See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/262993002

Flow Properties of MK-Based Geopolymer Pastes. A Comparative Study with

Standard Portland Cement Pastes

Article in Soft Matter · November 2013

DOI: 10.1039/c3sm51889b

CITATIONS READS

151 1,371

5 authors, including:

Aurélie Favier Jean-Baptiste d'Espinose de Lacaillerie

Mobbot École Supérieure de Physique et de Chimie Industrielles
28 PUBLICATIONS 1,867 CITATIONS 127 PUBLICATIONS 5,139 CITATIONS

SEE PROFILE SEE PROFILE

N. Roussel Julie Hot

Université Gustave Eiffel Institut National des Sciences Appliquées de Toulouse
216 PUBLICATIONS 15,978 CITATIONS 20 PUBLICATIONS 649 CITATIONS

SEE PROFILE SEE PROFILE

All content following this page was uploaded by Jean-Baptiste d'Espinose de Lacaillerie on 16 October 2014.

The user has requested enhancement of the downloaded file.

Soft Matter
PAPER

Flow properties of MK-based geopolymer pastes. A

comparative study with standard Portland cement
Cite this: Soft Matter, 2014, 10, 1134
pastes
Aurélie Favier,ab Julie Hot,a Guillaume Habert,b Nicolas Roussela
and Jean-Baptiste d'Espinose de Lacailleriecd

Geopolymers are presented in many studies as alternatives to ordinary Portland cement. Previous studies
have focused on their chemical and mechanical properties, their microstructures and their potential
applications, but very few have focussed on their rheological behaviour. Our work highlights the
fundamental differences in the flow properties, which exist between geopolymers made from metakaolin
and Ordinary Portland Cement (OPC). We show that colloidal interactions between metakaolin particles
are negligible and that hydrodynamic effects control the rheological behaviour. Metakaolin-based
geopolymers can then be described as Newtonian fluids with the viscosity controlled mainly by the high
Received 10th July 2013
Accepted 11th November 2013
viscosity of the suspending alkaline silicate solution and not by the contribution of direct contacts
between metakaolin grains. This fundamental difference between geopolymers and OPC implies that
DOI: 10.1039/c3sm51889b
developments made in cement technology to improve rheological behaviour such as plasticizers will not
www.rsc.org/softmatter be efficient for geopolymers and that new research directions need to be explored.

Fresh behaviour of cementitious suspensions such as

1. Introduction
The fundamentals behind the visco-elastic properties of
concentrated solid suspensions have been a subject of study for
many years.1 Beyond scientic curiosity, this interest has been
driven by the prevalence of concentrated suspensions in many
industrial formulations. Among them, the cementitious paste
used to cast concrete in the building industry is arguably by far
the most used today in volume.2 As a consequence, the “so
matter” properties of fresh Ordinary Portland Cement (OPC)
pastes have been the subject of particular attention since the
late 80's.3 It so happens that OPC pastes and OPC-based
concrete, despite being common and seemingly dull construc-
tion materials, actually hide a complex and challenging rheo-
logical behaviour that has only been grasped by combining the
knowledge of three disciplines: colloidal physics, inorganic
chemistry and uid mechanics.4
Indeed, signicant practical advances such as the develop-
ment of self-compacting concretes5 able to ow under the sole
effect of gravity, high-mechanical strength concretes6 able to
ow while having a solid volume fraction higher than 90% or
bre-reinforced concretes7,8 able to ow while containing
elongated rigid steel bres were only possible because of
fundamental questioning of granular paste rheology.9
a
IFSTTAR, Université Paris Est, France
b
Institute for Construction and Infrastructure Management, ETH Zurich, Switzerland
c
SIMM, UMR CNRS 7615, ESPCI ParisTech, France
d
Institute for Building Materials, ETH Zurich, Switzerland
concrete, cement pastes and grouts may be approximated over a
limited range of shear rates by using a simple linear Bingham
model.3,6,10
s ¼ s0 þ hp gc (1)
where s is the shear stress and g_ is the shear rate. From a
practical point of view, this approximation of the fresh concrete
paste behaviour with a Bingham model considerably simplies
the work as only two empirical parameters need to be deter-
mined to fully describe the rheological behaviour in the usual
range of industrial shear rates: the plastic yield stress s0 and the
plastic viscosity hp. These two parameters are shown to be
controlled by the competition between three major types of
interactions:11,12 colloidal interactions due to electrostatic and
van der Waals forces between particles, viscous forces in the
interstitial liquid squeezed between cement particles, and
direct contact forces.
In cement pastes, van der Waals colloidal forces dominate
hydrodynamic forces (both viscous and inertial) in the low
strain rate regime (i.e. less than several tens of s1) and give
rise to a shear thinning macroscopic behaviour. In the most
extreme cases, a viscosity inversely proportional to the shear
rate can be measured. At an intermediate strain rate (i.e.
around 100 s1), hydrodynamic viscous forces oen dominate
and can be at the origin of a plateau in viscosity. Finally, in the
case of highly concentrated systems (i.e. for shear rates of the
order of a couple hundreds s1 or higher), particle inertia
dominates the high strain rate response and may lead to shear

