3.1. Experimental Design Results
Table 3 summarizes the CS and WA values for each composition defined in the experimental design. In general, it can be observed that composition 2 (30% ST
C) exhibited the highest values of CS. Based on these data, statistical analysis was performed on the results.
Table 4 presents the main statistical parameters obtained from the analysis of variance (95% confidence level) of the experimental design conducted to determine the compositions of mortars containing ST
G and ST
C as partial replacements for PC. Linear, quadratic, and special cubic regression models were fitted to the experimental data for statistical analysis. The significance and predictive capability of the models can be evaluated through the F-test (F
cal). For a model to be significant and predictive, the calculated F-value (F
cal) should be at least four to six times greater than the tabulated F-value (F
tab) in the Fisher–Snedecor distribution [
35].
For the analysis of CS, the quadratic model showed the highest values of F
cal/F
tab (6.46), R
2 (90.17%), and R
2Adj (86.90%), making it the most suitable for predicting CS results. While for WA, the special cubic model showed the best results (F
cal/F
tab = 5.41, R
2 = 92.85%, and R
2Adj = 89.79%).
Figure 5a,b show the correlation between the observed and predicted values for CS (
Figure 5a) and WA (
Figure 5b). It can be observed that the models are suitable for describing the mechanical behavior and WA of the mortars containing ST
G and ST
C as partial replacements for PC. Equation (1) represents the quadratic model obtained from the regression analysis for CS, while Equation (2) represents the cubic model obtained for the WA of the mortars.
In Equation (1), it was observed that all components (PC, STC, and STG) and their interaction (CP × STC, CP × STG, and STC × STG) are statistically significant (pvalue < 0.05, with 95% confidence) and exhibit coefficients greater than zero, indicating that they all contribute to increasing the CS values. This behavior suggests that the replacement of PC with STG or STC did not, by itself, reduce the CS values of the mortars. Thus, any observed decrease is due to PC being the component with the highest contribution coefficient (7.337) to CS. In Equation (2), the CP × STC interaction was not statistically significant (i.e., pvalue (0.3498) > 0.05). Therefore, this interaction was disregarded during model adjustment. Furthermore, it was observed that the interactions between the PC and the ST (PC × STG and CP × STC × STG) had coefficients lower than zero, indicating that these interactions favor the WA value reduction.
The response surfaces obtained from the proposed models for CS and WA are illustrated in
Figure 6a,b. In
Figure 6a, it can be observed that regions with higher ST
C content exhibited higher CS results compared to regions with higher ST
G content, indicating that the calcination of ST favored the mechanical behavior of the mortars. However, within the studied range (maximum substitution of 30% of PC with ST), all mortar compositions showed CS values above the minimum limit of 2.4 MPa established by the ASTM C270 standard [
36].
Figure 6b indicates that the highest WA values are in regions with higher ST
G content. It is known that WA is directly related to the porosity of the material, so it is likely that the incorporation of ST
G increased the porosity of the mortars.
Based on the results of the experimental design, three mortar compositions named M0 (0% ST), M30SG (30% STG), and M30SC (30% STC) were selected to study the resistance to the AAR. M30SG and M30SC were chosen because they allowed for the highest amount of PC substitution with ST, while M0 was selected for comparison purposes.
3.2. Resistance to Alkali-Aggregate Reaction
Figure 7 shows the linear expansion values measured from the M0, M30S
G, and M30S
C samples immersed in a 1N NaOH solution for 28 days. The expansion values for the M0 mortar (without ST) were below the limits specified in the ASTM C1260 [
34] standard, which sets a maximum expansion value of 0.10% at 14 days and 0.20% at 28 days. This indicates that the QS does not exhibit deleterious potential. These results are in line with the chemical composition (
Table 1) and mineralogy (
Figure 2a) of QS, which indicates that the present silica is predominantly in the form of non-reactive silica, quartz, and mica materials. On the other hand, the potassium present in QS is associated with the mica structure, which imparts a low-reactivity alkaline character [
37]. For the M30S
G and M30S
C samples, it was observed that the use of calcined ST allows for the production of materials with deleterious potential but within the acceptable range (<0.20%) established by the ASTM C1260 [
34]. However, the material in its non-calcined form presents a value within the range considered deleterious by the ASTM, indicating that its use in mortar formulation should be avoided.
The MR, M0, M30SG, and M30SC samples were exposed to a 1N NaOH solution for 28 and 56 days to assess the evolution of mechanical behavior, and their compressive strength was determined.
