Enhancement of geopolymerisation reactivity and thermal resistance using mixed types of fly ash

In the 1970s, Professor Joseph Davidovits invented the term ‘geopolymer,’ an inorganic polymer with a three-dimensional (3D) network of inorganic molecules. It can be produced by a reaction between aluminosilicates and alkali activators at ambient or moderate curing temperatures (Davidovits, 2011; Lee et al., 2018; Provis and Deventer, 2009)

The most widely used aluminosilicate source is fly ash (FA) because of the large amount (∼900 million tons per year) produced globally (Malhotra, 2008). FA can be classified into class F (CaO < 18%) and class C (calcium oxide or lime (CaO) > 18%), according to ASTM-618 (Suraneni et al., 2021). Free Ca²⁺ plays a role in making hydrated calcium silicate (C–S–H) gel from the hydration reaction This reaction is well known to shorten the setting time (Samarakoon et al., 2019; Sobolev et al., 2017; Wattanasiriwech et al., 2017b). From previous works, initial and final setting times of class F FA-based geopolymer, were 4–8 times longer than class C FA-based geopolymers due to the smaller amount of free Ca²⁺ in class F FA (Nath et al., 2015; Nath and Sarker, 2014; Pangdaeng et al., 2014; Phoo-Ngernkham et al., 2014; Saha and Rajasekaran, 2017; Topark-Ngarm et al., 2015).

Although the high content of crystalline phases in class F FA is advantageous for fire retardation applications (Wattanasiriwech et al., 2017a; Xie and Kayali, 2016), the very long setting time of class F-based geopolymer is one of the barriers to its use. To overcome the problem, some techniques such as grinding to increase the surface area of the FA particle or adding Ca²⁺-bearing substances, for example, lime (CaO) or ground granulated blast furnace slag (GGBS) (Hardjito et al., 2008; Temuujin et al., 2009; Weil et al., 2007). These techniques are not well suited for all regions. For example, GGBS is available in only some countries (McNally and Sheils, 2012) and grinding requires high energy and thus increases the production cost.

The mix design for geopolymer preparation, such as the activator ratio of FA to alkali (S/L), is also vital in developing the properties, especially the compressive strength (Liew et al., 2012a). Geopolymer slurry prepared with a high S/L ratio had low workability and gave rise to problems in transporting and moulding. However, the low S/L ratio decreased the compressive strength of the final products due to excessive water contents, so the geopolymerisation process was sluggish (Yahya et al., 2015).

A new potential source of FA, from Hongsa thermal power plant (HS-FA) in Lao PDR, Thailand, has been explored to use as a precursor for geopolymer preparation in this research. This FA falls into class F, with quartz and mullite crystals. It has an extremely long setting time (initial 1941 min, final 3486 min) so has been problematic to use as a single precursor for geopolymer preparation. To overcome this problem, partial replacement of this HS-FA with class C FA from Mae Moh thermal power plant, Thailand (MM-FA) is proposed in this study. The very short setting times of the MM-FA should alleviate the sluggish setting of class F FA while maintaining the type of precursor material in the fine powder form. In this scenario, additional treatment of the precursor material such as grinding can be removed.

The effect of the S/L ratio on the properties of geopolymer paste was evaluated. The physical, mechanical, thermal and fire resistance properties of geopolymer were examined.

Two sources of FA – HS-FA and MM-FA – were used in this study. Owing to its low reactivity, only a small amount of HS-FA is used in cementation applications. Scanning electron microscopy (SEM) micrographs for the FAs used in this study are shown in Figure 1. The HS-FA was non-spherical with cavities and a rough surface. MM-FA showed a larger particle size distribution, but with a smoother surface. The chemical composition and surface area of the FA are indicated in Table 1. The chemical composition was analysed using an X-ray fluorescence technique (XRF; Horiba, MESA-500W). At the same time, the surface area of the FA powder was measured using the Brunauer–Emmett–Teller (BET) method (NOVA-e, Quantachrome). According to ASTM C 618-22 (ASTM, 2022), HS-FA falls into class F and MM-FA falls into class C. The X-ray diffraction (XRD) analysis and morphology of FAs have been reported elsewhere (Nurgesang et al., 2016; Phavongkham et al., 2021; Wattanasiriwech et al., 2021). HS-FA comprises quartz (SiO₂), mullite (3Al₂O₃·2SiO₂) and chabazite (NaAlSi₂O₆·3H₂O), while MM-FA comprises quartz and mullite, as well as some minor phases of lime, anhydrite (CaSO₄), quartz and magnesioferrite (Fe₂MgO₄) as the crystalline phases. To determine the amorphous contents in these FAs, an internal standard method was used by adding 10% calcium fluoride (CaF₂) to facilitate quantitative analysis of the amorphous phase (Rickard et al., 2011). The amorphous content of HS-FA was found to be 60% ± 5, while that of the MM-FA was 45% ± 7. The morphology of HS-FA particles was sub-rounded shape (round to angular) with high surface roughness. In contrast, MM-FA was generally spherical and smooth on the surface (Banchong et al., 2020).

