Cement and Concrete Composites 86 (2018) 9e18

Contents lists available at ScienceDirect

Cement and Concrete Composites

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

Compressive strength and microstructure of alkali-activated fly ash/

slag binders at high temperature
Z. Pan a, *, Z. Tao a, **, Y.F. Cao a, R. Wuhrer b, T. Murphy b
a
Centre for Infrastructure Engineering, Western Sydney University, Penrith, NSW 2751, Australia
b
Advanced Materials Characterisation Facility, Western Sydney University, Parramatta, NSW 2116, Australia

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

Article history: This paper reports the results of the compressive strength and microstructure of various alkali-activated
Received 24 July 2016 binders at elevated temperatures of 300 and 600
C. The binders were prepared by alkali-activated low
Received in revised form calcium fly ash/ground granulated blast-furnace slag at ratios of 100/0, 50/50, 10/90 and 0/100 wt.%.
27 August 2017
Specimens free of loading were heated to a pre-fixed temperature by keeping the furnace temperature
Accepted 25 September 2017
Available online 31 October 2017
constant until the specimens reached a steady state. Then the specimen was loaded to failure while hot.
XRD, SEM and FTIR techniques were used to investigate the microstructural changes after the thermal
exposure. The fly ash-based specimen shows an increase in strength at 600
C. On the other hand, the
Keywords:
Elevated temperatures
slag-based specimen gives the worst high-temperature performance particularly at a temperature of
Hot strength 300
C as compared to ordinary Portland cement binder. This contrasting behaviour of binders is due to
Geopolymers their different binder formulation which gives rise to various phase transformations at elevated tem-
Microstructure peratures. The effects of these transformations on the compressive strength are discussed on the basis of
Conversion experimental results.
© 2017 Elsevier Ltd. All rights reserved.

1. Introduction The excellent binding property of geopolymers is believed to be

the result of the formation of a three-dimensional “N-A-S-H” type
Geopolymers are inorganic polymeric materials that are pro- gel which is a network [Q4(Al)] consisting of alternating SiO4 and
duced by alkali activation of aluminosilicate raw materials (such as AlO4 tetrahedra linked by shared O atoms [8]. This three-
metakaolin and fly ash), which are transformed into reaction dimensional structure is incredibly resilient to exposure to high
products by partial dissolution and polycondensation in aqueous temperature. No change in structure was observed up to 600
C, as
alkaline environments. Due to their excellent mechanical proper- characterised by X-ray diffraction (XRD) and Nuclear magnetic
ties and low energy consumption of production from industry by- resonance (NMR) results [9]. In contrast, investigation of OPC
products, geopolymers are attracting increasing interest as an composites identified the chemical deterioration taking place at
environmentally-friendly alternative to traditional Ordinary Port- high temperatures because both chemically bound water and
land Cement (OPC) for use in construction applications [1]. In many interlayer water are lost due to the decomposition of calcium hy-
cases, previous investigations have shown that geopolymers droxide (CH) and calcium silicate hydrates (CSH) [10]. Due to the
outperform their OPC counterparts with respect to tensile strength difference in thermal evolution of chemical structure, the geo-
[2,3] as well as acid resistance [4] and impact resistance [5]. Geo- polymer is believed to have better fire resistance as compared to its
polymers also exhibit strong adhesive behaviour, which makes OPC counterpart [11e14].
them suitable to combine not only normal aggregates but also dune When geopolymers were exposed to elevated temperatures, it
sand [6] and residue sand [7] (red mud-residues from the alumina was found that the action of high temperatures (up to 680
C)
industry). increased the hot strength (tested in a hot state) of geopolymers
[15]. A similar phenomenon was also observed by Fernandez-
Jimenez and Palomo [16] who investigated bending strength of
* Corresponding author. geopolymers at elevated temperatures. At 600
C, the geopolymers
** Corresponding author. showed excellent fire resistance with a hot bending strength of
E-mail addresses: z.pan@westernsydney.edu.au (Z. Pan), z.tao@westernsydney. 120% of the original strength at ambient temperature. More
edu.au (Z. Tao).

https://doi.org/10.1016/j.cemconcomp.2017.09.011
0958-9465/© 2017 Elsevier Ltd. All rights reserved.
10 Z. Pan et al. / Cement and Concrete Composites 86 (2018) 9e18