1134 | Soft Matter, 2014, 10, 1134–1141 This journal is © The Royal Society of Chemistry 2014
Paper
thickening. In the case of deocculated highly concentrated
systems, a viscosity almost proportional to the shear rate can
even be measured. The transitions between these regimes are
governed by critical strain rates that depend on the cement
particle average size, the uid viscosity which can be modied
by polymer admixtures, the cement density, and the intensity
of the van der Waals forces, which can be controlled by
adsorbing polymers.13 However, the industrial success of
concrete raises new issues beyond its engineering properties.
Indeed, its environmental impacts, albeit low per unit of mass,
are leveraged by its massive consumption. For example,
concrete production is responsible for 5 to 10% of all anthro-
pogenic CO2 emissions.14–16 As a consequence, since the last
decade, a growing interest in alternative binders has grown.
Examples are sulfo-aluminate and magnesia cement.17 Geo-
polymers, which are based on the chemistry of alkali-activated
inorganic binders, are also considered as a highly potential
solution to reduce CO2 emissions in the eld of construction
materials.18,19 These binders are obtained by reaction of an
activated silica-alumina with an alkali silicate solution. In
order to develop a building material that could be used as a
cement replacement for specic applications, the under-
standing of the fundamentals at stake behind its fresh rheo-
logical behaviour is, as it was the case in cementitious systems,
primordial. Some studies on geopolymers based on industrial
wastes such as y ashes from the coal industry or blast furnace
slags from the steel industry20,21 showed that similarly to
cement pastes, these materials display visco-plastic behaviour.
Criado et al.22 and Palacios et al.23,24 studied the effect of
commercial rheology modifying polymers used in cement
applications on geopolymers obtained by alkali-activation of y
ashes and slags. They showed that these plasticizers have a low
efficiency in an alkali-activated system due to the fact that
superplasticizers such as polycarboxylates are altered aer long
exposure in a highly alkaline environment.25 However, recent
studies have identied PCE resistant in mild alkali activation
systems (0.5% NaOH),26 which could then be also more stable
in a highly alkaline environment.
In this study, we focus on a simpler model system where a
pure aluminosilicate phase is used and dissolved in a sodium
silicate solution in order to highlight the fundamentals of the
rheological behaviour of a fresh geopolymer. This model system
presents the interest of having a relatively simple chemistry
where Si, Al and Na are the predominant ions. Finally, in order
to address the potential use of these geopolymers as replace-
ments for Portland cement, we conduct a comparative study
between these geopolymer mixtures and standard OPC pastes.
As the rst approximation,27,28 we assume that fresh geo-
polymer pastes can be viewed as suspensions of particles
(metakaolin) in a continuous uid (sodium silicate solution).
In that sense, the physics which govern their rheological
behaviour is similar to that of a cement paste and grasped by
the physics of concentrated suspensions of rigid particles.29
Below, we will focus on two main parameters, the yield stress
and the viscosity, and evaluate through detailed rheological
measurements the main types of interactions that are
controlling them.
This journal is © The Royal Society of Chemistry 2014
Soft Matter
2. Experimental procedure
2.1. Material preparation
Geopolymer pastes were prepared by mixing a sodium silicate
solution with metakaolin (MK) particles. The chosen MK for
this research is Argical M-1000 from AGS Minéraux (France). Its
BET specic surface area is equal to 17 m2 g1 and the average
diameter in mass is approximately 10 mm. The alkaline solu-
tions were prepared from a commercial sodium silicate solution
with a SiO2/Na2O molar ratio of 3.2 and 65 wt% of water (VWR
International, France), sodium hydroxide pellets (Merck KGaA,
Germany) and distilled water. They were prepared by mixing an
appropriate amount of NaOH, commercial sodium silicate
solution and distilled water in a plastic bottle, which was
subsequently closed to avoid evaporation and carbonation. The
solutions were then allowed to cool for 24 h. Numerous studies
have shown that Si/Al close to 2 and Na2O/Al2O3 between 1 and
1.2 are the best molar ratios for structural applications30,31 and
development of mechanical properties.
Thus, geopolymer pastes were prepared by adding to an
alkaline solution with a H2O/Na2O molar ratio of 15 and a
constant SiO2/Na2O molar ratio of 1.15 the amount of MK
necessary to reach a Si/Al ratio of 1.8. The volume fraction of MK
is 0.3. The suspension was then mechanically mixed with a
stirrer for 5 minutes before immediate testing. The mass of MK
was adjusted relatively to the alkaline solution in order to
provide the ratio given in Table 1.
Furthermore, to better dene the rheological behaviour of
the interstitial phase, additional sodium silicate solutions of
various molar ratios (0.6 # SiO2/Na2O # 2 and 10 # H2O/Na2O
# 20) were studied independently of the geopolymer paste
preparation. They were obtained by dissolving silica (silica gel
60, 0.063–0.2 mm from Merck KGaA, Germany) in an aqueous
solution of NaOH prepared by dissolving NaOH pellets in
distilled water.
Portland cement pastes were prepared by mechanically
mixing cement powder (CEM I type cement from Lafarge Le
Havre) with water (the water/cement ratio (W/C) between 0.35
and 0.5 and f between 0.39 and 0.47) for 2 minutes. This range
of solid concentrations was selected because it allows for the
formation of a paste that remains stable under the effect of
gravity.32 In contrast to the industrial practice, most of these
cementitious systems were prepared without any rheology
modifying polymers in order to keep the system simple enough
to be compared with the geopolymer paste. However, in order to
include this aspect in the present study, some cement pastes
were also prepared with an addition of a commercial poly-
carboxylic ether13 (Tempo 12, Sika technology) typical of the
organic polymers used in the construction industry. The poly-
mer dosage was in the typical range of industrial dosages (i.e.
0.5% of the mass of cement).
2.2. Rheological measurements
The rheological measurements were carried out with a C-VOR
Bohlin® rheometer equipped with a vane geometry for the
geopolymer and Portland cement mixtures and parallel plate
Soft Matter, 2014, 10, 1134–1141 | 1135
Soft Matter Paper

