Dimensional stability of calcium aluminate and ... - LMC - EPFL

31 st Cement and Concrete Science Conference Paper Number 27
Novel Developments and Innovation in Cementitious Materials
12-13 September 2011
Imperial College London, United Kingdom
Dimensional stability of calcium aluminate and sulfoaluminate
systems
ABSTRACT
Julien Bizzozero, Christophe Gosselin, Karen Scrivener
Laboratory of construction materials, Swiss Federal Institute of Technology Lausanne (EPFL)
The aim of this study is to understand the dimensional stability of two expansive systems based on calcium
aluminate cement (CAC) and calcium sulfoaluminate cement (CSA). These alternative cements are
interesting for their performances and the lower CO2 emissions related to the manufacturing process.
Expansive and non-expansive compositions have been selected as a function of gypsum content for both
systems in order to compare the hydration kinetics, the microstructure and the dimensional stability. There is
a substantial difference in the expansion of the CAC and CSA systems when gypsum is added. The CSA
system has faster hydration reactions, a finer microstructure and the magnitude of expansion is lower. The
critical amount of calcium sulfate leading to high expansion has been optimized. However, the microstructure
(studied by XRD, TGA, SEM, MIP) between the low expansion and the high expansion system remains
comparable in terms of phases quantities. The only difference is the distribution of ettringite and aluminium
hydroxide much more finely intermixed in the CSA systems. The difference in the magnitude of expansion
could be explained by the crystallization pressure of ettringite and the supersaturation of the solution in the
nanopores formed few hours after the hydration of these cements.
1. INTRODUCTION
The use of alternative cements such as CAC and
CSA is motivated by the sustainable concerns
related to the CO2 emissions of the cement
production and the development of concretes and
formulated binders adapted to specific
applications. When gypsum is added to CAC, the
hydration patterns become closer to those
developed in plain CSA. However the present
study aims to demonstrate the fundamental
differences between the CAC and CSA system
when different amounts of gypsum are added.
In cements containing calcium sulfates and
aluminate phases, like CA, C4A3$ and C3A,
expansion can occur. Expansion is an important
property for applications where the permeability
and the crack development must be minimized.
One field of application of these binders is the floor
levelling. In these conditions the binder is exposed
to high surfaces compared to a really small
thickness. The drying shrinkage is here one of the
main causes of crack formation. For this reason
dimensional stability in this field is a priority.
However, if the expansion is too high, there can be
crack formation and even complete deterioration of
the cementitious sample. This expansion is usually
related to ettringite formation, which leads to
crystallization pressures that will cause expansion
of the cement paste, mortar or concrete.
Expansion exists in different forms and is a
relevant durability issue. For this reason it is
1
important to understand the main causes of this
phenomenon.
The objectives of this study are:
To find the expansion profile in function of
gypsum addition for CAC and CSA
systems with the analysis of different
compositions.
To analyse the reaction kinetics of the
different systems in order to evaluate the
effect of gypsum addition.
To study the microstructure and the phase
evolution.
To understand the mechanisms of
expansion.
2. Expansion theories
The expansive behaviour of expansive cements
has been studied during the past decades. Here
three different theories will be discussed: the
crystal growth theory presented by (Cohen, 1983),
the swelling theory presented by (Mehta, 1973)
and finally the crystallization pressure theory
discussed by (Scherer, 2002, 2004).
In the crystal growth theory, expansive particles
(i.e. CSA grains and CAC grains) act as a
nucleation site for ettringite. Expansion is due to
the growth of ettringite crystals from the surfaces of
the expansive particles, in other words there is a
solid state formation of ettringite. This generates
crystallization pressure and subsequent expansive

force. The expansive particles are covered by a
dense coating of ettringite whose further hydration
will increase the thickness of the coating layer
(Figure 1). When the thickness of the ettringite
layer exceeds that of the solution, the particles will
be in contact and exert pressure against the
others. This leads to expansion. This expansion
will persist until the gypsum or the expansive
particles will be depleted.
Figure 1. Ettringite coating around an expansive
particle (CSA cement grain) (Cohen, 1983)
In the swelling theory, the expansion is due to the
water adsorption or swelling of ettringite which is
formed by a through-solution mechanism, i.e. by
the hydrates precipitation into the pore solution.
This leads to ettringite formation into the bulk,
away from the anhydrous particles. Water
adsorption is possible because ettringite is
microcrystalline (has a colloidal size). The
formation of this type of ettringite occurs only if
calcium hydroxide (CH) is present in the solution. If
calcium hydroxide is absent, ettringite crystals will
be larger and swelling will not occur.
