Visible-light-driven photocatalytic degradation of diclofenac in water by nanoparticulate Cu₂ZnSnX₄ (X = S, Se) kesterites with kinetics, transformation products and ecotoxicity

Abstract

Diclofenac (DCF) is a prevalent aquatic contaminant poorly removed by conventional wastewater treatment plants. This study investigates the photocatalytic degradation of DCF under simulated sunlight using nanoparticulate kesterite semiconductors Cu₂ZnSnS₄ (CZTS) and Cu₂ZnSnSe₄ (CZTSe), focusing on efficiency, stability, degradation pathways, and environmental implications. CZTSe exhibited superior photocatalytic performance, achieving 83% DCF removal within 60 min with a pseudo-first-order rate constant of 3.0 × 10^− 2 min^− 1, whereas CZTS reached 68% removal with 1.7 × 10^− 2 min^− 1. Both photocatalysts maintained stable activity after four reuse cycles. Total organic carbon analysis confirmed progressive DCF mineralization, i.e., the conversion of organic carbon into CO₂ and inorganic species. HPLC–MS analysis identified six transformation products generated by DCF degradation, corresponding to the conversion of the parent molecule into intermediate compounds. In silico ecotoxicity assessment using predictive computational models indicated reduced bioaccumulation potential of DCF after treatment; however, some transformation products exhibited higher environmental persistence and retained or increased predicted toxicity. Reactive species experiments demonstrated that superoxide radicals (•O₂⁻) were the main reactive species, followed by hydroxyl radicals (•OH). These findings demonstrate that CZTS and, particularly, CZTSe are efficient and reusable visible-light photocatalysts for DCF removal while providing new insights into degradation pathways, mineralization, reactive oxygen species, and transformation-product ecotoxicity, underscoring the importance of assessing degradation by-products in advanced oxidation processes.

Supplementary Information

The online version contains supplementary material available at https://doi.org/10.1038/s41598-026-67543-5.

Introduction

In recent years, the presence of emerging water pharmaceutical active micropollutants (EWPhAMPs) in aquatic environments has attracted increasing attention due to their widespread use and continuous release from human and veterinary applications^1–5. Although many pharmaceuticals have relatively short half-lives, their constant input into water bodies makes them pseudo-persistent contaminants^1,3,4,6. Even at trace concentrations, these biologically active compounds may adversely affect aquatic organisms and pose potential risks to human health^1–3.

Among those EWPhAMPs, diclofenac (DCF, 2-[(2,6-dichlorophenyl) amino] phenyl acetic acid, C₁₄H₁₁Cl₂NO₂, M = 296.15 g mol^{− 1}) (see Figure S1 for the chemical molecular structure of diclofenac), a member of the nonsteroidal ant-inflammatory group of drugs associated with pain-relieving action^7–9, is one of the most frequently detected pharmaceutical contaminants in aquatic environments^9–15. Its extensive consumption, incomplete metabolism in humans, and resistance to hydrolysis and biodegradation promote its continuous release into wastewater treatment plants (WWTPs), where conventional treatment processes fail to achieve complete removal^8,12,15. Consequently, DCF is commonly detected in surface waters and groundwaters at concentrations ranging from ng L^{− 1} to µg L^{− 1}, with reports even from remote regions such as Antarctica^8,11,16. Due to its persistence, bioaccumulation potential, and toxicity to aquatic organisms at environmentally relevant concentrations, DCF has been included in the European Union Water Framework Directive Watch List for environmental monitoring^{10–12,17–19}.

The occurrence of diclofenac (DCF) in aquatic environments is largely attributed to the limited removal efficiency of conventional wastewater treatment plants (WWTPs)^6,8,10,18,20. These facilities were not designed to eliminate pharmaceutical micropollutants, and conventional physical and biological treatment processes are generally ineffective because of the low concentrations and diverse chemical characteristics of these compounds^{6,8,10,15,18,20}. In fact, these conventional WWTPs generally reduce less than 25% of the initial DCF pharmaceutical micropollutant concentration present in aqueous media^10,13,18.

Over the last decade, considerable efforts have been devoted to developing alternatives to conventional wastewater treatment plants (WWTPs) for the removal of emerging waterborne pharmaceutical and antimicrobial micropollutants (EWPhAMPs), particularly diclofenac (DCF), from aqueous media due to their adverse environmental and health impacts⁸. To complement existing WWTPs and achieve regulatory limits, several tertiary treatment technologies have been investigated, including nanofiltration, reverse osmosis, ozonation, adsorption, and advanced oxidation processes (AOPs)^{8,13,15,17,20}. Among these, AOPs have attracted particular attention because of their high efficiency in degrading a wide range of pharmaceutical contaminants^{8,13,15,17,20,21}. These processes—including UV/H₂O₂, Fenton and photo-Fenton reactions, ozone-based treatments, sonolysis, electrochemical oxidation, and photocatalysis—share the common mechanism of generating reactive oxygen species (ROS), particularly hydroxyl radicals (•OH), which oxidize organic pollutants; they mainly differ in the pathway used to produce these reactive species^{3,8,13,15,17,20,21}.

Among the advanced oxidation processes (AOPs) above discussed, semiconductor-based photocatalysis is anticipated to become the predominant method for wastewater treatment, owing to its advantageous features including facile synthesis, eco-friendly operation, ability for complete mineralization, and potential for reuse^11,22–27. The photocatalytic degradation process is considered a straightforward and cost-effective treatment method, as it enables the complete transformation of active pharmaceutical ingredients into less harmful by-products through efficient oxidation mechanisms, utilizing solely a semiconductor photocatalyst under light irradiation^{3,11,22,25–32}. To date, various semiconductor photocatalysts such as TiO₂³³, ZnO^7,34,35, SrTiO₃²², g-C₃N₄/BaBiO₃³⁶, Zn_1−x−yAl_xLa_yO¹⁶, V₂O₅³⁴ and CeFeO₃²⁸ have been applied for the photodegradation of DCF micropollutant in aqueous solution. Nevertheless, several widely studied photocatalysts, particularly TiO₂, ZnO, and SrTiO₃, exhibit limited efficiency because they predominantly absorb only ultraviolet (UV) light—which constitutes less than 4% of the solar spectrum—and also suffer from significant charge carrier recombination, thereby impeding their photocatalytic performance^23–26,37. So, all these facts hamper their adoption in industry for treatment of wastewaters^23–26,37.

In order to overcome these limitations, research is primarily focused on photocatalysts driven by visible light that use narrow bandgap semiconductors that exhibit optical absorbance across the visible spectrum^23–26. The semiconductor kesterite family constituted by: sulfur-based kesterite Cu₂ZnSnS₄ (CZTS), selenium-based kesterite Cu₂ZnSnSe₄ (CZTSe), and their alloys (Cu₂ZnSn(S_x,Se_1−x)₄ (CZTSSe), where 0 ≤ x ≤ 1), has recently gained popularity in photocatalytic degradation reactions to remove different pollutants from wastewater due to its good chemical and optical properties as a semiconductor^23,24. In fact, excellent optoelectronic qualities have been demonstrated by those kesterite semiconductor materials, such as broad optical absorbance over the entire visible region with high absorption coefficient (> 10⁴ cm^{− 1}) and optimal optical bandgap (0.9–1.6 eV), allowing the material to harvest maximum photons, good photo-stability and multi-dimensional symmetric carrier transport^23,24. Both, CZTS and CZTSe semiconductor materials have been successfully used as photocatalysts in the mineralization of different water pollutants such as: Congo red^23,24, methylene blue³⁸, methyl orange³⁹ and Rhodamine B⁴⁰ dyes. Recently, Tseberlidis et al. reported the photocatalytic activity of Cu₂ZnSnS₄ nanoparticles in the degradation of diclofenac in water under UV-visible and visible light provided both by a 300 W Xenon lamp⁴¹.

In our laboratories, both semiconductor CZTS and CZTSe kesterite based nanomaterials have been successfully developed through the hydrothermal and solvothermal synthesis, respectively, and further demonstrating their photocatalytic activity and effectiveness photocatalytic behavior in the photodegradation of organic Congo red azo dye under solar simulation irradiation^23,24. Nanomaterials exhibit considerable benefits in photocatalytic processes, attributable to their distinctive physicochemical characteristics, including enhanced specific surface area and superior electron transport efficiency^23,24. Moreover, a substantial part of photocatalysis costs stems from the energy demand of artificial radiation sources^42–44. This cost can be significantly reduced by employing natural sunlight, especially the visible spectrum, which constitutes the majority of solar irradiance^42–44. Recent literature emphasizes the development of visible-light and solar-driven photocatalysts as a sustainable alternative to UV-lamp-based systems (e.g., photocatalytic solar energy conversion systems that harness visible light rather than only UV-active processes)^42–44. Photocatalytic degradation under AM 1.5 is essential because it evaluates true solar applicability, while UV-only experiments often overestimate performance and lack environmental relevance^42–44.

