3.2. LG-MgO Characterization
The PSD in volume percentage for each LG-MgO sample under study is presented in
Table 2. Accordingly, the sample which presented the coarse size was MCB100. Meanwhile, PC8 and MCB100M samples presented similar values (same magnitude) for most of the percentages.
XRF results for the different LG-MgOs under study are shown in
Table 3. The samples with a higher content of MgO were MCB100 and MCB100M; meanwhile, PC8 presented higher values of CaO, SO
3, and loss on ignition (LOI) than MCB100 and MCB100M because this by-product was obtained as cyclone dust in the rotary kiln, and there are mineral phases that do not decompose during the calcination process, among the acid gases from the decomposition that react with alkaline particles, which explains the higher content of SO
3 determined by XRF [
31]. It can be detected that all the samples under study contained a small percentage of Al and Fe and about 3 wt.% of SiO
2.
XRD analysis was conducted for each LG-MgO sample. As expected, magnesium was present mainly as periclase (MgO) in all LG-MgO samples used as stabilizing agents. In addition to periclase, dolomite (CaMg(CO)2), quartz (SiO2), anhydrite (CaSO4), brucite (Mg(OH)2), calcite (CaCO3), and magnesite (MgCO3) were also determined as main mineralogical phases in the PC8 sample. Regarding MCB100 and MCB100M (both have the same nature), in addition to periclase, magnesite (MgCO3), quartz (SiO2), brucite (Mg(OH)2), dolomite (MgCa(CO3)2), calcite (CaCO3), and anhydrite (CaSO4) were also determined as main mineralogical phases.
MCB100 and PC8 samples were analyzed by TGA to estimate the percentage of the aforementioned mineralogical phases. Moreover, TGA results were compared with those XRF results. The TGA experiment results and the corresponding derivative weight are shown in
Figure 2, where blue and black lines are referred to as PC8 and MCB100 in the manuscript online version, respectively.
The weight loss percentage for each decomposition (
Figure 2a) and the temperature in the maximum rate of decomposition (
Figure 2b) can be observed to clarify the assignment of each mass loss step. These decomposition steps can be ascribed to the moisture (below 105 °C) and the water of crystallization loss (below 200 °C; attributed to the formation of CaSO
4·2H
2O), Mg(OH)
2 decomposition to MgO (from ≈ 250 °C to ≈ 450 °C), MgCO
3 decomposition to MgO and CO
2 (between ≈ 450 °C and ≈ 625 °C), CaMg(CO
3)
2 decomposition to MgO (between ≈ 625 °C and ≈ 750 °C), and CaCO
3 decomposition to CaO and CO
2 (between ≈ 760 and ≈ 1000 °C). The decomposition of CaSO
4 between 1100 and 1200 °C was also observed.
Considering the XRD and TGA decomposition results, the range temperature, and stoichiometry, the percentage of each compound was determined.
Table 4 presents the compound assignation of each decomposition according to the temperature at the maximum decomposition rate (T
max) as well as the weight loss percentage during the TGA test of each sample. The compounds’ weight percentages were calculated by considering, in each case, the decomposition reaction stoichiometry [
31] and the weight percentage of decomposition for each case (i.e., from
Figure 2 and
Table 4). Comparing the results obtained in
Table 3 and
Table 4, an estimation of MgO and CaO content in the samples was conducted (for instance in the case of PC8: 8.2 wt.% of CaMg(CO
3)
2 implies a 1.8 wt.% of MgO in XRF, 2.5 wt.% of CaO in XRF, and 3.9 wt.% of CO
2 or LOI in XRF). Therefore,
Table 5 summarizes the most relevant compounds in the samples under study. From the knowledge of the authors and considering the XRD phases detected, it would be concluded that the MgO amount was around 50% for PC8 and 80% in the case of MCB100 and MCB100M.
BET and the bulk density results as well as the CAT assay results are in
Table 6. According to the required time (s) to reach pH 9.0 (CAT), the reactivity of each LG-MgO was classified as follows: MCB100 is classified as medium reactive MgO, MCB100M as soft-burnt MgO (highly reactive), and PC8 is classified as dead-burnt MgO. The difference in the CAT results between both MCB100 samples should be attributed to the particle size distribution. Moreover, the greater the specific surface, the greater the reactivity.
Some of the alkali species contained in the LG-MgO upon contact with water would establish a solubility equilibrium resulting in their corresponding hydroxide. Thus, depending on the type of LG-MgO, the composition of the oxides could vary as well as their mineralogical phases, affecting the ANC curves results. In order to explain these pieces of evidence, three phases of the ANC curves and the equilibrium established in each of the samples are described [
25]. The different ANC curves for the three types of LG-MgO under study are shown in
Figure 3 and, for all the samples under study, the first phase was considered in a pH range between 12 and 10.5, the second phase was identified by a pH stabilization between 10.5 and 8.5 in a wide range of acid additions, and finally, the third phase was considered from 8.5 < pH < 4.
