Understanding the adsorption of iron oxide nanomaterials in magnetite and bimetallic form for the removal of arsenic from water

Arsenic decontamination is a major worldwide concern as prolonged exposure to arsenic (>10 µg L-1) through drinking water causes serious health hazards in human beings. The selection of significant, cost-effective, and affordable processes for arsenic removal is the need of the hour. For the last decades, iron-oxide nanomaterials (either in the magnetite or bimetallic form) based adsorptive process gained attention owing to their high arsenic removal efficiency and high regenerative capacity as well as low yield of harmful by-products. In the current state-of-the-art, a comprehensive literature review was conducted focused on the applicability of iron-based nanomaterials for arsenic removal by considering three main factors: (a) compilation of arsenic removal efficiency, (b) identifying factors that are majorly affecting the process of arsenic adsorption and needs further investigation, and (c) regeneration capacity of adsorbents without affecting the removal process. The results revealed that magnetite and bimetallic nanomaterials are more effective for removing Arsenic (III) and Arsenic (V). Further, magnetite-based nanomaterials could be used up to five to six reuse cycles, whereas this value varied from three to six reuse cycles for bimetallic ones. However, most of the literature was based on laboratory findings using decided protocols and sophisticated instruments. It cannot be replicated under natural aquatic settings in the occurrence of organic contents, fluctuating pH and temperature, and interfering compounds. The primary rationale behind this study is to provide a comparative picture of arsenic removal through different iron-oxide nanomaterials (last twelve yearsof published literature) and insights into future research directions.


Introduction
The United Nations General assembly (2010) report states that access to safe drinking water and sanitation is a human right. Rapid urbanization and industrialization polluted the majority of the water resources. Major pollutants are organic compounds, dyes, heavy metals (Pb, As, Hg, Cu, Fe, Ni, Pd, etc.), agricultural and industrial wastes, recalcitrant compounds, and spills (Lin et al., 2022;Moreira et al., 2021;Zamora-Ledezma et al., 2021;Moosavi et al., 2020;Qi et al., 2023;Bhardwaj et al., 2022.). It was noted that prolonged consumption of arsenic-containing water could lead to severe health conditions in humans, including arsenicosis, cardiovascular disease, diabetes, and cancer. (Rahaman et al., 2021;Shokoohi et al., 2021).
Arsenic exposure during pregnancy and early childhood can impair cognitive development in young adults (Tyler et al., 2014). Recently published literature revealed that approximately 230 million people from 108 screened countries, majorly from India, Bangladesh, Nepal, Vietnam, China, United States, and Canada, are affected by the elevated concentration levels of arsenic in groundwater set by the World Health Organization (WHO, maximum permissible limit = 10 µg L-1) (Shaji et al., 2021). Geogenic arsenic contamination accounts for over 90% of all arsenic pollution (Shaji et al., 2021). Arsenite or As(III) is most lethal and highly mobile among arsenate or As(V) (60 times) and methylated arsenic compounds (70 times) (De et al., 2009). Moreover, the uncharged nature of As(III) species over a wide pH range makes it's comparably very challenging to get rid of it in water as compared to As(V) (Qi et al., 2015). Rigorous efforts have been made to eliminate arsenic from contaminated water using conventional and novel technologies worldwide. Out of these, adsorption gained attention due to its flexible design systems, cost-effectiveness, and simple operating steps (Jaspal et al., 2020;Liu et al., 2020;Supplementary Figure S1). Also, researchers and scientists have reported different metal-based adsorbents (Fe, Ti, Zr, Cu, and Ce) for arsenic removal (McDonald et al., 2015;Sakthivel et al., 2017;Wen et al., 2019;Nazari et al., 2020;Peng et al., 2022). Out of these, Iron-based nanomaterials (nanoscale zerovalent iron, maghemite, and magnetite based NPs) and their functional derivatives exhibit excellent affinity towards water dissolved arsenic compounds (Bangari et al., 2020); and, are biocompatible and induced less/no toxic effect in various cell lines, plants and vital organs of different animal models (Iannone et al., 2016;Nguyen et al., 2020). Furthermore, a bimetallic sequence of nanosized iron with copper (Fe-Cu), cerium (Fe-Ce), manganese (Fe-Mn), titanium (Fe-Ti), and zirconium (Fe-Zr) can effectively be lowering the arsenic contamination in polluted water (Zhang et al., 2013;Wen et al., 2019;Sha et al., 2020;Rao et al., 2015;Ren et al., 2011). This study aims to do a detailed and comprehensive literature search on the removal of As from the water environment and analyze the last 12 years' research articles (review and original work), book chapters, and conference proceedings. This state-ofthe-art focuses on iron-based nanomaterials' uses for water's arsenic removal using iron-based nanomaterials and nanoparticles and bimetallic systems. The objective is to (i) compile information regarding well-addressed iron-based nanomaterials for arsenic adsorption (ii) identify factors that are majorly affecting the process of arsenic adsorption and need further investigation (iii) discuss the separation of arsenic laden exhausted adsorbents and their regeneration capability, (iv) provide brief insights on toxicity and biocompatibility of iron-based nanomaterials and (v) summarize and identified key issues (future directions).

