Treatment of Aqueous Bromate by Superparamagnetic BiOCl-Mediated Advanced Reduction Process

Bromate (BrO− 3 ) contamination in drinking water is a growing concern. Advanced reduction processes (ARPs) are reportedly promising in relieving this concern. In this work, UV/superparamagnetic BiOCl (BiOCl loaded onto superparamagnetic hydroxyapatite) assisted with small molecule carboxylic acid (formate, citrate, and acetate), a carboxyl anion radical (CO•− 2 )-based ARP, was proposed to eliminate aqueous BrO− 3 . Formate and citrate were found to be ideal CO •− 2 precursor, and the latter was found to be safe for practical use. BrO− 3 (10 μg·L−1, WHO guideline for drinking water) can be completely degraded within 3 min under oxygen-free conditions. In this process, BrO− 3 degradation was realized by the reduction of CO •− 2 (major role) and formyloxyl radical (minor role) in bulk solution. The formation mechanism of radicals and the transformation pathway of BrO− 3 were proposed based on data on electron paramagnetic resonance monitoring, competitive kinetics, and degradation product analysis. The process provided a sustainable decontamination performance (<5% deterioration for 10 cycles) and appeared to be more resistant to common electron acceptors (O2, NO− 3 , and Fe 3+) than hydrated electron based-ARPs. Phosphate based-superparamagnetic hydroxyapatite, used to support BiOCl in this work, was believed to be applicable for resolving the recycling problem of other metal-containing catalyst.


Introduction
As a toxic byproduct of disinfection, bromate (BrO − 3 ) in drinking water can find its sources from stock NaOCl and HOBr solution used as disinfectants [1], bromide (Br − ) oxidation by ozone, chlorine, or oxidative radicals (e.g., hydroxyl radical (HO•), sulfate radical (SO •− 4 )) [2], and contaminant of hypochlorite disinfectant used in water plants [3].Compared with the phase-transferring purification technologies (such as adsorption [4,5], ion exchange [6], and membrane filtration [7]), solving methods based on conventional non-radical-based reduction (medium-pressure UV irradiation [8], electrochemical reduction [9], zero-valent metal reduction [10], and Fe(II)/SI(IV) reduction [11]) or advanced reduction (involves strong reactive reducing radicals (RRRs) such as H/e − aq [12][13][14]) seem to be more thoroughly because of conversion of BrO − 3 to Br − in the latter methods.Advanced reduction processes (ARPs) involve fast degradation kinetics and exhibit a relatively weak sensitivity to dissolved oxygen (DO) compared with conventional non-radical-based reduction processes.As such, ARPs shed light on the efficient and economical removal of BrO 3 − from drinking water.
Besides HO•, SOCAs are other precursors of CO •− 2 .In the photocatalytic system, two mechanisms may underlie the conversion of organic carboxylic acids into organic radicals, such as CO •− 2 .First, SOCAs initially complex with the metal ion harboured by the photocatalyst.Subsequent intramolecular electron transfer from carboxylic acid to the photocatalyst conducted band and photogenerated hole (h + )-mediated oxidation lead to the formation of organic radicals [34].This mechanism considers that the reduction process occurs at the catalyst surface.In this mechanism, the energy of the highest occupied molecular orbital and ionization potential of SOCAs determine the formation potential of the reducing organic radicals.Second, ROSs (such as h + /HO•) generated on the interface of the photocatalyst and then reacted with the free SOCAs.As a result, reducing organic radicals, such as CO •− 2 , are produced [35].To date, only a few works have distinguished these two aspects in the mechanism expounding.In particular, the contributions of the absorbed ROSs on the photocatalyst surface and free ROSs in bulk solution to the formation of reducing organic radicals have not been discussed.
Given the above-mentioned literature review, an UV/BiOX/SOCA process was proposed to degrade BrO − 3 in drinking water.BiOCl, which was loaded onto superparamagnetic HAP (HAP SM ), was selected as a typical of BiOX for experimentation.This work intends to (1) probe the potential of combining photoexcited BiOX and SOCAs for CO •− 2 generation and BrO − 3 removal; (2) clarify the formation mechanism of reductive radical and the main site where BrO − 3 degradation occurs; and (3) evaluate the influence of several common electron acceptors (O 2 , NO − 3 , and Fe 3+ ), which may influence the decontamination efficiency of UV/BiOX/SOCA.