Scherer studied the expansion with a
thermodynamic approach and developed the
crystallization pressure theory. In porous materials,
cement paste, mortar, concrete and stone, there
can be a damage caused by the precipitation of
crystals from the liquid present in the pores. When
the supersaturation level of a given salt in the
liquid is reached, there is a precipitation of this salt.
Supersaturation depends on the concentration, on
the temperature and on the size of the pores. So a
crystal growing in a pore will encounter the pore
wall and exert pressure on it (Scherer, 2004).
When a crystal is surrounded by a film of solution
and there are small pores, high stresses appear.
High crystallization pressures require a high
supersaturation of the pore solution.
Supersaturation is given by the ratio between the
Ion Activity Product (IAP) and the solubility product
(Ksp) of the given phase. The species forming
ettringite are:
6 2 4 3 26
→ .3 .26
The resulting IAP for this phase is:
IAP=
. . . .
And the solubility product of ettringite is given by
(Warren and Reardon,1994):
2
10 . for AFt
A crystal in a supersaturated solution will grow until
the supersaturation will be consumed (IAP = Ksp).
Its growth produces a pressure on the pore walls
given by the Correns’ equation (2) (Correns, 1949):
(2)
Where R=8.314 J/K/mol is the gas constant, T is
the absolute temperature, vm is the molar volume
of the crystal (for ettringite vm = 707 cm 3 /mol).
3. Materials
3.1 CAC+C$H2 systems
The hydration of CAC systems with calcium sulfate
leads to the formation of ettringite and amorphous
aluminium hydroxide, see reaction (1). For this
study, gypsum (G) will be used and the respective
reaction is (2).
3CA+3C$Hx+(38-3x)HC3A.3C$.H32+2AH3 (1)
3CA +3C$H2+32HC3A.3C$.H32+2AH3
(2)
The calcium aluminate cement (CAC) used for this
project was supplied by Kerneos in France.
These compositions were measured by XRD and
quantified by the Rietveld refinement.
Table 1. Mineralogical composition of CAC cement
Anhydrous phase Cement notation [%wt]
Calcium aluminate CA 70
Gehlenite C2AS 20
Ferrites C3FT ~10
Perovskite CT
Spinel MgAl2O4
3.2 CSA+C$H2 systems
The main reaction occurring in CSA systems is
between the ye’elimite phase (C4A3$) and calcium
sulfate. This reaction (3) leads to the formation of
ettringite and amorphous aluminium hydroxide.
C4A3$+2C$H2+34HC3A.3C$.H32+2AH3
(3)
The calcium sulfoaluminate cement (CSA) used for
this project was supplied by Buzzi Unicem SPA in
Italy.
Table 2. Mineralogical composition of CSA cement
Anhydrous phase Cement notation [%wt]
Ye’elimite C4A3$ 50
Belite -C2A 20
Anhydrite + Gypsum C$ + C$H2 22
Calcite C
Aluminates C3A
Brownmillerite C4AF

4. Methods
All the experiments were carried out at 20°C.
Cement pastes were prepared with water/binder
ratio of 0.4.
All the mixes were prepared under the same
conditions. Additional gypsum (if present in the
mix) was added in water. Then cement was added
and mixed for 2 minutes using a paddle mixer
(1100rpm).
For expansion tests, the cement paste was cast in
special steel moulds of 1x1x4 cm 3 . The samples
were then cured for one day at high humidity
environment and demoulded after 24 hours. The
first measure of the length (~4 cm) was done with
an extensometer having a precision of ±1μm. The
samples were conserved into demineralized water.
Hydration kinetics were studied by isothermal
calorimetry with a TAM Air (3114/3236) from
Thermometric.
For SEM, XRD and MIP the samples were cast in
polystyrene cylinders of 30mm of diameter and
50mm high. After 24hours they were demoulded
and placed in recipients containing demineralized
water. For each age of curing (1,3,7,14 and 28
days) three slices were cut from the cylinders and
then introduced into isopropanol to stop hydration.
X-ray diffraction (XRD) analysis were done with a
Philips X’Pert Pro PANalytical (CuKα, λ=1. 54 Å)
working in Bragg-Brentano geometry with a 2θrange
of 5°-65°.
