Académique Documents
Professionnel Documents
Culture Documents
by
and
T.O. Mason
Department of Materials Science and Engineering
Northwestern University
Evanston, IL 60208 USA
Reprinted from Cement and Concrete Research, Vol. 34, No. 10, 1787-1793, October 2004.
SCIENCE@DIRECT
CEMENTAND
Pergamon CONCRETE
RESEARCH
Cement and Concrete Research 34 (2004) 1787-1793
Abstract
A scanning electron microscope (SEM) point-counting technique was employed to. study the hydration of plain portland and blended
cement pastes containing fly ash or slag. For plain portland cement pastes, the results for the degree of cement hydration obtained by the SEM
point-counting technique were consiste t with the results from the traditional loss-on-ignition (LOI) of nonevaporable water-content
measurements; agreement was within I0% . The standard deviation in the determination of the degree of cement hydration via point
counting ranged from 1.5% to 1.8% (one operator, one sample). For the blended cement pastes, it is the first time that the degree of
hydration of cement in blended systems has been studied directly. The standard deviation for the degree of hydration of cement in the blended
cement pastes ranged from 1.4% to 2.2%. Additionally, the degrees of reaction of the mineral admixtures (MAs) were also measured.
The standard deviation for the degree offly ash reaction was 4.6% to 5.0% and 3.6% to 4.3% for slag. All of the analyses suggest
that the SEM point-counting technique can be a reliable and effective analysis tool for use in studies of the hydration of blended cement
pastes.
2004 Elsevier Ltd. All rights reserved.
Keywords: Hydration; Blended cement; Fly ash; Granulated blast-furnace slag; Point counting
1. Introduction directly study the degree of reaction of both MAs and the
portland cement in blended cement pastes.
There are many experimental techniques used to measure In recent years, scanning electron microscopy (SEM)
the degree of hydration of portland cement, such as mea- quantitative techniques have been used in cement systems to
suring the heat of hydration, the nonevaporable water directly estimate the degree of hydration of plain portland
content, and the amount of calcium hydroxide produced in cement pastes [6-8]. These researchers collected backscat-
the hydration reactions. All of these methods are based on tered electron images (BEi) of cement pastes, and then
comparisons of the measured parameters with the predicted quantitatively analyzed them using a system software based
or measured parameters. for a fully hydrated paste. When on the different gray levels of the different cement phases.
reactive mineral admixtures (MAs) are incorporated into fgr example, in Ref. [6], for cement pastes with water-to-
portland cement, the system becomes more complex. It is cement mass ratio (w/c) = 0.45 cured at 20 C, the degree of
almost impossible to determine the degree of the ceme)lt hydration was about 70% at 1 day, and about 80% at 3 days.
hydration by the techniques used for plain cement paste. These results were much higher than those obtained from
Although some researchers [1-5] have measured the degree calcium hydroxide measurements and density determina-
of reaction of MAs by a selective dissolution procedure, tions. The large disagreement may have been caused by the
little information can be obtained in this manner about the selection of inappropriate thresholds in the gray scale
cement hydration in a. blended system. Thus, it would be images or by limiting the image resolution so as to be able
valuable to develop a technique that can simultaneously and to discriminate between different gray levels. In blended
cement pastes, discriminating between phases is even harder
due to the similar gray levels of some of the fly-ash, the
Corresponding author. Tel.: +!-301-975-6708; fax: +1-301-990-
6891.
slag, and the cement clinker particles. Another SEM imag-
E-mail address: edward.garboczi@nist.gov (E.J. Garboczi). ing technique, X-ray imaging, may be used to estimate the
0008-8846/$ - see front matter 2004 Elsevier Ltd. All rights reserved.
doi: I 0. 1016/j .cemconres.2004.01.01 4
1788 X Feng et al. I Cement and Concrete Research 34 (2004) 1787-1793
------------- - a : -----
l::,
) 40 , #< :- -- -
I .
ff - - T-10
cement and the degree of reaction of the MA in blended . I
----T-1H
cement pastes containing fly ash or slag. A similar optical- 20 - - ... ..... .. .
:
based point-counting technique is the basis for a number of 0 L....L-1- '-'---1- J.....L--'-.L...1.--'-J.....J.....L..L..1.....L..L...J.......L...J.-L...J.....J.-L...J......J.....J
ASTM standard test methods [9,10]. The technique accesses 0 20 40 60 80 100 120 140
the same field of view as does X-ray imaging, but in a Diameter, m
fraction of the time, so that enough fields of view for
statistical purposes can be examined in a reasonable length Fig. l. Cumulative particle size distributions for the T-ID and T-IH
portland cements. The minimum particle size measured was I m. Air
of time. This is an important consideration considering that dispersion and laser scattering techniques were used.
the point-counting technique requires a human operator.
