Recycling and Application of Mine Tailings in Alkali-Activated Cements and Mortars—Strength Development and Environmental Assessment

Mine tailings (MT) could represent a step forward in terms of the quality of the aggregates usually used in civil engineering applications, mostly due to its high density. The Portuguese Neves Corvo copper mine, owned by the Lundin Mining Corporation, produces approximately 3 million tonnes per year. Nevertheless, it cannot be used in its original state, due to its high levels of sulphur and other metals (As, Cr, Cu, Pb, Zn). This paper focuses on the stabilisation/solidification of high-sulphur MT, without any previous thermal treatment, using alkali-activated fly ash (FA). The variables considered were the MT/FA ratio and the activator type and concentration. A fine aggregate was then added to the pastes to assess the quality of the resulting mortar. Maximum compressive strengths of 14 MPa and 24 MPa were obtained for the pastes and mortars, respectively, after curing for 24 h at 85 ◦C. Thermogravimetric analysis, scanning electron microscopy, X-ray energy dispersive spectroscopy, X-ray diffraction, and infrared spectroscopy were used to characterize the reaction products, and two types of leaching tests were performed to assess the environmental performance. The results showed that the strength increase is related with the formation of a N-A-S-H gel, although sodium sulphate carbonate was also developed, suggesting that the total sodium intake could be optimized without strength loss. The solubility of the analysed metals in the paste with 78% MT and 22% FA was below the threshold for non-hazardous waste.


Introduction
The framework addressing the sustainability optimisation of the mining sector is currently undergoing a very significant revolution. It is abandoning the focus on the efficiency of operations, aiming to reduce the disposed waste, and shifting to a more global paradigm, encompassing not only the mining operations but also the complete life cycle of this vital industry [1]. Wastes resulting from mining operations can result either from crushing or other, more complex processing methods. The former, normally referred to as plain "mining waste", has been used in the construction industry (e.g., concrete, embankments) as it does not raise any special environmental concerns; the latter, usually referred to as "mine tailings" (MT), produces more worrying materials due to its high levels of different kinds of contaminants which can be washed away when in contact with water [2][3][4]. Nevertheless, the of toxic elements present in the stabilised mortars, and the values obtained were compared with the thresholds imposed by European standards.

Materials and Methods
Two types of mixtures were studied: binding pastes, based on a combination of mine tailings (MT) and activated fly ash (FA), and mortars, comprising the same MT-and FA-based pastes, but including also a fine aggregate. These mixtures were designed so as to study the mechanical and environmental stabilisation of the MT through alkali-activated FA and the influence on the mechanical behaviour of a medium-sized sand skeleton.

Materials
The mining waste was collected at the Neves Corvo mine, operated since 1988 by SOMINCOR (Sociedade Mineira de Neves Corvo S.A.), owned by the Lundin Mining Corporation. It was mostly included as an inert, in terms of alkaline activation reactions, and not as a precursor, and was recovered from one of the ponds located inside the mine compound (Figure 1), i.e., with a high water content. It was immediately transported to the laboratory, where it was separated from most of the constituent water and mixed with different percentages of FA, provided by the company PEGOP, responsible for the Portuguese thermo-electric power plant Central do Pego. The aggregate was a sandy soil (GS), collected on a bank of the river Tâmega, located in the region of Chaves in the north of Portugal.

Materials and Methods
Two types of mixtures were studied: binding pastes, based on a combination of mine tailings (MT) and activated fly ash (FA), and mortars, comprising the same MT-and FA-based pastes, but including also a fine aggregate. These mixtures were designed so as to study the mechanical and environmental stabilisation of the MT through alkali-activated FA and the influence on the mechanical behaviour of a medium-sized sand skeleton.