In this work, we present a comprehensive comparative assessment of kesterite-type semiconductors, CZTS and CZTSe, as efficient photocatalysts for the degradation and mineralization of diclofenac (DCF) in aqueous solution under simulated solar irradiation. Notably, and to the best of our knowledge, this is the first report addressing the photocatalytic mineralization of DCF employing nanoparticulate CZTS and CZTSe under solar-simulated conditions. Both materials exhibit outstanding performance, with CZTSe achieving 84% DCF degradation within 60 min, significantly outperforming CZTS (68%). This enhanced activity is further reflected in the mineralization efficiencies, reaching 34.2% for CZTSe and 24.3% for CZTS. Beyond degradation efficiency, the transformation products formed during the photocatalytic process were systematically identified, providing key insights into the degradation pathways. Importantly, the environmental relevance of these intermediates was evaluated through in silico toxicological predictions using VEGA software, encompassing mutagenicity, developmental toxicity, biodegradability, persistence in water, sediment and soil, and acute toxicity toward aquatic organisms (Daphnia magna, LC50, 48 h). These findings highlight the strong potential of kesterite-based photocatalysts for sustainable water remediation applications.

Experimental

Semiconductor photocatalysts synthesis

Cu₂ZnSnSe₄ (CZTSe) and Cu₂ZnSnS₄ (CZTS) nanoparticles were synthetized employing solvothermal and hydrothermal methods, respectively. The following chemical reagents were employed: copper chloride (CuCl₂, Winkler P.A., 99%), zinc chloride (ZnCl₂, Sigma-Aldrich P.A., 99%), tin chloride (SnCl₂, Winkler P.A., 99%), thiourea (CH₄N₂S, Merck P.A., 99%), elemental selenium (Se, Merck P.A., 99%), and ethylenediamine (en, Sigma-Aldrich P.A., 99%). The precursor solutions were individually transferred to Teflon vessels, which were then placed inside a stainless-steel reactor for hydrothermal or solvothermal treatment according to the in-detail protocols reported in our previously published works^23,24.

Physicochemical characterization of the grown CZTS and CZTSe nanoparticles

The physicochemical characterization techniques, procedures, and instrumentation used for the characterization of the CZTS and CZTSe nanoparticles were identical to those previously reported in our earlier works, which are cited accordingly below^23,24. Therefore, only a brief description is provided here.

The structural characterization of the CZTS and CZTSe phases was performed by X-ray diffraction (XRD) using a Bruker D8 ADVANCE diffractometer. The operating conditions were: CuKα radiation (30 mA, 40 kV, λ = 0.15418 nm), in Bragg-Brentano 45° mode, with a step of 0.01° and a step time of 34 s at room temperature^23,24.

Raman spectra were obtained using a Witec Alpha 300 confocal Raman microscope system equipped using an excitation laser wavelength of 785 nm and an electrically cooled CCD camera. The signal was calibrated using the 520 cm^{− 1} line of a Si wafer and a 20x objective. The laser power on the samples was 2 mW. The resolution was set to 4 cm^{− 1} and 10 scans with an integration time of 1 s were performed. Peak parameters have been obtained by least square Lorentz fitting using Origin 8.0 software^23,24.

The chemical composition of the CZTS and CZTSe samples was studied via X-ray photoelectron spectroscopy (XPS) using a Physical Electronics (PHI) VersaProbe II spectrometer equipped with an Al Kα radiation source (1486.6 eV and 47.3 W)^23,24.

Field emission scanning electron microscopy (FE-SEM) images of the CZTS and CZTSe samples were obtained on a Helios Nanolab 650 Dual Beam equipment from FEI Company. For this, the powders were supported on a conductive carbon tape inside the vacuum chamber. The analysis of the chemical composition of the formed structures was carried out using X-ray energy dispersion spectrometry (EDS). The equipment used was a QUANTAX 200 model from Bruker with XFLASH (EDS coupled to SEM equipment: Hitachi SEM SU-3500 of variable pressure with a detector 410-M). Samples for TEM were ultrasonically dispersed in 1 ml of ethanol. A small drop of the suspended solution was placed on a porous carbon film on a nickel screen and allowed to air dry. Transmission electron microscopy (TEM) and High-resolution transmission electron microscopy (HRTEM) images were obtained on a Talos F200X instrument^23,24.

The surface area of the samples using the Micromeritics 3FLEX instrument for the volumetric adsorption-desorption isotherm of N₂ at − 196 °C, using 0.05 g of each sample, which were degassed for 6 h at 180 °C, under vacuum using a Micromeritics SmartVacPrep instrument. The surface area was determined using the Brunauer-Emmett-Teller (BET) equation in the relative pressure range of 0.05 ≤ P/P₀ ≤ 0.25. The total pore volume was defined as the pore volume of a single point at P/P₀ = 0.99. The average pore diameter (d_p) was obtained from the desorption branch using the Barrett-Joyner-Halenda (BJH) method^23,24.

The optical properties were studied by UV-VIS molecular absorption spectra through the transmittance spectrum, using a SHIMADZU UV-2600 Spectrophotometer with a PC connection. The measurement range was from 400 nm to 900 nm at room temperature, with a scanning speed of 0.2 nm s^–1, 10 mg of CZTS or CZTSe nanoparticles suspended in ethanol were prepared^23,24.

Study of the DCF adsorption-desorption onto the different semiconductor photocatalysts under study

A study of adsorption-desorption of sodium diclofenac onto the CZTX photocatalyst (X: S or Se) was conducted. To this end, different concentrations of CZTS and CZTSe photocatalysts, with amounts of 55, 65, 105, and 175 mg L^− 1 for CZTS, and 75, 125, and 175 mg L^− 1 for CZTSe have been assayed. A 5 mg L^− 1 solution of diclofenac (DCF) was prepared, and the corresponding photocatalyst was added, maintaining constant agitation for 60 min in the dark. The adsorption-desorption process was monitored under dark conditions by extracting aliquots at regular intervals of 15 min. The samples were analyzed using UV-Visible spectrophotometry. The adsorption percentage of diclofenac (DCF) onto kesterite nanomaterials was determined based on the change in its concentration in solution before and after the adsorption process. In this method, (C_o) represents the initial concentration of diclofenac prior to its contact with the kesterite nanomaterials, while (C_t) corresponds to the remaining concentration of diclofenac in solution after a given contact time (t). The adsorption percentage (%Ads) was calculated using the following equation:

1

where C₀ corresponds to the initial concentration of diclofenac in solution prior to its adsorption onto the kesterite nanomaterials, and C_t corresponds to the remaining concentration of diclofenac in solution after being adsorbed onto these kesterite nanomaterials for a certain time t, where (C_o-C_t) represents the amount of diclofenac removed from the solution due to adsorption onto the surface of the kesterite nanomaterials.

Photodegradation of diclofenac sodium (DCF)

The photocatalytic experiments have been carried out in a Batch-type glass reactor, using 200 ml of a 5 mg L^− 1 pharmaceutical grade diclofenac solution and a CZTX (X: S or Se) nanoparticles load of 35 mg. The mixture was stirred in the dark for 30 min to generate an adsorption-desorption equilibrium followed by irradiation with a AM 1.5 G solar simulator (1000 W) ABET Technologies Model 11,002 SunLite (Figure S2 for this solar simulator spectrum).

The photodegradation was recorded by a HPLC with UV detector Agilent Technologies 1100 series with a C18 Kinetex 5 μm 100 Å 150 × 4.6 mm column. The wavelength of the UV detector was set at 276 nm, and mixture of 60% methanol (Panreac) and 40% formic acid (Panreac) at isocratic was employ as mobile phase. The flow was maintained at 1 mL min^− 1 and the injection were of 20 µL. Prior to photodegradation, a calibration curve of DCF was performed, and the Eq. 2 was employed to determine the photocatalytic degradation efficiency:

2

where C₀ corresponds to the initial concentration of diclofenac prior to exposure to irradiation and C_t corresponds to the concentration of diclofenac after being exposed to solar irradiation for a certain time t. Each test was conducted in triplicate, keeping the first 30 min in darkness for adsorption and then the next 60 min for photocatalytic degradation.

Photocatalyst regeneration study

The cyclability of the CZTX photocatalyst (X: S or Se) has been evaluated maintaining the experimental parameters described above. After each cycle, the nanoparticles were collected and washed with ethanol several times, dried for 30 min at 60 °C and then reused for subsequent cycles.

Total organic carbon (TOC)

The degree of mineralization of DCF photodegraded was monitored from the TOC through a Shimadzu total organic carbon analyzer (TOC-L).

The degree of sample mineralization was determined from the formula⁴⁵:

3

where MR is Mineralization rate (%); TOC_i is the TOC value before photocatalytic process, initial concentration of TOC in the sample (mg L^− 1); TOC is TOC value after the photocatalytic process (mg L^− 1).

