
is given in Table 2. inclusions in MOX up the temperature gradient.74
Bubbles nucleate by fission gas, mostly xenon, Inclusions were of radius of 0.5–3.25 mm in a temper-
precipitating from the solid matrix. The gas origi- ature range of 1973–2143 K and a temperature gradi-
nates as fission products, which diffuse to the bubble ent of 4900–3900 K cm1. Migration speeds averaged
from the surround solid. Their migration, coales- 0.2–9.2 Å s1 and were found to be inversely propor-
cence, and interlinkage are of particular importance tional to the radius of the inclusion, in conformation
for fuel swelling. The dominant migration mechanism with the surface diffusion mechanism.
is significantly affected by the bubble radius. High
internal gas pressure is required to balance the isostatic
3.21.3.3 Pores Migration
stress of the surrounding matrix, as calculated by
eqn [58]. The large pressure suppresses vapor phase The characteristics and behavior of pores differ quite
diffusion, while the surface diffusion mechanism significantly from those of bubbles. Unlike bubbles,
becomes more important due to small bubble radii. pores are not observed to be spherical, but rather are
This analysis of bubble migration has been applied typically observed to exist in the shape of a lens with
to nuclear fuels by several authors. Cornell and the minor axis aligned with the temperature gradient
Williamson observed the migration of Krypton bub- as shown in Figure 4. The irregular shape implies
bles in UO2 and verified that their migration also that the void is not in mechanical equilibrium with
adheres to the 1/rvoid relationship.73 In another case, the surrounding solid matrix since mechanical equi-
Michels et al. observed fission-gas-filled bubbles in an librium would require a spherical shape.
irradiated MOX element at a temperature range of Pores are typically filled with the fill gas used in
1933–2153 K and a gradient of 4900–3900 K cm1. fuel manufacturing. Helium is a common cover gas
The voids had radii of 15 mm and migrated between used in the manufacture of modern fuel and so pores
are typically filled with He at atmospheric pressure
in addition to CO2 and CO as may result from carbon
Table 2 Comparison of some characteristics between
contamination.9 As burnup increases, fission pro-
bubbles and voids observed in MOX fuels
ducts, notably xenon and krypton, may also be
Bubbles Voids found as they diffuse into the pore or are swept up
during pore migration. When they merge with a large
Size <1 mm diameter >1 mm in smallest
dimension fissure or the central void, the accumulated gaseous
Pressure 300 atm 3 atm inventory is released. Fission products, which are not
Shape Spherical Lenticular volatile and insoluble in the fuel matrix form preci-
Contents Gaseous fission Predominately cover gas pitates and these too may be captured in migrating
products
pores and deposited in the central region. Such is the
Origin Fission yield Initial porosity and cracks
case with Ru concentration.3
Matter Transport in Fast Reactor Fuels 651
Due to the low gas pressure found in voids and pore migration on the microstructure of the fuel is
their irregular shape, it is thought that the eva- thus significant and so will be dealt in greater detail
poration–condensation mechanism is the dominant in the following section.
material transport mechanism. This implies a quite The following section discusses some theories
complex system as the equilibrium partial pressure behind the shape and migration speeds of pores in
depends on the surface temperature and the solid greater detail.
composition, both of which are perturbed and altered
by the presence of the pore. The temperature gradi-
3.21.3.3.1 The origin of pores
ent in the region of the pore is perturbed due to the
Pores are initially present in the sample as part of
lower thermal conductivity within the low-pressure
the fabrication process. The density of modern fuels
gas-filled pores. The composition is altered due to
following the manufacturing process is less than
preferential evaporation, which redistributes the con-
the maximum, theoretical density in order to leave
stituents as the pore moves.
space to accommodate fission products. The empty
Pores also differ from bubbles in that they are very
volume is distributed approximately homogeneously
mobile in the fuel, migrating up the temperature
throughout the matrix as pores. As will be discussed
gradient until it reaches a macroscopic void. This
later, the migration speed is fast, and can completely
transient existence may explain why the spherical
remove the initial porosity within a matter of hours.
shape corresponding to mechanical equilibrium is
However, as indicated in Figure 5, pores continue to
never achieved. In contrast to bubbles, they are too
be created from cracks, and so the effect of their
large to be pinned to grain boundaries and in fact
migration continues as burn up increases. Ronchi
consume grains to form new columnar grains in their
and Sari suggested that the formation of pores from
wake. In accordance with the supposition of a vapor
the fabrication porosity was dependent on the pres-
transport migration mechanism, it was also found
sure of He in the fuel matrix and that large He
that pores migrate at approximately the same speed,
pressure could prevent the pores from forming len-
regardless of size. Considering that they do not get
ticular shapes.75 Lewis et al. showed that high-density
pinned to grain boundaries, this implies that they are
fuels did not, in general, exhibit pores and columnar
unlikely to coalesce with each other before reaching
grain growth.63,76
the highest-temperature solid region where they do
meet and form one of the most pronounced features
of fast reactor fuels, the central void. The effect of 3.21.3.3.2 The shape of migrating pores
During pore migration, the pore may change its
shape depending on the condensation rate at various
Tmax
points on the surface and eventually morph into the suitable low-energy site where they may attach per-
lenticular pores seen in Section 3.21.5. Studies have manently. Such a site includes ledges or pits in the
shown the appearance of lenticular voids at the head crystal lattice, effectively providing nearest neighbors.
of radially oriented columnar grains, with small Attachment may thus proceed as a series of moving
spheres between the grains, spaced roughly regularly ledges, as illustrated in Figure 6. The resulting evo-
at intervals approximately equal to the width of the lution of the pore shape is toward a disk shape.
pore. The radially oriented voids do not appear to The second factor considered by Oldfield and
be mobile61; however, they do appear to be related Markworth was the effect of the impurities in the
to the pore migration/shape. Explaining the shape pore. As the pore migrates, it will encounter fission
of the migrating pores has been a subject of much products dissolved in the solid matrix, caused by
examination and debate. Hypotheses to account for burnup. Volatile fission products will be transported
the shape of the pores and the origin of the trailing across the pore, whereas nonvolatile products will
bubbles include arguments based on vapor transport accumulate on the leading edge. The impurities at
dependence on the geometry of the pores, the attach- the trailing edge impede the rate of condensation and
ment kinetics of vapor molecules to the trailing edge, eventually destabilize the advancing solid front. The
and the presence of volatile impurities. front, now undergoing cellular growth, forms steep-
As discussed in Section 3.21.3.2, proper treat- sided grooves inside which impurities are trapped
ment of the evaporation–condensation mechanism when the grooves pinch off. The grooves then form
involves modeling both temperature and mass trans- a string of regularly spaced bubbles, which have been
port in and around the pore. The effect of pore postulated to stretch over the grain boundary as
geometry on heat and mass transport was investigated they relax mechanically.41 The combined effect of
by Sens for pores that were initially spherical, tubular, these two factors may yield the lenticular pores as
or disk-shaped solving the equations for vapor trans- observed, with a string of fission products-filled bub-
port and heat transport numerically, although surface bles in its wake, as shown in Figure 7.
energy was neglected.61 The simulation showed that
the initially spherical pores tended to elongate as
the leading edge moved faster than the trailing
edge. The tubular pores showed the same behavior Direction of void
and also a slower overall pore migration speed. The migration
elongation of spherical pores to cigar shape was
also predicted by Nichols.60,63 The disk-shaped
pores showed leading and trailing edges moving at
the same rate and a pore velocity greater than either
the sphere or the tube.
Net matter flux
Because of their observed role in fuel restructur-
ing, disk-shaped pores were investigated in more
detail. Sens’ simulations showed that during the
course of migration, the disk-shaped pores evolved
Direction of
towards the lenticular pores observed experimentally. step growth
Initially, the disk bends as a whole, until the curvature
at the trailing periphery is speculated to pinch off to
form spherical voids, which serves to flatten the trail- Gas phase
ing edge.61
Oldfield and Markworth performed a similar Ledge
analysis of pore migration based on a thermodynamic
Solid
argument for determining the driving force for solid
growth.58 Two effects that may affect the shape of Figure 6 Schematic of how pore shape may evolve due
the pore are discussed. First, the kinetics of vapor to attachment kinetics at the trailing edge. Growth direction
molecules attaching to the surface is considered. is indicated by arrows. Based on Olander, D. R.
Fundamental Aspects of Nuclear Reactor Fuel Elements;
A model was employed where vapor molecules Technical Information Center, Office of Public Affairs:
impinging on the surface temporarily attach and Springfield, VA, 1976; Oldfield, W.; Markworth, A. J. Mater.
may diffuse along the surface until they meet a Sci. Eng. 1969, 4, with permission from D. R. Olander.
Matter Transport in Fast Reactor Fuels 653
Hot surface
Pore
Vpore
Cool surface Jvap
Partial pressure
PH
of matrix vapor
PH⬘
PC⬘
PC
(assumed to be equal for U and Pu species) by a factor pressure of the matrix material in accordance with
of 6.7, which, via eqn [84], results in an increase in Raoult’s law. If the impurities are volatile, they will be
migration speed by the same factor.80 transported across the pore, but if they are nonvola-
Equation [84] implies some specific behavior tile, they will accumulate at the leading edge and
about pore migration. First, the velocity of the pore progressively decrease the matrix partial pressure.
is not dependent on the size (insofar as the tempera- In addition, the attachment kinetics discussed above
ture gradient is not further perturbed), which implies may also affect the rate of condensation on the trail-
that pores move at the same rate, an experimentally ing edge. This concept arises from the theory that
verified feature. Second, if the fill gas pressure is molecules in vapor attach most easily to ledges in the
assumed to behave ideally with temperature, and solid matrix. The relatively flat trailing edge implies a
is not dependent on the pore size, then the velo- retardation of this effect and so an excess in partial
city varies as T 5/2, which has been confirmed by pressure in the vapor may develop in order to over-
Ronchi and Sari.75 Finally, Ronchi and Sari have come this kinetic effect. The results of these two
also observed that the migration speed roughly effects on the equilibrium vapor pressure are shown
describes an Arrhenius plot with an activation energy schematically in Figure 8, where the vapor pressures
of 5
0.5 eV/molecule, comparable to their reported at the hot and cold sides, pH and pC, respectively, move
evaporation enthalpy of UO2, 5.8
0.1 eV/molecule, towards pH0 and pC0 . Olander performed a calculation
which indicates that the energetic effects proposed by of these combined effects assuming that the pore is
Oldfield and Markworth are not rate determining.75 infinite in transverse dimension and determined that
Clement pointed out a discrepancy between the the rate of migration may significantly decrease as a
velocity predictions of Nichols and the experimental consequence.9 Specifically, for a soluble impurity
results of Ronchi and Sari by as much as a factor of mole fraction at the leading edge of 0.01, a condensa-
400,32,75,78 which can be rectified by considering the tion coefficient (the fraction of impinging molecules,
sample as even slightly hyperstoichiometric due to which adhere to the solid) of 0.9, and a pore width of
the large increase in UO3 vapor pressure resulting 10 mm results in a decrease in the migration velocity
from such a shift. of 41%. If the pore width is increased to 20 mm, then
Similar to the discussion in Section 3.21.3.2 for the reduction is 59%.9 The calculations thus suggest
the general case of a void in the solid matrix, the a large impact of soluble fission products and attach-
presence of the pore perturbs the local temperature ment kinetics on the pore migration velocity, the mag-
field in such a way that the temperature gradient nitude of which depends on the width of the pore.
across the pore may be significantly larger than in Smaller pores may then be immobilized, whereas
the surrounding matrix. In his treatment, Nichols larger pores may continue unimpeded.
calculated as ðrT Þpore ¼ 3=2ðrT Þmatrix assuming
an insulating sphere. This calculation was superseded 3.21.3.3.4 Redistribution of constituents
by a numerical calculation by Sens on the same As discussed in Section 3.21.2.3, the different vapor
simulation as used to determine the pore shape pressures of species in the condensed phase may lead
described previously. His prediction, which is taken to unmixing as pores migrate. The magnitude of this
as superior, is ðrT Þpore ¼ 4ðrT Þmatrix . Sens briefly redistribution mechanism has been investigated by
considers the contribution of radiation to the heat numerous authors but remains somewhat unclear
transport in the pore. For his disk-shaped pore of especially in comparison to redistribution by vapor
20-mm width, he reports that the radiative contribu- phase through cracks.2,26,79–83
tion is below 10% of the total heat flux at 2000 K. He Guarro and Olander performed a sophisticated
further states that the radiative contribution becomes analysis considering the transport of the actinides
comparable in magnitude to the conductive flux at across the pores and also the distribution of those
temperatures approaching 10 000 K. actinides in the solid phase ahead and behind migrat-
Another modification to eqn [84] comes from ing pores.80 A schematic of the pore migration and
the reduction of the driving force, the difference in concurrent redistribution is shown in Figure 9 for a
partial pressure, resulting from the combined effects pore modeled as an infinite slab, with the U/Pu ratio
of impurity accumulation at the leading edge, and represented by q. The figure describes an initial
attachment kinetics at the trailing edge.9,58 The pres- startup spike of Pu depletion followed by a quasi-
ence of impurities dissolved in the solid matrix, as steady state migration of the pore lead by a Pu-rich
caused by burnup, lowers the equilibrium partial bow wave.
Matter Transport in Fast Reactor Fuels 655
q Pore at time t
Initial
after starting
pore location
r d
Vpore
qH
Bow wave
Trail
q0
bw
Z
Increasing temperature
Startup spike
Figure 9 Schematic of Pu redistribution as a result of pore migration. Modified from Guarro, S.; Olander, D. R. J. Nucl.
Mater. 1975, 57. Copyright (1975), with permission from Elsevier.
Initially, the hot and cold sides of the pore are at oxygen can migrate around the pore in this manner.80
the initial composition, q0. Preferential evaporation of Therefore, the model is justified for consideration of
U-bearing species from the hot side leaves it Pu-rich actinide separation, and the bulk stoichiometry is
and the cold side Pu depleted while advancing the considered to be quickly restored by this mechanism
pore slightly. The total vapor composition evolved despite the UO3 vaporization from the hot surface.
from the hot surface becomes slightly more Pu-rich The conclusion of the study by Olander and
due to the increased mole fraction of Pu in the solid, Guarro and Olander is that pore migration will not
resulting in more Pu being transported across the significantly contribute to actinide redistribution
pore and deposited on the cold side. This process considering the current models of pore migration
continues until the original composition is reached velocity.80,82 Olander also considered Pu migration
on the cold side, leaving the startup spike depicted in by the accumulation of Pu-rich vapor in the pore
Figure 9. volume, a process analogous to zone refinement, but
Once the initial startup spike is generated, the found that this effect was too small to be significant.82
pore migrates in a quasi-steady state manner where This conclusion is in disagreement with the model of
the hot surface is preceded by the bow wave of Pu Meyer.79,84 The model used by Clement and Finnis
caused by diffusion of Pu into the solid ahead of the supports the claim by Olander that pore migration
pore. The bow wave follows an exponentially decay- cannot contribute to Pu redistribution.85 They con-
ing Pu concentration with a characteristic depth clude that the pore migration velocity must be reduced
depending on the speed of Pu diffusion in the solid, in order for greater redistribution to occur, which may
DUPu, compared to the pore velocity, vp82: happen during operation as previously discussed.
DUPu
l bw ¼ ½86
vp
3.21.4 Bulk Phenomena
which, for example, is calculated to be 1 mm at
2000 K80 for MOX. This parameter is temperature We have now discussed mass transport occurring in
dependent in both DUPu and vp, but with increasing nuclear fuels on an atomic and microscopic level.
temperature the pore velocity increases faster imply- During the course of operation, macroscopic struc-
ing a shrinkage of the Pu enrichment wave. The Pu tures and characteristics will develop. In particular,
lost from the wave in front of the pore is deposited on this section addresses cracks, porosity evolution, and
the cold surface as the pore migrates, leading to a the consequence of melting.
gradually increasing concentration of Pu in the tail of
the pore.82
3.21.4.1 Cracks
The issue of solid state diffusion around the pore,
and thus the validity of the infinite slab model, may Cracks in the fuel pellet develop very quickly due to
be examined by considering the depth of the Pu wave thermal stresses. During operation, they may also be
compared to the width of the pore. Guarro and generated during power cycles, either in heating up
Olander noted that for actinides, the width is large or cooling down. As mentioned in Section 3.21.3.3,
enough to justify this assumption; however, note that cracks may form a source of migrating pores. These
656 Matter Transport in Fast Reactor Fuels
pores then remove some of the empty volume of the assumption that vapor pressure is always at equilib-
crack, leading to its healing. rium. Furthermore, what is actually calculated is
Cracks may be present in radial or circumferen- the matter flux in the free volume, from which the
tial directions. Radial cracks, in particular, play a pore migration speed is assumed to be the negative
role in redistribution as they offer long channels (an assumption implying fast attachment kinetics).
of void space, appropriate for mechanisms such as Therefore, there is no difference between consider-
carrier-gas-based redistribution, discussed in Section ing this model for migrating pores and for large
3.21.2.3. interconnected porosity.
1.0
0.6
0.4
0.2
0.0
340 345 350 355 360 365 370 375 380
Linear heat rating (W cm−1)
4000
100 kW m-1
Solid
3500 Liquid
80 kW m-1
3000
60 kW m-1
Temperature (K)
2500
40 kW m-1
2000
1500
1000
500
0.14
Stoichiometry deviation
0.12
0.10
40 kW m-1
60 kW m-1
0.08
0.06
80 kW m-1
0.04 100 kW m-1
0.02
0.00
0 1 2 3 4 5 6
Radius (mm)
Figure 12 Simulation of temperature and oxygen profiles in a centerline-melted sample of UO2.05 for a sequence of four
linear powers. Solid-phase thermomigration and the noncongruent separation of oxygen across the phase boundary are
shown. Reproduced from Welland, M.; Lewis, B. J.; Thompson, W. T. J. Nucl. Mater. 2011, 412, 342–349.
fuels typically contain 15–30 at.% Pu in substitutional constituent elements with reference to the mechan-
solution with the U. Most conventional reactors are isms previously discussed.
fuelled with UO2, which is transmuted to plutonium as
irradiation proceeds. Thus, it is sensible to discuss 3.21.5.1.1 MOX restructuring
UO2
x as a special case of (U, Pu)O2
x . MOX fuel is often divided into regions defined by
The fuel element may be produced in a variety of their restructured state. Proceeding outward from the
forms: pelletized, sphere-pac, or vipac. The differ- centerline they are:
ence lies in how the fuel is processed before insertion
1. Central void
in the reactor. With the exception of large amounts of
2. Columnar grain region
porosity in particle fuels, these fuel forms generally
3. Equiaxed grain growth
behave in a similar manner.2,90
4. As-fabricated microstructure
The low thermal conductivity of MOX fuels,
relative to metal fuels, for example, implies the Knowledge of the dimensions of each region is
possibility of high temperatures and large tempera- important as it affects the thermal conductivity and
ture gradients. Large, thermally driven mass trans- the heat generation. In addition, the microstructure
port effects are therefore expected and observed. The influences the material properties of the fuel such as
ensuing section summarizes some of the experiences diffusion of fission gases.
with MOX fuels beginning with the restructuring A schematic of the evolution of the MOX micro-
behavior, followed by the redistribution of the structure is given in Figure 13 in which burnup
Matter Transport in Fast Reactor Fuels 659
increases clockwise. An experimental micrograph is Olander’s calculations are shown in Figure 15; they
shown in Figure 14, which also shows the evolved show the variation of the radial extent of the columnar
structure. The presence of cracks is visible in the grains for a series of centerline temperatures, along
micrograph, and the healing process is depicted in with the temperature at the outer region.9
the schematic evolution. The results show that the temperature at the
We can use the analytical expressions of pore migra- columnar grain boundary is roughly independent of
tion given in eqn [84] to model the extent of columnar the centerline temperature, TC. The radius, however,
grain growth. On examination of irradiated fuel ele- is dependent on the centerline temperature. Olander
ments, the extent of the columnar grain growth is notes that these predictions give a larger radius and a
usually clearly determinable due to their high sensitiv- lower temperature than those of Nichols,63 and that
ity to temperature and temperature gradient. This Christiansen observed the temperature at the bound-
allows a fairly accurate prediction of the radial extent ary to be lower by approximately 200 K.91 However,
of the columnar grains and the temperature at that this calculation does not include the variation of
location. Calculations of this kind have been performed temperature profile with the evolving microstructure
by Nichols and Olander, and experimental results have and this effect would lower the rate of the boundary’s
been presented by Christensen.9,63,91 The results of movement.
0 MW
d/MT
M
~
0
M
W
d/
M
TM
~ 10
0
MWd
/MTM
~ 10
/MTM
5
MWd
MWd
/MTM
~ 10 1
~
TM
10
M
5
d/
M
W
W
M
d/
2
M
10
TM
~ 10 4
MWd /MTM
/MTM 3 MWd
~ 10
Figure 13 Expected evolution of microstructure in mixed oxide fuel, operating at 42.3 kW m1 shown at a sequence of
increasing burnups indicated in megawatt days per metric ton metal (MWD/MTM). Reproduced from Christensen,
J. A. Technical Report WHAN-SA-79; WADCO Corporation, 1970, with permission from Westinghouse.
Figure 14 Slice of a pelletized mixed oxide fast reactor fuel sample showing as fabricated region, equiaxed grain growth,
and columnar grain growth leading toward the central void. © European Atomic Energy Community, reproduced with
permission.
660 Matter Transport in Fast Reactor Fuels
The coupling of porosity evolution and heat trans- a function of radius, at a series of operation times.61 He
port implies a changing pore migration velocity with presents the results of his simulation of a 0.5 cm radius
the changing temperature profile, as described in pin of UO2 in a water-cooled thermal reactor operated
Section 3.21.3.3. Sens performed a calculation consid- at the linear power of 600 W cm1 in Figure 16. This
ering this coupled effect on pelletized UO2 calculating theory was later shown to reasonably describe the
the pore velocity, porosity, and temperature profile as observed migration rates of lenticular voids.75,92
2500
0 ⬚K
TC = 300
0.6 0 ⬚K 2300
250
T C=
2100
grain growth (x1)
0.4 1900
⬚K
00
1700
20
C =
T
0.2
0
1 10 102 103 104
Irradiation time (h)
Figure 15 The outer radius and temperature of the columnar grain region as a function of irradiation time and centerline
temperature Tc, calculated for UO2. Reproduced from Olander, D. R. Fundamental Aspects of Nuclear Reactor Fuel Elements;
Technical Information Center, Office of Public Affairs: Springfield, VA, 1976, with permission from D. R. Olander.
3000
t = 60 s
t = 15 min
t = 60 s
t = 30 min
10−4 t = 15 min
t=1h
Temperature (K)
t = 30 min
t=2h
1.00
Transport velocity (cm s−1)
Fractional porosity
2500
t=2h
t=1h
t = 30 min
10−5 t=1h t = 15 min
2000
t=2h
0.50
Fuel surface
Fuel surface
1500
10−6
t = 60 s 1000
Initial porosity
10−7 0.00
0.0 0.1 0.2 0.3 0.4 0.5 0.0 0.1 0.2 0.3 0.4 0.5
(a) Fuel radius (cm) (b) Fuel radius
Figure 16 Simulation results using the PREMOVSKI code for a fuel sample are a 1-cm diameter pin, initially at 89% TD, a
surface temperature of 873 K, and a linear power of 600 W cm1 in a thermal reactor showing (a) pore migration velocities
and (b) corresponding porosity and temperature evolution. Reproduced from Sens, P. F. J. Nucl. Mater. 1972, 43.
Copyright (1972), with permission from Elsevier.
Matter Transport in Fast Reactor Fuels 661
It is possible to apply the results presented in which may account for the discrepancy in the observed
Figure 16 to hyper/hypostoichiometric MOX by vs. predicted central void size.
considering the changes in vapor pressures above the Lackey et al. also investigated medium burnup
solid phases. In hyperstoichiometric (U0.8Pu0.2)O2 þ x, elements for thermal and fast reactors. A reassuring
Olander predicted that the increased vapor pressure of observance was that the redistribution was similar in
UO3 associated with hyperstoichiometric urania both cases, which is to be expected as the restructur-
would overcome the reduction in urania concentration ing process is understood to be thermally driven and
(only 80% compared to UO2) and result in an increase not dependent on the energy spectrum of the neu-
in the migration speed by a factor of 3.5.82 In con- tron flux. Interestingly, for the medium burnup ele-
trast, the reduction of UO2 concentration in hypostoi- ment at a fractional radius of approximately 0.9, there
chiometric MOX was predicted to lower the migration does not appear to be any porosity change; however,
speed by a factor of 5/7 by Guarro and Olander.80 vapor appears to have been deposited on the inner
Lackey et al. also performed a numerical simula- surface of the cladding. The deposition should lead
tion to study the porosity evolution for sphere-pac to an increase in the gap conductance as the gap is
fuels in thermal and fast neutron spectra fluxes.2 filled with solid material.2 The results of the experi-
They compare a sample of a low burn-up thermal ment are reproduced in Figure 18, which shows their
reactor fuel element with their simulation results; measured values for the thermal reactor pin, and a
these results are shown in Figure 17. Also reported micrograph of the fast reactor pin.
in the figure is the calculated plutonium redistribu- Ronchi and Sari performed experiments to deter-
tion, which is discussed in the following section. mine the rate of columnar grain growth in samples of
In commenting about this calculation Lackey et al. sug- (U0.85Pu0.15)O2. In this experiment, they used ‘old’
gested that axial transport of vapor may have occurred, samples of MOX in which they speculated that the
but they did not make a quantified measurement of this, decay of Pu with time would build up a pressure of
20 22
Initial smear density
18 21
Measured
Plutonium content (% Pu/(U+Pu))
20 Predicted
16
19
14
Predicted 18
Porosity (%)
12
Orientation 17
10 – 0⬚
16
– 120⬚ Quantimet
8 X – 240⬚ 15
– 0⬚, 120⬚, 240⬚ Manual
14
6 X
13
4
Columnar grains 12
2
Negligible 11
vapor transport
0 10
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
(a) Fractional fuel radius (b) Fractional fuel radius
Figure 17 Sample rod section (unetched), sphere-pac element, initially 81.4% smeared density, U0.85Pu0.15O2.00, irradiated
in a thermal reactor at 44.6 kW m1 to 0.7% FIMA. (a) Porosity measurements compared to model predictions, scatter near
central void is due to actual variation of porosity in this area. (b) Measured and calculated Pu/M ratio. Reproduced from
Lackey, W. J.; Homan, F. J.; Olsen, A. R. Nucl. Technol. 1972, 16. Copyright 1972 by the American Nuclear Society, La
Grange Park, Illinois.
662 Matter Transport in Fast Reactor Fuels
30
200 mm
Central Columnar
void grains
28
26
Orientation
24 – 0⬚
– 120⬚
22
20 Initial porosity
18
Porosity (%)
16
14
12
10
0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
(a) (b) Fractional fuel radius
Figure 18 Micrograph and measured porosity for pins irradiated to medium burnup showing deposition of material onto
inner surface of cladding. (a) Micrograph of (U0.8Pu0.2)O1.99 irradiated in fast neutron flux at 44.3 kW m1 to 5.7% FIMA
(unetched) and (b) Measured porosity distribution for (U0.85Pu0.15)O2.00 irradiated at thermal flux, 44.0 kW m1 to 4.2% FIMA.
Reproduced from Lackey, W. J.; Homan, F. J.; Olsen, A. R. Nucl. Technol. 1972, 16. Copyright 1972 by the American Nuclear
Society, La Grange Park, Illinois.
to nuclear reactors is the neutronic effects and the results using a fit parameter to determine U–Pu sepa-
change in heat generation, which, as mentioned in the ration as the pore migrates, and the pore migration rate
introduction to this chapter, may culminate in a pen- of Clement and Finnis.85,96
alty of the maximum linear heat rating by as much as Clear evidence of the vaporization mechanism
8.2 kW m1, 15% for a liquid metal fast breeder acting on the macroscopic scale was also observed
reactor MOX fuel element compared to the restruc- by Lackey et al. through comparison of a photo-
tured case without plutonium redistribution;1 how- micrograph and autoradiograph investigation of the
ever, as noted by Lackey et al.,2 this conclusion is sphere-pac element described in Figure 20. Compar-
incomplete as it misses the effects of relocation of ison of the results show that the space between the
oxygen to the outer radii and the lowering of the spheres near the periphery appears to have been
melting point by local Pu content. Naturally, the filled with a material of much lower Pu concentra-
fission process will also affect the Pu distribution, tion, as is visible on the a-autoradiograph. Thus, the
both from fission and transmutation processes, and appearance of the Pu islands remains from the origi-
this fact must be borne in mind.94 nal microstructure. This, along with a notable enrich-
Uranium and plutonium may redistribute as ment of Pu as the surface of the central void, indicates
a result of several overlapping mechanisms. Vapor that U-bearing species preferentially evaporated
transport may result in redistribution from either from the inner region and condensed in the free
macroscopic interconnected voids or via migrating space between the outer spheres.
closed porosity although the latter has previously The effects of sintering on closing vapor transport
been seen not to contribute largely to redistribution, pathways were addressed by Meyers, who emphasizes
in Section 3.21.3.3. Segregation by vapor transport the importance of this phenomenon.79 The model
relies on noncongruent evaporation of U- and includes vapor transport and sintering although it
Pu-bearing vapor species. The ratio of U/Pu in the discounts thermodiffusion, citing a much lower driving
vapor turns out to be dependent on the O/M in force compared to vapor transport. The author does
the solid, and there is a particular composition where acknowledge the buildup of Pu in front of a migrating
U/Pu ¼ 1. This was investigated by Bober et al. pore but, in contrast to Olander’s treatment, assumes
by experiments on cylindrical samples of (U85Pu15) that this will dissipate rapidly and so does not require
O2 x with a hole bored out in the middle.95 The the treatment discussed in Section 3.21.3.3.82 The
samples were subjected to an axial temperature gradi- results of his computer model reproduce the features
ent of maximum temperature 2673 K. Within the of the observed behavior well, and lead him to con-
borehole, redistribution was able to occur predomi- sider sintering to be of high importance in modeling
nately by preferential evaporation from the surface of restructuring and redistribution.
the sample into the empty space. The autoradiographs The appearance of a dense region in which segrega-
were then examined and, based on alpha activity, tion may no longer proceed via vapor transport and
the composition of U and Pu was observed. With pore migration implies that solid state thermodiffusion
the maximum temperature at 2673 K, experiments would be responsible for any further redistribution. To
with O/M 1.97 were seen to preferentially evapo- investigate the thermomigration of U–Pu, a series of
rate uranium oxides, whereas plutonium oxides pref- out-of-pile experiments were performed by Bober,
erentially evaporated for O/M 1.96. Sari, and Schumacher.3,26,40,97 The apparatus employed
Lackey et al. calculated actinide redistribution a gas-tight capsule in order to prevent evaporation of
through the use of their restructuring model and the fuel material or oxygen exchange with the environ-
results are shown in Figure 17.2 Redistribution was ment. A linear temperature gradient was applied with
included by considering the ratio of Pu/M in the a quoted maximum temperature of approximately
vapor phase depending on the temperature and local 2800 K, with an approximately linear gradient of up
O/M ratio as determined by the thermodynamic model to 1500 K cm1 in the sample. Experiments were per-
of Rand and Markin.48 Although the model employed formed with this apparatus on MOX samples with Pu/
by Lackey et al. is in fact based on pore migration, which M ¼ 15% and differing stoichiometry and the resulting
has been postulated not to result in much redistribution Pu distributions measured by alpha counting. Two
(Section 3.21.3.3), the results may still be considered experimental results are shown in Figure 21 and
valid for the same reason this porosity redistribution exhibit different profile characteristics corresponding
model is applicable as previously discussed. Ishii and to different dominant redistribution mechanisms for
Asaga modeled U–Pu redistribution and report good the experimental time and stoichiometry.
664 Matter Transport in Fast Reactor Fuels
0.25 in.
Photomacrograph a Autoradiograph
Figure 20 Composite photomicrograph and matching autoradiograph of a sphere-pac fuel element. Photomicrograph
(left) shows the extent of restructuring in filling in the gaps between spheres. Autoradiograph (right) reveals Pu islands, and
enrichment of Pu at central void periphery, indicating preferential vaporization of U. Inset shows fuel deposited from the vapor
as dendrites on the hot side of the spheres at the region between columnar and equiaxed grains. Reproduced from
Lackey, W. J.; Homan, F. J.; Olsen, A. R. Nucl. Technol. 1972, 16. Copyright 1972 by the American Nuclear Society, La
Grange Park, Illinois.
Temperature (⬚C) occurred at the hot surface into the space between
2500 2300 2100 1900 1700 the sample and the sample holder, resulting in an
additional Pu enrichment.
Initial O/M = 2.00
The second sample was heated for 48 h with
Initial O/M = 1.98
O/M ¼ 1.98. The stoichiometry was closer to the
20 congruent vaporization composition and therefore
PuO2 (mol %)
hot regions, contrary to the Pu enrichment at the The point defects are considered to be in dilute
central region shown in eqn [20].2 solution in the fuel matrix, and so eqns [42] and [46]
The mechanistic code of Lewis et al., which con- may be considered to apply. With this treatment of
siders fuel oxidation in the operation of defective fuel, the system, we can relate the mole fraction of vacan-
yields a similar conclusion. This work considers the cies/interstitials to the stoichiometry deviation by
presence of steam in contact with fuel resulting from a Nn ¼ x/2 and Ni ¼ x in the dilute limit. The mass
breach in the fuel sheath and resulting contact with the flux equations are thus given by:
coolant in a CANDU reactor. The model considers
Ni Q i
steam (actually gaseous D2O) due to the use of heavy Ji ¼ nDi rNi þ rT ½90a
fi kT 2
water in CANDU reactors; however, deuterium is
assumed to behave in a chemically identical fashion ð1 Nv ÞNv Q v
to hydrogen diffusion in the radial cracks, reaction with Jv ¼ nDv rNv þ rT ½90b
fv kT 2
the solid fuel, and thermodiffusion along with the
evolving temperature profile. The simulation results in which n represents the total local concentration of
show that H2/H2O (D2/D2O) ratio along the cracks is oxygen ions. For the steady state, this gives a form
not constant in the region of applicability.104–106 analogous to eqn [48]:
Olander performed a theoretical investigation on xð1 x=2Þ Q v
rx ¼ rT ½91
the idea of a constant ratio CO2/CO and determined 1 þ 2ð@ ln g=@ ln xÞ RT 2
that this assumption is valid for regions of lower
for hypostoichiometric (U, Pu)O2 x and
temperature but smoothly diminishes at higher tem-
peratures.103 Adamson proposed the limit for domi- x Qi
rx ¼ rT ½92
nance of carrier gas via the COCO2 transport be set 1 þ ð@ ln g=@ ln xÞ RT 2
to 1873 K in UO2 þ x, above which UO3 vapor pres- for hyperstoichiometric (U, Pu)O2 þ x .
sure is high enough to dominate oxygen redistribu- The results are presented in Figure 23 and com-
tion.47 The exact effect of UO3 diffusion is difficult to pared against other measurements,10,100,108,109 and fit
calculate due to sluggish reaction kinetics with solid to the curves given in eqn [93].
fuel implying that the vapor and the solid may not
Q v ¼ 9:45 105 þ 5:66 105 VPu 8:5 104 VPu
2
½93a
be in equilibrium, a requisite for the treatment
of the type described in Section 3.21.2.3. However, Q i ¼ 8:3 1033 expð17VU Þ ½93b
the effect may be seen as a departure from linearity
compared to eqn [48] although this may also be As can be seen from Figure 23, the heat of trans-
attributed to a varying carrier-gas ratio.47 port does not appear to be influenced by the U/Pu
At higher temperatures, the mobility of oxygen ratio in either the hyper- or hypostoichiometric case,
ions increases and so solid state thermomigration is rather only the U/Pu valence. Scatter increases
thought to play an increasingly important role. approaching the stoichiometric state as a result
Sari and Schumacher conducted a set of experi- of uncertainty in the measurement of oxygen con-
ments designed to determine the degree of oxygen centration as O/M approaches 2. The valence of
redistribution in MOX as a function of stoichiome- U and Pu is a useful parameter, which occurs because
try and U/Pu, and compared it with other experi- a deficit or excess of oxygen may preferentially
mental results.107 This required the minimization associate with plutonia or urania, respectively.48
of carrier-gas-assisted oxygen diffusion, which was Thus, hypostoichiometric fuel may be considered
done by coating the hypostoichiometric samples as a solid solution of UO2 and PuO2 x, whereas
with Mo, and tightly fitting the hyperstoichiometric hyperstoichiometric fuel would be a solid solution
samples in W or Mo containers. The authors of the of UO2 þ x and PuO2.26 It may be related to the
work acknowledge that while the decision to use tight deviation from stoichiometry quite readily110:
metallic capsules minimizes vapor phase transport,
2ðO=M 2Þ
the metallic sample containers may affect the electric VU ¼ 4 þ ½94a
potential that the sample generates by the thermo- 1q
electric effect, which should be considered part of the 2ðO=M 2Þ
heat of transport as discussed in Section 3.21.2.2. VPu ¼ 4 þ ½94b
q
Experiments were conducted with both radial and
axial heating apparatuses. where q ¼ Pu/M.
Matter Transport in Fast Reactor Fuels 667
O / (U + Pu) in (U0.8Pu0.2) O2 ± y
2.16 2.08 2.00 1.98 1.96 1.94
−41.8 x
x
x
x
x
0
4.5 4.4 4.2 4.0 3.8 3.6 3.4 3.2
VU VPu
Figure 23 The heat of transport of oxygen vacancies/interstitials in mixed oxide showing results from Sari and
Schumacher107 with (○) 15, (△) 20, (þ) 30, (x) 40, (□) 50, (◊) 85, (
) 100 mol% PuO2. Other data include (),108 (◘),100 and (◘) .10
Reproduced from Sari, C.; Schumacher, G. J. Nucl. Mater. 1976, 61, 192–202.
There has been some theoretical work on cal- from the TIP and eqn [5], including also a charge
culating the heat of transport of oxygen in MOX; current. The results of Kamata and Esaka are in
however, these efforts have met with little success. reasonably good agreement with the experimental
As discussed in Section 3.21.2.2, Qv or Qi can be data of Sari and Schumacher, indicating that indeed
calculated from either a thermodynamic or electro- the electrochemical contribution from the thermo-
chemical standpoint. Proponents of the thermo- electric effect may play an important, if not domi-
dynamic contribution tend to cite a change in sign nant, role in the perceived thermomigration. Janek
of the thermoelectric power at approximately and Timm have, however, criticized the work of
13731573 K.27 Later, Ruello et al. determined that Kamata and Esaka citing an unreasonable behavior
the sign of the Seebeck coefficient changes depend- near stoichiometry and report their own calculations,
ing on oxygen potential above 1200 K.111 Aitken first which are summarized in Figure 24 and show quite a
proposed that the heat of transport be expressed in good agreement.
terms of the partial molar enthalpy of oxygen in Once one considers that the thermoelectric
the solid phase and determined Q ¼ 1=2h O2 effect may enter into the experimental determina-
corresponding to hyper- and hypostoichiometric tion, one must consider that encapsulating the sample
fuel, respectively.38 However, as hO2 is always nega- in a metallic container may short-circuit the field
tive for MOX, it implies that oxygen always migrates generated by the sample and so perturb the measure-
down the temperature gradient, contrary to observa- ment of, for example, the experiment of Sari and
tions.27 Norris realized this and proposed that in fact Schumacher. Millot and Gerdanian treated the sys-
the partial molar enthalpy of interest should be half tem including the surrounding container in order to
the enthalpy of formation of an oxygen Frenkel pair, consider the effects of short-circuiting the thermo-
implying Q ¼ 1=2DhF which, since DhF is always electric field via interaction with tight-fitting metallic
positive, assures the correct sign of Q*, although containers.113 They determined that not only did
the magnitude was still incorrect.39 Finally, Sugisaki electrotransport contribute to the observed heat of
performed a similar calculation for Q* directly, via transport, but it is of the same sign and is of compa-
a computer simulation technique to calculate the rable magnitude for UO2.03.
excess energy associated with a defect movement;
however, while again predicting the correct sign, this 3.21.5.1.4 Fission products and other
did not reproduce the experimental results well.27 elements
A greater degree of success seems to be obtained The creation of fission products changes the fuel
by considering the thermoelectric effect. The justifi- from its original composition into a multicomponent
cation for this is that because UO2 is a mixed ionic system. Fission products may be present in the fuel in
and electronic conductor, it may negate the previous the form of solid precipitates, or gas bubbles, or
argument regarding the change of sign in the Seebeck in solid solution with the fuel matrix. All three
coefficient.112 Theoretical treatments are developed forms are affected by thermodiffusion, but the
668 Matter Transport in Fast Reactor Fuels
Q*o2-
−50
Qoexp/(kJ mol−1)
−100 Vo Oi
−150
Q*o
−200
1.90 1.95 2.00 2.05 2.10
O/M in (U,Pu)O2 + x
Figure 24 Investigation on the actual heat of transport of oxygen in mixed oxide. (a) The true heat of transport Qo and
ionic heat of transport Q02 by Janek and Timm compared to experimental Qexp0 from Figure 23. Modified from
Janek, J.; Timm, H. J. Nucl. Mater. 1998, 255. Copyright (1998), with permission from Elsevier.
application differs. Tracking the migration of all spe- as shown in eqn [25]. The concentration profile
cies would require a very large system of equations in demonstrates the result of thermodiffusion and void
the form of eqn [1], but such a complete treatment is migration acting in opposite directions. Specifically,
impractical due to computational requirement and the minimum is attributed to thermomigration acting
lack of knowledge of the cross diffusion coefficient. to drive Ce up the gradient, in opposition to the void
Moreover, it is not required since often the dilute migration, which preferentially moves Ce preferen-
limit is applicable. tially down the gradient, producing a maximum
We first consider fission products in solid solution nearer to the top of the sample. It is also noted that
in the fuel matrix, which amount to approximately some preferential Ce evaporation from the surface
30% of the yield from 239Pu, and comprise approxi- has occurred, resulting in a decrease in the average
mately 2.5% of the fuel after 5% burnup. These Ce/M ratio. The process heat of transport was calcu-
species are subject to solid state thermodiffusion sim- lated using eqn [39] as:
ilar to uranium and plutonium. If the flows are con-
Q UCe ¼ 100:4 kJ mol1 ½95
sidered to be in dilute solution, we may assume that
the fluxes are independent, and so proceed with cal- by consideration of the minimum concentration, with
culation of the flux of fission products using eqn [39]. a caveat that the value is influenced by the migrating
If we can consider that the diffusion coefficient of the voids and would be larger for pure thermodiffusion
fission product is much greater than that of U or Pu, (Figure 25).
then DF DU, DPu, and Q FM become approximately The transport of cesium, tellurium, and iodine in
equal to Q F , however, this is not necessarily the case. U–Pu fuels was measured in stainless steel capsules for
In order to investigate the redistribution of 100 h.4 Volatile fission products were found to migrate
cerium, experiments were conducted on samples down the temperature gradient via vapor transport.
of uranium–cerium MOXs.114 The case is analogous This was influenced by the O/M ratio, with a notable
to that of (U, Pu)O2 as cerium oxide forms a homo- effect on Cs in hyperstoichiometric fuel due to the
geneous solid solution with UO2; however, the mobil- formation of the less volatile cesium oxide.
ity of Ce is higher and so lower temperatures are The redistribution of americium has been
required to observe appreciable effects. In particular, observed recently and shows an increasing concen-
a case of (U0.8Ce0.2)O2 subjected to a linear thermal tration near the central void. Tanaka et al. irradiated
gradient of about 1273 K with a maximum tempera- MOX with O/M ¼ 1.98 and 3 and 5% Am for
ture of 2393 K showed a concentration distribution, 10 min.115 Visual investigation determined pore
Matter Transport in Fast Reactor Fuels 669
T (K)
1650 1750 1850 1950 2050 2150 2250 2270
15
CeO2 (wt.%) 14
13
12
11
10
9
1 2 3 4 5 6 7 8
Depth (mm)
Figure 25 Redistribution in the (U, Ce)O2 system showing thermodiffusion of Ce up the temperature gradient
superimposed on preferential CeO2 vaporization down the gradient.26 Reproduced from Beisswenger, H., Bober, M.;
Shumacher, G. J. Nucl. Mater. 1967, 21. Copyright (1967) with permission from Elsevier.
migration and central void formation had already Gulden as well as Bober and Shumacher suggest that
occurred, and chemical characterization showed an carbon diffusing through carbides may be treated as
increase in Pu and Am fractions near the central void. occurring through vacancies and interstitials through
Maeda et al. measured Am concentrations on samples an immobile metal lattice.26,117 The mass flux may be
with O/M of 1.98 and 1.95 and modeled the phenom- represented by eqn [46], and the steady state distribu-
ena assuming pore migration and thermodiffusion.46 tion by eqns [48c] and [48b].
Pore migration showed good agreement with the The heat of transport of carbon has been
observed Am profiles, whereas thermodiffusion was measured experimentally. Wallace et al. concluded
reported to occur too slowly to contribute during the that the heat of transport of carbon in b UC1.83 is
24-h irradiation time. They also reported more Am 1.5 kJ mol1; however, Gulden comments that in this
enrichment for the 1.98 sample, in line with the larger analysis, the thermodynamic factor, which is large
UO3 vapor pressure at this O/M. for UC2 was neglected.117,118 Gulden conducted
an experiment to measure the heat of transport
of carbon in UC2 via the migration rate of a UC2
3.21.5.2 U–Pu–C–N
disk through Pyrolytic carbon, leaving graphite in
Carbide and nitride fuels are an attractive alterna- its wake117:
tive to oxide fuels as they possess higher thermal
Q C ¼ 386
113 kJ mol1 ½96
conductivities and density of fissile elements. The
higher conductivity reduces the potential for high He also verified that another potential thermody-
centerline temperatures unless the linear power is namic driving force, the chemical potential between
scaled accordingly, which tends to produce less ther- the pyrolytic carbon and the rejected carbon, was
mally induced restructuring and redistribution.26 negligible.
A typical centerline temperature is 1500 K.116 Bubbles of fission gas are also produced in carbide
More details are available on the behavior of carbides fuels. In analyzing the mobility of spherical bubbles
over nitrides and so this section focuses more on compared to their radius as discussed in Section
carbide behavior. 3.21.3.2, Weeks et al. determined that volume diffu-
The diffusion rate of carbon in uranium carbides sion of carbon is important in bubble migration
is known to be greater than that of the metals (see in UC.119 This conclusion was later challenged by
Chapter 2.04, Thermodynamic and Thermophysical Matzke who proposed that while carbon may diffuse
Properties of the Actinide Carbides). Therefore, rapidly in the solid state, uranium bulk diffusion is
670 Matter Transport in Fast Reactor Fuels
slow. He also noted that the equilibrium vapor in vs. bubble size is shown in Figure 26, including an
contact with (U, Pu)C is predominantly metal, with experiment datum from a fast flux irradiation exper-
very little C.120,121 Therefore, he proposed a trans- iment.122 The activation energy for void migration is
port mechanism that relied on vapor phase transport thus the slower of the steps required to move a
of (U, Pu) across the void supported by fast carbon complete molecule and so the activation energy of
diffusion.69 The resulting graph of migration speed the pore migration is 5 eV, the heat of vaporization
of U from UC.
4 It was also noted that the vapor pressure above
2000 ºC dT/dx = 500 ºC cm-1 uranium carbide contains a much greater amount of
3 Pu than U.69,121 This of course has implications for
the vapor transport that may occur in this fuel type.
2 DVC
As void migration is considered to occur by vapor
Ds
transport of the metal species, this then implies
1
redistribution of actinides as the void migrates in
log velocity (Å s-1)
1 mm
1 mm
(a) (b)
Figure 27 Fuel samples irradiated to (7 at.%) in Phenix fast reactor showing (a) (U,Pu)C fuel with porous inner region
and (b) (U,Pu)N fuel with porous interior and formation of a noncentered central void. © European Atomic Energy Community,
reproduced with permission.
Matter Transport in Fast Reactor Fuels 671
Cracks
Zone of increased density phase and exists between 923 and 1023 K.124 The
phase is enriched in Zr, showing an increase from
0.24 the initial 23–42 at.% with a corresponding loss of
U from 61 to 38 at.%.123 The second region contains
0.22
a mix of g and z phases from 873 to 923 K.123 This
Crack
0.20 phase is enriched in U and depleted in Zr and does not
exhibit a gradient in concentration. It is also the
0.18
densest and contains fine bubbles.125 The third, lowest
Porous zone
0.16 temperature region contains a mixture of d and z
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 phases below 873 K.123 This phase is slightly enriched
Radial position (r/ro) in Zr and contains irregular voids originating from
grain boundary tearing due to anisotropic growth of
Figure 28 Pu/M graph measurements in a sample of
U0.8Pu0.2C0.8N0.2 irradiated to 3.9 at.% in a fast flux at
U grains. More details on this fuel type may be found
1250 W cm1.116 © European Atomic Energy Community, in Chapter 2.05, Phase Diagrams of Actinide Alloys
reproduced with permission. and Chapter 3.01, Metal Fuel of the current work.
(a) (b)
Figure 29 Pictures of void migration in C-rich fuels as revealed by a color etching technique. Trails are clearly visible
indicating variations of composition and structure trailing migrating voids.116 © European Atomic Energy Community,
reproduced with permission.
672 Matter Transport in Fast Reactor Fuels
(a)
0.5 mm
U
Zr
5 mm 50 mm 10 mm
0.5 mm
(a) (b)
Figure 30 Postirradiation analysis of U–Pu–Zr fuel with weight percents 71/19/10 at 1.9 at.% burnup, 42 kW m1 linear heat
rate, with a maximum calculated temperature of approximately 943 K. Side A results from optical metallography with an
overlay of constituent concentration from X-ray intensities to reveal three zone microstructure. Side B: from optical and
scanning electron microscopic enlargements of various zones showing (a) axial section, (b) intermediate zone, (c) central
zone, and (d) outer zone. Modified from Kim, Y. S.; Hofman, G. L.; Hayes, S. L.; Sohn, Y. H. J. Nucl. Mater. 2004, 327.
Copyright (2004), with permission from Elsevier.
Kim et al. conducted experiments in order to inner pyrolytic carbon layer, a silicon carbide layer,
determine ternary diffusion coefficients, heats of and outer pyrolytic carbon layer.127
transport, and enthalpies of solution for this sys- The dimensions of TRISO particles depend on
tem.123 Minor actinides, Am and Np, were also the intended design but are typically of the order
added to the U–Pu–Zr fuel and their migration stud- of hundreds of micrometers in radius. Thousands
ied. Tests indicated that the presence of the actinides of particles are collected into pebbles or compacts.
