Cement & Concrete Composites 32 (2010) 399–410

Contents lists available at ScienceDirect

Cement & Concrete Composites

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

Structure and strength of NaOH activated concretes containing fly ash or GGBFS

as the sole binder
Deepak Ravikumar, Sulapha Peethamparan, Narayanan Neithalath *
Department of Civil and Environmental Engineering, Clarkson University, Potsdam, NY 13699, USA

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

Article history: The influence of the concentration of the activating agent (4, 6, or 8 M sodium hydroxide solution), and
Received 10 September 2009 activator-to-binder ratio (0.40, 0.50, or 0.60) on the compressive strengths, pore structure features, and
Received in revised form 16 March 2010 microstructure of concretes containing Class F fly ash or ground granulated blast furnace slag (GGBFS) as
Accepted 17 March 2010
the sole binder is reported. The starting material contents and the curing parameters (temperature and
Available online 21 March 2010
curing duration) are optimized to provide the highest compressive strengths. Statistical analysis of the
compressive strength results show that the activator concentration has a larger influence on the com-
Keywords:
pressive strengths of activated concretes made using fly ash and the activator-to-binder ratio influences
Microstructure
Compressive strength
the compressive strengths of activated GGBFS concretes to a greater degree. Activated fly ash concretes
Fly ash and pastes are found to be more porous and contains a larger fraction of pores greater than 10 lm in size
Granulated blast furnace slag as compared to activated GGBFS mixtures. The differences in the microstructure and the reaction prod-
Cement-free binder ucts between activated fly ash and GGBFS pastes are detailed.
Porosity Ó 2010 Elsevier Ltd. All rights reserved.
Pore size

1. Introduction The starting material and activating agent type and concentra-
tion are the most important parameters that influence the prop-
Increasing emphasis on the environmental impacts of construc- erties of the alkali-activated end product [14,15]. The
tion materials such as Portland cement has provided immense production of cement-free binders requires a starting material
thrust in recent years to the increased utilization of waste and containing aluminum and silicon species that are soluble in
by-product materials in concretes. Activation of aluminosilicate highly alkaline solutions. Previous studies have shown that the
materials such as fly ash, blast furnace slag, and metakaolin using amount of vitreous silica and alumina present in the starting
alkaline solutions to produce binders free of Portland cement is a material plays a significant role in activation reactions and the
major advancement towards increasing the beneficial use of indus- properties of the reaction product [10,16]. The presence of cal-
trial waste products and reducing the adverse impacts of cement cium oxide in the source material also influences the properties,
production. These alkali-activated binder systems appear in the especially because of the formation of more than one reaction
vast available body of literature in a variety of names: alkali-acti- product [17]. Sodium hydroxide and sodium silicate are the more
vated cement [1] (predominantly used for binders containing large commonly used alkaline activating agents [18–20] even though
amounts of calcium such as slags), geopolymer [2–6], inorganic few studies have also been carried out with potassium hydroxide
polymer [7], hydroceramic, or low temperature aluminosilicate or sodium carbonate as the activating agent [13,21].
glass [8]. They are sometimes called polysialates also because of The dissolution stage during alkali activation of aluminosili-
the polymeric silicon–oxygen–aluminum framework [9]. In gen- cates is reported to be similar to that in the hydration of Portland
eral, concretes made using alkali activation of the starting alumi- cement at early ages [22,23]. The dissolved species and the active
nosilicate material do not contain Portland cement at all, and surface groups in the starting material undergo polymerization to
hence they are referred to generically as cement-free binder form the alkali-aluminosilicate gel with the aluminum cation in
(CFB) concretes in this study. The concretes containing alkali-alu- fourfold co-ordination with respect to oxygen [24,25] which hard-
minosilicate gel as the binder has been shown to have high com- ens to form the binder matrix. Mn[–(SiO2)z–AlO2]n–wH2O is the
pressive strengths, and resistance to fire, and chemical attack general empirical formula proposed for the reaction product
[10–13]. formed where M is the alkali ion, usually sodium or potassium, z
is 1, 2, or 3, and n is the degree of polymerization [9]. Only a small
* Corresponding author. Tel.: +1 315 268 1261; fax: +1 315 268 7985. fraction of the silica and alumina in the source material need to
E-mail address: [email protected] (N. Neithalath). dissolve and participate in the polymerization reaction for the

0958-9465/$ - see front matter Ó 2010 Elsevier Ltd. All rights reserved.
doi:10.1016/j.cemconcomp.2010.03.007
400 D. Ravikumar et al. / Cement & Concrete Composites 32 (2010) 399–410

