Oilwell Cement Clinkers

X-ray Microanalysis and Phase Composition

Christopher Hall and Karen L. Scrivener*
Schlumberger Cambridge Research, High Cross, Cambridge, United Kingdom; and Department
of Materials, Imperial College, London, United Kingdom

X-ray microanalyses have been made on a suite of nine oilwell Experimental

cement clinkers. Elemental data were obtained on alite, belite, and
ferrite phases and, for two clinkers only, also on aluminate. The alite The materials used were drawn from primary samples
and belite compositions are broadly similar to those previously (typically 1 to 5 kg) of production clinkers. They were
reported for other portland cements. Guest ion concentrations supplied from a number of plants in several countries.
suggest that several charge-balancing substitutions occur. Bulk The suite was chosen to represent a typical range of
MgO and SO3 levels determine the Mg and S contents of both clinkers for API class G or class H oilwell cements. We
phases, but for Fe and Al the substitution levels are not strongly denote the individual materials D1 through D9. D1 to
correlated with bulk composition. For ferrites, the data are generally D5 were taken from the same batches as materials for
consistent with the results of Bergstrom et al. (Adv Cem Res 1992,
which ferrite data were reported in our earlier paper [8]
4, 141–147): the Mg content varies widely, is controlled by the bulk
(see footnote, Table 1).
Bulk chemical analyses were carried out by standard
MgO, and is coupled with Si in a charge-balancing substitution for
X-ray fluorescence (XRF) methods (major and minor
Fe. Estimates of the total phase assembly using the directly deter-
oxides) and by wet methods for free lime and SO3. The
mined mineral compositions are compared with the predictions of a
results are given in Table 1.
modified Bogue calculation (similar to that of Taylor, Adv Cem
Specimens for microanalysis were prepared as de-
Res 1989, 2, 73–77). The agreement is variable and some refine-
scribed previously [8]. A sub-sample of several tens of
ments to the method of calculation are indicated. Advanced
grams of each clinker was crushed and homogenized.
Cement Based Materials 1998, 7, 28 –38. © 1998 Elsevier
Material from the 1 to 4 mm size fraction was mounted
Science Ltd.
in epoxy resin and polished to 1/4 mm. X-ray micro-
KEY WORDS: Cement, Clinker, Microanalysis, Alite, Belite,
analyses were made by energy-dispersive spectroscopy
Ferrite (EDS). Quantitative estimates of elemental composition
were obtained by comparison with stored standard
spectra and using the ZAF procedure. For each clinker
any X-ray microanalyses of cement clinkers at least 10 EDS analyses were made for each phase. For

M have been published previously [1–7] but

few of these are for oilwell cement materials.
a given phase, each analysis was made on a different
piece of clinker. In total, approximately 500 spot anal-
yses were obtained. In addition, X-ray diffraction (XRD)
In an earlier paper [8] we reported the results of an
X-ray microanalysis of the interstitial material in a suite powder patterns and scanning electron microscopy
of eight oilwell cement production clinkers. We have (SEM) back-scattered electron images were recorded.
now extended this work to the silicate phases. The For these anhydrous phases, at the accelerating voltage
of 15 kV used, each microanalysis relates to a volume
materials included in the later study that is described
somewhat less than 1 mm in each dimension. All analyses
here comprised five of the original suite and four
were taken well away from phase boundaries, as far as
additional clinkers. New analytical data were also ob-
could be judged from the two-dimensional section. For
tained on the interstitial material. For these nine clin-
alite and belite, analytical errors attributed to X-rays
kers, a fuller statistical analysis is now possible.
coming from adjacent phases (which may in some cases
be beneath the plane of the section and thus invisible) is
Address correspondence to: C. Hall, Schlumberger Cambridge Research, High
not considered a significant source of error. In the case of
Cross, Madingley Road, Cambridge CB3 0EL, United Kingdom. the other phases (the ferrite and particularly the alumi-
Received January 20, 1997; Accepted March 17, 1997
*Present address: Lafarge, Laboratoire central de recherche, 95 rue du Mont-
nate, where identified), the risk of contamination is
murier BP 15, 38291 Saint-Quentin-Fallavier, France. greater. Interstitial ferrite is irregular and aluminate has

© 1998 by Elsevier Science Ltd. ISSN 1065-7355/98/$19.00

655 Avenue of the Americas, New York, NY 10010 PII S1065-7355(97)00035-7
Advn Cem Bas Mat Oilwell Cement Clinkers 29
1998;7:28 –38

TABLE 1. Oxide analysis for clinkers

D1 D2 D3 D4 D5 D6 D7 D8 D9
Na2O 0.11 0.11 0.09 0.19 0.08 0.21 0.09 0.08 0.14
MgO 2.33 0.80 0.70 0.90 1.80 1.65 1.90 0.75 0.70
Al2O3 3.98 3.40 3.55 4.80 3.30 3.70 3.40 3.70 3.75
SiO2 21.68 22.60 22.95 21.65 22.95 22.40 23.20 22.00 23.60
SO3 1.15 1.10 1.02 0.17 0.60 0.81 0.52 0.20 0.20
P2O5 0.21 0.00 0.00 0.21 0.13 0.04 0.13 0.11 0.11
K2O 0.54 0.54 0.41 0.17 0.54 0.55 0.48 0.47 0.67
CaO 64.86 66.20 66.20 65.75 65.15 65.05 64.95 65.30 64.90
TiO2 0.30 0.0 0.16 0.24 0.18 0.19 0.17 0.16 0.17
Mn2O3 0.03 0.0 0.07 0.06 0.16 0.06 0.15 0.19 0.17
Fe2O3 4.66 5.20 4.25 5.10 4.55 4.95 4.65 6.75 4.65
loss 0.00 0.20 0.15 0.50 0.40 0.35 0.20 0.45 0.30
residue 0.13 0.0 0.01 0.09 0.15 0.02 0.08 0.11 0.02
Free lime 0.90 0.60 0.25 0.55 0.60 0.25 0.35 0.55 0.40
Total 99.85 100.15 99.55 99.74 99.84 99.96 99.84 100.16 99.36
Note: Samples D1, D2, D3, D4, and D5 are taken from the same batches as samples C5, C9, C7, C3, and C6, respectively, of Bergstrom et al. [8].