HPLC MS and ion exclusion chromatography

A HPLC JASCO system (JASCO Corp., Tokyo, Japan) accoupled to mass spectrometer (ESI-MS) LCMS-2020 (Shimadzu Corp., Kyoto, Japan) was employed to detection and identification of DCF transformation products in the photocatalytic degradation process.

Toxicity assessment

Environmental impact and the toxicity of DCF transformation products generated during DCF photocatalytic degradation was performed using in silico QSAR models implemented in VEGA. The following endpoints were evaluated: mutagenicity (Ames CONSENSUS v1.0.4), developmental toxicity (CAESAR v2.1.8), biodegradability (IRFMN v1.0.10), bioconcentration factor (BCF, Arnot–Gobas v1.0.1), environmental persistence in water, sediment, and soil (IRFMN v1.0.1), and aquatic toxicity (LC50 in Fathead minnow and Daphnia magna).

Prediction reliability was assessed using the applicability domain index (ADI, 0–1), which integrates structural similarity, prediction accuracy, and concordance with training set compounds. Biodegradability was classified as PRB (possibly readily biodegradable), PnonRB (possibly non-readily biodegradable), and nonRB (non-readily biodegradable), while reliability was categorized as LR (low), MR (medium), and GR (good).

Radical scavenging assay

Three different radical scavengers were employed for the study of generated oxidizing or reducing species in photocatalysis process. Ethylenediaminetetraacetic acid EDTA-2Na (Merck P.A.), Potassium dichromate (K₂Cr₂O₇) (Merck) and 2-propanol (IPA, Scharlab S.L.) were employed to inhibits holes in the valence band (h_vb⁺), electrons in the conduction band (e_cb^–) and the hydroxyl radicals (•OH), respectively. The scavenger concentration was set at 592 times the initial DCF concentration to ensure that competitive degradation would prioritize the scavenger, assuming the photocatalytic mechanisms targeted the intended ROS. For this, 2 mM of each scavenger was added to the photocatalysis system to evaluate the response in the presence or in the absence of different scavengers. Additionally, the effect of the presence of oxygen in the medium was evaluated by replacing it by argon (Ar). For the test with Ar, the solution was bubbled for 15 min and the atmosphere was maintained for the duration of the photocatalysis process. Each test was conducted in triplicate, keeping the first 30 min in darkness for adsorption and then the next 60 min for photocatalytic degradation.

Results and discussion

Structural, morphological and surface chemical study

In order to determine structural and morphological properties, and also chemical surface composition, the prepared sulfur-based kesterite Cu₂ZnSnS₄ (CZTS) and selenium-based kesterite Cu₂ZnSnSe₄ (CZTSe) samples have been characterized by X-ray diffraction (XRD), Raman spectroscopy, X-ray photoelectron spectroscopy (XPS), field emission-scanning electron microscopy (FE-SEM) and transmission electron microscopy (TEM).

Figure 1a and b shows the typical XRD patterns of hydrothermally grown CZTS and solvothermally grown CZTSe nanoparticulate samples, respectively, along with the corresponding standard tetragonal kesterite reference patterns (JCPDS #04-023-6315 for CZTS⁴⁶ and #00-052-0868 for CZTSe⁴⁷. The CZTS sample exhibits four main diffraction peaks at 2θ = 28.6°, 32.9°, 47.6° and 56.3°, indexed to the (112), (200), (220)/(204) and (132)/(116) planes of tetragonal crystalline kesterite structure of CTZS, while the CZTSe sample shows five intense peaks at 2θ = 27.1°, 45.0°, 53.4°, 65.6° and 72.5°, corresponding to the (112), (204), (312), (400)/(008) and (316) planes of tetragonal crystalline kesterite structure of CTZSe (Fig. 1a and b, respectively). Additional weak diffraction peaks observed in both patterns can also be assigned to the respective kesterite phases. These results confirm the formation of well-defined kesterite CZTS and CZTSe phases with polycrystalline nature. Moreover, XRD patterns reveal the presence of very weak diffraction peaks attributed to a minor SnO₂ secondary phase in both samples (Fig. 1a and b). In addition, the noticeable broadening of the CZTS and CZTSe diffraction peaks demonstrate their polycrystalline in nature. The average crystallite sizes, estimated from the (112) reflection using the Scherrer equation based on the full width at half maximum (FWHM)^48,49, were approximately 18 nm for CZTS and 35 nm for CZTSe, inferring their nanocrystalline character (please see Supporting Information, Eq. S1 for crystallite size calculation by using Scherrer equation).

Fig. 1.

(a, b) X-ray diffraction (XRD) patterns of nanoparticulate Cu₂ZnSnS₄ (CZTS) and Cu₂ZnSnSe₄ (CZTSe). Indexed reflections correspond to the kesterite phase and are compared with reference JCPDS patterns (#04-023-6315 and #00-052-0868, respectively, indicated by vertical thick blue bars). The cassiterite SnO₂ JCPDS pattern (JCPDS file No. 41-1445) is also provided for comparison (vertical thick green bars). Peaks marked with a green symbol (●) indicate the presence of cassiterite SnO₂. (c, d) Raman spectra (λ_exc = 785 nm) of CZTS and CZTSe nanopowders, showing characteristic vibrational modes; the main bands were fitted using two Lorentzian components (overall fit in cyan, individual contributions in red). (e, f) FE-SEM images of CZTS and CZTSe nanoparticles; higher-magnification views are shown as insets in (a, b). (g, h) TEM images of representative CZTS and CZTSe nanoparticulate samples. A typical squared shaped grain is indicated by yellow lines in panel (h).

Due to the structural similarity between kesterite Cu₂ZnSnS₄ (CZTS) or Cu₂ZnSnSe₄ (CZTSe) and their possible secondary phases, such as ZnS, Cu₂SnS₃, Cu_xSe, Cu₂SnSe₃ and ZnSe, the main X-ray diffraction peaks of these compounds often overlap, making phase identification by XRD alone inconclusive^50–55. Therefore, Raman spectroscopy was employed as a complementary technique to reliably identify secondary phases and confirm phase purity (Fig. 1c and d)^50–55. The first-order Raman spectra were analyzed by Lorentzian peak deconvolution in the 200–400 cm^{− 1} region for CZTS (Fig. 1c) and 150–230 cm^{− 1} for CZTSe (Fig. 1d). For CZTS, characteristic peaks at ~ 285 and ~ 339 cm^{− 1}, corresponding to the A and A₁ vibrational modes of kesterite CZTS, respectively, were observed^50,56,57, with no additional peaks associated with ZnS, Cu₂SnS₃, Cu₂S, SnS or SnS₂^50,58. Similarly, the Raman spectrum of CZTSe revealed two broad peaks centered at ~ 173 and ~ 190 cm^{− 1}, attributed to kesterite CZTSe, with the most intense A₁ mode at 190 cm^{− 1} related to purely anionic selenium vibrations^54,59,60. The absence of Raman signatures associated with impurity phases, such as Cu₂SnSe₃ or SnSe_x, confirms the high purity and single-phase nature of the synthesized materials, while the peak broadening is ascribed to phonon confinement effects, lattice defects, strain and reduced crystallinity inherent to the nanostructured samples^52,54,59,60. No Raman features attributable to impurity phases were detected, either within the narrow spectral window discussed in the main text or in the extended Raman spectra shown in Figure S5.

X-ray photoelectron spectroscopy (XPS) was employed to investigate the elemental composition and oxidation states of the synthesized nanoparticulate Cu₂ZnSnS₄ (CZTS) and Cu₂ZnSnSe₄ (CZTSe) samples. Survey spectra (Supporting Information, Figures S3a and S4a, respectively) confirmed the presence of Cu, Zn, Sn and the corresponding chalcogen (S or Se), together with minor C, O and N signals attributed to environmental contamination and residual CH₄N₂S precursor. High-resolution spectra of the relevant core levels (Cu 2p, Zn 2p, Sn 3d, S 2p or Se 3d), calibrated to the C 1s peak at 284.8 eV, were used to assess the chemical states (Supporting Information, Figures S3b-S3e and S4b-S4e for CZTS and CZTSe, respectively). In both samples, the Cu 2p spectra exhibited main components at ~ 931.6–933.7 eV with no shake-up satellites, indicating the presence of Cu⁺ and the absence of Cu²⁺, further supported for CZTS by the modified Auger parameter analysis of the Cu LMM transition (Supporting Information, Figure S3b). Zn 2p peaks centered at ~ 1021.4–1021.5 eV were assigned to Zn²⁺ species. The Sn 3d peaks at ~ 486.0–486.7 eV were assigned to Sn⁴⁺, as expected for the kesterite structure. Since Sn⁴⁺ in CZTS/CZTSe and SnO₂ exhibits similar binding energies, a minor surface SnO₂ contribution cannot be ruled out based solely on the Sn 3d spectrum⁶¹. For CZTS, the S 2p signal shows the characteristic spin–orbit doublet of sulfide sulfur in CZTS, with S 2p_3/2 and S 2p_1/2 components located at 161.3 and 162.6 eV, respectively. These peaks confirm the presence of S²⁻ in the Cu₂ZnSnS₄ lattice. A minor component at 168.8 eV, can be assigned to oxidized sulfur species (SO_x/sulfates), probably related to residual sulfur precursors or surface oxidation (Supporting Information, Figure S3e), whereas in CZTSe the dominant Se signal at ~ 54.3 eV was characteristic of metal selenides, with a minor SeO₂ contribution at higher binding energy (Supporting Information, Figure S4e). Overall, the identified valence states (Cu⁺, Zn²⁺, Sn⁴⁺ and S²⁻/Se²⁻) are consistent with the expected stoichiometry of kesterite-type CZTS and CZTSe⁶².