Research methodology
An in-depth study was conducted on several peer-reviewed journal papers (research articles and reviews) from several recognized databases (e.g., Sciencedirect, Hindawi publishing corporation, MDPI, Springer, Wiley, Taylor and Francis, ACS publications, and Royal Society of Chemistry) mostly from 2010 to 2021. Additional sources such as book chapters, conference proceedings, and technical journals were also referred to improve the quality of the presented study. The following search expressions were mainly used in search engine: "Iron supported magnetic nanoparticles" OR "Biocompatibility" OR "Iron supported arsenic adsorption" OR "Conjugation" OR "Adsorption" OR "Density functional theory" OR "Magnetite nanoparticles" OR "Desorption" OR "Nano iron (II) aluminum (III) oxide" OR "Biochar" OR "Magnetic composite" OR "Nanofibre filter" OR "Thermodynamics and kinetics parameters" OR "Film diffusion mechanism" OR "Exothermic" OR "Endothermic" OR "Adsorption isotherms models" OR "Langmuir isotherms" OR "Freundlich isotherm" or "Adsorption characteristics" OR "Bimetallic systems" OR "Thermal decomposition" OR "Co-precipitation" OR "Fe 3 O 4 nanoparticles embedded anodic aluminum oxide membrane "OR "Magnetic iron oxide-carbon nanocomposites" OR "Magnetite-silica nanocomposites" OR "Maghemite (γ-Fe 2 O 3 ) modified with Fe-Mn binary oxide" OR "Fe-Mn binary oxide incorporated magnetic graphene oxide" OR "Fe-Mn modified biochar composite" OR "Regeneration" OR "Nanosheet iron" OR "Regeneration of spent adsorbents". Mendeley software was used to compile and analyze the collected material as per the organization of the review and requirement. Firstly, selected journal papers were collected and categorized by exploring the above-stated keywords in the journal title, keywords, and abstract. Secondly, duplicate papers were eliminated. Lastly, papers of particular interest were gathered according to the various categories for further review, and document preparation ( Figure 1) shows the overall strategy adapted to preparing the current state-of-the-art).

Arsenic removal by adsorptive processes
Arsenic removal from water has been reported using a variety of adsorption processes and mechanisms (primarily based on intermolecular interactions). Such as surface complexation reactions or ion/ligand exchange, electrostatic interactions, Lewis acid-base interaction, filling up of size-specific pores present in the adsorbents, and oxidation of As(III) into As(V) before adsorption ( Figure 2) (Stumm, 1992;Singh et al., 2020;Wang et al., 2020;Wen et al., 2020). Mainly, arsenic removal from water involves both physisorption and chemisorption processes. In water, the surface of metal oxide adsorbents is covered by many hydroxyl groups due to the dissociative chemisorption of water molecules or the chemical composition of the adsorbent (Wang et al., 2020). On the other side, anionic arsenic species (solute ligand) also contain identical hydroxyl groups, which could replace the surface hydroxyl group present on the adsorbent surface through coordinative ligand exchange to form ligand surface complexes/inner-sphere complexes ( Figure 2A) (Stumm, 1992). Surface charges on the adsorbent change with the pH of the solution. At pH levels below point zero charge (pHpzc), adsorbent surfaces are positively charged due to a protonation reaction. Further, the adsorption of OH − above the pHpzc value turns the adsorbent surface negative (Bangari et al., 2019). In

FIGURE 1
Roadmap used for the presented study.

FIGURE 2
Pictorial representation shows the various mechanism involved in the adsorption process of arsenic using (A) magnetite NPs and (B) bimetallic iron oxide nanomaterials (where iron is coupled with metals oxides and or graphene oxides). Depending upon the iron nanomaterials, arsenic can be removed by R1 = adsorption, R2 = absorption, R3 = precipitation, R4 = dissolution, R5 = reduction, R6 = oxidation, or a combination of these. Finally, surface complexation was formed as an end step. As = arsenic, ad = adsorbent, ab = absorbent, aq = aqueous.

Adsorption based on iron oxide nanomaterials
Fe-based nanomaterials exhibit high surface area, chemical stability, adequate surface distribution of reactive sites, selectivity towards arsenic species, biocompatibility, and notable arsenic adsorption efficiency Nguyen et al., 2020;Nikić et al., 2021). Moreover, facile synthesis methods and efficient magnetic separation of Fe-based nanomaterials from treated water have increased the usability of this class of nanomaterials in arsenic removal processes (Chen et al., 2013a;Deng et al., 2019). Furthermore, as seen in Table 1 and Table 2, promising adsorption capacities of Fe-based nanomaterials after five to six adsorption-regeneration cycles lead to the cost-effective eliminationof arsenic from water (Jiang et al., 2012;Alotaibi et al., 2017).
Different synthesis approaches of Fe-based nanomaterials production include co-precipitation, proprietary copyrolysis, solvothermal, sol-gel, emulsion polymerization, hydrothermal, sonochemical, thermal decomposition and green synthesis methods (Yadav et al., 2020a;Otero-González et al., 2020;Singh et al., 2020;Torasso et al., 2021;Selvaraj et al., 2022). Green synthesis methods of Fe-based nanomaterials are environmentally benign and could be considered as a potential substitute of existing chemical and physical methods due to the affordability, abundance of biomaterials and simple operating procedures (Arslan et al., 2022;Xu et al., 2022). Biomaterials such as plant extracts (eg., green tea), vermicomposting leachate, residual food byproduct (e.g, banana peel) and microorganisms have been explored for synthesis of Fe based nanomaterials by several researcher while another study reported the utilization of incense sticks ash waste as precursor for iron oxide nanoparticle synthesis (Majumder et al., 2019;Yadav et al., 2020b;Chatterjee, et al., 2020;Arslan, et al., 2022;Chen et al., 2022).