mixture to solidify. The alkali-aluminosilicates produced from fly and 3.41 for fly ash and GGBFS, respectively. Apart from the high
ash and/or slag to be used as binding materials in concrete are lar- silica and alumina contents, GGBFS also has a high CaO content
gely X-ray amorphous, attributable to the low alkaline solution-to- (40%) while the CaO content in fly ash is very low (5.8%). X-ray
starting material ratio. Higher alkaline solution-to-fly ash ratio and diffraction (XRD) patterns of the starting materials are shown in
longer thermal curing durations have been reported to produce Fig. 1 and the particle size distributions (obtained using a laser par-
more crystalline reaction products [14,26]. Activation of alumino- ticle size analyzer) are shown in Fig. 2. The XRD pattern of fly ash
silicate materials with alkalis generally requires heat curing for shows a diffuse band corresponding to the glassy phase which
the formation of alkali-aluminosilicates especially when the acti- peaks at a 2h of about 23°. For GGBFS, an asymmetric diffuse band
vating solution does not contain soluble silica. A wide range of of glass is observed between 2h of 20° and 37°, with the peak at
temperatures ranging from 40 °C to 90 °C have been reported in or- 31°. The XRD patterns show that GGBFS is more amorphous which
der to produce alkali-activated binders with appreciable mechani- suggests better reactivity as compared to fly ash which contains
cal properties [18,27–29], with a general improvement in crystalline phases of silica and alumina. Particle size analysis
mechanical properties observed when higher temperatures are shows that GGBFS is finer than fly ash with 95% finer than 30 lm
used. compared to 60% for fly ash, which potentially influences its reac-
It has been reported that fly ash and ground granulated blast tivity. Analytic reagent-grade sodium hydroxide (NaOH) crystals
furnace slag (GGBFS) are very effective as starting materials for ce- were used to prepare the alkaline activating solutions of varying
ment-free binder concretes because of the soluble silica and alu- molar concentrations.
mina contents in these materials that undergo dissolution,
polymerization with the alkali, condensation on particle surfaces, 2.2. Mixing procedure
and solidification that eventually provides strength and stability
to these matrices. This paper focuses on the comparison of CFB Prior to the preparation of the concrete mixture, NaOH solu-
concretes and pastes made with fly ash or GGBFS as the starting tions of the desired concentrations were prepared and allowed
material, and activated with varying concentrations of sodium to stay at room temperature for 2 h. The mixing procedure in-
hydroxide solution. Since the mechanism of binding gel formation, volved initial mixing of the source material (fly ash or GGBFS),
even with the same activator, is different for both these starting fine aggregate, and the coarse aggregate (nominal maximum size
materials, there is a need to understand the effect of activator con- of 9.5 mm) for 2 min. NaOH solution of the desired concentra-
centration on the properties of CFB concretes made using different tion was then added, and mixed for further 4 min for CFB con-
starting materials. The relative influence of the concentration of cretes with fly ash and 2 min for CFB concretes with GGBFS to
the alkaline activator and the activator-to-binder ratio on the com- obtain a uniform mixture. The reason for using only 2 min of
pressive strengths and porosities of CFB concretes is evaluated in mixing for the GGBFS mixtures was the faster rate of setting
this paper so as to aid in the selection of mixture parameters when of NaOH activated GGBFS. The mixtures were then filled in cubi-
either fly ash or GGBFS is used as the starting material. Differences cal molds of 50 mm size and vibrated by placing on a table
in microstructure (morphology and phase composition) between vibrator until proper compaction was achieved. GGBFS mixtures
CFB pastes made with fly ash or GGBFS as the binding material were vibrated for 2 min as opposed to fly ash mixtures which
are also brought out in detail. needed only 1 min of vibration to achieve adequate compaction.
For the paste specimens to be used for microstructural analysis,
samples were prepared by mixing the activating solution and
2. Experimental program the binder in a blender for 2 min. Both the paste and concrete
specimens were removed from the molds after 24 h and sub-
2.1. Starting materials and activator jected to heat curing in a laboratory oven at either 60 °C or
75 °C for 24 h or 48 h.
To prepare CFB pastes and concretes in this study, a Class F fly
ash conforming to ASTM C 618 and a ground granulated blast fur-
nace slag (GGBFS) Type 100 conforming to ASTM C 989 were used
as the starting materials. The chemical compositions of fly ash and
GGBFS are given in Table 1. Both these binding materials are rich in
silica and alumina, which are required for the formation of the
strength imparting phases in CFB concretes. The silica-to-alumina
ratios (SiO2/Al2O3) by mass were found to be approximately 1.75

Table 1
Chemical composition and physical characteristics of the starting materials.

Composition/Property (% by mass)
Fly ash (%) GGBFS (%)
SiO2 50.2 36.0
Al2O3 28.7 10.5
Fe2O3 5.72 0.67
CaO 5.86 39.8
MgO 1.74 7.93
Na2O 0.96a 0.27
K2O 0.16
SO3 0.51 2.11
Loss on ignition (LOI) 2.80 3.00
Fineness (% finer than 30 lm) 60 95
Fig. 1. X-ray diffractions patterns of fly ash and GGBFS powder (Q: quartz, M:
a
Equivalent alkalis. mullite).
 D. Ravikumar et al. / Cement & Concrete Composites 32 (2010) 399–410 401

100 430 to 445 kg/m3 and the GGBFS content ranged from 545 to
575 kg/m3 for the different CFB concrete mixtures. CFB concretes
Fly ash
GGBFS were proportioned using the above determined starting material
80 contents, activator concentrations of 4 M, 6 M, and 8 M NaOH,
Percent passing (%)

and activator-to-binder ratios of 0.40, 0.50, and 0.60, to understand

the influence of these parameters on the compressive strength. A
60 total of 18 concrete mixtures were proportioned in this study.

2.4. Determination of strength, porosity, and pore sizes

40
After the heat curing duration, the concretes were allowed to
cool down and return to ambient temperatures. The compressive
20
strengths of the cubes were then determined in accordance with
ASTM C 109. No moist curing was adopted for any of the mixtures
after the heat curing procedure.
0
0.01 0.1 1 10 100 1000 10000
Particle size (µm) 2.4.1. Porosity from vacuum saturation method
The porosities of the CFB concretes and pastes were measured
Fig. 2. Particle size distribution of fly ash and GGBFS. in accordance with the procedure described in RILEM CPC 11.3
[30]. The 50 mm size CFB concrete or paste cubes were cut into
two equal halves after the heat curing duration. After measuring
2.3. Determination of optimal binder contents and mixture
the initial mass, the specimens were vacuum dried for 3 h and sub-
proportions
sequently saturated under vacuum for 1 h. The specimens were
kept submerged in water for a further 18 h after which they were
In order to obtain the binding material content (by volume) that
removed from water, surface dried, and weighed. The increase in
provides the highest compressive strength for the experimental
the mass of the specimen, represented as percentage of the volume
conditions used, CFB concretes were proportioned using fly ash
of the specimen, is reported as the porosity.
and GGBFS contents ranging from 15% to 30% by volume of con-
crete. The coarse aggregate to fine aggregate ratio by mass was
approximately 1.1 for all the concrete mixtures. NaOH solution of 2.4.2. Porosity and pore size of CFB pastes from image analysis
8 M concentration (prepared by dissolving 320 g of NaOH in 1 l Image analysis technique was used to determine the porosity
of water) was used as the activating agent at an activator-to-binder and pore size distribution of CFB pastes. The cut surfaces of CFB
mass ratio of 0.40. The specimens were heat cured at 75 °C for 48 h pastes were polished and divided into grids of 3 mm  3 mm size.
in a laboratory oven after demolding at 24 h. Fig. 3 shows the com- Optical microscope was used to capture images of these grids at a
pressive strengths of the CFB concretes as a function of either fly magnification of 20. A commercial image analysis and processing
ash or GGBFS contents. It can be seen that 18% fly ash content by software was used to extract the pore structure features. The cap-
volume and 25% GGBFS content by volume of the concrete mixture, tured color images were converted to grayscale images, and thres-
respectively provided the maximum compressive strengths in this holded to obtain binary images in which the solid (white) and pore
study. These volume fractions of the starting materials were used (black) areas are clearly differentiated as shown in Fig. 4a and b.
to proportion concretes for further experiments with different al- The sizes of all the images were 400 pixels  400 pixels (1 pix-
kali activator concentrations and activator-to-binder ratios. The to- el = 3/400 mm). Fig. 4a shows a binary image from a typical CFB
tal aggregate volume in the CFB concretes ranged from 65% to 70% paste containing fly ash as the starting material, and Fig. 4b shows
for CFB concretes with fly ash as binder and 55–60% for CFB con- the image of a typical paste containing GGBFS as the starting mate-
cretes with GGBFS as binder. The fly ash content ranged from rial. The pore area fractions (porosity) of these images were ob-
tained from the image analysis program. The frequency
distributions of the pore sizes (assuming the features to be circular
40 in shape) were also obtained for each image. The minimum pore
Fly ash CFB concrete size detected by the optical microscopy procedure was approxi-
GGBFS CFB concrete
mately 10 lm. A typical cumulative frequency distribution plot is
Compressive Strength (MPa)

