High-Temperature behavior of CaO-FeOx-Al2O3-SiO2-Rich alkali Activated Materials

: Alkali-activated materials (AAMs) provide an opportunity to up-cycle several residues into added-value materials. Although generally praised for their performance under thermal loads, the thermal behavior of AAMs is dictated by a multitude of factors and the performance of CaO-FeO x -rich systems may differ from geopolymers. Therefore, this work ascertains the high-temperature resistance of CaO-FeO x -Al 2 O 3 -SiO 2 -rich AAMs. Mortars were exposed to different heating rates ( ≤ 10 ◦ C/min) and temperatures ( ≤ 1100 ◦ C), and volume and mass loss, apparent density, compressive strength (CS), mineralogical composition, and morphology were evaluated. At low heating rates, the main effects noted were densiﬁcation and a gradual lightening of color as the temperature rose. CS underwent an abrupt decline at 750 ◦ C and recovered at higher temperatures, reaching a maximum value of 184 ± 13 MPa at 1100 ◦ C. With an increased heating rate to 10 ◦ C/min, the strength loss at 750 ◦ C persisted, but maximum CS was halved when ﬁring at 900 ◦ C. At 1100 ◦ C, a signiﬁcant reduction of CS was observed, but all samples maintained their integrity. Except for 1100 ◦ C at 10 ◦ C/min, all sintered-AAMs presented residual CS above 40 MPa. These results demonstrate that CaO-FeO x -Al 2 O 3 -SiO 2 -rich AAMs present interesting thermal behavior and can be potentially used to produce glass-ceramics or refractory materials from secondary resources.


Introduction
Increasing the use of alternative feedstock materials is crucial to reduce the current pressure on natural resources and to diminish the environmental impact of our industrial sectors. Moreover, using low-grade materials and residues as raw materials also represents an important and long identified economic opportunity for manufacturers. In the field of secondary resources, thermochemical conversion processes belong to a group of emerging technologies that aim to reclaim large volumes of low-value materials-such as industrial, municipal, and even mined solid wastes from historical landfills-and reintroduce them into materials' cycles in the form of added-value materials. The primary outputs of these processes are valuable energy carriers and metals, but an inorganic solid material is also generated. This inorganic fraction is molten and often vitrified by water quenching, afterwards being either disposed of or used in low-value applications (e.g., road paving). As thermochemical conversion processes continue to thrive, and their worldwide The chemical composition of the slag produced in thermochemical conversion processes is characterized by the presence of significant calcium and iron content, which, however, have a direct influence on the phase composition and morphology of the material after heating. Low calcium systems (i.e., geopolymers) are generally preferable for producing high-temperature resistance AAMs as calcium reduces the liquidus temperature of K 2 O-SiO 2 -Al 2 O 3 systems. Furthermore, the near absence of hydrated and chemically bonded water in those low-calcium structures [11,12] reduces the deleterious microstructural effects of dehydration, which results in more thermally stable AAMs. The presence of iron can, however, be deleterious to K 2 O-Al 2 O 3 -SiO 2 systems, depending on the amount of alumina present in the systems [10]. If insufficient alumina is available, the iron compounds formed are characterized by much lower melting temperatures relative to their iron-free analogs. Hence, the thermal performance of AAMs made from FeO x -Al 2 O 3 -SiO 2 precursors with low-aluminum content is expected to be significantly impaired relative to their iron-free equivalents. On the other hand, in alumina-rich K 2 O-Al 2 O 3 -SiO 2 systems, the introduction of iron is not expected to preclude the crystallization of highly stable and refractory phases such as mullite, leucite, and kalsilite, and can even promote the crystallization of refractory hercynite [10]. The presence of these phases enhances the thermal resistance of AAMs made from alumina-rich FeO x -Al 2 O 3 -SiO 2 precursors and are expected to deliver excellent fire-resistant properties [9,13].
Several authors have investigated the phase transitions and thermal characteristics of AAMs produced from precursors such as metakaolin and fly-ash, which can be mainly described as Al 2 O 3 -SiO 2 binary systems [14,15], and calcium or iron-rich ones, such as ground granulated blast furnace slag [15], ferronickel slag [9], and copper slags [7,16]. The thermal characteristics of more complex CaO-FeO x -Al 2 O 3 -SiO 2 quaternary systems have not yet been thoroughly investigated. As discussed, calcium-and iron-containing phases will play vital roles in determining the thermal performance of those AAMs. Thus, examining the properties of CaO-FeO x -Al 2 O 3 -SiO 2 -rich AAMs after thermal exposure (the sintered counterparts) is crucial to establish these AAMs as reliable temperature resistant materials, with a place in the market, and deliver a long-term, large-scale, upcycling solution for CaO-FeO x -Al 2 O 3 -SiO 2 -rich residues generated during thermochemical conversion processes.