The morphological and microstructural features (i.e.: detailed microstructure, size and shape morphologies) of the Cu₂ZnSnS₄ and Cu₂ZnSnSe₄ samples have been further investigated by field-emission scanning electron microscopy (FE-SEM) and transmission electron microscopy (TEM) (Fig. 1e and h). Figure 1e and f show low-magnification FE-SEM micrograph images of typical synthesized Cu₂ZnSnS₄ and Cu₂ZnSnSe₄ samples, respectively. The FE-SEM micrograph of Fig. 1e (corresponding to a typical nanostructured CZTS sample), shows a heterogeneous, porous surface with densely packed, irregular sub-micrometre grains forming 1–3 μm agglomerates. The rough, polycrystalline aggregates display intergranular voids, high surface roughness, and no preferred grain orientation. A high-magnification FE-SEM micrographs of these nanoparticulate sample is depicted as inset in this Fig. 1e, showing agglomerated, irregular polyhedral grains with sizes of approximately 1–3 μm. The grains are loosely packed, forming a porous microstructure with pronounced intergranular voids. Grain surfaces appear rough and partially faceted, with fine secondary particulates, while no evident cracks are observed. On the other hand, the FE-SEM micrograph of Fig. 1f (corresponding to a typical nanostructured CZTSe sample), shows a rough and heterogeneous surface composed of irregular, submicron-sized kesterite grains agglomerated into micrometer-scale clusters. The structure is loosely packed, with noticeable intergranular porosity and a high degree of surface roughness, indicative of a polycrystalline and porous morphology. The inset of Fig. 1f shows a high-magnification FE-SEM micrographs of this CZTSe sample, showing well-defined faceted micron-sized crystallites embedded in a finer-grained matrix. The microstructure exhibits angular plate-like grains, surface roughness, and intergranular porosity, indicative of anisotropic crystal growth and incomplete densification. The high-magnification TEM micrograph of a typical nanoparticulate Cu₂ZnSnS₄ cluster shown in Fig. 1g bright-field TEM image reveals thin, electron-transparent kesterite particles composed of multiple nanocrystalline domains with sizes of several tens of nanometers. Contrast variations arise from thickness differences and diffraction contrast between misoriented grains. Grain boundaries and defect-related contrast are visible, confirming a polycrystalline microstructure, while no distinct secondary phases are observed within the resolution of the image. High-magnification TEM micrograph of a typical nanoparticulate Cu₂ZnSnSe₄ cluster depicted in Fig. 1h shows agglomerated, irregularly shaped kesterite crystallites with sizes ranging from several tens to a few hundred nanometers. Contrast variations arise from thickness and orientation differences, indicating a polycrystalline microstructure with numerous grain boundaries and intergranular interfaces. Moreover, the squared-shaped feature highlighted by the yellow lines corresponds to a faceted kesterite grain, suggesting anisotropic crystal growth and well-defined crystallographic planes within the polycrystalline aggregate. Moreover, to further confirm the presence of all constituent elements in the synthesized nanopowders, EDS elemental mapping was performed for both Cu₂ZnSnS₄ and Cu₂ZnSnSe₄ samples. The corresponding elemental maps are provided in the Supporting Information (Figure S6). The results clearly confirm the presence of Cu, Zn, Sn, and S in Cu₂ZnSnS₄, and Cu, Zn, Sn, and Se in Cu₂ZnSnSe₄. In addition, the elemental maps reveal a homogeneous distribution of all elements throughout the samples, indicating good compositional uniformity.

Surface area of the Nanoparticulate Cu₂ZnSnS₄ and Cu₂ZnSnSe₄ samples

To study the porous nature of the nanoparticulate CZTS and CZTSe samples under study, nitrogen adsorption-desorption isotherms have been performed at -196 °C after outgassing the sample at 180 °C for 6 h; the results are shown in Fig. 2a and b. From these graphs it can be appreciated that the adsorption-desorption isotherms for both kesterite samples exhibit hysteresis loop like nature, which is the main feature of porous materials⁶³. Based on the Brunauer–Deming–Deming–Teller (BDDT) framework, most physisorption isotherms can be categorized into six distinct types^64,65. Additionally, the morphology of hysteresis loops provides valuable insights into the corresponding pore structures^64,65. According to this classification, both the CZTS (please see Fig. 2a) and CZTSe (Fig. 2b) samples exhibit a Type IV adsorption isotherm accompanied by a Type H3 hysteresis loop^64,65. The type IV isotherm confirms the mesoporous nature of the material, while the H3 hysteresis loop suggests the presence of slit-shaped pores originating from interparticle voids generated by aggregates of plate-like particles^64,65. Specific surface area of these samples has been obtained by using the Brunauer-Emmet-Teller (BET) gas adsorption method^64,65. The CZTS and CZTSe samples exhibit a BET surface area of ca. 29.5 m²/g ca. and 7 m²/g, respectively.

Fig. 2.

(a, b) Nitrogen adsorption–desorption isotherms at 77 K for nanocrystalline Cu₂ZnSnS₄ (CZTS) and Cu₂ZnSnSe₄ (CZTSe) (adsorption: blue symbols; desorption: red symbols). (c, d) UV–vis absorbance spectra of CZTS and CZTSe samples; insets show Tauc plots (αxhν)² vs. hν Tauc plots used to estimate the band gap energy (E_g), with linear fits (red dashed lines). (e, f) Diclofenac adsorption onto CZTS and CZTSe nanoparticulate photocatalysts as a function of kesterite dose. Error bars show the standard deviations of the measurements taken from three independent experiments.

Optical characterization

One of the most critical parameters in predicting the photocatalytic performance of a semiconductor is the determination of the optical bandgap energy (E_g). For this purpose, UV-visible spectrophotometry has been employed to analyze the absorption spectrum of both CZTS and CZTSe nanostructured samples. Figure 2c and d depict the optical absorption spectrum of typical nanoparticulate Cu₂ZnSnS and Cu₂ZnSnSe₄ samples, respectively. As shown in these Fig. 2c and d, both the prepared sulfur-based kesterite and selenium-based kesterite samples absorb light over the entire range of visible light wavelength and the absorbance even extends to larger wavelengths. So, this broad and strong optical absorbance makes this material an advantageous and suitable candidate for photocatalytic applications. The optical bandgap energy (E_g) and optical transition type of these nanoparticulate kesterite samples under study were determined using the Stern relation for near-edge absorption^26,49,66,67(please see Supporting Information Sect. 7, Eq. S2). For both chalcogenides, the absorption edge is well fitted by this model, indicating direct allowed electronic transitions (please see insets of Fig. 2c and d depicting the Tauc plots for the CZTS and CZTSe samples, respectively). These Tauc plots of (αhν)² versus hν yield a direct bandgap of approximately 1.5 eV for CZTS, whereas a significantly narrower bandgap of about 1.0 eV is obtained for CZTSe, consistent with reported literature values^{52,54,59,60,68–71}. The reduced bandgap and extended light-harvesting capability of CZTSe relative to CZTS highlight the impact of chalcogen substitution (S → Se) on the optoelectronic properties and underline the potential advantages of CZTSe for visible-light-driven applications^{52,54,59,60,68–71}.

