Construction and Building Materials 201 (2019) 563–570

Contents lists available at ScienceDirect

Construction and Building Materials

journal homepage: www.elsevier.com/locate/conbuildmat

Drying shrinkage of alkali activated binders cured at room temperature

Faris Matalkah a, Talal Salem b,⇑, Mamoon Shaafaey b, Parviz Soroushian b
a
Department of Civil Engineering, Yarmouk University, Irbid 2110, Jordan
b
Department of Civil and Environmental Engineering, Michigan State University, East Lansing, MI 48823, USA

h i g h l i g h t s

 Alkali activated binders cured at room temperature exhibit larger drying shrinkage compared to Portland cement binder.
 The use of additives to refine the alkali activated binder formulation lower the total drying shrinkage.
 The addition of polyethylene glycol to alkali activated binder significantly reduced the total drying shrinkage.

a r t i c l e i n f o a b s t r a c t

Article history: Alkali activated binders tend to exhibit higher drying shrinkage movements when compared with
Received 19 June 2018 Portland cement. An experimental investigation was undertaken in order to assess the drying shrinkage
Received in revised form 30 December 2018 of a basic alkali activated binder versus Portland cement, and to evaluate the effectiveness of different
Accepted 31 December 2018
additives in reducing the drying shrinkage of the alkali activated binder. The results indicated that the
Available online 7 January 2019
drying shrinkage of the basic alkali activated binder was about twice that of Type I Portland cement.
Some of the additives used to refine the alkali activated binder formulation were found to lower the dry-
Keywords:
ing shrinkage of the alkali activated binder to levels that were comparable with that of Portland cement.
Alkali activation
Aluminosilicates
Explanations were presented for the mechanisms through which these additives enhance the dimen-
Drying shrinkage sional stability of the alkali activated binder. SEM, TG/DTG and heat of hydration techniques were used
Additives to gain more insight into the mechanisms of action of different additives in alkali activated binders.
Geopolymer cement Ó 2019 Elsevier Ltd. All rights reserved.

1. Introduction
Alkali-activated binders, also known as geopolymers, have been
subject of growing investigations due their desired engineering
properties as well as reduced carbon footprint and energy content
when compared with Portland cement [1–4]. Some distinguishing
features of geopolymer binders, when compared with those based
on Portland cement, include: (i) high rate of strength development
[5]; (ii) fire resistance [6]; (iii) resistance to acids and sulfates [7];
and (iv) freeze-thaw durability [8]. Past investigations have indi-
cated that alkali-activated binders are prone to experiencing large
drying shrinkage movements [9–11]. Improvement of the dimen-
sional stability of alkali-activated binders would be an important
step for facilitating their broader market acceptance.
The phenomena involved in drying shrinkage of alkali-activated
binders have not been fully understood. It has been suggested that,
unlike Portland cement, water in geopolymer binders is not incor-
porated directly into the gel product; a small percentage of the
⇑ Corresponding author.
E-mail address: salemtal@msu.edu (T. Salem).
mixing water remains as interstitial water within the gel [12].
The excess of unbound or free water in the alkali-activated binder
can be removed from the hardened paste exposed to relatively low
relative humidity at ambient temperature. Excess moisture loss
under relatively normal environmental conditions can result in
extensive shrinkage cracking of geopolymer binders.
It is well established that the capillary and gel pore size distri-
butions have direct effects on the drying shrinkage of cementitious
binders [13,14]. It has been reported that, in calcium-rich alkali
activated binders, two hydration products are formed: calcium sil-
icate hydrates gel similar to that in Portland cement hydrates, but
with lower Ca-to-Si ratios, and a silica gel. The silica gel reportedly
contains high uncombined water content that could be eliminated
during the drying process, causing substantial shrinkage [15]. It
has been found that heat curing could reduce the drying shrinkage
of alkali-activated binders. This could be attributed to the partial
consumption of unbound water towards chemical reactions. For
alkali-activated binders cured at ambient temperature, loss of this
unbound water over time can cause significant drying shrinkage
strains, especially during the first two weeks [16,17].

https://doi.org/10.1016/j.conbuildmat.2018.12.223
0950-0618/Ó 2019 Elsevier Ltd. All rights reserved.
564 F. Matalkah et al. / Construction and Building Materials 201 (2019) 563–570

The work reported herein investigated the drying shrinkage of 100

alkali aluminosilicate binders cured at room temperature. Several 90
additives were used at different dosages to control drying shrink- Silica sand
80
age. These additives were selected to reduce the uncombined
70 Coarse
water, refine the pore structure, reduce the surface tension and
aggregate

