CEMENTand CONCRETERESEARCH. Vol. 2, pp. 213-223, 1972. PergamonPress, Inc.

Printed in the United States.

THE PASTEHYDRATIONOF 4CaO'3AI203.SO3 IN PRESENCEOF CALCIUM

SULPHATE, TRICALCIUMSILICATEAND DICALCIUMSILICATE
Mario Collepardi and Renato Turriziani
Istltuto dl Chimica Appllcata ed Industrlale,
Facolta Ingegneria-University of Rome
and
Aldo Marcialis
Istltuto dl Chimica Applicata • di Metallurgia
Facolta Ingegneria-University of Cagliari

(Communicated by P. J. Sereda)

ABSTRACT
Stoichiometry and kinetics of C4A3Spaste hydration in the presence
of CaSO4, CaSO4.2H20, C3S and of C2S have been studied. The stoichio-
metry depends on the type and on the amount of calcium s i l i c a t e , as
different quantities of calcium hydroxide are produced in the hydra-
tion of C3S or C2S. The type of calcium sulphate has no influence
on the stoichiometry of C4A3S hydration. The type of calcium sulphate
and calcium silicate do not affect the hydration rate of C4A3S. The
high rate of hydration of C4A3Sprecludes any practical utilization
of the expansive properties of a cement containing C4A3S. However,
the hydration rate can be strongly decreased by addition of organic
admixtures.

On a (tudl~ la stoechl~trle et la cin~tique d'hydratatlon en p&te

du C4A3S en pr(sence de C3S ou de C2S et de CaSO 4 ou de CaSO4.2H20.
La stoechio~trle de la r~actlon ne d~pend que de la nature et du
pourcentage de silicate, ~ cause de la dlff~rente quantlt~ de Ca(OH) 2
qui se form par hydratatlon de C3S ou de C2S. Le type de sulfate
de calcium ne influence pas la stoechl~trle d'hydratatlon du C4A3S.
La vitesse d'hydratation du C4A3S ne d~pend pas du type de sulfate
et de silicate de calcium. L'hydratatlon du C4A3S a lieu trop
rapidement pour qu'on pulsse utlliser pretlquement les proprl~t~s
expanslves d'un clment qul contlent du C4A3S. Toutefols on peut
fortement ~duire la vltesse de ~actlon par l'emploi d'additlfs
organiques. 213
214 Vol. 2, No. 2
CALCIUM SULFOALUMINATE, CALCIUMSILICATE, HYDRATION

Introduction
The anhydrous calcium sulphoaluminate C4A)~* can be used mixed with port-
land cement for the production of expansive cement.
The dimensional increase of a concrete made with this cement is linked
with the formation of hydrated trisulphoaluminate C3A.3CS.32H20 by hydration of
the C4A3S. Potentially, these expansive properties can be used either to Pro-
duce autocompressed reinforced concrete or to eliminate, or at least l i m i t , the
inconvenience caused by hydraulic shrinkage which shows i t s e l f after the molds
were removed. Whatever use is made of i t , i t is opportune to check not only
the nature of the hydrated product, but also the hydration speed of the anhyd-
rous sulphoaluminate.
In a previous study, l the influence of the addition of Ca(OH)2 and of
CaSO4.2H20 to C4A3S on the nature of the hydrated products obtained with varying
water/solid ratios was examined. In particular, in the hydration of paste, which
from a practical point of view is more important, i t was observed that for the
formation of hydrated trisulphoaluminate, i t is indispensable to add a certain
quantity of calcium sulphate. However, although the CaSO4/C4A3Smolar ratio
was such that, from a stoichiometric point of view, the anhydrous sulphoalumin-
ate should have been converted completely into ettringite, monosulphoaluminate
C3A.CS.12H20 was found in certain cases together with the hydrated trisulpho-
aluminate. I t is worth noting also that in the presence of Ca(OH)2 the hydrat-
ion speed of C4A~
_ was greater than that observed in the presence of CaSO4'2H20,
while an intermediate behaviour was revealed when lime and gypsumwere added
simultaneously to the anhydrous sulphoaluminate.
On the basis of these results and bearing in mind the different speeds with
which the Ca(OH)2 is liberated by the hydrolysis of C3S and C2S, i t was inter-
esting to examine, in the present study, the influence of the type of silicate
contained in the clinker of portland cement on the hydration of C4A3S. The pos-
s i b i l i t y of regulating the hydration process in anhydrous aulphoaluminate paste
was taken into consideration, either by varying the ratio C4A3S/silicate, o r by
changing the type of calcium sulphate present, or f i n a l l y by adding certain ad-
ditives of organic nature.
Experimental Procedure
A. Materials
The C4A3~ and the C3S were prepared according to the methods previously
described2, 3. The B-C2S was obtained by repeatedly heating the CaCO3 and SiO2

