Visible-Light Photocatalytic Abatement of Diclofenac Potassium over Bismuth Sulfoiodide

Abstract

The growing presence of pharmaceutical residues in aquatic systems underscores the need for effective and efficient treatment processes capable of eliminating persistent contaminants under mild conditions. In this study, bismuth sulfoiodide (BiSI) was investigated as a visible-light-responsive photocatalyst for the abatement of diclofenac potassium from aqueous media. Orthorhombic bismuth sulfoiodide nanorods were synthesized via a solution-phase method and comprehensively characterized using multiple analytical techniques. Powder X-ray diffraction (PXRD) confirmed the phase purity of crystalline BiSI, whereas UV–vis diffuse reflectance spectroscopy (UV–vis DRS) revealed strong visible-light absorption with an optical band gap of 1.65 eV. Scanning electron microscopy (SEM) and high-resolution transmission electron microscopy (HRTEM) images showed uniform nanorod morphology with an average width of 148 ± 46 nm and an average length of 2.84 ± 1.37 μm, indicating a high aspect ratio. Energy-dispersive X-ray spectroscopy (EDX) analysis confirmed the elemental composition of the material. X-ray photoelectron spectroscopy (XPS) confirmed the presence of Bi³⁺, S^2–, and I^– ions, consistent with the BiSI lattice structure. The photocatalyst demonstrated excellent performance under visible-light irradiation, achieving 95.2% removal of the parent diclofenac potassium, as determined by UV–vis spectroscopic analysis, within 120 min. The reaction was described by a pseudo-first-order Langmuir–Hinshelwood model with an apparent rate constant of 2.999 × 10^–2 min^–1; however, the moderate fitting indicates that this model does not represent ideal pseudo-first-order kinetic behavior.

1. Introduction

The discharge of pharmaceutical wastewater into aquatic ecosystems poses a significant environmental threat, as many pharmaceutical compounds are resistant to conventional treatment and remain biologically active after release. Wastewater generated from hospitals, residential zones, and pharmaceutical manufacturing represents a substantial pathway for the release of antibiotics, analgesics, hormones, and other hazardous pollutants into aquatic environments. These persistent pollutants cause chronic toxicity in aquatic and biological systems, disturb ecological balance, and drive antimicrobial resistance, even at minimal concentrations. Accordingly, advancing innovative technologies for pharmaceutical wastewater remediation is imperative.

Among the various therapeutic classes driving this ecological concern, non-steroidal anti-inflammatory drugs (NSAIDs) are particularly notorious, with diclofenac potassium standing out as one of the most frequently detected pharmaceutical contaminants worldwide. Diclofenac potassium is a widely detected pharmaceutical contaminant due to extensive usage, high solubility in water, and resistance to biodegradation. Residual diclofenac is detected in surface waters and municipal effluents at concentrations ranging from nanograms to micrograms per liter, posing toxicological risks to aquatic organisms. Insufficient removal generates hazardous intermediates, so treatment strategies must ensure complete mineralization.

The urgency to completely mineralize such robust molecules has exposed severe technical bottlenecks in our existing water management infrastructure. Traditional treatment methodologies – such as conventional activated sludge systems, physical adsorption, membrane filtration, chlorination, and routine ozonation – face notable operational and energetic limitations when treating complex chemical streams. , Biological systems typically fail to degrade these synthetic aromatics, − whereas physical separation strategies like adsorption or membrane filtration merely transfer the concentrated contaminants from the liquid phase to a solid matrix without achieving complete mineralization. , Furthermore, powerful oxidative options like chlorination often trigger the unintended generation of highly carcinogenic halogenated byproducts. To overcome these phase-transfer and byproduct hurdles, contemporary environmental research has pivoted decisively toward advanced oxidation processes (AOPs) designed to oxidatively mineralize contaminants. ,

Heterogeneous semiconductor photocatalysis represents one of the most prominent advanced oxidation processes, offering a green, economically viable framework that directly exploits solar energy to remediate contaminated water. This mechanism operates by utilizing light photons of matching or greater energy than the semiconductor’s band gap to excite electrons from the valence band to the conduction band, creating active electron–hole (e ^–/h ⁺ ) pairs. Once migrated to the catalyst surface, these photo-induced charge carriers interact with dissolved oxygen and ambient water molecules to generate highly reactive oxygen species (ROS), predominantly superoxide radicals (O₂•^–) and hydroxyl radicals (•OH). These unselective, highly energetic radicals systematically attack the stable aromatic rings of organic pollutants, dismantling their structural integrity and converting them into benign end products, including CO₂ and H₂O, via mineralization. − The ability to drive these intense chemical reactions under ambient conditions using sunlight positions photocatalysis as a highly sustainable pillar of next-generation water purification.

The high environmental relevance of diclofenac stems from this persistence, as standard wastewater treatment technologies fail to achieve complete elimination, leaving hazardous residues that continuously accumulate in food chains. When subjected to advanced treatment, the complete mineralization of this robust molecule is heavily dependent on specific structural degradation pathways. Photo-driven processes break down the contaminant by generating reactive oxygen species, such as superoxide and hydroxyl radicals. These aggressive intermediates initiate oxidative transformation through targeted hydroxylation of the aromatic rings, subsequent dechlorination, and cleavage of the central C–N secondary amine linkage. These sequential steps lead to ring-opening reactions, breaking the molecule down into smaller, lower-molecular-weight aliphatic intermediates before achieving absolute mineralization into CO₂ and H₂O. − ,,

A plethora of photocatalysts have been studied for pharmaceutical degradation, including TiO₂, ZnO, WO₃, carbon nitride (CN _x ), metal–organic frameworks (MOFs), covalent organic frameworks (COFs), perovskites, spinel ferrites, and bismuth-based semiconductors. TiO₂ is extensively employed; however, its activity is limited to the ultraviolet region due to its wide band gap, whereas ZnO undergoes photocorrosion under prolonged irradiation. Polymeric carbon nitride and related materials can operate under visible-light but commonly exhibit limited charge-separation efficiency. MOFs and perovskites offer promising adsorption and photocatalytic properties, yet many systems suffer from hydrolytic instability, moisture sensitivity, metal leaching, or complex synthesis. Therefore, developing robust, visible-light-responsive photocatalysts that possess structural stability, facile synthesis, and efficient charge transport remains a key research priority.

