Phosphoric acid-induced activation of sepiolite for enhanced As(III) adsorption: role of in situ deposition of nano-hydroxyapatite

Abstract A simple and effective approach was introduced to modify natural calcium-rich sepiolite (SEP) by in situ deposition of nanosized hydroxyapatite (nHAP) to improve SEP’s surface properties and adsorption potential to aqueous As(III). The obtained SEP-nHAP nanocomposites were characterized by means of X-ray diffraction, FTIR, SEM-EDS, and BET analysis. Adsorption of As(III) on SEP-nHAP was evaluated as a function of pH, contact time, coexisting anions, initial As(III) concentration, and temperature. Results showed that modification of SEP by nHAP deposition significantly increased the surface area of the SEP from 3.05 to 76.56 m2/g, and As(III) removal by SEP-nHAP nanocomposite was optimal at a pH range of 6.0–8.5 and at 2 h equilibrium time. The adsorption isotherm data fitted well to the Sips and Langmuir models with a maximum adsorption capacity of 17.34 mg/g, which outperformed not only the pure SEP but also many other reported adsorbents. In addition, the kinetics analysis revealed that the As(III) adsorption process was well described by the pseudo-second-order kinetic model. The primary mechanisms for As(III) adsorption included electrostatic interaction, the formation of complexation among As(III) species, and surface hydroxyl groups of nHAP. Furthermore, the SEP-nHAP adsorbent could be easily regenerated and reused with only a slight reduction in the As(III) adsorption capacity. Therefore, SEP-nHAP can be a suitable candidate for removing As(III) from groundwater. Graphical Abstract