Materials
The mining waste was collected at the Neves Corvo mine, operated since 1988 by SOMINCOR (Sociedade Mineira de Neves Corvo S.A.), owned by the Lundin Mining Corporation. It was mostly included as an inert, in terms of alkaline activation reactions, and not as a precursor, and was recovered from one of the ponds located inside the mine compound (Error! Reference source not found.), i.e., with a high water content. It was immediately transported to the laboratory, where it was separated from most of the constituent water and mixed with different percentages of FA, provided by the company PEGOP, responsible for the Portuguese thermo-electric power plant Central do Pego. The aggregate was a sandy soil (GS), collected on a bank of the river Tâmega, located in the region of Chaves in the north of Portugal. The chemical composition of the three materials, obtained by X-ray fluorescence (XRF), is shown in Error! Reference source not found.. The main elements of the MT are sulphur trioxide (41%) and iron (28%), although some heavy metals were also detected in smaller percentages. The FA presented mostly silica (45.2%) and alumina (28.5%). Finally, the aggregate, as expected, was basically silica (86.8%).  The chemical composition of the three materials, obtained by X-ray fluorescence (XRF), is shown in Table 1. The main elements of the MT are sulphur trioxide (41%) and iron (28%), although some heavy metals were also detected in smaller percentages. The FA presented mostly silica (45.2%) and alumina (28.5%). Finally, the aggregate, as expected, was basically silica (86.8%). The particle size distribution (PSD) of the MT and GS, determined using sieves and sedimentation (for particles below 0.063 mm), is presented in Figure 2, which also includes the PSD of the FA, determined by laser granulometry. The fines contents (particles smaller than 63 microns) represented 91%, 98%, and 4% of the MT, FA, and GS compositions, respectively. The density of the MT (32.85 kN/m 3 ) was significantly higher than those of the GS (25.81 kN/m 3 ) and FA (21.75 kN/m 3 ). The MT showed some plasticity, with a plastic limit of 10.4% and a liquid limit of 26.3%. The particle size distribution (PSD) of the MT and GS, determined using sieves and sedimentation (for particles below 0.063 mm), is presented in Error! Reference source not found., which also includes the PSD of the FA, determined by laser granulometry. The fines contents (particles smaller than 63 microns) represented 91%, 98%, and 4% of the MT, FA, and GS compositions, respectively. The density of the MT (32.85 kN/m 3 ) was significantly higher than those of the GS (25.81 kN/m 3 ) and FA (21.75 kN/m 3 ). The MT showed some plasticity, with a plastic limit of 10.4% and a liquid limit of 26.3%. In terms of mineralogy, the X-ray diffraction analysis of the FA showed a main presence of quartz and mullite. The amorphous content of the FA is evidenced by a halo between angles 17° 2θ and 33° 2θ. However, the precise quantification of the reactive content in coal fly ash type F is not straightforward, as demonstrated by Kuenzel and Ranjbar [41], who divided the FA into three In terms of mineralogy, the X-ray diffraction analysis of the FA showed a main presence of quartz and mullite. The amorphous content of the FA is evidenced by a halo between angles 17 • 2θ and 33 • 2θ. However, the precise quantification of the reactive content in coal fly ash type F is not straightforward, as demonstrated by Kuenzel and Ranjbar [41], who divided the FA into three different materials: reactive, partially reactive, and inert. They showed that the contribution of the cenospheres (i.e., partially reactive fraction) is restricted to the outer layer, which makes it almost impossible to accurately quantify the overall contribution. The diffractogram of the MT revealed, as expected, the presence of pyrite and quartz as the dominant minerals, with some arsenopyrite, muscovite, and chamosite. The GS mostly constitutes quartz.

Preparation, Analytical Characterisation, and Mechanical Testing
The MT was initially dried and submitted to a thorough de-flocculation process, by hand, upon arrival at the laboratory ( Figure 3). The FA was received with a water content lower than 3% and was thus considered ready for use. Preparation of the pastes and mortars started with the dry mixing of the solid components (FA+MT or FA+MT+GS) for 5 min in a Hobart counter mixer.
Appl. Sci. 2020, 10, x FOR PEER REVIEW 5 of 22 different materials: reactive, partially reactive, and inert. They showed that the contribution of the cenospheres (i.e., partially reactive fraction) is restricted to the outer layer, which makes it almost impossible to accurately quantify the overall contribution. The diffractogram of the MT revealed, as expected, the presence of pyrite and quartz as the dominant minerals, with some arsenopyrite, muscovite, and chamosite. The GS mostly constitutes quartz.

Preparation, Analytical Characterisation, and Mechanical Testing
The MT was initially dried and submitted to a thorough de-flocculation process, by hand, upon arrival at the laboratory (Error! Reference source not found.). The FA was received with a water content lower than 3% and was thus considered ready for use. Preparation of the pastes and mortars started with the dry mixing of the solid components (FA+MT or FA+MT+GS) for 5 min in a Hobart counter mixer. The activator was a combination of sodium hydroxide (SH) and sodium silicate (SS), with different SS/SH weight ratios of 0.5, 0.75, 1.00, and 1.25 (Error! Reference source not found.). The sodium hydroxide was supplied in pellets, with a specific gravity of 2.13 at 20 °C (99 wt.%), and was then dissolved in deionised water to previously defined concentrations of 7.5, 10.0, 12.5, and 15.0 molal. The SS presented a unit weight of 1.464 g/cm 3 , a SiO2/Na2O weight ratio of 2.0 (molar oxide ratio of 2.063), and a Na2O concentration in the solution of 13.0%. Four to six hours after mixing the SH and the SS, when the temperature of the resulting activator reached 20 °C ± 2 °C , it was mixed with the solid components, and additional mixing followed to guarantee homogenisation.
A crucial aspect of the preparation of the pastes was the definition of their optimal compaction properties, namely, the dry unit weight and its equivalent water content, which was achieved through a standard Proctor test. This Proctor test was performed on the paste with 20% FA and 80% MT and on the mortar with 20% FA, 30% MT, and 50% GS. A dry unit weight of γd = 1.974 g/cm 3 and a corresponding water content of ω = 13.0% were obtained for the paste, while values of γd = 2.064 g/cm 3 and ω = 9.0% were obtained for the mortar. The compositions of the pastes and mortars tested The activator was a combination of sodium hydroxide (SH) and sodium silicate (SS), with different SS/SH weight ratios of 0.5, 0.75, 1.00, and 1.25 ( Table 2). The sodium hydroxide was supplied in pellets, with a specific gravity of 2.13 at 20 • C (99 wt.%), and was then dissolved in deionised water to previously defined concentrations of 7.5, 10.0, 12.5, and 15.0 molal. The SS presented a unit weight of 1.464 g/cm 3 , a SiO 2 /Na 2 O weight ratio of 2.0 (molar oxide ratio of 2.063), and a Na 2 O concentration in the solution of 13.0%. Four to six hours after mixing the SH and the SS, when the temperature of the resulting activator reached 20 • C ± 2 • C, it was mixed with the solid components, and additional mixing followed to guarantee homogenisation.
A crucial aspect of the preparation of the pastes was the definition of their optimal compaction properties, namely, the dry unit weight and its equivalent water content, which was achieved through a standard Proctor test. This Proctor test was performed on the paste with 20% FA and 80% MT and on the mortar with 20% FA, 30% MT, and 50% GS. A dry unit weight of γ d = 1.974 g/cm 3 and a corresponding water content of ω = 13.0% were obtained for the paste, while values of γ d = 2.064 g/cm 3 and ω = 9.0% were obtained for the mortar. The compositions of the pastes and mortars tested are presented in Tables 2 and 3, respectively. Note that the slight variations in the FA/MT ratios in groups P2 and P3 were necessary to maintain the activator/ash ratio (Ac/FA) at a constant value of 0.75.  Each mixture was statically compacted inside a cylindrical mould 37 mm in diameter and with a height of 70 mm (EN 13286-53 [42]). After demoulding, they were stored under 85 • C and 20% relative humidity and tested after 24 h. The uniaxial compressive strengh (UCS) tests were carried out under monotonic displacement control at a rate of 0.2 mm/min. The results represent the average of three tested specimens.
The original FA and MT, as well as selected pastes, were characterised using scanning electron microscopy (SEM), X-ray diffraction (XRD), and Fourier transform infrared spectroscopy (FTIR). The FA and MT were also submitted to thermogravimetric analysis (TGA) by exposure to gradually increasing temperatures.

