Preparation of a novel Fe3O4/HCO composite adsorbent and the mechanism for the removal of antimony (III) from aqueous solution

A novel adsorbent (Fe3O4/HCO) was prepared via co-precipitation from a mix of ferriferrous oxide and a Ce-rich waste industrial sludge recovered from an optical polishing activity. The effect of system parameters including reaction time, pH, dose, temperature as well as initial concentration on the adsorption of Sb(III) were investigated by sequential batch tests. The Sb(III)/Fe3O4/HCO system quickly reached adsorption equilibrium within 2 h, was effective over a wide pH (3–7) and demonstrated excellent removal at a 60 mg/L Sb(III) concentration. Three isothermal adsorption models were assessed to describe the equilibrium data for Sb(III) with Fe3O4/HCO. Compared to the Freundlich and dubinin-radushkevich, the Langmuir isotherm model showed the best fit, with a maximum adsorption capacity of 22.853 mg/g, which exceeds many comparable absorbents. Four kinetic models, Pseudo-first-order, Pseudo-second-order, Elovich and Intra-particle, were used to fit the adsorption process. The analysis showed that the mechanism was pseudo-second-order and chemical adsorption played a dominant role in the adsorption of Sb(III) by Fe3O4/HCO (correlation coefficient R2 = 0.993). Thermodynamic calculations suggest that adsorption of Sb(III) ions was endothermic, spontaneous and a thermodynamically feasible process. The mechanism of the adsorption of Sb(III) on Fe3O4/HCO could be described by the synergistic adsorption of Sb (III) on Fe3O4, FeCe2O4 and hydrous ceric oxide. The Fe3O4/HCO sorbent appears to be an efficient and environment-friendly material for the removal of Sb(III) from wastewater.

as low adsorption capacity and difficulty in separating materials from the aqueous solution when they are saturated 2,5,6 . As frequently used as adsorbents, iron-based materials combine adsorption properties with useful magnetic properties. This makes the separation of solids from liquid phases straightforward using an external magnetic field and allows for recovery and reuse after regeneration 17,18 . The advantage of magnetic separation of adsorbents is a more effective approach to solid-liquid separation than normal separation methods such as sedimentation and filtration 19,20 . Also, magnetic separation is particularly useful when the aqueous solution contains nonmagnetic solid residues as would be found in complex waste water systems 19,21,22 .
The Fe 3 O 4 adsorbent system has previously been systematically studied as an excellent magnetic material 23 . However, it has a significant limitation in its lower sorption capacity restricting effective application 17 . Synthesis of iron-containing bimetal oxides can greatly enhance adsorbent performance by increasing the amount of surface pores, hydroxyl groups, and tunable surface charge 2,17,18 . Also, earlier studies have found that a Cerium-doped Fe 3 O 4 magnetic adsorbent tends to have high affinity surface hydroxyl groups and very promising adsorption capacity 17,[24][25][26] . And is considered to be an excellent adsorbent for antimony removal 17,27 . However, cerium is a rare metal widely used in modern devices and is expensive, limiting its application in adsorption systems. The polishing sludge identified, is particularly rich in condensed Cerium (Ce), mostly as hydrous ceric oxide (HCO: CeO 2 ·nH 2 O, also known as cerium hydroxide), and is a common component in chemical mechanical polishing (CMP) process wastewaters treatment by product, appearing also as a residue from the liquid crystal display (LCD) industry. China produces more than 5.0 × 10 4 tons a year according to incomplete statistics. The preparation of the sorbent using Fe 3 O 4 and the Ce-rich waste, whilst adding additional synthesis steps, uses lower amount of Ce than Cerium-doped Fe 3 O 4 magnetic adsorbent reported in previous research 17, 28 . In addition the sludge is stable and can be produced in a range of particle sizes 29 . The preparation of a successful sorbent based on Fe 3 O 4 /HCO also contributes to the aims of the "2030 Agenda for Sustainable Development", by reducing the generation of waste residues through recycling and the efficient use of secondary resources 2 . Additionally, the preparation and optimization of Fe 3 O 4 /polishing sludge adsorbent seem to be a key-role to control the removal efficiency of antimony by adsorption. To the best of our knowledge, the mechanism for antimony adsorption by Cerium-doped Fe 3 O 4 magnetic adsorbent is unclear 17,27 .