Table 1 Chemical composition given in molar ratios of the geopolymer paste

SiO2/Na2O (solution) H2O/Na2O (solution) Si/Al (MK + solution) Na2O/Al2O3 (MK + solution)

Geopolymer paste 1.15 15 1.8 1

geometry for the solutions. When pastes were tested, an initial

oscillatory pre-shear with a rate of 100 s1 was applied for 60 s
prior to each test to ensure that all samples were in the same
reference state of stress and strain.

3. Results and discussion

3.1. Macroscopic ow curves
As already stated, depending on their formulation, cement
suspensions can display in steady state ow a large variety of
behaviours, going from Newtonian (constant apparent viscosity)
to shear thinning (decreasing apparent viscosity with the shear
rate) or shear thickening (increasing apparent viscosity with the
shear rate).12,33
Fig. 1 shows the typical rheological behaviour which can be
obtained by modifying the solid volume fraction of a cement
paste as well as by adding a rheology-modifying poly-carboxylic
ether polymer. Here, a typical cement paste is meant to corre-
spond to a water cement ratio of 0.35 related to a solid volume
fraction of f ¼ 0.48. It illustrates the rheological behaviour of
pastes used for standard concrete application32 whereas the
paste with W/C ¼ 0.4 (f ¼ 0.44) illustrates the behaviour of very
uid products such as grouts for soil injections.
Fig. 2 shows the viscosity of a typical MK geopolymer in
comparison with the viscosity of a typical cement paste with
W/C ¼ 0.35. In the log–log scale, such as the one in Fig. 2, the
colloidal contribution can be represented by a slope-1 whereas
the viscous contribution can be represented by a horizontal
line. These respective contributions are plotted as dotted lines
in Fig. 2 for cement pastes.
When comparing the ow curve for the typical cement paste
with the one for a MK-based geopolymer, we observed quite
signicant differences. The viscosity of the geopolymer paste
was lower than that of the cement paste at low shear rates and
higher at high shear rates. Furthermore, it was almost shear rate
independent. This meant that the viscous contribution
Fig. 1 Apparent viscosity as a function of shear rate for cement pastes
(SP ¼ superplasticizer) relative to different applications.
Fig. 2 Apparent viscosity as a function of shear rate for geopolymer
paste (f ¼ 0.3 and f/fmax  0.7) and cement paste (f ¼ 0.47 and f/fmax
 0.7). Dotted curves describe the theoretical colloidal contribution
change according to the inverse of the shear rate and theoretical
viscous contribution independent of the shear rate.12
dominated over the contributions of both the colloidal inter-
actions and particle inertia over a larger range of shear rates
than in the case of cement pastes.
3.2. Evaluation of the colloidal contribution to the yield
stress
We then extrapolated the yield stress s0 from shear stress
measurements in the low shear rate range. In Fig. 3, the
evolution of shear stress for cement paste, geopolymer and
geopolymer constitutive silicate solutions is shown as a func-
tion of shear rate. Table 2 gives the estimated values of shear
stresses. Obviously, compared to the cement and MK suspen-
sions, the silicate solution could be considered as a purely
viscous uid as its yield stress is between one and two orders of
magnitude lower than that of the cement paste (Fig. 3). The
yield stress could be fully neglected compared to the contribu-
tion of viscous dissipation to the shear stress. The viscosity of
the silicate solution being almost constant, it behaved similarly
to the interstitial water in the cement paste also shown in Fig. 3
and can be described as a Newtonian uid.
Moreover, we measured the onset of ow as a function of the
shear strain for both the standard OPC paste and MK paste
using a vane test procedure.34 Let us recall that in this test, the
sample is sheared at a very low shear rate and the evolution of
stress with shear strain is recorded. Fig. 4 shows the result of
this vane test for cement and geopolymer pastes.
Fig. 4a shows results typical of a cement suspension. It
displays two characteristic critical strains. The rst one is