Bismuth-based semiconductors are promising alternatives due to their low toxicity, favorable electronic structures, and strong visible-light absorption. Photocatalysts such as BiVO₄, BiOX (X = Cl, Br, I), and Bi₂WO₆ have demonstrated effective pollutant removal under visible-light irradiation due to their hybridized valence-band structures and suitable redox potentials. Among emerging photocatalytic semiconductors, BiSI exhibits a narrow band gap, featuring an orthorhombic anisotropic crystal structure that enables broad visible-light absorption, moreover, the presence of Bi³⁺ ions enhances spin–orbit coupling which in turn facilitates charge-carrier separation and reduce charge recombination. Furthermore, BiSI offers low toxicity and strong chemical stability, making it especially suitable for practical environmental applications.

Despite these promising optoelectronic properties, the focus of the current literature has predominantly centered on the architectural design of complex BiSI-based heterojunctions and multi-component composites to drive environmental remediation paths. For example, recent structural configurations have coupled BiSI with secondary matrices – such as g-C₃N₄, bismuth oxyhalides, or carbonaceous frameworks – to construct type-II or Z-scheme pathways optimized for general organic dye and specific pharmaceutical breakdown. While these composite designs successfully offset fast charge recombination, they inherently complicate catalyst scalability, increase manufacturing costs, and often obscure the fundamental, intrinsic photocatalytic activity and long-term surface stability of the standalone chalcohalide crystal matrix. − Consequently, systematic exploration centered purely on pristine, single-component BiSI architectures for targeted nonsteroidal anti-inflammatory drug mineralization remains a critical missing link. − To bridge this knowledge gap, this study systematically investigates the performance of BiSI nanorods for the targeted elimination of diclofenac potassium under visible-light irradiation. The engineered nanorods demonstrated exceptional photocatalytic performance, eliminating 95.2% of the targeted pharmaceutical within 120 min via a pseudo-first-order Langmuir–Hinshelwood model. This work establishes a clear benchmark for utilizing single-phase ternary bismuth chalcohalides as highly stable, visible-light-driven photocatalyst obviating the need of complex heterojunction engineering.

2. Experimental Section

2.1. Materials and General Procedure

Bismuth nitrate pentahydrate (Bi­(NO₃)₃·5H₂O), potassium iodide (KI), thiourea (CH₄N₂S), and diclofenac potassium were purchased from Fisher Scientific and used as received. Powder X-ray diffraction was performed using a Rigaku MiniFlex benchtop X-ray diffractometer equipped with a CuKα radiation tube (λ = 1.542 Å), operated at 40 kV, over a 2θ range of 10–60°, with a scan rate of 2° min^–1. Sample surface morphology and elemental composition were analyzed using JEOL JSM 7000F scanning electron microscope equipped with an energy-dispersive X-ray spectroscopy detector. The powder sample was mounted vertically on an aluminum stub using double-sided carbon tape and loaded into the SEM stage and the instrument was operated under high vacuum at a 7.5 kV accelerating voltage. The diameter and crystallinity of BiSI were analyzed using a JEM-2100F Transmission Electron Microscope at 80 kV. High-resolution TEM imaging was conducted at 200 kV. Samples suspended in ethanol were dropped onto holey-carbon coated copper grids and dried for 2 h. Average nanorod size and size distribution were measured with ImageJ software (NIH). X-ray photoelectron spectroscopy was conducted using a Thermo Scientific Model K-Alpha instrument with monochromatic Al Kα radiation (1486.7 eV) at an output power of 36 W (12,000 V at 3 mA). The X-ray spot size was adjustable between 30 and 400 μm. The instrument was equipped with a hemispherical electron energy analyzer and a 128-channel detector system. Survey and high-resolution spectra were acquired at analyzer pass energies of 200 and 50 eV, respectively. The base pressure in the analysis chamber was typically 2 × 10^–9 mbar or lower. Samples were mounted on the sample platen using copper clips. Analytical areas were selected using a digital optical camera with magnifications ranging from 60–200×. All spectra were collected with the charge-neutralization flood gun activated, which used low-energy electrons and argon ions. The typical chamber pressure with the flood gun in operation was 2 × 10^–7 mbar. Depth profiling was performed using a Thermo Scientific EX06 argon-ion gun at 1000 eV and a medium current density. Data acquisition and processing were performed using the Thermo Scientific Advantage XPS software package The quantitative surface composition was estimated from the integrated peak areas after background subtraction and correction using the instrument-provided relative sensitivity factors. The binding-energy scale was calibrated using the C 1s signal at 284.8 eV. High-resolution spectra were fitted using mixed Gaussian–Lorentzian line shapes after background correction. For the Bi 4f/S 2p region, simultaneous deconvolution was performed because the S 2p doublet overlaps with the Bi 4f envelope. The resulting atomic percentages represent the near-surface composition probed by XPS and should not be interpreted as bulk stoichiometric values. Thermogravimetric analysis (TGA) was performed using a Mettler Toledo TGA 2 analyzer, with the activated sample placed in an alumina pan prior to analysis. The specific surface area was determined from the N₂ adsorption–desorption isotherm collected at 77 K (−196.15 °C) using a Micromeritics ASAP 2020 surface area analyzer. UV–vis diffuse reflectance spectroscopy of the sample was recorded using a Shimadzu UV–Vis spectrophotometer (UV-2600i Plus) equipped with an ISR2600Plus integrating sphere accessory. Barium sulfate (BaSO₄) was used to record the baseline as a non-absorbing reflectance standard. The reflectance data were converted to absorbance using the Kubelka-Munk function, F(R) = (1 – R)²/2R, where R is the measured reflectance. The optical band gaps were estimated by constructing Tauc plots of (F(R)hν)² versus photon energy (hν), assuming direct allowed electronic transitions, and extrapolating the linear region to the energy axis. ,