35 shown in Fig. 4c. The pore size corresponding to 50% of the cumu-
lative frequency distribution plot is taken as the characteristic size
of the pores greater than 10 lm (d50(>10 lm)).
30

2.5. Microstructural analysis

25
To understand the morphology and the composition of the reac-
tion products formed in NaOH activated CFB pastes, morphological
20 analysis was performed using scanning electron microscopy (SEM)
coupled with energy dispersive X-ray analysis (EDX) for elemental
18% (FA) 25% (GGBFS) composition analysis. X-ray diffraction (XRD) was used to deter-
15 mine the phase composition of the reaction products. The XRD pat-
12 16 20 24 28 32 terns of the powdered paste samples were recorded using a Bruker
Binder content by volume (%) DX-8 diffractometer. Cu Ka radiation (k = 1.504 Å) was used and
the tests were carried out in the 2h range of 5–40°, with a step size
Fig. 3. Determination of optimum starting material content for CFB concretes. of 0.02° and step time of 2 s.
402 D. Ravikumar et al. / Cement & Concrete Composites 32 (2010) 399–410

(c) 100

(a) (b) 90

Cummulative Frequency (%)

80
70
60
50
40
30
20
10
d50
0
0 40 80 120 160 200
Pore size (µm)

Fig. 4. (a) Binary image for CFB paste with fly ash, (b) binary image of CFB paste with GGBFS, and (c) a typical cumulative frequency distribution plot of pore sizes.

3. Results, analysis, and discussion ash and GGBFS CFB concretes are compared at the end of the
respective curing durations.
3.1. Compressive strength of CFB concretes as functions of curing Increasing the curing duration from 12 h to 48 h increased the
temperature and duration compressive strengths significantly in all the cases as seen from
Fig. 5. As observed from Fig. 5a, curing at 60 °C did not produce
Heat curing is required to achieve adequate mechanical proper- appreciable compressive strengths for CFB concretes made with
ties for CFB concretes activated using NaOH. In order to obtain the fly ash as the starting material, irrespective of the curing duration.
curing temperature and duration that provides the maximum com- Increasing the curing temperature to 75 °C resulted in significant
pressive strengths, CFB concretes made using an activator-to-bin- improvements in compressive strengths for these concretes. This
der ratio of 0.40, containing the optimal fly ash or GGBFS point to the fact that a curing temperature of 75 °C is necessary
contents (18% or 25% by volume, respectively), and activated using to produce sufficient amounts of sodium aluminosilicate gel neces-
8 M NaOH solution were subjected to temperatures of 60 °C and sary to impart strength to CFB concretes containing fly ash. While a
75 °C for 12, 24 or 48 h. Fig. 5a and b show the compressive curing temperature greater than 75 °C is expected to further in-
strengths of CFB concretes made using fly ash or GGBFS as the crease the strength [26], such an attempt is not made in this study
starting materials respectively, as functions of both curing temper- considering the even higher energy requirements.
ature and curing duration. The compressive strengths reported are From Fig. 5b, it can be seen that for CFB concretes made using
those determined at the end of the respective curing durations. The GGBFS as the starting material, heat curing for 48 h at 60 °C pro-
poly-condensation of the aluminosilicate gel is likely to continue duces appreciable compressive strengths, even though they are
even after this duration, and strength improvements might occur, lower than those cured at 75 °C for the same duration. The differ-
especially for mixtures that contain GGBFS as the starting material ence in compressive strengths between the specimens heat cured
because of the transformation of metastable amorphous gel to at 60 °C or 75 °C is not as distinct for CFB concretes made using
more stable crystalline phases [3]. However, this paper does not GGBFS as that for those made using fly ash, especially at longer
consider the above aspect and the compressive strengths of fly curing durations. This is indicative of the influence of processing

50
Compressive strength(MPa)

50
Compressive strength(MPa)

40
40

30
30

20 20

10 10

0 ) 0
48 urs 48 urs
)
Cur 75 24 (Ho Cur 75 24 (Ho
n
ing 12 tio ing tio
n
tem 60 ura tem 60 12
ura
per g D per
atur D
atur rin g
e (º Cu e (º rin
(a) C) (b) C) Cu

Fig. 5. Influence of curing temperatures and durations on the compressive strengths of CFB concretes with: (a) fly ash, and (b) GGBFS as the starting material.
 D. Ravikumar et al. / Cement & Concrete Composites 32 (2010) 399–410 403

