Cement and Concrete Research 43 (2013) 70–80

Contents lists available at SciVerse ScienceDirect

Cement and Concrete Research

journal homepage: http://ees.elsevier.com/CEMCON/default.asp

Role of slag in microstructural development and hardening of fly ash-slag geopolymer

Sravanthi Puligilla ⁎, Paramita Mondal
University of Illinois at Urbana Champaign, 205 N. Mathews Ave., Urbana, IL 61801-2352, USA

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

Article history: This paper presents the microstructural development and hardening rate of fly ash-slag geopolymers. The
Received 3 February 2012 solid precursors are treated with a potassium hydroxide-silicate activator of modulus 1.25. Ultrasonic wave
Accepted 11 October 2012 reflection and Proctor penetration methods are used to determine the rate of hardening. Microstructural de-
velopment is investigated using SEM-EDS, and the reaction rate is monitored using semi-adiabatic calorime-
Keywords:
try. Both UWR and Proctor penetration methods capture changes in the hardening rate due to changes in the
Microstructure (B)
Fly ash (D)
reaction mechanism. An increase in the rate of hardening is observed with the addition of slag. It is concluded
Granulated blast-furnace slag (D) that the calcium that dissolves from slag significantly influences both early and late age properties, and the
Acceleration (A) availability of free calcium ions seems to prolong fly ash dissolution and enhance geopolymer gel formation.
It is proposed that the hardening process is initiated by the precipitation of C-A-S-H and that rapid hardening
continues due to accelerated geopolymerization.
© 2012 Elsevier Ltd. All rights reserved.

1. Introduction
Davidovits defines geopolymer as an amorphous, three-dimensional
short range order inorganic polymer that forms when a highly concen-
trated aqueous alkali hydroxide-silicate solution is added to aluminosil-
icate raw materials (e.g. metakaolin, fly ash, slag) [1]. The highly
alkaline activator solution and high temperature curing were used
initially, but ongoing research in the field focuses on the develop-
ment of a user-friendly system [1–3]. To reach the full potential of
the geopolymeric binder (as a replacement of the portland cement
binder), the following are suggested: a combination of raw waste
materials like class F fly ash and slag with low concentration alkaline
activators (e.g. potassium hydroxide + potassium silicate), as well as
room temperature curing [1,3]. A study of a similar system reported
the successful development of a very high strength (100 MPa)
geopolymer, however, at the cost of low workability [2]. The loss of
workability was attributed to both the presence of free lime in the sys-
tem and the fineness of fly ash particles. The type, dosage, concentration
of alkali metal activator (Na or K), temperature of curing, and the pres-
ence of various ionic contaminants are several parameters known to af-
fect the hardening rate of fly ash geopolymer and change the setting
time from few minutes to few days [4–8]. The consequence of using
waste materials as precursors, then, is the emergence of an additional
challenge because the properties of the final product vary significantly
with the extreme variability of many raw materials.
Many researchers indicated the effects of calcium ion on early age
(loss of workability) as well as later age properties (high strength) of
⁎ Corresponding author. Tel.: +1 609 903 6082.
E-mail address: puligil1@illinois.edu (S. Puligilla).
metakaolin -or fly ash-based geopolymer [9–13]. Yip and van Deventer
reported the coexistence of geopolymeric gel and calcium silicate hy-
drate in metakaolin and slag blends [14,15]. The percentage of different
reaction products depends on various factors such as alkalinity, temper-
ature, ratio of aluminosilicate to calcium sources, rate of dissolution of
ions from aluminosilicate, and calcium sources [9,15–18]. The presence
of soluble calcium is known to significantly accelerate the hardening
process [9,19]. It has been proposed that the presence of calcium is crit-
ical to early strength development due to an enhanced aluminosilicate
gel formation [9,19,20]. However, the exact mechanism still remains
unknown. The interaction between calcium, silicate, and aluminate spe-
cies is very complex. A recent study proposed a phase diagram of CaO–
SiO2–Al2O3–Na2O based on a series of solution/precipitation studies
[21]. The role of calcium needs further verification in order to optimize
workability and strength.
In the current study, various combinations of slag and fly ash were
used to demonstrate the role of calcium on the hardening behavior of
fly ash-slag geopolymers. This study employed two different methods:
the Proctor penetration method and the ultrasonic wave reflection
method. In previous studies, several researchers measured the setting
times of metakaolin, fly ash- or slag-based geopolymers using a Vicat
penetrating needle (ASTM C191) [5,12,13,22–25]. Duxson et al. defines
the initial set of geopolymers as the time at which the rate of dissolution
of the raw materials is surpassed by the rate of condensation and pre-
cipitation of the aluminosilicate species [26]. Although some useful in-
formation, such as the initial and final set times can be estimated, one
serious drawback of using a Vicat needle is that it does not indicate
any change in penetration prior to the initial set, and it does not provide
any information about the hardening rate over time [27]. Detailed stud-
ies on setting/hardening behavior of geopolymer are important in order
to determine the time of transport, finishability, and time to remove