2.2. Synthesis of BiSI

BiSI nanorods were synthesized using solution-phase method under acidic conditions following a previously published procedure, with thiourea used as the sulfur source in place of thioacetamide (TAA), as shown in Figure . Initially, 0.485 g (1.00 mmol) of bismuth nitrate pentahydrate was dissolved in 50 mL of deionized water and heated to 393 K with continuous stirring. In parallel, 0.34 g (2.00 mmol) of potassium iodide and 0.76 g (10.00 mmol) of thiourea (KI-thiourea solution) were dissolved in another 50 mL deionized water and similarly heated to 393 K. The KI-thiourea solution was transferred to a 300 mL round-bottom flask equipped with a reflux condenser and maintained in an oil bath at 393 K. The preheated Bi­(NO₃)₃·5H₂O solution was then added dropwise under continuous stirring. The pH of the mixture was adjusted to 1.0–1.2 using glacial acetic acid (AcOH), and the reaction mixture was refluxed for 5 h. Upon completion of the reaction, the resulting black precipitate was collected by centrifugation and washed sequentially four times with ethanol and six times with deionized water until the supernatant became colorless. The purified solid was then dried at 333 K for 8 h to obtain BiSI as a black powder.

1.

Schematic illustration of BiSI synthesis.

2.3. Photocatalytic Removal of Diclofenac Potassium

The photocatalytic activity of BiSI was evaluated by monitoring the disappearance of diclofenac potassium under visible-light irradiation. An aqueous solution of diclofenac potassium (1 × 10^–4 M, 100 mL, pH 6.59) was prepared using deionized water. In a typical experiment, 40 mg of BiSI was dispersed in 100 mL of the diclofenac potassium solution contained in a 250 mL round-bottom flask. The suspension was stirred in the dark for 60 min to reach adsorption–desorption equilibrium. The adsorption was evaluated by comparing diclofenac potassium absorbance measurements obtained before and after the dark equilibration period. The removal efficiency (RE%) was calculated by eq as follows:

$$\mathrm{RE\; \%}=[({\mathrm{C}}_0-{\mathrm{C}}_{\mathrm{e}}){/\mathrm{C}}_0]\times \mathrm{100}$$ 1

where C₀ and C_e are the initial and equilibrium concentrations of diclofenac potassium (μg/L), respectively. After equilibration, the mixture was irradiated with a 500 W halogen lamp (positioned 10 cm from the reaction mixture) used as a visible-light source and continuously stirred at room temperature. At set intervals, aliquots were collected and filtered through 0.22 μm PTFE membranes and analyzed using a UH5700 UV–vis spectrophotometer equipped with a 10 mm quartz cuvette. Spectral measurements were performed over a wavelength range of 240–400 nm, and disappearance of diclofenac potassium was evaluated by monitoring the decrease in absorbance at 288 nm. Control experiments performed in the absence of BiSI under identical irradiation conditions showed no observable disappearance of diclofenac potassium, confirming that direct photolysis is insignificant under the studied conditions. The recyclability of BiSI was evaluated through consecutive photocatalytic cycles under the same photocatalytic conditions. Following each cycle, the photocatalyst was recovered by centrifugation, washed sequentially with deionized water and ethanol, dried at 333 K, and subsequently reused in a new diclofenac potassium solution. The structural stability of the recovered BiSI photocatalyst was further examined by powdered X-ray diffraction (PXRD) after the recyclability experiments (Figure B).

11.

(A) Reusability of BiSI nanorods over five photocatalytic cycles. (B) The post recycling PXRD pattern of BiSI.

The dominant reactive species responsible for the photocatalytic removal of diclofenac potassium were identified through scavenger experiments performed under conditions identical to those used in the photocatalytic activity tests. An aqueous solution of diclofenac potassium (1.0 × 10^–4 M, 100 mL) containing 40 mg of BiSI nanorods was transferred to a 250 mL round-bottom flask. Three separate scavenger experiments were then conducted individually by adding p-benzoquinone (BQ), triethanolamine (TEOA), and tert-butanol (t-BuOH), each at a final concentration of 1.0 mM, to selectively quench superoxide radicals (•O₂ ^–), photogenerated holes (h ⁺), and hydroxyl radicals (•OH), respectively. ,

3. Results and Discussion

3.1. Characterization of BiSI

The powder X-ray diffraction pattern of the synthesized BiSI, shown in Figure , exhibits well-defined diffraction peaks indexed to the orthorhombic BiSI phase with the space group Pnma, and peak positions show excellent agreement with literature values and corresponding diffraction patterns , as well as the simulated one derived from COD-1535800 database. The cell parameters of BiSI were a = 8.460(3) Å, b = 10.150(4) Å and c = 4.14(2) Å and α = β = γ = 90°. The most intense reflections observed around ∼23°, ∼29–30°, and ∼40° correspond to the characteristic planes of BiSI, confirming successful phase formation. No additional intense peaks attributable to secondary crystalline phases are detected within the instrumental resolution, suggesting that BiSI is the predominant phase in the sample. Minor variations in relative peak intensities compared to the simulated pattern may arise from preferred-orientation effects associated with the anisotropic growth of BiSI nanorods, which are commonly observed in one-dimensional BiSI nanostructures prepared from solutions. In addition, a few weak extra reflections or shoulders are observed, suggesting the possible presence of trace secondary Bi–S–I phases. Such minor secondary phases have been reported previously during solution-based growth of BiSI materials. , Nevertheless, the overall agreement between the experimental and simulated patterns confirms that orthorhombic BiSI is the predominant phase in the synthesized sample.

