Vous êtes sur la page 1sur 4

Scripta Materialia 113 (2016) 2730

Contents lists available at ScienceDirect

Scripta Materialia
journal homepage: www.elsevier.com/locate/smm

Features of cubic and tetragonal structures of UMo alloys:


Atomistic simulation
S.V. Starikov , L.N. Kolotova
Joint Institute for High Temperatures, Russian Academy of Sciences, Moscow 125412, Russia
Moscow Institute of Physics and Technology (State University), Dolgoprudny 141700, Russia

a r t i c l e

i n f o

Article history:
Received 15 August 2015
Received in revised form 2 October 2015
Accepted 6 October 2015
Available online 21 October 2015
Keywords:
Uranium
Alloy
Atomistic simulation
Phase transition

a b s t r a c t
Using molecular dynamics simulations, we study the structural properties of body-centered cubic (BCC) and
body-centered tetragonal (BCT) phases of UMo alloys. The local positions of uranium atoms in the BCC phase
correspond to the BCT structure. Thus, the BCC lattice exhibits cubic symmetry only on the scale of several
interatomic spacings, and it is therefore more correct to denote the high-temperature state of UMo alloys as
quasi-BCC. This structural feature occurs for pure uranium as well. This fact is the possible origin of the difculties
encountered in the description of the BCC phase of pure uranium by ab initio methods.
2015 Elsevier Ltd. All rights reserved.

Uranium is the main component in nuclear fuel. Pure uranium


is found in three crystal phases: a low-temperature orthorhombic
-phase, a high-temperature tetragonal -phase and a body-centered
cubic (BCC) -phase, which exists at higher temperatures [1,2]. The
BCC -phase exhibits the most useful properties for nuclear engineering.
Unfortunately, this phase is unstable at room temperature. It has been
shown that alloying U with Mo stabilizes BCC-like structures at low
temperatures [36], but the details of such procedures are still under
investigation [7,8].
A description of BCC -U by ab initio methods is another open
question in studies of U materials [911]. Ab initio calculations have
shown that this phase is highly mechanically unstable. This structural
instability can be found by calculation of the elastic constants [10,12]
and simulation of point defect behavior [13,14]. The reason for the
instability is currently unclear. It is important to note that while selfconsistent ab initio lattice dynamics enable investigations of the elastic
properties of the BCC-phase at high temperatures [11], lattice defect
stability still remains an open question.
Another important nding of ab initio calculations of uranium is the
energetic hierarchy of the structures. The calculations predict the
existence of a body-centered tetragonal (BCT) structure that is more
stable than the BCC lattice [15,2,16]. However, experiments on pure
uranium have not observed the predicted BCT phase. Conversely, the
BCT phase is observed in the experiments on the UMo alloy [7].

Corresponding author at: Joint Institute for High Temperatures, Russian Academy of
Sciences, Moscow 125412, Russia.
E-mail address: starikov@ihed.ras.ru (S.V. Starikov).

http://dx.doi.org/10.1016/j.scriptamat.2015.10.012
1359-6462/ 2015 Elsevier Ltd. All rights reserved.

Molecular dynamics simulation is a powerful tool for the study of


physical properties of matter at the atomistic level [1721]. In this
work, we study the structure of UMo alloys using molecular dynamics
simulations with a novel interatomic potential [22,23]. This potential
contains an angular-dependent term and was parameterized on the
basis of ab initio calculations (in the framework of density functional
theory) using the force-matching method [24,25]. It should be noted
that the potential reproduces the properties of the -phase rather
well and helps to explain the anomalously fast self-diffusion
observed in pure uranium at high temperatures [22]. This nding
demonstrates that pure uranium is accurately described by ab initio
calculations. Clearly, the problem of the instability of the uranium
-phase in ab initio simulations is not related to the insufcient
precision of the theoretical model.
All calculations in this work were carried out using the LAMMPS code
[26]. The AtomEye code is used for visualization of atomic dynamics
[27]. The simulation cell has the following size: 500 500 500,
where 0 is the lattice parameter of BCC (or BCT) structure. We use
the Langevin thermostat and NoseHoover barostat to control the
temperature T and pressure P, respectively. Nevertheless, a large fraction
of the calculations were performed in the NVE-ensemble in order to
obtain the statistics in the equilibrium state.
We rst obtained the atomistic model for the BCC structure of UMo
alloy. Here, two different approaches were used: (1) cooling from the
liquid state with crystallization and (2) random distribution of the
atoms (U and Mo) in the BCC lattice. Both methods were accompanied
by a variation of the simulation cell size in order to achieve zero
pressure in all directions (Pxx = Pyy = Pzz = 0). Nevertheless, we
found that a martensitic phase transformation occurs at low T and a