Magnetite nanoparticles and their composites
Magnetite (Fe 3 O 4 ) is one of the most studied iron oxides for several environmental remediation processes, including water treatment, due to its high reactivity towards various organic and inorganic pollutants and inherent magnetic properties. The presence of unpaired electrons in Fe 2+ and Fe 3+ exerts magnetic sensitivity to the iron oxide nanomaterial-based adsorbents, facilitating their easy separation from water treatment systems (Kokate et al., 2013;Usman et al., 2018). Numerous modifications of iron oxide with different functional groups, alloying with other elements, and researchers and scientists in recent documents have reported incorporation in several organic and inorganic substrates (De et al., 2009;Sha et al., 2020;Singh et al., 2020). Bangari et al. (2019) fabricated magnetite-coated boron nitride nanosheets (BNNS-Fe 3 O 4 ) with a specific surface area of 119 m 2 g -1 and removed~98.83% of As(V) from arsenic solution via exothermic chemisorption process ( Figure 3). The rapid increment was observed

Adsorption characteristics Separation and Regeneration
Process parameter Reference

As(III) As(V)
Spherical Frontiers in Environmental Science frontiersin.org  Membrane; NaOH (1 molL -1 ); 3(RC); Δη=26 for As(III) and 12 for As(V) within an initial 120 min, and the equilibrium was reached after 10 h (Table 1). Further, in another study, Bangari et al. (2020) observed that four folds increased the adsorption capacity of Fe 3 O 4 -functionalized boron nitride nanosheets (BNNS-Fe 3 O 4 nanocomposite) compared with bare BNNSs for arsenic solution having initial concentration range between 6 and 70 ppm. The Langmuir isotherm model governed the adsorption process with the maximum adsorption capacity of 30.3 mg g -1 for As(III) at pH 8, while the adsorption kinetics followed the pseudo-second-order model. A decrease in ΔG˚value from −15.35 to −16.92 (kJ mol -1 ) at 298-318K suggested the favorability of the BNNS-Fe 3 O 4 nanocomposite-mediated arsenic adsorption process at a higher temperature. Similarly, 160.65 mgg -1 of As(III) and 104.13 mgg -1 of As(V) were removed within 30 min by 1-butyl-3-methylimidazolium hexafluorophosphate ([BMIM][PF6])-(ionic liquid)-modified magnetic graphene oxide (MGO-IL) . The adsorption process was endothermic and followed pseudo-second-order reaction kinetics. Adsorption of As(III) obeyed the Langmuir isotherm model, whereas favorability of the Freundlich model was observed in the case of As(V). The adsorption process was carried out in a wide pH range (2-12), where higher values of adsorption of As(V) at lower pH (< 4) were observed due to the electrostatic interaction of negatively charged As(V) ionic species and positively charged MGO-IL. In addition, electrostatic interaction was also reported between the imidazolium ring of the ionic liquid (IL) and As(V) anion. While adsorption of As(III) at higher pH (9) was attributed to the combined effect of electrostatic attraction, surface complexation, and formation of hydrogen bonds (Figure 2A). The removal efficiencies (%) of various studies conducted by multiple researchers are illustrated in Figure 3.
Furthermore, Tripathy et al. (2020) fabricated l-cysteine functionalized mesoporous Magnetite (Fe 3 O 4 @Cy) nanosphere using a surfactant-assisted solvothermal method with enhanced stability and adsorption efficiency. The adsorption process was reported to be spontaneous, endothermic, followed by pseudosecond-order reaction kinetics, and fitted well in the Langmuir isotherm model. The maximum adsorption capacities of Fe 3 O 4 @ Cy nanoparticles were recorded as 20.0 and 34.0 mgg -1 for As(III) and As(V) ions, respectively. Rowley et al. (2019) investigated the adsorption capacity of 3-aminopro-pyltriethoxysilane (APTES)-Fe 3 O 4 nanoparticles modified polyethersulfone (PES) nanocomposite membranes for As(V) removal from water. The adsorption capacity of the PES-A-Fe 3 O 4 ) for As(V) gradually increased with an increasing percentage of APTES-Fe 3 O 4 nanoparticles in PES membranes, and the maximum adsorption capacity (14.6 mg g -1 ) was observed at 3% APTES-Fe 3 O 4 . Adsorption equilibrium was observed around 270 min. The  . The adsorption process attained the equilibrium in 40 min at 25°C and fitted well with the Langmuir isotherm model. Moreover, the adsorption process was endothermic (ΔH• = 14.5 kJ mol -1 ) and followed the linear pseudo-first-order. The negative value of Gibbs free energy change (ΔG°= −6.8214 kJ mol -1 at 298K) indicated the feasibility and spontaneous nature of As (V) adsorption Fe 3 O 4 @AAO membrane, which further increased with an increase in temperature. Additionally, the process could predict a suitable binding tendency between As(V) and Fe 3 O 4 @AAO membrane based on a change in entropy (ΔS°= 71.653J mol -1 K −1 ). The same study reported the removal of 84.2%-94.5% arsenic from real water samples at pH 6 in 40 min (Figure 3). Detailed information regarding process parameters, kinetics, and adsorbents characteristics of the referred studies was tabulated in Table 1 and Table 2. However, most of the work was carried out under controlled laboratory conditions by using arsenic as a single source of contamination at fixed temperatures and pH. Data regarding multi-mixtures of contaminants in combination with arsenic are missing and need to be further explored under pilotscale and environmental conditions.