2.

PXRD pattern of the synthesized BiSI compared with the simulated diffraction pattern over the 10–60° 2θ range, derived from the COD-1535800 database.

The UV–vis diffuse reflectance spectrum of the synthesized BiSI is presented in Figure A, where the spectrum exhibits a distinct absorption edge at approximately 751 nm. The optical band gap energy was determined using the Tauc approach, assuming a direct allowed transition, as BiSI is commonly reported as a direct band gap semiconductor. , The linear region of the (αhv)² versus photon energy plot (Figure B) was extrapolated to the energy axis, yielding a band gap value of approximately 1.65 eV, which is in good agreement with previously reported band gap values for BiSI, typically reported in the range of 1.3–1.8 eV, depending on the synthesis method, morphology, crystallite size, and sample thickness. , The steep absorption edge indicates efficient visible-light absorption, supporting its suitability for visible-light-driven photocatalysis. Although the PXRD analysis suggests BiSI as the major crystalline phase, the optical band gap should be interpreted as an effective band gap of the synthesized sample, since minor secondary phases or morphology-related effects may slightly influence the absorption profile. The broad absorption extending into the near-infrared region may be associated with the anisotropic nanorod morphology and/or defect-related electronic states, which are commonly observed in solution-processed semiconductor materials. ,

3.

(A) Diffuse reflectance spectrum of the synthesized BiSI and (B) Tauc plot for a direct allowed transition.

XPS analysis was employed to verify the chemical composition and oxidation states of the crystalline BiSI sample. The XPS survey spectrum in Figure A confirms the presence of Bi, S, and I as the constituent elements of the synthesized material. The characteristic Bi 4f doublet is observed in the 154–168 eV region, the S 2p signal appears around 156–170 eV, and the I 3d doublet is clearly detected at approximately 600 and 640 eV, consistent with the expected core-level binding energies of BiSI. Additionally, quantitative surface analysis indicates atomic percentages of approximately 9% Bi, 6.5% S, and 11% I. High-resolution analysis of the Bi 4f region (Figure B) reveals two well-defined peaks at ∼158.6 eV (Bi 4f_7/2) and ∼163.9 eV (Bi 4f_5/2), with a spin-orbit splitting of ∼5.3 eV, characteristic of Bi³⁺ in a chalcogenide environment. − The S 2p spectrum (Figure B and C) consists of the typical 2p_3/2 and 2p_1/2 components at ∼161.5 and ∼162.6 eV, respectively, with a spin-orbit separation of ∼1.2 eV, yielding S^2– binding energies typical of metal sulfides. Importantly, it should be noted that the S 2p doublet overlaps with the Bi 4f envelope due to their close binding energy positions; therefore, simultaneous fitting of both contributions was required to accurately deconvolute the spectra. Because the S 2p doublet overlaps with the Bi 4f region, sulfur assignment was supported by simultaneous deconvolution of the Bi 4f/S 2p envelope and by inspecting the survey spectrum for the S 2s contribution, where distinguishable (∼225–230 eV in Figure A). Moreover, the I 3d region in Figure D exhibits two sharp peaks at ∼619.0 eV (3d_5/2) and ∼630.5 eV (3d_3/2), with the expected spin-orbit splitting of ∼11.2 eV, characteristic for I^– anions, which also confirms the presence of iodide species. Minor signals from adventitious surface contamination are also observed, which are commonly detected in air-exposed samples and do not affect the identification of the BiSI phase. The additional surface contributions from C, O, and N signals are attributed to adventitious surface carbon, adsorbed oxygen-containing species, and residual nitrogen-containing species associated with the synthesis/handling process. The surface atomic percentages of Bi, S, and I are not exactly consistent with the nominal 1:1:1 BiSI stoichiometry. This discrepancy is expected in XPS analysis, as the technique probes only the near-surface region and is strongly influenced by surface adsorbates, signal attenuation, and potential surface termination effects. , In addition, accurate quantification of sulfur requires careful deconvolution due to the overlap of the S 2p doublet with the Bi 4f region. Therefore, the XPS results are interpreted as evidence for the surface chemical composition and oxidation states of BiSI, while PXRD provides confirmation of the dominant crystalline BiSI phase.

4.

(A) XPS survey spectrum of the synthesized BiSI showing the presence of Bi, S, and I elements. High-resolution (B) Bi 4f, (C) S 2p, and (D) I 3d XPS spectra.

Thermogravimetric analysis (TGA) of BiSI nanorods revealed multistep thermal decomposition, with the material exhibiting good thermal stability up to approximately 653 K. (Figure ) The initial mass loss of 0.58% between 298 and 653 K is due to the evaporation of adsorbed moisture and residual solvent molecules. The primary weight loss of 58.50% between 653 and 723 K, is mainly attributed to the decomposition of the BiSI lattice and volatilization of sulfur- and iodine-containing species. A gradual mass loss of 19.50% between 723 and 973 K is attributed to ongoing structural breakdown and removal of volatile fragments. Above 973 K, the mass remained nearly constant, resulting in a final residue of approximately 20.0% at 1073 K. The remaining residue is likely associated with the formation of thermally stable, bismuth-rich phases (e.g., bismuth oxide) or other nonvolatile species.