A mixture of sodium hydroxide (8 M NaOH) and sodium silicate (Na₂SiO₃) solutions at the ratio of 0.6 was used as an alkali activator. AR grade sodium hydroxide pellet was supplied by QReC, New Zealand (99% purity), while the sodium silicate solution with chemical composition of 16.50% sodium oxide (Na₂O), 35.25% silica (SiO₂) and 48.25% water (H₂O) was acquired from C. Thai Chemicals Co., Ltd.

The experiment is principally divided into two parts – the first part is the exploration of suitable HS-FA/MM-FA and the second part is the experiment to find an optimum S/L ratio.

The HS-FA was replaced with MM-FA from 10 to 60% with 10% intervals. The FA-to-alkali activator ratio (S/L) was kept at 1.25. The mixing process for geopolymer synthesis started with mixing HS-FA and MM-FA for 2 min. The alkali activator was added to the FA and thoroughly mixed for 3 min. The mixture was then moulded in cylindrical polyvinyl chloride (PVC) moulds with dimensions 15 mm dia. × 30 mm high and cured in an electric oven at 90°C for 24 h in a water-saturated atmosphere to avoid rapid drying. The specimens were further cured at 40°C for 72 h in the same atmosphere, followed by curing at 40°C for another 72 h in dry air. This curing condition is adapted from the work done by Nurgesang and co-workers (Wattanasiriwech et al., 2017a, 2017b).

The study on the strength development of geopolymer prepared with mixed HS-FA and MM-FA (HS/MM-FA) at different S/L ratios was further investigated. A ratio of HS-FA: MM-FA of 50 : 50 was used. The details of the mixture design and calculated total water content in the mix are provided in Table 2.

The setting time was measured, according to ASTM C 191, using a Vicat needle apparatus (ASTM, 2008). The density and porosity were measured following ASTM C 642 (ASTM, 2001). Compressive strength testing was carried out according to ASTM C 39 using a universal testing machine (UTM; Instron 5566) at the loading speed of 1 mm/min (ASTM, 2011). The microstructure of the polished surfaces was observed under SEM (LEO 1450 VP) at an accelerating voltage of 20 kV, while the phase development was examined using an XRD method with CuKα radiation at 15 kV, 20 mA. The weight change of the geopolymer was measured using a thermogravimetric analysis (TGA) instrument (Mettler-Toledo 851e, Switzerland). The geopolymers were heated in an alumina crucible in a nitrogen atmosphere from room temperature up to 1000°C at a heating rate of 20°C/min.

The fire resistance test was assessed in a furnace preheated to 1000°C. The test samples were 40 × 50 mm in size and 10 mm thick. The test was started by exposing one side of the panel to the heated furnace, and the temperature of the reverse side was measured every 2 min for 180 min. The experimental set-up and equipment are presented in Figure 2.

The geopolymerisation behaviour of the various samples was observed at different setting times, as shown in Figure 3. The final setting time of the geopolymers, with HS-FA (0%MM-FA)-based geopolymer as the reference, reduced significantly, in an exponential manner, with an increasing amount of MM-FA, while the initial setting time linearly decreased.

Nath et al. (2015) found that the enhanced reaction of calcium compounds in the mixture is effective to accelerate the setting time of the geopolymer. Geopolymer prepared with class F FA alone required more than 24 h to set. When GGBS and ordinary Portland cement (OPC) were mixed, both initial and final setting times of geopolymer pastes decreased significantly to only a few hours (Nath et al., 2015). Chindaprasirt and co-workers (Chindaprasirt and Phoo-Ngernkham, 2018) reported that the addition of calcium-rich compounds such as lime and OPC could shorten the setting time. The hydration reaction was accelerated by the presence of Ca²⁺ in MM-FA (Chindaprasirt et al., 2012). Therefore, it could be claimed that MM-FA provided dissolved species which increased its solidification rate in the second stage of polymerisation, so the initial setting time was speeded up. At the same time, the rate of setting increased significantly as the percentage of MM-FA was increased in the third stage of geopolymerisation, which led to rapid hardening and speeded up the final setting time. The results indicated that MM-FA mixed with HS-FA resulted in a faster setting of the geopolymer paste as proposed.