0008-8846/$ – see front matter © 2012 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.cemconres.2012.10.004
 S. Puligilla, P. Mondal / Cement and Concrete Research 43 (2013) 70–80
forms, as well as to understand the microstructural changes leading to
hardening.
The ultrasonic wave reflection (UWR) method has been effective-
ly used to study the setting/hardening behavior of several materials
that exhibit the transition from a liquid to a solid state. A few studies
have used Pressure-wave UWR to understand the properties of a
geopolymeric binder. One study used the ultrasonic wave measure-
ments to determine the relation between Si/Al ratio and elastic prop-
erties (elastic modulus and Poisson's ratio) of sodium-activated
metakaolin geopolymers [28,29]. In another study, the reaction
progress of metakaolin and slag blends with low and high concentra-
tions of alkali activator (NaOH) was successfully monitored using the
variation in the velocity of longitudinal waves (P-waves) [30]. The
propagation of pressure waves depends on changes in density, but
the propagation of shear waves is sensitive to the extent of a solid
network. In an effort to understand the effects of calcium in the hard-
ening process, the present study extended the use of the Shear-wave
ultrasonic wave reflection method (S-wave UWR) to investigate the
early age properties and hardening rate of fly ash-slag geopolymers.
Additionally, microstructural development was investigated using
scanning electron microscopy (SEM) with energy-dispersive X-ray
spectroscopy (EDS) and the reaction rate was monitored using
semi-adiabatic calorimetry.
2. Materials and methods
The materials used in this study were class-F fly ash, ground gran-
ulated blast furnace slag (GGBFS), potassium silicate and potassium
hydroxide. The fly ash was provided by Boral Material Technologies,
and the ground granulated blast-furnace slag was supplied by Lafarge
Corporation. The potassium silicate, produced by Pfaltz and Bauer,
and supplied by Fischer Scientific, has a solid content of 29% with
SiO2 = 20.75%, K2O = 8.3% and a modulus of 3.92. A Type I portland
cement, manufactured by Essroc Cement Corp. (USA), was used in
this study for purposes of comparison. The oxide composition of fly
ash, slag and cement is listed in Table 1. X-ray diffractograms were
recorded on a Siemens-Bruker D 5000 using CuKα radiation, scanned
from 5 to 70°2θ at a rate of 1°min −1. The X-ray diffraction patterns of
raw materials are presented in Fig. 1 and show, the main peaks of
quartz, mullite, and hematite. The broad amorphous hump in the
XRD pattern, which is characteristic of fly ash, is observed between
20° and 40°2θ. Slag is an amorphous glass, as evident from its XRD
pattern with a hump centered around 30°2θ.
The geopolymer was synthesized by mixing fly ash and slag with
the activator solution. The activator solution was prepared by mixing
water, potassium silicate, and potassium hydroxide such that SiO2/
K2O = 1.25 and water/solids = 0.32. The amount of slag in the mix-
ture varied from 0 to 15% (by total weight), keeping the total amount
of solids constant. The mixing procedure was as follows: the solid
precursors were weighed, and the dry powders were mixed for
30 s. The activator solution was then added and mixed for another
3.5 min at the same speed. It was observed that the geopolymer
mixes were greatly influenced by the ambient temperature and the
shear rate. To minimize the effects of temperature, the activator and
Table 1
Oxide composition of raw materials (weight in %).
Composition Fly ash Slag
SiO2 60.17 35.7
Al2O3 21.91 11.21
CaO 1.81 39.4
Na2O
K2O
0.81
2.13
0.26
0.48
MgO 1.28 10.74
Fe2O3 7.57 0.42
SO3 0.17 0.58
71
Fig. 1. XRD pattern of raw materials (fly ash and slag) where Q = Quartz, M = Mullite
and H = Hematite. Both raw materials show broad amorphous hump. The main crys-
talline phases observed in fly ash are quartz, mullite and hematite.
precursors were stored in a temperature-controlled room at 21 ±
2 °C, 24 h prior to the mixing process. To minimize the effects of
the shear rate, a constant amount of material was mixed. The mixes
are referred to as Geo Xs, where X indicates the percentage of slag
with respect to the total precursor and activator solution.
2.1. Monitoring hardening behavior
The rate of hardening of various geopolymer pastes was moni-
tored using the Proctor penetration method as well as S-wave UWR
measurements.
2.1.1. Proctor penetration method
As mentioned previously, the Vicat test provides only an estimate
of the initial and final setting times. Setting times are estimated by
measuring the penetration depth of a single needle while maintaining
a constant load. In contrast, the Proctor test, initially developed for ce-
ment mortar, measures the resistance offered by the medium against
a penetrating needle. Cement mortar does not provide any resistance
to penetration immediately following the mixing. As hydration con-
tinues and the microstructure develops, resistance to the penetrating
needle increases, and this helps in monitoring the rate of hardening/
strength gain. ASTM C403 describes the Proctor penetration test
method for cement mortars [31]. This method has been extended to
study the hardening behavior of cement pastes with slight modifica-
tions. It has proven to be successful in determining the rate of harden-
ing along with the determination of initial and final set values
[27,32,33].
The method is based on measuring the resistance against a penetrat-
ing needle as the medium transforms from a liquid to a solid state. The
same procedure is applied in order to study the hardening rate of
geopolymeric binders, in spite of the differences in chemistry and un-
derlying chemical reactions responsible for its hardening. Immediately
after mixing, geopolymer paste is placed in a plastic container and cov-
ered with a moist towel to prevent water evaporation. The penetration
resistance is measured by an Instron 4500 load frame. Six different
Cement
needles of varying cross-sections (645 mm2, 323 mm2, 161 mm2,
20.2 65 mm2, 32 mm2 and 16 mm2) are used as needed. Measurements
4.8
are initiated using the largest needle when the paste is liquid, and
63.3
needles of progressively smaller cross sections are used as the paste
g0.59 hardens, so as to maintain a penetration depth of 25 mm (1-inch) in
2.4 10±2 s (ASTM C 403) without exceeding the capacity of the load cell.
3.4 The penetration resistance offered by the paste is calculated by dividing
3.1
the resistance load with the area of the needle used for penetration.
72 S. Puligilla, P. Mondal / Cement and Concrete Research 43 (2013) 70–80
2.1.2. Shear-wave ultrasonic wave reflection method (S-wave UWR)
The inability of penetration-based methods to continuously mon-
itor the hardening process has prompted researchers to study the use
of acoustic methods. Non-destructive acoustic methods, including the
ultrasonic wave reflection method (UWR), have been used extensive-
ly in the past in order to estimate the elastic properties of concrete.
Recently, the process of monitoring the setting and hardening pro-
cesses of cementitious materials using the UWR has become impor-
tant [33–38].
These methods are based on the principle that shear waves cannot
pass through fluids but can only pass through solids. According to the
wave theory, when a wave is incident at the boundary between two
materials, some of it is reflected and some of it is transmitted. The
amount of shear wave energy reflected at the boundary between a
buffer material and paste depends on the interconnectivity of the
solid to transmit shear waves. When the paste is in a liquid state,
the entire shear wave energy is reflected at the interface between
the buffer material and the paste. Thus, the reflection coefficient is
one. The transition of the paste from a liquid to a solid changes its
acoustic impedance due to an increase in the interconnectivity of
the solid within the paste. As more and more solids form, the ability
of the paste to transmit shear waves increases, thus decreasing the re-
flection coefficient. Therefore, changes in the paste as a function of
time can be monitored with S-wave UWR using a buffer material of
constant acoustic impedance. Several buffer materials including
steel and PMMA were used throughout the literature [36,38]. Howev-
er, the sensitivity of the measurements is maximized when the differ-
ence between the acoustic impedance of the buffer and the test
material is small (high impact polystyrene in the case of cement
paste) [37,39].
Initiation of the setting/hardening process in systems like cement