conditions on the amount and structure of the reaction product (al- contributing to a portion of the increased strength. As could be ob-
kali-aluminosilicate gel) in CFB concretes made using fly ash or served from Fig. 3, the binder contents were chosen so as to max-
GGBFS as the starting material. While the presence of calcium imize the compressive strengths of CFB concretes using fly ash and
oxide in GGBFS would have facilitated the formation of some cal- GGBFS as the binder, and not on obtaining identical strengths.
cium bearing strength giving components even at lower tempera- For CFB concretes made with fly ash as the starting material, the
tures, the strength development of CFB concretes made with fly compressive strengths obtained by activation using 4 M and 6 M
ash is solely dependent on the formation of sodium aluminosilicate NaOH were low and almost similar, and activation with 8 M NaOH
gel. Based on these observations, the results reported further in resulted in increased compressive strengths, pointing to insuffi-
this paper are for specimens (both concretes and pastes) heat cient activation of fly ash at low concentrations of the activator.
cured at 75 °C for 48 h, since this combination of curing tempera- For CFB concretes with GGBFS as the starting material, Fig. 6b
ture and duration provided the highest compressive strengths. shows that activation with lower concentrations of NaOH (4 M
and 6 M) also provided reasonable compressive strengths unlike
3.2. Dependence of compressive strengths of CFB concretes on CFB concretes with fly ash. The relatively lower difference in
activator-to-binder ratios and molar concentration of NaOH strengths between CFB concretes made using GGBFS activated by
either 6 M or 8 M NaOH as compared to that of CFB concretes with
The influence of activator-to-binder ratios and molar concentra- fly ash is evidence of the relatively lesser influence of activator
tions of NaOH on the compressive strengths of CFB concretes con- concentration (within the ranges studied) when GGBFS is used as
taining fly ash or GGBFS are shown in Fig. 6a and b, respectively. the starting material.
These figures show that the compressive strengths increase with Since the optimum binder contents for CFB concretes made
increase in the molar concentration of NaOH used as the activator. with GGBFS as the starting material were higher than those for
Compressive strengths are found to decrease with increase in CFB concretes made with fly ash, the compressive strengths of dif-
activator-to-binder ratio even though a higher activator-to-binder ferent CFB concretes cannot be compared directly. Hence the mea-
ratio effectively contains a higher amount of the activating alkali. sured compressive strengths were normalized by the binder
This shows that the higher porosity created because of the in- content used in the particular mixture. Fig. 7a and b shows the nor-
creased amounts of the activating solution, and the substantial in- malized compressive strengths for CFB concretes made using fly
crease in number of pores greater than 10 lm as result of curing in ash and GGBFS respectively as the starting material. It can be
dry heat (stated in [31], and experimentally shown later in this pa- immediately noticed from these figures that, at higher activator
per) are more prominent than the effect of increased NaOH con- concentrations, the CFB concretes show very similar strengths
centration (that potentially could produce increased amounts of per unit binder content irrespective of the binder type and the acti-
reaction product) on the compressive strength in the range of acti- vator-to-binder ratio. This is an indication of better utilization of
vator concentrations and activator-to-binder ratios used in this fly ash particles in forming the reaction product when higher con-
study. A closer look at these figures also indicates that the strength centrations of the activator are used. At lower concentrations of the
drop with increasing activator-to-binder ratio is greater for CFB activator, CFB concretes with GGBFS show higher compressive
concretes made with GGBFS as the starting material. This might strengths than the ones made with fly ash. This once again shows
be due to the increased capillary porosity created in the cementi- that the activator concentration has a greater influence on the
tious material hydration product because of the increase in activa- compressive strength of CFB concretes with fly ash, which is con-
tor-to-binder ratio. firmed using statistical analysis of the compressive strength results
The activation of GGBFS with NaOH produced concretes of high- later in this section. While compressive strengths of the CFB con-
er compressive strengths than those produced by the activation of cretes with fly ash rely solely on the formation of sodium alumino-
fly ash. The higher reactivity and self-cementing property of GGBFS silicate gel, for CFB concretes with GGBFS, some calcium bearing
could partly be the reason for the higher compressive strengths of compounds also could be forming even at lower activator concen-
CFB concretes made using GGBFS. It needs to be also remembered trations. This could be a reason for their increased strengths at low-
that a higher volume of GGBFS could be incorporated into the CFB er activator concentrations as compared to CFB concretes
concrete mixtures for maximum compressive strength, thereby containing fly ash. Microstructural analysis of pastes made with

(a) 50 (b) 50
4M NaOH 4M NaOH
Compressive strength (MPa)

Compressive strength (MPa)

6M NaOH 6M NaOH
40 8M NaOH 40 8M NaOH

30 30

20 20

10 10

0 0
0.3 0.4 0.5 0.6 0.7 0.3 0.4 0.5 0.6 0.7
Activator to binder ratio Activator to binder ratio

Fig. 6. Compressive strengths of CFB concretes as a function of activator-to-binder ratio: (a) fly ash as the starting material, and (b) GGBFS as the starting material.
404 D. Ravikumar et al. / Cement & Concrete Composites 32 (2010) 399–410

(a) 1 (b) 1
Compressive strength normalised

Compressive strength normalised

4M NaOH 4M NaOH
with binder content (MPa/kg/m 3)

with binder content (MPa/kg/m 3)

6M NaOH 6M NaOH
8M NaOH 8M NaOH

0.1 0.1

0.01 0.01
0.3 0.4 0.5 0.6 0.7 0.3 0.4 0.5 0.6 0.7
Activator to binder ratio Activator to binder ratio

Fig. 7. Normalized compressive strengths of CFB concretes as a function of activator-to-binder ratio: (a) fly ash as the starting material, and (b) GGBFS as the starting material.