Introduction
Arsenic (As) contamination of groundwater and the related threats to human health have been a worldwide problem. [1] Depending on the pH and redox potential, carcinogen As exists primarily in two oxidation states in natural waters: arsenite [As(III)] and arsenate [As(V)]. Arsenate occurs as oxyanions (H 2 AsO 4 or HAsO 4 2-) in a pH range of 2-12, while arsenite remains as neutral undissociated species (H 3 AsO 3 ) at pHs below 9.2. [2] Generally, the toxicity of As is governed by its speciation. Since the pH of groundwater is often in the range of 6.0-8.0 and As(III) oxidation is often sluggish, [3] As(III) is the dominant species in groundwater, which is much more readily bio-accessible, soluble, and mobile than As(V). [4] Based on the mouse data of Vahter and Norin, [5] inorganic As(III) is more extensively absorbed from the gastrointestinal tract compared with inorganic As(V) at lower doses (e.g., 0.4 mg As kg À1 ). Delnomdedieu et al. [6] investigated the uptake of inorganic As(III) compared to inorganic As(V) in intact rabbit erythrocytes. They found that $76% of inorganic As(III) compared to $25% of inorganic As(V) was taken up within 0.5 h and that inorganic As(III) subsequently bound with intracellular glutathione, whereas inorganic As(V) entered through the phosphate pathway. It is accepted that As(III) is 60 times more toxic than As(V), which is attributed to the greater combining affinity of As(III) with thiol (-SH) group of protein for the soft-soft type of acid-base reaction. [7] Long term exposure to arsenic through drinking water can cause arsenicosis, a disease which is manifested by different types of cancer, hypertension, neurological complications, and cardiovascular disease. [2] Considering the high toxicity and carcinogenic effects, the U.S. Environmental protection agency (USEPA) and World Health Organization (WHO) have lowered the maximum permissible limit of As in drinking water from 50 down to 10 lg/L. [8] According to the Chinese Groundwater Quality Standard (GB/T 14848-2017), the maximum permissible limit is As < 10 lg/L for groundwater used in centralized drinking water and industrial or agricultural water. [9] Therefore, removal of As species from contaminated water is essential for the green development of the environment and human health. Various physical and chemical technologies have been developed for the remediation of As(III)-contaminated water, including coagulation, [10] reverse osmosis, [11] filtration, [12] precipitation, [13] etc. However, the major drawbacks of these methods are their high cost, intensive energy requirements, and formation of secondary waste products. In contrast, adsorption has been considered to be superior to other techniques due to treatment stability, easy operation, lower environmental impacts, and low cost. [4,8,14,15] Among a diversity of adsorptive materials that are capable of adsorbing As(III) from water, clay minerals are applied on a large scale. For instance, Manning and Goldberg [16] first reported the pH-dependent adsorption of As(III) on phyllosilicates (kaolinite, illite, and montmorillonite). Yu et al. [17] indicated that the calcined hydrotalcite had a higher As(III) adsorption capacity than the uncalcined hydrotalcite. Jovanovic et al. [18] reported that As(III) adsorption on natural zeolite had a maximum adsorption capacity of 0.97 mg/ g. Similarly, the As(III) adsorption by clinoptilolite-rich tuffs, [19] beidellite and zeolite, [20] montmorillonite, [21,22] red clay, [23] and sepiolite (SEP) [20] have been utilized for the removal of As(III) from water. However, natural clays sometimes provided very low adsorption affinity for As(III) due to the low surface area of natural adsorptive materials as well as the undissociated nature of As(III), which only had weak interactions between the adsorbate and adsorbent. [24] To address these issues, various surface modification processes were put in practice to promote As(III) adsorption, including Fe and Ce-modified zeolite, SEP, and beidellite, [20] surfactant modified montmorillonite, [21] H 2 SO 4 activated montmorillonite, [22] maghemite-modified SEP, [25] iron oxide-coated SEP, [26] iron oxide and manganese oxide pillared clays, [27] and Fe-Mn bimetal-modified kaolin. [28] In particular, the modification of clays with nanoparticles gained importance because nanoparticles exhibited some benefits such as small size, large surface area, high reactivity, and large number of active sites that facilitated better As(III) removal efficiency. [1,29] Furthermore, the clays could also serve as supporting materials, which could decrease aggregation of nanoparticles and enhance their reactivity. In the recent studies, nanoscale zero-valent iron-modified montmorillonite [30,31] and CeO 2 nanoparticle-decorated halloysite [32] have been reported to exhibit much higher As(III) adsorption capacities than the pristine clays. Nevertheless, the high cost, complex fabrication route, and potential environmental risk of zero-valent iron or metal oxide nanoparticles precluded their use in the modification of clays [1,33] and necessitated the development of more feasible, costeffective, and eco-friendly nanoparticles for modifying natural clays.
As a nontoxic, biocompatible, and environmental benign functional material, nano-hydroxyapatite (nHAP) exhibited great potential in the remediation of not only metal cations, [34] but also inorganic oxyanions such as Cr(VI) [35] and As(V). [36,37] But there have been very few studies directed toward the evaluation of the As(III) adsorption performance of nHAP. [38,39] In addition, SEP is a natural hydrated magnesium silicate clay mineral with a chemical formula of Si 12 Mg 8 O 30 (OH) 4 (OH 2 ) 4 Á8H 2 O, which is structurally composed of alternation blocks and interior tunnels along the fiber direction. [40] The nontoxicity, natural abundance, relatively low cost, high mechanical stability, high porosity, plentiful surface hydroxyl groups, and unique fibrous structure with interior channels allow feasible surface modification of SEP and assign SEP an important role in mineral, chemical, and environmental applications. [41,42] Moreover, approximately 10 million tons of SEP deposits exist in China (approximately 1/5 of the world's reserve). [43] The abundance and availability of SEP reserves together with its relatively low-cost guarantee its continued widely utilization. Furthermore, it was reported that the chemical composition of SEP varies geographically and the Chinese SEP is characterized by high calcium content (the Chinese SEP has a calcium oxide content as high as 30%). [40] The calcium-rich SEP has been reported to be modified by various nanoparticles for catalytic and adsorptive applications, [41,42] thus it is also expected to be an ideal material to be modified with nHAP through in situ precipitation. The use of SEP as the support material of nHAP exhibits two obvious advantages: first, the support is widely available and low-cost without further synthesis; second, the calcium ions required for nHAP preparation are derived from calcium-rich SEP itself without further addition of calcium salts.
Herein, this work aims to prepare a SEP-nHAP composite via H 3 PO 4 activation and in situ deposition of nHAP on the SEP surface, and then develop it into an As(III) adsorption material for the first time. The structure, morphology, and porosity of the prepared SEP-nHAP composite were studied using X-ray diffraction (XRD), Fourier transform infrared spectroscopy (FTIR), scanning electron microscopy coupled with energy dispersive spectroscopy (SEM-EDS), and N 2 adsorption-desorption characterizations. Batch adsorption experiments of As(III) as affected by several parameters (pH, contact time, coexisting ions, and temperature) were then systematically investigated on this adsorbent. In addition, the adsorption kinetics and isotherms were also evaluated to determine the mechanism of adsorption.

Materials
The SEP sample was supplied by Beijing Fuhong company (China). It was ground and sieved with a 100-mesh sieve before use. Chemical compositions (by weight percent) of the pristine SEP were SiO 2 (35.6%), MgO (16.7%), and the impurities were CaO (23.3%) (indicating the calcium-rich characteristic of the Chinese SEP), Al 2 O 3 (1.49%), and Fe 2 O 3 (0.44%). Analytical-grade chemicals were used without any additional purification. Deionized water was used throughout the experiments. The stock solution of As(III) was prepared by dissolving an accurately weighed amount of NaAsO 2 in deionized water.

Preparation of nHAP modified SEP
The in situ deposition of nHAP onto SEP was conducted through a H 3 PO 4 activation process of calcium-rich SEP. This process involved treatment of SEP with H 3 PO 4 which acted as a decalcifying agent, promoting partial dissolution and calcium release of the calcium-rich SEP. [44,45] H 3 PO 4 also reacted with the released calcium ions to form nHAP under alkaline conditions. In a typical preparation procedure, 10 g SEP was added into 100 mL of 2 mol/L H 3 PO 4 solution, and the dispersion was constantly stirred at room temperature in a flask for 2 h. Then, the pH of the dispersion was adjusted to 10.0 using 1 mol/L NaOH, and further stirred for another 2 h. The pH was maintained at 10.0 throughout the experiment. The mixture was then filtered through a vacuum pump and washed repeatedly with deionized water and dried at 100 C in an oven. The obtained sample was marked as SEP-nHAP and used in this study.

