Electron trap engineering in g-C₃N₄ with molecular Mo₃S₄ clusters for visible-light-driven photocatalysis

Abstract

In this study, a hybrid photocatalyst based on carbon nitride (g-C₃N₄) modified with the molecular cluster [Mo₃S₄Cl₃(en)₃]Cl was developed and evaluated for the visible-light-driven degradation of ciprofloxacin (CIP). The incorporation of the cluster introduced structural and electronic modifications in g-C₃N₄, as demonstrated by spectroscopic, microscopic, and electrochemical analyses. Transient photocurrent responses and Mott–Schottky plots revealed enhanced charge separation and a positive shift in flat-band potential, consistent with increased electron mobility. Impedance spectroscopy indicated reduced charge-transfer resistance, supporting more efficient carrier transport. The optimized sample containing 2.5 wt% cluster achieved complete degradation of CIP in 150 minutes and 65% mineralization under visible light. Kinetic modeling showed pseudo-first-order behavior with a 12-fold increase in the rate constant compared to pristine g-C₃N₄. The material maintained high performance under different pH conditions, catalyst dosages, and real water samples, including tap and river water. Interference studies with common ions revealed moderate inhibition, highlighting the importance of matrix composition. Fluorescence and absorbance-based tests identified hydroxyl radicals and singlet oxygen as the main reactive oxygen species (ROS). A mechanism is proposed in which the Mo₃S₄ cluster acts as an electron trap within the g-C₃N₄ matrix, facilitating charge separation and boosting ROS generation under visible light. The hybrid catalyst also exhibited good recyclability, maintaining over 50% of its mineralization capacity after five cycles. These results demonstrate the relevance of electron trap engineering in molecular–semiconductor hybrids for the efficient and robust photodegradation of emerging contaminants under environmentally realistic conditions.

---

Introduction

In recent decades, the accelerated growth of the global population, driven by industrialization and socioeconomic development, has contributed directly to the expansion of the pharmaceutical industry.¹ While this progress has improved access to medical care, it has also raised significant environmental concerns, particularly due to the uncontrolled release of pharmaceutical compounds into natural ecosystems. These substances are developed to prevent or treat illnesses and to enhance human health, undergoing metabolic conversion within the body once administered. Despite the key role of antibiotics in the treatment of bacterial infections, their widespread use has raised increasing concern due to their potential to cause long-term environmental and public health impacts.² In particular, small amounts of the original drug, whether unmetabolized or transformed into metabolites, are excreted into wastewater, which may subsequently be released into natural water bodies or reused for agricultural irrigation. Conventional wastewater treatment systems are typically designed to remove contaminants present at mg L⁻¹ levels.³ However, pharmaceutical residues are often detected in much lower concentrations, ranging from ng to a few µg L⁻¹, placing them in the category of micropollutants that are difficult to eliminate through standard processes.⁴ Antibiotics are especially problematic because of their high chemical stability, strong biological activity, mobility in aquatic systems, and persistence in the environment.⁵ Among the most critical consequences of their improper disposal is the development of antimicrobial resistance, a growing global health crisis in which bacteria acquire resistance to previously effective treatments, compromising the future efficacy of antibiotics in clinical settings.⁶ For these reasons, the monitoring and removal of pharmaceutical compounds have been included among the targets of the United Nations 2030 Agenda for Sustainable Development, as recognized by UNESCO.⁷

Ciprofloxacin (CIP), a fluoroquinolone widely prescribed for treating bacterial infections, exemplifies this issue.⁸ Its frequent detection in water bodies and resistance to degradation have raised concerns within the scientific and public health communities.⁹ In fact, CIP has been identified as a priority organic pollutant and classified as an emerging contaminant within the European Union.^10,11 Various treatment approaches have been explored to address its presence in aquatic environments, including ozonation, biological remediation, and heterogeneous photocatalysis.¹² Among these, heterogeneous photocatalysis has attracted particular attention due to its cost-effectiveness, adaptability, and potential for complete mineralization of persistent compounds.

Photocatalysis based on semiconductors has garnered significant attention as an environmentally friendly approach for degrading pollutants, owing to its ability to convert solar energy into chemical energy.¹³ Despite its potential, the low efficiency of many photocatalytic systems remains a major barrier to their widespread industrial application, highlighting the need for innovative materials with enhanced performance under solar or visible light.¹⁴ Recent studies have explored the use of modified semiconductors to tackle this issue. For example, Souza et al. employed Eu-doped Ag₂CrO₄ to degrade CIP under visible light, achieving approximately 27% mineralization of the initial residue after 120 min of irradiation.¹⁵ Similarly, Ribeiro et al. used lanthanum-doped Ag₃PO₄ for the same purpose, reporting a mineralization rate of 32.6% in just 90 min.¹⁶ While both catalysts demonstrated some activity, they exhibited limited stability over successive photocatalytic cycles, which compromises their practical usability. Other semiconductors such as ZnO, TiO₂, and CuO have also been investigated for CIP degradation.¹⁷ However, these materials typically require the use of high-intensity UV light to achieve satisfactory results. Such conditions are far from ideal in real-world applications, where low-energy, visible-light-driven processes are preferred. Consequently, the search for stable, efficient, and visible-light-active photocatalysts remains a central focus in the development of advanced water treatment technologies.

Graphitic carbon nitride (g-C₃N₄) is a metal-free, n-type polymeric semiconductor that has gained attention as an efficient photocatalyst under visible light, owing to its beneficial features such as chemical stability, low toxicity, visible-light responsiveness, narrow bandgap, suitable electronic properties, and straightforward synthesis.^18,19 These characteristics make it a versatile material for applications in catalysis, electronics, and renewable energy.^20–22 The intrigue surrounding g-C₃N₄ species stem from its making it substantial surface area and unique porosity networks, marking it an ideal substrate for immobilizing complex metal oxides. One of the main limitations of g-C₃N₄, however, lies in its relatively high rate of photogenerated charge recombination, which significantly hinders its photocatalytic efficiency.²³ To address this issue, a widely adopted approach involves coupling g-C₃N₄ with other functional materials to form heterostructures.^24–26 This strategy enhances the rates of charge migration and separation, and utilizes a broad spectrum of light, to improve photocatalytic efficiency. For example, Thuan et al.²⁷ developed a ZnO-doped g-C₃N₄ composite that exhibited superior photocatalytic performance to degrade CIP compared to g-C₃N₄ under visible-light irradiation. Similarly, Hu et al. reported enhanced degradation of CIP using TiO₂/g-C₃N₄ heterostructures, attributing the improvement to the increased generation of reactive species such as hydroxyl (·OH) and superoxide radicals (·O₂⁻).²⁸ Other studies have also shown that combining g-C₃N₄ with materials like Fe₃O₄ and Cu₂O leads to enhanced photocatalytic activity,^29,30 reinforcing the idea that the formation of heterostructures is a promising route to overcome the intrinsic limitations of g-C₃N₄ and to broaden its applicability in environmental remediation.

Among the different molecular systems explored for catalytic applications, Mo₃S₄ clusters (MSC) have emerged as attractive candidates due to their well-defined trinuclear structure, tunable electronic properties, and notable redox versatility.³¹ These clusters, composed of three molybdenum atoms bridged by sulfide ligands, have proved to be efficient catalysts in a wide range of organic transformations, including methanolysis of hydrosilanes,³² hydrogenation and semihydrogenation,^33–35 C–C cross-coupling reactions,³⁶ and hydrogen evolution processes.^37,38 A key advantage of these systems is their chemical robustness and stability under air, which greatly facilitates their handling and practical applicability in catalytic protocols.³⁹ Their modular structure also allows for the fine-tuning of activity and selectivity through ligand modifications. In recent years, the integration of molybdenum sulfide clusters with photoactive semiconductors has emerged as a promising strategy to enhance the photocatalytic performance of single-phase materials.^40–42 The incorporation of Mo₃S₄ units not only introduces additional reactive sites and promotes interfacial electron transport but also enables electronic modulation of the semiconductor through the creation of mid-gap trap states. This approach, rooted in the concept of electron trap engineering, facilitates directional charge migration and suppresses charge recombination, resulting in improved visible-light-driven photocatalytic activity. Such molecular-level control over the electronic landscape of the semiconductor matrix opens new pathways for designing hybrid materials with enhanced efficiency and selectivity in environmental remediation applications.

