The primary raw material used in this study is a highly amorphous granulated blast furnace slag (GBFS) from the Colombian factory Acerías Paz del Río, with a specific gravity of 2900 kg/m³ and a Blaine fineness of 399 m²/kg. The particle size range determined through laser diffraction is 0.1–74 µm, with a d₅₀ of 15 µm. The chemical composition of the GBFS is shown in Table 1.

Alkali activation of this precursor was carried out using three different activators:

Alkali-activated slag pastes were formulated with an activator dose of 5 g Na₂O per 100 g slag, and a water to binder (free water + water in the activator)/(anhydrous activator + slag) ratio of 0.23. The paste was cast in cylindrical molds of 30 mm diameter ×60 mm height, sealed, and cured under ambient conditions (temperature 25 °C and relative humidity 75%) for 24 h. Subsequently, the cylinders were demolded and cured in a humidity chamber at a temperature of 27±2 °C and relative humidity (RH) >90%, for 28 days.

For analysis of thermal performance, 28-day cured specimens were then held for 24 h at 50 °C to remove the excess superficial water, and heated in a furnace at temperatures of 200, 400, 600, 800 and 1000 °C, with a heating rate of 1 °C/min and a hold time of 2 h, and then cooled inside the furnace to 25 °C to avoid thermal shock and potential cracking of the samples. After thermal exposure, compressive strength was determined using a universal testing instrument (ELE International) at a displacement rate of 1 mm/min. Changes in the mass of the specimens were also determined.

Structural characterization of the heated pastes was conducted through X-ray diffraction using a RX Rigaku R-INT 2200 diffractometer. Fourier transform infrared (FTIR) spectroscopy was conducted via the KBr pellet technique, using a Shimadzu FTIR 8400 instrument, scanning from 2000 to 400 cm⁻¹, and thermogravimetry was carried out in a SDT-Q600 instrument in a nitrogen atmosphere, from 25 to 1100 °C at a heating rate of 10 °C/min.

Compressive strengths higher than 40 MPa are identified (Figure 1) for all the pastes after 7 days of curing. At both times of curing assessed, 7 and 28 days, increased mechanical strengths are identified when using the alternative activators based on SF and RHA compared with the strengths developed when using a commercial silicate solution. These differences are more significant at 28 days, where slags activated with a SF and RHA-based activators developed compressive strengths above 100 MPa. In a previous study activating metakaolin/slag blends with similar alternative activators (28), it was suggested that when using SFA and RHAA it is likely that at early age the mechanism of activation is dominated by the different rates of release of silicate in these systems, which influences the structural characteristics of the binders and therefore their mechanical performance. The results of this study are also consistent with the higher mechanical strengths observed in fly ash geopolymer systems (26) and alkali-activated slag/fly ash binders (27) when using alternative activators based on SF and RHA, respectively.

Compressive strength values of 28-day cured pastes exposed to temperatures between 200 °C and 800 °C are shown in Figure 2. Exposure to 200 °C seems to promote a slight increase in the compressive strength values when compared with the non-exposed specimens (Figure 1) of all the pastes, independently of the type of silicate activator used. Conversely, exposure at 400 °C led to a significant reduction of the mechanical strength (up to 50% lower when compared with non-exposed samples), especially of slag pastes activated with SF and RHA-based activators. However, it is important to note that the pastes still retain compressive strengths of up to 65 MPa and 52 MPa, when activated with SF and RHA-based activators respectively, after exposure to this temperature.

These values are substantially higher than those obtained in alkali-activated metakaolin/slag blended pastes (13), where exposure to 400 °C reduced the compressive strength by 87% compared with non-exposed samples. In this case, the reduction in the compressive strength is consistent with the temperature of dehydration of C-S-H type gels (4, 13, 22), which is the main reaction product responsible for the mechanical performance of alkali-activated slag binders. As the water in this phase is more tightly bonded to the gel structure than in geopolymer binders forming in the metakaolin/slag activated samples (13), higher temperatures of exposure are required to promote the complete dehydration of the C-S-H type gel, and the consequent destruction of the binding matrix. The alkali-activated slag samples exposed to 600 °C retain mechanical strengths slightly lower than specimens exposed to 400 °C, but still around 50 MPa.

The compressive strengths of samples activated with WG and SFA activators exposed to 800 °C and 1000 °C could not be determined, as the samples were easily crumbled by hand. This is consistent with the high volumetric contraction that can take place between 600 °C and 1000 °C in alkali-activated pastes (13) inducing the severe macrocracking of the specimens. Pastes activated with RHA-based activators showed macrocracks after 600 °C and dimensional changes at 800 °C and 1000 °C; however, the samples exposed at 800 °C do not exhibit the same level of decay (Figure 3) observed in specimens activated with WG and SFA, and report a residual strength of 5 MPa. This indicates that using the RHAA activator seems to enhance the strength retention upon heating of alkali-activated slag binders.