Characterization
The phase composition of the prepared SEP-nHAP was characterized using XRD (Bruker D8 Advance, Cu Ka radiation, k ¼ 1.5406 nm). The morphology and microstructure of the sample were investigated by a Hitachi S-3000N SEM-EDS (either point analysis or elemental mapping). Surface functional groups of the pristine SEP and SEP-nHAP were examined by FTIR spectroscopy (Nicolet iS5) in the range of 400-4000 cm À1 . The Brunauer-Emmett-Teller (BET) surface area and porosity of the samples were determined with a nitrogen adsorption-desorption instrument (Micromeritics ASAP2460).
By definition, the point of zero charge (pH PZC ) is the pH of the solution in contact with the solid when the net surface charge on the surface of an adsorbent particle is zero. Batch equilibrium method was employed to measure the pH PZC of SEP-nHAP adsorbent. [46] For this, 0.1 g of adsorbent was added into different vials containing 50 mL each of 0.1 mol/L NaNO 3 solution applied as an inert electrolyte. The initial pH values (pH initial ) were adjusted within the range of 2-10 using 0.1 mol/L HNO 3 or 0.1 mol/L NaOH. These vials were then placed on a reciprocating shaker for 24 h at room temperature to reach equilibrium. After 24 h, the final pH values (pH final ) of the suspensions were measured and plotted against the pH initial . The pH PZC values were obtained from the plateaus of the pH final versus pH initial plots.

As(III) adsorption experiments
For batch adsorption of As(III), SEP-nHAP adsorbent (25 mg) was accurately weighed into a 150 mL conical flask. Then, 50 mL of As(III) solution (10 mg/L) was added and the pH of the solution was adjusted and shaken by a reciprocating shaker. To evaluate the optimum conditions for As(III) batch adsorption, various influencing factors were studied including the effects of pH (3-10), adsorbent dosage (0.5-5 g/L), coexisting ions, contact time (0-3 h), As(III) concentration (0.5-10 mg/L), and temperature (283, 298, and 313 K). At the end of the adsorption process, the adsorbent was separated from the solution, and the supernatant was analyzed for the residual As(III) concentration using the hydride-generation atomic fluorescence spectrophotometer (AFS-2202, Haiguang, China). An arsenic hollow cathode lamp (197.2 nm) is used as the radiation source. The optimized operating parameters are as follows: negative high voltage of the photomultiplier: 350 V, lamp current: 65 mA, carrier argon flow: 350 mL/min, shield argon flow: 800 mL/min, and atomizer height: 8 mm. Prior to analysis, the aqueous samples were acidified with 1% HNO 3 and stored in acid-washed glassware vessels. Three replicates were performed for each As(III) adsorption experiment and the results given were the average values. The equilibrium adsorption capacity (q e ) for As(III) was calculated using the general equation q e ¼ (C 0 À C e )V/m. To investigate the adsorption processes of As(III) on SEP-nHAP, five kinetic models were used, including the pseudo-first order, pseudo-second order, Elovich, Boyd, and Intra-particle diffusion kinetic models. The Langmuir, Freundlich, Sips, Temkin, Redlich-Peterson, and Hill adsorption models were tested to interpret the isotherm data. The detailed information of these models can be found in the supplementary material.

