C~MENT and CONCRETE RESEARCH.

Vol. 16, pp. 875-88A, 1986. Printed in the USA.

0008-88A6/86
$3.00+00.
Copyright (c) 1986 Pergamon Journals, Ltd.

THE EFFECT OF TEMPERATURE UPON THE SETTING BEHAVIOUR OF

REFRACTORY CALCIUM ALUMINATE CEMENTS

S.M.
Bushnell-Watson and J.H.
Sharp
Dept.
of Ceramics, Glasses and Polymers,
University of Sheffield, Northumberland Road,
Sheffield, SI0 2TZ, UK.

(Refereed)
Received July 30, 1986)

ABSTRACT
A
setting
time parameter
has
been
determined
from
the heat
evolution versus time curve measured by means of a the rmocouple
embedded in the cement paste as well as by the Gillmore needle.
The setting time increases with increasing temperature, until it
reaches a maximum between 26 and 30C, for four out of the five
refractory
calcium aluminate
cements
studied.
This anomalous
retardation
is not, therefore,
restricted
to ciment fondu, as
reported elsewhere.
The rate of consumption of CA during the first
24 hours of hydration as determined by X-ray diffraction also
indicated a retardation at 30C compared with 20oc for Secar 51 and
Secar 71o
The cause of the anomaly cannot be the presence of C2S ,
which is virtually absent in most of the cements studied.
An
alternative explanation must be found, which should also explain
why
Secar
80
behaves
differently
from
the
other
cements
investigated.

Introduction
The term high alumina cement covers a family of cements ranging from
40-80% alumina content (I-3).
The less alumina rich of these materials
contain several phases such as calcium silicate and calcium aluminoferrite in
addition to the pure calcium aluminateso
These are the more extensively used
high
alumina
cements
and
the
term
ciment
fondu
is often
used when
distinguishing
this class of materials
from the more refractory calcium
aluminate cements which are essentially members of the CaO-AI203 binary
system.
As the hydration temperature is increased, it has been reported (1,3-5) that
ciment fondu exhibits an anomalous setting behaviour in the region of 30C.
Thus Robson (i) states that, "In the range 18-30C the setting time of most
high alumina cements becomes slower as the temperature rises and only above
30oc does the rate of set accelerate again." Recently published data (4,5)
875

876

Vol. 16, No. 6

S.M. Bushnell-Watson and J.H. Sharp

indicate that the retardation in the case of ciment fondu and other cements
containing C2S is substantial.
By contrast, the purer refractory cements
such as Secar 250 and Super Secar, did not show the same effect, but instead
the setting time decreased continuously as the temperature was raised.
From an X-ray diffraction investigation of the hydration of Secar 71 at 22 ,
30 and 40C (to be discussed later), evidence of an apparent retardation in
the rate of consumption of CA at 30C compared with that at 22oc was
observed.
This suggested to us that the setting behaviour of Secar 71 might
be more
similar to that reported
for ciment
fondu than for the more
refractory cements.
Hence we decided to carry out a thorough investigation
of the setting behaviour of commercial refractory calcium aluminate cements,
with alumina contents in the range 50-80%.

TABLE [
Chemical Analysis of Refractory Calcium Aluminate Cements
Secar 51
AI203
CaO
MgO
SiO 2
Fe203
Na20
FeO
Fe
TiO 2

50.4
36.6
0.l
6.7
1.9
0.1
[.0
0.15
2.15

Secar 71
7[.0
27.0
0.2
0.35
0.25
0.35
.
0.05

Alcoa CA-14

Alcoa CA-25

72
27
0.15
0.25
0.20
0.20

79.0
18.0
0.4
0.i
0.3
0.5

.
-

Secar 80
80.5
18.0
0.l
0.2
0.15
0.25

.
-

0.03

Experimental
(a) Materials
The
cements
examined
were
commercial
refractory
high
alumina
cements
manufactured by either Lafarge (the three Secar samples) or by Alcoa (CA-14
and CA-25).
Typical chemical analyses as supplied by these companies are
listed in Table [.
X-ray powder diffraction indicated the presence of CA,
CA2, CI2A 7 and unreacted alumina in all the samples except that Secar 51
contained no CA 2 but did contain C2AS.
An indication of the relative amounts
of the various phases detected in the cements is given in Table 2. It is not
possible to be more quantitative since small variations were observed from
one
batch
to another
(perhaps associated with different
particle
size
distributions of the various phases) and the X-ray procedure did not involve
the use of an internal standard.
There is, however, no doubt that the phases
reported are present, and that, for example, the CI2A 7 content of the CA-25
sample is appreciably higher than that of the Secar 80 sample.
(b) Methods
X-ray
diffraction
(XRD)
traces
were
obtained
using
a
diffractometer using Cu K - ~ radiation and a scanning rate
The hydrated samples examined by XRD were prepared by adding
water (lO or 15 ml.), already at the correct temperature,