lower and higher activator concentrations is dealt with in a forth- strength results detailed earlier, which were shown to decrease
coming section. with increase in activator-to-binder ratio. From Fig. 9a, it can be
In order to obtain a better understanding of the influence of two seen that the CFB concretes made using fly ash and activated
factors – activator concentration (a) and activator-to-binder ratio using 4 M NaOH exhibits the highest porosity values. The porosity
(b) – on the compressive strength (fc0 ), a 32 full factorial analysis decreases with increase in activator concentration, attributable to
was carried out with each factor at three levels. The levels of acti- the better activation efficiency and formation of increased
vator concentration (a) chosen were 4 M, 6 M, 8 M, and the levels amounts of reaction products. Similar trends are also observed
for the activator-to-binder ratio (b) chosen were 0.40, 0.50, and in Fig. 9b for CFB concretes made using GGBFS. The porosities
0.60. A quadratic model was found to better represent the com- of CFB concretes made using GGBFS are found to be lower than
pressive strengths of CFB concretes. Analysis of variance was car- those of CFB concretes with fly ash. The activator concentration
ried out to facilitate elimination of the parameter combination(s) is found to have a lesser influence on the porosity of CFB con-
in the model that do not significantly influence the compressive cretes made using GGBFS as compared to CFB concretes made
strengths. The quadratic equations obtained from the factorial using fly ash. This is in agreement with the trends for compres-
analysis for the compressive strengths of CFB concretes made with sive strength also.
fly ash and GGBFS, respectively are given in Eqs. (1) and (2), and The relationship between compressive strengths and porosities
the corresponding three-dimensional response surfaces are shown for all the CFB concretes (made using either fly ash or GGBFS as the
in Fig. 8a and b. binder, and with varying activator concentrations and activator-to-
binder ratios) is shown in Fig. 10. The porosity values ranged from
0 2
fcfly ash ¼ 26:4  7:7a þ 16:5b  9:0ab þ 1:3a ð1Þ 7% to 16%. The compressive strength–porosity relationship shows
an exponential trend, similar to that of conventional concretes.
0 2 For lower strength CFB concretes (<15 MPa), the trend seems to
fcGGBFS ¼ 65:0 þ 7:2a  221:7b  7:3ab ¼ 200:5b ð2Þ
deviate from that of the mixtures having a higher compressive
In these equations, the values of fc0 are in MPa, and the activator strength.
concentration in mol/l. A closer look at the above equations show A 32 factorial analysis was carried out to model the influence of
that the square of the activator-to-binder ratio (b) does not have activator concentration (a) and activator-to-binder ratio (b) on the
an influence on the compressive strengths of CFB concretes made porosity of CFB concretes made using fly ash or GGBFS. Based on
using fly ash, but is influential for CFB concretes made using the analysis of variance tests, it was found that a linear model suf-
GGBFS. Similarly the square of the activator concentration (a) does fices to represent porosity (%) as a function of activator concentra-
not influence the strengths of CFB concretes made using GGBFS but tion (mol/l) and activator-to-binder ratio. These relationships are
influences the strength of CFB concretes made using fly ash. The provided in Eqs. (3) and (4) for CFB concretes made using fly ash
accuracy of the developed models was confirmed from the normal and GGBFS respectively. The adequacy of the model was ascer-
probability plots of residuals, which fell on a straight line. tained from the normal probability plot of the residuals. Inspection
of the coefficients for the variables in Eqs. (3) and (4) suggest that
3.3. Pore structure of CFB concretes and pastes the activator-to-binder ratio (b) exerts significant influence on the
porosity of the CFB concretes, especially at higher activator con-
3.3.1. Porosities of CFB concretes centrations (a).
Fig. 9a and b shows the porosities of CFB concretes made using
/FA ð%Þ ¼ 5:3  0:8a þ 25:7b ð3Þ
fly ash or GGBFS as the starting materials as functions of the acti-
vator–binder ratio, at various activator concentrations. The poros-
/GGBFS ð%Þ ¼ 2:8  0:4a þ 21:9b ð4Þ
ities of CFB concretes were determined using a vacuum saturation
method as explained in earlier.
From Fig. 9a and b, it can be observed that the porosities of 3.3.2. Porosities and pore sizes of CFB pastes
CFB concretes increase with increase in the activator-to-binder The porosities of CFB pastes were determined using a vacuum
ratio as expected. This is in agreement with the compressive saturation method as well as using image analysis procedures on
 D. Ravikumar et al. / Cement & Concrete Composites 32 (2010) 399–410 405

Fig. 8. Response surfaces showing the influence of activator-to-binder ratios and activator concentrations on the compressive strengths of CFB concretes made with: (a) fly
ash, and (b) GGBFS as the starting materials.

polished sections. The porosities and pore sizes were investigated trends are observed between the paste porosities obtained either
only for those GGBFS and fly ash CFB pastes made using activa- using vacuum saturation method or image analysis method. How-
tor-to-binder ratios of 0.40 and 0.60, and an activator concentra- ever the porosities obtained using image analysis method are only
tion of 8 M. Fig. 11a shows the porosity (using vacuum about half of the porosities obtained from the vacuum saturation
saturation method) of CFB pastes made using fly ash or GGBFS as method. This is because the image analysis method captures only
the starting material. The porosities of CFB pastes made using those pores of size equal to or larger than 10 lm (because of the
GGBFS are seen to be lower than for those made with fly ash; a magnification used). However it can be safely stated from these re-
trend that was observed for concretes also. The porosities of CFB sults that about half of the total pore volume in CFB pastes made
concretes made using similar activator-to-binder ratios and activa- with either fly ash or GGBFS as the starting material is contributed
tor concentration are shown as symbols along with bars for the by pores of size less than 10 lm. The distribution of pore sizes
porosities of the paste in Fig. 11a. The porosities of the CFB pastes could be an influential parameter in the moisture and ionic trans-
are about three times more than those of concrete, because in CFB port in alkali-activated concretes.
concretes, the paste volume is only about a third of the total Fig. 12 shows the characteristic pore size (pore size correspond-
volume. ing to 50% of the cumulative frequency distribution plot) in CFB
The porosities of the CFB pastes obtained using the image anal- pastes made using fly ash or GGBFS. The d50 values as represented
ysis procedure is presented in Fig. 11b. The results are averages here correspond to the characteristic value of pore sizes for pores
from 10 images corresponding to each specimen and the error bars greater than 10 lm (d50(>10 lm)). Thus these values should not be
represent one standard deviation from the mean value. Similar regarded as true indicators of the actual pore sizes in the material.
406 D. Ravikumar et al. / Cement & Concrete Composites 32 (2010) 399–410

(a) 20 (b) 20
4M NaOH 4M NaOH
18 6M NaOH 18 6M NaOH
8M NaOH 8M NaOH
16 16
Porosity (%)

Porosity (%)
14 14

12 12

10 10

8 8

6 6

4 4
0.3 0.4 0.5 0.6 0.7 0.3 0.4 0.5 0.6 0.7
Activator to binder ratio Activator to binder ratio

Fig. 9. Porosities (measured by vacuum saturation method) of CFB concretes made using: (a) fly ash, and (b) GGBFS as the starting material.

50 50
Fly ash CFB concrete
GGBFS CFB concrete CFB FA paste
Compressive strength,fc (MPa)

CFB GGBFS paste

40 fc= 131.0 e -0.14 φ
R = 0.625 40

30 30
d50 (μm)

20 20

10 10

0 0
6 8 10 12 14 16 18 0.4 0.6
Porosity,Φ (%) Activator to Binder ratio

Fig. 10. Compressive strength–porosity relationship for CFB concretes. Fig. 12. Characteristic pore sizes (d50(>10 lm)) of CFB pastes.