28

S.V. Starikov, L.N. Kolotova / Scripta Materialia 113 (2016) 2730

low Mo concentration. This means that the initial BCC structure transforms into the BCT structure (this is true for both methods of the
model). After obtaining this BCC BCT transition, we have studied
the BCT lattice properties and constructed the stable BCT-phase in the
simulation cell.
We use radial distribution function G(r) for examination of the
structural details. Fig. 1 shows the obtained G(r) for four cases: two
UU distributions for U7.5 at.%Mo at T = 200 K (BCT-phase) and
T = 700 K (BCC-phase) and UU and MoMo distributions for the
BCC-phase of U40 at.%Mo at T = 200 K. Analysis of G(r) together
with the visualization of atom positions enables the exploration of the
alloy structures. The lattice parameters of the BCT phase fulll the
following condition: a = b N c. We found that the central atom in the
basic BCT cell is displaced. The displacement is along the c-axis and
has a magnitude of approximately 0.55 from the center of the basic
cell. In this case, the local U atom environment (in the region of two
coordination spheres) is similar to the -phase of pure uranium.
Such BCT structures can be obtained from the BCC state with a minimal
atomic reorganization and shows a lower energy than the BCC phase.
One of the key discoveries made in present work is that the local
positions of uranium atoms in the BCC phase show that the central
atom in the basic cell is displaced similar to that found in the BCT structures (i.e. small local anisotropy of the basic cell of atoms structure takes
place). In other words, in all studied cases, the uranium subsystem had a
tendency to form an -phase-like local atomic arrangement. In contrast
to the BCT phase, the BCC phase exhibits an absence of long correlations
in the anisotropy direction (i.e., the direction of central atom
displacement in the basic cell). The high-temperature structure
(or the structure with a high Mo concentration) is similar to that of
the BCC phase only on a scale of several interatomic spacings. Therefore,
it is more correct to denote the high-temperature state of the UMo
alloy as quasi-BCC (q-BCC). The described structural feature also occurs
for the pure -phase of uranium. Such peculiar local atomic ordering
may be the origin of the difculties encountered in the description of
-U by ab initio methods [9,10,12].
The dependence of the averaged BCT phase lattice parameters on the
Mo concentration is shown in Fig. 2. This gure depicts the simulation
results together with the data obtained from various experiments
[7,28]. The data are given for room temperature. The calculated
lattice parameters correspond to the average of all simulation cells
irrespective of local atomic arrangement. For high Mo concentrations,
the a = b = c condition is correct on average, but local anisotropy
for the uranium subsystem is also present. The anisotropy is produced
primarily by the displacement of the central atom in the basic cell.
The magnitude of this displacement is mostly independent of conditions such as temperature and composition. Examination of Fig. 2
shows that the simulation results are in qualitative agreement with

Fig. 2. The lattice parameters of BCT phase at room temperature (for various Mo
concentrations): 1 and 2 values measured in work [7]; 3 and 4 values of lattice
parameter measured at high concentrations of Mo [7,28]; 5 and 6 calculated data
from this work. The solid and empty symbols correspond to c and a, respectively.

the experimental data. For instance, our simulations resulted in a c/a


ratio of approximately 0.93 for U11 at.%Mo, while experimentally,
the c/a is approximately 0.97 for this alloy [7]. Conversely, density
functional theory yielded a c/a of approximately 0.82 for pure uranium
[2,29]. The difference between the calculations and the experimental
data may be because the BCT crystal has a complicated structure
consisting of domains. Each domain has its own anisotropy direction.
The size of the domains ranges between 10 and 50 nm, i.e., the domain
size is comparable to the coherence radius in X-ray diffraction. In this
case, some spatial averaging for a and c should occur in the experiments.
The hypothesized existence of such domains is based on the formation of domains observed in our simulation for UxMo alloys, for
x 2022 at.%. However, the size of the observed domains in the
simulations was approximately 15 nm.
Both studied structures exhibit areas of stability in the temperature
concentration phase diagram (Tx diagram). Thus, temperature
changes result in the phase transition found by simulations in the
NPT-ensemble. During the simulation, the following requirement
was enforced: Pxx = Pyy = Pzz = 0. The calculated line of the phase
transition between BCC and BCT structures is shown in Fig. 3. It
should be noted that the transition is accompanied by a change in
volume. As discussed above, we refer to the high-temperature
phase in Fig. 3 as q-BCC.