The fabrication of Mo₃S₄ units anchored g-C₃N₄, characterized by numerous active sites and exceptional stability, presents a promising but unexplored research field. This work investigates a design approach for developing a hybrid material composed of [Mo₃S₄Cl₃(en)₃]Cl cluster (en = ethylenediamine) and g-C₃N₄ for the first time, making them suitable candidates for advanced degradation of CIP under visible light. Structural and optical characterizations were conducted to investigate the effects of cluster incorporation, and photocatalytic performance was evaluated under various conditions, including real water matrices. The results demonstrated that the presence of the [Mo₃S₄Cl₃(en)₃]Cl cluster enhanced the activity and stability of g-C₃N₄, indicating a promising strategy for treating emerging contaminants.

Results and discussion

Characterization of hybrid g-C₃N₄/[Mo₃S₄Cl₃(en)₃]Cl

Following the incorporation of [Mo₃S₄Cl₃(en)₃]Cl clusters into g-C₃N₄, the resulting materials were structurally analyzed by X-ray Diffraction (XRD) and Fourier Transform Infrared Spectroscopy (FTIR). The XRD pattern of g-C₃N₄ is characterized by two main diffraction peaks: one around 12.8°, assigned to the (100) plane, which is associated with the in-plane structural repetition of tri-s-triazine units;⁴³ and a more intense peak at 27.6°, corresponding to the (002) plane, related to the interlayer stacking of the conjugated aromatic system along the c-axis (Fig. 1A).⁴⁴ The characterization of the [Mo₃S₄Cl₃(en)₃]Cl cluster by ¹H-NMR and HRMS (ESI-TOF) is provided in the SI (Fig. S1 and S2). Upon cluster incorporation, a broadening of the (002) peak is clearly observed, as shown in Fig. 1B. The full width at half maximum (FWHM) of this peak increases from 1.3° in the g-C₃N₄ to 1.9°, 2.1°, and 6.8° for the composites containing 2.5, 5.0, and 10.0 wt% of [Mo₃S₄Cl₃(en)₃]Cl, respectively. This increase in FWHM suggests a reduction in the crystallite size and a possible loss of long-range order in the layered structure, which may be attributed to the disruption of π–π stacking interactions between g-C₃N₄ layers. Importantly, no new diffraction peaks were detected after cluster incorporation, indicating that the overall g-C₃N₄ structure remains preserved, and no crystalline secondary phases were formed.

Fig. 1 (A) X-ray diffraction (XRD) patterns; (B) FWHM analysis; (C) FTIR spectra of the samples; (D) inset showing the FTIR region between 1150–1650 cm⁻¹ in detail.

This structural disorder is likely a result of the intercalation or close interaction between the [Mo₃S₄Cl₃(en)₃]Cl clusters and the polymeric layers of g-C₃N₄. Such interactions can induce lattice strain and introduce defects, thereby disturbing the regular π–π stacking of the g-C₃N₄ sheets. Feng et al.⁴⁵ synthesized g-C₃N₄ through the thermal condensation of urea under both enclosed and non-enclosed conditions, observing that the non-enclosed condition led to reduced interlayer connectivity due to the weakening of van der Waals forces between the stacked layers. In another study, Jiang et al.⁴⁶ reported that doping g-C₃N₄ with alkali metals such as Li, Na, and K led to an increase in the interlayer distance. This effect was attributed to the incorporation of metal ions into the nanoscale cavities of the g-C₃N₄ framework, where they interact with N-containing electron-donating functional groups, thereby weakening the interlayer interactions and expanding the layered structure. Similarly, the presence of the molecular [Mo₃S₄Cl₃(en)₃]Cl cluster may disrupt these van der Waals interactions, weakening interlayer bonding and promoting structural disorganization. These modifications are also expected to influence the electronic structure of the material, potentially enhancing charge separation and reducing the recombination of the photogenerated charge carriers.

The vibrational features of the samples were investigated by FTIR spectroscopy to gain insight into their functional groups and bonding environment (Fig. 1C). The spectrum for the [Mo₃S₄Cl₃(en)₃]Cl compound revealed bands associated with coordinated ethylenediamine ligands, notably broad N–H stretching vibrations in the 3100–3250 cm⁻¹ range, as well as C–H stretching modes from the –CH₂ groups between 2870 and 2950 cm⁻¹. Additional bands attributed to N–H bending and C–N stretching vibrations were observed throughout the 1000–1650 cm⁻¹ region, consistent with the presence of amine groups and the aliphatic framework of the ligand.⁴⁶ Low-frequency bands linked to Mo–S bonding appeared below 900 cm⁻¹, supporting the integrity of the Mo₃S₄ core.⁴⁷ In comparison, the FTIR spectrum of g-C₃N₄ showed the expected features of its polymeric structure.⁴⁸ Broad absorptions between 3070 and 3320 cm⁻¹ were attributed to N–H stretching vibrations, likely due to terminal amine groups. The fingerprint region exhibited a series of sharp bands between 1200 and 1650 cm⁻¹, corresponding to C–N and C=N stretching modes within the heptazine or tri-s-triazine units.⁴⁹ A distinct peak near 810 cm⁻¹ was assigned to the out-of-plane bending of triazine rings, a hallmark of the g-C₃N₄ structure.⁵⁰

After incorporation of the [Mo₃S₄Cl₃(en)₃]Cl cluster into the g-C₃N₄ matrix, no additional absorption bands exclusive to the cluster were distinctly visible. Nonetheless, slight shifts and broadening of the characteristic g-C₃N₄ bands, particularly in the 1200–1650 cm⁻¹ region, were observed (Fig. 1D). These changes indicate subtle modifications in the local chemical environment, likely arising from interactions between the ethylenediamine ligands and the nitrogen-rich surface of g-C₃N₄. Similar behaviors were observed when heterostructures were formed with other materials, such as BaTiO₃ and CeO₂.^51,52 Such interactions, possibly through hydrogen bonding or weak coordination, support the presence of chemical affinity between the two components, potentially leading to structural rearrangements and influencing the electronic properties of the resulting hybrid material.

The morphology of the materials was evaluated using Scanning Electron Microscopy (SEM) and Transmission Electron Microscopy (TEM) (Fig. 2). The samples were characterized by SEM, which revealed that the g-C₃N₄ consists of wrinkled lamellar structures. Upon anchoring the [Mo₃S₄Cl₃(en)₃]Cl cluster, no significant morphological changes were observed, indicating that the incorporation of the cluster does not visibly alter the material architecture. TEM analysis further confirmed this observation, showing that all samples are composed of entangled lamellar sheets with an average thickness of approximately 30 nm. To evaluate the chemical composition and verify the successful anchoring of the cluster onto the g-C₃N₄ matrix, Energy-Dispersive X-ray Spectroscopy (EDS) mapping was performed, targeting the O Kα₁, N Kα₁, S Kα₁, and Mo Lα₁ signals. For all samples, the lamellar domains were composed primarily of carbon and nitrogen, consistent with the structure of g-C₃N₄. Following cluster incorporation, clear signals of sulfur and molybdenum appeared on the lamellar surfaces, confirming the presence of the [Mo₃S₄Cl₃(en)₃]Cl cluster. Moreover, a progressive increase in the intensity and density of the S and Mo signals was observed in line with the increase in cluster loading, further supporting the effective and tunable anchoring of the cluster. Despite the presence of both elements being confirmed, their distinction remains challenging due to the close energy values of their emission lines (S Kα₁ at 2.309 keV and Mo Lα₁ at 2.293 keV), which results in partial spectral overlap during EDS analysis.