(a) 50 (b) 50
CFB FA paste CFB FA paste
CFB GGBFS paste CFB GGBFS paste
CFB FA concrete
40 CFB GGBFS concrete 40
Porosity (%)

Porosity (%)

30 30

20 20

10 10

0 0
0.4 0.6 0.4 0.6
Activator to Binder ratio Activator to Binder ratio

Fig. 11. (a) Porosity of CFB pastes and concretes measured by vacuum saturation method, and (b) porosities of CFB pastes obtained from image analysis method.
 D. Ravikumar et al. / Cement & Concrete Composites 32 (2010) 399–410 407

However this facilitates a comparison between the pore sizes in XRD was used for phase identification. Fig. 13a shows the micro-
CFB pastes made using fly ash or GGBFS, and the influence of acti- graph of CFB paste made using fly ash as the starting material
vator-to-binder ratio. The characteristic pore size of CFB pastes and activated using 4 M NaOH, while Fig. 13b shows the micro-
with fly ash is seen to be higher than those with GGBFS as the bin- structure of the resultant paste when activated using 8 M NaOH.
der. The higher porosity and larger pore sizes in CFB pastes made The activator-to-binder ratio in both these cases is 0.40. It can be
using fly ash as compared to those made using GGBFS also provide seen that the paste obtained by activation using a lower concentra-
a reason for the difference in compressive strengths of activated tion of the activator (4 M NaOH) shows more unreacted fly ash par-
concretes when these starting materials are used. ticles as compared to the paste obtained from activation using a
higher concentration of the activator (8 M NaOH). This is evidence
3.4. Microstructure of CFB pastes of increased gel formation when higher activator concentrations
are used, which can also be related to the compressive strength re-
3.4.1. CFB pastes made from fly ash sults discussed earlier. It is seen that the reaction product forms a
Scanning electron microscopy (SEM) was used to understand shell around the unreacted fly ash particles, which is particularly
the morphology of the reaction products in CFB pastes, while well observed in systems activated by a higher alkali concentration

(a) (b)

1 2

(c)

(d) (e)
Fig. 13. (a) and (b) Micrographs of fly ash CFB pastes activated using 4 M and 8 M NaOH, respectively, (c) EDX at two locations shown in (b), (d) micrograph at higher
magnification for paste activated using 8 M NaOH, and (e) XRD of paste activated using 8 M NaOH.
408 D. Ravikumar et al. / Cement & Concrete Composites 32 (2010) 399–410

such as the one shown in Fig. 13b. NaOH dissolves the soluble can be seen in Fig. 13b). Though a large concentration of OH is
phases in fly ash thereby releasing alumina and silica from the sur- available in the system at higher NaOH concentrations, the rate of
face of fly ash particles. These phases then react with the alkalis further crystal growth would be slower if the fly ash particles are
from activating agent and condense on the surface of fly ash form- completely encapsulated by the aluminosilicate gel. Further activa-
ing the aluminosilicate gel shell [32,33] shown by the arrows in tion is then controlled by diffusion as well as the pH gradients along
Fig. 13b. the thickness of the reaction product around the unreacted fly ash
A significant amount of unreacted fly ash particles (the core on core. Thus it is anticipated that even at further higher concentra-
which the reaction products are precipitated) indicates that the de- tions of the activating agent, the fly ash particles might not be com-
gree of reaction is only low to moderate. Even when a higher con- pletely utilized for the production of the reaction product, and the
centration of activating alkali (such as 8 M NaOH) is used, there is degree of reaction of the system might remain moderate.
likely a formation of a relatively thin layer of the reaction products The energy dispersive X-ray spectra (EDX) corresponding to the
on the surface of fly ash particle inhibiting its further activation (as points marked ‘‘1” and ‘‘2” in Fig. 13b are shown in Fig. 13c. The

(a) (b) (c)

1 2

(d)

(e)
Fig. 14. (a) and (b) Micrographs of GGBFS CFB pastes activated using 4 M and 8 M NaOH, respectively, (c) micrograph at higher magnification for paste activated using 8 M
NaOH, (d) EDX at two locations shown in (c), and (e) XRD of paste activated using 8 M NaOH.
 D. Ravikumar et al. / Cement & Concrete Composites 32 (2010) 399–410 409

