Facet-dependent photo-Fenton degradation of p-arsanilic acid and arsenic redistribution on hematite

Abstract

Iron minerals play a critical role in regulating arsenic fate through adsorption and photocatalysis. Organoarsenic compounds such as p-arsanilic acid (p-ASA) can be transformed into toxic inorganic arsenic, posing serious environmental risks. Although the surface reactivity of hematite is known to be facet-dependent, the facet-specific photo-degradation mechanisms of organoarsenics, especially the speciation transformation and fate of arsenic group remain unclear. In this study, the photo-Fenton catalytic degradation of p-ASA by H₂O₂ was systematically investigated using hematite nanoplatelets (HNPs, dominated by {001} facets) and nanocubes (HNCs, dominated by {012} facets) as catalyst. HNCs exhibited significantly faster and more complete degradation than HNPs, with a higher pseudo-first-order rate constant ((5.29 ± 0.79) × 10⁻¹vs. (7.62 ± 0.53) × 10⁻²). Notably, HNCs also promoted greater dearsenification, releasing more inorganic arsenic into solution (27.3% vs. 6.9%), indicating a higher potential environmental risk. In situ shell-isolated nanoparticle-enhanced Raman spectroscopy (SHINERS) revealed that molecular p-ASA persisted on HNP after reaction, whereas HNC surfaces were dominated by degradation products. Mechanistic studies identified hydroxyl radicals as the dominant reactive species, with higher concentrations generated in the HNC system. Its superior activity stemmed from a higher concentration of surface-labile Fe species and stronger co-adsorption of H₂O₂ (2 mmol L⁻¹) and p-ASA, which enhanced radical generation. This work demonstrates that the photocatalytic function of iron minerals is intrinsically facet-dependent, governing both pollutant transformation and arsenic speciation. These findings highlight the importance of facet analysis in understanding the environmental fate of organoarsenicals and suggest that facet engineering offers a strategic pathway for designing targeted remediation technologies.

---

Environmental significance

The environmental fate and risk of organoarsenic pollutants are shown to depend fundamentally on the exposed facets of ubiquitous iron oxides. Our study reveals a critical, and often overlooked, trade-off: the more photocatalytically active (012) facet of hematite nanocubes degrades contaminants more rapidly, yet it also mobilizes greater amounts of toxic inorganic arsenic into the solution. These insights establish a facet-dependent perspective that is crucial for accurately predicting arsenic speciation in dynamic environments and for designing targeted remediation strategies based on engineered mineral surfaces.

Introduction

Aromatic organoarsenic compounds, such as p-arsanilic acid (p-ASA) and roxarsone, have been extensively used as feed additives to promote livestock growth and control intestinal diseases.^1,2 However, over 90% of these compounds are excreted unchanged into the environment via manure,³ leading to widespread contamination of soils and water bodies. Reported concentrations reach up to 100 mg kg⁻¹ in soils and 5000 μg L⁻¹ in freshwater system in some regions.^4,5 More critically, these initially less toxic organoarsenic species can be transformed into highly mobile and toxic inorganic arsenic through mineral- or light-mediated processes in the environment.⁶ Despite their environmental prevalence and risk, the transformation pathways and behavior of organoarsenic compounds in soils and aquatic systems remain poorly understood.

Iron minerals are key reactive constituents in sediments and groundwater, possessing high specific surface areas and surfaces rich in hydroxyl groups.⁷ Beyond adsorption, they can drive redox reactions that govern the transformation of pollutants including organoarsenic. Representative iron oxides, like ferrihydrite, goethite, and hematite, exhibit strong affinities for organoarsenic species.^8–10 Hematite, a common crystalline iron oxide in soils, acts as both an adsorbent and a semiconductor capable of initiating photocatalytic reactions under light, generating reactive oxygen species (ROS) such as hydroxyl radicals.^11–14 Such photocatalytic processes significantly influence pollutant degradation and microbial activity¹⁵ and have been demonstrated in the transformation of dyes such as rhodamine B and methyl orange.^11,16,17 It is therefore plausible that under light exposure, hematite may not only adsorb organoarsenic compounds but also catalytically transform them, altering their environmental fate. Since organoarsenic degradation can yield more toxic inorganic arsenic,¹⁸ understanding hematite-mediated phototransformation-particularly the speciation and mobility of arsenic-is critical, yet remains underexplored.

The surface reactivity of hematite is highly facet-dependent, arising from variations in iron coordination and surface chemistry across different crystal planes.^19–21 For instance, Pb(II) adsorption differs markedly among facets: (001) and (1–12) form distinct inner-sphere complexes,²² while the (001), (012), and (110) facets exhibit varying adsorption capacities due to differences in site availability.^23,24 This facet dependence extends to photocatalytic performance: hematite dominated by (001) facets outperforms (110)-faceted hematite in degrading metronidazole and tetracycline under light,²⁵ and the (012) facet shows a higher photodegradation rate for rhodamine B than the (001) facet.¹⁵ Such differences are attributed not merely to site density but to the mismatch among surface cation sites. Facet-dependent behavior is also evident in photoelectrochemical water splitting, where hematite facets display distinct oxygen evolution reaction activities.^26–28 Although the catalytic reaction mechanisms on different crystal facets of hematite have been extensively investigated-for instance, the markedly distinct catalytic hydrolysis efficiencies of various hematite facets toward phthalic acid esters (PAEs), which exemplify facet-dependent reactivity²⁹-the ultimate environmental fate of arsenic-bearing compounds on these specific facets remains scarcely explored. This is particularly true regarding their influence on the transformation of heavy metal species and their interfacial partitioning behavior in environmental matrices. This knowledge gap significantly constrains our comprehensive understanding of the practical role that iron (oxyhydr) oxides play in regulating the environmental behavior of heavy metals such as arsenic. Previous studies have established that the markedly different reactivities of hematite facets originate from their distinct surface atomic structures.³⁰ Among the various crystallographic planes, the {001} and {012} facets are the most prevalent and commonly exposed surfaces of hematite. These two facets exhibit fundamentally different physicochemical properties; for instance, the {001} facet is characterized by a relatively low point of zero charge (PZC), whereas the {012} facet possesses a higher PZC, along with a high density of active sites and a strong affinity for a wide range of pollutants.³¹ Consequently, a systematic comparison of these two representative facets is sufficient to elucidate the divergent coordination modes of surface Fe atoms and their subsequent impact on catalytic performance. Although other surfaces, such as the {104} facet, are also present in hematite crystals, they were excluded from this study. This decision was made because these facets are both less representative of the primary exposed surfaces and synthetically more challenging to control, making a focused investigation on the {001} and {012} facets a more effective and rational approach.