pastes happens when the solid phases reach the percolation thresh-
old. It has been established, in the case of cement paste, that the re-
flection loss measured from UWR relates closely to the connectivity
of the solid phases [38]. The amount of shear wave reflection de-
creases as the connectivity of solid increases [34]. A few studies of
the wave-reflection method on cement mortars and cement pastes
have concluded that the wave reflection method is very sensitive to
the setting and hardening of cementitious materials and can be
used in monitoring the hydration process during the early age of mi-
crostructural development [34,38]. Similar studies on cement pastes
by Chung et al. have shown that the onset of early stiffening as mea-
sured from S-wave UWR, Proctor penetration resistance, and temper-
ature rise from calorimetric experiment, correlate very well [33].
Numerical simulations have shown that the setting time calculated
based on the reflection loss of shear waves matches very well with
the percolation threshold of the solid phase in the cement paste
[34,40]. Chung et al. have also shown that the UWR is sensitive to
study the sedimentation and flocculation of non-hydrating systems
[41]. UWR was used to study the flocculation of cement particles im-
mediately after mixing (before the initial setting). In the research
presented here, the method of S-wave UWR is extended to monitor
the hardening rate of geopolymer at an early age. In spite of the dif-
ferences in the reaction kinetics that govern the formation of C-S-H
(in cement paste) and aluminosilicate gel (in geopolymers), the re-
sults presented here indicated that the physical process of setting
may be regarded similarly and is governed by the percolation of
solid phases.
In the current study, the reflection coefficient is measured by
using a contact type 2.25 MHz S-wave transducer as shown schemat-
ically in Fig. 2. Immediately after mixing, the geopolymer paste is
poured into a high impact polystyrene (HIPS) container approximate-
ly sized, 50 mm × 60 mm × 50 mm (thickness 6.25 mm) and covered
to prevent the evaporation of water. HIPS acts as a buffer between the
transducer attached to the bottom of the container and the specimen.
The transducer is connected to a pulser/receiver unit. Lab View is
used to control, collect, and process the data. The rate of stiffening
is monitored for the first 24 h or until the specimen is completely
debonded from the buffer material. Debonding creates an air pocket
between the buffer and the specimen, which prevents the further
measurement and consequently, the acquisition of meaningful infor-
mation. In order to avoid confusion, this part of the data is truncated
from the UWR graphs presented in this paper. Further details of this
method can be found elsewhere [42].
2.2. Semi-adiabatic calorimetry
To monitor the ongoing chemical reaction in various geopolymer
mixes, a semi-adiabatic calorimeter is used to measure the variation
in temperature. The semi-adiabatic method uses a thermally insulat-
ed bottle (with some heat loss) and a type-T thermocouple to mea-
sure the temperature as a function of time. Since heat loss from
semi-adiabatic calorimeter will depend on laboratory conditions,
the tests are carried out in a climate controlled room at 21 ± 2 °C.
At a standard atmospheric pressure, 180 g of water takes about
3.5 h to drop the temperature by 0.5 °C, when the initial water tem-
perature is 25 °C and the outside room temperature is 21 ± 2 °C.
The rate of heat loss decreases as the temperature difference between
the sample and the outside environment decreases. Details of the ex-
perimental setup can be found elsewhere [43].
2.3. SEM imaging with EDS for microstructural development
Microstructural development is monitored by using a JEOL
JSM-6060 LV scanning electron microscope, equipped with an ener-
gy dispersive X-ray spectroscopy (SEM-EDS). The fractured surface
of different specimens at various ages is studied in a secondary elec-
tron mode (SE). Specimens are treated with acetone to stop the re-
action and then stored in a vacuum desiccator until tested. The
specimen surfaces are sputter coated with gold-palladium prior to
imaging.
Microanalysis and quantitative EDS is performed in a backscattered
mode (BSE). The samples are epoxy impregnated and polished prior to
testing. After epoxy impregnation, the samples are ground using SiC pa-
pers of various grit sizes, 21.8 μm, 15.3 μm, 8.4 μm, and 5 μm. A
glycol-based lubricant was used for polishing. The final steps involve
polishing the sample, using 3 μm and 1 μm diamond lapping films,
and this is followed by polishing using a 0.25 μm diamond paste on a
cloth until a glass-like finish is achieved. The samples are ultrasonically
cleaned in isopropyl alcohol after each round of polishing to remove any
debris. Microanalysis is performed at an accelerating voltage of 7 kV to
minimize the interaction volume for X-ray generation, as prescribed by
Lloyd et al. [44].
3. Results
3.1. Hardening rate of fly ash geopolymers with slag replacement
Fig. 3 shows the change in the S-wave reflection coefficient over
time for the fly ash geopolymers with varying slag content. Cement
paste of w/c = 0.33 (similar to the water-to-solids ratio of the
geopolymer pastes used in this study) is also presented for compari-
son. In the case of Geo 0s (fly ash geopolymer with 0% slag replace-
ment), the rate of hardening is observed to be very slow as the
S-wave reflection coefficient changes slowly and remains roughly
around 0.6 even after 1000 min. Conversely, when 1% (by total
weight of the mix) of slag is added to the mix (Geo 1s), a rapid
drop in the S-wave reflection coefficient is seen around 55 min after
mixing and continues until 420 min. After this, S-wave reflection co-
efficient suddenly reaches one. This is due to the complete debonding
of the specimen from the buffer surface. The debonding may be due to
drying of water, chemical shrinkage, or volume change associated
 S. Puligilla, P. Mondal / Cement and Concrete Research 43 (2013) 70–80
Fig. 2. Schematic diagram of UWR set up showing transducers, pulser and receiver units based on Chung et al. [39].
with the transition from liquid to solid, which limits the data collec-
tion. The behavior of Geo 5s, Geo 10s, and Geo 15s can be described
in a similar manner. As the amount of slag increases, curves move fur-
ther left and thus indicate early stiffening. Furthermore, in the pres-
ence of slag, the S-wave reflection curves of all the geopolymer
mixes show a change in the slope a second time, approximately a
few minutes to a few hours after mixing. This indicates that the hard-
ening rate is slowing down. This part of the curve is further analyzed
in conjunction with the Proctor penetration data (presented later in
this section). The reason for slope change could indicate a possible
change in the reaction rate. This is studied further and reported in
the section on semi-adiabatic calorimetry.
In addition to early stiffening, the increased amount of slag also in-
creases the compressive strength (Table 2). The development of mi-
crostructure is slow at ambient temperatures (due to the extremely
slow dissolution of class F fly ash) which is reflected in its low com-
pressive strength values and compares well with the values published
in the literature of similar systems [9,13,20,45]. In the case of Geo 10s
and Geo 15s, the initial rate of hardening from UWR curves is ob-
served to be the same. However, the later age compressive strength
(14 days) is higher in the case of Geo 15s (35 MPa) compared to
that of the Geo 10s (17 MPa).
Fig. 3. S-wave reflection coefficient vs. time plot for fly ash geopolymers with varying
slag content (0, 1, 5, 10 and 15% and cement paste of w/c = 0.33. Plot describes the
change in the rate of hardening with the varying amount of slag.
73
Fig. 4 shows Proctor resistance for all the mixes including cement
paste (w/c = 0.33), plotted as a function of time. Geo 0s offers nearly
zero resistance (below the sensitivity of the technique) until 300 min,
which indicates an extremely slow rate of hardening and microstruc-
ture development. This part corresponds to an S-wave reflection coef-
ficient of about 0.9–1.0 (Fig. 3). After ~ 300 min, the Proctor resistance
increases slowly and indirectly indicates that the microstructure is
gaining strength. It is important to note that as the amount of slag in-
creases, the onset of penetration resistance (hardening/strength gain)
is observed earlier. That is, the curves shift left as the amount of slag
increases, and this indicates faster stiffening. For example, Geo 15s