recently, an investigation of geopolymer concrete further However, the strength loss due to chemical changes (taking place at
confirmed the tendency of the strength evolution at elevated elevated temperatures) has not been investigated in that investi-
temperatures. Shaikh and Vimonsatit [17] studied the hot strength gation. More recently, Kurklu [26] has investigated the changes in
of fly ash based geopolymer concrete activated with 10 M NaOH microstructure and compressive strength of fly ash/slag system
solution and cured at the temperature of 60
C. At 600
C, the hot after thermal exposure. XRD analysis was conducted to monitor the
strength of geopolymer concrete was 110% of the original strength phase transformation at elevated temperatures. However, the
at ambient temperature. In contrast, OPC concrete exhibited about samples investigated were heat cured, which is regarded as a bar-
46.5% loss in compressive strength at 600
C, as a result of the rier for cast in situ applications.
dissociation of CH. Vickers et al. [18] investigated the hot strength It is noted that the previous investigations of AAS and fly ash/
of fly ash based geopolymers filled with alumina aggregate. The slag systems mainly focus on the residual strength [19e22],
addition of alumina aggregate reduced the thermal conductivity. As whereas the hot strength of these systems has received less
a result, crack formation was reduced due to smaller thermal attention although the hot strength of fly ash based geopolymer has
gradient across the fire test specimen. At elevated temperatures, an been investigated to some extent [11e13]. Therefore, the aim of this
increase in compressive strength was attributed to sintering study is to investigate the hot strength of AAS and slag/fly ash
forming a more compact microstructure. systems cured at ambient temperature. The microstructure alter-
It is noted that the fly ash-based geopolymer samples used in ation at elevated temperatures was investigated by means of XRD,
above studies [11e13] were manufactured by using heat curing Fourier-Transform Infrared Spectroscopy (FTIR) and Scanning
regime. Although this regime can easily be achieved in a typical Electron Microscope (SEM). For a purpose of comparison, the OPC
precast concrete factory, it is difficult to be implemented for cast in mixture and fly ash-based geopolymer mixture were also prepared
situ applications. In order to offset the limitation of heat curing, and tested under the same conditions. The results reported in this
there were a lot of attempts to develop ambient temperature cured investigation will assist engineers to understand the high-
geopolymers by combining fly ash with various additives such as temperature performance of various alkali-activated cement
Portland cement, blast-furnace slag, rice husk ash, silica fume and systems.
lime [19]. Among these additives, ground granulated blast-furnace
slag (GGBFS) is a commonly used additive. Incorporation of GGBFS
2. Experimental program
in fly ash to produce geopolymers has been tested with some
favourable results. Nath and Sarker [2] studied ambient tempera-
2.1. Characterisation of materials
ture cured fly ash/slag paste and found that the compressive
strength reached more than 50 MPa at 28 days. In comparison with
In this investigation, the cement used meets the requirement of
OPC concrete, slag blended fly ash geopolymers also showed better
ASTM C150 Type I cement. The class F fly ash used in this investi-
durability with good resistance to permeation [20] and sodium
gation is mainly glassy with some crystalline inclusions of mullite
sulfate attack [4].
and quartz. The granulated blast furnace slag is supplied with
Although fly ash and GGBFS are popular choices as raw materials
gypsum, which is pre-blended with slag. The oxide compositions of
for producing alkali-activated materials, they have marked struc-
the binder materials are summarised in Table 1. It is noted that fly
tural, chemical and physical differences. It is likely that the alkali
ash has very low calcium content. Therefore, it is expected that the
activation of slag (AAS) results in the dissolution of calcium and
fly ash-based geopolymer mainly consists of calcium-free gel
participation of Al to form a C-A-S-H type gel. Therefore, partial slag
structures.
substitution in alkali-activated fly ash will produce a binder system
The fine aggregate is natural river sand obtained from the bed of
consisting of both C-A-S-H and N-A-S-H gels [21]. With increasing
the Nepean River which is located in the south-west and west of
amount of slag, the dissolved calcium concentration will increase
Sydney, New South Wales, Australia. The sand has a fineness
and lead to the formation of CSH products [22]. When slag was used
modulus of 2.6. The alkaline activating agents used in this inves-
as the sole raw material in producing the binder, CSH gel is found to
tigation include sodium hydroxide, sodium silicate and sodium
dominate the reaction products. The CSH gel in AAS is formed in
meta-silicate. Sodium hydroxide flakes of 98% purity were supplied
pore solution where the silicate concentration has significant in-
by Orica Chemicals which is an Australian-based supplier of mining
fluence on the structure of CSH gel. Due to the high silicate con-
chemicals. Sodium silicate liquid and industrial grade-powdered
centration in the AAS pore solution, the CSH gel has lower C/S ratio
sodium meta-silicate were supplied by PQ Australia. Sodium sili-
than that of the OPC system [8].
cate liquid (grade D) has the chemical composition: 29.4% SiO2,
The different nature of CSH between OPC and AAS is at least
14.7% Na2O and 55.9% H2O, with a molecular modulus of Ms ¼ 2.06.
partially responsible for the different high-temperature perfor-
Sodium metasilicate has the chemical composition: 29% Na2O, 28%
mance of these two systems. Guerrieri et al. [23] investigated the
SiO2 and 43% H2O, with the Ms ¼ 1.0.
residual compressive strength of AAS and OPC paste after exposure
to elevated temperatures. Test results illustrate that the residual
strength of AAS was similar to that of the OPC mixture at 600
C. In 2.2. Mix proportion
contrast to this, a latest study [24] reported that AAS (after expo-
sure to 600
C) experienced more strength loss as compared to the The samples were prepared in fly ash/slag proportions of 100%/
OPC mixture. However, the results from the two investigations are 0%, 90%/10%, 50%/50% and 0%/100% (wt-%). The labels representing
hard to directly compare because the test conditions, shape and the above samples are F100, F90S10, F50S50 and S100, respectively.
size of the specimens were different. Except for studies on AAS
binders, tests have also been conducted on the high-temperature
Table 1
performance of fly ash/slag system by measuring the residual Chemical composition of materials.
compressive strength of pastes after thermal exposure [25]. It was
Materials Al2O3 SiO2 CaO Fe2O3 K2O MgO Na2O SO3 LOI
found that the high strength loss was associated with the low
ductility of samples. This is attributed to the fact that the sample Cement (%) 4.9 18.2 60.7 2.6 0.4 1.0 0.2 2.2 3.0
with low ductility has less capacity to accommodate the non- Fly ash (%) 30.6 48.4 2.7 12.1 0.3 1.3 0.2 e 1.7
Slag (%) 13.0 32.5 42.1 0.2 0.3 5.5 0.2 e 0.4
uniform thermal deformation, leading to a larger strength loss.
 Z. Pan et al. / Cement and Concrete Composites 86 (2018) 9e18 11

