3.1. Slag A and Slag R Characterization
A leaching test according to the European norm EN 12457-2 was necessary to verify the hazardous nature of Slag A and Slag R. The percentages of heavy metals, such as Ba, Pb, Cu, found in the eluate were compared to the limit values prescribed to dispose of the non-dangerous waste in a landfill (
Table 3). The legal limit values proposed in
Table 4 are those of Italian DM 27/09/2010 which was transposed by European Directive 1999/31/CE (Slovenian limits are the same). The amount of heavy metals was below the regulation limits, so the leaching test confirmed the non-hazardous nature of both the slags and the possibility of using these slags as raw materials to prepare alkali-activated materials.
In order to evaluate the release of ions in the eluate of slag A and slag R and their chemical stability in aqueous environment, pH and ionic conductivity measurements were collected at different times: 0, 5, 15, 30, 60, 120, 240, 480, 1440 min after immersion in water (
Figure 3). For both slags, pH values around 10–11 remained constant during the first 24 h. The conductivity values of Slag R were not constant, increasing very rapidly in the first 2 or 3 h to reach the value of approx. 300 mS/m within the first 24 h. The sharp increase in ionic conductivity is a consequence of the release of ions in the environment from part of Slag R. Slag A showed lower pH and conductivity with respect to Slag R, so it meant that slag A was more stable and less soluble in aqueous environment than Slag R.
The alkaline attack was carried out to determine the reactive fraction of Slag A and Slag R, in terms of Al and Si ion release. In particular, the interest is in the reactive Si/Al mass ratio calculated as ratio of the amount of Si and Al measured in the eluates. This ratio is considered a reactive ratio because it takes into account only the amount of Si and Al dissolved in the alkaline solution and not the total amount of these ions in the starting materials. It has been observed by the authors [
28,
29] for other kind of slag that the amorphous fraction is the more reactive with respect to the crystalline type. Confirming previous results, Slag A, with an amorphous fraction of about 56 wt.% presented higher contents of soluble Si and Al species in the eluate (
Table 4) occurrence that is directly related to the higher amorphous fraction than the Slag R (35 wt.%). This behavior can also be correlated to the higher specific surface of Slag A (7.61 m
2/g) with respect to Slag R (3.52 m
2/g), reported in Česnovar et al. [
21] which favors the slag reactivity and dissolution. Being the amorphous fraction of the slags, the most reactive fraction in alkali solution, the Si/Al mass ratio appears to be critical for the degree of reticulation of the final alkali-activated solid material. When the Si/Al mass ratio is below the value of 3 the corresponding materials are characterized by a 3D rigid network, suitable for a concrete, cement, or waste-encapsulating medium [
28,
31]. In the present study, the Si/Al mass ratio in alkali solution was similar for both slags, thus indicating that both slags could produce AAM with good mechanical performances even though the amount of SiO
2 and Al
2O
3 differs for the two slags.
The alkaline activation induced an extremely alkaline environment, so the proposed basic approach has proved to be useful to study the chemical stability of minor components of the slags that generally have a high environmental impact: the heavy metals. In addition, amphoteric elements, such as Sb, As, and Mo, were extremely interesting to investigate because they can be easily leached out in NaOH solution. The concentration of metals in the eluate is reported in
Table 4. From these values it appears evident that the releases of heavy metals are higher with respect to those in aqueous environment (see
Table 3). This behavior can be related to: (i) the strong conditions of the test in NaOH where the solution is at 80 °C with a concentration of 8M, and (ii) the different chemical behavior of each element. Among heavy metals, Ba and Zn are also the most released elements in water—even though in lower amount—while all the other elements are released in alkaline environment only.
Mineralogical analysis was performed in order to estimate the modifications in the crystalline phases after the alkaline attack (
Figure 4). Slag A was mainly characterized by Q-quartz (SiO
2), C-calcite (CaCO
3), and D-dolomite CaMg(CO
3)
2 and merwinite Ca
3Mg(SiO
4)
2. After the NaOH, 8M attack, a significant decrease of quartz (Q), was recorded while calcite (C), dolomite (D), and merwinite (MR) disappeared. The alkaline attack formed new phases: portlandite (Ca(OH)
2) (P), brucite (Mg(OH)
2 (B), and magnetite (Fe
3O
4) (M) (
Figure 4a). Main crystalline phases identified in Slag R were: Q-quartz (SiO
2) and C-calcite (CaCO
3) and D-dolomite CaMg(CO
3)
2. The last two have higher intensity with respect to Slag A as evidenced both by Rietveld analysis, reported in
Table 2, and by higher amounts of Ca and Mg in chemical analysis (
Table 1). After the alkaline attack, a significant decrease of quartz (Q), was recorded while calcite (C) and dolomite (D) disappeared, and the formation of portlandite (Ca(OH)
2) (P), brucite (Mg(OH)
2 (B), magnetite (Fe
3O
4) (M), and gehlenite (Ca
2Al
2SiO
7) (G) occurred (
Figure 4b). Formation of portlandite is particularly evident in Slag R where the content of Ca is higher with respect to Slag A. From XRD pattern appears that even if some phases decrease in intensity and others form, in general, crystalline phases present in the as-received slags remain almost undissolved after NaOH dissolution.