Although our preliminary work using in situ surface-enhanced Raman spectroscopy has revealed facet-dependent adsorption of organoarsenic on hematite,³² the photocatalytic degradation role of specific hematite facets-beyond adsorption-has not been systematically examined. In particular, the light-driven interfacial transformation of arsenic-bearing functional groups in organoarsenic compounds remains unclear. To address this gap, we investigate the photocatalytic degradation of p-ASA on two hematite morphologies with dominantly exposed facets: nanoplatelets (HNPs, dominated by {001} facet) and nanocubes (HNCs, dominated by the {012} facet). Beyond degradation kinetics, we elucidate the evolution of iron species, ROS generation, and-critically-the transformation and migration of arsenic species. This work provides a facet-resolved understanding of the photochemical behavior of aromatic organoarsenicals on hematite surfaces, offering new insights for assessing and controlling the environmental fate of these emerging contaminants.

Experimental

Chemicals

p-ASA (98%) was purchased from Macklin Chemicals (China). Ferric chloride hexahydrate (FeCl₃·6H₂O), hydrogen peroxide (H₂O₂), sodium chloride (NaCl), sodium bicarbonate (NaHCO₃), ethanol (C₂H₅OH), nitric acid (HNO₃), sodium hydroxide (NaOH), hydrochloric acid (HCl), sodium acetate (CH₃COONa), oleic acid (C₁₈H₃₄O₂), sodium oleate (C₁₇H₃₃COONa), potassium chloride (KCl), formic acid (HCOOH), thiourea, ascorbic acid and benzoic acid were supplied by Sinopharm Chemical Reagent (China). Methanol was supplied by Thermo Fisher Scientific Inc (USA).

Material characterization

The morphology of hematite was characterized by scanning electron microscopy (SEM, SU8010, Hitachi, Japan) and transmission electron microscopy (TEM, Talos F200XG2, Thermo Fisher Scientific, USA). The crystal structure of the hematite was analyzed by X-ray diffractometer (XRD, D8 ADVANCE 25, Bruker AXS, Germany). X-ray photoelectron spectra were received with X-ray photoelectron spectrometer (XPS, K-ALPHA, Thermo Scientific, USA). The concentration of p-ASA was quantified by high-performance liquid chromatography (HPLC, 1260, Agilent, USA) with the following conditions: methanol and 1% acetic acid solution (20 : 80, v/v) were used as mobile phase with a flow rate of 1 mL min⁻¹, sample injection volume 20 μL, and detection wavelength was set at 254 nm. The organic intermediate products were qualitatively determined using HPLC coupled with quadrupole-orbitrap mass spectrometry (HPLC-MS, Q-Exactive, Thermo Fisher Scientific, USA). The concentration of dissolved iron was quantified using inductively coupled plasma optical emission spectrometer (ICP-OES, EXPEC 6000, Focused Photonics Inc., China). The amount of generated inorganic arsenic in solution was determined by atomic fluorescence spectrometer (AFS, AF-3300, Beijing Beifen-Ruili Analytical Instrument, China). Raman spectra were collected with a confocal Raman microscope (Alpha 300-R, WITec, Germany). Electron paramagnetic resonance (EPR, EPR200M, CIQTEK Co., Ltd, China) spectroscopy was employed to detect radical species.

HNPs & HNCs synthesis

HNPs and HNCs were synthesized using solvothermal method referring to reported work.³³ Briefly, HNPs were prepared by dissolving 1.09 g FeCl₃·6H₂O in a mixture of ethanol (40 mL) and ultrapure water (2.8 mL), followed by addition of 3.2 g sodium acetate and stirring for 2 h at room temperature. The solution was then subjected to solvothermal treatment at 180 °C for 12 h. The resulting product was collected by centrifugation, washed with water and ethanol, and dried at 60 °C under vacuum to obtain brownish-red HNPs. HNCs were prepared by dissolving 1.28 g FeCl₃·6H₃O and 4.26 g sodium oleate in a solvent mixture containing ethanol (21.5 mL), oleic acid (2.6 mL), and water (18.5 mL), followed by solvothermal reaction at 180 °C for 12 h. The precipitate was washed with ethanol and vacuum-dried at 60 °C. Finally, the powder was annealed at 300 °C for 2 h.

Adsorption and degradation experiment

Hematite was suspended in water with a dosage of 1 g L⁻¹. Certain amount of p-ASA stock solution was added to the hematite dispersion to obtain the solution with final concentration of 10 mg L⁻¹. The mixture was stirred for 120 min in the dark for adsorption prior to photocatalytic tests. The photocatalytic reaction was initiated by turning on the light (PLS-SXE300D, 50 W, 320–780 nm, Beijing Perfectlight Technology Co., Ltd, China) and adding H₂O₂ (final concentration: 2 mmol L⁻¹). Samples were collected at 10, 30, 60, 120, and 240 min, centrifuged at 10 000 rpm for 5 min. The supernatant was analyzed by HPLC to determine the residual concentration of p-ASA. Part of the solution after reaction for 240 min was analyzed with HPLC-MS to detect the organic intermediates in the dispersion. The mass spectrometry conditions were set as follows: the ion source was a heated electrospray ionization (HESI) source operating in positive full MS-ddMS2 mode, with a scanning range of m/z 50–750. The spray voltage was set to 3.5 kV, the S-lens RF level was 50%, and the capillary temperature was maintained at 300 °C. The automatic gain control (AGC) target was set at 1 × 10⁶, the maximum injection time was 100 ms, and the mass resolution was 70 000. All other instrument parameters were maintained at their default settings.