Figure 4 shows the SEM images of geopolymers prepared with various percentages of MM-FA. It was observed that a denser matrix with lower porosity was obtained when MM-FA was incorporated into HS-FA. This indicates the faster reaction and better reactivity of the MM-FA. Some microcracks were observed in geopolymers prepared with 40–50% MM-FA, as shown in Figures 3(b), 4(a) and 4(c). A higher number of cracks with larger sizes were observed in geopolymer prepared with 60% MM-FA (Figure 4(d)) while the matrix was full of loosely packed granules of different sizes. Cracking in geopolymers occurs for many reasons, but mostly due to the removal of different types of water molecules. Water demand when making geopolymer paste depends on the surface area of the precursor materials. In this study, the MM-FA has a lower surface area than the HS-FA, so the water demand is lower.

The existence of unreacted granules of FA in the geopolymer was believed to be due to setting times that were too fast, as indicated in Figure 3.

The relict of unreacted FA granules, with cracking, in the 60% MM-FA was observed in all samples. However, the number of FA relicts tended to increase, whereas large air voids tended to decrease when a higher amount of MM-FA was used. The current authors’ previous work showed that HS-FA generated a massive amount of gas during curing, resulting in high surface porosity (Phavongkham et al., 2021). As the MM-FA content was increased, the setting time was progressively shortened. This resulted in the incomplete dissolution of FA, so the particle relicts could be observed.

Figure 5 presents the Fourier transform infrared spectroscopy (FTIR) spectra of FA geopolymers prepared with 0, 40, 50 and 60% MM-FA. The FTIR spectra of all geopolymers were slightly different from that of the original FA. HS-FA and MM-FA showed a band in the region of 455–467 cm⁻¹ and 1107–1120 cm⁻¹, assigned to T–O–Si bending and stretching (where ‘T’ denotes either Si or Al), respectively. The T–O–Si stretching vibration was more prominent than the bending mode; it was therefore logically used to indicate the degree of geopolymerisation. The T–O–Si bending had shifted to a lower frequency at around 992–996 cm⁻¹ in the geopolymers, implying that FA had reacted with an alkali activator. New reaction products were therefore formed. However, a shoulder at around 1010–1060 cm⁻¹ appeared in the 60% MM-FA geopolymer, indicating unreacted FA.

The broad spectra in the regions of 1645–1679 cm⁻¹ and 3402–3589 cm⁻¹ corresponded to the stretching and bending vibrations of O–H. A band in the region of 1408–1451 cm⁻¹, assigned to the stretching vibration of the C–O of the carbonate group, appeared, suggesting the presence of calcite (CaCO₃) as a result of the reaction between excess calcium oxide and atmospheric carbon dioxide (Khater, 2012).

The compressive strength of geopolymers prepared with different percentages of MM-FA at S/L of 1.25 is shown in Figure 6. It can be seen that the compressive strength decreased linearly with the increasing contents of MM-FA. Quartz and mullite crystalline phases are not highly reactive in geopolymerisation reactions, but they are strong phases that improve compressive strength (Xie and Kayali, 2016). There could be several reasons for the lower strength.

The reduction in compressive strength with increasing percentage of MM-FA was believed to be due to the reduced amount of HS-FA, which comprises a high amount of crystalline quartz and mullite phases. Besides, an increasing percentage of MM-FA let to the fast setting, resulting in a weak initial matrix framework. The very fast setting time of geopolymer hinders the geopolymerisation reaction and leads to the subsequent slow strength development (Chindaprasirt and Phoo-Ngernkham, 2018). This finding has been noted in previous work (Wardhono, 2018). Thus, the hardening of the geopolymer may occur rapidly, without completely dissolving the FA, which then becomes trapped inside the matrix. As evident from SEM micrographs (Figure 4), unreacted particles of FA remained. This was confirmed by the FTIR analysis in Figure 5, which shows a lower degree of geopolymerisation when MM-FA was used.

