Dried-granular sewage sludge, collected at a wastewater treatment plant (Serviço Municipal Autônomo de Água e Esgoto - SEMAE) in São José do Rio Preto city, SP, Brazil, was incinerated by a process called uncontrolled autocombustion in a cylindrical oven (200-liter volume). About 20 kg of dry sewage sludge were placed inside the oven. Only 1 minute of gas fire was necessary to start the combustion process. A thermocouple was placed in the center of the oven to record combustion temperature. After 15 h, the combustion process ended and about 8.6 kg of SSA was obtained (43% of the dry sewage sludge mass). Figure 1 shows the maximum average temperature of 774°C, which was generally reached after 3 h of combustion. The ash (SSA) from the uncontrolled combustion was ground for 50 minutes at an SSA/ball weight ratio of 0.10. The milled SSA had a mean particle diameter of 20.28 µm, with d(0.1), d(0.5) and d(0.9) being 1.58 µm, 11.17 µm and 52.45 µm, respectively. As can be seen in Table 1, SSA was mainly composed of SiO₂ (38.28%) and Al₂O₃ (35.47%). In accordance with the XRD pattern of SSA, which was showed in Figure 5, quartz (SiO₂), anhydrite (CaSO₄) and hematite (Fe₂O₃) were the main secondary phases. The insoluble residue content of SSA was 27.20%, that since the sum of SiO₂ (38.28%), Al₂O₃ (20.72%) and Fe₂O₃ (11.27%) percentages from the SSA chemical composition was 70.17%, and considering that the crystalline phase dissolution during the insoluble residue test is very low, could be inferred that at least significant percentage of the SiO₂ was amorphous, as well as a considerable amount of Al₂O₃. The presence of amorphous phases in the SSA was pointed out by the slight deviation of the baseline between 18-32° 2θ in the XRD pattern of SSA (See section 3.2.2.).

MK was supplied by Metacaulim do Brasil™ to produce geopolymeric pastes and mortars, whose chemical composition is presented in Table 1. NaOH pellets (98% of purity) and sodium silicate solution (waterglass, 61.4% H₂O, 29.7% SiO₂ e 8.9% Na₂O) were used to prepare the activating solution. Water was employed to adjust the water/binder ratio (binder = MK+SSA) of the geopolymeric pastes and mortars. Siliceous sand from Castilho city - SP, Brazil, was obtained to produce the geopolymeric mortars with a fineness modulus of 2.05 and a specific gravity of 2.67 g/cm³.

In Step 1, three different NaOH concentrations (8 mol.kg^-1 (ε=1.6), 10 mol.kg^-1 (ε=1.3), 12 mol.kg^-1 (ε=1.0)) in the activating solution were evaluated in the geopolymeric mortars with 10 wt.% SSA and 90 wt.% MK. The compressive strengths of those geopolymeric mortars cured for 3 and 7 days at room temperature are depicted in Figure 3.

The increasing NaOH concentration lowered the compressive strength of the SSA/MK-based geopolymeric mortars for both curing times (3 and 7 days). The mortars prepared using the 8 mol.kg^-1 NaOH and ε=1.6 (MK10 8(1.6)) activating solution yielded the highest compressive strengths (30.9 and 31.9 MPa at 3 and 7 days, respectively) compared to the other geopolymeric mortars. The literature reports that a NaOH concentration of approximately 8 M is often used to produce MK-based geopolymers (34, 35). Kuenzel et al. (36) assumed that the NaOH concentration around 8 M is enough for a proper dissolution of the reactive phases of metakaolinite. Moreover, it is well-known that as higher the NaOH concentration, faster is the MK geopolymerization, though a higher geopolymerization rate could lead to a lower polycondensation degree of the geopolymer, affecting the compressive strength development (37). The findings herein agree with such understanding. Hence, the dissolution of the amorphous phase of the system SSA/MK at 8 mol.kg^-1 NaOH lead to a proper geopolymerization ratio, which results in higher compressive strength development. At the highest NaOH concentration (12 mol.kg^-1), the compressive strength loss (3.8MPa), after 7 curing days, could be related to somewhat zeolite formation from the geopolymer matrix, which can occur for MK-based geopolymer systems even at room temperature (38). The following evaluations proposed in Step 2 were made with the geopolymeric mortars and pastes using the activating solution of 8 mol.kg^-1 NaOH (ε=1.6).