Determination of dissolved iron and arsenic species

Both dissolved iron in the solution and surface-adsorbed iron (Fe_labile) were analyzed at different reaction time intervals. The dissolved iron content was determined by analyzing the supernatant obtained after centrifuging the reaction solution at each time point using ICP-OES. Meanwhile, the Fe_labile on hematite was extracted with 0.4 mol L⁻¹ HCl from the solid samples collected at corresponding time points and subsequently quantified by ICP-OES referring to previous work.³⁴ Solid-phase digestion experiments were conducted on hematite samples to calculate the ratio of the dissolved and active iron to the total iron in hematite. The dried hematite solid was treated with concentrated HCl (6 mol L⁻¹) under heating and the iron concentrations in the extracts were determined by ICP-OES. During reaction, the amount of arsenic released into the solution was assessed by measuring the concentration of dissolved arsenic in the supernatant using AFS. For the total inorganic arsenic and As(III) concentration determination, the sample was mixed with/without thiourea and ascorbic acid before determination. While the surface-adsorbed inorganic arsenic was determination with similar procedure after the hematite was extracted with HCl (6 mol L⁻¹).³⁴ The surface-adsorbed arsenic species were characterized using surface-enhanced Raman spectroscopy (SERS) on the separated hematite particles. Specifically, the hematite samples after centrifugation were transferred onto clean glass slides, followed by the addition of SiO₂@Ag nanoparticles as the SERS substrate. Raman measurements were performed using a 532 nm laser source with a power of 10 mW, an integration time of 1000 ms, and 10 accumulations per spectrum. The synthesis and characterization of the SiO₂@Ag enhancement substrate were conducted according to procedures reported in our previous work.³⁵

Free radical determination

The generation of reactive radicals during the photocatalytic degradation of p-ASA was investigated through quenching experiments and EPR analysis. In the quenching tests, isopropanol (IPA), catalase (CAT), and superoxide dismutase (SOD) were employed as quenching agents for ˙OH, H₂O₂, and O₂˙⁻, respectively, at concentrations of 25 mmol L⁻¹, 200 mg L⁻¹, and 20 mg L⁻¹. For EPR measurements, 30 μL of the sample was mixed with 30 μL of DMPO (100 mmol L⁻¹, dissolved in deionized water/methanol solvent). A capillary tube was used to transfer an aliquot of the mixture into a quartz tube, which was then placed into the EPR resonator cavity. The measurements were conducted to detect DMPO-adduct signals of conventional radical species, including ˙OH and O₂˙⁻. The production of ˙OH in the system was quantitatively determined using benzoic acid as a probe. Benzoic acid (10 mg L⁻¹) was added into the reaction by replacing p-ASA. The concentration of the product after reaction with ·OH (p-hydroxybenzoic acid, p-HBA) is quantitatively determined by HPLC, and the cumulative yield of ·OH is calculated based on the amount of p-HBA.³⁶

DFT calculation

The adsorption configuration and energy of the adsorption of H₂O₂ on hematite were simulated using density functional theory (DFT) as implemented in the vienna ab initio simulation package (VASP). The electron exchange–correlation interactions were described within the generalized gradient approximation (GGA) using the Perdew–Burke–Ernzerhof (PBE) functional. The projector augmented-wave (PAW) method was employed to treat the interactions between ion cores and valence electrons.

The adsorption energy (E_ads) was calculated as follows:

E_ads = E_system − E_hematite − E_H2O2

where E_system, E_hematite, and E_H2O2 represent the total energies of the optimized adsorption system, the pristine hematite slab, and the isolated H₂O₂ molecule, respectively.

Results and discussion

Characterization of HNP and HNC

The X-ray diffraction (XRD) patterns confirm the phase purity of the as-synthesized hematite nanoplate (HNP) and nanocube (HNC), with no detectable impurity phases (Fig. S1 in SI). The distinct relative intensities of diffraction peaks at 2θ = 33.2° and 33.7° further indicate different dominantly exposed facets in the two morphologies. Electron microscopy reveals well-defined hexagonal platelets (HNPs) with an average diameter of ∼200 nm, and faceted cubic particles (HNCs) averaging ∼500 nm in size (Fig. 1) high-resolution TEM (HRTEM) of HNPs shows lattice fringes with an interplanar spacing of 0.25 nm and an inter-facet angle of 60°, consistent with the (001) facet of hematite.³⁷ The corresponding selected area electron diffraction SAED pattern confirms the dominant exposure of the (001) facet. For HNCs, HRTEM displays lattice spacings of 0.37 nm and an inter-facet angle of 90°, matching the structural signature of the (012) facet.³⁷ Statistical analysis based on geometric modeling reveals that the surface area fraction of {001} facets in HNPs is approximately 83.2%, with the remaining 16.8% attributed to {012} side facets, while HNCs are exclusively enclosed by {012} facets as detailed in the SI. These results collectively demonstrate the successful synthesis of two distinct hematite morphologies-HNPs and HNCs-with dominantly exposed (001) and (012) facets, respectively. This provides a well-defined material foundation for systematically investigating the facet-dependent photodegradation behavior of p-ASA on hematite surfaces.

Fig. 1 Morphology and crystal structure of the synthesized hematite nanocrystals. (a) SEM image, (b) TEM image, (c) high-resolution TEM image, and (d) corresponding SAED pattern of HNPs. (e) SEM, (f) TEM, (g) high-resolution TEM image, and (h) SAED pattern of HNCs.

Photodegradation kinetics of p-ASA on HNP and HNC

Both HNC and HNP exhibit certain adsorption capacities for p-ASA as reported in our previous work.³⁸ To better compare their photocatalytic degradation performance, a 2-hour dark adsorption period was implemented prior to light irradiation. As shown in Fig. S2, HNP and HNC exhibited distinct adsorption capacities toward p-ASA. Since this study focuses on the photocatalytic degradation process, all kinetic data presented hereafter were normalized using the p-ASA concentration at adsorption equilibrium as the initial concentration (c₀). The photocatalytic degradation reaction was then initiated by adding H₂O₂ and simultaneously turning on the light source. The degradation profiles of p-ASA on HNP and HNC are shown in Fig. 2a. In the absence of hematite, H₂O₂ induced only ∼10.1% degradation after 4 hours. By contrast, the addition of hematite-either alone or with H₂O₂-markedly enhanced the degradation, underscoring the essential catalytic role of hematite. When hematite alone, the 4 h degradation efficiencies were 37.7% for HNC and 32.1% for HNP, respectively, indicating a discernible facet-dependent activity. This difference became more pronounced in the presence of H₂O₂: in the HNC + H₂O₂ system, p-ASA declined to ∼10.2% of its initial concentration, whereas 26.5% of p-ASA remained in the HNP-mediated system, confirming the superior catalytic performance of HNC. These results clearly demonstrated the oxidative degradation capacity of the hematite-mediated photocatalytic system is facet-dependent, with HNC exhibiting significantly higher activity toward p-ASA degradation than HNP.