Raw Material Characterization and Mortar Preparation
A synthetic granular slag was produced with a chemical composition that mimics the expected inert vitrified material produced during the thermochemical conversion of combustible components of municipal and mined solid wastes. The detailed description of its production process is provided by Machiels et al. [1,2]. X-ray fluorescence (Bruker AXS S8 Tiger spectrometer, Billerica, MA, USA) was used to determine the bulk chemical composition of the slag, which can be generally described as a CaO-FeO x -Al 2 O 3 -SiO 2 quaternary system. The complete chemical composition of the slag is given in Table 1. The Fe oxidation state was quantified as described by Close et al. [17] and the iron phases were found to be predominantly in the bivalent oxidation state (92%). Aiming to deliver sustainable materials, the use of ordinary Portland cement and natural aggregates was completely avoided in the mortars' mix design; the vitreous slag was used both as the main precursor and aggregate. For this purpose, the granular slag was subjected to distinct pre-processing procedures to achieve different particle size distributions according to the envisioned usage ( Figure 1). Particle size distribution was determined using a laser particle analyzer (Malvern Mastersizer 3000, Malvern Panalytical, Malvern, UK). Finely powdered slag with a specific surface area of 1120 m 2 /kg (determined by the Brunauer-Emmett-Teller method, BET) and density of 2.97 g/cm 3 was used as the binders' main precursor (94.5 wt.% of binders' solid load). Silica fume (SF, Elkem, >95% SiO 2 ) with a BET specific surface area of approximately 22,200 m 2 /kg was used as an admixture to increase the availability of Si species and support strength development. Increasing the Si/Al ratio was also expected to promote the formation of highly thermally stable phases [14] and produce compact microstructures with low porosity levels after thermal exposure [18], thus improving the high-temperature resistance and residual strength of the produced materials. A potassium-based activating solution was used following a mix design previously optimized. Theoretical predictions [10] and empirical experimentations [6] in the available literature confirm the superior thermal performance of potassium-based AAMs relative to their sodium-based counterparts. Potassium hydroxide (14 M)  was used both as the main precursor and aggregate. For this purpose, the granular slag was subjected to distinct pre-processing procedures to achieve different particle size distributions according to the envisioned usage ( Figure 1). Particle size distribution was determined using a laser particle analyzer (Malvern Mastersizer 3000, Malvern Panalytical, Malvern, UK). Finely powdered slag with a specific surface area of 1120 m 2 /kg (determined by the Brunauer-Emmett-Teller method, BET) and density of 2.97 g/cm 3 was used as the binders' main precursor (94.5 wt.% of binders' solid load). Silica fume (SF, Elkem, >95% SiO2) with a BET specific surface area of approximately 22,200 m 2 /kg was used as an admixture to increase the availability of Si species and support strength development.
Increasing the Si/Al ratio was also expected to promote the formation of highly thermally stable phases [14] and produce compact microstructures with low porosity levels after thermal exposure [18], thus improving the high-temperature resistance and residual strength of the produced materials. A potassium-based activating solution was used following a mix design previously optimized. Theoretical predictions [10] and empirical experimentations [6] in the available literature confirm the superior thermal performance of potassium-based AAMs relative to their sodium-based counterparts. Potassium hydroxide (14 M) and potassium silicate solutions (23.8 wt.% SiO2, 9.5 wt.% K2O, and 66.7 wt.% H2O) were prepared by dissolving potassium hydroxide beads and anhydrous potassium silicate in demineralized water. The water content in the activating solution was adjusted (0.68 SiO2/K2O molar ratio and 83.4 wt.% H2O) to achieve the pre-defined solid over liquid ratio (S/L) of 2.85 and molar ratios in the binder composition. Figure 1. Particle size distribution of as-produced granular slag and after different pre-processing procedures. Powdered slag refers to finely milled slag used as binders' main precursor while fraction A and B designate roughly milled slags used as small size aggregates.
Based on our previous laboratory experimentation, the mortar mixture was designed for a binder-to-aggregate mass ratio of 0.88. Fuller distribution was applied to optimize aggregate particle size distribution (PSD). The PSD of the two fractions of the slag used as aggregates is shown in Figure 1. The density of the aggregate fractions was determined to be approximately 3.0 ± 0.03 g/cm 3 , and their mass ratio in the mortars' design was set to be 1.4:1. The characteristics of the reference mortar mix design are provided in Table 2.
The mortars were produced following the procedure described by Ascensão et al. [19]. In short, it involved: (a) manually mixing the powdered slag and silica fume in a plastic container; (b) mixing the binders' solid components (PS+SF) and the alkaline solution at 140 rpm for 600 s in a mortar mixer (Testing Bluhm & Beuerherdt Gmbh, Berlin, Germany); and (c) mixing the pre-mixed aggregate fractions with the binder for 120 s at the same speed. The mortars were cast into 5 × 5 × 5cm 3 metallic molds and cured for 24 h in controlled conditions (20 ± 0.5 °C and 95 ± 1.0% relative humidity). After demolding, the mortar specimens were individually sealed with a plastic film and thermally treated for 72 h at 50 °C and then kept at room conditions (20 ± 1.0 °C and 65 ± 5.0% relative humidity) until testing. Based on our previous laboratory experimentation, the mortar mixture was designed for a binder-to-aggregate mass ratio of 0.88. Fuller distribution was applied to optimize aggregate particle size distribution (PSD). The PSD of the two fractions of the slag used as aggregates is shown in Figure 1. The density of the aggregate fractions was determined to be approximately 3.0 ± 0.03 g/cm 3 , and their mass ratio in the mortars' design was set to be 1.4:1. The characteristics of the reference mortar mix design are provided in Table 2. The mortars were produced following the procedure described by Ascensão et al. [19]. In short, it involved: (a) manually mixing the powdered slag and silica fume in a plastic container; (b) mixing the binders' solid components (PS+SF) and the alkaline solution at 140 rpm for 600 s in a mortar mixer (Testing Bluhm & Beuerherdt Gmbh, Berlin, Germany); and (c) mixing the pre-mixed aggregate fractions with the binder for 120 s at the same speed. The mortars were cast into 5 × 5 × 5 cm 3 metallic molds and cured for 24 h in controlled conditions (20 ± 0.5 • C and 95 ± 1.0% relative humidity). After demolding, the mortar specimens were individually sealed with a plastic film and thermally treated for 72 h at 50 • C and then kept at room conditions (20 ± 1.0 • C and 65 ± 5.0% relative humidity) until testing.