Fig. 1. Calculated radial distribution functions for U7.5 at.%Mo (left gure) and U40 at.%Mo (right gure) alloys. Two dependences of GUU (r) at T = 200 K (BCT phase) and
T = 700 K (q-BCC phase) are shown for U7.5 at.%Mo. G UU (r) and G MoMo (r) are shown for U40 at.%Mo at T = 200 K. Vertical black lines mark the maximum of
G(r) corresponding to the perfect BCC lattice.

S.V. Starikov, L.N. Kolotova / Scripta Materialia 113 (2016) 2730

Fig. 3. Calculated line on Tx diagram for BCT q-BCC transition.

Mechanical hysteresis of the BCT phase is another peculiarity of


UMo alloy observed in the simulations. A summary of the relevant
data is presented in Fig. 4. Such hysteresis is most pronounced for
UxMo alloys with x 2022 at.%. For these alloys, the c/a ratio is
approximately 0.98. This difference between the lattice parameters
leads to the formation of small domains with various anisotropy directions. It is possible that such domains can be formed in
alloys with lower concentrations of molybdenum, but their size is
larger than the size of the simulation cell. We then applied a uniaxial
deformation to the alloy model. The deformation leads to atomic
reorganization until the domains disappear and the crystal assumes

29

a uniform BCT lattice with orientations along the deformation direction


(in several cases, domains with antiparallel orientation remain).
However, the reverse deformation does not result in a return to the initial state. The pressure corresponding to the nal structure is different
from the initial value. Domains are not formed for UxMo alloys
where x is approximately 1819 at.%, but mechanical hysteresis occurs.
The change in the anisotropy direction for the entire crystal caused by
the applied uniaxial deformation is the origin of such behavior.
An analogy exists between the phenomena studied in this work
and the ferromagnetic-paramagnetic transition. In our case, the local
anisotropy direction is similar to the magnetic moment of an atom
(or to electron spin in highly accurate theories). At a low level of disorder (corresponding to low values of temperature or Mo concentration),
the system exhibits long-range correlations in the anisotropy direction. Increased disorder leads to the formation of domains. At a high
level of disorder, long-range correlation in the anisotropy direction
is disturbed. Additionally, external action aimed at changing the
anisotropy direction (such as a uniaxial deformation) leads to
hysteresis. As the rst approximation, the temperature T tr of
ferromagneticparamagnetic transition may be calculated as /kB,
where is the energy needed for reorientation of single magnetic
moment and kB is the Boltzmann constant. In our case, reorientation
energy of single basic cell corresponds to a compression along a-axis
and an extension along c-axis. This formalism allows to obtain the
formula for the temperature of BCT q-BCC transition:
T tr

1
C 11 C 22 2C 12 ca2 c a;
4kB

where C 11 , C 22 and C12 are the elastic constants for the crystal
orientation like the one shown in Fig. 4c (i.e. c-axis is directed
along the x-direction). The calculation performed for a small system
without domains gives C 11 = 205 GPa, C 22 = 170 GPa and C 12 =
115 GPa. For the given elastic constants formula (1) describes the
calculated data (Figs. 2 and 3) with well precision: the deviation
from directly calculated Ttr is about 30%.
There is a considerable variation in the published data for the elastic
properties of UMo alloys [30,31]. In the present work we have simulated
elastic properties of U22 at.%Mo alloy. The effective elastic modulus
related to small deformation (y b 0.01) is 100 10 GPa, and macroscopic
isotropy takes place. This calculated data agree well with the recent
experiments [31]. However, the structure has isotropy only on scales
more than several nanometers, and the value of elastic modulus is caused
by domains reorientation. This question requires further investigation
and will be considered in future works.
The authors acknowledge A.Yu. Kuksin and D.E. Smirnova for their
suggestions and discussions, and Prof. G. Norman for the inspiration
and support. The calculations were carried out on the computer clusters
MVS-100K of the Joint Supercomputer Center of RAS and K-100
of Keldysh Institute of Applied Mathematics of RAS. The work was
supported by the Program for Basic Research of the Presidium of
the RAS No 2 (coordinator is G. I. Kanel) and the President RF Grant
MK-7688.2015.8.
References

Fig. 4. Calculation results for uniaxial deformation of U22 at.%Mo alloy at the room temperature. Compression takes place along y-direction. (a) Fragment of the calculation cell in
the initial state (two atomic planes are shown): black atoms U, white atoms Mo; blue
squares indicate atoms in one atomic plane; arrows indicate local anisotropy direction.
(b) State after deformation y = 0.04. (c) The dependencies of pressure components on
deformation: solid lines initial compression; dashed lines inverse deformation
to initial sizes of calculation cell. Atomic basic cell of BCT lattice is shown in sub-gure
(c), it corresponds to the solid blue square indicated in sub-gure (a).

