Evaluation of the Effect of Silica Fume on Amorphous Fly Ash Geopolymers Exposed to Elevated Temperature

: The properties of amorphous geopolymer with silica fume addition after heat treatment was rarely reported in the geopolymer ﬁeld. Geopolymer was prepared by mixing ﬂy ash and alkali activator. The silica fume was added in 2% and 4% by weight. The geopolymer samples were cured at room temperature for 28 days before exposed to an elevated temperature up to 1000 ◦ C. The incorporation of 2% silica fume did not cause signiﬁcant improvement in the compressive strength of unexposed geopolymer. Higher silica fume content of 4% reduced the compressive strength of the unexposed geopolymer. When subjected to elevated temperature, geopolymer with 2% silica fume retained higher compressive strength at 1000 ◦ C. The addition of silica fume in ﬂy ash geopolymer caused a lower degree of shrinkage and expansion, as compared to geopolymer without the addition of silica fume. Crystalline phases of albite and magnetite were formed in the geopolymer at 1000 ◦ C.


Introduction
Geopolymer is one of the alternatives to ordinary Portland cement (OPC), as a construction material, because it is more environmentally friendly [1,2]. The OPC production involves the burning of large amounts of fuel, thus resulting in almost 5% of total global CO 2 emissions [3]. Geopolymers are inorganic materials, introduced by Davidovits [4], and are made by mixing the solid aluminosilicate source with liquid alkaline activator, which is then cured at ambient temperature or a slightly higher temperature (<100 • C). The aluminosilicate sources, which consist of high silica and alumina, could be obtained from natural resources (kaolin and metakaolin) or industrial by-products (fly ash, bottom ash and rice husk ash). In addition, geopolymer could be used as an alternative to OPC because to its properties such as higher early strength, being superior against thermal resistance and lower CO 2 emission [5].
Geopolymers are generally recognized as superior systems with respect to thermal resistance. The amorphous aluminosilicate structure with three-dimensional arrangement enables geopolymers to possess ceramic-like properties and exhibit high thermal resistant [6]. Studies have found that geopolymers are incombustible and do not spall with extended heating as geopolymer contains lower Ca(OH) 2 content in comparison to OPC [7]. Besides, geopolymers also exhibit superior thermal stability as no smokes or toxic gases are released when they are heated [8].
The excellent thermal stability of geopolymers not only increases its endurance at high temperature, but also alters its other properties and geopolymer structure. Deterioration of strength of geopolymer at elevated temperatures was commonly reported [9,10]. However, an increase in mechanical strength of geopolymer when exposed to elevated temperature has also been reported [11]. Park et al. [12] observed the occurrence of further geopolymerisation reaction in class F fly ash geopolymer at 400 • C, which leads to increasing strength. The compressive strength dropped at a temperature beyond 400 • C, as result of the initiation and propagation of thermal cracks. In contrast, Wongsa et al. [13] found that higher residual strength of fly ash geopolymers incorporating natural mullite and zeolite was achieved at 900 • C, which is associated with the increase in crystallinity of the matrix. Heat treatment would initiate phase transformation in a geopolymer matrix. Crystalline nepheline (NaAlSiO 4 ) is most reported in the heated geopolymer [14,15]. Based on Rovnaník and Safránková [15], albite (NaAlSi 3 O 8 ) was also formed, except from nepheline, boosting the flexural strength of fly ash geopolymers. Payakaniti et al. [16] studied the relationship between thermal stability and the magnetic properties of high calcium fly ash geopolymer, and observed the formation of new phases of nepheline and zeolite and an increased hematite peak above 800 • C. The antiferromagnetic behaviour of hematite decreased the magnetic properties of geopolymers and eventually switched from ferromagnetic to paramagnetic properties when the temperature reached 1200 • C.
The incorporation of nanoparticles leads to the enhancement of physical structure and mechanical strength of geopolymer matrices. Silica fume is one of the potential nanoparticles that could be introduced into geopolymer as it is a highly effective pozzolanic and possesses extreme fineness and high silica content [17]. Incorporation of silica fume induces Si content, which improves the yield of geopolymer gels and results in a strength increment [18][19][20][21] with lower porosity and water absorption [22,23]. Liu et al. [21] reported superior compressive strength of 151 MPa when 30% of silica fume was incorporated into the fly ash/slag geopolymer concretes. It has been proven that the addition of silica fume results in improvement of mechanical strength, durability, and structure properties of geopolymers. Despite understanding the thermal resistance of pure geopolymers without additives, and the mechanical strength development of geopolymer with silica fume addition, the influence of silica fume on the properties of geopolymers at elevated temperatures is not known. The past literature mostly covers the evolution of microstructure and mechanical strength under the influence of a silica additive.
Therefore, this paper evaluates and compares the effect of the addition of silica fume on physical, mechanical properties and thermal resistance of fly ash geopolymers. A quantitative research study was performed on fly ash geopolymer with varying dosages of silica fume (0%, 2%, and 4%).