Characterization of the Photocatalyst
Peaks appearing in diffraction patterns for BiOCl-HAP SM (Figure S1) can be indexed to BiOCl (JCPDS No. 73-2060), HAP (JCPDS No. 09-0432), and Fe 3 O 4 (JCPDS No. 85-1436).SEM and TEM images show that a 3D-microflower morphology is formed for BiOCl-HAP SM (Figure 1a) and the HAP SM particles are uniformly embedded into the BiOCl structure (Figure 1b).Experiments were also conducted to directly test the magnetization of the photocatalyst.BiOCl-HAP SM powder (50 mg) was ultrasonically dispersed in 4 mL ultrapure water filled with a glass vial for 5 min before separation (Figure 2a).Once a magnet approached the glass vial, a clear and transparent solution was obtained within 30 s (Figure 2b).This occurrence indicates that the BiOCl-HAP SM powder can be separated and recovered easily from the solution.Figure 3 presents the magnetization (M-H) curves recorded with magnetic fields from −20 kOe to 20 kOe at room temperature.A saturation magnetization (Ms) of 16.8 emu•g −1 was measured for BiOCl-HAP SM .BiOCl-HAP SM attained an extremely narrow magnetic hysteresis loop corresponding to low values in remanence and coercivity.This result suggests the mixture's superparamagnetic properties, which is provided by the Fe 3 O 4 constituent (Figure S1). Figure 4 demonstrates the UV-visible (UV-Vis) diffuse reflectance (DR) absorption spectra of BiOCl-HAP SM .BiOCl-HAP SM exhibited a strong UV absorption.These results clearly show that micron-sized superparamagnetic BiOCl-HAP SM with outstanding UV absorptive ability was successfully prepared through the two-step method.

Characterization of the Photocatalyst
Peaks appearing in diffraction patterns for BiOCl-HAPSM (Figure S1) can be indexed to BiOCl (JCPDS No. 73-2060), HAP (JCPDS No. 09-0432), and Fe3O4 (JCPDS No. 85-1436).SEM and TEM images show that a 3D-microflower morphology is formed for BiOCl-HAPSM (Figure 1a) and the HAPSM particles are uniformly embedded into the BiOCl structure (Figure 1b).Experiments were also conducted to directly test the magnetization of the photocatalyst.BiOCl-HAPSM powder (50 mg) was ultrasonically dispersed in 4 mL ultrapure water filled with a glass vial for 5 min before separation (Figure 2a).Once a magnet approached the glass vial, a clear and transparent solution was obtained within 30 s (Figure 2b).This occurrence indicates that the BiOCl-HAPSM powder can be separated and recovered easily from the solution.Figure 3 presents the magnetization (M-H) curves recorded with magnetic fields from −20 kOe to 20 kOe at room temperature.A saturation magnetization (Ms) of 16.8 emu•g −1 was measured for BiOCl-HAPSM.BiOCl-HAPSM attained an extremely narrow magnetic hysteresis loop corresponding to low values in remanence and coercivity.This result suggests the mixture's superparamagnetic properties, which is provided by the Fe3O4 constituent (Figure S1). Figure 4 demonstrates the UV-visible (UV-Vis) diffuse reflectance (DR) absorption spectra of BiOCl-HAPSM.BiOCl-HAPSM exhibited a strong UV absorption.These results clearly show that micron-sized superparamagnetic BiOCl-HAPSM with outstanding UV absorptive ability was successfully prepared through the two-step method.