Figure 8 presents the CS values of the mortars after immersion in distilled water and the 1N NaOH solution for 28 days (
Figure 8a) and 56 days (
Figure 8b). Different letters indicate significant differences in CS values at a confidence level of 95%. The ascending order of the letters indicates the order of CS values, from highest to lowest. For the samples immersed in water, the CS values of M0 are statistically equal to those of MR (
p = 0.5244 at 28 days and
p = 0.6234 at 56 days), indicating that the substitution of commercial sand with QS did not cause significant changes in the mechanical behavior. When immersed in the 1N NaOH solution, the CS values of M0 were higher than those of MR (82.7% higher than MR at 28 days (
p = 0.0003) and 46.5% at 56 days (
p = 0.0103)). The partial replacement of PC with 30% ST
G resulted in a significant reduction in the CS values. Compared to the control group (M0), which does not contain tailings, the CS values decreased by 36.2% at 28 days (
p = 0.0211) and 33.7% at 56 days (
p = 0.0013). When immersed in the 1N NaOH solution, the reduction in CS values was even more significant (54.7% at 28 days (
p = 0.0283) and 37.2% at 56 days (
p = 0.0083)). On the other hand, the CS values of M30S
C were statistically equal to those of M0 at a 95% confidence level (
p = 0.9111 at 56 days in water and
p = 0.0924 and
p = 0.0670 at 28 and 56 days in the 1N NaOH solution). This result indicates that substituting PC with ST
C by up to 30% will not significantly affect the mechanical behavior of the mortars. In other words, this result suggests that ST
C is a sustainable and economically viable partial substitute for PC.
An increase in CS between the ages of 28 and 56 days was observed for the MR mortars (50.4% in water (
p = 0.0274) and 94.6% in the 1N NaOH solution (
p = 0.0008)), M0 mortars (52.7% in water (
p = 0.0046) and 55.3% in the 1N NaOH solution (
p = 0.0127)), and M30S
G mortar immersed in the 1N NaOH solution (47.1%, (
p = 0.0245)). This behavior was expected and can be attributed to the formation of calcium silicate hydrate (C–S–H) during the mortar curing process, contributing to the increase in mechanical strength. C–S–H has a three-dimensional gel structure that fills the empty spaces between the particles in the mortar, forming a cohesive and resistant structure. Over the curing time, C–S–H continues to develop, increasing its density and improving the mechanical properties of the mortar [
7,
38]. On the other hand, the M30S
G mortar immersed in water and the M30S
C mortar did not show a significant increase in CS values between the ages of 28 and 56 days (
p = 0.4714). This suggests that all the relevant reactions for gaining mechanical strength occurred during the first 28 days.
An increase in CS values was observed for the M0 mortar (39.6%, (
p = 0.0188)) and M30S
G mortar (93.8%, (
p = 0.0030)) after exposure to the 1N NaOH solution for 56 days. On the other hand, the other samples did not show significant differences in CS values (MR
p = 0.3702 and M30S
C p = 0.0508). This behavior indicates that the mortars containing tailings (QS, M30S
G, and M30S
C) developed in this study were not susceptible to the aggression of the alkaline solution, which is consistent with the results obtained in the expansion test (
Figure 7). This behavior suggests that M0, M30S
G, and M30S
C were able to withstand the negative effects of the 1N NaOH solution, which is a positive aspect in terms of the durability and performance of these materials in chemically aggressive environments.
On the other hand, it is noted that M30S
C showed higher CS values compared to M30S
G (
p = 0.0011 and
p = 0.0005 at 28 and 56 days in water and
p = 0.0036 and
p = 0.0126 at 28 and 56 days in the 1N NaOH solution). This difference can be attributed to the higher content of CaO (57.3%) present in the ST after calcination (CaCO
3 → CaO + CO
2). Previous studies [
39,
40,
41] have shown that CaO can accelerate hydration reactions due to increased alkalinity caused by the reaction between CaO and water to generate Ca(OH)
2. Consequently, increased alkalinity accelerates the dissolution of Ca
2+, Al
3+, and Si
4+ ions present in calcium-, aluminum-, and silicon-rich phases, resulting in the formation of hydration products such as C–S–H, which plays a crucial role in pore filling and strength gain. In fact, M30S
C exhibited lower porosity (17.7%) and higher density (2.19 g·cm
−3) compared to M30SG (18.9% and 2.11 g·cm
−3) (
Table 5).
Figure 9a,b show the differences in the microstructure of the mortars. In
Figure 9a, representing M30S
G, a higher presence of pores can be observed compared to M30S
C in
Figure 9b. This behavior is associated with the better packing provided by the calcined tailings, as indicated by the mortar density results (
Table 5). In fact, it can be observed that M30S
C had the highest density value among all the evaluated samples. This effect can be attributed to the smaller particle sizes exhibited by the ST after calcination (D
90 = 8.43 µm) compared to the natural tailings (D
90 = 16.38 µm). It is known that smaller particles provide a pore-filling effect, promoting better particle packing [
42,
43].
Still, in
Figure 9a, it is possible to observe the presence of elongated crystals in the form of needles, which can be associated with the ettringite phase. Ettringite is known to appear as clusters of elongated crystals that can group together, forming bundles or compact clusters. These crystals generate internal stresses and cracks around the aggregates, creating points susceptible to crack propagation. As a result, the material weakens [
35,
44]. On the other hand, in
Figure 9b, representing M30S
C, the presence of pores is less evident, the structure is more compact, and there is no presence of elongated crystals characteristic of ettringite. Therefore, M30S
C exhibits a structure less prone to crack formation compared to M30S
G. It is important to highlight that in the SEM images, the characteristic alkali-aggregate reaction gel was not observed in the pores or cracks, which is consistent with the results of the expansion test (
Figure 7), demonstrating expansion values within the acceptable limit (<0.20%).
