The removal of pollutants and natural organic compounds from acid mine drainage using a combination of bentonite clay and MgSO₄

Conventional wastewater treatment has been employed in colloidal suspensions of various types, where inorganic coagulants are widely used. However, there are some challenges which were identified attributed to those regents; that is, corrosion and scaling of the pipeline and other components of the treatment plant. However, it is ideal and also economical to investigate alternative technology and user-friendly reagent(s), taking into account their cost-effectiveness. Countries rich in mineral resources such as gold and coal face acid mine drainage (AMD) decanting problems. There have been numerous reported catastrophic cases related to AMD flooding, such as that which occurred from defunct underground mine workings near Krugersdorp (South Africa) in August 2002, leading to contamination of the surface water (McCarthy 2010). Based on global water scarcity attributed to climate change, in under-developed countries such as the African continent, decanting AMD can be an advantage as it can be treated for potable water and industrial supplies. The problems associated with the AMD are low pH and high toxic metals content, all being detrimental to the ecosystem. It is therefore imperative to investigate a cost-effective AMD treatment method in order to curb such water shortage crises. In some other instances, the load of AMD can be quite complex as it has high mineral dissolving capacity due to low pH. It flows vertically and horizontally, dissolving all soluble materials, hence treatment can be difficult at times. Research has shown that there are on-going studies investigating an ideal technology viable in the treatment of AMD. Apart from physical, chemical, and biological methods, other sophisticated technologies such as activated carbon adsorption, magnetic ion exchange, reverse osmosis, advanced oxidation process, membrane filtration, among others (Maree 2004; Kurniawan et al. 2006; Herrera et al. 2007; Sibrell et al. 2009; Vermeulen 2012; Gaikwad 2014; King 2015; Lhois 2015). Although some of the studies conducted by Gupta et al. (2012) and Saravanan et al. (2015a, 2015b, 2015c) have shown high performance in the removal of turbid materials, equipment and methodologies employed in some of them are costly and sophisticated.

Physico-chemical treatment is a common technique that has been employed in wastewater treatment, but its performance can be perturbed by high variability and complexity attributed to the presence of some exotic constituents such as toxic metals, toluidine dye, inorganic matter, humic materials, natural organic matter (NOM), etc. NOM is one of the constituents that aggravates wastewater treatment, and it is attributed to a number of its constituents, emerging from plants and animals (charcoal, lignin, wastes from microorganisms, effluent from agricultural depending on the biogeochemical reactions taking place in the environment (Bradby 2006; Listiarini et al. 2010). These factors pose a serious challenge in the treatment of wastewater consisting of NOM as the treatment system requires a suitable design, chemicals and treatment technique. Such colloidal suspension is associated with poor coagulation-flocculation, large sludge production and low removal efficiency. The NOM has a tendency to react with disinfectant to form disinfection byproducts such as trihalomethanes and haloacetic acid. It decomposes at high temperatures to form corrosive organic acids. Some of the studies (Marais et al. 2018) revealed that pre-treatment of wastewater for disinfection using ClO₂ produced better NOM removal compared to Cl₂, a method that seems to be costly.

In view of the fact that conventional treatment includes destabilization, hydrolysis, coagulation, flocculation, nucleation, crystallization, agglomeration and sedimentation, the complexity of some AMD decants poses a serious challenge in producing effluent discharge of good quality. The chemical reactions that occur during the coagulation-flocculation process; that is, destabilization and hydrolysis, have not been explicitly elucidated, thus posing a challenge in identifying effective physico-chemical interactive mechanisms attributed to production of treated effluent of good quality. In the present study, bentonite clay and monohydrate magnesium sulphate are dosed to a synthetic AMD sample (TSS, TDS, NOM, colour, toxic metals, inorganic matter). Bentonite clay commonly known for high salts adsorption due to charged porous tiny particles, whereas monohydrate magnesium is common due to its affinity for water and compatibility with most organic compounds. Another advantage associated with magnesium includes its ability to control calcium and alkalinity in a bulk water, and prevention of precipitation of calcium carbonate (Velimirov & Boehm 1976). It can replace calcium ions in CaCO₃ present in bulk water to form MgSO₄. The challenge prevalent in the removal of NOM is that high molecular weight hydrophobic NOMs has less solubility, where removal is by coagulation process. On the other hand, highly soluble low molecular weight hydrophilic NOMs are not easily removable as they dissolve completely throughout the solution (Uyguner and Beekbolet 2005; Matilainen et al. 2010; Nguyen et al. 2011; Ibrahim and Aziz 2014). NOMs are negatively charged compounds classified into aliphatic and aromatic compounds.