Leaching Tests
Two leaching test procedures were performed. The first, designated "LTT1", was based on standard EN 12457-4 [43]. It was performed on a specimen made from paste P2B. The fabrication and curing procedures were identical to the UCS. The element concentration on the eluate was determined following EN 16192 [44], and the values obtained were compared with the limits imposed by Council Decision 2003/33/EC [45]. The second test, designated "LTT2", was developed to study the leachability over time. After curing, the P2B specimen was submerged in distilled water, observing a liquid/solid weight ratio of 10:1. Two leachate samples of 50 mL each were then collected after 3, 7, 14, and 28 days. The pH was measured before the sampling. The values presented are the average of two replicates. Additionally, the soluble SO 3 content present in the water from the ponds, collected from the MT sample immediately upon arrival at the laboratory, was also measured. Finally, dry samples of MT and FA were left submerged in distilled water for 28 days, in separate containers, and their soluble SO 3 was also measured. The electrical conductivity of the four samples of water (i.e., from the pond, after submersion of original MT and FA for 28 days, and after submersion of a P2B monolith for 28 days) was also measured.

UCS
The MT was initially activated with a 12.5 molal sodium hydroxide solution, without FA (paste P0, Table 2), in order to assess the reactivity of the tailings. It immediately became clear, based on the structural integrity of these specimens, that chemical reactions did occur and generated some type of binding gel. This was confirmed by the UCS value of 1.8 MPa obtained. A parametric analysis was then developed to evaluate the influence of different parameters on compressive strength, namely, the ash/tailings weight ratio (FA/MT), the sodium hydroxide concentration (SH conc.), and the silicate/hydroxide weight ratio (SS/SH).
Regarding the ash/tailings ratio (with constant NaOH concentration and SS/SH ratio of 7.5 molal and 0.5, respectively), the results presented in Figure 4 show that an increase in the ash content produced an increase in compressive strength (UCS). The exception to this general trend was the paste with the lowest FA/MT ratio. This is not surprising, given that with a reduction in fly ash, the UCS tends to stabilise. Other authors [18], when studying the simultaneous activation of tungsten mining mud and red clay brick waste (precursor), using an activator solution based on SS and SH, obtained a compressive strength for the blends with a precursor content of 50% that was 3 times higher than that for the blends with only 10%. These authors were also able to prove that an increase in precursor content promoted the dissolution of some phases of the tungsten MT, with a positive effect on strength development. Given its higher strength among the P1 set, paste P1A was chosen for further analyses. The addition of the sandy aggregate to the paste produced a clearly positive influence on the mechanical response for every FA/MT ratio, representing a strength increase of between 14% (lowest FA/MT ratio of 0.25) and 42% (FA/MT = 0.33).
Regarding the influence of the NaOH concentration ( Figure 5), it became apparent that this variable was responsible for the most significant impact on UCS. An increase in concentration was responsible for an increase in UCS by favouring alkalinity levels which facilitate the dissolution of the species needed to form the final structural matrix. Cihangir et al. [16], when investigating the effect of the activator composition on the strength of high-sulphide tailings, obtained UCS values very similar to those here presented (between 1 and 2 MPa after 28 days, and between 2.5 and 4.0 MPa after 1 year). These authors also tested the use of Portland cement, which resulted in a compressive strength of 1.5 MPa after one year of curing. Moukannaa et al. [17], also focusing on NaOH properties for the activation of phosphate MT and metakaolin blends, reached a similar conclusion, i.e., an increase in NaOH concentration produces denser and stronger pastes. Although paste P2C reached the highest UCS among the P2 set, paste P2B was chosen for further analyses since its strength/cost compromise was considered more effective, i.e., the strength showed by paste P2B is enough for situations where this material might be used (e.g., structural embankments, pavement bases and sub-bases), and its cost is approximately 90% that of P2C. The influence of the aggregate phase on UCS was also positive. However, it is noticeable that the strength increase due to the addition of sand was more significant than that registered when analysing the FA/MT ratio, varying between 42% (lowest concentration of 7.5 molal) and 75% (12.5 molal).
