Novel hybrid metal loaded chelating resins for removal of toxic metals from acid mine drainage

Mining and mineral processing activities are very important with respect to the economic and social benefits for the country in which the mining activities are conducted. However, if not well managed, the environmental consequences of mining activities are massive because mining produces dumps containing metal sulphide ores covering vast areas of the mine site. When these metal sulphides, especially pyrite (FeS₂), are exposed to air and water they can easily result in acid mine drainage (AMD) (Mulopo 2015). In addition, the mine tailings and waste rocks are smaller in size than their natural geologic materials, thus have a greater surface area and are more prone to generation of AMD (Johnson 2003). The environmental pollution caused by AMD generated from gold mines is a serious worldwide problem due to the presence of other mineral ores that occur with the gold-bearing mineral deposits (Coelho et al. 2011), which upon oxidation, exacerbate AMD pollution.

In addition to low pH, AMD is characterized by high levels of metal ions, which have devastating effects on the environment. Toxic metals are recalcitrant bio-accumulative systemic toxins that can affect both soft and hard tissues in the human body (Simate & Ndlovu 2014), resulting in serious ill-health consequences that are translated to very high social costs for the population affected (Larsson et al. 2018). Metals contained in AMD impair the integrity of the whole aquatic ecosystems (Masindi et al. 2015) and, in addition, all areas affected by AMD and improper mine closures end up with deteriorated land that is unfit for both human settlement and farming. The communities around the AMD contaminated areas are therefore often forced to relocate to other safe areas due to health risks and consequences associated with AMD contamination (Ochieng et al. 2010).

Generally, remediation of AMD polluted water bodies is of paramount importance because once AMD is produced, it persists for centuries, and it is even worse in abandoned mines, polluting the environment in perpetuity (Kalin et al. 2006; Simate & Ndlovu 2014). Hence, removal of metals from acid mine water is crucial, even when the metals exist in trace levels.

The AMD problem in the environment has been a recognised environmental catastrophe, second only to global warming and stratospheric ozone depletion in ecological risk, by the United States Environmental Protection Agency (US EPA) since the year 1987 (Durand 2012; Moodley et al. 2017). Therefore, several AMD remediation technologies have been developed worldwide to address this problem adequately and sustainably. Available technologies that are commonly used for metal pollution remediation include reduction/oxidation (Zhang et al. 2017), precipitation (Park et al. 2014), membrane filtration (Fu & Wang 2011), ion-exchange (Wang et al. 2002) and adsorption (Luo et al. 2013). Chemical precipitation produces high-density sludge, containing metals, that risks secondary pollution upon disposal to the environment. Membrane filtration is another technology that has been reported but its limitation comes from the fact that it suffers from frequent fouling, leading to high maintenance costs. Ion-exchange boasts of high treatment capacity, fast kinetics and high removal efficiency and is a simple technology but it is very expensive and produces concentrated metal waste that is difficult to dispose of (Shilpa et al. 2016). Thus all these technologies, excluding adsorption, suffer from high capital and maintenance costs coupled with the risk of secondary pollution from the high quantities of chemical sludge and effluent produced (Sun et al. 2018). In adsorption, secondary pollution may be minimized by sequential recovery of the adsorbed metals as saleable products before disposing of the exhausted adsorbents. Adsorption-based remediation approaches are a lucrative technology because of their facile nature, fast kinetics, cost-efficiency and eco-friendliness (Acelas et al. 2015).

Transition metal oxides are known metal scavengers in the environment, and hence their application as adsorbent materials have proven to be beneficial in many applications (Zhou et al. 2018). Metal oxides, in particular, are among the best adsorbents because their size dimensions are in the range of nanoscales and they also have plenty of reactive atoms exposed on the surface (Zheng et al. 2011). These atoms react with the metal ions in the liquid phase, and hence they are removed from the solution through adsorption mechanisms. Although the metal oxide nanoparticles (MONPs) are excellent adsorbents for metal ions, they also suffer from aggregation and lack of mechanical strength to withstand flow-through systems (Yang et al. 2018). Therefore, MONPs are usually dispersed onto organic materials for support, producing hybrid metal oxide nanosorbents. The organic host also prevents their aggregation. Hybrid nanosorbents exhibit excellent adsorption efficiencies because both the organic support and the dispersed nanoparticles are involved in the adsorption process.