In the present study, bentonite clay, MgSO₄ and their flocculant were dosed in synthetic AMD sample respectively to determine their efficiencies. An advantage associated with inorganic coagulants includes their effectiveness in wastewater treatment at low concentration in the removal of turbid materials (TSS, TDS, NOM, toxic metals and colour). They also produce sludge that is more compact, with a small concentration of residual coagulant in treated wastewater (Stoll 2013). A study conducted by Ntwampe et al. (2013) on paint wastewater revealed that the turbidity removal efficiency using pure Fe salts is high. On the other hand, the choice of bentonite clay in this study is based on its abundance on the earth's crust and porous structure, which reduces turbid materials through sorption and intercalation (Brink 2009, 2012). The advantage of preparing the flocculant investigated in this study is its feldspar component, which is a fluxing agent; it also ameliorates adsorption efficiency.

The aim of the study is to determine the comparison of the efficiency of a flocculant of bentonite clay and MgSO₄ in the removal of composite turbid materials during rapid mixing and shaking. Another objective was to determine the removal of NOM that contains high specific UV₂₅₄ (SUVA₂₅₄) absorbance value. The last objective was to determine the removal efficiency of a flocculant on toxic metals.

In this study the coagulation-flocculation treatment has been applied to the AMD sample using dosages (20–60 mL) of 1.5 g bentonite clay or 0.05 M MgSO₄ respectively. Another set of experiments was conducted using a flocculant prepared by a combination of bentonite clay and MgSO₄. Three sets of experiments were conducted for reproducibility, but the results of two sets of experiments were used as those of the third were identical to those of the second. The statistical analysis of the turbid materials removal efficiency of the AMD samples dosed with a combination of bentonite clay and MgSO₄ is illustrated in Table A1 (Appendix). The pH, conductivity, turbidity/turbid materials and ORP of the samples were measured before and 1 hour after treatment. The treatment methods which were employed during the experiments include mixing and shaking; that is, rapid agitation (250 rpm for 2 minutes), slow agitation (100 rpm for 10 minutes). The order of the experiments is explained under experiments sub-section below.

The samples were collected from the Western Gold Mine in Krugersdorp (South Africa) in a 25 litre plastic drum. The sample was air-tight and stored at room temperature. The turbidity of the sample was raised by the addition of Vaal River water and 0.4 g of dye (100 mL AMD and 100 mL river water in 200 mL glass beaker). The pH, conductivity (EC), turbidity/TSS and ORP of the untreated AMD sample were 4.12, 6.37 mS/cm, 258 NTU (TSS of 882 mg/L) and 302 mV respectively. The composition of synthetic AMD sample is shown in Table 1.

Bentonite clay was obtained from the Yellowstar Bentonite mine, a bentonite mining and supplying company situated in Parys in the Free State. The chemical composition of bentonite clay is shown in Table 2.

Toludine blue dye purchased from Sigma Aldrich (South Africa) was used for coloration of the AMD. A 0.4 g of a dye was added to a litre of raw AMD sample and the solution was stirred thoroughly to ensure normal dispersion.