(1.2–1.3 wt%) did not greatly affect the migration of As the coolant passes around the surface of these
U, Pu or Zr. The results suggested that Am behaves compacts, a temperature gradient develops across
similarly to Zr by increasing in the central zone and the particles. The temperature gradient may induce
slightly in the outer periphery. Am was also found in kernel migration, also known as the Amoeba effect, an
regions of large porosity, with a notable reduction in image of which is shown in Figure 31. The effect has
the intermediate, dense zone. Np was not observed to been observed in all TRISO particles with a UO2
significantly redistribute.123 Later, the migration of kernel and also thoria-based fertile kernels, for exam-
some lanthanides (La, Ce, Nd, Pr) was also studied ple, (Th, U)O2, ThO2, (Th, U)C2, and ThC2.128,129 If
and compared to Am in U–Zr and U–Pu–Zr fuels. kernel migration proceeds far enough, it may result
The results of irradiation experiments suggest that in a breach of the structural coating layer, causing
they migrate by interdiffusion in the matrix and fuel-coating failure.
precipitation in pores.126 The kernel migration speed is often described by
the kernel migration coefficient (KMC):
3.21.5.4 TRISO Particles KMC ¼ KKMC exp DHm;kernel =kT ½97
TRISO particles are of interest for use in high- in which KKMC is a parameter that depends on the
temperature gas-cooled reactors as they can permit mechanism by which the kernel moves, and DHm,kernel
high coolant temperatures and fuel burnup. TRISO is an effective activation energy. For UCO and UC2
particles consist of a fuel kernel, which contains kernels, carbon is thought to migrate down the tem-
a fissionable element (uranium, plutonium, thorium, perature gradient via solid state thermodiffusion,
or another transuranic element) in an oxide, carbide, as discussed in Section 3.21.5.2.117,130,131 In this case,
or oxycarbide form. Surrounding the kernel is a the KMC would depend on Q C . In UO2-based
porous carbon buffer, which catches recoil fission kernels, however, thermodiffusion of C is not con-
fragments, accommodates internal gas buildup, and sidered to be rate controlling, and instead a more
dimensional change. Continuing outwards are an elaborate vapor-based concept, reliant on the presence
Matter Transport in Fast Reactor Fuels 673
(a) (b)
150 mm 100 mm
Figure 31 Migration of a UO2 kernel through the porous graphite buffer showing (a) full cross-section and (b) enlargement of
kernel. Crescent layers are visible behind the kernel movement, indicating a possible C transport mechanism. Also of note is the
filling of voids inside the kernel by carbon. Reproduced from Bullock, R. E.; Kaae, J. L. J. Nucl. Mater. 1983, 115. Copyright
(1983), with permission from Elsevier.
C
CO2
Dense PyC
CO CO2
O2-
Porous
buffer
Free
oxygen
CO2
Oxide kernel Hindered by
fission gas
Temperature
Radiolytic CO ® C + O2−
decomposition?
O2-
CO2
CO
C CO2 + C (deposit)
CO
Exothermic Surface reaction
adsorption Inhibited by fission products?
Figure 32 Schematic of potential mechanisms for UO2 kernel migration in coated particle fuels. Modified from
Wagner-Löffler, M. Nucl. Tech. 1977, 35, with permission from American Nuclear Society.
of oxygen from contamination or UO2 is generally transport mechanism made possible by the presence
considered.131 This hypothesis of oxygen playing a of O liberated from the kernel or present by contam-
key role in kernel migration is supported by observa- ination. The vapor may diffuse down the temperature
tions that the presence of an oxygen ‘getter’ such as gradient through the porous buffer and reduce to
ZrC eliminates this phenomenon. Similarly, using a deposit the carbon on the cold side, thereby pushing
UO2UC2 mixed kernel lowers the CO and CO2 the kernel up the temperature gradient. Evidence of
pressures, reducing the kernel migration rate.132 this mechanism is the appearance of crescent layers
Potential mechanisms for kernel migration via vapor behind the kernel shown in Figure 31. Oxygen must
phase transport of carbon are shown in Figure 32.131 then be returned to the hot side of the kernel in order
In this conception, carbon is transported down for the process to continue, although the mechanism
the temperature gradient by a carrier-based vapor for this return is not clear. The first mechanism is by a
674 Matter Transport in Fast Reactor Fuels
countercurrent flow of CO2 gas flowing in the oppo- 4. Aitken, E. A.; Evans, S. K.; Rosenbaum, H. S.; Rubin, B. F.
Trans. Am. Nucl. Soc. 1971, 14, 176.
site direction. The second is the uptake of the oxygen 5. Tourasse, M.; Boidron, M.; Pasque, B. J. Nucl. Mater.
by the kernel and solid state diffusion driven by either 1992, 188, 49–57.
differing oxygen partial pressure at the front and rear 6. Tanaka, K.; Maeda, K.; Sasaki, S.; Iskusawa, Y.; Abe, T.
J. Nucl. Mater. 2006, 357, 58–68.
surface or via thermomigration of the oxygen ions as 7. Maeda, K. J. Nucl. Mater. 2005, 344, 274–280.
described in Section 3.21.5.1. 127,130 8. Rondinella, V. V.; Wiss, T. Materials Today 2010, 13(12),
24–32.
9. Olander, D. R. Fundamental Aspects of Nuclear Reactor
Fuel Elements; Technical Information Center, Office of
3.21.6 Summary Public Affairs: Springfield, VA, 1976.
10. De Groot, S. R.; Mazur, P. Non-Equilibrium
Thermodynamics; General Publishing: Toronto, ON,
In summary, the causes and effects of some major 1984.
matter transport phenomena in nuclear fuel have 11. Fitts, D. D. Non-Equilibrium Thermodynamics;
McGraw-Hill: New York, 1962.
been reviewed. Fundamental thermodynamic driving 12. Philibert, J. Atom Movements: Diffusion and Mass
forces of matter transport were considered, and their Transport in Solids; Les Éditions de Physique: Orsay,
mathematical expressions derived. The emergence France, 1991.
13. Allnatt, A. R.; Lidiard, A. B. Atomic Transport in Solids;
and behavior of microscopic structures was then dis- University Press: Cambridge, 1993.
cussed, and their effect on restructuring and redistri- 14. Kreuzer, H. J. Nonequilbrium Thermodynamics and
bution determined. The bulk, macroscopic effects of Its Statistical Foundations; Oxford University Press:
Oxford, 1983.
matter transport demonstrated the results of restruc- 15. Bi, Z.; Sekerka, R. F. Phys. A 1998, 261, 95–106.
turing on scales comparable to the dimensions of the 16. Soret, M. Ch. Ann. Chim. Phys. 1881, 22, 293.
fuel itself. Finally, specific cases of different fuel 17. Dufour, L. Pogg. Ann. 1873, 148, 490.
18. Hartley, G. S. Trans. Faraday Soc. 1946, 42, 6.
types were given and described using the framework 19. Smigelskas, A. D.; Kirkendall, E. O. Trans. AIME 1947,
set forth by the other sections. 171, 130.
The goal of the author is that this work should 20. Manning, J. R. Diffusion Kinetics for Atoms in Crystals;
Van Nostrand: Princeton, NJ, 1968.
help in clarifying concepts and physical quantities 21. Darken, L. S. Trans. Met. Soc. AIME 1948, 175, 184–187.
associated with matter transport, and reveal how the 22. Darken, L. S.; Gurry, R. W. Physical Chemistry of Metals;
variety of matter transport phenomena in different McGraw-Hill: New York, 1953.
23. Matzke, Hj. J. Chem. Soc. Faraday Trans. 1987, 2,
fuel types are in fact manifestations of the same 1121–1142.
fundamental mechanisms. Looking forward, it seems 24. Lay, K. W. J. Am. Ceram. Soc. 1970, 53, 369–373.
apparent that future discussions must account for 25. Breitung, W. J. Nucl. Mater. 1978, 74, 10–18.
26. Bober, M.; Schumacher, G. Material Transport in the
radiation-induced restructuring and redistribution. Temperature Gradient of Fast Reactor Fuels; Academic
Of particular note is the mentioned rim structure, Press: New York, 1973; Vol. 7, Chapter 2, 121.
JOG, and white phase. In addition, the importance of 27. Sugisaki, M.; Sato, S.; Furuya, H. J. Nucl. Mater. 1981, 97,
79–87.
the thermoelectric effect on matter transport in 28. Wirtz, K. Ann. Phys. (Leipzig) 1939, 36, 295.
MOX may also be further investigated. 29. Wirtz, K. Phys. Z 1943, 44, 221.
30. Jones, C.; Grout, P. H.; Lidiard, A. B. Ber. Bunsenges.
Phys. Chem. 1997, 101, 1232.
31. Janek, J.; Timm, H. J. Nucl. Mater. 1998, 255, 116–127.
Acknowledgments 32. Nichols, F. A. J. Nucl. Mater. 1969, 30, 143–165.
33. Desai, T. G.; Millett, P.; Tonks, M.; Wolf, D. Acta Mater.
2010, 58, 330–339.
The author gratefully acknowledges M. Stan (ANL), 34. Shewmon, P. G. Diffusion in Solids; McGraw-Hill:
P. Van Uffelen (JRC-ITU), and V. Di Marcello New York, 1963.
(JRC-ITU) for their review of the manuscript. 35. Shewmon, P. G. Trans. Met. Soc. AIME 1964, 230,
1134–1137.
36. Shewmon, P. G. Diffusion in Solids; 2 ed.; McGraw-Hill:
New York, 1989.
References 37. Raleigh, D. O.; Sommer, A. W. J. Phys. Chem. 1962, 36,
381.
38. Aitken, E. A. J. Nucl. Mater. 1969, 30, 62–73.
1. Sha, W. T.; Huebotter, P. R.; Lo, R. K. Trans. Am. Nucl. 39. Norris, D. I. R. J. Nucl. Mater. 1979, 79, 118–127.
Soc. 1971, 14, 183–184. 40. Bober, M.; Sari, C.; Schumacher, G. J. Nucl. Mater. 1971,
2. Lackey, W. J.; Homan, F. J.; Olsen, A. R. Nucl. Technol. 39, 265–284.
1972, 16, 120–142. 41. Gill, S. P. A. Int. J. Heat Mass Transfer. 2009, 52,
3. Bober, M.; Schumacher, G. Transport Phenomena in 1123–1131.
Mixed Oxide fuel Pins; Technical Report KFK 1887; 42. Lupis, C. H. P. Chemical Thermodynamics of Materials;
Institute für Neutronenphysik und Reaktortechnik, 1973. Elsevier, 1983.
Matter Transport in Fast Reactor Fuels 675
43. Breitung, W.; Reil, K. O. Nucl. Sci. Eng. 1989, 101, 26. 85. Clement, C. F.; Finnis, M. W. J. Nucl. Mater. 1978, 75,
44. Biachi, M.; Chatillon, C.; Ducros, G.; Forment, K. J. Nucl. 193–200.
Mater. 2006, 349, 57–82. 86. Katsuyama, K.; Nagamine, T.; Matsumoto, S. I.; Ito, M.
45. Stølen, S.; Grande, T. Chemical Thermodynamics of J. Nucl. Sci. Tech. 2002, 39, 804–806.
Materials; Wiley: Hoboken, NJ, 2003. 87. Christensen, J. A. Technical Report BNWL-536; Battelle
46. Maeda, K.; Sasaki, S.; Kato, M.; Kihara, Y. J. Nucl. Mater. Northwest: Richland, WA, 1967.
2009, 389, 78–84. 88. Welland, M. J. Ph. D. Thesis, Royal Military College of
47. Adamson, M. G.; Carney, R. F. A. J. Nucl. Mater. 1974, Canada, Kingston, ON, 2009.
54, 121–137. 89. Welland, M.; Lewis, B. J.; Thompson, W. T. J. Nucl.
48. Rand, M. H.; Markin, T. L. In Thermodynamics of Nuclear Mater. 2011, 412, 342–349.
Materials; Internaltional Atomic Energy Agency: Vienna, 90. Hellwig, Ch.; Bakker, K.; Ozawa, T.; et al. Nucl. Sci. Eng.
1967; pp 637–650. 2006, 153, 233.
49. Adamson, M. G. J. Nucl. Mater. 1971, 38, 213–216. 91. Christensen, J. A. Trans. Am. Nucl. Soc. 1972, 15, 214.
50. Tikare, V.; Holm, E. J. Am. Ceram. Soc. 1998, 81, 480. 92. Kawamata, H.; Kaneko, H.; Furuya, H.; Koizumi, M.
51. Miodownik, M.; Martin, J. W.; Cerezo, A. Philos. Mag. A J. Nucl. Mater. 1977, 68, 48–53.
1999, 79, 203. 93. Potter, T. J.; Elyard, C. A. Proc. Br. Ceram. Soc. 1967, 7,
52. Chen, L. Q. Annu. Rev. Mater. Res. 2002, 32(1), 113–140. 273.
53. Shenyang, H.; Henager, C. H., Jr.; Heinisch, H. L.; 94. Zwicky, H. U.; Aerne, E. T.; Hermann, A.; Thomi, H. A.;
Stan, M.; Baskes, M. I.; Valone, S. M. J. Nucl. Mater. Lippens, M. J. Nucl. Mater. 1993, 202, 65–69.
2009, 392, 292–300; Nuclear Fuels and Structural 95. Bober, M.; Sari, C.; Schumacher, G. J. Nucl. Mater. 1971,
Materials 2, Proceedings of the Second Symposium on 40, 341–345.
Nuclear Fuels and Structural Materials for Next 96. Ishii, T.; Asaga, T. J. Nucl. Mater. 2001, 294, 13–17.
Generation Nuclear Reactors. 97. Bober, M.; Sari, C.; Schumacher, G. Trans. Am. Nucl.
54. Matthews, J. R. J. Nucl. Mater. 1979, 87, 356–366. Soc. 1969, 12, 603.
55. Veshchunov, M. S. Materials 2009, 2, 1252–1287. 98. Aitken, E. A.; Evans, S. K.; Adamson, M. G.
56. Speight, M. V. J. Nucl. Mater. 1964, 13, 207–209. A Thermodynamic Data Program Involving Plutonia and
57. Biebsack, J.; Diez, W. Phys. Stat. Sol. 1968, 27, 139–144. Urania at High Temperatures; Technical Report, 1971.
58. Oldfield, W.; Markworth, A. J. Mater. Sci. Eng. 1969, 4, 99. Bober, M.; Schumacher, G.; Geithoff, D. J. Nucl. Mater.
353–366. 1973, 47, 187–197.
59. Greenwood, G. W.; Speight, M. V. J. Nucl. Mater. 1963, 10, 100. Evans, S. K.; Aitken, E. A.; Craig, C. N. J. Nucl. Mater.
140–144. 1967, 30, 57–61.
60. Nichols, F. A. J. Nucl. Mater. 1968, 27, 137–146. 101. Fryxell, R. E.; Aitken, E. A. J. Nucl. Mater. 1969, 30,
61. Sens, P. F. J. Nucl. Mater. 1972, 43, 293–307. 50–56.
62. Gruber, E. E. J. Appl. Phys. 1967, 38, 243–250. 102. Perron, P. O. J. Nucl. Mater. 1968, 27, 237–238.
63. Nichols, F. A. J. Nucl. Mater. 1967, 22, 214–222. 103. Olander, D. R. J. Nucl. Mater. 1971, 44, 116.
64. Barnes, R. S.; Mazey, D. J. Proc. R. Soc. Lond. A 1963, 104. Lewis, B. J.; Thompson, W. T.; Akbari, F.; Thompson, D. M.;
275, 47–57. Thurgood, C.; Higgs, J. J. Nucl. Mater. 2004, 328,
65. Hu, S. Y.; Henager, C. H., Jr. Acta Mater. 2010, 58, 180–196.
3230–3237. 105. Higgs, J. D.; Lewis, B. J.; Thompson, W. T.; He, Z.
66. Kelly, R. Phys. Stat. Sol. 1967, 21, 451. J. Nucl. Mater. 2007, 366, 99–128.
67. Bourgeois, L.; Dehaudt, Ph.; Lemaignan, C.; Fredric, J. P. 106. Shaheen, K.; Lewis, B. J. Platform-based modelling of
J. Nucl. Mater. 2001, 295, 73–82. intact and defective fuel behaviour. In Proceedings of
68. Li, Y.; Hu, S.; Sun, X.; Gao, F.; Henager, C. H., Jr.; IAEA Symposium, Vienna, SFEN; 2009; pp 654–667.
Khaleed, M. J. Nucl. Mater. 2010, 407, 119–125. 107. Sari, C.; Schumacher, G. J. Nucl. Mater. 1976, 61,
69. Matzke, Hj. J. Nucl. Mater. 1975, 57, 180–186. 192–202.
70. Herring, C. Structure and Properties of Solid Surfaces; 108. Adamson, M. G.; Aitken, E. A.; Evans, S. K.; Davies, J. H.
Gomer, R., Smith, C. S., Eds.; University of Chicago Thermodynamics of nuclear materials. In Proceedings of.
Press: Chicago, 1953; p 5. IAEA Symposium; IAEA: Vienna; 1975; p 59.
71. Cahn, J. W. Acta Met. 1960, 8, 554–562. 109. Bober, M.; Schumacher, G. J. Nucl. Mater. 1973, 49,
72. Wernick, J. H. AIME Trans. 1957, 209, 1169. 322–324.
73. Cornell, R. M.; Williamson, G. K. J. Nucl. Mater. 1965, 17, 110. Lassman, K. J. Nucl. Mater. 1987, 150, 10–16.
200. 111. Ruello, P.; Petot-Ervas, G.; Petot, C.; Desgranges, L.
74. Michels, L. C.; Poeppel, R. B.; Niemark, L. A. Trans. Am. J. Am. Ceram. Soc. 2005, 88, 604–611.
Nucl. Soc. 1970, 13, 601. 112. Kamata, M.; Esaka, T. J. Appl. Electrochem. 1994,
75. Ronchi, C.; Sari, C. J. Nucl. Mater. 1974, 50, 91–97. 24, 390.
76. Lewis, W. B.; MacEwan, J. R.; Stevens, W. H.; Hart, R. G. 113. Millot, F.; Gerdanian, P. J. Nucl. Mater. 1980, 92,
Technical Report AECL-2019; Chalk River Nuclear 257–261.
Laboratories: Canada, 1964. 114. Beisswenger, H.; Bober, M.; Shumacher, G. J. Nucl.
77. Hales, D. R.; Horn, G. R. J. Nucl. Mater. 1963, 8, Mater. 1967, 21, 38–52.
207–220. 115. Tanaka, K.; Miwa, S.; Sato, I.; et al. J. Nucl. Mater. 2009,
78. Clement, C. F. J. Nucl. Mater. 1977, 68, 63–68. 385, 407–412.
79. Meyer, R. O. J. Nucl. Mater. 1974, 52, 326–328. 116. Benedict, U.; Saunders, R. A.; Schmidt, H. E.;
80. Guarro, S.; Olander, D. R. J. Nucl. Mater. 1975, 57, Whitlow, W. H. Programme progress report: Plutonium
136–144. fuels and actinide research; Technical Report 3438;
81. Olander, D. R. J. Nucl. Mater. 1973, 47, 251–254. Joint Research Centre, European Institute for
82. Olander, D. R. J. Nucl. Mater. 1973/74, 49, 21–34. Transuranium Elements, 1977.
83. Olander, D. R. J. Nucl. Mater. 1974, 52, 329–330. 117. Gulden, T. D. J. Am. Ceram. Soc. 1972, 55, 14–18.
84. Bramman, J. I.; Powell, H. J. J. Br. Nucl. Energy Soc. 118. Wallace, T. C.; Witteman, W. G.; Radosevich, C. L.;
1975, 14, 63. Bowman, M. G. Carbon Diffusion in the Carbides of
676 Matter Transport in Fast Reactor Fuels
Uranium; Technical Report LA-DC-8840; Los Alamos 125. Hofman, G. L.; Pahl, R. G.; Lahm, C. E.; Porter, D. L.
National Laboratory: Los Alamos, NM, 1968. Metal. Trans. A 1990, 21, 517.
119. Weeks, R. W.; Scattergood, R. O.; Pati, S. R. J. Nucl. 126. Kim, Y. S.; Hofman, G. L.; Yacout, A. M. J. Nucl. Mater.
Mater. 1970, 36, 223–229. 2009, 392, 164–170.
120. Storms, E. K. The Refractory Carbides; Academic Press: 127. Powers, J. J.; Wirth, B. D. J. Nucl. Mater. 2010, 405,
New York, 1967. 74–82.
121. Mardon, P. G.; Potter, P. E. Nucl. Met. 1970, 17, 842. 128. Maki, J. T.; Petti, D. A.; Knudson, D. L.; Miller, G. K.
122. Richter, K.; Coquerelle, M.; Gabolde, J.; Werner, P. J. Nucl. Mater. 2007, 371, 270–280.
Fuel and Fuel Elements for Fast Reactors; IAEA: Vienna, 129. Smith, C. L. Nucl. Tech. 1977, 35, 403–412.
1974; Vol. 1, p 71. 130. Lee, J. K.; Choi, Y. J. Nucl. Mater. 2006, 357, 213–220.
123. Kim, Y. S.; Hofman, G. L.; Hayes, S. L.; Sohn, Y. H. 131. Wagner-Löffler, M. Nucl. Tech. 1977, 35, 392–402.
J. Nucl. Mater. 2004, 327, 27–36. 132. Bullock, R. E.; Kaae, J. L. J. Nucl. Mater. 1983, 115,
124. Porter, D. L.; Lahm, C. E.; Pahl, R. G. Metal. Trans. A 69–83.
1990, 21A, 1871.
3.22 Modeling of Pellet Cladding Interaction
B. Michel, J. Sercombe, and C. Nonon
Commissariat à l’Energie Atomique, DEN, DEC, St Paul Lez Durance, France
O. Fandeur
Commissariat à l’Energie Atomique, DEN, DM2S, Gif-sur-Yvette, France
677
678 Modeling of Pellet Cladding Interaction
I-SCC Iodine-stress corrosion cracking of the advances in PCI modeling to date and is
MP Mid-pellet described in detail.
PCI Pellet–cladding interaction In the last part, the contribution of modeling to a
PIE Postirradiation examinations better understanding of PCI is detailed with simulation
PLEIADES Unified fuel performance software results. Among these results, the assessments of the
environment cladding residual displacements and of the cladding
PWR Pressurized water reactor loading are discussed through the multidimensional
SCC Stress corrosion cracking modeling. Finally, a PCI failure risk criterion based
SEM Scanning Electron Microscopy on a theoretical approach is proposed, its validation
SIMS Secondary Ion Mass Spectrometry and limits are discussed. In this approach, mechanical
aspects are considered to have a first-order effect on
the cladding failure compared to the chemical aspects
of stress corrosion cracking (SCC) damage process.
3.22.1 Introduction
200 μm
(a) (b)
Figure 3 Stress corrosion cracking location after a pellet–cladding interaction rupture in power ramp test. (a) Axial crack
observed on the external surface of the cladding. (b) Through wall crack observed in the optical microscope on a cross-
section at the interpellet plane.
microscopy, and SEM examinations.14 The database to completely fill the dishing of a UO2 fuel (see
examination analysis gives valuable information on Figure 5). But the kinetic and amplitude of all
the kinetics of the numerous mechanisms involved in these mechanisms are the function of power level,
the PCI process (see Figure 5): power increase, fuel mechanical properties, and
fuel rod burnup. Ramp test program with exactly
Fuel rod profilometry (diameter increase, ridging
the same ramp test sequence (ramp rate, maximum
evolution at interpellet (IP) and MP plane in func-
power level, and holding time),20 or ‘separate effect
tion of local power) – residual cladding strain, fission
experiments’ with specific instrumentation, such as
gas swelling, fuel creep
in-pile diameter measurements,26 on different types
Optical microscopy on cross-section and longitu-
of fresh or preirradiated fuel rods, is the most effi-
dinal section – pellet fracturing (number and
cient way to compare the contribution of mechan-
extension of each type of cracks), dishings filling,
isms of interest.
fission gas behavior and microstructure evolution,
Information from ramp test database analysis is
cladding damage shape and location, bonding con-
very useful to evaluate simulation capacity to quan-
ditions between fuel and cladding, etc.
tify the nonlinear model interactions which compete
SEM, EPMA, SIMS – fuel microstructure evolution,
during the power transient. In return, modeling tools
fission product distribution, inner-cladding surface
contribute to a better understanding of PCI, since they
observation25
help to separate the effect of each phenomenon or
Fuel fracturing at the periphery of the pellet occurs parameter on the fuel rod behavior under transient
during the short period of the transient phase, conditions, as discussed in Section 3.22.5. For a good
whereas most of the contribution of fuel creep and simulation capacity, the interaction between oxide and
gaseous swelling takes place during the first minutes cladding requires a computation in the two materials
of holding time at the ramp terminal level. A few with similar accuracy. While cladding mechanical
minutes at a power of about 45 kW m1 is sufficient properties can be easily achieved on irradiated
IP 6/7
1 mm
8 um
Géné 1
Géné 2
Géné 3
Géné 4
Géné 5
Géné 6
Géné 7
Géné 8
10 μm
materials,27 mechanical testings, such as compression effects. Neutronic and thermohydraulic problems are
tests24 or bending tests,28 are mainly performed on fresh strongly linked to computations at the scale of fuel
fuel. So, attention must be paid to conclusion deduced assembly or reactor core. Thermal and mechanical
from simulation, in function of the physical properties computations are more relevant to the fuel element
of the materials taken into account in the modeling. In scale commonly used for integrity assessment of
the same way, progress is still needed for developing structures, when physicochemical computation is
mechanistic models to predict cladding cracking by mainly linked to mechanisms at an atomistic scale
I-SCC. A critical point is the procedure used for SCC (densification and solid swelling processes, and gas-
tests, which must reproduce the loading mode of the eous fission products behavior).
cladding during power ramps (strain-controlled loading
and local mechanical loading).29 3.22.4.2 Nuclear Power Deposition
Computation
3.22.4 Numerical Simulation The power density distribution through the pellet is
computed with a simplified approach given by eqn [1].
3.22.4.1 Multiphysics Problem In this equation, the local power is a function of
The main phenomena involved in fuel behavior mod- the neutron flux FjFjðx; y; zÞ
eling under irradiation are detailed in Figure 6. These the mean neutron cross-section s pf
phenomena can be separated in two: on one side, the concentration of fissile atoms Cpf ðx; y; zÞ
neutronic and thermohydraulic problem, and on the (subscript pf indicates each type of fissile atom)
other side, a nonlinear multiphysics problem for fuel and the energy due to the fission of fissile atoms
element behavior under irradiation. The first problem Epf
is devoted to compute nuclear power deposition in the X
fuel and power evacuation from the fuel element pVlocal ¼ FjFjðx; y; zÞ pf Cpf ðx; y; zÞEpf
s ½1
pf
toward the energy conversion system. The second
problem is devoted to the computation of the temper- Input data used in eqn [1] to compute the mean
ature distribution in the fuel element, mechanical cross-section for one group of energy and the spatial
fields, and structural material changes due to irradiation distribution of fissile atoms concentration are taken
Neutronic computation
Time t
Thermal computation
Convergence test
Thermo-hydraulic computation
Heat transport to the energy
conversion system
Mechanical computation
Time
Physicochemical computation
Time t + Δt
The cracking strain rate, derived from eqn [9], can be 3.22.4.5.2 Cladding
written as follows: The nonlinear behavior of the cladding under irradi-
ni : ds ation is taken into account by a nonunified formula-
ni : defis ¼ ½10 tion coupling a creep law [15] and a plasticity model
ni : @@eR : ni
fis like the example presented in eqn [16].
The viscoplastic strain rate is derived from eqns [5]
dev
and [11] according to an associated flow rule: ÞÞnp eQ p =RT eBt
¼ Að J 2 ð s
dt
Þ
@J2 ðs þ C ðJ2 ðs ÞÞns e Qs =RT Fp ½15
decreep ¼ decreep ½11
@s
Þ as the second invariant of the deviatoric 2
with J2 ðs XÞ R0 0
J2 ðs X ¼ hep
and ½16
stress tensor and f ðJ2 ðsÞÞ an incompressible creep 3
function. In eqn [15], dev =dt is the creep strain rate, F the fast
In these equations, compressible creep due to neutron flux, ðA; np ; Qp ; B; C; ns ; Qs ; pÞ are material
material porosity evolutions under hydrostatic loading parameters, and R is the universal gas constant. In
conditions is not taken into account. Equation [11] can, eqn [16], X is the internal stress associated with
however, be easily replaced by a compressible creep kinematics hardening, R0 the yield stress, h the hard-
correlation, as proposed in Lucuta et al.31 In the latter, ening modulus, and ep the plastic strain.
Modeling of Pellet Cladding Interaction 685
Creep rate enhancement due to the effect of the strain rate is computed with the unit tensor and a
fast neutron flux F (E > 1 MeV) on the material is function of the pellet burnup. Solid swelling is
taken into account in the second term of eqn [15], linked to solid fission products, which tend to mod-
which models stationary creep. The irradiation- ify the crystallographic dimensions of uranium
induced hardening of the material is introduced in dioxide.40 Densification process is the result of the
the first term of eqn [15], which models primary creep annihilation of small porosities caused by the acti-
and by the dependency of the material parameters on vation energy of fission peaks.41
neutron fluence. To account for the material behavior e_
on the whole loading range, two different sets of SS D
¼ f ðBU ; . . .ÞI ½17
material parameters are used for creep at low stress
level (base irradiation) and creep at high stress level
3.22.4.6.2 Behavior of gaseous fission
(ramp test). The anisotropic behavior of the cladding,
products
particularly important for some alloys, can also be
The behavior of gaseous fission products is modeled
taken into account by using Hill’s equivalent stress27
by different approaches:
instead of the Von Mises equivalent stress (J2) in eqns
[15] and [16]. a semiempirical model for the simulation of inte-
gral tests,42
3.22.4.5.3 Pellet–cladding interface advanced models for detailed interpretation of
Friction is of primary importance with respect to laboratory results.41,43
stress or strain concentration in the cladding2,9,18
The multiphysics coupling developed in this chapter is
and fuel cracking.33,34 Measures of the friction coeffi-
based on the semiempirical model42 first developed for
cient between nonirradiated fuel materials and Zirca-
the one-dimensional approach (see Section 3.22.4.8).
loy materials are usually within 0.4–0.7,35,36 and
In this section, we will detail only for low burnup
almost independent of the contact pressure, temper-
constitutive equations needed to compute fission gas
ature, and oxide thickness. Some coatings have been
amount, which induces swelling and coupling equa-
developed in order to reduce friction at the pellet–
tions with the mechanical state. In the semiempirical
cladding interface.34 Irradiation can enhance fric-
approach, two sets of constitutive equations are used
tion37 and even lead to chemical bonding between
for nominal irradiation conditions and transient state
the pellet and the cladding38 at high burnup. Fuel rod
irradiation conditions, respectively.
codes usually include unilateral contact conditions
and axial adherence in part of the fuel rod as soon as
the pellet–cladding gap is closed. Chemical bonding 3.22.4.6.2.1 Steady-state irradiation conditions
has also been introduced for some specific applica- Under steady-state (ST) loading conditions, gaseous
tions.39 In 2D and 3D descriptions of the fuel pellet– swelling (GS) is derived from the empirical expres-
cladding system,10,12 unilateral contact is assessed by sion given by eqn [18], with k an adjustable parame-
the Lagrangian multiplier method of the CAST3M ter, r the density (kgU m3), and BU the burnup
Finite Element code, and a ‘Coulomb’ model is intro- (MWd kgU1). Due to the low stress level in the
duced to take into account friction-slip or adherence, pellet under nominal irradiation, there is no coupling
according to the following equation: effect with the mechanical state needed in eqn [18].
At the same time, other constitutive equations, not
0 if kFt k mkFn k
vt ¼ detailed here, are used to compute the quantity of
cFt if kFt k ¼ mkFn k
fission gas produced and its location (intragranular,
with vt the slip rate, Fn and Ft the normal and tangen- intergranular, or released).
tial forces, m the friction coefficient, and c a positive
number. tr eST
GS ¼ f ðT Þð1 e
krBU
Þ ½18
Balance equation for fission gas product transfer described in Figure 7. Equations [19]–[23] are used
to compute the fission gas rate through a conservation
dGbia principle between the amount of gas Cres available in
¼ f1 ðCbia ; r bia ; Cres Þ
dt the UO2 atomic network and the amount Crel released
Gbia in the rod plenum. Functions f1 and f2 (eqns [19] and
þ f2 ðCbia ; r bia ; Cres Þ ½19
t [20]) compute the amount of gas going from the UO2
atomic network to intragranular bubbles, respectively,
dCres through two different mechanisms: gas diffusion in
¼ f1 ðCbia ; r bia ; Cres Þ
dt UO2 and gas captured by moving intragranular bub-
f2 ðCbia ; r bia ; Cres Þ þ FFG ½20 bles. The variable t in eqns [19] and [26], equals to the
mean time needed for intragranular bubbles to move at
grain boundary, is used to compute the gas flow rate
dGbie dCres dGbia s
due to intragranular bubbles transport in eqn [19],
¼ þ FFG fth ðCbie ; r bie ÞDpor kT and to compute the bubbles concentration rate leav-
dt 0dt dt 1 ing the grain in eqn [26]. Fission gas production
B Gbie Gpor C during the transient stage is added as a source in
B C ½21
@ S g 4 fpor A eqns [20] and [21], which are, respectively, used for
Cbie
s
paðyÞr bie
3
the balance of gas in the UO2 atomic network and in
Vg 3
intergranular bubbles. A Heaviside function fth is used
dGpor in eqns [21]–[23] to avoid gas release to porosity and
s
¼ fth ðCbie ; r bie ÞDpor kT plenum when the surface concentration Cbie s
and the
dt 0 1 radius of intergranular bubbles are lower than thresh-
B old values. The gas flow rates between intergranular
B Gbie Gpor CC dGrel
@ A ½22 bubbles, manufacturing porosity, and rod plenum, are
S g 4 f dt
Cbie
s
paðyÞr 3bie por computed through a diffusion principle with two
Vg 3 parameters called Dpor and Drel, respectively, for the
porosity and the plenum. Gas state eqns [24] and [25]
dGrel s Gpor
¼ fth ðCbie ; r bie ÞDrel ðpH Þ pint ½23 are derived from the ideal gas law, and the gas tension
dt fpor surface effect is not neglected due to the small values
of bubble radius. In eqn [25], two shape factors, y and
Gas state equations a, are introduced to account for the lenticular geom-
2g 4 3 Gbia etry of intergranular bubbles. Function f3 in eqn [26]
phia þ pr ¼ kT ½24 compute the decrease in the intragranular bubble
r bia 3 bia Cbia
concentration rate due to coalescence, which is seen
2g siny 4 Gbie as a consequence of a nonuniform displacement of
phie þ paðyÞr 3bie ¼ kT ½25
r bie 3 Sg intragranular bubbles. The coupling effect with me-
Cbie
s
chanical hydrostatic pressure pH [29] is taken into
Vg
account through eqns [27] and [28], where the main
Constitutive equations for pressurized cavities’ strain mechanism of the bubble is assumed to be a
behavior in solid medium vacancy flux due to self UO2 diffusion ratio Du in
eqn [27] and surface grain boundary diffusion ratio
dCbia Cbia Dgb in eqn [28].
¼ f3 ðCbia ; r bia Þ ½26
dt t
1
pH ¼ tr½s ½29
dr bia Du O 3
¼ ðphia pH Þ ½27
dt kTr bia For each time step of the multiphysics algorithm,
this differential equation system is solved in order
dr bie Dgb O to compute the gaseous swelling rate as the sum of
¼ ðphie pH Þ ½28
dt kTr bie intragranular swelling and intergranular swelling:
4 s Sg 4
The unknown functions of the differential equation tr½eGS ¼ Cbia par 3bia þ Cbie par 3bie ½30
3 Vg 3
systems [19]–[28] are the ten physical variables
Modeling of Pellet Cladding Interaction 687
Porosity
Gpor
Intergranular bubble
Gbie
Intragranular bubble
Gbia
Cres
Gbia : Gas concentration per unit volume of fuel (mol m–3) located in intragranular bubbles
Cbia : Number of intragranular bubbles per unit volume of fuel
Cres : Gas concentration per unit volume of fuel (mol m–3) located in UO2 atomic network
rbia : Radius of intragranular bubbles (m)
Gbie : Gas concentration per unit volume of fuel (mol m–3) located in intergranular bubbles
rbie : Radius of intergranular bubbles (m)
Gpor : Gas concentration per unit volume of fuel (mol m–3) located in manufacturing porosity
Grel : Gas concentration per unit volume of fuel (mol m–3) released in the rod plenum
phia : Normal stress at the interface between gas and solid for intragranular bubbles
phie : Normal stress at the interface between gas and solid for intergranular bubbles
3.22.4.7.1 Approach
The meaning of the symbols is explained in Table 1.
The coupling problem between mechanical behavior
From a practical point of view, the time integra-
and gaseous fission products, behavior at time t can be
tion of the multiphysics problem is derived from an
decomposed as following:
incremental formulation, between times t and t þ Dt,
Mechanical equilibrium equations along irradiation loading time history.
8 Due to the fact that the constitutive gaseous swell-
< div s ¼ 0
> on O
ing equations fGS ðsÞ cannot be easily introduced
~
s n¼’ ~BC on @O’ ½31 in eqn [32], an external iteration process has to be
>
:
~
u ¼~ uBC on @Ou achieved between the mechanical problem [31] and
688 Modeling of Pellet Cladding Interaction
Linear power
→
U
Pmax
Thermal solver
T
T, Δegs Pcond
Convergence test
Mechanical solver
→
U, s t t + dt
Time
T, s
Time period
egs
→
t + dt ⎯
⎯→ T, U, s, Δegs
Figure 8 Numerical integration scheme for the multiphysics problem: internal state variables used for the coupling
algorithm.
Modeling of Pellet Cladding Interaction 689
Error estimator: linearization of iterative solutions concept fuels behavior. In this environment, all the
resources of the platform are shared between several
applications, each one being dedicated to a precise
Mechanical model fuel concept. Studies on fuel rods for pressurized
Hydrostatic stress
Fracture plane
2Drq
3D
Fuel rod
2Drz Pellet Clad
Slice n + 1 r
Thermomechanical interaction
Midpellet plane of slice n on slice n + 1
Pellet Clad
Slice n r
Tcooling
deformations of a fragment divided axially in sev- Regarding mechanical loading conditions, the inter-
eral stacked pieces. The boundary conditions con- nal pressure (gas pressure) is applied on the clad-
sidered in this model account for the geometrical ding inner surface and on the pellet fragment
symmetries of the problem and for the pellet– outer surface; the external pressure (water pressure)
cladding and fragment–fragment interactions. The is applied on the cladding outer surface. Pres-
mechanical reaction of the fissile column above sure ‘end-effects’ is also applied on the cladding
and under the meshed fragment is represented mid-plane. In this single-fuel fragment model, a
with a kinematic relation between the pellet and secondary crack network occurring inside the pellet
the cladding mid-planes (i.e., the generalized plane fragment can be assessed by the continuum damage
strain for the 2D r–y representation). This mid- approach presented in Section 3.22.4.5, as we can
plane locking condition is used only when the see in Figure 11 with 2D r–y plane strain
pellet–ladding gap is closed at least at one point. computation.
Modeling of Pellet Cladding Interaction 691
2.04E – 04
Orthoradial cracks strain 1.43E – 03
2.65E – 03
3.87E – 03
5.10E – 03
6.32E – 03
7.54E – 03
8.76E – 03
9.99E – 03
1.12E – 02
1.24E – 02
Radial cracks 1.37E – 02
1.49E – 02
1.61E – 02
1.73E – 02
1.85E – 02
1.98E – 02
2.10E – 02
2.22E – 02
2.34E – 02
Secondary cracks due Initial cracks due to 2.47E – 02
to transient loading nominal loading 2.59E – 02
For the axisymetric model, crack assessment is under irradiation (see Section 3.22.4.1). Thanks to
achieved through the same continuum damage this coupling approach, according to the spatial dis-
model used for secondary cracks in the single-fuel cretization issued from the finite element method,
pellet fragment. However, the axisymmetric hypoth- the two following separate algebraic equations need
esis leads only to a homogeneous solution in the to be solved to evaluate the evolution of a fuel ele-
circumferential direction, for which secondary crack ment under irradiation.
density cannot be assessed. Thanks to the fuel crack- The thermal equation leads to the following sym-
ing model, the 2D r–z representation can be interest- bolic algebraic system to determine the temperature
ing for the assessment of primary and secondary inside the fuel element, including temporal effects
ridges after irradiation. Finally, the 1D representation (e.g., power transients):
of the whole fuel rod, which was first developed at the
beginning of fuel performance codes, is a useful sim- d
At each time t ; ½CT ðT ; Lext ...Þ fT g
ulation to assess the global behavior under irradiation dt
and compute many integral values such as fission gas þ ð½KT ðT ; Lext ...Þ þ ½HT ðT ...ÞÞfT g
release and internal pressure. ¼ fFeq ðt ; Lext Þg ½37
3.22.4.8.3 Thermomechanical finite element where {T } represents the vector of the nodal
solver of the PLEIADES platform unknown temperatures for the studied fuel element,
Due to the complexity of nonlinear equations involved [CT] is the heat capacity matrix, which can be omit-
in the thermomechanical problem of PCI modelling, a ted ([CT] ¼ 0) in some cases, [KT] is the conductance
numerical solver is needed. To solve numerically both matrix (the Fourier’s law is assumed), the matrix [HT]
the thermal and mechanical problems, whatever the represents the heat surface convection transfer, espe-
dimension is, the retained method is the well-known cially through the pellet–cladding gap, and {Feq}
and general finite element method47,48 based on the represents the prescribed thermal flux and tempera-
Galerkin method, where the unknown degrees of ture boundary conditions, and also the known inter-
freedom are the temperature T and displacement U nal heat sources. These matrices are worked out from
of each node of the fuel element mesh (as described in the material properties of each part of the fuel rod
the previous Section 3.22.4.8.2). (Section 3.22.4.4), and are function of the tempera-
The thermomechanical coupling problem is ture (T) and other known loading parameters (Lext),
solved with a weak formulation consistent with the including the irradiation effects.
nonlinear algorithm proposed for the multiphysics The (quasistatic) mechanical equilibrium is writ-
problem associated with fuel element behavior ten by the next symbolic algebraic equation whose
692 Modeling of Pellet Cladding Interaction
resolution yields the mechanical fields inside the fuel 1D, 2D, and 3D thermal and mechanical simulations
rod (such as strain and stress): can be easily performed with the same framework in
this code.
At each time t ; ½KU ðU ; T ; Lext ; Vint . . .ÞfU g
¼ Feq ðt ; U ; Lext Þ ½38
3.22.5 Modeling Contribution to a
where {U } represents the vector of the nodal
Better Understanding of PCI
unknown displacements for the studied fuel element,
the matrix [KU] is the global stiffness, and the vector 3.22.5.1 Pellet–Cladding Gap Closure
{Feq} represents the prescribed force and displace- Mechanisms and Fuel Element Behavior
ment boundary conditions. This system is strongly Under Irradiation
nonlinear when the mechanical problem includes geo-
The mechanisms leading to pellet–clad gap closure
metrical nonlinearity, such as contact with or without
and ridge formation in the cladding during base irra-
friction (the matrix [KU] and the vector {Feq} are then
diations and power ramp tests can be schematically
dependent of the unknown solution {U }), and/or
decomposed as described below.50
time history effect through the knowledge of the
internal variables (Vint) due to the irreversible
mechanical material behaviors (Section 3.22.4.5). 3.22.5.1.1 Base irradiation (see Figure 12)
The temporal resolution of both these algebraic During the first power increase, the thermal gradi-
systems is done step-by-step in a sequential way for ent associated with the fuel fragmentation is at the
each time interval [tn, tnþ1] of the given time discre- origin of the hourglass shape of the pellet. The
tization (the known initial state corresponding consequence is a reduction of the gap at the IP
to time t0). The known loading parameters (Lext) plane (see Figure 12(a)). Then, during the power
are supposed to vary linearly on the time interval hold period, the fuel element dimensions will
[tn, tnþ1]. change due to the following phenomena:
First, the thermal problem is solved to determine densification and solid swelling in the pellet
the nodal temperature {Tnþ1} at the time tnþ1, cladding creep under a compressive stress state.
whereas the mechanical state is supposed to be
The competition between these geometrical changes
fixed and given by its value {Un} at the time tn.
leads to a gap decrease with mainly two steps of
A fixed-point iterative method is used to solve the
PCI:
nonlinear system obtained thanks to the implicit
Euler backward method. low interaction stage with a gap partially closed in
Second, the mechanical equilibrium is calculated the vicinity of the IP plane,
at the time tnþ1 (finding out the nodal displace- strong interaction stage with a gap entirely closed
ments {Unþ1}), assuming that the temperature and a significant contact pressure level (see Figure
field varies linearly with the time interval from 12(b)).
{Tn} to {Tnþ1}. The algorithm to solve the non-
During the interaction period, the pellet hourglass
linear mechanical system is based on an implicit
shape is printed in the cladding because of its inelas-
modified Newton–Raphson iterative method. It is
tic strains due to material creep under external pres-
worth noticing that the nonlinear material consti-
sure loading. Moreover, the cladding diameter
tutive laws (Section 3.22.4.5) are integrated at
decrease can also tend to reduce the extent of pellet
each iteration to obtain the history-dependent
hourglass magnitude, thanks to stress relaxation due
mechanical state (such as the stresses and the
to irradiation-induced creep in the pellet fragment.
internal variables Vint).
Through this analysis, it appears that the magnitude
In the PLEIADES platform, the calculation of the of clad primary ridges at the end of the base irradia-
temperature and the mechanical state at the end of a tion is the result of the competition between cladding
given time interval according to the algorithm and pellet creep. The development of a high-burnup
described earlier is realized whatever be the model structure in the pellets with pronounced gas swelling
dimension (Section 3.22.4.8.2) with two components can smoothen the radial deformation of the cladding
based on the Cast3m finite element code,49 devel- after base irradiation, as shown in Sercombe et al.13 for
oped at the Atomic Energy Commission CEA, since BWR fuel rods.
Modeling of Pellet Cladding Interaction 693
2.00E – 03
2.67E – 03
3.33E – 03
4.00E – 03
4.67E – 03
5.33E – 03
6.00E – 03
6.67E – 03
7.33E – 03
8.00E – 03
8.67E – 03
9.33E – 03
Time 1.00E – 02
1.07E – 02
1.13E – 02
1.20E – 02
1.27E – 02
1.33E – 02
1.40E – 02
Interpellet plane
(b) Gap closure
Figure 12 Pellet cladding gap closure and cladding ridging mechanisms under base irradiation.
3.22.5.1.2 Power ramp test (see Figure 13) Cladding expansion during ramp testing is first
The behavior of fuel rods during ramp testing induced at the IP level due to the hourglass of
depends on many factors: the geometry of the fuel the pellet resulting from the thermal gradient (see
pellet (height/diameter ratio of the pellet, dish vol- Figure 13(a)), but soon it is compensated by dish
ume, chamfer dimensions, etc.), the power history filling due to creep and fission gas swelling of UO2.
(maximum power, increase of power, power rate, If the height/diameter ratio of the pellet is large
duration of holding period, etc.), the thermomecha- (>1.5), the impact of creep on the MP plane will be
nical behavior of fuel and cladding (burnup of the small. Radial expansion will, therefore, be maximum
pellet, thermal expansion of the fuel pellet, cladding at the MP level since dish filling will limit radial
creep and plasticity, fuel creep, etc.), and fission gas expansion at the IP level20 (see Figure 13(b)). This
swelling in the fuel pellet. is the reason why the MP ridges observed in the
The diameter increase of the cladding during database can reach significant values (30 mm) and
power ramp is driven by the thermal expansion of often exceed their IP counterparts by a factor 2 or 3
the pellet and by fission gas swelling if the tempera- (see Figure 14). This is not the case with pellets of
ture of the pellet is high enough. The contribution of smaller height/diameter ratio (<1), as was shown with
gas swelling can be important particularly if the ALCYONE 3D in Sercombe et al.,13 since in this case,
holding period is long (>15–30 min) or if the fuel dish filling has consequences on the deformation of
rod has a high burnup. the MP plane due to axial creep. Hourglass-induced
694 Modeling of Pellet Cladding Interaction
Equivalent
inelastic strain
0.0
4.76E – 03
5.52E – 03
1.43E – 02
Power 1.90E – 02
2.30E – 02
3.66E – 02
Pmax 3.33E – 02
3.01E – 02
4.29E – 02
4.76E – 02
3.24E – 02
3.71E – 02
6.19E – 02
Pcond 6.67E – 02
7.14E – 02
7.62E – 02
8.10E – 02
8.57E – 02
8.05E – 02
8.52E – 02
0.10
Time
Time
Figure 13 Pellet viscoplasticity and cladding ridging mechanisms under the power ramp test.
strains, therefore, remain predominant in this config- considered in Figure 15 (BU < MWd kgU1).
uration, leading mainly to IP ridges. A threshold temperature can also be defined for com-
The kinetics of MP ridge development strongly plete filling up of the dishing (>70%) in the 3D simu-
depend on the maximum power and temperature lations. It is close to 1700 C and hence of the same
reached during ramp testing, since radial expansion order than the 1800 C for fission gas swelling activa-
of the pellet and dish filling depend on these two tion. Simulation results indicate furthermore that
parameters. Figure 15 gives the MP diameter 100% dish filling is usually reached within a few min-
increase of the pellet, as a function of the maximum utes during the holding period if power is sufficient.
pellet temperature during the transient part of the This interpretation of experimental results has
ramp test. In this figure respective contributions due been derived mainly from the 3D model of a single
to the total swelling and due to fission gas swelling pellet fragment presented in Section 3.22.4.8.2.
only are plotted. Also plotted is the calculated dish However, it was shown that the 2D r–z axisymetric
filling in function of the maximum temperature. model can also describe correctly the pellet ridging-
As can be seen, when the pellet maximum temper- hourglassing without the need for an empirical
ature is greater than 1800 C, fission gas swelling effect approach.51 Based on 3D simulations, a simplified
is significant at the fuel element scale. The contribu- pellet hourglassing model52 with partial reversibility
tion of gas swelling to the total diameter increase of the in case of power drop has also been developed for the
pellet during the transient stage, however, remains generalized plane strain approach of the 1D axisym-
small and does not exceed 25% for the fuel rods metric representation or the 2D r–y representation.
Modeling of Pellet Cladding Interaction 695
9.54
After ramp test
9.52
9.5
External diameter (mm)
9.48
Interpellet level Mid-pellet level
9.46
9.44
Total swelling
200 100
Fission gas swelling
Mid-pellet diameter increase during ramp
160 80
140 70
100 50
80 40
60 30
40 20
20 10
0 0
1500 1700 1900 2100 2300 2500
Maximum temperature in the fuel pellet (°C)
Figure 15 Calculated mid-pellet diameter increase due to total swelling and fission gas swelling only, dish filling in function
of the maximum temperature in the pellet.
Thanks to this simplified model, the pellet tempera- and power ramp tests, estimations by optical micros-
ture can be assessed with reasonable accuracy in the copy of dish filling of the pellet and of the number of
generalized plane strain approach. radial cracks after ramp tests (see Section 3.22.3).
The data from cladding outer profilometry are inter-
esting because they give quantitative elements on the
3.22.5.2 Analysis of Residual Displacement
pellet–cladding mechanical interaction, which can
After Irradiation
lead to a significant amount of residual strains in
PIE include measurements of the residual diameters the cladding. A comparison between computation
and height of IP and MP ridges after base irradiation and experimental measurements of the cladding
696 Modeling of Pellet Cladding Interaction
outer diameter before and after the power ramp test The behavior of the fuel pellet at the IP level is
is presented in Figure 16. As shown by Sercombe rather complex, as it results from a compromise
et al.,53 the 3D simulation has a good capacity to assess between the pellet thermal and gas-induced swelling
the external cladding profilometry variation after which tends to increase the diameter, and the filling
base and ramp test irradiations (residual diameters up of the dishing due to UO2 creep at high tempera-
and height of IP and MP ridges). ture which tends to reduce the diameter expansion.