shows a rapid increase in resistance, starting 10 min after mixing.
Similar to the trend described in Fig. 3, the rate of hardening for all
mixes was also observed to slow down from Proctor penetration
curves.
S-wave UWR response and Proctor penetration response for all
the geopolymer mixes are analyzed together to determine the time
of onset of hardening and the time when the rate of hardening slowed
down. Fig. 5 illustrates the analysis for Geo 1s and Geo 15s. The onset
of hardening is defined as the point at which the S-wave reflection
curve deviates from the initial slope of a linear fit line showing values
close to one and starting to drop rapidly for the first time. This is
noted as the ‘onset’ point, and is determined from various UWR
runs on the same mix with nominal variation (±5 min). The rapid
change in the curve indicates the initiation of microstructure devel-
opment. As seen in Fig. 5, the response of Geo 1s shows that the
onset of hardening/strength development starts around 55 min
after mixing. The response of Geo 15s shows that the hardening/
strength development starts within the first 10 min after mixing, indi-
cating rapid hardening. These results align with the findings that the
presence of soluble calcium accelerates the stiffening of geopolymers
[9,10,12,13,19]. Kumar et al. reported a decrease in the setting times
measured using the Vicat needle test with an increase in the amount
of slag replacement for fly ash-slag geopolymer made with activator
of higher alkalinity [13]. In this study, the onset of hardening is also de-
termined from the Proctor method as the point at which the Proctor
penetration curve deviates from the initial slope of a linear fit line
Table 2
3, 7 and 14 day compressive strength in MPa of Geo 0s, Geo 1s, Geo 15s, and CaO 1%.
Data is reported as an average and standard deviation of three measurements on
1-inch (2.54 cm) cubes.
Day Geo 0s Geo 1s Geo 15s CaO 1%
3 NA 1.5 ± 0.2 14.1 ± 0.7 1.1 ± 0.4
7 1.1 1.8 ± 0.1 28.8 ± 3.3 1.5 ± 0.1
14 1.2 ± 0.1 2.4 ± 0.3 34.8 ± 3.6 1.7 ± 0.2
74 S. Puligilla, P. Mondal / Cement and Concrete Research 43 (2013) 70–80
Fig. 4. Proctor penetration resistance in MPa vs. time for cement paste of w/c = 0.33
and fly ash geopolymers with variable amount of slag (0, 1, 5, 10 and 15%). The data
beyond 700 min is truncated to show the initial hardening phase.
showing values close to zero and starting to raise rapidly for the first
time. The times for onset of hardening determined from both methods
compare very well for all the geopolymer mixes as shown in Table 3.
(a)
(b)
Fig. 5. Correlation between S-wave UWR response and penetration resistance of a: Geo 1s,
b: Geo 15s. Results from S-wave UWR and penetration resistance method match very well.
As previously mentioned, both the Proctor and UWR curves show
a change in the slope a second time, indicating that the hardening
rate has slowed down. The point corresponding to this slope change
(noted as the ‘slowdown’ point in Table 3) is also determined from
both methods. As the slope change often occurred over a small region,
particularly for the UWR curves, the point associated with the slope
change is determined by the point of intersection between the two
linear lines of the best fit of two regions (before and after the slope
change). For Geo 1s, the rate of hardening slows down between 245
and 270 min, as seen from both Proctor penetration and UWR re-
sponse (Fig. 5), and between 50 and 75 min for Geo 15s. Little dis-
crepancy in the time of the slope change determined from both
methods is attributed to the sensitivity of the testing method. UWR
generally shows a very gradual change in the slope and the ‘slow-
down’ point determination by the point of intersection of linear fit
lines was more effective. The ‘slowdown’ point calculated from vari-
ous UWR runs on the same mix showed only ± 15 minute variation
on an average. In contrast, the Proctor penetration method shows a
sudden change in slope. Therefore, penetration resistance was mea-
sured at a close interval during the rapid hardening period after the
‘onset’ point. However, as this method is locally destructive, measure-
ment at a close interval after the second slope change was often not
possible due to limited availability of intact sample surface. Although
this introduced some uncertainty in the slope of the second linear fit
line, the intersection point did not vary by more than ±20 min and
still provided a good comparison with the ‘slowdown’ point deter-
mined from the UWR method.
3.2. Hardening rate of fly ash geopolymers with CaO replacement
As calcium is known to alter the hardening rate of geopolymer, the
effects of slag as mentioned above can be assumed to be largely due to
calcium coming from slag (as the CaO content of class F fly ash used in
this study is very low) [9,10]. To confirm this and to understand the
effects of calcium more carefully, CaO is added to fly ash geopolymers,
instead of slag, and the hardening rate is studied using UWR and
Proctor penetration tests. The results are compared with the mixes
of equivalent slag content (in terms of total weight of calcium
present). One critical difference between these two systems should
be noted here. The dissolution rate of CaO is very different from that
of the slag. This variability in the rate of dissolution is expected to cre-
ate a difference in the amount of free calcium ions and the duration of
availability in the system. Most of the CaO is consumed in the early
stages of reaction, whereas the dissolution of slag is much slower.
Slag dissolution also contributes magnesium, silicon, aluminum, sul-
fur and other minor elements. These may have further influence on
the rate of hardening; however, to study the effect of various ele-
ments is beyond the scope of the present study. Geopolymers made
with CaO replacement equivalent to 5% (and more) slag, were diffi-
cult to cast and compact in HIPS container because, they hardened ex-
tremely rapidly. Therefore, amount of CaO equivalent to 1% slag
replacement, which is 0.4% of CaO with respect to the total weight
of the fly ash, is used for UWR measurement. This particular mix
will be referred to as CaO 1%.
Table 3
The ‘onset’ point and the ‘slowdown’ point for various mixes from both S-wave UWR
and Proctor penetration method. Times are given in minutes.
Mixes UWR Proctor
Onset point Slowdown point Onset point Slowdown point
Geo 1s 56 270 55 245
Geo 5s 25 86 20 90
Geo 10s 15 55 15 55
Geo 15s 8 50 5–10 75
CaO 1% 105 240 30 235
 S. Puligilla, P. Mondal / Cement and Concrete Research 43 (2013) 70–80
Figs. 6 and 7 compare UWR measurements and Proctor resistance
of Geo 1s and CaO 1%, respectively. Measurements of Geo 0s and Geo
5s are also included in both figures for comparison. Differences in the
hardening rate due to the presence of calcium are evident from the
behavior of Geo 0s and CaO 1% in Fig. 6. As small as 0.4% by weight
of CaO added to fly ash is found to accelerate the stiffening rate of
the geopolymer significantly. As previously determined, the time for
onset of hardening in the case of Geo 1s is 55 min (Table 3), which
corresponds to a reflection coefficient of ~ 0.95. The S-wave reflection
coefficient of CaO 1% changes with a constant slope until 105 mins
pass, followed by a rapid drop. This rapid drop is observed at a reflec-
tion coefficient value of 0.89, which is lower than the value observed
for geopolymer mixes with slag replacement. Thereafter, the slopes of
curves for Geo 1s and CaO 1% match very closely during the rapid
hardening phase. For the GeO 1s, the hardening rate slows down at
270 min (from Fig. 6). For the CaO 1%, however, the slope of the
curve starts changing earlier (around 210 min). The slope seemed
to change more than once, and at around 500 min, the UWR curve be-
comes almost flat, thus indicating that the reaction has slowed down
to a greater extent in the case of CaO 1% compared to Geo 1s. This may
be on account of the complete consumption of calcium. The availabil-
ity of free calcium ions for extended time duration in the slag mix
promotes fly ash dissolution and geopolymer formation. Therefore,
the final strength of Geo 1s is also higher than the CaO 1% despite
the fact that the total amount of Ca was the same. In order to deter-
mine the ‘slowdown’ point for CaO 1%, the slope change at 240 min
was reported in Table 3; however, the continuous slope change of
the curve, as mentioned above, introduces a greater uncertainty
(± 30 min) in this case compared to the other mixes.
The mix with CaO 1% shows a rapid increase in Proctor resistance
starting at 30 min after mixing compared to 55 min in the case of Geo