This first phase is controlled by the solubility equilibrium of lime (CaO), close to pH 12.2 and characterized by an abrupt drop in pH caused by an addition of a small amount of nitric acid (mol H
3O
+), or the solubility equilibrium of brucite (Mg(OH)
2) close to pH 10.5 (K
ps Mg(OH) 2= 1.8·10
–11). Accordingly, while the pH of MCB100 was controlled by the portlandite, the pH of PC8 and MCB100M were controlled by the brucite. The different behavior between the MCB100 and MCB100M samples was attributed to particle size distribution. By decreasing the particle size (i.e., MCB100M), the reactivity increased, and the free lime was completely carbonated quickly. The reaction was displaced towards the control of Mg(OH)
2 in those stabilizing agents where the initial pH was controlled by portlandite (i.e., MCB100). This is because the added HNO
3 consumed the CaO in the first stage. The second phase was controlled exclusively by the solubility equilibrium of Mg(OH)
2, which acts as a buffering agent. The third phase on the ANC results was exclusively controlled by the equilibrium of HCO
3– and CO
32– [
32]. An abrupt decrease in the pH by adding a small amount of HNO
3 is observed, due to the low weight percentage of carbonated mineral phases (see
Table 5). It is important to mention that the Al
2O
3, SiO
2, and Fe
2O
3 present in the LG-MgO samples did not affect the remediation of CONSO, and scarcely influence in the ANC curve since these last oxides remained inert and virtually unchanged in their precipitated form during the stabilization process. In the light of these results, MCB100M, MCB100, and PC8 exhibit proper ANC, since it was 18.14, 15.98, and 13.86 mol H
+·kg
−1, respectively, in order to exceed the range of pH 10.5–8.5. Then, as a conclusion for the ANC assay for the LG-MgO samples, when MgO and specific surface of particles increase, the ANC and buffering capacity per unit mass of the stabilizing agent increase, as well. However, together with a high ANC value, it is also necessary to establish a reservoir of the alkali to guarantee a pH buffering capacity in the optimum range over a long period. In this regard, LG-MgO with very low reactivity (e.g., PC8), would be required to maintain the optimal conditions over time. In addition, it must be considered that PC8 is cheaper than MCB100M. Consequently, PC8 and MCB100M were selected as optimum LG-MgOs to stabilize the contaminated soil, due to the cost and because of the reactivity, respectively. PC8 is five to six times cheaper than MCB100M.
3.3. Remediation of the Contaminated Soil
The ANC comparative results after the remediation of the contaminated soil (REMSO) are shown in
Figure 4. As expected, it is observed that the buffering capacity of MCB100M used as the stabilizing agent was higher than PC8, because MCB100M contains a higher percentage of MgO (
Table 5) and also showed higher ANC per unit mass of stabilizing agent added (
Figure 3), which provide a greater reservoir of alkaline stabilizer. Consequently, in the pH range of minimum solubility of metal(loid)s (pH ≈ 8.5–10.0), MCB100M acquired higher buffering capacity than PC8. Increasing the addition of the stabilizing agent (PC8 or MCB100M) also increases the reservoir of alkali and the neutralizing capacity of the remediated soil.
It is well-known that both redox potential and pH deeply affect and interfere with the heavy metal solubility in soil. In this case, the pH effect was more significant than that of redox potential. Therefore, this study focused on the pH-dependent evaluation. The pH-dependent metal adsorption reaction and the pH between 9 and 11 were the mechanisms to control the release of heavy metals from the soil.
Table 7 shows the REMSO BLT EN 12457-4 mean results, determined per duplicate. It can be denoted that all LG-MgOs achieved the minimum concentration of metal(loid)s, considering the threshold landfill legislation (
Table 1). The potential release of As and Sb diminished when the pH of the medium increased, showing a pH-dependent behavior of both metal(loid)s, which mainly denoted the presence of trivalent As and Sb mineral species into CONSO. In these cases, the decrease in pH-dependent metalloids is mainly due to the formation of their corresponding insoluble hydroxides. As(III) and Sb(III) present the minimum solubility forming their corresponding hydroxide when pH is in the alkaline region and high pH (pH < 11 for As(III) and pH < 12 for Sb(III)). As(III) and Sb(III) hydroxides present lower solubility than AsO
43–, SbO
3–, and Sb(OH)
6–, which are soluble. The presence of calcium in the stabilizers helps the formation of calcium arsenates and calcium antimony with low solubility. The same argument justified the leaching decrease for most of the pH-dependent heavy metals. The concentration decrease observed in Cd and Zn was noticeable—greater than 95% in both cases. Surprisingly, the decrease in Pb concentration was lower than expected, compared to previous work [
22]. Unexpectedly, the minimum solubility was obtained at pH around 8.5, by adding the minimum amount of either LG-MgO used as a stabilizing agent. By increasing the addition of LG-MgO and increasing the pH of the medium, the total Pb concentration released increased as well. At pH close to 10.5, where minimal solubility would be expected if the Pb concentration was controlled by the solubility of lead hydroxide, the lead released was roughly equal to that of the non-stabilized soil. No significant differences were found when using either of the two LG-MgOs studied. This can be attributed to the content of lead sulfate (anglesite) in the polluted soil. This lead compound is found in large quantities in the spent lead–acid batteries [
33]. In this scenario, as indicated by Visual MINTEQ (v. 3.0/3.1) geochemical modeling software version 3.1, the Pb concentration is controlled by the solubility of lead sulfate. In this case, the concentration of the chemical species Pb(SO
4)
2–2 (aq) increases by increasing the pH of the medium.