Regeneration of adsorbent
Reusability experiments were carried out in a similar way to adsorption studies. A 0.1 g sample of SEP-nHAP was placed in 50 mL of 10 mg/L As(III) solution, and the mixture was shaken for 2 h at room temperature. After adsorption, the As(III)-loaded adsorbent was separated from the solution by centrifugation. For regeneration, the As(III)-loaded SEP-nHAP was desorbed with 50 mL of 0.1 mol/L NaOH for 2 h. The desorbed adsorbent was washed with water several times and subjected to the next cycle of adsorption. Figure 1 shows the XRD patterns of SEP before and after H 3 PO 4 activation. For the pristine SEP, the characteristic diffraction exhibited the peaks assigned for SEP, dolomite, calcite, and a small quantity of quartz. The impurity of the SEP used in this study was also observed in the literature on Chinese SEP. [40,47] After the activation with H 3 PO 4 , the characteristic peaks of SEP remained in SEP-nHAP, showing that the modification process did not have a remarkable influence on the structural framework of the SEP sample.  However, the intensity of calcite and dolomite peaks in SEP-nHAP was reduced compared with those of pristine SEP, indicating that H 3 PO 4 activation resulted in partial dissolution of calcite and dolomite. [45] In particular, some new peaks appeared at 2h ¼ 25.9 , 31.8 , 32.1 and 32.9 , corresponding to the characteristic diffraction peaks of nHAP (JCPDS card no. 09-0432), [48] suggesting the successful deposition of nHAP on SEP. Furthermore, these new diffraction peaks yielded relatively broad and overlapping reflections, indicating the low crystallinity and small size of the deposited nHAP. [49] FTIR spectra of the pristine SEP and SEP-nHAP are depicted in Figure 2. It was obvious that the major band positions did not change after H 3 PO 4 activation, implying that the basic crystal structure of SEP remained constant. The adsorption band observed at 3674 cm À1 was ascribed to the Mg-OH stretching vibration. [42] The adsorption bands that appeared around 3421 and 3178 cm À1 were relevant to the O-H stretching vibration of bound water, while the band at 1636 cm À1 corresponded to the -OH deformation of water, [40] and the stretching vibrations obtained at 1428 and 876 cm À1 could be characteristics of impurity carbonate in SEP. [40,42] The peaks around 948 and 1018 cm À1 represented the stretching of Si-O in the Si-O-Si groups of the tetrahedral sheets of SEP. The bands at 476 and 470 cm À1 were due to O-Si-O bending vibrations. [42] It was noteworthy that the peak intensity for carbonate (1428 and 876 cm À1 ) decreased remarkably in the spectrum of SEP-nHAP, indicating partial dissolution of carbonate induced by H 3 PO 4 activation. In particular, for the SEP-nHAP, some new absorption peaks appeared at 566 and 603 cm À1 , which were ascribed to the 4 bend vibrations of PO 4 3À, [48] while the bands at 1040 and 1096 cm À1 showed the as stretching of PO 4 3group ( 3 ). [49] These results suggested the formation of HAP was subjected to the H 3 PO 4 activation of calciumrich SEP and was in accordance with the XRD data.

Characteristics of SEP-nHAP
The representative SEM image in Figure 3a showed that the pristine SEP exhibited a smooth surface and a characteristic rod-like morphology with a diameter of several hundred nanometers and a length of several micrometers. The EDS spectrum of SEP (inset of Figure 3a) showed the main peaks of O, Si, Mg, and Ca, which were consistent with the literature. [50] After H 3 PO 4 activation, it was clear from the SEM images (Figure 3b) that the covering of dispersive and small particle-sized nHAP made the surface rough. It was considered that H 3 PO 4 activation not only facilitated the  disaggregation of SEP but also served as nucleation centers for the deposition of nHAP nanocrystals. [25,51] The higher magnification SEM image of the SEP-nHAP (Figure 3c) depicted nanoparticles deposited on the SEP surface, and no obvious aggregates of the nanoparticles were observed. As shown in Figure 3c, nHAP nanoparticles with a size distribution of around 100 nm were highly dispersed on the SEP surface. The elemental distribution of the SEP-nHAP surface was further confirmed by an EDS-mapping technique (Figures d-i), which showed that the composite contained the elements Si, Mg, Ca, P, and O, indicating the formation of nHAP on the SEP surface. These results proved the successful preparation of the SEP-nHAP composite and indicated that the surface of SEP was modified by nHAP particles. Figure 4 (a-d) displays nitrogen adsorption-desorption isotherms and pore size distribution curves of pristine SEP and SEP-nHAP composites. As can be seen, the BET surface area of the pristine SEP was 3.05 m 2 /g. This value was much lower than the reported SEP, [25,45] which was probably due to the presence of agglomerated dense bundles and carbonate impurities of natural calcium-rich SEP. The surface area of the SEP-nHAP sample was found to be 76.56 m 2 /g, which was 24 times higher than that of the pristine SEP. Previous studies also confirmed that acidic treatment increased the surface area of SEP. [47] The average pore diameter and pore volume of SEP were 9.66 nm and 0.0089 cm 3 /g, which changed to 7.66 nm and 0.1838 cm 3 /g in SEP-nHAP, respectively. The impurities such as calcite and dolomite in SEP were partly dissolved during the H 3 PO 4 treatment, which could result in the connections of original channels in the SEP structure, [52] thus enhancing the average pore volume. Moreover, the increased surface area of SEP-nHAP might be interpreted by the deposition of tiny nHAP particles on the SEP surface. From these results, it can be concluded that an SEP-nHAP composite with good surface properties was successfully prepared by a facile in situ deposition method. In general, a larger specific surface area of SEP-nHAP could bring more active sites and adsorptive interfaces for As(III) elimination. The results demonstrated that SEP-nHAP exhibited higher adsorption capacities than pristine SEP. Jovanovic et al. [18] also reported the poor adsorption capacity (< 2.0 lg/g) of natural SEP toward both As(V) and As(III) species, which might be due to the lack of active sites. In another study, Bektaş et al. [20] reported an adsorption capacity of 515.7 lg/g for As(III) using ferric nitrate-modified SEP. More recently, Ili c et al. [53] showed the enhanced As(III) uptake by hydrated iron(III) oxide-modified SEP with an adsorption capacity of 9.5 mg/g. Furthermore, the synthetic HAP adsorbent derived from carp fish scales was reported to exhibit significant potential for As(III) adsorption. [38] In this work, the SEP-nHAP adsorbent had a much higher adsorbed amount of As(III) than the pristine SEP at the given conditions, while the pristine SEP showed a quite weak adsorption of As(III). The difference on As(III) adsorption ability between SEP and SEP-nHAP illustrated that the nHAP particles deposited on the surface of SEP and provided the main activated sites for the As(III) adsorption.