Passing, %
decrease the water demand of the mortar mixture. 60
50
40
2. Materials and methods
Natural
30
sand
2.1. Materials 20
The alkali activated cement used in this investigation was developed in a previ- 10
ous work conducted by the same authors [18]. The aluminosilicate precursors used 0
in the mechanochemical process were coal fly ash (45 wt%), ground granulated
0.1 1 10 100 1000 10000 100000
blast furnace slag (25 wt%), and albite (15 wt%). The (dry) alkalis used as raw mate-
rials were sodium hydroxide (3 wt%), sodium silicate (7 wt%), and calcium oxide Parcle size, μm
(4 wt%). Sodium tetra-borate (Borax) was also added (1 wt%) to the cement formu-
lation to increase its set time. Fig. 1. Particle size distributions of silica sand, natural sand and coarse aggregates.
The ratio of SiO2/Na2O in the alkali activated cement is 3.96. The laboratory
grade alkalis purchased in powder form at 97–98% purity from Sigma Aldrich.
The specific gravities of sodium hydroxide, sodium silicate, quick lime and sodium 100
tetra-borate are 2.14, 1.39, 3.31 and 1.73 at 25 °C, respectively. Type I Portland Alkali acvated
cement was used as a reference for evaluation of the results produced with the 90 cement
alkali activated cement. The chemical compositions and Blaine fineness of the alkali 80
activated cement and the Portland cement used in this investigation are presented Portland
in Table 1. 70
cement
Standard silica sand (ASTM C778) with specific gravity of 2.65 was used to pre-
Passing, % 60
pare mortar mixtures. In concrete, crushed limestone with a maximum particle size
of 12.5 mm was used as coarse aggregate (specific gravity of 2.55), and natural sand 50
(specific gravity of 2.64) with maximum particle size of 4.75 mm as fine aggregate. 40
The particle size distributions of the silica sand, natural sand and coarse aggregates
are presented in Figs. 1 and 2 present the particle size distributions of the alkali 30
activated cement and Portland cement. 20
The additives introduced below were evaluated for reducing the drying shrink-
age of alkali activated cement. 10
0
 Polyethylene glycol (PEG) powder with molecular weight of 600 g/mol was
used as shrinkage reducing admixture at 2.5% by weight of cement [19].
0.1 1 10 100 1000 10000
 Gypsum powder (CaSO4. ½ H2O) was used at 5% by weight of cement to induce Parcle size, μm
expansion (for counteracting drying shrinkage) via formation of ettringite nee-
dles [20]. Fig. 2. Particle size distributions of the alkali activated cement and Portland
 Non-densified silica fume with 0.2 lm mean particle size, 15 m2/g specific area cement.
and 105% 7-day pozzolanic activity index was used at 5% by weight of cement
to enhance the microstructure of aluminosilicate hydrates by producing denser
hydration products [21,22].
 Limestone powder with mean particle size of 2 lm was used at 5% by weight of were stored in moist room in molds for 24 h. They were then
cement to produce a more uniform microstructure by acting as nucleation sites demolded and cured in lime-saturated water for 48 h. Subse-
for hydration products [23].
quently, the specimens were removed from water, wiped with
 Latex polymer (TRINSEO modifier A NA) with 47 wt% solid, pH of 10 and density
of 1.01 g/ml was used at 5% by weight of solid to delay moisture loss [24]. damp cloth and immediately a length comparator reading was
obtained for each specimen. The specimens were then stored at
All additives were added to the mixing water of mortar. The dosages noted 22 ± 2 °C and 50 ± 4% relative humidity, with length measurements
above were selected via trial-and-adjustment efforts to preserve the mechanical made over time up to 60 days.
properties of the control mix (without any additives). For the purpose of weight loss measurement versus time, con-
crete specimens were prepared with two different thicknesses of
3. Methods 15 and 50 mm, with 50 mm square planar dimensions. The con-
crete specimens were prepared with either alkali activated cement
The drying shrinkage of the mortar specimens was measured or Portland cement, limestone coarse aggregates of 12.5 mm max-
following the ASTM C596 procedures. This test method determines imum particle size, and natural sand of 4.75 mm maximum parti-
the change in length on drying of mortar bars made with a hydrau- cle size. The coarse aggregate: natural sand: cement: water weight
lic cement and graded standard sand. Drying shrinkage measure- ratios were 2: 1: 0.5: 0.25. The specimens were cured at room tem-
ments were performed on prismatic mortar bar specimens with perature and 95 ± 5% relative humidity for 24 h inside molds and
25  25 mm cross section and 285 mm length. Mortar mixture were then immersed in lime-saturated water for 48 h. The test
was prepared with sand-to-cement ratio of 2.0, and the water-to- started immediately thereafter (at 72 h of age). The specimens
cement ratio was adjusted to produce a fresh mix flow of used for weight loss measurement until the weight stabilized (after
110 ± 5% (as described in ASTM C596). After casting, specimens 21 days).

Table 1
Chemical compositions (wt%), Blaine fineness and specific gravity for the alkali activated cement and Portland cement.