Fig. 2 SEM, TEM, and EDS analyses of the materials. (A–C) g-C₃N₄; (D–F) C₃N₄–2.5MSC; (G–I) C₃N₄–5.0MSC; and (J–L) C₃N₄–10.0MSC.

To investigate the optical properties and possible structural modifications induced by the anchoring of the clusters, diffuse reflectance spectroscopy (DRS) analyses were performed (Fig. 3A). From the reflectance data, the E_gap of the samples was estimated using the Tauc plot method. g-C₃N₄ exhibits an indirect band gap,⁵³ and the estimated E_gap values ranged from 2.83 to 2.87 eV, which are in good agreement with values reported in the literature.⁵⁴ These results suggest that the fundamental electronic structure of g-C₃N₄ remains largely unchanged upon cluster anchoring.

Fig. 3 (A) Band gap determination from DRS; (B) PL spectra of the samples under visible excitation. (C) Powder EPR spectra obtained in argon atmosphere at 77 K. (D) Mott–Schottky plots. (E) Photocurrent transient response curves. (F) EIS spectra of C₃N₄ and 2.5MSC, 5.0MSC, 10.0MSC. Inset: Equivalent circuit.

Photoluminescence (PL) analysis was performed to gain insight into the electronic defects within the band gap and to evaluate the recombination dynamics of photogenerated charge carriers (Fig. 3B).⁵⁵ The PL emission in g-C₃N₄ originates from the radiative recombination of electrons and holes, typically involving π*–π transitions within the aromatic heptazine units, as well as transitions from the conduction band to mid-gap states introduced by structural defects such as vacancies, surface terminations, and heteroatom dopants.^56–58 Therefore, the emission profile provides important information on both the electronic structure and defect chemistry of the material.^59,60 In general, a decrease in PL intensity indicates a reduction in radiative recombination, suggesting more efficient charge separation and longer carrier lifetimes.⁶¹ In the present work, it was observed that as the concentration of anchored clusters increased, there was a systematic decrease in PL intensity. This quenching effect supports the hypothesis that the [Mo₃S₄Cl₃(en)₃]Cl cluster acts as an electron-trapping center, thereby delaying recombination processes and enhancing the lifetime of photogenerated charge carriers. Regarding emission wavelengths, the g-C₃N₄ sample exhibited a maximum PL emission at 588 nm (2.11 eV), which is typically attributed to deep-level defects such as nitrogen vacancies or interstitials.⁵⁷ With the introduction of 2.5 wt% cluster, this peak blue-shifted to 567 nm (2.18 eV), indicating a change in the defect states and the emergence of shallower defects, likely associated with edge or surface terminations. At higher cluster loadings (5.0 and 10.0 wt%), the emission red-shifted again to 579 nm (2.14 eV) and 584 nm (2.12 eV), respectively, suggesting the formation of new structural defects. This behavior suggest that the anchored clusters interact with native vacancies situated between the g-C₃N₄ lamellae, thereby altering the nature and distribution of defect states. These observations are consistent with structural modifications observed in XRD and FTIR analyses, providing further evidence that cluster anchoring leads to defect engineering within the material.

To further investigate the presence and nature of defect states, Electron Paramagnetic Resonance (EPR) measurements were carried out at 77 K. All photocatalyst samples exhibited a single, isotropic resonance signal with a g-factor of approximately 2.00 (Fig. 3C), characteristic of unpaired electron defects within the highly conjugated polymer matrix of g-C₃N₄.^62–64 To compare the relative concentration of these defects, the spectra were normalized by the weighted mass of each sample. This normalization revealed significant differences in signal intensity among the catalysts: C₃N₄ (15151), C₃N₄–2.5MSC (22325), C₃N₄–5.0MSC (13760), and C₃N₄–10.0MSC (16103). The 2.5MSC sample exhibited the most intense signal, suggesting the highest concentration of paramagnetic centers, whereas the other three catalysts displayed similar and comparatively lower defect densities.

This non-monotonic variation in signal intensity aligns well with the structural and electronic trends observed for the series. At low cluster loading (2.5 wt%), the [Mo₃S₄Cl₃(en)₃]Cl species are homogeneously anchored within the g-C₃N₄ matrix, promoting moderate structural disorder and the formation of shallow defect states, as evidenced by the broadening of the (002) plane and the partial quenching of the PL signal. These shallow defects act as active sites that facilitate charge separation, consistent with the enhanced photocurrent response observed for this composition (Fig. 3E). In contrast, higher loadings (5.0 and 10.0 wt%) lead to a progressive loss of XRD definition and a red-shift in the PL emission, indicating the formation of aggregated clusters and deeper defect states. Such aggregation likely causes partial passivation of intrinsic vacancies and reduces the number of unpaired spins detectable by EPR, resulting in lower signal intensities. These findings indicate that the 2.5 wt% cluster loading provides an optimal balance between defect generation, structural disorder, and charge carrier separation efficiency.

Electrochemical characterization techniques were employed to gain insight into the charge separation and transfer properties of the [Mo₃S₄Cl₃(en)₃]Cl-modified g-C₃N₄ samples. Mott–Schottky analysis of g-C₃N₄ and g–C₃N₄–MSC was performed to determine the semiconductor type and flat-band potential (E_FB) from potential-dependent space charge capacitance. Fig. 3D shows that the g-C₃N₄ and all the g–C₃N₄–MSC samples present positive slope characteristics of n-type. The E_FB values estimated from Mott–Schottky plots were −1.14, −1.02, −0.73, and −0.64 V vs. Ag/AgCl for the pristine g-C₃N₄, g-C₃N₄–2.5MSC, g-C₃N₄–5.0MSC, and g-C₃N₄–10.0MSC, respectively. The positive shift in the flat band potential by the [Mo₃S₄Cl₃(en)₃]Cl cocatalyst anchoring demonstrates the successful electronic modulation of g-C₃N₄, thereby enhancing the mobility of the photogenerated charge carriers and driving the reactions.^65,66 The schematic positions of the valence and conduction bands of the samples are depicted in Fig. S3. The band alignments were simulated by calculating the conduction band from the estimated flat band potential of the n-type semiconductor, which was considered 0.10 V higher than the flat band potential.^67,68 From band's positions observed, it's seen that the excited electron in the [Mo₃S₄Cl₃(en)₃]Cl cluster from the g-C₃N₄ structure, where the electron is easily transported, exhibits higher electron mobility.⁶⁸

The charge excitation and migration dynamics under light simulation were studied using transient photocurrent analysis, with switching on–off cycles of 30 seconds (Fig. 3E). The photocurrent transient profile obtained for the g-C₃N₄–2.5MSC sample demonstrates a higher photoresponse than that of the g-C₃N₄, highlighting the faster separation of charge carriers upon modification with the [Mo₃S₄Cl₃(en)₃]Cl. The lower photocurrent density observed for the g-C₃N₄ sample is due to the high recombination of the generated carriers. However, although great quantities of the [Mo₃S₄Cl₃(en)₃]Cl exhibit enhanced performance in the dark, the light response is inferior to that of the g-C₃N₄–2.5MSC, indicating the necessity of an ideal loading of the MSC cluster to achieve photo-response and effective charge separation from the heterostructure at the photocatalyst interface. As observed for the 5.0 and 10.0 wt% of [Mo₃S₄Cl₃(en)₃]Cl, the photoresponse is significantly lower. Moreover, the stability of the transients of the latter is inferior. This photocurrent profile is associated with high cocatalyst loading, which induces structural defects,^69,70 in concordance with the PL analysis, indicating a different defect nature for the higher MSC loading samples, thereby affecting the interface and diminishing its photocatalytic activity.⁶⁷ Moreover, the g-C₃N₄–10.0MSC sample presents a higher dark current, demonstrating the electrocatalytic nature of the [Mo₃S₄Cl₃(en)₃]Cl. The photocurrent response and PL profiles confirm the effectiveness of the [Mo₃S₄Cl₃(en)₃]Cl modification of g-C₃N₄, which works synergistically to diminish the charge recombination of the photogenerated carriers.