Effect of pH on as(III) adsorption by SEP-nHAP
The adsorption of contaminants from aqueous solutions is always dependent on the pH of the solution, which can affect the surface charge of adsorbents, degree of ionization, as well as the speciation of adsorbate species. [16] The pH dependence of As(III) adsorption on SEP-nHAP is shown in Figure 6a. The As(III) adsorption capacity of SEP-nHAP gradually increased in the pH range of 3.0-8.0 and dramatically decreased when the pH rose from 9.0 to 10.0. In the pH range of 3.0-6.0, As(III) adsorption was observed to increase rapidly from 7.2 to 12.7 mg/g, which was maintained in the pH range of 6.0-8.5 and only exhibited a slight increase to 14.1 mg/g at a pH of 8.0. This phenomenon could be ascribed to As(III) speciation variation and the surface charge change of SEP-nHAP, as the solution pH changed. At pH < 9.2, the dominated As(III) species is neutral H 3 AsO 3 . [2] In this case, the gradual increase in As(III) adsorption with the rise of pH could be due to the reason that H 3 AsO 3 preferred to form surface complexes through substitution of water molecules or hydroxyl groups on adsorbent surfaces. [54] With the further increase of pH values (pH > 7.0), H 3 AsO 3 species gradually converted to monoionic forms of H 2 AsO 3 -; thus, more As(III) uptake by SEP-nHAP adsorbent was expected because the surface of SEP-nHAP was positively charged below pH ZPC (pH < pH ZPC of 9.41, Figure 6b). Accordingly, electrostatic forces of attraction between the negatively charged H 2 AsO 3 and positively charged SEP-nHAP surface favored As(III) adsorption to a greater extent. [55] In addition, the increase in As(III) adsorption at a higher pH may also be attributed to the good stability of the adsorbent itself. [56] At pHs above pH ZPC , the anionic form of As(III), H 2 AsO 3 À , prevailed, and thus the repulsive forces between the negatively charged adsorbent surface and negatively charged H 2 AsO 3 À resulted in a remarkable drop of the As(III) sorption capacity. [20] Moreover, the decrease in adsorption at pH > 9.41 also might be due to the competition for the adsorption sites between hydroxyl ions and the predominant anionic As(III) species. To obtain the optimal adsorption of As(III), a pH of 8.0 was used in subsequent experiments. Based on the above-mentioned results, it was found that the favorable pH for the adsorption of As(III) species laid in the range of 6.0-8.5, which was consistent with the typical pH value of groundwater. [57] Thus, it is possible to use SEP-nHAP as a feasible decontamination material for groundwater. Figure S1 (supplementary material) shows the adsorption kinetics of As(III) on SEP-nHAP at pH 8.0. The plots in the figure depicted that the adsorption increased with time until equilibrium was attained. It was obvious that there were two adsorption stages. The adsorption was fast at the initial stage and became slow near equilibrium. The two-stage adsorption process has been previously reported for several adsorbents. [25,32,54,56] It was considered that the rapid step was quantitatively predominant and probably due to the abundant availability of active sites on the SEP-nHAP surface. The second step was slow and quantitatively insignificant, with gradual occupancy of the pore surface active functional sites. [51] In addition, the fast removal stage could be also attributed to the adsorption on the exterior surface of adsorbent, while with the gradual occupancy of these sites, the adsorption becomes less efficient in the second step. The data also indicated that the adsorption equilibrium time of As(III) by SEP-nHAP was approximately 2 h. Then a shaking time of 2 h was used for subsequent experiments.