To evaluate the behavior of the slags in acid environment the HCl attack was performed. This test allows quantification of the phases formed as a result of alkaline activation, thus measuring the reactive capacity of the slag to form a geopolymer network in alkaline environment. For this reason, the test was performed on the slags before and after alkaline activation to compare such behavior.
The insoluble fraction of Slag R was lower than the insoluble fraction of Slag A because the first reacted more than the second (
Figure 5). Slag R has a calcium content of about 13 wt.% and it was characterized by L.O.I. of about 20 wt.%, both these parameters justified by the presence of calcite and dolomite that were responsible of its high solubility in hydrochloric acid (carbonates content is higher than 30%). In contrast, Slag A has a calcium content of about 7 wt.% and L.O.I. of 14 wt.%, thus a lower solubility in HCl was recorded. The insoluble fraction of both slags was analyzed by XRD; the data indicate that only quartz remained in the insoluble phase while calcite and dolomite were dissolved during HCl test (
Figure 6a,b). Slag A and Slag R were characterized by the same behavior from mineralogical point of view after acid attack because of the similarity of the crystalline phases present, the difference is only related to carbonates amount.
3.2. Alkali Activated Materials Characterization
In order to estimate the consolidation of the structure of the alkali-activated materials, a sample of each AAMs formulation, A50NW and R50NW, was immersed in distilled water to perform an evaluation of the structural integrity of the samples. In order to qualitatively evaluate the efficacy of the consolidation process, the samples were observed by naked eye to see if they alter their aspect in this situation. Neither sample dissolves in water confirming the occurrence of alkali activation reaction.
This integrity test has been adopted in our laboratory as a common procedure to evidence the chemical stability of alkali-activated materials containing complex aluminosilicate powders with variable chemical and mineralogical compositions, such as incinerator bottom ash [
27] and mine tailings [
23].
The comparison of XRD patterns between slag A and R and their respective alkali-activated materials (
Figure 7a,b) show the presence of all minerals obtained from each precursor, but in a different quantity compared to the original amount. Alkaline activation products, such as amorphous calcium silicate hydrate gel C–S–H, are formed as part of the reaction between the activator and Ca
2+, abundant in the precursors, especially in slag R [
30,
32,
33].
As discussed above, since the alkaline environment can modify the solubility of some elements, such as Cr, Ni, As, Cd, a leaching test in water (according to EN 12457-2) on the alkali-activated materials was necessary to verify the eventual hazardous nature. The formation of the 3D network of Ca-Na-aluminosilicate as a consequence of the consolidation represents an alternation of silicon and aluminum atoms in four-fold coordination with oxygen. Being the electric charge of Al +3, with respect to +4 for Si, the AlO
4 tetrahedron is charged −1, thus attracting positive metallic cations, within the geopolymer network. If the number of AlO
4−1 in the 3D geopolymer network is high, the chemical stability of the bonded cations is high as well, thus the amount of metallic cations in the eluate will be low [
34].
The amount of heavy metals found in the eluate of A50NW and R50NW was compared to the limit values prescribed to dispose of non-dangerous waste in a landfill (same as indicated above for the slags) (
Table 5). All the release values are below the legal limit. An interesting comparison is the release values of slags before alkali activation in both the environments, NaOH and water. All the bivalent ions (Ba, Cd, Cu, Ni, Pb, and Zn) are well immobilized in a geopolymeric matrix. In particular, for Cd and Pb no release is shown. On the contrary, amphoteric elements, such as As and Cr, show a slight increase of release with respect to the corresponding slag in both the environments. These results confirm the findings of the authors [
27] for immobilization of incinerator fly ashes in metakaolin-based geopolymers, where for bivalent cations very low release are observed. In particular, for Cd, negligible release as found, as already reported by Izquierdo et al. [
35] and Zhang et al. [
36], who explained this behavior by considering the formation of cadmium hydroxide responsible for the immobilization of Cd within the geopolymer.
Particular attention was paid to the release of amphoteric elements Mo, Sb, and As before, and after alkaline activation of both slags (
Table 6). Specifically, the results related to the test on alkali-activated materials derive from leaching in water, while the values of the two slags derive from NaOH test. Since after few minutes in water the pH of the leaching medium becomes alkaline because of the nature of AAMs, these experimental results are comparable. These results show how the release of Sb and As of AAM slightly increases with respect to the corresponding slag, still remaining below the legal limits [
37,
38,
39].