In addition, HS-FA was non-spherical with cavities and a rough surface, thus requiring high liquid contents in processing (Banchong et al., 2020). Porous and less reactive particles containing some unburned matter increase the liquid demand necessary to prepare a workable mixture (Fernández-Jiménez and Palomo, 2003). The excess liquid thus occurred in a mixed design with a high amount of MM-FA due to the lower water demand. The lower surface area of MM-FA requires a smaller amount of liquid for wetting and dissolution than HS-FA. This experiment kept a constant S/L = 1.25 by weight, resulting in excess liquid in the system. Water played a role as a carrier for the transportation of the ions in the geopolymer systems (Liew et al., 2012b). However, the excess water could interrupt the gel structure and possibly slow down the geopolymerisation reaction, so a strength reduction with high water content was obtained.

The TGA analysis result for geopolymer, obtained from geopolymers prepared without MM-FA and with 50% MM-FA, is presented in Figure 7. The result shows that the weight loss of geopolymer prepared without MM-FA occurred at a temperature range of 25–150°C, 150–300°C and 600–800°C. It was observed that there was no further loss until 1000°C, where the remaining mass after heating was around 88%. For geopolymer prepared with 50% MM-FA, a similar trend was also seen. The main weight loss occurred at a temperature range of 25–300°C and around 88% remained after heating at 1000°C. The rapid decline in the weight of both geopolymers before 150°C and around 200°C was attributed to the fast evaporation of free water and bond water, respectively. The dehydroxylation reaction was also observed at around 600 and 700°C.

Figure 8 shows the appearance of cracked surfaces of the geopolymers prepared without MM-FA and with 50% MM-FA, respectively, after the fire resistance test. Catastrophic cracks on the surface of 50% MM-FA were observed after thermal exposure, while smaller crack sizes appeared on the surface of 0% MM-FA. The shrinkage and cracking resulting from the heating at high temperatures were caused by dehydration, dehydroxylation, phase transformation of the crystalline phase and viscous sintering (Zulkifly et al., 2017). However, with the higher porosity and greater strength in the 0% MM-FA, the samples showed a better ability to resist abrupt changes in temperature.

The fire resistance test involves very rapid heating of one side of a sample and measurement of the temperature on the reverse side. The reverse-side temperature of geopolymers prepared without MM-FA and with 50% MM-FA is presented in Figure 9. This shows that the temperature after the 180 min test of geopolymer prepared with 50% MM-FA was higher when compared to geopolymer prepared without MM-FA. The temperature at the reverse side of the geopolymer prepared without MM-FA slowly increased at the first step from around 400 to 490°C in 40 min. Afterwards, the temperature gradually decreased to around 440°C and remained constant until the end of the test. Geopolymer prepared with 50% MM-FA showed a temperature that gradually increased from around 460 to 570°C in 45 min and then remained constant at this temperature. Increase in the temperature at the first step for both geopolymers prepared without MM-FA and with 50% MM-FA was due to the evaporation of water. Geopolymers contain water in their framework, which can be confirmed by TGA/differential thermogravimetric analysis (DTG) curves, which showed a weight loss of free water and partial chemical bond water at around 150°C, as revealed in Figure 7. The evaporation of water during the fire resistance test is endothermic, so the input energy is used for this process. The temperature at the reverse side was thus slowly increased in this step.

The higher temperature on the reversed side of the geopolymer prepared with 50% MM-FA compared to the geopolymer prepared without MM-FA could be related to weight loss during heating, as identified by TGA analysis in Figure 7. However, the geopolymer prepared with 50% MM-FA showed a larger size of crack at the hot surface than geopolymer prepared without MM-FA, as presented in Figure 8. Cracks allowed the heat from the furnace to the reverse side by convection through the cracks rather than by conduction through the structure (Rickard et al., 2013). The higher temperature at the reverse side of the geopolymer prepared with 50% MM-FA could be because of the existence of cracks and, thus, a greater heat transfer.

In this section, the 50% MM-FA was selected. The stated ratio showed setting times close to those of the OPC standard (45 min for the initial setting time and 375 min for the final setting time).

Figure 10 presents the initial and final setting times of the geopolymer as a function of the S/L ratio at 50% MM-FA. It can be seen that both the initial and final setting times were significantly reduced as the S/L ratio was increased. The difference between the initial and final setting times was also reduced with the increase in the S/L ratio. The longest initial and final setting times of 735 and 1215 min were observed in the geopolymer prepared with a ratio S/L of 1.25. The shortest initial and final setting times of 136 and 240 min were observed in the geopolymer prepared with an S/L ratio of 2.00.