A stock solution was prepared using anhydous MgSO₄, which was prepared by heating hydrated MgSO₄ to approximately 200 °C. A quantity of the reagents (MgSO₄) was diluted in 1 L of demineralized water. A 0.05 M of Mg²⁺ ions (a concentration obtained from the study conducted by Fasemore 2004) were dosed to the colloidal suspension.

The equipment used for the jar tests was a BIBBY Stuart Scientific Flocculator (SW1 model), which has six adjustable paddles with rotating speeds between 0 and 350 rpm. A 200 mL sample of AMD containing 9.7 g (as measured by filtering 200 mL of the AMD) of solid particles was poured into each of the five 500 mL glass beakers for the test. It is evident that materials present in the AMD consist of turbid materials as the sampling was conducted in a catchment dam; that is, there was no settling by gravity as the AMD was stationary. Rapid mixing was set at 250 rpm for 2 min, followed by slow mixing at 100 rpm for 10 min, a normal standard recommended in a jar test.

A Merck Turbiquant 3000T Turbidimeter (Japan) was used to determine turbidity, or the suspended particles in the supernatant, using NTU as a unit of measure. It was calibrated with 0.10, 10, 100, 1,000, and 10,000 NTU standard solutions. The turbid material was calculated by NTU conversion; that is, by multiplying the turbidity readings by 3.42. The ORP were measured to determine the concentrations of oxygen and redox potential respectively. NB: The pH, EC and ORP were all measured using a SensoDirect Multimeter (South Africa) with an electrode filled with silver chloride solution and an outer glass casing with a small membrane covering at the tip. The equipment was calibrated with standard solutions at pH of 4.0 and 7.0 before use. Parameters were measured using their respective probes, and the instrument was fitted with a ‘temperature correction’ device.

Experiment (A): Jar test with bentonite clay and MgSO₄ dosage respectively, using rapid mixing. pH, conductivity, turbidity and ORP of the sample were measured. Five 500 mL glass beakers were filled with 200 mL samples of synthetic AMD sample with parameters mentioned above. The AMD samples were dosed with 20, 30, 40, 50 and 60 mL of the reagents mentioned above; treated in a jar test at 250 rpm for 2 minutes and reduced to 100 rpm for 10 minutes. The sample was allowed to settle for 1 hour, after which the pH, turbidity and ORP were measured. A similar set of experiments was conducted using a shaker for chemical dispersion.

Experiment (B): Jar test with a mixture of bentonite clay and MgSO₄ dosage involving using similar treatment method and measurements. A similar set of experiments was conducted using a shaker for chemical dispersion.

A KYKY-EM3200 digital scanning electron microscope (SEM; model EM3200) (China) was used to produce the SEM photomicrographs.

A Perkin Elmer Optima DV 7000 ICP-OES optical emission spectrometer (USA) was used to determine the metals in the supernatant of the AMD samples. It was calibrated with a standard solution between 2 and 50 mg/L of the salts mentioned above.

The measurement was conducted using a Cary 300 UV-Visible Absorbance at 254 nm (UV₂₅₄) to determine aromatic or C double bonds of organic matter. The solutions were stirred at 45 rpm for 10 min, followed by 30 min of quiescent settling. The samples were allowed to settle, after which filtration was conducted using 0.45 μm cellulose acetate membrane (Sartorius) before UV absorbance and DOC. Calibration carbon standard solutions were prepared with concentrations of 1.0, 5.0, 10, 20 and 30 mg/L using potassium hydrogen phthalate.

Langmuir and Freundlich models are normally employed in adsorption experiments to investigate adsorption capacity of an adsorbate onto an adsorbent; that is, turbid materials onto the flocculant. Pseudo-first and second order are common models, as shown by Equations (3) and (4).

A plot of log q_e vs log C_e should be linear if the model fits the experimental data.