Appl. Sci. 2020, 10, x FOR PEER REVIEW 8 of 22 Regarding the influence of the NaOH concentration (Error! Reference source not found.), it became apparent that this variable was responsible for the most significant impact on UCS. An increase in concentration was responsible for an increase in UCS by favouring alkalinity levels which facilitate the dissolution of the species needed to form the final structural matrix. Cihangir et al. [16], when investigating the effect of the activator composition on the strength of high-sulphide tailings, obtained UCS values very similar to those here presented (between 1 and 2 MPa after 28 days, and between 2.5 and 4.0 MPa after 1 year). These authors also tested the use of Portland cement, which resulted in a compressive strength of 1.5 MPa after one year of curing. Moukannaa et al. [17], also focusing on NaOH properties for the activation of phosphate MT and metakaolin blends, reached a similar conclusion, i.e., an increase in NaOH concentration produces denser and stronger pastes. Although paste P2C reached the highest UCS among the P2 set, paste P2B was chosen for further analyses since its strength/cost compromise was considered more effective, i.e., the strength showed by paste P2B is enough for situations where this material might be used (e.g., structural embankments, pavement bases and sub-bases), and its cost is approximately 90% that of P2C. The influence of the aggregate phase on UCS was also positive. However, it is noticeable that the strength increase due to the addition of sand was more significant than that registered when analysing the FA/MT ratio, varying between 42% (lowest concentration of 7.5 molal) and 75% (12.5 molal). Of the three factors considered, the SS/SH weight ratio appears to be the factor producing the lowest level of influence on strength development (Error! Reference source not found.). Even though this analysis is not straightforward, since there are other variables changing besides the SS/SH (i.e., the Ac/Sol and Na2O/SiO2 ratios and the water content), the results clearly show that a decrease in the SS/SH ratio did not produce a decrease in UCS. This also suggests that the sodium silicate content might be even further reduced without hindering the mechanical performance, which would represent a significant cost reduction or, possibly, a decrease in logistic complexity during live-scale applications if the SS is eliminated. Similar to the previous two analyses, the addition of the sandy skeleton showed a positive effect on UCS. Also similar is the fact that the trend of the UCS values obtained with the aggregate was very close to the trend obtained without it. In this case, the strength increase varied between 37% (SS/SH = 0.75) and 72% (SS/SH = 1.0). Of the three factors considered, the SS/SH weight ratio appears to be the factor producing the lowest level of influence on strength development ( Figure 6). Even though this analysis is not straightforward, since there are other variables changing besides the SS/SH (i.e., the Ac/Sol and Na 2 O/SiO 2 ratios and the water content), the results clearly show that a decrease in the SS/SH ratio did not produce a decrease in UCS. This also suggests that the sodium silicate content might be even further reduced without hindering the mechanical performance, which would represent a significant cost reduction or, possibly, a decrease in logistic complexity during live-scale applications if the SS is eliminated. Similar to the previous two analyses, the addition of the sandy skeleton showed a positive effect on UCS. Also similar is the fact that the trend of the UCS values obtained with the aggregate was very close to the trend obtained without it. In this case, the strength increase varied between 37% (SS/SH = 0.75) and 72% (SS/SH = 1.0). might be even further reduced without hindering the mechanical performance, which would represent a significant cost reduction or, possibly, a decrease in logistic complexity during live-scale applications if the SS is eliminated. Similar to the previous two analyses, the addition of the sandy skeleton showed a positive effect on UCS. Also similar is the fact that the trend of the UCS values obtained with the aggregate was very close to the trend obtained without it. In this case, the strength increase varied between 37% (SS/SH = 0.75) and 72% (SS/SH = 1.0).   Figure 7 shows the results of the thermogravimetric analysis (TGA) and respective derivative (DTG) of the tailings in the range 75 • C to 825 • C. The analysis of the fly ash is also included for comparison purposes. According to the literature [46], approximately 0.5% of the total weight loss usually registered during thermogravimetry of coal fly ash occurs below 500 • C, with an additional 2% after that, corresponding to the release of volatile matter from the sample and assuming that the TGA is carried in an oxidizing atmosphere (e.g., air) [47]. This behaviour is consistent with the result obtained for this particular FA. The MT started a significant weight decrease at the 375 • C mark, of more than 10%, resulting from the transformation of the pyrite's sulphur into gas [48]. When the temperature reached 585 • C, another stepped loss was noticeable, related with further decomposition of the pyrite [49].

Thermogravimetric Analysis
Error! Reference source not found. shows the results of the thermogravimetric analysis (TGA) and respective derivative (DTG) of the tailings in the range 75 °C to 825 °C. The analysis of the fly ash is also included for comparison purposes. According to the literature [46], approximately 0.5% of the total weight loss usually registered during thermogravimetry of coal fly ash occurs below 500 °C, with an additional 2% after that, corresponding to the release of volatile matter from the sample and assuming that the TGA is carried in an oxidizing atmosphere (e.g., air) [47]. This behaviour is consistent with the result obtained for this particular FA. The MT started a significant weight decrease at the 375 °C mark, of more than 10%, resulting from the transformation of the pyrite's sulphur into gas [48]. When the temperature reached 585 °C, another stepped loss was noticeable, related with further decomposition of the pyrite [49].