The liquid activator (a combination of sodium hydroxide solution cylinders) were submitted to a grinder for having smooth and
and sodium silicate solution) was used to prepare F100 and F90S10. parallel top and bottom faces. Compressive strength tests were
The concentration of sodium hydroxide solution was 8 M and the carried out using a universal testing machine at the age of 28 days.
mass ratio of sodium hydroxide solution to sodium silicate solution To measure compressive strength at elevated temperatures, the
was 2.5. The ratio of liquid activator to the binder (fly ash þ slag) specimens without pre-loading were submitted to very slow
was 0.45 for F100 and F90S10. heating in order to reproduce quasi-steady thermal conditions. In
It is noted that the dissolution and reaction of slag in strong the current study, compressive strength tests were conducted at
alkaline media are much faster, as compared to fly ash. As a result, 300 and 600
C. The tests were carried out in an electric split-tube
the liquid activator led to the quick setting of the fresh samples furnace equipped with a MTS actuator. The unstressed samples
with the high content of slag (F50S50 and S100). To avoid the flash were heated to the target temperature at 4
C/min. Our preliminary
set, the powdered activator (sodium metasilicate) was used to tests have proven that the temperature difference across the sec-
prepare F50S50 and S100. The activator dosage used was 4 wt.% of tion was less than 20
C when using this heating rate. This meets
the total mass of the binder and the water to binder ratio was 0.5. the requirement for minimising the risk of thermal self-stresses in
The hydrated lime, with a 1 wt.% of the binder, was added to both RILEM [27]. The samples were further remained at the target
the mixtures of F50S50 and S100. The OPC samples were also cast temperature for 1.5 h. This period was found to be sufficient for
for a purpose of comparison. The water to cement ratio was 0.5. reaching quasi-steady conditions in our preliminary tests. Then, the
When submitted to stress, a ductile material can deform to samples were loaded up to failure. The procedures proposed by
evenly distribute stresses throughout a region; but in a brittle RILEM [27] were followed for measuring strength at elevated
material, stress concentrations at defects promote crack propaga- temperatures. All samples were tested after 28 days.
tion, leading to a sudden failure. The brittleness of paste is several
times higher than that of mortars. As a result, 3e5 samples must be 2.5. XRD
tested for paste to obtain strength value with a standard deviation
less than 10%. For mortars, only 2e3 samples are required to obtain XRD patterns were obtained using a Bruker D8 Advance Powder
strength value with a similar standard deviation. It should be noted Diffractometer. The samples were ground just before running the
that the determination of compressive strength (in a hot state) is a tests. The grounded samples were then passed through a 200-mesh
time-consuming process. To reduce the number of specimens to be sieve. Diffraction analyses were made from 5 to 60 2q using copper
tested, only mortar samples were used to measure the compressive K radiation. The excitation voltage was 40 kV at 40 mA. Counting
strength. For all mortar specimens, the sand to binder ratio was time was 5 s.
fixed to 0.5. To have a fundamental understanding of the properties,
paste samples were prepared for XRD, SEM and FTIR measure- 2.6. SEM and EDS analyses
ments. Since these measurements were carried out in an un-
stressed state, it is expected that the brittleness of the paste has The morphology and microstructure of the pastes were
little influence on the variation of results. The composition of the observed with a JEOL 6510LV scanning electron microscope (SEM).
paste samples was the same as that of the mortar samples except The elemental composition of samples was observed using a Moran
for the lack of sand. Scientific microanalysis EDS (energy dispersive spectroscopy) sys-
tem. The samples used for SEM imaging were slices taken from the
2.3. Specimen preparation paste specimens. For heat treatment, both paste and mortar spec-
imens were submitted to the same heating regimes. The fracture
The mixing procedures used were summarised as follows. The surface of the samples was examined in secondary electron (SE)
binders (fly ash, slag or cement) and the liquid components (water mode and operated at around 15 kV. The samples were coated with
or alkali liquid) were mixed in a heavy duty mixer (made by Hobart carbon to make a conductive surface.
Corporation, an American manufacture) for 5 min. When making
mortar samples, the binders and sand were dry mixed for 2 min. 2.7. FTIR
Then the alkali liquid or water was added and wet mixing was
carried out for 4 min. For powder activated samples, the dry silicate Infrared spectra were recorded with a Bruker Vertex 70 Fourier
activator was pre-blended with slag and dry-mixed for 3 min to transform infrared spectrometer. The machine was operated in 32
ensure homogeneity of the mixture. The hydrated lime was mixed scanning times from 2000 to 400 cm1 at 4 cm1 resolution. The
with water in a ratio of 1:3. The lime slurry together with water was 200 mg of KBr pellet was prepared.
then added and wet mixing was carried out for 4 min. The mixture
was poured into the mould in three equal layers. Each layer was 3. Results
vibrated for 20e40 s on a vibration table.
The samples were sealed with cling wrap and cured for 1 day in 3.1. Compressive strength
a laboratory environment. Then the samples were removed from
the moulds and cured at an oven which was operated at a tem- The compressive strength of various binders is presented in
perature of 23
C and relative humidity of 100%. After 28 days, the Fig. 1 (a). At the age of 28 days, the slag-based mixture S100 acti-
samples were kept in a conditioning room until testing. This curing vated with powder-type activator developed strength of around
regime was conducted for all samples except F100. Regarding F100, 43 MPa which is similar to the strength of OPC control mixture at
the samples were kept in the moulds and covered with cling wrap the same age. When 50% of slag was replaced by fly ash, the
and then placed immediately in a preheated oven at 60
C. After 1 compressive strength of F50S50 decreased from 43 to 29 MPa. This
day, the samples were demolded and kept in the conditioning room is believed to be the result of low dissolution and reaction of fly ash
until testing. at ambient temperature and low concentration of the activator (4%
Na by mass of the binder). The strength of F50S50 was mainly
2.4. Compressive strength testing provided by the binding phase of C-S-H gel that decreased as the
reduced calcium supply at lower slag content, as compared to the
Prior to compressive strength tests, the samples (60  180 mm slag-based mixture. With the increased concentration to 8% Na by
12 Z. Pan et al. / Cement and Concrete Composites 86 (2018) 9e18