SiO2 CaO Al2O3 Fe2O3 MgO K2O Na2O SO3 Blaine Fineness, cm2/g S.G.
Alkali activated cement 35.2 28.1 13.6 4.03 3.73 1.14 8.89 0.53 3960 3.09
Portland cement 20.1 62.5 4.80 3.20 2.90 0.1 0.3 3.15 3500 3.15
 F. Matalkah et al. / Construction and Building Materials 201 (2019) 563–570 565

Scanning electron microscope (SEM) images of the hydration Days

products were captured using a JELO-5000. SEM in order to assess 0
the microstructural attributes and microcrack conditions of 0 10 20 30 40 50 60
cement pastes. The specimens were first coated with gold- -300
palladium alloy using a DESK II sputter coater, and then imaged
-600 Portland cement
in high-vacuum mode at an accelerating voltage of 10 or 15 kV.

Microstrain
Thermogravimetric analysis was performed on the cured paste -900
at 3, 7 and 28 days of age using a Perkin Elmer TGA 4000 at a heat-
ing rate of 15 °C/min from 50 to 950 °C under nitrogen atmosphere. -1200
The heat of hydration of the cement was measured by perform- Alkali acvated
ing the calorimetry test (ASTM C1679) using an I-cal 2000 HPC -1500 cement
calorimeter. For performance of this test, cement was mixed man-
-1800
ually with water for 30 s at a water/cement ratio of 0.35. The fresh
paste was placed inside the calorimeter, and its heat release was -2100
monitored over 24 h.
Fig. 4. Drying shrinkage versus time for alkali activated and Portland cement.

4. Results and discussion

4.1. Mass loss
Fig. 3 compares the mass loss versus time test data for concrete
specimens prepared using Portland cement or alkali activated
cement with different thicknesses. At 15 mm thickness, the weight
loss for concrete specimens prepared with alkali activated cement
is higher (5.5 wt%) when compared with those prepared with Port-
land cement (4%). At 50 mm thickness, Concrete specimens experi-
enced reduced weight loss; the total weight loss was more (3.8%)
for concrete prepared with alkali activated cement than that pre-
pared with Portland cement (3.3%). More work is needed to explain
these trends that may have been produced by the differences in the
initial moisture contents, chemical binding of water into the struc-
ture of hydrates, barrier qualities of the bulk material and surface
qualities for the two cement types considered in this investigation.
4.2. Drying shrinkage
Fig. 4 compares the measured values of drying shrinkage for
Portland and the alkali activated cement versus time. Alkali acti-
vated cement produced higher drying shrinkage (about double)
when compared with Portland cement. Shrinkage of Portland
cement stabilized after 45 days when it was less than 1000
micro-strain. Shrinkage of the alkali activated cement started to
stabilize after about 60 days when it had reached about 2000
micro-strain. There could be two reasons for the significantly
higher shrinkage of the alkali activated cement when compared
with Portland cement: (i) alkali aluminosilicate hydrates incorpo-
rate less chemically combined water, and thus leave higher quan-
tities of uncombined water that would be lost during drying and
cause shrinkage [9]; and (ii) while the interconnected pores among
Portland cement hydrates cover the size range of 5–1200 nm, those
in alkali activated cement hydrates generally range from 10 to
200 nm [25].The higher meso-pre content in alkali aluminosilicate
binders (74–82%) induces higher capillary stress and therefore
greater shrinkage during drying [26].
Fig. 5 compared the SEM images of Portland and the alkali acti-
vated cement pastes. Crystalline Portlandite and amorphous cal-
cium silicate hydrate could be identified in the hydrated Portland
cement paste. In the case of alkali activated cement, a relatively
dense gel of alkali aluminosilicate hydrate embodied unreacted
fly ash particles. Vacuum drying of the pastes prior to SEM imaging
caused more notable microcracking of alkali aluminosilicate
hydrates when compared with the hydration products of Portland
cement.
The difference in CASH phase morphology is attributed to many
factors, such as the water content, calcium/silicon ratio, degree of
hydration and the pore size distribution. But due to its structural
complexity at nanoscale, CASH is still not completely understood.
The pore structure of both cements is very different: alkali cement
has lesser pore volume than OPC for pores larger than 50 nm, but
vice versa for pores finer than 50 nm. Since the nanogranule in
alkali cement is finer and denser, its impermeability increases
and a thin hydration layer permits more dehydration [27]. Calcium

100 100
PC concrete PC concrete
99 99
AAC concrete AAC concrete
Remaining mass, %

Remaining mass, %

98 98

97 97

96 96

95 95

94 94
0 100 200 300 400 500 0 100 200 300 400 500
Hours Hours
(a) 15-mm thick specimens (b) 50-mm thick specimens

Fig. 3. Mass loss versus time for concrete specimens prepared with Portland cement (PC) and Alkali activated cement (AAC).
566 F. Matalkah et al. / Construction and Building Materials 201 (2019) 563–570

Fig. 5. SEM images of Portland cement paste (a) and the alkali activated cement paste (b).

