Three different methods are employed to produce synthetic wollastonite. These methods are the wet method, the solid-state reaction method, and the liquid-phase reaction method. All of the aforesaid methods are suitable to achieve a limited aspect ratio. Therefore, special methods should be developed to achieve the highest aspect ratio (6, 39). The production of SWM with an extremely high maximum aspect ratio (44:1) developed by utilizing natural materials is described as follows: At the first step of the three-step production process, 1:1 mol CaO and SiO₂ (56.08 and 60.08 g) and water (116.16 g) equal to the total weight of these two raw materials were mixed in a ball mill at 250 rpm for 30 minutes. After this step, called the mechanochemical process, the mixture placed in Teflon was kept in the oven at 200 °C and a pressure created by the autoclave for two days. At the second step, defined as the hydrothermal autoclave process, the tobermorite mineral was obtained. The tobermorite mineral, which was dehumidified at 100 °C for 24 hours, was subjected to pre-milling. Afterward, this mineral was kept in the ash oven at 1000 ^oC for 1 hour. Thus, in the third and last step, defined as the sintering process, SWM was obtained. After final milling, this material was used in FA+GGBFS-based EGC instead of FA.

Figure 1 presents the XRD peak changes of NWM and SWM. When the XRD peaks of these materials were examined, it was determined that the peaks of SWM were largely similar to natural wollastonite peaks. In particular, it can be said that the peaks at the 20-30 angular position [°2θ] levels, where peaks with high intensity occur, matched natural wollastonite peaks. The SEM images in Figure 2(a-b) show that almost all of the SWM particles had an acicular particle structure. In the SEM image in Figure 2(b), the SWM particle with the maximum aspect ratio was detected, and the maximum aspect ratio of SWM was determined as 44:1 using Pa 1/Pa 2 (4153 mm/94.26 nm). In light of the literature’s definition of a high aspect ratio as 20:1 (6, 38, 41), it is obvious that SWM with an extremely high maximum aspect ratio was developed. Table 1 lists the CaO and SiO₂ contents of SWM and NWM in different countries. It is seen that the CaO and SiO₂ contents of natural wollastonite vary between 42-50% and 46-55%, respectively, and these values for SWM are 45.74% and 50.71%, respectively. Therefore, XRD, SEM and XRF findings proved that this product, developed with a special technique, was wollastonite.

FA, GGBFS, NaOH, Na₂SiO₃, extra water, and PVA fiber were used in the design of the FA+GGBFS-based control EGC mix (SWM0). SWM was utilized at the ratios of 3%, 6%, and 9% instead of FA. Figure 3 shows the particle size distributions of FA, GGBFS, and SWM. Table 2 contains the physical properties and chemical compositions of these materials. Na₂SiO₃ and 13 mol NaOH were used as alkali activators in the design of EGC mixes. Na₂SiO₃ (water glass), with a specific gravity of 1.399, was purchased in liquid form from the manufacturer. A liquid Na₂SiO₃ solution, with a SiO₂/Na₂O molar ratio (module of Na₂SiO₃) of 2.5, was produced from 38.5% solids and 61.5% water (H₂O). The solid part of the Na₂SiO₃ alkaline activator contains 27.56% SiO₂ and 10.94% Na₂O oxide. For composites’ homogeneity, a fiber content of 2% by volume in excess of the calculated critical fiber content has been typically used in the mix design. The surface of the PVA fiber was coated with hydrophobic oil (1.2% by weight) to reduce the fiber/matrix interface bond strength. These decisions were made through ECC micromechanics material design theory and have been experimentally demonstrated to produce good ECC properties in previous research (42, 43). Therefore, PVA fiber was used as 2% by volume in EGCs.