Therefore, data concerning dish filling are of imp-
ortance to check whether the creep rates of the
fuel element are correctly estimated (overestimated
<10% Experimental values
20–30%
dish filling would lead to underestimated IP ridges
90
>90% and IP diameters). Figure 17 gives an overview
80 100% of the 3D calculated dish fillings (residual) after
Calculated dish filling (%)
9.54E – 03
9.52E – 03
External diameter (m)
9.50E – 03
9.48E – 03
Pellet Pellet
9.46E – 03
9.44E – 03
9.42E – 03
0.230 0.240 0.250 0.260 0.270
Axial position (m)
Figure 17 Experimental and computed axial variations of the cladding outer diameter after irradiation.
Modeling of Pellet Cladding Interaction 697
to that of experimental measurements. In particular, in Figure 18. Due to base irradiation and the pre-
small dish fillings (<30%) obtained on ramp tests conditioning period, the pellet–cladding gap is
with no holding periods are very well estimated. closed for a significant proportion of the interface.
This kind of comparison enables us to validate the In Figure 18(a), we can observe that the contact
nonlinear models presented in Section 3.22.4 which pressure (srr) is higher than the internal pressure
compete during PCI. (8 MPa) in the area where the gap is closed. Due to
contact pressure, the mean circumferential stress
level is at a value of 10 MPa despite an internal and
3.22.5.3 Analysis of Cladding Global
external pressure difference of 5.5 MPa, which should
Loading Under a Transient Power
lead to a compressive stress level of 55 MPa in a
Ramp Test
contact-free cladding tube. The circumferential stress
Cladding loading during the power ramp test has gradient through the cladding thickness can be mainly
been analyzed through the 3D simulation results of explained with a local bending due to an ovalization
two power ramp tests in Michel et al.10 corresponding of the cladding submitted by the pellet to a non-
to fuel rods A1 and B1 in Table 2. Because of the axisymetric displacement loading during the base
different pellet type used in fuel rod B1, cladding irradiation. This ovalization tends to increase the
rupture did not occur during the power ramp test circumferential stress applied on the inner-
despite a higher power level at the end of the tran- cladding surface in front of the radial crack of the
sient stage. The study of Michel et al.10 has mainly pellet fragment (see Figure 18); however, the magni-
focused on the impact of pellet mechanical fracture tude of this local bending is not significant compared
properties on the cladding loading. The choice fuel to the geometrical changes induced by the transient
rods A1 and B1 were motivated by the fact that the stage (see next section). Cumulated creep strain from
better PCI resistance exhibited by fuel rod B1, com- the beginning of irradiation has also been plotted in
pared to fuel rod A1, seems to be linked to the pellet Figure 18, these inelastic strains occurred under a
composition according to the experimental approach low stress level, thanks to the activation energy of the
developed in Section 3.22.3. The hypothesis devel- fast neutron flux (see eqn [11]).
oped in Michel et al.10 is that the AREVA-chromia Stresses and strains computed at the end of
doped UO2 pellet used in fuel rod B1 has lower the transient stage are presented in Figure 19, as
fracture properties than the standard UO2 pellet of expected, the temperature gradient through the
fuel rod A1. This assumption was partially justified by pellet (1680 C at the center and 410 C at the periph-
destructive postirradiation examination where crack ery) leads to a volume expansion with an hourglass
density in the AREVA-chromia doped UO2 pellet shape of the pellet fragment. These geometrical
was higher than in standard UO2.11 changes induce an enhancement of the pressure
Stresses and strains computed at the end of the contact and then, a tensile circumferential stress
preconditioning period for fuel rod A1 are presented in the cladding (a mean value of syy ¼ 465 MPa).
Rodlet Pellet–cladding Burnup (Unitary Maximum power level Time at final power Cladding
data) (Unitary data) (Unitary data) rupture
Zy4, Zircaloy-4.
698 Modeling of Pellet Cladding Interaction
Figure 19 Mechanical stress and strain fields in the cladding of rodlet A1 at the end of the transient stage: (a) srr stress (Pa)
at the pellet–cladding interface, (b) circumferential stress syy (Pa) in the cladding, (c) equivalent cumulated creep strain
variation in the cladding during the power transient.
Modeling of Pellet Cladding Interaction 699
VAL – ISO VAL – ISO surface nodes submitted to a symmetry boundary con-
> 4.51E + 08 > 1.06E − 03 dition in the circumferential direction (segments [AB]
< 5.97E + 08 < 1.14E − 02
4.53E + 08 1.14E − 03
and [CD] in Figure 10) has been set to zero in eqn [39].
4.59E + 08 1.63E − 03 The tangential stress, according to the definition of eqn
4.66E + 08 2.11E − 03 [39], is a pure shearing loading of the interface
4.73E + 08 2.59E − 03
h i h i
4.80E + 08 3.08E − 03
~ ~
f ½~r f ~y
4.87E + 08 3.56E − 03
4.94E + 08 4.04E − 03 ½sn ¼ R and ½st ¼ R
½N ðx; y; zÞds ½N ðx; y; zÞds
5.00E + 08 4.53E − 03
S
S
5.07E + 08 5.01E − 03
5.14E + 08 5.49E − 03 ½39
5.21E + 08 5.98E − 03
5.28E + 08 6.46E − 03 with [ ] a symbolic definition of a nodal vector, ~f , the
5.34E + 08 6.94E − 03 nodal force at the pellet–cladding interface, ~ r ;~
y unit
5.41E + 08 7.43E − 03
5.48E + 08 7.91E − 03 radial and circumferential vectors, N(x, y, z) the finite
5.55E + 08 8.40E − 03 element interpolation functions, and S the inner-
5.62E + 08 8.88E − 03 cladding surface.
5.69E + 08 9.36E − 03
5.75E + 08 9.85E − 03
This new interpretation of the cladding loading
5.82E + 08 1.03E − 02 proposed in Michel et al.10 (see Figure 21) is consis-
5.89E + 08 1.08E − 02 tent with the crack location observed in fuel rod A1
5.96E + 08 1.13E − 02
(see typical SCC location in Figure 2). Moreover, a
(a) (b) critical value of the shearing stress at the pellet–
Figure 20 Mechanical stress and strain fields in the cladding interface could be defined according to the
cladding of rodlet B1 at the end of the transient stage: rupture/nonrupture, respectively, observed for fuel
(a) circumferential stress syy (Pa) in the cladding, rods A1 and B1 (110 MPa in fuel rod A1 vs. 50 MPa
(b) equivalent cumulated creep strain variation in the in fuel rod B1). Another interesting point in this
cladding during the power transient.
interface loading concerns the interpretation of
the maximum tangential stress location, as shown in
In Figure 19 (a), we can see a pressure contact Figure 22, the triple point is at the intersection
concentration at the IP plane (srr ¼ 160 MPa) due between two lines defining, respectively, in the inter-
to the hourglass shape and the contact discontinuity face: the maximal displacement discontinuity and the
between two pellets. Regarding stress and strain distri- maximal sliding yield stress.
bution in the cladding, the high loading level at the
triple point (see rupture location in Section 3.22.2)
3.22.5.4 Analysis of the Cladding Local
can be understood through the results of Figure 19.
Loading
However, it is not possible to define a criterion which
leads to conclude that rupture will occur in this partic- 3.22.5.4.1 Stress and strain concentration in
ular area. Moreover, the results computed for the fuel the cladding
rod B1, presented in Figure 20, led to a higher loading The load transfer between the pellet and the cladding
level of the cladding at the end of the transient stage, has been analyzed in the literature through different
compared to the fuel rod A1, although no rupture approaches (see Section 3.22.2); the results of
occurred for the fuel rod B1 (see Table 2). Roberts2 are used to outline first-order parameters
This conclusion was the incentive to analyze the to consider for stress and strain concentration assess-
cladding loading from a different point of view, with ment in the cladding. An analytical solution is pro-
the normal and the tangential stresses, called sn and posed in Roberts2 for the 2D plane strain analysis of
st, respectively, in Figure 21, in the pellet–cladding an elastic cladding element submitted to a combined
interface at the end of the transient stage. These stres- internal pressure and shear loading. According to this
ses have been derived from the nodal forces, computed analytical solution, illustrated in Figure 23, it
in the Coulomb friction model according to the finite appears that a shear stress level of 16 MPa leads to
element solver of CAST3M, with eqns [39]. In order to an enhancement from 90 to 180 MPa the hoop stress.
take into account only the tangential loading due to the These results could not be extended easily to nonlin-
friction boundary condition, the circumferential com- ear material behavior. Hence, to show the impact of
ponent of the nodal forces computed for cladding inner the cladding viscoplasticity and of PCI loading on the
700 Modeling of Pellet Cladding Interaction
local stress concentration in the cladding, 2D displacement at the pellet crack tip resulting from
generalized plane strain computations have been per- the opening of the fracture can be estimated from
formed in Michel et al.10 with the 2D r,y fragment Figure 25 to be 12–15 mm. The opening of the frac-
model of ALCYONE (see Section 3.22.4.8.2). The ture leads in turn to the punching of the cladding
coarse and fine meshes (minimum size of element element by the pellet crack tip. A description of the
around 1 mm) used in the simulations, together with resulting local stress concentration in the cladding,
the applied boundary conditions, are described in therefore, requires a sufficiently refined finite ele-
Figure 24. The coarse mesh has radial and tangential ment mesh (element size at least five times smaller
discretizations similar to that of the 3D mesh. The than the opening of the pellet crack), as shown by
fine mesh presents, on the contrary, nonconstant Figure 26 in which the evolution of hoop stress in
radial and tangential element sizes, which are the thickness of the cladding element (line A–B
strongly reduced in the vicinity of the pellet fracture in Figure 24) is plotted for the simulations of fuel
plane. This refinement is applied to assess the local rods A1 and B1 performed with the coarse and fine
stress concentration in the cladding in contact with meshes. While the coarse mesh simulations provide
the tip of the pellet fracture. The 2D r,y simulations good estimates of the mean hoop stresses in the clad-
have been undertaken for fuel rods A1 and B1, with ding, only the fine mesh calculations are able to
the material models and properties used in the 3D describe the local stress concentration occurring at
simulations presented in Section 3.22.5.3. the cladding inner surface. The former leads to an
The deformation of the pellet and the cladding underestimation of the maximum hoop stresses by
elements in the vicinity of the pellet fracture tip, as about 100 MPa.
obtained at the end of the transient stage of the power The evolution of the normal and tangential stres-
ramp test from the simulation of fuel rod A1 with the ses at the pellet–cladding interface (line A–C in
fine mesh, are presented in Figure 25. Similar results Figure 24), as obtained from the 2D r,y simulation
have been obtained for fuel rod B1. The tangential of fuel rods A1 and B1 performed with the fine mesh,
Modeling of Pellet Cladding Interaction 701
VAL – ISO VAL – ISO are presented in Figure 27. Recalling that the 2D r,y
> 0.00E + 00 >–6.36E – 09 plane strain model describes the behavior of the MP
< 8.43E – 06 < 1.14E + 08
plane and not of the IP plane where SCC failures
6.59E – 08 8.88E + 05 occur, it explains why the local maximum normal and
4.61E – 07 6.22E + 06
shear stresses (respectively 100–150 and 20–40 MPa)
8.56E – 07 1.15E + 07
1.25E – 06 1.69E + 07
from the 2D r,y simulations are much smaller than
1.65E – 06 2.22E + 07 that obtained in the 3D simulations (respectively
2.04E – 06 2.75E + 07 190–240 and 50–110 MPa, see Figure 21). The stress
2.44E – 06 3.29E + 07 concentration shown in Figure 26 can, therefore, not
2.83E – 06 3.82E + 07 be used directly to compare the SCC behavior of fuel
3.23E – 06 4.35E + 07 rods A1 and B1. Hence, 2D simulations based on the
3.62E – 06 4.89E + 07
IP 3D shear stresses have been performed and are
4.02E – 06 5.42E + 07
4.41E – 06 5.95E + 07
presented in the next section.
4.81E – 06 6.48E + 07
5.20E – 06 7.02E + 07 3.22.5.4.2 Impact of the shear stress at
5.60E – 06 7.55E + 07 pellet–cladding interface
6.00E – 06 8.08E + 07 To quantify the impact of the interfacial shear stress
6.39E – 06 8.62E + 07 on local stress in the cladding, a two-dimensional
6.79E – 06 9.15E + 07
plane strain model of the cladding has been used
7.18E – 06 9.68E + 07
7.58E – 06 1.02E + 08
in Michel et al.10 (see Figure 28). To take into account
7.97E – 06 1.07E + 08
PCI loading during the transient stage of the power
8.37E – 06 1.13E + 08 ramp test, the following boundary conditions are used:
(a) (b)
an internal contact pressure (segment AB
Figure 28) equal to the mean value computed in
Figure 22 Loading parameters at the end of the transient
the 3D model (70 MPa for fuel A1 and 84 MPa for
stage at the pellet–cladding interface of fuel rod A1:
(a) circumferential component of the displacement fuel rod B1),
discontinuity between pellet and cladding (m), (b) sliding an external pressure (segment DC Figure 28) of
yield stress according to the coulomb friction model (Pa). the coolant fluid (13.5 MPa),
190
180 Smith and Ranjan
170 This work
6.0
160
4.0 150
Hoop stress (Mpa)
2.0 140
130
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
Shear stress (MPa)
2.0 120
Distance from inner surface of
110
4.0 cladding (m × 103)
100 Uniform value
6.0
90
8.0 Friction ratio = 1
80
10.0 70
12.0 60
14.0 Friction ratio = 0 50
16.0 40
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
18.0
Distance from inner surface of cladding
(m × 103)
Figure 23 Circumferential stress through the cladding thickness: impact of the friction ratio. Modified from Roberts,
G. Nucl. Eng. Des. 1978, 47, 257–266.
702 Modeling of Pellet Cladding Interaction
160
etry Normal stress A1
fs ymm 140 Normal stress B1
ne o
Pla 120
Shear stress A1
Shear stress B1
Stress (MPa)
100
Fracture plane C Cladding
80
Mid-pellet plane etry
mm
f sy 60
ne o
Pla
40
Fracture plane A B 20
with smax
r y the maximal value of the shear stress com-
puted at the pellet–cladding interface in the 3D
700 Fine mesh A1
Fine mesh B1 model and x the curvilinear distance from the pellet
Coarse mesh A1 edge normalized to unity for an angular portion of
22.5 . These simplifications have been proposed
650 Coarse mesh B1
Hoop stress (MPa)
D+
Cladding + C
A+
Edge
Pellet fragment
1.20E + 02 + B
Fuel rod B1 Middle
1.00E + 02 Fuel rod A1
Shear stress σrθ (Pa)
8.00E + 01
6.00E + 01
4.00E + 01
2.00E + 01
0.00E + 00
0.00E + 00 5.00E – 04 1.00E – 03 1.50E – 03 2.00E – 03
Distance from the pellet edge (m)
of the preconditioning period. The cladding behav- stress level greater in fuel rod A1 than fuel rod B1.
ior was the same as the 3D model (Section 3.22.5.3), The stress concentration zone in the cladding, shown
and thermal loading was also extracted from 3D in Figure 29, is
20–30 mm, this length is smaller
computations. than the finite element size used in the 3D mesh
The circumferential stress (syy) distribution (50 mm). In Figure 30, maximal values around the
through the cladding thickness (segment AD in triple point computed in the 3D model have been
Figure 28), computed at the end of the transient compared to maximal values and local mean values of
stage for both global loading of fuel rods A1 and B1, a 50 mm thickness zone in the 2D model. As shown in
is plotted in Figure 29. As shown in this figure, the Figure 30, the local mean values of the 2D model are
maximal stress in the cladding is equal to 1.24s0 and consistent with the maximal values of the 3D model.
1.14s0 for fuel rods A1 and B1, respectively, which is This means that the 3D mesh refinement is not suffi-
consistent with the experimental cladding rupture cient enough to assess stress concentration in the clad-
observed in fuel rod A1 (Table 2). As expected ding of fuel rod A1, and also that the local stress value
with the analysis of the interfacial shear stress computed in the 3D model could not be used easily to
(Section 3.22.5.3), despite a lower mean circumfer- discriminate PCI behavior of fuel rods A1 and B1.
ential stress level in fuel rod A1compared to fuel rod
B1, the stress concentration leads to a local stress 3.22.5.4.3 PCI failure criterion
state consistent with experimental results. Then, it According to the results of Michel et al.,10 as summar-
was concluded that the stress concentration in the ized in Section 3.22.5.4.2, the local stress level in the
cladding is proportional to the shear stress transmit- cladding can be indirectly derived from the global
ted at the pellet–cladding interface, and that the peak loading transmitted at the pellet–cladding interface.
stress reached in the cladding depends on the contact This global loading can be characterized by two
pressure and also on the interfacial shear stress. parameters: the contact pressure and the interfacial
These 2D results have also been used to under- shear stress. The former is linked to the mean cir-
stand why the 3D computation cannot assess a local cumferential stress in the cladding (as a tube under
704 Modeling of Pellet Cladding Interaction
1.05
0.95
0.9
0.85
0.8
0.75
0.00E + 00 5.00E – 05 1.00E – 04 1.50E – 04 2.00E – 04
Distance from pellet cladding interface (m)
Figure 29 Circumferential stress through the cladding thickness (segment AD in Figure 28) at the end of the transient
stage (normalized with the cladding yield stress s0).
0.186
Maximal shear stress/s0
0.163 A
0.140
0.116
0.047
0.023
A B
0.02
0.511 0.590 0.672 0.753 0.835 0.917 0.996 1.078 1.159 1.241 1.322
Mean circumferential stress/s0
Figure 31 Iso-loading curves at the end of the transient stage: maximal value of the local stress in the cladding inner
surface versus the cladding mean circumferential stress and the interfacial shear stress (normalized with the cladding yield
stress s0).
1.30E + 00
relaxation of stress concentration is not described.
In this first step model, this simplification seems
1.20E + 00 acceptable, as long as the PCI failure curve is identi-
Circumferential stress sqq /s0
interface, divided by the surface of a potential reduce the sensitivity of the loading parameter Wry
through wall crack in the cladding as detailed in to the number of time steps used to describe the
eqn [41] transient stage.
The use of the total energy transmitted at the
R 1
ð fy1 þ fy2 ÞðDuy2 Duy1 Þ pellet–cladding interface instead of a local shear
Spellet 2 stress was motivated by several points:
Wr y ¼ ½41
Scladding an energy parameter seems more adapted to rep-
resent the continuous process damage, leading to
where fy is the circumferential component of the
SCC initiation,
tangential nodal force at the pellet–cladding inter-
the Coulomb friction model can lead to local stress
face, Duy is the circumferential component of the
relaxation (due to its unilateral formulation)
tangential displacement discontinuity at the pellet–
despite a regular convergence process, and as a
cladding interface (see in Figure 33), subscripts 1 and
result, the total shear energy at the pellet–cladding
2 refer, respectively, to the value of a variable at the
interface is less dependent from numerical param-
beginning and at the end of the transient stage, and
eter than the maximal shear stress,
Spellet and Scladding are referential surfaces shown in
the strain localization between two pellet frag-
Figure 33. This parameter can be approximately
ments is taken into account in the calculus of the
considered as an upper bound of the energy density,
shear energy.
due to shear loading at the pellet–cladding interface,
available for the cladding damage process. In this Unfortunately, this parameter has also some draw-
approach, it is assumed that the shear energy backs because it is less intrinsic (compared to the
integrated on Spellet is available for a through wall pellet fragment geometry) than a stress parameter.
crack of a surface Scladding in the cladding, which However, this limitation seems acceptable in the
means that Wry can be approximately compared to framework of phenomenological PCI failure criterion.
the cladding toughness. Moreover, in eqn [41], the The experimental PCI failure curve derived
integration of the work of the shear forces has been from the power ramp test database, according to
simplified, with the assumption that circumferential the two-parameter criterion of Section 3.22.5.5.1,
friction forces (fy) are linear functions of the tangen- is given in Figure 34. In this figure, representative
tial displacement discontinuity Duy, in order to loading points of unfailed fuel rods are used to
Mid
-pe
llet
pla
ne
Pellet
Int
uq (pellet) uq (cladding)
er
pe
Spellet
Scladding
Figure 33 Referential surfaces used to compute the circumferential shear energy at the pellet–cladding interface.
Modeling of Pellet Cladding Interaction 707
Failed Zone 1
Unfailed
1.4
H1
0.4 I1 B1
0.2
0
0.6 0.7 0.8 0.9 1 1.1 1.2
Mean circumferential stress sm/s0
Figure 34 Experimental pellet–cladding interaction failure curve with a two-parameter criterion based on the
circumferential shear energy at the pellet–cladding interface (Wry) and the mean cladding circumferential stress (sm/s0).
define a limit loading curve, which can be consid- of the transient stage (Figure 35). In the latter, crack
ered as a lower bound assessment of the PCI failure extension in fuel pellets A1 and B1 can be schemati-
curve. The shape of the limit loading curve has cally described as follows:
been defined in order to be conservative if we
Fuel rod A1: one small radial crack started at the
refer to the results presented in Section 3.22.5.4.3.
IP plane in the middle of the pellet fragment with a
In Figure 34, it is interesting to note that a two-
cracking strain level of 0.4%.
parameter criterion enables to explain the strong
Fuel rod B1: two major radial cracks, one located in
impact of the interfacial shear loading and the non-
the middle of the pellet fragment with a cracking
rupture of fuel rods I1 and B1 despite a higher
strain level of 2% between the IP plane and the
maximum power level during the ramp test (see
MP plane, and another one located at an interme-
discussion in previous sections). Then, this phenom-
diate angle, between the middle and the edge of
enological criterion leads to a unified interpreta-
the pellet fragment, with a maximum cracking
tion of PCI behavior of fuels rods separated in two
strain level of 1.8% in the vicinity of the MP plane.
zones 1 and 2 in Figure 34.
The link between the crack extension level in
3.22.5.5.2 Different types of PCI behaviors the pellet fragment and PCI loading can be illu-
The analyses of the experimental power ramp test strated with the results of Figure 36. In the latter,
database leads to the conclusion that there are two we can see that strain localization due to crack
types of PCI behavior described by zone 1 and zone opening between pellet fragments (Figure 36(a)
2 in Figure 34. According to the interpretation of and 36(c)) is not greater in the fuel rod B1, although
Michel et al.,10 these different behaviors can be char- the radius increase during the transient stage is
acterized by the ratio between the interfacial shear almost twice smaller in fuel rod A1 than in fuel rod
stress loading and the mean circumferential stress B1 (Figure 36(b) and 36(d)). This unexpected result
in the cladding. The critical power of the ramp test is due to the fact that enhanced opening of radial
has an inverse proportionality with this ratio for cracks in fuel rod B1 tends to smoothen the displace-
loading points located between zone 1 and zone 2 ment discontinuity along the pellet fragment circum-
(Figure 34). This result, discussed through the local ference. From this, it was concluded that radial cracks
stress distribution in the cladding in previous sec- induced by the transient stage inside the pellet frag-
tions, has been linked to the mechanical fracture ment led to a loss of stiffness in the outer part of
properties through the analysis of the cracking strain the fragment, which explains the threshold effect
due to radial cracks in the pellet fragment at the end observed on the interfacial shear stress of fuel rod
708 Modeling of Pellet Cladding Interaction
B1, as shown in Figure 21. The distance to the slid- has first focused on stress concentration phenom-
ing stress at the pellet–cladding interface, plotted in ena in the cladding and then on the 3D behavior of
Figure 37, confirms that for fuel rod B1, the thresh- the fuel element under irradiation to catch the
old value of the interfacial stress is smaller than loading localization mechanism at the origin of
the limit imposed by the Coulomb friction model. the initiation and propagation of SCC at the triple
For instance, we can see in Figure 37 that the point.
distribution function peak is at about 0.09s0 of In the second part, the experimental approach of
the sliding stress in fuel rod B1, whereas the distri- PCI has been presented with a synthetic vision of the
bution function of fuel rod A1 is more asymmetric knowledge coming from postirradiation analyses of
with 80% of the surface where the difference industrial fuel rods and also power ramp test results.
between the interfacial stress and the sliding stress Due to the large number of nonlinear phenomena
is greater than 0.05s0. involved in PCI behavior, it is very difficult to get a
full understanding of the mechanisms at hand from
experimental results only. A detailed modeling
3.22.6 Conclusions approach is necessary. Among the other motivations
which have led to the development of PCI modeling,
The review of the phenomenological aspects of PCI, we can list: the difficulty and the cost of instrumented
detailed in the first part, shows that the modeling PCI experiments, the need for tools to develop rep-
approach has to deal with a multiphysics problem resentative laboratory tests, and also the ambition to
involving thermal, mechanical, and chemical– use numerical simulation to reduce the development
physical behavior under irradiation. It appears and qualification periods of a product and to optimize
through the bibliographic study that PCI modeling fuel rod design.
Modeling of Pellet Cladding Interaction 709
In Section 3.22.4, the numerical simulation for the generalized plane strain approach of the 1D
approach developed for PCI modeling has been axisymmetric representation or the 2D r–y represen-
presented with: the constitutive equations of the tation. Thanks to this simplified model, the pellet
multiphysics problem, and the PLEIADES software temperature can be assessed with a reasonable amount
environment. The PLEIADES platform provides fuel of accuracy in the generalized plane strain approach.
performance codes for the study of any reactor con- Experimental validation of the residual displacement
cept. Regarding PCI modeling, a computational assessment after irradiation has been achieved. This
scheme called ALCYONE is proposed to perform validation, focused on cladding diameter and pellet
multidimensional studies using a common database dishing height, shows the simulation capacity to quan-
of physical models and material laws. For this, the tify the interactions of nonlinear models which com-
thermomechanical solver is based on the CAST3M pete during PCI. Regarding these interaction effects,
Finite Element code, and all the physicochemical improved comprehension has been obtained for the
models can be shared by all simulations whatever following items:
the dimension is.
pellet and cladding properties impact on gap clo-
The contribution of modeling to facilitate a bet-
sure and cladding ridging mechanisms under base
ter understanding of PCI has been detailed with
irradiation,
simulation results. A detailed interpretation of fuel
gaseous swelling and pellet creep properties
element behavior under irradiation has been derived
impact on the magnitude of cladding ridges after
mainly from the 3D model of a single pellet frag-
power ramp test,
ment. However, it was shown that the 2D r–z axisy-
secondary crack network assessment in the pellet
metric model can also describe correctly the pellet
fragment during power ramp tests.
ridging-hourglassing without the need for an empir-
ical approach. Based on 3D simulations, a simplified Cladding loading during the power ramp test has
pellet hourglassing model with partial reversibility been analyzed through the simulation results with
in case of power drop has also been developed the following conclusions:
710 Modeling of Pellet Cladding Interaction
(a) (b)
30%
Fuel rod A1
Fuel rod B1
25%
20%
Surface fraction
15%
10%
5%
0%
–140 –133 –126 –118 –111 –104 –97 –89 –82 –75 –68 –60 –53 –46 –39 –31 –24 –17 –10 –2 5
Figure 37 Distance to the yield sliding stress computed with st msn at the pellet–cladding interface: (a) isovalue for fuel
rod A1, (b) isovalue for fuel rod B1, (c) distribution function of the surface fraction submitted to a stress st msn.
3D local stress computation in the cladding is not Shear stress at the pellet–cladding interface is a
consistent with experimental results because the key parameter because its distribution is consistent
mesh size of the finite element is not small enough with SCC rupture location and it has a first-order
to catch the stress concentration. effect on stress concentration in the cladding.
Modeling of Pellet Cladding Interaction 711
Based on the shear stress parameter, a PCI rupture In International Seminar on PCI, Aix en Provence, France,
Mar 2004.
criterion has been proposed and its critical values 15. Davies, J. H.; Hoshi, E. V.; Zimmerman, D. L. J. Nucl.
have been derived from an experimental database Mater. 1999, 270, 87–95.
of power ramp tests. 16. Mogard, H.; Howl, D. A.; Grounes, M. The international
TRANS-RAMP II fuel project – A study of the effects of
These simulation tools have already been used to rapid power ramping on the PCI resistance of PWR fuel.
In ANS Topical Meeting on LWR Fuel Performance,
improve the comprehension of PCI behavior of Williamsburg, VA, 1988.
advanced microstructure fuels. The sensitivity analy- 17. Chamiot-Bisson, F.; et al. High power and high speed
sis detailed in this chapter leads to the conclusion ramps in the OSIRIS reactor. In Proceedings of the Top
Fuel Conference (poster session), Salamanca, Spain,
that lower fuel pellet rupture properties can reduce 2006.
significantly the cladding loading during the power 18. Cox, B. J. Nucl. Mater. 1990, 172, 249–292.
ramp test. 19. Bourreau, S.; Brochard, J.; Bentejac, F.; Couffin, P.;
Verdeau, C.; Lansiard, S. Ramp testing of PWR fuel and
The general conclusion of this chapter is that, multi-dimensional finite element modelling of PCMI. In
thanks to the increasing power of computers and ANS Topical Meeting on LWR Fuel Performance, Park
software improvements, multidimensional modeling City, 2000.
20. Nonon, C.; Lansiart, S.; Struzik, C.; et al. Differential PCI
of PCI is now possible and has already been validated behaviour of PWR fuel rods under transient conditions. In
with experimental results and used for industrial Proceedings of the International Meeting on LWR Fuel
studies devoted to the assessment of the impact of Performance, Orlando, FL, 2004.
21. Julien, B.; Delafoy, C.; Rebeyrolle, V.; Beguin, S.;
material properties on PCI behavior. The challenges Lansiart, S. Performance of advanced fuel product under
of the future are: to develop 3D simulation to have a PCI conditions. In ANS Topical Meeting on LWR Fuel
local stress concentration assessment at a scale of the Performance, Orlando, FL, 2004.
22. Blanpain, P.; Thibault, X.; Pages, J. P. Recent results from
micrometer, and to improve multiphysics modeling in-reactor MOX fuel – Performance in France and
to have a more detailed representation of the trans- improvement program. In Proceeding of the ANS
port of corrosion products to the cladding inner International Topical Meeting on LWR Fuel Performance,
Portland, OR, 1997.
surface. For such challenges the use of high perfor- 23. Barker, M. A.; Stephenson, K.; Weston, R. Manufacture
mance computing is absolutely necessary. and performance of homogeneous microstructure
SBR MOX fuel. In Proceedings of the 2007 International
LWR Fuel Performance Meeting, San Francisco, CA,
2007.
References 24. Nonon, C.; et al. PCI behaviour of chromium oxide doped
fuel. In International Seminar on PCI, Aix en Provence,
France, 2004.
1. Gittus, J. H. Nucl. Eng. Des. 1972, 18, 69–82. 25. Desgranges, L. Characterisation of volatile fission
2. Roberts, G. Nucl. Eng. Des. 1978, 47, 257–266. products, including iodine, after a power ramp. In
3. Ranjan, G. V.; Smith, E. Nucl. Eng. Des. 1980, 56, International Seminar on PCI, Aix en Provence, France,
263–272. 2004.
4. Jackson, P. A. Nucl. Eng. Des. 1987, 101, 225–232. 26. Caillot, L.; Nonon, C.; Basini, V. Out-of pile and in-pile
5. Nerman, H. Nucl. Eng. Des. 1980, 56, 289–292. viscoplastic behaviour of mixed oxide fuels. In
6. Yu, A.; Walker, S. P.; Fenner, R. T. Nucl. Eng. Des. 1990, International Seminar on PCI, Aix en Provence, France,
121, 53–58. 2004.
7. Walker, S. P.; Yu, A.; Fenner, R. T. Nucl. Eng. Des. 1992, 27. Soniak, A.; L’Hullier, N.; Mardon, J. P.; et al. In Zirconium
138, 403–408. in the Nuclear Industry; Thirteenth International
8. Diard, O.; Leclercq, S.; Rousselier, G.; Azzouz, F.; Symposium, ASTM STP 1423; Moan, G. D., Rudling, P.,
Cailletaud, G. Modeling of pellet–cladding interaction Eds.; ASTM International: West Conshohocken, PA, 2002;
during power ramps in pressurized water reactors. In pp 837–862.
Transactions of SMIRT 16, Washington, DC, Aug 2001. 28. Evans, A. G.; Davidge, R. W. J. Nucl. Mater. 1969, 33(3),
9. Brochard, J.; Bentejac, F.; Hourdequin, N.; et al. Modelling 249–260.
of pellet–cladding interaction in PWR fuel. In Transactions 29. Leboulch, D.; Fournier, L.; Sainte Catherine, C. Testing
of SMIRT 16, Washington, DC 2001. and modeling iodine-induced stress corrosion cracking in
10. Michel, B.; Sercombe, J.; Thouvenin, G. Nucl. Eng. Des. stress-relieved Zircaloy-4. In International Seminar on PCI,
2008, 238, 1612–1628. Aix en Provence, France, 2004.
11. Marchal, N.; Campos, C.; Garnier, C. Comput. Mater. Sci. 30. Lassmann, K.; Hohlefeld, F. Nucl. Eng. Des. 1987, 103,
2009, 45, 821–826. 215–221.
12. Bentejac, F. TOUTATIS an application of the CAST3M 31. Monerie, Y.; Gatt, J. M. Mech. Mater. 2006, 38, 608–619.
finite element code for PCI three dimensional modelling. 32. Contesti, E.; et al. Nucl. Eng. Des. 1989, 116(3), 265–280.
In International Seminar on PCI, Aix en Provence, 33. Smith, E. The fuel-cladding interfacial friction coefficient in
France, 2004. water-cooled reactor fuel rods. In Transactions of SMIRT
13. Sercombe, J. J.; Agard, M.; Struzik, C.; et al. Nucl. Eng. 5, Washington, DC, 1979.
Technol. 2009, 41(2), 187–199. 34. Wood, J. C.; Surette, B. A.; Aitchison, I. J. Nucl. Mater.
14. Mougel, C. Power ramping in the OSIRIS reactor: Data 1980, 88, 81–94.
base analysis for standard UO2 fuel with Zy-4 cladding. 35. Nakatsuka, M. J. Nucl. Mater. 1981, 96, 205–207.
712 Modeling of Pellet Cladding Interaction
36. Shchavelin, V. M.; Kostochka, A. V.; Bolobolichev, A. M.; Performance Meeting, San Francisco, CA, Sept 30–Oct 3,
Kuznetsov, A. A.; Golovnin, I. S.; Bibilashvili, Y. K. 2007; Paper 1044.
Atomnaya Energiya 1984, 56(3), 134–136 [translated]. 46. Bentejac, F. Toutatis an application of the CAST3M finite
37. Shchavelin, V. M.; Kostochka, A. V.; Bolobolichev, A. M.; element code for PCI three dimensional modelling. In
Kuznetsov, A. A.; Golovnin, I. S.; Bibilashvili, Y. K. International Seminar on PCI, Aix en Provence, France,
Atomnaya Energiya 1986, 61(3), 175–178 [translated]. Mar 2004.
38. Nogita, K.; Katsumi, U. J. Nucl. Sci. Technol. 1997, 34(7), 47. Zienkiewicz, O. C.; Taylor, R. L. The Finite Element Method
679–686. for Solid and Structural Mechanics; Elsevier: Oxford, UK,
39. Suzuki, M.; Uetsuka, H.; Saitou, H. Nucl. Eng. Des. 2004, 2005.
229, 1–14. 48. Hughes, T. J. R. The Finite Element Method: Linear Static
40. Struzik, C.; Moyne, M.; Piron, J. P. High burnup modelling and Dynamic Finite Element Analysis; Dover: New York,
of UO2 and MOX fuel with METEOR/TRANSURANUS 2000.
version 1.5. In Proceedings of ANS Conference, Portland, 49. Cast3m Finite Element Code Web Site – http://www-
OR, 1997; pp 126–132. cast3m.cea.fr
41. Boulore, A. étude et modélisation de la densification en 50. Michel, B. PCMI assessment using 3D modelling. In
pile des oxydes nucléaires UO2 et MOX, thèse de Transaction of the 18th International Conference SMIRT,
l’ENSMSE et de l’INPG, 2001. Beijing, 2005.
42. Garcia, Ph.; Struzik, C.; Agard, M. The effect of fission gas 51. Helfer, T. Étude de l’impact de la fissuration des
swelling on cladding strains during power ramp tests. In combustibles nucléaires oxyde sur le comportement
Proceedings IAEA Conference on Fuel Chemistry and normal et incidentel des crayons combustible;
Pellet–Cladding Interaction Related to High Burn-up Fuel, Helfer, T., Ed.; École Centrale de Lyon, Soutenue le, Mar
Sweden, 1998; IAEA TECDOC 1179. 17, 2006.
43. Noirot, L. MARGARET: An advanced mechanistic model 52. Garcia, P.; Struzik, C.; Agard, M.; Louche, V. Nucl. Eng.
of fission gas behaviour in nuclear fuel. In Proceedings of Des. 2002, 216(1–3), 183–201.
the 2005 Water Reactor Fuel Performance Meeting, Kyoto, 53. Sercombe, J.; Michel, B.; Thouvenin, G.; et al. Multi-
Japan, Oct 2–6, 2005; Paper 1067. dimensional modeling of PCMI during base irradiation and
44. Plancq, D.; et al. A unified environment for ramp testing with ALCYONE V1.1. In Proceedings Top
multi-dimensional fuel performance modelling. In Fuel 2009, Paris, Sept 2009.
Proceedings of the International Meeting on LWR Fuel 54. Rousselier, G.; Leclercq, S.; Diard, O. Scenario for the
Performance, Orlando, FL, 2004. damage of PWR fuel cladding in situations of pellet–
45. Thouvenin, G.; et al. Multidimensional modeling of a ramp cladding interaction. In Transaction of the 17th
test with the PWR fuel performance code ALCYONE. In International Conference SMIRT, Prague, Czech Republic,
Proceedings of the 2007 International LWR Fuel Aug 17–22, 2003.
3.23 Metal Fuel Performance Modeling and Simulation
T. Ogata
Central Research Institute of Electric Power Industry, Komae, Tokyo, Japan
Yeon Soo Kim and A. M. Yacout
Argonne National Laboratory, Argonne, IL, USA
713
714 Metal Fuel Performance Modeling and Simulation
Gas plenum
Upper end plug Cladding Bond Na Fuel slug (U–Pu–Zr or U–Zr alloy rod) Bottom end plug
Figure 1 Schematic view of metal fuel pin.
In the peripheral region of the slug, where the tem- Rare-earth FPs move to the fuel slug surface
perature is relatively low, cavitation may occur and react with cladding. This reaction, called fuel–
because of anisotropic irradiation growth of the grains cladding chemical interaction (FCCI), reduces the
of the fuel alloy. Radial cracks may also take place at effective thickness of the cladding and increases the
the U–Pu–Zr fuel slug surface. Gas bubble formation, local cladding stress. In the metal fuel design, FCCI is
cavitation, and crack formation cause swelling of the one of the life-limiting factors along with the plenum
fuel slug and deteriorate its effective thermal conduc- gas pressure rise.
tivity. Fuel slug swelling narrows the gap between the The design and safety analysis of metal fuel pins
slug and cladding, and a part of the bond-sodium at require an understanding of the irradiation behavior
the gap is squeezed out to the gas plenum region. and a quantitative evaluation of its influence on the
Continuing neutron irradiation grows the gas bubbles, margin to fuel pin failure. The modeling and simu-
and enhances formation of cavities and cracks (collec- lation of the irradiation behavior provide the basis
tively called ‘pores’). Coalescence among the pores for them. The various phenomena in a metal fuel pin
may also occur, resulting in enlargement of the pores. during irradiation are, as described earlier, inter-
The fuel slug swells further and comes into contact linked directly or indirectly. The comprehensive
with the cladding at 1–2 at.% burnup. The progres- understanding of irradiation behavior needs appro-
sion of coalescence among the pores may cause for- priate modeling of the respective phenomena and
mation of the open pores that are connected to the the relationship among them. When a specific phe-
outside of the slug. Through the open pores, fission nomenon is examined in detail, a stand-alone model
gas comes outside and the bond-sodium goes into is often made, omitting the relation to the other
the slug. The released fission gas accumulates in phenomena under appropriate assumptions. In the
the gas plenum and increases its internal pressure. irradiation behavior analysis for fuel pin design,
The ingression of bond-sodium recovers the effective the phenomena having limited influence on the
thermal conductivity of the slug. The swollen fuel slug other phenomena can be treated by simplified
is restrained and compressed by the cladding, while it models.
pushes out the cladding. This action–reaction is called Comprehensive simulation of steady-state irradi-
fuel–cladding mechanical interaction (FCMI). The ation behavior of metal fuel is usually performed with
compressive stress applied to the swollen slug causes a computer code such as LIFE-METAL,2 ALFUS,3
decrease in the open pore volume, which relaxes MACSIS,4 and FEAST-METAL.5 Stand-alone mod-
FCMI. Further irradiation leads to an unignorable els have been developed for constituent migration,6
level of fuel volume increase due to accumulation of gas swelling, and cavitation7 (see Chapter 3.20,
nongaseous fission products (FPs), which is called Modeling of Fission-Gas-Induced Swelling of
‘solid fission product swelling.’ Nuclear Fuels). In this chapter, a model for constit-
The cross-section of the irradiated fuel slug usu- uent migration and computer codes LIFE-METAL
ally displays a few distinctive annular zones character- and ALFUS are described as examples of metal fuel
ized by differences in fuel microstructure, particularly modeling and simulation.
in size and number densities of pores. The annular
zone formation may lead to changes in the radial
distributions of the heat generation rate and effective 3.23.2 Models for Constituent
thermal conductivity, both of which affect the fuel slug Migration
temperature distribution.
3.23.2.1 Introduction
Thermochemical migration of U, Pu, and Zr in
their distinctive directions established by radial mul- One characteristic phenomenon of U–Zr or U–Pu–Zr
tiphase fields and the temperature gradient causes metallic fast reactor fuel during irradiation is the
their radial redistribution, which is known as ‘constit- migration of fuel constituents, which turns the uni-
uent migration’ or ‘constituent redistribution.’ Zr form as-fabricated alloy into a locally heterogeneous
tends to gather at the fuel center region where bcc fuel. This changes the properties locally and affects
solid solution (g-phase) emerges for a typical metal almost all other fuel behaviors.
fuel pin, while U moves in the opposite direction. Pu For U–Pu–Zr metallic nuclear fuels, this phenom-
does not migrate remarkably. The constituent migra- enon was first observed by Murphy et al.8 They
tion affects local solidus temperature and thermal reported that irradiated U–Pu–Zr fuel rods showed
conductivity of the slug. three concentric microstructural zones containing
716 Metal Fuel Performance Modeling and Simulation
different U and Zr concentrations in each zone, sug- concentric zones, namely, a Zr-enriched central
gesting that the change in composition was due to Zr zone, a Zr-depleted and U-enriched intermediate
migration. Later, Harbur et al.9 found that an out-of- zone, and a slightly Zr-enriched zone on the outer
pile annealing produced constituent redistribution, periphery. The annular zone structure was due not
indicating that a thermal gradient alone could only to local compositions, but also to differences in
generate a driving force sufficient for constituent porosity. Each annular zone lying in a different phase
redistribution (i.e., Soret-type migration). During has different fuel swelling, specifically different gas
the course of the integral fast reactor (IFR) fuel devel- bubble growth, which is the cause of the different
opment program, extensive irradiation tests were car- pore morphology for each zone.
ried out for U-based alloy fuels in which constituent The reduction in Zr concentration in the mid-
redistribution was characterized (see Pahl et al.10–12 radius region lowers the fuel melting point, whereas
and Chapter 3.01, Metal Fuel). the increase in Zr concentration in the fuel center
The typical as-fabricated U–Pu–Zr is shown in
Figure 3 together with constituent distributions.
Fuel alloy is a mixture of three phases in this case.
The concentration profiles are locally heterogeneous
because of the individual phases. However, in the
long range, it is marginally homogeneous without U
Pu
z (U,Pu)
0.5 mm
d-UZr2
20 mm a-U
Zr
(a)
1.0
Relative intensity
0.8
Pu
0.6 U
0.4
PU
0.2
Zr (b) DP-16 irradiated to 11% burnup
0.0
0 20 40 60 80 100 120 140 Figure 4 Cross-sectional optical photomicrograph of the
Distance (mm) postirradiation fuel and electron probe microanalysis
(b)
(EPMA) scans. The fuel elements shown here were
Figure 3 Characterization of injection casted U–15Pu– irradiated at similar temperatures in the EBR-II. (a) T179
10Zr. (a) Optical micrograph of as-cast U–15Pu–10Zr irradiated to 2% burnup and (b) DP-16 irradiated to 11%
(etched) and (b) Concentration profiles (EPMA). burnup.
Metal Fuel Performance Modeling and Simulation 717
increases fuel melting point there. This is fortuitous phase, a phase that provides a medium for species to
because the possibility of fuel melting is the highest diffuse. An example can be found in Figure 3, where
in the fuel center region. Zr migration to the fuel the a-U phase is the continuous phase. Fick’s diffu-
surface is sometimes evident by the presence of a sion is driven by the concentration gradient.
Zr-rich rind, which plays a positive role against FCCI. In a metallic fuel, Zr affinity for nitrogen and
Despite some irregularities, in general, the U pro- oxygen also drives Zr migration to fuel surface if
file is opposite to that of Zr, suggesting that U and Zr nitrogen and oxygen are present in the gap included
interdiffuse. The Pu concentration profile shows that during fabrication.
it remained virtually unchanged. Although Pu migra- The migration tends to decrease and eventually stop
tion itself is negligible, the presence of Pu in the alloy with burnup because the concentration gradient gen-
above 8 wt% enhanced the migration of other com- erates a driving force for back diffusion. The observed
ponents. As suggested by Porter et al.,12 this may be redistribution is a balance between these driving forces.
related to the increase in the gas bubble swelling rate The fuel morphology irradiated to 11 at.% shown in
in the high-Pu content fuels. The elevated tempera- Figure 4(b) shows only a slight difference from the one
ture in high-porosity fuel is also a condition favorable irradiated to 2 at.% shown in Figure 4(a). This sup-
for constituent migration. This hypothesis is sup- ports the prediction that migration quickly levels off
ported by the observation that the constituent early in life after a new distribution settles.
redistribution tends to saturate at a burnup of only
a few atomic percent, when rapid initial gas-driven
3.23.2.2 Phase Diagram
swelling results in an interconnected network of open
porosity that releases most of the fission gases subse- The ternary phase diagrams of U–Pu–Zr are avail-
quently generated and greatly slows continued fuel able.13 However, these ternary diagrams are not used
swelling. in the kinetic models; instead a pseudo-binary diagram
The underlying driving forces for constituent is used to simplify modeling with reasonable accu-
migration are as follows: racy. CALPHAD thermodynamic assessment14 of the
pseudo-binary phase diagram shown in Figure 5 was
1. temperature gradients in the fuel,
performed, based on the slightly modified results
2. solubility gradients in a phase and phase bound-
published in Kurata15 and the application of the
aries, and
Thermo-Calc software by Andersson et al.16 (see
3. additional chemical potentials applied by fuel–
Chapter 2.05, Phase Diagrams of Actinide Alloys).
cladding gap and cladding.
The complexity of the lower-Zr side in the phase
The fuel’s ever-present thermal gradients generate diagram poses a challenging obstacle in modeling in
heat fluxes. To compensate for the heat flux, one of that it is involved with the Zr-depleted fuel zone at
the fuel constituents, in this case Zr atoms, moves an intermediate radius. In particular, the existence of
in the direction opposite to that of the heat flux. the z-phase seems to cause additional difficulty in
Because the heat flux is in the radial positive direc- modeling. Unfortunately, the thermodynamic prop-
tion, the Zr atoms move to the fuel center. This is also erties of this fuel phase are not known. The phase
known as the Soret-type migration. In modeling, this diagram shown in Figure 5 is still too complex to be
effect is quantitatively measured by the phenomeno- implemented in a model. A simplified phase diagram
logical property of ‘heat of transport.’ In a typical shown in Figure 6 that still preserves the essential
fuel, the axial migration is not observed primarily properties has been used. The main features of the
because the thermal gradients in this direction are simplification are twofold. First, the phases on
much smaller (approximately two orders of magni- the U þ Pu-rich side were simplified by using the
tude smaller) than those in the radial direction. (a þ z)- and (b þ z)-phases for this region. (a þ z)
Meanwhile, the poly-phase fields associated with and (b þ z) are pseudo-single phases that possess
the U–Zr and U–Pu–Zr alloy provide a driving force characteristics of the mixture of a and z, and b and z,
for constituent migration. The solubility of the solute respectively. Second, the sloped phase boundaries
atoms changes across the phase fields, and therefore, between the large-phase fields of z þ g, z þ d þ g,
more precipitation occurs in some phases than in z þ d, and a þ z þ d are approximated by parallel
others, and within a phase field, the solubility of a lines to the x-axis.
species varies by the temperature distribution. This When applied to a fuel with different Pu content,
produces a concentration gradient in the continuous the phase diagram must be modified accordingly.
718 Metal Fuel Performance Modeling and Simulation
800
Pu0.16(U,Zr)0.84
b–U g
+g
700 b–U+
g+z
Temperature (⬚C)
b–U b–U
g+
+z z g+z
a – Zr
600 z+d
500
a-U+z+d z+d d + d – Pu
a–U+z
d – Pu + z + d + a – Zr
400
0.0 0.2 0.4 0.6 0.8
d + a – Zr
Pu0.16U0.84 Mole fraction of Zr
Figure 5 Pseudo-binary (U–Pu)–Zr phase diagram for the fixed Pu content at 19 wt%.
β+γ
3.23.2.3 Temperature Prediction
(β + ζ) + γ Temperature is one of the most important factors
that affect fuel constituent redistribution. Therefore,
(β + ζ)
γ+δ modeling constituent redistribution calls for a reli-
able prediction of fuel temperatures. Fuel tempera-
ture is affected by the thermal conductivity of the
(α + ζ) fuel, which in turn is changed by porosity evolution
δ
in the fuel matrix and sodium logging into open
(α + ζ)+ δ
pores, in addition to changes in Zr concentration.