The XRD results of the mortars immersed in distilled water and the 1N NaOH solution for 56 days (
Figure 10) confirm the presence of the ettringite phase (Ca
6A
l2(SO
4)
3(OH)
12·26H
2O), in addition to quartz (SiO
2) and portlandite (Ca(OH)
2), which are predominant phases. It is also possible to observe the presence of calcite (CaCO
3) and calcium silicate (Ca
2SiO
4) in smaller quantities. The peaks corresponding to the ettringite and quartz phases show an intensification with the exposure of the mortars to the 1N NaOH solution. This intensification is more pronounced in M30S
G. It is worth noting that the quartz present in the mortars comes from the aggregate (QS), while portlandite is formed during the hydration process of PC. Ettringite probably formed during the hydration of PC, as QS, ST
G, and ST
C do not contain sulfides in their composition (see
Figure 2a,b). There is also a reduction in the intensity of the peaks corresponding to portlandite with exposure to the 1N NaOH solution, indicating that the hydrate has been consumed. This reduction can be attributed to the chemical interactions between the alkaline solution and portlandite. These interactions can lead to the formation of more stable compounds, such as C–S–H, which is one of the main contributors to the mechanical strength development of the mortars [
38].
Figure 11 shows the TGA curves of the samples after immersion in water and after immersion in the 1N NaOH solution for a period of 56 days. It can be observed that MS30
G and M30S
C immersed in water exhibit more intense mass loss stages compared to those immersed in the solution. The most significant mass loss stages are observed in the following temperature ranges: 26 °C–385 °C, 385 °C–500 °C, and 500 °C–800 °C. These mass loss stages occur due to the dehydration or decomposition of the phases present in the mortars. The mass loss in the first region, between 26 °C and 385 °C, corresponds to the evaporation of free water as well as the decomposition of the C–S–H, ettringite (AFt), and AFm phases present [
45]. In the temperature range between 385 °C and 500 °C, dehydroxylation of portlandite occurs, which decomposes into calcium oxide and releases water (Ca(OH)
2 → CaO + H
2O). The mass loss observed in this region is related to the removal of water released during the dehydroxylation of portlandite. In the range of 500 °C to 800 °C, decarbonation of calcite takes place, decomposing into calcium oxide and carbon dioxide (CaCO
3 → CaO + CO
2). The mass loss measured in this temperature range is associated with the release of CO
2 during decarbonation [
35].
Based on the TGA/DTGA results, the quantification of the main phases present in the mortars up to 1000 °C was performed, and the results are presented in
Table 6. The percentages of the portlandite and calcite phases were calculated according to Equations (3) and (4). Regarding the C–S–H, ettringite, and AFm phases, it was not possible to calculate their individual percentages due to the overlapping decomposition peaks, as well as the simultaneous removal of free water during the decomposition process. Therefore, the percentages presented in
Table 5 for these compounds correspond to the gross decomposition, i.e., the overall weight loss in the first region (between 26 °C and 385 °C).
In these equations,
,
,
, and
represent, respectively, the molar masses of portlandite, calcite, water, and carbon dioxide;
and
are the calculated mass losses [
46].
Overall, a reduction in mass loss can be observed in the temperature range corresponding to the formation of C–S–H (26 °C–385 °C) with the partial substitution of PC by STG and STC. This behavior indicates that using 30% STG and STC as partial replacements for PC reduced the formation of C–S–H, which plays an important role in the mechanical strength gain of mortars. The reduction in C–S–H formation was more pronounced in M30SG (reduction of 44.9% and 116.3% compared to M0 when immersed in water and the 1N NaOH solution, respectively) than in M30SC (reduction of 13.6% and 43.1%). This difference justifies the lower CS values presented by M30SG. It is important to note that the mass loss values related to the C–S–H formation region are combined with ettringite and AFm compounds. Therefore, although M30SC showed a reduction in the mass loss in this temperature range (26 °C–385 °C), there was no reduction in the CS values compared to M0, indicating that many of the compounds formed in this region are C–S–H. In the CaCO3 decomposition range, a 152% increase in its content was observed in M30SG compared to M0 when immersed in water. However, when exposed to the 1N NaOH solution, there was a 30.7% reduction in this content. This increase in CaCO3 content suggests that M30SG is more susceptible to the carbonation phenomenon.
In general, it can be observed that the substitution of PC by STC did not cause significant changes in the mechanical behavior compared to the M0 sample (without tailings). This indicates that the STC can partially replace the PC without significantly compromising the mechanical performance of the mortars. On the other hand, the non-calcined ST should be used cautiously, as it reduced the CS values.