Loss-on-ignition (LOI) measurements of nonevaporable
water content for the plain portland cement pastes were cement atw/c=0.4. The slag blended cement (designated as
also perfonned to compare with the results from the SEM T-S) was made by mixing 30% (by mass fraction) slag with
point-counting technique. The accuracy of the LOI measure- 70% T-ID portland cement at a 0.4 water-to-solid binder (w/
ments is not known, but they could be a precise measure s) mass ratio. Both the slag and this cement, T-1D, were
because they are based on mass measurements, which can supplied by Dyckerhoff. The fly-a$h-blended cement (des-
be perfonned with great precision. ignated as T-L) was made by mixing 20% (by mass fraction)
fly ash and 80% T-IH cement at a 0.4 w/s mass ratio. Both
the fly ash and the T- lH cement were supplied by Holcim.
2. Experimental procedures Overall, four different cement pastes were studied.1
Fig. 1 shows the measured cumulative particle size
2.1. Materials distributions (PSD) for the T-ID and T-lH cements. The
minimum diameter is l m. These PSD curves were
Two kinds of Type I portland cements were selected for measured using a laser-scattering instrument with air dis-
the study and were designated as T-lH and T-ID. The persion (dry dispersion). A standard methodology for PSD
chemical compositions of the two cements, the Class F fly detennination, worked out in the context of several ASTM
ash, and the slag are given in Table 1 in tenns of oxide round robins, was used 11 . Fig. 2 shows similar PSD
mass percents. The LOI of the fly ash is low compared with curves for the slag and fly-ash materials, with the same
most Class F fly ashes because this fly ash was chosen for minimum diameter of 1 m. Wet dispersion in propanol was
its low carbon content and, hence, negligible effect on air used for these materials. The PSD of the blended cements
entrainment. used in this study would, of course, be a weighted average
of the relevant curves in Figs. 1 and 2. Io Fig. 1, the T-ID
2.2. Paste preparation cement was finer than the T-lH cement, and Fig. 2 shows that
the slag material was finer than the fly ash material.
Two plain portland cement pastes were made by mixing To eliminate entrapped air bubbles, house vacuum was
deionized water in a blender with either T-lH or T-lD employed during the mixing and casting processes, and a
vibration table was used during casting. To prevent
substantial bleeding, the samples were cast in sealed
Table I plastic molds and slowly rotated for 5 h on a roller.
Chemical compositions of cements, fly ash and slag used (mass fraction, %) By constantly changing the orientation of the samples
T- IH T- ID Fly ash Slag with respect to the direction of gravity, bleeding was
Si02 prevented. After this, the pastes were kept in the molds
AJi03 and put in sealed plastic bags at 20 C and 100% RH
Fe203 until the testing age was reached.
CaO
MgO
S03 I Certain commercial equipment and/or materials are identified in this
K20 report to adequately specify the experimental procedure. In no case does
Na20 such identification imply recommendation or endorsement by the National
LOI Institute of Standards and Technology, nor does it imply that the equipment
-
These are taken directly from the manufacturers' analyses. and/or materials used are necessarily the best available for the purpose.
X Feng et al./ Cement and Concrete Research 34 (2004) 1787-1793 1789
with the ethanol. After two weeks, the resin was cured at
100 70 C. The sample was then polished and carbon coated for
SEM point-counting analysis.
80 The point-counting procedure was modified from ASTM
Q)
(a) T-1D w/c::0.4, 28d, XlOOO (b) T-L w/s::0.4, 7d, XlOOO (c) T-S w/s::0.4, 28d, XlOOO
Fig. 3 . Typical fields of view used for point counting on plain cement paste and blended cement pastes.