Materials
The Class F fly ash used in this work was sourced from Manjung Coal-Fired Power Plant, Perak. The chemical composition of the fly ash is displayed in Table 1. The fly ash is made up mainly of SiO 2 (56.3%) and Al 2 O 3 (28.0%). Figure 1a demonstrates that the fly ash is highly amorphous with some crystalline inclusions of quartz, mullite and hematite. The fly ash particles revealed in Figure 1b have spherical shapes with smooth surfaces and wide size distributions.   Sodium hydroxide (NaOH) pellets with an assay (total alkalinity calculated as NaOH) of 98.0-100.5% were supplied by Progressive Scientific Sdn. Bhd., Selangor. Technical grade liquid sodium silicate (Na2SiO3) with 60.5% H2O, 30.1% SiO2 and 9.4% Na2O with 3.2 SiO2/Na2O ratios was supplied by South Pacific Chemicals Industries (SPCI) Sdn. Bhd., Malaysia. Nano-silica (silica fume) with an average particle size of 7 nm, surface area of 395 ± 25 m 2 /g and density of 36.8 kg/m 3 , supplied by Sigma-Aldrich Corporation, was added into fly ash geopolymers as the additive.

Geopolymer Formation
The binder-to-liquid ratio and Na2SiO3/NaOH ratio was fixed at 2.0 and 2.5, respectively, for the geopolymer mixture. The alkali activator was prepared by mixing NaOH 10 M solution with liquid Na2SiO3. Silica fume was first dry mixed with fly ash and then mixed with the alkali activator to achieve homogeneous slurry. The silica fume was added in 0%, 2% and 4% by weight of fly ash. The slurry was casted in molds with dimensions of 50 × 50 × 50 mm and then compacted to remove the entrapped air. The samples were first cured at room temperature (29 °C) for 24 h before oven-curing at 60 °C for another 24 h. This curing regime was selected as pre-curing at room temperature, followed by ovencuring, which are beneficial for strength improvement, based on Ranjbar et al. work [24]. After the curing process, the hardened geopolymer samples were sealed and kept under room temperature for 28 days.

Heat Treatment
The 28 day cured fly ash geopolymers were heat-treated in a furnace at 200 °C, 600 °C and 1000 °C, with a heating rate of 5 °C/min for 2 h. A set of geopolymer samples was left unexposed for comparison purposes.

Testing, Analysis and Characterization Method
The density of geopolymers was calculated by measuring the dimension and mass of the samples before and after exposure to elevated temperature, in accordance with ASTM C138/C138M. Compressive test was accessed according to ASTM C109/C109M by using universal testing machine (UTM) modeled Shimadzu UH-1000 kNI. The loading rate was fixed constant at 5 mm/min. Three samples were compressed to obtain the average strength value. The microstructure of fly ash and fly ash geopolymers were revealed with Sodium hydroxide (NaOH) pellets with an assay (total alkalinity calculated as NaOH) of 98.0-100.5% were supplied by Progressive Scientific Sdn. Bhd., Selangor. Technical grade liquid sodium silicate (Na 2 SiO 3 ) with 60.5% H 2 O, 30.1% SiO 2 and 9.4% Na 2 O with 3.2 SiO 2 /Na 2 O ratios was supplied by South Pacific Chemicals Industries (SPCI) Sdn. Bhd., Malaysia. Nano-silica (silica fume) with an average particle size of 7 nm, surface area of 395 ± 25 m 2 /g and density of 36.8 kg/m 3 , supplied by Sigma-Aldrich Corporation, was added into fly ash geopolymers as the additive.

Geopolymer Formation
The binder-to-liquid ratio and Na 2 SiO 3 /NaOH ratio was fixed at 2.0 and 2.5, respectively, for the geopolymer mixture. The alkali activator was prepared by mixing NaOH 10 M solution with liquid Na 2 SiO 3 . Silica fume was first dry mixed with fly ash and then mixed with the alkali activator to achieve homogeneous slurry. The silica fume was added in 0%, 2% and 4% by weight of fly ash. The slurry was casted in molds with dimensions of 50 × 50 × 50 mm and then compacted to remove the entrapped air. The samples were first cured at room temperature (29 • C) for 24 h before oven-curing at 60 • C for another 24 h. This curing regime was selected as pre-curing at room temperature, followed by ovencuring, which are beneficial for strength improvement, based on Ranjbar et al. work [24]. After the curing process, the hardened geopolymer samples were sealed and kept under room temperature for 28 days.