Bromate Degradation by UV/BiOCl-HAPSM in the Presence of Formate
The merit of BiOCl-HAPSM is its ability to integrate the magnetic performance of HAPSM and the outstanding photocatalytic capability of BiOCl.Given the affirmation of the mixture's magnetic characteristic (Figures 2 and 3), the photocatalytic efficacy of BiOCl-HAPSM was further tested.Formic acid/formate is a common precursor of reductive radicals in photoreductive processes [34,35].As such, formic acid/formate was selected herein to preliminarily evaluate the degradation efficiency of

Bromate Degradation by UV/BiOCl-HAPSM in the Presence of Formate
The merit of BiOCl-HAPSM is its ability to integrate the magnetic performance of HAPSM and the outstanding photocatalytic capability of BiOCl.Given the affirmation of the mixture's magnetic characteristic (Figures 2 and 3), the photocatalytic efficacy of BiOCl-HAPSM was further tested.Formic acid/formate is a common precursor of reductive radicals in photoreductive processes [34,35].As such, formic acid/formate was selected herein to preliminarily evaluate the degradation efficiency of

Bromate Degradation by UV/BiOCl-HAPSM in the Presence of Formate
The merit of BiOCl-HAPSM is its ability to integrate the magnetic performance of HAPSM and the outstanding photocatalytic capability of BiOCl.Given the affirmation of the mixture's magnetic characteristic (Figures 2 and 3), the photocatalytic efficacy of BiOCl-HAPSM was further tested.Formic acid/formate is a common precursor of reductive radicals in photoreductive processes [34,35].As such, formic acid/formate was selected herein to preliminarily evaluate the degradation efficiency of

Bromate Degradation by UV/BiOCl-HAP SM in the Presence of Formate
The merit of BiOCl-HAP SM is its ability to integrate the magnetic performance of HAP SM and the outstanding photocatalytic capability of BiOCl.Given the affirmation of the mixture's magnetic characteristic (Figures 2 and 3), the photocatalytic efficacy of BiOCl-HAP SM was further tested.Formic acid/formate is a common precursor of reductive radicals in photoreductive processes [34,35].As such, Catalysts 2017, 7, 131 5 of 16 formic acid/formate was selected herein to preliminarily evaluate the degradation efficiency of BrO − 3 in the SOCA-mediated UV/BiOCl-HAP SM process.As shown in Figure 5, 79.0%BrO − 3 was degraded by UV/BiOCl-HAP SM within 15 min of reaction time.Such performance is considered excellent, given that a UV lamp power of 10 W was used and the efficiency was about 35% as the manufacturer claims.That is, BiOCl-HAP SM possesses a good photocatalytic capability.Direct photolysis (UV alone, 16.7%; UV + Formate, 19.3%), adsorption (BiOCl-HAP SM + formate in dark, 1%), or reduction by formate (0.7%) caused slow BrO − 3 degradation (Figure 5).Hence, reductive radicals (e.g., CO •− 2 (−2.0 eV), H• (−2.3 eV), and O •− 2 (−0.28 eV)) and the conducted electron (e − CB ) were speculated to play important roles [23,36].Notably, high initial bromate concentration ([BrO − 3 ] 0 = 6.6 µM (1.0 mg•L −1 )) was used considering the inexpedience caused by the ultrafast degradation for low concentration (100% degradation of 10 µg•L −1 bromate within 3 min reaction time; Figure 6).Such concentration was also adopted to maintain the consistency of [BrO − 3 ] 0 across all experiments, including the kinetics and degradation product investigation.

Mechanism Clarification
To identify the radicals formed, electron paramagnetic resonance (EPR) tests were performed.The EPR data directly supported that only CO • was produced and no H•, O • and HO• was formed (Figure 7  BrO in the SOCA-mediated UV/BiOCl-HAPSM process.As shown in Figure 5, 79.0%BrO was degraded by UV/BiOCl-HAPSM within 15 min of reaction time.Such performance is considered excellent, given that a UV lamp power of 10 W was used and the efficiency was about 35% as the manufacturer claims.That is, BiOCl-HAPSM possesses a good photocatalytic capability.Direct photolysis (UV alone, 16.7%; UV + Formate, 19.3%), adsorption (BiOCl-HAPSM + formate in dark, 1%), or reduction by formate (0.7%) caused slow BrO degradation (Figure 5).Hence, reductive radicals (e.g., CO • (−2.0 eV), H• (−2.3 eV), and O • (−0.28 eV)) and the conducted electron ( e ) were speculated to play important roles [23,36].Notably, high initial bromate concentration ([BrO ]0 = 6.6 μM (1.0 mg•L −1 )) was used considering the inexpedience caused by the ultrafast degradation for low concentration (100% degradation of 10 μg•L −1 bromate within 3 min reaction time; Figure 6).Such concentration was also adopted to maintain the consistency of [ BrO ]0 across all experiments, including the kinetics and degradation product investigation.