5.

TGA analysis of BiSI.

The BET specific surface area of BiSI was 3.07 m² g^–1, whereas the Langmuir surface area was 5.22 m² g^–1. The single-point total pore volume was 0.0032 cm³ g^–1. Using the geometric 4V/A relationship, where V is the single-point pore volume and A is the BET surface area, an apparent average pore width of 4.15 nm was obtained. In comparison, BJH (Barrett-Joyner-Halenda) analysis gave a larger apparent average pore width of 39.5 nm.

Given the very low surface area and pore volume, these values are not considered representative of well-defined internal pores. Rather, they are likely associated with adsorption on the external surface and with voids formed between aggregated needle-like BiSI particles, as also suggested by the SEM images (Figure ). Synthesized BiSI can therefore be described as essentially nonporous, with its minimal adsorption capacity arising predominantly from external surfaces and interparticle spaces.

6.

SEM images of BiSI nanorods at low and high magnifications.

SEM images reveal that BiSI predominantly forms elongated, uniformly distributed rods (Figure ). The one-dimensional crystallites exhibit nanometer-scale widths and micrometer scale lengths, resulting in high aspect ratios. The rods have an average width of 148 ± 46 nm and an average length of 2.84 ± 1.37 μm. The particles typically exhibit widths under 200 nm and lengths ranging from approximately 1.5 to 4.0 μm (Figure ). The calculated aspect ratio is 19.2, which is indicative of pronounced anisotropic growth characteristic of bismuth chalcohalides. At low magnification, a dense and homogeneous distribution of rods is observed with no evidence of spherical particles or irregular agglomerates, suggesting highly uniform crystal growth and controlled nucleation. While the interconnected rod network creates an open macroscopic framework that facilitates mass transport and diffusion of the bulky diclofenac potassium molecules, it does not contribute significantly to high surface area. At higher magnification, the individual rods appear prismatic, with smooth surfaces and sharp edges, which represents high crystallinity. The absence of amorphous coatings or particulate residues demonstrates successful synthesis and effective post-treatment. Comparable morphologies have been reported for solution-based BiSI, attributed to directional bonding and the presence of stable crystal facets. High-aspect-ratio BiSI nanorods are expected to enhance photocatalytic performance by providing efficient charge transport pathways. ,

7.

Statistical size analysis of BiSI nanorods. (A) Length distribution of BiSI nanorods with log-normal fitting. (B) Width distribution of BiSI nanorods with log-normal fitting. (C) Aspect ratio (length/width) distribution with log-normal fitting. (D) Box plot of length and width showing median, interquartile range, and data spread.

Figure shows TEM and HRTEM images of synthesized BiSI. The material forms well-defined, one-dimensional rods, with lengths from several hundred nanometers to a few micrometers (Figure A). The nanorods have variable diameters and smooth surfaces. This indicates controlled anisotropic growth. High-resolution TEM (Figure D) reveals lattice fringes at the nanorod tips, confirming the crystalline nature. The interplanar spacing of 0.313(2) nm matches the (121) plane of orthorhombic BiSI. An extra spacing of 0.227(1) nm appears, possibly from a higher-index plane or minor lattice variation. The continuous lattice fringes confirm the formation of ordered crystalline domains within the nanorods.

8.

TEM images of BiSI nanorods illustrate a one-dimensional rod-like morphology (A, B) and partial aggregation (C). The HRTEM image (D) reveals distinct lattice fringes with interplanar spacings of 0.313(2) and 0.227(1) nm, which confirm the crystalline structure of the BiSI nanorods.

EDX analysis confirmed the presence of bismuth, sulfur, and iodine in the synthesized material. The representative region contained 81.9 wt % Bi, 5.50 wt % S, 3.50 wt % I, and 9.20 wt % O, corresponding to 33.70 atomic% Bi, 14.60% S, 2.40% I, and 49.30% O (Figure ). The oxygen signal is attributed primarily to surface oxidation and adsorbed oxygen-containing species following exposure to air.

9.

EDX spectrum, quantitative elemental composition, and corresponding elemental maps of the synthesized BiSI sample, confirming the presence and spatial distribution of Bi, S, I, and O within the rod-like structures.

Local variations in elemental composition are expected for rod-like micro- and nanostructures because of differences in surface roughness, particle overlap, sampling depth, and particle orientation during analysis. Despite these variations, all regions analyzed shared the same elemental fingerprint – bismuth, sulfur, and iodine – demonstrating compositional uniformity across the sample. The persistent presence of these three elements, along with the absence of significant foreign impurities, confirms the material’s chemical purity. The elemental mapping and EDX spectra are provided in Figure . The elemental composition obtained via EDX differs from that of XPS because EDX probes a significantly larger subsurface interaction volume. Conversely, XPS is a surface-sensitive technique that primarily detects the outermost atomic layers. Consequently, factors such as surface oxidation, adsorbed chemical species, particle geometry, and local compositional variations may contribute to the observed discrepancy between the two methods. Similar surface-composition variations have been reported for BiSI single crystals analyzed by XPS.

3.2. Photocatalytic Activity

To establish adsorption–desorption equilibrium at the photocatalyst–solution interface, the BiSI nanorod–diclofenac potassium suspension was stirred in the dark prior to visible-light irradiation. Within the first 40 min, the absorbance of diclofenac potassium at 288 nm decreased by approximately 17% and subsequently remained constant, confirming that equilibrium had been reached (Figure A).

10.