Crack width of
50 μm
Surface cracks

500 μm

500 μm

(a) 90% Relave humidity (b) 50% Rela ve humidity

Fig. 6. Optic microscope images of alkali activated cement pastes cured in different humidity conditions.

Days
0
0 10 20 30 40 50 60

-300
Control Silica fume
PEG Latex
-600
Limestone Gypsum
Microstrain

-900

-1200

-1500

-1800

Fig. 7. Fractured section of an alkali activated cement mortar bar specimens dried
-2100
for 28 days at room temperature and 50% relative humidity prior to fracturing.

Fig. 8. Drying shrinkage versus time for alkali activated mortar without and with
different additives.

and silicon ratios impact the cluster morphologies in diverse ways:

the former contributes to an ordered structure while the latter is
known to increase the amorphous state of the structure. In alkali
cement, there are nearly no crystal hydration products, which
attributes towards the stability of intrinsic CASH phases so it has
high shrinkage as compared to the OPC [28].
Since the curing reactions of alkali activated cement releases
some of the water involved in reactions, supply of moisture during
curing may be less critical when compared with Portland cement.
Curing at room temperature and 50% relative humidity (when
compared with 90% relative humidity), however, was observed to
produce surface microcracks (Fig. 6). Another consideration is the
high barrier qualities of the alkali aluminosilicate hydrates, that
could reduce the rate of moisture loss from alkali activated cement
mortar and concrete, thereby intensifying the moisture gradient
and thus the internally restrained drying shrinkage stresses near
surface. Fig. 7 shows the fractured section of an alkali activated
 F. Matalkah et al. / Construction and Building Materials 201 (2019) 563–570 567

Fig. 9. SME for alkali activated cement paste without additive and with different additives.

cement mortar specimen that was dried at room temperature and
50% relative humidity for 28 days prior to fracturing to expose its
interior. The interior core of this prismatic specimen is observed
to be still wet after 28 days of air-drying.
4.3. Effects of additives
Fig. 8 presents the drying shrinkage test results for alkali acti-
vated mortar with different additives. Addition of 5 wt% gypsum
decreased the shrinkage movements after 60 days of drying by
about 8%. This could be due to the increased dissolution of
aluminum compounds in the presence of gypsum [29] (mainly in
Ca-dominated systems) which could yield more crosslinked alkali
aluminosilicate hydrates, the contributions of calcium from gyp-
sum towards more thorough room-temperature curing, alteration
of the pore system characteristics [30], and the formation of ettrin-
gite with expansive effects that partially compensate for the drying
shrinkage effects. It should be noted that addition of gypsum to
alkali activated aluminosilicate has been reported to reduce the
flowability of the concrete mix and thus reduce the compressive
strength [31,32].
The addition of limestone powder and latex reduced the total
shrinkage by 17% and 26%, respectively. Silica fume, on the other
hand, increased the 60-day drying shrinkage by 11%. The trends
noted here for limestone powder and silica fume are consistent
with those reported in the literature for the shrinkage movements
568 F. Matalkah et al. / Construction and Building Materials 201 (2019) 563–570

100 DTG 0.0 involves extensive vacuum-drying. Formulation of the hydraulic

cement with silica fume produces a denser paste with reduced
95 620 °C
760 °C
-0.5 microcracking (Fig. 9b). This could be attributed to the role of silica
410 °C
fume particles as micro-fillers, and also as heterogeneous nucle-

Derivave weight, %/min

390 °C

90 -1.0 ation sites for hydration products [33]. Formulation of the cement
Limestone
Control with PEG is observed in (Fig. 9c) to produce a paste with larger
Weight, %

pores and minimal microcracking after extensive vacuum drying.

85 Silica fume -1.5
The use of gypsum produced a tendency towards ettringite forma-
tion, but the extent of microcracking was significant (Fig. 9d).
80 -2.0
Gypsum Addition of limestone powder along with gypsum could lead to
Latex stabilize the ettringite, leading to minimize the microcracks forma-
75 -2.5 tion. The use of latex and limestone powder led to some reduction
PEG TG
175 °C in the extent of microcracking after vacuum drying (Fig. 9e and f,
70 -3.0 respectively). The use of limestone powder seemed to produce a
0 100 200 300 400 500 600 700 800 900 1000 denser microstructure. The addition of fine limestone could lead
Temperature, °C to form hemi- and monocarboaluminate AFm phases, where they
can hold more crystal water and hence will cause less shrinkage
Fig. 10. TG/DTG test data for alkali activated cement pastes prepared without
(control) and with different additives. [37].