C = CaO; A = A1203; S = S03; S = SiO2

Vol. 2, No. 2 215
CALCIUM SULFOALUMINATE, CALCIUMSILICATE, HYDRATION

at 1600°C, using B203 (0.5%) as a stabiliser. The CaSO4 was obtained by the
dehydration of Ca$O4.2H20 at gO0°C. The Ca(OH)2 was prepared through the hyd-
ration of the CaO obtained by heating the CaCO3 at gSO°C.
The above mentioned materials were mixed by grinding them in an agate mill
for 2 hours, using CCI4 as a dispersing agent which was later eliminated by
evacuation. The percentage compositions of the mixtures are set out on Table I.

Tabl e 1
Composition on a percentage basis of the anhydrous samples subjected to hydrat-
ion. Specific surface = 4000 cm2/g.
Sample C3S C2S C4A3S CaSO4.2H20 CaSO4 Ca(OH)
2 CaF2
N°
1 50.9 15.1 34.0
2 50.9 15.1 26.9 7.1
3 50.9 15.1 34.0
4 50.9 15.1 26.9 7.1
5 25.1 56.6 18.3
6 25.1 44.7 18.3 11.9
7 ]8.3 25.] 56.6

In all the samples, the CaSO4/C4A3S molar ratio was ftxed equal to 8. The
CaSO4.ZH20 present in the mixtures nos. 1-3-5 was substituted with an equal num-
ber of moles of CaSO4 respectively in the mixtures nos. 2-4-6; in order not to
modify the percentage composition of the other components, a given weight of
inert material (CaF2) was added equal to the difference between the weight of
CaSO4.2H20 and that of CaSO4.

B. Procedure of hydration and methods of analysts.

2g. of each sample were mixed for 3 minutes with 2 ml of water. The pastes
were kept in polyethylene containers and thermostated at 25°c+0.1°C.
After a preestablished period of time, the hydration reaction was stopped
according to the methods described in a previous study. 3 Then the determination
of the heat loss was carried out on a part of each sample, whereas the remaining
part was treated with a solutton saturated with Ca(OH)2 and containing ]00 g/1
of NaNO3 in order to extract the calctum sulphate not reacted. 4 The weight ratio
between the solution and the sample was selected equal to 1000, to ensure the
extraction of all the calcium sulphate not reacted.
With preliminary tests i t was established that in order to completely ex-
tract the anhydrite, i t was necessa~ to agitate the suspension for 5 minutes
216 Vol. 2, No. 2
CALCIUM SULFOALUMINATE, CALCIUMSILICATE, HYDRATION

and then leave i t to stand for at least 2 hours. Consequently, in order to

standardize the experimental methods, this treatment was adopted even for samp-
les containing dihydrate (gypsum) even though the extraction of this compound
can be completed in less time.
After f i l t e r i n g the suspension in a C02-free atmosphere, the amount of cal-
cium sulphate extracted was determined by nephelometric methods, whereas the
solid residue was repeatedly washed with lO0 ml absolute ethyl alcohol per l g
of the sample and then evacuated with a rotary pump. The calcium sulphate that
passes in the lime saturated solution is attributed to the CaSO4.2H2O or to the
--m

