3.1. Carbonated and Desulfurization Slags Characterization
In order to determine the potentially reactive fraction that is present in the slags, as a precursor in the production of alkali-activated materials, the alkaline attack was performed. The reactive Si/Al mass ratio, calculated from the number of Si and Al ions released into the alkaline solutions, provided information about the suitability of the raw materials to form the alkali-activated materials. The concentrations of these elements are reported in
Table 4. The results showed that both the slags present very low releases of Si and Al, with respect to the reference material metakaolin (Si = 194 ppm, Al = 144 ppm, and Si/Al = 1.35, see
Supplementary Materials). The c-BOF slag presented a higher number of Si and Al ions in the leachate, and a more proper ratio with respect to the De-S slag. For both the slags, the Si content in the solutions is higher with respect to Al, according to both the chemical compositions (
Table 1), in comparison to metakaolin.
The Si/Al mass ratio is correlated with the reticulation degree of the alkali-activated materials; in fact, when the Si/Al mass ratio is below the value of three, the final materials are characterized by a 3D rigid network, which would be a proper matrix for a cement, mortar, or concrete [
31,
37]. Fletcher et al. [
37] observed that the raw materials that are characterized by a Si/Al mass ratio below the value two, do not show the typical properties of geopolymers. The previous results confirm the findings of the authors, for the significance of the relation between the Si/Al mass ratio and qualifying the reactivity of raw materials that are suitable in the realization of AACs.
Furthermore, heavy metals and amphoteric elements, such as Sb, Se, and As, are below the detection limits, and Mo was only slightly released (
Table 4).
Mineralogical analysis was performed to evaluate if the crystalline phases in the starting slags changed or disappeared after the alkaline attack (
Figure 3a,b). Carbonated slag (
Figure 3A) was characterized by Q-quartz (SiO
2), C-calcite (CaCO
3), and P-portlandite (Ca(OH)
2). After immersion in the NaOH 8M solution, no significant variation was observed (
Figure 3a). For desulfurization slag, a flattening of the band between 25 and 40 2θ, corresponding to the aluminosilicate amorphous phase, is visible, confirming the higher reactivity of the glassy fraction in an alkaline environment (
Figure 3B). The higher reactivity of the amorphous phase in the NaOH 8M solution was previously evidenced by the authors for incinerator bottom ash, showing a significant decrease in the amorphous hump after the treatment [
38].
In order to investigate the eventual hazardous nature of carbonated BOF and De-S slags, the leaching test, according to the European standard EN 12457-2, was carried out. The concentrations of heavy metals, such as Pb, Cd, and As, into a leachate (
Figure 4), were compared to the limit values that were included in the council decision 2003/33/EC prescribed to place not dangerous waste in landfill (law limit in water).
The concentrations of heavy metals found in the eluate were lower than the law limit values, confirming that these slags are not dangerous raw materials and can be safely used for the alkali-activated materials.
During the leaching test (EN 12457-2), the pH and ionic conductivity of the c-BOF and De-S slags were measured at the following different times: 0, 5, 15, 30, 60, 120, 240, 480, and 1440 min, after immersion in water (
Figure 5). These measurements have been useful to analyze the chemical stability of raw materials that are in contact with water, as already reported in the previous paper [
9,
29]. Starting from pH = 7, all the slags showed an increase until the constant pH value around 12, during the first 24 h; instead, their conductivity grew rapidly in the first hour, to reach constant values. C-BOF slag’s conductivity increased continuously, reaching the value of 600 mS/m. Although De-S slag showed the highest conductivity values, it was the most stable of the two slags because it immediately released the largest number of ions in the aqueous environment, yet rapidly reached a constant value (800 mS/m) (see the plateau in
Figure 5). On the contrary, carbonated slag was characterized by a continuously increasing trend in the release of ions during the 24 h of the test. This behavior shows that the leaching of ions is progressive for c-BOF, while for De-S, after the first minutes, the slag is stable without the leaching of ions. The progressive release of ions into NaOH solution, reached the higher values of the heavy metals content (see
Table 4) for c-BOF.