Fig. 2 Photocatalytic degradation kinetics of p-ASA on HNP and HNC. (a) Concentration profiles of p-ASA during photocatalytic degradation. (b) Langmuir–Hinshelwood pseudo-first-order kinetic fitting for the photodegradation data. (c–f) Dependence of the initial reaction rate constant on (c) hematite dosage, (d) p-ASA concentration, (e) H⁺ concentration, and (f) H₂O₂ concentration (pH = 7).

The photodegradation kinetics of p-ASA were well described by a pseudo-first-order Langmuir–Hinshelwood (L–H) model (Fig. 2b). The fitted rate constants were 0.78 h⁻¹ for HNC and 0.52 h⁻¹ for HNP in the presence of H₂O₂. Without H₂O₂, the rate constants were substantially lower (0.10 h⁻¹ for HNP and 0.17 h⁻¹ for HNC). Increasing the H₂O₂ concentration enhanced the p-ASA degradation rate (Fig. S3a). At 2 mmol L⁻¹ H₂O₂, the degradation rate increased by approximately 5.2-fold for HNP and 4.6-fold for HNC. Acidic conditions further accelerated the reaction (Fig. S3b). At pH 3, the rate constants reached 1.2 h⁻¹ for HNC and 0.51 h⁻¹ for HNP (Fig. S3c and d), indicating not only a pronounced pH enhancement but also a more distinct facet-dependent difference under acidic media.

The overall degradation reaction can be empirically simplified as:

a p-ASA + b hematite + c H⁺ + d H₂O₂ + hv → degradation products (1)

Accordingly, the initial degradation rate of p-ASA can be described by:

r_int = −d[p-ASA]/dt = k_init[p-ASA]^a[hematite]^b[H⁺]^c[H₂O₂]^d (2)

The initial degradation rate (r_initial) was obtained from the early-stage degradation data (within the first 20 min under the experimental conditions). Experiments were performed by varying hematite loading, p-ASA concentration, pH, and H₂O₂ concentration. As shown in Fig. 2c, the slope of the linear fit after logarithmic transformation of eqn (2)-with only the catalyst concentration varied-yields the reaction order b with respect to hematite. This value reflects the facet-specific influence on the photocatalytic activity. The obtained orders were b_HNP = 0.16 and b_HNC = 0.40, indicating that hematite exposing the (012) facet (HNC) exerts a stronger influence on the reaction kinetics than the (001)-faceted HNP. The reaction order with respect to p-ASA was slightly higher for HNC (a_HNC = 0.67) than for HNP (a_HNP = 0.59) (Fig. 2d). In contrast, the contributions of H⁺ were relatively small (c_HNP = 0.07, c_HNC = 0.05) (Fig. 2e), suggesting that although acidic conditions promote degradation, protons play a limited direct role in the rate-determining step. The order with respect to H₂O₂ was also higher for HNC (d_HNC = 0.26) than for HNP (d_HNP = 0.16) (Fig. 2f), implying more efficient utilization of H₂O₂ on the HNC surface-a point that will be further examined in subsequent sections.

Based on these results, the empirical rate equations for p-ASA photodegradation under simulated solar irradiation can be written as follows:

HNP: r_HNCinit = −d[p-ASA]/dt = k_init[p-ASA]^0.67[HNP]^0.40[H⁺]^0.05[H₂O₂]^0.26 (3)

HNC: r_HNPinit = −d[p-ASA]/dt = k_init[p-ASA]^0.59[HNP]^0.16[H⁺]^0.07[H₂O₂]^0.16 (4)

The apparent rate constants (k_init) derived from eqn (3) and (4) using the measured r_init values under various conditions are (7.62 ± 0.53) × 10⁻² for HNP and (5.29 ± 0.79) × 10⁻¹ for HNC. This indicates that the photocatalytic degradation proceeds faster on HNCs exposing the (012) facet than on HNPs exposing the (001) facet. It should be noted that both r_init and k_init are empirical parameters; while they effectively describe the overall initial degradation rate under the investigated conditions, they do not resolve the individual contributions of elementary reaction steps.

Arsenic speciation transformation during photocatalytic reaction

In the environmental transformation of organoarsenic compounds such as p-ASA, light- or heat-driven cleavage of the arsenic moiety can release inorganic arsenic, whose toxicity often far exceeds that of the parent compound.^39,40 Therefore, beyond tracking the disappearance of p-ASA, detailed analysis of arsenic speciation is essential for accurate environmental risk assessment. In this work, we employed a multi-method approach to monitor arsenic transformation and distribution during photocatalysis (as shown in Fig. 3a). In addition to HPLC quantification of residual p-ASA, inorganic arsenic species in solution were analyzed kinetically by atomic fluorescence spectrometry (AFS). Surface-adsorbed arsenic was determined after chemical extraction, and its oxidation state was probed by X-ray photoelectron spectroscopy (XPS). Furthermore, using our previously developed shell-isolated nanoparticle-enhanced Raman spectroscopy (SHINERS) technique,³⁸ we performed in situ analysis of arsenic-containing species on the hematite surface. The combined data provide a comprehensive, real-time picture of arsenic speciation and migration (Fig. 3b–g).

Fig. 3 Transformation and distribution of arsenic species during photocatalytic degradation of p-ASA on HNP and HNC. (a) Schematic illustration of the multi-method analysis used to track arsenic speciation. Temporal variations in the concentrations of dissolved p-ASA, dissolved inorganic arsenic, and surface-adsorbed inorganic arsenic for (b) HNP and (c) HNC. High-resolution As 3d XPS spectra of (d) HNP and (e) HNC before and after the reaction. In situ SHINERS spectra collected on (f) HNP and (g) HNC after 0, 30, 60, 120, and 240 min of photocatalytic reaction.

