Thermal Behaviour and Microstructure of Self-Cured High-Strength Plain and Fibrous Geopolymer Concrete Exposed to Various Fire Scenarios

Abstract

The fire resistance of construction materials is an essential part of safety requirements in the construction industry. In this work, experimental investigations were conducted to understand the thermal behaviour, spalling, transfer characteristics, strength, and microstructures of self-cured high-strength plain (HSGC) and steel-fibre-reinforced geopolymer concrete (S–HSGC) under severe fire scenarios with peak temperatures of 275, 560, and 825 °C; the peak was maintained for a period of 120 min after reaching it. Forty-eight standard cylindrical specimens for each mixture were prepared to test and analyse their time–heat response, gradients, visual appearance, spalling, density change, water absorption, and compressive strength before and after fire exposure. Additionally, Scanning Electron Microscopy (SEM) along with Energy Dispersive X-ray Analysis (EDX) were utilised to analyse the internal structures and phase transformations. The thermal analysis showed that no cases of explosive spalling were recorded during sample exposure to various fires, while the used hook-end steel fibres had an influence on the considered test variables. The sample cores almost reached the target heat, and the thermal saturation degree at the peak ranged from 55 to 97%. The experimental findings also revealed slight surface cracking after exposure to 560 °C fires, while the surface cracking was more obvious for specimens exposed to 825 °C. Moreover, the residual compressive strength of the S–HSGC at various fires was noticeably 10.20% higher than that of the HSGC. Also, state-of-the-art research data were used to discuss the prediction model’s performance. The SEM and EDX results showed that the self-cured geopolymerization process was effective and successful in producing gels, in addition to the significant phase transformations in microstructures at different fires. This study presented sophisticated data on the behaviour of HSGC and S–HSGC exposed to fires up to 825 °C.

1. Introduction

In recent years, the building industry has demonstrated a great deal of interest in the usage of High Strength Concrete (HSC) due to the structural performance advantages it offers over standard plain concrete [1]. Over the last few decades, fire disasters have frequently been reported worldwide and have seriously threatened private and public property and the safety of the beneficiaries. It was considered that fire-induced spalling, which weakens the load-bearing capacity of structural elements, was the primary cause of the structural collapse. Due to its denser structure, HSC has lower resistance to internal vapour pressure at high temperatures compared to normal strength concrete, which makes spalling failure typical for HSC structures exposed to accidental fires. The tunnel’s HSC linings were severely damaged by spalling after being exposed to fire, and, as a result, the main reinforcement rebar’s tensile strength was deteriorating [2].

During the concrete’s exposure to increased temperatures, its free water starts to evaporate between 100 and 200 °C, and as the temperature rises, chemical water (in the microstructure of the cementitious matrix) is released [3,4]. Because the vapours created in the HSC cannot be removed because of its limited permeability, the internal vapour pressure inside the concrete grows as the temperature rises. More than the tensile strength of the concrete, this water vapour pressure may reach 8 MPa [5]. As a result, capillary cracks emerge in the concrete, and as they expand (macro-crack), the concrete breaks cause explosive spalling because of the restricted tensile strength of the concrete [6,7]. As a result of the limited thermal conductivity of concrete, the exterior surface has a higher temperature than the inside layers of the fire-exposed concrete. This variance in heat generates internal stress in the mixture components. Tensile stress and cracking may be exacerbated due to variances in the coefficient of thermal expansion of the concrete components, nonlinear expansion of the components, and changes in distinct phases in the aggregates [8]. In contrast, early microcracks (before and after the hardening process) start to develop in the initial phase of cracking at the weakest zone, which is the interfacial transition zone (ITZ) between aggregate and cement mortar in old concrete [9,10]. Therefore, the increase in heat contributes to the development of those cracks and then causes deterioration in microstructures.

Moreover, heating to approximately 450 °C causes chemical degradation of the cement paste, resulting in contraction along the ITZ. A chemical reaction occurs at around 550 °C that yields calcium oxide from the breakdown of calcium hydroxide and calcium carbonate in concrete. However, the calcium silicate hydroxide gel, which decomposes around 600 °C, is the most critical component of the cement paste and the source of its greatest strength [11,12]. In contrast, the aggregates expand during the early heating phases, but the cement paste contracts due to the loss of water, resulting in stress from drying-induced shrinkage [13]. As a result of these changes, the concrete’s strength deteriorates at a different level, which is influenced by various factors, including fire temperatures, duration of exposure, type and rate of cooling, structural element type and size, and the fire-resistant properties of the concrete materials [14]. Therefore, the study of construction material properties’ deterioration after fires should cover most aspects, thereby accessing the maintenance possibility or not of the structure’s elements.

HSC production consumes a lot of virgin materials and emits greenhouse gases [15]. In general, early studies found that one tonne of carbon dioxide (CO₂) is produced for every tonne of concrete, but new techniques and cleaner energy have decreased this to 0.6 tonnes [16]. Cement manufacturing is responsible for 7–10% of worldwide CO₂ emissions [17,18]. To achieve significant environmental improvements, these days, it has become necessary to find suitable alternative construction materials with environmentally friendly properties. Alkali-activated or geopolymer concrete (GPC) is an environmentally friendly alternative to classical ordinary Portland cement concrete, which emits approximately 70% fewer greenhouse gases than Portland cement concrete [19]. It is made from by-product aluminosilicate components, such as fly ash (FA), ground granulated blast furnace slag (slag), and clay ceramic waste, or agricultural waste, such as bottom ash and rice-husk ash with geopolymerization by an alkaline activator (AL), resulting in the formation of an amorphous to semi-crystalline three-dimensional silicoaluminate polymeric structure called geopolymeric gels [20]. According to Davidovits [21], FA-based GPC requires a high-temperature treatment and is kept constant for at least 24 h in order to accelerate the creation of geopolymeric gels and achieve acceptable performance compared to that of classical concrete. Hardjito et al. [22] looked into the strength and setting time of type F FA GP paste and found that it was unable to set in 24 h at room temperature (self-curing) and that its strength ranged from 1.6 to 20 MPa when heat-cured at 65–85 °C of the samples. The heat-curing method leads to boosted costs, greenhouse gas emissions, and functional issues, and it also prevents in situ application of GPC. However, self-curing, also known as the ambient-curing method, is considered a good choice because of its many benefits, such as decreased production costs, CO₂ emissions, and increased mobility of GPC, which can be easily implemented in construction sites. Self-cured GPC could be developed by increasing the fineness of the FA binder and incorporating calcium source materials, such as steel slag [23]. According to Ali and Tayşi [24], both FA and slag were utilised as the source components and stimulated with a combination of Na₂SiO₃ and a constant concentration of NaOH solution. That mixture was then combined with a constant ratio of fine and coarse aggregate to produce a GPC. They investigated the influence of the AL ratio on the mechanical properties and workability of GPC. Their samples were kept at ambient temperature to be self-cured. They found that the AL ratio has a significant effect on the workability and strength of GPC.

On the other hand, GPC has shown good resistance to fires, especially the types presented with FA as a precursor, which has been applied as a heat-resistant construction material in road tunnels and has presented good fire performance without a chance of explosive spalling and toxic gas emissions [25,26,27]. Kong and Sanjayan [28] illustrated a 25% decrease in the compressive strength of cubic-shaped moulds of metakaolin-based geopolymer samples after 10 min of exposure to 800 °C. Kong et al. [29] investigated the influence of heat on GPC and proposed that the loss of compressive strength at increased heats might be related to the thermal mismatch between the geopolymeric gels and coarse aggregates. In the case of GPC, the spalling of concrete that occurs when it is subjected to fire was also observed. Zhang et al. [30] studied the impact of temperatures up to 600 °C on the geopolymeric paste and mortar. They found the cracks on the samples’ surfaces increased with temperature, and the presence of sand limited cracks under 200 °C but produced crack development at 200–600 °C. Pan [31] investigated the influence of aggregate size on spalling of GPC in fire and discovered that GPC holding 10 mm aggregate spalled, but GPC produced by the presence of 14 mm aggregate did not record explosive spalling. Regardless of the previous mechanically excellent characteristics, the geopolymer materials, such as paste, mortar, and concrete, present a weak resistance to cracking due to their brittleness, which causes a lack of ductility [32]. However, many different ways of controlling explosive spalling and the brittleness and non-ductile tensile performance of geopolymeric materials have been suggested in previous studies. One of the earliest and most effective methods is to incorporate randomly distributed steel fibres in the mix to enhance the flexural capacity and tensile strength; this leads to a much higher energy absorption capacity and also improves the thermal heating shock resistance [33].