Impedance spectroscopy analyses were conducted further to investigate the charge transfer properties of the samples. Fig. 3F displays the Nyquist plots along with the equivalent circuit used to fit the parameters and estimate the charge transfer resistance (R_ct) for the g-C₃N₄ and g-C₃N₄–MSC samples. As shown in Fig. 3F, the [Mo₃S₄Cl₃(en)₃]Cl modification significantly decreases the quasi-semicircle radius by demonstrating enhanced electronic conductivity. Table S1 summarizes R_ct values, confirming the lower charge transfer resistance of the [Mo₃S₄Cl₃(en)₃]Cl-modified samples. Although the g-C₃N₄–2.5MSC presents only a slight improvement in the R_ct, its photocatalytic performance is enhanced due to its ability to accelerate charge transport while maintaining its photocatalytic response, resulting in superior photocatalytic performance. Higher-loaded samples exhibit better charge transfer properties but diminish the photoresponse,⁶⁷ suggesting that the excess charge is likely undergoing recombination under illumination, thereby enhancing only the electrocatalytic performance in dark. Collectively, EPR results indicate that an optimal concentration of paramagnetic defects introduced by MSC clusters enhances charge separation and photocatalytic performance, whereas excessive defects lead to recombination and diminished photoresponse.

Molybdenum sulfide species have been explored as cocatalysts for photocatalytic processes, such as hydrogen production,^65,71,72 in which the electronic transfer properties of g-C₃N₄ can be improved. Fig. S4 presents the polarization curves recorded in dark for the g-C₃N₄ and g-C₃N₄–MSC loaded electrodes, performed to investigate the influence of [Mo₃S₄Cl₃(en)₃]Cl incorporation on the catalytic activity towards H₂ evolution. As the [Mo₃S₄Cl₃(en)₃]Cl loading increases at the material, a rise in the maximum current density is observed, along with a positive shift in the onset potential for H⁺ reduction. This behavior represents a significant improvement compared to g-C₃N₄, indicating that [Mo₃S₄Cl₃(en)₃]Cl anchoring enhances charge generation and transfer capabilities. Additionally, the reduced overpotential for H₂ production suggests that [Mo₃S₄Cl₃(en)₃]Cl facilitates electron transfer from the C₃N₄ structure to the electrolyte.

Taken together, these findings indicate that under visible-light excitation, photoexcited electrons in g-C₃N₄ are transferred to the [Mo₃S₄Cl₃(en)₃]Cl cluster, while the holes remain in the valence band of g-C₃N₄. The cluster thus acts as an efficient electron trap and co-catalyst, facilitating interfacial charge transfer and suppressing electron–hole recombination. In contrast to a conventional type-II heterojunction, where electrons continuously flow to the conduction band of a secondary semiconductor, and to a direct Z-scheme, where electrons from one component recombine with holes from the other to preserve high redox potentials, the present system follows a distinct trap-state-mediated pathway. Here, the molecular cluster introduces discrete sub-band trap levels that transiently capture the photogenerated electrons from g-C₃N₄, delaying recombination and directing them selectively toward surface reduction reactions. This charge migration pathway can therefore be more accurately described as a trap-state-mediated system,⁷³ in which the [Mo₃S₄Cl₃(en)₃]Cl cluster functions simultaneously as an electron reservoir and co-catalyst, facilitating interfacial charge transfer and prolonging charge carrier lifetime. This distinction highlights that, unlike in type-II or Z-scheme systems, the captured electrons are stored and re-emitted from localized molecular traps rather than directly recombining or flowing across a semiconductor interface. This mechanism effectively promotes spatial separation of charge carriers and accounts for the superior photocatalytic performance of the hybrid, particularly at 2.5 wt% cluster loading (Fig. S5).

The colloidal behaviour of the materials in aqueous solution was further investigated through dynamic light scattering (DLS) and zeta potential measurements. These analyses are crucial for photocatalytic applications, as they offer insights into the particle size distribution, stability, and surface charge, all of which can significantly impact catalyst–analyte interactions, dispersion, and overall efficiency. The DLS results revealed that g-C₃N₄ exhibited a relatively large hydrodynamic diameter of approximately 1100 nm, suggesting significant agglomeration in solution. Upon anchoring the [Mo₃S₄Cl₃(en)₃]Cl cluster, this size was markedly reduced to approximately 580 nm, 610 nm, and 780 nm for increasing cluster concentrations (Fig. 4A). This reduction indicates that the Mo₃S₄ cluster facilitates the deagglomeration of g-C₃N₄ lamellae, likely by interacting with intrinsic vacancies within the material. Such interactions disturb the π–π stacking between layers, as also suggested by the XRD and PL results. This deagglomeration effect is particularly beneficial for photocatalysis, as it enhances the accessibility of active surface sites, thereby improving charge transfer and catalytic activity.

Fig. 4 (A) Hydrodynamic size determined by DLS; (B) zeta potential values as a function of pH and cluster concentration.

Additionally, the zeta potential of the samples was measured as a function of pH, ranging from 6 to 10 (Fig. 4B). For all samples, a clear trend was observed: the zeta potential increased as the pH decreased, indicating changes in surface charge and protonation states under different conditions. Furthermore, as the concentration of [Mo₃S₄Cl₃(en)₃]Cl increased, the zeta potential values also increased, suggesting that the cluster contributes to a more positively charged surface. This shift in surface charge can significantly influence the interaction between the photocatalyst and the analyte, potentially affecting the selectivity and efficiency of the photocatalytic process.

Photocatalysis

Photocatalytic degradation experiments of CIP were conducted to evaluate the efficiency of the synthesized materials under visible light. Initially, all samples were kept under dark conditions for 60 minutes to ensure the establishment of adsorption/desorption equilibrium. Following this period, the suspensions were exposed to visible light for up to 180 minutes, and the concentration of CIP was monitored throughout the process by UV-Vis spectroscopy (Fig. 5A). The initial pH of the CIP solution was 8.3. In the photolysis of CIP, less than 7% degradation was observed after 180 minutes of light exposure, indicating that CIP is relatively stable under the tested conditions. In contrast, g-C₃N₄ showed significant adsorption capacity, removing approximately 40% of the initial CIP content during the dark phase. Upon visible light irradiation, an additional 20% degradation was achieved, resulting in a total removal of about 60% of the initial CIP concentration. For the sample containing 2.5 wt% of the [Mo₃S₄Cl₃(en)₃]Cl cluster, a lower adsorption capacity was observed, with around 24% adsorption within the first hour in the dark; however, complete degradation of CIP was achieved after 150 minutes of light irradiation. Interestingly, the samples with 5.0 and 10.0 wt% cluster loading exhibited increased adsorption capacities of approximately 47% and 62%, respectively. This variation in adsorption behavior is likely related to changes in the surface charge of the materials upon cluster anchoring, as previously confirmed by zeta potential measurements. Regarding photocatalytic efficiency, complete degradation of CIP was achieved within 180 minutes for both the 5.0 and 10.0 wt% cluster-containing samples. To further confirm that the 2.5 wt% cluster composition was indeed the most effective, an additional sample with 1.25 wt% cluster was synthesized and its photocatalytic activity was assessed (Fig. S6). This sample exhibited inferior photocatalytic performance compared to the 2.5 wt% material, validating that there is an optimal concentration of clusters that maximizes catalytic activity.