In this work, we focus on: (1) synthesis of Fe 3 O 4 /HCO by a modified coprecipitation method, and subsequent detailed characterization; (2) sorption of aqueous Sb(III) by Fe 3 O 4 /HCO and evaluation of capacity and the effects of pH, reaction time, the amount of adsorbent, reaction temperature and initial concentration; (3) tests of models of isothermal adsorption, reaction kinetics and thermodynamics leading to proposals for the mechanism for the adsorption of Sb(III) on Fe 3 O 4 /HCO. We believe this is the first report of the successful preparation of iron-based adsorbents with polishing sludge and its initial application addressing a pressing environmental issue of Sb(III) contamination. experimental Synthesis and characterization of fe 3 o 4 /Hco adsorbent. Polishing sludge material was collected from the wastewater treatment plant of Lansi Technology (Hunan) Co., Ltd. The air-dried bulk analysis (w/w) was: moisture content of sludge is 80.5%, and the other main components are cerium oxide (7.8%), silicon dioxide (4.5%), aluminium oxide (3.8%), calcium oxide (2.8%) and other material (0.6%).
The Fe 3 O 4 /HCO adsorbent was prepared using the following modified coprecipitation method 17,30 . Firstly, a 1000-ml three-necked flask was purged with nitrogen for 10 min and 10 g air dried polishing sludge was added followed by 50 ml of an aqueous solution containing 5.56 g FeSO 4 ·7H 2 O and 50 ml of an aqueous solution with 10.8 g FeCl 3 . Secondly, the flask was placed in a water bath at 60 °C and 200 ml 7% aqueous ammonia solution slowly added whilst being agitated at 350 rpm. Thirdly, after continuous stirring for 2 h under the nitrogen atmosphere, the resulting slurry was separated by rapid centrifugation, decanted and washed with deionized water and ethanol followed by drying at 80 °C for 24 h. Finally, the dry mixture was ground into fine powder with a mortar and pestle to pass a 100-mesh sieve, and then used for Sb(III) removal studies.
Particle morphology and crystallinity of Fe 3 O 4 /HCO were characterized using scanning electron microscope (SEM, JSM-6380LV, JEOL Ltd.) and X-ray diffraction (XRD) patterns (D8 Advance, Brook AXS Ltd., Germany). Elements on the surface of Fe 3 O 4 /HCO were analyzed using Energy Dispersive Spectrometer (EDS) (Bruker XFlash 5010, Germany). The XRD was used to identify compounds present in the solid sorbent before and after adsorption of Sb(III). X-ray photo-electron spectroscopy (XPS) spectra focused on Ce, Fe and Sb sorbed onto the Fe 3 O 4 /HCO using a PHI 5000 Versa probe system (Thermo Scientific: Esala 250Xi). All the binding energies were associated with the C 1s peak at 285.1 eV and XPS peak fit version 4.1 was used to analyze the spectral data. N 2 adsorption-desorption isotherms were used to test the surface area and the pore structures of Fe 3 O 4 /HCO. The specific surface area, pore volume as well as pore diameter of Fe 3 O 4 /HCO were measured by N 2 adsorption at 77 K using a QuadraSorb Station 1 Instruments (Anton Paar GmbH).