Even with the high amount of Pb contained in CONSO, both LG-MgOs succeeded in reducing around 70% the leaching content of Pb with 5 wt.% LG-MgO. Then, all values were below the inert or non-hazardous threshold landfill classification, except for Pb where its concentration is above the limits for non-hazardous waste. Thus, according to the BLT EN 12457-4, the optimal percentage for both LG-MgO was 5 wt.%.
Although the behavior of both LG-MgOs was very similar in the BLTs, the column test was tested only with PC8 as a stabilizing agent. The main reason is because of the by-products price, where PC8 is five to six times cheaper than MCB100M. Then, the PCTs (CEN/TS 16637-3) were carried out by using CONSO mixed with 5, 10, and 15 wt.% of PC8, separately. Results per duplicate are shown in
Table 8 and, in general, increasing the percentage of PC8 increases the pH value in all the eluates (Es). Unlike the BLT, when adding 5 wt.% of PC8, the pH value of eluates decreases progressively as the L/S ratio of 10 L/kg increases. The first five eluates were within the range of minimum solubility of pH-dependent metal(oids)s (pH 8.5–10.5), but upon eluate number 6 (E6), the pH value decreased until 7.60 for E7 (eluate number 7). This decrease in pH value is due to the low reactivity of PC8 (as shown in
Figure 4) and the consumption of the MgO contained in this LG-MgO (
Table 5). On the other hand, when adding 10 and 15 wt.% of PC8, the pH value increased during the first four eluates (E1-E4) until a value of 10.1 for 10 wt.% of PC8 and around 10.4 for 15 wt.% of PC8. After the E4, the pH values obtained in E7 decreased until a pH of 8.46 adding 10 wt.% of PC8 and 9.39 adding 15 wt.% of PC8. Taking into account the pH values of the last fraction number (E7), with an L/S ratio of 10 L/kg, the minimum percentage needed to maintain the pH in the range of minimum solubility of pH-dependent metal(loid)s was 15 wt.% of PC8 into the CONSO to obtain a proper REMSO, as LG-MgO acts as a buffering agent in the 9–11 pH range. Below and above this pH range, the pH-dependent metal(loid)s increase their mobility, and they may be re-dissolved and released into the aqueous media.
When comparing the BLT with the PCT using PC8 as a stabilizing agent, the pH values in the batch analysis were 8.54, 9.40, and 9.90, from 5, 10, and 15 wt.% of PC8, respectively. In contrast, for the same L/S ratio, the pH values of E7 for 5, 10, and 15 wt.% of PC8 were 7.60, 8.46, and 9.39, respectively. This fact can be justified by the low reactivity of PC8, which requires longer reaction times, or more vigorous conditions (i.e., BLT conditions) to achieve the equilibrium of solubility. Therefore, although the solid-liquid (S/L) ratio for both leaching tests were 1/10, the obtained results following both procedures did not coincide. Thus, while the BLT evaluates the maximum leaching potential of the pollutants (i.e., heavy metals and metalloids), the PCT evaluates the behavior of the remediated soil over a long period of time. In this case, the use of a higher percentage of PC8 ensures a higher reservoir of stabilizing agents and optimal pH conditions over a long period of time. Again, with the addition of 15 wt.% of PC8 and although all eluates (E1-E7) showed a pH value within the optimal stabilization range, the concentration of Pb was above the threshold set for hazardous waste, very high for the first L/S ratios. This fact must be justified by the presence of lead sulfate in the polluted soil, which controls the solubility of this heavy metal. As mentioned above, the contaminated soil was in lands which included slag from lead and/or fine filtering powder. This is the main reason for the high concentration of Pb in the first eluates (E1–E4). The rest of the pH-dependent heavy metals and metalloids showed concentrations in all eluates below the threshold set established for non-hazardous wastes, with most of these below limits for inert waste.