High-Temperature Processing
After 28 d of curing, mortar specimens were heated from room temperature up to 1100 • C in a chamber furnace (Carbolite RHF). Five different target temperatures were defined: 105, 600, 750, 900, and 1100 • C. These target temperatures were chosen on the grounds of previous studies [5,7] that demonstrated the existence of relevant chemical transformations occurring in these temperature ranges. For each target temperature, a thermal profile with a heating rate of 1 • C/min [5,20,21] and an isothermal step of 1 h at the maximum temperature was followed. The heated specimens were left to cool down naturally inside the closed furnace. This procedure was replicated with a heating rate of 10 • C/min to assess the impact of heating rate on AAMs mineralogy, morphology, and post-heating properties. A minimum of two specimens per heating rate and maximum temperature were tested.

Physical-Mechanical Characterization
The weight (0.1-g accuracy) and dimensions (0.01-mm accuracy) of each specimen were measured before and after thermal exposure to evaluate the mass loss and the volumetric changes that occurred during thermal treatment (and eventual anisotropic features), respectively. The apparent density of the mortars was determined by the relation between the weight and volume of each sample. Compressive strength was determined using a Universal Testing Machine (Controls) according to EN196-1:2016 [22].

Crystallinity and Morphological Characterization
The crystallinity of the starting slag, silica fume, and mortars produced was assessed by X-ray diffraction (XRD), carried out on a conventional Bragg-Brentano Bruker D8 Advance diffractometer equipped with Lynxeye detector (Cu Kα radiation λ = 1.54059 Å, divergence slit 0.5 • , Soller slit set 2.5 • + 2.5 • , 5-70 • 2θ, step/size 0.02 • , and t/step 0.04 s.) and phase identification by EVA software (database ICDD-PDF-2). Samples from green and thermally treated mortars were collected from mechanical tests, then ground, sieved (≤45 µm), and kept sealed prior to testing. Scanning electron microscopy (SEM-EVO ® MA 15) equipped with energy dispersive X-ray spectrometry (EDS, AZtecEnergy, Karnataka, India) was used to further evaluate the differences in each sample's morphology before and after thermal exposure. All backscattered electron images (BSE) were acquired using a 20-kV acceleration voltage and a working distance of 10.0 mm.

Physical-Mechanical Characterization
The AAMs' residual properties were characterized as a function of temperature (up to 1100 • C) and heating rate (1 and 10 • C/min). Figure 2 illustrates the color change, cracking, and deformation of the mortars after high-temperature exposure. Below 750 • C, there were no visible effects on the surface of the specimens independent of the heating rate used. Similar observations have been made by Khan et al. [23] in fly ash/GGBFS mortars. Only a slight lightening of the color could be observed from 20 • C ( Figure 2a) to 600 • C (Figure 2b), which can be attributed to gradual dehydration and some Fe oxidation as shown by Rickard et al. [24]. In fact, Figure 3 shows that mass loss mainly occurred in this temperature range. The mass loss at 105 • C can be attributed to the evaporation of free water located in existing pores. Previous works [7] have reported a higher mass loss in this temperature range, which can be explained-along with the differences in binder chemistry-by the curing conditions (50 • C for 3 d) and curing duration (28 d) prior to thermal testing employed in this work. Thus, part of the water has already evaporated before the experiment when following this procedure. The further mass loss observed at temperatures up to 600 • C can be attributed to the egress of water present in smaller sized pores, interstitial water, and chemically bonded water in hydroxyl groups [11,14]. No anisotropic features were detected, and a uniform volumetric contraction that did not exceed 20 vol% was observed. Rickard et al. [24] reported significant lower shrinkage values in Fe-rich AAMS (3% at 900 • C) while He et al. [25] refer to shrinkage values of 5% in FA geopolymers fired at temperature of 800 • C. The influence of the heating rate started to become evident from 750 • C, as shown in Figure 2c-h. The specimens heated at lower rates (1 • C/min) changed color from a light grey to a gradually more brownish and yellow color (see Figure 2c,e,g). No cracks were visible on the surfaces of these samples, and the absence of color gradients indicates uniform mineralogical transformations [24]. The mineralogical transformations that occurred at these temperatures explain the changes in color and are discussed later in this work.
Appl. Sci. 2022, 12, x FOR PEER REVIEW 6 of 17 this temperature range, which can be explained-along with the differences in binder chemistry-by the curing conditions (50 °C for 3 d) and curing duration (28 d) prior to thermal testing employed in this work. Thus, part of the water has already evaporated before the experiment when following this procedure. The further mass loss observed at temperatures up to 600 °C can be attributed to the egress of water present in smaller sized pores, interstitial water, and chemically bonded water in hydroxyl groups [11,14]. No anisotropic features were detected, and a uniform volumetric contraction that did not exceed 20 vol% was observed. Rickard et al. [24] reported significant lower shrinkage values in Fe-rich AAMS (3% at 900 °C) while He et al. [25] refer to shrinkage values of 5% in FA geopolymers fired at temperature of 800 °C. The influence of the heating rate started to become evident from 750 °C, as shown in Figure 2c-h. The specimens heated at lower rates (1 °C/min) changed color from a light grey to a gradually more brownish and yellow color (see Figure 2c,e,g). No cracks were visible on the surfaces of these samples, and the absence of color gradients indicates uniform mineralogical transformations [24]. The mineralogical transformations that occurred at these temperatures explain the changes in color and are discussed later in this work. (e,f) 900 °C with heating rates of 1 and 10 °C/min, respectively; (g,h) 1100 °C with heating rates of 1 and 10 °C/min, respectively. Mortars exposed to 105 °C are not shown here, but no significant modifications of specimens' appearance could be seen relative to the ones kept at 20 and 600 °C (a). At 600 °C there were no visible differences between specimens heated with distinct heating rates, with (b) being a representative example of both. Molten material at the surface of mortars heated up to 1100 °C at 10 °C/min is illustrated in (h). (e,f) 900 • C with heating rates of 1 and 10 • C/min, respectively; (g,h) 1100 • C with heating rates of 1 and 10 • C/min, respectively. Mortars exposed to 105 • C are not shown here, but no significant modifications of specimens' appearance could be seen relative to the ones kept at 20 and 600 • C (a). At 600 • C there were no visible differences between specimens heated with distinct heating rates, with (b) being a representative example of both. Molten material at the surface of mortars heated up to 1100 • C at 10 • C/min is illustrated in (h). Appl. Sci. 2022, 12, x FOR PEER REVIEW 7 of 17 The curves are drawn as a guide to the eye. Due to deformation of the samples fired at 1100 °C with a heating rate of 10 °C/min data could not be collected.