The carbonated and desulfurization slags presented low percentages of insoluble residual fraction in HCl. The high amount of calcium oxide, 54 and 55 wt.%, respectively, justifies the high solubility in hydrochloric acid (
Figure 6). The high solubility in HCl is related to both the dissolution of the amorphous gel, formed during alkali activation, and to the presence of crystalline phases that are soluble in HCl, such as calcite. The contribution of the soluble crystalline phases leads to a lower amount of insoluble fraction when compared to metakaolin, which shows 36% (see
Supplementary Materials).
The soluble fractions of all the raw materials were analyzed by ICP/AES, in order to obtain information about the presence of elements that dissolved from the solid structure of the materials. For such a reason,
Table 5 does not show the law limit for these leachate values, since this is a structural leaching and not a chemical leaching test. All the metals show low values of release, except chromium, because its leaching increases after carbonation, due to pH reduction.
The insoluble residual fraction of all the slags was analyzed by XRD (
Figure 7), in order to observe the modifications of the crystalline phases after the acid attack.
Figure 7 shows the dissolution of calcite (CaCO
3) and portlandite (Ca(OH)
2) after the HCl test 1:20 for both the slags.
The acetic acid test was performed in order to assess the behavior and stability of the investigated AACs to acid rains. This analysis was also carried out on slags, to evaluate the role of geopolymerization, in terms of the chemical stability of the corresponding ACCs. Thus, the goal of the acetic acid attack was to compare the metals that were released into distilled water and the corresponding amount released after the contact with the acid solution (
Table 5), and the possible modification of the crystalline phases after the test.
The release of heavy metals was compared with the law limits that are contained in the European Directive 1991/271/CEE concerning urban waste water treatment.
Table 5 shows that the presence of each metal in the liquid fraction is below the law limits, meaning that the leachate that was left free after the contact between the acid rains and materials was not dangerous for urban water.
3.2. AAM and AAC Characterization
The structural and chemical stability of the AAM (M1) and AACs (M2 and M3) were analyzed by an integrity test, as described in
Section 2.3. After immersion in the distilled water for 24 h, all the specimens were stable, recording a weight loss of 6–8 wt.%, confirming the occurrence of the geopolymerization reaction (
Figure 8).
The integrity test in water was performed to assess the structural stability of the samples, while the quantitative evaluation of the chemical stability was measured by the weight loss (
Figure 8). The weight loss range was between 6.5 to 8.2%, but the wide variability of the samples generated an error of approx. ±2%. The reinforced samples, with basalt fibers (specimen M2), seems to have higher weight loss with respect to the reference AAM (specimen M1) that did not contain any fibers, but, due to the sample variability, such difference cannot be considered significant (
Figure 8).
The capability of the AAMs materials to immobilize heavy metals into their matrix, was evaluated by the leaching test that is prescribed in the European standard EN 12457-2. The content of heavy metals in the eluate of M1, M2, and M3 was compared to the limits that are prescribed to dispose of non-dangerous waste in landfill (council decision 2003/33/EC) (
Figure 9). All the compositions present leaching values below the law limit for landfill for not dangerous waste, confirming that the stability that is reached with the occurrence of geopolymerization is higher than the enhancing of ions mobility, due to the alkaline pH that is typical of alkali-activated materials.
The presence of the fibers in the specimens did not influence the release of the heavy metals in the eluate; the values found of M2 and M3 were very similar to those of M1. A slight increase in the metals release in the reinforced composites is evident for Cd, Ni, and Pb. The presence of the fibers does not affect the ions’ release, which is in line with the absence of influence on the cold consolidation. Other authors stated that harmful water environments (similar to the alkaline environment that is typical of geopolymers) did not weaken the effect of the composite fiber in the asphalt matrix [
39]. An interesting point is the comparison of the release values of the amphoteric elements Mo, Sb, and As, from the as-received slags (
Figure 4) and their corresponding AAMs in the distilled water (
Figure 9). The results showed the increase in Sb, Mo, and As after the alkaline activation, still remaining below the law limits, as already observed for other kinds of metallurgical slags [
9].