Ion-exchange resins have been used as support material for metal oxide nanoparticles producing hybrid nanosorbents for both cationic and anionic inorganic pollutants from wastewaters as suggested by (Cumbal et al. 2003). Much work has been reported on the development of ion-exchange resin supported iron (Fe) oxides for remediation of both organic and inorganic pollutants from polluted aquatic environments or wastewater (Acelas et al. 2015). It has also been reported that zirconium (Zr) and titanium (Ti) oxides embedded onto polymeric resins for support can enhance metal removal from wastewaters (Zheng et al. 2011; Zhang et al. 2013). To the best of our knowledge, there is no study that reports on the synthesis of hybrid adsorbents by supporting MONPs on a commercial weak acid chelating resin for metal removal from AMD.

The chelating resin is a good candidate for use as support for the MONPs because it has been reported that chelating resins surpass both cation and anion exchange resins in metal removal from solutions (Dabrowski et al. 2004). Chelating resins not only bind metal ions through their ion-exchange sites but are also able to bind them through coordination with their ligand atoms, a feature that both cation and anion exchangers lack. Thus, it was anticipated that supporting MONPs on chelating resins can produce composite adsorbents of excellent metal removal efficiency from wastewaters.

In this study, a macroporous polystyrene cross-linked with DVB weak acid chelating (methylaminophosphonate functional group) resin was used as a support for Fe, Zr and Ti oxides nanoparticles, producing new hybrid adsorbents namely: HFO-P, HZO-P and HTO-P, respectively, which were used for the removal of metal ions from the complex AMD. It was envisaged that the high porosity, fixed negative charges and chelating functional groups of the host resin would enhance the adsorption by increasing the diffusivity of the metal ions.

All chemicals used in the experiments were of analytical grade purchased from Sigma-Aldrich, Johannesburg, South Africa. A commercial weakly acidic chelating macroporous polystyrene cross-linked with divinylbenzene (DVB) resin of the di-Na⁺ form (R–NH–CH₂–PO₃²⁻) was used as a host material for Fe, Zr and Ti oxides, producing the hybrid organic/inorganic nanocomposites HFO-P, HZO-P and HTO-P, respectively. Three 50 g portions of the host resin were weighed into a 500 mL conical flask pre-washed according to standard operation procedures. The resins were each washed using 250 mL deionized (DI) water by agitation for 2 h using an orbital shaker set at a speed of 150 rpm. This procedure was carried out three times with fresh DI water each time. Solutions of 1 M ammonium ferric sulphate dodecahydrate (NH₄Fe(SO₄)₂·12H₂O), 0.1 M zirconium(IV) sulphate hydrate (Zr(SO₄)₂·xH₂O) and 0.1 M titanium(III) sulphate hydrate (Ti₂(SO₄)₃·H₂O) were prepared using DI water and used to load the respective metal ions into the host resin. The full synthesis procedure of the hybrid materials was carried out in three steps. The first step was the loading of the metal into the 50 g washed resin exchange sites by shaking with 250 mL loading solution in an orbital shaker set at 150 rpm at room temperature for 24 hours. The supernatant was then decanted and discarded appropriately. The second step involved the simultaneous desorption and precipitation of the metal ions loaded into the resins by shaking with 250 mL of 1 M each of a binary sodium sulphate (NaSO₄)-sodium hydroxide (NaOH) solution under the same conditions as in the previous step and the solution was then decanted and discarded. The resin beads were washed with DI water more than 10 times to remove all residual binary solution. The clean resins were finally dried in an oven set at 40 °C overnight for 16 h to stabilize the incorporated hydrated metal oxides. Hence, the hybrid chelating ion-exchange metal oxide resins (HLIX) were produced for the purpose of this study.

Physical characterization was carried out on both the host and the hybrid resins. The crystallinity of the resins was studied using X-ray diffraction (XRD) analysis using a Rigaku Smartlab X-ray Diffractometer at room temperature using Cu–Kα radiation (λ = 0.154059 nm) operated at 45 kV and 200 mA in a 2θ range of 5–90° and speed 2° min⁻¹. A resin bead for each nanocomposite was dissected, stuck on double carbon tape and coated with carbon using a spatter coater model Quorum Q150T ES. The hybrid resin's morphology and qualitative qualities were studied on the cross-sectional area using high resolution scanning electron microscopy (HRSEM) model JEOL JSM-7800F Field Emission Scanning Electron Microscope coupled with Thermo Scientific Ultra-dry energy-dispersive X-ray spectroscopy (EDS) detector. Functional groups on the adsorbents' surfaces were investigated using a Fourier transform infrared (FTIR) spectroscope (PerkinElmer FTIR spectrometer Frontier, Spectrum 100 spectrometer)) following the attenuated total reflection (ATR) method in the range 400–4,000 cm⁻¹ at a resolution of 4 cm⁻¹.