Mechanism Clarification
To identify the radicals formed, electron paramagnetic resonance (EPR) tests were performed.The EPR data directly supported that only CO • was produced and no H•, O • and HO• was formed (Figure 7

Mechanism Clarification
To identify the radicals formed, electron paramagnetic resonance (EPR) tests were performed.The EPR data directly supported that only CO •− 2 was produced and no H•, O •− 2 and HO• was formed (Figure 7).DO stripping at the start of experiments rationalizes inability to trap the O  7).Given the above argument, CO •− 2 is the only trapped species in the bulk solution.This observation is consistent with several reported results, which found that CO •− 2 was the only active species during the TiO 2 photocatalytic oxidation of formate [37].Considering that CO •− 2 presents a much stronger reducing power than that of e − CB (−0.25 V, [37]) and a weak BrO ).Thus, a signal from DMPO-H• should be observed if H• is located in bulk solution.However, ESR tests did not detect the H• (Figure 7).Given the above argument, CO • is the only trapped species in the bulk solution.This observation is consistent with several reported results, which found that CO • was the only active species during the TiO2 photocatalytic oxidation of formate [37].Considering that CO • presents a much stronger reducing power than that of e (−0.25 V, [37]) and a weak BrO adsorption onto the surface of BiOCl-HAPSM particles (making the particle surface reduction by e negligible), the CO • in the bulk solution was thus considered as the main contributor of BrO degradation in the UV/BiOCl-HAPSM/formate.Given the fast reaction rate constant of BrO with CO • (  , = 1.37 × 10 9 M −1 •s −1 , Appendix A and Figure S1), the superior degradation of BrO by UV/BiOCl-HAPSM/formate can be reasonably explained.Obviously, two pathways for formate are involved in the degradation process of bromate.In the first pathway, HO• near the catalyst's surface (HO • ) diffuses into bulk solution to form free HO• (HO ) and formate directly reacts with HO to produce CO • (Equations ( 1)-( 3), [38]).This occurrence was proven by the ESR tests (Figure 7).In the second pathway, formate first adsorbs (complexes) onto the BiOCl-HAPSM particle surface (Equation ( 4)).This action is followed by ligand-to-metal charge-transfer (Equation ( 5), [34]) or direct hole transfer reaction (Equation ( 6), [34]).Ultimately, formyloxyl radical ( HCOO • ) is produced.HCOO • can reduce bromate because of its low reduction potential [34].HCOO • also injects an electron to the conduction band of BiOCl-HAPSM to terminate the reaction (Equation ( 7), [39]).Previous studies proposed that the decomposition of HCOO • produces H• and then H• transforms into CO • by reacting with formate (Equations ( 8) and ( 9), [37,40]).However, Perissinotti et al. [39] negated the occurrence of Equations ( 8) and ( 9) on the basis of nontrapping of the H• signal.Similarly, no EPR signal of H• was observed in our work (Figure 7).Thus, the H•-derived formation of CO • can be excluded.Regarding the second pathway, the adsorption (complexation) of formate with BiOCl-HAPSM is a premise step for the final CO • production.Therefore, the concentration evolution of formate in bulk solution was monitored in the BiOCl-HAPSM/formate system with and without competitors (Figure 8).Formate readily adsorbed onto BiOCl-HAPSM and the presence of 4 mM F well inhibited formate's adsorption onto BiOCl-HAPSM (<4% for 20 min) (Figure 8a).Correspondingly, BrO degradation readily decreased by about 11% when the adsorption of formate onto BiOCl-HAPSM was almost inhibited (Figure 8b).Given these Obviously, two pathways for formate are involved in the degradation process of bromate.In the first pathway, HO• near the catalyst's surface (HO• s ) diffuses into bulk solution to form free HO• (HO f ) and formate directly reacts with HO f to produce CO •− 2 (Equations ( 1)-( 3), [38]).This occurrence was proven by the ESR tests (Figure 7).In the second pathway, formate first adsorbs (complexes) onto the BiOCl-HAP SM particle surface (Equation ( 4)).This action is followed by ligand-to-metal charge-transfer (Equation ( 5), [34]) or direct hole transfer reaction (Equation ( 6), [34]).Ultimately, formyloxyl radical (HCOO •− ) is produced.HCOO •− can reduce bromate because of its low reduction potential [34].HCOO •− also injects an electron to the conduction band of BiOCl-HAP SM to terminate the reaction (Equation ( 7), [39]).Previous studies proposed that the decomposition of HCOO •− produces H• and then H• transforms into CO •− 2 by reacting with formate (Equations ( 8) and ( 9), [37,40]).However, Perissinotti et al. [39] negated the occurrence of Equations ( 8) and ( 9) on the basis of nontrapping of the H• signal.Similarly, no EPR signal of H• was observed in our work (Figure 7).Thus, the H•-derived formation of CO •− 2 can be excluded.Regarding the second pathway, the adsorption (complexation) of formate with BiOCl-HAP SM is a premise step for the final CO •− 2 production.Therefore, the concentration evolution of formate in bulk solution was monitored in the BiOCl-HAP SM /formate system with and without competitors (Figure 8).Formate readily adsorbed onto BiOCl-HAP SM and the presence of 4 mM F − well inhibited formate's adsorption onto BiOCl-HAP SM (<4% for 20 min) (Figure 8a).Correspondingly, BrO − 3 degradation readily decreased by about 11% when the adsorption of formate onto BiOCl-HAP SM was almost inhibited (Figure 8b).Given these results, one can conclude that the free formate acts as the source of CO •− 2 through HO•-induced hydrogen abstraction and the adsorbed formate primarily serves as origin of reductive HCOO •− (−2.1 V, [39]) which also contributes to the bromate degradation: BiOCl BiOCl-HAP SM + HCOO − (Formate) → ≡Bi-− OOCH(Complex), ≡Bi-− OOCH electron injecting into the conducted band h Catalysts 2017, 7, 131 7 of 15 hydrogen abstraction and the adsorbed formate primarily serves as origin of reductive HCOO • (−2.1 V, [39]) which also contributes to the bromate degradation: h + +H 2 O→H + +HO • diffusing to bulk solution to form free HO• (HO ) , ( 2) ≡Bi-O OCH HCOO  + BiOCl-HAP , ( 5)