During the leaching test, the consolidation performance was investigated by pH and ionic conductivity measurements, which were taken as indicators of the chemical stability of alkaline-activated materials. For all the samples, there was not a significant modification of pH values; M1, M2, and M3 showed a pH range from about 11.2, after a few minutes from immersion, to around 12.2, after 24 h. The fibers that were added to the M1 matrix did not influence the behavior in terms of ions release (
Figure 10). For all the three alkali-activated materials, there was a significant increase in the conductivity values during the 24 h, in a more gradual way [
31]. In particular, M1 showed lower values of ionic conductivity (from 400 mS/m to 900 mS/m) with respect to M2 and M3, containing basalt and cellulose fibers, respectively. M2 showed values ranging from 400 mS/m to 1200 mS/m, and M3 from 450 mS/m to 1350 mS/m (
Figure 10). The presence of the fibers in the AAM matrix did not favor the consolidation of the material, and it was possible to observe, from the conductivity values, the release of a significant number of ions in the aqueous environment for the AACs, probably Na
+ coming from unreacted activating solutions. This behavior can be related to a lower degree of consolidation of M2 and M3, with respect to M1. These results were confirmed by the weight loss values (
Figure 8) and by the ionic conductivity values (
Figure 10), while the differences in the release of other metals (
Figure 9) are less than 0.1 mg/kg. From the results of the previous analyses, sample M1 (reference specimen without fibers) was characterized by the best properties, in terms of chemical stability, whilst the presence of the fibers enhanced the release of ions. Moreover, M3 showed higher values of conductivity during the first 24 h; the presence of fibers involved no improvements in the properties of the material, probably due to the incomplete dispersion of the fibers within the matrix, as observed by the SEM micrographs.
To investigate the behavior of the AAM (M1) and AACs (M2 and M3) on the attack of acid rains, an acetic acid test was performed. This test was also carried out on the as-received slags for the sake of comparison. The results showed that the raw materials did not release heavy metals on the urban waste water, so they were suitable to realize AACs. The same considerations are valid for the alkali-activated materials. The soluble fractions, the eluates of the acetic acid test, were analyzed by ICP/AES.
Table 6 shows that the release of heavy metals was below the limits contained in the council decision 2003/33/EC, concerning urban waste water treatment, so these materials are non-dangerous for the environment.
The percentage of reaction products that was generated during alkali activation was determined by acid attack with HCl solution, according to the literature [
37]. After the acid attack, insoluble and soluble fractions were produced. In particular, the insoluble fraction has to be considered as the part of the consolidated AAM/AAC that had not reacted with the alkaline solutions. From
Figure 11, it was possible to observe the insoluble fractions of M1, M2, and M3, which are very similar, especially considering the error bar. M2 was characterized by a lower insoluble phase with respect to M1 and M3, but taking into account the natural compositional variability, no significant differences are recognized. As for all the metallurgical slags, the presence of crystalline soluble phases increases the solubility of the samples, with respect to metakaolin [
9]. In the literature [
40], the effect of steel fibers on the durability performance of alkali-activated materials were studied, showing that the incorporation of fibers has no significant effect on the acid attack. Fiber-reinforced alkali-activated materials present a mass loss for chemical attacks, but the loss in the Portland materials was approximately twice that recorded for the AAMs.