The main objective of this study was to investigate the efficiency of a flocculant consisting of bentonite clay and MgSO₄ in the removal of a composite AMD sample. Although AMD contains SO₄²⁻ ions, more acidity was added to increase ionic strength with diprotic salt, which dissociates two-fold, behaving as a buffer during hydrolysis; this ameliorates the rate of hydrolysis, unlike in monoprotic metal salts (Ntwampe 2013). Research (Naceradska et al. 2019) states that the most effective flocculation is achieved at low pH levels due the reduced electrostatic repulsion between colloids, leading to a greater chance of polymer bridging due to expansion of the polymer chains.

Figure 1 and Figure A1 (Appendix) represent the pH, conductivity and ORP of the AMD samples dosed with bentonite clay and MgSO₄ and a flocculant (a combination of bentonite clay and MgSO₄), with mixing and shaking respectively.

The pH of the samples dosed with bentonite clay, with mixing and shaking (Figure 1) show a slight decreasing inconsistent trend from 4.12 (untreated AMD) to the ranges 4.08–4.12 and 4.07–4.10 respectively. The deviation of the pH values between the two sets of experiments is insignificant. The pH of the samples dosed with MgSO₄ with mixing and shaking show a slight inconsistent decreasing trend from 4.12 (untreated AMD) to the ranges 4.08–4.12 and 4.07–4.10 respectively. The deviation of the pH values between the two sets of experiments is insignificant. The pH of the samples dosed with MgSO₄ with mixing and shaking show a lower decreasing trend in the ranges of 4.0–3.68 and 3.90–3.68 respectively. MgSO₄ yielded a slightly lower pH changing trend compared to bentonite clay. This is attributed to pH depression in the colloidal suspension caused by increasing SO₄²⁻ ions from MgSO₄ with dosage. On the other hand, bentonite clay does not exhibit protonation during mechanical agitation. The results also show that the effect of deprotonation during hydrolysis of Mg²⁺ ions is insignificant. The pH of the samples dosed with a flocculant with mixing and shaking (Figure A1) shows a decreasing trend in the ranges 4.04–3.70 and 3.81–3.68 respectively. The samples dosed with shaking exhibit a lower changing trend compared to those dosed with mixing, including those of the samples dosed with bentonite clay and MgSO₄ respectively (Figure 1).

The EC of the samples dosed with bentonite clay with mixing and shaking show an increasing trend from 6.37 (untreated AMD) to the ranges of 6.45–6.68 and 6.42–6.62 mS/cm respectively. On the other hand, the EC of the samples dosed with MgSO₄ also show a slightly higher increasing trend in the ranges 6.55–6.70 and 6.50–6.99 mS/cm respectively. This is obviously attributed to dissolution of MgSO₄ to form Mg²⁺ and SO₄²⁻ ions, thus increasing the ionic strength of a solution. The EC of the samples dosed with a flocculant with mixing and shaking (Figure A1) show an increasing trend in the ranges 6.56–6.75 and 6.62–6.81 respectively. The results are slightly higher compared to those of the samples dosed with bentonite clay and MgSO₄ respectively (Figure 1).

The ORP of the samples dosed with bentonite clay with mixing and shaking shows a slight decreasing trend from 302 mV (untreated AMD) to the ranges 300–291 and 298–288 mV respectively. The changing trend is indicative of oxidation reaction during destabilization-hydrolysis, predominantly on the toxic metals and other oxidizable constituents. On the other hand, the ORP samples dosed with MgSO₄ with mixing and shaking show a slightly lower decreasing trend compared to that of the samples with bentonite clay; that is, in the ranges 296–277 and 294–280 mV respectively. The observation is also attributed to oxidation of oxidizable constituents, which also indicate the higher oxidation potential of the MgSO₄ compared to bentonite clay. The results obtained in the pH and ORP measurements show a correlation between the two: the samples with low pH show low ORP. This indicates that the rate of destabilization-hydrolysis is directly proportional to the rate of oxidation. The ORP of the samples dosed with a flocculant with mixing and shaking (Figure A1) show a slightly increasing trend in the ranges 6.54–6.65 and 6.48–6.86 mV respectively when compared to that of the samples dosed with bentonite clay and MgSO₄ respectively (Figure 1).