To this end, further research on the behaviour of GPC under fire is needed due to the lack of sufficient experimentally supported knowledge about different points. To fill some gaps in the literature, this research sought to investigate the thermal behaviour, spalling, transport properties, and compressive strength of self-cured HSGC and S–HSGC after exposure to various severe fire scenarios with peaks of 275, 560, and 825 °C. The fire peak temperatures were selected to achieve a simulation close to reality at a highly aggressive level while, at the same time, allowing for the study of the geopolymerization phase changes. SEM and EDX were conducted on unheated and heated samples in order to evaluate the self-curing geopolymerization process, microstructure characteristics, and elemental composition, as well as the chemical analysis. In addition, the performance of a proposed prediction model was discussed with a cocktail of state-of-the-art research data. This research helps to better understand the behaviour of a sustainable construction material when subjected to fire. In practise, the findings of this comprehensive study would encourage the sustainable use of self-cured HSGC and S–HSGC for purposes of fire shielding and other applications in infrastructures as well as high-rise buildings.

2. Materials and Methods

2.1. Materials and Geopolymer Production

In the HSGC mixture, FA-class F and slag, corresponding to ASTM C 618 [34] and ASTM C 989 [35], respectively, were used as the binder components. Their chemical compositions were investigated via X-ray fluorescence (XRF) analysis; the outcomes of this analysis are shown in

Table 1. The chemical compositions of FA, slag, and cement.

Binder	Component (%)
	SiO2	Al2O3	Fe2O3	CaO	MgO	K2O	Na2O	SO3	LOI	F
Fly ash	62.40	21.14	7.85	1.57	1.76	0.73	2.45	0.10	2.07	227
Slag	40.40	10.60	1.28	34.19	7.63	2.40	0.17	0.68	2.74	418
Cement	21.25	4.30	1.80	64.30	1.80	0.70	0.17	3.70	1.50	394

LOI: Loss of ignition, F: Blaine fineness (m²/kg).

. Locally manufactured sodium hydroxide (NaOH) and sodium silicate solution (Na₂SiO₃) were used (European products), which were procured in flake form with 98.5% purity. The Na₂SiO₃ consisted of Na₂O (10.6%), SiO₂ (26.5%), and H₂O (66.1%), with a fresh density of 1.390 g/mL at 25 °C. Commercial local crushed sand with a maximum size of 4 mm and crushed rocks with a maximum aggregate size of 12 mm were also used in the mixes. The specific gravity of the fine and coarse aggregates was 2.4 and 2.75, respectively. Figure 1 shows the gradation curves of the fine and coarse aggregates.

Dramix^® hooked-end steel fibres (Belgium manufacturer) were used in the current experimental work. These fibres are designated as a 3D type with a single hook-end for each side. The geometrical and physical characteristics are detailed in

Table 2. Properties and geometry of hooked-end steel fibre.

Diameter dƒ (mm)	Length Lf (mm)	Aspect Ratio (λƒ = Lƒ/dƒ)	Density (g/cm3)	Tensile Strength ƒt (MPa)	Modulus of Elasticity E (GPa)
0.55	30	55	7850	1345	200

. Finally, high-range water-reducing MasterGlenium^® RMC 303 was used to regulate the workability of the HSGC and S–HSGC mixes. The slump of all mixtures ranged between 90 and 110 mm. The primary component of the geopolymer binder is an aluminosilicate-rich material. When this material is combined with a solution of silicate and sodium hydroxide, a homogeneous mixture is formed that is considered a geopolymeric paste. After this paste is mixed with fine and coarse aggregates, it produces GPC and then necessitates high-temperature curing to obtain high strength and durability.

Utilising a combination of slag and FA as binder materials has allowed for the creation of HSGC at self-curing. The mixture proportions of HSGC and S–HSGC are tabulated in

Table 3. Mix proportions of self-cured HSGC and S–HSGC (Kg/M³).

Mix Code	FA	Slag	Activators		Aggregate		Steel Fibre	SP ‡	Ex. W a
			S.H *	S.S †	Fine	Coarse
HSGC	127	381	76	152	771	981	-	19	22.5
S-HSGC							78.50	24	22.5

*: Sodium hydroxide solution (NaOH) 12 Molarity, †: Sodium silicate (Na₂SiO₃), (Na₂SiO₃/NaOH) Mass ratio = 2.5, ‡: Superplasticiser MasterGlenium^® RMC 303, a: Extra water.

. First of all, the coarse aggregate was prepared to be in a saturated surface dry condition before being utilised in pouring to prevent the AL solution from being absorbed. The sodium hydroxide solution was prepared in the laboratory by mixing sodium hydroxide flakes with potable water, and a12 M NaOH solution was used. The AL was a uniform sodium-hydroxide-based solution prepared by mixing NaOH solution with Na₂SiO₃. To prepare the AL solution, Na₂SiO₃ and NaOH were mixed together to meet the optimum mass ratio set at 2.5. The solution was then rested for 24 h to allow the AL to reach the lab temperature. Because a standard code for the mix design of HSGC is not available, we resorted to the ACI-211 [1] and followed those guidelines to design the mix. The HSGC mixture consisted of a constant binder ratio (B) containing 75% slag and 25% FA with 74% of both coarse and fine aggregates blending and an AL/B ratio of 0.45. For the HSGC specimens, the crushed sand and the aggregate were mixed for 2–4 min using a high shear capacity concrete mixer, and then the FA and slag were added, followed by another 2–4 min of mixing. Then, AL was added to the dry mix, and the wet components were mixed jointly for approx. 3–5 min. Finally, a high-range water-reducing superplasticizer was added gradually with continuity in mixing over approximately 3–8 min to obtain a homogeneous mixture. The steel fibres were used to prepare the S-HSGC mixture with a volume fraction of 1%. The fibres were soaked in water to melt the glue and to disconnect them from the fibre pieces, and then they were dried at 40 min to be ready to mix. The steel fibres were spread throughout the mixture while being continuously mixed to achieve a homogeneous mix. The material mixing procedures and the production of HSGC and S–HSGC samples, as well as the casting processes, are illustrated in Figure 2.

2.2. Specimens’ Preparation and Casting

According to ASTM C39-18 [36] test standards, 100 × 200 mm cylindrical specimens were selected to carry out all tests. The test samples were divided into four heat-exposure groups. A total of forty-eight standard test specimens were cast for each mixture, obtaining, altogether, ninety-six samples, as every mixture includes a group of nine identical samples for every heating stage to conduct the tests. The first group was tested under ambient conditions, while the other three groups were subjected to various fires of 275, 560, and 825 °C. To obtain the thermal distributions in both the HSGC and S–HSGC specimens, several thermocouples (K–type) were installed at the cores of the cylindrical samples before the casting process to record the temperature over time, as schematically drawn in Figure 2a. Special techniques were used to keep the thermocouple in the core of the samples, as displayed in Figure 2b. Finally, to maintain the high accuracy of the results and to prevent significant variations between the sample values in this work, all samples of each group were poured using a single mixing batch for both HSGC and S–HSGC. The cylindrical test specimens were covered directly after casting with heavy-duty nylon bags. Then, after, the specimens were demoulded and kept at an ambient temperature of 21 to 24 °C, allowing them to self-cure until the test day at 28 days.