When a higher heating rate was employed (10 °C/min), less distinguishable color changes were observed. The specimens heated up to 750 and 900 °C presented a less vivid brownish color relative to the ones heated at lower heating rates. Some color gradients were observed across the specimens' surfaces, especially when heated to 900 °C. The differences in color indicate that temperature gradients may have led to uneven transformations as suggested elsewhere [24]. In addition, some cracks and superficial degradation were visible on the surface of samples. The rapid increase of temperature accelerates dehydration processes, but water release occurs at higher temperatures due to lower heat and mass transfer rates (Figure 3), giving less time for steam to escape, which in turn results in excessive vapor pressure and leads to crack propagation [15]. Khan et al. [23] also reported severe cracks on Ca-rich fly ash/GGBFS mortars fired at 5 °C/min, suggesting some similarities in the effects of increased heating rates. However, all samples heated up to 900 °C maintained their structural integrity, regardless of the heating rate used. The samples heated to up 1100 °C with a heating rate of 10 °C/min present considerable molten material on their bottom surfaces (rotated in Figure 2h). Apart from the excessive vapor pressure generated, the large temperature gradients across the samples' cross-sections resulted in differential shrinkage, which increased the stresses formed and crack propagation (Figure 4), allowing for the runoff of molten material. Similar findings were reported by Fernández-Jiménez et al. [26], who observed that high heating and cooling rates form new or intensified pre-existing cracks due to the high thermal strain generated, ultimately leading to local failure. The viscosity of the melts also decreases as the temperature rises, which further facilitates their flow. Hence, the amount of molten material is expected to be significant at temperatures higher than 900 °C, as the increased flowability of the melts and the high number of voids and cracks in the mortar structures resulted in molten material coming out of the surfaces, as seen in Figures 2h and 4. When a higher heating rate was employed (10 • C/min), less distinguishable color changes were observed. The specimens heated up to 750 and 900 • C presented a less vivid brownish color relative to the ones heated at lower heating rates. Some color gradients were observed across the specimens' surfaces, especially when heated to 900 • C. The differences in color indicate that temperature gradients may have led to uneven transformations as suggested elsewhere [24]. In addition, some cracks and superficial degradation were visible on the surface of samples. The rapid increase of temperature accelerates dehydration processes, but water release occurs at higher temperatures due to lower heat and mass transfer rates (Figure 3), giving less time for steam to escape, which in turn results in excessive vapor pressure and leads to crack propagation [15]. Khan et al. [23] also reported severe cracks on Ca-rich fly ash/GGBFS mortars fired at 5 • C/min, suggesting some similarities in the effects of increased heating rates. However, all samples heated up to 900 • C maintained their structural integrity, regardless of the heating rate used. The samples heated to up 1100 • C with a heating rate of 10 • C/min present considerable molten material on their bottom surfaces (rotated in Figure 2h). Apart from the excessive vapor pressure generated, the large temperature gradients across the samples' cross-sections resulted in differential shrinkage, which increased the stresses formed and crack propagation (Figure 4), allowing for the runoff of molten material. Similar findings were reported by Fernández-Jiménez et al. [26], who observed that high heating and cooling rates form new or intensified pre-existing cracks due to the high thermal strain generated, ultimately leading to local failure. The viscosity of the melts also decreases as the temperature rises, which further facilitates their flow. Hence, the amount of molten material is expected to be significant at temperatures higher than 900 • C, as the increased flowability of the melts and the high number of voids and cracks in the mortar structures resulted in molten material coming out of the surfaces, as seen in Figures 2h and 4.  Appl. Sci. 2022, 12, x FOR PEER REVIEW 8 of 17 In samples heated to 1100 °C with a lower heating rate, no molten material was visible after thermal exposure. The more gradual heating process diminished crack formation, which allowed the molten material to be retained inside the monolith, positively affecting the mortars' structural stability and high-temperature performance. Although mortars heated up to 1100 °C with heating rates of 10 °C/min retain their cubic shape and were tested for their compressive strength, they were excluded from further physical analysis (e.g., mass loss). The physical and mineralogical modifications that occurred upon heating had clear repercussions on AAMs' residual properties. From room temperature to 600 °C a slight increase of compressive strength after thermal exposure was observed, as shown by Figure 5. The improvement of compressive strength after thermal exposure to temperatures that do not exceed 600 °C is not related to melting and crystallization processes (see Section 3.3), but most likely to the increment of on-going polymerization reactions. In previous work [27,28], the