CaSO4 not reacted, since the concentration of the ions SO4 _in the aqueous
solution extracted is almost nil when i t is treated with C4A3S, with ettringite,
or hydrated monosulphoaluminate.
On the solid sample, minus CaSO4.2H20 or CaSO4, the percentage OfoC4A3S
o

(3.77A) was determined by means of X-ray analysis, using Mg(OH)2 (3.38A)as an

internal standard. The absence of peaks at 7.56A and at 3.50A, of the gypsum
and anhydrite respectively, confirmed that the extraction of the calcium sul-
phate had completely taken place and, at the same time, guaranteed that the peak
o

of the gypsum at 3.79A does not superimpose i t s e l f on that of the C4A3~ chosen
for the quantitative X-ray analysis.
Both the percentage values of CaSO4.2H20 not combined and those of C4A3S
are calculated in relation to the calcinated sample, having corrected the results
referred to the dried out samples with the value of the weight loss.
The determination of the c i t r i c acid used as a set retarder additive was
obtained by using the spectrophotometer UNICAM (mod. SP 800 for the visible and
ultraviolet) measuring the absorption of the aqueous solution by the wave length
of 420 millimicrons. 5

Results and Discussion

In Fig. l are set out at different hydration times, the percentage values
of the non-reacted calcium sulphate (referred to as CaSO4"2H20) present in the
hydrated mixtures from no.s I-4. These results show that, in the period of time
used for the experiment, the rate of the reaction of the calcium sulphate, while
not depending appreciably on the thermic treatment undergone by the dihydrate
(gypsum) varies considerably with the type of calcium silicate added to the an-
hydrous sulphoaluminate. In the presence of C3S, the combination speed of the
calcium sulphate is much more rapid; between 3 and 72 hours of reaction the
CaSO4.2H2O passes from 16.4% to 12.5% in the mixture containing the dicalcium
silicate, and from 12.5% to 0.0% in the presence of tricalcium silicate. On the
other hand, from the data set out on Fig. 2 one deduces that the hydration rate
Vol. 2, No. 2 217
CALCIUM SULFOALUMINATE,CALCIUMSILICATE, HYDRATION

• SAMPLE Ne. 1
• ,, i, 2
e • 3
N O • • &

so

m
ii
u
,7 l0

! ~ ~ |
10 100
TIME OF HYDRATION (hr)

FIG. 1
Percentage of gypsum vs. time.
M .

O SAMPLE Ne. 1
x • • 3

0 II

TIME OF HYDRATION ( h r )
FIG. 2
Percentage of C4A3S vs. time.
of C4A3S does not depend substantially on the type of stltcate present. The d i f -
ferent Influence of the type of stltcate on the reaction rate of the calcium
sulphate and of the tetracalctum sulphoalumtnate must be ascrlbed to the d i f f e r -
ent stotchtometry of the reactton as the results indtcate on Table 2. The wetght
ratio between the quantity of C4A3S and the CaS04.2H20 combined between 3 and 72
218 Vol. 2, No. 2
CALCIUM SULFOALUMINATE, CALCIUMSILICATE, HYDRATION

Table 2
C4A3] and CaSO4.2H20 contents unreacted after 3 and 72 hrs of hydration
expressed as percentages of the calcined weight of the hydrated sample.
The figures appearing in the lowest row are contained by the two pre-
ceding rows.