1136 | Soft Matter, 2014, 10, 1134–1141 This journal is © The Royal Society of Chemistry 2014
Paper Soft Matter

Fig. 3Shear stress as a function of shear rate for geopolymer (H2O/Na2O ¼ 15; f ¼ 0.3), cement paste (W/C ¼ 0.35 and f ¼ 0.47), the geo-
polymer suspending fluid (silicate solution) and the cement suspending fluid.

Table 2 Estimated yield stress values for the geopolymer, cement around 103 whereas the second one is between 102 and 101.
paste, geopolymer suspending fluid (silicate solution) and cement Recent studies have shown that the peak associated with the
suspending fluid
smallest critical strain nds its origin in the breaking of
Yield stress the calcium silicate hydrates (C–S–H) percolating between the
cement grains. The second peak nds its origin in the collapse
Geopolymer paste 2 Pa of the colloidal interaction network.35 Above this second critical
Standard cement paste 20 Pa strain, the system is considered to be owing. In the geopolymer
Sodium silicate solution 0.01 Pa
system though, we observed only one peak (Fig. 4b). It was
Cement interstitial uid 0.001 Pa
located between 102 and 101. This suggests that only one type
of interaction between particles is involved, and that once the
particles have moved sufficiently far from their initial positions,
this interaction vanishes and the ow starts. Recently, Favier
et al.36 have shown that the elastic modulus observed in the MK-
based geopolymer is due to the formation of an inter-grain gel at
a very early stage and is not due to colloidal interactions
between grains. They observed that the critical strain of a geo-
polymer (as shown in Fig. 4b) is controlled by the critical strain
of an inter-grain aluminosilicate gel. As a consequence, geo-
polymer pastes seem to have their yield stress controlled solely
by the percolation of an aluminosilicate gel between MK parti-
cles, and no colloidal interactions between MK particles are
involved.
Furthermore, as already observed above, geopolymers have a
yield stress far lower (<0.1 Pa) than that of typical cement paste
(>10 Pa). This suggests that the attraction between C–S–H is
much stronger than the cohesion of the early aluminosilicate
gel produced in the early stages of the geopolymeric reaction.
From these differences, an important practical consequence
can be anticipated. As the role of cement rheology modifying
industrial admixtures is to reduce the yield stress by mini-
mizing colloidal interactions between particles, such admix-
tures would be of no effect for geopolymer pastes since their
yield stress is not related to colloidal interactions but to the
percolation of the early reaction products.

3.3. Evaluation and comparison of the viscosities of fresh

Fig. 4 Vane test experiments (a) for the standard cement paste plotted
from (N. Roussel et al. 2012) (W/C ¼ 0.4 and f ¼ 0.44) after 20 minutes
and (b) for the geopolymer paste (H2O/Na2O ¼ 15, f ¼ 0.3) after 10
minutes at room temperature.
cement and geopolymer pastes
From a theoretical point of view, geopolymer pastes can be
considered as suspensions of rigid MK grains in a suspending
sodium silicate Newtonian solution. Their viscosities can be