Substitutes of Formate
Formic acid/formate is currently the most efficient hole scavenger for the semiconductor photocatalytic process [37].This property is ascribed to the simple one-carbon molecular structure of formic acid/formate.As such, the formic acid/formate transforms into CO • straightforwardly and involves minimal intermediate products (Equation ( 3)).However, formic acid/formate remaining after reaction will pose a healthy risk and demand post-treatment [23].Green precursors that produce no second pollution are thus needed.Citrate and acetate are reasonable alternatives because they are nontoxic, readily available [41,42] and hold a low potential to form chlorinated disinfection byproducts during chlorination as aliphatic carboxylic acids [43].BiOCl-HAPSM/citrate demonstrated a performance in BrO degradation equivalent to that of UV/BiOCl-HAPSM/formate under the same conditions (Figure 9).By contrast, while UV/BiOCl-HAPSM/acetate presented a much poorer performance than that of UV/BiOCl-HAPSM/formate.This result is analogous to the finding of Marinho et al. [44], who reported citrate significantly promoting the photoreduction of Cr(VI) by TiO2/UVA-vis.

Substitutes of Formate
Formic acid/formate is currently the most efficient hole scavenger for the semiconductor photocatalytic process [37].This property is ascribed to the simple one-carbon molecular structure of formic acid/formate.As such, the formic acid/formate transforms into CO •− 2 straightforwardly and involves minimal intermediate products (Equation ( 3)).However, formic acid/formate remaining after reaction will pose a healthy risk and demand post-treatment [23].Green precursors that produce no second pollution are thus needed.Citrate and acetate are reasonable alternatives because they are nontoxic, readily available [41,42] and hold a low potential to form chlorinated disinfection byproducts during chlorination as aliphatic carboxylic acids [43].BiOCl-HAP SM /citrate demonstrated a performance in BrO − 3 degradation equivalent to that of UV/BiOCl-HAP SM /formate under the same conditions (Figure 9).By contrast, while UV/BiOCl-HAP SM /acetate presented a much poorer performance than that of UV/BiOCl-HAP SM /formate.This result is analogous to the finding of Marinho et al. [44], who reported citrate significantly promoting the photoreduction of Cr(VI) by TiO 2 /UVA-vis.
Catalysts 2017, 7, 131 8 of 15 The enhancement of photocatalytic reduction incited by organic hole scavengers mainly depends on the following three aspects [45]: (I) adsorption capacity of hole scavengers onto the photocatalyst surface; (II) reduction potential of the formed reducing radical ( CO • , HCOO • , CH CH OH , and CH OH ); and (III) accumulation of intermediate products that consume the photogenerated active species (e.g., h + , e , and HO•) but do not produce reducing radicals.The influence of (I) and (II) were certified by the observation that the BrO degradation rates in the presence of the three SOCAs (formate, citrate, and acetate) are correlated with their redox potentials or complex formation constants [46].For (III), suspicion should fall on the difference in chemical structure of the three SOCAs.In the presence of oxygen, acetate can generate CO • through HO•induced dehydrogenation and oxygen addition (Equation ( 10)) and subsequent decarboxylation reaction (Equation ( 11)).Under oxygen-free or anoxic conditions, only a small fraction of acetate can transform into oxalate, and the major products are glyoxylic acid