1790 X Feng et al. I Cement and Concrete Research 34 (2004) 1787-1793
products and voids, which are counted as "all others" in this The volume fraction of unhydrated cement at hydra-
technique. Fig. 3(b) and (c) shows fly-ash- and a slag-blended tion time t, Vct)cem, is calculated by dividing the number
cement pastes, respectively. In both microstructures, the gray of points counted for this phase by the total number of
levels of some fly-ash and some slag particles were very close points counted (3000 in the current study). However, one
to those of hydration products, while the gray levels of the needs to apply a gypsum correction. All the gypsum
others were very close to the cement clinker particles. To help contained in the cement is substantially reacted by about
distinguish between these phases, the examiner can use 24 h after mixing [13). Therefore, it will not be counted
particle morphology and, if necessary, use the X-ray signal as a cement linlq:r phase during point counting. The
to show the elemental abundance at a questionable point. gypsum does not contribute to the cement paste LOI
Using these indicators, as well as the gray level, one can measurement because it is not present, but it does
readily discriminate among the fly-ash and slag particles, contribute to the initial volume of cement. Therefore,
hydration products, and cement clinker particles at each grid the initial volume fraction of portland cement needs to be
point. In Fig. 3(b), there is one fly-ash particle falling at grid reduced by the amount of gypsum originally present in
points that is marked "2", two cement clinker particles that the portland cement when calculating the degree of
are marked " 1", and the unmarked intersections are counted hydration of the cement paste at time t:
as "all others". In Fig. 3(c), there are five cement clinker
particles marked " 1", one slag particle marked "3 ", and the
unmarked grid points have been counted as "all others".
This standard deviation in the phase volume fraction is (a) T-1H w/c=0.4
80,-------........-------
then converted to the standard deviation in the degree of
cement hydration: ,ft. 70-
6
...J 60-
(8)
rS 50
R2 = 0 .90
where a"' is the measured standard deviation for the degree 40 -j--- - --,.- - - -1,,---- - -T,.;._. ...J
of hydration as determined by point counting. This is done 40 50 60 70 80
a, by point-coun ting, %
similarly for the blended cements and standard deviation in
the degree of reaction in the MA. (b) T-1D w/c=0.4
BOT-"------------'----,
,ft. 70
3. Results and discussion
g 60
3.1. Degree of hydration of plain cement pastes obtained .tj 50 1-+i y = 1.06x
from LOI and SEM point-counting R2 =0.85
40 +-- - -,.,-......;.._,,....,.- - - "T.-, - --1
The estimated degrees of hydration of the plain portland 40 50 60 70 80
a, by point-counting, %
cement pastes made from the T-lH and T-1D cements as a
function of age, determined by LOI and SEM point-counting, Fig. 4. Regression analysis relating SEM point counting and LOI
are listed in Table 2. The LI column in the table stands for the determinations of o: for the plain portland cement pastes.
difference between those two measurements (SEM point
counting - Lon, divided by the LOI results and multiplied
by 100%. The results are consistent with each other as the regression fit indicates a multiplicative factor. If the intercepts
deviations for both cement pastes are within 10%.' are not forced to be 0, then, the regression equations become:
One should note that, in the LI columns, the value of LI is
almost always negative. This indicates that the SEM point- T ,..;. lH cement paste:
counting results forthe degree of hydration are systematically cx% (LOI)
lower than the LOI measurements. This may be partly due to = 0.91 X cx%(SEM) + 0.08, R2 = .91
other materials releasing mass in the temperature range105-'-
1000 C, like calcium carbonate, which releases carbon T - ID cement paste:
dioxide, which would tend to make the LOI results higher
than the point-counting results. The calcium carbonate could CX%(LOI)
come from the partial carbonation of the CH present. = 0.94 X cx%(SEM) + 0.07, R2 = .86
Fig. 4 shows the regression analysis between the data
obtained by LOI and SEM point counting for the two plain The R2 factors are now somewhat closer to one, and the
cement pastes. The values of R2 for these regressions are .90 slopes are now a little less than one. The positive intercepts
for the T-lH cement and .85 for the T-1D cement, with the are slightly larger than zero, which means that the degree of
regression lines forced to go through the origin. Note that cement hydration determined by LOI is.a little larger than
both slopes are close to and a little larger than unity. The fact that determined by SEM due to an additive factor. This is
that the point-countip.g measurements seem to be systemat- reasonable because some of the very fine particles of the
ically lower than the LOI measurements could be due to a cement clinker may be miscounted due to the limited
multiplicative factor or to an additive factor. This kind of rellolution and because of the overall uncertainty in phase
identification. Fig. 4 shows the actual standard deviations in
the degree of cement hydration as determined by point
Table 2
counting, calculated according to Eqs. (7) and (8). The
Degree of hydration of plain cement pastes by LOI and point counting standard deviation in the point-counting determination of
Age (days) T-lH (w/c=0.4) T-lD (w/c=0.4)
the degree of hydration ranged from 1.5% to .1.8%.