2.3. Fire Exposure Details

The fire exposure tests were conducted in an electric programmable kiln equipped with heaters on four sides in addition to the ceiling and a maximum heating capability of 1350 °C, a heating rate of 7–18 °C/min, and a 1350 × 1300 mm cross-section × 600 mm height. In ordinary HSC elements, explosive spalling begins between 250 and 500 °C due to water content and low permeability, which increase the chances of not only explosive spalling but also high drying shrinkage [37]. In previous research, no problems have been reported regarding explosive spalling in GPC [38]. However, the potential for explosive spalling is still a concern, though the authors used a steel grid box to prevent any damage to the electric furnace heaters, as illustrated in Figure 3a.

There are two categories of heating processes that are generally available to study the effect of high temperatures on concrete characteristics. One is a steady rate of heating, and the other is heating per a fire standard temperature (rise time) curve corresponding to a fire. The steady rate of heating is generally used for studying the characteristics of construction materials after exposure to high temperatures, termed the residual characteristics [39].

In this work, a heating rate of 6–15 °C/Min was used to reach the target temperatures, which were then maintained constantly to ensure that the cores of the heated samples would reach the maximum target temperatures [40]. The ignition of the fires was started at 23 °C (ambient lab temperature) without the presence of any pre-heating of the test samples, and the samples’ temperatures were close to the lab ambient temperature when placed in the furnace in order to keep them moist and increase the chance of explosive spalling, which provides a simulation close to reality and a highly aggressive level. Once the target temperature is attained, the furnace temperature is stabilised for around 120 min to allow for heat saturation and to achieve the thermal steady state. After that, it is allowed to naturally cool down to reach the lab climate.

Moreover, the selected maintenance time (120 min) has a significant impact on the explosive spalling of concrete exposed to high heat, which can contribute to increasing the chance of the condition occurring. According to Chena and Liu [41], the first explosive spalling in the HSC samples developed when the temperature approached 400 °C; when the peak was maintained for 40 min with the increased maintained time of peak temperature, a series of explosive spalling occurred serially.

These fires were chosen after analysing many actual electric, hydrocarbon, and car fires, as well as fire reports [42]. Therefore, the utilised fires were selected depending on the fire scenario in underground parking garages, as reported by Bamonte and Felicetti. The chosen fire technique is realistic for imitating the peak temperatures caused by a fire accidentally in a parking garage. As reported by Bamonte and Felicetti, the maximum heat generated by six cars simultaneously igniting in a parking garage is between 500 and 600 °C [43], with the addition of a more extreme fire scenario to the current study, which is an 825 °C scenario to expand the investigation and, at the same time, to investigate changes in microstructures. Additionally, the approved fire scenarios were compared with Eurocode 2 (EN 1992-1-2:2004) [44] and ASCE Manual No. 78 [45], as it was found that the approved fires are more conservative, as illustrated in Figure 4. It is important to note that the selected heating rate is below the ISO-834 standard [46] and ASTM E119 requirements [47]. The ISO 834 standard thermal repercussions, however, were shown to be more severe than those generated by real fires [48,49].

During the fire simulation preparation, the samples were placed in fixed locations within the furnace to ensure even heat distribution and to prevent any heat radiation between samples. Figure 3a shows the furnace dimensions, the sample-occupied locations, and the steel grid box. The K-type thermocouple sensors positioned in the core of both the furnace and the cylindrical samples were linked to an electronic high-precision data acquisition system to monitor and record the thermocouple signals, as illustrated in Figure 3b. The last stage included programming the furnace according to selected fires and maintaining the peak temperature consistently for approximately (thermostatic time) 120 min. Afterwards, the fire exposure process ended, and all test samples were allowed to cool down naturally at room temperature by opening the furnace ventilation window and then a quarter of the door. The heating and cooling regimes of the specimens are plotted in Figure 4. Subsequently, the compressive strength-bearing capacity tests were carried out to determine the residual strengths.

2.4. Test Methods

First of all, the visual appearances of the samples after each fire exposure were examined. The compression test was carried out in accordance with ASTM C39-18 [36] using a universal machine with a 3000 kN test size. The ends of the cylinder had been formed using sulphur mortar to give parallel, smooth surfaces so as to obtain a uniform distribution of load on the top and bottom surfaces. Figure 5 shows a specimen in the compressive test setup and the locations of the LVDTs. In all tests, nine samples were tested for each fire exposure stage in addition to the unheated stage, and the average results of three samples for each test are reported and discussed in this work. The transfer tests were performed after fire tests. The specimens were weighed before and after each fire exposure in order to obtain the weight change during fires and determine the densities. The density loss ratio at each fire was calculated according to Equation (1).

$$Density loss\left(\%\right)=\frac{{\rho }_{24}-{\rho }_T}{{\rho }_{24}}\times 100$$

(1)

where ${\rho }_{24}$ and ${\rho }_T$ are the initial and final densities before and after fire exposure, respectively.

Due to the importance of water absorption capacities tests on ordinary concrete elements, many studies have been conducted [50,51,52,53]. However, there is a need to study the effect of fires on the water absorption capacities of GPC. As we know, every fire has starting and ending points, and there are many ways to extinguish fires; one of the essential methods is to use water. Thus, water absorption is a crucial factor in evaluating the quality of concrete before and after the fires. The total water absorption tests were conducted according to ASTM C 642-21 [54], which is considered a more conservative test.

They were performed by oven-drying the exposed-to-fire and non-exposed specimens in the furnace at 105 ± 5 °C for 24 h until a steady dry mass was obtained by weighing to determine (W_d). After they were completely soaked for 24 h, the specimens were then allowed to drain for two minutes before wiping off noticeable surface water using a damp cloth and weighing again to determine (W_s). The water absorption of each group was calculated using Equation (2). The water absorption results determined in this study are the average of three specimens of each group for all mixtures.

In order to investigate the composition microstructure, the small pieces from the tested samples of both the HSGC and S–HSGC without exposure and with exposure of 275 °C, 560 °C, and 825 °C were carefully chosen for further analysis.

$$Water Absorption \left(\%\right)=\frac{W_s-W_d}{W_d}\times 100$$

(2)

where W_s and W_d are the masses of the saturated and dry concrete specimens.

The microstructure characterisations were performed with the aid of a high-resolution field emission scanning electron microscope with a sub-nanometre resolution (ZEISS–Gemini SEM 300) with an Oxford instruments Energy Dispersive Spectroscopy (EDS) detector with variable electron beam intensity, as shown in Figure 6. The specimens after exposure to the fires as well as the ambient (unheated) specimens were cut and polished into small pieces with approximate dimensions of 10 × 10 mm. Then, after, three pieces from the surface, core, and bottom for each sample were selected and dried at 55–75 °C for 72 h. Following the drying time, the specimens were placed in the test device and then vacuumed for 24 h. Chemical elemental analyses were also performed with a built-in energy-dispersive X-ray spectroscopy (EDX) detector on the whole sample area or selected spots of the detected specimens. Specific concerns included the elemental compositions of the individual particles and the geopolymerization gel bonding particles as well as the ITZ, the interface of the binder–aggregates–fibres.