small grain size, usually around 1 mm or less. Spot clinkers, between one quarter and one third of the total
analyses that departed dramatically from stoichiometry Al resides in the silicate phases. Overall minor oxide
(.10%) were discarded as being unreliable. Even so, the levels are in the range of 2.7 to 4.4 wt % for alite and 3.6
results for the aluminate phase should still be treated with to 6.2 wt % for belite.
some caution.
It is difficult to give a precise value for the accuracy of Variation in Composition Within Clinkers
EDS microanalysis. The microscope employed in this
study is used exclusively for the examination of cemen- The spread of local compositions as sampled by indi-
titious materials and the accuracy of the standards vidual spot analyses is a measure (albeit a rough one) of
correction has been optimized for these materials. We the structural homogeneity of the phase. For both alite
consider that the accuracy of individual element spot and belite phases the compositions are quite erratic, as
analyses probably approaches 0.1 atom or weight % (1 can be seen from the variances of the main minor oxides
part in 103). However, when several analyses are made such as MgO, Al2O3, and Fe2O3 (see also the cluster
on a phase, the observed compositional variations from plots of Figure 1). These variances are generally smaller
point to point are much greater than this, so that than those of the minor oxides of the ferrite phase.
measurement errors contribute relatively little to the
statistical variance. Consideration of charge balance Guest Ions and Substitution Patterns
and stoichiometry suggest that the overall accuracy of Statistical analysis of the individual spot compositions
the mean values from each phase were perhaps better reveals several internal correlations between the con-
than 0.1%, so that values are quoted to 0.01%. centrations of certain elements in the composition data
set. We have calculated the correlation matrix for the
Silicate Phases entire alite spot analysis dataset, and likewise for the
belite dataset. The more significant correlation coeffi-
Composition cients (r . 0.5) are given in Table 4. For alite, we find
In Tables 2 and 3 we show the compositions of alite and strong inverse correlations between Mg and Ca and
belite phases in the suite of clinkers examined. These between Al and Si; and a positive correlation between
are the means of the individual spot analyses for each Al and Fe.
clinker. Variances are also tabulated. Taken as a whole For belite, there is a similar strong inverse correlation
these compositions (both means and ranges) are similar between Al and Si, and a positive correlation between
to those previously reported for construction cements Al and Fe. There is also a clear inverse correlation
[1–7]. However, the alite Fe in these Fe-rich clinkers is between Si and S.
rather higher, as is the Al, in spite of the low bulk Al. In The correlations reflect the complexities of the under-
the belite, the sulphate content is notably higher than lying substitution patterns in these calcium silicate
reported elsewhere. In both silicate phases the Mg is, as minerals. To help more detailed discussion, we express
usual, very variable from clinker to clinker, but there is the composition in terms of the parent structural for-
much less variation in Al and Fe contents. In these mula units: for alite, Ca3SiO5, normalizing the atomic
30 C. Hall and K.L. Scrivener Advn Cem Bas Mat
1998;7:28 –38

TABLE 2. Alite oxide compositions in the clinker suite

D1 D2 D3 D4 D5 D6 D7 D8 D9 Mean
Na2O 0.01 0.0 0.08 0.04 0.0 0.05 0.0 0.0 0.07 0.01
0.29 0.30 0.23 0.27 0.29 0.12 0.18 0.25 0.26 0.25
MgO 1.19 0.23 0.29 0.19 1.13 0.87 1.07 0.30 0.28 0.62
0.33 0.28 0.23 0.22 0.40 0.23 0.20 0.27 0.29 0.49
Al2O3 1.47 1.25 1.10 1.07 1.20 1.30 1.14 1.32 1.29 1.25
0.40 0.29 0.39 0.38 0.42 0.38 0.36 0.32 0.59 0.41
SiO2 24.56 24.92 24.73 25.99 25.18 25.30 25.09 24.80 25.05 25.06
0.52 0.63 0.57 0.56 0.69 0.88 0.47 0.50 0.93 0.76
SO3 0.38 0.46 0.30 0.25 0.11 0.35 0.12 0.17 0.12 0.24
0.22 0.18 0.26 0.22 0.17 0.30 0.15 0.18 0.21 0.24
Cl 0.04 0.07 0.02 0.09 0.08 0.0 0.04 0.0 0.04 0.03
0.07 0.06 0.10 0.09 0.10 0.09 0.08 0.10 0.07 0.09
K2O 0.06 0.04 0.09 0.03 0.07 0.13 0.03 0.14 0.15 0.09
0.11 0.10 0.10 0.08 0.05 0.13 0.13 0.13 0.10 0.11
CaO 71.07 71.50 72.43 71.33 71.21 70.70 71.37 72.15 71.37 71.46
0.90 0.46 0.74 0.63 0.80 0.79 0.71 0.57 0.84 0.88
TiO2 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
0.15 0.18 0.24 0.20 0.14 0.17 0.20 0.14 0.21 0.19
Mn2O3 0.08 0.20 0.07 0.06 0.07 0.12 0.12 0.17 0.18 0.10
0.21 0.18 0.26 0.27 0.21 0.24 0.25 0.24 0.20 0.24
Fe2O3 1.13 1.44 1.09 0.98 1.04 1.12 1.12 1.00 1.52 1.18
0.45 0.36 0.40 0.39 0.41 0.52 0.45 0.21 0.58 0.46
Notes: For each element, the mean and standard deviation of the X-ray spot analyses are expressed as oxide and normalized to 100 mass percent; the last column
gives standard deviations for each element for the entire dataset, not means of the standard deviations for each clinker.

composition to 5 O atoms; and for belite, Ca2SiO4, isomorphous divalent substitution MgCa21. Recent nu-
normalizing the atomic composition to 4 O atoms. This clear magnetic resonance (NMR) data [9] indicates that
is shown in Tables 5 and 6. Al substitutes almost exclusively in tetrahedral Si sites
in a synthetic alite and various clinkers. This substitu-
The observed inverse correla-
ION SUBSTITUTION IN ALITE. tion requires charge balancing. For the solid solution
tion between Ca and Mg in alite is consistent with the C3S–Al2O3, Midgley and Fletcher [10] proposed the