All of the above analyses suggest that the results from
oc% SEM o:%LOI LI( %) cx%SEM oc% L0 I LI(%)
SEM point counting are close to those from the traditional
3 52.2 53.9 -3.1 50.8 51.4 - 1.1 LOI method and that the differences are within a reasonable
7 54.8 60.0 -8.7 53.8 59.8 - 9.9
14 62.9 64.3 -2.2 57.8 62.4 -7.3 range. However, the SEM point-counting results are sys-
28 64.2 66.8 - 4.0 59.9 66.1 -9.4 tematically lower than the LOI measurements. The uncer-
60 68.7 68.6 0.2 66.8 71.3 -6.3 tainties of the LOI measurements are much lower than those
90 67.2 71.6 - 6.1 68.8 68.8 0 for the point-counting procedures and would not show up on
1792 X Feng ei al. I Cement and Concrete Research 34 (2004) 1787-1793
50
the scale used in Fig. 4. Thus, SEM point counting, when
carefully used, can. be an effective technique to determine 0
_._FA
<{
the degree of hydration ofplain cement paste. Having tested 40 -- - Slag
0
this procedure on plain portland cement pastes, attention is 0C: 30
now turned to its much more important use on blended n<ll
a: 20
Q)
cement pastes, as LOI measurements are not meaningful at
present for blended cements. 0
Q)
10
Cl
3.2. Degree of hydration of port/and cement in blended Q)
cement systems
0 0
10 100 1000 10000
Age.hrs
Fig. 5 shows the degree of hydration of the portland
cement in blended cement pastes compared with the degree Fig. 6. Degree of reaction of MA in blended cement systems.
of hydration of plain cement pastes, as determined by point
counting only. The incorporation of MAs can significantly used in the manufacture of a fly-ash-blended cement by
accelerate the cement hydration in blended systems as early Holcim, so that a fly ash that retarded cement hydration
as 3 days, especially for the slag-blended cement. This effect would not have been so used. Hence, it is not surprising that
is more significant between 7- and 28-day hydration, and is no apparent retardation with the fly ash present was seen
less significant at later ages. There are two possible reasons experimentally.
for this acceleration: (1) the relatively higher w/c in the
blended system and (2) the pozzolanic reaction between the 3.3. Degree of reaction of MAs in blended cement systems
MA and the calcium hydroxide released from the cement
hydration in the fly-ash-blended cement. The standard The degrees of reaction of MA in the blended systems
deviation for the degree of cement hydration in the blended are shown in Fig. 6, along with the standard deviations.
cement pastes, as determined by point counting, ranged These degrees of reaction come from a direct point-counting
from 1.4% to 2.2%. measurement. At all ages, slag reacts much faster than fly
One should note that, often, the addition of fly ash retards ash due to its inherent hydraulic and higher pozzolanic
portland cement hydration. This particular fly ash had been reactivity. The standard deviation for fly ash reacted is
from 4.6% to 5.0% , while for slag reacted, it is from
3.6% to 4.3%. The higher standard deviation for fly
ash than for slag is due to the lower replacement of fly ash
(a) T-1H and T-L
>, 90 ; (20%) in the blended cement pastes than for slag (30%). In
..c Eq. (8), similar point-counting standard deviations will be
g 80 divided by smaller numbers, thus giving larger standard
o
... C:
70 1 deviations for the extents of reaction. These results are
E
:I: :,
0
comparable with the results from selective dissolution
0 '-r 60 methods as performed by other researchers [2 3]. For
<1>-
<I> .E
... 0
50 example, with a 20% fly ash replacemen,t a 10% fly-ash
gi C.
0 - T-L extent of reaction at 28 days was reported by Hardt! [2]. Our
40 1
work here shows an 11% reaction extent, with a standard
10 100 1000 10000
Age.hrs deviation of 4.8%. With a 30% slag replacement, Lumley
et al. [3] found a 35% slag degree of reaction at 28 days,
(b) T-10 and T-S while in our work, a 28 3.9% reaction extent was mea-
90 - - - - - - - - - -
sur.ed. Standard deviations for the literature measurements
were not given.