4.4. Thermogravimetric analysis

of hydraulic cements with predominantly alkali aluminosilicate
chemistry [33]. The reduction of drying shrinkage by latex modifi- Fig. 10 presents the TG/DTG test data for hydrated cement
cation could be attributed to the blocking of capillary pores with pastes prepared without (control) or with different additives. Most
latex polymer that reduces the interconnectivity of pores and thus of the weight loss occurred for all specimens at temperatures
hinders moisture movements [34]. The resulting reduction in the below 200 °C; this can be attributed to the loss of physically
rate of moisture loss would translate into reduced drying shrink- adsorbed water and some fewer stable hydrates. This trend has
age. The drop in drying shrinkage with the addition of limestone been observed for the hydrates of cements with similar chemistry
powder could be attributed to the role of limestone powder parti- [18]. A rather continuous mass loss was noted within the 200–
cles as nucleation sites for hydration products, some precipitation 950 °C temperature range, which can be attributed to dehydration
of CAH that speeds up the hydration of process [35] to reduce the of the hydration products [38,39] and hydrotalcite-like phases
uncombined water content that can cause drying shrinkage, and [39,40], and decomposition of carbonates [41]. The addition of dif-
the micro-filling action of limestone powder particles. The addition ferent additives produced new DTG peaks. In the presence of poly-
of 2.5% PEG reduced the total shrinkage by 44%, which makes the ethylene glycol the new peak occurred within the 300–420 °C
activated cement in par with Portland cement as far as drying temperature range [42,43]. With latex polymer, the new peak
shrinkage is concerned. The effectiveness of PEG in reduction of occurred at about 410 °C [44-46]. Both these peaks can be attribu-
drying shrinkage could be attributed to: (i) the drop in surface ten- ted to the decomposition of the organic polymer inclusions in the
sion of water held in fine pores, which reduces the stress caused by context of the inorganic polymer matrix. When limestone powder
water evaporation from these pores [36]; (ii) refinement of the was added, a significant DTG peak appeared around 760 °C, which
pore size distribution in favor of pores with 0.1–1.0 lm size that can be attributed to the decomposition of calcium carbonate [47].
exhibit a capillary stress much lower than that in smaller pores; With silica fume used as the additive, weight loss within the 310–
this redistribution of the pores is due to the decrease of the capil- 650 °C temperature range was lower than that of the control paste,
lary stress of the water that PEG induces during the mixing process which could be due to the formation of thermally stable hydration
[19]. products in the presence of silica fume [48]. With the addition of
Fig. 9 shows the SEM images of hydrated pastes of cements for- gypsum, greater weight loss was noted below 200 °C, which can
mulated with different additives. The control cement formulated be attributed to the dehydration process of gypsum. Above
without any additives produces a paste that experiences extensive 200 °C, however, the paste incorporating gypsum exhibited
microcracking (Fig. 9a) after a sample preparation process that improved thermal stability. This can be attributed to the role of

100 Control PEG 400

Total heat release, Kcal/kg
Heat of hydraon, mW/g

Gypsum Limestone
80
300
Silica fume Latex
60
200
40
Control PEG
100 Gypsum Limestone
20
Silica fume Latex
0 0
0 20 40 60 0 5 10 15 20 25
minutes Hours
(a) Heat release versus me (b) Total heat release

