Three coal wastes were obtained from a coal washing plant located in Santa Catarina (Brazil): one from a conveyor belt (coal gangue; CG), one from the decantation basin (coal sludge; CS), and one from a landfill area (spontaneous combustion coal gangue; SPC).

The raw materials were dried at a temperature of 105°C ± 5°C for 24 h and the ground in a ball mill. The SPC was milled for 360 min and the CG and CS for 30 min. The different grinding times were necessary to ensure particles smaller than 75µm (determined by dried sieving). The CG and CS materials were calcined at 700°C for 1 h in a muffle furnace and after this process the samples were referred to as CGT and CST, respectively (the SPC was not calcined).

Ordinary Portland cement (OPC) (36) and metakaolin (MK) were used to obtain the cement alkaline samples. Kaolin was calcined at 800°C for 2 h to obtain metakaolin. Sodium hydroxide (NaOH - P.A. 97%) and sodium silicate (Na₂SiO₃ - P.A. with 12.7% Na₂O, 30.4 % SiO₂ and 57.0% H₂O in solution) were used in the preparation of the alkaline solution (AS).

The chemical compositions were determined by X-ray fluorescence (XRF) on a Shimadzu instrument (model 7000). As shown in Table 1, the wastes are rich in Al₂O₃ and SiO₂, with Si/Al ratios of 2.70, 1.60 and 1.90 for the SPC, CST and CGT, respectively. The loss on ignition (LOI) is associated with presence of carbon and organic material and after calcination the LOI values for CST and CGT were lower compared with SPC.

The particle size distributions of the materials were measured on a Microtac S3500 and the results are given in Figure 1. The D₅₀ value was approximately 5.7 μm for CST and MK, whereas for CGT and SPC it was 10 μm. The Blaine test determines the fineness of a material and the values obtained were highest for CST, followed by SPC and CGT.

The mineralogical content was determined by X-ray diffraction (XRD) on a Phillips X-Pert 1 analyzer, with Cu Kα1 (λ = 1.54056), a scan step of 0.02° (1°/min) and 2θ angle of 3° to 60°. The mineralogical composition in Figure 2 shows that the raw wastes were composed of quartz, muscovite and kaolinite, and after calcination the kaolinite peaks disappear due to dehydroxylation caused by thermal stress (14–15).

The compositions and molar ratios of the alkali-activated materials (AAMs) are shown in Table 2. The alkaline solution (AS) was prepared by dissolving NaOH pellets in Na₂SiO₃ to achieve a silica modulus of 1.5 (SiO₂/Na₂O). No extra water was added. The dry powders were mixed and then, in a ratio of 1:1 (powders: AS), they were mixed with the AS for 5 min. The AAMs were placed in a hermetically sealed cylindrical mold (20 x 40 mm) and cured at 50°C for 24 h. They were then demolded, sealed with PVC plastic and kept at room temperature until testing. A loss of workability was observed for all 25%OPC pastes and thus a content of 38% of extra water was added to these mixes. Reference pastes were produced using MK and OPC. The former was mixed with AS and the latter was obtained with the same H₂O/solids ratio (0.38) as the AAM systems.

The compressive strength test was performed on the samples at ages of 1, 7 and 28 days, in an Instron 2245 testing machine, with a load cell of 50 kN and velocity rate of 3kN/s. Six cylindrical micro-samples of AAM were tested at each age. The software provided the stress-strain curve, which allowed the elastic modulus to be obtained (37).

Microstructural analysis was performed on crushed cylindrical samples that were immersed in acetone for 24 h, filtered and oven dried. Samples with 100% of coal wastes and 25% of OPC, after 1 and 28 days of curing, were selected. The objective of choosing systems with OPC was to evaluate the potential for CaO to improve the mechanical strength and alter the microstructure. However, the AAMs with MK showed a better mechanical performance.

The mineralogical content was determined by XRD on a Phillips X-Pert 1 analyzer, with Cu Kα1 (λ = 1.54056), a scan step of 0.02° (1°/min) and 2θ angle of 3° to 60°. Fourier transform infrared (FTIR) spectroscopy was performed on a JASCO FTIR 4200 instrument in transmission mode, using KBr pellets (1:60, powder: KBr). Measurements were taken from 400 cm⁻¹ to 4000 cm⁻¹, with a resolution of 4 cm⁻¹. Scanning electron microscopy (SEM) was performed on a JEOL JSM-6369 LV microscope and the samples were prepared by fragmentation and gold coating.