Fig. 5 (A) Photocatalytic degradation profiles of CIP; (B) kinetic fitting using a pseudo-first-order model; (C) TOC removal over time; (D) stability of the photocatalyst over five reuse cycles.

To determine the photocatalytic degradation kinetics, the experimental data were fitted to a pseudo-first-order kinetic model (Fig. 5B).⁷⁴ All kinetic fits yielded correlation coefficients (R²) greater than 0.950, indicating a good fit to the proposed model. For g-C₃N₄, the rate constant (k) was calculated to be 0.0023 ± 0.0003 min⁻¹. Upon cluster anchoring, the k values significantly increased to 0.0292 ± 0.0015 min⁻¹, 0.0185 ± 0.0011 min⁻¹, and 0.0142 ± 0.0006 min⁻¹ for the 2.5%, 5%, and 10% cluster-containing samples, respectively. These results demonstrate an over 12-fold enhancement in the reaction rate for the 2.5 wt% sample compared to pure g-C₃N₄, clearly confirming the role of the cluster in boosting the photocatalytic degradation of CIP. Table 1 summarizes the key parameters of this study, alongside those of previously reported systems for CIP photodegradation, highlighting the improved efficiency achieved with the synthesized photocatalysts.

Table 1 Comparative analysis of photocatalytic degradation parameters for CIP using different catalytic systems

Photocatalyst	[Catalyst] (mg mL^−1)	CIP concentration (mg L^−1)	Irradiation source	Time (min)	k (min^−1)	TOC removal (%)	Ref.
TiO2/g-C3N4	0.25	6.63	UV-Vis (500 W)	60	0.0259	—	75
g-C3N4/ZnIn2S4	1	25	UV-Vis (350 W)	60	0.0452	—	76
Black Ti^3+/TiO2 : N	0.4	0.5	Visible LED (5 W)	140	0.076	82	77
BiOCl/diatomite	0.5	10	UV-Vis (350 W)	240	—	42.9	78
g-C3N4/Fe2O3	0.3	25	UV lamp (36 W)	60	0.0470	93.86	79
Ag/AgBr/BiVO4	_	10	Visible (300 W)	120	0.0192	52.33	80
CeO2/Ag/AgBr	1	10	Visible (300 W)	120	0.0201	—	81
La2Ti2O7/Bi5O7I	0.5	5	Visible (300 W)	40	0.0940	—	82
ZnO/CD NCs	0.33	12	UV (30 W)	120	0.030	—	83
g-C3N4/[Mo3S4Cl3(en)3]Cl	1	10	Visible (30 W)	180	0.0292	65	This work

---

The photocatalytic degradation of CIP primarily targets the chromophore structure responsible for its characteristic UV-Vis absorption, indicating that the compound has been decomposed when broken down. However, the loss of the chromophore does not necessarily imply complete mineralization. To assess the extent of mineralization, total organic carbon (TOC) measurements were carried out for all samples. As shown in Fig. 5C, photolysis resulted in negligible TOC removal, indicating that the molecule remained largely intact despite exposure to the light. In contrast, g-C₃N₄ achieved around 10% mineralization, indicating partial breakdown beyond chromophore degradation. Notably, the hybrid materials containing the [Mo₃S₄Cl₃(en)₃]Cl cluster demonstrated significantly enhanced mineralization, reaching 65%, 58%, and 50% for the 2.5, 5.0%, and 10.0 wt% cluster-loaded samples, respectively. These results clearly show that both the degradation efficiency and mineralization potential are improved through the anchoring of the cluster onto the g-C₃N₄ surface. This enhancement is likely attributed to an increase in charge separation and longer carrier lifetimes, as previously discussed. Furthermore, Fig. S7 illustrates that TOC removal increases progressively with irradiation time for the sample with 2.5% of the cluster.

The stability and reusability of the photocatalyst were assessed through consecutive photocatalytic cycles under visible light (Fig. 5D). After each cycle, the material was recovered, washed with distilled water, and reused under the same experimental conditions. As shown in the results, even after five successive cycles, the catalyst maintained a mineralization efficiency above 50%, despite a gradual decrease in performance. This slight reduction may be attributed not only to partial surface fouling but also to material loss during recovery steps between cycles. Importantly, Fig. S8 presents TEM images of the recycled catalyst, revealing no significant morphological changes compared to the fresh material, further confirming the structural integrity of the photocatalyst throughout the reuse process. These findings support the material's good stability and practical potential for repeated use in water purification applications. Additionally, EDS analysis confirms that the cluster remains present in the recycled material, indicating that the elemental composition is preserved after multiple cycles.

XPS analysis of the fresh [Mo₃S₄Cl₃(en)₃]Cl, Fig. S9, confirm the successful incorporation of the cluster onto the g-C₃N₄ surface through Mo–N and Mo–C bonds. In the C 1 s region, the peak at 288.2 eV corresponds to N–C=N bonds from the g-C₃N₄ framework, while the feature at 286.3 eV is assigned to C–NH bonds from the cluster ligands.^84,85 The N 1 s spectrum shows the characteristic C–N=C and N–(C)₃ signals, consistent with the formation of a stable heterojunction between the cluster and the support. The Mo 3d spectrum of the fresh material displays Mo⁴⁺ (230.5 eV) and Mo⁶⁺ (232.9 eV, 235.1 eV) species,^86,87 indicating an active [Mo₃S₄] core partially covered by a thin oxide layer. The S 2p spectrum reveals anion sulfide (S²⁻) as the dominant species and a minor contribution from oxidized sulfur (SO₃²⁻), confirming partial surface oxidation. After the recycle test, the C 1s and N 1s spectra show minor changes, but the Mo–C and Mo–N peaks decrease in intensity, suggesting surface modification. The Mo 3d signal reveals oxidation of Mo⁴⁺ cations to higher oxidation states (Mo⁵⁺/Mo⁶⁺), and the S 2p signal disappears, indicating surface leaching or transformation of the [Mo₃S₄] cluster. This oxidation process may be associated with interfacial charge transfer between the cluster and g-C₃N₄, suggesting that Mo centers participate in redox interactions during photocatalysis. In this context, the [Mo₃S₄Cl₃(en)₃]Cl cluster acts as an efficient co-catalyst and electron trap, promoting charge separation and stabilizing photogenerated electrons, whereas the loss or transformation of sulfur species could further facilitate electron migration across the interface and contribute to the enhanced photocatalytic performance. However, EDS still detects Mo and S in the bulk, confirming that this degradation is mainly confined to the surface. The depletion and oxidation of surface Mo–S sites explain the reduced catalytic performance after recycling, highlighting the limited surface stability of the g-C₃N₄/[Mo₃S₄Cl₃(en)₃]Cl catalyst under the applied reaction conditions.

The optimization of the photocatalytic conditions was further explored using the most effective catalyst (i.e., g-C₃N₄–2.5MSC) with a focus on evaluating the influence of pH on the degradation process (Fig. 6A). The results revealed that under more alkaline conditions, the adsorption capacity of the material toward CIP remained comparable to that observed at the initial pH (8.3). However, the photocatalytic performance was significantly reduced, with degradation reaching only 60% after 180 minutes of irradiation. In contrast, under acidic conditions, a notable increase in adsorption was observed, rising to approximately 44% during the dark equilibrium phase. This enhancement in adsorption was accompanied by a substantial improvement in photocatalytic activity, achieving up to 94% degradation of CIP within 180 minutes of exposure to visible light. These findings can be understood by considering the speciation of CIP in aqueous media, which is governed by its two pK_a values: pK_a1 = 6.09, corresponding to the deprotonation of the carboxylic acid group, and pK_a2 = 8.62, corresponding to the deprotonation of the piperazinyl nitrogen (Fig. 6B).¹⁶ Although the surface of the hybrid material also becomes more positively charged at low pH, indicated by zeta potential, other factors such as hydrogen bonding, van der Waals interactions, and possible π–π stacking may contribute to enhanced adsorption despite the expected electrostatic repulsion. These interactions may facilitate the accumulation of CIP molecules on the catalyst surface, explaining the observed increase in adsorption (∼44%). Nevertheless, even with improved adsorption under acidic conditions, the total photocatalytic degradation reached 94%, which was slightly lower than the complete degradation (100%) observed at pH = 8.3.