The mechanism of geopolymerisation occurs in three steps: (a) step 1 – breaking down the aluminosilicate bonds of the precursor material and forming monomers; (b) step 2 – rearrangement of the monomers to form oligomers through condensation by removing water molecules; (c) step 3 – polycondensation of oligomers to form a 3D framework of aluminosilicate (Pundiene et al., 2020; Zhang et al., 2018).

Water acts as the medium for the dissolution of aluminosilicate material in step 1. Water serves as a carrier for the aluminate and silicate monomers that formed oligomers and polymers in step 2 and step 3, respectively. However, water was removed after forming oligomers and polymer networks, as indicated in step 2. The initial setting time occurred in step 3, the high water content presented in the system could hinder the polycondensation of the oligomers (Liew et al., 2012b). In this study, the water content was continually decreasing, as increasing the ratio (Table 2) resulted in faster geopolymerisation. This is similar to Siyal et al. (2016), who reported that increasing the ratio of S/L in mixes could reduce water content and fasten the geopolymerisation reaction by decreasing the initial setting time. In this study, it was shown that increasing the S/L ratio could shorten the curing time when HS-FA was used in a combination with MM-FA as an aluminosilicate precursor for geopolymerisation.

In comparison, 60% MM-FA geopolymer prepared with S/L = 1.25 and 2.00, respectively, contained the highest and lowest water content in the mix (see Table 2). They were thus chosen for the TGA/DTG analysis. Figure 11 shows that the weight loss of geopolymer prepared with S/L = 1.25 occurred at temperature ranges of 25–130°C, 150–300°C and 550–700°C, and a small weight loss occurred at temperatures higher than 700–1000°C. At the same time, a similar trend was noticed in the geopolymer prepared with S/L = 2.00. The remaining weights of 88.6 and 89.9% were found in geopolymer prepared with S/L = 1.25 and 2.00, respectively. The remaining weight of the geopolymer prepared with S/L = 2.00 was slightly higher, indicating water evaporation was lower than the geopolymer prepared with S/L = 1.25. There are three stages of water in a geopolymer structure: (a) physically bonded water (free water), which is present on the surface of the geopolymer; (b) chemically bound water (zeolitic water), which is present at the interlayer of the geopolymer network; and (c) the hydroxyl group (OH⁻), which is present at the surfaces and edges of each of the geopolymeric micelles (Davidovits, 2015).

The DTG curve of geopolymer prepared with S/L = 1.25 explained that the weight loss at 100°C can be attributed to the fast removal of physically bound water (free water). The weight loss before 200 and 600°C is attributed to removal of chemically bound water and the hydroxyl group, respectively. Meanwhile, a similar trend was noticed in the geopolymer prepared with S/L = 2.00; however, the dehydroxylation was observed at 580°C with a smaller degree of weight loss.

The bulk density and porosity of geopolymers with different S/L ratios are shown in Figure 12. When the S/L ratio increased, the bulk density increased, while the porosity decreased. The increasing ratio of S/L in geopolymer systems reduces water content, Si species content, OH⁻ and Na⁺ ions within the alkali activator (Liew et al., 2012b; Ng et al., 2018). The increase in bulk density with increasing the S/L ratio in this study was due to the lower water loss. This was confirmed by TGA/DTG analysis in Figure 11, where around an 11.4% weight loss of free water in geopolymer prepared with S/L = 1.25 was higher than the 10.9% weight loss of free water in geopolymer prepared with S/L = 2.00. Besides, the increase in bulk density was confirmed by reducing porosity when increasing the S/L ratio. A porosity of 33.37% was obtained in geopolymers prepared with S/L = 1.25 and this gradually reduced to 32.71, 31.37 and 31.03% in geopolymers prepared with S/L = 1.43, 1.66 and 2.00, respectively.

The compressive strength of geopolymer prepared with different S/L ratios is shown in Figure 13. Increasing the S/L from 1.25 to 2.00 led to improved compressive strength from 11.79 to 30.54 MPa. Increasing the S/L ratio led to a decrease in concentrations of Na⁺ and OH⁻ ions, Si species and water content in the mix design. These factors have a significant effect on the geopolymerisation reaction and the compressive strength. Liquid alkali (NaOH + Na₂SiO₃) contained water. Sodium silicate provided Na⁺ ions and additional Si species, which supported the geopolymerisation rate (Criado et al., 2008). Sodium hydroxide served in the dissolution of aluminosilicate precursor and Na⁺ ions presented in cavities to balance the negative charge of Al³⁺ tetrahedra in the geopolymer networks.