Prior to illumination, 25.8% and 32.4% of p-ASA adsorbed onto HNP and HNC surfaces, respectively, consistent with the stronger adsorption capacity of HNC³⁸ (Fig. 3b and c). To track arsenic fate throughout the reaction, the total arsenic was partitioned into four fractions: dissolved p-ASA, adsorbed p-ASA, dissolved inorganic arsenic (As(III) and As(V)), and adsorbed inorganic arsenic. As the photocatalytic reaction proceeded, p-ASA concentration declined due to oxidative degradation, while partial regeneration of surface sites allowed re-adsorption from solution. Concomitantly, cleavage of the arsenic group released inorganic arsenic into both solution and adsorbed phases. After 4 h, the dissolved inorganic arsenic accounted for only 6.9% of total arsenic in the HNP system, but reached 27.3% in the HNC system. Speciation analysis (Fig. S4) revealed distinct pathways: for HNP, trace As(III) (15 μg L⁻¹) appeared early, while As(V) rose to a maximum of 67 μg L⁻¹ before dropping to ∼15 μg L⁻¹, suggesting re-adsorption of generated inorganic arsenic. In contrast, for HNC, As(III) reached 48 μg L⁻¹ initially and then decreased, becoming undetectable at 4 h, whereas As(V) accumulated progressively to 295 μg L⁻¹. This indicates rapid initial degradation of p-ASA to As(III), followed by oxidation to As(V) that remained largely in solution. The adsorbed inorganic-arsenic fraction accounted for 43.0% of total arsenic with HNP and 52.9% with HNC. Relative to the overall p-ASA removal (79.4% for HNP, 92.7% for HNC), adsorption contributed >54.2% and 57.1%, respectively (Fig. S5). Thus, although HNC exhibited higher degradation efficiency, it also released more inorganic arsenic into solution, representing a greater potential risk of inorganic arsenic pollution.

XPS analysis of the hematite surfaces before and after reaction (Fig. 3d and e) showed that the As 3d signal at 44.7 eV, assigned to organo-arsenic in p-ASA, decreased significantly after degradation. Deconvolution revealed a new component at 45.7 eV, attributable to adsorbed inorganic arsenic, consistent with the extraction data. Arsenic mass-balance calculations indicated that after 4 h, about 29.6% of arsenic in the HNP system was neither in solution nor as adsorbed inorganic species; this fraction is attributed to residual molecular p-ASA on the surface. For HNC, this residual portion was only 12.6%.

To probe the interfacial species in situ, SHINERS spectra were collected at different reaction stages (Fig. 3f and g). Initially, the Raman signals of adsorbed p-ASA differed between HNP and HNC, reflecting facet-dependent adsorption configurations³⁸ On HNP, On HNP, the characteristic p-ASA vibrations—including the As–C stretch at ∼1085 cm⁻¹, the ring breathing mode at ∼1000 cm⁻¹, and the As–O bands in the 780–820 cm⁻¹ region—weakened within the first 30 min but then stabilized, with some features persisting even after 240 min. A band near 600 cm⁻¹ even intensified over time, possibly due to As–C stretching from accumulating intermediates.⁴¹ This suggests that besides adsorbed p-ASA, degradation products (including As-containing species) accumulated on the HNP surface. On HNC, spectral changes were more pronounced. Early in the reaction (0–30 min), the SHINERS profile evolved, indicating a change in adsorption configuration. As the reaction proceeded, characteristic p-ASA peaks became barely discernible after 60 min, and a broad, featureless background emerged in the 1200–1700 cm⁻¹ region, consistent with the accumulation of organic degradation products and minimal residual p-ASA. This aligns with the lower surface-retained p-ASA inferred from mass balance. The more complete disappearance of p-ASA features on HNC corroborates its superior degradation efficiency and greater release of inorganic arsenic into solution, highlighting the critical role of facet-specific surface reactivity in governing the interfacial degradation pathway. HPLC-MS analysis of the solution confirmed the formation of p-aminophenol (retention time 1.63 min; Fig. S6), an intermediate generated by cleavage of the arsenic group from p-ASA. This supports the lability of the As–C bond and the formation of aromatic intermediates, part of which adsorb on hematite.

In summary, photocatalytic degradation of p-ASA is more complete on HNC, leading to higher release of dissolved inorganic arsenic. On HNP, degradation is less extensive, leaving more p-ASA in molecular form on the surface and resulting in lower inorganic arsenic mobilization. These facet-dependent pathways highlight the importance of interfacial speciation in determining the environmental fate and risk of organoarsenic compounds.

Interfacial iron behavior during the degradation process

To elucidate the facet-dependent interfacial mechanisms of p-ASA degradation, the behavior of iron species was monitored alongside arsenic transformation. In iron oxide systems, surface coordinatively unsaturated (labile) iron (Fe_labile) is recognized as a key active site for activating H₂O₂ and generating hydroxyl radicals (˙OH).^34,42,43 Therefore, we quantified both dissolved iron in solution and the solid-phase Fe_labile concentration during the reaction. As shown in Fig. 4a, the concentration of dissolved iron remained consistently low (< 1 mg L⁻¹) throughout the reaction, indicating negligible contribution from homogeneous iron species. Post-reaction XRD and TEM characterization (Fig. 4b–d) confirmed that both hematite morphologies retained their crystallinity and surface structure, consistent with the minimal iron dissolution. In contrast, the concentration of Fe_labile showed distinct trends between the two facets. For HNP, the Fe_labile content decreased from 0.75% to ∼0.5% of total iron during the reaction. The initial higher level likely reflects surface activation by H₂O₂ and light. This labile iron not only participates in radical generation but may also facilitate the immobilization of released inorganic arsenic. In the HNC system, Fe_labile levels remained higher throughout (0.8–1.2% of total iron). The greater abundance of surface-active iron on HNC promotes more efficient H₂O₂ activation, leading to higher radical yields and consequently faster and more complete degradation of p-ASA.

Fig. 4 Evolution of iron species during the photocatalytic degradation. (a) Temporal changes in the concentrations of dissolved iron and surface-adsorbed labile iron (Fe_labile). (b) XRD patterns of HNP and HNC after the reaction. TEM images of (c) HNCs and (d) HNPs collected after 4 h of reaction. High-resolution Fe 2p XPS spectra of (e) HNP and (f) HNC after reaction.