Porosity growth in U–Pu–Zr fuel during irradia-
Figure 6 Simplified pseudo-binary phase diagram of tion differs by zone. Typically, the central region with
U–19Pu–10Zr. the g-phase has relatively small pores that are formed
by agglomeration of fine fission gas bubbles. The
In particular, the temperatures at the phase bound- intermediate-radius region has the smallest pores,
aries between g1 þ g2 and (b þ z) þ g and between which are generally spherical. The pores in the fuel
(b þ z) þ g and (a þ z) þ d must be changed. For periphery, where the a-U and d-phases are dominant,
example, if the Pu content decreases from the typical are different in morphology. These are most abun-
content of 19 wt% to 15 wt%, the phase boundary dant in fraction and are generally nonspherical
Metal Fuel Performance Modeling and Simulation 719
" #
because of swirling swelling. After fuel–cladding 1 kNa =kf pNa
contact occurs, porosity tends to saturate because fpNa ¼ 1 3 ½2
2=e þ ð3 2=eÞ kNa =kf 1 pg
of swelling of solid FPs. Therefore, beyond 1
at.% burnup, porosity distribution is considered where e is 1.72, kNa and kf are thermal conductivities
unchanged. of liquid sodium and metallic fuel, respectively,
Porosity decreases fuel thermal conductivity, and pNa and pg are fractions of porosity with and
increasing fuel temperatures. However, in a fuel pin without sodium logging, respectively.
with a liquid-sodium bonded gap, subsequent sodium The sodium fraction in the pores reaches a maxi-
logging in the open pores improves fuel thermal mum of 80% at the fuel periphery. Based on the
conductivity. The effect of porosity formation on typical fuel fabrication and operation parameters
overall fuel thermal performance is a complicated given in Table 1, fuel temperature predictions for
topic for modeling. Model predictions of evolution DP-16 pin at several burnup points are shown in
of porosity in DP-16 with burnup are shown in Figure 8. Fuel temperatures generally decrease
Figure 7. with burnup as the power decreases, including the
A model for the thermal conductivity reduction fuel surface temperature.
due to porosity and sodium logging into fuel porosity No discernable changes at the zone boundaries
is reported by Bauer and Holland.17 The reduction can be seen. Sodium logged in the periphery pores
factor for fuel thermal conductivity by pores that are increases the thermal conductivity in the outer zone
free of sodium logging, applicable to the central and
intermediate zones where no sodium logging occurs,
Table 1 Fuel fabrication and operation data for a typical
is given by fast-reactor fuel element
fp ¼ ð1 pÞ1:5e ½1 Fuel alloy nominal composition (wt%) U–19Pu–10Zr
235
U enrichment (%) 56.99
where e is 1.72 for metallic fuel and p is the porosity
Fuel slug length (mm) 342.9
fraction. Fuel slug radius (mm) 2.16
Liquid sodium, filled in the as-fabricated fuel– Fuel slug density (g cm3) 15.8
cladding gap, infiltrates into open pores in the fuel Fuel smeared density (%) 72.3
as fuel volume increases by swelling and increases Fuel-cladding gap material Liquid Na
Sodium level above fuel (mm) 6.35
the overall fuel thermal conductivity above that
Cladding thickness (mm) 0.381
of the fuel alloy, depending on the available porosity. Fuel-cladding gap width (mm) 0.762
The effect of sodium infiltration is modeled also by BOL Linear heat rate (W cm1) 420
using the Bauer–Holland model in the pores. The Subassembly coolant temperatures ( C) Inlet: 371
thermal conductivity of fuel during irradiation is Outlet: 486
Subassembly coolant flow rate (l s1) 7
expressed as follows:
0.8
10 at.% BU 750
2 at.% BU BOL
0.6 0.2 at.% BU 2 at.%
700
10 at.%
Porosity
0.4 650
T (⬚C)
600
0.2
550
0.0
0.0 0.2 0.4 0.6 0.8 1.0 500
r/R 0.0 0.2 0.4 0.6 0.8 1.0
r/R
Figure 7 Model predictions of radial porocity distribution
in DP-16 (shown in Figure 4(b)). Relative radius r/R is a Figure 8 Predictions of radial temperature profile in
radial position from the center, normalized by the fuel slug DP-16. Relative radius r/R is a radial position from the
radius R. center, normalized by the fuel slug radius R.
720 Metal Fuel Performance Modeling and Simulation
restoring the thermal conductivity of the fresh fuel Within a single-phase field, as was given by
while the dense intermediate zone with the low-Zr Hofman et al.,19 the Zr interdiffusion flux is given by
concentration also increases the thermal conductiv-
~ ~ eff @CZr Q~Zr CZr @T
ity. The thermal conductivity at the porous center JZr ¼ DZr þ ½4
zone decreases about 16%. In general, thermal con- @r RT 2 @r
ductivity gradually changes, so no abrupt changes in where D ~ Zr
eff
is the effective interdiffusion coefficient
temperature profiles occur. of Zr in U–Pu–Zr, Q~Zr the heat of transport of Zr in
U–Pu–Zr, R the gas constant, and T the temperature.
Equation [4], a pseudo-binary equation for the Zr flux,
3.23.2.4 Modeling
is convenient for use in simplifying the model.
There are models to predict the phenomenon for This is justified as long as the proper diffusion proper-
U–Zr and U–Pu–Zr alloy fuels. Ogawa and Iwai18 ~ Zr
ties, D eff
and Q~Zr , are provided. With reference to the
analyzed the U–Zr system by solving the 1D Fick’s phase diagram, the single phase occurs only at high
equation numerically. A qualitative suggestion regard- temperatures, above 650 C, and the corresponding
ing the effect of the addition of Pu in U–Zr was made. phase is the g-phase. Therefore, eqn [4] applies solely
No experimental data were utilized to verify the to the g-phase.
model predictions. Hofman et al.19 analyzed their post- Within a dual phase field (strictly speaking, these
irradiation data to develop a computer model that are pseudo-dual phase fields, such as (b þ z) þ g and
predicts the redistribution in a binary U–Zr fuel. (a þ z) þ d, the driving force for diffusion is affected
Model results were parametrically compared with by the solubility of Zr in the precipitation phases,
measured U and Zr redistribution profiles to obtain (b þ z) and (a þ z), in the present case. This changes
estimates for the necessary kinetic and/or thermoki- the Zr concentration in the continuous phase, given
netic coefficients. Ishida et al.20 extended a model by by g and d, in the present case. The terminal solubil-
Marino21 for the U–Zr binary system to the U–Pu–Zr ity of Zr in (b þ z) and (a þ z) is described by
ternary system. Assuming that Pu was equally parti-
DH s
tioned in U and Zr, they assessed a quasibinary system CZr ¼ CZr exp
0
½5
numerically. RT
A kinetics model, RECONT,6 applicable to fuel where DH s are the partial molar enthalpies of solution
constituent redistribution in U–Zr or U–Pu–Zr alloy of Zr in the fuel for each phase field. Therefore, the
fuels was developed by extending the model used for corresponding flux equation in a dual phase field is
U–Zr binary alloy fuels. This model was improved obtained by substituting eqn [5] into eqn [4] as follows:
later to extend the application range to higher
DH s þ Q~Zr @T
burnup. J~Zr ¼ D
~ Zr
eff
CZr ½6
RECONT treats U–Pu–Zr as a U–Zr pseudo- RT 2 @r
binary system by partitioning Pu in U in that Pu From eqns [4] and [6], it is noticeable that the Zr flux in
migration is relatively small. However, the presence of a single phase is caused by the concentration gradient
Pu is indirectly taken into account because it enhances and the temperature gradient. In a dual phase field,
the kinetics of uranium and zirconium migration. however, only the temperature gradient has a direct
U and Zr migrate in directions opposite to each other’s. effect, whereas the concentration gradient has an indi-
Since axial temperature gradients are negligible rect effect. Because the temperaturegradient is always
compared to radial gradients and, more importantly, negative, the sign of DH s þ Q~Zr determines the
axial constituent redistribution has not been ex- direction of the Zr flux. If it is positive, the Zr flux
perimentally observed, only radial redistribution is also becomes positive, and the direction of Zr migration
considered. The 1D continuity equation for Zr in is toward the fuel surface; alternatively, a negative Zr
cylindrical coordinates is flux would indicate Zr migration toward the fuel center.
@CZr 1 @ r J~Zr Casting eqn [3] in finite-difference form gives
¼ þs ½3
@t r @r i1; j
r i1 J~Zr
iþ1; j
r iþ1 J~Zr
i; j i; j 1
where CZr is the Zr concentration, J~Zr is the interdif- CZr ¼ CZr þ 2Dt þ Dts j ½7
ðr iþ1 Þ2 ðr i1 Þ2
fusion flux of Zr, and s is the Zr production rate by
fission. The Zr yield per U atom fission is 0.29 and where Dt is the time-step size, i the ith radial ring
that for Pu atom fission is 0.20. and j the j th time-step. Equation [7] is solved using
Metal Fuel Performance Modeling and Simulation 721
the flux equations given by eqns [4] and [6]. The to estimate these data, using the measured data for
time-step size is restricted by the stability criterion T179 and DP-16. In this section, a brief summary is
given to describe the results.
ðDr Þ2 The effective interdiffusion coefficients D ~ Zr
eff
Dt < ½8 for
~ eff
2D Zr U–Pu–Zr are not available. An iterative method was
to generate a valid calculation, where Dr is the radial used to estimate them on the basis of U–Zr binary
~ Zr
spatial-step size and D eff is the effective interdiffusion alloy data available in Hofman et al.19 The interdiffu-
coefficient of Zr. To satisfy this criterion, the com- sivities of the binary system were augmented by an
puter program divides the time-step further and iter- order of magnitude. This enhancement is possibly
ates the whole calculation process internally for the due to the addition of Pu, rather than the effect of
sub-time-steps. radiation-enhanced diffusion (RED). Since tempera-
The solution procedure is as follows: tures are relatively high, thermal diffusion effects
should be dominant, where thermal annealing of de-
1. The fuel cross-section is divided into radial rings fects generated by fission is effective. Diffusivity
at intervals of Dr. change at high burnup is not included in the current
2. Time is divided while preserving the characteris- model, but will be explored in a future study.
tic time criterion Dt < Dr 2 =2D~ Zr
eff
. The enthalpies of solution of Zr in the dual phases
3. Fuel temperatures are calculated. (b þ z) þ g and (a þ z) þ g of U–Pu–Zr were
4. The phase field is determined. estimated by using the Muggianu equation for the
5. For a given time interval and for the radial ring, Gibbs energy of the ternary solution phases and data
temperature- and concentration-dependent D ~ Zr
eff
, for the excess Gibbs free energy of Zr solution in the
DH s , and Q~Zr are calculated. g-phase of the U–Pu–Zr alloy, GZr E
, given as a func-
6. J~Zr and direct production s of Zr are calculated. tion of temperature and the Zr content. These data
7. CZr is updated. were used to provide the phase diagram shown in
Figure 5. The enthalpy of solution can be calculated
The model prediction of constituent redistribution in by the relation
DP-16 is compared with measured data in Figure 9.
The model prediction has overall consistency with s ¼ GZr @GZr
E
DH E
T ½9
the measured data. @T
The kinetic parameters that must be known to Heats of transport for the phase regions are simulta-
model the phenomenon are diffusion coefficients, en- neously obtained with other kinetics parameters such
thalpies of solution of constituents in the second phase as diffusivities and enthalpies of solution. A set of new
in the multiphase fields, and heats of transport of the values have been found to provide reasonable pre-
constituents. However, these data are mostly not avail- dictions for both lower burnup tests such as T179 and
able in the literature. Only scarce data for the U–Zr for higher burnup tests such as DP-16.
binary fuel are available. A parametric study was done
reactors has been introduced only recently. Experi- In previous models,22,23 Am migration behavior
ments such as the X501 test from the Integral Fast was explained by a mechanism in which Am vapor
Reactor (IFR) program have shed some light on this moves through interconnected pores down the radial
subject.22,23 The wavelength dispersive spectroscopy temperature gradient. This mechanism reflects the
(WDS) measurements of G582 fuel element from observation that Am was located mainly where large
the EBR-II X501 test are shown in Figure 10(a). interconnected pores were available. However, it is also
The fuel was fabricated by adding 2.1 wt% Am and easier for Am atoms to precipitate in the pores than
1.3 wt% Np in U–20Pu–10Zr and was irradiated to a in the fuel matrix because Am has limited solubility
burnup of 6 at.% at the peak fuel centerline and in U–Pu–Zr alloy. Furthermore, the vapor transport
surface temperatures of 700 and 540 C, respectively. mechanism cannot explain the observation that Am
The irradiation achieved 9.1% transmutation of Am. precipitates are also found in the fuel region where
The WDS measurements show typical U, Pu, and Zr the temperature is higher than in the fuel periphery
redistribution shown in Figure 3, and Am is found region. Thus, Am interdiffusion through the fuel
mainly as precipitates in the regions where large matrix and subsequent precipitation in the pores
pores were observed. However, Np is more uniformly appears to be a more plausible mode. The migration
distributed, much like Pu. mechanism is probably complex, involving some vapor
transport and surface diffusion in interconnected pores
in addition to interdiffusion in the fuel alloy.
U Lanthanides (LA ¼ La, Ce, Pr, Nd) also accumu-
late in the fuel, with yields of 0.18 for Nd, 0.17 for Ce,
Pu 0.06 for La, and 0.05 for Pr. LA react with stainless
steel cladding, and the reaction products lower
Zr
the cladding melting point. Some of them appear to
have migration behavior similar to that of MA.
Therefore, test data with LA provide valuable insight
into prediction of the behavior of MA-bearing fuel.
In particular, considering the high cost and difficulty
Am of dealing with actinides in experiments, the use of
Np surrogate LA seems practical.
Fuel
The DP-11 fuel element from the X447 test
Cladding
(a) center contained U–10Zr and was irradiated in EBR-II to a
peak burnup of 10 at.% at the fuel centerline and
surface temperatures of 820 and 660 C, respectively.
La The fuel temperatures of the DP-11 are 100 C
higher than those of G582 shown in Figure 10(a).
The DP-11 was tested at a higher-than-typical temper-
Ce ature to obtain a margin in temperature with respect to
LA migration and reaction with cladding. The DP-11 is
Pr believed to have faster LA migration than more typical,
lower temperature tests such as G582. The reaction
between LA and the cladding observed in the DP-11
Nd is a direct effect of the high temperature.
As shown in Figure 10(b), the redistribution pro-
Sm files of LA have large spikes, implying that these
peaks are due to precipitates. As given in Table 2,
most of the lanthanides have limited solubility in
(b) Cladding ID Fuel center
U–Pu–Zr, so they precipitate in the fuel matrix.
Figure 10 Minor actinide and lanthanide redistribution. More spikes are seen at the fuel periphery region
(a) U–20.2Pu–9.1Zr–2.1Am–1.3Np (wt%) of G582 from X501 than the fuel center region, which is similar to the
test irradiated to 6 at.% burnup. (b) Lanthanide
redistribution profiles in irradiated U–10Zr of DP-11 pin in
behavior of Am in U–Pu–Zr. The spike strength of
X447 test irradiated in EBR-II to 10 at.% burnup at Sm is weaker than those of La, Ce, Pr, and Nd
cladding temperature 650 C. because of its lower concentration.
Metal Fuel Performance Modeling and Simulation 723
Pu 10 U 24 U <1
Zr <1 Zr 18 Pu 10
Np 32 Np 16 Np 0
Am 1 Am 11 Am NA
U Cm NA Pu Cm 1 Zr Cm NA
La 0 La 1 La 0
Ce 0 Ce 11 Ce 0
Pr 0 Pr 1 Pr 0
Nd 0 Nd 2 Nd 0
Plenum
region
more similar to Am and LA, Cm is expected to have a
behavior similar to those of Am and LA. Like LA, the
solubility of Am in U, Pu, or Zr is very low. Am and LA
Sodium
height
migration behavior may be generally similar. Accord-
ing to the correlation proposed by Johansson,24 Am is
close to Pr and Nd in chemical and structural proper-
Gap
ties, suggesting an analogous behavior in migration
between these elements.
region
Fuel
3.23.3 LIFE-METAL
section
Axial
3.23.3.1 Background
The LIFE-METAL fuel performance code2,25 has
been developed to predict the behavior of metallic
fuel rods in fast reactor environment as a function of
reactor operating history. The code has evolved from Gap ring
Clad
the LIFE series of codes26 that perform steady-state Fuel slug radius thickness
Wastage ring
and design-basis-transient analyses for the thermal, Mechanical ring
Thermal/
mechanical, and irradiation behavior of nuclear fuel Thermal ring
mechanical ring
rods. The original code was developed for UO2 and
mixed oxide fuels for use in fast reactor systems, q + dq/dr dr
analysis. Axial variations in operating conditions are part of the Fuel Cycle Research and Development
accounted for by using powers and fast fluxes for up (FCR&D) modeling and simulation effort. This
to nine fuel axial nodes and one plenum node. Ther- interest is a motivation for a future activity for
mally, the axial nodes are coupled through the calcu- improving the code models to accommodate possible
lated coolant temperatures. Axial heat conduction is advanced designs that include MA and possibly
ignored and there are no provisions for mechanical achieve higher burnups.
coupling between axial nodes. A detailed mechanical The following sections describe general code dif-
analysis is performed for both fuel and cladding, uti- ferences from previous versions, including a brief
lizing the generalized-plane-strain assumption for description of specific models that are associated
each axial segment and incorporating a large strain with significant metallic fuel performance character-
capability. The solution procedure involves iteration istics and their treatment in the code. Discussion of
on local total strain within each time step, and the code validation and possible future developments are
solution procedure is explicit in time. also presented.
LIFE-METAL code development has been asso-
ciated with the IFR program,29 where the code was
3.23.3.2 Code Structure and Models
the focus of the program activities related to predic-
tion of fuel-pin behavior under normal operating This section briefly describes the general code struc-
conditions. Predictions of interest about the nuclear ture and key models of interest to metallic fuel
design are changes in fuel length and fissile content behavior.
due to burnup and breeding. Thermal predictions
about fuel temperature, design margins to fuel melt- 3.23.3.2.1 Code structure and
ing, and design margins to low-melting-temperature thermomechanical analysis
alloy (e.g., U–Fe) formation are also of interest. As mentioned before, the original version of the LIFE
Mechanical predictions useful to designers are code26 was aimed at simulation of oxide fuel behav-
FCMI and FCCI, cladding deformation and design ior. Later code versions, including LIFE-METAL,
margin to significant coolant flow area reduction, and maintained the same general structure of the oxide
cladding damage and design margin to cladding fail- code with different materials properties included and
ure due to fission gas pressure loading. specific models being added or removed, according to
The code has been extensively used for planning the type of fuel. Detailed description of the code
steady state and transient experiments at EBR-II, structure is presented elsewhere.26,35
given the initial conditions for transient fuel perfor- In converting LIFE-4CN to LIFE-METAL, a
mance codes such as FPIN code30 in addition to number of fuel properties and models were changed
performance evaluation of new fuel designs and and/or added. Fuel property correlations for U–5Fs
out-of-pile experiments.31,32 For example, the code and U–xPu–yZr were first developed as continuous
was used to evaluate the expected performance of functions of temperature, porosity, alloy composition,
EBR-II Mk-V fuel design25,33 and was used to check stress, fission rate, burnup, etc. Here, Fs stands for
the performance against prespecified design criteria. fissium, a mixture of metals: 2.46Mo, 1.96Ru, 0.28Rh,
LIFE-METAL code development was frozen in 1994 0.19Pd, 0.1Zr, and 0.01Nb (in wt.%), which is the
with the termination of the IFR program. Recently, equilibrium composition of residual materials left in
Toshiba Corporation and Central Research Institute the melt-refining process (see Chapter 3.01, Metal
of Electric Power Industry (CRIEPI) has initiated the Fuel). The physical and thermal properties include
preapplication review of its Super Safe, Small, and density, phase-change temperatures, specific heat,
Simple (4S) fast reactor in the United States, through thermal expansion, thermal conductivity, and alloy
informing the United States Nuclear Regulatory and FP distributions. The mechanical properties
Commission (NRC) on the reactor’s systems and include elastic moduli, fracture strength, and thermal
fuel design. The code has been used in relation to and fission-enhanced creep.
this activity where evaluation of 4S reactor fuel Physical, thermal, mechanical, and irradiation
design34 was performed. In addition, recent develop- property correlations for test and design cladding
ments in the US fast reactors program have brought materials (e.g., austenitic 316, austenitic D9, and fer-
back interest in the code as a metallic fuels code that ritic HT9) are included in the code. Some work has
can be used for verification and validation of future been done during LIFE-METAL development to
detailed 3D finite element code to be developed as improve the correlations (e.g., D9 creep and swelling)
Metal Fuel Performance Modeling and Simulation 725
in order to get a better fit to the database within the thermal creep properties which impose a nominal
operating range of interest for metallic fuels. Wastage inner clad temperature that is <620 C. This leads
correlations for sodium–cladding interaction and to peak fuel temperatures that are far from the
time and strain-failure correlations are also included. melting temperature even in regions where Zr is
Models were developed for Ni depletion from D9 depleted, though Zr reduction lowers the margin
cladding and carbon depletion from HT9 cladding for fuel melting during a transient.
due to the FCCI. Empirical eutectic penetration A more significant effect of redistribution is its
correlations, based upon fuel behavior test apparatus impact on fuel swelling rates and axial fuel growth.
(FBTA)36 and whole-pin-furnace (WPF)32 test data, The multiple radial zones present across fuel radius
are also incorporated. as shown in Figure 4, due to the redistribution phe-
In order to extend the LIFE code to simulation of nomenon, have different growth rates, and are
metallic fuel, in addition to changes in the basic fuel believed to play a role in observed metallic fuel
properties, a number of models were removed or anisotropic growth.1 This anisotropic fuel growth, in
became no longer important in the metallic fuel ver- which axial growth rate is lower than radial growth
sion. Those phenomena include fuel restructuring and rate, was observed in U–Zr-based fuel (anisotropy
hole formation, crack formation and healing, oxygen increases with increasing Pu content as shown in
and porosity migration, hot pressing and sintering, and Figure 12) and where constituent redistribution
gap conductance. Other phenomena such as FCCI, takes place. Another indirect effect of the redistribution
fission gas driven swelling and gas release, migration phenomenon is its possible relation to the fuel–
of fuel constituents, transport of FPs, and fuel thermal cladding gap closure time, in which varying redistribu-
and irradiation creep remained important. However, tion levels and associated differences in radial
approach to each of those phenomena is different swelling can axially vary the time for gap closure.
between oxide and metallic versions of the code. Of Kim et al.38 describe associated lock-up effects and
those phenomena, the important metallic fuel (U–Zr the impact on fuel axial growth, in which the part
and U–Pu–Zr) issues encountered in previous US below the spot in which the gap closure occurs stops
designs are related to FCCI, fuel constituent redistri- to contribute to axial growth. This variation in gap
bution and swelling, and fission gas release.1,37 Those closure time is shown in Figure 13.
phenomena impact cladding degradation and fuel Axial fuel growth estimates in the original LIFE
swelling rates during operations. The following sub- code are based on the assumption of isotropic fuel
sections briefly discuss each of these issues and its growth during irradiation. LIFE-METAL adds the
implementation in the LIFE-METAL code. effects of anisotropic growth using a simple linear
growth model as follows. The axial swelling behavior
3.23.3.2.2 Constituent redistribution
Constituent redistribution in a metallic U–Pu–Zr 12
alloy fuel under fast reactors operating conditions is
a commonly observed irradiation phenomenon as
10
discussed in detail in Section 3.23.2. There are a
number of performance degrading issues associated
with this phenomenon. The reduction of Zr fraction 8
Axial growth (%)
12
Node 9
Node 8
10
Node 6 8
Case 1
Node 5
Case 2 6
Node 4
Node 3 4
Node 2
Node 1 2
0 2000 4000 6000 8000 10 000
Time (day) 0
Case 1: variable axial power shape during operation
0 2 4 6 8 10 12
Case 2: cosine shape axial power LIFE-METAL DL /L0 (%)
Figure 13 Predicted time for the fuel–cladding gap 4
closure for low-power (40 W cm1) fuel pins by the lock-up
model. Reproduced from Kim, Y. S.; Yacout, A. M.; Hofman, 3
G. L.; Ryu, H. J. Trans. Am. Nucl. Soc. 2007, 96, 709.
–4
For Bu > 0:75 at:%; DL=Lo ð%Þ 0 2 4 6 8 10 12 14 16 18 20
Peak Bu (at.%)
¼ 2:75 þ m2 ðBu 0:75Þ ½11
Figure 14 Predictions versus axial expansion (DL/Lo)
where m1 and m2 are calibration constants, and in each data for U–xPu–10Zr fuel.
case Bu is the peak burnup in the fuel element (at.%).
Figure 14 shows a comparison of LIFE-METAL pre-
dictions of fuel axial expansion and measured data for formation of compounds having relatively low melt-
U–Zr–Pu fuel. Note that the existing database for axial ing points in both fuel and cladding. During normal
fuel growth shows considerable scatter, as shown in the operations, FCCI is characterized by solid-state inter-
figure, which can be explained largely by the scatter in diffusion although liquid phase formation might be
the fuel operating temperatures. expanded during transients reaching 700–800 C.
A more rigorous treatment of the redistribution In general, FCCI in metallic fuel can be character-
phenomenon is given in Section 3.23.2, where the ized by two stages of interaction as shown in Figure 16,
time and temperature dependence of concentra- where the different possible cladding wastage mechan-
tion of redistributed elements and associated impact isms are indicated.39 Prior to accumulation of lantha-
on fuel swelling are considered. Future plans for nide FPs at the fuel–cladding interface (due to FPs
the LIFE-METAL code include implementation of migration), FCCI is characterized by a ferretic layer
those models and tying it to fuel swelling, improving formation. This layer can be a result of Ni depletion in
the mechanistic aspect of the code. austenitic cladding or decarburization of the martensi-
tic cladding. This type of interaction has solid-state
3.23.3.2.3 Fuel–cladding chemical interaction diffusion type time and temperature dependence.
An example of FCCI, as shown in Figure 15, where Ultimately, rare-earth FPs (lanthanides) control
interdiffusion between metallic U–10Zr fuel and FCCI as they migrate to the fuel–cladding interface
cladding components takes place, represents a twofold in amounts that are functions of both burnup and
potential problem for metallic fuel. It both weakens extent of radial migration in fuel (see Section 3.23.2
the cladding mechanical properties and leads to the for more details on lanthanides migration in metallic
Metal Fuel Performance Modeling and Simulation 727
Fuel Cladding fuel). This radial migration increases with both fuel
temperature and presence of Pu in the fuel. Modeling
of FCCI and FP migration should be accompanied by
10 mm
better understanding of the possible complex phase
relations between cladding and fuel constituents and
some of the FPs. Efforts to experimentally establish
those phase relations were started at Argonne National
Laboratory in the early 1990s as part of the IFR pro-
gram.40 The effort continued in Japan14,41 after the
termination of the IFR program. However, much
remains unfinished in this area of predicting the
phase relations to match the experiments, which can
now be complemented with extensive advanced phase
diagram simulations.42 In addition to the understand-
ing of the FCCI phenomena, efforts are currently
directed toward different possibilities to reduce its
effect on cladding performance. These include the
use of a barrier of zirconium, vanadium, or zirco-
nium-nitride on the cladding inner surface. Another
possibility is to investigate fuel alloy additions to slow
Rare earths Ferrite Martensite
Optical, Etched
down or impede diffusion of certain FPs that are
important to FCCI43 as discussed in Section 3.23.2.
Figure 15 Interdiffusion between U–10Zr fuel and HT9 Empirical diffusion-type models that account for
cladding irradiated to 6 at.% burnup at 620 C.
FCCI have been implemented in the LIFE-METAL
Scratches on
C/Ni depleted zone
outer diameter
on outer diameter
Scratches on
inner diameter
Original cladding
Inner diameter
Load bearing
Ligament
Original cladding
Outer diameter Steady-state
After steady-state plus transient
Figure 16 Cladding wastage mechanisms. Reproduced from ANL-IFR-244. Integral Fast Reactor Program Annual
Progress Reports FY 1993, Oct 1994.
728 Metal Fuel Performance Modeling and Simulation
before the fuel slug comes into contact with the compressible FP swelling is calculated from a model
cladding. The leveling-off in length increase is thus that assumes a number density cb of bubble nucleation
determined by the fuel smeared density. As men- sites, which is independent of burnup but dependent
tioned before, the behavior shown in the figure is on temperature according to
an anisotropic behavior in which the axial growth
c b ¼ Bb eðQ b =RT Þ ½18
proceeds more slowly than the radial growth, and
3
the extent of anisotropy increases with increasing where cb is in cm , Bb and Qb are empirical constants.
the fuel’s Pu content. This axial growth behavior is The model also includes a nonequilibrium effect by
represented in the LIFE-METAL code by a correla- allowing bubbles to grow to their equilibrium size at
tion and no mechanistic model is implemented. a rate proportional to the creep rate. Thus, more
Recently, a model that combines the predictions of bubble nucleation sites are predicted to exist at lower
constituent redistribution with axial lock-up effects temperatures, and the fewer gas atoms per bubble
was used to predict metallic fuel axial growth.38 (along with the reduced rate toward equilibrium) will
However, this model still uses fixed axial growth result in smaller bubbles and less swelling strain.
rates that are dependent on burnup, as the two dis- As experimental data for U–Zr and U–Pu–Zr fuel
tinct axial growth regimes shown in Figure 12 are became available, the different sets of LIFE-METAL
represented by two fixed growth rates. calibration constants were derived. Figure 18 shows
Although kinetic fission gas release models (such LIFE-METAL predicted and measured fission gas
as a model by Rest7; also see Chapter 3.20, Model- release in U–10Zr. Similar comparisons have been
ing of Fission-Gas-Induced Swelling of Nuclear made for U–Pu–Zr fuel data.
Fuels) are available, none has been implemented in
the LIFE-METAL code. Instead, fission gas release
3.23.3.3 LIFE-METAL Validation
fractions are calculated by a phenomenological corre-
lation. Future activities in this area will attempt to The following validation discussion is based on the last
implement a mechanistic fission gas release model code validation activities performed by Billone.44 This
into the code. In addition, a more detailed modeling validation effort has been extensive as it used post
of fission gas release considering fuel microstructure irradiation examination (PIE) data that are available
evolution is subject of future activities. from a large number of metallic fuel-pin irradiations at
A summary of LIFE-METAL implementation of EBR-II and FFTF.46 PIE data include fission gas
an empirical fission gas release model is given by release, fuel volumetric and fuel length change, clad-
Billone et al.2 In this model, the rate of increase of ding diametral change, and cladding wastage. Axial
retained fission gas in a fuel ring, G_ s , is found by a profiles are available for fuel radial growth at low
simple balance between the rate of gas production, burnup (prior to and including initial fuel–cladding
G_ p , and the rate of gas release, contact) and for cladding radial growth for a wide
G_ r ¼ Fg Gs ½15 range of burnups and fast fluences. Some data that
are available on a more limited basis are radial and
G_ s ¼ G_ p Fg Gs ½16 axial variations in U, Pu, Zr, fission gas porosity, axial
1 variations in fraction of porosity filled (logged) with
The fractional gas release rateFg (s ) is given as
follows: Na, and depth of C-depleted and Ni-depleted zones
h i in HT9 and D9, respectively. Fairly complete sets of
Fg ¼ 1 e aBu 1 þ cg ðP 0:1Þ2=3 data are available for 80 fuel-pin irradiations (111 pins
in total were used in the validation). Limited data
Rg e ðQg =RT Þ þ f ðes esi Þ ½17 (e.g., fuel length change, cladding diameter change)
where Bu is burnup in at.%, P is porosity (0:1 P), T is are available for hundreds of irradiated fuel pins.
the temperature in K, and es is the fuel swelling strain The validation database includes three cladding
0 ðes esi Þ. a, cg , Rg , Qg , f and esi are empirical types (cold-worked, austenitic D9 and 316 stainless
parameters to be determined from in-reactor data. steels and HT9 ferritic–martensitic steel) and eight
For 0:3 esi, the second term is orders of magnitude fuel compositions (U–10Zr, U–3Pu–10Zr, U–8Pu–
larger than the first term, giving rapid gas release when 10Zr, U–19Pu–6Zr, U–19Pu–10Zr, U–19Pu–14Zr,
fuel swelling of gas bubbles becomes >30%. U–22Pu–10Zr, and U–26Pu–10Zr, where the num-
The LIFE-METAL fuel-swelling model includes bers represent weight percents). The data from the
incompressible and compressible FP swelling. The 111 pin irradiations fall into one or more of the
latter part is due to gas bubble growth. The strain by following categories: fission gas release (FGR), fuel
730 Metal Fuel Performance Modeling and Simulation
Table 3 Summary of fabrication, operating conditions, and PIE data for LIFE-METAL validation
Peak Profile
PICT is peak inner cladding temperature, FGR is fission gas release, (DL/Lo)f is fuel axial strain, and (DD/Do)c is cladding diametral strain.
Metal Fuel Performance Modeling and Simulation 731
for cladding strain, fuel length change, and fission basis of detailed neutronics and thermal hydraulics
gas release. This specific experiment was a vehicle calculations to provide the parameters for each irra-
for validating the FCMI model in LIFE-METAL. diated pin at any axial location.
Figure 1925 shows the predictions (lines) versus data
(points) as a function of smeared density for two differ-
ent exposures (0.59 full power years and 1.25 full power 0.25
years). As shown in the figure, the agreement is good 0.20
H T9
for all of the low-burnup data, while high-burnup 0.15
data for the peak HT9 and D9 cladding strains for 75% 1.8
1.2
4
T470
0.59 Full power years
3.5 1 T473
1.25 Full power years LIFE-METAL
Diametral clad strain
3
0.8
Diametral strain (%)
2.5
0.6
2
0.4
1.5
1 0.2
0.5
0
0 0.2 0.4 0.6 0.8 1 1.2 1.4
0
68 70 72 74 76 78 80 82 84 86 x/Lo
Smear density (%)
Figure 21 Measured and LIFE-METAL predicted diametral
Figure 19 Measured and predicted peak cladding cladding strain profile for X425 experiments. Reproduced
diametral strain in X441 elements as a function of smeared from ANL-IFR-169. Integral Fast Reactor Program Annual
density and exposure time. Progress Reports FY 1991, June 1992.
732 Metal Fuel Performance Modeling and Simulation
3.23.3.4 LIFE-METAL Code Status and 3. an empirical model to incorporate the effect of
Future Activities large radial crack formed in the U–Pu–Zr ternary
fuel slug,
The latest calibration of the LIFE-METAL code was
4. a model to calculate the volume increase of the
performed just before the termination of the IFR
fuel alloy matrix due to accumulation of nongas-
project in 1994. Sets of verification test problems that
eous FPs (solid FP swelling),
correspond to data from different EBR-II experiments
5. an empirical correlation to estimate the cladding
are available and have been used systematically to
wastage by rare-earth FP,
verify the code calculations.
6. a temperature analysis model with effective ther-
Minor changes have been done to the code since
mal conductivity of the porous slug infiltrated by
its calibration. Those changes did not affect the
the bond sodium.
code’s calibration and were mainly aimed at correct-
ing a code error associated with FCCI for fuels with These models are linked in ALFUS so that the rela-
long irradiation periods. tions among in-reactor phenomena in metal fuel pins
As for future plans for improving the code, can be simulated properly. Input data for ALFUS are
modeling effort will be required to reevaluate the fuel pin specification and irradiation condition, such as
effect of constituent redistribution, FP migration, linear power rate and its axial distribution, coolant
and fission gas release, and improve LIFE-METAL’s flow rate per pin, and coolant inlet temperature.
existing models by including mechanistic models A set of model parameters, the physical and mechani-
to account for the synergistic effects of those phe- cal properties of the fuel alloy and cladding, is also
nomena.6,7 Those code improvements can provide required. The ALFUS code outputs temperature dis-
better predictions about fuel swelling and fuel ther- tribution and stress–strain state in the slug and clad-
mal behavior, including fuel thermal conductivity. ding, fission gas release, margin to cladding creep
New models for FCCI will need to be developed rapture (cumulative damage function, CDF), etc.
focusing initially on rare-earth FP diffusion and Changes in the melting temperature (solidus and
their interactions with cladding; and phase diagrams liquidus) of the fuel slug due to the fuel constituent
for alloys of rare earth and cladding constituents migration are important when the design limit of the
are needed. fuel pin is evaluated, so ALFUS includes the thermo-
Metallic fuel remains to be a candidate for use as diffusion model20 to simulate the fuel constituent
a transmutation fuel containing minor actinide ele- migration. However, this model is not used since the
ments such as Am and Np in addition to U, Pu, and stress–strain calculation in the current version of
Zr.52 Inclusion of models for the behavior of MA in ALFUS is relatively insensitive to changes in the
the LIFE-METAL code will be a high-priority task local compositions of the fuel slug.
for the future code development, which is dependent The following sections describe the analytical
on the availability of experimental data and a better models in some detail, including the model para-
knowledge of phase diagrams for those elements with meters and the fuel alloy properties, and then present
other fuel components. the results of ALFUS validation. Finally, the stress–
strain behavior of metal fuel pins is summarized.
3.23.4 ALFUS
3.23.4.2 Models in ALFUS
3.23.4.1 Background
3.23.4.2.1 Calculation flow in ALFUS
For a comprehensive understanding of the mechani-
The calculation flow in ALFUS is schematically
cal behavior of metal fuel during irradiation, the
shown in Figure 22. First, the temperature distribu-
computer code ALloyed Fuel Unified Simulator
tion in the fuel pin at time step t ¼ tnþ1 is calculated
(ALFUS)3,53 was developed. The ALFUS code in-
from a linear power rate and a coolant flow rate at the
cludes the following analytical models:
present t ¼ tnþ1 , and from the slug–cladding gap and
1. a stress–strain analysis model, including gas bubble porosity in the slug at the previous time step t ¼ tn.
compressibility and open pore volume decrease, Second, gas swelling and gas release are calculated
2. a mechanistic model to describe coalescence and based on the local temperature at t ¼ tnþ1 and the
growth of the gas bubbles, formation of the open stress state at t ¼ tn . After strain increments due to gas
pores, and fission gas release, swelling, solid FP swelling, and thermal expansion
Metal Fuel Performance Modeling and Simulation 733
Cladding
n
x
–1 1
n o
iþ1
The derivatives @Ni =@r and @Ni =@z is given by unknown variable dsnþ1 in eqn [19] as
( @N ) 8 9 n o
i < @Ni = ð½Ce þ ½Cs þ ½Cc Þ dsnþ1iþ1
@r 1 @
¼ ½J n o
@Ni : @Ni ; ½23 ¼ ½Bn Dunþ1 iþ1
½Ce sinþ1 fsn g
@z @
n o n o n o n o
th;iþ1 ss;iþ1
where the Jacobian matrix ½J is given by Denþ1 þ Denþ1 þ Desw;i
nþ1 þ De cr;i
nþ1 ½25
2 3
P
4
@Ni P
4
@Ni where fug is the element nodal displacement and
6 @ ri @ zi 7
6 i¼1 i¼1 7 D means increment from the previous time tn to the
½J ¼ 6 7
4P 4
@Ni P4
@Ni
5 present tnþ1 . Volumetric strain is represented by e
@ ri @ zi ½24 without
n {} in theo following text. Thermal expansion
i¼1 i¼1 th;iþ1
strain Denþ1 is given as a function of temperature.
Considering the strain increments at the i þ 1th itera- Components of strain vector feg are summarized in
tion at the present time tnþ1 , we can express the Table 4. Substituting eqn [25] into eqn [19] to eliminate
Metal Fuel Performance Modeling and Simulation 735
n o n o
iþ1
dsnþ1 iþ1
, we can obtain the solution Dunþ1 . Then, the where Ci is the number density (n m3) of the class i
bubble. From eqns [27], [30], and [31],
stress vector and each component of the strain at the 21 1 1 3
time step t ¼ tnþ1 is calculated after the i þ 1th iteration. 3 3 3 0
The stress–strain matrices ½Ce and ½Cs correspond 61 1 1 7
X 1 desw 6 07
to the elasticity of the fuel alloy and volumetric change i 63 3 3 7
½Cs ¼ 6 7
of closed gas bubbles, respectively. The other stress– 3 dsm 6 1 1 1 7
i 4 3 3 3 05
strain matrix ½Cc represents the response by creep of
0 0 0 0
the fuel alloy and volume decrease of the open pores.
Each stress–strain matrix is given as follows. 4ri2 ðri3 43
mi Av bÞCi
For the elastic strain–stress matrix ½Ce , ¼
27ri2 sm þ 36gri þ 18g 4
3
mi Av b=ri2
n o n o
2 3
el;iþ1 el;i iþ1
Denþ1 Denþ1 þ ½Ce dsnþ1 1 1 1 0
6 7
61 1 1 07
2 3 6 7
1
E E 0 6 7
E 61 1 1 07
6 7 4 5
6E E
1
E 0 7
½Ce ¼ 6 7 ½32
6 E 1
0 7 0 0 0 0
4 E E 5
0 0 0 2ð1þÞ ½26 The stress–strain matrix ½Cc in eqn [25] for the creep
E
deformation is defined as
where E is the elastic modulus and n is the Poisson’s
n o n o
ratio. The stress–strain matrix ½Cs in eqn [25] for the cr;iþ1
Denþ1 cr;i
Denþ1 þ ½Cc dsnþ1 iþ1
volume change of closed gas bubbles is defined as
n o
n o
sw;iþ1
denþ1 iþ1
½Cs dsnþ1 ¼ Dtnþ1 e_ nþy
c;iþ1 ½33
n o n o
sw;iþ1
Denþ1 sw;i
Denþ1 ½27
where
n othe time increment Dtnþ1 ¼ tnþ1 tn , and
_e c;iþ1 is the creep strain rate vector at a stress
nþy
As described in Section 3.23.4.2.3, the volume fsnþy g estimated at a time tnþy ¼ tn þ yðtnþ1 tn Þ,
of a closed gas bubble (class i ) is calculated by the (0 y 1). According to Nakajima et al.,54 the
Van der Waals equation: matrix ½Cc is given by
sm þ
2g 4 3
pr mi Av b ¼ mi RT ½28 h ii 0 i !
ri 3 i 0 0
@sj
½Cc ¼ yDtnþ1 F1 sj sk þ F2
nþy @sk nþy
where g is the surface tension (N m1), ri the radius
i !
of a class i bubble (m), mi the gas amount in a class i i
fnþy
9 @f ½34
bubble (mol), Av the Avogadro number (atoms mol1), F1 ¼ 2
R gas constant ( J K1 mol1), T the temperature (K), 4 si @s nþy s inþy
nþy
b ¼ 85 1030 (m3), and sm the average stress: i
3fnþy
sr þ sy þ sz F2 ¼ i
sm ¼ ½29 2s nþy
3
Differentiation of eqn [28] yields where the function f is the uniaxial creep strain
dri ri3 4p
3
mi Av b rate of the fuel alloy given by eqn [38]; f ¼ e_ c ðs Þ.
¼ ½30 The equivalent stress s is given by eqn [39], and the
dsm 3ri2 sm þ 4gri þ 2g4p3
mi Av b=ri2
deviatoric stress fs0 g is defined as
When we consider the volume change of one gas
bubble, 2s @ s
f s0 g ¼ ½35
desw desw dri 3 @s
i
¼ i
dsm dri dsm n o
and Note that Decr;iþ1
nþ1 in eqn [33] includes the strain
deswi increment for the volume decrease of the open pores,
¼ 4pri2 Ci ½31
dri as shown in the next section.
736 Metal Fuel Performance Modeling and Simulation
3.23.4.2.2.2 Volume decrease of the open pores hydrostatic stress applied to the open pores because
Total swelling increment fDesw g (hereinafter, itera- the gas plenum pressure Ppl in eqn [39] is equal to the
tion and time step counters are dropped) in eqn [25] ‘internal’ pressure of the open pores. The hot press
is defined as parameter a means the compressibility of the open
pores. Substituting eqn [39] into eqn [37], we obtain
fDesw g ¼ Decls þ fDeopn g;
½36 e_ c ð
sÞ
fDeopn g ¼ fDeopnþ g þ fDeopn g fecr g ¼ t
2 s 8 9
2 3>sr >
The vector De cls
is the isotropic strain increment < >
2 þ 6a 1 þ 6a 1 þ 6a 18a > =
due to the closed bubble development, and fDeopn g is 4 1 þ 6a 2 þ 6a 1 þ 6a 18a 5 s
>sz>>
the ‘net’ isotropic strain increment due to open pore 1 þ 6a 1 þ 6a 2 þ 6a 18a >: ;
Ppl ½40
volume change, which consists of two components:
fDeopnþ g and fDeopn g. The increment fDeopnþ g In eqn [40], the value of 1 þ 6a must be negative. (For
corresponds to open pore increase by interconnec- example, consider the case of fsgT ¼ fsr ; 0; 0; 0g. The
tion of the closed bubbles. The gas swelling model y must be opposite to that of sr .) Therefore, a
sign of Decr
explained later gives Decls and fDeopnþ g. The should be in the range of 0 a 1=6. The hot press
increment fDeopn g means the open pore collapse, parameter a is assumed to be dependent on the open
which is calculated with creep strain increment pore swelling eopn
sw and takes the maximum value when
fDecr g as follows. Since the open pore has connection eopn
sw is as large as 10%:
with the external of the slug and does not include 8
> 0:0; eopn
sw ¼ 0
high-pressure fission gas unlike the closed bubble, it >
>
may collapse when compressive stress is applied. >
< 1:5
1 eopn
Here, volume decrease due to the open pore collapse a¼ sw
; 0< eopn
sw < 0:1
>
> 6 0:1
is assumed to be controlled by creep mechanism and >
>
: ½41
calculated in a manner similar to the hot press model 1=6; 0:1 eopn
sw
proposed by Rashid et al.55 When the creep strain
component without volume change is denoted by The dependency of a on eopn sw was determined on the
fDecreep g, the sum of fDeopn g and fDecreep g is basis of the following consideration of the compressibil-
expressed by the following flow rule: ity of the porous material which consists of the same-
sized spherical shells (see Figure 24).
fDecr g fDecreep g þ fDeopn g When hydrostatic pressure Pex is applied to the
porous material as shown in Figure 24, the equivalent
@s
¼ e_ c ðsÞ Dt
@si ½37
where e_ ð
c
sÞ is the equivalent creep strain rate and si
P
(i ¼ r ; y; z) is the stress in the slug. The stress si
takes a negative value when it is compressive. The
equivalent creep strain rate for the U–Pu–Zr ternary
alloy is given by
(
ð5000 s4:5 Þe52 000=RT 923K
s þ 6:0
e_ ð
c
sÞ ¼ ½38
s3 e28 500=RT
0:08 > 923K
P P
according to Gruber and Kramer.56 The equivalent
in eqn [37] is given by
stress s
1
s ¼ ððsr s Þ2 þ ðs sz Þ2 þ ðsz sr Þ2
2
1=2
2
þ 6rz Þ þ 3aðsr þ sy þ sz þ 3Ppl Þ
½39
P
where the shear stress trz is usually negligible. The Figure 24 Closely-packed spherical shells, forming
term ðsr þ sy þ sz þ 3Ppl Þ represents the net effective porous material.
Metal Fuel Performance Modeling and Simulation 737
coupling eqn [43] with eqn [44] yields Creep strain rate e_ c for U–(Pu–)Zr alloys at 973 K is
given in McDeavitt57 by
1
epore nþ1
2n
Specimen Driving force (MPa) Linear strain Ratea (/s) Fractional closed poreb(%) a
a
Read from plot data in McDeavitt57
b
Calculated from data reported in McDeavitt57 by (% closed porosity)/(100-(% closed porosity)).
738 Metal Fuel Performance Modeling and Simulation
which is essentially identical to eqn [38]. The equiva- This gas swelling model comprises two parts (see
HIP in eqn [48] is given in the form similar
lent stress s Figure 25); One of the submodels describes migration
to eqn [42]: of fission gas atoms and intragranular gas bubbles into
pffiffiffiffiffi the grain or phase boundary (intragranular gas model).
HIP ¼ 3 3aPeff
s ½50
Another submodel describes growth of the grain-
where Peff is the HIP driving force in Table 5. Substi- boundary bubbles due to gas supply from the grain
tution of eqns [49] and [50] into eqn [48] gives the hot and coalescence with the existing grain-boundary bub-
press parameter: bles (grain-boundary gas model). The open pore forma-
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi tion due to interconnection of the grain-boundary
_el
a¼ ½51 (closed) bubbles and fission gas release through the
7:8 106 Peff 3
open pore are also treated in this submodel. Gas
Values of a calculated from eqn [51] are also presented swelling volume is calculated as the total volume of
in Table 5, which indicates that a is in the range from the closed bubbles and the open pores.
1/220 to 1/90. Although the function of eqn [41] gives
larger a values than this range, eqn [41] seems accept- 3.23.4.2.3.1 Intragranular gas model
able because the open pore in the irradiated fuel slug is The intragranular gas model for oxide fuel by Wood
really ‘open’ to the outside of the slug and considered and Matthews59 is modified and applied to calculate
more compressible than the closed pore in the sintered the concentrations of the fission gas atoms and the gas
U–Zr specimen. bubbles in the grain. The following assumptions are
made in this model:
3.23.4.2.3 Gas swelling model
The
clsgas
swellingopnþ
model gives the strain increments (a) A part of the gas atoms generated by fission forms
De and fDe g in eqn [36] due to the closed gas bubbles in the grain. Some gas atoms may be
bubble and open pore formation, respectively, and also trapped by the existing gas bubbles. Some of the
calculates the amount of fission gas released into the gas bubbles may be destroyed by a fission fragment
gas plenum. This swelling model, which was applicable and dissolved in the fuel alloy matrix. Fission gas
to analysis of uranium metal swelling,58 was modified atoms and gas bubbles migrate in the spherical
and incorporated into the ALFUS code.53 grain by the diffusion mechanism (Figure 25).