4. Conclusion
This procedure was first checked by comparing SEM Cement-Based Materials and the Environmental Protection
point-counting with LOI measurements on two plain port- Agency for partial funding of this research.
land cements. It was found that the degree of hydration of
plain portland cement pastes obtained from SEM point-
counting techniques were consistent with the results from
References
the traditional LOI method, with differences between the
two methods no more than 10%. The SEM point-count- [I] L. Lam, Y.L. Wong, C.S. Poon, Degree of hydration and gel/space
ing results were systematically lower than the LOI measure- ratio of high-volume fly ash/cement systems, Cem. Coner. Res. 30
ments. The standard deviation in the point-counting (2000) 747-756.
determination of the degree of hydration of plain portland [2] R. Hardt!, The pozzolanic' reaction of fly ash in connection with
cement ranged from 1.5% to 1.8%. This is for a single different types of cement, in: H. Justnes (Ed.), Proceedings of the
I 0th International Conference on Chemistry of Cement, vol. m. Amar-
operator, on a single small sample cut from a larger sample. kai and Congrex Goteburg, Gothenburg, Sweden, 1997, pp. 8- 15.
Once this technique was established, it was used to measure [3] J.S. Lumle R.S. Gollo G.K. Moir .F.
the degrees of hydration of cement in blended cement action of the sla in some blends with oortland
systems for the first time. The standard deviations for the Res. 26 (1996) 139- 151.
degree of hydration of cement in the blended cement pastes [4] J.I. Escalante L.Y. Gomez K.K. Johal n bl.
Mendez, Reactivity of blast-furnace slag in portland cement blends
were within 1.4% and 2.2%. The same SEM point- hydrated under different conditions, Cem. Coner. Res 3 I (2001)
counting technique was also used to measure the degree of 1403'- 1409.
reaction of the MAs in the blended cements. The results [5] A.F. Battagin, Influence of degree of hydration of slag on slag cements.
were comparable with previous results obtained using a in: A.K. Mullick (Ed.}, Proceedings of tlie 9th International Conference
on Chemis of Cement vot m. National Co 1 fo
selective dissolution method. The standard deviations for
Building Materials, New Delhi, India, I992, pp. 166- 172.
the degrees of fly-ash reaction ranged from 4.6% to [6] J.C. Gomes, J. Cabrera, S. Jalali, The degree of cement hydration
5.0% and, for the slag-blended cements, from 3.6% determined by backscattered electron imaging, in: M. Cohen, S.
to 4.3%. Mindess, J. Skalny (Eds.), Materials Science of Concrete-The
Results of the present study indicate that the SEM point- Sidney Diamond Symposium, American Ceramic Society, Wester-
ville, OH, 1998, pp. 109-126.
counting technique can be a reliable and eff ctive analysis [7] J.E. Ash M.G. Hall J.I. Lan ford M. Mellas s ati s e
tool to measure the degree of hydration of the portland of hydration of portland cement pastes, Cem. Coner. Res 23 0993)
cement and MAs in blended cement pastes. These measure- 399- 406.
ments will allow models like the Virtual Cement and [8] "i<::oT;ellsen, R.J. Detwiler, O.E. Gjo rv, Back scattered electron im-
Concrete Testing Laboratory [15] to be validated on these aging of cement pastes hydrated at different temperatures, Cem.
Coner. Res. 20 (I 990) 308-311.
materials. With the growing use of blended cements utiliz-
[9] ASTM Cl356 M-96, Standard Test Method for Quantitative Determi-
ing waste stream materials, this combined measurement and nation of Phases in Portland Cement Clinker by Microscopic Point-
prediction capability will allow better and more controlled count Procedure.
use of these materials. [10] ASTM C457-98 , Standard Test Method for Microscopical Determina-
tion of Parameters of the Air-void System in Hardened Concrete.
[11] C.F. Ferraris, V.A. Hackley, A.I. Aviles, Measurement of Particle Size
Distribution in Portland Cement Powder: Analysis of ASTM Round-
Acknowledgements Robin Studies, ASTM Cem., Coner. Aggreg. June, 2004.
[12] D. Bentz, X. Feng, C. Haecker, P. Stutzman, Analysis of CCRL pro-
The authors would like to thank Holcim, and Dyckerhoff ficiency cements 135 and 136 using CEMHYD3D, NIST Internal
GmbH for supplying materials for this study and Dr. Chiara R rt 6545 Au st 2000.
[13] H.F.W. Taylor, Cement Chemistry, 2nd edition, Thomas Telford, Lon-
Ferraris and Mr. John Winpigler of the Inorganic Materials
don, 1997, p. 188.
Group at NIST for performing specific gravity and particle [14] A. Papoulis, Probability, Random Variables, and Stochastic Processes,
size distribution measurements on the various powders. The McGraw-Hill, New York, 1965.
authors would also like to thank the Center for Advanced [ I 5J http://vcctl.cbt.nist.gov.
The author has requested enhancement of the downloaded file. All in-text references underlined in blue are linked to publications on ResearchGate