U D C 666.942.015.42: 621.385.

833

The analysis of belite in Portland cement

clinker by means of an electron-probe
micro analyser *
by K. E. Fletcher
BUILDING RESEARCH STATION

SUMMARY
It is believed that the rate of hydration of dicalcium
silicate in a Portland cement clinker will depend upon
the type and amount of stabilizing agent present in the
crystal lattice. Complete analyses are given for a belife
grain in a normal British Portland cement clinker and
in a sulphate-resisting Portland cement clinker. Lattice
substitutions are suggested which are consistent with the
analyses obtained. Belife grains in the normal and
sulphate-resisting Portland cement clinkers d!Uered in
the relative proportions of Al and Fe present. The ratio
of Al to Fe in the be/ife grains corresponded closely to
the ratio of Al to Fe in the cement clinker. Difficulties
which arise in the electron-probe microanalysis are
discussed.

Introduction
Dica1cium silicate is a major phase in Portland
cements. It can occur in various polymorphic forms,
depending partly upon the minor elements present in
the lattice and partly upon the thermal history of the
sample(l). t Its normal occurrence in cements is predominantly as the beta modification, although the
presence of some alpha prime, or even alpha dica1cium
silicate is possible (2). It is believed that the rate of
hydration of the dica1cium silicate present in a Portland
cement clinker will depend upon the particular minor
elements which are present in the structure. Attempts
have been made to separate dica1cium silicate from
cement clinkers in order to carry out a chemical
analysis, and one analysis (3) has been reported for beta
dica1cium silicate separated by differential hydration
of a clinker. This was, however, a rather unusual type

* Crown copyright.
t The term 'belite' indicates the dicalcium silicate phase normally
obs7rved as ~ounded grains by reflected-light microscopy in a
pohshed sectIOn of a Portland cement clinker.

of clinker and it also seemed likely that a gel was

present on the surface of the dicalcium silicate grains.
The electron-probe microanalyser makes it possible
to obtain a direct analysis of belite grains in a Portland
cement clinker. Various workers have examined Portland cement clinkers by means of electron-probe
microanalysis, (4-7) using different instruments, and the
limitations of the method appear quite clearly from
these accounts. It is relatively easy to obtain distribution pictures for Ca, Si, AI, Fe, K, Mn and Ti. Such
pictures show clearly the presence of the silicate phases
and interstitial material. They also show K to be concentrated in the dica1cium silicate phase, and Ti and
Mn to be concentrated in the interstitial material.
When differentiation of the interstitial material is
possible, Ti and Mn can be shown to be concentrated
III the ferrite phase. In some cases, K ZS0 4 has been
shown to be present.
It is much more difficult to carry out quantitative
analysis. The main reason is that, at a reasonable
excitation voltage of 20 keV, a probe of I fLm diameter
has been found to cause excitation in a surface region
of about 5 fLm diameter and to a depth of about 15 fLm
in the sample. Such analyses as are obtained will not
even represent the average composition of a particular
volume of the sample, because the depth in the sample
from which excitation occurs will depend upon the
excitation voltage for the particular spectral line which
is being measured. Thus SiKG(t with an excitation
voltage of 1740 keV will be excited at a greater depth
in the sample than CaKG(l with an excitation voltage
of 3691 keV.
Peterson has presented uncorrected values for Al z0 3,
KzO, Na 2 0 and S03 in alite and belite grains in
Portland cements, by the use of standards such as
Ca 4AI 2 Fe z0 1 (), Ca 3Al z0 6 and K 2S0 4 But the ratio he
obtained for Si in belite to Si in alite varied from 113
to 130. If an accurate analysis is to be obtained with167