Figure 2 shows turbid material removal efficiencies of bentonite clay, MgSO₄ and a flocculant dosed in AMD samples mixing and shaking.

Turbidity removal efficiencies of bentonite clay with mixing and shaking (Figure 2) are in increasing trend in the ranges 60–74 and 63–70% respectively. Turbidity removal efficiencies of MgSO₄ with mixing and shaking (Figure 2) show a lower increasing trend compared to those of the samples dosed with bentonite clay, in the ranges of 48–60 and 47–58% respectively.

On the other hand, turbidity removal efficiencies a flocculant is higher compared to those of the samples dosed with bentonite clay and MgSO₄ respectively, in the ranges of 91–96 and 88–97% respectively. Efficiencies of the samples with higher dosages during shaking are slightly higher compared to those of the samples with mixing. According to the results (Figure 2), MgSO₄ is not an ideal coagulant, but it has an ability to replace other metal ions in a compound to mitigate their detrimental effect, such as replacement of Ca²⁺ in scale-forming CaCO₃. On the other hand, the rate of hydrolysis still occurred in the system dosed with MgSO₄, as some of the OH⁻ ions that resulted in Mg(OH)₂ species are released during the cleavage of the water molecules (Equation (1)), thus regulating the pH of the colloidal suspension. In addition, hydrophobicity or hydrophillicity plays a pivotal role by determining the rate of particle-liquid separation during destabilization-hydrolysis (Ntwampe et al. 2015a, 2015b). The efficiencies obtained in the samples with mixing are slightly higher compared to those in the samples with shaking, which is attributed to uniform chemical dispersion by steady agitation of a shaker allowing the formation of larger flocs. That also confirms that severe shear forces due to high impeller speed result in re-stabilization (Sharp et al. 2006a, 2006b), resulting in poor removal of turbid materials.

In the case of bentonite clay, optimal removal of turbid materials is attributed to its ion exchange capacity, which is dependent upon both the surface and the negative charge; where the former is attributed to the pH level and the latter to the isomorphous exchange of charges in the tetrahedral and octahedral sheets of montmorillonite. The former is balanced due to exchangeable cations whereas the latter is caused by protonation and deprotonationation of the edge sites. Cation exchange capacity (CEC in meq/100 g clay), the ability of a cations to replace another cation attached on the surface of the clay at certain pH values, is another factor that adds to its performance (Figure 2). Such a cation is also an index cation and has to be larger in order to replace an exchangeable cation. The quantity of multivalent toxic metal cations obviously has to be greater than the CEC of the clay, hence it was inevitable for turbid materials to be optimally adsorbed (Figure 2), and such observation is invoked by the observation obtained in the study by Bergaya et al. (2006).

Figure 3 shows the amount of NOM, and total organic carbon (TOC) and dissolved organic carbon (DOC) are applied in the determination using UV₂₅₄ for characterization (Matilainen et al. 2010). It has been deemed unnecessary to determine specific ultraviolet absorbance due to high turbidity removal efficiencies shown by the results (Figure 2), showing that turbidity consists predominantly of high molecular weight NOM, SUVA being greater than 4 (Matilainen et al. 2010).