Philips PW-1730
of 2 2e/minute.
boiled distilled
to cement powder

Vol. 16, No. 6

87~
SETTING TIP~, ANOMALY, RETARDATION, CA CEMENT

TABLE 2
Mineralogical Composition of Refractory Calcium Aluminate Cements
Secar 51
CA
CA 2
CI2A 7
A
C2S
C2AS
C4AF

+++
nd
+
+
tr
+
tr

Secar 71

Alcoa CA-14

+++
++
tr/+
+
nd
nd
nd

+++
+++
tr
+
nd
nd
nd

+++ major amount present

++ substantial amount present
+
minor amount present

Alcoa CA-25
+++
+
+
++
nd
nd
nd

Secar 80
+++
++
tr
+++
nd
nd
nd

tr trace amount detected

nd not detected

(30g) and mixing by hand for two minutes.

The pastes were placed in a water
bath and samples periodically removed, ground with a large excess of acetone
to prevent further hydration, dried at room temperature and then examined.
Two different setting time parameters were determined.
The first involved
the use of a Gillmore needle, which weighed 0.25 Ib and had a point which is
1/12 in.
in diameter.
The time that elapsed between mixing the cement until
this needle failed to penetrate the surface of the cement was recorded as the
initial setting time.
The surface of the cement paste was impressed at
regular intervals with the needle, until no obvious surface mark was left.
This measurement of the setting time is obviously subjective, but reasonably
good reproducibility could be obtained and a clear pattern of behaviour was
observed.
The second method was designed to allow a more accurate and less subjective
measurement of a parameter related to setting to be determined by means of
the embedded thermocouple method (3-6).
As the hydration of a calcium
aluminate cement proceeds there is initially a small generation of heat.
There then follows a dormant period after which bulk precipitation occurs and
a corresponding increase in heat evolution is observed.
By embedding a
thermocouple into the hydrating paste, the onset of this heat evolution can
be detected and used as a measure of the setting time of the cement.
The
cement paste was prepared as before and poured into a plastic tube, which was
already immersed in the water bath and so had equilibrated at the desired
temperature.
A fine copper/constantan thermocouple was already placed inside
the tube to allow the paste to be poured round it to ensure that the
thermocouple bead was properly embedded.
A second thermocouple was placed
into the water bath and connected back-to-back with that in the paste so that
the
temperature
of
the cement
was measured
relative
to
that
of its
surroundings.
The heat evolution curve obtained by the embedded thermocouple method for a
Secar 71 paste prepared with a w:c ratio of 0.5 and hydrated at 35oc is shown
in Figure i.
Since it is difficult to determine precisely the time at which
the heat evolved departs from the base line, the setting time parameter
reported
in this paper
is the extrapolated
onset
time determined
as
illustrated
in Figure I.
This procedure was found to give excellent
reproducibility and was the preferred method.

878

Vo!. 16, No. 6

S.M. Bushnell-Watson and J.H. Sharp

Secar 71 hydrated at 35oC

\,.

\\

TI

t.