The pseudo strain-hardening criteria used in the micromechanical model by Li and Leung (44) have been adopted for ECC design following a J-integral energy approach suggested by Marshall and Cox (45). Previous studies on EGCs have used the micromechanical model developed by Li et al. (44, 46), providing a solid theoretical design basis (47, 48). Despite the outstanding performance of EGCs in previous studies, challenges still exist regarding the mixture design of EGCs, which includes even more factors to be considered in comparison with the traditional EGC design (21). However, most studies on EGCs used the model in question mainly on already developed mixes to explain the origin of strain-hardening behavior and sometimes for optimization purposes (49). Thus, in line with the literature, the origin point of the matrix for EGC design is based on a PVA fiber-reinforced geopolymer composite that exhibits good deflection strain-hardening behavior (50). Furthermore, the consistent rheological properties of ECC and EGC mixes are among the most important for better fiber distribution and workability. In fact, the strain-hardening behavior of these composites depends on proper fiber distribution (51).

After the samples were poured into the mold, they were cured under laboratory conditions for 24 hours. Geopolymers produced with GGBFS have cracking problems in oven curing and, therefore, generally gain strength in ambient curing. This results in the inability to fully utilize the high reactivity of CaO (27). Therefore, FA+GGBFS-based EGCs, which were kept in the mold under laboratory conditions for 24 h immediately after production, were kept in water at 60 °C until the test age due to the improvement of the higher early compressive strength compared to ambient curing at particularly high GGBFS/FA ratios (30). In all FA+GGBFS-based EGC mixes, the ratio of Na₂SiO₃/NaOH was kept constant at 2.5, and the ratio of total alkali activator to total binder was 0.33. SWM was used instead of FA at the ratios of 3%, 6%, and 9%. Table 3 presents the mixing ratios of EGCs. The mixture nomenclature is very simple. Numbers 0, 3, 6, and 9 at the end of SWM indicate the percent by weight of SWM used instead of FA.

The workability properties of the fresh EGC matrices were determined with a mini-slump flow diameter. A truncated mini-slump cone with a height of 60 mm and a diameter of 70 mm at the top side and 100 mm at the bottom side was put on a smooth plate and then filled with ECC and EGC (without/with PVA fiber) and finally, lifted upward. When the composite flow was completed, the diameter of this shape was determined by two measurements. The deformations in the slump flow of EGCs were defined as the flow diameter when the mortar stopped flowing (6, 27).

To define the compressive strength of EGCs, 50-mm cube samples were produced. A minimum of three compression cubes were utilized to determine the average strength with respect to ASTM C109 (52). The compressive test was carried out on a test device (Brand: Utest Materials Testing Equipment and Model: UTC-4731) with a loading speed of 0.9 kN/s and a capacity of 3000 kN.

Cylindrical samples with a length of 200 mm and a diameter of 100 mm were produced for the water sorptivity test. The pieces with a length of 50 mm and a diameter of 100 mm cut from these samples were utilized to obtain three samples for each test age. Figure 4(a) shows the application of this test. The samples that arrived on the test day were kept in an oven at 50±5 °C. One of the two surfaces of the samples was chosen to allow water absorption, and the side of the selected surface was covered with silicone to prevent water permeability. After being weighed as dry weight, the samples were put in a tiny ball pool with a thin water layer of 5 mm according to ASTM C1585 standard (53). The samples’ weights were measured at certain minutes after drying the wet surfaces with a dry cloth. Therefore, the water sorptivity coefficient of the samples was calculated by the amount of water absorbed over time per unit cross-sectional area. According to the test procedure, results were calculated with the determined average of the three samples’ data using the equation below:

The rapid chloride permeability test was performed in accordance with ASTM C1202 (54) on three discs with a height of 50 mm and a diameter of 100 mm, cut from a Ø150 300 mm cylinder. After curing, the samples were kept in a vacuum for 2 hours. The samples, whose vacuum process was completed, were placed in the test device with 3% NaCl on one side and 0.3 M NaOH solution on the other side, as seen in Figure 4(b). The resistance to chlorine ion penetration of EGCs was determined by the charge passing through the composite according to ASTM C1202 (54). Thus, during the 6-hour experiment, the amount of electric current passing through the disk was recorded by applying a constant potential difference of 60 V. The test results were evaluated by taking the average of the three EGC samples at 7 and 28 days.