The X-ray diffractograms of the silicate-activated slag binders after 28 days of curing and exposed to 800 °C and 1000 °C are shown in Figure 4. In all the activated samples assessed, the formation of the crystalline phase gismondine (CaAl₂Si₂O₈·4H₂O, PDF # 020-0452), and a partially ordered calcium silicate hydrate (C-S-H) type phase with a riversideite 9Å type structure (Ca₅Si₆O₁₇·H₂O, PDF # 029-0329, but probably with some Al substitution), is identified. This indicates that the sodium silicates produced with SF and RHA do not modify the phase assemblage of alkali-activated slag binders compared to the WG activator, as the reaction products forming in these systems are consistent with what has been observed in previous studies using a slag from the same source (8, 31). Traces of gehlenite (Ca₂Al₂SiO₇, PDF # 035-0755), calcite (CaCO₃, PDF # 01-083-0577) and quartz (SiO₂, PDF # 046-1045) are also observed. These compounds have been identified in the unreacted slag (8), and calcite and quartz do not seem to participate in the activation reaction.

Upon exposure to 800 °C, a significant reduction in the intensity of the main reflection peak assigned to the C-S-H type phase (29.4° 2θ) is observed in all the activated pastes, along with a significant increase in the intensity of the peaks assigned to gehlenite, which is likely to result from partial devitrification of the remnant slag grains in the specimens. The crystalline phase wollastonite-2M (CaSiO₃, PDF# 027-0088), a dehydration product of C-S-H type gels, and traces of what seems to be a dehydrated form of the zeolite levyne, are also observed in all the samples. It is important to note that the intensities of the reflections assigned to wollastonite are significantly higher in specimens activated with RHA-based activator compared with binders produced with WG and SFA activators, indicating that the fraction of C-S-H forming in RHA-activated slag binders might be much higher in this binders compared with pastes produced with the other activators used.

In alkali silicate-activated slag binders, Puertas (11) identified the formation of åkermanite, gehlenite and nepheline after exposure to 1200 °C, while Rovnaník et al. (22) identified formation of åkermanite, diopside and wollastonite at temperatures of exposure >800 °C. On the other hand, Rashad et al. (16) observed the formation of åkermanite, merwinite and gehlenite in sulfate-activated slag samples after exposure to 800 °C. The slag used here has a very low Mg content compared to the materials used in those studies, and so gehlenite forms in preference to these other Mg-rich phases during its partial devitrification.

The total mass loss values of AAS-WG, AAS-SFA and AAS-RHA (Figure 5) samples upon heating to 1100 °C were all 22±2%, indicating that the commercial and alternative activators are promoting the formation of similar amounts of reaction products upon the activation of the slag. The thermograms of pastes show that the mass loss occurs in two stages: the first one between 37 and 600 °C, where the major loss (18%) occurs between 37 °C and 260 °C, associated with the evaporation of the free water, and the loss of water molecules leading to dehydration of the C-S-H type gel present in the alkali-activated slag binders. The second stage is identified around 600 to 800 °C attributed to the decomposition of carbonates (4). In this temperature range, it is also likely that there is some crystallisation of the poorly ordered component present in these binders, most likely leading to wollastonite formation, which occurs at ~780 °C (32), as this is the main crystalline phase identified via XRD (Figure 5) in these samples exposed above 800 °C.

After exposure of the binders to 800 °C, the mass loss below 300 °C is no longer identified in any of the silicate-activated slag pastes, consistent with the complete removal of the free water and the chemically/adsorbed water in the reaction products during thermal treatment. The only feature identifiable in the thermograms is a peak at 664 °C, consistent with the decomposition of the residual carbonate-rich compounds present in these samples, including those which result from some minor CO₂ uptake onto the products between temperature exposure and TG analysis. Similar results are observed for the specimens after exposure to 1000 °C, where minor traces of decomposition of carbonates are again identified.

It is also notable that in the carbonate region between 600 to 800 °C, the 50 °C samples show two peaks whereas the heated samples show only one; this aids in the identification of the lower-temperature peak at ~660 °C to the products of atmospheric CO₂ uptake by the binder and/or its pore solution, while the sharp higher-temperature peak (~760 °C) is related to the more highly crystalline calcite present in the slag due to its weathering prior to binder production, which is decarbonated during thermal exposure and does not rapidly re-form under atmospheric conditions.

Figure 6 shows the FTIR spectra of the different silicate-activated slag binders as a function of the temperature of exposure. In 28-day cured samples before high temperature exposure, a high intensity broad band assigned to the stretching vibration modes of H-OH groups (between 3200 and 3600 cm⁻¹) in the hydrated reaction products is identified. A small vibration band observed at ~1640 cm⁻¹ corresponds to the bending vibration modes of H-OH bonds, while the band between 1200–950 cm⁻¹ is associated with the asymmetric stretching vibration mode of Si-O-T bonds (T: tetrahedral Si or Al) (33, 34). The shoulder at 875 cm⁻¹ is assigned to the asymmetric stretching of AlO₄ groups present in the binders (35, 36). Traces of carbonates are also identified at 1480 cm⁻¹, assigned to the asymmetric stretching mode of the O-C-O bonds of CO₃²⁻ groups (37). The band at ~660 cm⁻¹ is assigned to Si-O-Si bending vibration modes, and vibrations between 400–500 cm⁻¹ correspond to the deformation of SiO₄ tetrahedra (34).