To verify the degree of reactivity, one gram of AAM was immersed in 250 ml of HCl, with a ratio of 1:20 for 3 h, filtered, dried at a temperature of 105°C and calcinated at 1000°C for 1 h. The ratio between the residual and initial masses was then calculated, as proposed in the literature (38).

The pastes select to produce the mortars were 100CST, 75CST25OPC, 75CGT25OPC and 75SPC25OPC, the first due to its better mechanical properties and the others to observe the influence of the presence of CaO on the mortar samples. Standard sand was used [2.5/0.15 mm] (39), in a ratio of 1:3 (AAM:sand, by mass). The mortars were renamed as M100CST, M75CST, M75CGT and M75SPC, respectively. Polycarboxylate was added at content of 1%, calculated by mass of the powder material.

The AS was added to the powder mixes. The mixtures were homogenized for 5 min and then cast in cylindrical steel molds (5 × 10 cm) and protected with a plastic film. The curing process was the same as that described for the pastes.

The compressive strength was evaluated, at 28 days, using a SoloTest machine, with a stress rate of 0.5 MPa/s. The ultrasonic pulse velocity (UPV) test was performed on a Pundit Lab® test instrument (model 6.0), with 200 kHz frequency transducer, at ages of 7, 14, 21 and 28 days, as described in the literature (40). The dynamic elastic modulus (Ed) was calculated using Equation [1], with an assumed Poisson’s ratio (μ) of 0.2. The density (ρ) was obtained from the mass/volume ratio while UPV (υ) was determined by testing.

The capillary water absorption was evaluated, at 28 days, by selecting three samples of each mortar. Before the test, the samples were oven dried at 50°C for 72 h. This test consists of measuring the variation in the amount of water in a Mariotte tube (41), which is related to the water absorbed by the sample. The capillary water absorded amount is ploted as function of √t. By means of this test the sorptivity of the mortar can be calculated using Equation [2], where ‘Vt’ is the total volume absorbed (m³), ‘A’ is specimen area (m²), ‘S’ is sorptivity (m. s^−1/2), and ‘t’ is time (h). After the capillary water absorption had been determined, the samples were dried at 50° C for 24 h the water absorption test was performed (42).

The compressive strength tests were performed at sample ages of 1, 7 and 28 days and the results are shown in Figures 3, 4 and 5. The systems showed an increase in compressive strength, over the cure time. Better mechanical performance was observed for all CST-AAMs compared to CGT and SPC, which can be explained by differences in the chemical properties, fineness (particle size and Blaine results), and mineralogical content (the CST samples tend to have a greater amount of amorphous phase).

Values of 68.5 MPa, 24.6 MPa and 6.9 MPa were obtained for 100CST, 100CGT and 100SPC, respectively, at the age of 28 days. The addition of MK leads to an increase in the mechanical strength in all cases, due mainly to the amorphous phase of MK. Figure 6 shows the results for the static elastic modulus calculated from the stress x strength data. It was found that the addition of metakaolin improved the mechanical strength and stiffness, in all cases, to a greater degree than the OPC.

At an early age, the additions improved the mechanical properties, mainly in the case of the CGT cements. Mixtures of SPC-MK showed good mechanical performance only with MK contents of over 50%. Metakaolin is a more reactive material, mainly due to its chemical (Si/Al content) and mineralogical (amorphous phases) characteristics (26), which can explain its excellent results. The addition of calcium as OPC can improve the compressive strength (25), which is attributed to the microstructure formed.

Some mixtures were selected to perform reaction degree (RD) analysis and Figure 7 shows the results. The CST-AAMs present higher RDs, more than 60% for all cases and ages, which is similar to the value for 100MK. In contrast, the SPC-AAMs provided lower values. The 100CGT RD value was also below 60%, but showed an improvement when mixed with supplementary materials. These results could be related to the chemical, fineness and mineralogical properties of the materials, which may explain the mechanical properties of AAMs.

The main factors that may have influenced the mechanical results obtained for the pastes are: i) the chemical components of CST residues with a higher alumina content and Si/Al ratio (1.60) than the other residue; ii) an increase in the particle finesses, since the values for the CST used in this study were more than 40% and 200% higher compared with the SPC and CGT, respectively; and, iii) the mineralogical composition, since XRD patterns for the coal sludge waste after calcination the kaolinite peaks had disappeared, suggestion the formation of semi-crystalline phases. Also, it should be noted that the Si/Al ratio (Table 2) may be related to the mechanical properties. Mixes with the lowest Si/Al ratio reached a higher compressive strength, especially those comprised of CST with metakaolin addition.