The results (Figure 3) are reflective of the removal of turbid materials, predominantly aromatic NOM as stated (Thebe et al. 2000). However, the present study focuses on the composite removal of turbid materials. The results show the lowest decreasing absorbance in the samples dosed with a flocculant (both mixing and shaking). On the other hand, the results show higher absorbance in the samples dosed with MgSO₄ (mixing slightly higher than shaking), whereas that of the samples dosed with bentonite clay was slightly lower than that of the samples dosed with MgSO₄. Low UV₂₅₄ absorbance is indicative of low turbid materials content in treated effluent of the samples dosed with a flocculant, which also indicates higher UV₂₅₄ in the treated effluent of the samples dosed with MgSO₄ is indicative of higher turbid materials content. Those results correlate with those obtained in Figure 2; that is, flocculant showing high removal efficiency compared to both bentonite clay and MgSO₄. The results also confirm that removal of the hydrophobic fraction of turbid materials (NOM) was through the coagulation-flocculation phenomenon (Thebe et al. 2000). The results also invoke the changing trend of the ORP of the samples dosed with a flocculant (Figure 1), showing higher rate of oxidation of toxic metals, resulting in precipitation of metal hydroxides. It is also conceivable to observe optimal NOM removal using a flocculant consisting of bentonite clay and MgSO₄ reagents as they exhibit a positive charged property (cationic) in ionic colloidal suspension. On the other hand, the negatively charged NOM is attributed to its hydrophilicity and hydrophobicity (anionic); the cationic component of the reagents reacts with the anionic components of the NOM to form settleable products. The flocculant showed high efficiency in the removal of colour, a parameter known to be removable by Fe salts (Aboulhassan et al. 2006).

Furthermore, the removal of toxic metals such as Co, Cu, Ni, Pb, Sb, Se and Zn from 4.117, 6.722, 4.34, 6.155, 4.083, 5.897 and 6.814 mg/L (untreated AMD, Table 1) to 1.138, 0.059, 0.375, 1.218, 1.092, 0.795 and 1.423 mg/L (Table 3) respectively, also invoke the results obtained (Figures 2 and 3). The observation indicates that the colloidal suspensions in the system were more hydrophobic and less hydrophilic, which requires more physisorption and less chemisorption reactions. However, some of the minerals dissolved due to low pH and required chemisorption, hence chemical treatment was necessary. Turbid material removal values show high removal efficiencies; that is, from 244 mg/L (untreated AMD) to a range of 3.6–9.3 mg/L, a percentage in a range of 96.1–98.9%.

Despite the particle size of the bentonite clay (Table 4), the minerals attached to the porous sites result in a higher surface charge, which increases sorption efficiency, a process that occurs due to increasing interaction between approaching opposite charges (binding sites and the turbid material).

Table 4 illustrates three different particle sizes; that is, 60, 180 and 220 μm, which provide bentonite clay with a larger surface area suitable for optimal adsorption of turbid materials. Despite the surface area having been identified as one of the factors that ameliorates thermal-kinetic reaction, excessive milling/attrition is also uneconomical as it may cause wear and tear and high energy consumption. Particle size of 220 μm was recommended in this study as it was envisaged that it enabled adequate mineral liberation for electrochemical reactions, including ideal particle sizes for optimal sorption. On the other hand, the minerals attached to the porous sites result in a higher surface charge, which increases sorption efficiency, a process that occurs due to increasing interaction between approaching opposite charges (binding sites and the turbid material). The optimal removal of turbid materials from the AMD sample (Figure 2 and Table 3) invokes the use of 220 in highly concentrated wastewater.

Figure 4 represents adsorption kinetics of the pollutants present in the AMD during mixing and shaking using a flocculant employing the pseudo-second order model. It was deemed unnecessary to plot the pseudo-first order model due to the accuracy of the second order model and limited number of graphs required.

The experimental data shown by plotting t/q_t vs. t (Figure 4) shows correlation regression (R²) of 0.967 (96.7%) for adsorption capacity of turbid material present in the AMD samples dosed with a flocculant treated with shaking. On the other hand, the correlation regression for t/q_t vs. t of the AMD samples dosed with a flocculant treated with mixing is 0.959 (95.9%). The correlation regressions for t/q_t vs. t between mixing and shaking are close to a unit (1); this indicates that the pseudo-second order is the best fit for the data of both sets of experiments.

Figure 5 represents the adsorption isotherm of the Langmuir and Freundlich adsorption models showing adsorption capacities of the turbid materials present in the samples. The results plotted are obtained from the experimental data of AMD sample dosed with a flocculant with shaking.