Fig. 1. 28-day compressive strength (a) Original strength of various binders (F100: 100% fly ash; F90S10: 90 wt.% fly ash/10 wt.% slag; F50S50: 50 wt.% fly ash/50 wt.% slag; S100:
100% slag; OPC: cement paste); (b) Relative change in strength at elevated temperatures.

mass of the binder, the compressive strength of the mixture having

10% slag/90% fly ash was improved by 29%, although only 10% slag
was included in this mixture. The higher concentration resulted in
higher pH value of the activator where the dissolution of fly ash
particles was rapid. The addition of slag provided additional cal-
cium bearing compound, promoting the formation of a low calcium
N-(C)-A-S-H gel. This gel served as the main strength provider for
F90S10. Without the inclusion of slag, the fly ash-based geo-
polymers did not develop any meaningful strength (Results are not
presented in the paper) when cured at ambient temperature. As is
known, the dissolution and reaction of fly ash in alkaline media is
very favoured at temperatures higher than 60
C. When the curing
was carried out at 60
C, the fly ash-based geopolymer F100 ach-
ieved a compressive strength of about 45 MPa.
When the samples were exposed to high temperatures, the hot
strength was determined after heating the specimens in quasi-
steady conditions, which is a well-known approach [28] Fig. 2. XRD results (Aragonite-A, Calcite-C, Mullite-M, Quartz-Q, Maghemite-Mg,
commonly used to minimise the risk of dangerous thermal self- Vaterite-V, Gypsum-G).

stresses. In comparison with its original strength tested at

ambient temperature, the relative changes in hot strength of each
The element gypsum has been introduced in the milling process
sample are plotted in Fig. 1 (b). Among all the samples, the largest
which occurs in a factory for producing Portland cement.
strength loss was observed in the slag-based samples. The reduc-
In the activated mixtures, various crystalline phases are
tion was 65% at temperatures between 20 and 300
C. From 300 to
observed, as shown in Fig. 3. The fly ash/slag ratio plays significant
600
C, the compressive strength of S100 decreased further to 35%
roles in the appearance of these phases. In the alkali activated slag,
of the strength of the corresponding mixture at ambient temper-
a low intensity peak centred at a 2q angle around 29.5
is identified,
ature. When exposed to 600
C, both the OPC and S50F50 samples
indicating the poorly crystalline CSH gel. The peak at 22.7
2q
showed decrease in strength, but the OPC samples demonstrated a
matches reasonably well with the C4AH13 which has also been
larger decrease. The strength losses were 48% and 33% in OPC and
identified as a reaction product of AAS in the earlier study [29]. The
S50F50, respectively. The samples containing 10% or no slag
presence of CH was characterised by two diffraction peaks at 34.3

retained the hot strength at temperatures below 300
C. At tem-
2q and 47.2
2q. The small reflections for aragonite and calcite were
peratures between 300 and 600
C, the fly ash-based sample
also identified. With the decreased content of slag, the main peak
demonstrated the best performance with a strength increase. At
remaining near 29.5
2q, representing CSH type gel, together with
600
C, the hot strength of F100 reached around 58 MPa, which was
traces of calcite. However, the intensity of CSH peak is much higher
about 1.6 times the original strength. The mechanisms of these
and shaper in F50S50, as compared with S100. This demonstrates
changes will be discussed in the last section after presenting the
that the inclusion of fly ash alters the reaction kinetics in the alkali-
results of thermal structural evolution.
activated slag system. At reduced Ca2þ availability because of the
lower slag content, the evolution of the CSH may be delayed. The C-
3.2. XRD results A-S-H type gel will be formed by inclusion of aluminium in bridging
positions to replace the silicon [30].
The XRD patterns of raw fly ash and slag are shown in Fig. 2. The The X-ray diffractograms of F90S10 and F100 are similar but
presence of crystalline phase quartz and mullite is identified in the different from the observations of the slag-based sample. The CSH
unreacted fly ash. A small amount of crystalline phase containing peak is not identified, suggesting the presence of different binding
iron oxides, which has a structure similar to maghemite, is also phases in the samples containing mainly fly ash. The X-ray dif-
observed. The unreacted slag shows the peaks related to aragonite fractogram of these samples shows quartz, mullite and hematite
and gypsum. The traces of calcite and vaterite are also identified.
 Z. Pan et al. / Cement and Concrete Composites 86 (2018) 9e18 13

Fig. 3. XRD patterns of alkali-activated fly/slag pastes, as a function of the fly ash/slag ratio and exposure temperature: (A) 100% slag; (B) 50 wt.% slag/50 wt.% fly ash; (C) 10 wt.%
slag/90 wt.% fly ash; and (D) 100% fly ash (Aragonite-A, Calcite-C, Ca(OH)2-△, C4AH13-,, C-S-H->, Mullite-M, Quartz-Q, Maghemite-Mg).