Microstructural analysis was performed on samples with waste only and with OPC replacement (to assess the influence of calcium on the microstructural development). Figures 8, 9 and 10 show the diffractograms for selected alkali-activated cement samples. After activation, the X-ray diffractogram exhibits a halo from 20° to 35° (2θ), suggesting the formation of sodium aluminosilicate gel (N-A-S-H) (13, 38, 43). Peaks associated with crystalline phases, such as quartz, muscovita and pyrite (for CGT-AAMs), are present. The addition of OPC displaces the halo to higher 2θ values since the calcium oxide modifies the structure of vitreous siliceous phases (44). The chemical and mineralogical properties of CST can influence the microstructural development of AAMs and the amorphous phase is more clearly observed than in the case of CGT and SPC.

The FTIR transmission spectra are shown in Figures 11, 12 and 13. The presence of the broad band for O-H asymmetric stretching (3000–3700 cm⁻¹) and the H-O-H deformation band (1600 cm⁻¹), indicate chemically-bound water, which is associated with the formation of hydrated compounds (9, 45). The broad band between 800 cm⁻¹ and 1200 cm⁻¹ is assigned to the asymmetric stretching vibration of Si-O-T (T is Al or Si). The main constituents in these materials are tetrahedral SiO₄ and AlO₄ in different structural orders (46). The initial waste presents the Si-O-T band close to 1030–1050 cm⁻¹, but after alkali activation this band is shifted to lower frequencies, which indicates the formation of gel-type aluminosilicates (45).

The bands at between 800–500 cm⁻¹ are associated with the tetrahedral vibration of T-O. The interpretation of bands in this range can be difficult and the spectra show weak intensity for all samples. This range corresponds to secondary building units (SBU) and aluminosilicate systems (45, 46). The main bands are observed at 520–550 cm⁻¹ and 760–780 cm⁻¹, associated with octahedral aluminum and the stretching vibration of Si-O-T (9), respectively.

Frequencies between 470–450 cm⁻¹ are related to the bending vibration of T-O and after alkaline activation the position of this band is almost the same and its intensity is similar to that on the spectra for the initial waste (46). Weak stretching vibrations of N-O and O-C-O (1450–1389 cm⁻¹) are present, which indicates traces of carbonates (47). Strong bands in the ranges of 1027–1014 cm⁻¹, 1030–1021 cm⁻¹, and 1029–1022 cm⁻¹ are observed for CST-AAMs, CGT-AAMs and SPC AAMs, respectively. The chemical and mineralogical characteristics of CST, CGT and SPC affect the shifts in the spectra and the Si/Al ratio influences the changes observed (46). During the curing time, there is a slight change in the main band, due to the reaction processes. Initially there is a greater release of aluminum species, due to the Al-O bond being weaker than Si-O, and as the reaction continues more silicon species are available. These processes are reflected on the FTIR spectrum, firstly with a change to the lower region followed by a return to higher wavenumbers (46). No alteration in the FTIR spectra was noted with the addition of OPC, since the calcium content is low. Pyrite minerals are associated with bands lower than 500 cm⁻¹ (48), since sulfur is heavier than oxygen, and the frequencies of cation-anion vibrations occur in regions of low energy (49). In Figure 12, vibration bands can be observed at 418–420 cm⁻¹, which could indicate S-S bonds.

SEM/EDS analysis of 100CST, 100CGT, 100SPC and 75SPC25OPC was performed to identify the reaction products and their morphology. Figure 14 shows the main images selected. The EDS was carried out at 2 to 6 points for each compound. The main chemical elements are Si, Al and Na along with traces of Ca and Fe, suggesting the formation of N-A-S-H gels. The Si/Al ratio ranges from 1.5 to 6.8 and Na/Al from 0.10 to 7.0, depending on the compound. The 100CST sample (Figure 14a) showed the lowest ratios and a more homogeneous and dense matrix compared with the other samples analyzed. The presence of the Fe ion was identified, notably in the case of 100CGT (Figure 14b), with an Fe/Si ratio of around 0.10. According to Djobo et al. (50), during alkali activation the aluminum ion can be replaced by iron. The incorporation of iron can be attributed to its presence in the vitreous form in the precursors. Davidovits et al. (51) report the possibility of iron-containing ferro-siliates structures (-Fe-O-Si-O-Al-O-).

The 100SPC sample (Figure 14c) showed a heterogeneous matrix and some circular forms, which suggests the initiation of the nucleation of the gels, but not polymerization. Based on this observation, some particles may have reacted (50), producing a less dense and resistant matrix. Needle-shaped morphologies are seen for 75CST25OPC (Figure 14d), containing Na, Al, Si and traces of Ca. The mean Ca/Si ratio for this system is 0.15, suggesting the incorporation of calcium into the matrix and the formation of N-(C)-A-S-H. Figure 14e shows plate-shaped forms (13, 50), in which iron ions are present.