Magazine of Concrete Research: Vol. 20, No. 64 : September 1968

out resort to rather uncertain correction procedures,
then it is necessary to have a number of standards of
which the compositions and homogeneity are well
established, the compositions being close to those of
the sample being analysed.
A private communication from K. T. Ball has indicated that, although individual alite and belite grains
are reasonably homogeneous in minor element composition, differences certainly exist between different
grains in the same polished section. For this reason,
an analysis which will be useful for indicating possible
substitutions in the lattice must be carried out on a
single grain. Midgley(8) has recently analysed a single
grain of alite in the sulphate-resisting Portland cement
clinker used for this investigation. Alite exists as fairly
large crystals from 50 to 100 fim in diameter and, by
the judicious selection of fairly large crystals, the
possibility that the beam will excite elements outside
the grain boundary can be avoided.
Belite exists as rather smaller crystals from 25 to
50 fim in diameter, and the problem of penetration is
more acute. In the present work, reasonably accurate
analyses have been obtained by selecting the larger
grains and by carrying out a preliminary analysis for
Si with exciting voltages of both 20 and 10 keY; penetration should be very much less at 10 keY. Only when
results from both exciting voltages agreed was a complete analysis attempted.

Experimental data
A Cambridge Instrument Co. Geoscan electronprobe microanalyser was used. This instrument has a
75 take-off angle which, while minimizing absorption
of emergent characteristic X-rays, enables the spectrometer to receive X-rays generated well below the
surface, thus accentuating penetration problems.
There was a choice of three analysing crystals: mica,
quartz and lithium fluoride. The analysis was invariably carried out with a flow proportional counter.
Two commercial clinkers were used, an ordinary
Portland cement and a sulphate-resisting Portland
cement. Polished surfaces were prepared by the
methods given by Midgley and Taylor(9), final polishing being by 1 fim diamond paste in oil. The sulphateresisting clinker was found to be electrically conducting, but the ordinary Portland cement and the standards had to be coated with a 100 A layer of carbon.
A preliminary investigation of belite grains in these
cements indicated the presence of Ca, Si, K, Na, Mg,
Fe, AI, Ti and Mn. Neither S nor P could be detected,
although both have been detected in the belite phase
in some other, rather unusual clinkers. The standards
used are given in Table I.
Complete analysis of a belite grain was attempted
only when the determinations of Si at accelerating
voltages of 20 and 10 keY had given good agreement.
This minimized the risk of selecting a grain where
excitation was occurring outside the grain boundary.
168

TAB LEI:

Standards used for electron-probe

microanalysis.
Standard

Elements
Ca, Si

beta dicalcium silicate with the

composition: 6S-4% CaO; 34'8% Si0 2 ;
0'06% B20 3

K,Al, Mg, Mn

beta dicalcium silicate containing

02S% K 20, 018% Al,03, 091 % MgO,
0-43% MnO,

Al F
, e

tricalcium silicate containing

l'OS % Fe,03, 0'8S % AI,03

Ti

tricalcium silicate containing

031 % TiO,

Na

tricalcium aluminate containing

3'0% Na 20

------------------------~----~---------

TAB L E 2:

The analyses of belite grains by

electron-probe microanalysis.
Normal
Sulphate-resisting
9S%
Portland cement Portland cement confidence
(% by weight)
(% by weight)
limits

CaO
SiO,
K 20
Na,O
MgO
Fe,03
Al,03
TiO,
Mn,03
P,Os
S03
Al,O,,/Fe 20 3

632
321
03
03
04
08
17
03
003
< O'IS
< 010
212

629
33-2
O'S
03
02
I'S
11
02
006
< O'IS
< 010
073

O'S
O'S
O'OS
O'IO
O'OS
O'IO
O'IO
OOS
0'03

The other elements were then estimated using an accelerating voltage of 20 keY and a constant specimen
current of about 100 millimicroamperes. In attempting
a multi-element analysis of a single grain, it is difficult
to obtain enough results to give a reliable standard
deviation for each element. The total area which can
be used is limited and surface contamination builds up
during the analysis. Therefore, the results quoted are
the means of at least four determinations, but the
standard deviations are estimated from similar observations made on other dicalcium silicate grains.