3. Experimental Results

3.1. Time–Temperature Profile Response

Figure 7a–c illustrates the temperature as a function of the fire exposure time, including the furnace air temperature (Ta) and the sample core temperature (Tc), for fire exposure scenarios of 275, 560, and 825 °C, respectively. The furnace could reach the soaking target temperatures with a slight error of 2 to 12 °C, depending on the temperature level. Although the surface temperatures of the HSGC and S–HSGC commonly evolved in a parallel manner, it seemed that their core fires’ heat evolved in slightly varying ways, whereas the heat had transferred softly in the samples. As predicted, the cores heated or cooled at a slower rate than the surrounding furnace air. However, the soaking period was long enough for the entire specimen to reach a uniform homogeneous heat. The target exposure heats were almost reached at the core of the cylinders for all of the various target fires, except the 275 °C group. In the meantime, the heat was slowly growing and continued building up into the 275 °C group cores; however, the cores were unable to reach the target heat, and the core temperatures barely reached 154 and 158 °C for HSGC and S–HSGC, respectively, as illustrated in Figure 7a. This is attributed to the dispersion that occurred from the evaporation of free water and water generated during the geopolymerization reaction in addition to small amounts of gel water [55], which lasted until the end of the fire exposure period. It is crucial to remember that the coarse aggregates were soaked prior to casting to reach a saturated–dry surface condition and to prevent the consumption of AL water, which could affect the mixture’s workability. Furthermore, the hardened samples were exposed to fire without preheating, which resulted in an increase in the amount of sample-free water. It could be concluded that this phenomenon most likely happens at the start of the evaporation of free waters trapped in the test sample voids and part of the chemically bound water, as reported in [3,28], at temperatures between 100 and 200 °C, also, which was confirmed in [56].

This process consumes a significant portion of the heat delivered to the cores. Thus, the quantity of heat delivered to them is decreased, which is synchronised with the appearance of a single small plateau, as can be clearly observed in the time–temperature relationships shown in Figure 7a. Some previous researchers have identified this issue, such as Noumowé and Gall [57], who reported that the centres of concrete cylindrical samples were unable to reach the target temperatures during heating up to 200 °C, and they attributed this to the escape of free water by evaporation. Likewise, Noumowé [58] examined the temperature differential of high-strength ordinary concrete at exposure temperatures of up to 200 °C with a very low heating rate. The recorded temperature lag was 90 °C. Although the heating rate was lower and the soaking time was longer than those adopted in the current study, the 90 °C temperature lag is greater than that measured in the current investigation. The different temperature lags can be due to the experimental and curing circumstances as well as the primary effect of the type of concrete mixture. It can be summarised that the geopolymeric concrete has a higher conductivity than ordinary concrete at high temperature exposure. This higher conductivity can be attributed to the increased concentration of metal ions, such as silicon, aluminium, and iron, as well as the loss of ignition in the matrix of the geopolymer binders compared to the matrix of the ordinary Portland concrete, as shown in Table 1. It can be noted that the temperature lag phenomenon was less or gradually reduced for the case of the higher temperature scenarios at 560 and 825 °C. Thus, the samples’ cores almost reached the required heat, which is clear in the time–temperature relationships illustrated in Figure 7b,c, respectively. As expected in the fire scenarios at 560 and 825 °C, the heat rose slower in the core than that of the furnace air heat, and the heating curve started to grow until it reached the 105–195 °C stage. A single small plateau was formed at 560 °C due to the evaporation of free and some chemically linked water, which reflects the dehydroxylation process of (OH–) chains and the degeopolymerisation of geopolymeric gels [55]. As can be seen in Figure 7c, the plateaus were more clearly visible in the 825 °C fire curves, and they can be reliably diagnosed. For instance, the first occurred at 121 to 160 °C, and the second occurred at 466 to 482 °C. On the contrary, ordinary Portland concrete often exhibits two plateaus in the time–temperature curve at 100–200 °C, which is attributed to free water loss. The second occurs at 300–400 °C, which is known as Ca (OH)₂ as well as C–S–H gels dissolution, which represents endothermic peaks and thereby leads to the destruction of the mixture matrix, as reported in [3,59]. In this context, the water evaporation phase has significant effects on shrinkage and generates high internal stress in ordinary Portland concrete, as previously mentioned and described in the introduction section and as reported by Kodur [60]. Davidovits [21] reported that the evaporation of the water is not the source of the destructive stresses, despite the fact that the free water makes up to about 60% of the total water content in the geopolymer matrix. At high temperatures, the evaporation of the remaining 40% of the water content contributes approximately 90% of the overall shrinkage. Consequently, the minor intensity and the large exothermic phase in the thermal curve between 200 and 700 °C are related to the slow and partial degradation of the geopolymer paste matrix caused by the release of the remaining water. The heat continues building up after the small plateau; thus, the convection is high enough in the furnace space for the incoming heat to be greater than the outgoing heat, as shown in Figure 7b,c. A similar phenomenon was also reported in previous research [61]. It can be realised from Figure 7a–c that the rising rate of the temperature was higher in the S–HSGC test samples compared to the HSGC samples. The core heat of the S–HSGC reached its peak earlier compared to the HSGC specimens. Therefore, it can be noticed that the rate of heat flow in the steel-fibre-reinforced geopolymer as well as the conductivity were higher than those without steel fibre. This increase in thermal conductivity is attributable to the fact that the thermal conductivity of steel fibre is approximately fifteen times that of non-metallic construction materials [59].

3.2. Thermal Gradients

The thermal gradient (∇T) in concrete structures plays a major role in characterising the thermomechanical response and the microstructure changes. The thermal gradients in HSGC and S–HSGC were directly calculated by high-resolution thermocouples reading in the centre and on the surface of the specimens during the various fire scenarios. In the present work, the temperature gradient in the horizontal direction was determined where it critically depends on the shortest distance of fire travelled, which is 50 mm, as illustrated in Figure 2a. The thermal gradients (∇Th) were calculated by the temperature difference (∆T), noted as (furnace air subtracted from sample core temperatures), which is divided by the distance from the sample centre to the surface for each value, as in Equation (3).

$$\nabla T=\frac{(T_a-T_c)}{0.5D}$$

(3)

where $\nabla T$, T_a, T_c, and D are the thermal gradient (°C/m or °F/m), the furnace air temperature (°C or °F), the sample core temperature (°C or °F), and the diameter (mm), respectively.

Given the importance of the thermal gradient in affecting the internal thermal stresses in the specimens and explosive spalling, two different thermal gradient relationships are presented as a function of the concrete surface heat or as a time-variant function. The resulting thermal gradients are shown in Figure 8a–c as a function of the samples’ surface temperatures and in Figure 9a–c as a function of the exposure fire time for HSGC compared with S–HSGC at various fire scenarios.

As seen in Figure 8a, it can be detected that the thermal gradients directly increased with increasing air temperature until reaching their maximum and then beginning to decrease due to the beginning of heat convection storage during the soaking period. In addition, a slight gap was observed between the HSGC and S–HSGC thermal gradient curves because of the presence of steel fibres. Therefore, it can be concluded that there was a slight decrease in the thermal gradient. The maximum thermal gradient of the 275 °C fire scenario occurred at 2520 and 2440 °C/m when the surface heat was approximately 280 and 281 °C for HSGC and S–HSGC, respectively. Afterward, it is evident that the thermal gradient started decreasing with the increase in the surface heat; nonetheless, it was unable to approach the equilibrium stage, and it continued to decrease until the end of the fire scenarios, as shown in Figure 8a. In the 560 °C fire scenario, as illustrated in Figure 8b, it can be observed that the relationship between the thermal gradient and the surface heat increased linearly until the maximum thermal gradient; after that, it was detected that the inflection stage occurred when the surface temperature had stabilised. It could be seen that the maximum thermal gradients were varying between 8722 and 8568 °C/m for corresponding surface heats of 555 and 553 °C for HSGC and S–HSGC samples, respectively. In the meantime, the thermal gradient values started to go linearly down until they approximately reached the equilibrium stage. This could be attributed to the stability of high convection in the furnace for a longer period of time. The relationships of the thermal gradient with the surface heat of test specimens exposed to 825 °C had generally similar behaviours to those in a 560 °C fire scenario, as illustrated in Figure 8c. It could be seen that the maximum thermal gradients varied between 10,472 and 9836 °C/m for corresponding surface temperatures of 789 and 785 °C for HSGC and S–HSGC samples, respectively.