TABLE 3. Belite oxide compositions

D1 D2 D3 D4 D5 D6 D7 D8 D9 Mean
Na2O 0.10 0.08 0.11 0.25 0.17 0.19 0.07 0.05 0.10 0.12
0.19 0.22 0.31 0.25 0.19 0.21 0.31 0.29 0.17 0.25
MgO 0.06 0.0 0.0 0.0 0.09 0.14 0.17 0.0 0.0 0.0
0.28 0.21 0.21 0.22 0.37 0.28 0.29 0.31 0.28 0.30
Al2O3 1.75 1.56 1.63 1.69 1.35 1.84 1.64 1.36 1.06 1.55
0.30 0.21 0.23 0.34 0.22 0.29 0.38 0.50 0.37 0.41
SiO2 30.69 31.44 30.92 32.09 31.59 31.28 31.52 32.25 32.77 31.64
0.60 0.97 0.60 0.79 0.24 0.75 0.75 1.21 1.06 1.05
SO3 1.69 1.38 1.28 0.73 0.58 1.48 0.88 0.49 0.31 0.97
0.38 0.31 0.48 0.43 0.16 0.98 0.31 0.44 0.22 0.67
Cl 0.01 0.05 0.04 0.06 0.04 0.01 0.02 0.00 0.03 0.03
0.07 0.07 0.09 0.12 0.04 0.11 0.10 0.08 0.09 0.09
K2O 0.39 0.37 0.59 0.46 0.41 0.64 0.47 0.58 0.77 0.54
0.23 0.12 0.31 0.22 0.12 0.39 0.14 0.23 0.27 0.28
CaO 63.55 63.54 63.65 63.42 64.16 62.49 63.38 63.93 63.61 63.49
0.64 0.27 0.95 0.77 0.73 0.74 0.84 0.92 0.76 0.86
TiO2 0.18 0.0 0.01 0.0 0.00 0.0 0.04 0.0 0.0 0.01
0.19 0.20 0.22 0.21 0.15 0.19 0.14 0.10 0.18 0.19
Mn2O3 0.02 0.0 0.14 0.02 0.0 0.03 0.09 0.17 0.09 0.07
0.20 0.24 0.20 0.25 0.13 0.20 0.16 0.20 0.21 0.21
Fe2O3 1.55 1.82 1.71 1.43 1.60 1.93 1.72 1.44 1.55 1.63
0.32 0.46 0.39 0.33 0.40 0.56 0.46 0.60 0.65 0.50
Notes: For each element, the mean and standard deviation of the X-ray spot analyses are expressed as oxide and normalized to 100 mass percent; the last column
gives standard deviations for each element for the entire dataset, not means of the standard deviations for each clinker.
Advn Cem Bas Mat Oilwell Cement Clinkers 31
1998;7:28 –38

TABLE 4. Correlation coefficients expressing the more signif-

icant correlations between the concentrations of individual
elements in the clinker phases
Phase Correlation Coefficient
Alite (n 5 142)
Mg, Ca 20.57
Al, Si 20.70
Al, Fe 10.62
Si, Fe 20.52
Belite (n 5 136)
Al, Si 20.72
Al, Fe 10.54
Si, S 20.68
Ferrite (n 5 173)
Mg, Fe 20.61
Si, Fe 20.63
Included are all Pearson product-moment correlation coefficients .0.5 (highly
significant); n is the sample size.

for the cations and the framework elements are also

included in Table 5. The simplest explanatory model
consistent with the microanalytical data is that Al and S
substitute for Si and this substitution is charge-balanced
by a combination of Fe, Na, K, and Mn substituted for
Ca. The alite S levels in these clinkers are always low
(,0.5 wt %) and the SO4 ion does not appear to play an
important part in the alite substitution chemistry.

ION SUBSTITUTION IN BELITE. In belite (in contrast to alite) S

does have an important role in the substitution pat-
terns. It is known that SO3 is concentrated in the belite
FIGURE 1. Cluster plots showing typical variation of some phase but the levels found here are very variable and
minor oxide concentrations in alite and belite spot analyses. frequently high (range 0.3 to 1.7 wt %) [12]. Statistically
(a) MgO and Al2O3 in alite (clinkers D1 h, D3 E, and D6 å
only are included for clarity); (b) SO3 and Al2O3 in belite significant inverse correlations are found between the
(clinkers D1 h, D2 ■, and D9 Ç). concentrations of Si and Al (20.72) and between Si and
S (20.68). In most clinkers Al and Fe spot concentra-
tions are also positively correlated, although the overall
simple substitution 4Al(3Si)21, with excess Al occupy- correlation coefficient for the entire suite is only 0.54.
ing a vacant lattice site. Hahn et al. [11] argued also for The cation charge subtotals are given in Table 6. They
the substitution FeCa21 in the system C3S–Fe2O3. The
behave much as the alite totals do, except that the Si 1
positive correlation between Fe and Al and the inverse
Al 1 S charge never exceeds 4.0, even though the S
correlation between Ca and Fe concentrations in our
levels are invariably quite high. Recently obtained
clinker alites suggest (for these Fe-rich production clin-
NMR data [9] show that Al substitutes exclusively for Si
kers at least) that the substitution FeCa21 may provide
one mechanism for charge compensation. In any case, in a synthetic substituted dicalcium silicate. The present
in production clinkers such as these with numerous data therefore support a charge-balancing substitution
guest ions, several other contributions to overall charge involving S/Al for Si. However, considerable Fe is also
balance can easily be envisaged. These cannot be dis- present in these belites. The location of the Fe must
cerned from microanalytical data alone. Therefore, we remain an open question but the overall charge-balance
employ the conventional cation sums (Ca 1 Mg 1 K 1 is found to be closer to the C2S parent if Fe is partitioned
Na 1 Fe 1 Mn) and (Al 1 Si 1 S 1 Ti), which are equally between the Ca-type and tetrahedral (Si-type)
shown in Table 5. For these clinkers, these sums vary sites. There is also a weak inverse correlation (20.41)
somewhat both above and below the stoichiometric between the belite Ca and Fe spot concentrations. Both
values of the parent tricalcium silicate. The charge sums features suggest some substitution of Fe into Ca sites.
32 C. Hall and K.L. Scrivener Advn Cem Bas Mat
1998;7:28 –38