This journal is © The Royal Society of Chemistry 2014 Soft Matter, 2014, 10, 1134–1141 | 1137
Soft Matter
predicted using empirical expressions such as the Krieger–
Dougherty relationship.37 The general form of these relation-
ships can be written as:
 n
f
h ¼ h0  1  (2)
fmax
with
n ¼ ½h  fmax
where h0 is the solution viscosity, f is the solid volume fraction,
[h] is the intrinsic viscosity, and fmax is the dense packing
fraction. The viscosity of the paste is therefore controlled by two
parameters depending respectively on the suspending intersti-
tial liquid and on packing properties of the grains. Below, these
two contributions are studied separately.
3.3.1. Evaluation of the interstitial liquid contribution. In a
MK-based geopolymer, the sodium silicate solution, which can
be considered as the liquid phase, has a composition that is
determined through two molar ratios: H2O/Na2O and SiO2/
Na2O. It is known from studies on water glass that these ratios
strongly inuence the viscosity of the solution.27,38 In Fig. 5, the
viscosities of different sodium silicate solutions are presented.
This gure conrms that the sodium silicate solutions used
for geopolymer formulations are Newtonian uids39 but with a
viscosity which is 10 to 100 times higher than the viscosity of
f sodium silicate solution remains limited (around 0.25)
water. As a consequence, for a similar (i.e. for a similar
fmax
jamming of the system), the viscous dissipation in geopolymer
pastes is expected to be 10 to 100 times higher than that
f tion to the macroscopic viscosity of the mixture.
measured in a standard cement paste of similar ratio.
fmax
Fig. 5 Viscosity as a function of shear rate at 20  C (a) for different
H2O/Na2O silicate solutions (SiO2/Na2O ¼ 1.06) (b) for different SiO2/
Na2O silicate solutions (H2O/Na2O ¼ 16).
1138 | Soft Matter, 2014, 10, 1134–1141
Paper
3.3.2. Evaluation of the solid fraction contribution. To
isolate the contribution of the solid fraction to the viscosity, we
consider here the relative viscosity, which is the ratio between
the effective viscosity of the geopolymer and the viscosity of the
sodium silicate solution. In this way, one can focus exclusively
on the solid fraction contribution as shown in eqn (2). This
relative viscosity is shown in Fig. 6 for cement and geopolymer
pastes with varying solid volume fractions.
Fig. 6 shows that the relative viscosity diverges as the solid
volume fraction f approaches the one corresponding to the
maximum packing fraction. The exact value of this maximum
packing fraction was difficult to measure as, to the best of our
knowledge, no appropriate procedure has been developed for
that purpose.29 The results of Fig. 6, however, suggest that the
MK powder has a lower fmax in the range of 0.5–0.6 (i.e. lower
packing properties) in comparison with deocculated cement
powder (in the range of 0.6–0.7). This could be rationalized by
the fact that MK grains have a plate shape, which is very dele-
terious for ow and for packing properties.40 As a consequence,
for the same solid volume fraction, the grain contribution to the
viscosity is expected to be much higher for geopolymers than for
cement paste. However, it is interesting to note that when
chemical ratios of Al2O3/SiO2, Na2O/Al2O3 and H2O/Na2O are
selected to provide optimal strength properties to the hardened
geopolymer, the volume fraction of MK introduced in the
compared to the one used in typical cement pastes (>0.4). As a
consequence, even if MK's fmax is small, the limited volume of
solid grains in the suspension will reduce the grain contribu-
Another aspect of rigid grain contribution to the macro-
scopic rheological behaviour of a suspension lies in the direct
frictional contacts between particles. In Fig. 7 this point is
highlighted by plotting for both cement and geopolymer pastes
the ratio between normal and shear stress when both materials
are sheared at 100 s1. This comparison is made at identical
f/fmax, which in terms of contact contribution is a far more
relevant parameter than f. The normal to shear stress ratio is
Fig. 6 The relative viscosity as a function of the solid fraction for a
geopolymer paste (H2O/Na2O ¼ 15) and a deflocculated cement paste
with 0.5% superplasticizer (SP).