Grain-boundary bubble
Di
ffu
sion
Coalescence
Intragranular bubble
ion
fus
Dif
Open pore
Spherical grain
The number density of the gas bubbles is repre- migrating gas atoms is expressed in Wood and
sented by a single bubble size. Matthews59 as
(b) The gas bubble formation (nucleation), gas atom b
kb2 ¼ 4pr b C ½55
trapping by the gas bubbles, and the resolution of
the gas into the matrix are in an equilibrium state. Here, Cb is estimated from the number density of the
(c) The gas atoms and bubbles that arrive at the grain bubbles at the previous time step. The radius of
boundary form the grain-boundary bubbles or flow the bubble r b in eqn [55] is calculated by the Van
into the existing grain-boundary bubbles, and do der Waals equation (see eqn [28]), assuming an equi-
not come back to the grain. This means that the librium state among internal pressure, surface tension,
concentrations of both the gas atoms and the intra- and external stress (average stress in the fuel slug) of
granular bubbles are zero at the boundary. the bubble. The gas amount m0 in the bubble was
assumed to be m0 ¼ 1 1022 so that the bubble
According to Wood and Matthews,59 assumption (a)
radius r b takes a value of 1 nm. From eqn [54],
is expressed as
bK Dg kb2 þ K
@Cg 1 @ 2 @Cg Cg ¼ Ct ; and Cgb ¼ Ct ½56
¼ 2 Dg r þ Kg Dg kb2 Cg Dg kb þ b
2 Dg kb2 þ b
@t r @r @r
Coupling eqns [52], [53], and [56] yields
KCt þ bCgb ½52
@Ct 1 @ 2 @Ct
@Cb 1 @ @Cb ¼ 2 ðD1 þ D2 Þr þ Kg ½57
Av m0 ¼ Av m0 2 Dgb r 2 @t r @r @r
@t r @r @r
where
þ Dg kb2 Cg þ KCt bCgb ½53
bK Dg kb2 þ K
D1 ¼ Dg ; D2 ¼ Dgb ½58
where Cg is the concentration of fission gas atoms Dg kb þ b
2 Dg kb2 þ b
dissolved in the matrix (atoms m3); Cgb , the concen-
During normal reactor operation, change in the irra-
tration of gas included in intragranular bubbles (atoms
diation conditions, such as fission rate and tempera-
m3), Cgb ¼ Av m0 Cb ; Cb , the number density of intra-
ture, is much slower than diffusion of the gas atom and
granular bubbles (bubbles m3), Ct ¼ Cg þ Cgb ; Av ,
the bubble, so the system can be assumed to be in a
the Avogadro number (atoms mol1); m0 , the gas
steady-state: @Ct =@t ¼ 0. The diffusion equation [57]
amount in one intragranular bubble (mol); Dg , the
can be solved under the boundary conditions: Ct ¼ 0
diffusion coefficient of the fission gas atom (m2 s1);
at the grain boundary (r ¼ rG , assumed to be 2 mm)
Dgb , the diffusion coefficient of the intragranular bub-
from assumption (c), and @Ct =@r ¼ 0 at the center of
ble (m2 s1); Kg , the generation rate of the fission gas
the spherical grain (r ¼ 0). In ALFUS, the approxi-
atoms (atoms m3 s1); Dg kb2 Cg , the rate of gas atom
mate numerical solution is used that Matthews and
trapping by the bubble (atoms m3 s1); KCt , the
Wood60 proposed.
bubble nucleation rate (atoms m3 s1); and bCgb
The diffusion coefficient Dg of the gas atom in
the rate of the bubble resolution by the fission frag-
eqns [52] and [53] is tentatively assumed to be equal
ment (atoms m3 s1).
to the self-diffusion coefficient of g-uranium61:
Wood and Matthews59 assumed that the gas bub-
bles were immobile for the oxide fuel, so that the first 26700
term in the right-hand side of eqn [53] was not Dg ¼ 1:19 107 exp ½59
RT
included in the equation. In metal fuel, however, the
diffusion of gas bubbles will not be negligible, because no data on diffusion of the noble gas atom in
because the ratio T =Tm of the fuel temperature T metals is available. The diffusion coefficient Dgb of
(on average) to the melting temperature Tm for metal the intragranular bubble in eqn [53] can be related to
fuel is higher than that for the oxide fuel. the surface diffusion coefficient Ds of the bubble
From assumption (b), inner surface by considering the jump distance of a
bubble due to individual jumps of molecules on its
Dg kb2 Cg þ KCt ¼ bCgb ½54 inner surface,62 which yields a form of
Values of K and b were estimated as K ¼ 4 103
(s1) and b ¼ 6 102 (s1) based on Wood and 3 a0 4
Dgb ¼ Cadj Ds ½60
Matthews.59 The sink strength kb2 of bubbles for the 2p rb
740 Metal Fuel Performance Modeling and Simulation
where a0 is the lattice constant of the fuel alloy (f ) When the bubbles interconnect and form the open
(¼3.01010 m). The surface diffusion coefficient pore, the fission gas included in the bubbles is
Ds of the fuel alloy in eqn [60] is assumed to be immediately released into the gas plenum. The
Ds ¼ 1000Dg , which gives the value close to the intragranular and grain-boundary bubbles may col-
correlation evaluated by Gruber and Kramer.63 The lide with the open pore. It leads to gas release and
parameter Cadj in eqn [60] was introduced to account increase in an open pore size. The fission gas atoms
for various factors affecting the gas bubble migration, may also be released through the open pore.
such as fission fragments, multiphase structures of
The grain-boundary (closed) bubbles are classified
the fuel alloy, and an internal gas pressure in the
into M 1 classes by gas amount mi (mol) included in
bubble. The value of Cadj was adjusted as described
each bubble so that growth of the bubbles can be
in Section 3.23.4.2.8; Cadj ¼ 20. In this case,
numerically described. Although the characteristics
Dgb
77Dg for the intragranular bubble.
of the open pore will be different from that of the
As far as eqn [59] is used, Dg takes a value in the
closed bubble, the open pore is categorized into the
range of 1013 to 1015, and Cb 1015 to 1016
class M ‘bubble’ for the convenience of numerical
in the analyses of EBR-II test fuels and the prototypic
modeling. The formation of the open pore (bubble
fuel. Therefore, the value of Dg kb2 is on the order of
interconnection) is, however, treated in a manner
106 ( b), so that the gas concentrations in the
different from the growth of the closed bubble, as
form of the gas atoms and bubbles can be estimated
mentioned later.
from eqn [56] as
The change in the number density of the closed
K K
Cg
1 Ct
0:93Ct and Cgb
Ct bubble of class i can be described by considering fission
b b ½61 gas supply from the grain, growth of lower (i 1) class
0:07Ct
bubbles, growth to upper class (i þ 1) bubbles, and
Therefore, the ratio of fission gas transfer by the gas collision with upper class bubbles (assumption (d));
atom diffusion to that by the bubble diffusion can be
dC1 X M
estimated as ¼ G1 R0i;j Rj ;1 ; i¼1 ½63
dt
Dg @Cg =@r r ¼rG Dg 0:93ð@Ct =@r Þr ¼rG j ¼1
0:17
Dgb @Cgb =@r r ¼r 77Dg 0:07ð@Ct =@r Þr ¼rG dCi X
i1 X i
¼ Gi þ R0i1;j R0i;j
G
½62 dt j ¼1 j ¼1
½64
This indicates that 15% of the total fission gas X
M
where ri is the radius of class i bubble and Di is the as follows. Since the open pore is treated as the
diffusion coefficient of class i bubble. The factor Fi;j in highest-class (M) bubble, the factor FM;j in eqn
eqn [66] means a fraction of the class i bubbles which [72] is given by the following conservation of gas
rank up to the upper class (i þ 1) when one class i amount:
bubble collides with the lower class ( j) bubble, and can
mM þ mj ¼ mM þ FM;j mM ½73
be given by the conservation of the amount of the
bubble gas: Note that ‘gas amount’ mM of the open pore is the
input parameter that controls the rate of rank-up of
mi þ mj ¼ ð1 Fi;j Þmi þ Fi;j miþ1 : ½67
the closed bubble to the open pore. The value of mM
The diffusion coefficients Di in eqns [65] and [66] was adjusted so that the fission gas release data can
should be determined by considering the characteris- be reproduced, as explained in Section 3.23.4.2.8.
0
tics of the grain or phase boundary. However, the The factor FM1;j in eqn [70] represents the fraction
bubble mobility on the boundary in the fuel alloy is of the bubble interconnection to coalescence of one
not known well at this stage. It was assumed, therefore, class j bubble with the class M 1 bubble. When
that Di is given by eqn [60]. every coalescence of the class j bubble with the class
M 1 bubble contributes to the open pore forma-
3.23.4.2.3.2.2 Open pore The formulation similar tion (the bubble interconnection, i.e., rank-up to the
to eqns [63]–[67] was assumed for the open pore class M), the conservation of gas amount included in
formation by interconnection of the gas bubbles. For the bubbles is expressed as
the class M 1 bubble (the highest class of the closed
bubble) and the class M ‘bubble’ (the open pore), mM1 þ mj ¼ ð1 FM1;j ÞmM1 þ FM1;j mM ½74
dCM1 X
M 2 where FM1;j is essentially the same as Fi;j in eqn
¼ GM1 þ R0M2;j 0
[67]. In this case, FM1;j ¼ FM1;j . On the other
dt j ¼1
hand, when gas swelling is not enough to form the
X
M1 X
M open pore and a part of the coalescence with
R0M1;j Rj ;M1 the class M1 bubble contributes only to increase
½68
j ¼1 j ¼M1 of the number density of the class M 1 bubbles,
the gas amount conservation is expressed as
dCM X
M1 X
M1
¼ GM þ R00M1;j þ R0M;j ½69 0
mM1 þ mj ¼ ð1 FM1;j ÞmM1
dt j ¼1 j ¼1
þ FM1;j fM1;j mM ½75
R0M1;j ¼ 4pdðDM1 þ Dj Þ 0
The factor FM1;j is calculated by coupling eqn [74]
ðrM1 þ 0
rj ÞCM1 Cj FM1;j ½70 with eqn [75]. The function fM1;j can be considered
as a probability of open pore formation. It appears
R00M1;j ¼ 4pdðDM1 þ Dj Þ reasonable to assume that the probability fM1;j
increases with the gas swelling level. In this study,
ðrM1 þ rj ÞCM1 Cj FM1;j fM1;j ½71 fM1;j is given by
8
R0M;j ¼ 4pdðDM þ Dj ÞðrM þ rj ÞCM Cj FM;j ½72 >
> 0:0; eSW eSW
< SW SW 1
e e1
The second term in the right-hand side of eqn [69] fM1;j ¼ eSW eSW ; e1 < e < eSW
SW SW
>
> 2 1
2
represents the open pore formation by interconnec- : SW SW
1:0; e2 e ½76
tion between the grain-boundary bubbles. The first
and third terms indicate evolution of the existing where esw is the calculated gas swelling and esw
1 is the
open pores by the fission gas supply from the grain threshold gas swelling for open pore formation. The
and by coalescence with the grain-boundary bubbles, breakaway swelling level esw2 was taken as e2 ¼ 0:33
sw
respectively. For the values of mi assumed currently based on a theoretical consideration by Barns.65 The
(see Section 3.23.4.2.8), DM
DM1 DM2 and value of esw
1 can be determined either by irradiation
R0M;M1 R0M;M2 , which means that the open pores experiences or by thermodynamic assessment of
are treated as immobile pores. open pore morphology, as in the case of UO2
Based on assumption (e), F-factors in eqns [70]– fuel.66 Tsai et al.67 reported that fuel pins of 85%
[72] are different from that determined by eqn [67] smear density, where the fuel slug is allowed to swell
742 Metal Fuel Performance Modeling and Simulation
only 18% before the slug–cladding contact, deformation before the slug–cladding contact is
achieved 10 at.% peak burnup without cladding anisotropic; The U–Zr binary fuels of about 75%
failure and showed a fractional fission gas release smear density show an axial elongation of 8–10% of
of 60%. This suggests that the value of esw
1 should as-fabricated slug length, while radial growth is about
be much less than 0.18 (¼18%). Based upon this 16% at the slug–cladding contact.45 This anisotropic
irradiation experience, esw
1 is set to 0.10. deformation can be related to the tearing at the grain
or phase boundaries in the peripheral region of the
3.23.4.2.3.2.3 Fission gas release The amount of slug due to anisotropic irradiation growth of a-U
fission gas retained in a unit volume of the fuel alloy crystals.7,45 In the case of the U–Pu–Zr ternary fuel,
is expressed as more pronounced anisotropy was observed and sug-
X
M 1 gested to be dependent upon the Pu content and
Mretain ¼ Cg þ Cgb þ mi C i ½77 radial temperature gradient in the fuel slug.45 The
i¼1 anisotropy of the ternary fuel slug can be attributed
and fission gas release during a time interval Dt is calcu- to large radial crack formation due to the brittle
lated from the change in the retained gas amount. Con- nature of the ternary slug.45 The cracks increase
sistent with eqn [77], fission gas release rate dMrelease =dt the slug radial strain and promote the anisotropy.
can also be derived from eqn [69] multiplied by ‘virtual’ In order to incorporate the effect of the cracks on
gas amount mM in the open pore, namely, the stress–strain state, the concept of the effective
slug radius was introduced. The effective radius r eff
dMrelease dCM X
M1
is defined (see Figure 26) as
mM ¼ mM G M þ m M R00M1;j
dt dt j ¼1 r eff ¼ r0 þ dr slug þ dr crack ¼ r slug þ dr crack ½81
X
M 1 where r0 is the as-fabricated slug radius, and dr slug
is
þ mM R0M;j the radial strain increment due to thermal expansion,
j ¼1
½78
elasticity, creep, gas, and solid FP swelling. The
The first term in eqn [78] represents release of the fission increment dr crack is due to the cracks and the tearing.
gas atom from the grain through the existing open pores. After the slug including the tears and cracks comes
The second term means gas release when the grain- into contact with the cladding, swelling into the tears
boundary (closed) bubbles are interconnected. The and cracks will occur. In fact, extrusion of inner zone
third represents release of the gas in the grain-boundary fuel into the radial crack was observed.45 This process
bubbles through the existing open pores. will relax radial FCMI stress. During this stage of the
contact (Stage II in Figure 26), the area of the contact
3.23.4.2.3.2.4 Gas swelling Volumetric strain interface continues to increase without significant
increment of the slug due to the closed bubble evo- FCMI stress, and then the slug will stick to the clad-
lution during a time interval Dt is given by ding. Using r eff , the following contact conditions are
adopted in the ALFUS calculation (see Figure 26);
X
M 1
dCi
nþ1 ¼ Dt
Decls vi ½79 (I) r eff < ri : no restraint by the cladding (no contact),
i¼0
dt
(II) r slug < ri r eff : axial restraint by the cladding,
where i ¼ 0 means the intragranular bubble. The but no radial restraint,
volume vi of one class i bubble is calculated from (III) ri r slug : both axial and radial restraints by the
the gas amount mi in the bubble based on the Van der cladding,
Waals equation. Volumetric strain increment due to
where ri is the inner radius of the cladding. When the
open pore formation is given by
aforementioned condition is applied to the contact-
opnþ dCM state judgment, the value of the anisotropic radius in-
Denþ1 ¼ vM Dt ½80
dt crement dr crack at the beginning of Stage II ðr eff ¼ ri Þ
The open pore volume vM is calculated by the expe- must be known, which is estimated as follows.
dient of using the same scheme as the closed bubbles. As shown in Ogata,53 the sum of the strain com-
ponents except tearing and crack strains is almost
3.23.4.2.4 Effect of radial cracks isotropic until the slug–cladding contact,
so that its
After the fuel slug comes into contact with the clad- strain vector can be expressed as eiso ; eiso ; eiso .
ding, further axial growth is restrained. The slug Assuming that the sum of the tearing and crack
Metal Fuel Performance Modeling and Simulation 743
Cladding
r eff r eff
r slug r slug
dr crack dr crack
Cladding
gap slug
strains
crack iscrack
highly anisotropic and can be expressed as where r0 ð¼ ri r0 Þ is the initial gap width
e ;e ; 0 (no axial component), local axial between the slug and cladding. The factor f crack can
elongation can be expressed as 100eiso (%), and be estimated as follows; eqn [83] can be transformed to
slug
dr slug ¼ eiso r0 ; dr crack ¼ ecrack r0
slug
½82 eiso
f crack
¼ 1 iso
e þ ecrack contact
Here, the anisotropy factor f crack is introduced,
which is defined as a ratio of the anisotropic strain
eiso contact
to the total strain at the slug–cladding contact ¼ 1 qffiffiffiffiffi ½85
ðr eff ¼ ri Þ: SD 1
100
ecrack where SD is the initial smear density (%) of the
f crack
iso ½83
e þ ecrack contact fuel pin;
!
where variables in the parentheses ðÞcontact take the slug 2
r0
values at the slug–cladding contact. The anisotropy SD ¼ 100 ½86
ri
factor f crack is assumed to be a function of the initial
Pu content CPu and the temperature gradient in the Values of f crack for various fuel pins were calculated by
peripheral region of the slug. The temperature gradi- using eqn [85], and plottedin Figure
27 as a function
ent can be shown to be proportional to the value of of CPu and q=D0 . Values of eiso contact in eqn [85] were
q=D0 : the liner power rate divided by the diameter of estimated from the experimental data45,68,69 of the
the slug. Substitution of eqns [81] and [82] into [83] axial elongation of the whole slug instead of the local
yields elongation. As can be seen in Figure 27, f crack values
crack gap saturate at f crack
0:8 when CPu 19 (wt%) and
dr ¼ f crack r0 ½84
contact q=D0 750 (W cm2). In the calculations described
744 Metal Fuel Performance Modeling and Simulation
Elements Generation (mol) Molar volume (cc mol1) Volume change (cc)
thermal conductivity of the slug due to evolution of Using the conductivity correction factor fcon, the
the gas bubble and infiltration of the bond sodium. effective thermal conductivity of the fuel slug is
The thermal conductivity of the fission gas in the expressed17 as
closed bubble and the open pore is much less than
keff ¼ fcon kf ½90
that of the fuel alloy, so that the effective conductivity
of the irradiated slug is degraded as fractions of the where kf is the thermal conductivity of the unirradi-
bubbles and open pores increase. On the other hand, ated fuel alloy. The correction factor fcon is approxi-
Hofman and Walters1 reported that the bond sodium mated by the following expression,17 which was
infiltrates the irradiated slug, probably through the derived for spherical gas-filled and irregular
open pore. Since the thermal conductivity of sodium sodium-filled pores:
is comparable to that of the U–Pu–Zr alloys, bond 2 3
kNa
sodium infiltration recovers the degraded conductiv- 6 1 7
6 PNa kf 3=2
ity of the irradiated slug. These effects should be fcon ¼ 61 3 7 7 1 Pg ½91
4 1 Pg 2 2 kNa 5
incorporated into the effective thermal conductivity þ 3
of the slug. e e kf
746 Metal Fuel Performance Modeling and Simulation
where kNa is thermal conductivity of sodium, Pg although values or correlations for these parameters
volume fraction of gas-filled pore, and PNa is include uncertainties due to effects of the various
the volume fraction of sodium-filled pore. The factors not considered in the models.
shape factor e is taken as e ¼ 1:72. In the ALFUS The models in ALFUS include other parameters
model, a part of the existing open pores is assumed and fuel alloy properties, values of which were as-
to be infiltrated by the bond sodium. Using the vol- sumed tentatively because of the lack of relevant data.
ume fraction FN of the sodium-filled open pores to The diffusion coefficient Dg of the gas atom was
the total open pores, the volume fractions Pg and PNa assumed to be equal to the self-diffusion coefficient
are expressed as of g-uranium (eqn [59]). The larger number of
the bubble class M is preferable to simulate the
ecls þ ð1 FN Þeopn
Pg ¼ ½92 actual evolution of the bubbles, but M ¼ 6 was
1 þ etotal assumed to cut down the computing time. The gas
FN eopn amounts mi of the closed bubbles were determined as
PNa ¼ ½93 m1 ¼ 4.0 1021, m2 ¼ 1.6 1019, m3 ¼ 6.4 1018,
1 þ etotal
m4 ¼ 2.6 1016, and m5 ¼ 1.0 1014, because the
In eqns [92] and [93], etotal denotes the total volumet- observed bubble size is on the order of micron
ric strain of the slug, ecls is volumetric strain due meter. In this case, the bubble radii of class 1 to
to the closed bubbles, and eopn volumetric strain M1 are in the range of 4 nm–1 mm. The diffusion
due to the open pores. The volume fraction FN is coefficients Di of the grain-boundary bubbles were
estimated as FN ¼ 0:5. As shown in the next section assumed to be given by eqn [60] because of insuffi-
(Figure 37), the open pores occupy 20–70% of the cient knowledge of the bubble mobility on the
porosity: the volume fraction of the sum of the closed boundary.
bubbles and the open pores to the swelled fuel. It On the premise that the tentative values and cor-
means that about 10–35% of the porosity is assumed relations mentioned earlier are used, values of the
to be infiltrated by the bond sodium. This estimation multiplication factor Cadj for the bubble diffusion
is consistent with the following observation results: coefficient (eqn [60]) and the parameter mM (‘gas
15–20% of the porosity1 and 21–39% of the amount’ of the open pore; eqn [73]) were adjusted,
porosity17 (calculated from Table I in Bauer and as follows. Fission gas release data (Figure 28) show
Holland17) are infiltrated by the sodium. sharp increase in the amount of the gas released after
The thermal conductivity kf of the unirradiated an incubation period (1 at.% peak burnup). At the
U–Pu–Zr alloy is given as a function of Pu content higher burnup, the gas release levels off at an asymp-
CPu , Zr content CZr , (in atom fraction), and tempera- totic value of 60–80%. The calculated gas release
ture T (K); over the whole irradiation period increases with the
kf ¼ 16:309 þ 0:02713T 46:279CZr value of the factor Cadj, because a larger value of Cadj
½94 increases the collision and growth rate of the bubbles.
þ 22:985CZr
2
53:545CPu
This correlation was made by fitting the experimen-
tal data for the U–Zr alloys74–76 and for the U–Pu–Zr 100
alloys.77
Fractional fission gas release (%)
Incubation period
80
a: the hot press parameter for the open pore vol- U–10 wt% Zr
U–8 wt% Pu–10 wt% Zr
20
ume decrease (Section 3.23.4.2.2), U–19 wt% Pu –10 wt% Zr
esw1 : the threshold gas swelling for open pore for-
0
mation (Section 3.23.4.2.3), 0 2 4 6 8 10 12 14 16 18 20
f crack : the anisotropy factor in the crack model Peak burnup (at.%)
(Section 3.23.4.2.4), and Figure 28 Fission gas release data68 measured for 72%
e_ sol : the solid FP swelling rate (Section 3.23.4.2.5), smear density pins of the lead test assemblies.
Metal Fuel Performance Modeling and Simulation 747
On the other hand, the parameter mM has an influ- Table 7 Specification and irradiation condition70 of fuel
ence on the increasing rate of the gas release at the pins analyzed in the present study
early stage of irradiation. At this stage, the second Assembly No. X441a X425
term in the right-hand side of eqn [69] is dominant
compared to the other terms, and varies sensitively Cladding material HT9 HT9
Clad outer diameter (nm) 5.84 5.84
with a value of mM as expected from eqns [71] and
Clad wall thickness (mm) 0.38 0.38
[74]. As a result of the calculations by ALFUS Fuel alloy composition U–19Pu–10Zr U–19Pu–10Zr
for various values of Cadj and mM, Cadj ¼ 20 and (wt%)
mM ¼ 1.015 1014 were appropriate to simulate Slug outer diameter 4.67 4.32
the measured data presented in Figure 28. If values (mm)
Fuel slug length (mm) 343 343
or correlations of Dg, Di, M, and mi are determined
Smear density (%) 85 72
from theoretical and/or empirical considerations in Peak linear power 510 400
the future, the adjusted values of Cadj and mM should (W cm1)
be revised. Peak cladding temp ( C) 600 590
At the present stage, each of the parameter values a
Listed only for 85% smear density pin although X441 includes
and the property correlations assumed tentatively different smear density pins.
cannot be validated separately. In order to validate
the ALFUS code as a whole, therefore, the results of
the code calculation are compared with the experi- 100
mental data of the fuel irradiation behavior, which
Fractional fission gas release (%)
ALFUS calculation
3.23.4.3 Validation of ALFUS 40 Measured data68
U–10 wt% Zr
In order to validate ALFUS, calculations were per- 20 U–8 wt% Pu–10 wt% Zr
formed for the U–19Pu–10Zr (in wt%) fuel pins of U–19 wt% Pu–10 wt% Zr
the EBR-II test assemblies designated as X425 and
0
X441. The X425 assembly consisted of 72% smear 0 2 4 6 8 10 12 14 16 18 20
density pins and was irradiated until the cladding Peak burnup (at.%)
breach occurred at 19 at.%.50,70 The X441 assem- Figure 29 Calculated result of fractional fission gas
bly, which included different smear density pins (70, release for U–19Pu–10Zr (wt%) pin of X425 assembly:
75, 85%), was irradiated up to 10 at.% burnup.70 measured data68 are from 72% smear density pins of the
lead test assemblies.
The cladding material of both assemblies was the
low-swelling martensitic steel, HT9. The values
of the model parameters and fuel alloy properties for 72% smear density pins of the lead test assem-
discussed in Section 3.23.4.2 are used in all the blies designated X419, X420, and X421.68 Although the
following ALFUS calculation. Specifications and gas release data for the X425 pins have not been pub-
irradiation conditions of the fuel pins analyzed in lished, Pahl et al.70 reported that they are consistent
this section are summarized in Table 7. with the lead test data. The calculated curve reasonably
represents the trend of these measured data. The cal-
3.23.4.3.1 Fission gas release culated burnup of the gas release onset (the incubation
Fractional fission gas release calculated for the X425 period) is related to the threshold gas swelling esw
1 for
fuel pin is shown in Figure 29. Release of fission open pore formation and the gas swelling rate. The
gas starts at 0.7 at.% peak burnup, which corre- agreement between the calculated and test results as
sponds to the onset of interconnection of the closed shown in Figure 29 indicates that ALFUS calculates a
bubbles (open pore formation), and levels off at reasonable level of the gas swelling rate.
higher burnup. This leveling-off behavior of the gas
release is due to stable existence of the open pores 3.23.4.3.2 Axial elongation of the fuel slug
over higher burnup as indicated later (Figure 37). The slug axial elongation calculated for the same
Figure 29 also shows the measured gas release data X425 pin is shown in Figure 30, again in comparison
748 Metal Fuel Performance Modeling and Simulation
4 18.9 at.%
1.5
2
0
0 2 4 6 8 10 12 14 16 18 20
1.0
Peak burnup (at.%)
10.4 at.%
30 3.0
25 ALFUS calculation
Measured data44
Radial FCMI stress (MPa)
2.5
20
10
10.4 at.%
5 1.5
10 at.%
0
0 0.2 0.4 0.6 0.8 1
1.0
Bottom Axial position Top
(normalized by whole slug length)
5 at.%
although the calculation overpredicts the data at the 0
top part of the 15.8 at.% burnup pin. The results for 0 0.2 0.4 0.6 0.8 1.0
the X441 pins of 85% smear density are shown in Bottom Axial position Top
(normalized by whole slug length)
Figure 33, where large cladding strain was observed
for the 10 at.% burnup pins. The calculated curves Figure 33 Calculated results and measured data44 of
(thick solid lines) in Figure 33 generally agree cladding diametral strain at 5 and 10 at.% peak burnup
(X441, smear density 85%).
with the measured data although slight overpredictions
occur for both the top and bottom parts of the 10 at.%
burnup pins. Similar discrepancies between measured
and calculated results are also observed in the LIFE- 2.5
METAL code predictions,70 and can be attributed ALFUS calculation
Peak cladding diametral strain (%)
45 45
Smear density ~85% Smear density ~75%
40 40
Total
35 35
Swelling (%dV/V0)
30 30
Swelling (%dV/V0)
25 25 Op
en
por
Total 20 e
20
15 FP 15
Solid
Gas bubble
10 Gas bubble
10
5 FP
5 Solid
Open pore
0
0 0 2 4 6 8 10
0 2 4 6 8 10
(a) Peak burnup (at.%)
(a) Peak burnup (at.%)
80 8
80 8 Smear density ~75%
Smear density ~85%
70 7
70 7
60 6
50 5
FCMI stress (MPa)
50 5 um
en
40 Pl 4
40 4
30 3
30 3
um
en 20 2
Pl 2
20
10 FCMI 1
10 1
0 0
0 0 2 4 6 8 10
0
0 2 4 6 8 10 (b) Peak burnup (at.%)
(b) Peak burnup (at.%)
Figure 36 Calculated histories of (a) radially averaged
Figure 35 Calculated histories of (a) radially averaged swelling components and (b) FCMI stress and plenum
swelling components and (b) FCMI stress and plenum pressure at the axial position of 45% of the slug length from
pressure at the axial position of 45% of the slug length from the bottom (X441, smear density 75%).
the bottom (X441, smear density 85%).
increase in FCMI stress as indicated in Figure 35(b). In this case, although decrease in the open pore vol-
In the case of lower smear density (Figure 36), the ume accommodates the solid FP swelling until 13
open pore swelling amounts to more than 20%, and at.% burnup, the remaining volume of the open pores
can serve as a buffer against the solid FP swelling. becomes too small at the higher burnup. This leads to
Consequently, FCMI stress remains at a low level up significant increase in the FCMI stress as indicated in
to 10 at.% burnup (Figure 36(b)). Figure 37(b). This mechanism explains the increase in
the measured cladding strain data after 10 at.%
3.23.4.3.3.3 Effect of solid FP accumulation at burnup, which is shown in Figure 31.
higher burnup
At a very high burnup, significant FCMI can occur
even in the case of low smear density pins because of 3.23.5 Summary and Outlook
the accumulation of the solid FPs. This effect of the
solid FP swelling at higher burnup is illustrated in The constituent migration model and the simulation
Figure 37, which shows the calculated swelling his- codes LIFE-METAL and ALFUS are capable of
tory in the very high burnup pin of the X425 assembly. simulating metal fuel irradiation behavior to some
Metal Fuel Performance Modeling and Simulation 751
45
Smear density ~72%
Understanding of radial migration of rare-earth
40
fission products and FCCI.
Total In the future, the progress in the development of
35
the reprocessing of irradiated metal fuel may lead
30
Swelling (%dV/V0)
5 5
References
0 0
0 5 10 15 20
(b) Peak burnup (at.%)
1. Hofman, G. L.; Walters, L. C. In Material Science and
Technology, A Comprehensive Treatment; Cahn, R. W.,
Figure 37 Calculated histories of (a) radially averaged Haasen, P., Kramer, E. J., Eds.; VCH Verlagsgesellschaft:
swelling components and (b) FCMI stress and plenum Weinheim, Germany, 1994; Vol. 10A.
pressure at the axial position of 45% of the slug length from 2. Billone, M. C.; et al. In Proceedings of the International
Conference on Reliable Fuels for Liquid Metal Reactors,
the bottom (X425, smear density 72%).
Tucson, AZ, Sept 7–11, 1986; American Nuclear Society:
La Grange Park, IL.
3. Ogata, T.; Yokoo, T. Nucl. Technol. 1999, 128, 113.
4. Hwang, W.; et al. Nucl. Technol. 1998, 123, 130.
extent. Improvements in the models and codes are 5. Karahan, A.; Buongiorno, J. J. Nucl. Mater. 2010, 396,
recommended with respect to the following points: 283–293.
6. Kim, Y. S.; Hofman, G. L.; Hayes, S. L.; Yacout, A. M.
Incorporation of the swelling models for respec- J. Nucl. Mater. 2006, 359, 17.
tive annular zones formed in the fuel slug 7. Rest, J. J. Nucl. Mater. 1993, 207, 192.
8. Murphy, W. F.; Beck, W. N.; Brown, F. L.; Koprowski, B.;
corresponding to a-uranium zone, intermediate Neimark, L. A. Postirradiation examination of U–Pu–Zr
zone, and g-uranium zone. fuel elements irradiated in EBR-II to 4.5 atomic percent
Coupling of the fuel constituent migration model burnup, Report ANL-7602; Argonne National Laboratory,
1969.
and swelling models, then introducing them into a 9. Harbur, D. R.; Anderson, J. W.; Maraman, W. J. Studies on
comprehensive simulation code, after measure- the U–Pu–Zr alloy system for fast breeder reactor
ment of dependency of fuel alloy mechanical prop- application, Report LA-4512; Los Alamos Scientific
Laboratory, 1970.
erties on the composition. 10. Pahl, R. G.; Lahm, C. E.; Villareal, R.; Beck, W. N.;
Measurement of compressibility data of porous Hofman, G. L. Recent irradiation tests of
metal fuel alloy. uranium–plutonium–zirconium metal fuel elements.
In Proceedings of the International Conference on Reliable
Evaluation of the solid FP swelling rate based on Fuels for Liquid Metal Reactors, Tucson, AZ, Sept 7–11,
the analysis of the forms of FPs in the fuel alloy. 1986; pp 3–36.
752 Metal Fuel Performance Modeling and Simulation
11. Pahl, R. G.; Porter, D. L.; Lahm, C. E.; Hofman, G. L. 39. ANL-IFR-244. Integral Fast Reactor Program Annual
Metall. Trans. A 1990, 21A, 1863. Progress Reports FY 1993, Oct 1994.
12. Porter, D. L.; Lahm, C. E.; Pahl, R. G. Metall. Trans. A 1990, 40. Leibowitz, L.; Blomquist, R. A.; Pelton, A. D. J. Nucl.
21A, 1871. Mater. 1991, 184, 59–64.
13. O’Boyle, D. R.; Dwight, A. E. In Proceedings of the 41. Nakamura, K.; Ogata, T.; Kurata, M.; Yokoo, T.;
4th International Conference on Plutonium and Other Mignanelli, M. A. J. Nucl. Mater. 2002, 304, 63–72.
Actinides, Santa Fe, NM; Material Society: New York, 42. Turchi, Private communication. Lawrence Livermore
1970; p 720. National Laboratory, 2008.
14. Saunders, N.; Miodownik, A. P. In CALPHAD, Calculation 43. Kim, Y. S.; Hofman, G. L.; Yacout, A. M.; Turchi, P. In
of Phase Diagrams: A Comprehensive Guide; Cahn, R. W., Global 2009, Paris, France, Sept 6–11, 2009.
Ed.; Pergamon Materials Series; Pergamon: Oxford, 1998; 44. Billone, M. C. Unpublished work, Argonne National
Vol. 1. Laboratory, 1994.
15. Kurata, M. CALPHAD 1999, 23(3–4), 305–337. 45. Hofman, G. L.; et al. Metall. Trans. 1990, 21A, 517.
16. Andersson, J. O.; Helander, T.; Höglund, L.; Shi, P.; 46. Crawford, D. C.; Porter, D. L.; Hayes, S. L. J. Nucl. Mater.
Sundman, B. CALPHAD 2002, 26(2), 273. 2007, 371, 202.
17. Bauer, T. H.; Holland, J. W. Nucl. Technol. 1995, 110, 407. 47. Pahl, R. G.; Porter, D. L.; Crawford, D. C.; Walters, L. C.
18. Ogawa, T.; Iwai, T. J. Less-Common Metals 1991, 175, 59. J. Nucl. Mater. 1992, 188, 3.
19. Hofman, G. L.; Hayes, S. L.; Petri, M. C. J. Nucl. Mater. 48. Yacout, A. M.; Salvatores, S.; Orechwa, Y. Trans. Am.
1996, 227, 277. Nucl. Soc. 1993, 69, 193.
20. Ishida, M.; et al. Nucl. Technol. 1993, 104, 37. 49. Yang, W. S.; Yacout, A. M. Assessment of the
21. Marino, G. P. Nucl. Sci. Eng. 1972, 49, 93. SE2-ANL Code for EBR-II temperature predictions. In
22. Meyer, M. K.; Hayes, S. L.; Crawford, D. C.; Pahl, R. G.; Proceedings of the 7th International Meeting on Nuclear
Tsai, H. In Proceedings of ANS Conference on Accelerator Reactor Thermal Hydraulics (NURETH-7), Saratoga
Applications in the New Millennium, Reno, NV, Nov 11–15, Springs, NY, Sept 10–15; 1995; Vol. 3, p 2394.
2001. 50. Yacout, A. M.; Salvatores, S.; Orechwa, Y. Nucl. Technol.
23. Kim, Y. S.; Hofman, G. L.; Hayes, S. L.; Sohn, Y. H. J. Nucl. 1996, 113, 177.
Mater. 2004, 327, 27. 51. Yacout, A. M.; Yang, W. S.; Hofman, G. L.; Orechwa, Y.
24. Johansson, B. Phys. Rev. B 1975, 11, 1367. Nucl. Technol. 1996, 115, 61.
25. ANL-IFR-169. Integral Fast Reactor Program Annual 52. Hill, R. N. Private communication, Argonne National
Progress Reports FY 1991, June 1992. Laboratory, 2006.
26. Jankus, V. Z.; Weeks, R. W. Nucl. Eng. Des. 1972, 18, 83. 53. Ogata, T.; et al. J. Nucl. Mater. 1996, 230, 129.
27. Boltax, A.; et al. In International Fast Reactor Safety 54. Nakajima, T.; et al. Japan Atomic Energy Research
Meeting, Snowbird, UT, 1990; American Nuclear Society: Institute Report JAERI-M 9251, 1981.
Hinsdale, IL, 1990; Vol. II, pp 427–436. 55. Rashid, Y. R.; et al. Nucl. Eng. Des. 1974, 29, 1.
28. Liu, Y. Y.; Zawadzki, S.; Billone, M. C.; Nayak, U. P.; Roth, T. 56. Gruber, E. E.; Kramer, J. M. In 13th International
Development of LIFE-4CN. In Post-SMiRT5 Second Symposium on Radiation-Induced Changes in
International Seminar on Mathematical/Mechanical Microstructure (Part I), ASTM STP 955, 1987; p 432.
Modeling of Reactor Fuel Elements, Berlin, Germany, 57. McDeavitt, S. M. PhD Thesis, Purdue University, 1992.
Aug 20, 1979. 58. Tsuboi, Y.; et al. J. Nucl. Mater. 1992, 188, 312.
29. Chang, Y. I. Nucl. Technol. 1989, 88, 129. 59. Wood, M. H.; Matthews, J. R. J. Nucl. Mater. 1980, 91, 35.
30. Hughes, T. H.; Kramer, J. M. The FPIN2 Code – An 60. Matthews, J. R.; Wood, M. H. Nucl. Eng. Des 1980,
application of the finite element method to the analysis of 56, 439.
the transient response of oxide and metal fuel elements. 61. Brandes, E. A., Ed. Smithells Metals Reference Book,
In Proceedings of Conference on Science and Technology 6th ed.; Butterworth: London, 1983.
Fast Reactor Safety, Guernsey, UK, May 1986; 427. 62. Olander, D. R. Fundamental Aspects of Nuclear Reactor
31. Liu, Y. Y.; Tsai, H.; Billone, M. C.; Holland, J. W.; Fuel Elements, TID-26711-P1; US Technical Information
Kramer, J. M. J. Nucl. Mater. 1993, 204, 194. Center, US ERDA: Springfield, VA, 1976.
32. Kramer, J. M.; Liu, Y. Y.; Billone, M. C.; Tsai, H. C. J. Nucl. 63. Gruber, E. E.; Kramer, J. M. J. Am. Ceram. Soc. 1987,
Mater. 1993, 204, 203. 70(10), 699.
33. ANL-IFR-246. Integral Fast Reactor Program Annual 64. Hayns, M. R.; Wood, M. H. J. Nucl. Mater. 1977,
Progress Reports FY 1994, Dec 1994. 67, 155.
34. Yacout, A. M.; Tsuboi, Y.; Ueda, N. Development of 4s and 65. Barns, R. S. J. Nucl. Mater 1964, 11, 135.
related technologies(2): Long life metallic fuel. In 66. Beere, W.; Reynolds, G. L. J. Nucl. Mater. 1973, 47, 51.
Proceedings of ICAPP’09, Tokyo, Japan, May 10–14, 67. Tsai, H.; et al. In Proceedings of 3rd JSME/ASME Joint
2009; paper 9195. International Conference on Nuclear Engineering, Kyoto,
35. Jankus, V. Z.; Weeks, R. W. LIFE-I: A Fortran-IV Computer Japan, Apr 23–27, 1995; Vol. 2, p 849.
Code for the prediction of fast reactor fuel element 68. Pahl, R. G.; et al. In Proceedings of International Fast
behavior, USAEC Report ANL-7736; Argonne National Reactor Safety Meeting, Snowbird, UT, 1990, American
Laboratory, 1970. Nuclear Society: La Grange Park, IL, Vol. IV.
36. Tsai, H. Fuel/cladding compatibility in irradiated 69. Crawford, D. C.; et al. Trans. Am. Nucl. Soc. 1994, 71.
metallic fuel pins at elevated temperatures. In 70. Pahl, R. G.; et al. In Proceedings of International
International Fast Reactor Safety Meeting, Snowbird, UT, Conference on Fast Reactor and Related Fuel Cycles,
Aug 12–16, 1990. Kyoto, Japan, Oct 28–Nov 1, 1991; Atomic Energy Society
37. Yacout, A. M.; Hofman, G. L.; Lambert, J. D. B.; Kim, Y. S. of Japan: Tokyo, Japan.
Fuel behavior modeling issues associated with future 71. Croff, A. G. ORNL/TM-7175, Oak Ridge National
fast reactor systems. In GLOBAL 2007: Advanced Nuclear Laboratory, 1980.
Fuel Cycles and Systems, Boise, ID, Sept 9–13, 2007. 72. Cohen, A. B.; et al. J. Nucl. Mater. 1993, 204, 244.
38. Kim, Y. S.; Yacout, A. M.; Hofman, G. L.; Ryu, H. J. Trans. 73. Pahl, R. G.; et al. J. Nucl. Mater. 1993, 204, 141.
Am. Nucl. Soc. 2007, 96, 709. 74. Takahashi, Y.; et al. J. Nucl. Mater. 1988, 154, 141.
Metal Fuel Performance Modeling and Simulation 753
75. Argonne National Laboratory. Chemical Technology 78. Yacout, A. M.; Orechwa, Y. Trans. Am. Nucl. Soc. 1992,
Division Annual Technical Report for 1986, ANL-87–19, 65, 191.
1987. 79. Chin, B. A. In Topical Conference on Ferritic Alloys for Use
76. Touloukian, Y. S.; et al. Ed. Thermal Conductivity, Metallic in Nuclear Energy Technologies, Snowbird, UT, 1983;
Elements and Alloys; IFI/Plenum: New York, 1970; Vol. I. p 593.
77. Argonne National Laboratory. Reactor Development 80. Puigh, P. J. In 12th International Symposium on Effects of
Progress Report, ANL-7230,1969. Radiation on Materials, ASTM STP 870; 1985; p 7.
3.24 TRISO Fuel Performance Modeling and Simulation
K. Verfondern
Institute for Energy Research – Safety Research and Reactor Technology (IEF-6), Jülich, Germany
CP Coated particles
Abbreviations
CRP Coordinated Research Program
AVR Arbeitsgemeinschaft Versuchsreaktor
EDN Equivalent dido nickel
BISO Bi-structural
755
756 TRISO Fuel Performance Modeling and Simulation
FIMA Fissions per initial metal atoms Program. Since the 1970s, irradiated fuel elements
FZJ Forschungszentrum Jülich from the German high-temperature pebble bed reactor
HEU High-enriched uranium ‘‘Arbeitsgemeinschaft Versuchsreaktor’’ (AVR) and
HTGR High temperature gas cooled reactor from material test reactor (MTR) tests have been
IAEA International Atomic Energy Agency investigated in accident simulation tests.2,3 Similar
IPyC Inner pyrocarbon efforts have been taken in the United States and
JAEA Japan Atomic Energy Agency Japan with regard to their respective HTGR fuel
KMC Kernel migration coefficient designs. The experimental programs were accompa-
KORA Korrosionsapparatur (corrosion nied by the development of computer codes to help
apparatus) in the understanding of the physical phenomena in
LEU Low-enriched uranium fission product behavior and to provide the appropriate
LOFC Loss-of-forced-convection tools for its prediction in specific HTGR designs.
LTI Low temperature isotropic The source of radioactive materials is the large
MTR Material test reactor inventory of fission and activation products built up
OPyC Outer pyrocarbon in the reactor fuel. A typical HTGR power plant
PIE Post irradiation examination design generates thermal power up to 600 MW,
PyC Pyrocarbon or 2 1019 fissions per second. In a reactor’s
THTR Thorium-Hochtemperaturreaktor 40-year operational lifetime, 5 1028 fission prod-
TRISO Tri-structural uct atoms, along with a large inventory of activation
products, are created. Fortunately, most of the fission
products have very short half-lives and decay very
quickly (in terms of a fuel element, core residence
3.24.1 Introduction time of three years), or form stable oxide compounds
in the fuel kernel. In modern HTGRs, the principal
The ultimate goal of nuclear systems is to retain all containment barrier is the so-called ‘tri-structural’
radioactivities in all potential states of normal and (TRISO) coated particle, which has proven effective
abnormal operation. In most reactors, active safety during various irradiation tests and real-time experi-
systems inside the primary circuit accomplish this ments.4 It has been explained already in Chapter 3.07,
goal typically by means of redundant shutdown TRISO-Coated Particle Fuel Performance and
systems and decay heat-removal systems. With the Chapter 3.08, Advanced Concepts in TRISO Fuel.
introduction of small, modular-type, high-temperature Even under the best manufacturing conditions, a
gas-cooled reactors (HTGRs), this classical approach small fraction of coated fuel particles will be defec-
has been modified in that it now relies entirely on the tive, which together with contamination in the fuel
ability of coated particle fuel within the fuel element materials, are potential sources of fission product
to retain all key radionuclides during design-basis release into the coolant gas. Furthermore, under
accidents. The HTR-Modul proposed by Siemens abnormally high temperatures and power surges,
Interatom1 was designed such that a high degree of coated fuel particles may start to fail and release
safety was derived from the ability of the all-ceramic fission products. Condensable radionuclides released
fuel to retain their fission products under normal and from the core are plated out in the cooler regions of
accident conditions, the safe neutron physics behavior the main power system, causing radiation dangers
of the core, the chemical stability of the core, and the to operating personnel and to the public as well if
ability of the design to dissipate decay heat by natural released from the closed coolant circuit. It is there-
heat-transfer mechanisms – convection, radiation, and fore imperative to predict, with a reasonable level of
conduction – without reaching excessive temperatures. certainty, the fission product release from the fuel
The maximum fuel temperature reached within the elements and the plant under any expected operating
HTGR core would not exceed a value that results in or accident conditions.
irreparable damage to coated particle fuel. The fabrication and operation of HTGR fuel has
Demonstrating the capability of spherical fuel always been accompanied by intensive efforts of math-
elements to withstand a severe depressurization acci- ematical modeling of the performance of the fuel,
dent without any measurable loss of fission product was taking into consideration, as far as possible, the physical
a primary objective of the German Fuel Development phenomena that may occur. Modeling can assist both at
TRISO Fuel Performance Modeling and Simulation 757
the fabrication stage in identifying optimal fuel designs, 3.24.2 The HTGR Fuel Element Design
and at the irradiation stage in finding appropriate irra-
diation conditions or postcalculating the performance The basic fuel unit is the coated particle common to
of irradiated fuel, and finally at the accident simulation all HTGR designs. The particle of 1 mm diameter
stage in assessing the fuel and fission product release has a kernel to contain the fissile material protected
behavior under elevated temperature conditions. by a sequence of coating layers (Figure 1). The core
This chapter will first describe the general design of a modular-type reactor for 400–600 MW(th) will
of an HTGR fuel element and its components. In contain between 109 and 1010 individual fuel particles.
the first main section, the physical phenomena char- The particles are embedded in a graphite matrix
acterizing coated particle behavior and the underly- to form the fuel elements. Various designs have been
ing model approaches as well as the simulation of proposed according to different purposes and exten-
the particle behavior with computer codes and sively tested over the years. At present, the SiC
examples of calculations as part of the verification TRISO-coated fuel particle has been accepted by
and validation work will be explained. In the second most countries as the reference concept, but R&D is
part, the modeling of the fission product release continuing for further improvement and extension of
behavior from a TRISO-coated particle will be the fuel envelope.
described. This part will explain in detail the classi- The tasks of the single components of a coated
cal approach of diffusive transport through the vari- particle are as follows:
ous particle materials. It will be completed by a
description of alternative methods of fission product The spherical fuel particle consists of a dense heavy
release prediction, including some special cases such metal oxide or carbide spherical kernel. It represents
as release from the particle kernel or fission gas and a first partial barrier to radionuclide release by
iodine release. immobilizing many of the fission products and
5
Graphite shell 4
6
Graphite matrix
Coated fuel 4
60 mm particle
Coating
Kernel Particle 1
delaying the diffusive release of others, and also coming into contact with one another, thereby dam-
substantially reducing release from the particle due aging their coatings during the pressing of the fuel
to radioactive decay before release from the kernel. spheres or compacts. With TRISO particles, the
The first layer in contact with the kernel is the overcoating is around 200 mm thick.
buffer layer, a porous layer with 50% of theo- There are two major designs of a fuel element.
retical density 2.2 Mg m3. The purpose of the A spherical fuel element for the pebble-bed HTGR
buffer layer is to provide void volume for gaseous is formed by pressing the overcoated particles with
fission products to limit pressure buildup within matrix material into a 50 mm-diameter inner sphere
the coated particle. Additionally, it serves to containing between 9,000 and 17,000 coated particles.
decouple the kernel from the inner pyrocarbon The inner fuel sphere is then placed within a protec-
layer to accommodate kernel swelling mechani- tive 5-mm thick layer of matrix material. In the block-
cally, thereby reducing the buildup of stress in type reactor, the coated particles are pressed into
the outer coating layers during irradiation. The cylindrical or annular fuel compacts inserted in a
buffer layer also represents a sacrificial layer for hexagonal graphite block. The fuel block contains
absorbing energetic fission products recoiling from bore holes that are, in the US design, either coolant
the kernel surface, thus protecting the inner pyr- channels or filled with the fuel compacts. In the Japa-
ocarbon layers. nese design, the bore holes of a fuel assembly are filled
The inner high-density, isotropic layer of with fuel rods (‘pin-in-block’) to contain the compacts,
pyrolytic carbon (IPyC), with an average density while the coolant flows through the annular gap
of 1.9 Mg m3, forms the first load-bearing bar- between the rod and inner surface of the bore hole.
rier against the pressure exerted by fission pro- Just by its huge mass in an HTGR, the graphitic
ducts within the fuel kernel and buffer layer, material (Figure 2) represents an effective sink for
thereby reducing the pressure on the SiC layer. metallic fission products (cesium, strontium, and less
Although an intact IPyC layer forms a practically for silver) escaping the fuel particles. On the other
impenetrable barrier for fission gases and iodine, it hand, under the abnormal operating conditions of an
becomes increasingly pervious to palladium, cesium, ingress of oxidizing media (air and water) into the
silver, and strontium at higher temperatures. active core, corrosive reactions may lead to a partial
The purpose of the silicon carbide (SiC) layer with degradation of core structures.
a density of 3.20 Mg m3 density is to prevent the A set of standards for modern high-quality HTGR
release of fission products into the graphite matrix, fuel has been defined in terms of low levels of heavy
and then into the reactor helium stream. The SiC metal contamination, manufacture-induced particle
layer thus acts as a pressure-retaining containment defects during fuel body and fuel element making,
and as the principal metallic fission product reten- irradiation/accident-induced particle failures, and
tion barrier in the coated particle. limits on fission product release from intact particles.4
The outer high-density pyrocarbon layer (OPyC)
is a further diffusion barrier for gaseous and metal-
lic fission products. It also serves as a protective
layer for the SiC during particle handling and
sphere/compact formation, and provides bonding
surface for the overcoating described subsequently.
By mass
During irradiation, IPyC and OPyC layers shrink at
Matrix 93.6%
first, possibly expanding again if sufficiently high UO2 3.3%
fast-neutron dose levels are reached. The interac- Inner PyC 0.6%
tion between the IPyC and OPyC high-density Outer PyC 0.9%
pyrocarbon layers, and the SiC layer sandwiched SiC 1.1%
between them, plays an important role in keeping Buffer 0.5%
the SiC layer under compression during irradiation.
Frequency
100 %
50 %
Fuel elements
0 with 0,1,2,3,...
0 1 2 3 4 ... 0 1 2 3 4 5 6 7 ... 0 1 2 3 ... SiC-defects
AVR 19 AVR 21 AVR 21–2
Figure 3 Coated particle defects during sphere production 1981–1985.
With regard to manufacture-induced defects, a total of irradiation tests with spherical fuel elements (i.e., with-
102 defective particles detected out of the 3.3 million out the tests with loose particles) comprising 276 680
particles examined, resulted in an expected (mean) particles. Taking into account the different burnup
value of 30 106. The design value to be guaranteed categories, the derived failure fractions at the 95%
by the manufacturer was fixed at 60 106. For the confidence level are 1.3 105 for particles with 5%
German reference fuel element for the HTR-Modul, FIMA and 4.6 105 for particles with 10% FIMA.
this value corresponds to 0.7 defective particles per fuel
sphere or two defective particles in three fuel elements.