Mineralogical and Microstructural Characterization
Error! Reference source not found. shows the XRD diffractograms of selected pastes and anhydride FA and MT. Based on the size of the humps in the FA and MT diffractograms, associated  Figure 8 shows the XRD diffractograms of selected pastes and anhydride FA and MT. Based on the size of the humps in the FA and MT diffractograms, associated with their amorphous content, it appears that the MT has a higher crystalline content than the FA. The FA band at around 1010 cm −1 was partially dissolved in the reactions, indicating the presence of soluble material, while the peak at 410 cm −1 was maintained or partially dissolved (it is difficult to interpret since it is overlapping with an intense peak from the MT).

Mineralogical and Microstructural Characterization
Regarding the MT, the original peak at 1060 cm −1 , associated with Si-O or Al-O, was still present in paste PRef, but not in the remaining three pastes, suggesting that parts of the silicon and aluminium species dissolved from the MT were able to form new compounds. The intensity of the 872 cm −1 peak (MT) was maintained in PRef, but showed a slight decrease in P1A, P2B, and P3B, suggesting that some Si-O-T links were broken. A similar situation was found at 1102 cm −1 (typical of S-O asymmetric stretching vibrations in SO4 tetrahedra), i.e., it was still present in the PRef paste but was eliminated in the remaining pastes, suggesting the presence of soluble sulphur species. The intensity of the peak at 610 cm −1 , representing O-S-O deformation vibrations in SO4, increased for pastes PRef and P2B, suggesting the formation of new molecules involving dissolved S, and slightly decreased for pastes P1A and P3B relative to the original MT. Bands at 529, 460, and 410 cm −1 (MT), The pastes showed peaks associated with the crystalline phases of the FA (quartz and mullite) and MT (arsenopyrite, chamosite, muscovite, quartz, and pyrite), only with a lower intensity due to a dilution effect. This indicates that such phases are mostly inert. Some of the peaks associated with the chamosite phase (Fe 2+ ; Mg; Fe 3+ ) 5 Al(Si 3 Al)O 10 (OH;O) 8 on the MT showed lower intensities after activation (paste P0), suggesting that these phases partially participated in the reactions. This intensity reduction appears to have also occurred on the remaining four pastes analysed; however, in this case, the original MT peaks are diluted with the FA peaks.
The only new phase that was found to develop in the activated MT (paste P0) was burkeite (Na 6 (CO 3 )(SO 4 ) 2 ), which is basically a sodium sulphate carbonate. Although the burkeite peaks were less intense when FA was added to the precursor combination, this mineralogical phase was still visible, suggesting that the released S ions combined with the sodium, forming sodium sulphate.
The FTIR spectra of the same pastes are presented in Figure 9.  SEM and EDX of P0, PRef, P1A, P2B, and P3B are presented in Error! Reference source not found. and Error! Reference source not found., respectively. The presence of Fe in the P0 is, as expected, very intense. The gel phase, characterised in Error! Reference source not found., essentially constituted Na, Al, Si, and S. Based on these semi-quantitative values, it is possible to conclude that a N-A-S-H-type gel was developed, responsible for the strength gain. However, at this stage, the possible inclusion of sulphur ions in the gel structure was also considered, even if the exact role of such ions was not yet established. The FA band at around 1010 cm −1 was partially dissolved in the reactions, indicating the presence of soluble material, while the peak at 410 cm −1 was maintained or partially dissolved (it is difficult to interpret since it is overlapping with an intense peak from the MT).
Regarding the MT, the original peak at 1060 cm −1 , associated with Si-O or Al-O, was still present in paste PRef, but not in the remaining three pastes, suggesting that parts of the silicon and aluminium species dissolved from the MT were able to form new compounds. The intensity of the 872 cm −1 peak (MT) was maintained in PRef, but showed a slight decrease in P1A, P2B, and P3B, suggesting that some Si-O-T links were broken. A similar situation was found at 1102 cm −1 (typical of S-O asymmetric stretching vibrations in SO 4 tetrahedra), i.e., it was still present in the PRef paste but was eliminated in the remaining pastes, suggesting the presence of soluble sulphur species. The intensity of the peak at 610 cm −1 , representing O-S-O deformation vibrations in SO 4 , increased for pastes PRef and P2B, suggesting the formation of new molecules involving dissolved S, and slightly decreased for pastes P1A and P3B relative to the original MT. Bands at 529, 460, and 410 cm −1 (MT), indicating O-Si-O vibrations, were also found in the pastes, with similar intensities. A brief summary of the bands involving sulphur is presented in Table 4. Table 4. Summary of the sulphur-related peaks presented in Figure 9. SEM and EDX of P0, PRef, P1A, P2B, and P3B are presented in Figures 10 and 11, respectively. The presence of Fe in the P0 is, as expected, very intense. The gel phase, characterised in Figure 10, essentially constituted Na, Al, Si, and S. Based on these semi-quantitative values, it is possible to conclude that a N-A-S-H-type gel was developed, responsible for the strength gain. However, at this stage, the possible inclusion of sulphur ions in the gel structure was also considered, even if the exact role of such ions was not yet established. In terms of morphology, the micrographs indicate fairly compacted material due to the formation of gel with a SiO2/Al2O3 molar ratio of approximately 2 (Error! Reference source not found.). Nevertheless, several intact pyrite and ash particles are still visible. The P2B paste presented an apparent lower amount of unreacted FA particles (Error! Reference source not found.c) and a higher volume of gel compared with the remaining three pastes. This higher reaction degree, responsible for the highest UCS reached by this paste (Error! Reference source not found.), is a consequence of the higher concentration of the activating solution (12.5 M, Error! Reference source not found.).
Based on the gel composition, assessed from a total of 120 EDX points from which a representative sample was selected and presented in Error! Reference source not found. (with corresponding data summarised in Error! Reference source not found.), a typical N-A-S-H gel was formed, although with significant percentages of sulphur included. The presence of sulphur in the gel indicates that this element had some influence on the composition of the gel structure, probably encapsulated in the aluminosilicate matrix [50]. The high sulphur percentages found in paste P2B were always associated with high sodium contents, indicating the possible development of sodium sulphate, which was found in the mineral form of burkeite (Error! Reference source not found.). The majority of the gel points from the remaining three pastes presented a significantly higher content of Si+Al than Na+S, meaning that the formation of sodium sulphate was less intense in these pastes.  Table 5. Characterisation of the points presented in Figure 11. In terms of morphology, the micrographs indicate fairly compacted material due to the formation of gel with a SiO 2 /Al 2 O 3 molar ratio of approximately 2 (Table 5). Nevertheless, several intact pyrite and ash particles are still visible. The P2B paste presented an apparent lower amount of unreacted FA particles ( Figure 11c) and a higher volume of gel compared with the remaining three pastes. This higher reaction degree, responsible for the highest UCS reached by this paste (Figure 5), is a consequence of the higher concentration of the activating solution (12.5 M, Table 2). Appl. Sci. 2020, 10, x FOR PEER REVIEW 14 of 22    Table 5).
Based on the gel composition, assessed from a total of 120 EDX points from which a representative sample was selected and presented in Figure 11 (with corresponding data summarised in Table 5), a typical N-A-S-H gel was formed, although with significant percentages of sulphur included. The presence of sulphur in the gel indicates that this element had some influence on the composition of the gel structure, probably encapsulated in the aluminosilicate matrix [50]. The high sulphur percentages found in paste P2B were always associated with high sodium contents, indicating the possible development of sodium sulphate, which was found in the mineral form of burkeite ( Figure 8). The majority of the gel points from the remaining three pastes presented a significantly higher content of Si+Al than Na+S, meaning that the formation of sodium sulphate was less intense in these pastes.