1s (Fig. 7). The Proctor resistance for CaO 1% increases rapidly until
235 min and slows down thereafter. In the case of Geo 1s, the Proctor
resistance increases rapidly until 245 min after which it slows down.
By comparing the Proctor and UWR curves of CaO 1%, it is clear that
the onset of hardening is indicated earlier by the Proctor resistance
curve (30 min) than by the first slope change of the UWR curve
(105 min). Notably, the measured reflection coefficient for CaO 1%
at 30 min is 0.95. This value is the same as the measured reflection
coefficient at the ‘onset’ point for all other mixes. Based on numerical
simulations of cement pastes, it has been proposed in the past that a
single value of reflection loss can be defined that corresponds to an
Fig. 6. Effect of calcium on hardening rate of fly ash geopolymers: S-wave UWR of CaO
1%, Geo 0s, Geo 1s and Geo 5s. Slope changes are indicated by intersection of linear fit
lines for Geo 1s and CaO 1%.
75
Fig. 7. Proctor penetration resistance vs. time of Geo 0s, CaO 1%, Geo 1s and Ge0 5s.
initial set time [34]. For the geopolymer mixes studied in this paper,
a reflection coefficient of 0.95 seemed to be associated with the
same level of connectivity in the solid phase (microstructure develop-
ment) as it starts to offer resistance to a penetrating Proctor needle.
This result indicates that, regardless of the difference in chemical re-
action, the physical process of setting in geopolymer mixes is also
governed by the percolation of solids. The fact that the slope of the
UWR curve for CaO 1% did not change until a later time (105 min)
may indicate that the initial reaction mechanism responsible for the
onset of hardening continued for an extended period at the same
rate. This may be due to the continued availability of calcium ions.
3.3. Semi-adiabatic calorimetry
Semi-adiabatic calorimetry is used to measure temperature changes
in different mixes during the first 1500–3000 min after mixing and will
give an indication of the ongoing chemical changes. Temperature mea-
surement began within 30 s after mixing. Temperature–time curves
shown in Fig. 8 compare well with the published literature on alkali ac-
tivated slag systems, metakaolin-slag/CH blends, and fly ash-slag blends
[6,13,16,17,46–48]. In the case of Geo 0s, there is only one peak
(24.1 °C) after which the temperature decreases continuously and
reaches room temperature (22 °C) within1000 min. The initial peak
corresponds to the wetting of raw materials, which, in this case, is not
followed by any further rise in temperature, indicating an insignificant
amount of product formation. Geo 5s, Geo 10s and Geo 15s also show
an initial rise in temperature, reaching a peak within the first 80, 40,
and 30 min, respectively.
The first peak corresponds to the wetting of particles, dissolution
of fly ash and slag particles, and the complexation of silicate units
with calcium and potassium in the solution [6]. Moreover, in the
case of Geo 5s, Geo 10s and Geo 15s, the first calorimetry peak ap-
pears during the rapid hardening period before the ‘slowdown’
point, as detected by both UWR and Proctor. In other words, the
peak position correlates well with the ‘slowdown’ point measured
earlier. Therefore, the first calorimetry peak is also associated with
the precipitation of the initial reaction products that cause rapid
hardening. Furthermore, both UWR and Proctor penetration methods
proved to be very sensitive to changes in the rate of reaction. Harden-
ing after the ‘slowdown’ point (from UWR and Proctor) corresponds
to the descending side of the first peak and the beginning of the in-
duction period. Thus, the reduction in the rate of stiffening is indeed
due to the change in chemical reaction during the induction period.
76 S. Puligilla, P. Mondal / Cement and Concrete Research 43 (2013) 70–80
Fig. 8. Temperature vs. time curves of various geopolymer mixes obtained using semi-adiabatic calorimetry. Hatched region under curves in the inserted panel indicates the rapid
hardening phase determined from UWR and Proctor (between ‘onset’ point and ‘slowdown’ point).
In the case of Geo 5s, the temperature decreases after the first
peak and reaches room temperature within 1500 min. Geo 10s and
Geo 15s show a second peak after a long induction period, which is
related to the further precipitation of reaction products. The second
peak does not appear within the time frame of the experiment for
Geo 5s, indicating a slower reaction rate. As the second peak occurs
earlier in the case of Geo 15s, compared to Geo 10s, the peak position
seems to be highly dependent on the amount of slag in the system.
3.4. Scanning electron microscopy
Fig. 9 shows SEM micrographs of fly ash-based geopolymers with
and without slag replacement. Geo 15s and Geo 0s represent two ex-
treme cases of microstructural development and are chosen for SEM
characterization in an effort to understand the effects of slag. Fig. 9a
and b shows the micrographs of Geo 0s and Geo 15s, respectively,
at 3 h after mixing. As the figures indicate, the difference between
the two microstructures is evident. Within 3 h, Geo15s show clear ev-
idence of reaction product formation on fly ash spheres whereas Geo
0s does not show any sign of product formation. It appears, from
Fig. 9a, that there was minimal dissolution of the fly ash spheres in
Geo 0s. The microstructure seems to resemble a fractured sample of
densely-packed fly ash spheres without any interconnectivity. This
is not surprising based on the calorimetry and UWR results reported
earlier. In the case of Geo 15s (Fig. 9b), the particle morphology of
fly ash spheres did not change markedly, but they seemed to be
connected by a gel-like network. The presence of slag enhances prod-
uct formation, resulting in a gel-like network with increased inter-
connectivity. The difference in microstructures further explains the
reason behind a faster hardening rate of Geo 15s compared to Geo 0s.
Fig. 9c and d and e and f shows microstructure of Geo 0s and Geo
15s at 24 h and 14 days, respectively. In the case of Geo 15s, the mi-
crostructure developed continuously with age, and by 14 days, it
seemed to be quite dense. As shown in Fig. 9f, the fly ash spheres
are densely covered with product, and there is no clear distinction be-
tween them. In the case of Geo 0s, the microstructure development at
24 h (Fig. 9c), is comparable to that of Geo 15s at 3 h (Fig. 9b).
However, compared to that of Geo 15s, the final microstructure
shows less product formed on the fly ash surface even after 14 days.
3.5. Quantitative elemental spot analysis
Quantitative EDS analysis is performed on a Geo 15s sample in
order to study the chemical makeup of reaction products formed in
the system. Two different ages are chosen: (i) 1 h as it falls within
the ascending part of the first peak from calorimetry when the mix
is still hardening rapidly and, (ii) 14 days as it represents a later age
sample. All of the ratios that are calculated from EDS microanalysis
results are compared with the ratios set by Garcia-Lodeiro et al. in
order to distinguish between C-A-S-H and (Ca,K)-A-S-H and are
given below [21]:
C  A  S  H : 0:72bCaO=SiO2 b1:94; 0bAl2 O3 =SiO2 b0:1
ðCa; KÞ  A  S  H : 0bK2 O=Al2 O3 b1:85; 0bCaO=SiO2 b0:3; 0:05bAl2 O3 =SiO2 b0:43:
EDS spot analysis on a 1 h old sample indicates the presence of
K-A-S-H geopolymer (average of 15 points) with an average of,
K2O/Al2O3 = 0.47 ± 0.44, CaO/SiO2 = 0.05 ± 0.03, and Al2O3/SiO2 =
0.23 ± 0.09. Some points also indicated possible calcium substitution
in the geopolymer. EDS analysis of a 1 hour old sample with a repre-
sentative spectrum and an average atomic ratios of various elements
is given in Fig. 10. A distinctive and segregated C-A-S-H phase is not
found in the BSE images, which is similar to an earlier observation
made on class F fly ash-slag geopolymers and Class C fly ash
geopolymer [19,49]. The interaction volume for X-ray generation is
kept to minimum by working at a low voltage. However, the re-
sponse from EDS spot analysis could still come from multiple phases
whose size ranges from few nanometers to 1 μm, much smaller than
the interaction volume (~ 1 μm in diameter for the voltage of 7 kV as
predicted by Lloyd et al. [44]). Early in the reaction, as calcium dis-
solves, C-A-S-H can form randomly throughout the microstructure
and not necessarily around slag particles. This C-A-S-H can then pos-
sibly act as nucleation site for geopolymerization, developing an
intermixed microstructure of C-A-S-H and K-A-S-H at an early age.
 S. Puligilla, P. Mondal / Cement and Concrete Research 43 (2013) 70–80 77