Adsorption experiments were carried out using the batch equilibrium technique. Synthetic solutions containing about 50 mg/L each of Al(III), Fe(II) and manganese (Mn)(II) (the main metal constituents of AMD) were prepared in 3,000 mg/L sulfate (SO₄²⁻) and used to model the adsorption efficiency of the synthesized nanosorbents. The adsorption experiments were carried out at the natural pH of the synthetic solution, pH 1.7–1.8, to mimic the acidic conditions of AMD. The batch tests were carried out in 100 mL Erlenmeyer flasks using 50 mL of synthetic solution in an orbital shaker set at 200 rpm, at 25 °C. Each flask was covered with parafilm to avoid evaporation during the adsorption process. A fixed amount of adsorbent (20 mg to the nearest unit) was used to study the effect of contact time (5, 10, 20, 30, 60, 180 and 360 min) on Al(III) adsorption. The removal of Al is important because in AMD, at pH < 3, Al exists as ions, which pose serious environmental risks. The Al(III) is as toxic as the so-called heavy metals; it causes bone disease, Alzheimer's disease, neuronal atrophy and cancer in humans, while in plants it inhibits nutrient uptake by the roots (Jaishankar et al. 2014). Effect of adsorbent dosage (0–12 g/L) for the pre-determined times on adsorption efficiency was conducted. Effect of initial solution pH (1.5–3.5 and 9.5–11.5) on the adsorption of Al(III) was investigated at the pre-determined time and resin dosage; Al(III) precipitates between these two pH ranges, and hence its adsorption cannot be conducted in solutions in this pH range. The pH was adjusted appropriately by dropwise addition of either 1 M HCl or 1M NaOH solutions. All pH measurements were taken using an XS PC80, EU designed, pH meter. The rate of Al(III) adsorption on the hybrid nanosorbents was investigated using the pseudo-first-order, pseudo-second-order and the intraparticle diffusion kinetics models. Three two-parameter isotherm models, Freundlich, Langmuir and Temkin, were used to study the mechanisms of the adsorption. Regeneration of the loaded hybrid resins was carried out using NaCl–NaOH binary solution (3% w/v of each) and 0.5 M hydroxylamine hydrochloride (NH₂OH.HCl) solution as regenerant solutions. The loaded resins were shaken at 180 rpm with 50 mL of regenerant solution for 6 h.

Finally, HFO-P, HZO-P and HTO-P were used to remove ten selected metals (Al, cadmium (Cd), cobalt (Co), chromium (Cr), copper (Cu), Fe, Mn, nickel (Ni), lead (Pb), zinc (Zn)) from environmental AMD obtained from a defunct gold mine located in the western part of Johannesburg, South Africa. The level of metal ions in the solutions was determined using inductively coupled plasma-optical emission spectrometry (ICP-OES, Agilent Technologies 700 Series. All experiments in this work were carried out in triplicate and the average results were reported. The hybrid adsorbents were also applied to a blank solution of dilute H₂SO₄ (sulphuric acid) containing 3,000 mg/L SO₄²⁻ to investigate their contribution in the change in pH observed after the adsorption process. Blanks were also run together with the samples to establish the effect the glassware had on the adsorption of the metal ions.

The adsorption of metal pollutants onto the polymer-supported metal oxide sorbents can either be through physical or chemical processes, that is physisorption or chemisorption, respectively. The physical sorption mechanisms are ion-exchange and electrostatic interactions. Hybrid metal oxide nanocomposites have surface H⁺ ions, especially on the hydrated MONPs, and Na⁺ ions on the surface of the polymeric material. These Na⁺ and H⁺ can be ion-exchanged with the metal cations in solution, hence depositing them on the adsorbent surface (Zhou et al. 2018). This type of metal removal is most likely to occur in low pH solutions, where the composite metal oxide surface is protonated. Electrostatic (Coulombic) interactions between the sorbent and metal cations are likely to occur at high pH conditions where the abundant hydroxide (OH⁻) deprotonates the surface of the adsorbent to give it a negative charge. Metal ions are attracted onto counter-charged adsorbent surface, and hence removed from solution (Guaya et al. 2016). This removal mechanism is also possible with metal oxyanions at low pH where the adsorbent surface is positive due to protonation.