Bimetallic iron oxide nanomaterials
Bimetallic nanomaterials exhibit novel molecular architecture and possess synergistic chemical and physical characteristics of constituent metals (Scaria et al., 2020). Improved reactivity, thermal stability, unique texture, magnetic properties, and tunable surface properties of bimetal oxides have drawn much interest from various researchers to evaluate the efficiency of this class of materials in catalytic removal of heavy metals and various oxyanions from contaminated water . Moreover, bimetallic nanomaterials or nanoalloys are inexpensive, readily available, and environmentally benign, making them preferable for designing affordable water treatment strategies (Zhang et al., 2013;Liu et al., 2014;Siddiqui et al., 2019;Sha et al., 2020). The structure of bimetallic (A m B n ) NPs mainly depends on the nature of the constituent elements, synthesis pathways, A and B, their relative bond strengths, surface energies, relative atomic sizes, charge transfer, specific electronic/magnetic effects, and electronic perturbations in metal adatoms (Ferrando et al., 2008;Liu et al., 2014;Gilroy et al., 2016;Scaria et al., 2020).

Frontiers in Environmental Science
frontiersin.org Wen et al. (2019) synthesized nanosized magnetic mesoporous Fe-Ce bimetal oxides (Nanosized-MMIC) to study the adsorption behavior for As(V). They reported that the adsorption of As(V) onto the nanosized-MMIC was attributed to the effective electrostatic interaction and formation of inner-sphere complexes between the adsorbent-adsorbate interface ( Figure 2B). Nitrogen adsorptiondesorption isotherm suggested the generation of ordered mesoporous metal oxide with 143.61 m 2 g -1 surface area and 0.564 cm 3 g -1 pore volume. The batch mode adsorption followed pseudo-second-order reaction kinetics, with the maximum adsorption capacity for As(V) being 111.17 mgg -1 at room temperature at pH 3. Another study reported by Wen et al. (2020) performed better adsorption of As(III) and As(V) on the magnetic ordered mesoporous Fe/Ce bimetal oxides (OMICs) compared to individual adsorption capacities of constituent pure mesoporous iron oxides and mesoporous cerium oxides. The adsorption process followed pseudo-second-order kinetics with maximum adsorption capacities of As(III) and As(V) for optimized adsorbent (calcined at 350°C for 3 h) were reported as 281.34 and 216.72 mg g -1 , respectively (Table 2). Chen et al. (2013b) reported that the maximum adsorption capacities of Ce-Fe oxide decorated multiwalled carbon nanotubes (CF-CNTs) for As(V) and As(III) were 30.96 and 28.74 mg g -1 , respectively. Partial oxidation of As(III) to As(V) before the adsorption has been reported. Further, As(V) adsorption onto CF-CNTs was attributed to the combined effect of adsorbate-adsorbent electrostatic interaction, surface complexation, and formation of both monodentate and bidentate complex on the adsorbent (CF-CNTs) surface.
Fe-Mn binary oxide could effectively oxidize As(III) to arsenate As(V) and remove both As(V) and As(III) from water (Zhang et al., 2014). X-ray absorption near-edge structure results indicated the existence of Mn atoms in a mixed-valence state of +3 and +4. MnO x (1.5 < x < 2) content was inferred to be responsible for the oxidation of As(III) to As(V) via two-step reduction of Mn(IV) to Mn(III) and subsequent Mn(III) to Mn(II) while adsorption of As(V) was attributed to FeOOH content in Fe-Mn binary oxides. However, no significant contribution of pure FeOOH was reported in the oxidation of As (III) in the presence of air. The study further confirmed the formation of the inner-sphere bidentate binuclear corner-sharing complex with As-M (M = Fe or Mn) interatomic distance of 3.22-3.24 Å in adsorbate-adsorbent surfaces. Mag-Fe-Mn particles (0.1 gL -1 ) having maghemite core surrounded by amorphous Fe-Mn binary oxide coating lowered the concentration of As (III) from 200 µgL -1 to below 10 µgL -1 at pH 7 within 20 min with adsorption capacity 47.76 mgg -1 (Shan and Tong, 2013) (Figure 4). Complete oxidation of As (III) into As(V) was reported in the pH range 4-8. Heterogeneous adsorption of As (III) followed the pseudo-second-order kinetic model and fitted well in the Freundlich isotherm model and the two-site Langmuir model. Reduced graphene oxide (RGO) supported starch stabilized Fe-Mn binary oxide (Starch-FeMnOx) showed adsorption capacities of 78.74 mg g -1 and

FIGURE 4
Arsenic removal via bimetallic iron oxide nanomaterials.