Heat Treatment
The 28 day cured fly ash geopolymers were heat-treated in a furnace at 200 • C, 600 • C and 1000 • C, with a heating rate of 5 • C/min for 2 h. A set of geopolymer samples was left unexposed for comparison purposes.

Testing, Analysis and Characterization Method
The density of geopolymers was calculated by measuring the dimension and mass of the samples before and after exposure to elevated temperature, in accordance with ASTM C138/C138M. Compressive test was accessed according to ASTM C109/C109M by using universal testing machine (UTM) modeled Shimadzu UH-1000 kNI. The loading rate was fixed constant at 5 mm/min. Three samples were compressed to obtain the average strength value. The microstructure of fly ash and fly ash geopolymers were revealed with scanning electron microscopy (SEM) modeled JEOL JSM-6460 LA. Perkin Elmer Fourier Transform Infrared Spectroscopy (FTIR) was used to analyze the functional group of fly ash, fly ash geopolymer and bond vibration frequencies. The samples were scanned from 650 cm −1 to 4000 cm −1 with a resolution of 4 cm −1 . X-ray diffraction (XRD) was performed to characterize the mineralogical changes using Shimadzu X-ray Diffractometer modeled XRD-6000. To investigate the shrinkage or expansion of fly ash geopolymer samples, LINSEISL76 Platinum Series dilatometer was used to investigate its thermal properties. Figure 2 displays the surface conditions of unexposed fly ash geopolymers with varying contents of silica fume. It could be clearly seen that more pores were found on the surface of geopolymer without the addition of silica fume ( Figure 2a). However, porosity of geopolymer decreased as the content of silica fume increased. Silica fume behaved as a pore-filler as it had a very fine particle size [19], which could fill the empty space within the geopolymer matrix. In addition, some visible white spots could be observed on the surface of geopolymers, which was known as the efflorescence effect [25]. scanning electron microscopy (SEM) modeled JEOL JSM-6460 LA. Perkin Elmer Fourier Transform Infrared Spectroscopy (FTIR) was used to analyze the functional group of fly ash, fly ash geopolymer and bond vibration frequencies. The samples were scanned from 650 cm −1 to 4000 cm −1 with a resolution of 4 cm −1 . X-ray diffraction (XRD) was performed to characterize the mineralogical changes using Shimadzu X-ray Diffractometer modeled XRD-6000. To investigate the shrinkage or expansion of fly ash geopolymer samples, LIN-SEISL76 Platinum Series dilatometer was used to investigate its thermal properties. Figure 2 displays the surface conditions of unexposed fly ash geopolymers with varying contents of silica fume. It could be clearly seen that more pores were found on the surface of geopolymer without the addition of silica fume ( Figure 2a). However, porosity of geopolymer decreased as the content of silica fume increased. Silica fume behaved as a pore-filler as it had a very fine particle size [19], which could fill the empty space within the geopolymer matrix. In addition, some visible white spots could be observed on the surface of geopolymers, which was known as the efflorescence effect [25].  Figure 3 depicts the physical appearances of exposed fly ash geopolymers with a different percentage of silica fume addition. In general, the color changed from dark grey to light grey when heated to 200 °C (Figure 3b), while it further changed to beige when exposed to 60 °C ( Figure 3c). The color changes at 60 °C were mainly associated with the oxidation changes of iron oxide content in fly ash [26]. Moreover, the color of the sample further changed to dark brown at 1000 °C (Figure 3d), which was related to the recrystallisation of the amorphous phase in the geopolymer matrix [27].  Figure 3 depicts the physical appearances of exposed fly ash geopolymers with a different percentage of silica fume addition. In general, the color changed from dark grey to light grey when heated to 200 • C (Figure 3b), while it further changed to beige when exposed to 60 • C ( Figure 3c). The color changes at 60 • C were mainly associated with the oxidation changes of iron oxide content in fly ash [26]. Moreover, the color of the sample further changed to dark brown at 1000 • C (Figure 3d), which was related to the recrystallisation of the amorphous phase in the geopolymer matrix [27].