The chemical sorption mechanisms are chelation/complexation and reduction/electron transfer (Zhou et al. 2018). Complexation is chemisorption whereby the H⁺ on the functional group (like carboxylic acid (–COOH), phosphonic acid (–PO.2OH), hydroxyl (–OH), metal hydroxide (MOH)) of the adsorbent is exchanged with metal ions in solution, which then chelates with the O atom. This process is sometimes called hydroxylation because a water molecule is released in the process. If the metal ion being removed from solution does not have a vacant d-orbital, like Al(III), the complex is produced by chelation of the O atom of the Al hydration radius (aquo –OH) with the central metal for the metal oxide adsorbent. This mechanism is also called ligand exchange because the metal ion must release its aquo –OH ion and replace it with the O atom of the adsorbent or vice versa. Removal of metal ions from solution through electron transfer is also possible. Some metal oxide adsorbents, like titanium dioxide (TiO₂) in the presence of light, exert reducing properties on their surfaces. The metal oxide produces electrons that are then transferred to the metal cations, reducing them to oxidation state 0. The produced metal atoms are then deposited on the adsorbent surface, and hence removed from the solution. In organic/inorganic nanocomposites the pores present in the polymeric support material provide enhanced adsorption sites through increasing the surface area of the adsorbent. The bigger the pore size, the higher the diffusion rate, and hence the better the removal of metal ions. Therefore, macroporous polymeric materials are an excellent choice as support for metal oxide nanoparticles for metal pollution remediation in the environment.

Figure 1 shows the EDS analysis results for the hybrid HFO-P, HZO-P and HTO-P adsorbents. In light that the scanning electron microscopy–energy dispersive X-ray spectroscopy (SEM–EDS) analysis was conducted on the cross-sectional area of a single sliced bead of each hybrid resin, the results illustrated in Figure 1 show that all the metal oxides were successfully incorporated within the host resin. The energy dispersive X-ray mapping show the metal distribution pattern, indicated by particles, on the cross-sectional area of each sliced hybrid resin bead, which is a confirmation that the hybrid chelating resin/metal oxide nanoadsorbents were produced.

The FTIR-ATR results of the three nanocomposites and their host resin are illustrated in Figure 2(a). The broadband observed at 3,300 cm⁻¹ and peaks at 1,600 cm⁻¹ in all the adsorbents are due to O–H vibrational stretching and bending, respectively. The peaks observed at 1,100 cm⁻¹ and 950 cm⁻¹ may be ascribed to the two P–O stretching of the phosphonate groups of the host material and its hybrid nanocomposites. In another study, in different media, the phosphonate P–O stretching at 1,016 cm⁻¹ and 966 cm⁻¹ has been reported (Bingol et al. 2018). Thus, the FTIR-ATR results show that the incorporation of the metal oxides did not change the surface functional groups of the host resin. That is, there is no bond formed between the metal oxide and the resin functional groups, but the MONPs are trapped by the cross-linkages of the DVB within the resin bead. This is an intended outcome because both the host resin and the dispersed Fe, Zr or Ti oxide nanoparticles should complement each other in the adsorption process. The host resin is envisaged to enhance metal ion permeation, by the Donnan membrane effect, concurrently concentrating the analyte within the resin beads for effective adsorption by the dispersed metal oxide nanoparticles. The XRD spectra illustrated in Figure 2(b) show that the hybrid nanosorbents are all amorphous in nature. The broadband at 2θ = 23° for HTO-P is attributed to the distorted (poor) crystal lattice of TiO₂ (Borai et al. 2015), which is categorised as amorphous; crystalline lattices give sharp resolute peaks. Being amorphous is a desirable feature of adsorbents. This is because the more amorphous an adsorbent material is, the greater the surface area and the more active sites are available, and hence higher adsorption efficiency is envisaged.

The adsorption efficiencies of HFO-P, HZO-P and HTO-P hybrid nanosorbents for Al in a multi-element (Al³⁺, Fe²⁺, Mn²⁺) solution as a function of contact time are shown in Figure 3. Generally, the removal rate of Al(III) was found to be very high in the first 30 min for all three adsorbents; thereafter it slowed down for the rest of the time. This observation may be attributed to the abundance of adsorption sites available on the hybrid resins' surfaces, which pick up the analyte very fast from solution. The enhanced adsorption rate may also be facilitated by the high concentration of the Al(III) ions in the solution which induces a high concentration gradient to the surface of the nanosorbents by mass transfer. The consecutive slower adsorption rate may be ascribed to the increased number of occupied adsorption sites and the decreased Al(III) concentration in the solution. These two factors lead to a slower adsorption rate of the Al(III) from the solution to access the few available adsorption sites on both the exterior and interior surfaces of the resin bead. Ultimately, all the adsorption sites are filled and the adsorbent becomes exhausted and no more metal removal is observed. The optimum adsorption time for HFO-P, HZO-P and HTO-P were 60, 360 and 180 min, respectively. Results on the removal of the coexisting Fe(II) and Mn(II) from the solution by the adsorbents are supplied in Figure S1 in the Supporting information. The removal of Al(III) was higher than that of the coexisting ions in the solution. This observation may be attributed to the smaller radius and higher charge of the Al ion compared to the two other coexisting analytes in the solution, which are Fe(II) and Mn(II) ions.