XPS analysis of the hematite surfaces before and after reaction further revealed changes in iron oxidation state (Fig. 4e and f). Prior to the reaction, the Fe 2p spectra of both samples were dominated by peaks at 710.4 eV (Fe 2p_3/2) and 724.0 eV (Fe 2p_1/2), characteristic of Fe(III), along with their satellite features at 718.6 eV and 732.6 eV.⁴⁰ After photocatalytic reaction, while the overall peak position remained unchanged, indicating that iron predominantly existed in the trivalent state (Fe(III)), the subtle variations observed at 709 eV and 724 eV suggest minor changes in the oxidation state of Fe during the reaction, implying the formation of Fe(II) on the surface.⁴⁴ This confirms that a surface Fe(III)/Fe(II) redox cycle operates under light, which is crucial for sustained radical generation. The slightly higher Fe(II) signal on HNC correlates well with its superior degradation performance, underscoring the link between surface iron redox activity and catalytic efficiency. To directly compare the radical generation capacity of the two facets, ESR spectroscopy and quenching experiments were conducted, as discussed in the following section.

Free radical analysis

Quenching experiments were conducted to identify the reactive oxygen species (ROS) responsible for the p-ASA photodegradation. The effect of IPA (˙OH quencher), CAT (H₂O₂ quencher), and SOD (O₂˙⁻ quencher) on the degradation mediated by HNP and HNC are shown in (Fig. 5a and b), with the corresponding inhibition data summarized in Table 1. Degradation by HNC was more strongly inhibited than that by HNP for all quenchers. For example, in the presence of 25 mmol L⁻¹ IPA, the initial degradation rate (r_init) for HNC dropped from 61.1 to 39.4 μM h⁻¹, whereas for HNP, it decreased from 47.6 to 38.0 μM h⁻¹. Similar trends were observed with CAT and SOD. These results indicate that hematite exposing the (012) facet (HNC) generates a higher concentration of ROS under light. The more pronounced suppression by IPA relative to SOD highlights the dominant role of hydroxyl radicals (˙OH), produced via H₂O₂ decomposition, in driving the degradation. ˙OH reacts rapidly with organic molecules via hydrogen abstraction, initiating efficient chain-like degradation⁴⁵ in contrast, O₂˙⁻ possesses weaker oxidative capacity and readily undergoes side reactions, leading to a lesser contribution. Notably, even with excess quencher, degradation was not completely halted. This suggests that in addition to solution-phase radicals, surface-bound radicals likely participate in the reaction.

Fig. 5 Role of reactive oxygen species in the photodegradation of p-ASA. Photodegradation kinetics of p-ASA on (a) HNCs and (b) HNPs with presence of different ROS quenchers. (a and b) Degradation kinetics on (a) HNC and (b) HNP in the presence of different ROS quenchers. (c) EPR spectra of DMPO-trapped radicals in the reaction systems. (d) Cumulative ˙OH yield quantified using benzoic acid as a probe.

Table 1 Photocatalytic degradation efficiency of p-ASA with different scavengers

	Scavenger	r init × 10^−6	Inhibition efficiency (%)
HNP	Without scavenger	47.6	—
	20 mg L^−1 SOD for O2˙^−	39.9	15.9
	25 mmol L^−1 IPA for ˙OH	38.0	20.3
	CAT + SOD	29.6	37.7
HNC	Without scavenger	61.1	—
	20 mg L^−1 SOD for O2˙^−	45.1	27.0
	25 mmol L^−1 IPA for ˙OH	39.4	35.9
	CAT + SOD	25.5	69.7

---

The EPR spectrum using DMPO as a spin trap confirmed the generation of ˙OH, showing the characteristic 1 : 2 : 2 : 1 quartet signal (Fig. 5c). No discernible signal for O₂˙⁻ was detected, verifying ˙OH as the primary radical species. Quantification of ˙OH using benzoic acid as a probe (Fig. 5d) showed that the HNC system produced a greater amount of ˙OH within the first 4 h, consistent with its faster degradation kinetics. At later stages, both systems generated comparable levels of ˙OH, indicating that although HNPs initially produce radicals more slowly, the sustained presence of H₂O₂ allows continued radical generation over time.

DFT simulation and reaction mechanism

To elucidate the facet-dependent generation of active radicals, we analyzed the surface oxygen species of hematite via O 1s XPS and performed DFT simulations to examine H₂O₂ adsorption. The deconvoluted O 1s spectra (Fig. 6a and b) show the components of lattice oxygen (O_latt), surface oxygen (O_suf), and adsorbed oxygen species (O_ads) before and after the photoreaction.^38,44 Although the O_ads component was initially low for both samples, it increased significantly after reaction due to H₂O₂ adsorption and activation. The post-reaction proportion of ROS was 28.3% for HNP and 36.5% for HNC, confirming the superior capacity of HNC to catalytically generate ROS. This is consistent with DFT simulations (Fig. 6c and d). The optimized adsorption energy of H₂O₂ on the (012) facet (−1.21 eV) is stronger than on the (001) facet (−0.94 eV). On the (012) surface, H₂O₂ adopts a monodentate configuration stabilized by hydrogen bonding, whereas the (001) facet favors a different adsorption mode. Concurrently, the increase in the O_suf component (associated with –OH, O²⁻, and O₂²⁻) indicates the formation of surface oxygen vacancies in the presence of H₂O₂, which further promotes catalytic activity.

Fig. 6 Surface oxygen speciation and H₂O₂ adsorption on hematite facets. High-resolution O 1s XPS spectra of (a) HNP and (b) HNC before and after the photocatalytic reaction with p-ASA. (c and d) DFT-optimized adsorption configurations and corresponding adsorption energies of H₂O₂ on the (001) and (012) facets of hematite.