TABLE 5. Mean molar compositions of alite phase, atoms per five oxygen atoms
D1 D2 D3 D4 D5 D6 D7 D8 D9 Mean
Na 0.001 0.0 0.006 0.003 0.0 0.004 0.0 0.0 0.005 0.001
Mg 0.067 0.013 0.016 0.011 0.064 0.049 0.061 0.017 0.016 0.035
Al 0.066 0.056 0.049 0.048 0.054 0.058 0.051 0.059 0.058 0.055
Si 0.932 0.947 0.942 0.982 0.953 0.956 0.952 0.944 0.954 0.951
S 0.011 0.013 0.008 0.007 0.003 0.010 0.003 0.005 0.003 0.007
Cl 0.002 0.004 0.001 0.005 0.004 0.0 0.002 0.0 0.002 0.002
K 0.003 0.002 0.004 0.002 0.003 0.006 0.002 0.007 0.007 0.004
Ca 2.890 2.911 2.956 2.887 2.890 2.864 2.904 2.944 2.913 2.906
Ti 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
Mn 0.002 0.006 0.002 0.002 0.002 0.0 0.003 0.004 0.005 0.002
Fe 0.032 0.041 0.031 0.028 0.030 0.040 0.032 0.029 0.043 0.034
cation 3.00* 2.97 3.02 2.93 2.98 2.96 3.00 3.00 2.99 2.98
subtotals 1.01† 1.01 0.99 1.04 1.01 1.02 1.01 1.01 1.01 1.01
charge 6.02* 5.99 6.06 5.89 6.00 5.95 6.03 6.03 6.01 6.00
subtotals 3.99† 4.03 3.94 4.11 3.99 4.05 3.98 3.98 4.00 4.01
Total cation charge is calculated assuming formal charges: Fe, Mn 13; S 16.
*Na 1 Mg 1 K 1 Ca 1 Mn 1 Fe.
†Al 1 Si 1 S 1 Ti.

Distribution of Minor Oxides Between Alite results [8]. The overall picture is of highly variable
and Belite interstitial compositions, with large clinker-to-clinker
Figure 2 shows the partitioning of the minor oxides in variance and substantial textural variation within indi-
alite and belite phases in the same clinker. K and S are vidual clinker nodules. This is also consistent with the
strongly concentrated in the belite; Mg is concentrated observations of Richardson et al. [13], who showed on a
somewhat in the alite. Na levels are too low for definite single clinker specimen that composition swings equal
conclusions, although this element appears to associate to the entire sample range occurred over traverses of a
with K. Al and Fe are not strongly selected, although Fe few microns.
has a weak preference for belite. These observations In Table 7, we give the oxide compositions of the
agree qualitatively with those of Ghose and Barnes [5a, ferrite phase in each clinker; in Table 8 the compositions
5b] and Kristmann [6]. are expressed as atoms per 10 oxygen atoms.
In these Fe-rich oilwell clinkers, aluminate is either
absent or rare. However, in two cases (D4 and D8),
Interstitial Material aluminate material was identified in the back-scattered
The new microanalytical data on the interstitial material electron image in sufficient amount to allow X-ray
confirm to a large extent the previously published microanalysis. Aluminate compositions for D4 and D8

TABLE 6. Mean molar compositions of belite phase, atoms per four oxygen atoms
D1 D2 D3 D4 D5 D6 D7 D8 D9 Mean
Na 0.005 0.004 0.006 0.014 0.010 0.011 0.004 0.003 0.006 0.007
Mg 0.003 0.0 0.0 0.0 0.004 0.006 0.007 0.0 0.0 0.0
Al 0.059 0.053 0.056 0.058 0.046 0.062 0.056 0.046 0.036 0.053
Si 0.884 0.905 0.895 0.924 0.914 0.901 0.911 0.933 0.947 0.913
S 0.037 0.030 0.028 0.016 0.013 0.032 0.019 0.011 0.007 0.021
Cl 0.001 0.002 0.002 0.002 0.002 0.000 0.001 0.0 0.001 0.001
K 0.014 0.013 0.022 0.017 0.015 0.023 0.017 0.021 0.028 0.019
Ca 1.962 1.960 1.974 1.957 1.989 1.928 1.962 1.978 1.971 1.965
Ti 0.004 0.0 0.0 0.0 0.0 0.0 0.001 0.0 0.0 0.0
Mn 0.001 0.0 0.003 0.0 0.0 0.001 0.002 0.004 0.002 0.001
Fe 0.034 0.040 0.037 0.031 0.035 0.042 0.037 0.031 0.034 0.036
cation 2.02* 2.01 2.04 2.01 2.05 2.01 2.03 2.03 2.03 2.03
subtotals 0.98† 0.99 0.98 1.00 0.97 0.99 0.99 0.99 0.99 0.99
charge 4.07* 4.04 4.09 4.03 4.12 4.03 4.08 4.07 4.06 4.06
subtotals 3.93† 3.96 3.91 3.96 3.87 3.98 3.93 3.93 3.94 3.94
Total cation charge is calculated assuming formal charges: Fe, Mn 13; S 16.
*Na 1 Mg 1 K 1 Ca 1 Ti 1 Mn 1 Fe.
†Al 1 Si 1 S.
Advn Cem Bas Mat Oilwell Cement Clinkers 33
1998;7:28 –38

with charge balancing by coupled substitution of Si, Ti for

further Fe: that is, 2FeMg21(Si,Ti)21. Richardson et al. [13]
suggested that Mn (as a divalent cation) also may substi-
tute for Fe (Mn levels are generally very small). There was
no strong evidence for isomorphous substitution of Mg
for Ca. In our previous paper [8], we showed that the
molar quantities Am and Fmeq fall compactly on the join
between the corresponding Taylor ferrite compositions
[14]. (Am is 21 Al2O3 expressed as mol/100 g clinker; Fmeq
is the sum of the amounts of 21 Fe2O3, MgO, SiO2, TiO2 and
1
2
Mn2O3, also expressed as mol/100 g clinker.) For com-
parison with earlier data, Figure 3 shows a plot of Am
versus Fmeq. The tendency noted previously for composi-
tions to extend to higher Fe contents than that represented
by Taylor’s low Al ferrite is more pronounced in this
dataset. This suggests that for oilwell cement composi-
tions the low Al ferrite (the effective end member of the
ferrite solid solution series) may be somewhere around
FIGURE 2. Distribution of minor oxides between alite and Ca2Al0.65Fe1.35O5. We have also calculated from the spot
belite phases. analyses the indices x, which describes the extent of the
substitution 2Fe Mg21 (Si, Ti)21, and Am/Fmeq, which is
the generalized A/F ratio of the ferrite. The mean values
are given in Table 9, expressed as oxides and in Table 10
for each clinker are given in Table 11. The anomalous
as atoms per formula unit related to the parent struc-
value previously reported [8] for D4 can now be attributed
ture Ca3Al2O6.
to the contribution of C3A to a few spot analyses.