The percentage removal of Al(III) ions from solution increased with increasing sorbent dosage for all three hybrid adsorbents as illustrated in Figure 4. HZO-P removed 100% Al(III) ions at an adsorbent dosage of 8 g/L, HTO-P adsorbed 98.2% Al(III) ions at dosage 12 g/L and HFO-P removed 62.4% Al(III) ions at dosage 12 g/L. The increase in adsorption with increasing adsorbent dosage is attributed to the increase in adsorbent surfaces offering additional fresh adsorption sites for the uptake of the analyte (Zhang et al. 2017). The lower efficiency of HFO-P can be attributed to the possible destruction of the nanocomposite adsorbent through protonation–dissolution illustrated in Equations (12) and (13). The occurrence of protonation–dissolution of the HFO-P adsorbent is confirmed in Figure S2(a) in the Supporting information. The negative percentage Fe removed shows that there was more Fe in the solution after the adsorption process than in the initial feed solution. The higher Fe level in the final solution is attributed to the leaching of the hydrated Fe oxide that is incorporated in the hybrid adsorbent into solution; which is the only source of the additional Fe in solution. The destruction of the inorganic part of the composite adsorbent in highly acidic media is supported by the little Mn adsorbed by HFO-P as the only organic component involved in its adsorption.

HZO-P and HTO-P are chemically very stable; they do not dissolve in the low pH solution, and hence do not suffer from adsorbent destruction through protonation, like HFO-P. Although the removal efficiency of the adsorbents increased with dosage, the capacity (q) decreased with increasing adsorbent dosage. This result may be ascribed to the depleted Al(III) ions in the solution such that the additional active sites of the adsorbent have no analyte to adsorb. This observation may also be attributed to the overlapping of the resin beads due to overcrowding, causing some adsorption sites to be unavailable to for adsorption of the solution, which already has very little analyte, leading to a lower mass analyte adsorbed per gram of resin.

Thereafter, the adsorption efficiency dropped with the increase in solution pH, showing that regeneration of the HFO-P can be achieved using acidic solutions of about pH 3. Similar results are reported by Pan et al. (2010) who efficiently regenerated HFO-001 at pH 3. Adsorption is highly unfavourable in highly alkaline solutions of pH > 11. This result was because the pH_pzc (10.2) of the hybrid adsorbent was exceeded and its surface charge was negative, which repelled the Al(III) complex ion (Al(OH)₄⁻), which prevails in alkaline pH solution.

Figure 5(b) shows that the HZO-P removed Al(III) ions from solutions of pH 2.01 and 2.40 yielding solutions of pH 6.57 and 7.41, respectively. Adsorption of Al(III) ions onto HZO-P was pH-dependent because the removal efficiency was highest in acidic pH (1.5–3.5), achieving 100% Al(III) removal, and very low in alkaline pH. It is noteworthy that hydrolysis of Al(III) at solution pH 4 might have contributed to its total removal from solution rather than just by adsorption. The pH-dependency observed may be ascribed to the joint adsorption onto both the fixed negatively charged functional groups of the host resin and the dispersed hydrated zirconium oxides. The Al(III) cations (AlSO₄⁺ and Al³⁺) in the solution, which are dominant in low pH), exchange sites with the mobile Na⁺ of the host resin, and on the HZO NPs, the Al(III) ions exchange sites with the surface H⁺. The adsorption of Al(III) onto the HZO-P in alkaline pH was unfavourable as shown by the low efficiency observed in Figure 5(b). This observation may be attributed to repulsive forces between the adsorbate and adsorbent because they are both negatively charged at alkaline pH. The adsorbate, Al(III), exists as Al(OH)₄⁻ at high pH and the adsorbent Zr(IV) oxide species in alkaline conditions exist as Zr(OH)₅⁻, which is coupled with the negative fixed charges of the host resin. Thus, it may be concluded that the pH-dependence mechanism for adsorption kinetics of HZO-P was the one most followed. This result also informs that HZO-P may be amenable to efficient regeneration using highly alkaline solutions of pH > 11 where adsorption is unfavourable. The estimated pH_pzc (10.8) confirms that the HZO-P assumes a negative surface charge in highly alkaline pH. The negatively charged surface exerts repulsive forces to the adsorbed Al(III), and hence their desorption.