The Cembureau method suggested by RILEM (55) was applied to measure the apparent gas permeability of EGCs at 7 and 28 days. Figure 4(c) displays the test setup. The test sample belonging to gas permeability was produced as a disc with a thickness of 50 mm and a diameter of 150 mm, cut from 300 mm thick and 150 mm diameter cylinder samples. On the test days, the samples were dried at 50±5 °C. Then, they were allowed to cool until ambient temperature was reached in a sealed container. Inlet gas pressures were applied as 150, 250, and 300 kPa. Oxygen gas was used to provide the permeating medium. Gas permeability coefficients were determined for each level of these gas pressures. As given by RILEM (55), the apparent gas permeability coefficient was determined by the average of coefficients. The coefficient can be determined by modified Darcy’s equation (Equation [2]) as follows:

Three samples for each EGC mixture were tested, and the average of them was determined as K_a.

Freeze-thaw tests of EGC mixes were performed according to procedure A specified in ASTM C666 (56). Six 400x100x100 mm prisms for the freeze-thaw test were produced from each EGC mix. All samples were demolded at the age of 24 hours and in water at 60 °C for 13 days. After 14 days, three beams for each mix were placed in the freeze-thaw cabinet. Accordingly, the samples placed in the cabinet were left completely in water and exposed to 3-4 freeze-thaw cycles within 24 hours. As seen in Figure 5, freeze-thaw cycles were carried out in 6.5 hours, with the temperature decreasing from 4 °C to -18 °C and increasing from -18 °C back to 4 °C. On the other hand, 14 days after manufacturing, three samples with a span length of 344.8 were tested under the four-point bending test up to failure, and the average mid-span beam deflection capacity and ultimate flexural strength were calculated. These results were used as control to base the pre-cycle flexural performance of the samples.

At the end of every of 50 freeze-thaw cycles, the samples were taken out of the cabinet. After being kept for 1 day, they were photographed and weighed, and the ultrasonic pulse velocity (UPV) values were measured. UPV measures the time of the first arrival of an ultrasonic wave from one side of the sample to another. The test result of UPV assesses the relative quality of concrete or geopolymer in the presence of imperfections (i.e., voids, cracks, and the effectiveness of its repairs). Then, the samples were put back into the freeze-thaw cabinet. Each time, the chambers were cleaned and filled with fresh water. All of the EGCs completed 300 cycles and were subjected to the four-point bending test, their residual bending performances were found, and flexural strength-mid-span beam deflection graphs were obtained. Furthermore, after the samples’ flexural performance was measured, the crack numbers and widths were determined with the help of a handheld microscope.

After 300 freeze-thaw cycles, the relative dynamic modulus of elasticity (RDME) of EGCs was determined with the formulae below. The first formula proposed to find P_c is given in Equation [3].

Although not explained in ASTM C666 (2015), RDME, which uses UPV measurements instead of basic transverse frequency values, can also be calculated using Equation [4] (57).

Equation [4] was used to determine the RDME in the present study.

For each EGC mix, three 285x25x25 mm bar samples were cast to monitor the drying shrinkage and the mass loss (180 days after the initial curing of 1 day in the mold and 27 days in own cure (in water at 60 °C) according to ASTM C157) (58). As seen in Figure 6(a), the samples were then kept at 50±5% RH and 23±2 °C in an environmental chamber. During this period, the weights and length changes (Figure 6(b)) of the samples were determined on certain days.

The microstructural properties of EGCs were determined with the help of samples obtained from matrices. TGA/DTA analysis was applied to SWM0, SWM6, and SWM9 mixes. TGA analysis was carried out on 15 mg powder samples in the temperature range of 25-1000 °C, under nitrogen atmosphere conditions with a scanning speed of 10 °C/min. DTA curves were drawn with the obtained data after this analysis.