It can be concluded that there were evident points in the thermal gradients’ relationships with the surface heat. The thermal gradient exhibited an almost linear increase at the start of fire ignition. Then, after, the gradient countlessly grew up into the fire-growing stage until it reached its maximum values before the soaking period. Finally, intensive gradient decrease occurred until approaching the point of surface heat stabilisation. Therefore, it effectively contributed to the heat reaching the cores of the samples at this stage, whereas the process of endothermic heat is higher than the process of exothermic heat and allowed faster heat travel until reaching the thermal stability stage. Thus, when the endothermic process equalled the exothermic process, the equilibrium stage was reached. While conduction accounts for the vast majority of the heat flow, the convective heat transfer within the sample was improved by moisture escaping to the core of the sample. These findings have also been reported in real fires at underground parking garages [43]. The maximum thermal gradient of S–HSGC is slightly lower than that of HSGC because of the presence of the steel fibre, although it is lower than the ordinary concrete thermal gradient but within the normal range.

As illustrated in Figure 9a–c, the thermal gradient between the core and the surface as a function of the exposure time inside the cylindrical samples had a more complicated history after starting and passing the peak point. Moreover, it can be seen in these relationships that there were perturbations in the rates of thermal gradient decrease that happened at varied times throughout the fire exposure. These perturbations enable us to approximately pinpoint the concrete phase change, which is related to the vaporisation and movement of free and chemically bound water throughout the fire exposure.

It can be detected from Figure 9a–c that the maximum thermal gradient occurred at each fire after 28, 78, and 108 min from the fires’ ignition points for the 275, 560, and 825 °C fire scenarios, respectively. As seen in Figure 9a, it can be clearly noticed that the progression of evaporation of free water after the thermal peak began after 28 min from the fire’s ignition, which is in sync with the thermal gradient rate of 35 °C/m/min. Furthermore, the evaporation process continued to be clearly observed, and until the 150th minute, the perturbations also appeared, corresponding with a rate of 14 °C/m/min. These perturbations could be attributed to the latent heat consumption associated with free water evaporation. In Figure 9b, it can be observed that the perturbations, which represent the evaporation of free water and vapor escape to the core, occurred before reaching the peak thermal gradient, and it almost identically started with the 560 °C fire stage at 28 min after fire ignition. However, the thermal gradient increased until it reached the peak after 78 min of exposure with a record rate of 106 °C/m/min, and then it achieved the equilibrium stage at 223 min of exposure.

The development of thermal gradient curves at the 825 °C fire stage is displayed in Figure 9c. It can be detected from the figure that the thermal gradients increased with time until it reached the peak at 106 min with a rate of 99 °C/m/min, and then it continued to decrease after the peak stage until reaching the thermal equilibrium stage at 269 and 257 min from the ignitions points for HSGC and S–HSGC, respectively. Generally, the thermal gradients between the core and furnace heats, as a function of the fires’ exposure time for, together, HSGC and S–HSGC, decreased with the temperature increase, and the thermal gradients of HSGC were a bit higher than those of the S–HSGC. This may be linked to the presence of steel fibre reinforcing in geopolymer, which aided in minimising fracture formation and propagation as well as producing a denser internal matrix.

3.3. Thermal Gradients and Saturation Degree at the Heat Peaks

The maximum heat values of the samples’ cores and the furnace air heat corresponding to them were diagnosed for each mixture (HSGC and S–HSGC) at various fire exposures, and their cores’ heat saturation degree (SD) and thermal gradients at heat peak (∇Tp) were also determined; the details and values are presented in

Table 4. Maximum heat values of air and cores, thermal gradients, and heat saturation degrees at peak.

Mixture Code	Unit	HSGC				S-HSGC
Temperatures (T)	°C	24	275	560	825	24	275	560	825
	°F	75	527	1040	1517	75	527	1040	1517
Furnace air heat at peak (Tap)	°C	24	280	566	832	24	281	565	834
	°F	75	536	1051	1530	75	538	1049	1533
Sample core heat at peak (Tcp)	°C	24	154	518	775	24	159	524	805
	°F	75	309	964	1427	75	318	975	1481
Heat Peak Saturation Degree † (SD)	%	100	55	92	93	100	57	93	97
Thermal gradients at the heat peak ‡ (∇Tp)	°C/M	832	2520	960	1140	834	2440	820	580
	°F/M	1530	4536	1728	2052	1533	4392	1476	1044

^† S_D = [Sample Core Temperature/Furnace Air Temperature] × 100, ^‡ ∇T_p = [Peak Furnace Air Temperature—Peak Sample Core Temperature]/0.5 × Sample diameter, D: Sample diameter is 100 mm.

. In order to conveniently compare the steel fibre effect, the degree of saturation was used to plot the relationship between different fire scenarios and the heat saturation degree, as illustrated in Figure 10a. The degree of saturation was calculated simply by dividing the temperatures inside the specimens’ cores at their peak by the corresponding heat of the furnace air. The degree of heat saturation was 55%, 92%, and 93.1% for HSGC and 57%, 93%, and 97% for S–HSGC, respectively. A previous study [47] showed that the recorded degree of saturation ranged from 39 to 41% when the concrete samples were exposed to the ISO–834 fire standard. Therefore, this confirms that the fire scenarios used in the current study were more aggressive and contributed to better understanding the variance between the thermal behaviours of geopolymeric and conventional concretes. The larger the saturation degree, the higher the heat transfer rate, and vice versa.

The steel fibre mixture had a slightly higher range of SD values than the normal GPC, particularly in the 560 and 825 °C fire scenarios, which means that it is one of the influential factors in the transfer of heat. Furthermore, the obvious decrease in ∇Tp when steel fibres were used is present in Figure 10b, where the ∇Tp difference between HSGC and S–HSGC at the 560 °C and 825 °C fire scenarios can be more clearly observed. The S-HSGC samples had a lower gradient and a higher saturation degree than the HSGC samples. This is due to the fact that steel fibres in the concrete act as good thermal conductors when compared.

3.4. Spalling and Visual Appearance

No explosive spalling was observed in the surface after various fire scenarios of the heated specimens of both HSGC and S–HSGC. It seems that the water vapour pressure and the internal thermal pressure resulting from it were lower than the splitting tensile strength for both of the tested GPC samples. This attribute is indicated by the high bonding between the geopolymer gel and other mixture composites, as well as with steel fibres. Also, this helped the heat transfer occur in a stable manner through the geopolymer matrix. On the other hand, it indicates that spalling is not predominant in geopolymeric test samples owing to the decreased thermal incompatibility between the geopolymeric matrix and its aggregate components, as described earlier [29]. Figure 11 displays the sample’s visual appearance changing after exposure to various fires within the ambient samples as a control. On the other hand, according to Chen and Liu [41], the classical HSC suffered explosive spalling inside the furnace at 400 °C and severe spalling at 600 °C, and there was evidence of significant rupture on the sample surfaces and splattering fragments; at 800 °C, most of the samples completely exploded. Ju et al. [62] also observed explosive spalling of samples inside the furnace.