FIG. I

Heat Evolution versus Time

Results
(a) XRD analysis of unreacted CA
As part of an investigation of the effect of temperature on the hydration of
refractory calcium aluminate cements, samples of Secar 71 were hydrated at
22 , 30 and 40oc with a w:c ratio of 0:5.
The XRD peak height for the
unreacted CA was observed to be greater after 7 hours hydration at 30C than
at 22C.
A similar observation was made in the case of Secar 51, but not in
the case of Secar 80 (Figure 2).
The effect was only seen during the initial
stages of hydration and beyond 2A hours the consumption of CA increased with
increasing temperature.
(b) Initial setting times determined by the Gillmore needle
The Gillmore needle was used to determine the initial setting time for each
of the three Secar cements and the results are shown in Figure 3.
These
results confirm the setting behaviour suggested by the XRD data.
Both Secar
51 and Secar 71 cements clearly exhibit a retardation in set at 30C, whilst
Secar 80 does not.
The retardation is very dramatic and in the case of Secar
51 the setting time is I0 hours longer at 30C than at 20C.
(c) Setting time parameter determined b~ the" embedded thermocouple method
Once again the three Secar cements were examined first and the results are
shown in Figure 4.
Many more determinations were made because of the more
convenient method of measurement.
Excellent reproducibility was obtained;
repeat determinations were made frequently, which never differed by more than
15 minutes and usually by no more than 5 minutes.
The results shown in Figure 4 provide further confirmation of the anomalous
setting retardation of Secar 51 and Secar 71 as the temperature is raised
from 10 to 30oc.
The behaviour of $ecar 80 is again different from that of
the other two cements.
The

plots

constructed

from

the

two

methods

of

determining

setting

time

Vol. 16, No. 6

879
SETTING TIME, ANOMALY,

RETARDATION,

CA CEMENT

Decrease in CA as Hydration Progresses

$Qcar 51

4ooc~', \\\

Secar ? I
o
o

Secar 51

Secer eO

:/

TIME (In scale]

40C

30oc~

2oc

"

Secar 7 1

qr
i'

24c

Secar 8 0
M

o
1~

~'s

7days

TU~

10

20

30

40

50

H y d r a t i o n T e m p e r a t u r e (oC)

FIG. 2
XRD peak height of
CA versus time (log scale).

FIG. 3
Initial setting time versus
temperature of hydration as
determined by the
Gillmore needle.

parameters are in very good agreement.

It is to be expected that the actual
times measured
are
slightly
different,
since
the methods
determine
a
different stage in the setting process.
This observation confirms that made
by George (4) who reported excellent correlation between results obtained by
use of the Vicar needle and embedded thermocouple methods.
In addition to affecting the setting time, the temperature at which hydration
occurs affects the shape of the heat evolution curve determined by the
embedded thermocouple method.
This is illustrated in Figure 5.
As the
temperature is raised to 28C, the evolution of heat becomes more gradual,
resulting in a shallow curve.
Above 30oc there is a dramatic change in the
shape of the curve and a definite peak is clearly resolved.
On further
increase in temperature this peak becomes progressively sharper.
The results presented so far were not what was expected from reading the
relevant literature.
Although Secar 51 is intermediate between ciment fondu
and
the
"pure"
refractory
calcium
aluminate
cements,
it
contains
a

880

Vol. 16, No. 6

S.M. Bushnell-Watson and J.H. Sharp

Secer

5 1

Secar

7 1

Secer 80

it

01
~"

e
u~

(O

ot
0

i
10

20

30

40

50

Hydration T e m p e r a t u r e (C)

l'hr

0.5hr

FIG. 4
FIG.
Setting time parameter versus
temperature of hydration as
determined by the embedded
thermocouple method.

Heat evolution curves

for Secar 71 hydrated at
various temperatures.

significant amount of SiO 2 and might, therefore, be expected to display

similar behaviour to ciment fondu (as indeed it does).
Secar 71, however, is
very similar
to Secar 250, yet its behaviour reported here is totally
different from that reported for Secar 250 (5).
Since Secar 71 does not
contain more than trace amounts of C2S , the presence of this phase cannot be
the cause of the anomalous retardation observed in the case of this cement by
means of three different experimental methods.
To determine whether the exceptional behaviour is associated with Secar 71 or
with Secar 80, the study was extended to include the equivalent cements
marketed by Alcoa, CA-14 and CA-25.
The setting time parameter as determined
from the heat evolution curves is shown in Figure 6.
Both of these cements
exhibit a definite increase in setting time between I0 and 30C, leading to a
maximum setting time around 28C.
Their behaviour is closely similar to that
reported for Secar 51 and Secar 71 in the present study, for ciment fondu
(4,5) and for pure CA (7,8).
It is concluded,
therefore,
that the
exceptional behaviour is that displayed by Secar 80.