Adsorption and photocatalytic activity performance

Photocatalysis is a surface driven process, and the micropollutant molecules must be close enough to the photocatalyst surface in order to react with the short live reactive oxygen species (ROS)^72–74. In fact, to allow the transfer of photogenerated electrons and holes and in turn begin a redox reaction, relevant reactant species firstly need to be adsorbed onto a semiconductor photocatalyst^72–74. So, as the photocatalytic reaction occurs in the adsorbed phase, the adsorption ability of a photocatalyst plays a role that is likely to have a great influence on the overall efficiency^72–74. The contribution of adsorption equilibria to the degradation results has been studied for DCF removal using either CZTS or CZTSe nanoparticles from 55.0 to 175 mg L^{− 1} and 75.0 to 175.0 mg L^{− 1}, respectively, at a constant DCF concentration of 5 mg L^{− 1}. Before irradiation the suspension must reach the adsorption-desorption equilibrium the adsorption-desorption equilibrium is reached within 30 min. Thus the photocatytic experiments could start after 30 min of adsorption process. Figure 2e and f show the degree of DCF adsorption results on both kesterite materials under study (after 60 min), showing very poor DCF adsorption onto both these CZTS and CZTSe photocatalysts. Moreover, these results show that the DCF adsorption degree onto both kesterite material surfaces increases with the rise in the dose of both CZTS and CZTSe nanoparticles. In both systems, the maximum adsorption rate was achieved at a 175 mg L^{− 1} of kesterite nanoparticles. This can be attributed to the increase of the surface area, which provides more active adsorption sites⁷⁵. Moreover, at this kesterite nanoparticle concentration, an adsorption capacity of 30% and 20% onto the CZTS and CZTSe photocatalyst surfaces, respectively, can be appreciated (Fig. 2e and f). This behavior can be related to the higher surface area exhibited by CZTS (29.4 m² g^{− 1}) in comparison to that of CZTSe (7.0 m² g^{− 1}), as has been previously reported (please see above).

The photocatalytic degradation of DCF has been evaluated by employing the two nanoparticulate CZTX-based photocatalysts (X: S or Se), in the presence of low DCF concentration aqueous solution, under AM1.5G simulated sunlight (100 mW cm^{− 2}) illumination. Moreover, a comparison of their performance as photocatalysts on the photodegradation of DCF has been made. However, to eliminate the effect of adsorption on catalytic performance and obtain the adsorption-desorption equilibrium (adsorption process), all photocatalytic experiments are first carried out for 30 min under dark conditions in the presence of the two nanoparticulate CZTX-based photocatalysts under study. As observed from Fig. 3a and b, after 30 min of constant stirring in the dark (equilibrium was reached within 30 min), both DCF solutions containing each one the respective CZTX-based photocatalyst reveal a slight decrease in the C/C_o ratio, which is attributed to the adsorption effect of the DCF molecules on the surface of the photocatalyst^39,75–77. Here, and throughout this work, the error bars represent the standard deviation of the mean calculated from three independent batches. It can be seen that about 28.6% and 19.8% of DCF as been absorbed by the nanoparticulate Cu₂ZnSnS₄ and Cu₂ZnSnSe₄ photocatalyst, respectively, within 30 min. The greater DCF adsorption amount onto the nanoparticulate Cu₂ZnSnS₄ photocatalyst than onto the Cu₂ZnSnSe₄ one is in line with that displayed in Fig. 2e and f and discussed above. Figure 3a and b compare the removal efficiency of DCF (5.0 mg L^{− 1}) under illumination without using the nanoparticulate CZTX-based photocatalysts (blank experiment, photolysis process), and in the presence of both nanoparticulate CZTX-based photocatalysts (photocatalytic degradation); panels (a) and (b) depict the obtained results with the presence of CZTS and CZTSe, respectively. The blank experiment shows that the DCF concentration did not exhibit an appreciable change with increasing irradiation time up to 60 min (only a 2% degradation can be seen), which indicated the negligible effect of the photolysis on DCF removal (Fig. 3a and b). Further, in the presence of the respective kesterite photocatalysts and under visible light irradiation, both photocatalysts systems show obvious DCF degradation activity because the DCF pollutant shows a sharp quasi-linear C/C_o decline with the reaction time and obviously different from dark conditions. Moreover, a significantly difference in the degradation efficiency between the two CZTX-based photocatalysts under simulated solar irradiation can be appreciated (Fig. 3a and b). In fact, it can be seen that a ca. 68% of DCF photodegradation has been reached after 60 min in the presence of CZTS photocatalyst, whereas when CZTSe has been employed, a DCF degradation of ca. 84% can be obtained for the same time of exposure to simulated sunlight, these values are also listed in Table 1. The better performance of CZTSe photocatalyst compared to the performance of CZTS one can be attributed to photocatalyst CZTSe exhibiting a lower bandgap value (1.0 eV) compared to that of CZTS (1.5 eV) (please see above), which allows CZTSe to harvest a greater proportion of the solar spectrum. As has been said above in the Experimental section, in all these DCF photocatalytic degradation studies, the DCF abatement evolution with solar simulator irradiation time has been monitored by evaluating the DCF concentration in solution through UV-vis absorbance spectroscopy, and results are depicted in Figure S7 for both CZTS and CZTSe photocatalysts. The spectra exhibit only the characteristic DCF absorption band at 275 nm and its progressive attenuation. Notably, the absence of both the peak at 235 nm and the broad shoulder in the 310–350 nm region — typically assigned to carbazole derivatives in the literature — indicates that such intermediates were not formed under the present conditions⁴¹. It must be pointed out that the carbazole family degradation products exhibit comparable toxicity and similar persistence in the environment as that of DCF⁴¹.

Fig. 3.

Photocatalytic degradation of diclofenac in aqueous solution using nanoparticulate kesterite photocatalysts. (a, b) Relative concentration (C/C_o) as a function of illumination time in the absence (circles) and presence (squares) of Cu₂ZnSnS₄ (CZTS) and Cu₂ZnSnSe₄ (CZTSe), under AM1.5G simulated sunlight (100 mW cm^-2). (c, d) Corresponding first-order kinetic plots for CZTS and CZTSe over 60 min of illumination. (e, f) Recyclability of CZTS and CZTSe photocatalysts under the same irradiation conditions. Error bars show the standard deviations of the measurements taken from three independent experiments.

Table 1.

Photocatalytic degradation efficiency of DCF, reaction first-order rate constant (k) and the coefficient of determination (R²) values obtained onto the two different kesterite systems under study.

Semiconductor photocatalyst	Efficiency (%)	ka (min− 1)	R 2a
Cu2ZnSnS4 (CZTS)	68.0	1.7 × 10− 2	0.8502
Cu2ZnSnSe4 (CZTSe)	84.0	3.0 × 10− 2	0.9631

^aObtained from plots depicted in Fig. 3c and d.

When comparing the DCF removal efficiency results obtained through the use of CZTS and CZTSe nanostructured photocatalysts with those reported in the literature for other photocatalysts, significant differences in performance are observed. For instance, in the photodegradation of diclofenac using TiO₂, only 25% efficiency is achieved after 60 min of UV light exposure⁷⁸. In contrast, when employing a heterojunction such as TiO₂/g-C₃N₄, the degradation efficiency reaches 35% within the same period, significantly outperforming TiO₂ alone⁷⁸. Other heterojunctions, such as TiO₂-SnO₂ or TiO₂-CdS, achieve DCF degradation rates of 40–45% at 60 min^79,80. These comparisons suggest that TiO₂ demonstrates lower efficiency compared to CZTX (X: S or Se) photocatalysts, and even when incorporated into heterojunctions, it only approaches the percentage obtained with CZTS but does not match the efficiency of CZTSe, which exceeds 80% for DCF elimination.

The kinetics of the photocatalytic reactions corresponding to the DCF photodegradation process for both CZTX-based photocatalysts has been further studied. Langmuir–Hinshelwood kinetics is the most commonly used kinetic expression to explain the kinetics of the heterogeneous catalytic processes^75,81. The photocatalytic reaction rate could be described by following Eqs^75,81. :

4

where C_t is the concentration of DCF at irradiation time t, and k is the rate constant of the reaction. Equation 4 could be transformed in following Eqs^75,81. :

5

The rate constant is driven from the slope of ln (C/C_o) vs. t as is displayed in Fig. 3c and d. Using this model, the obtained kinetic pseudo-first-order rate constant k and the coefficient of determination (R²) for the CZTS and CZTSe photocatalytic systems are listed in Table 1. Results show that the reaction rate of DCF photodegradation onto CZTSe is highest with k = 3.0 × 10^{− 2} min^{− 1}, followed by photocatalytic degradation rate onto CZTS with k = 1.7 × 10^{− 2} min^{− 1}. Based on these results, it was more suitable for CZTSe to photodegrade DCF under visible light irradiation compared with CZTS, the degradation rate onto CZTSe photocatalyst is about two folds higher than that onto CZTS one. Furthermore, these values are comparable to those reported for DCF photodegradation using Degussa TiO₂ photocatalyst (k = 1.5 × 10^{− 2} min^{− 1}) and not substantially different from those reported for TiO₂ heterojunctions (k = 0.6 × 10^{− 2} min^{− 1})^82,83.