This journal is © The Royal Society of Chemistry 2014
Paper
Fig. 7 Ratio between normal stress and shear stress as a function of
time for geopolymer paste (H2O/Na2O ¼ 15, f ¼ 0.3 f/fmax  0.7) and
for a standard cement paste (f ¼ 0.47 and f/fmax  0.7) at room
temperature at 100 s1.
oen considered in the literature as a good indicator of the
contribution of frictional direct contacts between non-deform-
able grains. In geopolymer pastes, this ratio is one hundred
times lower than that commonly measured in the case of
cement pastes, which conrms the fact that very low to no grain
contribution can be measured. This, moreover, suggests that
even if some temporary contacts do occur between MK grains,
there does not exist any percolated sustained contact network
within the system. As a consequence, from the point of view of
rheology, a geopolymer paste, in strong contrast to a cement
paste, does not behave as a granular suspension since its
viscosity is nearly exclusively controlled by the viscosity of the
interstitial uid h0 and not by the details of the direct frictional
contacts between inclusions.
4. Conclusions and perspectives
In this study, it is shown that:
- Colloidal interactions between MK grains suspended in an
alkaline silicate solution are negligible and only low energy
interactions between grains in the presence of an interstitial gel
are at the origin of the small yield stress of the suspension.
- Hydrodynamic viscous dissipation in the Newtonian highly
viscous sodium silicate solution is the dominating mechanism
at the origin of the macroscopic viscosity of the suspension.
- Direct frictional contact contribution can be neglected as
despite the fact that the maximum packing fraction of a MK
powder is low, in a geopolymer the standard solid volume
fraction needed to achieve adequate hardened properties
f consequence, ratios are constrained to 10 to 20 for H2O/Na2O
maintains the ratio in the semi-dilute regime.
fmax
Table 3 Summary of different contributions
Yield stress
MK-based geopolymer paste Near 0 Pa weak particle interactions
Typical cement paste Near 10–20 Pa van der Waals
attractive interactions
This journal is © The Royal Society of Chemistry 2014
Soft Matter
Geopolymers can therefore be seen as a semi-dilute
suspension of non-Brownian, non-colloidal non-deformable
particles in a highly viscous Newtonian uid.
The summary of the comparison of this system with stan-
dard cement pastes is presented in Table 3.
The technical options to alter the rheology of the mixture will
then be completely different from those used for cement.
(1) Grains and interfaces: In cement paste, the most
commonly used technique to control the rheology is the use of
deocculating agents.41–44 They actually reduce the magnitude
of the attractive colloidal interactions between cement particles
by being selectively adsorbed on the surface of the cement
grains in order to create a steric effect.11 For our model system of
pure MK-based geopolymers, this action is of no interest as no
colloidal interaction between grains of MK exists. The only weak
interaction, responsible for the small elastic modulus at the
early stage, is the dissolution and precipitation reaction of
alumina silicate gel between grains.36 Reducing the precipita-
tion of this gel would prevent the polymerisation reaction,
which in turn would delay setting.
Polymer admixtures can also impact the rheology though
other physical means. First, the polymers adsorbed on the
grains can create bridging forces that are attractive forces, but
this effect increases the viscosity and the yield stress of the
system. Also, polymers that are not adsorbed remain in the pore
solution and can cause an increase of its viscosity. They then
create lubrication forces between the grains, but this is not
necessary in model geopolymer systems as direct contact effects
are very limited. Polymers in solution can also induce depletion
forces. Depletion forces are mainly attractive, but the occur-
rence of repulsive depletion forces has also been reported.45
However, the high signicance of these repulsive depletion
forces in cement pastes is still under discussion and, further-
more, are limited to the case of large polymers which also have
the effect of increasing the viscosity of the interstitial uid.