Table 4 shows the average slump diameters of each mix. It is known that PVA fiber can significantly reduce the slump flow of EGC depending on the increased friction value between fibers and between the matrix and fibers, in addition to the decreased packing density (59, 60). As a result of a lot of trial and error, EGCs, with/without PVA fiber, were produced with slump flow diameters very close to each other. In other words, the slump flow diameters of EGCs with/without PVA fiber ranged from 16.4±0.2 cm to 32.4±0.2 cm, respectively. To this end, as seen in Table 3, SWM9, SWM6, and SWM3 mixes could be produced with 34%, 7%, and 3% higher water content than SWM0 mix, respectively. Therefore, it can be said that for both EGCs with/without PVA fiber, workability decreased with the increasing SWM content. It is known that FA has a smooth surface and spherical particle structure. Thus, FA increases the workability of EGCs (6). Unlike FA, the acicularity of SWM can form network structures by overlapping between each other, which results in the decreased workability of mixes (40, 61). Thus, the probability of such network structures increases with the increasing amount of SWM due to the interlocking of SWM needles (23, 35, 62). Moreover, this can be identified as a result of the interlocking of SWMs due to the acicular particle structure (6, 38). Nevertheless, it should be noted that water only improves the workability of EGCs and does not take part in the geopolymerization reactions different from the chemical reaction of cementitious composites (63). In other words, the role and existence form of water in the geopolymer are related to the way of geopolymerization. Qualitative analysis indicates that water is released during the reaction (64). The evaporated water leaves discontinuous nanopores, improving the performance of the geopolymer. Thus, water plays no role in chemical reaction processes but simply enhances the mixture’s workability during production. This differs from the hydration of cementitious composites, where water is involved in a chemical reaction (63).