From a practical application perspective, the recyclability of the semiconductor photocatalyst is a crucial factor, as it not only significantly lowers costs but also helps prevent secondary pollution. Recyclability of the CZTX-based nanoparticulate materials has been performed under simulated sunlight after complete deterioration of the DCF molecule; the nanoparticles recovered by filtration, washed multiple times, dried at 60 °C, and reused again. This has been repeated for 4 cycles, and the result of the CZTX-based nanoparticulate materials reusability in terms of % degradation in each cycle is presented in Fig. 3e and f. As can be shown in these figures, a similar trend is exhibited by both photocatalysts under study. These results indicated no loss of photocatalytic efficiency up to the first 2 cycles, and a slight decrease in efficiency can be observed during cycles 3 and 4 when the photocatalyst. In the case of the CZTS photocatalyst, the reduction in DCF removal has been about 13% after 4 times (from 68.0 to 55%), and in the case of the CZTSe one has been about 13.7% (from 83.0 to 69.3%), which sums up to 4 h of irradiation. Although these results suggest good operational stability under the investigated conditions, the slight loss of activity observed after the fourth cycle does not allow us to completely rule out possible structural or surface modifications of the photocatalysts. In addition, some loss of material during the recovery, washing and drying steps is expected when handling nanoparticulate photocatalysts and may also contribute to the decrease in performance. Therefore, the results demonstrate that both CZTX-based photocatalysts exhibit good reusability over four cycles, while further structural characterization after cycling (e.g., XRD or Raman analysis) would be required to fully assess their long-term stability.

The operational pH of the solution is a critical parameter that dictates the electrostatic interactions at the semiconductor-pollutant interface. Recent studies on anionic pollutant degradation have highlighted that weakly acidic conditions (around pH 5.0) typically optimize degradation rates; at this point, the catalyst surface stays positively charged (below its pH _PZC), maximizing attraction toward anionic molecular structures, whereas extreme basicity induces electrostatic repulsion and suppresses radical generation^84,85. This electrostatic alignment is consistent with the literature consensus for diclofenac (pKa approx 4.15–4.5), where several TiO₂-based heterojunctions achieve maximum performance at pH 5.0 due to the attraction between the protonated active sites and the deprotonated drug anion (DCF)^78,79. In the present work, however, photocatalytic evaluations were deliberately performed within a near-neutral pH range (6.5–7.0). While a lower pH might technically enhance initial adsorption, real-world wastewater matrices contaminated with pharmaceuticals typically fluctuate between pH 5.5 and 7.5. Therefore, assessing the performance of bare CZTS and CZTSe under unadjusted, near-neutral conditions provides a far more environmentally representative and practically scalable baseline for real remediation scenarios, avoiding the chemical cost and complexity of large-scale effluent acidification.

The extent of organic pollutant mineralization is a crucial factor in avoiding secondary pollution^76,86. A total organic carbon (TOC) study was conducted to provide an accurate assessment of the photocatalytic process effectiveness and to determine the degree of DCF mineralization by measuring the complete conversion of the pollutant and its by-products into CO₂ and H₂O^45,76,86. In fact, the higher the degree of mineralization, the more efficient the photodegradation process is, because organic pollutants (in present case DCF) decompose to simple inorganic compounds^45,76,86. The mineralization capability of both CZTX-based nanoparticulate materials to DCF molecules has been evaluated by monitoring the variation of TOC in the reaction system during photocatalytic degradation process. The changes of TOC during photocatalytic degradation of DCF in the presence of CZTS and CZTSe as well as in the absence of a photocatalyst are shown in Fig. 4a and b, which was in accordance with the tendency shown in Fig. 3a and b. The gradual decrease of TOC represented the gradual disappearance of organic carbon. In the basis of the obtained TOC values, the mineralization degree (MR) of DCF has been calculated by using Eq. 3, and the resulting MR values as a function of the time of exposure to simulated sunlight are displayed for both the CZTS and CTZSe photocatalyst as inset in panels a and b of Fig. 4a and b, respectively. The CZTS photocatalyst achieved 68% photodegradation of DCF, but the TOC analysis shows only of a 24.3% mineralization degree. The low degree of mineralization can be attributed to the chemical stability of DCF and its transformation intermediates, which also exhibit resistance to degradation and therefore necessitate the use of more potent oxidizing agents for complete mineralization. In contrast, the CZTSe photocatalyst resulted in a 34.2% reduction in TOC, signifcantly improving mineralization compared to CZTS (MR = 24.3%). The observed enhancemet is ascribed to the elevated oxidizing potential of the CZTSe valence band, which facilitates a more efficient mineralization pathway during the photodegradation process. When contextualizing mineralization performance within the literature, direct TOC comparisons require caution due to diverging operational parameters like treatment time and lamp power. For instance, Aziz et al.⁸⁷ achieved complete DCF degradation using bare TiO₂ but reported only 16.0% mineralization after 90 min of UVA exposure due to persistent toxic chlorinated intermediates. Conversely, the advanced TiO₂-CdS Z-scheme heterojunction by Elangovan et al.⁷⁹ reached a high mineralization degree of 86.0% but required a prolonged treatment time (240 min) and a high-power (400 W) light source. Similarly, John et al.⁷⁸ achieved 93.49% DCF removal within 90 min using a TiO₂/g-C₃N₄ heterostructure, but their system relied on an intense 1000 W lamp and omitted the exact numerical percentage of TOC mineralization despite tracking 10 intermediates. In this context, the mineralization degrees obtained in this work using bare, single-phase kesterites (24.3% for CZTS and 34.2% for CZTSe) after only 60 min under standard simulated solar light (AM1.5G, 100 mW cm^{− 2}) present a highly competitive baseline. These results demonstrate that even without complex heterojunction engineering or high-energy inputs, these nanoparticles possess a strong intrinsic capacity to drive rapid primary DCF removal (68%–83%) while initiating organic core breakdown.

Fig. 4.

(a, b) Variation of total organic carbon (TOC) during photocatalytic degradation of diclofenac under AM1.5G simulated sunlight (100 mW cm^-2), in the presence of Cu₂ZnSnS₄ (CZTS, green squares) and Cu₂ZnSnSe₄ (CZTSe, blue squares), and in the absence of photocatalyst (black squares, (a)). Insets show the corresponding mineralization degree. Error bars show the standard deviations of the measurements taken from three independent experiments. (c–f) HPLC–MS spectra of diclofenac photodegradation products using CZTS (c, e) and CZTSe (d, f): (c, d) before irradiation (0 min) and (e, f) after 60 min of illumination.

In order to further explore the photodegradtion processes of DCF by the CZTX-based nanoparticulate materials, high-performance HPLC-MS has been employed to identify the DCF transformation products (DTPs) generated during the photocatalysis of diclofenac. The mass spectra and the catalyzed photodegradation products of DCF are presented in Fig. 4c and f. The structural specifications of the different intermediates have been developed according to the mass-to-charge ratio (m/z ratio). In the presence of the CZTS photocatalyst and prior to the solar simulator irradiation, a strong m/z peak located at m/z = 295 can be seen (Fig. 4c), which could be ascribed to the starting material DCF. Same HPLC-MS result has been observed without illumination for the CZTSe photocatalyst system (Fig. 4d). After 60 min of irradiation and for both CZTX-based nanoparticulate photocatalysts, the peak of DCF disappears and new peaks appeared (Fig. 4e and f). This suggests that the mineralization process has occurred and is dependent on the specific photocatalyst material used. In the case of CZTS (Fig. 4e), the degraded DCF residue spectrum shows six peaks (corresponding to six new different fragments) and they were named as a-f, respectively, at m/z = 94, 127, 145, 166, 189 and 237 which could be ascribed to the split parts of DCF C₆H₆O (a), C₆H₅ClO (b), C₆H₄Cl₂ (c), C₈H₉NO₃ (d), C₇H₅Cl₂NO (e) and C₁₂H₉Cl₂N (f), respectively. Indicating that DCF molecules have been completely degraded into six intermediates, corresponding to compounds a to f. Moreover, when using CZTSe photocatalyst, the HPLC-MS spectrum of the photodegraded DCF residue (Fig. 4f) shows six peaks, and they were named as a-g, respectively, at m/z = 83, 94, 127, 145, 166 and 237 which could be attributed to the fragment parts of DCF C₆H₁₀ (g), C₆H₆O (a), C₆H₅ClO (b), C₆H₄Cl₂ (c), C₈H₉NO₃ (d) and C₁₂H₉Cl₂N (f), respectively. Indicating that by using CZTSe photocatalyst, DCF molecules have been completely degraded into six intermediates, corresponding to compounds a to d, f and g. These fragments and their possible structures have been previously reported by other authors in the context of DCF mineralization^87–90. Moreover, a clear difference between CZTS and CZTSe can be observed, as previously discussed, with CZTSe showing a higher degree of mineralization compared to CZTS. This is reflected in the HPLC-MS fragment analysis, where smaller alkene-type structures with lower molecular weights were observed for CZTSe. It is worth noting that, in both cases (CZTS and CZTSe), the appearance and subsequent disappearance of these fragments were observed throughout the 60-minute photodegradation process, with six fragments identified for each photocatalyst at the 60-minute mark. Then the photodegradation of DCF, mediated by oxidizing agents such as h⁺ and •OH, seems to follow various transformation routes: decarboxylation, aromatic ring elimination, dehalogenation, deamination and progressive fragmentation. These transformations occur sequentially and in parallel, leading to the mineralization of the original compound^87–90. It is relevant to highlight that the intermediate products resulting from the degradation of DCF are usually less toxic than the original drug⁹¹.