The dark green colour was observed on the samples at 24 degrees; the dark green colour of the test specimens may be associated with the crystalline phase of gehlenite, as reported by Aziz et al. [63]. It was noticed that after exposure to 275 °C, no important variations were detected on the outer surface or in the colour of the specimens. However, the variation in colour was noted, and eye-visible cracks were observed after exposure to 560 °C. The colour of the samples changed from dark green at room temperature to a lighter grey at 275 °C and exhibited a slightly blackish grey colour at 560 °C, but they changed to light yellow at 825 °C owing to the dehydration of the slag directors under high heat. Also, the colour change is attributed to the loss of free water and some of the chemical water, as well as the phase transformation of geopolymer. At 825 °C, a significant increase in the erosion of the samples was observed. Furthermore, eye-visible cracks and fragility in the coarse aggregate were observed in the samples’ surfaces. The cracks were likewise caused by the breakdown of the geopolymeric matrix and a change in the phase when the temperature went up, and it became worse in the 825 °C groups. In the S–HSGC at the 275 °C fire, no significant effects were observed, but in the 560 °C fires, the situation was slightly different in the samples, as the steel fibres prevented the growth of the cracks and eliminated the upgrade of the cracks from micro to macro. Moreover, in the S–HSGC samples at the 825 °C fire scenarios, there were other concerning effects that contributed to the erosion and decomposition of the geopolymeric matrix. Nonetheless, the most substantial alteration might have occurred in steel fibres owing to their probable phase transition, which is formed when the fibres are heated to around 725 °C. A phase transition could occur, and the body-cantered cubic crystalline structure of ordinary carbon steel fibres becomes a face-cantered cubic crystalline structure. This degree is a guidance and is subject to changes depending on the chemical composition and purity of the steel fibres [64]. Due to the increased packing factor of the face-cantered cubic structure, the negative impact of this change is the fibres shrinking. This leads to a loss of adhesion between the geopolymeric paste and the steel fibre surface. The partial deboning of the steel fibres was observed in the outer layer of the S–HSGC samples at 825 °C. Despite that, the fibres kept a good characteristic length, and the hooks were anchored in the geopolymer paste. The hooks contributed to enhancing the steel fibre anchor implantation in the geopolymer paste and decreasing the chance of explosive spalling.

3.5. Transfer Characteristics

3.5.1. Density at Various Fires

The HSGC and S–HSGC proportions were based on a target density of 2300 and 2450 kg/m³, respectively, tested at an ambient temperature (unheated). The 28-day densities were calculated from the average of three cylindrical samples for ambient (unheated) conditions and after each fire scenario. It can be realised that the hardened densities at ambient (unheated) conditions for both HSGC and S–HSGC were found to be acceptable when compared with high-strength ordinary concrete densities [1]. The loss in the hardened densities of construction materials assists in identifying their porosity degree because highly porous concrete will efficiently shed moisture at rising heats. The hardened densities of HSGC and S–HSGC, as well as a hardened densities loss factor, are presented as a function of the fire temperature, as shown in Figure 12a, b. In general, it was observed that the densities of the HSGC and S–HSGC samples exhibited a gradual decrease with the increase of heat. The residual densities of HSGC after exposure to 275 °C still retained, approximately, their original unheated values with a loss factor of 4.3%, and the corresponding value for the 560 °C fire scenarios was 7.3%, as shown in Figure 12a. On the other hand, the density loss ratios of S–HSGC samples after exposure to 275 °C and 560 °C were 3.3% and 5.24%, respectively, as shown in Figure 12b, where they are lower than HSGC. The density loss occurs during the fires due to the evaporation of free water as well as the chemically bonded water, which coincides with many phase changes in geopolymer mixtures. There are several types of water in the GPC that are released after exposure to heat, in addition to the absorbed water in the aggregate. Through the formation of the geopolymerisation matrix, the free water or physically bonded water are released during the reaction and evaporate between 25 and 105 °C. After that, the chemically linked or zeolitic water in the alkaline amorphous aluminosilicate hydrate gels (N–A–S–H) and the calcium amorphous silicate hydrate gels’ (C–S–H/C–A–S–H) water will partially escape between 105 and 300 °C. When the fires’ heat ranges from 300 °C to 500 °C, the hydroxylation (OH⁻) chains of the micelle on geopolymeric gels surface suffer the dehydroxylation of the (OH⁻) chains [55,65]. As illustrated in Figure 12a, b, significant density loss was observed at 825 °C, and the density loss ratios of the HSGC and S–HSGC samples were 14.8% and 11.5%, respectively. This is attributed to a dramatically increased decomposition of the (N–A–S–H) and (C–S–H/C–A–S–H) gels’ structure, which started at temperatures between 400 and 600 °C [66], in addition to the development of crack patterns and their scattering on the heated specimen surface. Supporting results were observed by Kong and Sanjayan [29], where the percentage residual density remaining after 800 °C exposure was 89%. Behera et al. [67] also found that 87% of the density was retained after exposure to 800 °C. Figure 12b illustrates that the addition of 1% of double-hooked-end steel fibre to HSGC has a slight effect on mass loss for test samples exposed to 275 °C fires, a moderate effect for the 560 °C fire samples, and a significant influence in the samples for the 825 °C fire. This indicates that the steel fibres had eliminated the progression of disintegration in the samples due to high heat shrinkage in the geopolymer matrix by acting as bridging elements, and that they restrained heat shrinkage in the geopolymeric matrix.

As seen in Figure 13, the proposed prediction model of the density loss ratio has been compared with the loss ratios obtained from selected state-of-the-art studies. Pliya et al. [68], Behfarnia and Shahbaz [69], Khaliq and Mujeeb [70], Jiang et al. [71], Vafaei et al. [51], Li et al. 2014 [72], Behfarnia and Shahbaz [69], Luhar et al. [73], and Guo et al. [74] are included. The proposed prediction models in this study were found to be more accurate than the others and lower than the HSC curves obtained by Vafaei et al. [51] and Behfarnia and Shahbaz [69]. This is again attributed to the different phase transformation of water in geopolymer matrices under the effect of fire, which does not happen in ordinary normal concrete.

3.5.2. Water Absorption

Water absorption is one of the important influences in evaluating the durability of concrete at ambient (unheated) temperatures, and it certainly has an importance after exposure to fires. For this reason, a relationship was drawn between the water absorption ratio and the various fire temperatures of HSGC and S–HSGC. These relations were also compared with the equivalent ordinary cement concrete as a control, which was reported by [51], as illustrated in Figure 14. In general, it has been noticed that geopolymeric mixers have lower absorption characteristics when compared with normal concrete. This is attributed to the primary difference in geopolymeric mixture microstructures in addition to the mechanism and speed of the geopolymerisation reaction [51]. The geopolymerisation in the geopolymer mixture, which occurs as a product of the reaction of aluminosilicate resources, includes slag and FA, which represent the mixture binder, along with the active liquid, in addition to sand and gravel, which may also contain a small percentage of aluminosilicate materials. As a result of these interactions, a homogeneous condensed high-viscosity mixture with strong adhesion is produced, with the slag reaction contributing to the establishment of internal C–S–H/C–A–S–H gels abundant at an early duration due to the high surface area of the slag, resulting in a high reactivity, and the N–A–S–H gels produced from FA contribute to the intensification of the internal structures and micropores, which fill very slowly with time. Moreover, by using a Na₂SiO₃/NaOH ratio of 2.5, the resulting continuous geopolymeric matrix is denser. As seen in Figure 14, it can be noticed that there is not much variance in the water absorption ratios at 24 °C, which represents the value at ambient heat and the heat of the 275 °C fire. The stability in the water absorption could be accredited to the slight increase in the dissolution of the reactive phase of the slag and FA gel, leading to an imperative geopolymerisation and densification of the geopolymer matrix. The water absorptions of HSGC were 1.21%, 1.25%, 6.73%, and 10.82% at ambient temperatures, 275 °C, 560 °C, and 825 °C, respectively.

In contrast, it was observed at the fires’ temperatures of 560 °C and 825 °C that the water absorption increased with the increase in the fire temperature due to the heat’s effects on the characteristics of microstructural evolution. In particular, this is attributed to the starting decomposition of the geopolymeric gels, which opened the pore system of the geopolymer paste and increased the disintegration ratio and the total porosity. These findings are similar to those obtained by Rashad et al. [66], who stated that the increased pores at high heat exposure were possibly caused by the collapse of the geopolymeric matrix caused by geopolymer decomposition, while Zhang et al. [75] attributed it to phase transformations. The water absorption values were comparatively lower in the case of S–HSGC than those of HSGC and less than those of HSC, as presented by Vafaei et al. [51]. As realised in Figure 14, the water absorption ratios of S–HSGC were 1.05%, 1.13%, 5.33%, and 9.40%, respectively, corresponding with 24 °C, 275 °C, 560 °C, and 825 °C. Also, it can be observed that the water absorption values of S-HSGC are lower than those of the concrete reinforced with 0.5% steel fibre content (SFRC), which was presented by Sideris et al. [76]. This may be attributable to the random distribution of tiny, short, discrete steel fibres that attempt to repair the majority of the microcracks, thereby minimising the number of continuous voids and eliminating the progression of disintegration decomposition in the structures of the samples. Moreover, there was a good relationship between the fires’ temperatures and the water absorption, with correlation coefficients of 0.96 and 0.98 for HSGC and S–HSGC, respectively.