authors have shown that in very similar alkali-activated systems, exothermic reactions have not completely ceased after more than 20 d of curing. Elevated temperatures stimulate these late-stage reactions and prompt the formation of more mechanically resistant binding phases. Along with the absence of significant crack formation, this beneficial effect overcomes the deleterious impact of dehydration and improves AAMs' mechanical features. The increase of compressive strength in this temperature range is often associated with densification [7,29]. However, it can be seen in Figure 6 that In samples heated to 1100 • C with a lower heating rate, no molten material was visible after thermal exposure. The more gradual heating process diminished crack formation, which allowed the molten material to be retained inside the monolith, positively affecting the mortars' structural stability and high-temperature performance. Although mortars heated up to 1100 • C with heating rates of 10 • C/min retain their cubic shape and were tested for their compressive strength, they were excluded from further physical analysis (e.g., mass loss). The physical and mineralogical modifications that occurred upon heating had clear repercussions on AAMs' residual properties. From room temperature to 600 • C a slight increase of compressive strength after thermal exposure was observed, as shown by Figure 5.  In samples heated to 1100 °C with a lower heating rate, no molten material was visible after thermal exposure. The more gradual heating process diminished crack formation, which allowed the molten material to be retained inside the monolith, positively affecting the mortars' structural stability and high-temperature performance. Although mortars heated up to 1100 °C with heating rates of 10 °C/min retain their cubic shape and were tested for their compressive strength, they were excluded from further physical analysis (e.g., mass loss). The physical and mineralogical modifications that occurred upon heating had clear repercussions on AAMs' residual properties. From room temperature to 600 °C a slight increase of compressive strength after thermal exposure was observed, as shown by Figure 5. The improvement of compressive strength after thermal exposure to temperatures that do not exceed 600 °C is not related to melting and crystallization processes (see Section 3.3), but most likely to the increment of on-going polymerization reactions. In previous work [27,28], the authors have shown that in very similar alkali-activated systems, exothermic reactions have not completely ceased after more than 20 d of curing. Elevated temperatures stimulate these late-stage reactions and prompt the formation of more mechanically resistant binding phases. Along with the absence of significant crack formation, this beneficial effect overcomes the deleterious impact of dehydration and improves AAMs' mechanical features. The increase of compressive strength in this temperature range is often associated with densification [7,29]. However, it can be seen in Figure 6 that The improvement of compressive strength after thermal exposure to temperatures that do not exceed 600 • C is not related to melting and crystallization processes (see Section 3.3), but most likely to the increment of on-going polymerization reactions. In previous work [27,28], the authors have shown that in very similar alkali-activated systems, exothermic reactions have not completely ceased after more than 20 d of curing. Elevated temperatures stimulate these late-stage reactions and prompt the formation of more mechanically resistant binding phases. Along with the absence of significant crack formation, this beneficial effect overcomes the deleterious impact of dehydration and improves AAMs' mechanical features. The increase of compressive strength in this temperature range is often associated with densification [7,29]. However, it can be seen in Figure 6 that the apparent density slightly decreases at 600 • C. Although more resistant and probably denser polymeric chains have been formed, this decrease is the combined effect of water loss and shrinkage. As mass loss (≈8.5 wt.%) was significantly higher than the volumetric contraction up to 600 • C (<2 vol%), apparent densities decrease slightly. The increment of on-going reactions is further supported by comparing the mechanical strength of samples thermally treated up to 600 • C with different heating rates. As can be seen, lower heating rates-meaning more prolonged residence time at elevated temperatures-favor the formation of more resistant AAMs without significantly modifying the samples' mass and volume (Figure 3) or promoting the significant formation of new crystalline phases (Figure 7). At 750 • C, considerable deterioration of AAMs' mechanical properties occurred with both heating rates. This process may be related with modifications on the iron-and aluminosilicate networks occurring even before sintering [30]. The shrinkage of the aluminosilicate network induces internal cracks and increases the mean pore size, which leads to poorer mechanical performances [30]. Nonetheless, totally dehydrated charge balance cations (Na + or K + ) act as flux at high temperatures and reduce the temperature at which melting and crystallization start. This can imply the formation of partially sintered phases at around 750 • C [26]; however, these are not fully able to compensate for the decline in compressive strength caused by dehydration-related shrinkage.