EDX for point ‘‘1” is that of the unreacted core in a fly ash particle, is stated as the main hydration product in alkali-activated slag ce-
whereas the EDX for point ‘‘2” shows the elemental composition of ments [34]. The higher Ca to Si ratio (1.4) at this point lends cre-
the reaction product. The Na to Al ratio of the Point ‘‘1” was deter- dence to this premise. The Na to Al ratio for the reaction product is
mined to be 0.15, and that of Point ‘‘2” 0.83, indicating the higher rather high (1.37) as compared to that in the reaction product of
Na content in the reaction product. The Si to Al ratio for the reac- activated fly ash paste (0.80), which could be attributable to the
tion product (Point ‘‘2”), determined from quantitative EDX was lower amounts of alumina in GGBFS. The EDX spectra for CFB
found to be approximately 2.0, which is in the range of reported pastes containing GGBFS are found to be significantly different
values (1.8–2.0) for the aluminosilicate gel produced by the activa- both in the peak intensities and in composition (especially the
tion of fly ash using NaOH [22]. Fig. 13d shows a magnified view of presence of calcium) from those containing fly ash, showing that
a partly reacted fly ash plerosphere in a paste activated with 8 M the reaction products formed from both these starting materials
NaOH that has retained its original shape. The activation process are different in structure and composition.
is seen to have resulted in the formation of the aluminosilicate The XRD pattern for the CFB paste with GGBFS as the binder is
gel outside the larger particle as well dissociation of the smaller shown in Fig. 14e. The paste is more amorphous than the reaction
spheres inside, forming more reaction products. A few randomly product of fly ash activation. At a 2h angle of 29°, a peak that could
oriented and platey crystals, plausibly of hydroxy sodalite, are be attributed to either calcite or C–S–H is observed. Occurrence of
found in this micrograph. calcite can be attributed to the fact that carbonation is likely to oc-
Fig. 13e shows a typical XRD pattern of CFB pastes made fly ash cur during the mixing of very high amounts of alkalis to a binder
as the starting material and 8 M NaOH as the activator. The XRD that contains significant amounts of calcium oxide [35]. The pres-
spectrum shows that the characteristic peaks of quartz and mullite ence of calcite at very early ages could be the cause of premature
remain even after alkaline activation with 8 M NaOH and heat stiffening of CFB paste made with GGBFS which was observed dur-
treatment for 48 h. This shows the relatively low reactivity of these ing the time of mixing. No sodalite peaks are observed in the CFB
phases in alkaline conditions. Only the glassy phases in fly ash are paste with GGBFS, as was observed in the case of CFB paste with
the source of Al and Si to form the aluminosilicate gel. However a fly ash.
previous study [33] has shown that the quartz and mullite phases
dissolve completely when very high activator solution-to-fly ash
ratios (3 to 6) are used. Such high ratios, commonly used for zeo- 4. Conclusions
lite synthesis from precursor materials, cannot be typically used
for concretes because of the adverse impact of higher amount of This paper has provided extensive experimental results and
water in the activating solution on the compressive strengths analysis on the compressive strengths, porosities, and microstruc-
and porosities. Lower alkaline solution-to-fly ash ratios such as ture of cement-free binder (CFB) pastes and/or concretes contain-
the ones used here reduce the rate of crystal growth, and conse- ing either fly ash or GGBFS as the sole binder and activated using
quently reduce the formation of crystalline phases. In the case of NaOH solutions of different concentrations. The following conclu-
low alkaline solution-to-fly ash ratios, a reaction product is quickly sions are made based on this study:
formed that engulfs the fly ash particle as shown in the micro-
graphs discussed above, slowing down further activation of the 1. The curing temperature was found to have a profound influence
fly ash particles, and thus resulting in only low to moderate de- on the compressive strengths of CFB concretes containing fly
grees of reaction. The only crystalline aluminosilicate phase that ash as compared to those containing GGBFS. A curing tempera-
is detectable in the XRD spectrum is that of hydroxy sodalite (Na6(- ture of 75 °C and a curing duration of 48 h resulted in maximum
Si6Al6O24)8H2O), which is a zeolite. Other crystalline aluminosili- compressive strengths of the CFB concretes. CFB concretes with
cate phases such as Na-phillipsite have been reported [33] at fly ash showed lower compressive strengths when heat cured at
high activating solution-to-fly ash ratios, longer activating dura- 60 °C irrespective of curing duration whereas CFB concretes
tions, and higher activating temperatures than those used in this with GGBFS showed reasonable compressive strengths even at
study. lower curing temperatures.
2. Compressive strengths of CFB concretes were found to increase
3.4.2. CFB pastes made from GGBFS with increase in activator concentration and decrease in activa-
Fig. 14a shows the micrograph of CFB paste made using GGBFS tor-to-binder ratio. The CFB concretes made using GGBFS as the
as the starting material and activated using 4 M NaOH, and Fig. 14b starting material showed higher compressive strengths com-
corresponds to CFB paste activated using 8 M NaOH. The activator- pared to those made with fly ash because of the self-cementing
to-binder ratio was 0.40 for both these pastes. Individual GGBFS property of GGBFS and higher binder content used. When the
particles cannot be isolated in these micrographs. The reaction compressive strengths were normalized with fly ash or GGBFS
product formed is much more compact, homogeneous, and uni- content, it was observed that the CFB concretes made using
form when compared to the pastes with fly ash as the starting GGBFS and fly ash had similar compressive strength when acti-
material. A lower concentration (4 M NaOH) of the activator is vated using higher concentration of the activator. At lower con-
found to create a microstructure that is more porous than that in centrations of the activator, CFB concretes with GGBFS showed
a paste made using a higher (8 M NaOH) activator concentration higher compressive strengths. A 32 factorial analysis and the
(Fig. 14b). This shows that the use of a more concentrated alkali in- quadratic equations for the response surface for compressive
creases the amounts of reaction product formation, and conse- strengths showed that the activator concentration influences
quently reduces the porosities, and increases the compressive the compressive strengths of CFB concretes made using fly
strengths. ash to a greater degree, whereas the activator-to-binder ratio
Fig. 14c is a micrograph at higher magnification of the CFB paste exerts more influence on the compressive strengths of CFB con-
made using GGBFS activated with 8 M NaOH. The EDX spectra at cretes made using GGBFS.
two points marked ‘‘1” and ‘‘2” in Fig. 14c are shown in Fig. 14d. 3. Porosities of CFB concretes were found to decrease with
The EDX spectrum of the Point ‘‘1” shows that it is likely a GGBFS increase in activating agent concentration and activator-to-bin-
particle with some reaction products on the surface. In the matrix der ratio. The strength–porosity relationships for CFB concretes
of reaction products, a few areas showing network of fibrous or fla- were also found to be exponential, similar to those for normal
ky structures are apparent (Point ‘‘2”), plausibly C–S–H gel, which concretes. Statistical analysis showed that a linear relationship
410 D. Ravikumar et al. / Cement & Concrete Composites 32 (2010) 399–410