Vol. 16, No. 6

881
SETTING T I ~ ,

ANOmaLY,

RETARDATION,

CA CLMENT

+4

AIoll C A 2 S
t~4

v,-

AIcoe

W:C

O.S

W:C

0,33

CA14

(D
0

lm

E
0

cn

~P
c~

L
\

r
t

0
0

10

20

30

40

50

10

20

30

40

i50

Hydration Temperature (0C)

HydratlonTempermture(oC)

FIG. 6

FIG.

Setting time parameter versus

temperature of hydration as
determined by the embedded
thermocouple method.

Setting time parameter versus

temperature of hydration of
Secar 71 as determined by the
embedded thermocouple method.

All the pastes used to obtain the data so far presented were prepared using a
w:c ratio of 0.5.
Since this is a higher value than that used in previous
studies, a further series of setting time parameters was determined for Secar
71 using a w:c ratio of 0.33.
The results are shown in Figure 7 and it can
be seen that they are closely similar to those obtained with a w:c ratio of
0.5.
The
anomalous
retardation
in setting
time
parameter
was again
observed.
Discussion
It is clearly established by means of the three different experimental
methods that a maximum setting time is observed between 26 and 30C for four
of the five refractory calcium aluminate cements investigated.
The anomalous
setting
behaviour
reported
by George
et
ai.(3-5)
is not,
therefore,
restricted to ciment fondu, but is the normal behaviour to be expected during
the hydration of high alumina cements.
The fifth cement studied, Secar 80,
shows a different setting behaviour, in that the setting time decreases with

882

Vol. 16, No. 6

S.M. Bushnell-Watson and J.H. Sharp

increasing temperature especially between 15 and 25C.

It might be expected
that Secar 80 and Alcoa CA-25, with similar chemical analyses, would behave
in a similar manner, but this is not so.
The behaviour reported by George et
al.(5) for Secar 250 and Super Secar is different from that reported here
for the supposedly similar cements Secar 71 and Secar 80.
Although the
setting time of these cements also decreases with increasing temperature,
they behave differently from each other and in a different manner from Secar
80.
This study also demonstrates that it is meaningless to quote a setting time
for high alumina cements, as is frequently done in trade literature, without
stating the temperature at which it has been determined.
The first explanation for the effect was proposed by George et al. (4,5) who
suggested that the presence of C2S in ciment fondu might be responsible for
the observed retardation in setting.
This explanation was subsequently
realised not to explain all the observations made (3,5).
Furthermore, a
study of the hydration of pure CA (7,8) showed that this also displayed a
maximum setting time at 27C.
The results presented here for Secar 71, Alcoa
CA-14 and CA-25 establish that the presence of C2S in the cement is not the
cause of the anomalous setting behaviour and some other explanation must be
found.
It is possible that admixtures incorporated in commercial cements may affect
their setting behaviour.
We have no access to information about these,
except that they are present at least in some cements (9). Indeed we had
speculated that the explanation for the anomalous setting behaviour might be
the presence of an admixture, until we became aware of the similar anomaly in
the setting of pure CA (7,8).
Clearly there were no admixtures affecting
this sample, so that the exceptional behaviour is that shown by Secar 80 (and
by Secar 250 and Super Secar).
It is conceivable that these differences are
associated with the presence of admixtures, but we cannot confirm or negate
this hypothesis.
Any mechanism must explain several observations.
Firstly, why there is an
increase in the setting time in the region I0-30C with a maximum setting
time in the range 26-30C f o r pur e CA, ciment fondu, Secar 51, Secar 71,
CA-14 and CA-25.
Secondly, why there is a different relationship between
setting
time and temperature
for Secar 80, Secar 250 and Super Secar.
Thirdly, the dramatic changes in the profiles of the heat evolution curves
around 30C shown in Figure 5.
Fourthly, why the delay in hydration is only
an initial effect; beyond about 16 hours the degree of hydration increases
with increasing temperature (Figure 2), as would be expected from a "normal"
relationship between setting time and temperature.
Whilst we would agree with George's most recent conclusion
(3) that a
coherent explanation for the retardation remains elusive, it is clear that it
must be associated with either the rate of dissolution of the anhydrous
phases or the nucleation and precipitation of the various hydrates or both.
Murat et al(7) emphasised that at temperatures below 15C only CAHIo is
formed, that at temperature above 27C only C2AH 8 is formed, whereas the two
hexagonal hydrates are formed simultaneously in the intermediate temperature
range.
This is an oversimplification,
since we have observed CAHI0 as a
minor hydration product at 40C and even 50oc.
Moveover, the phases formed
depend on other factors such as the Ca:A1 ratio in solution as well as
temperature.
More recently, Guilhot and Soustelle(8) slightly modified the