Figure 14. Water absorption at different fire scenarios compared to ordinary cement high-strength concrete reports by [51].

Figure 14. Water absorption at different fire scenarios compared to ordinary cement high-strength concrete reports by [51].

3.6. Compressive Strength-Bearing Capacity

The results of the compressive strength testing of HSGC and S–HSGC are presented in Figure 15. The gradual degradation in compressive performance was observed in HSGC and S–HSGC during the increase in fire heat. This decrease was due to the beginning of phase changes in geopolymeric mixtures that included the evaporation of free water and chemical water as well as the release of hydroxylation (OH⁻) chains by partial separation from polymerisation and dihydroxylation (C–S–H/C–A–S–H) besides (N–A–S–H) geopolymeric mixture gels, which thus caused a partial decomposition in the gels. Moreover, the drying of the geopolymer’s surface layer, while the cores remained hydrated, led to softening of the (C–S–H/C–A–S–H) gels and gradually reducing the Vander-Waal’s forces between the gel particles, resulting in a decrease in the compressive strength. In regard to that, the mixture was recorded to have a higher residual compressive strength value than ordinary concrete. This matter can be attributed to the sintering process [56] as well as further geopolymerisation reaction of unreacted FA remnants to produce more FA gels, but with a limited percentage due to the incorporation of FA at 25% in this study. This result is in agreement with the slight decrease in hardened densities and water absorption ratios discussed previously. Then again, the hooked-end steel fibres’ presence demonstrated a tangible enhancement in the compressive strength residual factor by a 10.20% increment, which is higher than those without steel fibres when exposed to the identical fire scenarios. These findings indicate that adhesion between the steel fibres and the geopolymer matrix still exists, with exerted friction between them.

4. Microstructural Analyses

4.1. Scanning Electron Microscope

Firstly, a scanning electron microscope (SEM) was utilised to determine the microstructure approach of FA and slag powders. FA SEM photos illustrated spherical and circular forms with varied particle sizes, as shown in Figure 16a, whereas slag SEM images displayed irregular flake-shaped forms with angular and sharp edges, as shown in Figure 16b. The difference in particle shape and distribution contributed to the workability performance, whereas the angular form of the slag particles as well as the high reactivity when reacting with the AL formed a high-density gel at an early duration, which limited the workability. However, the presence of spherical FA particles significantly limited the intensity of the reaction and contributed to increasing workability. The combination of the particles (75% slag and 25% FA) showed good interaction stability and workability.

SEM micrographs in Figure 17a–d compare the microstructure changes as a function of fire exposure at 500×, 1000×, 2000×, and 4000× magnifications in addition to the non-heated samples. In general, it can be discovered that the geopolymerisation procedure was effective and successful, as all of the examined samples had a mostly amorphous structural network formed by aluminosilicate chains. As seen in Figure 17a, the majority of the slag and FA particles underwent geopolymerisation as a result of the AL and reaction products; however, a portion of the particles preserved their original form. It is also shown that the denser matrix inclusion of the C–S–H-C–A–S-H gels was generated faster due to the high reactivity of the slag particles and the N–A–S–H gel from the FA particles that reacted with a lower reactivity speed. It was further detected that the gels coexisted together. The homogeneous and well-compacted dense microstructure is attributed to the higher reaction products created by the slag, which has fine particles with a high surface area. The complete integration of slag into the mixture seems to increase interlineal matrix compactness and decrease the number of voids in the HSGC microstructure, as seen in Figure 17a. For these reasons, the thermal conductivity, strength capacity, and thermal behaviour of the samples were significantly improved.

As soon as the liquid phase was able to reach the FA particles, the observed open pores were immediately filled with aluminosilicate gels. The increase in the calcium complex in the dissolved binder created a reaction product of slag and FA, which contributed to the improvement in the materials’ strength. In addition, this enhancement in the microstructure matrix’s compactness resulted in reduced pore size and total porosity, as detected in the water absorption test results (as discussed previously in Section 3.5.2). Figure 17a shows that there are some unreacted or partially reacted FA particles with spherical shapes in the geopolymer matrix. The small percentage of unreacted particles that appeared was attributed to the utilisation of the solution concentration (molarity of NaOH M12) in AL.

Figure 17a also shows zoomed images of the formation of reactants in the form of intense particles along with the figuration of countless small granule-shaped reactants at the surface of the FA and slag particles. In addition, a close inspection of the ITZ revealed a good interfacial structure with strong bonds, together with a very dense interfacial transition region between the gels and the other mixture components. The high quality of ITZ allowed the internal pressure to be transferred sustainably, which improved the thermal and transfer properties. Additionally, it has been claimed that the bond strength of geopolymer paste with aggregates and steel fibre is greater than that of ordinary concrete. Fang and Zhang [77] also reported a denser ITZ in geopolymers, which is stronger than that in cement concrete. The unheated SEM image has hardly any noticeable flaws, except for some tiny cracks that can be visualised in the geopolymeric gels, which is attributed to the autogenous and drying shrinkage and high reactivity, which could be typical for slag-based geopolymer [63].

In the meantime, due to the dense matrix of the SEM samples, a slight number of tiny cracks might be induced through the drying procedure before the SEM test. As can be seen in Figure 17b, the SEM images of samples heated to 275 °C revealed a thick geopolymer gel structure and a structure with few cracks due to the fast water evaporation sideways with additional sharp inner thermal stress. The further geopolymerisation of unreacted particles was observed due to heat; however, deterioration and cracks occurred in the geopolymeric gel structure at this stage. This situation was also compatible with the deterioration of the strength capacity of the samples. Topal et al. [78] observed that at 200 °C, the geopolymeric gel microstructure started degrading. They linked the decrease in the sample strength to these deteriorations. After exposure to 560 °C, the gels shrunk and caused the microstructure of the geopolymeric paste to be more porous, and it had microcracks with wider spacing. These cracks were possibly the result of the evaporation of the chemically linked or zeolitic water in the alkaline amorphous aluminosilicate hydrate gel N–A–S–H and calcium amorphous silicate hydrate gels C–S–H/C–A–S–H starting at 105 °C until 300 °C (as detected in the temperature gradient curves). Moreover, the dehydroxylation of the hydroxylation OH⁻ chains from the microstructure matrix at 300 °C to 500 °C high heats caused an increase in the number and breadth of microcracks. As evidently shown in Figure 17c, the white needle-shaped bundles began to appear after exposure to 275 °C and increased after 560 °C. This phenomenon indicates the dehydration of calcium silicate hydrate to calcium silicates and lime [79]. Additionally, a wider ITZ between the gels and the unreacted FA particles was observed, as shown in Figure 17c. Simultaneously, there are tiny holes on the surface of unreacted FA particles that are spotted, showing that the process of sintering started at around 560 °C. The porosity of the samples significantly increased, and pores became evident after exposure to heat at 560 °C, which resulted in considerable strength capacity degradation and increased water absorption.

At 825 °C, the samples were severely damaged, and the dehydration and degeopolymerisation of the N–A–S–H and C–S–H/C–A–S–H gels continued and exhibited a transformation to a smoother texture due to re-crystallisation and viscous sintering phases, as shown in Figure 17d. These transformations can have a negative impact on microstructure compactness, resulting in weakened gel skeletons, as demonstrated by the significant reduction in skeleton density in Figure 17d. As can also be observed in the figure, the porous microstructure looked honeycomb-shaped, which confirmed the degeopolymerisation and dehydration of gels, and these findings were compatible with the increase in water absorption ratios. Moreover, it can be noticed that the unreacted particles of FA were disappearing due to the sintering reaction and partial melting; thus, the partial melting of aluminosilicate gel N–A–S–H produced from unreacted FA particles in a sodium-rich solution at 825 °C might seal small cracks and gaps, and it also improves inner particle bonding, whereas this phenomenon is limited in current research due to the low content of FA in the binder, as it depends on the amount of FA in the total binder of the mixture, as reported previously in [80].