The effect of pH on the adsorption behaviour of the HTO-P hybrid adsorbent on Al(III) ions is shown in Figure 5(c). The HTO-P adsorbed the metal in both acidic and alkaline solutions. At pH 2–3 the adsorbent achieved >99.9% removal of the Al(III) from solution, yielding a final solution pH of 3.7–5.7. The final water product may be used for other uses, like agricultural and industrial activities, rather than for drinking purposes. In the acidic conditions, the Al ions are most likely to have been adsorbed through the ion-exchange mechanism on the positive surface of HTO-P, which has an estimated pH_pzc of 8. HTO-P also removed more than 97% Al(III) at alkaline pH (9.90) and the pH of the final solution was 8.87. In alkaline conditions Al(III) exists as a negative complex, Al(OH)₄⁻, and the surface charge of HTO-P at solution pH 9.90 is negative (above pH_pzc of 8). Al(III) must therefore have been adsorbed through ion-exchange between the negative surface and the negatively charged analyte. This observation is in consensus with Barakat (2005) who found that HTO removed Cu(II) ions completely at pH > 8. The high adsorption efficiency of HTO-P in both acidic and alkaline pH makes it an attractive adsorbent for the alleviation of metal pollution from waters. At pH > 11, Al(III) adsorption onto HTO-P was unfavourable, thus regeneration can be achieved using highly alkaline solutions. In conclusion, the metal loaded HFO-P, HZO-P and HTO-P nanocomposites can be regenerated using highly alkaline solutions (pH > 12) because metal adsorption is not favourable on the three nanoadsorbents at this pH range. Moreover, Figure S3 in the Supporting information shows that the HZO-P and HTO-P hybrids are stable in very acidic solution compared to HFO-P.

Table 1 illustrates the kinetics results for adsorption of Al(III) ions in both single and multi-element solutions. The models are illustrated in the Supporting information in Figure S4. The adsorption of Al(III) was achieved through pseudo-second-order kinetics as depicted by the coefficient of determination, which was closer to 1 (R² > 0.9) than the other two models – the pseudo-first-order and the intraparticle diffusion models. This finding was confirmed by comparing the calculated equilibrium capacity for the pseudo-first-order and the pseudo-second-order models with that obtained experimentally, which is shown in Table 2. The pseudo-second-order gave values are close to those obtained experimentally and hence it best represented the adsorption data. Table 1 shows that the initial Al(III) adsorption rates (V₀) were higher in the single element solution than in the multi-element solution. This observation may be because there was no competition for adsorption sites for the Al³⁺ in the single element solution, while in the ternary solution all three elements (Al, Mn and Fe) competed for adsorption sites on the surface of the adsorbent. The decreased rate of adsorption of Al ions in the multi-element system may be because the Al ions are smaller than the Fe and Mn ions and hence they take longer to reach the surface of the adsorbents for the adsorption process to take place.

The lower coefficient of determination obtained for the intraparticle diffusion model reported in Table 1 indicates that the model was not followed by the adsorption process. This result was confirmed by the fact that none of the linear plots of the systems passed through the origin. The poor fit of the kinetics data into the intraparticle diffusion model may be attributed to the smaller size of the Al atoms compared to those of the coexisting Fe and Mn in the solution. The pore diffusion model best fits when the analyte particles are big and are in elevated concentrations in a well-mixed solution (Kajjumba et al. 2018). The adsorption of Al(III) by the adsorbents was therefore not controlled by the intraparticle diffusion model.

Table 3 presents the results for the adsorption isotherm study of the nanocomposites. The models are provided in the Supporting information in Figure S5. The results indicate that the adsorption of Al(III) on HFO-P occurred through a single layer because it fitted well in the Langmuir model with a good coefficient of determination (R² = 0.9891) and the separation factor R_L = 0.050. The adsorption of Al(III) on HTO-P also fitted the Langmuir model (R² = 0.9504), with R_L = 0.005, better than the Freundlich and Temkin models. The R_L (0.050 and 0.005) between 0 and 1 indicates that Al(III) adsorption onto HFO-P and HTO-P was favourable, as mentioned earlier. Adsorption through the Langmuir model is usually due to formation of strong bonds between the adsorbent and the analyte. This observation suggests that Al(III) adsorbed onto HFO-P and HTO-P through chemisorption, while the adsorption of Al(III) onto HZO-P best fitted the Freundlich model (R² = 0.9001) with n = 1.902, implying favourable Al(III) adsorption through physical attraction onto the adsorbent surface (1 < n < 10). Moreover, best fit into the Freundlich model implies that there were some interactions between the adsorbed metal ions on the surface of the adsorbent. This observation is not surprising because the coexisting Fe and Mn oxides are good metal scavengers. It can be inferred that the adsorbed Fe and Mn became hydrated on the surface of the adsorbent and provided new adsorption sites for metal ions in the solution, resulting in multilayer adsorption. The Temkin model fitted poorly for all three adsorption systems, as implicated by the lowest coefficients of determination. This result is in agreement with literature attesting that adsorption from a liquid cannot be appropriately represented by the Temkin model; however, this model is excellent for representing gas phase equilibrium (Febrianto et al. 2009). The chi-square statistic (χ²) was in good agreement with the isotherm of best fit as per the coefficient of determination for HFO-P and HZO-P but not for HTO-P. The discrepancy observed with the HTO-P adsorbent may be attributed to the Ti oxide having complicated adsorption mechanisms, which is beyond the simple adsorption addressed in this study.