Frontiers in Environmental Science
frontiersin.org 55.56 mg g -1 for As(III) and As(V), respectively (Lou et al., 2017). Starch-FeMnOx/RGO exhibited 1.99 and 2.39 times more adsorption capacities for As(III) and As(V) than bare FeMnOx. Another study (Cui et al., 2014) also reported complete oxidation of As(III) and lowering of 200 µg L -1 As(III) to below 10 µg L -1 in 75 min by magnetic porous Fe-Mn binary oxide nanowires (0.05 g L -1 ) within the pH range 3-9. The adsorption capacity of Fe-Mn binary oxide nanowires was highest (171 mg g -1 ) for the smallest Fe/Mn molar ratio 1:3 due to the enhanced peroxidation of As(III) in comparison to that in other Fe/Mn molar ratios (1:1 and 1:2). The Freundlich isotherm model and pseudo-second-order sorption kinetics are being used to explain the adsorption mechanism (Cui et al., 2014). Fe-Mn binary oxide encapsulated within a polystyrene anion exchanger D201 (D201-Fe/Mn) exhibited a maximum adsorption capacity of 44.91 mgg -1 and 13.17 mgg -1 for As(III) and As(V), respectively (Li et al., 2012). D201-Fe/Mn mediated arsenic adsorption was less sensitive to reaction pH over a wide pH range of 4-10 than Fe(III) oxide-loaded D201. Wen et al. (2015b)showed that pre-oxidation of As(III) followed by adsorption of As(V) by Fenton-like magnetic mesoporous iron cerium bimetal oxide (MMIC) removed arsenic from water. After 60 min, there was approximately total oxidation of 1000 ppb A(III), and after 180 min, there was the total elimination of arsenic. XPS data and analysis of reactive oxidizing species indicated that As(III) was oxidized by the surface-bound •OH radicals of MMIC assigned to Fe 2+ and Ce 3+ . Further, free •OH generated by iron ions released from MMIC into bulk solution also influenced the arsenic adsorption process (Wen et al., 2015b).
As per the literature discussed above, the bimetallic system works better in terms of arsenic removal as well as it takes less time to complete the reaction. However, bimetallic systems depend highly on reaction variables such as temperature, pH, absorbent, and adsorbate concentration.

Parameters influencing the adsorption performance
The reaction parameters crucially impact the efficiency of the arsenic adsorption process are initial arsenic concentration, adsorbent dose, contact time, temperature, pH, and competing anions in water. Herein, the effect of such factors on the removal of arsenic in iron oxide (magnetite) and bimetallic iron oxide nanomaterials mediated adsorption process will be discussed in light of recently published literature.

Initial arsenic concentration and adsorbent dosages
The adsorption efficiency for a particular adsorbate-adsorbent system was directly proportional to the adsorbent concentration, which may be due to the abundance of vacant binding sites, and the reaction reaches the adsorption equilibrium quickly (Bangari et al., 2019;Siddiqui and Chaudhry, 2019). Initially, arsenic adsorption increases rapidly and reaches the saturation plateau at equilibrium after the saturation of binding sites present on the adsorbent surface (Dutta et al., 2020). However, excessive adsorbent doses reduce the adsorption efficiency due to a large amount of unoccupied binding sites and less availability of surface area (Wang et al., 2020). Wang et al. (2016) examined the adsorption capacity of magnetic bio-sludge (MS) containing activated sludge and magnetite (Fe 3 O 4 ) nanoparticles in different initial arsenate concentrations (10, 50, 100, and 500 µg L -1 ). The adsorption process was performed using 0.7 gL -1 of adsorbent dose for 5 h at 25°C. Almost complete As removal was observed at 10 µgL -1 initial As(V) concentrations. Chaudhry et al. (2016) observed significant changes in the arsenic adsorption efficiency of binaryFe(III)-Sn(IV) mixed oxide-coated sand (ITOCS) at different initial As concentrations. They suggested decrement could be associated with the saturation of binding sites on ITOCS. Sahu et al. (2016) reported a significant increment in As(V) removal with the increase in initial arsenic concentration from 0 to 10 mgL -1 by Ce-Fe bimetal mixed oxide. However, this study reported no prominent increase in As(V) removal after 10 mgL -1 of initial arsenic concentration.

Contact time and temperature
Process efficiency is greatly influenced by adsorbate-adsorbent contact time as shorter adsorption equilibrium signifies higher removal of As in smaller time intervals. Similarly, the reaction temperature can affect the activation energy of the adsorbatadsorbent system, thus altering the adsorption performance. In endothermic adsorption, the reaction rate and adsorption capacity increase with the temperature rise, while exothermic processes are favored at low temperatures Wang et al., 2020). Liu et al. (2015a) reported spontaneous, endothermic adsorption of both As(III) and As(V) on crystalline magnetite nanoparticles (MNPs). Rapid adsorption was observed in the initial 5 min with an adsorption rate of 7.7 mmol g -1 h -1 and 6.7 mmol g -1 h -1 for As(V) and As (III), respectively, at pH 5. Furthermore, a slight increase in the maximum adsorption capacity for As(V) from 0.20 mmol g -1 to 0.25 mmol g -1 at 283 and 328 K was reported, while for As (III), the maximum adsorption capacity increased from 0.21 mmol g -1 to 0.23 mmol g -1 at 283K and 313 K respectively. Similarly, (Lin et al., 2017) observed a gradual increase in Fe-Mn modified biochar composite adsorption capacity from 8.09 to 8.45 mg g -1 for 288-308K. Also, another study conducted by Bangari et al. (2019) reported that adsorption of As onto Magnetite-coated boron nitride nanosheets (BNNS-Fe 3 O 4 ) was endothermic between 10°C and 30°C whereas the process becomes exothermic after 30°C (30°C-40°C). This suggests that the initial increase in temperature provided enough energy for chemisorption, resulting in BNNS-Fe 3 O 4 possessing a greater arsenic adsorption capability. However, a decrease in adsorption capacity was attributed to breaking adsorbate-adsorbent bonds with a further increase in temperature (beyond 30°C). A similar pattern of removal efficiency of Ce-Fe bimetal oxide for As(V) has been reported (Sahu et al., 2016).