Environmental Performance
Based on the results already presented, namely, the mechanical performance and the activator content used, paste P2B was determined to be the most adequate to study the potential of the technique to fix the contaminants present in the MT. This study was based on the use of two different methodologies (LTT1 and LTT2) to obtain an eluate solution from the solid pastes. Three different materials were submitted to the LTT1 test, namely, the original MT without any addition, the activated MT (P0), and the MT+FA paste designated P2B. Only the latter was submitted to LTT2.
The leachable concentrations obtained during test LTT1 were compared with the threshold values presented in Council Decision 2003/33/EC for inert waste, non-hazardous waste, and hazardous waste ( Figure 12). In general, activation of the original MT (paste P0) produced a decrease in the leached concentration, the exceptions being As, which showed a significant increase after the AA reactions, and chlorides and sulphates. The eluate collected from paste P2B improved upon the performance showed by the P0 paste, although elements Cr and Pb presented slightly higher solubility, above even the values presented by the original MT. Nevertheless, the addition of FA resulted in increased performance and complied with the most severe limits imposed by the European standard (inert waste disposal).
The leachable concentrations obtained during test LTT1 were compared with the threshold values presented in Council Decision 2003/33/EC for inert waste, non-hazardous waste, and hazardous waste (Error! Reference source not found.). In general, activation of the original MT (paste P0) produced a decrease in the leached concentration, the exceptions being As, which showed a significant increase after the AA reactions, and chlorides and sulphates. The eluate collected from paste P2B improved upon the performance showed by the P0 paste, although elements Cr and Pb presented slightly higher solubility, above even the values presented by the original MT. Nevertheless, the addition of FA resulted in increased performance and complied with the most severe limits imposed by the European standard (inert waste disposal). The solubility of paste P2B was evaluated for up to 28 days (readings taken after 3, 7, 14, and 28 days) by submerging a monolith-type specimen in deionised water (LTT2 test). The leaching solution was not refreshed after each sample collection, meaning that the leachate ions were able to accumulate. The decision of not refreshing the solution was made to enable the detection of a potential leaching asymptotic level, as described by Zheng et al. [51]. The results are presented in Error! Reference source not found., together with the threshold values for non-hazardous waste. The solubility of paste P2B was evaluated for up to 28 days (readings taken after 3, 7, 14, and 28 days) by submerging a monolith-type specimen in deionised water (LTT2 test). The leaching solution was not refreshed after each sample collection, meaning that the leachate ions were able to accumulate. The decision of not refreshing the solution was made to enable the detection of a potential leaching asymptotic level, as described by Zheng et al. [51]. The results are presented in Figure 13, together with the threshold values for non-hazardous waste. The pH of the solution decreased with time (Error! Reference source not found.), which could be a consequence of the consumption of OH − ions during the dissolution of Si and Al species [52], reinforced by the reduced amorphous calcium content [53]. Cihangir et al. [16] obtained a pH reduction throughout 360 days, from 12.5 to 11.2, of copper-zinc tailings stabilised with sodiumsilicate-activated slag, which was attributed to the formation of acid via the oxidation of pyrite. Although in this case the pH reduction occurred in different curing conditions (20 °C and 85% RH), a similar origin (pyrite oxidation) is plausible.  The pH of the solution decreased with time (Table 6), which could be a consequence of the consumption of OH − ions during the dissolution of Si and Al species [52], reinforced by the reduced amorphous calcium content [53]. Cihangir et al. [16] obtained a pH reduction throughout 360 days, from 12.5 to 11.2, of copper-zinc tailings stabilised with sodium-silicate-activated slag, which was attributed to the formation of acid via the oxidation of pyrite. Although in this case the pH reduction occurred in different curing conditions (20 • C and 85% RH), a similar origin (pyrite oxidation) is plausible. The leaching LTT2 test results were compared with the soluble SO 3 values measured directly from the water collected in the pond and from a sample of MT that was submerged in water for 28 days upon arrival to the laboratory ( Figure 14). The value obtained after 28 days is comprehensively lower that the value obtained from the pond, while the value obtained in the LTT2 test is even lower (approximately 45%) than the 28-day solubility, confirming the effectiveness of the stabilisation in terms of environmental performance. The figure also presents the conductivity measured in the same samples. In this case, the value measured in the water from the LTT2 test was significantly increased relative even to the pond water. This is most likely explained by the addition of fly ash, which generated a conductivity value in the submersion water higher than that in the water from the MT.