The quality in the manufacture of the late (‘modern’) 3.24.3 The Modeling of TRISO-Coated
German fuel during the years 1981–1985 is shown Particle Performance During Reactor
in Figure 3 expressed as the number of particle defects Operation and Under Accident
during sphere manufacture.5,6 Conditions
A respective irradiation-induced particle failure
3.24.3.1 Physical Phenomena in Coated
fraction was derived by Ragoß5 based on five irradia-
Particle Behavior
tion experiments with fuel from a production in 1981,
covering a total of 212 000 UO2 TRISO-coated Temperature, fast fluence, and burnup trajectories are
particles. None of them failed during the tests all strongly influencing parameters of fuel performance.
corresponding to an expected (mean) value of In the following sections, the major failure mechanisms
zero. A statistical treatment led to a failure fraction for coated particles and other phenomena affecting
of <2 105 at a 95% probability. A design value coated particle behavior will be described.7 Perfor-
for the irradiation-induced failure fraction was ‘arbi- mance models for many of these phenomena have
trarily’ chosen by Interatom to be higher by a been developed and correlated with experimental data.
factor of 10, that is, 2 104 at the target burnup of
9% FIMA for the HTR-Modul.5 It should be noted 3.24.3.1.1 Pressure buildup from carbon
that about half of the particles reached a burnup monoxide formation
above the target value; the rest of the particles ended With stoichiometric oxide fuel particles, carbon mon-
up with a burnup between 7.5% and 8% FIMA. oxide is produced from excess oxygen liberated upon
With further irradiation tests from Phase II, the fissioning of the heavy metal because the fission
total number of investigated coated particles products in the aggregate are thermochemically inca-
increased to 357 500, and still no particle failure pable of binding all of the oxygen. Each 235U fission
was observed during irradiation. In a IAEA report,4 a within UO2 leads to fission products that, at maxi-
statistical evaluation is given based on all the above mum, may combine with only 1.62 of the two
760 TRISO Fuel Performance Modeling and Simulation
oxygen atoms released. The remaining oxygen com- During the heating (accident) phase, a modification
bines with other materials, with the oxidation of of the above equation is applied, taking account of an
carbon to CO as the favored reaction. Oxycarbide additional dependence on the heating temperature:8
fuel is designed such that UC2 is converted to UO2 8500
from the reaction with the liberated oxygen. log Of ¼ 10:08 þ 2 log tB 4040
TB
The suppression of excessive CO formation dur- ½1b
ing the irradiation of oxide-based fuel particles is an 1 1
extremely important particle design objective, espe- T TB þ 75
cially for higher burnups. Consequently, the thermo- with an upper limit fixed at
chemical calculations presented earlier need to be
confirmed experimentally by measuring the CO Of 0:625 ½1c
contents of irradiated fuel particles over a range of The addition of 75 K to the irradiation temperature TB
burnups. accounts for the fact that TB is defined as a surface
Proksch conducted such measurements on irra- temperature, whereas the temperature relevant for the
diated UO2 8 and (Th,U)O2 particles,9 which gener- release is about 75 K higher. The dependence on the
ally confirmed the thermochemical calculations. The heating/accident temperature T (although, no further
results were found to be strongly dependent on the CO is being produced under accident conditions)
irradiation history, and to a great extent, on the type accounts for the observed fact that under elevated
of kernel. From experiments with UO2 kernels cover- temperature conditions, part of the CO stored in the
ing an irradiation time up to 550 equivalent full kernel is being released into the free volume of the
power days and a temperature range between 950 particle (but will not be readsorbed in the case of
and 1525 C, the following relationship for the num- decreasing heating/accident temperatures).
ber of liberated oxygen atoms per fission, Of, was Studies on the gas behavior in the fuel, including
derived (Figure 4): carbon monoxide, will be conducted within the
8500 framework of the AGR fuel development and quali-
log Of ¼ 10:08 þ 2 log tB ½1a fication program at the Idaho National Laboratory.10
TB
Furthermore, Russia will concentrate on the mea-
where TB is the irradiation temperature (K) and tB is surements of the CO inventories of high-burnup,
the irradiation time (s). substoichiometric PuO2 x particles.
0.7
Number of oxygen atoms per fission
ays
UO2
ays
000 d
0.5
00 d
ays
s
day
00 d
⬚C, 1
, 10
ys
500
da
~11 C, 10
0 ⬚C
1100
s
00
ay
⬚C,
100
d
,5
⬚
00
00
900
⬚C
0.3 ,5
00
0⬚
C
10
90
0.1 (Th,U)O2
1.0E – 08
1.0E – 09
1.0E – 10
1.0E – 1 1
4.0 5.0 6.0 7.0 8.0 9.0
Inverse temperature (104 / T(K))
Figure 5 Kernel migration coefficient data from the testing of UO2 fuels.
be described, in its general form, according to the The elastic modulus of SiC is much larger than that
following Weibull equation: of PyC, and the dimensional changes and creep strains
in SiC are comparatively smaller than for PyC. The
Fp ðt ; T Þ ¼ 1 expfln2ðst =s0 Þm g ½4a
modeling requires a knowledge of the properties of all
where s0 is the SiC tensile strength (Pa) which is the materials involved and their change by neutron fluence.
stress corresponding to a 50% failure probability, Figure 7 shows an example for the evolu-
st is the stress induced in the SiC layer due to the tion of stresses in the three coating layers. Initially,
internal gas pressure (Pa), and m is the Weibull stresses in the SiC layer are compressive because
parameter. of the attempted shrinkage of the two PyC layers,
The reference approach in the US modeling also but as the gas pressure increases, they become
combines stresses in a spherical SiC shell with a tensile until failure occurs at the fracture stress of
Weibull failure distribution, but is more of empirical 650 MPa. Material properties assumed for the two
character to approximate the failure probability by: PyC layers are the same. However, while stresses
in the OPyC vary little during the course of the
Fp ðt ; T Þ ¼ 1 expf½ða þ bT ÞTfbn 5 g ½4b
irradiation, they will decrease in the IPyC layer
where a (K1), b (K2), n are constants, and fb is Bu/26 because of an increasing compressive component
with burnup Bu in [% FIMA]. The constants are from increasing gas pressure inside the particle.11
defined for fissile and fertile particles and those
with missing buffer or failed OPyC layer.18 3.24.3.1.5 Fission product attack
In a more sophisticated modeling approach, Past irradiation experiments indicate that fission
the mechanical performance of a coated particle is products, in particular, noble metals, such as Ru, Rh,
assessed by calculating the stresses within each of the Pd, and Ag, can be transported from the kernel to the
three load-bearing layers (IPyC, SiC, and OPyC) to inner surface of the SiC, where they interact and may
determine whether or not fracture stresses of the damage or even fail the SiC layer. In older uranium
single layers will be exceeded during irradiation.11 carbide kernels, rare earth fission product migration
The stresses induced by the internal gas pressure was of concern, while in UO2 kernels, palladium and
onto the SiC in the simple model are reduced if the some of the rare earths may be an issue. In UCO
presence of the two PyC sandwich layers is consid- kernels, the mobility of fission products is usually
ered: the reason being the shrinkage of pyrocarbon limited due to the generation of either carbide or
under neutron irradiation. The effect of irradiation oxide compounds. However, palladium transport has
creep, however, is working against these stresses. still been observed in UCO-coated particle fuel.
Burnup (%)
0 2 4 6 8 10 12 14 16
500
400
Tangential stress (MPa)
300
SiC layer
200 failed
100
0
-100 IPyC
SiC
-200 OPyC
-300
-400
0 10 20 30 40 50 60 70 80 90
Neutron dose (energies > 0.1 MeV) ( × 1024 neutrons m-2)
Figure 7 Example of stresses in the three load-bearing layers during irradiation. Reproduced from Martin, D. G. Nucl. Eng.
Des. 2002, 213, 241–258.
764 TRISO Fuel Performance Modeling and Simulation
Although the quantity of Pd is small compared with where d0 is the initial thickness of the SiC layer (m),
the mass of the SiC layer, the reaction is highly t is the time (s), _ is the volume corrosion rate
localized and may lead to a complete penetration of according to Montgomery20 (Figure 8),
the SiC layer for longer times at high temperatures.
7 179500
Also, silver has been observed to migrate through _ ¼ 5:87 10 exp ½ms1 ½6
RT
apparently intact particles and be released into the
reactor coolant system. The migration of the fission R is the gas constant, R ¼ 8.3143 J mol1 K1, and T is
products is thought to be a function of time at temper- the temperature (K).
ature and burnup as well as temperature gradient,
which means that it may play a more important role
3.24.3.1.6 Thermal decomposition of the
in prismatic reactors. In LEU fuels, the concentration of
SiC layer
Ag and Pd is much greater than in HEU fuel of similar
At temperatures above 2000 C, silicon carbide will
burnups due to higher yields from plutonium fissions.
decompose into its constituents.21 Particle failure
CO generated during irradiation of UO2 kernels
resulting from thermal decomposition of the silicon
can attack the SiC layer at temperatures above
carbide is determined by applying a semiempirical
1400 C if the IPyC layer is either permeable or
relationship:
cracked. Microprobe examinations confirmed ther-
mochemical calculations, according to which silicon Fd ðt ; T Þ ¼ 1 expfaxb g ½7
is transported to the kernel in the form of SiO gas,
which then reacts with the fission products, thereby where the constants a and b have been empirically
making the coating retention ineffective.19 derived from experimental data:
The corrosive attack on the SiC by fission pro-
ducts is interpreted in the modeling by a thinning a ¼ ln 2 and b ¼ 0.88 for loose fuel particles
rate of the SiC layer. For small thicknesses (‘thin a ¼ 0.0001 and b ¼ 4 for particles embedded in a
shell’), it is: fuel sphere
d0 x is a so-called ‘action integral,’ which includes the
_ Þ
d ¼ d0 ð1 t ½5 temperature-time history of the particles:
ð1 þ t_ =d0 Þ
Tirr = 1050–1200 ⬚C
Tirr = 800–915 ⬚C
SiC corrosion rate (s-1)
10-6
Tirr = 650–750 ⬚C
10-8
10-10
4 5 6 7
104 / T a (K-1)
10−2
Tirr = 1050–1200 ⬚C
Tirr = 800–915 ⬚C
10−4
SiC decomposition rate (s−1)
Tirr = 650–750 ⬚C
10−6
Benz lab. data
10−8
10−10
3 4 5 6
104 / Ta (K−1)
soon as the stress induced in the SiC layer by the 650 650
log Gs ¼ 0:556 þ and log Gm ¼ 0:394 þ ½11
internal gas pressure exceeds the tensile strength of TB TB
the SiC. The strength is a material parameter whose
mean value and statistical Weibull distribution can be The stress induced in the SiC layer, st, is determined
derived, for example, from SiC ring crack tests.17 with the following equation valid for a ‘thin shell’ or
Both tensile strength and Weibull modulus are ‘soap bubble’ pressure vessel:
dependent on the fast neutron fluence. The following rp rp
st ¼ ¼ ð1 þ t _ =d0 Þ½Pa ½12
relationship (with lower limits given) has been 2d 2d0
derived from experimental data on a single German where r is the mean radius of the SiC layer (m), d is
particle batch (which was denoted EO 1607), where the thickness of the SiC layer (m), and p is the gas
strength and Weibull modulus were measured before pressure inside the particle (Pa).
and after irradiation (Figure 10): The internal gas pressure, p, of the aforementioned
equation is calculated by applying the ideal gas law to
s0 ¼ s00 ð1 G=Gs Þ
196MPa ½10a the generation of fission gases Xe, Kr, and reaction
gas CO. The amount depends on various parameters
m0 ¼ m00 ð1 G=Gm Þ
2 ½10b given by particle properties and operating conditions:
where s00 is the SiC tensile strength or the stress, at ðFd Ff þ Of ÞBu RT
p¼ ½Pa ½13
which there is a 50% failure probability, in the unir- Vf =Vk Vm
radiated stage (Pa), m00 is the Weibull parameter for
where Fd is the release fraction of fission gases (Xe
the unirradiated stage, and G, Gs, Gm are fast neutron
and Kr) from the kernel into the void volume23
fluences (n m2 EDN, 1 n m2 EDN ¼ 1.52 n m2
E > 0.18 MeV or 29 fJ ¼ 1.67 n m2 E > 0.1 MeV or ðti þ ta Þf ðti þ ta Þ ta f ðta Þ
Fd ¼ ½14a
16 fJ) with ti
1.0
0.9
0.8
0.7
Cumulative frequency
0.6
0.5
0.4
EO1607
irradiated to
0.3 2.5 ´ 1025 m-2
at 116z ⬚C
0.2 EO1607
unirradiated
0.1
0.0
0 100 200 300 400 500 600 700 800 900 1000 1100 1200
Strength (MPa)
Figure 10 Tensile strength measurements of SiC specimens taken from the German particle batch denoted EO 1607.
Unirradiated: s00 ¼ 834 MPa, m00 ¼ 8.02; Irradiated: s0 ¼ 687 MPa, m0 ¼ 5.98.
TRISO Fuel Performance Modeling and Simulation 767
with for the time interval ( C). The initial values will be zero,
1
X t1 ¼ Fp;d ðt1 Þ ¼ 0. This procedure also allows the treat-
6 expfn2 p2 tg
f ðtÞ ¼ 1 1 ½14b ment of temperature transients.
t n¼1 n4 p4
and 3.24.3.2.2 Other approaches
0 0 In a currently ongoing IAEA project, CRP-6, follow-
ti ¼ D ðTB ÞtB and ta ¼ D ðT Þt ½14c
ing the previously mentioned CRP-2, the enormous
where D0 is the reduced diffusion coefficient of the progress in terms of modeling fuel performance under
fission gases (s1), TB is the irradiation temperature normal operating and accident conditions is being
(K), T is the heating temperature (K), tB is the dealt with.24 Most of these more recent modeling
irradiation time (s), t is the heating time (s), Ff is approaches to describe fuel performance during nor-
the yield of stable fission gases, Ff ¼ 0.31, Of is the mal operation and under accident conditions are more
number of oxygen atoms per fission (see below), sophisticated in that they simulate the complete
Bu is the heavy metal burnup [FIMA], Vf is the TRISO-coated particle and include the analysis of
void volume (m3), typically 50% of the buffer vol- mechanical and thermal stresses acting inside the par-
ume, Vk is the kernel volume (m3), and Vm is the ticle coating. The result of the codes’ calculations is
molar volume of the heavy metal in the kernel the quantification of the particle failure probability
(m3 mol1). It is for with most codes, including the subsequent simulation
UO2 : Vm ¼ 0:2672ðkgmol1 Þ=10960ðkgm3 Þ of radionuclide transport and release.
Stress and strain analyses in each of the three
¼ 2:43796 105 load-bearing layers are conducted by the application
of displacement equations calculating the overall
UC0:29 O1:63 : Vm ¼ 0:2682ðkgmol1 Þ=10700ðkgm3 Þ
strain change in tangential and radial directions,
¼ 2:50654 105 compatibility equations relating the radial displace-
ments to strains, and the equation of equilibrium
In case of a transient irradiation temperature his-
between tangential and radial stresses. Results can
tory, the CO production term Of contributing to the
be compared with debonding stresses to judge on
overall pressure inside the particle, will be stepwise
the potential formation of gaps between layers.
integrated (in analogy to the ‘free fall’ relationship):
Thermal submodels consider heat conduction
1
Of ¼ gðTB ÞtB2 ½15 through the various particle material zones, taking
2 into account porosities and whether or not gaps from
The final Of value obtained at the end of the irradiation debonding layers have evolved in the course of neutron
time can be taken to derive, in reverse, a fictive average irradiation. The result is a temperature distribution
irradiation temperature TB using the previously inside the particle. Mechanical submodels determine
described Of relationship. TB then corresponds to that the distribution of stresses, strains, and displacements
temperature, which would result in the same Of value, if resulting from the thermal loads, pressure loads from
it were kept constant over the same irradiation time. the inventories of gaseous species, or mechanical loads
8500 given by kernel swelling or irradiation-induced dimen-
TB ¼ ½K ½16
10:08 logðOf =tB2 Þ sional changes of the pyrocarbon layers.
The overall failure fraction, Ftotal, is determined by a However, for two reasons, not all particles will fail
combination of the two single and independent failure at the same burnup value. First, because the layers are
probabilities, Fp and Fd, according to: considered to be brittle materials, their fracture stres-
ses will be variable from particle to particle. This
Ftotal ¼ 1 ð1 Fp Þð1 Fd Þ ½17 implies that, as the tangential stress increases during
The failure fractions Fp,d are typically functions of time irradiation, so also will the fraction which has failed.
and temperature. If a time increment is defined, The second reason is due to the statistical variations in
t ¼ t2 t1 , the accumulated failure fraction at time the characteristics of individual particles. Therefore, the
t2 is given by models typically apply statistical methods, for example,
Monte Carlo, to account for distributions in geome-
Fp;d ðt2 Þ ¼ Fp;d ðt1 Þ þ Fp;d ðt2 ; Tm Þ Fp;d ðt1 ; Tm Þ ½18
try, material properties, and stresses. The numerical
where t1, and t2 are the beginning and the end of the solution methods used are either finite difference or
time increment (s) and Tm is the average temperature finite element schemes.
768 TRISO Fuel Performance Modeling and Simulation
100
10-1
10-2
Fractional release krypton
Tirr = 1070 ⬚C
10-3
10-4
*** Tirr = 980 ⬚C
***** * * * *
10-5
Heating temperature 2000
Temperature (⬚C)
10-6 1500
* * * * ** 1000
10-7
* * Experimental data krypton 500
10-8 * 0
0 50 100 150 200 250 300 350 400 450 500
Time (h)
Figure 11 PANAMA postcalculation of gas release in heating experiment FRJ2-K13/4 with a fuel sphere heated at 1600
and 1800 C.
TRISO Fuel Performance Modeling and Simulation 769
100
Experimental data 85/18
10-1 X Experimental data 90/2 Irradiation temperature
* Experimental data 90/5
1000 ⬚C
Fractional release 10-2
10-3
820 ⬚C
10-4 XXXX
XXX X X X X X X X X
Temperature (⬚C)
2000
10-6 Heating temperature 1500
X
* * 1000
10-7 X
** *
X * *
* 500
X
10-8 * 0
0 50 100 150 200 250 300 350
Time (h)
Figure 12 PANAMA postcalculation of gas release in heating experiments AVR 85/18, 90/2, and 90/5 with the fuel spheres
heated according to the transient temperature predicted for the LOFC case for the HTR-Modul.
core or further away. The PANAMA calculations irradiation temperatures, and two sets of SiC strength
are definitely not a proof that the fuel spheres con- data, the totally eight predictive calculations for the
sidered have taken a certain path through the AVR failure fractions ranges between 4 108 and 0.43, thus
core, but at least they would be in agreement with showing tremendous uncertainty margins. The calcu-
the respective estimations of the irradiation tem- lation of some 107 random particles with the French
peratures of 725 C (of spheres near a buttress) and ATLAS code, using a finite element method to calcu-
842 C (of spheres in the middle between buttresses), late the thermal and mechanical performance in con-
respectively.25 nection with a Monte Carlo method, resulted in the
A third example of fuel performance calcula- assessment of a first particle to fail when the burnup
tions is shown in Figure 1326 describing a (real) pre- reached approximately 18% FIMA. Further work will
diction for the irradiation experiment HFR-EU1,27 be necessary (and is planned within the national and
a currently conducted test of fuel spheres to be irra- IAEA programs) to resolve these differences. In practi-
diated to very high burnups. The figure includes the cal terms, however, this is of no concern when fuels are
predicted particle failure fraction as a function of irradiated to 10% FIMA and no irradiation-induced
burnup (or irradiation time) applying different com- failures are predicted or observed.
puter models.24
The results exhibit an excessively broad range:
The German code PANAMA predicts the first 3.24.4 The Behavior of Fission
particle to fail (which is about equivalent to reach- Product Release from TRISO-Coated
ing a failure fraction level of 104) at a burnup Particle Fuel
between 14% FIMA (Tirr ¼ 1100 C) and 20%
3.24.4.1 Safety Relevance
FIMA (Tirr ¼ 950 C). The Russian GOLT-V1 results,
revealing a strong dependence on material properties, The transport of any metallic fission or activation
show the first particle failure between 14 and 16% product could be calculated if the transport para-
FIMA. Calculations with the UK code STRESS3 in meters of the nuclide were available. But it is imprac-
connection with the statistical code STAPLE result tical (and unnecessary) to evaluate the production,
in a failure fraction exceeding the level of 104 (or one transport, and release of all species generated in a
failed particle) near 14% FIMA. The Korean code nuclear reactor. Therefore, only the radiologically
COPA predicts the same result for the first particle to most significant nuclides are analyzed. Key radionu-
fail, but then calculates a lower failure probability for clides are selected based on the combination of their
higher burnups. With the US model PARFUME fission yield, their transport and release properties,
applying two options of calculating CO pressure, two and their radiological hazard level.
770 TRISO Fuel Performance Modeling and Simulation
1.E+00
GOLT-V1 (Russia)
1.E–02
ATLAS (France)
PANAMA (Germany)
1.E–04
Failure fraction
1.E–05
1.E–06
1.E–07
1.E–08
1.E–09
1.E–10
0 5 10 15 20
Burnup (% FIMA)
Figure 13 Prediction of particle failure probability during the irradiation experiment HFR-EU1 as a function of burnup with
various fuel performance codes.
The radiologically most relevant radioisotopes 900 C, plates out on metallic surfaces at tempera-
monitored during accident simulation testing are tures of 800 C, and diffuses into the bulk metal,
listed in Table 1. These include the long-lived stron- where it is effectively captured. As a result, silver
tium (90Sr), silver (110mAg), cesium (134Cs, 137Cs), and does not present an important concern with regard
krypton (85Kr) activation and fission products, and to offsite dose, but can be a dominant contributor
the short-lived iodine (131I) and xenon (133Xe) fission to occupational dose depending on component
products. Isotopes of the same chemical species are maintenance requirements.
assumed to result in the same release fractions. More Strontium is retained to a large extent in oxide
nuclides may be added to the earlier list depending kernels during normal operation even when coat-
on the expected fuel conditions. ings are defective and is slowly released at tem-
The relative characteristics of the metallic fission peratures beyond modular HTGR accident
products of radiological interest for LEU UO2 conditions (i.e., approaching 1800 C for several
TRISO fuel under HTGR service conditions, are as days). Additionally, it is strongly adsorbed in
follows: matrix material and little is released from spherical
fuel elements. Different from other species, it is
Silver is known to begin to be released from difficult to measure, and the results shown have an
TRISO fuel at temperatures above 1000 C of uncertainty of an order of magnitude compared to
normal operating temperatures, and to be released around 10% for others.
from intact particles in significant fractions after Cesium is also released from particles with defec-
several days at 1600 C. It deposits in graphite at tive silicon carbide layers in normal operation and
TRISO Fuel Performance Modeling and Simulation 771
can be released from the fuel spheres depending on extremely slow, and release fractions from intact coated
the local conditions. However, there is a significant particles are negligible. Iodine is considered to show a
delay in release from the sphere due to a holdup in similar transport behavior as the fission gases.
the matrix. In the AVR, the LEU UO2 TRISO Even under the best manufacturing conditions, a
spheres were typically net absorbers of cesium small fraction of coated fuel particles will be defec-
due to the low silicon carbide defect fractions. tive. Furthermore, under operating conditions and
also under abnormally high temperatures, there is
The primary isotopes of cesium of interest are 134Cs
a certain chance for coated fuel particles to fail.
and 137Cs, whose inventories developed differently as
In most models, a defective/failed particle is repre-
a function of burnup, but whose release fractions
sented conservatively by a bare fuel kernel. It is
under accident conditions are very similar. The com-
assumed that fission products from failed and defec-
bination of 85Kr and 137Cs data in the early phase of
tive particles do not have to travel through coating
the heating tests allows the determination of both
layers and only have to escape the kernel to be
through-coating failure fractions and failure fractions
released to the matrix material.
of particles with silicon carbide defects and an intact
Particles with defective or missing coating layers,
pyrocarbon layer.
particularly SiC, can be modeled by a modification of
the diffusion coefficients.
3.24.4.2 Fission Product Transport
Phenomena 3.24.4.2.2 Inventory buildup
A wide spectrum of fission product nuclides is created
The schematic in Figure 14 summarizes the most by the fission of uranium and plutonium resulting from
important physical phenomena that occur in an successive b-particle emissions from the fission frag-
HTGR fuel element regarding the transport and ments. Most have very short half-lives or form stable
release of fission products. oxide compounds in the fuel kernel. The inventory
buildup of fission products during reactor operation
3.24.4.2.1 Types of particles is more or less linear for long-lived isotopes, whereas
In fully intact TRISO-coated particles, fission products short-lived isotopes soon reach an equilibrium state.
that are formed have to be transported through all the Activation products such as 134Cs and 110mAg are cre-
coating layers before being diffused through the matrix ated when a fission product captures a neutron to form
material and released into the coolant. Different from a new nuclide. Some of these activation products,
metallic fission products, the diffusion parameters whose inventory buildup is different from fission pro-
for fission gases in PyC are very small. Therefore, ducts, are very important for HTGR maintenance and
the transport rates of gaseous fission products are safety. For example, the 110mAg isotope is the product
772 TRISO Fuel Performance Modeling and Simulation
Pore
D1 D
2
Figure 14 Fission product transport and release phenomena. Cp, Coated particles.
of an (n,g)-reaction of the stable 109Ag whose decay Table 2 Uranium inventories outside the fuel kernels
chain by b emissions is shown below:
Uranium outside fuel kernel Inventory fraction
b 109 b 109 b 109 b 109 b
40 Zr ! 41 Nb ! 42 Mo ! 43 Tc ! 44 Ru !
109
U in buffer 1.0 103
U in IPyC 1.0 104
109 b 109 b 109 ðn;gÞ 110m b 110 U in SiC 1.0 106
45 Rh ! 46 Pd ! 47 Ag ! 47 Ag ! 48 Cd
U in OPyC 1.0 106
Due to decay effects on fission product transport U in matrix graphite 1.0 107
and the ability to diffuse through the layers of a coated
fuel particle, radiologically important fission products
released from fuel elements are divided into relatively typical to the German reference fuel element and
short-lived gaseous fission products and long-lived have been used in the calculations. The complement
metallic fission products. With regard to the silver to 100% of the sum of the above fractions is set as the
transport through a TRISO coating, which is compar- initial kernel inventory.
atively fast at temperatures above 1000 C, a part of the In a study conducted by Interatom,5,28 evaluating
silver may be released from the fuel in the form of the burn-leach tests on as-manufactured spherical fuel
stable (inactive) 109Ag and no longer be subjected to an elements with modern LEU TRISO particles, the
activation to 110mAg. Different theoretical models and fraction of free uranium, Ufree/Utot, that is, uranium
sets of transport parameters (e.g., diffusion coefficients not covered by an intact SiC coating, was generally
and correlations) describe the behavior of both fission found to be either a value near the detection limit
product groups. (typically 1 106), or a multiple of the uranium
inventory of a single-coated particle. These measure-
3.24.4.2.3 Initial distribution in accidents ments indicate that in most cases, the measured free
The initial distribution of the specific fission product uranium fraction is due to the fabrication-induced
nuclide considered is equivalent to the initial fissile defective particles, whereas the quasihomogeneously
uranium distribution. By far, most of the fissile mate- distributed contamination in the matrix material is
rial is concentrated in the particle kernels surrounded comparatively small.
by an intact coating. The particles are homogeneously The natural contamination of the matrix material
dispersed in the fuel zone of the spherical fuel ele- with uranium exists in traces only close to the detection
ment. However, there are small fractions of uranium limit of 106; the fraction of the fissile material assum-
outside the kernel in the coating and matrix graphite ing a 10% enrichment is, thus, in the order of 107.
due to the manufacturing process or natural contam- All inventories are assumed to be homogeneously
ination, respectively. The data given in Table 2 are distributed within the respective material zones.
TRISO Fuel Performance Modeling and Simulation 773
3.24.4.2.4 Recoil R is the radius (m) of the kernel (or the coated
Due to their high kinetic energy, fission products particle or the graphite grain), and d is the mean
travel a finite distance after birth before being recoil distance (m).
stopped by the surrounding material. Consequently,
some of the fission products formed near the surface
3.24.4.2.5 Chemisorption effect of metallic
of a fuel kernel are released directly into the buffer
fission products
region. This phenomenon is called recoil effect. Also,
At the boundary between the fuel element and cool-
a part of the fission products present in the OPyC
ant, the transition of diffusing atoms occurs due to
layer (due to uranium contamination during the fab-
adsorption and desorption (evaporation) processes,
rication process), may escape directly into the adja-
respectively. In most practical cases, the processes
cent matrix material. Similarly, some of the fission
are sufficiently fast that a local equilibrium may be
products formed in the grains of the graphite matrix
assumed between the concentration adsorbed at the
(heavy metal contamination is the source of such
solid surface and that in the adjacent boundary layer
fission products) escape the grains directly without
in the gas (‘vapor pressure’). Both are connected by a
diffusion onto the grain boundaries.
steep concentration difference of several orders of
Direct recoil after fissioning from the kernel
magnitude, as is shown in Figure 15. These concen-
into the buffer layer is an additional release mecha-
trations are put into relation by the so-called sorption
nism. It is independent of irradiation parameters,
isotherms, which are a function of the temperature
particularly of temperature, and can be derived from
and at high surface concentrations (in the so-called
purely geometrical considerations. At lower tem-
Freundlich regime, additionally of the surface con-
peratures (below normal operating temperatures),
centration itself ). In practice, adsorption is mainly of
diffusion transport is very slow and releases are
importance in the case of porous solids that have a
dominated by recoil.
large specific surface. The vapor pressure is expressed
The recoil contribution can be calculated as fol-
as an exponential function of temperature and sorbate
lows (e.g., Müller29):
" concentration.
#
3d 1 d 2 The sorption effect is strongly dependent on the
Frecoil ¼ 1 ½19 concentrations of the nuclide considered in the bulk
4R 12 R
and coolant, on the type of radionuclide and the
where Frecoil is the release fraction into the buffer thermodynamic conditions, and also on the type of
(or the matrix material or the grain boundaries), graphitic material. The sorption effect is furthermore
Release Deposition
Sorption:
Cbl= α Cw
Boundary layer
Boundary layer
Cw
Fission product flux:
δ Cw
j = −D δ r = b (Cbl−Cgas)
Concentration
Cw
Cbl j
Cgas
Graphite Graphite
Cbl
Cooling gas
Distance
Figure 15 Concentration profiles for adsorption and desorption processes.
774 TRISO Fuel Performance Modeling and Simulation
influenced by the coolant velocities. It is less pro- expected to mobilize a great part of the chemisorbed
nounced during normal operation with high-speed metallic fission products that form stable compounds
coolant flows along the graphite surfaces, but will with the water. Although the penetration by water is a
become more significant at low gas velocities such slow process, in safety analyses, the assumption is
as in the scenario of an unrestricted core heatup in made that all metallic fission products in graphite
the depressurized reactor. regions, which get into contact with liquid water,
Coked phenolic resin binder, a component of will be released.32 The kinetics of most reactions is
the fuel element matrix graphite, has a high sorption determined by an activation energy, typically
capacity for cesium and strontium. The high values of described in the form of an Arrhenius approach.
sorption enthalpies show that Cs and Sr are strongly Calculation models have been applied to assess air
bound by chemisorption. The large quantity of ingress accidents. The contributing transport pro-
matrix material in the core indicates a high potential cesses are convection, conduction, and radiation,
for cesium and strontium retention during normal with their shares depending on the conditions of
operation and in the case of a core heatup accident. flow rate, oxygen concentration, temperature, ther-
This potential increases with fast neutron irradiation mal gradient, local distribution of chemical heating,
at temperatures below 1100 C, and decreases if and local and global burning.
irradiation temperatures reach 1400 C.30 In con- The release behavior of fission products is influ-
trast, the sorption capacity of graphite for silver and enced by their chemical state.33 Experience acquired
iodine is much smaller. from the accident in the Three Mile Island (TMI)
nuclear power station, for example, has shown that
3.24.4.2.6 Influence of chemical reactions on the release of iodine was, by several orders of magni-
release behavior tude, less than expected, explained by chemical reac-
Chemical reactions of the impurities generally pres- tions within the aqueous phase. Distinguishing
ent in the primary (and secondary) coolant, such as between metals and nonmetals, the former represents
H2O, CO2, CO, H2, CH4, or N2, can occur over the the dominant part of the fission products. For the
total temperature range of an HTGR, from roughly HTR-Modul, the estimated ratio is
300 to 1200 C.31 Corrosive reactions with the core
ðCs þ Rb þ Sr þ BaÞ=ðI þ Br þ Te þ SeÞ 7:5
graphite are of principal importance because they may
degrade the strength of core structures and may also
lead to the formation of flammable gas mixtures. The
3.24.4.2.7 Influence of hydrolysis on release
reactions with the graphite are of either thermal,
behavior
radiolytic, or sorptive nature. While in colder core
The fractional releases of fission gases from exposed
areas, radiolytic reactions dominate, thermal reactions
fuel kernels under irradiation can be enhanced if the
are essential at higher temperatures.
kernel is hydrolyzed by reaction with trace amounts of
The fission products bonded to the graphite sur-
water vapor that may be present in the helium cool-
face may interact directly with the water, steam, or
ant.16,34 The magnitude of the effect depends upon the
oxygen, or may be released with the gasified graphite.
kernel composition, with UC2 showing the largest
The latter process would even liberate fission pro-
enhancement and UO2 showing the least; the effect
ducts trapped inside graphite grains. Porosity or
for UCO is intermediate. With kernels containing
cracks developing in the graphite through the oxida-
uranium carbide, the water reacts to convert the car-
tion processes may offer a faster way for the air or
bide phase completely to oxide along with kernel
water to reach the fuel particles. It was found that
swelling and increased porosity. With UO2 kernels, a
matrix material is more reactive than nuclear-grade
hyper-stoichiometric uranium-oxide phase may be
structural graphite.
formed; in any case, kernel porosity is increased with
The water–carbon reaction is an endothermal
enhanced gas release.
process that consumes nuclear decay heat. Therefore,
The general response of the exposed kernel to the
high temperatures and high material loss are not
introduction of water vapor consists of three distinct
expected. Reaction rates are dependent on the tem-
stages (Figure 16):35
perature. Changes in the chemical form of fission
products or changes in the properties of the graphite 1. a transient release of stored fission gas with a
material could change the transport rates through the concomitant increase in the steady-state fractional
matrix. Liquid water in contact with graphite is release,
TRISO Fuel Performance Modeling and Simulation 775
10−2
186 Pa H2O or in spherical coordinates:
2
@c @ c 1 @c
Stages 0 1 2 3 ¼D þ lc þ S ½21b
@t @r 2 r @r
10−3
100 110 120 150 where l is the decay constant (s1), and S is the fission
t (days)
product production rate (atoms m3 s1). The follow-
Figure 16 Reaction of fission gas release from a ing boundary conditions apply:
particle kernel upon contact with water. Reproduced
from Myers, B. F. The effect of water vapor on the release Concentration gradient is 0 at radius r ¼ 0
of fission gas from the fuel elements of high-temperature,
gas-cooled reactors: A preliminary assessment of @c
¼0 ½22a
experiments HRB-17, HFR-B1, HFR-K6, and KORA; Report @r r ¼0
ORNL/TM-4294; Oak Ridge National Laboratory: Oak
Ridge, TN, 1995.
Continuity of flux and concentration is given at
the interface between two adjacent materials with
diffusion coefficients D1 and D2:
2. a period of constant steady-state release, and, upon
removal of the water vapor, @c @c
c1 ¼ c2 and D1 ¼ D2 ½22b
3. a monotonic decline in the fractional release to @r 1 @r 2
prehydrolysis values (or nearly so).
For fission product transport in the fuel element
3.24.4.3 Simulation of Fission Product matrix graphite, a third boundary condition
Transport and Release applies describing the mass transfer at the fuel
surface which is given by
3.24.4.3.1 Diffusive transport of metallic
fission products @c
D ¼ bðc bl cgas Þ with c bl ¼ f ðcw ;T Þ ¼ acw ½22c
The transport of metallic fission products through @r r ¼rs
the fuel materials is modeled as a transient diffusion
process. The transient diffusion equation is typically where b is the mass transfer coefficient from the
solved numerically with appropriate boundary and surface to the coolant, cbl is the concentration in
interface conditions. the coolant immediately above the graphite surface
The rate of migration of a species through a (atoms m3), cgas is the mean concentration in the
homogeneous medium is defined by its mass flux coolant (atoms m3), and cw is the concentration in
and a concentration gradient as the driving force. the graphite near the surface (atoms m3). a is the
According to Fick’s first law, as described in ratio between boundary layer and wall concentration
Crank,36 the flux of atoms diffusing through a medium and is determined by means of sorption isotherms.
is proportional to the concentration gradient: This third boundary condition defines the release
rate from the surface of the fuel sphere into
@c
Jx ¼ D ½20 the coolant. In most cases, free evaporation from the
@x surface can be assumed, which is expressed by
where J is the diffusion flux (atoms m2 s1), D is the b ! 1 or cw ¼ 0 ½22d
diffusion coefficient (m2 s1), c is the concentration
(atoms m3), and x is the position or length (m). The transport speed is determined by the diffusion
The second Fickian law describes the time- coefficient D (m2 s1) which is typically dependent
dependent change of the concentration field by dif- on the temperature. For gas-in-gas binary diffusion,
fusion. Assuming the diffusion coefficient D to be a the diffusion coefficient is approximately proportional
776 TRISO Fuel Performance Modeling and Simulation
to T 1.5. For transport in solids, an Arrhenius-type (due to its numerous adsorption sites), and do not
temperature dependence37 of the form lose their trapping character if occupied. The pro-
cesses are described by a trapping coefficient, m, and
Q a reemission coefficient, b, whose reciprocal values
D ¼ D0 exp ½23a
RT indicate the average times of a particle to be trapped
or mobile, respectively. Under quasiequilibrium con-
is assumed, where D0 is a preexponential factor ditions, the trapping-emission process approaches
(m2 s1), Q is the activation energy ( J mol1), R is the Fickian diffusion process with the effective diffu-
the universal gas constant, R ¼ 8.3145 J mol1 K1, sion coefficient
and T is the absolute temperature (K).
0 Deff
The transport of mobile fission metals is certainly Deff ¼ ½25
1 þ m=b
structure-sensitive and more complex than classical
Fickian diffusion and likely a combination of lattice A calculation study has shown that the cesium
diffusion, grain boundary diffusion, pore diffusion, etc., release from the active core is larger by a maximum
further complicated by effects such as irradiation- factor of eight, if diffusion trapping is taken into
enhanced trapping and adsorption. Consequently, account, but still remains with ¼ 108 insignificant
any quoted diffusion coefficient should be called an from the radiological point of view. Therefore, the use
‘effective’ diffusion coefficient, which implies that the of effective diffusion coefficients may be justified.
overall migration process is approximately described
by Fick’s laws.38 3.24.4.3.2 Release of long-lived or stable
An effective diffusion coefficient can be deduced fission products from the particle kernel
experimentally for specific temperatures or tempera- In the equivalent sphere model,38 the primary fission
ture ranges, since the diffusion of fission product product retaining object is a UO2 particle kernel
atoms increases as the temperature is raised according (or a grain) that is modeled as a sphere with radius a.
to the Arrhenius equation. Sometimes, experimental The accumulated fractional diffusive release F of long-
findings suggest the incorporation of additional lived or stable fission products (decay constant l 0)
dependencies, such as neutron fluence, G, or concen- is given by the transient solution of the diffusion equa-
tration of the considered species, c, or the use of two tion to yield
temperature ranges with different activation energies 6 X 1
1 expfn2 p2 D0 t g
such as one for the normal operation phase and F ¼1 ½26
D0 t n¼1 n4 p4
another for the accident temperature range:
X where D 0 is the reduced diffusion coefficient (s1) and
Qi
Deff ¼ DðT ; G; c; . . .Þ ¼ D0;i exp ½23b t is the irradiation time (s).
i¼1;2
RT F is shown as the lower curve in Figure 17 as a
function of diffusion time t ¼ D0 t . F can be effi-
Burn-up dependence may be taken into account by ciently approximated by
the following equation: rffiffiffiffiffiffiffi
D0 t 3 0
D0 ¼ C1 ½1 þ ð1 þ nÞC2 F n ½24 F 4 D t for D0 t 0:35 ½27a
p 2
where C1, C2, and n are derived from experiments and and
F the burnup expressed in fissions per initial metal 1
F 1 for D0 t > 0:35 ½27b
atom, 1% FIMA 10 MWd ton1 (FIMA). 15D0 t
The use of effective diffusion coefficients, that is, During accident condition heating tests, the release of
the assumption of the Fickian diffusion model, is a long-lived or stable fission products from the particle
simplified simulation of the transport mechanism. At kernel or a spherical grain can be approximated by
least for cesium, a trapping mechanism was found solving the diffusion equation, but without source/
to describe more realistically the transport process sink terms, with the boundary condition
in graphite.39 The diffusion/trapping/reemission
@c
model assumes that migrating particles are captured ¼ 0; cðr ¼ a;t Þ ¼ 0 ½28
@r r ¼0
in traps for some time, before they are reemitted and
become mobile again. Traps are uniformly distributed, where a is kernel/grain radius. The initial condi-
predominantly in the polymeric binder carbon tion of a uniform concentration in the sphere
TRISO Fuel Performance Modeling and Simulation 777
1E + 0
Fractional release
1E − 1
1E − 2
1E−4 1E−3 1E−2 1E−1 1E+0
τ = D't
Figure 17 Fractional release of long-lived fission products during irradiation (lower curve) and during heating (upper curve)
as a function of diffusion time t ¼ D0 t according to the equivalent sphere model.
cðr ;t ¼ 0Þ ¼ constant represents the case of prior zero concentration at the surface, the R/B ratio is
cold irradiation. During an isothermal heating, it is calculated from
assumed that gas, which diffuses to its surface, is rffiffiffiffiffi rffiffiffiffiffi rffiffiffiffiffi! qffiffiffiffiffiffiffiffiffi
released from the sphere. The out-of-pile fractional R D0 l D0
¼3 coth / T1=2 ½31
release F after a time t is then given by B l D0 l
X
1
expfn2 p2 D0 t g
Fout-of -pile ¼ 1 6 ½29 where l is the decay constant (s1), and D 0 is the
n¼1
n2 p2 reduced diffusion coefficient (s1). For small R/B
ratios of less than about 10%, the release can be
0
where D ¼ D=a the reduced diffusion coefficient
2
approximated by
(s1). rffiffiffiffiffiffi
R D0
Fout-of -reactor is efficiently approximated by 3 ½32
B l
rffiffiffiffiffiffiffi
D0 t The R/B ratio is approximately proportional to the
F 6 3D0 t for D0 t 0:15 ½30a
p square root of the half life T1/2. Figure 18 shows, as
and an example, the measured fission gas release from
Peach Bottom fuel (BISO-coated HEU (Th,U)C2)
6 exp fp2 D0 t g in the upper curves and as-manufactured Fort St.
F 1 for D0 t > 0:15 ½30b
p2 Vrain fuel (TRISO-coated HEU (Th,U)C2) in the
The release curve, applicable to a heating test of a lower curves.16
bare kernel or broken particle, is shown as the upper General Atomics has developed a modified Booth
term in Figure 17. formula to account for the more complex character
of the diffusive release mechanism and it is even
3.24.4.3.3 Release of short-lived fission more complex. The reference gas release model
products from the particle kernel equation includes additional temperature and
For short-lived fission gases, the time t available burnup-dependent terms:16
rffiffiffiffiffiffi
for diffusion from the sphere becomes very short, so R D0
that p2D 0 t 1. A respective equation was derived 3 f ðT Þf ðBuÞ ½33a
B l
by Booth40 and modified or refined by other authors. with
Some approaches are described in IAEA.4 For short-
lived fission gases with decay constants l
106 (s1) Q 1 1
f ðT Þ ¼ A þ ð1 AÞ exp ½33b
at steady-state assuming a constant birth rate and a R T T0
778 TRISO Fuel Performance Modeling and Simulation
10–2
10–4
Kr
Xe
10–6
10–3 10–2 10–1 100 101 102 103
Nuclide half-life (h)
Figure 18 R/B ratios for various xenon and krypton isotopes measured on US fuel versus their half lives. Reproduced from
Hanson, D. L. A review of radionuclide release from HTGR cores during normal operation; EPRI Report 1009382;
Electric Power Research Institute: USA, 2004.
For design purposes, however, a simpler approach is inventory still left in the kernel, this moment is ‘time
taken in recognition that the transient release phase is zero,’ from which an independent diffusive release
short-lived, and that a new steady-state R/B is estab- calculation will start. The particle failure function,
lished if the water is persistent, or the R/B returns to which is an input to the code, is actually a step
its original value if the water is removed. For fully function with a maximum of ten steps.
hydrolyzed UCO fuel, a correlation similar to the The procedure of calculation within a time step is
aforementioned reference GA gas release model is performed in two consecutive parts: first, the diffusive
used but with different fitting constants, and a ‘hydro- release is determined from intact and defective par-
lysis increase factor’ of 1.7 is included.18 ticles and from the graphite grains. The sum of those
single releases is the overall source term for the fuel
element matrix graphite (grain boundaries). In the
3.24.4.4 Simulation of Fission Product
second part, the diffusive transport through the
Transport and Release Behavior
matrix and release into the coolant is calculated.
Most of the currently used models are based upon the The diffusion calculation is conducted based on rela-
numerical solution of Fick’s laws of diffusion repre- tive values. The inventory of a fission product is built
senting the ‘classical’ approach for studying the fis- up during the normal operation phase according to
sion product transport in the particle kernel, the its decay constant until reaching 100% at the end of
coating layers, and the matrix graphite of the fuel irradiation and the beginning of the accident phase,
element as a function of the temperature–time his- respectively. For long-lived species, the buildup is
tory during normal operation and under core heatup (almost) linear with irradiation time, while short-
accident conditions. lived species quickly reach the equilibrium state.
The GETTER program,46 a German develop-
3.24.4.4.1 Diffusion model ment and currently applied in South Africa, is also
The German computer model FRESCO exists in two based on Fick’s laws of diffusion, and calculates
versions. The ‘core’ version44 was developed first to the dynamic release of fission products from fuel.
describe the fission product release behavior in a Differential equations are solved numerically with
pebble bed core during a core heatup accident specific boundary conditions. The code includes sub-
phase taking into account the redistribution of fission models which calculate the burnup and power history
products by the coolant flow and the sorption effects of a fuel sphere in the core and the temperature
at the graphite surfaces of both the fuel elements and distribution in the spheres on the base of given neu-
the top and bottom reflectors. This FRESCO(-I) tron cross-sections and gas temperatures. Fission
version is strongly dependent on the input of tran- product production is determined by the power pro-
sient temperature and coolant flow distribution as a duced in each core region. The transport of metallic
result of a thermodynamic model calculation. fission products are calculated for contamination
A follow-up version, FRESCO-II45, concentrates sources in the matrix material, intact coated particles,
on a single, representative spherical fuel element and and from failed coated particles.
includes, in addition to the heating/accident phase, In the case of activation products (134Cs, 110mAg),
also the phase of irradiation/normal operation. The GETTER is run two times: at first for the release of
main purpose of this version is to simulate a complete the parent nuclide (133Cs, 109Ag) and afterward for
irradiation and heating experiment for a spherical the activation product, whereby the time-dependent
fuel element. The main input data for the code are, concentration profiles of the parent nuclides define
apart from the fuel design, the initial fuel distribu- the source term for the second run. For gas precur-
tion, the transport data, the particle failure function, sors (137Xe ! 137Cs or 90Kr ! 90Sr), a Booth-type
and the temperature–time history of the fuel. calculation similar to the NOBLEG model is applied.
The FRESCO diffusion model distinguishes The NOBLEG code,47 also used in South Africa,
between two types of particles, intact particles and calculates the release of short-lived fission products.
defective/failed particles. A defective or failed parti- It is, in principle, based on Booth’s ‘equivalent sphere’
cle is represented by a bare kernel only, meaning that model, and calculates the steady-state releases of
fission products are released immediately into the fission gases from fuel.48 It contains a thermohy-
matrix graphite. At the moment of failure of a parti- draulic routine that calculates fuel temperatures and
cle, the fission product content inside the coating fission product generation based on the power pro-
layers is released into the matrix graphite. For the duced and coolant temperatures in each core region.
780 TRISO Fuel Performance Modeling and Simulation
The transport of gaseous fission products is deter- gap between the compact outside and the sleeve,
mined for graphite contamination and for defective taking into account the desorption/adsorption pro-
particles. cesses at the surfaces.
For the release of fission gases from the uranium
(or thorium) contamination, the graphitic matrix
material is treated as a three-component system. 3.24.4.4.2 Integrated particle failure and
The components are the graphite grains of the raw release model
material, the amorphous, nongraphitized binder coke In a joint US/German statistical modeling approach,
between the grains, and the open pore system filled different from diffusion, the release of cesium from
with helium. Primary fission products are distributed a TRISO particle is assumed to be the direct indi-
homogeneously in these components by direct recoil. cator of a failed SiC layer. There is no diffusive
Gas atoms diffuse from the recoil sites in the grains of transport of cesium within the silicon carbide. The
the solid components to the open porosity of the fuel observed delay of fission gas release compared to
sphere. With regard to defective coated particles, a cesium is caused by slow krypton transport through
four-component system is used with the components’ the remaining intact OPyC which is highly perme-
kernel ‘grains,’ two material components of the buffer able for cesium. In its original 1985 version,51 two
layer (‘grains’ and ‘amorphous constituents’), and degradation mechanisms for the SiC layer are mod-
helium-filled open pores in the kernel and in the eled, which correspond to those implemented in
porous buffer layer. The birth rates of primary fission the PANAMA code, corrosion of the SiC layer
products in the four components are calculated with dominant in the lower temperature range, and SiC
the known relations for the recoil stopping ranges in thermal decomposition dominant at temperatures
different materials. above 2000 C.
The US code TRAFIC-FD49 is a core survey code The occurrence of SiC failure is a statistical phe-
for calculating the full core release of metallic fission nomenon. The combination of a layer thickness dis-
products and actinides. TRAFIC-FD is a finite dif- tribution with a degradation rate distribution can be
ference solution to the transient diffusion equation approximated by a Weibull distribution, including
for multihole fuel element geometry with a convec- the dependence on fuel type, irradiation tempera-
tive boundary condition at the coolant hole surface. ture, burnup, and fast neutron fluence. The Weibull
The effect of neutron fluence on graphite sorptivity distribution is fitted to the observed cesium release
is modeled explicitly. The temperature and failure curves from the fuel elements. Cesium retention in
distributions required as input are supplied by an fuel kernel, pyrocarbon, and matrix graphite is
automatic interface with the SURVEY code. TRA- neglected. Also, release from the ‘free’ uranium in
FIC-FD contains component models for fission metal the fuel element (manufacturing defects and heavy
transport in kernels, coatings, compact matrix, and metal contamination) is not considered.
fuel-element graphite.16 A revised version of the US/FRG statistical model
JAEA in Japan uses the FORNAX code50 to simu- of 198852 is related to a new derivation of the model
late the release of metallic fission products from the coefficients based on more heating experiments. The
particles and fuel compact, respectively. Similar to basic philosophy has been kept. Compared to the
FRESCO, it is based upon the diffusive transport original version, the revised statistical model contains
through the various materials, including the prompt a stronger dependence on the irradiation tempera-
release from the particle kernels by recoil. The ture and a weaker dependence on the fast fluence.
FORNAX code is able to handle three different These dependencies do not have a physical meaning,
kinds of particles: intact particles, bare kernels rather these are the result of the fitting procedure.