Fig. 11. Heat of hydration for alkali activated cement paste without an additive (control), and with different additives.
 F. Matalkah et al. / Construction and Building Materials 201 (2019) 563–570
gypsum in enhancing the hydration (geopolymerization) reactions
[49]; the presence of sulfate can improve the release of aluminum
to solution, which enhances the degree of crosslinking [50] and the
thermal stability of hydration products.
4.5. Reaction kinetics
The heat of hydration test data collected for alkali activated
cement pastes prepared without and with different additives are
presented in Fig. 11. For the cement incorporating gypsum, the pri-
mary peak at few minutes was followed by a minor and broad peak
at about 50 min (Fig. 11a). All other hydraulic cements exhibited
only one peak at few minutes, after which the heat release dimin-
ished gradually over several minutes. The main peak appearing
after few minutes was most pronounced for the cement incorpo-
rating limestone powder and gypsum, followed by the cements
incorporating silica fume, latex polymer, no additive (control),
and PEG. The pronounced peak observed with limestone powder
and silica fume could be caused by their high specific surface area
that facilitates nucleation of hydration products. As far as the
cumulative heat release is concerned, the highest values were
obtained for cements incorporating limestone powder and gyp-
sum, followed by cements incorporating silica fume, latex, no addi-
tive (control), and finally PEG. While the inorganic additives raised
the cumulative heat release, the organic additives considered here
either retained or reduced the cumulative heat release associated
with the hydration reactions.
The effect of gypsum on increasing the rate of heat release can
be attributed to ettringite formation [30,51]. In the case of lime-
stone powder, this effect could have been caused by the nucleation
of hydration products on its relatively high specific surface area,
and the stimulation of some chemical reactions by limestone pow-
der [52,53]. The distinctly fine particle size and reactivity of silica
fume could be used to explain its effects on raising the rate of heat
release as well as the cumulative heat released [54,55]. The effects
of latex can be attributed to the blocking of capillary pores with
latex polymer that hinders moisture movements [34] which
enhance the reaction between the alkali aluminosilicate hydrates
yield to more CAAASAH gel that release more heat [50].
The addition of PEG is observed in Fig. 11 to reduce the rate of
heat release, which lowers the intensity of the initial peak; a weak
and broad peak also appears at about 50 min in the presence of
PEG. This could be expected because PEG makes the capillary pore
surfaces hydrophobic (thus reducing the surface tension of water
in these pores [56]), which slows down the exothermic hydration
reactions. The presence of silica fume is observed in Fig. 11 to
accelerate the exothermic hydration reactions. Silica fume particles
are particularly fine, and its structure is amorphous [54,55,57]. It is
thus expected to accelerate the exothermic reactions involving sil-
ica that lead to precipitation of alkali aluminosilicate hydrates.
5. Conclusions
Alkali activated binders cured at room temperature tend to pro-
duce larger drying shrinkage movements than Portland cement.
Restrained shrinkage cracking can thus be a concern when Port-
land cement is replaced with alkali activated binders. An experi-
mental investigation was undertaken to assess the drying
shrinkage of a basic alkali activated (alkali aluminosilicate) cement
formulation, and to evaluate the effects of incorporating different
additives into the formulation on the drying shrinkage and other
characteristics of the alkali activated cement. The following con-
clusions were derived based on the experimental data generated
in the project.
569
 Refinement of the cement size distribution by adding 5 wt% sil-
ica fume (with relatively fine particle size) led to a slight rise in
the drying shrinkage of the alkali activated cement. The addi-
tion of PEG, on the other hand, produced a significant reduction
of drying shrinkage when used at 2.5 wt% concentrations. Mod-
ification of alkali activated cement with PEG produced drying
shrinkage movements that were comparable with those of Type
I Portland cement. Addition of 5 wt% gypsum or limestone pow-
der decreased the shrinkage movements of alkali activated
cement. The addition of and latex reduced the total shrinkage.
The reduction of drying shrinkage by latex modification could
be attributed to the blocking of capillary pores with latex poly-
mer that hinders moisture movements, and thus reduce the
drying shrinkage movements. The altered particle size distribu-
tion and pore structure could also contribute towards improve-
ment of dimensional stability.
 The use of inorganic additives for shrinkage reduction raised the
cumulative heat release associated with exothermic hydration
reactions. The rise in heat release resulting from the addition
of gypsum could be attributed partly to ettringite formation.
With limestone powder, the rise in cumulative heat release
could be attributed partly to the role of limestone powder par-
ticles with high specific surface area as nucleation sites for