Guest Ions in Ferrites

Substitution patterns in Fe-rich portland cement intersti- Guest Ions in Aluminates
tial material were discussed in detail by Bergstrom et al. The crystal structure refinements of Takéuchi et al. [15]
[8]. The dominant feature is the substitution of Mg for Fe, have established the main substitution patterns in C3A.

TABLE 7. Ferrite oxide compositions

D1 D2 D3 D4 D5 D6 D7 D8 D9 Mean
Na2O 0.06 0.01 0.17 0.17 0.0 0.17 0.05 0.0 0.07 0.06
0.24 0.21 0.29 0.25 0.28 0.34 0.27 0.21 0.25 0.27
MgO 5.19 2.29 2.59 2.52 4.12 4.65 4.86 1.93 1.62 3.29
0.84 0.35 0.34 0.21 0.77 1.24 1.25 0.37 0.41 1.51
Al2O3 16.08 14.11 14.53 18.51 14.62 16.91 15.00 16.98 14.40 15.66
1.04 1.62 1.67 1.44 1.29 1.18 1.59 2.49 2.91 2.25
SiO2 5.75 3.80 5.15 5.17 5.68 5.49 5.83 4.05 3.42 4.90
0.47 0.60 0.64 1.08 0.59 0.76 1.68 2.36 2.22 1.62
SO3 0.90 0.36 0.35 0.20 0.14 0.18 0.13 0.08 0.06 0.29
0.49 0.26 0.21 0.12 0.24 0.33 0.18 0.22 0.21 0.38
Cl 0.04 0.01 0.03 0.07 0.05 0.07 0.03 0.03 0.03 0.03
0.10 0.11 0.10 0.08 0.12 0.11 0.09 0.09 0.07 0.10
K2O 0.18 0.11 0.13 0.13 0.06 0.17 0.21 0.07 0.26 0.15
0.19 0.16 0.13 0.14 0.07 0.13 0.16 0.10 0.44 0.20
CaO 47.92 46.34 47.84 48.14 47.77 47.34 47.90 47.74 46.26 47.49
0.69 0.63 1.20 1.24 0.82 0.73 1.36 1.36 1.48 1.27
TiO2 0.79 0.49 0.40 0.69 0.55 0.34 0.30 0.49 0.35 0.50
0.18 0.18 0.20 0.36 0.24 0.20 0.28 0.25 0.26 0.29
Mn2O3 0.21 0.59 0.63 0.34 0.96 0.29 0.65 1.34 0.83 0.67
0.20 0.30 0.30 0.27 0.29 0.24 0.46 0.55 0.60 0.52
Fe2O3 22.88 31.90 28.17 24.06 26.12 24.38 25.04 27.35 32.68 26.95
0.92 1.09 1.94 2.62 1.26 1.66 2.97 2.30 4.11 3.97
Notes: For each element, the mean and standard deviation of the X-ray spot analyses are expressed as oxide and normalized to 100 mass percent; the last column
gives standard deviations for each element for the entire dataset, not means of the standard deviations for each clinker.
34 C. Hall and K.L. Scrivener Advn Cem Bas Mat
1998;7:28 –38

TABLE 8. Mean molar compositions of ferrites, atoms per 10 oxygen atoms

D1 D2 D3 D4 D5 D6 D7 D8 D9 Mean
Na 0.001 0.001 0.027 0.036 0.0 0.026 0.008 0.0 0.011 0.010
Mg 0.602 0.276 0.309 0.295 0.488 0.543 0.573 0.230 0.197 0.390
Al 1.475 1.344 1.368 1.709 1.368 1.563 1.398 1.601 1.381 1.468
Si 0.447 0.307 0.412 0.404 0.451 0.431 0.460 0.323 0.277 0.390
S 0.052 0.022 0.021 0.012 0.008 0.011 0.007 0.005 0.004 0.016
Cl 0.004 0.001 0.003 0.008 0.005 0.007 0.007 0.005 0.004 0.004
K 0.018 0.011 0.013 0.013 0.006 0.017 0.021 0.007 0.027 0.015
Ca 3.997 4.017 4.099 4.043 4.065 3.978 4.059 4.093 4.041 4.044
Ti 0.046 0.030 0.024 0.041 0.033 0.020 0.018 0.029 0.022 0.029
Mn 0.012 0.037 0.038 0.020 0.058 0.017 0.039 0.082 0.052 0.040
Fe 1.341 1.943 1.696 1.421 1.562 1.440 1.492 1.648 2.010 1.617
Cation 4.03* 4.03 4.14 4.08 4.06 4.02 4.09 4.09 4.08 4.07
Subtotals 3.98† 3.96 3.87 3.90 3.97 4.02 3.99 3.92 3.94 3.95
Charge 8.02* 8.05 8.24 8.13 8.12 8.00 8.15 8.18 8.12 8.11
Subtotals 11.98† 12.00 11.80 11.89 11.93 12.01 11.89 11.89 11.94 11.93
Total cation charge is calculated assuming formal charges: Fe, Mn 13; S 16.
*Na 1 K 1 Ca.
†Mg 1 Al 1 Si 1 S 1 Ti 1 Mn 1 Fe.