Scanning electron microscopy (SEM) was done
using a FEI Quanta 200 with an electric beam
generated by a tungsten (W) filament submitted to
a tension of 15 kV. The cement paste samples
were impregnated with epoxy resin under vacuum
conditions in order to fill the open porosity with the
resin and after were polished and coated with
carbon.
Mercury intrusion porosimetry (MIP) was done on a
Porotec machine with pressure capacity of 400
MPa. Massive samples of about 1.5 grams were
used for the analysis.
Pore solution analyses were done on cylindrical
samples of 5cm diameter and 10cm high. The
cement paste was casted in plastic bottles of the
same size and stored for 24hour is sealed
conditions. After the samples were cured for 6 days
under a reduced amount of water and finally tested
after 7 days of hydration. The applied pressure
was 560MPa in order to extract 10ml of pore
solution. The cations were analysed with ICP-OES
and the anions with ion chromatography.
3
5. Results
5.1 Expansion tests
Figure 2 shows the expansion in function of the
time for different compositions of CAC with
gypsum. The number on the right of the system
name (e.g. B-G1.0) is the molar ratio between C$
and CA. With the increase of the calcium sulfate
content there is an increase in expansion. For a C$
content below 0.7 the expansion seems to be
really low and above this ratio the expansion
continues on over time.
Figure 3 shows the expansion for the systems with
CSA and gypsum. Here again the expansion
increases with the calcium sulfate content and
there is a critical behaviour. The composition with
the higher amount of gypsum broke after 4 days.
There is not a soft transition between low and high
expansion for CSA systems. The range of
expansion of these systems is lower than for CAC
systems.
Expansion [%]
Expansion [%]
5.0%
4.0%
3.0%
2.0%
1.0%
0.0%
0 50 100 150
Time [days]
2.0%
1.5%
1.0%
0.5%
crack threshold
B‐G1.2
B‐G1.1
B‐G1.0
B‐G0.92
B‐G0.85
B‐G0.82
B‐G0.67
B‐G0.54
B‐G0.43
B‐G0.25
Figure 2. Expansion of CAC with gypsum.
4 days: sample broken
crack threshold
0.0%
0 50 100 150
Time [days]
B‐G0_ref
C‐G1.0
C‐G0.96
C‐G0.92
C‐G0.82
C‐G0.75
C‐G0.70
C‐G0.56
C‐G0.54
C‐G0.5
Figure 3. Expansion of CSA with gypsum.
High expansion
Low expansion
High expansion
Low expansion

5.2 Isothermal calorimetry
Isothermal calorimetry is useful to measure the
hydration kinetics of the different systems. Figure 4
and Figure 5 show that CSA systems have faster
kinetics than CAC systems. Both systems show a
similar behavior: the second peak is delayed when
the gypsum content increases.
Normalized heat flow [mW/g of binder]
35
30
25
20
15
10
5
B‐G‐0.25
B‐G0.67
B‐G0.82
B‐G0.92
B‐G1.1
0
0 10 20
Time [h]
30 40
Figure 4. Hydration kinetics of CAC with gypsum.
Normalized heat flow [mW/g of binder]
35
30
25
20
15
10
5
C‐G0.5_ref
C‐G0.54
C‐G0.56
C‐G0.92
C‐G1.0
0
0 10 20
Time [h]
30 40
Figure 5. Hydration kinetics of CSA with gypsum.
5.3 XRD analysis of CAC systems
For XRD results and SEM results only the CAC
systems are presented in this paper. During the
presentation both CAC and CSA systems will be
compared and discussed.
Figure 6 and Figure 7 show the consumption of the
anhydrous phases, CA and gypsum, and the
formation of ettringite on the left axis. On the right
axis the expansion is presented. There is more
ettringite precipitation in the high expansion
system. For the high expansion system it is
important to note that when ettringite formation
reaches a plateau, the expansion continues on
with the time. For CSA systems the behavior is
similar.
4
Intensity [cts]
Intensity [cts]
4.E+07
3.E+07
2.E+07
1.E+07
0.8%
0.7%
0.6%
0.5%
0.4%
0.3%
0.2%
0.1%
0.E+00
0.0%
0 10 20 30
Time [days]
40 50 60
4.E+07
3.E+07
2.E+07
1.E+07
Expansion [%]
CA
Gypsum
Ettringite
Expansion
Figure 6. Low expansion system (B-G0.67).
0.8%
0.7%
0.6%
0.5%
0.4%
0.3%
0.2%
0.1%
0.E+00
0.0%
0 10 20 30
Time [days]
40 50 60
Expansion [%]
CA
Gypsum
Ettringite
Expansion
Figure 7. High expansion system (B-G0.82).