On the other hand, in the AMMs specimens the Mo presents an increase of leaching values that slightly exceeds the legal limit for landfill for non-dangerous waste.
These results are in line with literature [
39] showing good retention of Pb, Cd, Cr, Cu, Ni, Zn, even when the leachable contaminants content was measured by conducting compliance leaching tests according to EPA Test Method 1311 (toxicity characteristic leaching procedure-TCLP) [
40].
The pH and ionic conductivity measurements were additional experimental measures to assess the reticulation process and the three dimensional reticulation extent. The alkali-activated material A50NW obtained from Slag A presented a very stable pH values and low values of ionic conductivity (from 385 to 1251 mS/m), indicating a low amount of ions released from the geopolymeric network (
Figure 8). The stable pH values indicated that the excess of OH
− not reacted with the aluminosilicate powders was released in the very first seconds of the contact between the solid and water. Subsequently, no other OH
− groups were released. R50NW alkali-activated material was characterized by a similar behavior of pH values recording values almost constant (11.7–12.2) while conductivity slightly increased (from 445 to 1615 mS/m). The A50NW presented a slightly higher chemical stability which confirmed the more efficient reticulation of the structure with respect to R50NW (
Figure 8). This behavior confirms the hypothesis noted above that slag with higher amorphous fraction also shows higher reactivity in alkaline environment.
As conclusion, we can state that these values of pH and ionic conductivity are typical for alkali-activated materials, with good reticulation and almost complete reaction of the alkaline solution added for activation. Similar trends are already reported in literature for geopolymers based on metakaolin [
41] and alkali-activated materials from metakaolin added with residues [
42].
Testing in HCl was performed to quantify the phases formed as a result of alkaline activation and to establish the reactive capacity of slag in alkaline environment with the formation of a geopolymer. The main reaction product formed in this process is an amorphous/semicrystalline aluminosilicate gel. After the test, the insoluble fraction corresponds to the part that had not reacted with the alkaline solutions because HCl provokes the dissolution of the reaction products formed during alkali activation of slag as indicated in previous study [
43].
The results of the test on alkali-activated materials containing A and R slags show that the insoluble fraction of R50NW was less than the insoluble residue of A50NW (
Figure 9c). It confirmed the results of the same test on the slags. The results are distorted by the presence of carbonatic crystalline phases in the alkali-activated materials-carbonates that almost did not react during the activation. Yet, these phases are soluble in HCl test, so the soluble fraction sums up the percent of carbonates and the reaction products formed during activation. This is also confirmed by the comparison between slags and the corresponding AAMs after test in HCl. Slag A (
Figure 9a) shows lower solubility, both as in the as-received state and after alkali activation, with respect to Slag R, and corresponding AAM (
Figure 9b). This fact is related to the lower carbonatic content. For Slag R, the presence of unreacted fraction of carbonates, in the slag as well as in the corresponding AAM, leads to higher solubility. The aluminosilicate fraction created during the alkaline activation process is less soluble with respect to original calcite and dolomite. For AAM with high carbonatic content, this test needs a different discussion with respect to materials that do not contain carbonates.
The insoluble residue resulting from the HCl test was analyzed by XRD to evaluate the modifications in crystalline phases after the acid attack. X-ray analysis showed that only quartz remained. The calcite and dolomite were dissolved during HCl test (
Figure 10a,b), contributing to the amount of soluble fraction. Both alkali-activated materials were characterized by the same behavior of their respective slags. XRD patterns shows the presence of broad band between 20 and 40° in 2θ typical of an amorphous fraction that has not been dissolved in HCl, hence it is composed of Al-O-Si network. This behavior is particularly evident in AAM containing slag A, which is richer in amorphous phase (56%) with respect to slag R (35%). Such glassy fraction is more reactive in alkaline environment forming C–(N–)A–S–H–gel, as already observed by Adesanya et al. [
16] and more insoluble in HCl acid with respect to carbonatic fraction. Carbonatic fraction in AAM plays the role of inert phase and do not contribute to alkali activation. The formation of C–(N–)A–S–H–gel is confirmed by EDS analysis which shows that the chemical composition of the gels formed in the two alkali-activated materials is Na = 7.7, Ca = 10.4, Al = 2.3, and Si = 11.2 for A50NW and Na = 8.3, Ca = 8.3, Al = 1.4, and Si = 10.2 for R50NW (
Figure 2). The more significant difference is related to Ca content, because notwithstanding the higher amount of Ca in slag R, it corresponds to crystalline phases insoluble to alkali activation and therefore not included in amorphous gel.