The compressive strength of geopolymer mortars MK0, MK10, MK20 and MK30 at 3, 7, 28, 90 and 720 days were measured and are depicted in Figure 4. The geopolymeric mortar containing no SSA (MK0) content was set as a reference.

The reference geopolymeric mortar (MK0) presented a higher compressive strength for all the curing times (3 to 720 days), except for sample MK10, whose compressive strength was similar at 720 days. The compressive strength of the MK0 mortar was 66.9±2.9 MPa, while that recorded for the MK10 mortar was 60.7±4.3 MPa, both recorded at 720 days. According to the Tukey test, the difference between the average compressive strength of these mortars (MK0 and MK10) was not significant at the 0.05 level.

Furthermore, the increasing SSA content in the geopolymeric mortars decreased the compressive strength for the first and last curing ages. Although SSA had displayed a reactive and synergic behaviour in Portland cement systems in accordance with previous works (22,28), it showed different behaviour in the alkali-activation system. According to Cheng et al. (39), MK-based geopolymer develops high compressive strength, mainly because of the large surface area and high reactivity of MK, which lead to greater geopolymerization due to increasing dissolved aluminosilicate phases. Zhu et al. (40) evaluated the replacement of MK by other material as reactive as MK, in thermal-treated geopolymer production, that was RHA. According to the authors, a replacement content of MK by RHA in 20% offered a significant compressive strength enhancement around 62.5% and 21.7% at 7 and 28 curing days at 50°C, respectively. In such a study, the compressive strength enhancements were attributed to the enrichment of the content and nature of the gel due to the dissolution of the silica from RHA. In this current work, increasing MK replacement with SSA, that presented a lower content of amorphous phases containing alumina and, mainly, silica compared to MK, reduces compressive strength likely due to a sum of a dilution effect and a changing of the nature of the gel formed, which could occur when the SiO₂/Al₂O₃ and Na₂O/Al₂O₃ ratios of the system are changed (41). Such effects could explain the decrease of the compressive strength higher than the MK replacement percentage in the mortars. However, at intermediate curing times, the increasing SSA content had no significant effect on compressive strength.

Although a partial MK replacement with SSA lowered the geopolymerization rate, the reaction progressed for an extended duration. The geopolymerization progress was clearly identified by an increasing compressive strength gain by rising the SSA content between 3 and 720 days. The compressive strength gain was 96.4%, 96.7% and 143.6% for geopolymeric mortars MK10, MK20 and MK30, respectively, but was only 43.9% for MK0 for the same time frame. This behaviour means that the SSA reaction rate was slower than for MK, and the geopolymerization process for SSA/MK systems required a longer curing time. Zhang et al. (42) studied the influence of MK replacement by fly ash on the reaction process in the MK-based geopolymers. According to the authors, the MK replacement of 10% by fly ash increased the reaction extent due to the lower dissolution of fly ash compared to the MK. The results obtained in the study herein are endorsed by such a report.

The XRD patterns of the geopolymeric pastes with an equivalent mix proportion of mortars MK0, MK10 and MK30 were determined at both 90 curing days and 25°C. According to the XRD patterns in Figure 5, the crystalline phases of SSA were quartz (SiO₂, PDFcard#331161), hematite (Fe₂O₃, PDFcard#130534) and anhydrite (CaSO₄, PDFcard#371496), as mentioned previously in item 2.1.1, while those of MK were quartz (SiO₂, PDFcard#331161), kaolinite (Al₂Si₂O₅(OH)₄, PDFcard#140164) and muscovite (KAl₃Si₃O₁₀(OH)₂, PDFcard#210993). Regarding the XRD patterns of the geopolymeric pastes, all the pastes presented the same crystalline phases identified in the XRD pattern of the precursors because of the non reacted phases of MK and SSA. As reported by Belmokhtar (43), the structures of muscovite and quartz are not affected during geopolymeric, which means that those phases do not react and, consequently, do not offer any contribution to developing cementing gel. This result was corroborated by the compressive strength results: the samples containing SSA had lower compressive strengths because of the low reactivity of ash. In addition, in the XRD patterns of all samples containing SSA was not identified zeolite peaks, pointing out any zeolite formation from the geopolymer matrix, which could lead to compressive strength loss. Such a result endorses that the lower compressive strength of the samples MK10 and MK30 are more related to a dilution effect and a changing the nature of gel formed due to SSA addition.