and glycolic acid for the reaction of acetate with HO• [47].Oxalate may form as a result of further oxidation of glyoxylic acid.Oxalate has been reported to transform into formate through the photo-Kolbe mechanism during semiconductor photocatalytic process [48].This observation implies that the intermediate products that consume photogenerated active species (e.g., h + , e , HO•) but do not produce reducing radicals are generated when acetate acts as CO • precursor.These points hence partially explain the relatively weak promotion of acetate.According to the work of Meichtry et al. [49], the direct attack of HO• on the hydroxyl group of citrate will yield CO • and another product (1,3-acetonedicarboxylic acid).The abstraction of the α-hydrogen (−CH − COO ) of citrate also yields glyoxylic acid, which can further oxidize into oxalate ( CO • precursor).These occurrences may explain the approximating enhancement of citrate as formate.Thus, citrate can be used as a harmless substitute for formate:

Reusability of BiOCl-HAPSM
Catalyst reusability is often assessed when considering a catalyst's possible practical application.Thus, the BiOCl-HAPSM powders were recycled in successive tests of bromate degradation (Figure 10).Recovered BiOCl-HAPSM particles showed sustainable high activity in bromate degradation.Almost no obvious change (< 5%) was observed after 10 cycles when the unavoidable loss of BiOCl-HAPSM powders was considered.No significant release of bismuth ion, iron ion, and phosphate ion (PO ) CB , and HO•) but do not produce reducing radicals.The influence of (I) and (II) were certified by the observation that the BrO − 3 degradation rates in the presence of the three SOCAs (formate, citrate, and acetate) are correlated with their redox potentials or complex formation constants [46].For (III), suspicion should fall on the difference in chemical structure of the three SOCAs.In the presence of oxygen, acetate can generate CO •− 2 through HO•-induced dehydrogenation and oxygen addition (Equation ( 10)) and subsequent decarboxylation reaction (Equation ( 11)).Under oxygen-free or anoxic conditions, only a small fraction of acetate can transform into oxalate, and the major products are glyoxylic acid and glycolic acid for the reaction of acetate with HO• [47].Oxalate may form as a result of further oxidation of glyoxylic acid.Oxalate has been reported to transform into formate through the photo-Kolbe mechanism during semiconductor photocatalytic process [48].This observation implies that the intermediate products that consume photogenerated active species (e.g., h + , e − CB , HO•) but do not produce reducing radicals are generated when acetate acts as CO •− 2 precursor.These points hence partially explain the relatively weak promotion of acetate.According to the work of Meichtry et al. [49], the direct attack of HO• on the hydroxyl group of citrate will yield CO •− 2 and another product (1,3-acetonedicarboxylic acid).The abstraction of the α-hydrogen (−CH 2 − COO − ) of citrate also yields glyoxylic acid, which can further oxidize into oxalate (CO •− 2 precursor).These occurrences may explain the approximating enhancement of citrate as formate.Thus, citrate can be used as a harmless substitute for formate: 3+ , residual chlorine (also an electron acceptor) is usually found in the drinking water to depress the microorganism growth.To decrease the influence of residual chlorine, pretreatment by UV photolysis is feasible and convenient: e Catalysts 2017, 7, 131 10 of 15