Upon exposure to high temperatures, the intensity of the broad band assigned to the stretching vibration modes of H-OH groups decreases (Figure 6). In the case of binders activated with the commercial silicate, this band is no longer observed after exposure to 200 °C. Conversely, in specimens activated with SFA and RHAA, this band is still identified in pastes exposed to temperatures as high as 400 °C and 600 °C respectively. This might suggest that the water in the reaction products forming when using the SF and RHA based activators is more tightly bonded than when using commercial sodium silicate, as higher temperatures of exposure are required for its complete removal from the paste. This could be a potential consequence of a denser microstructure of the binding phases forming when using the alternative activators produced.

In WG and SFA activated specimens exposed to temperatures between 200 °C and 800 °C the formation of a distinctive peak centred at 1384 cm⁻¹ is identified. It is well known that zeolites can separate CO₂ from gas mixtures and can be good candidates for CO₂ capture, especially at moderate temperatures (120 °C–400 °C) (38). Stevens et al. (39) assigned this peak to bidentate carbonate sites associated with chemisorbed CO₂ in zeolites, whose intensity increased with the temperature at which CO₂ sorption took place. Considering that the formation of gismondine has been identified in these binders (Figure 4), it is likely that some absorption of CO₂ is occurring during testing, and as the exposure to high temperatures leads to the decarbonation of the samples, the consequent reduction in the intensity of the broad band at 1420–1440 cm⁻¹ allows the identification of this phase. In specimens produced with RHA the peak at 1384 cm⁻¹ is not clearly identified, as this peak overlaps with the resonance of the O-C-O bonds whose intensity does not decrease upon high temperature exposure.

The spectra in Figure 6 do show differences in the main bands in the ranges of 600–400 cm⁻¹ and 1000–900 cm⁻¹ as a function of the activator used and the temperature of exposure. In the WG-activated slag paste (Figure 6A), this band is centred at 995 cm⁻¹ after 28 days of curing. Upon exposure to temperatures up to 600 °C this band shifts towards lower wavenumbers. This is likely associated with the depolymerization of the C-A-S-H product, promoting the formation of a less crosslinked reaction product upon heating. In the specimen exposed to 800 and 1000 °C the formation of bands at 1016, 960 and 900 cm⁻¹ is observed, assigned to the triply degenerate antisymmetric stretch of the SiO₄ present in wollastonite (40), as identified in the X-ray diffractogram (Figure 4). The bands observed at 723 and 446 cm⁻¹ are assigned to the CaO₆ sites present in wollastonite.

In the paste activated with the silica fume based activator (Figure 6B) the T-O-T band is observed at 992 cm⁻¹ after 28 days of curing. Upon heating the vibration mode shifted towards lower wavenumbers, similar to the commercial silicate activated slag binders. This is consistent with the fact that similar reaction products have been identified in both systems through X-ray diffraction (Figure 4) and thermogravimetry (Figure 5). In samples exposed to 800 and 1000 °C, vibrations bands assigned to the formation of wollastonite are also observed; however, these modes seem broader than in the commercial silicate activated paste, which suggests a lower degree of ordering in the structure of the wollastonite formed through decomposition of the C-S-H type gel in this formulation.

Conversely to the changes identified in the samples activated by WG and SFA, changes in the wavenumber of the T-O-T band are not identified when the RHAA specimens are heated to 200 °C. In samples exposed to 400 °C, a slight shift in the vibration mode of the T-O-T band towards higher wavenumbers (from 976 cm⁻¹ to 987 cm⁻¹) is observed, when compared with specimens heated up to 200 °C. No further changes in the position of this band are observed when the sample is exposed to 600 °C. This indicates that the exposure to high temperature of slag activated with RHAA induced less disruptions in the C-S-H formed in these binders than in the others tested, and it is most likely promoting the densification of this phase between 400 °C and 600 °C. This is consistent with the fact that the same compressive strength was retained in the RHAA samples exposed to these two temperatures (Figure 1). In samples exposed to 800 °C, however, similar results to those identified in samples activated by WG and SFA are observed, in agreement with the formation of wollastonite in all of the samples (Figure 4C).

These FTIR results elucidate that, even though other techniques seem to show that similar reaction products are forming in the activated slag pastes with all three activators, the exposure to high temperature induced different changes in the main binding phase forming in these materials. This demonstrates that there might be significant differences in the nano-porosity and chemical composition of the C-S-H type gel, depending on the silicate activator used, as consequence of the variable availability of Si at different times of reaction. This directly influences the degree of water release and the consequent shrinkage of this type of gel upon high temperature exposure. Therefore, further investigation in this area is required.