Fig. 6 (A) Influence of solution pH on photocatalytic efficiency; (B) speciation of CIP based on its pK_a values; (C) effect of catalyst dosage; (D) performance in untreated tap and river water; (E) impact of common ionic species on degradation efficiency.

Fig. 6C shows the effect of catalyst loading on the photocatalytic degradation of CIP. As expected, the amount of CIP adsorbed increased proportionally with the catalyst mass, indicating more available surface area for interaction. However, a reduction in photocatalytic efficiency was observed at both lower (25 mg) and higher (75 mg) catalyst concentrations. At lower concentrations, the generation of photogenerated charge carriers is limited due to the reduced number of active sites, which compromises the degradation process. On the other hand, at higher catalyst loadings, the suspension becomes more turbid, which can hinder light penetration and decrease the effective activation of the photocatalyst. As a result, both insufficient and excessive catalyst amounts can lead to suboptimal photocatalytic performance.

While photocatalytic degradation in distilled water provides a good indication of a material's potential, it does not fully replicate real environmental conditions. To better assess the applicability of the system, additional experiments were carried out using untreated tap water and natural river water (Fig. 6D). These water sources contain various dissolved substances: tap water may include residual chlorine, bicarbonates, and metal ions from plumbing, whereas river water often contains organic matter, suspended solids, microorganisms, and diverse inorganic ions. In tap water (pH = 7.1), the adsorption of CIP decreased significantly to around 8%, and the overall photocatalytic degradation was slightly reduced to 87% after 180 minutes of visible light exposure. Similarly, in river water (pH = 6.8), adsorption was limited to about 9%, with a corresponding photocatalytic efficiency of 82%. These reductions can be attributed to competitive adsorption and light scattering or absorption caused by impurities and background constituents in the water, which may interfere with catalyst/analyte interactions and hinder effective light activation. Interestingly, as shown in Fig. S10, adjusting the pH of both tap and river water samples to 8.3 did not lead to significant improvements in photocatalytic efficiency compared to their original pH values. This suggests that the presence of interfering species, rather than pH alone, is the primary factor influencing adsorption and degradation efficiency in real water matrices.

To better understand the influence of common environmental interferents, the photocatalytic performance of the material was evaluated in the presence of individual ions at a concentration of 10 ppm, including Zn²⁺, Cu²⁺, Ca²⁺, Mg²⁺, HCO₃⁻, CO₃²⁻, SO₄²⁻, and NO₃⁻ (Fig. 6E). These species were selected based on their relevance to real water systems, particularly tap water, which often contains metal ions such as Ca²⁺, Mg²⁺, and residual bicarbonates, and river water, which may include a broader range of inorganic anions and dissolved metals from both natural and anthropogenic sources. The results showed that all tested ions caused a reduction in photocatalytic degradation efficiency, indicating that competitive adsorption, surface deactivation, or light scattering effects may be involved. Among the metal ions, Zn²⁺ had the lowest impact, with degradation efficiency remaining at 93%. In contrast, Cu²⁺, Ca²⁺, and Mg²⁺ led to more significant decreases, with efficiencies dropping to 76%, 80%, and 81%, respectively. These effects may be attributed to partial occupation of active sites or interactions with charge carriers, reducing the number of available reactive species for degradation. More pronounced effects were observed with CO₃²⁻, SO₄²⁻, and NO₃⁻ ions, which further reduced efficiency, yielding values between 64% and 69%. These anions can interfere with radical-mediated pathways or alter the local ionic environment around the catalyst, ultimately hindering the photocatalytic process. These findings align with the observed performance losses in untreated tap and river water, confirming that the presence of background ions can significantly influence the efficiency of photocatalytic systems and should be considered when designing applications for real-world water treatment.

Mechanism

ROS photogeneration. To better understand the photocatalytic mechanism, a series of scavenger experiments were performed to identify the reactive oxygen species (ROS) involved in the degradation of CIP (Fig. 7A). Selective scavengers were used to quench ·OH, ·O₂⁻, and singlet oxygen (¹O₂) species, as well as to suppress high electron density (e⁻) and low electron density (h⁺). In all cases, a noticeable decrease in degradation efficiency was observed, indicating that all these species might be involved in the photocatalytic process. The most significant effects were observed when ¹O₂ and ·OH were scavenged, highlighting their dominant contribution.

Fig. 7 (A) Effect of different scavengers on degradation efficiency; (B) detection of ·OH using coumarin probe; (C) detection of ¹O₂ using DMA; (D) proposed photocatalytic mechanism for the g-C₃N₄/[Mo₃S₄Cl₃(en)₃]Cl clusters.

However, due to the complex cascade of reactions involved in ROS generation, scavenger experiments alone cannot offer a fully accurate mechanistic process. Therefore, specific probe molecules were employed to monitor ROS production directly. Formation of ·OH species was tracked using coumarin, which reacts with ·OH to form 7-hydroxycoumarin, a compound with characteristic fluorescence.⁸⁸ A gradual increase in emission intensity was observed over time, confirming the continuous generation of ·OH radicals (Fig. 7B). For ¹O₂, 9,10-dimethylanthracene (DMA) was used, as this molecule selectively reacts with ¹O₂, leading to a decrease in its absorbance.⁸⁹ A progressive reduction in DMA absorbance was recorded during light exposure, confirming the formation of ¹O₂ by the catalyst (Fig. 7C).

The enhanced activity observed in the g-C₃N₄/[Mo₃S₄Cl₃(en)₃]Cl hybrid material can be attributed to both structural and electronic modifications induced by the anchored cluster (Fig. 7D), which partially disrupts the π–π stacking between g-C₃N₄ layers, increasing the surface area and improving dispersion in solution. This disaggregation effect creates more accessible active sites for the adsorption and subsequent transformation of pollutants. On the other hand, the molecular Mo₃S₄-based cluster introduces localized energy levels within the band gap of g-C₃N₄, which act as trapping centers for photogenerated electrons to delay recombination, effectively prolonging the lifetime of charge carriers and enhancing their participation in redox reactions. Upon visible light excitation, electron-deficient regions oxidize water molecules to generate ·OH and protons (H⁺).⁵² The electrons, in turn, are funnelled through the Mo₃S₄ cluster trap states and transferred to adsorbed oxygen molecules in electron-rich regions, forming ·O₂⁻. These can either react with protons to yield hydroperoxyl radicals (·OOH) or undergo further reduction to generate ¹O₂.⁹⁰ These ROS are highly oxidative and attack CIP at multiple sites, leading to progressive fragmentation of its molecular structure and, ultimately, to mineralization. The synergistic effect of the modified electronic structure of the hybrid material is thus central to its superior photocatalytic activity.

CIP degradation. The photodegradation of CIP is a complex process that may involve the generation of multiple intermediate species. In this work, the main intermediates have been identified using high-resolution mass spectrometry (HRMS) based on their m/z values (see Fig. S10). According to the reactive species previously detected, several tentative pathways for CIP photodegradation using the hybrid g-C₃N₄/[Mo₃S₄Cl₃(en)₃]Cl material are proposed, as illustrated in Fig. 8.

Fig. 8 Schematic representation of the tentative CIP photodegradation pathways using the hybrid g-C₃N₄/[Mo₃S₄Cl₃(en)₃]Cl material.