(a) (b)

(c) (d)

(e) (f)

Fig. 9. SEM micrographs of Geo 0s and Geo 15s at different ages. a: Geo 0s—3 h, b: Geo 15s—3 h, c: Geo 0s—24 h, d: Geo 15s—24 h, e: Geo 0s—14 day, f: Geo 15s—14 day. SE mode
was performed at an accelerating voltage of 15 kV.

The possible presence of calcium substituted geopolymer cannot be

Fig. 10. BSE image of Geo 15s 1 hour old sample showing a typical spectrum of the
phase present along with average atomic ratios of various elements.
eliminated either. Early age samples could have a mixture of C-A-S-H
and potassium geopolymer and/or calcium substituted potassium
geopolymer.
The quantitative microanalysis of a 14 day-old sample, indicated
the presence of two phases, C-A-S-H and (Ca,K)-A-S-H. The average
composition of C-A-S-H (average of 5 points) is K2O/Al2O3 = 1.34 ±
0.50, CaO/SiO2 = 0.73 ± 0.06, and Al2O3/SiO2 = 0.12 ± 0.03 and that
of (Ca,K)-S-H (average of 15 points) is K2O/Al2O3 = 0.61 ± 0.46,
CaO/SiO2 = 0.23 ± 0.13, and Al2O3/SiO2 = 0.37 ± 0.15. The EDS analy-
sis of a 14 day-old sample with a representative spectrum and an av-
erage atomic ratio of various elements is given in Fig. 11. Later age
samples indicate a calcium substitution in the geopolymer as the
EDS spot analysis shows the presence of (Ca,K)-A-S-H extensively.
As the microstructure becomes denser with time, reaction becomes
diffusion-controlled and an inward reaction of slag occurs. This ex-
plains the presence of a distinctive C-A-S-H phase in the later ages.
The composition of C-A-S-H has K2O/Al2O3 = 1.34, indicates an alkali
binding as the C-A-S-H has higher tendency for alkali binding than
C-S-H [50]. Assuming that Al substitutes Si in tetrahedral sites of
C-S-H, alkalis are absorbed into C-S-H for charge balancing and for
78 S. Puligilla, P. Mondal / Cement and Concrete Research 43 (2013) 70–80
every one Al atom, one alkali atom will be substituted to balance the
charge (K/Al ~ 1,K2O/Al2O3 ~ 1). This is because the binding of alkali is
predicted to occur by a valence compensation mechanism, where an
imbalance in charge is created by the substitution of Al in Si tetrahe-
dral sites and is compensated by the inclusion of alkali [50].
4. Discussion
The present study indicates that the presence of calcium in a fly
ash-based geopolymer causes early onset of hardening. Based on
the observations of existing literature and those made in this study,
this study proposes the possible mechanisms and the role of calcium
during the hardening of geopolymer. The dissolution of silicon and
aluminum species from fly ash is very slow for the conditions used
in this particular study (low alkaline activator concentration and am-
bient temperature curing) and evident from the low reactivity of Geo
0s [47]. As calcium ions dissolve from raw materials (slag or CaO),
they preferentially combine with Si ions in the solution to precipitate
as calcium silicate hydrate, which could be rich in alkalis [6,14]. The
precipitation of C-S-H is preferred over Ca(OH)2 due to its very low
solubility [47]. Compatibility studies between N-A-S-H and C-A-S-H
gels by García-Lodeiroet al. have shown that in the presence of Ca,
the precipitation of C-A-S-H and N-A-S-H is dependent on the pH of
the system. At a high pH (>12), C-A-S-H is preferred over an alumi-
nosilicate geopolymer [21]. The precipitation of C-A-S-H in the sys-
tem forms a basic skeleton of percolating solids which define the
time for the onset of hardening.
The present study also shows that in the presence of calcium, the
hardening rate of fly ash geopolymers increases even when all of the
other parameters are kept constant. This rapid hardening is not only
due to C-A-S-H formation at an early age but is also due to an increase
in the rate of geopolymerization. In this study, an EDS microanalysis
performed at an early age indicates the presence of K-A-S-H
geopolymer. Rapid geopolymerization in the presence of calcium is
also supported by the SEM micrographs presented in this paper. Re-
cently, a 3-D nanotomography study showed that aluminosilicates
form on or near fly ash particles [51]. In the current study, a SEM mi-
crograph of Geo 15s at 3 h (Fig. 9b) conclusively shows the product
formation on fly ash particles. Such microstructure is observed in
SEM micrographs as early as 45 min after mixing for Geo 15s, during
the rapid hardening phase (micrograph not included in the paper as it
Fig. 11. BSE image of Geo 15s 14 day old sample showing typical spectra of two differ-
ent phases along with average atomic ratios of various elements.
is not possible to capture a micrograph for Geo 0s at that early age for
comparison).
Yip et al. suggested that, in the presence of calcium, C-A-S-H pre-
cipitates and acts as nucleation sites, and this promotes the rapid for-
mation of geopolymer gel [9]. Yip's hypothesis also suggests a
formation of calcium hydroxide formation that could potentially act
as a nucleation site for a geopolymer formation [9,19]. X-ray diffrac-
tion patterns obtained in a parallel study did not indicate the forma-
tion of Ca(OH)2 or CaCO3. The presence of definite regions of
C-A-S-H is not detected from the microanalysis, owing to a large in-
teraction volume of EDS and the intermixed microstructure of
C-A-S-H and geopolymer. Spectroscopic techniques could reveal the
presence of C-A-S-H in the early age samples which remains out of
the scope of this study. Additionally, as Si is preferentially consumed
to form C-A-S-H, the rate of fly ash dissolution is enhanced to maintain
the Si ion concentration, resulting in a release of aluminates. The forma-
tion of C-A-S-H also consumes water that can increase the alkalinity of
the system and further encourage the dissolution of fly ash particles
[20]. This process increases the rate of poly-condensation and alumino-
silicate geopolymer formation. Furthermore, a rapid geopolymer gel
formation, due to calcium acting as a charge balancer, cannot be elimi-
nated at this point. A recent study has confirmed the preferential ad-
sorption of Ca into the geopolymer until Ca ions are exhausted,
replacing some of the anions and forming (Ca, M)-A-S-H gels where
M = K, Na [52]. However, this could not be confirmed based on the
EDS experiments done as part of this study at an early age. The presence
of a small amount of calcium reported in the average chemical formula
could be a result of intermixed C-A-S-H and K-A-S-H or due to the cal-
cium substitution in the K-A-S-H geopolymer.
Based on the calorimetry data and SEM micrographs, the presence of
calcium not only increases the rate reaction but also the extent of reac-
tion and the product formation. Comparing the slag mixes with the CaO
mixes, the presence of free calcium seems to initiate hardening in both
systems. After the initial rapid stiffening, it is observed from both UWR
and Proctor tests that the stiffening rate of geopolymer with CaO slows
down more significantly compared to that with slag. Since CaO dis-
solves much faster, most of the calcium may be consumed sooner. How-