(representing through-coatings failures), and – at a The SiC thermal decomposition process is now
stage in between – particles with a degraded SiC assumed to be the controlling failure mechanism as
layer simulated by a larger diffusion coefficient. The well at lower temperatures. The matrix graphite of
code is designed to model the fuel geometry in a the fuel element is recognized to retain a significant
block reactor. It usually treats a complete fuel column amount of cesium simulated by a so-called ‘matrix
at a time with the vertical temperature, burnup, and release fraction’ (MRF), which is defined as the ratio
neutron fluence profiles as input data. Besides the of the release from the fuel element over the release
diffusion process, it also includes the transport of from the coated particles as function of the fast neu-
fission products through the gaseous phase in the tron fluence.
TRISO Fuel Performance Modeling and Simulation 781
Ct
Ct
102 102
T (⬚C)
Partial pressure (Pa)
Partial pressure (Pa)
100 0 100
2400 T (⬚C)
0
22 00
20 0
10–2 0 10–2 240
180 0 0
0 0 22
16 0
10–4 0 10–4 200
140 0
180
0 0
10–6 120 10–6 160
Measure-
ments 0
10–8 100
0 10–8 140
Measure-
ments
10–10 10–10 1200
800
10–12
0
0
600 margin 100 margin
10–14 10–14
10–4 10–2 100 102 10–4 10–2 100 102
(a) Concentration (mmol kg–1) (b) Concentration (mmolt kg–1)
Figure 20 Sorption isotherms for cesium (a) and strontium (b) over A3 matrix material. Shaded areas indicate additional
uncertainty margins recommended for safety analyses.
TRISO Fuel Performance Modeling and Simulation 783
as well as for normal operation and core heatup acci- Figure 21 shows the fractional release of 137Cs
dents for numerous designs of small- and medium-sized versus heating time for several cases as part of a
HTGRs. The purpose of these calculations was first verification study. The upper two red curves describe
of all a model verification to ensure that the theory the cesium release from a bare 500 mm UO2 diameter
and principles are clearly understood and a model kernel when heated at 1600 C, comparing the ana-
validation by comparing the computational results lytical solution according to Booth (thin line) with
with well-documented experiments. Additionally, alter- the corresponding FRESCO calculation (thick line).
native calculation models may be used in benchmark- It can be seen that the numerical solution is some-
ing exercises, for example, for simple geometries and what below the analytical one. The difference in the
assumptions that allow comparisons between the ana- order of a few percent can be reduced by increasing
lytical solutions (i.e., Booth equation) and numerical the number of nodes inside the kernel for which
solution (Fick’s laws). A comparison between differ- concentrations are calculated. The blue curve is also
ent codes would complete the verification process. the result of heating at 1600 C, but now applies to a
Validation usually requires more effort, as real-life kernel coated with buffer and PyC layer. The curve
experiments and benchmarks must be evaluated. increases at a slower rate, but after about 100 h,
A thorough evaluation of independent data available also reaches the 100% release level. Its difference
from suitable irradiation and postirradiation heating to the red curve represents the retention potential
tests is required. of the buffer and PyC layers at 1600 C, heating
Based on conclusions from these verification and showing a certain delay in the release.
validation efforts, the models were used to make The influence of the heating temperature can be
predictive calculations for specific reactor designs, seen when comparing the red curve for 1600 C with
mainly in the frame of safety and risk analyses. But the black curve for 1200 C, which has the same steep
codes have also been used to judge important input increase at the beginning, but remains at a lower
data such as diffusion coefficients and, if necessary, release level, reaching about 46% of its initial cesium
to replace reference relationships by new recommen- inventory after 200 h, whereas in the case of 1600 C,
dations. A few examples from a study with the all cesium has already escaped after about 60 h.
German diffusion code FRESCO will be given in The two lower green release curves in the figure
the following, which is part of the still ongoing refer to the case of the full TRISO-coated particle,
benchmark exercise with fission product release both showing the retention capability of the coating
codes within the previously mentioned CRP-6.55 layers for cesium. The difference between the thick
1E + 00
1E – 01
1E – 02
Fractional release cesium
1E – 03
1E – 04
1E – 05
1E – 08
0 20 40 60 80 100 120 140 160 180 200
Heating time (h)
Figure 21 Predicted fractional release of 137Cs from a bare kernel during heating at 1200 and 1600 C, a kernel coated with
buffer and PyC at 1600 C, and a TRISO-coated particle at 1600 C (no preceding irradiation phase).
784 TRISO Fuel Performance Modeling and Simulation
1E + 00
1E – 01
1E – 02
Fractional release
1E – 03
1E – 04
1E – 05
1E – 06
Cesium Strontium Silver
1E – 07 Cs meas Sr meas Ag meas
Failure fraction
1E – 08
0 50 100 150 200
Heating time (h)
Figure 22 Predicted and measured fractional release of metallic fission products during heating test of the spherical fuel
element HFR-K3/3 at 1800 C.
significant release from the coated particles during increases. In contrast, in the case of higher core
the 1600 C heating. In contrast, in the two spheres temperatures (solid lines), the Cs release from the
from the HFR-K3 irradiation test that have experi- coated particles is so high that concentrations in
enced more severe irradiation conditions (higher the fuel and reflector graphite soon reach its capacity
burnup and neutron fluence), the particles are in a limits such that all additional Cs released from the
stage of releasing cesium through the still intact fuel directly contributes to the release from the active
TRISO coating at the end of the heating test. The core and is transported through the primary circuit.
profiles in the matrix graphite calculated with the For practical calculations within safety analyses
FRESCO code (where the cesium release from the for small modular HTGRs, however, the sorption
coated particles into the matrix was adjusted to the effect is typically neglected for conservative reasons.
measurements that allowed to isolate the diffusion
transport in the matrix) show a good agreement
with the measurements. It means that the ‘‘effective’’ 3.24.5 Further Work
diffusion approach is sufficiently well-describing
cesium transport behavior in matrix material. A sim- With the renewed interest in HTGRs worldwide,
ilarly good agreement was achieved for three heated numerous theoretical and experimental activities
spheres with respect to strontium. have been resumed to further understand the behavior
The particular effect of sorption could clearly of fuel performance and fission product release from
be seen in a calculation study considering a core HTGR fuel. Improved testing and analytical methods
heatup accident scenario for the THTR-300 for today allow a much deeper insight into the physical and
two different accident temperature histories with chemical phenomena that take place in the fuel. Also,
maximum fuel temperatures of 2080 and 2360 C, new developments in computer modeling have started,
respectively. Figure 24 shows the comparison of which may replace in future the classical approaches,
FRESCO-I calculational results for cesium release which were often handicapped by constraints in terms
from the coated particles (gray), from the core cavern of CPU time and array at the time they were devel-
(brown), and the inventory in the top reflector graph- oped, and which appear ‘old-fashioned’ today.
ite (green). In the case of lower core temperatures The model calculations on coated particle fuel per-
(dashed lines), the matrix graphite provides a very formance require the knowledge of various properties of
high potential of cesium sorption supported by the the particle’s constituent materials and also the way
top reflector whose inventory gradually increases. in which these values vary over the course of the irradi-
Therefore, the release from the core cavern remains ation. More experimental work on material properties
on a low level over 280 h, before it eventually is necessary to reduce uncertainties in the input data
10–8
0 100 200 300
Time (h)
Figure 24 FRESCO-I prediction of cesium release behavior in the THTR-300 core for two different accident temperature
distributions.
786 TRISO Fuel Performance Modeling and Simulation
and answer still open questions, especially with regard One of the most recent heating tests, HFR-K6/3,
to the desire for attaining higher burnups in future. has shown surprisingly low krypton and cesium
Plans to investigate material properties are being ad- release values, which are largely overpredicted by
dressed by fuel development programs in the United the model calculations. This extremely good accident
States, South Africa, Republic of Korea, China, and the condition performance of the very latest manufacture
European Union. There is also a need for continuous of German TRISO fuels should be the starting point
interaction between modelers and experimenters to of further studies: if the results of this type can be
ensure that what is observed is being modeled. substantiated reproducibly, new HTGR designs
The determination of the property ‘SiC strength’ might benefit from potentially higher accident tem-
and its statistical distribution is central to all fuel peratures than the presently established 1600 C
particle mechanical modeling, since it is part of the limit. The new limit has to be quantified taking into
primary failure criterion. The specific problem is account the burnup and neutron fluence levels
that SiC strength has not been measured on any of reached during irradiation.
the high performance UO2 TRISO fuels. The refer-
ence data for SiC strength (836 MPa) and respective
Weibull modulus (8.02) for the nonirradiated state, 3.24.6 Summary and Conclusions
typically used in predictive (PANAMA) calculations,
correspond to a former German-coated particle Since the end of the 1950s, the coated fuel particle
batch production, EO 1607, which appears to repre- has been developed as the basic fuel unit common to
sent some kind of average data. It was also the only all HTGR concepts, having reached today a high
particle batch, for which SiC strength and Weibull level of quality. A statistical evaluation of defective/
modulus were measured after neutron irradiation. failed SiC in German UO2 TRISO fuel is given in
A respective relationship is usually interpolated or Table 3. Performance requirements for future high-
extrapolated to the fast fluences considered, however, temperature reactors establish a basis for the future
with no assured data basis. development of advanced coated particle fuels.
From the postcalculation of heating tests with the The failure mechanisms and performance limits of
German FRESCO code and the comparison with reference HTGR fuels as a function of particle
the measurements, it appears that some important design parameters are relatively well established.
reference diffusion coefficients still vary over a broad Continuing improvements in fuel performance and
range. It is obvious that in particular, the strontium data fission product transport models are further increas-
are largely overpredicted in most cases. While the ing the understanding of the behavior and limitations
retention capability of UO2 and matrix material for of this particle design. HTGR gas turbine and pro-
strontium is visible, it is most probably the diffusive cess heat applications under development require
transport through the silicon carbide layer that appears significant increases in fuel temperatures relative
very conservative and should undergo a thorough to the steam cycle applications developed earlier.
review. While the LEU SiC TRISO fuel system performance
Item Number of particles Number of failed Expected failure One-sided upper 95%
tested, N particles, n level, n/N limit of failure level
Manufacture
All UO2 LTI TRISO 2 202 200 86 3.91 105 4.67 105
Fabrication after 1985 528 400 6 1.14 105 2.24 105
Irradiation testing
All LEU UO2 irradiation 586 964 9 1.53 105 2.68 105
tests
UO2 sphere irradiation 506 120 0 0 5.92 106
tests
Accident condition testing
All heating tests to 1620 C 303 840 5 1.65 105 3.46 105
1600 C isothermal tests 221 840 0 0 1.35 105
capability is sufficient to support initial deployment 9. Proksch, E.; Strigl, A.; Nabielek, H. J. Nucl. Mater. 1985,
136, 129–135.
of these applications, achievement of their full eco- 10. Crawford, D. C.; et al. J. Nucl. Mater. 2007, 371, 232–242.
nomic potential may be enhanced by the development 11. Martin, D. G. Nucl. Eng. Des. 2002, 213, 241–258.
of advanced fuels with higher temperature capability. 12. Miller, G. K.; et al. J. Nucl. Mater. 2004, 334, 79–89.
13. Wimmers, M. Das Verhalten kugelförmiger
National engagement as well as bilateral or multi- HTR-Brennelemente bei der Massenerprobung im
national cooperation in HTGR fuel development are AVR-Reaktor; AVR Report; Arbeitsgemeinschaft
ongoing and are expected to further improve fuel Versuchsreaktor: Düsseldorf, Germany, 1977.
14. Kovacs, W. Candidate fuel property document (LEU);
performance and the ability to make reliable predic- Report 906040/B; San Diego, CA, 1981.
tions. The benchmark exercises conducted within the 15. Huddle, R. A. U.; Jones, K. W.; Mayr, K. The influence of
IAEA Coordinated Research Program #6 are excel- fabrication parameters on the incidence of spearhead
attack in coated particle fuel; Dragon Report D.P. 652;
lent examples. They have proved to be a useful basis Winfrith: UK, 1969.
for verification and validation in establishing the 16. Hanson, D. L. A review of radionuclide release from HTGR
reliability of code predictions. The confidence in cores during normal operation; EPRI Report 1009382;
Electric Power Research Institute: USA, 2004.
the applicability of these codes is strengthened by 17. Bongartz, K.; et al. J. Nucl. Mater. 1976, 62, 123–137.
validation along the experimental programs that 18. GA, Fuel Design Data Manual. General Atomics Document
include both irradiation and heating tests. 901866, Issue F, 1987.
19. Minato, K.; et al. J. Mater. Sci. 1991, 26, 2379–2388.
Furthermore, any new HTGR fuel development 20. Montgomery, F. Fission product SiC reaction in HTGR
program will, however, have to ascertain that the fuel; Report GA-905837, General Atomics: San Diego, CA,
fuel is consistent with the internationally established 1981.
21. Benz, R. Kinetics of decomposition of CVD SiC in modified
standards as described here. Additional irradiation TRISO-coated fuel particles at temperatures of
and postirradiation testing and examination will be 1600–2200 C; Technical Note IRW-TN-124/82; Research
required for newly manufactured fuel, in order to Center Jülich: Germany, 1982.
22. Verfondern, K.; Nabielek, H. PANAMA Ein
expand and improve the existing statistical database Rechenprogramm zur Vorhersage des Partikelbruchanteils
for fuel performance analysis. The additional data von TRISO-Partikeln unter Störfallbedingungen; Report
will demonstrate successful fuel designs, cover the Jül-Spez-298; Research Center Jülich: Germany, 1985.
23. Allelein, H. J.; et al. The behavior of HTR fuel under
full range of design and operating requirements of irradiation. In Proceedings of the 7th International
future reactors, strengthen the statistical viability Conference on Structural Mechanics in Reactor
of the performance base, and confirm expected Technology SMIRT, Chicago, IL, 1983.
24. Verfondern, K.; Lee, Y.-W. Advances in HTGR fuel
in-reactor performance. Fuel-test programs will be technology – A new coordinated research program.
continual efforts in that they may be open to new International Congress on Advances in Nuclear Power Plants
test requirements or new understandings of fuel ICAPP’05, Seoul, Republic of Korea, 2005; Paper 5050.
25. Nabielek, H.; et al. Prediction of temperature and fission
performance. product release from HTR fuel Under accident conditions.
In Proceedings of the International Conference on Design
and Safety of Advanced Nuclear Power Plants ANP’92,
References Tokyo, Japan 1992; Paper P1-9.
26. Phelip, M.; et al. The CRP-6 benchmark on HTGR fuel
behavior under normal operation. In International
1. Lohnert, G.; et al. Nucl. Eng. Des. 1988, 109, 257–263. Congress on Advances in Nuclear Power Plants ICAPP’06,
2. Schenk, W.; Pitzer, D.; Nabielek, H. Fission product Reno, USA, 2006.
release profiles from spherical HTR fuel elements at 27. Fütterer, M. A.; et al. Irradiation results of AVR fuel
accident temperatures; Report Jül-2234; Research Center pebbles at increased temperature and burn-up in the HFR
Jülich: Germany, 1988. petten. In Proceedings of the 3rd International Topical
3. Freis, D.; et al. Post irradiation testing of high temperature Meeting on High Temperature Reactor Technology
reactor spherical fuel elements under accident conditions. HTR2006, Johannesburg, South Africa, 2006; Paper
Proceedings of the 4th International Topical Meeting on B00000035.
High Temperature Reactor Technology HTR2008, 28. Nabielek, H.; et al. Burn-leach: The most important test in
Washington, DC, 2008; Paper 58203. the manufacture of HTGR fuel. In International Congress
4. IAEA. Fuel Performance and Fission Product Behaviour in on Advances in Nuclear Power Plants ICAPP’06, Reno,
Gas Cooled Reactors; Report IAEA-TECDOC-978; USA, 2006.
International Atomic Energy Agency: Vienna, Austria, 29. Müller, A. Freisetzung gasförmiger Spaltprodukte
1997. (Kr, Xe, J) aus Brennelementen für gasgekühlte
5. Ragoß, H. Description of fuel behavior performance used Hochtemperaturreaktoren; Report Jül-1295; Research
in the German licensing procedures for the HTR-module; Center Jülich: Germany, 1976.
Technical Note 38.07247.9; Interatom: Bensberg, 30. Hilpert, K.; et al. High Temp. High Press. 1988, 20, 157–164.
Germany, 1989. 31. Nieder, R. Grundzüge der HTR-Chemie. In Proceedings
6. Nabielek, H.; et al. Nucl. Eng. Des. 1990, 121, 199–210. of the Seminar on Chemie und Entsorgung, Report
7. Miller, G. K.; et al. J. Nucl. Mater. 2001, 295, 205–212. Jül-Conf-43; Research Center Jülich: Germany, 1981;
8. Proksch, E.; Strigl, A. J. Nucl. Mater. 1982, 107, 280–285. 19–29.
788 TRISO Fuel Performance Modeling and Simulation
32. Moormann, R. Source term estimation for small sized Core-Aufheizung; Report Jül-1791; Research Center
HTRs; Report Jül-2669; Research Center Jülich: Jülich: Germany, 1982.
Germany, 1992. 45. Krohn, H.; Finken, R. FRESCO-II: Ein Rechenprogramm
33. Moormann, R. Untersuchungen zum chemischen zur Berechnung der Spaltproduktfreisetzung aus
Verhalten der Spaltprodukte bei HTR- kugelförmigen HTR-Brennelementen in Bestrahlungs-und
Kernaufheizstörfällen. In Proc. Reaktortagung Ausheizexperimenten; Report Jül-Spez-212; Research
Kerntechnik, München, Germany, 1985; pp 163–166. Center Jülich: Germany, 1983.
34. Myers, B. F. A review of selected aspects of the effect of 46. Röllig, K. Rechenprogram GETTER; Report GBRA 052 477;
water vapor on fission gas release from uranium Westinghouse Reaktor GmbH: Ladenburg, Germany, 2001.
oxycarbide; Report ORNL/TM-12641; Oak Ridge National 47. Röllig, K. Description of the computer code NOBLEG;
Laboratory: Oak Ridge, TN, 1994. Report GBRA 054 350; Westinghouse Reaktor GmbH:
35. Myers, B. F. The effect of water vapor on the release of Ladenburg, Germany, 2001.
fission gas from the fuel elements of high-temperature, 48. Röllig, K. Nucl. Technol. 1977, 35, 516–523.
gas-cooled reactors: A preliminary assessment of 49. Tzung, F. TRAFIC-FD, a finite difference program to
experiments HRB-17, HFR-B1, HFR-K6 and KORA; compute release of metallic fission products from
Report ORNL/TM-4294; Oak Ridge National Laboratory: an HTGR core: Code theory and users manual;
Oak Ridge, TN, 1995. Report CEGA-001904; CEGA Corporation: San Diego, CA,
36. Crank, J. The Mathematics of Diffusion, 2nd ed.; Oxford 1992.
University Press: New York, 1975. 50. Sawa, K.; et al. J. Nucl. Sci. Technol. 1992, 29, 842–850.
37. Jost, W. Diffusion in Solids, Liquids, Gases, 3rd Printing 51. Goodin, D. T.; Nabielek, H. US/FRG accident condition
with Addendum; Academic Press: New York, 1960. fuel performance models; Report DOE-HTGR-85107;
38. Nabielek, H.; et al. Performance limits of coated particle General Atomics: San Diego, CA, 1985.
fuel; Part III: Fission product migration in HTR fuel; Report 52. Goodin, D. T. US/FRG accident condition fuel
DP-828 (Part 3), Dragon Project; Winfrith: UK, 1974. performance models; Report DOE-HTGR-85107;
39. Hensel, W.; Hoinkis, E. J Nucl. Mater. 1995, 224, 1–11. General Atomics: San Diego, CA, 1989; Revised version.
40. Booth, A. R. A method of calculating fission gas diffusion 53. Myers, B. F. Cesium diffusion in silicon carbide during post
from UO2 fuel and its application to the X-2-F loop test. irradiation anneals; Technical Note HBK-TN-01/84;
Report CRDC-721; Atomic Energy of Canada Limited, 1957. Research Center Jülich: Germany, 1984.
41. Ogawa, T.; et al. Nucl. Eng. Des. 1991, 132, 31–37. 54. Verfondern, K.; Müller, D. Code development for fission
42. Myers, B. F.; Morrissey, R. E. Licensing topical report: The product behavior in HTR fuel for safety assessments;
measurement and modeling of time-dependent fission Internal Report KFA-ISF-IB-10/89; Research Center
product release from failed HTGR fuel particles under Jülich: Germany, 1989.
accident conditions; Report GA-A15430; General 55. Verfondern, K.; Nabielek, H. Fission product release
Atomics: San Diego, CA, 1980. from HTGR fuel under core heatup accident conditions.
43. Myers, B. F. Experiment HFR-B1: A preliminary analysis of In Proceedings of 4th International Topical Meeting on
the water vapor injection experiments in capsule 3; Report High Temperature Reactor Technology HTR2008,
ORNL/TM-11846; Oak Ridge National Laboratory: Oak Washington, DC, 2008; Paper 58160.
Ridge, TN, 1991. 56. Schenk, W.; Nabielek, H. Kugelbrennelemente mit
44. Krohn, H. Freisetzung von Spaltprodukten aus dem TRISO-Partikeln bei Störfalltempera-turen; Report
Core eines Kugelhaufenreaktors bei Störfällen mit Jül-Spez-487; Research Center Jülich: Germany, 1989.
3.25 Modeling of Sphere-Pac Fuel
M. A. Pouchon and L.-Å. Nordström
Paul Scherrer Institut, Villigen PSI, Switzerland
Ch. Hellwig
Fachgebietsleiter Kernbrennstoff-Technologie, Baden, Switzerland
789
790 Modeling of Sphere-Pac Fuel
Both particle fuel concepts imply important structural accommodation of the volume increase.
changes in the physical properties compared to pellet With these good conducting matrices, the initial
fuel (see also Chapter 3.11, Sphere-Pac and VIPAC structurally reduced thermal conductivity becomes
Fuel). The reason is the macroscopic structure of the irrelevant. These are the major findings of the AC-3
particle arrangement. This structure introduces small project,4 where carbide pellet and Sphere-pac fuel of
pathways (touching points of the particles) a carbide matrix were both irradiated in the US fast
for thermal flow on one hand, and a large deforma- flux test facility (FFTF). With this study, the Sphere-
tion potential (soft mechanical behavior) on the other pac fuel proves its potential, especially when applied
hand. However, during reactor operation, the tem- in a fast reactor.
perature rises, and the particles sinter together. This The modeling codes have to address these particle
restructuring is more pronounced than in case of the fuel specific features. This chapter mainly concen-
pellet fuel. trates on the modeling of Sphere-pac fuel and there is
The initially much lower thermal conductivity only a short mention of some attempts being per-
and the larger surface compared to the pellet fuel are formed in modeling the Vipac fuel too.
the two major drawbacks for particle fuel. Because
of sintering, this initial disadvantage becomes less 3.25.1.2 Important Terms and Definitions
important during reactor operation. As the tempera- for Particle Fuel Modeling
ture is highest in the fuel center, the restructuring is
In modeling, the following important features arise
most important here, resulting in an almost pellet-like
from the particle (and here especially Sphere-pac)
structure. Additionally, a hole is formed at the center
fuel, whose terms are defined as follows:
(pore migration toward the temperature gradient).
At the border, the structural character of the Sphere- Smear density: The smear density (in %) is the
pac remains. Bart et al.4 and Hellwig et al.5 impressively material density taking into account the particle
show this behavior for carbide fuel being irradiated in porosity and the interstices between the particles
a fast reactor (collaboration program with the United as empty space.
States, the AC-3 irradiation at fast flux test reactor Neck ratio: The neck ratio describes how strongly
(FFTF)), and for oxide fuel being irradiated for the two adjacent spheres are sintered together. The
FUJI program. For the prediction of the thermal neck ratio is defined by the quotient of the dia-
behavior, and for the more general fuel performance, meter of the contact area (‘neck’) and the
two computer codes have been developed at Paul midsphere-to-midsphere distance of the two spheres
Scherrer Institute (PSI); these are the SPACON (see Figure 2). Alternatively, this definition can
code6 and the SPHERE code7 which are the main also be applied on necks formed between a sphere
subjects of this chapter. and the cladding wall.
An attempt to solve the disadvantage of the Porous-pellet-like structure: The thermal conductivity
initially lower thermal conductivity is made in the of the porous-pellet-like structure is calculated
Cer-met Sphere-pac concept. The idea was first by the thermal conductivity of the fuel material
developed by Van der Linden8 in the Netherlands, corrected with a factor for porosity. Several
and it was later taken up again by Pouchon and models11,12 are available for the determination of
coworkers9,10 for an inert matrix fuel, where espe- the correction factor. However, most models were
cially the yttria-stabilized zirconia-based matrix suf- developed for porosities below 10% and cannot be
fers from a rather low thermal conductivity. easily extrapolated to the higher porosities of sin-
Especially, when considering fuel matrices such as tered Sphere-pac beds. Thus, a suitable porosity
carbides, nitrides, or carbonitrides, the Sphere-pac correction model should be chosen carefully.
concept becomes very favorable. These matrices offer The transition from sintered Sphere-pac bed to a
the large advantage of being good thermal conductors. porous-pellet-like structure is normally defined by
Unfortunately the swelling behavior is disadvanta- the breakdown of the long gas channels between the
geous. However, used in a Sphere-pac arrangement, spheres to single gas reservoirs. This breakdown
the volume increase can be accommodated by the takes place at a neck ratio of 50% if perfect geomet-
interspaces between the spheres. In other words, rical boundary conditions (BCs) are applied.
the Sphere-pac arrangement behaves as a soft one, Because of imperfections such as slightly varying
and the swelling induced fuel–cladding mechanical sphere diameter and ovality, the transition to the
interaction (FCMI) is reduced because of the porous-pellet-like structure is normally set to 45%.
792 Modeling of Sphere-Pac Fuel
At PSI, the infiltration filling was commonly (see Section 3.25.5.1), which besides the basic models
applied for binary Sphere-pacs. Only this type of also incorporate mechanistic models for sintering
binary packing is considered in this chapter. Packing (sphere necking), fission gas behavior and release,
densities can then be defined for each fraction sepa- fuel-restructuring (pore migration and grain growth),
rately: fmat ¼ Vmat/Vclad for the coarse fraction (or etc. For a more detailed description of the models see
matrix); finf ¼ Vinf/(Vclad – Vmat) for the fine fraction Sections 3.25.3 and 3.25.4. Section 3.25.5.2 shows
(or infiltrate); and ftot ¼ fmat þ (1 – fmat)finf for the the modeling and validation of the PSI codes; the
total packing density (see also Chapter 3.11, corresponding experiments are also described in
Sphere-Pac and VIPAC Fuel for concrete values). Chapter 3.11, Sphere-Pac and VIPAC Fuel.
From the filling procedure, it follows that these pack-
ing densities are not free experimental parameters
but are essentially determined by the relative size 3.25.2 Modeling of the Thermal
and shape of particles and container. Properties
Table 1 Comparison of heat transfer processes of Sphere-pac fuel and pellet fuel
Sphere-pac fuel pin Effect on Effective Pellet fuel pin Effect on Effective
heat transfer range heat transfer range
Fuel (sphere) density The higher Fuel center to Fuel (pellet) density The higher Fuel center
the better cladding the better to pellet rim
Packing density The higher Gap size The lower Pellet rim to
the better the better cladding
Fill gas pressure and The higher Fill gas pressure and The higher
thermal conductivity the better thermal conductivity the better
Size and surface roughness The lower Surface roughness of The lower
of the fuel fractions the better fuel pellet the better
Formation of neck at The more
sphere–sphere contact the better
points
Formation of neck at The more Fuel–cladding Gap closure The sooner Pellet rim to
sphere–wall contact the better interface the better cladding
points
Modeling of Sphere-Pac Fuel 793
rc (f < 2/3) rs
rc = rs (f > 2/3) rc
R R
2R
rn,s–s rn,s–w
Sphere Wall
Figure 2 Schematic of the sphere–sphere cell (left) and the sphere–wall cell (right) with their geometrical characteristics:
the sphere radius R, the (relative) cell radius rc, the solid radius rs, and the neck radii rn (for sphere–sphere and for
sphere–wall contact). The solid radius is rs is equal to 1 for packing densities f 2/3, and equal to the cell radius rc for packing
densities f > 2/3. The cells have cylindrical symmetry.
and the neck radius, different values can be assumed incorrectly extended over the whole cell height,
depending on the type of cell, sphere–sphere (s–s) or without taking full advantage of the conducting
sphere–wall (s–w); see Figure 2. capacity of the remaining sphere material. The
model solves this problem by starting with the ana-
lytical one-dimensional treatment and then using a
3.25.2.2 Modeling of Heat Transport in
correction factor that is derived from a semianalytical
the Basic Cells
radial heat flow relaxation procedure, accounting for
Any heat transport model for the basic cell has to the bottleneck effect. This iterative procedure is out-
take into account two to three different zones: (i) lined in the following.
Sintered zone, ranging from the centerline to the
neck radius. Only solid-state thermal conductivity 3.25.2.2.1 Heat conduction assuming parallel
takes place in this zone. This zone has to be taken heat flow (one-dimensional case)
into account only in case of necking; otherwise, it In the case of parallel heat flow, the following BCs are
reduces to a simple contact point. (ii) The gas-filled valid: The cylinder circumference of both cell types
zone, ranging from the neck radius to the solid radius. (sphere–sphere and sphere–wall) is adiabatic by
Here the heat transport is determined by thermal definition (BC I). The upper and lower planes of
conductivities of solid and gas and a heat transfer the sphere–sphere cell are isothermal (BC II). The
resistance at the gas–solid interfaces. The gas is lower plane of the sphere–wall cell is (approximately)
assumed transparent for thermal radiation exchanged isothermal (BC III). A schematic of the cells is shown
between the solid surfaces. (iii) The bypass zone, in Figure 3.
ranging from the solid radius to the cell radius. BC II lets us conclude that the temperature dif-
In this zone, the heat transport is determined by ference across the cell, Dncell, is constant over the cell
the gas conductivity alone. It is present only if the radial cross-section. It is the sum of the temperature
packing density is less than two-thirds. drops in the different axial sections and interfaces,
In a first approach, we now apply an analytical caused by the heat flux dq in the cross-section
one-dimensional treatment to assess the axial heat dA ¼ 2pr dr . Within the sphere, it depends only
transport in such a basic cell. However, if the thermal on the thermal conductivity of the sphere material.
conductivity of the sphere material exceeds that of For the interstice, not only does the gas conductivity
the surrounding gas – as is generally the case and is and the heat transfer value at the sphere surfaces
assumed in this model – it is evident that the heat have to be considered, but also the thermal radiation
transport is restricted mainly to the region of the between the sphere surfaces. While for narrow gas
contact point, acting as bottleneck. An analytical gaps (such as the interstices in the Sphere-pac
one-dimensional treatment of such a cell neglects fuel), heat transfer by convection can generally be
the radial heat flow vector and thus normally under neglected, the temperature jump distance (calculated
predicts the heat transport because this bottleneck is according to Harding11) must be considered as an
Modeling of Sphere-Pac Fuel 795
dr r r dr rt,2
ls ls
R
Cylindrical
shell
j,arad Parabolic
li
ΔJcell
shell
j,arad
li
Cell mid plane rm,1
lw
rm,2
R
Wall ls
material Figure 4 Qualitative illustration of the heat flow lines in a
sphere–sphere cell with cylindrical symmetry, as assumed
to be realistic (left) and as assumed in the two-dimensional
Sphere–wall unit cell Sphere–sphere unit cell heat flow model (right). The section between the radii rm,1
Figure 3 One-dimensional model applied on the and rm,2 forms a ring area, where the heat enters the model.
sphere–sphere cell (right) and the sphere–wall cell (left). Equivalently, the heat exits the cell (on the cell top plane) in
For both cases, the cylindrical shell formed in the the ring formed between the radii rt,1 and rt,2.
infinitesimal radial section dr represents the conducting
path for the heat dq, leading through the different media.
ls, lw, and li are the corresponding thermal conductivities Figure 4, on the left side, shows qualitatively
of the spheres, the wall material, and the interstitial gas, j is
the expected heat flow pattern in a cell with partly
an additional distance due to temperature jump and surface
roughness, and arad is the heat transfer by radiation. sintered spheres of high thermal conductivity, sur-
rounded by a low-conducting gas. We now approxi-
mate the shape of the heat flow lines within the solid
spheres by parables symmetric to the cell axis. In the
additional distance to the geometrical width of the interstice, we stay with the assumption of the one-
gas gap. Similarly, the surface roughness of technical dimensional approach, that is, straight lines parallel
surfaces (spheres and cladding) has to be added to the to the cell (see Figure 4, on the right side). Any
gas gap as an additional distance (for details, see Botta two adjacent parabolic lines now define a shell
and Hellwig6). of two concentric paraboloids through which the
With these definitions and simplifications, and heat is conducted from the neck to the sphere mid-
considering that in the gas-filled interstices the heat plane. The heat enters the sphere in the ring area
transport by conduction and the heat transport by within the radii rm,1 and rm,2 (using the cylindrical
radiation are parallel processes, the temperature symmetry, the mid plane of the unit cell is divided
drop in any radial zone dr of the basic cell (see into rings), and leaves the sphere in the ring area
Figure 3) can be calculated. within the radii rt,1 and rt,2 (see Figure 4). The
heat flow resistance in this shell is obviously different
3.25.2.2.2 Two-dimensional approximation of from that of the cylindrical shell presumed in the
heat conduction one-dimensional calculation.
In the two-dimensional approximation, the con-
strained heat flow in the contact area at the 3.25.2.2.3 Evaluation of the thermal
sphere-to-sphere contact is allowed to diverge and conductivity and the heat transfer value
eventually spread out over the whole available The evaluation of the thermal conductivity (for
cross-section of the sphere. In the cell top plane sphere–sphere cells) and the heat transfer value
(or – in case of a wall – in the depth of one sphere (for sphere–wall cells) requires an iterative procedure,
radius from the wall surface), the heat flow can be with the exception of the bypass zone, which
assumed to be rather homogeneously distributed. is assumed to conduct the heat along parallel lines.
The thermal BCs I, II, and III of the two basic cells, The objective of this iteration is the achievement
as defined above, can thus be extended by the of constant heat flow densities in the cell top and
following: the heat flux is constant within the solid bottom planes. First, the one-dimensional heat flows
radius of the sphere-to-sphere contact place of the q(i ) in each of the radial zones except for the bypass
sphere–sphere cell and within the solid radius of zone are calculated (each zone is divided into a num-
the lower plane of the sphere–wall cell. ber of subzones for accuracy improvement). A first set
796 Modeling of Sphere-Pac Fuel
of radii, defining the parabolic boundaries in the cell size fractions, but also the solid radii and the neck
top plane, can then be determined by distributing all radii are calculated as average values on the basis of
heat flux densities in the cell top plane equally. the average sphere radius.
The ratio of the heat flow resistance of the para- The modeling of a two-fractional package
bolic shell to that of the cylindrical shell delivers now fabricated by infiltration filling is more demanding.
a first set of conductivity correction factors (i.e., one However, the package can be divided into basic
correction factor for each cylindrical shell). The one- sphere–sphere and sphere–wall cells of one-
dimensional calculation is then repeated with the fractional packages, with the sphere–sphere cell
corrected conductivities; a new set of radii, defining representing the contact of two equal-sized spheres,
the parabolic boundaries in the cell top plane, can be and the sphere–wall cell representing the contact of a
determined in the next iteration. This is repeated sphere with a cladding wall or the contact of a small
until convergence is reached. Then the heat flow in sphere with a coarse sphere as the coarse sphere with
the bypass qby is calculated and added, yielding a total a diameter greater than eight times the small sphere
heat flow in the cell of qtot ¼ q þ qby . The effective can be considered as a wall (see Figure 5). A new
thermal conductivity of the cylindrical cell of the geometrical parameter is needed to describe the
outer radius Rrcell and height 2R can now be calcu- package, that is, the penetration radius rp. This
lated assuming a temperature drop D#cell: value describes how close the small spheres can
2qtot penetrate into the interstice of two adjacent spheres
lcell ¼ ½1 (rp,s–s) or a sphere and a wall (rp,s–w). Other para-
Rrcell pD#cell
2
meters have now to be extended also to the infiltration
The thermal conductivity of a sphere–wall cell can be fraction. By separation of the complex two-fractional
modeled by basically applying the same method as
for calculating the thermal conductivity of a sphere–
sphere cell. This allows an analytical determination rc
Coarse–coarse
of the heat transfer coefficient a, as shown in detail in
cell rs
Botta and Hellwig.6 rp,s–s
rn,s–s
3.25.2.2.4 Application to two-fractional
packages
The model described above can now be extended for
modeling two-fractional packages. For two-fractional
Fine–fine
packages, we have to consider the applied fabrication cell
technique: spheres of different size fractions filled
into a pin in parallel form a multifractional package Fine–coarse
where spheres of different sizes are arranged arbi- cell
trarily in relation to each other. On the contrary, a
two-fractional package formed by infiltration filling
as described in Section 3.25.1 consists of a stable
package of the large spheres with the interstices filled
with small spheres. For this arrangement, normally
a higher thermal conductivity can be expected as
the large spheres form a network with a relatively
small number of bottlenecks, that is, sphere-to-sphere
contact points. rn,s–w
Fine–wall
For modeling the thermal conductivity of a paral- cell
rp,s–w
lel filled multifractional package, we simply consider
the package to be comparable with a one-fractional Coarse–wall cell
package of spheres of an average diameter. The
as-fabricated packing density shall be taken into
Figure 5 Illustration of a sphere–sphere cell (top) and a
account in the determination of the cell radius rc as sphere–wall cell (bottom) in a two-fractional package with
described in Section 3.25.2.1. Not only the sphere their characteristic radii. The basic one-fractional cells are
radius R is calculated as an average of the different marked within the binary package.
Modeling of Sphere-Pac Fuel 797
basic cell into the one-fractional basic cells, the same the smaller pores into bigger ones to minimize sur-
procedures as discussed above for calculation of ther- face energy. Compared to the initial and intermediate
mal conductivity and heat transfer are applicable. stages of sintering, the disappearance of the remain-
More details are available in Botta and Hellwig.6 ing porosity is a slow process.
The situation can be compared to a soap bubble, Sintering mechanisms to be modeled are the fol-
where surface energy is causing a stress, which must lowing (compare Figure 6):
be balanced by the gas pressure inside. As two bub-
1. plastic and elastic deformation (bulk deformation),
bles are brought together, they merge to minimize the
2. surface diffusion (surface transport),
total surface energy.
3. grain boundary diffusion (bulk transport),
For a UO2 sphere having a diameter of 70 mm, the
4. volume diffusion (contains both bulk and surface
sintering stress on the sphere surface far away from
transport mechanisms),
the neck is 0.057 MPa. On the neck surface, with a
5. evaporation–condensation (surface transport),
neck ratio of 0.05, the sintering stress is –23 MPa.
6. thermal creep (bulk deformation),
In a similar way, the atomic bonding at the grain
7. irradiation creep (bulk deformation).
boundary is disrupted, which creates grain boundary
energy. During sintering, grains grow and coalesce to Spheres can be sintered with or without shrinkage
minimize grain boundary energy. and accompanying densification, depending on which
mechanism is exploiting the available driving forces.
3.25.3.1.2.2 External mechanical load For example, if temperature is increasing rather slowly,
The sintering stress is supplemented with an external necks can be formed by surface diffusion, that is,
mechanical load. At the beginning of irradiation, the without shrinkage. When higher temperatures are
necks are small, and the external load is concentrated reached, curvature differences have already decreased,
and amplified on the small contact areas. Theoreti- and the driving force for other sintering mechanisms
cally, in a point contact, the stress would be infinite. which cause shrinkage, have already been exploited.
When the neck grows, the stress in the contact area Which type of mechanism, surface or bulk trans-
decreases, and approaches gradually the applied port, dominates, is important as shrinkage at the begin-
external stress. ning of the irradiation could increase fuel temperatures
The stress between spheres is further reduced as by creating a gap between the fuel and the cladding.
new contacts between spheres are formed. Further, hot
isostatic pressing could be assumed. The stress is 3.25.3.1.3.2 Initial plastic and elastic deformation
hydrostatic; that is, the components of the stress are (plastic flow)
the same in all directions. In reality, the stresses in the Sintering stress is supplemented by an external
fuel are not uniform, causing the necks in the pressing mechanical load. At the beginning, the contact areas
direction to be larger than in perpendicular direction. between the spheres are small, and the external
mechanical load is amplified in these contacts. Even
3.25.3.1.3 Sintering mechanisms with low external loads, the stress at the contact
3.25.3.1.3.1 Classification points can exceed material yield strength. Plastic
The name of the sintering mechanism tells via which flow increases the contact area until the stress has
route the mass is flowing into the neck area. The diminished to a value which is approximately three
mechanisms can be divided in two categories: surface times the yield strength.21 This process is rapid and
transport and bulk transport. important only at the beginning of life. In SPHERE,
Surface transport mechanisms are moving material it is assumed to be instantaneous.
from the sphere surface to the neck surface, and
therefore, are not causing shrinkage or densification. 3.25.3.1.3.3 Thermal and irradiation creep
The driving force for these mechanisms is the differ- As the stress between spheres has decreased lower
ence in surface curvature between the neck and the than three times the yield strength, further deformation
surface far away from the neck. Curvature differences is caused by slower viscous creep (diffusional flow).
are decreased by mass flow, and the total surface Creep is usually caused by several different mechan-
energy is minimized. External mechanical load has isms. Important are the diffusion of atoms along the
no effect on these mechanisms. grain boundaries (Coble creep), the diffusion through
Bulk transport mechanisms move material from the the grains from the grain boundaries in compression to
sphere interior to the neck, causing spheres to grain boundaries in tension (Nabarro–Herring creep),
approach each other (i.e., shrinkage). These mechan- and the dislocation climb at high stresses and tempera-
isms are affected by external mechanical load, and tures (power law creep) (see also Section 3.25.3.2.1).
they are more important at higher temperatures. As these processes are diffusional, they are highly
Modeling of Sphere-Pac Fuel 799
Surface diffusion:
Evaporation–condensation: As temperature increases, less
Vapor pressure over the tightly bounded atoms in the
concave surface is higher than surface defects break away
over the convex neck area and diffuse to neck
Volume diffusion:
Grain boundary diffusion: From the surface
Of ‘new’ grain boundary. Because of
the high number of defects the atomic Volume diffusion:
diffusion is high From the ‘new’
grain boundary
‘New’ grain boundary:
Formed by the contact of
spheres and their missaligned
crystal structures
Thermal and irradiation creep:
Initial plastic deformation:
After the stress in contact has
In initial small contacts, even low
declined <3.yield strength, further
external loads exceed the material
defomation is caused by slower
yield strength. Fast plastic flows
creep (e.g., stress-assisted volume
increases the contact area until the
diffusion through the grains)
stress <3.yield strength
Figure 6 Mass transport routes during sintering. Data from International Atomic Energy Agency. Nuclear Fuel Behaviour
Modelling at High Burnup and Its Experimental Support; IAEA-TECDOC-1233; IAEA: Vienna, 2001; with permission
from IAEA.
temperature sensitive. Unlike the other sintering causing it to be an excellent path for atomic diffusion
mechanisms, creep and initial plastic and elastic defor- transporting material from the inner part of the neck
mations can be governed by empirical macroscopic to the neck surface (Figure 6).
deformation formulas.21 Here, the term ‘grain boundary diffusion’ is used
solely to describe the transport along this ‘new’ grain
3.25.3.1.3.4 Surface diffusion boundary between the spheres. Diffusion along the
The sources of atoms in surface diffusion are the defects grain boundaries inside the spheres is not as impor-
on the sphere surface on atomic scale. Less tightly tant, and it is therefore, neglected.21
bound atoms are breaking away as temperature is The activation energy of this process is usually
increasing. For surface diffusion, the activation energy somewhere between the activation energies of surface
is low compared to other mass transport mechanisms, diffusion (low) and volume diffusion (high). During
causing it to be more important at lower temperatures. sintering, free surface area and curvature gradients
Because material is replaced from the sphere sur- decrease. At the same time, new grain boundary areas
face to the neck surface (Figure 6), surface diffusion are formed. As a consequence, surface diffusion loses
does not cause spheres to approach each other (i.e., its importance to grain boundary diffusion later
no shrinkage). in life.
Zone III: The transition zone between the slow Describing the total effect with a simple Arrhenius-
and fast sintering regions. type equation is thus always a rough estimate and must
Zone IV: Slowly sintering fuel, where tempera- be considered more as a mathematical fit for existing
tures are low (<800 C), and irradiation creep is the creep data. As creep processes are diffusional, they are
dominating sintering mechanism. highly sensitive for temperature and for existence
In Zone II, fuel has proceeded through the initial of higher carbide phases. Creep measurements are
and intermediate stages of sintering to the final stage. difficult to perform, they show large scatter, and
As the irradiation proceeds, this stage is reached are partly done with macroscopic samples not suitable
at continuously lower temperatures. The original for modeling microscopic local creep in single spheres
sphere structure is not visible, and fuel resembles (the effect of porosity, cracks, and inhomogeneous
pellet fuel. Pore network has collapsed into discrete deformation in macroscopic samples).
pores in fuel matrix, and smaller pores coalesce and It is recommended that the calculation takes into
shrink to form bigger pores. account whether the stress is lower or higher than the
In this zone, higher densities and shrinkage transition stress, that is, the border value between
are reached, if bulk transport mechanisms have linear and power law creep. Five different models
been dominating during the sintering (see Section are listed below for uniaxial thermal creep rate. Fur-
3.25.3.1.3). However, the amount of remaining poros- thermore, an option to calculate time dependent
ity depends also on several other processes than the creep is also shown.
transport mechanisms. For example, if the fuel grains
grow faster than the pores can move, the pores on 3.25.3.2.2 Transition stress
the grain boundaries might be trapped inside the At transition stress (stran), the creep rate changes
fuel grains. If this happens, pore mobility is further from a linear stress dependence to a creep rate pro-
decreased, and the pores can disappear only with portional to stress to a power n. The transition stress
slower diffusion controlled mechanisms, and densifica- is only mildly dependent upon temperature but more
tion slows down. strongly affected by grain size. The transition stress is
The transition Zone III is relatively narrow, owing defined by23
to the small temperature difference needed to go
from slow to rapid sintering, and the large tempera- stran ¼ 16:5474=rgrain
0:5714
½4
ture gradient in fuel. In this zone, the effect of pres-
where stran is transition stress (MPa) and rgrain is fuel
sure on necking is pronounced.
grain size (mm).
In Zone IV, fuel temperature is less than 800 C,
When the actual stress s is less than transition
sintering is slow, and fuel never reaches the interme-
stress stran, the power law creep is zero. For stresses
diate stage of sintering. The original sphere structure
greater than stran, stran is used in the linear part of the
is clearly visible, and the porosity is in the form of a
creep rate equation, and s in the power law term.
tunnel-like network. On polished fuel cross-sections,
In practice, in Sphere-pac fuel, s stran, and power
the sizes of the spheres, the necks, and the pores
law creep is zero.
appear smaller than their true values because of the
randomness of the selected cross-section.
3.25.3.2.3 Time-independent thermal
creep rate
The thermal creep rate deS/dt can be constant
3.25.3.2 Creep
over time, or, it can be enhanced immediately after
3.25.3.2.1 Steady-state thermal creep introducing a higher stress (see below). For the time-
Creep is usually caused by several different mechan- independent steady-state thermal creep rate deS/dt,
isms. Important are the calculation code uses five models for different
diffusion of atoms along the grain boundaries (Coble fuel materials. These are given in Table 2.
creep),
diffusion through the grains from the grain bound- 3.25.3.2.4 Time dependence of thermal
aries in compression to grain boundaries in tension creep rate
(Nabarro–Herring creep), and When the stress is higher than in any other previous
dislocation climb at high stresses and temperatures time step, the strain rate is enhanced and is time
(power law creep). dependent; compare the following23:
802 Modeling of Sphere-Pac Fuel
Table 2 Material-specific models for steady-state ther- gas. The driving force for this mechanism is the
mal creep rates deS /dt; these models are implemented in variation of the vapor pressure of the solid with
the code
temperature causing the molecules to evaporate on
Model Fuel type Reference the hotter side and condensing on the colder side.
The net effect is the pore movement toward the fuel
1 UO2 MATPRO23 Recommended
2 Carbide Hall et al.24 and Recommended
rod center.30
Tokar25 The gas composition and pressure in a pore are
3 Nitride UNCLE-T code assumed to correspond to the gas in rod free space
4 MOX (not MATPRO23 Recommended at the moment the pore is closed, that is, at the
UO2) moment when the fuel is assumed to turn from
5 MOX JNC Recommended
the initial sphere structure to solid sintered body.
Thus, the gas mixture within a pore consists primar-
ily of helium and quantities of xenon and krypton,
deT deS which increase with burnup. Small error is made, as
¼ 1 þ 2:5 exp 1:2E 6tcreep ½5 in reality the fraction of fission gas grows bigger
dt dt
as the pore sweeps the gas out of the fuel matrix
where deT/dt total creep rate (s1), deS/dt is steady- on its way.
state (time-independent) creep rate (s1), tcree is time As the porosity is conserved, in every radial fuel
since the stress increase to a level higher than in any ring, the change in porosity must equal the net flux of
previous time step (s). the porosity:
Unlike in UO2, linear creep is assumed to be zero
in carbide fuels. In that case, the creep rate is calcu- @p 1 @
¼ rJp ½7
lated with a power law creep equation: @t r @r
de DHcreep;th where p is smear porosity of the fuel ring, r is radius
¼ Acreep;th sn exp ½6 at the calculation node (cm), and Jp is flux of
dt RT
porosity (cm s1)
where deT/dt is total creep rate (s1), Acreep,th is The porosity flux Jp is as follows:
4.36 107 (s1 MPa2.4), tcree is time since the stress
increase to a level higher than in any previous time Jp ¼ v p p ½8
step (s), s is stress (MPa), n is 2.4, DHcreep,th is activa-
where vp is the pore velocity (cm s1)
tion enthalpy (¼120.880 cal mol1), R is gas constant
For the pore velocity, the model of Olander30 is
(¼1.985 cal K1 mol1), and T is temperature (K).
chosen and recommended.