The distinct surface atomic structures of the two facets underlie these differences. Hematite crystal has a rhombohedral hexagonal structure in which two-thirds of the octahedral sites are occupied by Fe³⁺ ions.⁴⁴ The predominantly exposed (001) facet on HNP presents a mixed termination of iron and oxygen layers, with a limited density of low-coordinate surface iron cations (∼4.6 atoms nm⁻²).⁴⁶ In contrast, the (012) facet on HNC exhibits a ridge-and-valley topography that exposes a higher density of five-coordinated iron sites (Fe_5c, ∼7.3 atoms nm⁻²).⁴⁷ These coordinatively unsaturated iron cations act as Lewis acid sites, providing more active centers for H₂O₂ and p-ASA adsorption and reaction.⁴⁸

Based on the combined experimental and theoretical evidence, a surface-mediated radical generation pathway is proposed. Under light irradiation, labile Fe(II) sites are generated on hematite (eqn (5)). These sites strongly adsorb H₂O₂ (a Lewis base) to form surface complexes such as ≡Fe(III)OFe(II)(H₂O₂)* or ≡Fe(III)OFe(III)(H₂O₂)* (eqn (6)–(7)). Due to the reducibility of Fe(II), electron transfer to H₂O₂ produces ˙OH and regenerates Fe(III) (eqn (8)). The Fe(III)/Fe(II) cycle can be sustained by further reaction with H₂O₂ (eqn (9)), continuously supplying ˙OH for p-ASA oxidation.⁴⁹ The higher density of under-coordinated Fe sites on the (012) facet (HNC) facilitates stronger H₂O₂ adsorption, more efficient electron transfer, and consequently greater ˙OH production, explaining its enhanced photocatalytic activity compared to the (001) facet (HNP).¹¹

≡Fe(III)OFe(III) + hv → ≡Fe(III)OFe(II) (5)

≡Fe(III)OFe(II) + H₂O₂ ↔ ≡Fe(III)OFe(II)(H₂O₂)* (6)

≡Fe(III)OFe(III) + H₂O₂ ↔ ≡≡Fe(III)OFe(III)(H₂O₂)* (7)

≡Fe(III)OFe(II)(H₂O₂)* ↔ ≡Fe(III)OFe(III) + ·OH + OH⁻ (8)

≡Fe(III)OFe(III)(H₂O₂)* ↔ ≡Fe(III)OFe(II) + ·OOH + H⁺ (9)

Conclusions

To better understanding the environmental behavior of organoarsenic on hematite under visible light illumination, hematite crystals with exposed (001) and (012) facets were synthesized and used for the photocatalytic degradation of p-ASA under simulated sunlight. The degradation followed pseudo-first-order kinetics with clear facet dependence, showing higher rates on HNCs than HNPs. Hydroxyl radicals (·OH) were identified as the main reactive species via quenching and EPR analyses. Multi-method characterization revealed distinct arsenic transformation pathways: HNP promoted surface retention of arsenic through adsorption and mineralization, yielding less dissolved inorganic arsenic, whereas HNC enabled more complete degradation but also released more mobile inorganic arsenic. Thus, the environmental fate of p-ASA is governed jointly by facet-specific photocatalytic activity and surface immobilization capacity. This work provides mechanistic insight into facet-controlled photodegradation of organoarsenic compounds on iron oxides, highlighting how interfacial structure dictates not only degradation efficiency but also arsenic speciation and mobility-key factors for environmental risk assessment. The findings support the design of facet-tuned iron-based catalysts for simultaneous pollutant degradation and contaminant sequestration.

Author contributions

Y. G. and Y. S. performed the experiments and conducted formal analysis. Y. G. wrote the original draft. S. X., Z. Z., and Q. S. contributed to methodology and manuscript review. L. O. conceived and supervised the study, acquired funding, and reviewed the manuscript. All authors participated in the intellectual development of the work.

Conflicts of interest

There are no conflicts to declare.

Data availability

The data supporting this article have been included as part of the supplementary information (SI).

Supplementary information: the SI includes Fig. S1 XRD patterns of HNP and HNC. Fig. S2 Integrated performance of hematite in the dark adsorption and subsequent photocatalytic degradation of p-ASA. Fig. S3 effects of reaction conditions on the photocatalytic degradation of p-ASA. Fig. S4 Time-dependent concentrations of inorganic arsenic species (As(III) and As(V)) released during the photocatalytic degradation of p-ASA. Fig. S5 post-degradation distribution of arsenic species in the solid and liquid phases. Fig. S6 analysis of degradation intermediates.

Acknowledgements

The authors acknowledge the funding support from the Research Fund of Shenzhen Science and Technology Program (JCYJ20250604183434047), Guangdong Basic and Applied Basic Research Foundation (2026A1515011982), the open project program in Hunan Provincial Key Laboratory of Geochemical Processes and Resource Environmental Effects, Geophysical and Geochemical Survey institute of Hunan (GRE202501G), Hubei Key Laboratory of Resources and Eco-Environment Geology (HBREGKFJJ-202408), and the Natural Science Foundation of Hubei Province (2025AFA048).