(A) UV–vis absorption spectra of diclofenac potassium recorded during dark equilibration in the presence of BiSI nanorods. The decrease in absorbance at 288 nm corresponds to approximately 17% adsorption, with equilibrium reached after about 40 min. (B) UV–visible absorption spectra obtained during the removal of diclofenac potassium with BiSI photocatalyst. (C) Diclofenac potassium removal efficiency over BiSI nanorods as a function of time under visible light irradiation. Data from three separate trials are presented as mean ± SD.

To track the reaction kinetics, temporal UV–vis absorption spectra were acquired at regular intervals. The intense absorption maximum of diclofenac potassium at 288 nm exhibited a steady decline as visible-light irradiation progressed, demonstrating continuous and effective photocatalytic removal. The absorbance values at 288 nm and their corresponding removal efficiencies across various irradiation intervals are summarized in Figure B.

To evaluate the statistical reliability of the photocatalytic performance, the diclofenac potassium degradation experiments were performed in triplicate under identical reaction conditions. The reported values are expressed as the mean ± standard deviation of three independent replicates. As shown in Table , the concentration of diclofenac potassium decreased progressively with increasing irradiation time. Furthermore, the minimal margins of the error bars across all irradiation intervals (Figure C) demonstrate the high reproducibility of the photocatalytic process. Furthermore, with the main absorption band, the gradual reduction of the broad shoulder in the ca. 300–340 nm region suggests progressive disappearance of the conjugated aromatic structure of diclofenac potassium, indicating its transformation into lower-molecular-weight intermediates.

1. Absorbance at 288 nm and Corresponding Removal Efficiency of Diclofenac Potassium during Visible-Light Irradiation in the Presence of BiSI Nanorods.

Irradiation time (min)	Absorbance at 288 nm	Removal efficiency (%)
0	0.83	0.00
15	0.50	39.83 ± 0.35
45	0.40	51.80 ± 0.35
90	0.17	79.50 ± 0.44
120	0.04	95.20 ± 0.41
135	Negligible	99.99 ± 0.27

The BiSI nanorods demonstrated robust reusability under visible-light irradiation, maintaining high photocatalytic performance over five consecutive cycles. Specifically, the diclofenac potassium elimination efficiencies were recorded at 99.9%, 88.0%, 84.3%, 82.4%, and 80.1% for the first through fifth cycles, respectively (Figure A), indicating that the photocatalyst retained approximately 84% of its initial activity. The steady, marginal decline in efficiency is primarily attributed to minor photocatalyst mass loss during successive recovery and washing steps – with recovery rates measuring 97.0%, 94.0%, 91.3%, 89.3%, and 87.3% across the cycles – as well as the potential accumulation of reaction intermediates on the active surface sites. Crucially, the post-cycling PXRD patterns (Figure B) confirmed the preservation of the original crystalline framework, retaining all characteristic diffraction peaks of BiSI without the emergence of secondary crystalline phases.

Because the halogen lamp also produces heat, a control experiment was conducted under thermal conditions to mimic the heat generated by the lamp. No decrease in the absorption peaks was observed, confirming that heating alone did not contribute to the removal process.

3.3. Mechanistic Study

The exceptional photocatalytic performance of BiSI is intrinsically linked to its narrow optical band gap and intense visible-light absorption. Upon excitation with photons possessing energy equal to or exceeding this band gap, electrons are promoted from the valence band (VB) to the conduction band (CB), leaving behind positively charged holes in the VB. These photogenerated charge carriers subsequently migrate to the BiSI surface to drive interfacial redox reactions. Analogous pathways for reactive species generation have been comprehensively documented across various semiconductor photocatalytic systems. ,,

$$\mathrm{BiSI}+h\nu \to \mathrm{BiSI}({{\mathrm{e}}^{-}}_{\mathrm{CB}}+{h^{+}}_{\mathrm{VB}})$$

$${{\mathrm{e}}^{-}}_{\mathrm{CB}}+{\mathrm{O}}_2\to •{{\mathrm{O}}_2}^{-}$$

$${h^{+}}_{\mathrm{VB}}+{\mathrm{H}}_2\mathrm{O}/{\mathrm{OH}}^{-}\to •\mathrm{O}\mathrm{H}$$

These reactive species initiate the oxidative transformation of diclofenac through hydroxylation, dechlorination, C–N bond cleavage, and aromatic ring opening, thereby forming smaller oxidized intermediates. , These transformation pathways have been extensively elucidated by chromatographic and mass spectrometric analyses. ,−

The roles of various reactive species were investigated by introducing p-benzoquinone (BQ), triethanolamine (TEOA), and tert-butanol (t-BuOH) as scavengers for •O₂ ^–, h⁺, and •OH, respectively. These scavengers are widely used in photocatalytic trapping experiments to identify the dominant active species. Without any scavenger, BiSI removed 95.2% of diclofenac after 120 min of visible-light irradiation. The removal efficiency declined to 36.4% in the presence of BQ, indicating strong inhibition of the photocatalytic process. Since BQ selectively scavenges •O₂ ^– species, this significant reduction demonstrates that superoxide radicals are the primary reactive species involved in photocatalytic oxidation of diclofenac potassium. ,,

The addition of TEOA lowered the removal effectiveness to 47.2%, demonstrating that photogenerated holes also contribute significantly to the photocatalytic process. These holes may participate in the direct oxidation of adsorbed diclofenac molecules and the formation of secondary reactive oxygen species. In contrast, the removal efficiency remained relatively high (81.5%) in the presence of t-BuOH (Figure A and B). The comparably mild suppression by t-BuOH suggests that hydroxyl radicals participate in the photocatalytic process but are not the primary oxidizing agents. Based on these scavenger experiments, the apparent contribution of the reactive species follows the order:

$$•{{\mathrm{O}}_2}^{-}>h^{+}>•\mathrm{O}\mathrm{H}$$

12.