Time of Sample N° l Sample N° 3

hydration C4A3S C4A3S
(hr) %C4A3S %CaSO4.2H20 CaSO4.2H20 %C4A3S %CaSO4.2H20 CaSO4.2H20

3 5.7 12.5 6.2 16.4

72 0.0 0.0 0.0 12.5
5.7 12.5 0.46 6.2 3.9 l .59
M.W. C4A3S/M.W. 8 (CaSO4.2H20) = 0.44
M.W. C4S3S/M.W. 2 (CaSO4.2H2O) = 1.77

hours is 0.46 in the pastes containing C3S, and 1.59 in those containing C2S.
The above mentioned r a t i o for the tricalcium s i l i c a t e pastes is thus more or
less coincident with theoretical one of 0.44 deducible from the following react-
ion:
C4A3S + 6Ca(OH)2 + 8(CaSO4.2H20) + 74H20 = 3(C3A'3CS'32H20) (1)
I f , instead, the hydration of the sulphoaluminate occurs in the presence
of the dicalcium silicate, the ratio (I.59) is contained between the value of
0.44, corresponding to the reaction (1), and 1.77 relative to reaction (2), but
much nearer the following:
C4A3~ + 2(CaSO4.2H20) + (28+2x)H20 = C3A.3CS.32H20 + 2(A1203.xH20) (2)

The different stoichiometric behaviour of C4A3S, when hydrated with C3S and
with C2S is attributable to the different quantities of lime present in the two
cases, because of the different hydrolysis rate of C3S compared with that of C2S.
The quantity of lime obtained by hydration of C3S is such that C4A3~ reacts
completely according to the scheme of reaction (1), whereas the quantity of lime
produced by hydrolysis of C2S is so small that the C4A3S reacts mostly according
to the scheme of reaction (2) and only in minor degree according to the reaction
(1).
By means of X-ray analysis of the s t i l l moist samples, i t was observed that
besides the ettringite a small quantity of hydrated monosulphoaluminate was also
present, although in same cases the calcium sulphate s t i l l exists as CaSO4.2H2O.
Vol. 2, No. 2 219
CALCIUM SULFOALUMINATE, CALCIUMSILICATE, HYDRATION

This result, in accordance with what has been set out in a previous study, 1 can
be explained i f one considers that during the paste hydration, i t is not possi-
ble to achieve a real chemical equilibrium.
From a practical point of view i t is interesting to emphasize that in the
pastes containing C3S and thus very rich in Ca(OH)2, the quantity of e t t r i n g i t e
which is formed per unit of weight of C4A3S, is noticeably greater than that ob-
tained in the presence of C2S. However, the results set out on Table 2 show
that in both cases the hydration rates of the hydrated sulphoaluminate is such
that i t is not possible to regulate the rate ofexpansion of the concrete w i t h
a view to compensating for the contractions caused by the hydraulic shrinkage,
since the speed of the reaction is too rapid and in th e space of 3 days the hy-
dration of the C4A3S is more or less complete.
In order to slow down the process of hydration of the anhydrous sulpho-
aluminate, certain tests were carried out to estimate the retardant effect of a
number of organic additives.
The mixtures nos. S and 6, subjected to hydration for 6 hours, with or
without additives contain: calcium hydroxide, anhydrous sulphoaluminate and
gypsum or anhydrite mixed according to the molar ratios deduced from the stotch-
iometery of reaction {1). Since in these samples the calcium silicates are not
present, i t is possible to value d i r e c t l y the influence of the additives used
on the only possible hydration process, which is that regarding the tetracalcium
sulphoaluminate.
The values of the percentages of gypsum and C4A3S not combined after var-
ious hydration times {Table 3) show that all the additives used slow down the
hydration speed considerably, whatever type of calcium sulphate is used. The
most efficacious in slowing down the rate of the reaction were mallc acid and
citric acid. The latter was used as a set retarder additive in the hydration of
misture no. 7 in which the percentages of anhydrous suophoaluminate and gypsum
are equal to that of mixture no. 5, while the lime was completely substituted by
tricalcium silicate.
The results of these tests are set out in Fig. 3, where the contents of
the unreacted calcium sulphate are shown as function of time. It was noted that,
in accordance with the data on Table 3, the citric acid has a retarder effect on
the reaction rate of the calcium sulphate even in the presence of the tricalcium
silicate. It is worth noting that in the absence of additives the percentage of
CaSO4.2H20 not combined between 6 and 340 hours decreases linearly with the time
log. Instead, when c i t r i c acid is present, the curve that shows the percentage
values versus the time logarithm reveals a rising rate with the hydration process.
220 Vol. 2, No. 2
CALCIUM SULFOALUMINATE, CALCIUMSILICATE, HYDRATION