Adsorption experiments. The influence of experimental variables on the adsorption isotherm and kinetics were assessed using a sequential batch test. Aliquots of concentrated Sb(III) solution and diluted with deionized water were added to a 500-ml Erlenmeyer flask to a total volume of 200 ml and 0.80 g Fe 3 O 4 /HCO was then added to the mixture and pH was adjusted using either 0.1 mol/L HCl or 0.1 mol/L NaOH solution. Adsorption was conducted at 150 rpm at temperature of 25 °C. After the reaction reached equilibrium, samples were filtered (0.45μm filter), and the concentration of Sb(III) determined in solution using hydride generation atomic fluorescence spectrometry (see below). All experiments were completed in triplicate and the adsorption tests were performed in the dark 31 In which q (mg/g) is the adsorption capacity; c i (mg/L) and c e (mg/L) are the ion concentrations of the solution before and after the reaction, respectively; V (mL) is the volume of the solution; M (g) is the adsorbent mass used during the reaction process. In order to analyse the adsorption mechanism, four classic kinetic models, named as the Pseudo-first-order model, Pseudo-second-order model, Elovich model and Intra-particle diffusion model 30,32,33 (Eqs (S6-S9)), were used to test fit to the experimental data.
Reagents and analytical methods. A quantity of antimony potassium tartrate was weighed and dissolved in deionized water to prepare antimony standard bulk solution of 1.0 g/L Sb (III). Experimental Sb solutions were obtained by appropriate dilution. The reagents used in the experiment are analytical or superior grade reagents, and the experimental water was deionized water. Hydride generation atomic fluorescence spectrometry was utilized to determine the concentrations of antimony (III), following the method of Leuz A et al. 34 . The detection limit of this method was 1 μg/L. All samples were measured within 24 h after the adsorption experiment, and deionized water was used for the blank. The Sb(III) recovery using this test was over 90.0% and analytical error on Sb determination was <1%.  17 and Zhang et al. 26 . Further work is needed to evaluate the potential to synthese Fe 3 O 4 /HCO with higher Ce/(Fe + Ce) molar ratio.

Results and Discussion
The detailed XRD characterization of the sorbent is given in the supplementary data (Fig. S1). Ce and Fe and mixed oxide phases are identified. The peaks were observed at 2θ = 26.62°, 28.06°, 33.11°, 47.49°, and 56.40°, represent reflections from the (220), (311), (400), (511), and (440) planes, is different from other studies 17   www.nature.com/scientificreports www.nature.com/scientificreports/ adsorbent. The synthesis of FeCe 2 O 4 might play a major role in the adsorption of Sb(III) and its identification is an important innovation in this study.
The specific surface area (S BET ) and pore volume of the Fe 3 O 4 /HCO was found to be 83.496 m 2 /g and 0.098 cm 3 /g, respectively. It was three times greater than the original Fe 3 O 4 (S BET 28.0 m 2 /g) and was much larger than hydrated ferric oxides supported by polymeric anion exchange 35 , indicating that the new sorbent has a significantly higher accessible surface area for adsorption of Sb(III).
Effect of pH, react time, amount of adsorbent, temperature and initial concentration. The pH and point of zero charge (pH pzc ) are the most important parameters affecting Sb removal efficiency in adsorption technology 6,29 . As shown in Fig. 2, the removal of Sb(III) by Fe 3 O 4 /HCO varies with pH. The rate of Sb(III) removal initially increased in our experiment before decreasing as pH increased from 2 to 9, but only slightly changed (90.00-91.98%) over the pH 3 to 7 range. This was identified as optimal pH for Fe 3 O 4 /HCO to adsorb Sb(III). The measured zeta potentials of the Fe 3 O 4 /HCO suspensions was about 6.8, which was consistent with the literature results 17 . When the pH is 2, Sb(III) exists in the form of Sb(OH) 2 + 1 , competing with H + and Sb(OH) 2 + 29 which reduced the removal efficiency Sb(III). Across a pH range from 2 to 9, Sb(III) is predominantly in the form of [H 3 SbO 3 ] or Sb(OH) 3 1 which can result in the precipitation of Fe-Sb(III) 36 and CeSbO 3 37 with Fe 3 O 4 /HCO to enable a higher rate of Sb(III) removal. It was obvious that pH is close to the pH pzc and makes the adsorbent surface uncharged and attracts the neutral Sb(OH) 3 . When pH > 9, oxidation of Sb(III) is enhanced 34 , and the increase in pH can inhibit the production of iron oxyhydroxide and the solubility of iron ions 18 , resulting in a decrease in the removal efficiency for Sb(III). These findings are consistent with those of Fan et al. 29 . Compared with other iron-based adsorbents 38,39 (shown in Table 1), the adsorption of Sb(III) onto Fe 3 O 4 /HCO can occur over a wide pH, which introduces versatility and enhances potential application.