which are also identified in the anhydrous fly ash samples. The exposure has little influence on these phases. The formation of
formation of new crystalline reaction products is not identified. zeolite and hydrotalcite like crystal phases is not observed in these
The effects of high temperature exposure on XRD patterns are samples. At elevated temperatures, the XRD patterns of these
also presented in Fig. 3. In the slag-based sample, the high tem- samples are fairly similar to that at ambient temperature.
perature exposure has distinct effects on structural evolution. At
300
C, the main feature of XRD patterns is the decomposition of 3.3. FTIR results
C4AH13. This is demonstrated by the disappearance of the peaks at
22.7
2q and 55.2
2q. A reduced intensity of the CSH peak is Fig. 4 shows the IR spectra of the unreacted raw materials. The IR
observed in S100 exposed to 300
C when compared with S100 spectra of both materials contain two broad bands in the ranges of
before exposure. This suggests that the structural evolution of CSH 1200-950 cm1 and 520-450 cm1. These bands are the indication
gel taking place at temperatures below 300
C. In the temperature
range of 300e600
C, the density of CSH peak did not show any
notable changes. At 600
C, the major changes, as shown by X-ray
diffractogram, are the decomposition of CH and calcite. When
F50S50 samples were exposed to 300
C, it is impossible to identify
any notable changes in XRD patterns. As the exposure temperature
was increased up to 600
C, a significant reduction in the intensity
of the peak (at 29.5
2q) is observed. This differs from the obser-
vation of major peak in the slag-based sample where the intensity
of the major peak remained almost the same. This difference again
indicates the different nature and structure of main binding phases
between F50S50 and S100. Besides the major peak, the reduction in
intensity of the peaks related to calcite is also observed. This is
consistent with the observations of the decomposition of calcite
occurred at around 600
C.
The samples mainly containing fly ash were thermally stable.
The presence of quartz and mullite in F90S10 and F100 was iden-
tified at all tested temperatures, indicating that high temperature Fig. 4. FTIR spectrum of raw materials.
14 Z. Pan et al. / Cement and Concrete Composites 86 (2018) 9e18

of internal vibrations of tetrahedral SiO4 and AlO4 groups. The band of the band at 877 cm1 indicates the decomposition of calcite at
in the high wavenumber is associated with anti-symmetric T-O 600
C. The weak band at 860 cm1 is indicative of a slightly
(T ¼ Si, Al) stretching vibrations. This band is observed at 978 cm1 increased vitreous phase.
in the slag and 1092 cm1 in the fly ash. A shift of this band to a The spectrum of F90S10 shows no change at temperatures up to
lower wavenumber suggests that the amorphous phase in the slag 600
C. On the other hand, the spectra of the fly ash only mixture
has a higher degree of crosslinking, as compared to the fly ash. This show some small changes. At 600
C, the shift of T-O stretching
is a result of high calcium content in its structure. The band in the band from 1022 cm1 to 1028 cm1 indicates a slight increase in the
range of 520-450 cm1 is due to O-S-O bending vibration. In the crosslinking degree of the gel. The narrowing of the band at
slag, a small band at 1621 cm1 is linked with stretching of OeH 457 cm1 suggests the formation of more condensed tetrahedral
groups and bending vibrations of molecular water. The presence of species, as a result of on-going geopolymerization at elevated
carbonates is associated with the bands at 1422-1427 cm1. Traces temperatures.
of gypsum are also identified, as evidenced by the absorption at
670 cm1. In the fly ash, the identified crystalline phases include 3.4. SEM
quartz and mullite. A broad band at 1092 cm1 and a weak band at
797 cm1 correspond, respectively, to the asymmetric and sym- The study of microstructure using backscattered electron (BSE)
metric of stretching vibrations of Si-O-Si in quartz. The band at imaging shows different morphologies of samples with different
554 cm1 is assigned to the octahedrally coordinated aluminium in slag/fly ash ratios. The micrograph of the slag-based mixture shows
mullite. All these crystalline phases are also observed in the XRD the crystallized hexagonal plates having a high content of Al, as
patterns of unreacted raw materials, as discussed before. identified by the EDS result. They are likely to be C4AH13 hydrate
Upon activation, the two main bands in S100, which correspond which has also been identified by the XRD pattern and IR spectra.
to the internal vibrations of SiO4 and AlO4 group, are different from With the inclusion of fly ash in the binder, the binding phases show
those of unreacted slag. The width of these bands decreases as the amorphous-looking morphology. Dispersed pores are randomly
reaction progresses. A small band of AFm-phase occurs at 421 cm1 distributed in the F50S50. Both unreacted fly ash and slag particles
[31]. This is consistent with the XRD result which indicates the are embedded in the matrix. However, the binding phases seem to
presence of C4AH13 in the slag-based mixture. In samples with slag loosely attach to the surface of the unreacted particles. The sample
content of 50%, the T-O stretching band has shifted slightly towards presents a less dense morphology than that observed in the slag-
higher wavenumbers from 966 cm1 for the slag only mixture to based binder. When the binder contains mainly fly ash, a number
1002 cm1 for the mixture with 50 wt.% fly ash/50 wt.% slag. This of cavities are observed throughout the paste. Some of them are
shift is attributed to the formation of a C-A-S-H type gel, as a result heterogeneously distributed in the matrix while others are
of the simultaneous activation of fly ash. The broad bands at observed on the surface of the unreacted fly ash particles. This is a
1423 cm1 and 877 cm-1 correspond, respectively, to the asym- typical feature of microstructure which is observed in the fly ash-
metric stretching vibrations and bending vibrations of the O-C-O based binder. The only difference between F90S10 and F100 is
bonds of CO2 3 group. This is a reflection of the presence of car- that the angular particles (associated with slag) are embedded in
bonates. In the mixture containing mainly fly ash, the intensity of the matrix of F90S10. In F100, only spherical particles (associated
1
the CO23 bands between 1418 cm and 1490 cm1 decrease. The with fly ash) are embedded in the matrix, which is indicative of the
weak peak at 877 cm1 is not identified. incomplete geopolymerization.
The spectrum of F90S10 is very similar to that of fly ash-based The effects of thermal exposure on the microstructure are pre-
mixture. This suggests that the inclusion of a small amount of sented in Figs. 6 and 7. The exposure to 300
C has induced
slag in the geopolymers seems to have no fundamental effects on distinctive changes of morphology of the slag-based binder. At
its final structure. A comparison of spectra between F90S10 and 300
C, the well-distributed hexagonal crystalline are not identi-
F100 shows some slight differences. Firstly, the CO2 3 bands are not fied. Some remnant slag particles are loosely embedded in the
observed in F100, suggesting that geopolymers have high resis- matrix. The pores can be identified in the matrix as well as at the
tance to atmospheric carbonation. Secondly, the T-O stretching interface between the matrix and the particles. A detailed exami-
band centred at a slightly higher wavenumber, as compared to nation of the image shows the traces of platy crystalline with high
F90S10. This indicates the slight differences of geopolymer gels calcium content. These crystal products are indicative of the pres-
between F90S10 and F100. The incorporation of Ca2þ into the N-A- ence of CH, which agrees with the XRD data of S100 heated at
S-H gel may lead to the formation of an N-(C)-A-S-H type gel in 300
C. With further increased temperature of exposure, the
F90S10. disappearance of these crystalline indicates the decomposition of
The structural changes of alkali-activated mixtures at elevated CH. At 600
C, S100 shows a weak microstructure with a number of
temperatures are also presented in Fig. 5, as characterised by the cracks. In the sample with 50 wt.% fly ash/50 wt.% slag, the thermal
infrared spectra results. When the slag-based mixture was exposed exposure resulted in an increase in porosity. At 300
C, the
to 300
C, the AFm band at 421 cm1 completely disappeared. This disconnected cracks are identified throughout the matrix. At
is consistent with the XRD observation that the vanish of C4AH13 600
C, the coalescence of cracks led to a major crack which nearly
peak also occurs at 300
C. A small shoulder at 877 cm1, together passes through the matrix. The microstructure of F90S10 seems to
with the bands between 1418 cm1 and 1454 cm1, are attributed have no change at all tested temperatures. In the fly ash-based
to the carbonation of soluble alkali species. At 600
C, the absorp- mixture, exposure to the temperature range of 300e600
C led to
tion band at 715 cm1 was assigned to the symmetric stretching a reduced number of remnant particles, indicating the on-going
vibrations of the Si-O-Si(Al) bridges. The main band at 966 cm1 reaction of fly ash during the thermal exposure. As a result, the
has shifted to 1002 cm1. This is similar to the shift of the main matrix at 600
C shows a denser morphology than those observed
band between S100 and F50S50 at ambient temperature, indicating at either ambient temperature or 300
C.
that the main binding phase in S100 is mainly dominated by a C-A-
S-H type gel at 600
C. The main band (at 1002 cm1) of F50S50 4. Discussion
heated up to 600
C, on the contrary, showing bare change of the
position. At 600
C, the band at 877 cm1 split into two bands at When the cementitious materials are exposed to elevated
877 cm1 and 860 cm1, respectively. The reduction in the intensity temperatures, occurrence of physical and chemical transformations
 Z. Pan et al. / Cement and Concrete Composites 86 (2018) 9e18 15