Physical Observation
However, there were severe cracks observed in geopolymer without additive, whereby minor cracks observed in the geopolymer with silica fume addition at 1000 • C. It is suggested that, by the incorporation of silica fume, it tended to solve the problem of cracks with only a small amount (2% and 4%) of silica fume. Magnetochemistry 2021, 7, x FOR PEER REVIEW 5 of 15 However, there were severe cracks observed in geopolymer without additive, whereby minor cracks observed in the geopolymer with silica fume addition at 1000 °C. It is suggested that, by the incorporation of silica fume, it tended to solve the problem of cracks with only a small amount (2% and 4%) of silica fume. Figure 4 illustrates the density changes of fly ash geopolymers with varying percentages of silica fume at elevated temperatures. In general, the change in density of unexposed geopolymer, regardless of the addition of silica fume, was not significant. The unexposed geopolymer had a density of 1844.6 kg/m 3 . Increasing silica fume content up to 2% slightly raised the density to 1893.2 kg/m 3 . Inclusion of silica fume enhanced the geopolymerisation reaction due to the induced Si content, which rendered more reaction products and hence increased density. The observation was supported by Adak et al. [18]. The density decreased to 1875.6 kg/m 3 with 4% of silica fume, which was associated with the hindered geopolymerisation reaction at high silica contents. When excess silica content reacted with insufficient alumina content, the reaction products and the density reduced. It was further supported by Ghanbari et al. [28] who produced a metakaolin geopolymer with the addition of silica fume.  Figure 4 illustrates the density changes of fly ash geopolymers with varying percentages of silica fume at elevated temperatures. In general, the change in density of unexposed geopolymer, regardless of the addition of silica fume, was not significant. The unexposed geopolymer had a density of 1844.6 kg/m 3 . Increasing silica fume content up to 2% slightly raised the density to 1893.2 kg/m 3 . Inclusion of silica fume enhanced the geopolymerisation reaction due to the induced Si content, which rendered more reaction products and hence increased density. The observation was supported by Adak et al. [18]. The density decreased to 1875.6 kg/m 3 with 4% of silica fume, which was associated with the hindered geopolymerisation reaction at high silica contents. When excess silica content reacted with insufficient alumina content, the reaction products and the density reduced. It was further supported by Ghanbari et al. [28] who produced a metakaolin geopolymer with the addition of silica fume.

Density Measurement
In the case of exposed geopolymers, all geopolymers displayed a reduction trend in the density with increasing temperature. The density loss corresponded to the mass loss in geopolymer during sintering. Mass loss is linked with the moisture loss of chemically and physically bonded water below 600 • C [29]. In comparison, the density loss of geopolymers without silica fume (9.3-28.1%) was higher than the geopolymer with silica fume addition (5.8-27.6%). This was because the geopolymers without silica fume (10.4-21.7%) have higher mass loss than the geopolymer with silica fume addition (11-20.7%). It meant that, by the inclusion of silica fume, it reduced the mass loss of geopolymers and hence resulted in reduced density loss.  In the case of exposed geopolymers, all geopolymers displayed a reduction trend in the density with increasing temperature. The density loss corresponded to the mass loss in geopolymer during sintering. Mass loss is linked with the moisture loss of chemically and physically bonded water below 600 °C [29]. In comparison, the density loss of geopolymers without silica fume (9.3%-28.1%) was higher than the geopolymer with silica fume addition (5.8%-27.6%). This was because the geopolymers without silica fume (10.4%-21.7%) have higher mass loss than the geopolymer with silica fume addition (11-20.7%). It meant that, by the inclusion of silica fume, it reduced the mass loss of geopolymers and hence resulted in reduced density loss. Figure 5 shows the compressive strength evolution of fly ash geopolymer before and after being exposed to elevated temperatures. Unexposed geopolymers without silica fume attained a compressive strength of 14.3 MPa. Incorporation of 2% of silica fume did not significantly improve the compressive strength (14.5 MPa), despite the reduced observable pores and increased density, as shown in Figures 2 and 4, respectively. However, obvious degradation of compressive strength to 8.6 MPa was recorded for the geopolymer with 4% silica fume. This could be due to the increased workability of geopolymer mixture at higher silica fume content. The observation was supported by Das et al. [30], who reported a reduction of compressive strength due to increased flowability with increasing silica fume from 1% to 3%. Increased workability restricted the contact between the aluminosilicate and alkali activator.  Figure 5 shows the compressive strength evolution of fly ash geopolymer before and after being exposed to elevated temperatures. Unexposed geopolymers without silica fume attained a compressive strength of 14.3 MPa. Incorporation of 2% of silica fume did not significantly improve the compressive strength (14.5 MPa), despite the reduced observable pores and increased density, as shown in Figures 2 and 4, respectively. However, obvious degradation of compressive strength to 8.6 MPa was recorded for the geopolymer with 4% silica fume. This could be due to the increased workability of geopolymer mixture at higher silica fume content. The observation was supported by Das et al. [30], who reported a reduction of compressive strength due to increased flowability with increasing silica fume from 1% to 3%. Increased workability restricted the contact between the aluminosilicate and alkali activator.  In the case of exposed geopolymer without silica fume, increasing temperature raised its compressive strength up to 200 °C (67 MPa) and the strength reduced after 200 °C. The reason of the strength gained at 200 °C was related to the quicker of geopolymerisation reaction as temperature of 200 °C could stiffen the binder at a faster rate [31]. The compressive strength dropped above 200 °C and was associated with the high-density loss ( Figure 4) and crack formation (Figure 3). Particularly at 1000 °C, the geopolymer samples exhibited wide visible cracks and the highest density loss (28.1%). The formation of crack would destroy the sample and hence be unbeneficial to the strength performance of the geopolymer.