5.4 SEM BSE images of CAC systems
Figure 8 and Figure 9 show the microstructure of
low expansive and high expansive CAC systems
respectively. The two main hydrates, ettringite and
amorphous aluminium hydroxide can be easily
distinguished. Ettringite is light grey and AHx is
dark grey in the hydrated matrix. The cracks are
mainly due to the sample preparation technique.
There is no substantial difference between the low
expansive and the high expansive system.
However there is a difference between CAC and
CSA systems, the hydrates are more intermixed in
CSA systems so it is more difficult to distinguish
them.
B-G0.67
Figure 8. Low expansion system (B-G0.67).
B-G0.82
Figure 9. High expansion system (B-G0.82).

5.5 MIP, comparison of CAC with CSA systems
Figure 10 shows the cumulative pore size
distribution of the 4 studied systems. CAC systems
have a bigger cumulative porosity than CSA
systems. The pore size distribution is comparable
for all the systems with a pore size of the
nanometric range (around 4 nm).
Porosity [%]
25
20
15
10
5
14 days
0
0.001 0.01 0.1 1 10 100
Pore Radius [µm]
CAC systems
CSA systems
B‐G0.67
B‐G0.82
C‐G0.92
C‐G1.0
LE
HE
LE
HE
Figure 10. Cumulative pore size distribution of CAC
and CSA systems.
5.6 Pore solution analysis
Figure 11 shows that with the increase of gypsum
content there is a strong increase of the
supersaturation which leads to an increase in the
crystallization pressure at 7 days of hydration.
There is a threshold of crystallization pressure at
around 70 MPa above which there is uncontrolled
and high expansion.
Ettringite saturation index [‐]
12
10
8
6
4
2
0
CAC+Gypsum
CSA+Gypsum
7days
0 0.2 0.4 0.6 0.8 1
C$/CA [mol/mol]
Figure 11. Supersaturation of ettringite in function of
the gypsum content.
6.0 Discussion
The previous results show that there is no
substantial difference between low expansive and
high expansive systems in the phase assemblage,
in the microstructure and in the pore size
distribution. However, the supersaturation index of
ettringite indicates a difference between the low
expansion and high expansion systems.
90
80
70
60
50
40
30
20
10
0
Crystallization pressure [MPa]
5
With the presented results it is possible to discuss
the different expansion theories. The crystal growth
theory seems to not apply to these systems
because there are no expansive particles and the
ettringite is formed through-solution. The swelling
theory seems also to be not so accurate because
in the studied systems there is expansion even if
no lime is present.
Finally the crystallization pressure theory seems to
be in accordance with the present results because
with an increase of the supersaturation index of
ettringite, which is the expansive phase, there is an
increase in expansion. It has to be noted that not
all the ettringite participate to the expansion. Only
a fraction of it does. Moreover the ettringite has to
form in a confined space to cause expansion
(pores of nanometric size).
7.0 Conclusions
There is a threshold between low and high
expansion at C$/CA molar ratio of 0.7 (40 to 50
%mol of C$).
There is a comparable behaviour for CAC and
CSA systems.
There is no noticeable difference in hydrates
content, microstructure and porosity between
low expansion and high expansion systems.
There is a higher supersaturation in the high
expansion systems, which causes high
crystallization pressures.
8.0 References
Cohen, M.D.,1983. Theories of Expansion in
Sulfoaluminate-Type Expansive Cements -
Schools of Thought. Cement and Concrete
Research, 13(6):809-818.
Cohen, M.D.,1983. Modeling of Expansive
Cements. Cement and Concrete Research,
13(4):519-528.
Mehta P.K. ,1973. Mechanism of expansion
associated with ettringite formation, Cement and
Concrete Research, 3(1):1-6.
Scherer, G.W.,2002 Factors affecting crystallization
pressure. In Internal Sulfate Attack and Delayed
Ettringite Formation (K.L. Scrivener and J.P.
Skalny, eds.),RILEM:Villars, Switzerland,139-
153.
Scherer, G.W.,2004. Stress from crystallization of
salt. Cement and Concrete Research,
34(9):1613-1624.
Warren, C.J. and Reardon, E.J.,1994. The
solubility of ettringite at 25°C. Cement and
Concrete Research,24(8):1515-1524.
Correns, C.W.,1949. Growth and Dissolution of
Crystals under Linear Pressure. Discussions of
the Faraday Society, (5):267-271.