Fig. 5. FTIR spectrum of alkali-activated fly/slag pastes, as a function of the fly ash/slag ratio and exposure temperature: (A) 100% slag; (B) 50 wt.% slag/50 wt.% fly ash; (C) 10 wt.%
slag/90 wt.% fly ash; and (D) 100% fly ash.

results in substantial changes of compressive strength. On the basis have disappeared, leading to a strength loss. The F50S50 decreased
of the changes in microstructure, this section would provide in strength by 33%, which is slightly lower when compared to the
explanation for the changes in strength. strength loss occurred in the OPC binder. This is attributed to the
In the slag-based mixture, the metastable C4AH13 converted to absence of CH in F50S50.
C3AH6 due to the heat exposure. The SEM image shows that the In the mixture containing mainly fly ash, the main binding
hexagonal plates which were observed abundantly in the sample phase is an N-A-S-H type gel (Fig. 7 (B1)) and/or a low calcium N-
before exposure had disappeared at 300
C. The specific gravity of (C)-A-S-H type gel (Fig. 7 (A1)). Both gels show the structural sta-
C3AH6 is higher than that of C4AH13. Therefore, the conversion will bility up to 300
C. As a result, the strength of F90S10 and F100
lead to a reduction in volume (of solid product) and an increase in remains unchanged. At a higher temperature, the replacement of
porosity (of the specimen). Partly for this reason, a large strength Na by Ca is known to lower the thermal stability of zeolite materials
loss was observed at 300
C. Simultaneously, intensity of the CSH [34]. This might also explain why F90S10 decrease in strength in the
peak decreased at 300
C. This may indicate the slight decompo- range of 300e600
C. In the same temperature range, however, a
sition of CSH gel, which has also been previously observed in AAS large strength increase is observed in F100. Together with this
[32]. The loss of absorbed water (in CSH gel) will interrupt adhesive strength increase, a shift in the T-O band towards higher fre-
forces between gels, leading to the deterioration in strength. At quencies indicates the increased Si content in N-A-S-H gel. This is
600
C, the intensity of the CSH peak shows little change. However, indicative of the progressive reaction of fly ash remnant particles;
the peaks related to calcite and CH disappeared, indicating the exposure to the high temperature further promotes the formation
decomposition of these hydration products. These phase changes of N-A-S-H gel. As the reaction progresses, the matrix becomes
are also commonly occurred in OPC binder [33], which are believed denser due to the formation of more geopolymer gels filling in the
to be responsible for the strength loss in the temperature range of space, leading to the strength increase at temperatures between
300e600
C. In this temperature range, it is noted that the strength 300 and 600
C. At 600
C, the hot strength of F90S10 is only 50%
loss of AAS is less than that of OPC, indicating the lower content of that of F100. This distinctive difference suggests that main reaction
calcite and CH presented in AAS. In F50S50, the structure of the gel products may be different between F100 and F90S10. In F90S10, the
is mainly dominated by a C-A-S-H type gel (Fig. 6 (B1)) and/or N-A- dominant binding gel is N-(C)-A-S-H gel rather than N-A-S-H gel.
S-H gel. At a temperature of 300
C, the structure of the gel does not Compared to F100 cured at 60
C, F90S10 can develop strength at an
show significant changes except a small reduction in the intensity ambient temperature. This suggests that the formation of N-(C)-A-
of the peak at 29.5
2q which may be an indication of a slight S-H gel is not very temperature sensitive. As a result, the exposure
decomposition of C-A-S-H gel. The strength loss is 15%, which is to higher temperature may not promote further reaction of fly ash
lower than that of AAC or OPC binder. At 600
C, the calcite seems to remnant particles in F90S10. Further research is needed on this
16 Z. Pan et al. / Cement and Concrete Composites 86 (2018) 9e18