Discussion
The MT recovered from the Neves Corvo operation is a very interesting material in terms of density and mechanical behaviour, thus showing high potential for the construction industry. However, it bears a prohibitive environmental toll that needs to be significantly mitigated before it can be used. This motivation is also supported by the financial and environmental costs that are associated with the current landfilling of MT.
Although the MT was successfully activated with a 12.5 molal sodium hydroxide solution, without the addition of a second (and more effective) precursor or a thermal pretreatment, it still presented a low UCS, and its metal ion immobilisation was less than ideal. The reaction products included sodium sulphate carbonate in the form of burkeite (Na6(CO3)(SO4)2).
In order to improve both the mechanical and environmental performance, a second residue (FA) was added to the MT as a precursor. The results showed that an increase in the FA/MT ratio produced only a slight strength improvement, justifying the choice of an average FA/MT ratio (which favours the disposal of MT) for the environmental tests. The most influential variable on UCS was the activator's NaOH concentration, with the 15 molal paste showing the highest strength. However, the values obtained with concentrations below 15 molal are considered acceptable for several civil engineering applications (e.g., road base and sub-base; structural embankments; backfill material); thus, paste P2B (12.5 molal) was chosen for environmental testing.
The FTIR spectra showed that S-O bonds present in the original (anhydride) MT were broken in three of the four pastes tested (including P2B), suggesting the release of sulphur ions during the Figure 14. Comparison of the soluble SO 3 and electrical conductivity measured in the water from the mine ponds, in the water where the original MT and FA were submerged for 28 days, and in the water where a P2B monolith was submerged for 28 days.