4.2. EDX

The EDX spectrum analysis results obtained from the selected areas of samples exposed to different fire scenarios with ambient (unheated) samples showed the main chemical elements contributing to the geopolymeric microstructures, as illustrated in Figure 18a–d. At ambient temperature samples, the major chemical elements (O, Na, Al, Si, and Ca) of regular slag and FA composites, as well as C, Mg, and Fe, were found in the different spots of the SEM samples. These elements are directly connected to the geopolymerisation reactions that make the strong Si-Al bonds and Na-Al-Si bonds, in addition to the Ca-Si bonds that give the geopolymer matrix strength. The EDX results confirmed the SEM findings of samples tested at ambient temperature, and the EDX analyses demonstrated the production of N–A–S–H and C–S–H and C–A–S–H gels created by geopolymerisation–synthesis in the microstructures of samples. The appearance of a high percentage of silicon/aluminium (Si/Al) in the EDX at different areas indicates that the geopolymer matrix had a more homogeneous and close-grained structure with a high degree of compactness and good bonding characteristics (observed Si/Al levels varied between 3.61 and 3.94). In addition, the percentage of Na/Al ranged from 0.96 to 1.89 and showed good combinations of FA particles with better geopolymerisation, contributing to the increase in the product of geopolymeric gels.

The Ca/Si ratios were 1.02, 1.22, and 2.44; this ratio is in agreement with the SEM results, which can be attributed to the high content of Ca in the slag, which is responsible for the production of high-density gels by geopolymerisation of C–A–S–H and hydration of C–S–H, which contributes to the development of microstructures strength. Singh et al. [81] correlated the Si/Al ratio with the compressive strength, and they found the optimum ratio was 4 at 40 MPa. As the Si/Al ratio increases, a better homogeneity of the mixture and, therefore, a decrease in porosity are seen with improvements in strength. Similar chemical elements were seen in fire-exposed samples, and the Si/Al ratios were calculated to be 6.89, 9.94, and 3.71 for 275, 560, and 825 °C fires, respectively, whereas the Ca/Si ratios were 4.42, 13.40, and 1.81, respectively. It can be observed that the Ca/Si ratios increased significantly at 275 °C and 560 °C, which confirmed the dramatically increased dehydroxylation and decomposition of the C–S–H and C–A–S–H gel structures, as seen in SEM images. It could be concluded that the increased Ca/Si ratio causes more calcium-oriented weak bonds and causes serious damage, with a deterioration in the compressive strength capacity. Moreover, the Si/Al ratios increased until 560 °C; this was attributed to the decomposition of the aluminosilicate network structures due to the evaporation of chemical waters. At 825 °C, the Si/Al ratio dramatically decreased, which is attributed to the partial sintering process [82]. On the other hand, the Na/Al ratio was 0.43, which confirms that the sintering or phase transformation process in FA unreacted particles leads to a connecting microstructure matrix that is responsible for retained strength [82]. Niklioć et al. [83] stated that the matrix crystallisation at 800 °C formulated a highly porous structure and kept up the strength of the matrix binder, which is due to the sintering effect. The crystalline phase contributes to maintaining the cohesion of the matrix and preventing complete collapse, as supported by SEM images. These findings are in agreement with the results of the residual compressive strength-bearing capacity discussed earlier.

5. Conclusions and Further Research

In this experimental work, the influence of exposure to various fire scenarios at 275, 560, and 825 °C on the thermal behaviour, transport properties, and microstructure characteristics of plain and steel-fibre-reinforced high-strength GPC was investigated. The following conclusions can be drawn from the experimental results:

• From the cylinders’ core time–temperature profiles, it was found that the cores heat or cool at a slower rate than the surrounding furnace air. The used 2 h soaking period was long enough to reach a uniform homogeneous heat except at the 275 °C temperature. A small plateau was formed in the 560 °C scenario at the 105–195 °C stage due to the evaporation of free and some chemically linked water, and a second plateau appeared between 481 and 518 °C, which reflects the dehydroxylation process of (OH–) chains and the degeopolymerisation of geopolymeric gels. The plateaus are more clearly visible in the 825 °C fire curves, where the first occurred at 121–160 °C and the second occurred at 466–482 °C.
• The thermal gradient almost exhibited a linear increase at the start from the beginning of fire ignition, and it countlessly went up into the fire-growing stage until the soaking period. The intensive decreases occurred at the beginning of heat convection storage during the soaking period until reaching the equilibrium stage. The saturation degrees were 55, 92, and 93% for HSGC, while slightly higher saturation degrees of 57, 93, and 97% for S–HSGC were recorded for fire temperatures of 275, 560, and 825 °C, respectively.
• Unlike ordinary HSC, no explosive spalling was recorded for the plain and fibrous high-strength geopolymer specimens. The colour of the unheated samples was dark green, whereas the colours of the specimens exposed to 275, 560, and 825°C were lighter grey, slightly blackish grey, and light yellow, respectively. Furthermore, a few surfaces with visible thermal cracks were observed after exposure to 560 °C, and more thermal cracking and fragility in the coarse aggregate were detected on the surfaces of the samples heated to 825 °C. The presence of hook-end steel fibre in S–HSGC significantly improved the post-fire behaviour and also the post-cooling behaviour after 825 °C.
• The hardened density decreased with the increase in temperature. For HSGC, the losses were 4.30% at 275 °C, and they increased to 14.8% at 825 °C. On the other hand, the losses of S–HSGC after fire exposure to 275 °C and 825°C were 3.3% and 11.5%, respectively. The water absorption capacities of HSGC and S–HSGC were lower than those of high-strength ordinary concrete. The water absorption values of the HSGC ranged from 1.25% to 10.82%, whereas those of S–HSGC ranged from 1.13% to 9.4%.
• The inclusion of 75% slag with 25% FA achieved self-curing high compressive strength with dense microstructure morphology, whereas adding 1% of hooked-end steel fibres led to 21.7% strength improvement. After exposure to various fires, a gradual deterioration in compressive strength was observed, but in varying proportions. The presence of steel fibre limited the deterioration of the specimens exposed to 560 °C and 825 °C.
• SEM analysis revealed C–A–S–H and N–A–S–H to be the primary strength-contributing geopolymerisation reaction products, besides C–S–H, which was also detected. In addition, some non-reacted particles or partially reacted FA particles were also observed. At 275 °C, a thick geopolymer gel structure was revealed, and a few cracks were gradually formulated, which indicates the start of microstructures degrading. At 560 °C, the gels shrank and became more porous, and more cracks were observed. The dehydration of calcium silicate hydrate to calcium silicates and lime was observed at 275 °C and increased at 560 °C. At 825 °C, the microstructures were severely damaged, and the dehydration and degeopolymerisation of the N–A–S–H and C–S–H/C–A–S–H gels continued and exhibited a transformation to a smoother texture due to re-crystallisation and viscous sintering phases. The EDX results confirmed the SEM findings and were suitably compatible with thermal and transfer characteristics as well as compressive strength results.
• The current work is a seed for studying the thermal behaviour and characteristics of self-cured high-strength plain and fibrous geopolymer concrete after exposure to different fires. The experimental results are considered promising for the possibility of using HSGC in places prone to fires. However, in fire accidents, there is a need for further studies on the thermal behaviour of HSGC using different FA replacement ratios, as well as studies on the impacts of the various kinds of coarse aggregates on GPC fire performance. On the other hand, the investigation should be extended to evaluate the serviceability and structural performance of beams, columns, and slabs.