Fig. 6. BSE images of alkali-activated fly ash/slag pastes before and after exposure: (A) 100% slag and (B) 50 wt.% fly ash/50 wt.% slag.

interesting topic. 1. Conversion of calcium aluminate hydrates (principally C4AH13)

In terms of fire performance, this investigation studied four al- occurred in the slag-based sample at temperatures between 20
kali activated binders containing 0%, 10%, 50%, and 100% replace- and 300
C. This event was largely responsible for the significant
ment of fly ash by weight with slag. An OPC binder was also tested. strength loss of alkali-activated slag (S100). At temperatures
The fly ash-based binder shows the best performance followed by between 300 and 600
C, this mix shows an average 8% strength
F90S10, F50S50, OPC, and S100. The F100 was cured at 65
C for loss compared to 20% strength loss of the ordinary Portland
24 h, which is suitable for precast production. The 28 day strength cement (OPC) sample. This is associated with the low content of
reached 45 MPa in geopolymer binder containing 10% slag and 90% calcite and calcium hydroxide presented in the alkali-activated
fly ash when cured at ambient temperature. At 600
C, the strength slag. It should be noted that the test trends observed are valid
loss was only 17% in F90S10, while a much higher strength loss was only for one specific mix. Further study is required to investigate
observed in the OPC binder, which was nearly 50%. These results effects of mixing parameters (e.g., the nature of the activating
suggest that F90S10 may be a good alternative to the OPC binder for solution and the water to binder ratio) on the high-temperature
construction of structures suffering from high risks of fire. performance of alkali activated slag.
2. At 300
C, 50 wt.% fly ash/50 wt.% slag (F50S50) samples show
better performance with an average 12% strength loss compared
5. Conclusions to the 25% strength loss of ordinary Portland cement samples,
indicating the C-A-S-H gel is more thermal stable. At tempera-
The following main conclusions can be drawn from the current tures between 300 and 600
C, the strength loss of this mix is
investigation: associated with the decomposition of calcite.
 Z. Pan et al. / Cement and Concrete Composites 86 (2018) 9e18 17

Fig. 7. BSE images of alkali-activated fly ash/slag pastes before and after exposure: (A) 90 wt.% fly ash/10 wt.% slag and (B) 100% fly ash.

3. The N-A-S-H gel is the major binding phase in alkali-activated Acknowledgments

fly ash (F100) while the N-(C)-A-S-H gel may be the dominant
binding phase in 90 wt.% fly ash/10 wt.% slag (F90S10) samples. The authors are grateful for the financial support from Austra-
Both gels exhibited structural stability at 300
C. Heating sam- lian Research Council Linkage Grant No. LP160101484 and Western
ples to 600
C promote the reaction of fly ash remnant in the Sydney University ECA award. Authors would like to acknowledge
alkali-activated fly ash, leading to a strength increase. In the the contributions from the laboratory staff Mr Murray Bolden and
same temperature range, the thermal stability of N-(C)-A-S-H Mr Robert Marshall. The authors would also like to thank the
gel decreased due to the Na/Ca exchange. The strength loss was Advanced Materials Characterisation Facility and staff at WSU.
17% in 90 wt.% fly ash/10 wt.% slag samples.
4. The investigated alkali-activated binders show different fire
performance, which is dependent on their binder formulation. References
The heat-cured F100 shows the best performance followed by
[1] Z. Pan, et al., Damping and microstructure of fly ash-based geopolymers,
F90S10, F50S50, OPC, and S100. Compared to the heat curing J. Mater. Sci. 48 (2013) 3128e3137.
process, the ambient temperature curing has wider application [2] P. Nath, P.K. Sarker, Effect of GGBFS on setting, workability and early strength
properties of fly ash geopolymer concrete cured in ambient condition, Constr.
and less energy consumption. It is more feasible to use F90S10 as
Build. Mater. 66 (2014) 163e171.
an alternative to the ordinary Portland cement binder for con- [3] G.S. Ryu, et al., The mechanical properties of fly ash-based geopolymer con-
struction of structures requiring high fire resistance. crete with alkaline activators, Constr. Build. Mater. 47 (2013) 409e418.
[4] I. Ismail, et al., Microstructural changes in alkali activated fly ash/slag geo-
polymers with sulfate exposure, Mater. Struct. 46 (3) (2013) 361e373.
[5] M. Khandelwal, et al., Effect of strain rate on strength properties of low-
18 Z. Pan et al. / Cement and Concrete Composites 86 (2018) 9e18