Consequently, we do not expect the use of traditional
admixtures to reduce the viscosity of model MK based
geopolymers.
As a word of caution, it must be understood nevertheless that
these conclusions cannot be extrapolated to y-ash and slag
based geopolymers in which some colloidal interactions can
exist due to the presence of calcium.
(2) Liquid viscosity: As discussed above, the only parameter
which controls the properties of the geopolymer is the viscosity
of the alkaline solution. This viscosity is controlled by the
chemical proportions of species in the solution, which are
imposed in order to achieve adequate hardened properties. As a
Solution viscosity Solid fraction
10 to 100 times higher than water f  0.2–0.3, fmax  0.5–0.6 for usual
geopolymers
Water 0.001 Pa s f  0.4–0.5, fmax  0.6–0.7
Soft Matter, 2014, 10, 1134–1141 | 1139
Soft Matter
Fig. 8 Viscosity of silicate solution depending on the cation for a
formulation H2O/Na2O ¼ 20 and SiO2/M+2O ¼ 1.6 at 20  C (M+ ¼ K+ or
Na+).
Fig. 9Range of evolution of shear stress with the shear rate for the
geopolymer and OPC formulation range.
and 1 to 2 for SiO2/Na2O.46–48 However, one option which has not
been studied here is to change the nature of the alkali ion.27
Actually, changing from sodium to potassium silicate reduces
the viscosity by 5 for an equivalent formulation (Fig. 8).
It would then be possible to produce geopolymers with a
lower viscosity when potassium ions are used instead of
sodium.46 However, replacing sodium with potassium increases
signicantly the price of the mixture, thus limiting its applica-
tions to niche applications such as re or acid resistant
ceramics48 or nuclear waste containment.46
If no changes of alkali are made, in Fig. 9, we show the
rheological behaviour limit we can meet when formulating a
geopolymer binder with adequate hardened properties and
compare it with the range of OPC formulations. Due to its
rheological behaviour, the MK-based geopolymer does not seem
suitable as a binder in mortar or adhesive applications where a
high yield stress is needed49 or in applications requiring fast
ows such as pumping or spraying because of its high viscosity.
Applications where it could be substituted for cement are self-
compacting or self-levelling applications such as oor screed
where a low yield stress allows the material to ow and spread
under its own weight while the high viscosity improves the
stability of the coarse inclusions during casting.
Once again, as a word of caution before generalizing the
results of the present study, it is important to note that the
mechanisms governing the formation of an alumina-silicate
gel at an early stage are strongly dependent on the chemical
composition of the mixture and therefore the precursor
1140 | Soft Matter, 2014, 10, 1134–1141
Paper
selected. The presence of Ca (expected for y-ash or slag based
geopolymers) could dramatically change the chemistry of the
system and the interaction forces between particles. The
presence of divalent cations can result in attractive ion corre-
lation forces such as the one present between C–S–H parti-
cles50,51 whereas monovalent cations only induce repulsive
forces.52,53 We might then expect to enhance the colloidal
interactions between grains as well. In that case, the use of
plasticizers would be of certain interest as long as the chemical
stability of the organic plasticizer can be ensured in alkaline
solutions.25
References
1 P. Coussot, So Matter, 2007, 3, 528–540.
2 M. Schneider, M. Romer, M. Tschudin and H. Bolio, Cem.
Concr. Res., 2011, 41, 642–650.
3 G. H. Tattersall and P. F. G. Banll, The rheology of fresh
concrete, Pitman Advanced Publishing Program, 1983.
4 N. Roussel, Understanding the Rheology of Concrete,
Woodhead Pub Ltd, 2012.
5 J. Yammine, M. Chaouche, M. Guerinet, M. Moranville and
N. Roussel, Cem. Concr. Res., 2008, 38, 890–896.
6 F. de Larrard and T. Sedran, Cem. Concr. Res., 2002, 32, 1699–
1704.
7 P. Rossi, Cem. Concr. Compos., 1992, 14, 3–16.
8 L. Martinie, P. Rossi and N. Roussel, Cem. Concr. Res., 2010,
40, 226–234.
9 P. Coussot, L. Tocquer, C. Lanos and G. Ovarlez, J. Non-
Newtonian Fluid Mech., 2009, 158, 85–90.
10 C. Hu and F. de Larrard, Cem. Concr. Res., 1996, 26, 283–294.