hydration products, that could further advance the exothermic
reactions. A similar reasoning could be used to explain the role
of silica fume in rising the cumulative heat release. The use of
organic additives either retained or reduced the cumulative
heat release. The addition of PEG slowed down the hydration
reactions probably because it generated hydrophobic pore
surfaces.
 Thermogravimetric analysis of cement pastes incorporating dif-
ferent additives provided further support for the physical and
chemical effects of these additives that are used in this investi-
gation to explain their effects on drying shrinkage.
 This work investigated on the effect of several additives on the
drying shrinkage of alkali activated cement binder. However,
the authors recommend studying the effect on the mechanical
properties of alkali activated binders. Noting that if additives
can reduce the shrinkage but at the same time sacrifice the
mechanical strength of concrete.
 As a final conclusion, improvement of the dimensional stability
in term of drying shrinkage of alkali activated binder would be
an important step for facilitating its broader market acceptance.
Conflict of interest
The authors declare that they have no conflict of interest.
Acknowledgment
The authors are thankful to the U.S. Department of Energy for
its financial support of the project under Award No. DE-
SC0015197. The authors declare that they have no conflict of
interest.
References
[1] J.S. van Deventer et al., Chemical research and climate change as drivers in the
commercial adoption of alkali activated materials, Waste Biomass Valorization
1 (1) (2010) 145–155.
[2] A. Palomo, M. Grutzeck, M. Blanco, Alkali-activated fly ashes: a cement for the
future, Cem. Concr. Res. 29 (8) (1999) 1323–1329.
[3] D.M. Roy, Alkali-activated cements opportunities and challenges, Cem. Concr.
Res. 29 (2) (1999) 249–254.
[4] F. Pacheco-Torgal et al., Durability of alkali-activated binders: a clear
advantage over Portland cement or an unproven issue?, Constr Build. Mater.
30 (2012) 400–405.
570 F. Matalkah et al. / Construction and Building Materials 201 (2019) 563–570
[5] A. Fernández-Jiménez, J. Palomo, F. Puertas, Alkali-activated slag mortars:
mechanical strength behaviour, Cem. Concr. Res. 29 (8) (1999) 1313–1321.
[6] H.Y. Zhang et al., Development of metakaolin–fly ash based geopolymers for
fire resistance applications, Constr. Build. Mater. 55 (2014) 38–45.
[7] F. Puertas et al., Alkaline cement mortars. chemical resistance to sulfate and
seawater attack, Materiales de Construccion 52 (267) (2002) 55–71.
[8] Y. Fu, L. Cai, W. Yonggen, Freeze–thaw cycle test and damage mechanics
models of alkali-activated slag concrete, Constr. Build. Mater. 25 (7) (2011)
3144–3148.
[9] C. Kuenzel et al., Ambient temperature drying shrinkage and cracking in
metakaolin-based geopolymers, J. Am. Ceram. Soc. 95 (10) (2012) 3270–3277.
[10] Z. Zuhua et al., Role of water in the synthesis of calcined kaolin-based
geopolymer, Appl. Clay Sci. 43 (2) (2009) 218–223.
[11] C.D. Atisß et al., Influence of activator on the strength and drying shrinkage of
alkali-activated slag mortar, Constr. Build. Mater. 23 (1) (2009) 548–555.
[12] D. Perera et al., Influence of curing schedule on the integrity of geopolymers, J.
Mater. Sci. 42 (9) (2007) 3099–3106.
[13] Bazant, Z.P. and F.H. Wittmann, Creep and shrinkage in concrete structures.
1982.
[14] Z.P. Bazant, R. L’Hermite, Mathematical Modeling of Creep and Shrinkage of
Concrete, Wiley, New York, 1988.
[15] S.-D. Wang, K.L. Scrivener, Hydration products of alkali activated slag cement,
Cem. Concr. Res. 25 (3) (1995) 561–571.
[16] S. Wallah, B.V. Rangan, in: Low-calcium Fly Ash-Based Geopolymer Concrete:
Long-Term Properties, Res. Report-GC2, Curtin University, Australia, 2006, pp.
76–80.
[17] M. Chi, R. Huang, Binding mechanism and properties of alkali-activated fly
ash/slag mortars, Constr. Build. Mater. 40 (2013) 291–298.
[18] F. Matalkah, P. Soroushian, Synthesis and characterization of alkali
aluminosilicate hydraulic cement that meets standard requirements for
general use, Constr. Build. Mater. 158 (2018) 42–49.
[19] M. Palacios, F. Puertas, Effect of shrinkage-reducing admixtures on the
properties of alkali-activated slag mortars and pastes, Cem. Concr. Res. 37
(5) (2007) 691–702.
[20] V. Sata, A. Sathonsaowaphak, P. Chindaprasirt, Resistance of lignite bottom ash
geopolymer mortar to sulfate and sulfuric acid attack, Cem. Concr. Compos. 34
(5) (2012) 700–708.
[21] A.A. Ramezanianpour, M.A. Moeini, Mechanical and durability properties of
alkali activated slag coating mortars containing nanosilica and silica fume,
Constr. Build. Mater. 163 (2018) 611–621.
[22] F. Okoye, J. Durgaprasad, N. Singh, Effect of silica fume on the mechanical
properties of fly ash based-geopolymer concrete, Ceram. Int. 42 (2) (2016)
3000–3006.
[23] X. Gao, Q. Yu, H. Brouwers, Properties of alkali activated slag–fly ash blends
with limestone addition, Cem. Concr. Compos. 59 (2015) 119–128.
[24] W.K. Babcock, D.A. Ruefenacht, Utilization of Latexes with Aluminous Cement
and Gypsum Composition, Google Patents, 1989.
[25] P. Duxson et al., Understanding the relationship between geopolymer
composition, microstructure and mechanical properties, Colloids Surf., A 269
(1–3) (2005) 47–58.
[26] T. Häkkinen, The influence of slag content on the microstructure, permeability