calcium fly-ash-based geopolymer mortar under dry condition, Arabian J. admixtures and additives on the properties of alkali-activated fly ash, Mater.
Geosciences 6 (7) (2013) 2383e2389. Des. 53 (2014) 1005e1025.
[6] S. Chuah, et al., The properties of fly ash based geopolymer mortars made with [20] Z. Li, S. Liu, Influence of slag as additive on compressive strength of fly ash-
dune sand, Mater. Des. 92 (2016) 571e578. based geopolymer, J. Mater. Civ. Eng. 19 (6) (2007) 470e474.
[7] A. van Riessen, et al., Bayer-geopolymers: an exploration of synergy between [21] M. Chi, R. Huang, Binding mechanism and properties of alkali-activated fly
the alumina and geopolymer industries, Cem. Concr. Compos. 41 (2013) ash/slag mortars, Constr. Build. Mater. 40 (2013) 291e298.
29e33. [22] N.K. Lee, H.K. Lee, Reactivity and reaction products of alkali-activated, fly ash/
[8] C. Li, H. Sun, L. Li, A review: the comparison between alkali-activated slag (Si þ slag paste, Constr. Build. Mater. 81 (2015) 303e312.
Ca) and metakaolin (Si þ Al) cements, Cem. Concr. Res. 40 (9) (2010) [23] M. Guerrieri, J. Sanjayan, F. Collins, Residual strength properties of sodium
1341e1349. silicate alkali activated slag paste exposed to elevated temperatures, Mater.
[9] K. Dombrowski, A. Buchwald, M. Weil, The influence of calcium content on the Struct. 43 (6) (2010) 765e773.
structure and thermal performance of fly ash based geopolymers, J. Mater. Sci. [24] H.T. Türker, et al., Microstructural alteration of alkali activated slag mortars
42 (9) (2007) 3033e3043. depend on exposed high temperature level, Constr. Build. Mater. 104 (2016)
[10] Z. Pan, Z. Tao, T. Murphy, R. Wuhrer, High temperature performance of 169e180.
mortars containing fine glass powders, J. Clean. Prod. 162 (2017) 16e26. [25] M. Guerrieri, J.G. Sanjayan, Behavior of combined fly ash/slag-based geo-
[11] P.K. Sarker, S. McBeath, Fire endurance of steel reinforced fly ash geopolymer polymers when exposed to high temperatures, Fire Mater. 34 (4) (2010)
concrete elements, Constr. Build. Mater. 90 (2015) 91e98. 163e175.
[12] H.Y. Zhang, et al., Thermal behavior and mechanical properties of geopolymer [26] G. Kürklü, The effect of high temperature on the design of blast furnace slag
mortar after exposure to elevated temperatures, Constr. Build. Mater. 109 and coarse fly ash-based geopolymer mortar, Compos. Part B Eng. 92 (2016)
(2016) 17e24. 9e18.
[13] W.D.A. Rickard, A. van Riessen, Performance of solid and cellular structured fly [27] 129 -MHT, R, Compressive strength for service and accident conditions, Mater.
ash geopolymers exposed to a simulated fire, Cem. Concr. Compos. 48 (2014) Struct. 28 (7) (1995) 410e414.
75e82. [28] F. Lo Monte, P.G. Gambarova, Thermo-mechanical behavior of baritic concrete
[14] L. Zuda, et al., Alkali-activated aluminosilicate composite with heat-resistant exposed to high temperature, Cem. Concr. Compos. 53 (2014) 305e315.
lightweight aggregates exposed to high temperatures: mechanical and wa- [29] Z. Li, et al., Effects of mineral admixtures and lime on disintegration of alkali-
ter transport properties, Cem. Concr. Compos. 32 (2) (2010) 157e163. activated slag exposed to 50
C, Constr. Build. Mater. 70 (2014) 254e261.
[15] Z. Pan, J.G. Sanjayan, Factors influencing softening temperature and hot- [30] F. Puertas, et al., A model for the C-A-S-H gel formed in alkali-activated slag
strength of geopolymers, Cem. Concr. Compos. 34 (2) (2012) 261e264. cements, J. Eur. Ceram. Soc. 31 (12) (2011) 2043e2056.
[16] A. Ferna ndez-Jime nez, et al., New cementitious materials based on alkali- [31] A. Hidalgo, et al., Microstructure development in mixes of calcium aluminate
activated fly ash: performance at high temperatures, J. Am. Ceram. Soc. 91 cement with silica fume or fly ash, J. Therm. Analysis Calorim. 96 (2) (2009)
(10) (2008) 3308e3314. 335e345.
[17] F.U.A. Shaikh, V. Vimonsatit, Compressive strength of fly-ash-based geo- [32] P. Rovnaník, P. Bayer, P. Rovnaníkova , Characterization of alkali activated slag
polymer concrete at elevated temperatures, Fire Mater. 39 (2) (2015) paste after exposure to high temperatures, Constr. Build. Mater. 47 (2013)
174e188. 1479e1487.
[18] L. Vickers, Z. Pan, Z. Tao, A. van Riessen, In situ elevated temperature testing of [33] M.W. Hussin, et al., Performance of blended ash geopolymer concrete at
fly ash based geopolymer composites, Materials 9 (2016) 445e458. elevated temperatures, Mater. Struct. 48 (3) (2015) 709e720.
[19] A.M. Rashad, A comprehensive overview about the influence of different [34] D.W. Breck, Zeolite Molecular Sieves, John Wiley & Sons, New York, 1974.