Discussion
The MT recovered from the Neves Corvo operation is a very interesting material in terms of density and mechanical behaviour, thus showing high potential for the construction industry. However, it bears a prohibitive environmental toll that needs to be significantly mitigated before it can be used. This motivation is also supported by the financial and environmental costs that are associated with the current landfilling of MT.
Although the MT was successfully activated with a 12.5 molal sodium hydroxide solution, without the addition of a second (and more effective) precursor or a thermal pretreatment, it still presented a low UCS, and its metal ion immobilisation was less than ideal. The reaction products included sodium sulphate carbonate in the form of burkeite (Na 6 (CO 3 )(SO 4 ) 2 ).
In order to improve both the mechanical and environmental performance, a second residue (FA) was added to the MT as a precursor. The results showed that an increase in the FA/MT ratio produced only a slight strength improvement, justifying the choice of an average FA/MT ratio (which favours the disposal of MT) for the environmental tests. The most influential variable on UCS was the activator's NaOH concentration, with the 15 molal paste showing the highest strength. However, the values obtained with concentrations below 15 molal are considered acceptable for several civil engineering applications (e.g., road base and sub-base; structural embankments; backfill material); thus, paste P2B (12.5 molal) was chosen for environmental testing.
The FTIR spectra showed that S-O bonds present in the original (anhydride) MT were broken in three of the four pastes tested (including P2B), suggesting the release of sulphur ions during the dissolution stage of the reactions. X-ray diffractograms detected the presence of burkeite in the pastes, a new mineralogical phase which is essentially sodium sulphate, as a result of the combination of the sulphate and sodium ions.
Based on the SEM/EDX data, direct activation of the MT (i.e., without FA) produced N-A-S-H gel, responsible for the strength development showed by this paste. The same gel was also developed in the FA/MT pastes. Since alkalis could neutralise the acidic environment generated by the oxidation, the lower the alkali content, the faster it became depleted during this neutralisation process; thus, a lower activation degree of the FA was achieved. Therefore, the dissolution process was taken further in the P2B paste than in any of the remaining pastes. Additional evidence of this behaviour was obtained when measuring the pH of the solution used for the LTT2 test, performed during 28 days on a P2B specimen (Table 6), since the pH reduction can be explained by the progressive oxidation of pyrite.
A comparison between both types of leachate tests revealed some similarity in terms of aggressiveness, since the values registered with the LTT1 test (which takes approximately 24 h to develop) are, in general, similar to the values obtained after 3 days in the LTT2 test. The solubility of the Cr, Cu, Pb, and Zn ions did not change significantly between 3 and 14 days, suggesting a rapid surface removal of these species after only 3 days. The exception to this trend was As, which presented a relevant solubility increase between 3 and 14 days, up to a point where even the 2003/33/EC normative As threshold for non-hazardous waste was exceeded. After 28 days, the solubility increase was even more dramatic and, in this case, occurred also with the Pb and Zn ions. This process is probably governed by diffusion [54], although a possible failure of the physical encapsulating structure (i.e., local breakage that exposed some previously protected ions) should not be ignored [55].
The As leached from the original MT was below the thresholds proposed for inert, non-hazardous, and hazardous waste. However, after activation (P0), the increase in As solubility was very significant, surpassing not only the inert limit but also the non-hazardous limit. The difficulty in fixating As ions in aluminosilicate matrixes is well known [54,56], although this behaviour tends to improve with time [38,57]. It is related with the oxidation of the As ion, capable of assuming different oxyanoin forms, which possess very high solubility in alkaline environments but, nevertheless, are not bound to the final aluminosilicate matrix [57]. Therefore, it is possible that the increased alkalinity, as a result of the addition of the activator, favoured the release of As ions. This increased arsenic solubility was reverted again for paste P2B, suggesting that the addition of a more effective precursor to the original MT (fly ash) resulted in the production of a higher volume of AA gel, which was then capable of containing the As ions more effectively than paste P0.
It is somewhat worrying that the LTT1 test, which is strongly based on standard EN 12457-4, was not able to detect the solubility levels of this particularly threatening element, revealed only by the LTT2 test, although it is reasonable to assume that the LTT2 conditions might be too extreme and, thus, hardly representative of the real field environment.
Regarding the immobilization of the Cr, Cu, Pb, and Zn species, with all of them showing leachate values below the normalized limits, Lee et al. [50] sustained that geopolymers can immobilise lead ions through a combination of physical encapsulation and chemical bonding. Zhang et al. [55] observed a similar combination when studying the geopolymer immobilisation of Cr, Cd, and Pb, and concluded that the predominant mechanism depends on the solubility level of the heavy metal ions, i.e., highly soluble salts are more spread throughout the matrix, while less soluble salts form higher-volume nuclei which are then involved by the gel matrix. However, García-Lodeiro et al. [58] showed that the inclusion of such metals like Cr and Pb in cementitious gel may decline in the presence of chloride ions (due to their interaction and subsequent formation of soluble chlorides), which would then increase their leaching. Studies regarding the immobilisation of Cr in alkali-activated fly ash systems have shown that this ion has a negative effect on the activation mechanism and, subsequently, on the hardening process, and is thus not very efficiently contained in the geopolymeric matrix [55,59]. Nevertheless, the maximum Cr solubility values found in the present work are well under the non-hazardous waste limit established in the 2003/33/EC normative. The above-mentioned papers also concluded that Pb is very effectively immobilised in aluminosilicate gel, which is in accordance with the present data, even if a substantial increase was obtained between 14 and 28 days.
The leaching potential of some inorganic ions (Ca, Fe, K, Mg, and Na) not listed in the acceptance criteria defined in the normative 2003/33/EC are presented in Table 7. Decreasing release rates were registered for the Ca, Mg, and Na species as a consequence of their fixation, while the Fe and K species showed increasing solubility, most likely due to the oxidation of pyrite. It is noteworthy that the test was "static" (water was not changed after sampling), which allowed the saturation of the leachant and subsequent ion precipitation. No specific crystalline phases containing As, Cr, Cu, Pb, or Zn were detected in the XRD diffractograms or FTIR spectra, which might indicate that no specific chemical compound was formed between these metals and, for instance, silicon-a conclusion that was confirmed by the EDX analysis. However, it is also plausible that the concentration of these elements in the original MT is too low, meaning that even if any compound was formed after activation, it is possible that it was below the detection limit. Such a lack of association was also observed by Lee et al. [50] when studying the effect of lead on geopolymers with a Si/Al ratio of 2. Although the sulphate leaching presented in Section 3.3 was below the normative limits, it is still significant nonetheless, implying that full isolation of these ions was not achieved. Nevertheless, it can be concluded that it is possible to stabilise acidic mine tailings from copper and zinc explorations using alkali-activated binary blends.

Conclusions
In terms of mechanical behaviour, the stabilisation of mine tailings from copper extraction using alkali-activated FA is an effective procedure which can be considered for low-strength applications (<20 MPa), like embankments or road bases. This procedure has the advantage of avoiding thermic treatment to increase the amorphisation level, while requiring only 25%-30% FA, i.e., 70%-75% of the paste will be tailings. This is significant since it represents an alternative to traditional landfilling, and it also reduces cement consumption. Considering that the above-mentioned applications require high volumes of material, this solution represents a suitable alternative to conventional binders in terms of mechanical properties, environmental friendliness, and cost-effectiveness. A UCS of 15 MPa was obtained after a curing period of 24 h at 85 • C using a 15 molal SH activator (P2C). However, the P2B paste was chosen for further analyses, since its strength/cost compromise was considered to be more effective. The results showed that strength increase is related with the formation of N-A-S-H gel, although sodium sulphate carbonate is also developed, suggesting that the total sodium input could be optimized without strength loss. The solubility of all the metals analysed in the P2B paste (78%MT/22%FA) was below the threshold for non-hazardous waste. However, additional environmental studies are required to validate this technique, especially considering the SO 3 contents in the tailings.