Normally, the adsorption efficiency does not provide enough information for an adsorbent to be used in a practical application and the adsorbent should be regenerated for re-use. Several binary solutions were trialled as regenerant solutions for the new adsorbents and the NaCl–NaOH solution exhibited the best efficiency. The regeneration of HFO-P, HZO-P and HTO-P was then conducted using NaCl–NaOH binary solution (pH 12.5) as the regenerant on a batch system. Figure 6 shows that 85.6%, 92.2% and 89.4% recovery of Al(III) from HFO-P, HZO-P and HTO-P, respectively, was achieved. This result is expected because Al(III) adsorption is not favourable at pH > 11 for all three nanocomposites, as indicated by Figure 5. The higher Al(III) recovery from the HZO-P compared to HFO-P and HTO-P may be because the Al was physically (weakly) bound onto HZO-P while chemically (strongly) bound to the other two, as has been reported in the isotherm study.

It is noteworthy that the coexisting Fe(II) and Mn(II) were not detected in the regenerant effluent. The interpretation of this observation may be in two-fold. First, this result may indicate that Fe(II) and Mn(II) were adsorbed on different sites through a different mechanism from that through which Al(III) was adsorbed onto the nanocomposites. For instance, Al(III) was adsorbed onto the cation exchange sites of the host resin through Coulombic attractions, while Fe(II) and Mn(II) were adsorbed onto the metal oxide nanoparticles surfaces through both electrostatic attractions and coordination. In this case, the NaCl–NaOH binary solution could only regenerate the host resin not the metal oxide adsorbents and hence the absence of the metal oxide sequestrated Fe(II) and Mn(II). On the other hand, the Fe(II) and Mn(II) may have been desorbed, and then immediately precipitated out of solution by the NaOH of the regenerant solution. This result may be expected because the pH of the regenerant was high enough (pH 12.5) to precipitate both the Fe(II) and the Mn(II) ions. Nevertheless, in this experiment, the absence of the d-block elements in the regenerant effluent may be attributed to their sequestration by the dispersed hydrated metal oxide (HMO), while Al(III) was sequestrated by the host resin. Evidence of the possible Al(III) ions sequestration by the host resin was confirmed when hydroxylamine hydrochloride (NH₂OH.HCl) solution (pH 3) was used as a regenerant for the loaded HFO-P resin. The hydroxylamine hydrochloride solution failed to desorb the Al(III) (desorption <1%) from the HFO-P; instead, it dissolved the embedded Fe(III) oxide concomitantly releasing the adsorbed Mn(II) into solution (Table 4). This inference is made from the fact that the regeneration effluent had a higher Fe level than was adsorbed and almost all the Mn adsorbed was desorbed.

The sequestration of the d-block elements by the HMOs was also confirmed by the results supplied in the Supporting information in Figure S2(a). No Fe(II) was adsorbed onto HFO-P; instead, the Fe bled into solution due to the instability of hydrated Fe oxides in low pH conditions. As a result, very little Mn(II) was adsorbed by the nanocomposite while 100% removal of the Fe(II) and Mn(II) was achieved for the HZO-P and HTO-P hybrid systems. The Zr and Ti oxides are chemically stable and are able to withstand harsh acidic conditions, and hence adsorbed all the Fe(II) and Mn(II) from the solution. In conclusion, the NaCl–NaOH solution only regenerates the ion-exchange sites of the host resin; therefore, a suitable regenerant solution for the HMOs should be sought.