(A) Effect of reactive-species scavengers on the visible-light removal of diclofenac potassium over BiSI. (B) UV–vis spectra of diclofenac potassium after 120 min of visible-light irradiation over BiSI with and without scavengers. The strongest inhibition by BQ indicates that •O₂ ^® is the dominant reactive species.

The significant contributions of •O₂ ^– and h ⁺ observed for BiSI are consistent with those reported for a recently developed visible-light-driven photocatalytic system for diclofenac potassium, in which radical-trapping experiments likewise identified •O₂ ^– and h ⁺ as the dominant reactive species. These findings support a mechanism in which photoexcited conduction-band electrons reduce dissolved oxygen to generate •O₂ ^– radicals, while photogenerated holes provide an additional oxidation pathway. Hydroxyl radicals (•OH) also participate in the process but play a comparatively minor role. Furthermore, the pronounced inhibition observed in the presence of BQ, together with the negligible direct photolysis in the absence of BiSI, confirms that diclofenac potassium transformation proceeds predominantly through a visible-light-driven photocatalytic process mediated by BiSI nanorods.

3.4. Kinetic Analysis of Diclofenac Potassium Removal

A pseudo-first-order kinetic model was employed to evaluate the apparent kinetics of diclofenac potassium degradation over BiSI nanorods under visible-light irradiation. The kinetic analysis was performed after the adsorption–desorption equilibrium had been established in the dark. Following 60 min of dark equilibration, t = 0 was defined as the onset of visible-light irradiation, and C₀ corresponded to the diclofenac potassium concentration at that time rather than the initial concentration before adsorption. The concentration at irradiation time t (C_t) was determined from the absorbance at 288 nm using the Beer–Lambert law. Photocatalytic reactions in batch systems are commonly described using pseudo-first-order or Langmuir–Hinshelwood kinetic models, particularly at relatively low pollutant concentrations. However, the suitability of these models should be assessed based on the goodness of fit to the experimental data.

The pseudo-first-order kinetic equation is expressed as

$${\ln (\mathrm{C}}_0{/\mathrm{C}}_{\mathrm{t}}{)=\mathrm{k}}_{app}\mathrm{t}+\mathrm{b}$$ 2

where C₀ is the diclofenac potassium concentration at the onset of visible-light irradiation (after dark adsorption–desorption equilibrium), C_t is the concentration at irradiation time t, k _app is the apparent pseudo-first-order rate constant (min^–1), and b is the intercept obtained via unconstrained linear regression.

Since the concentration of diclofenac potassium was measured by its absorbance at 288 nm, the Beer–Lambert relationship was used to approximate the concentration ratio C₀/C_t using the corresponding absorbance ratio A₀/A_t. At extended irradiation times, the absorbance values became very low, so even small experimental variations could significantly influence the calculated ln­(C₀/C_t) values. Figure shows a plot of ln­(C₀/C_t) against irradiation time. Linear regression on the complete experimental dataset yielded the following equation:

$${\ln (\mathrm{C}}_0{/\mathrm{C}}_{\mathrm{t}})=0.02999\mathrm{t}-0.45629$$ 3

13.

Pseudo-first-order kinetic plot for the visible–light photocatalytic removal of diclofenac potassium over BiSI nanorods.

The slope has an apparent rate constant of k _app = 2.999 × 10^–2 min^–1 and a coefficient of determination of: R² = 0.8125.

The positive curvature confirms the gradual decrease in diclofenac potassium concentration with increasing irradiation time. However, the R² value of 0.8125 indicates only a moderate linear relationship, while the intercept of −0.45629 deviates substantially from zero. These results suggest that the experimental data do not strictly follow ideal pseudo-first-order kinetics over the entire irradiation period.

Accordingly, the pseudo-first-order model should be considered an approximate empirical description of the overall removal behavior rather than definitive evidence of a single first-order reaction mechanism. A kinetic rate expression alone cannot unambiguously establish a specific photocatalytic pathway, as similar mathematical forms may arise from different combinations of adsorption, surface reactions, charge transfer processes, and mass transport limitations.

The most pronounced deviation from linearity occurred during the later stages of irradiation, where minor absolute differences in absorbance at low residual diclofenac concentrations resulted in disproportionately substantial changes in ln­(C₀/C_t). Consequently, the data point acquired at 135 min exerts a disproportionate influence on the slope, intercept, and R² coefficient of the fitted line. The kinetic value at this final interval may be altered by the combined effects of an exceptionally low residual absorbance and potential spectral overlap from intermediate transformation products generated during irradiation. Given that these intermediates were not explicitly identified in the current study, this hypothesis is proposed solely as a plausible analytical explanation for the observed kinetic discrepancy.

Nonideal kinetic behavior may also arise from variations in diclofenac surface coverage, competition for accessible active sites, fluctuations in dissolved oxygen or reactive species availability, catalyst aggregation, and changes in light utilization during irradiation. Photocatalytic reaction rates are governed by coupled surface adsorption and light-induced charge-carrier processes; therefore, a single apparent rate constant may not adequately describe the entire reaction course. , Thus, the computed k _app should be interpreted as an apparent overall rate constant under the specific experimental conditions employed in this study. The moderate R² value and nonzero intercept indicate that the pseudo-first-order model only provides an approximate description of diclofenac potassium degradation over BiSI and does not reflect ideal first-order kinetics.