adequately describes the influence of activator concentration [9] Alvarez-Ayuso E, Querol X, Plana F, Alastuey A, Moreno N, Izquierdo M, et al.
Environmental, physical and structural characterization of geopolymer
and activator-to-binder ratio on porosities of CFB concretes.
matrixes synthesized from coal (co-)combustion fly ashes. J Hazard Mater
The porosities of CFB pastes were found to be about three times 2008;154:175–83.
higher than those of CFB concretes, mainly because the paste [10] Duxson P, Provis JL, Luckey GC, Mallicoat SW, Kriven WM, van Deventer JSJ.
volume is only about a third of the concrete volume. The char- Understanding the relationship between geopolymer composition,
microstructure and mechanical properties. J Coll surf A: Physiochem Eng
acteristic size of pores larger than 10 lm (d50(>10 lm)) in CFB Aspects 2005;269:47–58.
pastes was found to be higher for the paste made using fly [11] Bakharev T. Resistance of geopolymer materials to acid attack. Cem Concr Res
ash as the starting material. 2005;35:658–70.
[12] Bakharev T. Durability of geopolymer materials in sodium and magnesium
4. Microstructural observations revealed the formation of alkali- sulfate solutions. Cem Concr Res 2005;35:1233–46.
aluminosilicate gel as a shell around the fly ash particles [13] Kong DLY, Sanjayan JG. Damage behavior of geopolymer composites exposed
whereas the reaction product was more homogeneously dis- to elevated temperatures. Cem Concr Compos 2008;30:986–91.
[14] Criado M, Fernández-Jiménez A, de la Torre A, Aranda MAG, Palomo A. An XRD
tributed in pastes made using GGBFS. The reaction product shell study of the effect of SiO2/Na2O ratio on the alkali activation of fly ash. Cem
formed around the fly ash inhibits further activation, resulting Concr Res 2007;37:671–9.
in incomplete utilization of the fly ash particles. The presence [15] van Jaarsveld JGS, van Deventer JSJ, Lukey G. The characterization of source
materials in fly ash-based geopolymers. Mater Lett 2003;57:1272–80.
of the crystalline phases of fly ash in the XRD pattern of the [16] Silva PD, Crenstil KS, Sirivivatnanon V. Kinetics of geopolymerization: role of
paste shows that only glassy phases in fly ash take part in gel Al2O3 and SiO2. Cem Concr Res 2007;37:512–8.
formation. Hydroxy sodalite was the only crystalline alumino- [17] Temuujin J, van Riessen A, Williams R. Influence of calcium compounds on the
mechanical properties of fly ash geopolymer pastes. J Hazard Mater 2009.
silicate phase detected in CFB pastes with fly ash. The reaction
doi:10.1016/j.jhazmat.2008.12.12.
product in the GGBFS paste showed the presence of significant [18] Hardjito D, Rangan BV. Development and properties of low-calcium fly ash-
amounts of calcium, in addition to the sodium, alumina and sil- based geopolymer concrete. Curtin research report. <http://www.geopolymer.
ica phases. The XRD pattern for the CFB paste made using org/library/technical-papers/17-development-and-properties-of-low-calcium-
fly-ash-based-geopolymer-concrete>.
GGBFS was mostly amorphous. The micrographs and the XRD [19] Yang KH, Song JK, Ashour AF, Lee ET. Properties of cementless mortars
patterns provide an insight into the differences in the reaction activated by sodium silicate. Constr Build Mater 2009;22:1981–9.
product morphology and composition when fly ash or GGBFS [20] Sathonsaowaphak A, Chindaprasirt P, Pimraksa K. Workability and strength of
lignite bottom ash geopolymer mortar. J Hazard Mater 2009. doi:10.1016/
are used as the starting material in CFB concretes or pastes. j.jhazmat.2009.01.12.
[21] Shi C. Strength, pore structure and permeability of alkali-activated slag
mortars. Cem Concr Res 1996;26:1789–99.
[22] Fernández-Jiménez A, Palomo A, Criado M. Microstructure development of
Acknowledgements alkali-activated fly ash cement: a descriptive model. Cem Concr Res
2005;35:1204–9.
This study was partly funded by the Green and Energy Efficient [23] Xu H, van Deventer JSJ. The geopolymerization of alumino-silicate minerals.
Int J Miner Process 2000;59:247–66.
Building Products and Systems program of the New York State En- [24] Davidovits J. Geopolymer chemistry and sustainable development. The
ergy Research and Development Authority (NYSERDA). The materi- Poly(sialate) terminology: a very useful and simple model for the promotion
als for this study were provided by Holcim Cement and Graymont and understanding of green-chemistry. In: Davidovits J, editor. Proceedings of
the world congress geopolymer, Saint Quentin, France; 2005. p. 69–73.
Materials. A part of the experimental study was carried out at the
[25] Mozgawa W, Deja J. Spectroscopic studies of alkaline activated slag
facilities of the Center for Advanced Materials Processing (CAMP) Geopolymers. J Mol Struct 2009. doi:10.1016/j.molstruct.2008.12.026, 200.
at Clarkson University. The authors acknowledge all these entities [26] Bakharev T. Geopolymeric materials prepared using Class F fly ash and
for the support. elevated temperature curing. Cem Concr Res 2005;35:1224–32.
[27] Chindaprasirt P, Chareerat T, Sirivivatnanon V. Workability and strength of
coarse high calcium fly ash geopolymer. Cem Concr Compos 2007;29:
References 224–9.
[28] Provis JL, van Deventer JSJ. Geopolymerisation kinetics. 2. Reaction kinetic
[1] Palomo A, de la Fuente JIL. Alkali-activated cementitious materials: alternative modeling. Chem Eng Sci 2007;62:2318–29.
matrices for the immobilization of hazardous wastes. Part I. Stabilization of [29] Panias D, Giannopouloua IP, Perraki T. Effect of synthesis parameters on the
boron. Cem Concr Res 2003;33:281–8. mechanical properties of fly ash-based geopolymer. J Coll Surf A: Physicochem
[2] Davidovits J. Geopolymers: inorganic polymeric new materials. J Therm Anal Eng Aspects 2007;301:246–54.
1991;37:1633–56. [30] RILEM CPC11.3. Absorption of water by immersion under vacuum. Mater
[3] van Deventer JSJ, Provis JL, Duxson P, Lukey GC. Reaction mechanism in Struct 1984;17:391–4.
geopolymeric conversion of inorganic waste to useful products. J Hazard Mater [31] Kovalchuk G, Jimenez AF, Palomo A. Alkali-activated fly ash: effect of thermal
2007;139:506–13. curing conditions on mechanical and microstructural development – Part II.
[4] Yunsheng Z, Wei S, Qianli C, Lin C. Synthesis and heavy metal immobilization Fuel 2007;86:315–22.
behavior of slag geopolymer. J Hazard Mater 2007;143:206–13. [32] Fernández-Jiménez A, Criado M. Microstructure development of alkali-
[5] Sun W, Zhang YS, Lin W, Liu ZY. In situ monitoring of the hydration process of activated fly ash cement: a descriptive model. Cem Concr Res 2004;35:
K-PS geopolymer cement with ESEM. Cem Concr Res 2004;34:935–40. 1204–9.
[6] Wang H, Li H, Yan F. Synthesis and mechanical properties of metakaolinite- [33] Rios CAR, Williams CD, Roberts CL. A comparative study of two methods for
based geopolymer. Coll Surf A: Physicochem Eng Aspects 2005;268:1–6. the synthesis of fly ash-based sodium and potassium types zeolite. Fuel
[7] Sofi M, van Deventer JSJ, Mendis PA, Lukey GC. Engineering properties of 2009;88:1403–16.
inorganic polymer concretes (IPCs). Cem Concr Res 2007;37:251–7. [34] Taylor HFW. Cement chemistry. 2nd ed. London: Thomas Telford; 1997.
[8] Rahier H, Van mele S, Wastiels J. Low-temperature synthesized aluminosilicate [35] Astutiningsih S, Liu Y. Geopolymerisation of Australian slag with effective
glasses. Part II. Rheological transformations during low-temperature cure and dissolution by the alkali. In: Davidovits J, editor. Proceedings of the world
high-temperature properties of a model compound. J Mater Sci 1996;31:80–5. congress geopolymer, Saint Quentin, France; 2005, p. 69–73.