The more important of these are the simple charge- duction clinkers all of these substitutions may contrib-
conserving replacements of MgCa21 in octahedral ute to the observed solid solution. However, the re-
sites and FeAl21 in tetrahedral sites on the Al6O18 quirement of overall charge balance must be satisfied.
rings; and the coupled charge-balancing substitutions Our microanalytical data are rather sparse since alumi-
(Na, K)2Ca21h21, where h denotes a normally vacant nate was identified in only two of these clinkers, but
site at the center of the ring; and AlSi21Ca21(Na, they agree for the most part with analyses reported
K)21. The microanalytical data of Han et al. [16] on elsewhere [2,7,17]. The principal guest ions are Fe and
multicomponent synthetic aluminates are consistent Si, with lesser amounts of Mg, Na, and K. Approxi-
with Takéuchi’s crystallographic conclusions. In pro- mately one in eight of the tetrahedral sites is occupied
by Fe and an equal number by Si. Kristmann [6] found
similar high levels of Fe but our Si concentrations are
TABLE 9. Aluminate oxide compositions
greater than reported elsewhere. The amounts of Si are
D4 D8 much higher than the amounts of Na 1 K, so that excess
Na2O 1.58 0.46 Si charge could be balanced by a Ca deficit [2]. How-
0.54 0.33
MgO 1.47 1.12
0.43 0.86
Al2O3 27.75 29.00 TABLE 10. Mean molar composition of the aluminate phase,
2.14 1.06 atoms per six oxygen atoms
SiO2 5.57 4.97 D4 D8 Mean
0.98 1.37
SO3 0.18 0.20 Na 0.139 0.040 0.090
0.27 0.26 Mg 0.099 0.076 0.088
Cl 0.04 0.04 Al 1.482 1.550 1.516
0.11 0.10 Si 0.253 0.225 0.239
K2O 0.74 1.34 S 0.006 0.007 0.006
0.24 1.29 Cl 0.003 0.002 0.003
CaO 55.29 56.94 K 0.043 0.078 0.060
1.56 2.83 Ca 2.685 2.767 2.725
TiO2 0.0 0.0 Ti 0.0 0.0 0.0
0.23 0.13 Mn 0.002 0.004 0.003
Mn2O3 0.04 0.13 Fe 0.255 0.203 0.229
0.17 0.26 cation 2.97* 2.96 2.97
Fe2O3 7.45 5.95 subtotals 1.99† 1.98 1.99
2.91 2.35 charge 5.76* 5.82 5.79
subtotals 6.24† 6.18 6.21
Notes. For each element, the mean and standard deviation of the X-ray spot
analyses are expressed as oxide and normalized to 100 mass percent; the last Total cation charge is calculated assuming formal charges: Fe, Mn 13; S 16.
column gives standard deviations for each element for the entire dataset, not *Na 1 Mg 1 K 1 Ca 1 Mn.
means of the standard deviations for each clinker. †Al 1 Si 1 S 1 Ti 1 Fe.
Advn Cem Bas Mat Oilwell Cement Clinkers 35
1998;7:28 –38

TABLE 12. The more significant correlations between elemen-

tal compositions of individual phases and bulk compositions
(linear correlation coefficients)
Alite Belite Ferrite
Mg 0.97 0.72 0.96
Al ns ns 0.83
Si ns ns ns
S total 0.76 0.91 0.74
S corrected* 0.85 0.95 0.81
K 0.65 ns ns
Fe 0.72 ns ns
ns 5 not highly significant (,0.6).
*S corrected for soluble sulphates (see text).
Summary of regression equations for elements and minerals with r .0.8 (all wt
%):
MgO(alite) 5 0.64MgO(bulk) 2 0.23
MgO(ferrite) 5 2.0MgO(bulk) 1 0.64
Al2O3(ferrite) 5 2.8Al2O3(bulk) 1 5.0
SO3(belite) 5 1.23SO3(bulk, total) 1 0.24
SO3(alite) 5 0.72SO3(bulk, corrected) 1 0.15
SO3(belite) 5 3.36SO3(bulk, corrected) 1 0.55
SO3(ferrite) 5 1.43SO3(bulk, corrected) 1 0.10.
FIGURE 3. Am vs Fmeq, ferrites (spot analyses of all clinkers).
Large open squares mark the Taylor ferrite compositions. for estimating phase compositions when bulk data only
are available. The significant correlations are summa-
rized in Table 12. In calculating these correlations, the
ever, the charges calculated on that basis do not balance raw bulk oxide data have been corrected for free lime
(Table 10). It is possible that there may be a contribution and soluble sulphates. The measured free lime has been
from divalent Fe incorporated into Ca-type sites. This subtracted from the total CaO. The soluble sulphates
remains an open question. We have also noted earlier present have been computed according to the rules
that the aluminate microanalyses may be subject to given by Taylor [14] and the SO3 equivalent to these
some errors arising from contamination by other soluble sulphates has been subtracted from the total
phases. SO3 given by bulk oxide analysis in Table 1. The sum of
the oxides is then normalized to 1.00.
Relation Between Bulk Composition
and Composition of Individual Phases Alite
There is strong positive correlation between alite Mg
We tested the data for correlations between the mean and bulk MgO, and somewhat weaker ones for S, Fe,
compositions of individual phases and the bulk com- and K. For Mg, we find
position for all clinkers and all elements. There are
several strong, statistically significant, positive correla- wt % MgO ~alite! 5 0.64 MgO ~bulk! 2 0.23
tions. These are valuable in providing empirical rules
which may be compared with Taylor’s equation [14]
MgO 5 0.667 MgO (bulk), up to a maximum of 2 wt %.
TABLE 11. Composition indices Am/Fmeq and x for Maki et al. [18] find MgO 5 0.58 MgO (bulk).
ferrite phase We find that the alite S, while low, is significantly
Am/Fmeq x correlated with the bulk SO3 corrected for soluble
sulphates and also with the total bulk SO3 (see Table
D1 0.60 0.226
D2 0.52 0.125 12). Thus,
D3 0.56 0.158
D4 0.79 0.175 wt % SO 3 ~alite! 5 0.72 SO 3
D5 0.53 0.198
D6 0.64 0.205 ~bulk, corrected for estimated soluble sulphates!
D7 0.55 0.211
D8 0.70 0.143 1 0.15.
D9 0.55 0.108
It is known that high bulk clinker SO3 encourages the
Am/Fmeq 5 1 Al2O3/(1 Fe2O3 1 MgO 1 SiO2 1 TiO2 1 1 Mn2O3) and x 5
2 2 2
a/[2(1 1 a)], where a 5 (MgO 1 SiO2 1 TiO2 1 1 Mn2O3)/1 Fe2O3 with all formation of the alite M1 polymorph and so this corre-
2 2
quantities expressed as mol/100 g clinker. lation may have some predictive value [1].
36 C. Hall and K.L. Scrivener Advn Cem Bas Mat
1998;7:28 –38

contents are very low and the mean is close to zero. Mg

substitution in belite is not therefore of any mineralog-
ical significance. The S content, which is variable and
much higher, is well estimated by the regression equa-
tion

wt % SO 3 ~belite! 5 3.36 SO 3
~bulk, corrected for estimated soluble sulphates!
1 0.55.