Figure 7 represents the compressive strength values of cubes for EGCs with/without PVA fiber at 7 and 28 days. Low-calcium systems, such as FA-based ones, tend to generate an alkali aluminosilicate (N-A-S-H) gel with a highly crosslinked (mainly Q⁴), disordered pseudo-zeolitic structure. When GGBFS (high calcium precursor) is additionally used, C-A-S-H gel with a tobermorite-like structure is another reaction product (65, 66). These gels can coexist in binders based on blends of high-calcium and low-calcium precursors (65), although the stability of the gel coexistence at high alkalinity is still the subject of discussion (65). Garcia-Lodeiro et al. (67) found a tendency toward C-A-S-H gel as the final stable product when pH remained > 12 in mixtures of synthetic C-S-H and N-A-S-H gels. In other words, GGBFS used together with FA in the geopolymerization process indicated that the increase in strength performance was mostly provided by GGBFS with higher reactivity, the development of the microstructure, and the gel phase of geopolymers. GGBFS is the basic precursor of alkaline activation for FA+GGBFS-based geopolymers at ambient curing (68). Additionally, geopolymers with GGBFS can reach considerably higher strength with oven curing. However, the material has a cracking problem (27). The geopolymer containing GGBFS, which is heated under the effect of temperature, expands just like concrete. While composites such as concrete or geopolymer containing GGBFS, which start to dry from the surface, dry normally, since the inner part does not interact with air, it starts to dry later and expand from the inside due to the effect of temperature. This causes such composites to exert outward pressure. However, since the surface of the concrete or geopolymer containing GGBFS has already dried, it is subjected to pressure from the inside. The concrete or GGBFS containing geopolymer shell, which is dry on the outside, starts to crack, and these cracks cannot be reversed. To prevent this, the concrete is frequently irrigated, especially in hot weather. This is the main reason for the hot water curing of the FA+GGBFS-based geopolymer developed in the present study. Therefore, the researchers (27) suggested hot water curing to take advantage of the high strength performance of GGBFS. Thus, high cubes of EGCs with/without PVA fiber varied between 88.1-111.1 and 95.1-122.6 MPa at 7 and 28 days, respectively. The compressive strength of EGCs without PVA fiber, named SWM3, SWM6, and SWM9 mixes, was 6.3%, 12.1%, and 1.9% higher than that of SWM0, respectively, whereas SWM3, SWM6, and SWM9 mixes incorporating PVA fiber had 4.6%, 11.5%, and 1.2% higher compressive strength than that of SWM0 at 7 days, respectively. In comparison with SWM0 at 28 days, the increment ratio of compressive strength was determined as 6.5%, 11.2%, 2.6%, 6.7%, 11.0%, 4.0% for SWM3, SWM6, and SWM9 with/without PVA fiber, respectively. As seen in Figure 7, the addition of PVA fiber reduced bulk density by increasing flaws and interfaces in the geopolymer paste, thus decreasing the compressive strength of EGCs (69, 70). The replacement ratio of SWM has more impact on the compressive strength of EGCs. In other words, the highest compressive strength was observed in SWM6. This may be explained by the improving effects of SWM, which is used instead of FA in the geopolymer matrix, explained by the fibrous structure due to the acicular particle morphology of SWM (71). It can also be said that SWM, with the feature of bridging microcracks due to its fibrous structure, enhanced bond strength between the matrix/microfiber at the interface of EGCs (72). As seen in Figure 3 the grain size distribution of SWM was higher than that of FA, and, therefore, SWM may have increased the pore volume of EGCs. However, it is known that the pore discontinuity in the matrix provided by wollastonite occurs due to its needle-like particle structure. The main reason for this discontinuity is that wollastonite forms a pore structure where liquids cannot reach normal pressure. Furthermore, the presence of SWM in the pore system, which has no chemical effect on the pore solution due to its inert nature, is thought to affect the formality of the pore network folds of ECC with its needle-like particle structure (38, 41). This agrees with the studies by Öz and Güneş (38) and Yücel et al. (6), showing that the mechanical performance of the cementitious composite developed with 9% SWM decreased beyond this substitution level. Likewise, there was a decreasing trend with the usage of 9% SWM in EGCs, although SWM9 with/without PVA fiber has a higher strength than that of SWM0. The excessive amount of SWM in the geopolymer matrix was the reason for the decreased compressive strength since workability loss originated from the high surface area of SWM for the desired slump flow diameter of FA+GGBFS-based EGCs (41). This information was accepted by Patankar et al. (72), indicating that the excessive amount of water for geopolymerization did not have any positive effect on the strength properties of geopolymers. It was also noted that the parameters defined as fiber dispersion in the paste and the type and amount of fiber had a significant effect on the interaction between the matrix and fiber. Hence, the large amount of SWM with inert characteristics no longer contributed to the improvement of the compressive strength for FA+GGBFS-based EGCs (11).

Figure 8 presents the water sorptivity coefficient of FA+GGBFS-based EGCs with PVA and SWM. The 7-day and 28-day water sorptivity coefficients were in the range of 0.1706-0.2071 and 0.1089-0.1251 mm/min^0.5 for FA+GGBFS-based EGCs, respectively. The water sorptivity coefficients of SWM3, SWM6, and SWM9 mixes were 7.8%, 17.6%, and 12.3% lower than that of SWM0 at 7 days, respectively. On the other hand, the improvement ratio of the water sorptivity coefficient in comparison with SWM0 was determined as 5.9%, 13%, and 7% for SWM3, SWM6, and SWM9 at 28 days, respectively. Similar to the current study, Mathur et al. (73) stated that wollastonite microfibers reduced the water absorption amount of cementitious composites. Thus, it can be concluded that the lowest water sorptivity coefficient among FA+GGBFS-based EGC mixes was obtained from SWM6 mix, while SWM0 had the highest water sorptivity coefficient. As a result, it can be thought that SWM enhanced the compactness of the microstructure in the EGC matrix (73-75). Hence, a pore discontinuity could be ensured with the pore structure formation of SWM in the matrix, which liquids cannot reach at normal pressures (41, 73). Moreover, using SWM at a 9% replacement level deteriorated the permeability performance of SWM9 with respect to SWM6. The high amount of SWM in EGCs was the reason for the weakened matrix bond in the interfacial transition zone (33). As a result, the pores created by the high amount of SWM resulted in the increased water sorptivity coefficients of EGCs.