Compressive Strength Test
Geopolymer with 2% silica fume also possessed compressive strength increment up to 49.7 MPa at 200 °C. The compressive strength further reduced to 25.0 MPa at 600 °C. At In the case of exposed geopolymer without silica fume, increasing temperature raised its compressive strength up to 200 • C (67 MPa) and the strength reduced after 200 • C. The reason of the strength gained at 200 • C was related to the quicker of geopolymerisation reaction as temperature of 200 • C could stiffen the binder at a faster rate [31]. The compressive strength dropped above 200 • C and was associated with the high-density loss ( Figure 4) and crack formation (Figure 3). Particularly at 1000 • C, the geopolymer samples exhibited wide visible cracks and the highest density loss (28.1%). The formation of crack would destroy the sample and hence be unbeneficial to the strength performance of the geopolymer.
Geopolymer with 2% silica fume also possessed compressive strength increment up to 49.7 MPa at 200 • C. The compressive strength further reduced to 25.0 MPa at 600 • C. At 1000 • C, the compressive strength remained (26.5 MPa), which might be due to the crystalline phase formation [27]. However, for the geopolymer with 4% silica fume, the compressive strength increased up to 24.3 MPa at 200 • C, 34.3 MPa at 600 • C and further reduced to 18.7 MPa at 1000 • C. The sharp spike in the compressive strength at 200 • C, regardless of the addition of silica fume, was mainly due to the quasi equilibrium of geopolymerisation reaction under room temperature. Thus, when heat was applied, the temperature-controlled dissolution and geopolymerisation occurred and consequently enhanced the compressive strength of the final products [24]. This subsequently caused the further increment in the compressive strength of geopolymer with 4% silica fume up to 600 • C. The higher silica content tended to slow down the kinetics of geopolymerisation reaction [24] and hence a higher temperature was required to improve the mechanical strength. The strength reduction at 1000 • C for geopolymer with 4% silica fume was due to the higher degree of expansion, which will be discussed in the next section.
In summary, the strength retention of geopolymer with 2% of silica fume at elevated temperature proved that addition of a low amount of silica fume helped to improve the thermal stability of the geopolymer. Figure 6 depicts the dilatometry performance of fly ash geopolymers with varying silica fumes up to 1000 • C. Geopolymers without silica fume shrunk from room temperature to 700 • C and then expanded up to 1000 • C. Similar trend was observed for geopolymer with the addition of silica fume up to 800 • C. Beyond 800 • C, shrinkage was observed. In general, the dilatometric curves of geopolymers can be divided into three zones. The first zone (up to 300 • C) was due to the establishment of capillary contraction when the water was evaporated. The second zone (300 • C-700 • C) was associated with the dehydroxylation reaction. The third zone (700 • C-1000 • C) corresponded to the viscous sintering effect and phase transformation.  As seen from Figure 6, the geopolymer with silica fume content shrunk at a slightly lower temperature than the geopolymer without silica fume. This was related to the very fine particle of silica fume, which could strengthen the formation of rings and chains of polysialates that shrink by 100 °C. The degree of shrinkage of geopolymer up to 700 °C with silica fume was lower compared to the reference geopolymer (without silica fume). In the contrary, in the temperature range of 600 to 800 °C, the expansion of geopolymer with 2% silica fume was the lowest compared to the counterpart without silica fume and As seen from Figure 6, the geopolymer with silica fume content shrunk at a slightly lower temperature than the geopolymer without silica fume. This was related to the very fine particle of silica fume, which could strengthen the formation of rings and chains of polysialates that shrink by 100 • C. The degree of shrinkage of geopolymer up to 700 • C with silica fume was lower compared to the reference geopolymer (without silica fume). In the contrary, in the temperature range of 600 to 800 • C, the expansion of geopolymer with 2% silica fume was the lowest compared to the counterpart without silica fume and with 4% of silica fume, which consequently caused the highest strength attained for geopolymer with 2% silica fume, as shown in Figure 5. The higher degree of expansion of geopolymer without silica fume and with 4% silica fume was the most probable reason for strength reduction at 1000 • C. Figure 7 illustrates the phase identification of unexposed fly ash geopolymers with varying percentages of silica fume. Geopolymers are highly amorphous. The existence of main crystalline phases of mullite and quartz and minor hematite phase in geopolymer was originated from fly ash (Figure 1a). However, a new crystalline phase of zeolite appeared in the geopolymer. The zeolite formation was recognized as the secondary reaction products during the geopolymerisation process in accordance with Rashad and Zeedan [32]. It is observed that the intensity of mullite and quartz (26.5 • 2θ) was high in the geopolymer with 4% of silica fume. The high content entailed that lower dissolution of aluminosilicate and progress of chemical reaction to form the geopolymer network, as aforementioned, and prove the low compressive strength of the geopolymer with 4% of silica fume.  Figure 8 displays the phase analysis of exposed fly ash geopolymers. The diffraction peaks of quartz and mullite at 26.5° 2θ remained in the geopolymers, even though they were heated to elevated temperatures, which verified that they were highly stable. Quartz and mullite did not decompose below a temperature of 1000 °C as they have a high melting point of 1840 °C and 1670 °C, respectively. Besides, a minor peak of hematite disappeared at 1000 °C was believed to be transformed into magnetite at 600 °C, which leads to the beige color appearance of the geopolymer sample (Figure 3c). The occurrence of this conversion process was common under elevated temperature [33]. A new crystalline peak of albite appeared at 1000 °C. The presence of albite was supposed to retain the compressive strength of geopolymer at an elevated temperature, as supported by Alehyen et al. [27].  Figure 8 displays the phase analysis of exposed fly ash geopolymers. The diffraction peaks of quartz and mullite at 26.5 • 2θ remained in the geopolymers, even though they were heated to elevated temperatures, which verified that they were highly stable. Quartz and mullite did not decompose below a temperature of 1000 • C as they have a high melting point of 1840 • C and 1670 • C, respectively. Besides, a minor peak of hematite disappeared at 1000 • C was believed to be transformed into magnetite at 600 • C, which leads to the beige color appearance of the geopolymer sample (Figure 3c). The occurrence of this conversion process was common under elevated temperature [33]. A new crystalline peak of albite appeared at 1000 • C. The presence of albite was supposed to retain the compressive strength of geopolymer at an elevated temperature, as supported by Alehyen et al. [27].