Adsorption kinetics
The pseudo-first order, pseudo-second order, Elovich, Boyd, and Intra-particle diffusion kinetic models were applied to further investigate the adsorption performances of As(III) on SEP-nHAP, as shown in Figures S1 and S2  (supplementary material). Pseudo-first order and pseudosecond order were used to determine the rate of the adsorption process. Elovich model was used to determine the rate of the reaction and the nature of the adsorption process. Intra-particle diffusion model and Boyd model were applied to determine the rate limiting step. The calculated kinetic parameters are presented in Table S1 and S2 (supplementary material). It was found that the pseudo-second-order model fitted the As(III) adsorption data much better than the other kinetic models according to the relatively higher correlation coefficient (R 2 > 0.95), which suggested that the chemical interactions played an important role in the As(III) adsorption process. In addition, the values of experimental adsorbed capacities (q e , exp) were much closer to the calculated values (q e , cal) from the pseudo-second-order model. This further confirmed the applicability of the pseudosecond-order model in the adsorption of As(III) using the  [15] Calcined LDHs 250 cmol/kg 0-4 cmol/L 24 4.0 10 298 [17] Natural zeolite 0.97 0.5-100 6 10.0 7 - [18] Fe-SEP 0.48 250-5000 lg/L 1 4.0 -298 [20] Fe-Beydellite 0.79 250-5000 lg/L 1 4.0 -298 [20] Modified montmorillonite 10.6-15.03 1-2500 24 10 5 298 [22] Red Clays 0.292 10 4 2.5 10 298 [23] Magnetic iron oxide anchored SEP 50.35 0.1-50 24 0.2 7 298 [25] Iron oxide-coated pumice 0.579 100-1000 lg/L 84 2.0 7 303 [26] Iron SEP-nHAP adsorbent. Previous studies have also showed that the adsorption of As(III) onto Fe-Mn bimetal modified kaolin clay, [28] halloysite-CeO x nanohybrid, [32] magnetic iron oxide nanocrystal-anchored SEP, [25] mesoporous zirconia nanostructures, [54] and mesoporous magnesium oxide hollow spheres [56] fitted the pseudo-second-order kinetics well. The Elovich model considers that the real solid surface is energetically heterogeneous and that the desorption process and interactions between adsorbed species does not influence the adsorption kinetic significantly. [48] The Elovich model is also illustrated in Figure S1 (supplementary material) and the constants obtained from the Elovich equation are introduced in Table S1. It can be noted that the Elovich mode did not exhibit better adaptability with the kinetics data.
As the above kinetic models were not able to determine the diffusion mechanisms and identify the possible rate controlling procedure that affected the kinetics of adsorption, Intra-particle diffusion and Boyd's film-diffusion model were further examined. Based on the Boyd model, if the plot of B t versus t produces a straight line that passes through the origin, pore diffusion is the rate limiting step; Otherwise, the adsorption process is intra-particle diffusion controlled. [58] Figure S2 (left panel, supplementary material) shows that the B t Àt plot is straight lines but not completely pass through the origin, indicating that the particle diffusion is the controlled rate of As(III) adsorption process with involvement of film diffusion in the mechanism of adsorption. This is combatable with the results obtained from Intra-particle diffusion kinetic model as illustrated in Figure  S2 (right panel, supplementary material), which shows two steps with different slopes for HAP-nHAP adsorbent. These two linear portions in the Intra-particle model suggest that the adsorption process consists of both surface adsorption and intraparticle diffusion. While the initial linear portion of the plot is the indicator of boundary layer effect, the second linear portion is due to intra-particle diffusion. [59] The constants of Intra-particle diffusion model for the two steps are presented in Table S2 (supplementary material), which indicated that the Intraparticle diffusion model fitted the experimental data very well for As(III) adsorption process.

Effect of competing ions
It was generally accepted that competition of natural groundwater constituents with arsenic for the surface sites mainly arise from anions, especially oxyanions, due to the anionic nature of inorganic arsenic in water. [60][61][62] The adsorption performance of SEP-nHAP adsorbent toward As(III) in the presence of various coexisting anions (1-50 mg/L of Cl À , NO 3 À , HCO 3 À , CO 3 2À , and SO 4 2-) that were commonly found in groundwater was evaluated to investigate its potential application in treating As(III) contaminated groundwater. The experimental tests were performed in a binary system containing 10 mg/L of As(III) and various concentrations of each competing ion. The obtained results (Figure 7) demonstrated that the As(III) adsorption was only slightly influenced in the presence of other coexisting ions, revealing the proper selectivity of the proposed SEP-nHAP adsorbent for As(III) removal.