Vol. 16, No. 6

883
SETTING TIME, ANOMALY, RETARDATION,

CA CEMENT

temperature ranges quoted above and concluded that the behaviour depends very
largely on nucleation phenomena which still remain poorly understood.
Our observation that beyond 16 hours the extent of hydration increases with
inceasing temperature leads us also to favour an explanation involving the
difficult nucleation of C2AH 8.
Other authors (6,10,11) have shown evidence
for a nucleation barrier to the formation of the hexagonal C-A-H products,
while Coffin and George(12) have suggested that the formation of sub-critical
embryos of CAHI0 hinders the development of critical embryos of C2AH 8.
In
the temperature range under consideration there is no doubt that the favoured
hydration products from pure CA and high alumina cements are C2AH8, alumina
gel and gibbsite, AH 3.
Conclusions
i.

A maximum setting time in the range 26-30C has been observed with Secar
51, Secar 71, Alcoa CA-14 and Alcoa CA-25.
This is similar to the
behaviour already reported by other workers for pure CA and ciment fondu
and is, therefore,
the behaviour
t o be expected for high alumina
cements.

2.

This maximum setting time was not observed for Secar 80.
Instead the
setting time decreased with increasing temperature especially between 15
and 25oc.

3.

The anomalous setting behaviour of high alumina cements cannot be due to

the presence of C2S, which is virtually absent in Secar 71, CA-14 and
CA-25.

4.

Whenever a setting time is quoted for a high alumina cement,

temperature at which it has been determined should be stated.

5.

The variation in the setting time parameter of Secar 71 determined by

the embedded thermocouple method is virtually independent of the w:c
ratio.

6.

The retardation in setting time between

with difficult nucleation of C2AH 8.

20 and 30C may

the

be associated

Acknowledgements
One of us (SMBW) wishes to thank I.C.I.
Agricultural Division and SERC for a
grant under the CASE scheme.
We are grateful to Lafarge Aluminous Cement
Co., and I.C.I.
for the supply of samples, and to Mr.
L.
Lord (I.C.I.),
Dr.
C.
Fentiman
(Lafarge), Dr.
P.F.G.
Banfill and Miss S.
Gill
(University of Liverpool) for helpful discussions.
References
I.

T.D. Robson, High

Record (1962).

2.

K.M. Parker
(1982).

3.

C.M. George in "Structure

Applied Science Publishers,

and

Alumina

J.H.

Cements

Sharp,

Trans.

and

J.

and Performance
p.415 (1983).

Concretes,

Brit.

of

London

Ceram.

Cements",

Soc.

Ed.

Contractors

81,

P.

35-42

Barnes,

884

Vo!. 16, No. 6

S.M. Bushnell-Watson

George,

4.

C.M.

5.

C.M. George, B. Cottin and R. Ranc,

Turin, pp.138-149 (1982).
Rodger

Trans.

and

D.D.

J.

and J.H. Sharp

Brit.

6.

S.A.

Double,

7.

M. Murat, P. Galtier, B. Guilhot

Set. C, 291, 113-115 (1980).

8.

B. Guilhot and M.
p.97-Iii (1982).

9.

C.M. George, 'Developments

talk given
to the South
Johannesburg, May 1980.

Soustelle,

II. P.
Barter
pp.112-124,

and D.
Bertrandie,
Turin ([982).
George,

Int.

Soc.,
Seminar

Concr.

82-90

on Calcium

Res.,

and A. Negro,

Seminar

79,

C.R.

on Calcium

14,
Acad.

D.

Int.

Bertrandie,

Int.

Cem.

Seminar

Seminar

on

Calcium

(1984).

Sci.,

Paris,

Aluminates,

Turin,

Concr.

on

(1980).

Aluminates,

73-82

in Refractory Hydraulic Cements',

African
Institute
of Refractory

D. M~n~trier

and C.M.
(1982).

Cem.

Int.

10. P. Barter,
(1974).

12. B. CottEn
pp.160-170

and

Ceram.

based on a
Engineers,

Res. ~,

Calcium

545-556

Aluminates,

Aluminates,

Turin,