Phase Identification
ing point of 1840 °C and 1670 °C, respectively. Besides, a minor peak of hematite disappeared at 1000 °C was believed to be transformed into magnetite at 600 °C, which leads to the beige color appearance of the geopolymer sample (Figure 3c). The occurrence of this conversion process was common under elevated temperature [33]. A new crystalline peak of albite appeared at 1000 °C. The presence of albite was supposed to retain the compressive strength of geopolymer at an elevated temperature, as supported by Alehyen et al. [27].   Figure 9 depicts the IR spectrum of fly ash. The main band (1029 cm −1 ) and a minor band (729 cm −1 ) were recognized as asymmetric stretching vibration (Si-O-Si and Si-O-Al) [34] and symmetric stretching vibration (Si-O-Si) [35], respectively. Another band of 2348 cm −1 was assigned as symmetric axial deformation of the CO 2 [36], whereby the region at 1529 cm −1 was verified as the stretching vibration (C-O) [37]. In addition, the bands of 3608 cm −1 and 1694 cm −1 were assigned as OH stretching [38] and bending [39] vibrations, respectively.  Figure 9 depicts the IR spectrum of fly ash. The main band (1029 cm −1 ) and a minor band (729 cm −1 ) were recognized as asymmetric stretching vibration (Si-O-Si and Si-O-Al) [34] and symmetric stretching vibration (Si-O-Si) [35], respectively. Another band of 2348 cm −1 was assigned as symmetric axial deformation of the CO2 [36], whereby the region at 1529 cm −1 was verified as the stretching vibration (C-O) [37]. In addition, the bands of 3608 cm −1 and 1694 cm −1 were assigned as OH stretching [38] and bending [39] vibrations, respectively.  Figure 10 represents the IR spectra of unexposed fly ash geopolymers with different contents of silica fume. The absorption bands of geopolymer were almost similar with fly ash (Figure 9). The additional band at ~3300 cm −1 was also corresponded to the O-H stretching vibration [38]. The main band of fly ash at 1029 cm −1 shifted to a lower wavenumber (~980 cm −1 ). The shift implied that the formation of amorphous aluminosilicate gel phase after the activation of fly ash amorphous phase by alkali solution [38].  Figure 10 represents the IR spectra of unexposed fly ash geopolymers with different contents of silica fume. The absorption bands of geopolymer were almost similar with fly ash (Figure 9). The additional band at~3300 cm −1 was also corresponded to the O-H stretching vibration [38]. The main band of fly ash at 1029 cm −1 shifted to a lower wavenumber (~980 cm −1 ). The shift implied that the formation of amorphous aluminosilicate gel phase after the activation of fly ash amorphous phase by alkali solution [38].  Figure 10 represents the IR spectra of unexposed fly ash geopolymers with different contents of silica fume. The absorption bands of geopolymer were almost similar with fly ash (Figure 9). The additional band at ~3300 cm −1 was also corresponded to the O-H stretching vibration [38]. The main band of fly ash at 1029 cm −1 shifted to a lower wavenumber (~980 cm −1 ). The shift implied that the formation of amorphous aluminosilicate gel phase after the activation of fly ash amorphous phase by alkali solution [38]. Figure 10. IR spectra of un-exposed fly ash geopolymers with varying silica fume content addition.