To contextualize the practical position of emerging chalcohalides in modern wastewater remediation, a quantitative appraisal of conventional photocatalytic frameworks reported for diclofenac degradation reveals a wide spectrum of operational demands and corresponding efficiencies. For instance, pristine titanium dioxide (TiO₂) consistently provides rapid elimination rates exceeding 95%, yet its functional window is severely bound to shortwave ultraviolet (UV) illumination (λ < 387 nm), which drastically increases the energetic requirements of the system. , When shifted under sustainable solar or visible light distributions, standard semiconductors often require protracted reaction windows or complex chemical modifications to achieve matching kinetics. Metal-organic frameworks (MOFs) and perovskites have demonstrated notable pollutant degradation rates (ranging from 80% to 92%) under visible irradiation, but they generally dictate the use of low initial contaminant metrics ∼10–20 mg/L) and are hampered by protracted reaction intervals extending past 180 min or the necessity of toxic co-catalyst loading. Graphitic carbon nitrides offer a highly scalable pathway, yet they frequently log moderate degradation capacities (∼75%–88%) over extended photolysis durations due to quick charge-carrier annihilation paths. Bismuth-based architectures like BiVO₄ or Bi₂WO₆ have bridged some of these visible-light bottlenecks, yielding up to 90% removal, but they still struggle to match the fast pseudo-first-order kinetics demanded by complex pharmaceutical streams without multi-step composite layering. Table shows quantitative benchmarking of representative photocatalytic systems for the degradation of diclofenac.

2. Comparative Performance of Representative Photocatalytic Systems Reported for Diclofenac Degradation under Different Irradiation and Reaction Conditions.

Photocatalyst	Light source	Reaction time	Initial diclofenac concentration	Degradation efficiency	Key operational consideration	Reference
TiO2 P25	150 W medium-pressure mercury lamp.	15 min	30 mg L–1	Complete removal	Prompt removal of 0.10 g L–1 TiO2 P25 at pH 6.2 required an artificial UV-rich source and immersion-well reactor design
Ag3PO4/g-C3N4 (30 mol % Ag3PO4 relative to g-C3N4	300 W Xe lamp, λ ≥ 400 nm	12 min	1 mg L–1	Complete DCF removal; 43.2% TOC removal	Use of 0.10 g L–1 catalyst resulted in partial reduction of Ag3PO4 to metallic Ag, but not total mineralization as DCF disappeared
Ag–BiOI–rGO nanocomposite	300 W halogen lamp, visible light	80 min	10 mg L–1	Approximately 99–100%	Ag and decreased graphene oxide modification are required; multicomponent synthesis and probable Ag leaching should be considered
g-C3N4 nanosheets	Four 10 W white light LEDs	150 min of irradiation	3 mg L–1	Apparent rate constant, k = 0.031 min–1 in tap water and 0.015 min–1 in ultrapure water	The catalyst loading was 0.65 g/L–1.Approximately 5% adsorption occurred during the 40 minute dark equilibration stage, and the photocatalytic rate depended heavily on the water matrix
NH2-MIL-125/g-C3N4 composite, MOF3N4 mass ratio = 50:50	UV-LED irradiation at 384 nm	120 min	10 mg L–1	Complete DCF disappearance	This technique used near-UV radiation rather than visible light. To increase activity, a Ti-based MOF was combined with g-C3N4
TiO2–carbon microspheres	Simulated solar light	180 min	5 mg L–1	Complete DCF disappearance; approximately 50% mineralization	Preparation involves carbonization and solvothermal TiO2 deposition, with a long irradiation time and low starting contaminant concentrations
Cu2ZnSnS4 nanoparticles	300 W xenon lamp fitted with a UV-exclusion filter; visible-light-only irradiation at 1 sun	120 min	10 mg L–1	80 ± 10% DCF degradation; final DCF concentration = 1.9 ± 0.2 mg L–1	A catalyst concentration of 1.0 g L–1 was utilized. Catalyst recycling was yet to be investigated, and the authors recognized optimization of catalyst loading and recyclability as important follow-up studies
BiSI nanorods	500 W halogen lamp with UV cutoff	120 min	1.0 × 10–4 M diclofenac potassium, approximately 33.4 mg L–1	95.2%	predominantly orthorhombic BiSI, narrow band gap (1.65 eV); robust visible harvest	This work

Table presents a comparison of representative semiconductor systems reported for diclofenac degradation, based on irradiation source, initial pollutant concentration, reaction time, and removal efficiency.

4. Conclusion

The synthesized BiSI nanorods exhibited strong visible-light-driven photocatalytic activity toward the removal of diclofenac potassium from aqueous solution. Structural and spectroscopic analyses confirmed the formation of crystalline orthorhombic BiSI with a band gap of 1.65 eV, indicating strong absorption across the visible spectrum and suitability for solar-energy-driven applications. Electron microscopy revealed uniform rod-like morphologies with high aspect ratios, providing abundant exposed active sites and facilitating efficient charge transport via shortened diffusion pathways.

Under visible-light irradiation, the BiSI photocatalyst achieved rapid and sustained removal of diclofenac potassium, reaching nearly complete removal within 135 min. The removal kinetics were approximately described by a pseudo-first-order model, suggesting that interfacial oxidation may contribute to the photocatalytic process. The enhanced photocatalytic efficiency is attributed to the favorable electronic structure of BiSI, which promotes effective photogenerated charge separation and supports the formation of reactive oxygen species responsible for eliminating pharmaceutical pollutants from aqueous solutions.

In summary, BiSI is identified as a promising bismuth-based photocatalyst for solar-driven water purification. Its strong visible-light response and high photocatalytic efficiency make it a viable candidate for the treatment of pharmaceutical contaminants in water. Future studies should focus on catalyst mineralization efficiency, and performance under real wastewater conditions to further evaluate its environmental applicability.

The authors declare no competing financial interest.