[1]
[2]
[3]
[4]
[5]
[6]
[7]
[8]
[9]
[10]
[11]

C.-S. Yoo, H. Cynn, P. Sderlind, Phys. Rev. B 57 (1998) 1035910362.


R.Q. Hood, L.H. Yang, J.A. Moriarty, Phys. Rev. B 78 (2008) 024116.
A. Landa, P. Sderlind, P.E.A. Turchi, J. Nucl. Mater. 414 (2011) 132137.
Y.S. Kim, G.L. Hofman, J. Nucl. Mater. 419 (2011) 291301.
A. Landa, P. Sderlind, P.E.A. Turchi, J. Nucl. Mater. 434 (2013) 3137.
Y.S. Kim, G.L. Hofman, A.M. Yacout, T.K. Kim, J. Nucl. Mater. 441 (2013) 520524.
I. Tkach, N.-T. Kim-Ngan, S. Makov, M. Dzevenko, L. Havela, A. Warren, C. Stitt,
T. Scott, J. Alloys Compd. 534 (2012) 101109.
T.A. Pedrosa, A.M.M. dos Santos, F.S. Lameiras, P.R. Cetlin, W.B. Ferraz, J. Nucl. Mater.
457 (2015) 100117.
P. Sderlind, Phys. Rev. B 66 (2002) 085113.
C.D. Taylor, Phys. Rev. B 77 (2008) 094119.
P. Sderlind, B. Grabowski, L. Yang, A. Landa, T. Bjrkman, P. Souvatzis, O. Eriksson,
Phys. Rev. B 85 (2012) 060301.

30

S.V. Starikov, L.N. Kolotova / Scripta Materialia 113 (2016) 2730

[12] B. Beeler, C. Deo, M. Baskes, M. Okuniewski, J. Nucl. Mater. 433 (2013) 143151.
[13] B. Beeler, B. Good, S. Rashkeev, C. Deo, M. Baskes, M. Okuniewski, J. Phys. Condens.
Matter 22 (2010) 505703.
[14] G.-Y. Huang, B.D. Wirth, J. Phys. Condens. Matter 24 (2012) 415404.
[15] M. Freyss, T. Petit, J.-P. Crocombette, J. Nucl. Mater. 347 (2005) 4451.
[16] D.E. Smirnova, S.V. Starikov, V.V. Stegailov, J. Phys. Condens. Matter 24
(2012) 015702.
[17] S. Starikov, Z. Insepov, J. Rest, A. Kuksin, G. Norman, V. Stegailov, A. Yanilkin, Phys.
Rev. B 84 (2011) 104109.
[18] D.E. Smirnova, A.Y. Kuksin, S.V. Starikov, V.V. Stegailov, Z. Insepov, J. Rest, A.M.
Yacout, Model. Simul. Mater. Sci. Eng. 21 (2013) 035011.
[19] X.-F. Tian, H.-X. Xiao, R. Tang, C.-H. Lu, Nucl. Inst. Methods Phys. Res. B 321
(2014) 2429.
[20] H.-X. Xiao, R. Tang, X.-F. Tian, C.-S. Long, Chin. Phys. Lett. 31 (2014) 047101.
[21] H. Xiao, C. Long, X. Tian, S. Li, Mater. Des. 74 (2015) 5560.

[22] D. Smirnova, A. Kuksin, S. Starikov, J. Nucl. Mater. 458 (2015) 304311.


[23] D.E. Smirnova, A.Y. Kuksin, S.V. Starikov, V.V. Stegailov, Phys. Met. Metallogr. 116
(2015) 445455.
[24] F. Ercolessi, J.B. Adams, Europhys. Lett. 26 (1994) 583.
[25] P. Brommer, F. Gahler, Model. Simul. Mater. Sci. Eng. 15 (2007) 295.
[26] S.J. Plimpton, J. Comp. Physiol. 117 (1995) 119.
[27] J. Li, Model. Simul. Mater. Sci. Eng. 11 (2003) 173177.
[28] V. Sinha, P. Hegde, G. Prasad, G. Dey, H. Kamath, J. Alloys Compd. 506 (2010)
253262.
[29] P. Sderlind, Adv. Phys. 47 (1998) 959998.
[30] D.E. Burkes, R. Prabhakaran, T. Hartmann, J.-F. Jue, F.J. Rice, Nucl. Eng. Des. 240
(2010) 13321339.
[31] D. Brown, D. Alexander, K. Clarke, B. Clausen, M. Okuniewski, T. Sisneros, Scr. Mater.
69 (2013) 666669.

Vous aimerez peut-être aussi