Functional Group Identification
The addition of silica fume did not alter or produce a new absorption band in geopolymer. Increasing the silica fume content increased the wavenumber from 983 cm −1 to Figure 10. IR spectra of un-exposed fly ash geopolymers with varying silica fume content addition.
The addition of silica fume did not alter or produce a new absorption band in geopolymer. Increasing the silica fume content increased the wavenumber from 983 cm −1 to 990 cm −1 , associated with the formation of geopolymer network that is rich in Si bonds [40]. The intensity of this band for the reference geopolymer and geopolymer with 2% silica fume is almost similar, while the intensity reduced (higher percentage transmittance) in geopolymer with 4% silica fume. The observation complied with the compressive strength result in Figure 5, whereby the compressive strength of geopolymer with 4% silica fume was lower than the counterpart with no silica fume and 2% of silica fume.
When subjected to 200 • C (Figure 11), increasing silica fume content up to 4%, the intensity of the band at~980 cm −1 decreased with increasing silica fume content, which was also consistent with the recorded compressive strength, as shown in Figure 5. However, at 1000 • C, the geopolymer with 2% silica fume has the highest intensity of this band due to increased Si-O-Al bonds which were further verified by the phase identification ( Figure 8) as albite (NaAlSi 3 O 8 ) was formed at 1000 • C in addition to the sodium aluminium silicate hydrate (N-A-S-H).
In addition, as referred to in Figure 11, the absorption bands due to OH stretching vibration (~3300-33,600 cm −1 ) and OH bending vibration (~1630 cm −1 ) reduced in intensity at high temperature exposure. The indicated the dehydration of structure with increasing temperature. The intensity of C-O bonds at~2340 cm −1 and~1470 cm −1 also reduced due to the breakage of C-O at elevated temperatures. Based on Figures 10 and 11, there are differences in the intensity of bands at~3300 cm −1 and~1630 cm −1 of the unexposed and exposed geopolymer samples. It was supposed that the intensity of these bands was higher in the unexposed samples compared to those that were exposed to elevated temperatures. The discrepancy of result was due to the different stages of experimental work done. Therefore, the comparison of the bands between unexposed and exposed geopolymer samples was not discussed.
When subjected to 200 °C (Figure 11), increasing silica fume content up to 4%, the intensity of the band at ~980 cm −1 decreased with increasing silica fume content, which was also consistent with the recorded compressive strength, as shown in Figure 5. However, at 1000 °C, the geopolymer with 2% silica fume has the highest intensity of this band due to increased Si-O-Al bonds which were further verified by the phase identification ( Figure 8) as albite (NaAlSi3O8) was formed at 1000 °C in addition to the sodium aluminium silicate hydrate (N-A-S-H). Figure 11. IR spectra of fly ash geopolymers with varying silica fume content addition after being exposed to 200 °C and 1000 °C.
In addition, as referred to in Figure 11, the absorption bands due to OH stretching vibration (~3300-33600 cm −1 ) and OH bending vibration (~1630 cm −1 ) reduced in intensity at high temperature exposure. The indicated the dehydration of structure with increasing temperature. The intensity of C-O bonds at ~2340 cm −1 and ~1470 cm −1 also reduced due to the breakage of C-O at elevated temperatures. Based on Figure 10 and Figure 11, there are differences in the intensity of bands at ~3300 cm −1 and ~1630 cm −1 of the unexposed and exposed geopolymer samples. It was supposed that the intensity of these bands was higher in the unexposed samples compared to those that were exposed to elevated temperatures. The discrepancy of result was due to the different stages of experimental work done. Therefore, the comparison of the bands between unexposed and exposed geopolymer samples was not discussed. Figure 12 illustrates the SEM micrographs of unexposed fly ash geopolymers with varying silica fume contents under magnification of 1000×. In general, some remnant fly ash particles indicated by the spherical particles with smooth surface, pores and cracks Figure 11. IR spectra of fly ash geopolymers with varying silica fume content addition after being exposed to 200 • C and 1000 • C. Figure 12 illustrates the SEM micrographs of unexposed fly ash geopolymers with varying silica fume contents under magnification of 1000×. In general, some remnant fly ash particles indicated by the spherical particles with smooth surface, pores and cracks were observed in the matrix of all geopolymers. The reference geopolymer revealed a relatively loose matrix and some microcracks (Figure 12a). With the addition of 2% silica fume (Figure 12b), the microstructure became denser and homogeneous with less unreacted fly ash. Yet, the microstructure of the geopolymer with 4% of silica fume was comprised of a loose microstructure with a large amount of unreacted fly ash and wider cracks (Figure 12c). Figure 13 illustrates the SEM micrographs of exposed fly ash geopolymers (magnification 500×). Exposure to high temperature led to transformation in the microstructure of geopolymers and affected its mechanical strength. Based on Figure 13a, the geopolymer matrix was dense, with fewer cracks, remnant fly ash particles and pores after being exposed to 200 • C. Therefore, the dense microstructure contributed to the improvement of compressive strength of the geopolymer at 200 • C ( Figure 5). However, with increasing temperature up to 1000 • C, solidifying melt occurred and intervening matrix was observed. At the same time, large pores were observed in the exposed geopolymer to 1000 • C. This might be the reason for the degradation in the compressive strength at high temperature. The observation was supported by Škvára et al. [41] when subjected to fly ash geopolymer to high temperatures. Referring to Figure 13, the geopolymer with 4% silica fume was comprised of larger pores and hence recorded lower compressive strength than geopolymer with 2% silica fume.