Alternatively, and just as well,

wt % SO 3 ~belite! 5 1.23 SO 3 bulk, uncorrected

FIGURE 4. MgO (ferrite) vs MgO (bulk). Symbols: E Ghose
and Barnes [5a, 5b]; ■ Bergstrom et al. [8]; å Harrisson et al.
1 0.24.
[2]; Ç Kristmann [6]; 1 Yamaguchi and Takagi [22]; h new
data. Regression line based on new data only is wt % MgO The mean Al and Fe contents of the belite do not
(ferrite) 5 2.0 MgO (bulk) 1 0.64. correlate significantly with the bulk clinker Al and Fe
contents. However, the ratio A/F is highly correlated
(r 5 0.93) with the bulk A/F, so that the factors that
Ghose and Barnes [5a, 5b] found that in clinkers favor the incorporation of these two impurity ions
having widely varying A/F, the alite and belite A/F apparently act similarly on both. This observation
varied correspondingly. For these oilwell clinkers, with agrees qualitatively with the recent data of Fukuda et al.
a much more restricted range of A/F, the alite and [19], although we find that the belite A/F is somewhat
belite A/F cluster around 1.0 and there is no evidence higher than the bulk value, rather than lower. The
of a significant correlation with the bulk A/F for either regression equation is
silicate. The alite Al levels are surprisingly high in view
of the very low bulk Al content. The positive correlation A/F ~belite! 5 1.20 A/F ~bulk! 1 0.03
(Table 3) between Al and Fe contents suggests that the
alite Al level follows the high Fe level established by the for the set of nine clinkers.
high bulk Fe. This may be driven by the charge-balance
requirement. Ferrite
Our new data confirm that the ferrite MgO content
Belite follows closely the bulk MgO. Analyses from several
The only two elements for which correlations between sources are shown together in Figure 4. Our new data
belite guest ion level and bulk concentration are statis- and our previous data are almost indistinguishable and
tically significant are Mg and S. However, the Mg in good agreement with most others. Only Ghose and

TABLE 13. Comparison of amounts of clinker phases calculated from modified Bogue calculation and from X-ray compositions
of individual phases (see text)
Alite Belite Ferrite Aluminate
mB X-ray mB X-ray mB X-ray mB X-ray
D1 69.3 67.0 8.5 15.5 19.2 17.2 0 —
D2 68.7 65.6 15.7 15.2 13.9 13.3 0 —
D3 69.4 65.1 16.4 21.8 12.2 12.9 0 —
D4 70.0 75.1 11.0 4.5 16.9 17.5 0.3 2.7
D5 65.3 65.6 17.1 19.3 14.7 14.6 0 —
D6 67.5 73.5 14.1 10.7 16.3 15.7 0 —
D7 62.3 61.8 20.2 22.8 15.5 15.1 0 —
D8 55.6 52.7 29.0 32.1 13.1 13.4 0 1.5
D9 66.4 68.2 14.7 14.1 17.3 17.4 0 —
The modified Bogue (mB) calculation follows closely that of Taylor [14]. There are minor differences in the assumed MgO and TiO2 contents of the ferrite phase.
Columns labelled X-ray give the results of an unweighted least-squares solution of the equation b 5 Cp (see text).
Advn Cem Bas Mat Oilwell Cement Clinkers 37
1998;7:28 –38