The chloride permeability values of EGCs at 7 and 28 days are graphically shown in Figure 9. The rapid chloride permeability coefficients of FA+GGBFS-based EGCs changed between 4412-5011 and 3055-3483 at 7 and 28 days, respectively. The total charge passing through FA+GGBFS-based EGCs, including SWM, was in the range of 2000-4000 C at 28 days, thus leading to the moderate classification of rapid chloride permeability in accordance with ASTM C 1202 (54). In the development of EGCs, the substitution percentage of SWM had a significant effect during the activation process, in addition to the constant parameters such as the alkali activator ratio and curing temperature. Thus, the highest rapid chloride permeability coefficient was obtained from SWM0, while SWM6 had the lowest one. The studies by Ransinchung et al. (75), Kalla et al. (74), and Kumar and Ramujee (71) also supported this result. The pore discontinuity provided by SWM, with its needle-like particle morphology, can be considered the main reason for this improvement (73). Owing to its inert structure, SWM does not enter any chemical reaction with the pore solution, which preserves its physical structure during geopolymerization. Therefore, the unique structure of SWM may affect the formality of the pore network folds of EGCs. It can also be said that the bridging capacity in the pore structure provided by SWM positively affects the pore network folds against the rapid chlorine permeability (38). However, the high replacement ratio of SWM9 caused the poor compaction of fresh EGCs. Then, the porous structure formation occurred due to the heterogeneous interaction between SWM and the matrix, which caused the deficiencies of the poor system to result in the increased rapid chloride permeability coefficient.

Figure 10 presents the details of the apparent gas permeability coefficient of EGCs at 7 and 28 days. The lowest gas permeability coefficients among EGCs were obtained for SWM6 as 5.71 x10^-16 and 4.79 x10^-16 m² at 7 and 28 days, respectively. As expected, the highest gas permeability coefficients at 7 and 28 days were obtained from SWM0 as 6.55 x10^-16 and 5.75 x10^-16 m², respectively. It is known that the durability properties of geopolymer depend on the microstructural performance of the matrix and the pore structure of the surface (39). From a physicochemical perspective, this may be explained by SWM having good adsorptive properties with the rough surface reducing the water absorption of EGCs by its reinforcing effect (39, 74, 75). Additionally, SWM can become a center of the formation relationship with FA and GGBFS particles of composites, and thus, the “constraint” mobility can continue (76). On the other hand, SWM with acicular particles, bridging ability, and inert structure, which does not have any chemical effect on the pore solution, enhanced the pore network of the material and resulted in pore discontinuity in the geopolymer matrix (74, 75). Despite this result, the poor gas permeability performance of SWM9 in comparison with SWM3 and SWM6 can be attributed to the decreasing paste strength of EGCs in the interfacial transition zone of SWM (33). Consequently, the gaps created by the weakness of the matrix resulted in the deterioration of the gas permeability coefficients for EGCs.