References

1. C. Anderson, Arsenic: Industrial, biomedical, environmental perspectives, 1983, pp. 89–97.
2. E. D'Angelo, G. Zeigler, E. G. Beck, J. Grove and F. Sikora, Sci. Total Environ., 2012, 438, 286–292.
3. P. W. Aschbacher and V. J. Feil, J. Agric. Food Chem., 1991, 39, 146–149.
4. D. D. Giri, N. Srivastava, B. C. Ruidas, M. Y. Areeshi, S. Haque and D. B. Pal, Biomass Convers. Biorefin., 2023, 13, 13357–13367.
5. Y. Kong, F. Lu, Z. Huang, Y. Nie, Z. Chen and J. Ma, Sep. Purif. Technol., 2024, 343, 127121.
6. T. Yang, L. Wang, Y. Liu, J. Jiang, Z. Huang, S.-Y. Pang, H. Cheng, D. Gao and J. Ma, Environ. Sci. Technol., 2018, 52, 13325–13335.
7. Y. Zhu, Q. Xie, F. Deng, Z. Ni, Q. Lin, L. Cheng, X. Chen, R. Qiu and R. Zhu, Sep. Purif. Technol., 2023, 325, 124702.
8. J. R. Garbarino, A. J. Bednar, D. Rutherford, R. Beyer and R. Wershaw, Environ. Sci. Technol., 2003, 37, 1509–1514.
9. T. Fresno, E. Moreno-Jiménez, P. Zornoza and J. M. Peñalosa, J. Environ. Manage., 2018, 205, 142–150.
10. Y. Gao, Y. Luo, Z. Pan, Z. Zeng, W. Fan, J. Hu, Z. Zhang, J. Ma, Y. Zhou and J. Ma, Water Res., 2024, 249, 120967.
11. X. Huang, Y. Chen, E. Walter, M. Zong, Y. Wang, X. Zhang, O. Qafoku, Z. Wang and K. M. Rosso, Environ. Sci. Technol., 2019, 53, 10197–10207.
12. M. A. Jafar Khan and R. J. Watts, Water, Air, Soil Pollut., 1996, 88, 247–260.
13. K. M. Lee, C. W. Lai, K. S. Ngai and J. C. Juan, Water Res., 2016, 88, 428–448.
14. A. D. Bokare and W. Choi, J. Hazard. Mater., 2014, 275, 121–135.
15. C. Li, Y. Zhang, Y. Zheng, C. Shi, Y. Lu, Y. Zhang and S. Yuan, Sci. Total Environ., 2024, 919, 170855.
16. X. Wang, J. Wang, Z. Cui, S. Wang and M. Cao, RSC Adv., 2014, 4, 34387–34394.
17. W.-C. Huang, L.-M. Lyu, Y.-C. Yang and M. H. Huang, J. Am. Chem. Soc., 2012, 134, 1261–1267.
18. W.-Q. Ding, L. Labiadh, L. Xu, X.-Y. Li, C. Chen, M.-L. Fu and B. Yuan, Chemosphere, 2023, 339, 139687.
19. H. E. Rudel, M. K. M. Lane, C. L. Muhich and J. B. Zimmerman, ACS Nano, 2020, 14, 16472–16501.
20. C. M. Eggleston, A. G. Stack, K. M. Rosso, S. R. Higgins, A. M. Bice, S. W. Boese, R. D. Pribyl and J. J. Nichols, Geochim. Cosmochim. Acta, 2003, 67, 985–1000.
21. K. S. Tanwar, C. S. Lo, P. J. Eng, J. G. Catalano, D. A. Walko, G. E. Brown Jr, G. A. Waychunas, A. M. Chaka and T. P. Trainor, Surf. Sci., 2007, 601, 460–474.
22. J. Bargar, G. Brown Jr and G. Parks, Geochim. Cosmochim. Acta, 1998, 62, 193–207.
23. M. R. Noerpel, S. S. Lee and J. J. Lenhart, Environ. Sci. Technol., 2016, 50, 12283–12291.
24. P. Chen, M. P. Prange, D. Song, S. Lee, S. S. Lee, J. Liu, Y. Zhao, X. Wang, M. H. Engelhard and Y. Zhou, Environ. Sci. Technol., 2025, 59, 11862–11874.
25. Z. Yang, Y. Li, X. Wang, J. Li, J. Wang and G. Zhang, J. Environ. Sci., 2024, 142, 204–214.
26. O. Zandi and T. W. Hamann, Nat. Chem., 2016, 8, 778–783.
27. H. Wu, T. Yang, Y. Du, L. Shen and G. W. Ho, Adv. Mater., 2018, 30, 1804341.
28. W. Li, K. R. Yang, X. Yao, Y. He, Q. Dong, G. W. Brudvig, V. S. Batista and D. Wang, ACS Appl. Mater. Interfaces, 2018, 11, 5616–5622.
29. X. Jin, D. Wu, C. Liu, S. Huang, Z. Zhou, H. Wu, X. Chen, M. Huang, S. Zhou and C. Gu, Nat. Commun., 2022, 13, 6125.
30. J. Liu, S. Inoué, R. Zhu, H. He and M. F. Hochella, Geochim. Cosmochim. Acta, 2021, 307, 151–167.
31. H. Mei, Y. Liu, X. Tan, J. Feng, Y. Ai and M. Fang, J. Hazard. Mater., 2020, 399, 123028.
32. W. Chen, G. Guo, L. Huang, L. Ouyang and Q. Shuai, J. Hazard. Mater., 2024, 464, 132976.
33. X. Huang, X. Hou, F. Song, J. Zhao and L. Zhang, J. Phys. Chem. C, 2017, 121, 1113–1121.
34. G. Zhang, Z. Yuan, L. Lei, J. Lin, X. Wang, S. Wang and Y. Jia, Chem. Geol., 2019, 525, 380–389.
35. L. Ouyang, M. Wang, L. Zhu, H. Tang and Q. Shuai, Nano Res., 2023, 16, 3046–3054.
36. Z. Cui, X. Zhang, B. Hu, L. Zhu and L.-Z. Huang, Water Res., 2025, 277, 123293.
37. S. Cao, X. Zhang, X. Huang, S. Wan, X. An, F. Jia and L. Zhang, Environ. Sci.: Nano, 2019, 6, 3280–3291.
38. W. Chen, G. Guo, L. Huang, L. Ouyang and Q. Shuai, J. Hazard. Mater., 2024, 464, 132976.
39. L. Wang and H. Cheng, Environ. Sci. Technol., 2015, 49, 3473–3481.
40. Y. Yuan, X. Yuan, L. Luo and X. Xie, ACS ES&T Water, 2022, 2, 830–840.
41. R. López, F. J. Pereira, D. Suárez and A. J. Aller, Microchem. J., 2022, 175, 107186.
42. A. Sheng, X. Li, Y. Arai, Y. Ding, K. M. Rosso and J. Liu, Environ. Sci. Technol., 2020, 54, 7309–7319.
43. J. Liu, A. Sheng, X. Li, Y. Arai, Y. Ding, M. Nie, M. Yan and K. M. Rosso, Environ. Sci. Technol., 2022, 56, 3801–3811.
44. N. Chen, Y. Wan, G. Zhan, X. Wang, M. Li and L. Zhang, Chem. Eng. J., 2020, 384, 123254.
45. X. Xie, Y. Hu and H. Cheng, Water Res., 2016, 89, 59–67.
46. A. K. Patra, S. K. Kundu, A. Bhaumik and D. Kim, Nanoscale, 2016, 8, 365–377.
47. X. Huang, X. Hou, X. Zhang, K. M. Rosso and L. Zhang, Environ. Sci.: Nano, 2018, 5, 1790–1806.
48. X. Zhou, J. Lan, G. Liu, K. Deng, Y. Yang, G. Nie, J. Yu and L. Zhi, Angew. Chem., Int. Ed., 2012, 51, 178–182.
49. X. Huang, X. Hou, J. Zhao and L. Zhang, Appl. Catal., B, 2016, 181, 127–137.

---