Solution pH and coexisting anions
Iron oxide contains surface hydroxyl groups that get protonated or deprotonated with the pH of the solution (Deng et al., 2019).

Frontiers in Environmental Science
frontiersin.org Cheng et al. (2015) observed the point zero charge of nanosized magnetic iron oxide calcined at 250°C (FeMag-250) at about pH 7.5 and reported the maximum adsorption capacity (~50 mg g -1 ) of FeMag-250 for As(III) at pH 7. The study suggested that the abundance of unaltered hydroxyl groups present on the adsorbent surface at pH 7 facilitates the adsorption of H 3 AsO 3 , while the scarcity of unaltered hydroxyl groups with further protonation at lower pH might reduce the adsorption capacity (~32 mg g -1 ) of FeMag-250 for As(III) at pH 3. Further, a reduction in the adsorption capacity (~24.1) at higher pH could be attributed to the repulsion between arsenite anion and negatively charged adsorbent surface. Mag-Fe-Mn particles could remove 95% of As(III) within the pH range 4-8, while the removal efficiency of the adsorbent was reduced by 30% in the pH range 9-11 (Shan and Tong, 2013). Fe-Mn binary oxide nanowires removed 99% arsenic within the pH range three-nine and maintained residual arsenic below 10 µg L -1 (Cui et al., 2014). At a higher initial, As(III) concentration (5 mg L -1 ), arsenic removal was found to be pH-dependent, and maximum arsenic removal efficiency was noted as 36.1% in acidic conditions, which further decreased with an increase in the pH of the solution. Dutta et al. (2020) reported the removal of~94% As(III) and 98% As(V) from 1000 µg L -1 arsenic solution by polyaniline Microsphere/Fe 3 O 4 nanocomposite (1 g L-1, pHpzc = 7.7) within optimum pH range of 3-7. However, adsorption of arsenic was inhibited above pH 9 due to the apparent electrostatic repulsion between negatively charged arsenic species and the adsorbent in excessive OH − environment.
Insignificant effect of coexisting Cl − , NO − 3 and SO 2− 4 was reported for As(III) adsorption by magnetic porous Fe-Mn binary oxide nanowires during the coexistence of PO 3− 4 significantly decrease As(III) removal efficiency from 98.8% to 42.5% for 0-2 mM PO 3− 4 concentration (Cui et al., 2014). Humic acid (HA) lowered the arsenic removal efficiency of Fe-Mn binary oxide nanowires leading to increased residual arsenic in the solution. It was reported that the residual arsenic concentration was below 10 µg L -1 for HA concentration <2 mg TOC L -1 , while a gradual increase of residual arsenic was observed for HA concentrations more than 2 mg TOC L -1 . However, the authors suggested that the residual arsenic concentration could be maintained below 10 µg L -1 by increasing the adsorbent dose. Lou et al. (2017) observed mild inhibition in the adsorption efficiency of Starch-FeMnOx/RGO composite-mediated arsenic adsorption for the coexistence of SO 2− 4 (10 mM), while the coexistence of PO 3− 4 declined the adsorption efficiency to 18.4% and 2.5% for As(III) and As(V), respectively. Furthermore, the study reported a decrease in the removal efficiency of As(III) and As(V) to 83.5% and 45.0% in the presence of 10 mM HCO − 3 .