EGCs can deteriorate under various environmental loads (e.g., freeze-thaw cycles). However, it is expected that the permeability of EGCs can be much less in comparison with similarly strained concrete due to its tight crack width, which would be useful for the durability of cementitious composites (4). The effects of SWM amount on the mass change (mass loss) and UPV measurement of FA+GGBFS-based EGCs samples subjected to 50, 100, 150, 200, 250, and 300 freeze-thaw cycles are shown in Figures 11-12, respectively. Moreover, Figure 13 displays the appearance of SWM6 sample at the end of 0, 50, 150, and 300 freeze-thaw cycles. As a result of freeze-thaw cycles, it was determined that the presence of SWM in the mixtures of FA+GGBFS-based EGCs reduced the mass loss and UPV values. For example, after 300 freeze-thaw cycles, the mass losses of SWM0, SWM3, SWM6, and SWM9 mixes were obtained as 2.27%, 1.94%, 1.58%, and 2.15%, respectively. Likewise, the UPV measurement at the end of 300 freeze-thaw cycles of these mixes decreased by 55.19%, 54.04%, 52.23%, and 54.41%, respectively. SWM used instead of FA partially increased the void ratio of EGCs due to its larger grain size than that of FA. However, the increase in UPV measurement for SWM3, SWM6, and SWM9 with respect to SWM0 was considered evidence that SWM provided pore discontinuity in the matrix (33, 72, 73). On the other hand, an increase in the durability of EGCs can be related to the reduced permeability of the matrix system due to the formation of discontinuous pores. Furthermore, the increased porosity of the matrix with the use of SWM up to a 6% replacement level with FA is likely to accommodate frozen water (ice) without stress formation in EGC. Therefore, the array of interactions within the composite matrix, including SWM, led to the unique pore size distribution, resulting in a composite capable of resisting freeze-thaw (73). Moreover, the downward durability trend for UPV measurement and mass loss of SWM9 may be explained by the limited geopolymerization with excess SWM.

Figure 14 shows the variations of the relative dynamic modulus of elasticity (RDME) values depending on the number of freeze-thaw cycles. For all freeze-thaw cycles in every 50 cycles, the lowest RDME values were obtained from SWM0. It was determined that SWM6 performed the best in terms of RDME. Unfortunately, the durability factor (DF) value of all FA+GGBFS-based EGCs at the end of the 50^th cycle was obtained as “60” smaller than the value specified in the ASTM C666 standard (56). Therefore, the DF values of EGCs were determined as 0.

Mid-span beam deflection and flexural strength values and curves of EGCs at the end of 14 days and 300 freeze-thaw cycles are presented in Figures 15-17, respectively. According to the results, it is not difficult to state that the flexural strength of geopolymer samples is related to compressive strength. As seen in Figures 15-17, the mid-span beam deflection and flexural strength values of EGCs increased up to a 6% SWM replacement level, while SWM9 had lower mid-span beam deflection and flexural strength values than those of SWM6 but higher than those of SWM0. The mid-span deflections of SWM3, SWM6, and SWM9 increased by approximately 18.3%, 39.6%, and 28.8%, respectively, compared to SWM0 for 14 days. The flexural strength values were determined as 4.3%, 8.9%, and 2.0%, respectively. As predicted, the flexural strength of all EGCs decreased due to the damage at the end of 300 freeze-thaw cycles. This trend reached quite high values for EGCs. However, the ductility capacity of FA+GGBFS-based EGCs incorporating SWM increased significantly, contrary to expectations. Thus, SWM improved the flexural performance of EGCs up to a 6% content, just as before freeze-thaw cycles. The mid-span deflections of SWM3, SWM6, and SWM9 mixes increased by approximately 12.7%, 25.6%, and 20.5%, respectively, with respect to SWM0 after 300 freeze-thaw cycles. The flexural strength values were determined as 6.9%, 12.6%, and 3.2%, respectively. The bridging ability of SWM improved the resistance of EGCs during freeze-thaw cycles, resulting in higher flexural performance of FA+GGBFS-based EGCs. It is also evident from Figures 16 and 17 that the acicular structure of SWM ensures bridging the micro-cracks of FA+GGBFS-based EGCs (39). However, a higher content of SWM in SWM9 ensures no further improvement in flexural strength. Hence, the lacking interfacial adhesion of SWM and EGC paste reduced the flexural strength at the end of 300 freeze-thaw cycles (74, 75).