The results obtained from the experiment (Figure 5) show a correlation regression plotting Log q_e vs. log C_e applying the Freundlich model to be 0.989 (98.9%); whereas the correlation regression applying the Langmuir model is 0.829 (82.9%). This shows that the experimental data exhibits a best mathematical fit for Freundlich adsorption isotherms compared to the Langmuir model.

Figure 6 represents the SEM micrographs of the sludge of the AMD sample dosed with a flocculant with mixing (Figure 6) whereas Figure A2 represents the SEM micrographs of sludge of the AMD sample dosed with a flocculant with shaking.

The SEM micrographs of the samples dosed with a flocculant with shaking (Figure 6(a)) show sponge-cake-like dense non-spherical structures surrounded by smaller structure. The images show the swollen structures that are protruding as a result of sorption. On the other hand, the micrograph of the AMD sludge with a flocculant dosage and shaking (Figure 6(b)) also shows sponge-cake like dense structures mostly concentrated on the left side with some smaller structures on the right side close to one another. Figure 6(b) shows more voids around the dense structures, which is indicative of floc rupture due to high shear stresses, whereas there are fewer voids shown in Figure 6(a). Despite the optimal turbid materials sorption, efficiencies revealed by the flocculant, Figure 6(b) shows less area covered by the structural images compared to Figure 6(a), showing multiple layers consisting of dense and small structures responsible for the removal of turbid materials.

According to the chemical composition of the bentonite clay (Table 2), it is evident that the rate of reactivity taking place in a flocculant was high, more especially considering ionic exchange (CEC) and electrochemical reaction. The adsorption efficiency of the flocculant (Figure 2) and toxic metals removal (Table 3) and UV₂₅₄ absorbance (Figure 3) invoke the morphological swollen structure of the micrographs (Figure 6) that optimal sorption was attributed to physico-chemical phenomenon of the flocculant.

The determination of the efficiency of a combination of bentonite clay and MgSO₄ was successfully investigated. Intuitively, the use of MgSO₄ as a reagent in a highly concentrated AMD sample has proven beyond the envisaged reactions dynamics involved during destabilization-hydrolysis, a determinant to nucleation, aggregation and sedimentation. On the other hand, the efficiency of bentonite clay in co-removal of turbid materials (TSS, TDS, NOM, colour, toxic metals) was extensive. The experimental results show that optimal removal of turbid materials is not dependent upon the use of costly sophisticated process reagents (coagulants/flocculants), but economically viable reagents that are abundant and effective.

A flocculant (bentonite clay and MgSO₄ exhibit different reactivity compared to that of each reagent individually, i.e. depressing the pH of a solution, increasing the EC and showing redox reaction. Based on highly turbid material removal efficiency exhibited by a flocculant, it shows high destabilization-hydrolysis potential, high NOM, toxic metals, colour and inorganic matter. Turbid materials removal results for the samples dosed with a flocculant with mixing are slightly identical to the results of the samples dosed with the same flocculant with shaking, the former showing a slightly lower performance compared to the latter. Rapid mixing is attributed to production of treated effluent of poor quality due to floc rupture being responsible for de-flocculation and re-stabilization. The finding indicates that gentle mixing (shaking) is an ideal method to employ for mechanical agitation. The flocculant has a high sorption capacity as demonstrated by the crystal morphology of the SEM micrographs.

Based on the physical and chemical properties of both bentonite clay and MgSO₄, i.e. porosity, T-O-T structure, high ionic exchange property (CEC), acidity, high solubility, oxidizing potential, among other, the removal of turbid materials is a physico-chemical phenomenon. Dense (sponge-like) flocs shown by SEM micrographs are indicative of the maximum mass transfer of the colloidal particles through sorption.

The determination of this study reveals that a combination of bentonite clay and MgSO₄ is an ideal cost-effective flocculant to be employed in the treatment of composite AMD.

All relevant data are included in the paper or its Supplementary Information.