In silico assessment of DCF transformation products (DTPs) toxicity

The ecotoxicological profile of DCF and its DTPs (corresponding to the fragments previously identified by HPLC-MS, as detailed in Table 2) was evaluated using VEGA QSAR predictions^92–95. A general decrease in bioaccumulation potential was observed after transformation, with BCF values decreasing from 3.12 (DCF) to 0.08–2.91 in the DTPs, indicating reduced bioconcentration potential.

Table 2.

In silico prediction of ecotoxicity, mutagenicity, bioaccumulation, and persistence of diclofenac and intermediate products by VEGA software.

ID	Ames1	Tox2	Biodeg3	BCFmAG4 Log(L/Kg)	PerWat5 days	PerSed6 days	PerSoil7 days	ToxFis8 mg/L	ToxDaph9 mg/L
DCF	NON-Mutagenic	Toxicant (MR)	nonRB	3.12	0	49	40	0.99	1.06
a	NON-Mutagenic	Toxicant (MR)	RB	0.38	2	72	2	99.89	31.61
b	NON-Mutagenic	Toxicant (MR)	nonRB	0.91	5	45	23	76.72	5.70
c	NON-Mutagenic	NON-Toxicant (MR)	nonRB	2.18	73	708	229	10.69	5.60
d	Mutagenic	Toxicant (MR)	PRB	0.15	4	49	5	459.27	8.55
e	Mutagenic	Toxicant (MR)	nonRB	0.08	2	49	40	23.8	0.50
f	Mutagenic	Toxicant (MR)	nonRB	2.91	5	226	708	8.01	0.99
g	NON-Mutagenic	NON-Toxicant (MR)	RB	2.07	7	71	23	29.85	82.99

¹ Mutagenicity (Ames test) CONSENSUS model (version 1.0.4), VEGA.

² Developmental Toxicity model (CAESAR) (version 2.1.8), VEGA.

³ Ready Biodegradability model (IRFMN) (version 1.0.10), VEGA.

⁴ BCF model (Arnot-Gobas) (version 1.0.1), VEGA, log(L/Kg).

⁵ Persistence (water) quantitative model (IRFMN) (version 1.0.1), VEGA, days.

⁶ Persistence (sediment) quantitative model (IRFMN) (version 1.0.1), VEGA, days.

⁷ Persistence (soil) quantitative model (IRFMN) (version 1.0.1), VEGA, days.

⁸ Acute Toxicity Fish (Fathead minnow LC50 (96 h)), EPA (version 1.0.10), VEGA, mg/L.

⁹ Daphnia Magna LC50 48 h (EPA) (versión 1.0.9), VEGA, mg/L.

* PnonRB possible non-readily biodegradable, PRB possible readily biodegradable, nonRB non-readily biodegradable, LR low reliability, MR medium reliability, GR good reliability.

In contrast, environmental persistence showed a variable trend. Water persistence increased from 0 days for DCF to values ranging from 2 up to 73 days in the DPTs. Similarly, some IPs exhibited a marked increase in persistence in sediment and soil, reaching up to 708 days, compared to 49 and 40 days for DCF, respectively, suggesting the formation of more persistent species⁹⁶.

Toxicological predictions revealed that most DTPs remained toxic, with only one compound classified as non-toxic. Notably, three intermediates were predicted as mutagenic, whereas two were identified as readily biodegradable (RB/PRB), indicating partial improvement in environmental degradability⁹⁷.

Overall, although photocatalytic degradation of DCF using CZTS or CZTSe reduces bioaccumulation potential, it may lead to the formation of DTPs intermediates with increased persistence and, in some cases, enhanced toxicological concern, highlighting the importance of assessing transformation products.

Mechanism of photo-catalytic activity

In a first step, when a semiconductor is irradiated with photons of energy equal or greater than its band gap energy the photons are absorbed. Subsequently, the excitation of photo-generated electrons (e⁻) into conduction band takes place, leaving behind energetic photo-generated holes (h⁺) in the valence band (Fig. 5)^98,99. These electrons and holes can either recombine or migrate towards the photo-catalyst surface participating in several redox reactions. The photocatalytic effects are exerted by these redox reactions. Several reactive species are generated through the reactions with the holes and electrons, which are considered to be involved in the actual oxidative and reductive reactions in photocatalysis^98,99. Oxidation reaction is initiated in the valence band, while reduction reaction is initiated in the conduction band of the semiconductor photo-catalyst^98,99. The species to which oxygen converts with high reactivity are generally called reactive oxygen species (ROS)^98,99. ROS comprise a highly active group of oxidative substances, including superoxide anion radical (•O²⁻), hydrogen peroxide (H₂O₂), singlet oxygen (¹O₂), and hydroxyl radical (•OH) among others^98,99. These ROS play a pivotal role as oxidizing agents in the decomposition processes of organic pollutant compounds. Photo-catalytic activity is closely related to the type and amount of the reactive oxygen species originating from the photo-generated electron and holes. The valence band edge energy and conduction band edge energy of semiconductors determine the type of ROS generated during photo-excitation as it will be further discussed (vide infra). The formation of ROS can be shown by a sequence of chain oxidation and reduction reactions. Since both reduction and oxidation take place concurrently in photo-catalysis, ROS may be produced sequentially both from O₂ and H₂O as illustrated in Fig. 5. Namely, ROS of •OH, H₂O₂, •O²⁻, and ¹O₂ would be generated in this order by the stepwise oxidation of H₂O^98,99. On the other hand, the stepwise reduction of O₂ generates •O²⁻, H₂O₂, and •OH^98,99. These transitory ROS can initiate and promote a number of reactions which end up in complete mineralization of the pollutants into harmless products such as CO₂, water and salts. The basic degradation pathway is explained by the following equations presented and discussed below^73,98–101.

Fig. 5.

Schematic diagram for photo-generated charge carriers separation as well schematic representation of reactive oxygen species (ROS) generation by photo-catalytic reduction and oxidation steps of oxygen and water, and process of photo-oxidation of an organic pollutant during the visible light-active photo-catalytic process onto a semiconductor photocatalyst. The recombination of the electron-holes is also realistic during the photoreaction (for clearness not indicated). (h⁺ = holes, e⁻ = electrons, •O₂⁻ = superoxide anion radical, O₂ = oxygen, H₂O₂ = hydrogen peroxide, •OH = hydroxide radical, OH⁻ = hydroxide, OP = organic pollutant, CO₂ = carbon dioxide).

As it has been schematically depicted in Fig. 5, upon visible-light irradiation, the semiconductor photocatalyst generates electron–hole pairs:

6

The photogenerated holes can directly oxidize the pollutant or react with surface H₂O/OH⁻ to produce hydroxyl radicals (•OH):

7

8

while conduction-band electrons reduce dissolved oxygen to superoxide radicals (•O₂⁻):

9

These reactive oxygen species, together with photogenerated holes, are generally recognized as the main oxidative agents responsible for pollutant degradation. The overall process ultimately leads to the mineralization of organic contaminants into CO₂, H₂O and inorganic ions. Finally, the interaction of the hydroxyl radicals with the adsorbed DCF molecules should lead to the degradation of this DCF pollutant:

10

A more detailed description of the ROS generation pathways is provided in the Supplementary Information (Sect.  9).

In photo-catalysis, the efficiency of the photo-induced degradation varies over a wide range depending on the kinds of semiconductors (photo-catalysts) used and the solution conditions employed.

Radical scavenging assay and plausible photocatalytic mechanism of DCF degradation

Photocatalytic activity is closely related to the type and amount of the ROS originating from the photogenerated electron and holes. As has just been said above, in a general way, •OH, •O₂⁻, and h⁺ are the major reactive species responsible for the photo-oxidation of environmental pollutants under visible light irradiation^{74,77,102–104}. So, it is imperative to determine the role of those species in the degradation process to design an effective photocatalyic system¹⁰¹. Then, in order to evaluate the presence, relevance, and influence of each of those reactive oxidative species generated by the photocatalysts in the degradation process, chemical compounds called scavengers, which act as selective captors or inhibitors of these reactive species, have been applied to different AOPs^{74,77,102–104}. These scavengers are designed to react rapidly and specifically with a radical, generating a stable species that does not interfere with the reaction, thereby eliminating the effect of the radical in the degradation^{74,77,102–104}. Introducing known radical scavenging species is a common technique for studying photocatalytic mechanisms, as changes in degradation rate in the presence of a specific scavenger can indicate which photogenerated species are involved in the rate-limiting steps of the reaction^{74,77,102–104}.