Table 3
Percentages of calcium sulphate (expressed as CaSO4.2H20) and of C4A3S unreacted
after 6 hours of hydration. Figures are calculated as percentages of the cal-
cined weight of the hydrated sample.

Mixture N° 5 N° 6 " N° 5 N° 5 N° 5 N° 5 N° 5 NQ 5 No 6

Organic None None Citric Ethylenediamine- T a r t a r i c Malonic Malic Lactic Citric

admixture acid* sodium t e t r a c e t a t e * * a c i d * * * acid*** acid*** a c i d * * * acid*

%
CaSO4.2H20 39.4 40.0 54.5 49.5 47.5 49.6 57.3 41.7 55.4

C4A3S 20.4 20.9 26.3

w
Added as the I% of the anhydrous weight of the mixture.

Added as the 3% of the anhydrous weight of the mixture.

Added in equimolar amounts as referenced to the weight of c i t r i c acit

The same behaviour is noticeable in the curves shown on Fig. 4 which concern the
content of C4A3S not combined at the various times.
A possible explanation of this phenomenon is probably to be found in the
decrease in the concentration of the c i t r i c acid in the liquid phase during the
hydration of the paste with the consequent decrease of the retarding effect on
the hydration of the anhydrous sulphoaluminate. The progressive decrease of the
concentration of the citrate could be caused either by the precipitation of the
calcium citrate (8.5 g/l) or by the formation of increasing quantities of "tober-
morite gel" which could thus adsorb increasingly greater quantities of c i t r i c
acid. This second hypothesis seems to be encouraged by the results set out on
Table 4. They refer to the concentration of c i t r i c acid present in the aqueous
solution agitated for varying periods of time with different quantities "tober-
morite gel" free of Ca(OH)2. One notices that the concentration of citrate in
the liquid phase diminishes considerably with the increase of both the time of
contact and the quantity of "gel" present in the solution.
With regard to the stoichiometry of the possible reactions depending on when
the hydration of the sulphoaluminate occurs, one notes that in the mixture no. 7
the weight ratio between C4A3S and CaSO4.2H20 that have reacted between 6 and 340
hours is about l , independently of the presence of the c i t r i c acid. This value
is considerably higher than that found in sample l , notably richer in C3S. This
indicates that an equal proportion of weight of C4A3Swhich is hydrated in the
mixture no. 7 makes a smaller quantity of gypsum disappear than that which is
Vol. 2, No. 2 221
CALCIUM SULFOALUMINATE, CALCIUMSILICATE, HYDRATION

SO

" ~ ~ ' ~ C ~

0
eq
z

O
(,.1

2o

10 • SAMPLE lee 7
O " " " WITH CITRIC ACID

| l
I0 100
TIME OF" HYDRATION (hr)

FIG. 3
Percentage of gypsum vs. time

• SAMPLE lb. 7
o " • - WITH CITRIC ACID

)
' I
10 100
TIME OF HYDRATION ( h r )

FIG. 4
Percentage of C4A3S vs. time
222 Vol. 2, No. 2
CALCIUM SULFOALUMINATE, CALCIUM SILICATE, HYDRATION

Table 4
Concentration of the c t i r i c acid in solution brought into
contact with the "tobermorite gel."