Appl. Sci. 2022, 12, x FOR PEER REVIEW 9 of 17 the apparent density slightly decreases at 600 °C. Although more resistant and probably denser polymeric chains have been formed, this decrease is the combined effect of water loss and shrinkage. As mass loss (≈ 8.5 wt.%) was significantly higher than the volumetric contraction up to 600 °C (<2 vol%), apparent densities decrease slightly. The increment of on-going reactions is further supported by comparing the mechanical strength of samples thermally treated up to 600 °C with different heating rates. As can be seen, lower heating rates-meaning more prolonged residence time at elevated temperatures-favor the formation of more resistant AAMs without significantly modifying the samples' mass and volume (Figure 3) or promoting the significant formation of new crystalline phases (Figure 7). At 750 °C, considerable deterioration of AAMs' mechanical properties occurred with both heating rates. This process may be related with modifications on the iron-and aluminosilicate networks occurring even before sintering [30]. The shrinkage of the aluminosilicate network induces internal cracks and increases the mean pore size, which leads to poorer mechanical performances [30]. Nonetheless, totally dehydrated charge balance cations (Na + or K + ) act as flux at high temperatures and reduce the temperature at which melting and crystallization start. This can imply the formation of partially sintered phases at around 750 °C [26]; however, these are not fully able to compensate for the decline in compressive strength caused by dehydration-related shrinkage.  An additional factor that has been reported to contribute to the loss of strength at temperatures around 800 • C is the expansion of coarse aggregates [31]. Due to the type and granulometric distribution of the amorphous aggregates used in this work, however, a similar phenomenon seems unlikely to be significant. The effects of the network shrinkage could be seen at the macroscopic level by the steepest increase in volumetric contraction observed around 750 • C (Figure 3). At this temperature, increasing the heating rate was found to be detrimental to AAMs' residual strength. Along with higher crack formation due to the higher vapor pressure generated, the considerable difference in mechanical performance between mortars heated at different rates may indicate more deteriorated network structures. At 750 • C, AAMs start to lose their predominantly amorphous character, leading to the appearance of some broadened peaks of magnetite (Fe 3 O 4 ) and esseneite (CaAlFeSiO 6 ) (Figure 7). These crystals could not be clearly identified in morphological analysis and does not seem able to compensate for the decrease of AAMs' mechanical strength; however, they may have contributed to some degree to the densification of the samples. As mass loss was mainly associated with dehydration processes occurring mostly before 300 • C, the mass remained practically unchanged at temperatures higher than 600 • C. Therefore, the apparent density followed a similar trend to the volumetric contraction. A minor increase of mass was visible from 750 • C and above (Figure 3) that can be attributed to further oxidation and crystallization of iron species into magnetite and afterwards to hematite. The reference mortar presented a density of 2.46 ± 0.02 kg/m 3 and a compressive strength of 85 ± 3 MPa after 28 d of curing. Using a heating rate of 1 • C/min the maximum apparent density was observed at 900 • C (2.78 ± 0.03 g/cm 3 ); however, the decrease of apparent density by 1100 • C was not detrimental to mechanical performance as the samples resulting from a heating rate of 1 • C/min presented the maximum compressive strength recorded, 184 ± 13 MPa ( Figure 5). An additional factor that has been reported to contribute to the loss of strength at temperatures around 800 °C is the expansion of coarse aggregates [31]. Due to the type and granulometric distribution of the amorphous aggregates used in this work, however, a similar phenomenon seems unlikely to be significant. The effects of the network shrinkage could be seen at the macroscopic level by the steepest increase in volumetric contraction observed around 750 °C (Figure 3). At this temperature, increasing the heating rate was found to be detrimental to AAMs' residual strength. Along with higher crack formation due to the higher vapor pressure generated, the considerable difference in mechanical performance between mortars heated at different rates may indicate more deteriorated network structures. At 750 °C, AAMs start to lose their predominantly amor- In those samples, when the temperature exceeded 750 • C and the crystallization of new phases became more intense, the compressive strength decrease was fully reversed. The compressive strength of sintered samples was progressively increased as the temperature rose, with an increase relative to the parent mortar of 85% and 115% at 900 and 1100 • C, respectively. Such strength gains contrast with the losses reported by several authors in low-iron systems, with proclivity towards lower mechanical performances as the temperature rises [23,32]. Saxena et al. [32] investigated the influence of the Si/Al ratio in the thermal performance of fly ash geopolymer mortars and reported loses in compressive strength ranging from 54-77% when firing at 1000 • C. Conversely, Villaquirán-Caicedo and de Gutiérrez [33] reported significant increases in the mechanical strength of metakaolin geopolymers, but the maximum strength reported was approximately 160 MPa and required firing at 1200 • C for 2 h.
Increasing the heating rate to 10 • C/min not only provoked a higher decrease of compressive strength at 750 • C, but also significantly impaired the performance recovery at higher temperatures. On those samples, the maximum compressive strength was halved (97 ± 8 MPa) and attained at 900 • C. Despite the recovery from the decay observed at 750 • C and the 13% increase relative to the parent mortar, much higher compressive strengths were observed in samples heated up to 900 • C with the lower heating rate (158 ± 13 MPa). At 1100 • C, the mechanical strength of samples heated with different heating rates became even more divergent. Unlike samples heated at 1 • C/min that yielded the maximum strength at this temperature, when high heating rates were imposed, a considerable degradation of the AAMs' performance occurred. As shown in Figure 3, these samples were not dimensionally stable, with some molten material coming out of the samples' structure. This explains the much lower compressive strength. The reason for material run-off in some rapid heated samples might be that crystallization is not fast enough, while the viscosity decreases due to the increasing temperature. At the same time, capillary forces generated at the solid surface may further favor some material outflow from the molten core. At 1 • C/min, there seems to be a better match between the heating and crystallization rate, as the significant residence time and formation of more liquid phases does not promote such phenomena. With the exception of samples exposed to 1100 • C with a heating rate of 10 • C/min (17 ± 9 MPa), all sintered samples presented compressive strength values higher than 40 MPa.

Crystallinity
The mineralogical composition and the high-temperature mineralogical transitions are of utmost importance to fully describe the thermal behavior of AAMs and to safely classify them in terms of high-temperature resistance. At room temperature, AAMs are predominantly amorphous to semi-crystalline materials, and the amount and type of crystalline phases present are mainly dependent on the parent precursors. The powdered slag (PS) and silica fume (SF) used in this work as starting materials were almost completely XRD amorphous materials. Only one crystalline peak of small intensity was detected in the XRD patterns for SF, identified as moissanite (SiC) which was an impurity (SF and PS XRD patterns available in [27]).