Generally, the amount of OH⁻ and Na⁺ ions must be sufficiently high to induce the complete dissolution of the aluminosilicate source and balance the negative charges, so an ideal M₂O/Al₂O₃ ratio (where M is Na or K) should be around 1.00 (Kovalchuk et al., 2007). However, a high M₂O/Al₂O₃ ratio would have excess Na⁺ ions resulting in low compressive strength (Criado et al., 2008). Increasing the S/L ratio led to a decrease in water and Na⁺ contents and thus speeded up the polymerisation process. This was evident by the reduction of the setting times and porosity, as shown in Figure 10 and 12. Therefore, geopolymer prepared with low water content in the mix design possibly leads to a speedy geopolymerisation reaction and improved strength.

As a result, it was clear that, by increasing the ratio of S/L, a higher compressive strength was obtained. In addition, it was found that the compressive strength of geopolymer prepared with a mixture of HS-FA and MM-FA as a precursor material led to improved compressive strength as compared to geopolymer that used only HS-FA or MM-FA. A compressive strength of 26.54 MPa was obtained when geopolymers were prepared using HS-FA alone. In contrast, geopolymer prepared with only MM-FA was reported to have a compressive strength of 12 MPa (Nurgesang, 2016) due to its flash setting.

The reactivity of FA in geopolymerisation was typically influenced by the contribution of the crystalline phase to the overall composition, the particle size and geometry (angularity) of the particles (Klima et al., 2022). HS-FA-based geopolymer has extremely slow curing at ambient temperature. Mixing MM-FA, classified as class C, with HS-FA led to an enhanced geopolymerisation reactivity due to the hydration of calcium ions forming C–S–H (Wattanasiriwech et al., 2017a). By mixing the two types of FA, with the optimisation of the processing parameters, an enhancement of mechanical properties could be obtained.

The hot side of 50% MM-FA geopolymer prepared with S/L = 1.25 and 2.00 after the fire resistance test is presented in Figure 14. It can be seen that a catastrophic crack occurred in the geopolymer prepared with S/L = 1.25. In contrast, after thermal exposure, the geopolymer prepared with S/L = 2.00 showed cracks of a much smaller size, and only in the hot–cold region. However, neither structural disintegration nor burning flame and smoke were observed in either of the samples. The catastrophic crack on the surface of the geopolymer with S/L = 1.25 was due to thermal shock, which was a function of fracture strength, thermal conductivity, thermal expansion and the Young's modulus (Zulkifly et al., 2017). With only a slight difference in the porosity of the two geopolymer samples, the Young's modulus, thermal conductivity and thermal expansion of the samples could be in the same value range. The thermal shock resistance (TSR) for this case was thus dependent mainly on the fracture strength of the materials. The geopolymer with S/L = 2, therefore, showed a better TSR due to its superior fracture strength.

Two geopolymers were chosen for the fire resistance test, as presented in Figure 15. The result revealed that the temperature at the reverse side after the test lasting 180 min for a geopolymer prepared with S/L = 1.25 was slightly lower than that prepared with S/L = 2.00. The reverse-side temperature of both samples was slowly increased in the first 30 min from room temperature to around 570–580°C and remained constant until the end of the experiment. The thermal stability of FA-based geopolymer was determined by the Si/Al ratio in the amorphous part. A study on class F FA-based geopolymers showed that the higher the Si : Al molar ratio is, the better is the strength retention (or increase) after exposure to high temperature (Rickard et al., 2011). The mix design of the geopolymer paste was found to have vital effects on its thermal stability. Geopolymers with high added Si content had lower thermal stability and greater loss of strength due to swelling of the silicate phase at high temperatures in the sodium silicate-activated samples (Klima et al., 2022; Phavongkham et al., 2021). However, another study showed that class F FA-based geopolymer pastes had poorer thermal performance and lower retention ratio of strength than class C FA-based geopolymer pastes after being exposed to above 800°C (Jiang et al., 2020).

The reactivity of a geopolymer can be adjusted using mixed types of FA. Using 50 : 50 HS-FA to MM-FA at the proper solid-to-liquid ratio, initial and final setting times of 136 and 240 min were obtained. The fire resistance test at 1000°C showed maximum temperatures on the reverse side of ∼580°C. The materials showed small surface cracks on the hot side without structural disintegration, suggesting potential use for fire retardation panels. The compressive strength of the geopolymer was at a maximum at 30.54 MPa.

The authors gratefully acknowledge the financial support of the National Science Research and Innovation Fund (NSRF) through Mae Fah Luang University.