11 R. J. Flatt, Mater. Struct., 2004, 37, 289–300.
12 N. Roussel, A. Lemaı̂tre, R. J. Flatt and P. Coussot, Cem.
Concr. Res., 2010, 40, 77–84.
13 R. J. Flatt, I. Schober, E. Raphael, C. Plassard and
E. Lesniewska, Langmuir, 2009, 25, 845–855.
14 P. Capros, N. Kouvaritakis and L. Mantzos, Economic
Evaluation of Sectoral Emission Reduction Objectives for
Climate Change Top-down Analysis of Greenhouse Gas
Emission Reduction Possibilities in the EU, European
commission, 2001.
15 P. Friedlingstein, R. A. Houghton, G. Marland, J. Hackler,
T. A. Boden, T. J. Conway, J. G. Canadell, M. R. Raupach,
P. Ciais and C. Le Quéré, Nat. Geosci., 2010, 3, 811–812.
16 UNSTATS, United Nations Stat. Div., Springer, 2010.
17 R. J. Flatt, N. Roussel and C. R. Cheeseman, J. Eur. Ceram.
Soc., 2012, 32, 2787–2798.
18 C. Shi, A. F. Jiménez and A. Palomo, Cem. Concr. Res., 2011,
41, 750–763.
19 E. M. Gartner and D. E. Macphee, Cem. Concr. Res., 2011, 41,
736–749.
20 A. Palomo, P. F. G. Banll, A. Fernández-Jiménez and
D. S. Swi, Adv. Cem. Res., 2005, 17, 143–151.
21 M. Palacios, P. F. G. Banll and F. Puertas, ACI Mater. J.,
2008.
22 M. Criado, A. Palomo, A. Fernández-Jiménez and
P. F. G. Banll, Rheol. Acta, 2009, 48, 447–455.
This journal is © The Royal Society of Chemistry 2014
Paper
23 M. Palacios, Y. F. Houst, P. Bowen and F. Puertas, Cem.
Concr. Res., 2009, 39, 670–677.
24 M. Palacios and F. Puertas, Cem. Concr. Res., 2005, 35, 1358–
1367.
25 M. Palacios and F. Puertas, Mater. Constr, 2010, 54, 65–86.
26 D. Marchon, U. Sulser, A. Eberhardt and R. J. Flatt, So
Matter, 2013, 9, 10719–10728.
27 A. Bourlon, PhD thesis, Université Pierre et Marie Curie, 2010.
28 A. J. G. Bourlon, J.-B. d' Espinose, E. Lecolier and
D. Pasquier, ACI- Twelh Int. Conf. Recent Adv. Concr.
Technol. Sustain. Issues, 2012, SP-289, pp. 1–9.
29 P. Coussot, Rheometry of pastes, suspensions, and granular
materials, Wiley-Interscience, Hoboken (N.J.), 2005.
30 P. Duxson, J. L. Provis, G. C. Lukey, S. W. Mallicoat,
W. M. Kriven and J. S. J. van Deventer, Colloids Surf., A,
2005, 269, 47–58.
31 M. Rowles and B. O'Connor, J. Mater. Chem., 2003, 13, 1161–
1165.
32 A. Perrot, T. Lecompte, H. Kheli, C. Brumaud, J. Hot and
N. Roussel, Cem. Concr. Res., 2012, 42, 937–944.
33 P. Coussot and C. Ancey, Phys. Rev. E: Stat. Phys., Plasmas,
Fluids, Relat. Interdiscip. Top., 1999, 59, 4445–4457.
34 H. A. Barnes and Q. D. Nguyen, J. Non-Newtonian Fluid Mech.,
2001, 98, 1–14.
35 N. Roussel, G. Ovarlez, S. Garrault and C. Brumaud, Cem.
Concr. Res., 2012, 42, 148–157.
36 A. Favier, G. Habert, J. B. d' Espinose de Lacaillerie and
N. Roussel, Cem. Concr. Res., 2013, 48, 9–16.
37 I. M. Krieger and T. J. Dougherty, Trans. Soc. Rheol., 1959, 3,
137–152.
This journal is © The Royal Society of Chemistry 2014
View publication stats
Soft Matter
38 J. G. Vail, Soluble silicates: their properties and uses, Reinhold,
1952.
39 A. Bourlon, L. Barré, J.-B. d'Espinose de Lacaillerie, D. Frot,
E. Lécollier and D. Pasquier, So Matter, 2013, submitted.
40 H. Had, PhD thesis, Université Paris-Est, 2012.
41 P.-C. Aı̈tcin, S. Jiang, B. Kim, P. Nkinamubanzi and N. Petrov,
Bull. Lab. Ponts Chaussees, 2001.
42 S. K. Agarwal, I. Masood and S. K. Malhotra, Constr. Build.
Mater., 2000, 14, 253–259.
43 P.-C. Aı̈tcin, S. L. Sarkar, M. Regourd and D. Volant, Cem.
Concr. Res., 1987, 17, 995–999.
44 C. Jolicoeur and M.-A. Simard, Cem. Concr. Compos., 1998,
20, 87–101.
45 R. I. Feigin and D. H. Napper, J. Colloid Interface Sci., 1980,
75, 525–541.
46 J. Davidovits, Geopolymer chemistry and applications,
Geopolymer Institute, 2008.
47 F. Pacheco-torgal, S. Jalali, J. Labrincha and V. John, Eco-
Efficient Concrete, Woodhead Publishing Limited, 2013.
48 J. L. Provis and J. S. J. Van Deventer, Geopolymers structure,
processing, properties and industrial applications,
Woodhead, CRC Press, Oxford, Boca Raton, FL, 2009.
49 C. Brumaud, PhD thesis, Université de Paris Est, 2011.
50 M. Yang, C. M. Neubauer and H. M. Jennings, Adv. Cem.
Based Mater., 1997, 5, 1–7.
51 R. J.-M. Pellenq and H. Van Damme, MRS Bull., 2004, 29,
319–323.
52 G. Lu and K. Wang, ACI Mater. J., 2010, 107, 12–19.
53 R. J.-M. Pellenq, J. M. Caillol and A. Delville, J. Phys. Chem. B,
1997, 101, 8584–8594.
Soft Matter, 2014, 10, 1134–1141 | 1141