Parameters in eqn [6] are based on reviews in
Blank26 and Matzke27 and original measurements
3.25.3.3.2 Olander’s model for pore velocity
done by Hall28 on hyperstoichiometric (U0.85Pu0.15)
The pore velocity in pure UO2 is calculated as fol-
C1.05. Equation [6] is based on data with high-density
lows (see Olander30 eqn [14.11]):
samples, and thus can be directly implemented to
microscopic creep in single spheres.26
4 106 O DHvap DSvap DHvap 1 dT
For a comparison of the above creep model vp
¼ Dg e k e kT ½9
kT kT T dr
(eqn [6]) with often-referred creep measurements
of U–Pu carbides done by Tokar25 (see Wallin29 and where vp
is pore velocity in pure UO2 (Å s1), Dg
Figure 6). diffusion coefficient of the UO2 molecule in the
He/Xe gas mixture at temperature T and pressure p
3.25.3.3 Pore Migration and Restructuring 3=2
T 1
A pore migration (fuel-restructuring) model used in Dg
½10
2000 p
the PSI predicting codes is described below.
diffusion of UO2 molecules from the hot side of DXe is diffusion coefficient of the UO2 molecule in Xe
the pore to the cold side through the contained at 2000 K, 1 atm ¼ 0.9 cm2 s1 and DHe
is diffusion
Modeling of Sphere-Pac Fuel 803
coefficient of the UO2 molecule in He at 2000 K, 3.25.3.3.3 The finite difference solution of
1 atm ¼ 11 cm2 s1, yXe is mole fraction of Xe, P is porosity equation
gas pressure in the pore SPHERE uses following finite difference equivalent
of eqn [7] in cylindrical geometry:
T =Tsint psint ðatmÞ ½12
t t
pit þDt pit 1 riþ1 viþ1 piþ1 ri vit pit
Tsint is the temperature at which the pore has closed, ¼ ½14
dt ðriþ1 þ ri Þ=2 dr
that is, solid sintered body has formed (the neck ratio
in finest fraction ¼ 0.45), psint is the gas pressure when where pit is porosity in fuel ring i at time moment t, ri
the pore has closed, DHvap is heat of vaporization of is radius of fuel ring i, pore velocity in fuel ring i at
UO2 ¼ 567 kJ mol1, DSvap is entropy of vaporization time moment t, and vit calculated from eqns [9]–[13].
of UO2 ¼ 150 Jmol1 K1, O is atomic volume of
UO2 ¼ 41 Å3 3.25.3.4 Grain Size and Grain Growth
The reduction of pore velocity because of
the combined effects of impurity accumulation One of the most important parameters for fission gas
on the hot face and condensation rate limitations on release (FGR) is the fuel grain diameter, as it gives
the cold face is given by eqn [14.18] in Olander30: the distance that the fission gas has to travel before
reaching the grain boundary. The bench-marking
( " #1 )
DHvap 1 dT against PSI experiments showed the grain size to be
1 yI 1 þ d even a more important parameter than temperature
vp kT T dr p
¼ ½13 concerning the FGR.
vp
Dg pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi The as-fabricated grain size, being the starting
1þ 1E 7J =erg 2pmkT
akT d point for the growth, has to be investigated and known
where vp is the final, reduced pore velocity (Å s1), from the quality inspection of the fuel following the
yI mole fraction of soluble impurity at the hot side fabrication process. The fuel grain size model integrates
of the pore, ðdT =dr Þp ¼ 4ðdT =dr Þ (eqn [14.1] in the grain growth from a start value until the grain size
Olander30), ðdT =dr Þp mean temperature gradient reaches a limiting value, after which it is kept constant.
in the gas contained in the pore, ðdT =dr Þ unperturbed The explanation for that is that the calculated grain
matrix temperature gradient, d pore thickness (typing diameters for a typical fuel pin lifetime were found to
error in Wallin29; should be given in cm), m mass of the be much larger than those observed in postirradiation
UO2 molecule ¼ 238 þ 32 ¼ 270, a is condensation examinations (PIEs). Therefore, the predicted grain
coefficient defined as the fraction of the molecules diameters are limited to a maximum value which is a
hitting the solid surface that stick to the surface ¼ 0.9. default value or can be specified in input.
An example has been given in Olander30 using In modeling the AC-3 experiment (see Section
following values: 3.25.5.2.1), the start value for grain size (10 mm) was
taken from pretest technical specifications, and the
yI ¼ 0:01 maximum upper limit for grain growth (100 mm)
d ¼ 0:001cm from ceramographic sections taken from PIEs.
The grain growth kinetics has to be considered by
With these values, the velocity reduction predicting tools. The model presented below, refers
vp =vp
¼ 0:40. If the pore thickness would be to an atomic process in which large grains grow at
0.002 cm, vp =vp
¼ 0:59. The pore thickness d ¼ 0.001 the expense of smaller ones. The since long time
cm is typical for a solid sintered body in Sphere-pac used cubic rate law, in which the grain growth is
fuel, and thus, this value was used for predicting in limited by pore migration, presented among others
PSI. The value yI ¼ 0.01 has been used so far at PSI, in Nordström31 is used:
but a possible future development might be to change
D3 D03 ¼ at eðRT Þ
Q
probability that a gas atom is captured by a bubble or The volume per atom (Vatom) depends on gas
an as-fabricated pore is pressure according to van der Waals equation of state:
gb ¼ 4pDrb Nb ½20 kT kT
Vatom ¼ þB ¼ þB ½25
p 2g=r b þ s
gp ¼ 4pDrp Np ½21
where kT/p is ideal gas law dependence and B van
where rb is intragranular bubble radius (cm), rp intra- der Waals volume of xenon atom, 85 Å3 atom1
granular pore radius (cm), Nb intragranular bubble (Olander30). At high pressures, Vatom approaches B,
concentration (bubblescm3), and Np intragranular which thus can be regarded as the volume occupied
pore concentration (pores cm3) by the atom proper.
As the absorption probability depends on The number of atomsper bubble is then
bubble and pore sizes and concentrations, they must (Olander30),
be integrated at the same time with the diffusion Bubble volume 4prb3 =3
eqn [17]. mb ¼ ¼ ½26
Vatom kT =ð2g=rb þ sÞ þ B
3.25.4.1.2.4.2 Bubble and pore radii (rb, rp) The In eqn [26], mb is known (from eqn [23]), and bubble
bubble radius is calculated by first solving the num- radius rb can thus be solved.
ber of atoms in the bubble. Then, assuming that the The pore radius is calculated using a similar
bubble is in mechanical equilibrium with its sur- method except that a correction is done for the initial
rounding, the radius can be calculated. He gas amount in the pore.
The number of atoms in bubbles in a unit volume
(cb) is solved from the following equation: 3.25.4.1.2.4.3 Bubble concentration (Nb) A for-
mer method to calculate the bubble concentration
@c b assumed that, at the end of nucleation stage, the gas
¼ g b cm bc b ½22
@t production rate exactly equals the rate of gas atom
The number of atoms in a single bubble (mb) is then resolution from the bubble, and that the bubble concen-
tration stays constant throughout the whole irradiation.
cb
mb ¼ ½23 However, the calculated concentrations were not in
Nb line with empirical observations (e.g., Wood and Mat-
When the number of atoms in a bubble is known, the thews37,38). Therefore, two new, alternative models are
bubble radius can be calculated. As a new bubble suggested: stationary trapping and empirical formula.
nucleates, it faces an immense pressure from the
129.4.1.2.4.3.1 Stationary trapping To keep things sim-
surrounding lattice, and thus, deviation from ideal
ple, stationary trapping conditions were assumed.
gas law must be considered. In fact, the deviation is
That is, the amount of trapped gas is assumed to
small when the diameter is greater than 100 nm, and
remain unchanged. At the end, the bubble concentra-
therefore, would not have any effect in larger pores
tion should eventually saturate at the value (White
(diameter 1 mm), but to keep things simple, the
and Tucker36):
same calculation routine with the correction is used
for both bubbles and pores. Production rate of bubbles 2F 1:52a
Nb ¼ ¼
When the bubble is in equilibrium with its sur- Destruction rate of bubbles 2F pðZ þ rb Þ2 lf
rounding, the gas pressure inside is balanced by the
1:52a
hydrostatic stress in the fuel and the surface tension ¼ ½27
forces that tend to contract the bubble. As new atoms pðZ þ rb Þ2 lf
enter the bubble, the bubble grows until the pressure where a is number of bubbles generated/fission
has again dropped to a level that is equal to the (¼24), lf is the length of the fission spike (cm)
contracting forces: (¼6 mm), and Z range of influence about the path of
2g fission spike (cm) (¼1 nm)
Gas pressure ¼ p ¼ þs ½24
rb
3.25.4.1.2.4.3.2 Empirical formula Bubble concentra-
where g is the surface tension (¼surface energy) of tion has a strong effect on trapping of the gas (eqn
the fuel (erg cm2 or Jm2) and s external hydro- [20]). In eqn [27], both a and Z have large uncertain-
static stress in the fuel. ties, and rb is the result of the calculation containing
806 Modeling of Sphere-Pac Fuel
also uncertainties. Therefore, an alternative is to esti- a perfect sink because the resolution from the grain
mate Nb using an empirical formula. The best results boundary back to the grain (intergranular resolution)
in modeling experiments (e.g., the PSI-Halden exper- increases the gas concentrations close to grain bound-
iment IFA-550.9, Section 3.25.5.2.3) was achieved ary. Turnbull’s approximation (Turnbull35 and White
with Baker’s empirical formula (Baker39 or White and Tucker36) is adopted. The gas concentration at
and Tucker36): the grain boundary is set equal to the gas concentra-
tion at resolution depth:
Nb ¼ 1:52 1021 =T 3:3 1017 ½28
cm ¼ bgb lNgb surface =ð2DÞ at the grain boundary ½33
3.25.4.1.2.4.4 Pore concentration (Np) The as-
fabricated intragranular porosity and pore radius where bgb is resolution probability from the
(1 mm) have to be specified. grain boundary (s1), l is average distance gas
The pore concentration is calculated from atoms are knocked back to fuel matrix (cm), and
Ngb surface is number of atoms per grain surface
Np ¼ pI = 4prp3 =3 ½29 area (atoms cm2).
In Turnbull,35 Ngb surface and the product bgbl have
where pI is as-fabricated intragranular porosity.
been approximated by a method of ‘trial and error’ in
order to reproduce the Halden (Vitanza) 1% thresh-
3.25.4.1.2.4.5 Intragranular resolution (b) Reso-
old for FGR. Ngb surface can be calculated in different
lution of fission gas from intragranular bubbles back
ways. Another way is shown in Section 3.25.4.1.4.
to matrix is caused either by simple collisional
However, the value of the product bgbl has large
knock-on, or through high-temperature and pressure
uncertainties. After personal communication with
shocks created by fission spikes. In UO2, fission spikes
Turnbull, the product bgbl was determined using a
are able to destroy a single fission bubble completely.
similar method than in Turnbull,35 that is, its value has
As the bubbles are destroyed, the gas they contained
been determined to reproduce the FGR measured
is resolved back to fuel matrix. Actually, without
after a base irradiation test in Gösgen pressurized
resolution, FGR would be negligible.
water reactor (PWR) (Section 3.25.5.2.2). This
The last term (bcb) in diffusion eqn [17] represents
resulted in the formula to be used:
the resolution of the gas trapped in bubbles; b is the
fraction of the gas in bubbles that is resolved per F
bgb l ¼ bgb l ref 1:5 ½34
second. This is the same as the fraction of bubbles Fref
destroyed, or the fraction of unit volume in which where (bgbl)ref is Turnbull’s value for bgbl at the refer-
bubbles are destroyed by fission spikes per second. ence fission rate (cm s1) and Fref is the reference
A ‘touch’ of a fission spike is assumed to be enough to fission rate (9.21 1012 fissions cm3 s1).40
destroy a bubble, and therefore, one fission fragment For different models and codes, the value of bgbl is
destroys all bubbles inside a cylindrical volume: slightly different. This is reflected in the factor 1.5 in
b 0 ¼ 2pðZ þ r b Þ2 lf ½30 eqn [34]. However, using the same calculation model
for other fuel rods, its value is not expected to change.
where b0 is resolution volume per fission (cm per 3
This assumption must be proofed when calculations
fission) and 2 is two fragments per fission. are done in other fuel tests. It is worth mentioning
If eqn [30] is corrected for the bubble size distri- that the preliminary value given in eqn [34] is rather
bution (as shown in White and Tucker36), it gets close to the value given in Turnbull.35
the form
b 0 ¼ 3:03pðZ þ rb Þ2 lf ½31 3.25.4.1.2.5 Intergranular gas (gas at the grain
The resolution rate b caused by all fission fragments boundary)
in a unit volume per unit time is 129.4.1.2.5.1 Gas flux to grain boundary The gas
flux to grain boundary consists of two parts:
b ¼ b0 F ½32
1. Diffusional flux from the grain, which is according
to Fick’s law:
3.25.4.1.2.4.6 Boundary condition of the diffusion
equation When solving the diffusion equation in @cm
Fluxdiff ¼ D ½35
the grain, the grain boundary cannot be considered as @r
Modeling of Sphere-Pac Fuel 807
2. Gas swept by moving grain boundary, which is whole grain surface, but in light of PSI experimental
according to the model of Kogai41: test results, this was not considered necessary.
In the models presented in Turnbull35 and White
@r
Fluxsweep ¼ aA ½36 and Tucker,36 interlinkage and gas release are
@t
initiated when the number of atoms per grain surface
where a ¼ 0.001 and A is total gas inventory of grain. area (Ngb surface) reaches a maximum value around
Because the model specified in eqn [36] has not 5.0 1015 atoms cm2. This condition is slightly dif-
been confirmed, it is recommended to keep it as an ferent from the one shown above in eqn [33], but at
option. Moreover, according to White,36 the diffusion the end, the maximum concentrations calculated are
and resolution effects predominate in grain boundary quite close to 5.0 1015 atoms cm2.
gas accumulation.
3.25.4.1.3.2 Total gas release in a fuel ring
3.25.4.1.2.5.2 Atoms per grain boundary bubble The total flux out of all the fuel grains in a radial fuel
Dividing the grain surface area by the number of ring is calculated from the following equation:
grain bubbles on it, we get the area per grain bound- Fluxring ¼ 4prgrain
2
Ngrains Fluxdiff þ Fluxsweep ½40
ary bubble. The radius rgb max of this is as follows:
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi where rgrain is fuel grain radius, Ngrains the number of
1 grains in the fuel ring
rgb max ¼ ½37
pNgb bubbles Mass in the fuel ring
where Ngb bubbles is number of grain boundary bub- Mass of a single fuel grain
bles per unit surface area. Vring rsmear; ring
¼ ½41
As the diffusion along the grain boundary is fast rsmear; grain 4=3prgrain
3
compared to diffusion inside the grain, atoms arriving
to this area are assumed immediately to enter the rsmear, ring is smear density of the fuel ring, rsmear, grain
grain boundary bubble. Then, the number of atoms and smear density of the grain.
per grain boundary bubble (mgb bubble) can be calcu- Before the grain boundary bubbles are interlinked,
lated from the equation the total flux Fluxring is added to the gas in the grain
boundaries. After the grain boundary bubbles have
@mgb bubble
¼ prgb2
max 2Fluxdiff þ Fluxsweep ½38 interlinked forming a tunnel network (rgb bubble ¼ rgb
@t
max), the grain boundaries are saturated, and Fluxring
The diffusional flux is multiplied with two as it con- is excess gas which must be vented off. Gas release is
tains also the flux from the adjacent grain. initiated by adding Fluxring to the gas release.
The bubble radius is calculated in much the same After part of the gas in the tunnel has been vented
way as the intragranular bubble radius, that is, assum- off, the gas pressure in the tunnel decreases, and the
ing thermodynamic and mechanical equilibrium with tunnel collapses. The rest of the gas in the tunnel is
the surrounding matrix. Because the grain boundary isolated and forms new bubbles. As the new bubbles
energy and the surface energy of UO2 differ, the grow, they interlink again, and gas is released, and the
grain boundary bubbles have lenticular shapes, which tunnel collapses. Thus, even after release has initiated,
slightly complicates the calculation. The recom- some gas is constantly retained at grain boundaries.
mended method is a standard one found in open This is taken into account by using the model described
literature (see, e.g., White and Tucker36). in White and Tucker.36 It is assumed that the gas
amount in the grain boundaries increases until its max-
3.25.4.1.3 Gas release imum value is reached just before interlinkage. After
3.25.4.1.3.1 Condition for gas release interlinkage, it is not released but stays in grain bound-
(interlinkage) aries, that is, the release consists only from the excess
Grain boundary bubbles are assumed to interlink gas entering grain boundaries after the interlinkage.
when they cover the whole grain surface, that is:
3.25.4.1.3.3 Gas release from the columnar
rgb bubble ¼ rgb max ½39
grain area
where, rgb bubble is radius of grain boundary bubble (cm). At high temperatures (1600–1700 C), fuel evapo-
Interlinkage (and gas release) could be established rates at the hotter side of a pore and condenses at
also before the grain boundary bubbles cover the the cooler side resulting in pore movement toward
808 Modeling of Sphere-Pac Fuel
the rod center. Moving pores leave behind columnar the bed load, the more pronounced is the initial
grains. It can be assumed, that all fission gas (in fuel necking (caused by plastic deformation) and the
matrix, pores, in grain boundary, etc.) in that fuel ring lower is the initial center temperatures. After a
is swept into moving pores and released into rod open short time, diffusion causes the differences in necking
volume. If later pore movement ceases, normal fission (and therefore in center temperature) to become
gas accumulation continues. insignificant. A real irradiation would, of course, con-
However, a minor shortcoming of the model is that sist of a preceding start ramp, for example, about
the FGR model assumes only equiaxed grains and 10 W cm1 h1 can be assumed. During such a start
cannot take into account the special shape of the ramp, different processes occur simultaneously. Differ-
columnar grains. This has no effect by normal thermal ences in thermal expansion cause a well-defined bed
reactor temperatures, which are too low to create load. This bed load is – at least partly – released as
columnar grains. In fast breeder reactor (FBR) condi- diffusion processes and plastic creep occur at tempera-
tions, the shortcoming is not important because gas is tures above about 1400 C. Plastic deformation is only
released from the columnar grains mainly by moving a reaction to bed load and therefore it stops when the
pores and not by diffusion. If temperature decreases bed load decreases. Diffusion processes are causing
later to a lower level, the release mechanism in colum- necking, which can cause shrinkage, even without
nar grain zones would again be diffusion. However, the any bed load. An irradiation test with Sphere-pac
effect of this is limited because of the following: fuel specifically devoted to investigate the restructur-
ing effects during the start-up phase is presented
The characteristic distance that an atom must dif-
below (see Section 3.25.5.2.4 for the FUJI program).
fuse to arrive on the grain boundary is the width of
To be modeled is how strong the loading of the
the columnar grain, not its length. Therefore,
cladding is during the whole irradiation period, and
assuming a spherical grain, the fission gas diffusion
how pronounced the necking is in the outer fuel region
to the grain boundary is only slightly overestimated.
after the start ramp. The necking in the inner fuel
The grains have been ‘cleaned up’ of the fission gas
region is of no real importance as experiments showed
by moving pores.
that the fuel is nearly immediately modified to porous-
Uncertainties in later much smaller diffusional
pellet-like fuel. On the contrary, the necking in the outer
release have only a minor effect on total gas release.
fuel region could be of importance later in life when
3.25.4.1.4 Final comments FGR deteriorates the conductivity of the surrounding
At PSI, modeling the irradiation of Sphere-pac fuel gas. It is also important whether a gap can be formed
with the above described mechanistic model for FGR during the early stage of irradiation due to shrinkage.
has so far been very successful, as the comparison of 3.25.4.2.1.2 Calculation of bed pressure
the experimental results with experimental data has The prediction of the bed load is not satisfactorily
demonstrated. This model could predict about the solved for the PSI thermal modeling codes. However,
same fractional FGR as that measured. Furthermore, the sintering model described above (see Section
the predicted onset and evolution of gas release also 3.25.3.1) requires such a value on the mechanical
could clearly be correlated to the, with pressure trans- pressure. Presently, such a value has to be specified
ducer, measured values as well as the power history; and is assumed to stay constant during the irradiation.
compare Section 3.25.5.2. However, two parameters Generally, such a value and its time dependency are
need to be set: the temperature above which columnar not known; that is, its calculation requires an integral
grains are formed (set as 1600 C), and a parameter for thermomechanical model.
gas resolution from grain boundaries. As a solution, a subroutine can be added which
gives an estimate of the bed load in fresh fuel during
3.25.4.2 Mechanical Properties the initial start-up ramp by using the results of
an external mechanistic code (e.g., PINTEMP, see
3.25.4.2.1 Bed pressure
Section 3.25.5.1), calculated for a similar fuel. The
3.25.4.2.1.1 Influence of shrinkage and bed load
procedure is as follows:
on fuel behavior
It has been shown that the bed load has an effect on 1. After the fuel center temperature has been calcu-
the initial sintering. Calculations made in PSI were lated, the subroutine finds the corresponding dif-
performed by assuming 300 W cm1 linear heat rate ference in thermal expansion of fuel and cladding
without any start ramp. They showed that the higher (calculated by PINTEMP).
Modeling of Sphere-Pac Fuel 809
2. The net expansion is calculated by subtracting weight. This load can be given either to the next
the shrinkage from the difference in thermal section (which would result in a high bed pressure
expansion. for the lower sections), or it can be given to the
3. Finally, the subroutine finds the bed load (calcu- cladding by friction (which would cause the lowest
lated by PINTEMP) corresponding the net expan- and most homogeneous bed pressure). Assuming the
sion. If the net expansion is negative, the bed load second case, as this will yield the lowest pressure,
is set to the minimum pressure of 0.012 bar (see we can write
below).
Fw ¼ Ff ½42
The above option to calculate the bed load can be
where Fw is force due to the weight of a fuel section
chosen or the bed load values given in input file are
and Ff is force due to friction.
used as before. If this new option is selected, the
shrinkage caused by plastic and elastic deformations d2
Fw ¼ p lrg ½43
is set as zero as they are already taken into account 4
in PINTEMP calculations. where d is inner diameter of the cladding, l length of
The procedure to estimate the bed load has its the fuel section, r density of the fuel, and G is accel-
limitations, as it is valid only for fresh fuel during the eration due to gravity.
initial up-ramp when the necks are small. Also, an
approximate linear relationship between load and Ff ¼ mðpdlÞLbed ½44
expansion is assumed. As the present external mech- m is friction coefficient, d inner diameter of the clad-
anistic code takes into account only instantaneous ding, l length of the fuel section, and Lbed bed load.
plastic and elastic necking and neglects the slower Assuming the parameters d ¼ 0.006 m, r ¼ 8 kg dm3,
processes, which are integrated in the thermal code, g ¼ 10 N kg1, and m ¼ 0.1 (typical for metal/ceramic
the neck sizes in both code calculations differ at the friction), a bed load of 0.012 bar is achieved.
same center temperature. However, this difference Any decrease in friction will cause higher bed load
was found to be only a few percents. because of the weight of the fuel column and the
loading of the plenum spring.
3.25.4.2.1.3 Estimation of the minimum
bed pressure 3.25.4.2.2 Cladding deformation
A rough estimation of the lower limit for the bed One very characteristic difference between pellet and
pressure can be given by the idea that in a fuel pin Sphere-pac fuel pins is the clad behavior, shown in
each Sphere-pac section must carry at least its own Figure 8. The pellet clad starts with a pronounced
As fabricated
10.84 PIE after base irr, 0° + 90°
PIE after ramp, 0° + 90°
10.82 S04 (sphere-pac) Transuranus-calc, EOL
Clad outer diameter (mm)
10.80
10.78
10.76
10.74 N06 (pellet)
10.72
10.70
10.68
10.66
50 100 150 200 250 300 350 400 450 500
Axial position (mm)
Figure 8 Comparison of the clad radial behavior for Sphere-pac and pellet fuel, as fabricated, after basic irradiation
as well as ramp experiment. Reproduced from Nordström, L. Å. Comparative Irradiation and Transient Test of Pellet and
Sphere-pac Fuel (IFA-550.11). In EHPG-Meeting, OECD Halden Proceedings HPR-359, Storefjell, Noway, Sept 2002, with
permission from OECD Halden Reactor Project.
810 Modeling of Sphere-Pac Fuel
creep down (no internal support from fuel) until gap are described in detail in Wallin et al.43 There
closure, followed by a creep out caused by the fuel are also several PSI internal documents available
swelling. TRANSURANUS42 is a computer program describing the newest version SPHERE-3, for
for the thermal and mechanical analysis of pellet-fuel example, Wallin,32 and how to use it.44
rods, developed at the Institute for Transuranium Besides implementing the models presented in
Elements (ITU). Using this code to compute pellet previous sections, the newest SPHERE-3 implemen-
fuel behavior shows a gap closure just at the end of tation offers the interesting option to conduct a
base irradiation, confirmed by the very flat profilo- series of successive calculations (RESTART option),
metry curve. During ramp, the fuel swelling provokes that is, the possibility to make subsequent calcula-
creep out of the cladding, also seen in the measured tions with the same fuel, mainly with new settings
curve. However, the ramp time may be too short to and under other conditions. This enables, for exam-
make that still more obvious. The TRANSURANUS ple, the predicting of test fuel rods, which after
calculated EOL diameter fits the measured data very basic irradiation in a commercial reactor, followed
well. On the other hand, the Sphere-pac fuel, without by refabrication or reinstrumentation, were ramped
any fuel-to-clad gap, prevents a cladding creep down. in a test reactor, although the fill gas and plenum
From the beginning, the fuel swelling results in a clad volume in rod or the reactor type was changed;
diameter growth, which however is very modest, as compare Sections 129.5.2.2 and 129.5.2.3.
the main part of the mechanical stresses caused by
the fuel–cladding interaction can be reduced and
compensated by the deformation of the spheres in 3.25.5.2 Experimental Validations of the
their mutual contact points. SPACON and SPHERE Codes
In no PSI experiments with Sphere-pac fuel, a Over the years, many in-pile experiments with
clad–fuel gap could be observed after irradiations Sphere-pac fuel were carried out in different reactors
were made at very different linear heat rates and to and in cooperation with various external partners.
different burnups. The objective was not only to increase and prove
the PSI knowhow of Sphere-pac fuel, but also to
bench mark the PSI modeling tools, particularly,
3.25.5 Implementation and Validation the SPHERE code (see Section 3.25.5.1). Experi-
3.25.5.1 The Predicting Tools SPACON ments were carried out with different Sphere-pac
and SPHERE fuel material, for example, UO2, MOX, carbide, and
nitride. The test fuel and rods were fabricated in PSI
At PSI, some computer codes were developed for the Hotlaboratory. An overview and short description of
modeling of Sphere-pac fuel. The aim was to create the most important experiments are given below.
a tool for design and layout of different experiments
with Sphere-pac fuel, and of course, also to get a
3.25.5.2.1 AC-3
better theoretical understanding of all the effects
The experimental results of this comparative study of
and processes taking place during the irradiation in
pellet and Sphere-pac carbide fuel can be found in
a reactor. Several in-core experiments were carried
Chapter 3.11, Sphere-Pac and VIPAC Fuel.
out in order to validate and bench-mark the codes
As a part of the validation process of SPHERE,
and their single models (compare Section 3.25.5.2).
it was decided to compare the code calculations with
The heat transport mechanisms in fuel pins
the irradiation test results of the AC-3 experiment.
containing sphere-pac fuel are addressed with the
The pin 3L115, which operating at highest rating
theoretical model SPACON6 implemented in a
reached a peak burnup of 8.2%FIMA (fissions per
C code. The corresponding theory is presented in
initial metal atom), was selected to be modeled with
Section 3.25.2.
SPHERE-2. The following observations were made
The thermal behavior of Sphere-pac fuel is
in the comparison with ceramographic sections29:
predicted with the code SPHERE, written in
FORTRAN. The code, together with many of the The neck ratio showed good agreement for the
physical models, was presented in different meetings, restructured radial part (27% and 18–21% for
for example, Wallin et al.7 The phenomenological the SPHERE-2 and measured, respectively), but
fission gas model in SPHERE and some results of was calculated to be considerably smaller at the
the qualification process against experimental data outer fuel area close to the cladding.
Modeling of Sphere-Pac Fuel 811
The agreement between the quantity of the fuel less coincidental, as the available data for the FGR
that had lost its sphere structure and the quantity model was very limited.
transformed into porous-pellet structure was very
good, 52.5% and 48% respectively for SPHERE-2.
Furthermore, the SPHERE calculation gave an 3.25.5.2.2 Siemens-PSI Gösgen PWR
overall fuel structure similar to that observed in This experiment was followed by the PSI-Halden
PIE and the calculated transition zone was in cor- BWR irradiations (see Section 3.25.5.2.3), which
rect radial position (see Figure 9). are designed as ramp tests. For experimental details
At the end of the irradiation, the calculated FGR see Chapter 3.11, Sphere-Pac and VIPAC Fuel. The
fraction was 5.17%. At the PIE a release fraction of comparison with the code modeling is presented in
4.88% was measured. However, this seemingly Nordström et al.45 A summary of the most important
good agreement between the measured and calcu- results are shown in Table 3.The modeling predictions
lated FGR fractions could be considered more or are presented together with the ramp experiment.
2000
s 45
9 day
r 1900
Afte ys
r 1 8 da 40
Afte 1800
1700 35
End o
1600
iation
18
30
f irrad 1500
End o
f irrad
day
9
da
1400 ys s
25
iation
1300
20
1200
1100 15
Fuel temperature (°C) Neck ratio (%)
1000
10
900
800 5
Pin radius (mm) Pin radius (mm)
700
Radius of compact sintered body:
– Calculated by SPHERE (upper graph)
– Measured from ceramographics (below)
Figure 9 Comparison of neck ratio and size of compact sintered body at end of life as calculated (top), with measurements
on ceramographic section at corresponding position by postirradiation examination (bottom). Data from International
Atomic Energy Agency. Nuclear Fuel Behaviour Modelling at High Burnup and Its Experimental Support; IAEA-TECDOC-
1233; IAEA: Vienna, 2001; with permission from IAEA.
812 Modeling of Sphere-Pac Fuel
Table 5 Neck ratios for the different fractions as a function of fuel radius
Position from fuel Coarse– Medium– Medium– Fine–coarse (%) Fine–medium (%) Fine–fine (%)
center (mm) coarse (%) coarse (%) medium (%)
1 30 45 30 100 75 55
2 40 25 80 60 45
3 30 20
4 15 60 50 30
Estimated from the fuel ceramography taken from the PSI-Halden BWR experiment.
Modeling of Sphere-Pac Fuel 813
results. Only the neck growth initiated by irradia- 3.25.5.2.4.1.1 Segment I11L The fuel centerline
tion creep was underestimated. temperature is following the power history in shape
The RESTART-options in SPHERE-3 worked and reaches the maximum of about 2550 C for
without any problems, and were very useful the maximum power of 540 W m1, with 427 C
for modeling the present fuel pin reinstrumenta- on the cladding outer surface by the end of irradia-
tion case. tion (EOI) after 46 h. The fractional FGR that starts
at a fuel centerline temperature of about 1500 C is
slightly decreasing during the power and tempera-
3.25.5.2.4 FUJI
ture dip lasting for 10 h and finally reaches the rela-
In the FUJI experiment, which was a collaboration
tively high value of 40%. However, the absolute
between JNC (Japan Nuclear Cycle Development
values are negligible after such a short irradiation.
Institute) (today JAEA (Japan Atomic Energy Agency))
The restructuring takes place rather quickly at this
in Japan, PSI in Switzerland, and Nuclear Research
high fuel temperature and after 36 h at a fuel center-
and Consultancy Group (NRG) in the Netherlands,
line temperature of 2200–2300 C, a compact sintered
four kinds of fuel, that is, MOX Sphere-pac, MOX
fuel body (necking ratio 45%) is created and increased
Vipac, MOX pellet, and Np-MOX Sphere-pac fuel
toward 60% of the fuel radius at EOI. On the other
were fabricated at PSI and then simultaneously irra-
hand, the time is too short and the temperature not high
diated under identical conditions in the high flux
enough to establish a notable void (hole) in the fuel
reactor (HFR) in Petten. One of the main aspects was
center; just a first start was predicted. The predicted
the irradiation behavior during the start-up phase.
results are seen in Figure 10. Figure 11 shows the
Special emphasis was given to this phase, because
detailed ceramography of the same Sphere-pac rod.
during the initial ramp, the contact areas between the
spheres were small, and the external mechanical load 3.25.5.2.4.1.2 Some comments to the SPHERE
was amplified in the contacts. Thus, potentially, the predictions of Rod I11L47 The detailed ceramo-
bed load has a strong effect on sintering during the graphy for segment I11L can be seen in Figure 11.
initial phase. The MOX always contained 20% Pu and The outcome of the different sintering mechanisms on
for the Np-MOX additionally 5% Np. the fuel, as calculated with SPHERE, is plotted in the
First results show that the particle fuel undergoes figure to enable a direct comparison. Furthermore, the
a substantial structure change already at the very calculated maximum for the radial fuel temperature
beginning of the irradiation when the maximum distribution, in this case at EOI, is also shown. Here, a
power is reached. The changes in microstructure, comparative evaluation can be done only with some
that is, the formation of a central void and the densi- uncertainties and the result commented as follows:
fication of fuel, decrease the fuel central temperature.
The calculated central void, 1.4% of fuel radius, is
Thus, the fast and strong restructuring helps to
prevent central fuel melting at high power levels. comparable with PIE in only one direction (to the
For further experimental details of this study, see left of the center position in the figure) but under-
Chapter 3.11, Sphere-Pac and VIPAC Fuel. predicted in all other directions; an average value
The experiments are combined with fuel perfor- according to the ceramography picture seems to be
mance calculations performed with codes specially around 6%.
The calculated compact sintered body (58.6%) is
developed to predict particle fuel behavior, one of
them being the PSI code SPHERE.49 somewhat overpredicted and a value around 50%
would have better fitted the PIE.
The region with columnar grains, that is, changed
3.25.5.2.4.1 Modeling results from the as-fabricated equiaxed grains, is well pre-
Modeling calculations with SPHERE were made for dicted, or slightly overpredicted.
the three Sphere-pac segments I11L, R11L, and The neck ratio for the fine sphere fraction is com-
R22U (see Table 8 in Chapter 3.11, Sphere-Pac parable with the PIE, while it is overpredicted for
and VIPAC Fuel). From each segment, one sample the coarse fraction.
was cut out for the PIE. The calculations were made
on the identical axial positions in order to get as 3.25.5.2.4.1.3 Conclusions The test results show
relevant a base for the comparison as possible. The that significant restructuring takes place in particle
result for Segment I11L is presented below; for the fuel during the first days of irradiation, if the power
other segments see Nordström.49 and temperature are high enough. The thermally
814 Modeling of Sphere-Pac Fuel
2800 70
2600
35 Neck ratio
Neck ratio (%)
Coarse
30
1500 Fine
25
20
1000
15
500 10
5
0 0
0 0.5 1 1.5 2 2.5 3 3.5
Calculated structure Radius (mm)
Figure 11 FUJI segment I11L; fuel-restructuring effects calculated with SPHERE, compared to a detailed ceramography
at the same axial position. Reproduced from Nordström, L. Å.; Hellwig, Ch.; Nakamura, M.; Bakker, K. PIE and modelling
results for the FUJI irradiation test of Sphere-pac MOX fuel. In EHPG-Meeting, OECD Halden Proceedings HPR-364,
Lillehammer, Norway, Oct 2005, with permission from OECD Halden Reactor Project.
driven processes (sintering and pore migration) the densification of sintered fuel, and the forma-
depend mainly on the achieved fuel temperatures tion of a central void increase the fuel thermal con-
and can take place within a few hours. The sintering ductivity significantly. As a result, the reduction in
mechanisms (i.e., necking between adjacent spheres), fuel centerline temperature for the very high
Modeling of Sphere-Pac Fuel 815
temperature levels reached in the FUJI experiment pin), developed by JAEA, was revised in this study to
can be as much as 250 C (segment R11L) already develop a Vipac MOX fuel performance analysis code.
after 90 or 135 h of irradiation. This is very advanta-
geous, as it considerably reduces the risk of fuel 3.25.6.1.1 For the nonsintered region
melting at high power levels. For the thermal conductivity of the nonsintered
The evaluation indicates that the thermal driven region in Vipac fuel, a model51 that was on the basis
sintering processes are more or less finished after less of the Hall and Martin model16 was applied and the
than 100 h, at that high temperature level, and will not effect of necking between fuel granules was taken into
continue in radial direction much further outward in account. The model contained two simplified geo-
the fuel because the temperature close to the cladding metrical arrangements. One is a cubic packed sphere
is too low and setting a limit for the thermal sintering array having a porosity of 0.476 (1p/6), which con-
effects. This state was not reached in segment I11L. sisted of an infinite layer of regularly piled up spheres
Modeling of the irradiation behavior of the FUJI parallel to an assumed heat flow. The other is a square
Sphere-pac segments with the fuel performance code packed infinite cylinder array which consists of an
SPHERE-3, developed at PSI, was very successful. The infinite layer of regularly piling up cylinders parallel
comparison with PIE results showed that it was possi- to the heat flow. The porosity of the arrangement is
ble to model all mechanisms of fuel restructuring for geometrically evaluated as 0.215 (1p/4). The Hall
MOX fuel with high contents of Pu (20%), in reason- and Martin model estimates the thermal conductivity
ably good agreement with the experimental results. of packed granules with several size classes on the
Including the irradiation experiments reported on basis of a combined cubic packed sphere array and a
earlier, SPHERE has so far been validated against square packed infinite cylinder array.
base irradiation of UO2 fuel up to a burn up of The thermal conductivity of the cubic packed
about 50 MWd kgU1, and power transients up to a arrangement is (see Ishii et al.51)
power of 53 kW m1, and now with FUJI, the start-up 2 3
dþm
irradiation of MOX fresh fuel go to extremely high p d 1 ð1 þ mÞln
6 1þm 4 p7
temperatures. The predictions of the very different kc ¼ kg 6
4 þ 7 ½45
thermal driven aspects have all been in good agree- 2ðd 1Þ 2
4 5
ment with experimental results.
and that of the square packed infinite cylinder array
is (see Ishii et al.51)
3.25.6 Modeling of Vipac Fuel sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
2 3
dþm
For the Vipac modeling, only the thermal conductiv- 6 p 2 arcsin 7
61 þ m 2ð1 þ mÞ p 7
ity is treated. It creates however a basis for other 6
ksq ¼ kg 6 pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 7 ½46
properties, treated earlier in detail for the Sphere- 41 d ðd þ mÞð2 d þ mÞ 2ð1 dÞ7 5
pac fuel, such as sintering behavior and FGR. This
chapter also covers a highly Vipac specific concept, where kg, ks are thermal conductivity of the gas and solid
the addition of a metallic getter, which importantly phases, d ¼ kg/ks, g is temperature jump distance cal-
influences the thermal conductivity. culated by52 C1 kg T C2 Pg1 , where D ¼ average diame-
Pp ffiffiffiffiffi P p3 ffiffiffiffi
oi
ter of granules calculated by 3
oi = di where
,
3.25.6.1 Basic Thermal Conductivity i i
oi is the weight and di the mass fraction of the ith
The thermal conductivity is difficult to be modeled, component, Pg is the gas pressure, T the temperature
as the particle shape is not known (being random). (in K), and C1 and C2 are adjustable parameters, and
Direct geometrical approaches as conducted in m ¼ g/D.
SPACON are therefore not possible. JAEA is having For a material of porosity x, a logarithmic interpola-
some interest in Vipac fuel and is also conducting tion will lead to the following equation:
programs together with Research Institute of Atomic 8 8 99
Reactors (RIAR) (see, e.g., Chapter 3.11, Sphere- < X < Y x xj = =
k ¼ exp lnðki Þ ½47
Pac and VIPAC Fuel) and PSI (see Chapter 3.11, : : x xj ;;
i¼s; c; sq; g j ¼s; c; sq; g; j 6¼i i
Sphere-Pac and VIPAC Fuel). The pellet-type
MOX fuel performance code ‘CEDAR’50 (Code for where s, c, sq, g are indices for solid, cubic packed,
Thermal and Deformation Analysis of Reactor fuel square packed infinity cylinders, and gas, k, ki are
816 Modeling of Sphere-Pac Fuel
overall and fractional thermal conductivity for i¼s, c, type. The small touching point between the particles,
sq, g, and x, xi are overall and fractional densities for the important restructuring behavior, plus the high
i¼s, c, sq, g. specific surface result in the need to develop model-
An example of experimental evaluation of this can ing tools for understanding the physical behavior of
be found in Ishii et al.51 The factors C1 and C2 are these advanced nuclear fuel types.
derived numerically. For the Sperhe-pac fuel, modeling tools have been
developed at PSI, the two most important ones being
3.25.6.1.2 Introduction of metallic particles SPACON and SPHERE (see Section 3.25.5.1). These
(getter concept) modeling tools have extensively been validated and
The getter concept, which is an introduction of metal- have proven their capability; Section 3.25.5.2 shows
lic (usually U) particles, is performed in order to four examples where modeling is compared with
decrease the oxygen potential in the fuel. This helps experimental results (mostly described in Chapter
to lower the chemical interaction with (corrosion of ) 3.11, Sphere-Pac and VIPAC Fuel). The most
the cladding. The thermal conductivity of these metal- important aspects here are the initial thermal conduc-
lic particles is much higher than that of the ceramic tivity, which is a weak point of this fuel concept, the
phase; therefore, the introduction of the getter particles restructuring and the consequent evolution of the
increases the thermal conductivity. The introduction of thermal conductivity, plus the FGR.
the particles in Ishii et al.51 is modeled by layers being For Vipac fuel a modeling approach for the thermal
arranged parallel or perpendicular to the thermal flow conductivity is shown in Section 3.25.6. The main
direction. This leads to the following two equations: difference here is the randomness of particle shapes
1 and sizes compared to the spheres arranged in assumed
kjj ¼ ð1 V Þkf þ Vkg ; k? ¼ ½48 packing order. Additionally, the getter concept, where
ð1 V Þ=kf þ V =kg
metallic uranium particles are included in the pack for
representing the most extreme conditions; a superpo- regulating the oxygen potential, is considered for its
sition leads to influence on the thermal conductivity.
1 Particle fuel is still an innovative concept. How-
kmix ¼ ½49
ð1 C3 Þ=kjj þ C3 =k? ever, this chapter shows that modeling tools exist to
where k||,? is conductivity in parallel and perpendicu- predict its behavior under various conditions. These
lar direction, kmix conductivity of mixture between models represent a solid basis to plan future irradia-
parallel and perpendicular mode, V volume fraction tion experiments and to develop a basic understand-
of the getter, kf, g conductivity of fuel and getter, and C3 ing of the fuel behavior.
is fraction of parallel conduction mode.
Ishii et al.51 shows examples of measured thermal
Acknowledgments
conductivity data compared with the behavior mod-
elled using above equation.
The compilation of this overview article was partially
In the case of a high temperature condition
financed by the European Commission through the
beyond 1408 K, U particles contained in the UO2
FP7 F-BRIDGE project (Contract No. 211690). The
compact should melt. Once melting occurs, the ura-
IAEA and the OECD Halden Reactor project are
nium can move easily through the spaces between the
thanked for their courtesy to reproduce figures
UO2 granules and provide favorable heat conduction
from their reports.
channels in the compact such as bridged bondings, in
other words, quasinecks between UO2 granules.
In Ishii et al.51 these high temperature effects for References
the metallic inclusions are derived, on the basis of the
necking between the particles. Basically, this results 1. McGeary, M. K. J. Am. Ceram. Soc. 1961, 44, 513, DOI:
10.1111/j.1151–2916.1961.tb13716.x.
in extensions of eqns [45] and [46]. 2. Hellwig, Ch.; Heimgartner, P.; Tomita, Y.; Kono, S.
Sphere-pac Filling for Irradiation Tests of the FUJI Project.
In Proceedings of the Global 2003, New Orleans, LA, 2003;
3.25.7 Conclusions 1348–1353.
3. Shigetome, Y.; Kono, S.; Hellwig, Ch.; Heimgartner, P.
Vipac fuel fabrication for irradiation tests of the FUJI
Both particle fuel concepts imply a very different project. In Proceedings of the Global 2003, New Orleans,
physical behavior compared to the solid pellet fuel LA, Nov 16–20, 2003.
Modeling of Sphere-Pac Fuel 817
4. Bart, G.; Botta, F.; Hoth, C.; Ledergerber, G.; Mason, R.; 31. Nordström, L. Å. SPHERE-1: Physical Models applied in
Stratton, R. J Nucl. Mater. 2008, 376(1), 47–59. the Code, PSI TM-43–97–12; Paul Scherrer Institut:
5. Hellwig, Ch.; Bakker, K.; Ozawa, T.; et al. Nucl. Sci. Eng. Würenlingen (CH), Mar 1997.
2006, 153(3), 233–244, URL http://www.new.ans.org/ 32. Wallin, H. SPHERE-3: A New Code Version. New Features
pubs/journals/nse/a_2609. Explained, PSI TM-43–99–39, Sept 1999.
6. Botta, F.; Hellwig, Ch. Nucl. Sci. Eng. 2000, 135, 1–12. 33. Speight, M. V. Nucl. Sci. Eng. 1969, 37, 180.
7. Wallin, H.; Nordström, L. Å.; Hellwig, Ch. The Sphere-pac 34. Turnbull, J. A.; Friskney, C. A.; Findlay, J. R.; Johnson, F. A.;
Fuel Code SPHERE-3. In Proceedings of IAEA TCM on Walter, A. J. J. Nucl. Mater. 1982, 107(2–3), 168–184,
Nuclear Fuel Behaviour Modelling at High Burnup, DOI: 10.1016/0022–3115(82)90419–6.
Windermere, UK, June, 2000; CD-ROM. 35. Turnbull, J. A. An assessment of fission gas release and
8. Van Der Linden, A. Cer–Met Sphere-pac Fuel Potential, the effect of microstructure at high burn-up, OECD Halden
Netherlands Energy Research Foundation, Petten, Reactor Project HWR-604, Technical report; Institutt for
ECN-RX-90–031, June 1990. Energiteknikk, Halden, Norway, Mar 1999.
9. Pouchon, M. A.; Nakamura, M.; Hellwig, Ch. J. Nucl. Sci. 36. White, R. J.; Tucker, M. O. J. Nucl. Mater. 1983, 118, 1.
Technol. 2002, (Supp 3), 830–833. 37. Wood, M.; Matthews, J. J. Nucl. Mater. 1980, 91(1), 35–40.
10. Pouchon, M. A.; Nakamura, M.; Hellwig, Ch.; Ingold, F.; http://dx.doi.org/10.1016/0022–3115(80)90029-X.
Degueldre, C. J. Nucl. Mater. 2003, 319, 37–43. 38. Matthews, J.; Wood, M. J. Nucl. Mater. 1980, 91(2–3),
11. Harding, J. H.; Martin, D. G.; Potter, P. E. Thermophysical 241–256, http://dx.doi.org/10.1016/0022–3115(80)90224-X.
and Thermochemical Properties of Fast Reactor 39. Baker, C. J. Nucl. Mater. 1977, 66, 283.
Materials, EUR 12402 EN; Commission of the European 40. Turnbull, J. A. Personal communication.
Communities, 1989. 41. Kogai, T. J. Nucl. Mater. 1997, 244(2), 131–140,
12. Bakker, K.; Kwast, H.; Cordfunke, E. H. P. J. Nucl. Mater. http://dx.doi.org/10.1016/S0022–3115(96)00731-3.
1995, 226, 128, DOI: 10.1016/0022–3115(95)00087–9. 42. Lassmann, K. J. Nucl. Mater. 1992, 188, 295–302.
13. Schotte, W. AIChE J. 1960, 6, 63, DOI: 10.1002/ 43. Wallin, H.; Nordström, L. Å.; Hellwig, Ch. Fission Gas Model
aic.690060113. of the Fuel Code SPHERE-3. In Proceedings of IAEA/NEA/
14. Zehner, P.; Schlünder, E. U. Chemie Ingenieur Technik CEA International Seminar on Fission Gas behaviour in Water
1970, 42, 933, DOI: 10.1002/cite.330421408. Reactor Fuels, Cadarache (F), Sept 2000.
15. Zehner, P.; Schlünder, E. U. Chemie Ingenieur Technik 44. Wallin, H. SPHERE-3: User’s Manual, PSI TM-43–99–48;
1972, 44, 1303, DOI: 10.1002/cite.330442305. Dec 1999.
16. Hall, R. O. A.; Martin, D. G. J. Nucl. Mater. 1981, 101, 172, 45. Nordström, L. Å.; Wallin, H.; Hellwig, Ch. A Comparative
DOI: 10.1016/0022–3115(81)90455–4. Irradiation Test of Pellet and Sphere-pac Fuel (IFA-550.9).
17. Dalle Donne, M.; Sordon, G. Fusion Technol. 1990, In EHPG-Meeting, OECD Halden Proceedings HPR-351,
17, 597. Loen, Norway, May 1999.
18. Okazaki, M.; Yamasaki, T.; Gotoh, S.; Toei, R. J. Chem. 46. Nordström, L. Å. Comparative Irradiation and Transient
Eng. Jpn. 1981, 14, 183. Test of Pellet and Sphere-pac Fuel (IFA-550.11).
19. Masamune, S.; Smith, J. M. I&EC Fundam. 1963, In EHPG-Meeting, OECD Halden Proceedings HPR-359,
2, 136. Storefjell, Norway, Sept 2002.
20. Wallin, H. SPHERE-1: Review of sintering models, 47. Nordström, L. Å.; Nakamura, M. PIE and Modelling
PSI TM-43–97–20, Paul Scherrer Institut: Würenlingen Results for the Irradiation and Transient Test of Sphere-
(CH), Nov 1997. pac UO2 Fuel (IFA-550.11). EHPG-Meeting, OECD Halden
21. Matthews, J. R. J. Nucl. Mater. 1979, 87, 356–366. Proceedings HPR-362, Sandefjord, Norway, May 2004.
22. Barsoum, M. Fundamentals of Ceramics, McGraw-Hill: 48. Nordström, L. Å.; Botta, F. Comparative FGR Ramp Tests
New York, 1997. with Pellet and Sphere-pac Fuel, IFA-550.9.
23. MATPRO-version 11 (revision 2), A Handbook of Materials In EHPG-Meeting, OECD Halden Proceedings
Properties for Use in the Analysis of Light Water Reactor HPR-347, Loen, Norway, May 1996.
Fuel Rod Behavior, NUREG/CR-0479; 1981. 49. Nordström, L. Å.; Hellwig, Ch.; Nakamura, M.; Bakker, K.
24. Hall, R. O. A.; Mortimer, H. J.; Mortimer, D. A. J. Nucl. PIE and modelling results for the FUJI irradiation test of
Mater. 1987, 148, 237–256. Sphere-pac MOX fuel. In EHPG-Meeting, OECD
25. Tokar, M. J. Am. Ceram. Soc. 1973, 56, 173. Halden Proceedings HPR-364, Lillehammer, Norway,
26. Blank, H. In Materials Science and Technology; VCH: Oct 2005.
New York, 1994; Vol. 10 A. 50. Mizuno, T.; Nagai, S.; Itaki, T.; Nagae, N.; Tanaka, K.;
27. Matzke Hj, B. Science of Advanced LMFBR-Fuels; Komatsu, J. Fuel pin performance and reliability analysis
North Holland Physics Publishing: Amsterdam, 1986; code in PNC. In: Proceedings of the International
ISBN 0–444–86997–2. Conference on Reliable Fuels for Liquid Metal Reactors,
28. Hall, R. A. Report AERE-R-7326, Harwell, UKAEA. Tucson, Arizona; 1986; pp 5–28.
29. Wallin, H. Irradiation of Mixed Carbide Fuel (AC-3 Test): 51. Ishii, T.; Yuda, R.; Hirai, M.; Tsuboi, Y.; Ukai, S. J. Nucl.
Modelling of Pin 3L115 with the Fuel Code SPHERE-2, Sci. Technol. 2004, 41(12), 1bv204–1210. http://dx.doi.
PSI TM-43–99–10; Paul Scherrer Institut: Würenlingen org/10.3327/jnst.41.1204.
(CH), Feb 1999. 52. Lanning, D.; Hann, C. R. Review of Methods Applicable to
30. Olander, D. R. Fundamental Aspects of Nuclear Reactor the Calculation of Gap Conductance in Zircaloy-clad UO2
Fuel Elements; Energy Research and Development Fuel Rods, BNWL-1894/UC-78b, 1975.
Administration (ERDA), ISBN 0–87079–031–5, 1976; Vol. 1
Springfield, VA.