The effect of reaction time on Sb(III) removal from water (Fig. 3a) suggests a two steps process: a fast stage from 0 to 2 h and a slow stage after 2 h. The reason for this is likely to be that initially surface adsorption takes place between 0-2 h. As the adsorption continues the binding sites on the adsorbent surface are saturated, therefore subsequent adsorption is by internal diffusion stage 40 which reduced the rate of adsorption. The adsorption equilibrium time (2 h) of Fe 3 O 4 /HCO for removal Sb(III) is much less than that of many iron-loaded adsorbents, such as hematite coated magnetic nanoparticles 41 , iron (III) loaded orange peel residue 42 , and quartz sand loaded iron oxide 21 . The reaction equilibrium time of the subsequent experiments in this study was set to 2 h 43 .
Results for the effect of Fe 3 O 4 /HCO loading on Sb(III) removal ( Fig. 3b) showed removal rates to increase from 89.68% to 92.75% with Fe 3 O 4 /HCO dosing from 2 g/L to 4 g/L. As the dose of Fe 3 O 4 /HCO rose to 12 g/L, the removal of Sb(III) slowly increased to 94.64%. This result indicates that Fe 3 O 4 /HCO loading has only a slight impact on the removal of Sb(III) when the addition of Fe 3 O 4 /HCO exceeds 4.0 g/L, a finding consistent with previous results of Sun,et al. 32,44 . As the adsorbent dosage increases, its surface can provide more functional groups and adsorption sites to enhance removal rate of Sb(III) 18 , while Sb(III) removal would not be improved when the adsorption equilibrium was reached. This indicates that an optimal dosage can be determined when other conditions remain stable 18 . Therefore, considering the cost of the adsorbent and the removal efficiency of Sb(III), the optimum loading of Fe 3 O 4 /HCO in our study was 4.0 g/L, this being less than that of iron (III) loaded orange peel residue (5 g/L) 42 , iron (III) and zirconium (IV) loaded orange peel residue (5 g/L) 42 and composite material of biomorphic Fe 2 O 3 /Fe 3 O 4 /C with eucalyptus(10 g/L) 45 . This has positive practical application.
The effect of temperature shows a modest increase in sorption with increasing temperature. As shown in Fig. 3c, the removal of Sb(III) by Fe 3 O 4 /HCO adsorbent was 92.18%, 93.41% and 94.32% at 20 °C, 25 °C and 30 °C, respectively. The finding is consistent with studies of Sb(III) removal using ferric salts or ferric salt modified adsorbents 18,32 . Due to an endothermic reaction occurring when metal ions were adsorbed by the iron matrix, adsorption and removal efficiency, adsorption capacity and adsorption rate of most iron-based inorganic adsorbents will improve with an increase in temperature 35 . By taking operating costs into account, the optimal reaction temperature for this study was 25 °C.
As shown in Fig. 3d, an initial increase in concentration of Sb(III) from 10 mg/L to 60 mg/L resulted in a decrease in the removal of Sb(III) from 95.08% to 91.39%. When concentration increased to 200 mg/L, a significant decreased of Sb(III) removal to 51.10% was observed. This identifies good operating conditions at Sb(III) concentrations less than 60 mg/L. In addition, when the initial concentration of Sb(III) ranged from 150 mg/L to A comparison of the adsorption capacity of different iron-loaded composites for Sb(III) from water is shown in Table 1. The adsorption capacity of Fe 3 O 4 /HCO synthesized in this study is lower than the values of some adsorbents, such as Ce-doped (0.5) 17 , hematite coated magnetic nanoparticles 41 , iron (III) loaded orange peel residue 42 , and iron (III) and zirconium (IV) loaded orange peel residue 42 . However, it is significantly higher than a   www.nature.com/scientificreports www.nature.com/scientificreports/ number of other adsorbents, for instance, ferric chloride modified sand 21 and iron oxide loaded carbon nanotubes 46 . Also, the equilibrium time for Sb(III) adsorption Fe 3 O 4 /HCO for was only 2 h, considerably shorter than many other adsorbents.