A ternary phase diagram (Figure 15) was produced for the chemical compositions of the precursor materials. This model was suggested by Ismail et al. (47) for FA-GGBS alkali-activated cement, considering the Al, Si and Ca contents. The right side of the graph shows the AAMs with the formation of N-A-S-H. The upper part contains the silica-rich (> 70% Si) materials, which have the lowest values for mechanical performance. The Al-rich (30–40% Al) materials in the lower part, which have better mechanical performance, are those produced with CST and metakaolin addition. Towards the left side of the graph are the materials with the addition of Portland cement, which increases the Ca/Si ratio from zero to 0.15 (approximately). Materials containing in the region of 5–20% Ca tend to incorporate this ion, forming the N-(C)-A-S-H, as evidenced by the SEM.

The dynamic Young’s Modulus (Ed) values were obtained at 7, 14, 21 and 28 days, and on the last day the samples were submitted to compressive strength (fc) tests. Figure 16 shows the results obtained. The M100CST presented higher values than the other mortars. The compressive strength values were 39.4 MPa, 14.2 MPa, 4.6 MPa and 2.6 MPa for M100CST, M75CST, M75CGT and M75SPC, respectively. Mortars with Portland cement added needed extra water, which adversely affected the mechanical performance. The chemical characteristics of the waste may have influenced the mechanical properties. For the hybrid mortars, the Ed value decreased during the curing time and the M75CGT and M75SPC specimens showed degradation. The Ed values were 21.9 GPA, 9.1 GPA, 5.4 GPA and 4.5 GPA for M100CST, M75CST, M75CGT and M75SPC, respectively.

The ultrasonic pulse velocity (UPV) is inversely related to the time a stress wave travels through a solid. Cracks, voids, density and water content are factors that can affect the UPV. Table 3 shows the results for the specimens at 7, 14, 21 and 28 days. M100CST shows a slight increase in the UPV (2.5%) over the test ages, but the M75CST, M75CGT and M75SPC present reductions of 12.1%, 13.9% and 22.4%, respectively. These results indicate the quality of the mortars, as the appearance of cracks, voids and a poorly developed microstructure may lead to a reduction in the UPV. The bulk density is similar for all mortars, but the open porosity and water absorption show differences (<10% for M100CST and >20% for the others). The UPV is directly related to the compressive strength and the compactness of the samples (52). Figure 17 shows the correlations between the UPV and the compressive strength, water absorption (w.a.) and sorptivity (S). The higher the UPV, the higher the compressive strength and the lower the water absorption and sorptivity will be.

The degradation in M75CGT and M75SPC mortars may be related to the mineralogical composition. One hypothesis in this regard is the reaction of sulfide minerals, such as pyrite, which can lead to degradation of the material. However, durability tests would need to be performed to verify this behavior.

Capillary water absorption is shown in Figure 18 and the hybrid mortars have higher absorption and lower saturation time. M75CGT and M75SPC have saturation times of less than 2 h whereas for M75CST and M100CST the saturation times are 6 h and 24 h, respectively. This behavior is related to the high open porosity presented by the hybrid mortars (> 20%). The mechanical and physical characteristics can be related, as shown in Figure 19. The mortars with higher sorptivity have higher porosity and lower mechanical resistance.

The development of microstructures and the appearance of cracks may be the main factors affecting the mechanical and physical performance. Visual surface detachment was observed for M75CGT and M75SPC, in agreement with the results obtained, mainly for the UPV, suggesting the appearance of cracks and degradation of the specimens.

The high porosity may be related to the low compressive strength (53). Hasselman (54) suggested a linear correlation between these properties for refractory materials, as presented in Eq. [3] (55), where σ₀ is the strength at zero porosity, c is an empirical constant and p is porosity.

Figure 20 shows the correlations obtained. The resulting equations are σ = -2.91p + 67.50 (r² = 0.88) and E = -1.40p + 35.41 (r² = 0.93) for the compressive strength and modulus of elasticity, respectively. These equations could aid the prediction of the mechanical characteristics of the mortars studied. Figure 21 shows the mass variation (MV) over time. Although the MV is similar for all mortars at 28 days, the mechanisms involved may differ in each case. The main factor influencing the MV is the loss of free water, but for M75SPC and M75CGT the degradation of the specimens may also play a role, as seen from the UPV analysis. In the case of M100CST, the variation may be related to the formation of the reaction gels (25), since these presented low porosity and a higher UPV.