Results
The analyses of belite obtained by electron-probe
microanalysis are given in Table 2. The elemental
analyses have been expressed in the conventional manner as the appropriate oxides. The ratio of AI 20 3 to
Fe 2 0 3 in the belite grains corresponds closely with the
ratio of Al 20 3 to Fe 20 3 in the cement clinker. The
ratios were 212 and 190 respectively for the ordinary
Portland cement clinker and 073 and 0'67 respectively
for the sulphate-resisting cement. This could be expected because the belite grains formed in a liquid of
which the AI/Fe ratio was very close to that of the

The analysis of belite in Portland cement clinker by means of an electron-probe microanalyser

original cement. It does indicate that the dicalcium
silicate lattice shows little selectivity for AI3 + or Fe 3+.
In Table 3 the conventional chemical analyses of the
Portland cement clinkers are given.
TAB L E 3: Chemical analysis of the Portland
cement clinkers used for this investigation.
Normal
Portland cement

Sulphate-resisting
Portland cement

6575
2356
0'21
051
096
255
484
024

6408
2306
040
028
088
633
423
028

020
019
190

024
024
067

CaO
SiO z
KzO
NazO
MgO
FezO,
AlzO,
TiO z
MnzO,
P2 0 S
SO,
AlzO,/FezO,

TAB L E 4: Electron-probe analyses of belite

grains expressed as molar ratios based on a beta
dicalcium silicate unit cell of Ca sSi 4 0 w
Molar ratios
Element

Normal
Portland cement

Ca
K
Na
Mg
Total
Si
Fe
AI
Ti
Mn
Total

788
0'04
007
007
806

-----

SuI phate-resist ing

Portland cement
777
007
006
003
793
-- - - - - -

374
007
023
003
0003

-._----- - - -

383
013
015
002
0005

407

413

1600

16'00

In Table 4 the electron-probe belite analyses have

been expressed as molar ratios based on the conventional unit cell(lO) of beta dicalcium silicate containing
16 oxygen atoms, that is CaSSi 40 16 The elements have
been placed in two groups, those likely to occupy
calcium sites and those likely to occupy tetrahedral
silicon sites. (11-17)
The results in Table 4 show that this grouping
corresponds reasonably well with the analyses obtained. Because of the accumulated errors of the
analysis, a more detailed conclusion is not attempted.
Some reservations must, however, be made. If it is
assumed that the basic oxygen network is unchanged,
then, in order to preserve over-all electrical neutrality,
some of these elements must also occupy interstitial
sites in the structure. The amount of FeZ t detected in
a Portland cement clinker is normally very low, but
this depends upon the composition and thermal history
of the clinker and there is no direct evidence about the
iron present in the dicalcium silicate. Any iron which
is present as Fe z + will be expected to replace Ca z +.

Conclusions
Detailed analyses have been given for belite in a
normal Portland cement clinker and in a sulphateresisting Portland cement clinker. The minor elements
which were found to be present were K, Na, Mg, Fe,
AI, Ti and a trace of Mn. The main difference between
belite in the normal and in the sulphate-resisting
cements was the ratio of Al to Fe present. This was
found to correspond closely to the ratio of Al to Fe
in the cement clinker.
The results suggest that K, Na and Mg mainly
occupy Ca sites and that Fe, AI, Ti and Mn mainly
occupy Si sites. The accumulated errors of the analysis
preclude a more detailed allocation. Sand P, which
have been detected in some special clinkers, could not
be found in the belite phase in these particular clinkers.

* The amount of 0 has been calculated by making the assumption common to cement chemistry that the elements can be
represented as their appropriate oxides, as given in Table 2.

ACKNOWLEDGEMENT
This paper deals with work forming part of the proKramme of the Building Research Station, and is
published by permission of the Director.

Lattice substitutions

REFERENCES

---~--

0*

In beta dicalcium silicate, solid solution can occur

by direct replacement of Ca or Si and by the occupation of interstices in the structure.
There are many alternative ways in which the elements which have been found can be fitted into the
dicalcium silicate structure and the assumption that
there is some distortion of the structure is necessary.
This assumption is probably justified since the total
number of cations entering the structure is smallabout I atom for every 24 calcium and silicon atoms.
The distortion can be accommodated by a slight rearrangement over a large number of neighbouring
groups without affecting the basic structure.

I.

2.

3.
4.