Adsorption isotherms. Model definition and accuracy of fit for the isothermal adsorption model is related
to the type of adsorbent, the valence state of the Sb ion, the initial concentration, pH, and a number of other factors 18,29,32 . The results for the fit to Langmuir, Freundlich and D-R model are shown in the Fig. 4. The parameters of the three adsorption isotherm models are listed in the Table 2.
As shown in Fig. 4a, the linear relationship between C e /q e and C e at 20 °C, 25 °C and 30 °C indicates that the Langmuir model has strong fit at each temperature (R 2 > 0.99). The maximum adsorption capacity and b value for Sb(III) at 25 °C were 23.171 mg/g and 0.209 L/mg, respectively.
As shown in Fig. 4b, a linear relationship between lg(C e ) and lg(q e ) for the Freundlich model at 20 °C, 25 °C and 30 °C was a slightly poorer fit (R 2 > 0.82). The values of K f as well as 1⁄n are related to the adsorbent, adsorption mechanism and reaction temperature, which can be calculated by the relationship between lg(C e ) and lg(q e ). The isothermal adsorption form can be determined according to the value of 1⁄n 31 . At 25 °C the K f and 1⁄n was 5.108 and 0.387, respectively. As the value of 1⁄n was less than 0.5, which indicated that Sb(III) was easily adsorbed by Fe 3 O 4 /HCO 32 . This is also illustrated that Fe 3 O 4 /HCO is an excellent adsorbent for adsorbing antimony.  www.nature.com/scientificreports www.nature.com/scientificreports/ The linear relationship between q e and ε 2 of the D-R model at 20 °C, 25 °C and 30 °C are shown in Fig. 4c. The value of q s and β values at 25 °C was 80.743 mg/g and 3.335E −9 mol 2 /KJ 2 , respectively. In addition the average adsorption energy E (kJ/mol), which could be determined from the D-R model, is the free energy change as one mole of ions transfers from the solution to the sorbent surface 47 . Using Eq. (S4), E ranged from 113.283 to 138.145 kJ/mol at 20-30 °C. According to the scale of the force and the E value between the adsorbed substance and the adsorbent, the adsorption process can be classified as physical adsorption (1 kJ/mol ≤ E ≤ 8 kJ/mol), ion exchange (9 kJ/mol ≤ E ≤ 16 kJ/mol) as well as chemical adsorption (E > 16 kJ/mol) 48 . Therefore, the adsorption of Sb(III) to Fe 3 O 4 /HCO was a chemisorption process 29 , which is also consistent with the conclusions by Deng et al. for adsorption of Sb(III) using Fe(III)-modified humus sludge 32 .
In summary, the Langmuir model has the best fit (R 2 > 0.99) for the removal of Sb(III) by Fe 3 O 4 /HCO. This is consistent with the adsorption of Sb(III) by iron-based matrices 18,29,30,32 . We conclude that sorption reactions take place on the surface of iron-based adsorbents, as monolayer adsorption 29 . The Langmuir model obtained the maximum adsorption capacity of Fe 3 O 4 /HCO for removing Sb(III) (up to 23.171 mg/g at 25 °C), higher than many iron-based adsorptive substrates 21,46 . All of the 1/n values was determined using the Freundlich model were less than 0.5, indicating that Sb(III) in an aqueous solution is readily adsorbed by Fe 3 O 4 /HCO 32 . Results for the D-R model further indicate that Sb(III) adsorption by Fe 3 O 4 /HCO is a chemisorption process 41 .