The average crack width and crack number values at the end of the 14^th day and 300 freeze-thaw cycles of EGCs are given in Figure 18. It was revealed that the number of cracks changed as the ductility of composites was enhanced by SWM. Likewise, the crack width of EGCs decreased with the increasing ductility.

The drying shrinkage and mass loss variations of FA+GGBFS-based EGCs up to 180 days of drying are graphically depicted in Figures 19 and 20, respectively. The drying shrinkage of SWM0 was determined as 2140.44 με at the end of the 180-day drying period. According to these findings on SWM9, SWM3, and SWM6, the drying shrinkage of FA+GGBFS-based EGCs decreased in the aforesaid order, but all of the EGCs incorporating SWM exhibited lower shrinkage than SWM0. Hence, the lowest drying shrinkage among FA+GGBFS-based EGCs was obtained as 2000.18 με from SWM6 after the 180-day drying period. Moreover, similar test results to the present study were reported by other studies, indicating that the bridging ability of micro-cracks and contribution to pore discontinuity, as well as the inert characteristics of SWM, were accompanied by lower shrinkage strain in SWM3, SWM6, and SWM9 in comparison with SWM0, as confirmed by the decrease in drying shrinkage with the increasing SWM content in EGCs (35, 72, 77). It can be stated that SWM, with crystal or needle grain forms, has certain roughness forms around itself, associated with surrounding materials, and forms the matrix of the main composites as micro-reinforcing. This reduces the degree of their mobility independently of one another. Therefore, the drying shrinkage of SWM6 was noticeably decreased by SWM with high adhesion (76). As seen in Figure 20, the % mass loss changes of EGCs exhibited similar trends with the drying shrinkage results, especially until the 15th day. While the highest mass loss was obtained from SWM0 as 17.07%, SWM6 had the lowest mass loss of 15.33%. However, after day 15, the drying shrinkage increased steadily despite the fact that the % mass loss remained almost constant for all EGCs. This may be explained by the fact that drying shrinkage deformation is associated with factors other than mass loss (78). Hence, mass loss alone cannot represent sufficient data on the variations of drying shrinkage for FA+GGBFS-based EGCs (79).

The TGA/DTA curves of FA+GGBFS-based EGCs are shown in Figures 21-22, respectively. According to previous studies (38, 80), the decrease in water in the N-A-S-H gel for FA-based geopolymers can be attributed to the peak occurring in the 50-150 °C range due to the reduction in mass. Moreover, for alkali-activated GGBFS-based composites, the initial mass loss at 50-200 °C indicates the presence of a C-S-H-like solid phase and dehydration in the geopolymer paste (81, 82). The N-A-S-H and C-S-H gel contents of EGCs were determined by the mass losses determined by the peak temperatures of the TGA curves in the range of 20(initial)-200(final) °C in the DTA graphs. The gel contents of SWM0, SWM6, and SWM9 were calculated as 9.27%, 8.77%, and 7.97%, respectively. The analysis results indicated that the gel contents of EGCs with SWM did not change in parallel with compressive strength. The compressive strength of SWM6 and SWM9 was higher than that of SWM0. The gel content of EGCs gradually decreased with the increasing SWM content. These results showed that SWM was not involved in the reactions between the alkali activator and GGBFS/FA in the chemical reaction zones due to its inert structure (38, 39). It can be stated that its presence in the regions prevented chemical reactions. Moreover, the decrease in FA can be thought to be another reason for the decreased total gel content. Based on the gel contents calculated with the TGA/DTA results and compressive strength values, although the gel contents decreased with SWM, its improving effects on compressive strength can be attributed to its superior physical shapes in the microstructure. Due to its acicular particle structure, SWM can bridge micro-cracks, leading to a higher load-carrying capacity (35, 38, 39). Additionally, the improvement in the durability and dimensional stability performance of EGCs can be explained by the inert characteristics of SWM. In fact, although it is predicted to increase the pore volume, SWM preserves its acicularity by not participating in the chemical reaction, thus providing pore discontinuity with this physical structure (6, 38, 39).