Microstructural Analysis
were observed in the matrix of all geopolymers. The reference geopolymer revealed a relatively loose matrix and some microcracks (Figure 12a). With the addition of 2% silica fume (Figure 12b), the microstructure became denser and homogeneous with less unreacted fly ash. Yet, the microstructure of the geopolymer with 4% of silica fume was comprised of a loose microstructure with a large amount of unreacted fly ash and wider cracks (Figure 12c).  Figure 13 illustrates the SEM micrographs of exposed fly ash geopolymers (magnification 500×). Exposure to high temperature led to transformation in the microstructure of geopolymers and affected its mechanical strength. Based on Figure 13a, the geopolymer matrix was dense, with fewer cracks, remnant fly ash particles and pores after being exposed to 200 °C. Therefore, the dense microstructure contributed to the improvement of compressive strength of the geopolymer at 200 °C ( Figure 5). However, with increasing temperature up to 1000 °C, solidifying melt occurred and intervening matrix was observed. At the same time, large pores were observed in the exposed geopolymer to 1000 °C. This might be the reason for the degradation in the compressive strength at high temperature. The observation was supported by Škvára et al. [41] when subjected to fly ash geopolymer to high temperatures. Referring to Figure 13, the geopolymer with 4% silica fume was comprised of larger pores and hence recorded lower compressive strength than geopolymer with 2% silica fume. and (c) 4% of silica fume addition after exposed to 1000 °C.

Conclusions
This paper investigated the influence of silica fume on the thermal performance of fly ash geopolymers. The addition of silica fume (2%) did not boost up the compressive strength of fly ash geopolymer, while higher silica fume content of 4% degraded the compressive strength due to the impeded dissolution and geopolymerisation reaction. In general, all exposed geopolymers, regardless of the addition of silica fume, showed improvement in strength at 200 °C and then deterioration of strength with further increase in ex- Figure 13. SEM micrographs of fly ash geopolymers with (a) 2% of silica fume addition after exposed to 200 • C, (b) 2% and (c) 4% of silica fume addition after exposed to 1000 • C.

Conclusions
This paper investigated the influence of silica fume on the thermal performance of fly ash geopolymers. The addition of silica fume (2%) did not boost up the compressive strength of fly ash geopolymer, while higher silica fume content of 4% degraded the compressive strength due to the impeded dissolution and geopolymerisation reaction. In general, all exposed geopolymers, regardless of the addition of silica fume, showed improvement in strength at 200 • C and then deterioration of strength with further increase in exposed temperature. Even so, the addition of 2% silica fume showed no degradation of compressive strength at 1000 • C compared to that exposed to 600 • C. In addition, the extent of shrinkage and expansion of geopolymers with 2% silica fume was the lowest among all geopolymers, which showed the improvement of thermal stability of geopolymer due to effect of silica fume. Albite and magnetite peaks were formed in the geopolymer with silica fume at 1000 • C and the intervening matrix was produced due to partial melting at high temperature.