Not surprisingly, the AAMs made thereof exhibited an entirely amorphous structure at room temperature, with a pronounced hump between 20-40 • 2θ in their XRD patterns. Heating the samples to 105 • C did not lead to any crystallization regardless of the heating rate used (not shown here for the sake of brevity). The first signs of the formation of crystalline phases were visible in the samples heated at 600 • C. Operating at the heating rate of 1 • C/min, no crystalline peak could be detected, as shown by Figure 7a; however, increasing the heating rate to 10 • C/min likely promoted the formation of some crystalline phases, visible from the peaks at 2θ~26-28 • (Figure 7b). Given the weakness of the signals, a clear identification is impossible, but the positions are consistent with the main diffraction lines of leucite (KAlSi 2 O 6 ) polymorphs (Leucite 1, PDF#76-2298, and Leucite 2, PDF#85-1626). Leucite is a well-known product of the thermal transformation of K-based AAMs [9].
The formation of crystalline phases was more pronounced as the temperature increased, stimulated by the separation of iron oxide phases. As with many waste-derived iron-rich glass-ceramic systems [34,35], a temperature increase to 750 • C involved the precipitation of magnetite (Fe 3 O 4 , PDF#89-0951), whereas a pyroxene phase (esseneite, PDF#78-1546), i.e., an Al-doped calcium-iron silicate, crystallized at 900 • C. Magnetite, embedding iron in both Fe 2+ and Fe 3+ forms (Fe 3 O 4 =FeO·Fe 2 O 3 ), is a typical nucleating agent, separating at 650-800 • C from 'basaltic' glasses (glasses of the SiO 2 -Al 2 O 3 -Fe 2 O 3 -CaO system, including also alkali oxides), later transformed into pyroxene glass-ceramics. Magnetite may persist or be reabsorbed in the solid solution [34].
Heating to 900 • C and above promoted some oxidation, shown by the fact that both additional phases detected, i.e., CaAlFeSiO 6 (PDF#78-1546) and hematite (Fe 2 O 3 , PDF#89-0599) host only Fe 3+ ions [34]. Finally, it can be noted that the different heating rates determined a different evolution at the maximum temperature (1100 • C), with the appearance of calcium alumino silicate phases: while the sample heated at 1 • C/min exhibited a limited precipitation of anorthite (CaAl 2 Si 2 O 8 , PDF#73-0264), the sample heated at 10 • C/min featured both anorthite and gehlenite (Ca 2 Al 2 SiO 7 , PDF#72-2128). Differences in the phase assemblage, with consequences for the thermo-elastic stresses developed upon cooling, could be a further explanation for the lower strength of samples heated at 10 • C/min: in particular, gehlenite is recognized for its high coefficient of thermal expansion [36].
The formation of esseneite and magnetite as the predominant phases in all samples can be seen as a natural consequence of the abundant availability of Si, Ca, and Fe species provided by the mix design. The addition of silica fume and potassium silicate solution could promote the formation of these thermally stable and refractory phases, which favor the formation of AAMs with high thermal performance [6,7,9,10,14]. The residual properties of the developed materials were enhanced with the adoption of slower heating rates. The physico-mechanical phenomena that occurred during faster heating processes, such as the higher vapor pressure and increased shrinkage and degradation of binding phases, seem to overwhelm possible positive effects from crystal formation. To further investigate such phenomena, morphological analyses were performed in samples heated up to 1100 • C, where the most significant crystallographic differences were observed. Figure 8 shows representative backscattered electron imaging micrographs of the reference mortar before being exposed to thermal treatment. A binder phase was formed (Figure 8a), but the existence of embedded unreacted particles confirms that complete dissolution was not achieved. The high solid-to-liquid ratio used (2.85) limited the precursors' dissolution, but the samples show a homogeneous matrix in which undissolved particles acted as small-sized aggregates, (Figure 8b). Extensive microcracks have been observed with undissolved particles acting as physical barriers that limited crack development and propagation. Similar microcracks have been reported in the literature, and their origin associated with drying shrinkage processes [28]. EDS analyses revealed a binder phase predominantly composed of Si, Al, Fe, Ca, and K, while the spectrum of undissolved particles was in line with the chemical composition of the slag used as the main precursor and as aggregates (Table 3). Some spherical pores were also observed that can be attributed to entrapped air during mixing. phase predominantly composed of Si, Al, Fe, Ca, and K, while the spectrum of undissolved particles was in line with the chemical composition of the slag used as the main precursor and as aggregates (Table 3). Some spherical pores were also observed that can be attributed to entrapped air during mixing. The effects of high-temperature exposure and heating rate on mortar morphology are illustrated in Figure 9. Independently of the heating rate used, a more compact microstructure with few large cracks was formed relative to the parent mortar. Partial softened aggregates have been reported to contribute to form homogenous mortars [23] and morphological analysis indicates that undissolved particles previously visible in green samples were almost completely molten (Figure 9a,b). Their previous homogeneous structure was transformed into a texture of finely exsolved phases. However, the contours of undissolved particles remain identifiable as white lines. A clearer picture of interfacial transition zones between molten aggregates and previous existing binders is better seen at higher magnification (Figure 9c,d). EDS analyses revealed the predominance of iron elements in the white areas outlining the aggregates' contours, which suggests the concentration of oxidized iron phases at these locations (Table 3 and Figure 9d). At higher magnifications, the morphological differences between samples heated with distinct heating rates become more evident.    temperatures that may have increased the amount of molten material, and the use of potassium-based activating solutions that assist the formation of very viscous vitreous phases [26] may have all contributed to sealing pre-existing pores and cracks. More homogeneous microstructures were formed, in agreement with the enhanced compressive strength of these mortars after being exposed to high temperatures (Figure 3). At the highest heating rate (Figure 9d), the thermal strains generated upon heating were more difficult to absorb by the existing microstructure at room temperature. Apart from the already mentioned local failure and runoff of molten material, the formation of micropores along interfacial transition zones was also visible. A less compact and more heterogeneous structure was formed at the microscopic level, and the less effective bonding between the newly crystallized phases can explain the lower mechanical strength of mortars heated at 10 °C/min at all temperatures above 600 °C.