The observed degradation pathways can be classified into three main routes, i.e. oxidation reactions, hydroxylation, and decarboxylation processes. In pathway A, the ciprofloxacin molecule undergoes oxidation of both the piperazine ring and the quinolone moiety, leading to the formation of intermediate CIP I.³⁰ Then, after defluorination and oxidation of the aromatic ring, the compound CIP II is obtained. Hydroxylation processes drive pathways B and C. In the former one, hydroxyl radicals attack the ciprofloxacin molecule, reducing the piperazine ring to generate CIP III.⁹¹ This species further decomposes into CIP IV, which after cleavage of both the cyclopropyl and carboxyl groups, results in compound CIP V.

According to Pathway C, ·OH similarly attack the ciprofloxacin structure to obtain intermediate CIP VI via loss of cyclopropyl and ethylene groups.⁹² Then, after continuous decomposition intermediates CIP VII and CIP VIII can be generated. Finally, in Pathway D, the ciprofloxacin molecule undergoes decarboxylation and cyclopropyl group cleavage to form CIP IX, also mediated by ·OH.⁹³ This compound upon further hydroxylation may undergo different routes: dehydrogenation on the piperazine group to obtain CIP X, or ring opening of the quinolone system to yield intermediate CIP XI. Then, the final products of the proposed pathways can be converted into non-identified small molecules. Upon attack by free radicals, these species are further mineralized into products such as H₂O, CO₂, NH₄⁺, F⁻, etc.

Experimental

Synthesis

Synthesis of g-C₃N₄. g-C₃N₄ was synthesized via the thermal decomposition of urea (CH₄N₂O, 99%, Sigma Aldrich). In a typical procedure, 10 g of urea was weighed and placed into a sealed porcelain crucible. The precursor was then thermally treated in a muffle furnace at 550 °C for 3 h with a heating rate of 5 °C min⁻¹.

Synthesis of [Mo₃S₄Cl₃(en)₃]Cl. The [Mo₃S₄Cl₃(en)₃]Cl cluster compound was prepared according to the published procedure.³² A solution of the trinuclear Mo₃S₄Cl₄(PPh₃)₃(H₂O)₂ (500 mg, 0.362 mmol) cluster precursor in dry CH₃CN (40 mL) was reacted at room temperature under an inert atmosphere with ethylenediamine (80 µL, 1.191 mmol) for 4 hours. The resulting solid was filtered, washed with CH₃CN (30 mL), and dissolved in CH₃OH. Next, 3.0 mL of a 0.5 mol L⁻¹ HCl solution was added, leading to the partial precipitation of a green solid. The precipitate was filtered, washed with cold CH₃OH (20 mL) and CH₂Cl₂ (50 mL) to remove the organic impurities, obtaining 188.0 mg (70%) of the desired complex. ¹H-NMR (400 MHz, DMSO-d₆): δ = 8.07 (m, 3H, NH, H_A), 6.84 (m, 3H, NH, H_B), 4.51 (m, 3H, NH₂, H_C), 3.54 (br, 3H, CH₂, H_D; 3H, NH₂, H_E), 3.10 (br, 3H, NH₂, H_F), 2.67 (br, 6H, CH₂, H_G). HRMS (ESI-TOF) (20 V, CH₃CN) m/z [M]⁺ Calc for Mo₃S₄C₆H₂₄N₆Cl₃: 702.7156. Found 702.7170. Elemental analysis calc. (%) for Mo₃S₄C₆H₂₄N₆Cl₄: C 9.8, H 3.3, N 11.4, S 17.4. Found C 9.2, H 3.6, N 11.2, S 16.9.

Incorporation of [Mo₃S₄Cl₃(en)₃]Cl into g-C₃N₄. Solid g-C₃N₄ (50 mg) was dispersed in 25 mL of an aqueous green solution of the [Mo₃S₄Cl₃(en)₃]Cl (MSC) cluster salt at three different concentrations of 2.5, 5.0, and 10 wt% to promote its adsorption onto the g-C₃N₄ surface. The mixture was stirred overnight at room temperature and in the dark to ensure equilibrium between adsorption and desorption processes. At the end of the process, a light-green powder was separated by filtration, and the resulting solution appeared completely transparent, indicating the adsorption of the cluster. The light-green solid was thoroughly washed with water and dried in an oven at 60 °C overnight. The three materials obtained were named as C₃N₄–2.5MSC, C₃N₄–5.0MSC, and C₃N₄–10.0MSC depending on the initial metal sulfide cluster (MSC) concentration.

Characterizations

XRD patterns were recorded using a Bruker D4-Endeavor diffractometer equipped with Cu Kα radiation (λ = 1.5406 Å). Data collection was performed in the 2θ range of 10° to 70°, with a step size of 0.02°. FTIR spectra were acquired at ambient temperature using a Jasco FT/IR-6200 instrument, operating over a spectral range of 600 to 4000 cm⁻¹, with a resolution of 4 cm⁻¹ and averaging 32 scans per sample. The surface morphology of the materials was investigated by SEM using a LEO 440i Leica–Zeiss system at an accelerating voltage of 15 kV. At the same time, internal structure and nanoscale features were examined via TEM on a JEOL JEM-2100Plus microscope coupled with an Oxford INCA 250 EDS detector. Measurements of zeta potential and DLS were performed using a Zetasizer Nano ZS (Malvern Instruments, UK). For zeta potential evaluation, pH values were adjusted using 0.1 M NaOH (Synth) and glacial acetic acid (Sigma-Aldrich, purity >99.7%). DRS measurements were carried out using a Cary 5G UV-Vis spectrophotometer (Varian, USA). The optical band gap energy (E_gap) of the materials was estimated by transforming the reflectance data into Tauc plots, applying the Kubelka–Munk function. PL spectra were recorded under steady-state conditions using a Fluorolog 3-11 spectrofluorometer (Horiba), with an excitation wavelength set at 350 nm.

Elemental analyses were performed with a Euro EA 3000 Elemental Analyzer. Mass spectra were registered in a QTOF Premier instrument operated in the V-mode at a resolution of ca. 10 000 (FWHM) and a triple quadrupole mass spectrometer, both of them were equipped with an orthogonal Z-spray–electrospray interface (Waters, Manchester, UK). The temperature of the source block was set to 100 °C, and the desolvation temperature was set to 120 °C. A capillary voltage of 3.3 kV was used in the positive scan mode, and the cone voltage was set to U_c = 20 V. Sample solutions in CH₃CN were injected with a syringe pump directly connected to the ESI source at a flow rate of 10 µL min⁻¹. The observed isotopic pattern of each compound perfectly matched the theoretical isotope pattern calculated from their elemental composition by using the MassLynx 4.1 program. ¹H-NMR spectrum was recorded on a Bruker Avance III HD 400 MHz spectrometer.

XPS analysis was performed on a Scienta-Omicron ESCA (Al Kα hν = 1486.6 eV). Casa XPS software (version 2.3.15) was used for elements deconvolution. EPR spectra were acquired on a EMX Plus Bruker X-band EPR spectrometer (∼9.8 GHz of MW frequency) with a TM110 cylindrical front-grid cavity installed. The same parameters were used for all measurments: 0.6325 mW of MW power, 100 kHz of modulation frequency and 1 G of modulation amplitude. For the detection of solid paramagnetic centers, spectra were recorded using a Wilmad liquid N₂ Dewar (77 K) placed inside the resonator. To compare the signals of structural defects between the samples, each photocatalyst was properly weighed and the signal was normalized by the mass inserted in the EPR quartz tube.