Synthesis and Characterization of BiOCl-HAP SM
HAP SM was synthesised as follows.First, 0.025 mol FeCl 2 •4H 2 O and 0.05 mol FeCl 3 •6H 2 O were dissolved in 50 mL ultrapure water, and 0.03 mol H 3 PO 4 was diluted with 50 mL ultrapure water.Then, the above-mentioned solution was mixed with continuous stirring at 200 rpm for 10 min.This mixture was stripped with nitrogen for 10 min to solution A. Afterwards, 0.5 mol Ca(OH) 2 was dissolved in 100 mL 70 • C Milli-Q water to produce solution B. Solution A was dosed into solution B with a peristaltic pump at a speed of 1.2 mL•min −1 .The whole process was maintained in continuous stirring (400 rpm) and thermostatic (70 • C) under nitrogen protection.Ammonia solution was used to adjust the solution pH to 12-13.The final reaction slurry was transferred to a 1000 mL beaker, sealed with sealing film and settled for 72 h at 25 • C. The deposits were rinsed repeatedly with Milli-Q water, dehydrated at 70 • C under vacuum, and then ground.Pure HAP crystals were synthesised following the same steps of the synthesis of n-Fe-HA, except replacing the solution A by 100 mL H 3 PO 4 solution (0.3 M).
The obtained HAP SM (0.5 g) was added to 50 mL 0.01 M ethylene glycol solution of Bi(NO 3 ) 3 .The mixture was ultrasonicated for 10min to ensure uniform dispersion.Then, 50 mL 0.01 M ethylene glycol solution of KBr was dosed into the above solution with a peristaltic pump at a flow rate of 0.6 mL•min −1 with mechanically stirring at room temperature.The mixed solution was then autoclaved at 160 • C for 12 h.The formed precipitate was washed repeatedly with copious amounts of ultrapure water and anhydrous ethanol until the conductivity of waste liquid was maintained constant.The precipitate was then dried under vacuum for 12 h at 60 • C. The X-ray powder diffraction pattern of HAP SM powders was tested using a Rigaku Dmax-2000 diffractometer (Rigaku Co., Tokyo, Japan.The morphology and microstructure of the synthesized BiOCl-HAP SM particles were observed by scanning electron microscopy (SEM; FEI Quanta FEG-650) and transmission electron microscopy (TEM; JEOL JEM-1230).X-ray photoelectron spectroscopy (XPS) analysis was conducted using an ESCALAB 250Xi spectrometer.Magnetization measurements were performed using a Lakeshore 7304 vibrating sample magnetometer (Lake Shore Co. Ltd., Columbus, OH, USA) at 25 • C.

Experimental Procedures
A similar photoreactor described in a previous work was used to perform experiments [22].This photoreactor is cylindrical (4 cm internal diameter) and has a total volume of 1.8 L (1.5 L samples).The UV source is a Heraeus low-pressure mercury UV lamp (no ozone generation, light emission at 253.7 nm), which was put in a quartz tube located at the center of the reactor and caused a incident UV intensity of 4.0 mW•cm −2 in the completely mixed sample.Sodium bromate (1 mg•L −1 ) was first dissolved with oxygen-free water preloaded in the reactor.Afterwards, precalculated amount HAP SM -BiOCl (0.5 g•L −1 ) and SOCAs (4 mM) were added.The resultant solution was mixed thoroughly.Oxygen-free water was prepared by helium gas purging (30 min at 1.0 L•min −1 ).In most of the reactions, borate was used to adjust the solution pH.Collected samples were filtered through a 0.22 µm membrane before analysis.All experiments were operated in a thermostatic room (25 • C).The UV lamp was turned on for 20 min until stable prior to experiments.

Figure 2 .
Figure 2. BiOCl-HAP SM dispersion in water (a) without or (b) with a magnet.

Table 1 .
Scavenging of CO • by different electron acceptors.