Adsorption isotherms and thermodynamic parameters
The equilibrium adsorption isotherm is of fundamental importance in the design of adsorption systems. The obtained As(III) adsorption equilibrium data at 283, 298, and 313 K were analyzed using nonlinear Langmuir, Freundlich, and Sips, Temkin, Redlich-Peterson, and Hill isotherm models. The fitting plots for these models are shown in Figures S3 and S4 (supplementary material). The adsorption constants obtained from the isotherms are listed in Table S3 (supplementary material). The parameters calculated from the model fitting results showed that the isothermal adsorption process followed both the Langmuir model and Freundlich model since all the correlation coefficients (R 2 ) were higher than 0.94, suggesting that the adsorption of As(III) on the surface of adsorbents was composed of both homogeneous and heterogeneous adsorption. The theoretical maximum monolayer adsorption capacities (q m ) were found to be 17.34 and 17.91 mg/g at 298 and 313 K, respectively, much higher than that at 283 K, suggesting the adsorption  reaction was more favorable at higher temperatures. The Freundlich constant K F is considered to be a parameter related to the adsorption capacity of the adsorbent. In this work, the maximum value of K F was observed at 313 K and the minimum value was observed at 283 K (Table S2, supplementary material). The 1/n value is related to strength of the adsorption, by which the adsorption process could be divided into five categories, namely pseudo-irreversible, strongly favorable, favorable, pseudo-linear, and unfavorable levels, corresponding to 1/n < 0.01, 0.01 < 1/n < 0.1, 0.1 < 1/ n < 0.5, 0.5 < 1/n < 1, and 1/n > 1, respectively. [63] It was found that most values of 1/n were around or below 0.5 in this study, suggesting that the As(III) adsorption process of SEP-nHAP is a favorable process. The Sips isotherm is a combination of the Langmuir and Freundlich isotherms, and can reduce to either at its limits. When 1/n equals unity, the Sips isotherm returns to the Langmuir isotherm and predicts homogeneous adsorption. Alternatively, as either K S approaches zero, the Sips isotherm reduces to the Freundlich isotherm. Equilibrium adsorption data (Table S2, supplementary material) showed that the Sips model exhibited the best accordance with the equilibrium data based on the R 2 values (R 2 > 0.98). As demonstrated in Table S2 (supplementary material), all 1/n values of the Sips model were close to 1, which meant that the As(III) adsorption isotherms obtained in this work were more of the Langmuir model than that of the Freundlich model. Temkin isotherm model is an empirical equation that contains a factor accounting for the interaction between the adsorbent (SEP-nHAP) and adsorbate (As(III)). This equation (given in the Supplementary Material) has been expressed based on chemical adsorption and the assumption that adsorption heat of As(III) species in the layer decreases linearly rather than logarithmically as the As(III) adsorption capacity increases, because the Temkin isotherm constant (B) is related to As(III)-adsorbent interactions. [64] Fitting of the experimental data to the nonlinear form of Temkin isotherm is shown in Figure S4 (supplementary material). The obtained parameters are listed in Table S3. The Redlich-Peterson (R-P) isotherm model combines the properties of Langmuir and Freundlich isotherms. This isotherm model can be used either in homogeneous or heterogeneous systems. The Redlich-Peterson isotherm constants for adsorption of As(III) onto SEP-nHAP are shown in Table S3. The isotherm constant K R and a R increased with temperature and reached maximum at 313 K, indicating a favorable adsorption at higher temperature. It is worth noting that n values were close to unity, which suggested that the adsorption data could preferably be fitted with Langmuir model. These results were in accordance with the Sips model. The Hill isotherm model is derived to describe the adherence of different species onto homogeneous substrates. The graphical representations of nonlinear form of Hill model applied to experimental data were shown in Figure S4 and the values of isotherms parameters were included in Table S3 (supplementary material). Based on the regression coefficient values and also considering the parameters values of the isotherms model, the Hill model also could well describe the experimental results. Hill model considers a biosorption process as "a cooperative phenomenon with adsorbates at one site of the adsorbent influencing different binding sites on the same adsorbent" and a homogeneous adsorbent surface. [65] The maximum adsorption capacity calculated by Hill nonlinear equation for As(III) is 16.80-23.79 mg/g, which were close to the experimental data. Overall, it was found that the capability of models used for data correlation of As(III) adsorption isotherms on SEP-nHAP descended in the order: Sips > Langmuir > Redlich-Peterson > Hill > Temkin > Freundlich.
Furthermore, the comparison of the maximum adsorption capacities (q m ) obtained in the present study with those of other adsorbents reported in the literature is listed in Table 1. It indicated that the prepared SEP-nHAP adsorbent exhibited a higher adsorption capacity toward Ar(III) than Fe/Al bimetals [4] and many natural or modified clays. [18,[20][21][22][23][26][27][28]30] These results suggested the suitability of the proposed SEP-nHAP adsorbent for As(III) removal. In comparison, the q m of SEP-nHAP was lower than other adsorbents, such as magnetic iron oxide-anchored SEP, [25] montmorillonite-supported NZVI, [31] and halloysite-CeO x nanohybrids. [32] Nevertheless, those adsorbents had some drawbacks, including the addition of expensive raw materials, tedious synthetic routes, and the use of some environmentally incompatible and toxic additives, which made SEP-nHAP still a competitive adsorbent against those materials. However, it should be noticed that q m is a function of the initial adsorbate concentration and adsorbent dosage used. [28] Generally, higher q m was gained at higher initial concentrations and lower adsorbent dosages. Most of the previous adsorbents that reported higher q m compared to SEP-nHAP might have used higher initial As(III) concentrations.
The values of thermodynamic parameters help to test the spontaneous occurrence of a given process as well as the viability of the operation at a given temperature. Thermodynamic parameters (Table 2) such as the standard free energy change (DG 0 ), standard enthalpy change (DH 0 ), and standard entropy change (DS 0 ) were calculated based on the method recommended by Garcia-Delgado et al.. [66] Enthalpy (DH 0 ) was determined from the slope of the ln(1/ C e ) versus 1/T profiles ( Figure S3, supplementary material).
The negative values of DG 0 and positive values of DH 0 indicated that As(III) adsorption was governed by a spontaneous and endothermic nature. The DS 0 values were obtained as positive, which have ensured increased randomness during the As(III) adsorption on SEP-nHAP.