Separation and regeneration of spent adsorbents
Separation of the arsenic-laden adsorbents from water is inevitable for the practical applicability of the adsorbent (Yavuz et al., 2006;Zhao et al., 2018) as it is important to avoid the generation of secondary pollutantsoriginating from the exhausted adsorbents (Hao et al., 2018). Several separation methods, such as centrifugation, filtration, gravitational sedimentation, and magnetic separation, have been considered to isolate the spent adsorbents from water (Lata and Samadder, 2016). The selection of the suitable desorbing agent depends upon the adsorbate-adsorbent interaction mechanism and the nature of the adsorbents. It has already been stated that the adsorption capacity of iron oxide-based adsorbent strongly depends upon their point of zero charges (PZC) values and the pH of the treatment system. Thus, alkaline solutions are often used for the desorption of anionic pollutants, while acidic solutions remove cationic contaminants from adsorbent surfaces (Wang et al., 2020). Numerous studies ( Figure 5) have reported using NaOH, HCl, NaCl, and NaOCl to regenerate iron nanomaterialbased adsorbents (Rajesh Kumar et al., 2016;Yu et al., 2018;Nikić et al., 2021). However, NaOH was commonly used for arsenic desorption from iron nanomaterial-based adsorbents. Spherical polystyrene-supported nano-Fe 3 O 4 beads (PS-Fe 3 O 4 ) of 350-400 nm in diameter were successfully separated from water by a low-field separation method under a magnetic field of intensity <0.035 T in 15 min (Jiang et al., 2012). Desorption of arsenate from magnetically separated exhausted adsorbent was carried out by dispersing the spent adsorbent in 0.10 mol L -1 NaOH solution for 2 h, followed by centrifugation and cleaning with deionized water. The adsorption efficiency of PS-Fe 3 O 4 for As(V) was 93.8% after the sixth adsorptionregeneration cycle, while the adsorption capacities decreased from 10.6 to 9.5 mgg -1 . Powell et al. (2020) removed more than 94% of arsenic-laden iron oxide nanoparticles (As-Fe 3 O 4 ) using a magnetic nanoparticle recovery device (MagNERD) after arsenic treatment. Zhang et al. (2019) reported a promising saturation magnetization value of 27.79 emu g-1 of ionic liquidmodified magnetic graphene oxide (MGO-IL). The authors reported rapid magnetic isolation of MGO-IL from aqueous solution with a negligible decrease in adsorption capacity 8.06% Frontiers in Environmental Science frontiersin.org for As(III) and 14.14% for As(V), respectively, after the fifth adsorption-regeneration cycle. Kumar et al. (2014) separated a superparamagnetic GO-MnFe 2 O 4 hybrid (GONH) from an aqueous solution using an external magnetic field. Efficient recovery of 99% As(V) and 93% As(III) from spent GONH in 1M NaOH solution has been reported in the literature. Similarly, significant retention (95%) of the removal efficiency was reported in nanosized iron oxide (FeMag) for As(III) after five adsorption-regeneration cycles by treating with 1 molL -1 NaOH solution for 6 h (Cheng et al., 2015). Desorption of anionic arsenic species from the FeMag surface was attributed to the hydroxyl ion exchange and electrostatic repulsion on the adsorbent-adsorbate interface. Dutta et al. (2020) reported~83% retention of initial adsorption efficiency for As(III) in polyaniline Microsphere/ Fe 3 O 4 nanocomposite after the third regeneration cycle at 300 ± 3 K. Similarly, 86% and 90% removal efficiency for As(III) were reported for iron oxide-rice husk hybrid biochar nanocomposite and iron oxide-wheat husk hybrid biochar nanocomposite after the fourth regeneration cycle using NaOH (0.1 N) and NaCl (0.1 N) . Similarly, Sha et al. (2020) reported only a 5.1% loss of adsorption efficiency by magnetic graphene oxide (MRGO)-Fe-Mn binary oxide (FMBO) after five recycling cycles.
It was reported that higher desorption (97.58%) for As(V) could achieve in Ce-Fe bimetal oxide by increasing the concentration of NaOH (0.5M) solution (Sahu et al., 2016). A ternary solution composed of NaOH, NaCl, and NaClO could efficiently regenerate D201-Fe/Mn (Li et al., 2012). The authors reported that NaOH and NaCl could desorb arsenic from the exhausted adsorbent while NaClO oxidized Mn(II) to Mn(IV). Similarly, Mag-Fe-Mn particles obtained via magnetic separation were regenerated using a ternary solution consisting of NaOH, NaCl, and NaClO (Shan and Tong, 2013). Mag-Fe-Mn particles exhibited 87% desorption efficiency and 98% of adsorption efficiency for As(III) throughout five adsorption-regeneration cycles.

Insight on toxicity and biocompatibility
Possible migration of nano solids (pristine and pollutantladen) during water decontamination processes could impose potential risk upon its end users and various organisms in natural systems (Morillo et al., 2016;Simeonidis et al., 2016). Hence, a closer assessment of the toxicity of nanomaterials is a dire necessity before the mass application of nanomaterialmediated water purification systems. Accumulated research suggested that iron oxide nanoparticle-mediated toxicity greatly depends on the pathways of the entry of nanoparticles into organisms (Frtús et al., 2020;Chrishtop et al., 2021). The toxicity of iron oxide nanoparticles could be assigned to the production of reactive oxygen species and subsequent oxidative stress in nanoparticle-mediated metabolic processes (Wu et al., 2014;Frtús et al., 2020). Oral

FIGURE 5
Reuse cycles of adsorbents.