N° Time ,, Solution
(days) Weigh ratio "Tobermorite 9el concentration
Solution I~/l )

1 0 .... I0.000
2 2 0.05 0.152
3 2 0.I0 0.120
4 3 0.05 0.122

consumed in mixture no. I. This result is wholly justified i f one considers the
different weight ratio C3S/C4A3S of the two samples and thus the different con-
tent of Ca(OH)2 potentially useable according to the scheme of reaction (1).
While in sample no. l the percentage of C3S i n i t i a l l y present is such as to
guarantee a Ca(OH)2 potential/C4A3S molar ratio notably superior to 6, in the
mixture no. 7, on the contrary, the Ca(OH)2 obtainable for the hydration of all
the C3S is not stoichiometrically sufficient to convert all the sulphoaluminate
in ettringite according to the scheme of reaction (1). For this reason in sam-
ple no. 7 an appreciable quantity of sulphoaluminate hydrates according to the
scheme of reaction (2) with a considerably lower consumption of calcium sulphate.

Conclusion
The hydration rate in sulphoaluminate paste does not depend appreciably on
the type of calcium sulphate nor on the type of calcium s i l i c a t e present in the
portland cement clinker. However, in the presence of C3S the hydration of the
sulphoaluminate proceeds with a higher consumption of gypsum and with the pro-
duction of a greater quantity of e t t r i n g i t e than that observed in pastes contain-
ing C2S. This can be attributed to the varying quantity of Ca(OH)2 liberated by
hydrolysis of the two s i l i c a t e s and consequently to the d i f f e r e n t stoichiometry
of the hydration reactions.
In addition, to a decrease of the C3S/C4A3S weight r a t i o corresponds a de-
crease of the hydrated lime which combines with the sulphoaluminate and with
gypsum. Consequently the stoichiometry of the hydration of the anhydrous sulpho-
aluminate varies, thus diminishing the quantity of e t t r i n g i t e obtainable from
the unit of weight of C4A3S. Therefore, with the purpose of converting all the

To simplify one allows that the C3S hydrates according to the scheme:
2C3S + 6H20 : C3S2.3H20 + 3Ca(OH)2.
"Vol. 2, No. 2 223
CALCIUM SULFOALUMINATE, CALCIUMSILICATE, HYDRATION

anhydroussulphoaluminate into the greatest possible quantity of ettringite, i t

is opportune that the C3S/C4A3S molar ratio in the expansive cement is not less
than 2 from which i t guarantees that the value of the Ca(OH)2 potential/C4A3S
~r
molar ratio does not descend below 6.
However, independently of the quantity of C3S present, the greater part of
C4A3S hydrates within the f i r s t 6 hours. This creates grave problems because
of the intense development of hydration heat and because of the brevity of the
setting time. In addition, during the hardening of the paste and of the con-
crete, i t is not possible to use, in any appreciable measure, the potential ex-
pansive properties of a cement containing C4A3S, this being for the greater part
combined during the f i r s t hours of hydration. I f , however, the hydration is
conducted in the presence of a number of additives, for example c i t r i c acid, i t
is possible to slow down the hydration rate of the C4A3S so that the expansive
capacities of the cement are developed so as to extend the setting time well
beyond that normally required.
Acknowled~ments
This paper was carried out with the financial support of Consiglio Nazion-
ale delle Richercheof Italy.
References
I. R. Turriziani and M. Collepardi, Ind. Ital. Cem. 37, 785 (1968).
2. M. Collepardi, G. Rossi and G. Usai, Ind. Ital. Cem. 37, 657 (1968).
3. R. Turriziani and F. Massazza, Ann. Chim.(Rome) 56, I172 (1966).
4. R. Turriziani, A. Rio and A. Celani, Ind. Ital. Cem. 32, 313 (1962).
5. F. D. Snell and C. T. Snell, Colorimetric methods of analysis, p. 396, 3 ed.,
D. Van Nostrand Company, Inc. New York, 1953.

Expansive cements of the Klein type which besides C4A3S contain fair quantit-
ies of f r e e CaO in the anhydrous c l i n k e r h a v e n o t been d e a l t with in t h i s study.