5.

NURSE, R. w. The dicalcium silicate phase. Proceedings of

the Third International Symposium on the Chemistry of
Cement, London, 1952. London, Cement and Concrete
Association, 1954. pp. 56-77.
YAMAGUCHI, G., KAWAMURA, S. and SODA, Y. Strength of
mortars made from the modifications of Ca 2 Si0 4 Review
of the Seventeenth General Meeting of the Japan Cement
Engineering Association, Tokyo, May 1963. pp. 39--43.
MIDGLEY, H. G. Work referred to on page 75 of reference I.
MOORE, A. E. Examination of Portland cement clinker by
electron probe micro-analysis. Silicates Industriels. Vol.
30, No.8. August 1965. pp. 445-450.
PETERSON, O. Untersuchung von Portland-Klinker mit der
Mikrosonde. (Investigation of Portland cement clinker with
the micro-probe.) Zement-Kalk-Gips. Vol. 20, No.2.
February 1967. pp. 61-64.

169

Magazine of Concrete Research: Vol. 20, No. 64 : September 1968

6.

7.

8.

9.

10.
11.

170

and REBUFFAT, P. Microanalisi

del clinker di portland mediante sonda elettronica. (Electron probe microanalysis of Portland cement clinker.) La
Ricerca Scientijica. Vol. 36, No.7. 1966. pp. 567-572.
YAMAGUCHI, G., TAKAGI, S., YOSHIZUMI, M. and OHTA, T.
Estimation on the phase composition and fine structure of
interstitial phase in Portland cement clinkers by the
electron-microprobe technique. Review of the Twentieth
General Meeting, Technical Session of the Cement Engineering Association, Tokyo, May 1966. pp. 39-41.
MIDGLEY, H. G. The composition of alite (tricalcium silicate)
in a Portland cement clinker. Magazine of Concrete
Research. Vol. 20, No. 62. March 1968. pp. 41-44.
MIDGLEY, H. G. and TAYLOR, H. F. w. Optical microscopy.
The chemistry of cements (ed. H. F. W. TAYLOR). London,
Academic Press, 1964. Vol. 2, Chapter 20. pp. 223-247.
MIDGLEY, C. M. The crystal structure of dicaIcium silicate.
Acta Crystallographica. Vol. 5. 1952. p. 307.
LOCHER, F. w. Hydraulic properties and hydration of glasses
of the system CaO-AI 2 0 3-SiO z' Chemistry of Cement. Proceedings of the Fourth International Symposium, Washington,
1960. NBS Monograph 43, Vol. 1. pp. 267-275.
AMICARELLJ, Y., SERSALE, R.

12.

13.

14.

15.

16.
17.

and FLETCHER, K. E. The role of alumina and

magnesia in the polymorphism of tricaIcium silicate.
Transactions of the British Ceramic Society. Vol. 62, No. 11.
1963. pp. 917.
WOERMANN, E., HAHN, T. H. and EYSEL, w. Chemical and
structural studies of the solid solutions of tricalcium silicate. Zement-Kalk-Gips. Vol. 16, No.9. 1963. pp. 370-375.
NURSE, R. w. The effect of phosphate on the constitution
and hardening of Portland cement. Journal of Applied
Chemistry. Vol. 2, No. 12. 1952. pp. 708.
FLETCHER, K. E., MIDGLEY, H. G. and MOORE, A. E. Data on
the binary system 3CaO.AI2 0 r Na 20.8CaO.3AI 20 within
the system CaO-AI203-Na20. Magazine of Concrete
Research. Vol. 17, No. 53. December 1965. pp. 171-176.
BRAGG, L. and CLARINGBULL, G. F. Crystal structures of
minerals. London, O. Bell and Sons, 1965. pp. 409.
FLETCHER, K. E. The effect of Fe 3 + and ALl + on the polymorphism of tricaIcium silicate. Transactions of the British
Ceramic Society. Vol. 64, No.8. 1965. pp. 377-385.
MIDGLEY, H. G.

Contributions discussing the above paper should be in the hands

of the Editor not later than 31 December 1968.