Frontiers in Environmental Science
frontiersin.org administration of iron oxide nanoparticles significantly increases the nanoparticle load on the liver and kidneys, while iron oxide nanoparticles entered through the sensory route could cross the blood-brain barrier in animals and cause neurotoxicity (Chrishtop et al., 2020). Iron oxide nanoparticles at high concentrations could impose severerisk due to higher accumulation in various organs and difficulties in their efficient excretion from the organism's body (Mou et al., 2015;Chrishtop et al., 2020).  Ledda et al. (2020) designed sub-5 nm silica-coated superparamagnetic iron oxide (sub-5 SIO-Fl) nanoparticles for oncology-related theranostics applications. This study observed the quick absorption of nanoparticles within 24 h of treatment; however, a gradual decrement in uptake was observed after 48 h. Additionally, no sign of kidney, spleen, stomach, or liver necrosis was reported until 7 weeks afterthe nanoparticles injection. Nguyen et al. (2020) observed a significant decrement in copper toxicity by using Fe-Cu bimetallic oxide nanoparticles (500 μg mL -1 ) for the treatment of rat kidney proximal tubule epithelial cells (NRK-52E). About 94% survival rate was observed in Fe-Cu bimetallic oxide nanoparticles treated NRK-52E cell line while the control (CuO) shows great decrement in survivability (8%). Whereas approximately 100% survival rate was observed for Fe 3 O 4 nanoparticles on the same cell lines using similar concentrations. Fe-Cu nanocomposite treatment showed faster healing of diabetic wounds in methicillin-resistant S. aureus (MRSA) infected Wistar albino rats than iodine-treated and untreated groups over 21 days (Das et al., 2019). Further, the same literature reported 0.29% and 0.32% hemolysis in nanocomposite-treated human erythrocytes for nanocomposite powder and nanocomposite-impregnated cotton, respectively, is relatively lower than the permissible limit (5%) for blood containing biomaterial.
The accumulated data suggested the biocompatibility and low toxicity of iron oxide nanoparticles in target organisms. However, more data is required to evaluate the toxicity and biocompatibility of bimetallic iron oxide nanoparticles. It has been observed that iron oxide nanoparticle-mediated toxicity could be assigned to its physical and chemical characteristics. Indeed, iron oxide nanoparticle-induced toxicity is target specific and closely related to the properties of nanoparticle coating materials. Possible causes of iron oxide toxicity could be summarized as nanoparticle-induced oxidative stress, ROS production, autophagy, apoptosis, cell migration, inhibitory effect on osteoclastogenesis, inflammation and cell death, transient phenotypic changes of macrophages, cell cycle arrest, activation of dendritic cells in different cell models (Frtús et al., 2020). Research on the molecular mechanisms of nanoparticle-cell interaction is necessary to evaluate nanoparticle-induced toxicity and its biocompatibility. Impact assessment of external stimuli, such as light water chemistry on the surface-bound ligands of nanoparticles, could be helpful in understanding the dynamics of nanoparticle release in nanomaterial-mediated water decontamination systems (Frtús et al., 2020;Nagar and Pradeep, 2020).

Conclusion
Arsenic adsorption was found to be predominated by surface complexation (electrostatic interaction, chemisorption, hydroxyl radical influence adsorption process) and ligand exchange onto the nanomaterials surface, while oxidation of As(III) to As(V) before adsorption has been reported in case of some manganese-containing bimetallic iron oxide nanomaterials. Bimetallic systems are more effective and require less time for reaction completion. The Fe-Mn bimetallic oxide nanomaterials showed promising trends in arsenic removal due to the synergistic oxidation and adsorption of highly toxic As(III) from water. The aqueous medium's pH significantly impacts the As adsorption process, which can be effectively described by the Langmuir and Freundlich isotherm models. The remarkable magnetic properties of iron-based nanomaterials facilitate their easy magnetic separation from treated water which makes magnetic nanomaterials advantageous over non-magnetic adsorbents. Results revealed that magnetite and bimetallicbased adsorption could be safely used up to four to five and three to four regeneration cycles. However, the presented document revealed that majorly studies were carried out under controlled laboratory conditions, but research is needed to verify them under natural environmental conditions. Furthermore, this review article will help scientists and researchers decide which combinations of iron nanomaterials were used for arsenic removal from water.

Future research directions and key challenges
Despite notable achievements in arsenic removal by nanomaterials, specific challenges still need to be addressed before the successful implementation of nanomaterial-mediated Frontiers in Environmental Science frontiersin.org arsenic removal processes on a large scale. The design and implementation of magnetite and bimetallic iron oxide nanomaterials with suitable adsorption characteristics and the applications in As elimination from water should be addressed coherently to optimize. Despite the fact that the applications have been extensively described in the literature, several study scopes (open research topics) remain to be explored.
1. In most of the investigations, the adsorbent was constructed by incorporating nanophase into a suitable substrate or host to enhance the stability of adsorbents and reduce the migration of nanomaterials into product water. However, the biocompatibility and toxicity of nanoparticle coating materials and hosts/substrates need to be evaluated to eliminate the secondary risk associated with magnetite and bimetallic iron oxide nanomaterials. 2. There is a need for further improvements in the adsorption capability of magnetite and bimetallic iron oxide nanomaterial, whether from adsorption rate or capacity. This can be achieved by optimizing particle/pore shapes and sizes, imparting suitable functional groups onto the surface depending on the As properties, or by modulating the synthesis conditions to maximize active adsorption sites. Most of the studies were carried out under controlled laboratory conditions, and minimal studies have been reported using real water matrices, thus making the practical application of adsorbents difficult. Hence, more research is needed to demystify the uncertainties in the effectivity and hydrodynamic stability of magnetite and bimetallic iron oxide nanomaterials in complex real water matrices under regulated conditions (such as low initial concentration of As, neutral pH, atmospheric temperature, and the presence of other pollutants); and ranked them based on their importance. 3. Regeneration and reusability of spent adsorbents (of magnetite and bimetallic iron oxide nanomaterial) determine the cost efficiency of water decontamination processes. Although most of the adsorbents cited here showed promising adsorption capacity for arsenic removal after four to five adsorptionregeneration cycles ( Figure 5), improvements are still required to enhance the affordability and sustainability of the adsorbents. 4. Leaching tests are also needed to successfully implement ironbased adsorption processes under natural environmental conditions. 5. Life cycle assessment of nanomaterials and long-term monitoring of arsenic-laden adsorbents could be helpful to evaluate the environmental sustainability of nanomaterials enabled As adsorption processes. 6. There is a need to understand the compatibility of iron-based nanomaterials as per the characteristics of water where arsenic is present.