ever, slag dissolves slowly and maintains the calcium ion supply for a
longer period. The rate of stiffening seems to decrease as the availability
of calcium ions decreases. By comparing mixes with slag and CaO, the
availability of free calcium ions for a longer period seems to prolong
fly ash dissolution and increase the extent of aluminosilicate gel forma-
tion. This can also explain the observed increase in the compressive
strength of fly ash-based geopolymer in the presence of calcium from
slag, similar to the results reported by others [9,15,20,45].
Yip et al. proposed that the increase in the compressive strength
might be due to the coexistence of C-S-H and geopolymeric gel [9,15].
At a high initial pH, the presence of Ca could degrade the K-A-S-H gel
forming on fly ash spheres as K-A-S-H gel is only stable in the absence
of calcium ions. Calcium ions can preferentially form (Ca, K)-A-S-H as
the pH of the system lowers in the later age [21]. Yip et al. also reported
that the presence of Ca could disrupt the three-dimensional ordering of
geopolymers. A recent study, however, reports that Ca ion substitution
into a geopolymer can significantly alter the chemical composition
without disrupting the gel structure [9,52]. This could complicate the
differentiation of calcium-based geopolymers from others using only
imaging techniques. The EDS microanalysis performed in this study on
a later age sample indicates calcium substitution in the geopolymer.
The average composition of microstructure resembles that of the (Ca,
K)-A-S-H. In addition, the presence of C-A-S-H indicates the ongoing
diffusion-controlled chemical reaction even after a dense microstruc-
ture has been formed. Therefore, it can be concluded that the initiation
of hardening is due to C-S-H/C-A-S-H formation and the hardening con-
tinues due to a rapid formation of a combination of C-A-S-H, K-A-S-H
and (Ca, K)-A-S-H depending on the availability of Ca ions and pH of
the system.
 S. Puligilla, P. Mondal / Cement and Concrete Research 43 (2013) 70–80
5. Conclusions
This study shows that calcium dissolving from slag is important
for both early and late age properties. Slow reaction rate and low
strength development has been confirmed when fly ash with low cal-
cium content is activated with low concentration alkali activator and
cured without any heat treatment. Previous studies have reported ac-
celerated hardening when calcium was added to fly ash, but for the
first time, two different methods, a continuous ultrasonic wave
based method and a penetration method are used to carefully moni-
tor the hardening over time. It is observed that the drop in the
S-wave reflection coefficient is directly related to the strength gain
measured from the Proctor test and the microstructural development
observed from the SEM micrographs. Both the UWR and Proctor
method were found to be sensitive to the rate of reaction and clearly
indicate the onset of hardening due to initial product formation, and a
slowing down of the hardening at the beginning of the induction
period.
It is proposed that the hardening process is initiated by the precip-
itation of C-S-H/C-A-S-H, and rapid hardening continues on account
of an accelerated geopolymerization. Unlike in the case of fly ash
geopolymers, the microstructure of slag substituted geopolymers
shows a basic network formation due to the formation of aluminosil-
icate geopolymer at a very early age. Product formation was observed
not only around the slag particles but also on the fly ash spheres. This
formation emphasizes that the presence of free calcium enhances the
rate and the extent of fly ash dissolution and extent of product forma-
tion. A slow dissolution of calcium from slag effectively increases
compressive strength as rapid geopolymerization continues for a lon-
ger duration. With a higher amount of slag replacement, calorimetry
curves also showed a second peak due to further product formation.
EDS microanalysis indicated the presence of K-A-S-H at an early age
and calcium substituted geopolymer, (Ca, K)-A-S-H at a later age.
Acknowledgments
The authors would like to thank Prof. Leslie J. Struble, Prof. John S.
Popovics and Prannoy Suraneni for their invaluable help with the use
of UWR and Proctor instruments. X-ray diffraction and SEM-EDS anal-
ysis were carried out in part in the Frederick Seitz Materials Research
Laboratory Central Facilities, University of Illinois, which are partially
supported by the U.S. Department of Energy under grants DE FG02 07
07ER46453 and DE FG02 07ER46471. Special thanks to anonymous
reviewers for insightful comments that helped to improve the
manuscript.
References
[1] J. Davidovits, Geopolymer Chemistry and Applications, Institut Geopolymere,
Saint-Quentin, France, 2008.
[2] H.W. Nugteren, V.C.L. Butselaar-Orthlieb, M. Izquierdo, High strength geopolymers
from fractionated and pulverized fly ash, in: World of Coal Ash (WOCA) Conference,
Lexington, KY, USA, 2009.
[3] M. Izquierdo, X. Querol, J. Davidovits, D. Antenucci, H. Nugteren, C. Fernández-
Pereira, Coal fly ash-slag-based geopolymers: microstructure and metal leaching,
J. Hazard. Mater. 166 (1) (2009) 561–566.
[4] J.G.S. van Jaarsveld, J.S.J. van Deventer, Effect of the alkali metal activator on the
properties of fly ash-based geopolymers, Ind. Eng. Chem. Res. 38 (10) (1999)
3932–3941.
[5] V. Živica, Effects of type and dosage of alkaline activator and temperature on the
properties of alkali-activated slag mixtures, Constr. Build. Mater. 21 (7) (2007)
1463–1469.
[6] S.A. Bernal, J.L. Provis, V. Rose, R. Mejía de Gutierrez, Evolution of binder structure
in sodium silicate-activated slag-metakaolin blends, Cem. Concr. Compos. 33 (1)
(2011) 46–54.
[7] W.K.W. Lee, J.S.J. van Deventer, Effects of anions on the formation of aluminosili-
cate gel in geopolymers, Ind. Eng. Chem. Res. 41 (18) (2002) 4550–4558.
[8] J.J. Chang, A study on the setting characteristics of sodium silicate-activated slag
pastes, Cem. Conr. Res. 33 (2003) 1005–1011.
[9] C.K. Yip, G.C. Lukey, J.L. Provis, J.S.J. van Deventer, Effect of calcium silicate sources
on geopolymerisation, Cem. Concr. Res. 38 (4) (2008) 554–564.
79
[10] S. Antiohos, S. Tsimas, Activation of fly ash cementitious systems in the presence
of quicklime. Part I. Compressive strength and pozzolanic reaction rate, Cem.
Concr. Res. 34 (5) (2004) 769–779.
[11] K. Dombrowski, A. Buchwald, M. Weil, The influence of calcium content on the
structure and thermal performance of fly ash based geopolymers, J. Mater. Sci.
42 (9) (2007) 3033–3043.
[12] W.K.W. Lee, J.S.J. van Deventer, The effect of ionic contaminants on the early-age
properties of alkali-activated fly ash based cements, Cem. Concr. Res. 32 (4)
(2002) 577–584.
[13] S. Kumar, R. Kumar, S.P. Mehrotra, Influence of granulated blast furnace slag on
the reaction, structure and properties of fly ash based geopolymer, J. Mater. Sci.
45 (3) (2010) 607–615.