and mechanical properties of concrete Part 1 Microstructural studies and basic
mechanical properties, Cem. Concr. Res. 23 (2) (1993) 407–421.
[27] W. Shen et al., Morphology difference between the alkali activated cement and
portland cement paste on multi-scale, J. Wuhan Univ. Technol.-Mater. Sci. Ed.
23 (6) (2008) 923–926.
[28] D. Hou, H. Ma, Z. Li, Morphology of calcium silicate hydrate (CSH) gel: a
molecular dynamic study, Adv. Cem. Res. 27 (3) (2015) 135–146.
[29] K. Boonserm et al., Improved geopolymerization of bottom ash by
incorporating fly ash and using waste gypsum as additive, Cem. Concr.
Compos. 34 (7) (2012) 819–824.
[30] A.A.M. Neto, M.A. Cincotto, W. Repette, Mechanical properties, drying and
autogenous shrinkage of blast furnace slag activated with hydrated lime and
gypsum, Cem. Concr. Compos. 32 (4) (2010) 312–318.
[31] Y. Jun, J.E. Oh, Use of gypsum as a preventive measure for strength
deterioration during curing in class F fly ash geopolymer system, Materials 8
(6) (2015) 3053–3067.
[32] F. Collins, J. Sanjayan, Early age strength and workability of slag pastes
activated by NaOH and Na2CO3, Cem. Concr. Res. 28 (5) (1998) 655–664.
[33] A.M. Rashad, Influence of different additives on the properties of sodium
sulfate activated slag, Constr. Build. Mater. 79 (2015) 379–389.
[34] H. Wu et al., Laboratory evaluation of abrasion resistance of Portland cement
pervious concrete, J. Mater. Civ. Eng. 23 (5) (2010) 697–702.
[35] I. Soroka, N. Setter, The effect of fillers on strength of cement mortars, Cem.
Concr. Res. 7 (4) (1977) 449–456.
[36] B. Rongbing, S. Jian, Synthesis and evaluation of shrinkage-reducing admixture
for cementitious materials, Cem. Concr. Res. 35 (3) (2005) 445–448.
[37] T. Matschei, B. Lothenbach, F. Glasser, The AFm phase in Portland cement, Cem.
Concr. Res. 37 (2) (2007) 118–130.
[38] O.A. Abdulkareem et al., Effects of elevated temperatures on the thermal
behavior and mechanical performance of fly ash geopolymer paste, mortar and
lightweight concrete, Constr. Build. Mater. 50 (2014) 377–387.
[39] A. Gruskovnjak et al., Quantification of hydration phases in supersulfated
cements: review and new approaches, Adv. Cem. Res. 23 (6) (2011) 265–275.
[40] E. Kanezaki, Thermal behavior of the hydrotalcite-like layered structure of Mg
and Al-layered double hydroxides with interlayer carbonate by means of
in situ powder HTXRD and DTA/TG, Solid State Ionics 106 (3) (1998) 279–284.
[41] A. Hidalgo et al., Microstructural changes induced in Portland cement-based
materials due to natural and supercritical carbonation, J. Mater. Sci. 43 (9)
(2008) 3101–3111.
[42] L. Feng et al., The shape-stabilized phase change materials composed of
polyethylene glycol and various mesoporous matrices (AC, SBA-15 and MCM-
41), Sol. Energy Mater. Sol. Cells 95 (12) (2011) 3550–3556.
[43] L. Feng et al., The shape-stabilized phase change materials composed of
polyethylene glycol and graphitic carbon nitride matrices, Thermochim. Acta
612 (2015) 19–24.
[44] S. Noor et al., Morphology, chemical interaction, and conductivity of a PEO-
ENR50 based on solid polymer electrolyte, Ionics 16 (2) (2010) 161–170.
[45] W.A.K. Mahmood, M.M.R. Khan, M.H. Azarian, Sol-gel synthesis and
morphology, thermal and optical properties of epoxidized natural rubber/
zirconia hybrid films, J. Non-Cryst. Solids 378 (2013) 152–157.
[46] S.-A. Riyajan et al., Crosslinking of epoxidized natural rubber with terephthalic
acid: effect of Terephthalic Acid and Cassava Starch, J. Polym. Mater. 31 (2)
(2014) 145.
[47] U. Avila-López, J. Almanza-Robles, J. Escalante-García, Investigation of novel
waste glass and limestone binders using statistical methods, Constr. Build.
Mater. 82 (2015) 296–303.
[48] P. Chindaprasirt, P. Paisitsrisawat, U. Rattanasak, Strength and resistance to
sulfate and sulfuric acid of ground fluidized bed combustion fly ash–silica
fume alkali-activated composite, Adv. Powder Technol. 25 (3) (2014) 1087–
1093.
[49] U. Rattanasak, K. Pankhet, P. Chindaprasirt, Effect of chemical admixtures on
properties of high-calcium fly ash geopolymer, Int. J. Miner. Metall. Mater. 18
(3) (2011) 364–369.
[50] K. Pimraksa, P. Chindaprasirt, Lightweight bricks made of diatomaceous earth,
lime and gypsum, Ceram. Int. 35 (1) (2009) 471–478.
[51] Taylor, H., Cement chemistry. 2nd. Ed., T Telford, London, 459pp, 1997.
[52] V.S. Ramachandran, Thermal analyses of cement components hydrated in the
presence of calcium carbonate, Thermochim. Acta 127 (1988) 385–394.
[53] J. Pera, S. Husson, B. Guilhot, Influence of finely ground limestone on cement
hydration, Cem. Concr. Compos. 21 (2) (1999) 99–105.
[54] H. Cheng-Yi, R.F. Feldman, Hydration reactions in Portland cement-silica fume
blends, Cem. Concr. Res. 15 (4) (1985) 585–592.
[55] C.M. Dobson et al., 29 Si MAS NMR study of the hydration of tricalcium silicate
in the presence of finely divided silica, J. Mater. Sci. 23 (11) (1988) 4108–4114.
[56] K.J. Folliard, N.S. Berke, Properties of high-performance concrete containing
shrinkage-reducing admixture, Cem. Concr. Res. 27 (9) (1997) 1357–1364.
[57] A.M. Rashad, M.H. Khalil, A preliminary study of alkali-activated slag blended
with silica fume under the effect of thermal loads and thermal shock cycles,
Constr. Build. Mater. 40 (2013) 522–532.