In the present work, to detect the ROS active species generated in the photocatalytic system, different scavengers were introduced into the solution of DCF, including ethylenediaminetetraacetic acid (EDTA-2Na), potassium dichromate (K₂Cr₂O₇) and isopropanol alcohol (IPA), which were respectively employed as scavengers for hole (h⁺), electron (e⁻) and hydroxyl free radical (•OH), and which are depicted in detail in Table S1 in the Supplementary Section. Moreover, some experiments have been carried out by bubbling argon gas in the DCF solution with the semiconductor photocatalyst in order to remove the molecular oxygen and then eliminate its presence. Avoiding then, the generation of the superoxide anion radical (•O₂⁻) through the electrochemical reduction of the molecular oxygen by photoexcited conduction band electrons (Fig. 5). In fact, several times, molecular oxygen is considered as an electron scavenger^105,106. Obtained results of the effect of the different assayed scavengers on the photodegradation of DCF over the CZTX-based nanoparticulate photocatalysts, (presented as the relative concentration (C/C_o) changes of DCF molecule versus the solar simulator irradiation time), in comparison to the process without scavengers, are presented in Fig. 6a and b. Moreover, Fig. 6c and d show the influence of the different assayed scavengers on the photocatalytic efficiency of DCF degradation over the CZTX-based nanoparticulate photocatalysts, after 60 min AM1.5G simulated sunlight (100 mW cm^{− 2}) illumination. It must be pointed out that in these Fig. 6c and d, photocatalytic experiments with no scavengers have been used as a blank with a normalized degradation efficiency of 100%. It can be observed that when the photocatalysis experiments were carried out in the absence of dissolved oxygen (under argon-saturated conditions), the DCF photodegradation efficiency decreased significantly, exhibiting low values of approximately 43% and 38% when CZTS and CZTSe photocatalysts were used, respectively. The last, revealing that the dissolved molecular oxygen in the reaction solution played a critical role in DCF photodegradation process for these CZTX-based nanoparticulate photocatalysts. Molecular oxygen adsorbed onto the CZTX-based photocatalyst surface prevents the recombination of e⁻ / h⁺ pairs by trapping conduction band electrons with the formation of •O₂⁻ radicals according to Eq. S8^101,105,106. Additionally, the effective production of •O₂⁻ radicals can also form H₂O₂ (according to Eq. S11), which in turn can be transformed into •OH radical according to Equations S12 to S14^99,100,107. In fact, the H₂O₂ produced does not accumulate proportionally to its formation because it simultaneously undergoes decomposition through a series of surface reactions, leading to the generation of additional •OH radicals¹⁰⁸. The generation of hydrogen peroxide (H₂O₂) ultimately results in the formation of hydroxyl radicals (•OH)¹⁰⁹. So, it could be stated that the presence of molecular oxygen also determines the production of •OH radicals via multistep reduction of molecular oxygen^110–112. As previously mentioned, •OH radicals are the main reactive oxygen species driving most oxidation reactions in photocatalytic systems. This is largely due to their extremely high reactivity, with rate constants for their reactions with most pollutants typically ranging from 10⁶ to 10⁹ M⁻¹s⁻¹ (almost those of the diffusion limit)^98,110. In brief, those results indicate that the formation of hydroxyl radicals, from the oxidation reaction of dissolved oxygen with electrons from the conduction band, is crucial in the generation of these oxidizing radicals. Furthermore, the absence of dissolved oxygen favors the recombination of the electron-hole pair, making the photodegradation process even less efficient^113–115.

Fig. 6.

Effect of radical scavengers on the photocatalytic degradation of diclofenac under AM1.5G simulated sunlight (100 mWcm^-2). (a, b) Relative concentration (C/C_o) as a function of illumination time using Cu₂ZnSnS₄ (CZTS) and Cu₂ZnSnSe₄ (CZTSe), respectively, in the presence of EDTA-2Na (green filled circle symbols), IPA (red up-pointing filled triangle symbols), K₂Cr₂O₇ (royal filled squared symbols), without scavengers (blue filled squared symbols), and under Ar atmosphere (molecular oxygen removed), (orange filled squared symbols). (c, d) Normalized photocatalytic degradation efficiency after 60 min for CZTS and CZTSe, highlighting the effect of each scavenger and the corresponding reactive species involved. Error bars show the standard deviations of the measurements taken from three independent experiments.

As previously mentioned, the reduced photocatalytic performance observed with each scavenger underscores the crucial role of the corresponding reactive species. With this in mind, we proceed with the analysis of the results presented in Fig. 6. As shown in Fig. 6c, it is obvious that the DCF degradation only slightly changes after EDTA addition (from 100% to 80%), indicating that h⁺ ROS plays a minor role in the photocatalytic degradation of DCF by using the CZTS photocatalyst. However, K₂Cr₂O₇ and IPA markedly inhibited DCF degradation, reducing the proportion of DCF removed from normalized 100% to 58% and 56%, respectively. This outcome underlines the significant role of e⁻ in tandem with •OH in the degradation of DCF by nanoparticulate CZTS. In the case of CZTSe photocatalyst, as shown in Fig. 6d, the DCF degradation slightly changes after K₂Cr₂O₇ addition (from 100% to 75%), from which it can be inferred that e⁻ plays a minor role in the photocatalytic degradation of DCF. Nevertheless, both EDTA and IPA significantly suppressed DCF degradation, lowering the normalized removal efficiency from 100% to 47% and 39%, respectively. These findings emphasize the critical involvement of valence band holes (h⁺), in concert with hydroxyl radicals (•OH), in the photocatalytic degradation of DCF mediated by nanoparticulate CZTSe photocatalyst.

So, it must be concluded that the decrease in the removal rate in the presence of argon/scavengers presents the following trend:

1. CZTS photocatalyst:

Ar > IPA > K₂Cr₂O₇ > EDTA.

• b)CZTSe photocatalyst:

Ar > IPA > EDTA > K₂Cr₂O₇.

These findings illustrated that the •O₂⁻ (and/or •OH) and •OH radicals acted as dominant active species for photocatalytic degradation of DCF under visible light illumination by using both CZTS and CZTSe photocatalysts. These results are consistent with several recently published works, reporting that •O₂⁻ followed by •OH radicals were in sequence the preliminary reactive oxygen species responsible for DCF photodegradation^28,116. Moreover, in agreement with other works, where major contribution of •O₂⁻ radical was described for the photodegradation of DCF^7,8,117.

Overall, the proposed photocatalytic mechanism and the predicted toxicity of the identified transformation products demonstrate that DCF degradation is accompanied by the formation of intermediates with distinct environmental behaviors. Although successive oxidation reactions generally reduce the bioaccumulation potential of these compounds, they do not necessarily ensure complete detoxification, as some intermediates may retain structural features associated with persistence or mutagenic potential. These findings highlight the importance of evaluating photocatalytic performance not only in terms of DCF removal efficiency but also considering the formation and toxicity of transformation products.

Conclusions

This study reports, for the first time, the successful application of nanoparticulate Cu₂ZnSnS₄ (CZTS) and Cu₂ZnSnSe₄ (CZTSe) as visible-light photocatalysts for diclofenac (DCF) removal under simulated sunlight. Among the two materials, CZTSe exhibited the highest photocatalytic activity, achieving 83% DCF removal within 60 min, while CZTS reached 68%, both outperforming benchmark TiO₂ reported under comparable conditions. Both photocatalysts also showed excellent stability, maintaining their activity after four consecutive reuse cycles. Total organic carbon analysis confirmed the progressive mineralization of DCF, while HPLC–MS identified six transformation products formed during photocatalytic degradation. Although photocatalysis reduced the bioaccumulation potential of DCF, in silico analysis indicated that some transformation products may exhibit greater persistence and potential toxicological risks, highlighting the importance of assessing degradation by-products in addition to pollutant removal. Scavenger experiments demonstrated that •O₂⁻ (and/or •OH) and •OH radicals are the main reactive oxygen species involved in DCF degradation over both kesterite photocatalysts. These findings demonstrate the potential of CZTS and CZTSe as efficient and stable photocatalysts for water remediation while providing new insights into the degradation mechanism and the environmental implications of the generated transformation products. Overall, this work establishes CZTS and CZTSe, particularly CZTSe, as promising visible-light photocatalysts for DCF removal and provides a comprehensive framework linking photocatalytic efficiency with degradation pathways, mineralization, reactive oxygen species, and the environmental risks associated with transformation products. From a practical perspective, diclofenac removal alone does not guarantee complete detoxification of the treated effluent. Therefore, a secondary treatment step, together with experimental validation of the QSAR-based toxicity predictions, may be required before wastewater discharge.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

PSN, RH, PG, EM, RSG, GR, ENA and EAD wrote the main manuscript text, each one contributes in their area of ​​expertise and they have contributed by creating the figures and tables of the manuscript . All authors reviewed the manuscript.

Funding

This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.

Data availability

All data supporting the findings of this study are available within the paper and its Supplementary Information.

Declarations

Competing interests

The authors declare no competing interests.