Selected metals, with initial concentration provided in Table 5, were removed from environmental AMD using the three hybrid adsorbents and the host resin. All three nanoadsorbents adsorbed all the selected metals from AMD, as illustrated in Figure 7. AMD waters are characterized by a very low oxygen content and high H⁺, ultimately providing a reducing environment. The surfaces of the HLIX nanosorbents were protonated by the abundant H⁺ in the AMD and became positively charged. The dissolved metals in the AMD were then bound through ion-exchange with the surface H⁺ of the dispersed MONPs and with the mobile Na⁺ on the phosphonate head of the aminophosphonate parent resin, producing outer sphere complexes (Coulombic interactions). Binding of metal ions in the AMD through the formation of coordinate covalent bonds by a donated pair of electrons by an oxygen atom of both the HMO and the –PO₃²⁻, and also by the N atom of the amino group, producing inner sphere complexes, is another possible removal mechanism. This is the synergy that gives the synthesized HLIX nanosorbents the superior performance in metal removal from acidic/harsh complex industrial wastewater, AMD.

The adsorption efficiency was in the order HTO-P > HZO-P > HFO-P for all the metals with the exception of Co, Mn and Ni, where HFO-P adsorbed more than HZO-P. This observation suggests that there is a special affinity between Fe and the other magnetic metals, which needs to be investigated. The high removal efficiency of HTO-P may be attributed to the large hydroxyl group density (eight of them around each molecule) for titanium dioxide, which uses high specific adsorption sites for the metal ions. Titanium dioxide is also photoresponsive; that is, when light waves strike on titanium oxide electrons are produced that quickly migrate to the surface causing the adsorbent to have negative energy (Kuvarega et al. 2014). Through electrostatic attractions, these electrons pull metal cations from the solution, reduce them and ultimately deposit them as atoms on their surfaces and hence their adsorption, in addition to the common metal removal mechanisms by HMOs discussed earlier. The least removal efficiency for HFO-P may be because iron oxide is susceptible to destruction through protonation–dissolution in low pH conditions, leading to decreased adsorption sites for the metal ions. Iron oxides are also very unstable in anoxic acidic environments, which are conditions typical of AMD. The Fe(III) in HFO is reduced and produces Fe(II) ions and causes bleeding of the adsorbent, which ultimately reduces the active adsorption sites of the HFO-P. These phenomena are not experienced by the chemically stable Zr and Ti oxides.

Comparison of the metal removal efficiency of the novel hybrid adsorbents to their host resin is illustrated in Figure 8. Comparing the performance of the host resin with the synthesized hybrid resins revealed that HZO-P and HTO-P materials offer a higher removal efficiency for all the metals investigated in the complex environmental AMD, while with the HFO-P hybrid, the host resin removed Cr, Fe, Pb and Zn better. This result may be due to the shielding of the host active adsorption sites for the metals by the dispersed hydrated iron oxides. From this observation, it may be inferred that, in addition to providing mechanical support to the weak MONPs, the host material also complemented the metal oxides in the adsorption of metals and hence the enhanced adsorption efficiency of the hybrid resins. Indeed, the two components of the HLIX work in synergy in the metal ion removal from the AMD. HZO-P and HTO-P are therefore more promising adsorbents for AMD pollution remediation than HFO-P, which may be limited in extremely acidic pH water.

Oxides of Fe(III), Zr(IV) and Ti(IV) were successfully precipitated within macroporous polystyrene cross-linked with DVB weak acid chelating resin producing HFO-P, HZO-P and HTO-P nanocomposite adsorbents, respectively. In addition to providing mechanical support to the weak MONPs, the host material also complemented the metal oxides in the adsorption of metals and hence the enhanced adsorption efficiency of the nanocomposites compared to the host performance. Al³⁺ was sequestrated by the host resin through chemical ion-exchange, while the d-block metal ions, Fe²⁺ and Mn²⁺, were adsorbed onto the metal oxide surfaces by coordination. The components of the hybrid materials (host resin and the HMOs) should therefore be regenerated sequentially using different regenerant solutions. The adsorption/desorption selectivity is advantageous in that it would enable sequential recovery of the metals, which then can be separately precipitated as saleable products to cover the high cost of the chelating resin host. These hybrid adsorbents are very attractive because they can remove Mn(II) from low pH solutions and raise the pH of the solution being treated without addition of basic chemicals, which are expensive and prone to causing secondary pollution through problematic sludge. However, the increase in pH was smaller in real AMD due to the complexity of the sample. Thus, HFO-P, HZO-P and HTO-P hybrid adsorbents can provide a turnkey solution for AMD pollution remediation to the mining industry. The choice of adsorbent for application would be determined by the pH of the AMD polluted water to be treated.

Authors are grateful for laboratory working space from College of Agriculture and Environmental Science, use of ICP–OES from Analytical Chemistry Department, use of XRD and HRSEM from the Physics Department, and funding from the Nanotechnology and Water Sustainability Research Unit, all from University of South Africa Science Campus, Johannesburg, South Africa.

The Supplementary Material for this paper is available online at https://dx.doi.org/10.2166/wst.2020.285.