Barnes [5a, 5b] reported somewhat lower ferrite MgO at D8, there is close agreement between the amounts of the
corresponding bulk concentrations. The least squares four clinker phases calculated from the X-ray data and
regression equation based on our new data is that computed by the modified Bogue method. In all cases
the ferrite is well predicted. However, little confidence can
wt % MgO ~ferrite! 5 2.0 MgO ~bulk! 1 0.64. apparently be placed in the computed amounts of alite
and belite. This is attributable to the inherent sensitivity of
The evidence for an upper limit is inconclusive. In the computed alite/belite ratio to the assumed Ca concen-
Figure 4, ferrite MgO and bulk MgO are linearly related tration, particularly in the alite. In the modified Bogue
to the highest concentrations. The highest ferrite MgO calculation, the total Ca (by XRF) is corrected for soluble
concentrations reported here and elsewhere appear to sulphate and free lime. One case where this arises is D4:
be approximately 5 wt %, corresponding to approxi- the bulk analysis shows little SO3 and consequently the Ca
mately 2.3 wt % bulk. This bulk concentration is about adjustment for soluble sulphate is small. However, the
that at which cement microscopists report the appear- X-ray microanalysis does not show especially low SO3.
ance of free periclase [7,20]. In an unpublished study by The inconsistency probably arises from sampling error.
Schlumberger of 76 oilwell cements collected world- The data suggest that several adjustments to the
wide, the bulk MgO content was very variable and model phase compositions and some of the rules for
often above this: the range was 0.52 to 4.51 wt % (mean calculating them may be justified, at least for oilwell
1.23 6 0.87 SD). Unfortunately, there are no materials in cements having low A/F, such as A/F , 1. By compar-
the microanalytical suite with sufficiently high MgO ing the observed clinker mineral compositions and
contents for us to be entirely confident about an upper those assumed in the modified Bogue calculation, we
limit to Mg substitution. conclude that (1) belite MgO is invariably very low and
may be assumed to be zero; (2) alite Al2O3 is in the
Phase Composition range of 1 to 1.5%, rather than 1%, as assumed; (3) belite
Al2O3 is in the range of 1 to 2%, lower than 2.1%, as
Since we have experimental data both on the composi- assumed; (4) belite Fe2O3 is higher (range 1.4 to 2.0%) than
tion of the individual clinker phases and on bulk assumed (0.9%); (5) alite SO3 lies in the range of 0.1 to
composition, it is possible to estimate the phase assem- 0.5%, rather than 0; (6) belite SO3 is much higher than
bly. Thus, for each clinker we have a 3 3 10 composi- assumed, lying between 0.3 and 1.7%; and (7) ferrite SO3
tion matrix C, comprising the mean oxide compositions is also higher than assumed, in the range of 0 to 0.9%.
of each clinker mineral by EDS X-ray microanalysis, Our results suggest that the scheme for estimating the
and a vector b of bulk oxide data by XRF. We seek the sulphate subsystem is not yet satisfactory. Further
solution to the equation b 5 Cp, where p 5 [f1 f2 f3] and study of the distribution of the sulphate ion between
fi are the mass fractions of alite, belite, and ferrite, the various mineral phases is needed. In most cases,
respectively, in the clinker. With 10 elements common SO3 is partitioned between at least five minerals. The
to the X-ray microanalysis and the bulk oxide analysis, rules stated by Taylor [14] and based on the work of
the problem is overdetermined and a least-squares Pollitt and Brown [21] apparently underestimate the
solution can be obtained. Both the bulk oxide analysis amount of SO3 in the silicate and ferrite minerals, at
and the X-ray phase analyses are individually accurate least in this class of cement. However, little is known
and the main source of error lies in the sampling about how the minor sulphate minerals such as aphthi-
statistics, since the X-ray microanalytical data are not talite, calcium langbeinite, and anhydrite are physically
necessarily representative of the bulk sample. distributed in the clinker microstructure. Until it becomes
For two clinkers, D4 and D8, SEM backscattered clearer where and in what form these minerals crystallize,
images showed clear evidence of a separate aluminate we cannot exclude the possibility that EDS microanalyses
phase in the interstitial material. To refine the calcula- of some of the major clinker phases contain contributions
tion of the phase assembly for these two clinkers only, from these minor sulphate minerals. For example, the
the mean X-ray aluminate compositions were included high S contents of the belites, which we find could arise
as a fourth component mineral. from intimately mixed sulphate minerals, dispersed per-
haps in the interlamellar boundaries.
Modified Bogue Calculation
These estimated phase assemblies can be compared with
the results of calculations made directly from the oxide
Conclusions
data alone, using the modified Bogue method described From our observations, we conclude that (1) silicate min-
by Taylor [14]. The results are given in Table 13 (see eral compositions in oilwell cements are generally similar
footnote for minor differences between our calculation to those of other portland cements; (2) aluminate is
method and that of Taylor). In four cases, D3, D5, D7, and usually absent or present only in very small quantities; (3)
38 C. Hall and K.L. Scrivener Advn Cem Bas Mat
1998;7:28 –38

ferrite is very variable and is heavily substituted; and (4) 8. Bergstrom, T.B.; Hall, C.; Scrivener, K.L. Adv. Cem. Res.
refined phase calculation based on Taylor’s modified 1992, 4, 141–147.
Bogue method, while generally a good estimator of the 9. Skibsted, J.; Jakobsen, H.J.; Hall, C. Journal of the Chemical
Society: Faraday Transactions 1994, 90, 2095–2098.
phase composition, does not reliably predict alite/belite
10. Midgley, H.G.; Fletcher, K.E. Trans. Br. Ceram. Soc. 1963,
ratios. It can probably be improved appreciably by minor
62, 917–937.
changes to rules for the composition of individual phases. 11. Hahn, T.; Eysel, W.; Woermann, E. Proceedings of the 5th
International Symposium on the Chemistry of Cement; Tokyo,
1968, Part I, Vol. 1, 61– 66.
Acknowledgment 12. Gies, A.; Knofel, D. Cem. Concr. Res. 1987, 17, 317–328.
We thank several cement manufacturers for the supply of production 13. Richardson, I.G.; Hall, C.; Groves, G.W. Adv. Cem. Res.
materials and chemical analyses. The X-ray microanalytical data were 1993, 5, 15–21.
obtained by Stephen Laing. 14. Taylor, H.F.W. Adv. Cem. Res. 1989, 2, 73–77.
15. Takéuchi, Y.; Nishi, F.; Maki, I. Zeitschrift für Kristallogra-
phie 1980, 152, 259 –307.
References 16. Han, K.S.; Gard, J.A.; Glasser, F.P. Cem. Concr. Res. 1981,
1. Taylor, H.F.W. Cement Chemistry; Academic Press: Lon- 11, 79 – 84.
don, 1990. 17. Fletcher, K.E. Mag. Concr. Res. 1969, 21, 3–14.
2. Harrisson, A.M.; Taylor, H.F.W.; Winter, N.B. Cem. Concr. 18. Maki, I.; Fukuda, K.; Yoshida, H.; Kumaki, J. J. Am. Ceram.
Res. 1985, 15, 775–780. Soc. 1992, 75, 3163–3165.
3. Moranville-Regourd, M.; Boikova, A.I. Ninth International 19. Fukuda, K.; Maki, I.; Ito, S.; Yoshida, H.; Kato, C. J. Am.
Congress on the Chemistry of Cement; New Delhi, 1992, Vol. Ceram. Soc. 1994, 77, 3027–3029.
1, pp 3– 45. 20. Campbell, D.H. Microscopical Examination and Interpreta-
4. Ghose, A. PhD thesis, University of London, 1980. tion of Portland Cement Clinker; Construction Technology
5a.Ghose, A.; Barnes, P. World Cement Technology 1980, 11, Laboratories: Skokie, IL, 1986.
441– 443. 21. Pollitt, H.W.W.; Brown, A.W. Fifth International Sympo-
5b.Ghose, A.; Barnes, P. Cem. Concr. Res. 1979, 9, 745–755. sium on the Chemistry of Cement; Tokyo, 1968, Vol. 1, pp
6. Kristmann, M. Cem. Concr. Res. 1978, 8, 93–102. 322–333.
7. Odler, I.; Abdul-Maula, S.; Nüdling, P.; Richter, T. Zement- 22. Yamaguchi, G.; Takagi, S. Fifth International Symposium on
Kalk-Gips 1981, 34, 445– 449. the Chemistry of Cement; Tokyo 1968, pp 181–225.