Adsorption kinetics. The adsorption kinetics model can describe the potential rate of control and adsorption mechanism of material transfer and chemical reactions during the adsorption process. Adsorption kinetics are dominated by the physical as well as chemical properties of the adsorbent in the adsorption process of the adsorbent 11 . In this study, Pseudo-first-order, Pseudo-second-order, Elovich and Intra-particle diffusion models are used to analyze the kinetic characteristics of Fe 3 O 4 /HCO adsorption to remove Sb(III) 30,32,33 . As shown in Fig. 5a and Table 3, although a high correlation coefficient (R 2 = 0.971) was recorded for the Pseudo-first-order model curve, the fitting of the curve tail was poor, indicating that the Pseudo-first-order model is not an appropriate simulation for Sb(III) adsorption onto Fe 3 O 4 /HCO. However, the Pseudo-second-order model showed the strongest fit with the experimental data (R 2 = 0.993) in all adsorption kinetics. In addition, there is little difference between the theoretical value of q e (23.145 mg/g) and the experimental value (21.396 mg/g). Therefore, it is proposed that the kinetics of adsorption of Sb(III) by Fe 3 O 4 /HCO can be more accurately described by the Pseudo-second-order model 32 .
In addition, results for the fit to the curve between q t and lnt (Fig. 5a) and the calculated parameters (Table 3) highlight that Elovich model also has a good fit with the experimental data (R 2 = 0.992) 31 .
The intra-particle diffusion model describes the diffusion relationship between the adsorbate and the pores of the adsorbent 29,49 . As shown in Fig. 5b, the fit of the plots of q t and t 0.5 can be divided into two steps, including a fast initial and slow later adsorption stages. The difference between the slopes of the first and second phases, indicate that a gradual phase exists in which the surface adsorption process is controlled by thickness of the boundary layer. The two intercepts (α1, α2) represent the thickness of the theoretical boundary layer of the two stages. As shown in Table 3, the significant difference between α1(−0.139) and α2 (10.176) demonstrates that the pore diffusion rate is not a unique control process. Therefore, the adsorption rate of Sb(III) by Fe 3 O 4 /HCO is determined by the boundary layer effect and the external mass transfer effect 29,31 . thermodynamic studies. Gibbs free energy ΔG° (kJ/mol), standard enthalpy change ΔH° (kJ/mol) as well as standard entropy change ΔS° (J/(mol.k)) are the main parameter of adsorption thermodynamics. In the study, Eqs (2) and (3) were used to calculate the ΔG°, ΔH 0 and ΔS 0 for Sb(III) adsorption onto Fe 3 O 4 /HCO at 20 °C, 25 °C as well as 30 °C (Table 4). Adsorption properties were also investigated, as well as its spontaneity. www.nature.com/scientificreports www.nature.com/scientificreports/ In which R is the molar constant, 8.314 J/(mol.k); T is the absolute temperature, K; K 0 is the equilibrium constant of adsorption thermodynamics. K 0 can acquire according to the method stated by Zheng 47 and the Freundlich equation was used to fit the parameters to calculate K 0 , namely, K 0 = Kf.
As shown in Table 4, the value of K 0 increased (from 4.062 to 5.183) as temperature increased (from 293.15 K to 303.15 K), suggesting that the adsorption process of Sb(III) on Fe 3 O 4 /HCO was an endothermic reaction 50 . The values of ΔG 0 at 293.15 K, 298.15 K and 303.15 K were −3.183 kJ/mol, −3.704 kJ/mol and −3.736 kJ/mol, respectively. All ΔG 0 less than 0 indicates that adsorption of Sb(III) on Fe 3 O 4 /HCO was a spontaneous process 50 . Furthermore, the decrease of ΔG 0 with temperature increasing implied that the degree of spontaneous adsorption could be enhanced with increasing temperature. The values of ΔH 0 and ΔS 0 were 33.427 kJ/mol and 11.75 J/ (mol.k), respectively. ΔH 0 > 0 further indicates that the adsorption process is endothermic. ΔS 0 > 0 demonstrates which adsorption occurred on the surface of the Fe 3 O 4 /HCO adsorbent as a process of random improvement on the solid-liquid surface, and the arrangement of the adsorbed Sb(III) on the surface of Fe 3 O 4 /HCO was chaotic, probably owing to the release of water molecules from hydrated Sb(III) 33 .