The insoluble fraction that was collected after the acid attack was analyzed by XRD, to investigate the crystalline phase changes (
Figure 12a–c). All the residues after the HCl test showed the dissolution of calcite and dolomite, which significantly contributed to increasing the amount of soluble fraction. The presence of the quartz crystals after the acid attack confirmed that it was not soluble in hydrochloric acid. The fraction of amorphous content in all the mixes remained stable and comparable among the developed materials after exposure to the acid. All the mixes contained ca. 60–65 wt.% amorphous content after the acid attack, showing good stability of the C–S–H gel in the materials. Note that this relative fraction slightly increased compared to the non-exposing samples, since the dissolution of portlandite and calcite led to the higher proportion of the C–S–H phase.
In all the cases, it should be noted that the broad peak, indicating the presence of the amorphous phase, was not completely dissolved in HCl, but shifted its position towards low two-theta degrees, with respect to the as-prepared AACs, moving from 25–40° 2θ to 20–30° 2θ after the acid attack. This particular behavior can be explained with the modification of the chemical composition of the amorphous gel phase, which becomes richer in silica after the leaching of Al out from the aluminosilicate geopolymer matrix. The Si-rich amorphous glass is typically localized at lower two-theta values (10–35° in 2θ) with respect to the aluminosilicate amorphous phases [
41]. This behavior can also explain the increase in the insoluble fraction of AAM and AAC, with respect to the as-received slags after HCl attack, as reported in
Figure 11.
Moreover, the insoluble fractions that were obtained by the HCl test were analyzed by scanning electron microscopy (SEM), to investigate the microstructure of AACs before and after the acid attack. In particular,
Figure 13a shows M1 AAM, where calcite was formed at the interface surrounding the slag grains, as the result of the carbonation process for BOF slag, which are used as aggregates in the formulation. There was a fraction of unreacted BOF slag in the binder, likely due to the covering of calcite from the carbonation process, preventing further reaction of the slag during alkali activation. The acid attack led to the dissolution of portlandite, and the C–S–H gel remained in the matrix after acid exposure, as visible in Fig 13b.
Figure 13c showed the effects of M2 AAC (presence of basalt fibers in the matrix); the basalt fibers seem to dissolve partly, to form some C–N–A–S–H phases. This was likely due to the high alkali environment that can cause the dissolution of basalt fibers, or may also come from the acid attack. This, on one hand, can increase the bonding between basalt fibers and the AAM matrix. On the other hand, the contribution of fibers in delaying crack propagation seems delayed, since the debonding process was less effective. In contrast, the cellulose fibers remained stable in the AAM matrix. Moreover, there was no clear damage observed on the fibers after the acid exposure. This observation can explain the better increase in the compressive and flexural strength of the cellulose fiber-reinforced De-S slag-based AAM compared to that of the basalt fiber-reinforced AAC [
6].
The soluble fraction that was generated by the acid attack of c-BOF and De-S slags, and their AACs were analyzed by ICP/OES, to observe if the release of heavy metals was decreased in the acid environment (
Figure 14). The carbonated slag releases a higher content of heavy metals, in particular, Cr, Zn, and Pb, than the desulfurization slag. Instead, the release of As, Cd, and Mo was the same for carbonated and desulfurization slags. The geopolymeric network reduced the release of all metals with respect to the corresponding slags; as an example, Pb, which reduced its values from 0.7 mg/L to a value below 0.1 mg/L.
The residual M1, M2, and M3 after exposure to HCl, were subjected to thermal analysis (TG/DTG) (
Figure 15). The mass loss was very small, where M1 and M3 are identical, with a weight loss around 6% and M2 around 5%; this confirms the SEM data showing that the cellulose fibers remained stable in the AAM matrix. While M2 exhibited the least mass loss among the three mixes. The weight losses were attributed to the dehydration (ca. T < 200 °C) and the dehydroxylation, which was typical for geopolymeric matrices for T = 200–400 °C. In addition, there was a very slight thermal event at ca. 800 °C, which was attributed to the decarbonation of leftover calcite. This indicates that likely not all calcite has dissolved after the acid exposure.