Diffuse halo signals were noted at about 16° - 34° and 19° - 32° of the 2θ angle in the XRD patterns of MK and SSA, respectively, which indicated the presence of amorphous phases (44) in the composition of the precursors. These amorphous halo signals were displaced to 18° - 38° in the XRD patterns of geopolymeric pastes MK0, MK10 and MK30. This displacement denoted the occurrence of a geopolymeric reaction, which yielded new amorphous phases as N-A-S-H gel (34, 44, 45). When comparing the XRD pattern of pastes MK0 and the others containing SSA (MK10 and MK30), no significant difference was found for pastes MK10 and MK30, except for the remaining crystalline phases from SSA.

The microstructural analyses were carried out using SEM characterization on pastes MK0, MK10 and MK30. Paste MK10 was chosen to be analysed by SEM for its low SSA content, along with paste MK30 because it had the highest SSA content. The micrograph images of pastes MK0 were taken as the reference. The micrograph images of samples MK0, MK10 and MK30 at both 90 curing days and 25°C are presented in Figure 6. At a magnification between 700 and 10000x, geopolymeric pastes MK0, MK10 and MK30 presented a dense microstructure, described in the literature as the result of N-A-S-H type gel forming by a geopolymeric reaction (46-48). The XRD analysis did not reveal the presence of zeolitic phases by the chemical evolution of N-A-S-H gel (30). The SEM micrographs confirmed the absence of crystalline zeolitic-type particles.

The non-formation of these zeolitic phases agrees with compressive strength increasing for long curing times (30, 49). There was a significant difference in the microstructure organisation among pastes MK0 (Figure 6a and 6b), MK10 (Figure 6c and 6d) and MK30 (Figure 6e and 6f). Many cementing particles (N-A-S-H type cementing gel) were observed for MK0 (Figure 6b). When a small amount of MK was replaced with SSA (MK10), the same type of particles were observed, but in smaller numbers (Figure 6d). These particles were bonded to the amorphous matrix (probably formed with N-A-S-H gel and the unreacted SSA particles). Finally, for sample MK30, fewer characteristic gel particles were observed in MK0 and the bonding matrix among them was larger. The changes in the microstructure of the cementing matrix of geopolymers due to the presence of SSA could explain the differences observed in mechanical strength development.

In order to analyze the potential contribution of SSA in the geopolymeric reaction, SSA-activated pastes were prepared (maintaining the same proportions in the activating solution and the activator/SSA ratio than those for MK/SSA systems previously described. A 28-days (room temperature cured) SSA activated paste was prepared and low strength was reached (<10MPa): this value means that SSA had a limited reactivity in the activating conditions selected. However, there is a contribution to the development of binding gels. SEM micrographs of the SSA and the activated SSA samples were taken (Figure 7). Figure 7a shows the irregular morphology of SSA particles and figure 7b shows the presence of platelet-like particles attributed to decomposed clay compounds. Figure 7c shows the microstructure of the binding gel formed by geopolymeric reaction; Figure 7d shows a detailed microstructure of the N-A(F)-S-H gel (due to the presence of iron compounds in the SSA, the gel also contained this element) and a reacted plate-like SSA particle. The gel was formed by the agglutination of nanoparticles (less than 200 nm) and its composition (EDS) presented a SiO₂/(Al₂O₃+Fe₂O₃) molar ratio of 3.6 and (Al₂O₃+Fe₂O₃)/Na₂O molar ratio of 0.62.