Desorption. In this study, desorption of the adsorbed Sb(III) ions from sorbent was also studied in a series of batch experiments. The efficiencies of the different eluents are shown in Table 5. Compared with HCl and NaOH, the repetitive adsorption rate of Fe 3 O 4 /HCO to Sb(III) ions after EDTA and water repeatedly desorption was very low. In a 2 cycle adsorption-desorption process, after desorption by EDTA and pure water, the removal rate of Fe 3 O 4 /HCO to Sb(III) ions decreased to less than 80%(65.27% and 78.21%). After 3 and 4 cycles of adsorption-desorption process with HCl and NaOH desorption, the removal of Sb(III) ions by Fe 3 O 4 /HCO is still close to 80%(79.91% and 79.22%). Compared to HCl, NaOH is cheaper and safer. Thus NaOH solution was used as a desorption agent. Sb(III) ions desorption from Fe 3 O 4 /HCO created the removal process economical both adsorbent and Sb(III) ions were regenerated and recycled effectively.
Adsorption mechanism. From the characterization of before and after the adsorption of Sb on the Fe 3 O 4 / HCO, the mechanisms proposed for the Sb(III) adsorption on Fe 3 O 4 /HCO are illustrated in Fig. 6. The possible reactions in the adsorption process are speculated as Eqs (4)~ (8). And the preferred adsorption mechanisms between Sb(III) and Fe 3 O 4 /HCO was concluded as following: Models Initial Sb(III) concentrations (mg/L) 10 50 100    (Fig. 6a). In addition, Sb(III) may be adsorbed by the inner sphere complex of Fe 3 O 4 and other spherical complexes 36 (Eq. (4)). 2. Adsorption of Sb by HCO. When the pH of aqueous solution is 6.7, HCO is a hydrated metal oxide with zero surface charge. The XRD diffraction pattern (Fig. S2) confirms that the compound CeSbO 3 exists in the residual precipitate after adsorption. Therefore, the second reaction mechanism of Fe 3 O 4 /HCO adsorbing Sb(III) is speculated as shown in Fig. 6b. CeSbO 3 was synthesized by the reaction of HCO with H 3 SbO 3 (HCO + H 3 SbO 3 → CeSbO 3↓ + H 2 O), and the main mechanism of HCO adsorbing anions in water is the exchange reaction of anionic ligands 37 (Eq. (5)). 3. FeCe 2 O 4 was used to hydrolyze HCO and FeOOH, and then they reacted with Sb (III). In the preparation of Fe 3 O 4 /HCO, two Ce 3+ ion replaced Fe 3+ at octahedral sites in a lattice structure (Fe 3 O 4 + HCO + OH − = FeCe 2 O 4 + H 2 O). FeCe 2 O 4 was hydrolyzed in aqueous solution, and electron and ion was transferred occurred between phase interface and aqueous solution, forming a double-electron layer structure. This results in the in situ formation of an amorphous hydrated iron oxide which has a larger specific surface area 19 . Ligand exchange as well as adsorption of Sb (III) occurred on the iron oxide film (Eqs (6)~ (7)). 4. From other studies, it can be concluded that Ce(IV) itself being a good oxidizing agent (Ce 4+ /Ce 3+ = 1.72) in acidic medium 27 . As shown in the Fig. S4, Ce(IV) can oxidizes the surface sorbed Sb(III) to Sb (V) (Eq. (8)), It is basically consistent with the Sb 3d XPS Spectra. And Ce(IV) itself getting reduced to Ce 3+ according to the underlying redox reaction.