[14] C.K. Yip, J.S.J. van Deventer, Microanalysis of calcium silicate hydrate gel formed
within a geopolymeric binder, J. Mater. Sci. 38 (18) (2003) 3851–3860.
[15] C.K. Yip, G.C. Lukey, J.S.J. van Deventer, The coexistence of geopolymeric gel and
calcium silicate hydrate at the early stage of alkaline activation, Cem. Concr.
Res. 35 (9) (2005) 1688–1697.
[16] S. Alonso, A. Palomo, Alkaline activation of metakaolin and calcium hydroxide
mixtures: influence of temperature, activator concentration and solids ratio,
Mater. Lett. 47 (1–2) (2001) 55–62.
[17] S. Alonso, A. Palomo, Calorimetric study of alkaline activation of calcium hydroxide–
metakaolin solid mixtures, Cem. Concr. Res. 31 (1) (2001) 25–30.
[18] P.J. Williams, J.J. Biernacki, L.R. Walker, H.M. Meyer, C.J. Rawn, J. Bai, Microanalysis
of alkali-activated fly ash-CH pastes, Cem. Concr. Res. 32 (6) (2002) 963–972.
[19] R.R. Lloyd, J.L. Provis, J.S.J. van Deventer, Microscopy and microanalysis of inor-
ganic polymer cements 2: the gel binder, J. Mater. Sci. 44 (2) (2009) 620–631.
[20] J. Temuujin, A. Van Riessen, R. Williams, Influence of calcium compounds on the
mechanical properties of fly ash geopolymer pastes, J. Hazard. Mater. 167 (1–3)
(2009) 82–88.
[21] I. García-Lodeiro, A. Palomo, A. Fernández-Jiménez, D.E. Macphee, Compatibility
studies between N-A-S-H and C-A-S-H Gels. Study in the ternary diagram
Na2O–CaO–Al2O3–SiO2–H2O, Cem. Concr. Res. 41 (2) (2011) 923–931.
[22] T.W. Cheng, J.P. Chiu, Fire-resistant geopolymer produced by granulated blast fur-
nace slag, Miner. Eng. 16 (3) (2003) 205–210.
[23] P. De Silva, K. Sagoe-Crenstill, The effect of Al2O3 and SiO2 on setting and harden-
ing of Na2O–Al2O3–SiO2–H2O geopolymer systems, J. Aust. Ceram. Soc. 44 (1)
(2008) 39–46.
[24] J.J. Chang, A study on the setting characteristics of sodium silicate-activated slag
pastes, Cem. Concr. Res. 33 (7) (2003) 1005–1011.
[25] K. Komnitsas, D. Zaharaki, V. Perdikatsis, Effect of synthesis parameters on the
compressive strength of low-calcium ferronickel slag inorganic polymers, J. Haz-
ard. Mater. 161 (2–3) (2009) 760–768.
[26] P. Duxson, G.C. Lukey, F. Separovic, J.S.J. van Deventer, Effect of alkali cations on alumi-
num incorporation in geopolymeric gels, Ind. Eng. Chem. Res. 44 (4) (2005) 832–839.
[27] L.J. Struble, T.Y. Kim, H. Zhang, Setting of cement and concrete, Cem. Concr.
Aggreg. 23 (2) (2001) 88–93.
[28] J. Lawson, B. Varela, R.S.P. Panandiker, M. Helguera, Determining the elastic prop-
erties of geopolymers using non-destructive ultrasonic techniques, Dev. Strateg.
Mater. Ceram. Eng. Sci. Proc. 29 (10) (2008) 143–153.
[29] J. L. Lawson, On the determination of the elastic properties of geopolymeric ma-
terials using non-destructive ultrasonic techniques, Master's thesis, Rochester In-
stitute of Technology, Rochester, US, 2008
[30] A. Buchwald, R. Tatarin, D. Stephan, Reaction progress of alkaline-activated
metakaolin-ground granulated blast furnace slag blends, J. Mater. Sci. 44 (20)
(2009) 5609–5617.
[31] ASTM Standard C403/C403-08, Standard Test Method for Time of Setting of Concrete
Mixtures by Penetration Resistance, ASTM International, West Conshohocken, PA,
2008.
[32] C.W. Chung, M. Mroczek, I.Y. Park, L.J. Struble, On the evaluation of setting time of
cement paste based on ASTM C403 penetration resistance test, J. Test. Eval.
8 (2010) 527–533.
[33] C.W. Chung, J.S. Popovics, L.J. Struble, Early Age Stiffening of Cement Paste Using
Ultrasonic Wave Reflection, Transition from Liquid to Solid: Re-examining the Be-
havior of Concrete at Early Ages (ACI SP-259-1), in: Kyle Riding (Ed.), American
Concrete Institute, Farmington Hills, MI, 2009, pp. 7–16.
[34] Z. Sun, G. Ye, T. Voigt, S.P. Shah, K. van Breugel, Early age properties of portland
cement pastes investigated with ultrasonic shear waves and numerical simula-
tion, in: International RILEM Symposium on Concrete Science and Engineering:
A Tribute to Arnon Bentur, RILEM Publications SARL, 2004.
[35] K.V. Subramaniam, J. Lee, Ultrasonic assessment of early-age changes in the
material properties of cementitious materials, Mater. Struct. 40 (3) (2007)
301–309.
[36] T. Ozturk, O. Kroggel, P. Grubl, J.S. Popovics, Improved ultrasonic wave reflection
technique to monitor the setting of cement-based materials, NDT and E Int. 39 (4)
(2006) 258–263.
[37] C.W. Chung, P. Suraneni, J.S. Popovics, L.J. Struble, Setting time measurement
using ultrasonic wave reflection, ACI Mater. J. 109 (2012) 109–118.
[38] T. Voigt, S.P. Shah, Properties of early-age portland cement mortar monitored
with shear wave reflection method, ACI Mater. J. 101 (6) (2004) 473–482.
[39] C.W. Chung, J.S. Popovics, L.J. Struble, Using ultrasonic wave reflection to measure
solution properties, Ultrason. Sonochem. 17 (1) (2010) 266–272.
[40] D.P. Bentz, P.V. Coveney, E.J. Garboczi, M.F. Kleyn, P.E. Stutzman, Cellular autom-
aton simulations of cement hydration and microstructure development, Modell.
Simul. Mater. Sci. Eng. 2 (1994) 783.
[41] C.W. Chung, J.S. Popovics, L.J. Struble, Flocculation and sedimentation in suspen-
sions using ultrasonic wave reflection, J. Acoust. Soc. Am. 129 (2011) 2944.
80 S. Puligilla, P. Mondal / Cement and Concrete Research 43 (2013) 70–80
[42] C. W. Chung, Ultrasonic wave reflection measurements on stiffening and setting
of Cement Paste, Ph.D. thesis, University of Illinois, Urbana Champaign, US, 2010.
[43] S.S. Go, C.W. Chung, L.J. Struble, H.C. Lee, Pozzolanic activity of Hwangtoh clay,
Constr. Build. Mater. 24 (12) (2010) 2638–2645.
[44] R.R. Lloyd, J.L. Provis, J.S.J. van Deventer, Microscopy and microanalysis of inor-
ganic polymer cements. 1: remnant fly ash particles, J. Mater. Sci. 44 (2) (2009)
608–619.
[45] F. Puertas, S. Martínez-Ramírez, S. Alonso, T. Vazquez, Alkali-activated fly ash/slag
cements: strength behaviour and hydration products, Cem. Concr. Res. 30 (10)
(2000) 1625–1632.
[46] H. Zhou, X. Wu, Z. Xu, M. Tang, Kinetic study on hydration of alkali-activated slag,
Cem. Concr. Res. 23 (6) (1993) 1253–1258.
[47] C. Shi, P.V. Krivenko, D.M. Roy, Alkali-Activated Cements and Concretes, Taylor
and Francis, Abingdon, UK, 2006.
[48] A. Fernández-Jiménez, F. Puertas, Alkali-activated slag cements: kinetic studies,
Cem. Concr. Res. 27 (3) (1997) 359–368.
[49] J.E. Oh, J. Moon, S.G. Oh, S.M. Clark, P.J.M. Monteiro, Microstructural and compo-
sitional change of NaOH-activated high calcium fly ash by incorporating
Na-aluminate and co-existence of geopolymeric gel and C-S-H (I), Cem. Concr.
Res. 42 (5) (2012) 673–685.
[50] S.Y. Hong, F.P. Glasser, Alkali sorption by C-S-H and C-A-S-H gels: part II. Role of
alumina, Cem. Concr. Res. 32 (7) (2002) 1101–1111.
[51] J.L. Provis, V. Rose, R.P. Winarski, J.S.J. van Deventer, Hard X-ray nanotomography
of amorphous aluminosilicate cements, Scr. Mater. 65 (4) (2011) 316–319.
[52] I. García-Lodeiro, A. Fernández-Jiménez, A. Palomo, D.E. Macphee, Effect of calci-
um additions on N-A-S-H cementitious gels, J. Am. Ceram. Soc. 1940 (7) (2010)
1934–1940.