Reusability study
Recyclability and regeneration of adsorbents are important factors for its economic use and applicability in an aqueous media. Therefore, it is of great significance to investigate adsorbents that possess high adsorption capacities as well as a worthwhile desorption property. To evaluate the reusability of the SEP-nHAP adsorbent, four consecutive adsorption/desorption cycles were conducted using 0.1 mol/L NaOH as the desorption reagent. At high pHs, the surface hydroxyl groups of adsorbents became deprotonated and negatively charged, resulting in the efficient desorption of negatively charged arsenic species. [57] Figure 8 showed that As(III) adsorption capacities of SEP-nHAP reduced gradually from 14.1 to 13.3, 12.4, 11.5, and 10.8 mg/g, suggesting that SEP-nHAP could be used repeatedly with a slight loss of adsorption capacity for As(III).

Cost analysis
The H 3 PO 4 activation of natural calcium-rich SEP leads to the in-situ deposition of nHAP onto the SEP surface, which can help to reduce the cost for preparing SEP-nHAP composites since there is no need to use additional calcium salts. China hosts one of the world's major SEP reserves, and the wholesale price of raw SEP is approximately $180-$300 per ton. [67] The 2 mol/L H 3 PO 4 used in the experiment was prepared by diluting analytical grade phosphoric acid (85%, $8.90 per liter), and other raw materials can be cheaply obtained from industrial suppliers. Compared with other adsorbents such as single-walled carbon nanotubes ($441.85 Â 10 6 per ton), multi-walled carbon nanotubes ($2.5 Â 10 6 per ton), carboxyl multi-walled carbon nanotubes ($70 Â 10 6 per ton), C18 silica ($400 Â 10 4 per ton), [68] activated alumina ($660-1190 per ton), coconut-shell-based activated carbon ($880-1320 per ton), and coal-based activated carbon available in China ($540-710 per ton), [69] the cost of our prepared SEP-nHAP adsorbents is much lower.

Adsorption mechanisms
In the As(III) adsorption process, the proposition of adsorption mechanisms is still a foremost challenge. It is considered that numerous factors, including the speciation of adsorbates, physicochemical characteristics of adsorbents, and the interactions of adsorbates with adsorbent surfaces, are of great importance in the establishment of adsorption. [70] In order to explain the As(III) adsorption behavior, it is important to know that the dominant As(III) species in the adsorbate solution were H 3 AsO 3 and H 2 AsO 3 À at pHs below 9.2. [71] The pH-value dependence of As(III) adsorption (Section 3.3) onto SEP-nHAP could be explained by the pH ZPC (9.41) of the adsorbent. At a pH below 7.0, the significant increase of As(III) adsorption was due to the formation of complexation among neutral H 3 AsO 3 and surface hydroxyl groups on the SEP-nHAP surface. [54] At 7.0 < pH < 9.41, the monoionic H 2 AsO 3 adsorption occurred through electrostatic attraction with positively charged SEP-nHAP adsorbent. [55] The As(III) adsorption dropping at pH values above 8.0 was due to the ionization of H 3 AsO 3 , which resulted in more competition between negatively charged arsenite and OH À anions. Moreover, at pH > 9.41 (pH ZPC ), the increasing coulombic repulsion between As(III) species and the negative surfaces of SEP-nHAP might be another main reason for the decline in As(III) removal in basic media. [20] Similar types of mechanisms for As(III) adsorption onto clinoptilolite-rich tuffs, [19] mesoporous zirconia nanostructures, [54] cerium modified chitosan, [70] and Mg-Fe-Cl layered double hydroxide [72] have also been reported in literature. Based on the above-mentioned analysis, it is suggested that the As(III) adsorption mechanism on the SEP-nHAP composite was a combination of different processes, such as electrostatic interaction and surface complexation. Figure 9 represents a plausible mechanism of As(III) adsorption onto SEP-nHAP adsorbent. However, an extensive study is required to confirm the proposed mechanism.

Conclusions
In summary, a novel SEP-nHAP nanocomposite was successfully prepared through simple activation of natural calcium-rich SEP with H 3 PO 4 . Relevant characterization demonstrated that the deposition of nHAP onto the SEP surface did not affect the SEP structure, and the obtained material exhibited excellent adsorption abilities toward As(III) compared with the pristine SEP and other reported adsorbents. The adsorption process of As(III) by SEP-nHAP was a function of the pH, contact time, initial As(III) concentration, and temperature. The optimal pH for As(III) adsorption laid in the typical pH range of groundwater (6.0-8.5), which favored the decontamination of groundwater. The adsorption kinetics and isotherms revealed that the pseudo-second-order kinetic model and the Sips isotherm model could well quantify As(III) adsorption on SEP-nHAP. Thermodynamics studies revealed that As(III) adsorption occurred spontaneously and the reaction was endothermic in nature. In addition, the selective adsorption tests exhibited that the as-prepared SEP-nHAP had a remarkable selectivity toward As(III) against other competing anions. Moreover, As(III) loaded on SEP-nHAP could be desorbed with a NaOH solution and the regenerated adsorbent would still provide a high adsorption affinity for aqueous As(III). The present findings highlighted that the modification of natural calcium-rich SEP with H 3 PO 4 activation is an eco-friendly and sustainable approach to produce a new SEP-nHAP nanocomposite adsorbent with enhanced physicochemical characteristics and efficient As(III) removal performance from contaminated groundwater.