Morphological Analysis
These differences in the mortars' morphology show that heating rate plays a vital role in the sintering process. Increasing the heating rate affects not only the phases crystallized but also their morphology and distribution, which in turn affects the dimensional stability of the mortars at high temperatures and their properties after cooling. The crystallization of new phases was particularly relevant at temperatures over 900 °C, but strength only increased if a low heating rate was employed. Nevertheless, all mortars were able to largely surpass the threshold temperature of 400 °C without any visible signs of degradation or spalling. This behavior clearly shows the superior thermal performance of the developed products relative to mainstream construction materials. Figure 9. Backscattered electron imaging micrographs of mortars exposed to 1100 °C: (a,c) with a heating rate of 1 °C/min; (b,d) with a heating rate of 10 °C/min.

Conclusions
In this work, the effect of temperature and heating rate on the post-heated properties of CaO-FeOx-Al2O3-SiO2-rich AAMs was investigated. The mix design of the AAMs employed a high content of a low-value residue (85.9 wt.%). This was driven by the aim to Figure 9. Backscattered electron imaging micrographs of mortars exposed to 1100 • C: (a,c) with a heating rate of 1 • C/min; (b,d) with a heating rate of 10 • C/min. The effects of high-temperature exposure and heating rate on mortar morphology are illustrated in Figure 9. Independently of the heating rate used, a more compact microstructure with few large cracks was formed relative to the parent mortar. Partial softened aggregates have been reported to contribute to form homogenous mortars [23] and morphological analysis indicates that undissolved particles previously visible in green samples were almost completely molten (Figure 9a,b). Their previous homogeneous structure was transformed into a texture of finely exsolved phases. However, the contours of undissolved particles remain identifiable as white lines. A clearer picture of interfacial transition zones between molten aggregates and previous existing binders is better seen at higher magnification (Figure 9c,d). EDS analyses revealed the predominance of iron elements in the white areas outlining the aggregates' contours, which suggests the concentration of oxidized iron phases at these locations (Table 3 and Figure 9d). At higher magnifications, the morphological differences between samples heated with distinct heating rates become more evident.
Mortars heated at 1 • C/min exhibit a continuous interfacial transition zone with no visible cracks or microscopic pores. The progressive increase in temperature that reduces vapor pressure and cracks formed upon heating, the prolonged residence time at high temperatures that may have increased the amount of molten material, and the use of potassium-based activating solutions that assist the formation of very viscous vitreous phases [26] may have all contributed to sealing pre-existing pores and cracks. More homogeneous microstructures were formed, in agreement with the enhanced compressive strength of these mortars after being exposed to high temperatures (Figure 3).
At the highest heating rate (Figure 9d), the thermal strains generated upon heating were more difficult to absorb by the existing microstructure at room temperature. Apart from the already mentioned local failure and runoff of molten material, the formation of micropores along interfacial transition zones was also visible. A less compact and more heterogeneous structure was formed at the microscopic level, and the less effective bonding between the newly crystallized phases can explain the lower mechanical strength of mortars heated at 10 • C/min at all temperatures above 600 • C.
These differences in the mortars' morphology show that heating rate plays a vital role in the sintering process. Increasing the heating rate affects not only the phases crystallized but also their morphology and distribution, which in turn affects the dimensional stability of the mortars at high temperatures and their properties after cooling. The crystallization of new phases was particularly relevant at temperatures over 900 • C, but strength only increased if a low heating rate was employed. Nevertheless, all mortars were able to largely surpass the threshold temperature of 400 • C without any visible signs of degradation or spalling. This behavior clearly shows the superior thermal performance of the developed products relative to mainstream construction materials.

Conclusions
In this work, the effect of temperature and heating rate on the post-heated properties of CaO-FeO x -Al 2 O 3 -SiO 2 -rich AAMs was investigated. The mix design of the AAMs employed a high content of a low-value residue (85.9 wt.%). This was driven by the aim to reduce the production cost and environmental impact of the produced materials, thus providing an up-cycling solution for the large volumes of CaO-FeO x -Al 2 O 3 -SiO 2 -rich slags generated in thermochemical conversion processes.
The results demonstrated the superior thermal performance of AAMs produced from these residues relative to mainstream construction materials and acknowledged the vital role of both of temperature and heating rate. Lower heating rates had a beneficial effect on AAMs' mechanical performance. The maximum strength was observed when specimens were exposed to 1100 • C at 1 • C/min (184 ± 13 MPa), more than doubling the compressive strength of green mortars (85 ± 3 MPa). Increasing the heating rate from 1 to 10 • C/min was found to promote deterioration processes that resulted in extensive crack development and even the deformation of samples at 1100 • C. Along with the formation of less homogeneous microstructures, these deleterious effects were found to result in more fragile structures after thermal exposure, which limits sintering to 900 • C if high heating rates are involved.
Crystallographic analyses revealed the crystallization of thermally stable phases, indicating that the refractory and high-temperature resistant features of the developed CaO-FeO x -Al 2 O 3 -SiO 2 -rich AAMs could be potentially further enhanced by fine-tuning the mix design.
Comparing to benchmark materials based on OPC, no signs of spalling were visible, and the developed materials could withstand temperatures up to 900 • C with heating rates up to 10 • C/min while maintaining their structural integrity and considerable mechanical strength (>40 MPa).