The (photo)electrochemical tests were performed in a PGSTAT302N (Autolab, Mehtrom®) potentiostat and a solar simulator light source (100 mW cm², Newport®). For the determination of the flat-band potential (E_FB), the prepared samples were deposited on clean fluorine-doped tin oxide (FTO) substrates (Sigma Aldrich) using spray-drying (6 mg mL⁻¹ of the photocatalyst in ethanol for 60 s). Mott–Schottky analysis was performed by measuring the electrochemical impedance spectra in potentiostatic scan mode, ranging from 1.0 to −1.0 V vs. Ag/AgCl, with an amplitude of 10 mV and a frequency of 1 kHz, in a 0.50 M Na₂SO₄ (Sigma Aldrich, purity 99%) aqueous solution (pH 7). The electrochemical tests were performed in a three-electrode cell with an Ag/AgCl (KCl-saturated) electrode as the reference electrode and a carbon rod as the counter electrode. The transient photocurrent density polarization was performed at 1.10 V vs. Ag/AgCl. Electrochemical Impedance Spectroscopy analysis was performed at 1.10 V vs. Ag/AgCl with a frequency range of 1 kHz to 0.1 Hz. For the LSV measurements, the photocatalyst was coated on carbon paper TGP-H-60, which was PTFE-treated. The polarization curves were registered from 0.1 to −1.0 V vs. Ag/AgCl at 5 mV mV s⁻¹ in 0.50 M Na₂SO₄ solution. The potentials were converted to the RHE scale by applying the following equation:

E_RHE = E_Ag/AgCl + 0.0592pH + 0.197

Photocatalysis

General procedure. Photocatalytic degradation tests were performed using a 10 ppm CIP (98%, Sigma Aldrich) aqueous solution. In each experiment, 50 mg of the photocatalyst sample was dispersed in 50 mL of the CIP solution. Initially, the suspensions were stirred in the dark for 60 minutes to assess adsorption equilibrium. Subsequently, the samples were exposed to visible light for 180 minutes under continuous stirring. Aliquots were withdrawn at predetermined intervals and analyzed at CIP maximum absorption wavelength (276 nm) using a UV-Vis spectrophotometer (JASCO V-660). Visible light irradiation was provided by a standard 30 W IP67 LED floodlight, delivering a light intensity of 1.7 mW cm⁻² at the reactor surface (20 cm), as measured using a Solar Hukseflux radiometer. The catalytic system was maintained at 25 °C. After 180 minutes of irradiation, the TOC content of the solutions was measured using a GE Sievers InnovOx analyzer to evaluate the extent of mineralization. Mass spectra were registered in a QTOF Premier instrument operated in the V-mode at a resolution of ca. 10 000 (FWHM) and a triple quadrupole mass spectrometer, both equipped with an orthogonal Z-spray–electrospray interface (Waters, Manchester, UK). The temperature of the source block was set to 100 °C, and the desolvation temperature was set to 120 °C. A capillary voltage of 3.3 kV was used in the positive scan mode, and the cone voltage was set to U_c = 20 V. Sample solutions in CH₃CN were injected with a syringe pump directly connected to the ESI source at a flow rate of 10 µL min⁻¹. The observed isotopic pattern of each compound perfectly matched the theoretical isotope pattern calculated from their elemental composition by using the MassLynx 4.1 program.

For pH-dependent studies, the solution pH was adjusted using 0.1 M NaOH (Synth) or glacial acetic acid (Sigma-Aldrich, >99.7%). Tap water was collected at Universitat Jaume I, and lake water samples were obtained from “La font de la Presola” (Les Useres, Castelló, 40.166833951338475, −0.156654273143594). To assess the influence of common inorganic ions, additional tests were conducted by adding 10 ppm (equal to CIP concentration) of the following salts: CuCl₂, ZnCl₂, CaCl₂, MgCl₂, NaHCO₃, K₂CO₃, Na₂SO₄, and NaNO₃. Scavenger experiments were also conducted to identify the primary reactive species involved in CIP degradation. Equimolar amounts of scavengers relative to CIP were added: tert-butanol (TBA) for ·OH, p-benzoquinone (p-BQ) for ·O₂⁻, ascorbic acid (AA, 99%, Sigma Aldrich) for singlet oxygen (¹O₂), AgNO₃ for high electron density, and ammonium oxalate (AO, Sigma Aldrich) for low electron density. All the photocatalytic tests were performed in triplicate.

Detection of ·OH and ¹O₂via spectroscopic methods. To monitor the formation of ·OH, a solution of coumarin (15 µL of a 1000 ppm stock, Sigma Aldrich) was diluted in 10 mL of acetonitrile and then exposed to the catalytic system under study. Samples were collected at specific time intervals and analyzed using a Shimadzu RF-5301 PC spectrofluorophotometer. The fluorescence measurements were performed with an excitation wavelength of 332 nm, and the emission spectra were recorded in the 350–600 nm range, using a 5 nm slit width. ¹O₂ generation was assessed using DMA (Sigma Aldrich) as the probe. A 2 mL aliquot of a 1 × 10⁻⁴ mol L⁻¹ solution in acetonitrile (98%, Aldrich) was used, and the mixture was similarly exposed to the photocatalytic conditions. At predetermined intervals (0-30 minutes), aliquots were withdrawn and analyzed using a Jasco V-660 UV-Vis spectrophotometer, monitoring the absorbance in the 300–450 nm range to track the degradation of the probe molecule.

Conclusions

This study successfully demonstrates the development of a novel photocatalytic system based on g-C₃N₄ modified with the molecular [Mo₃S₄Cl₃(en)₃]Cl cluster for the efficient degradation of the emerging pharmaceutical contaminant CIP under visible light. Through a combination of structural, spectroscopic, and surface charge analyses, it was shown that the anchoring of the Mo₃S₄ cluster induces both electronic and morphological modifications in the g-C₃N₄ framework. These changes include the disruption of π–π stacking interactions between the lamellar sheets and the introduction of trap states within the band gap, leading to enhanced charge separation and suppressed electron recombination.

Electrochemical analyses further supported these findings, revealing that the introduction of the [Mo₃S₄Cl₃(en)₃]Cl cluster led to a positive shift in the flat-band potential, consistent with enhanced electron mobility. Transient photocurrent measurements revealed faster photoinduced charge separation for the g-C₃N₄–2.5-MSC sample, whereas electrochemical impedance spectroscopy indicated a reduction in charge transfer resistance compared to pristine material. Despite the better charge transport properties in higher-loaded samples, their photocurrent response decreased due to increased recombination, highlighting the importance of optimizing cluster concentration. Polarization curves recorded in the dark further revealed a rise in current density and a positive shift in the onset potential for hydrogen evolution in modified samples, confirming the electrocatalytic contribution of the [Mo₃S₄Cl₃(en)₃]Cl cluster.

The optimized material containing 2.5% cluster achieved complete degradation of CIP in 150 minutes and a mineralization rate of 65%, representing a significant improvement over g-C₃N₄. Kinetic modeling confirmed a pseudo-first-order behavior with a twelve-fold increase in the rate constant. The material maintained high efficiency under various experimental conditions, including variations in pH and catalyst loading. Additionally, it demonstrated promising results in real water matrices, including untreated tap and river water, even in the presence of competing ions and natural organic matter. It highlights the robustness of the hybrid system under environmentally relevant conditions. The recyclability test demonstrated the stability and reusability of the heterojunction under environmentally relevant conditions, retaining over 50% of its mineralization capacity after five reuse cycles. Post-catalysis TEM imaging confirmed that the material maintained its morphology, suggesting minimal structural degradation during the process. The observed decrease in activity was attributed primarily to partial mass loss during recovery and minor surface passivation.

Interference studies revealed that the presence of common ions can moderately reduce photocatalytic efficiency. ROS analysis confirmed that ·OH and ¹O₂ are the primary oxidative species involved in the degradation process. The generation of these species was validated using both scavenger molecules and selective probes, which demonstrated increased ROS production during irradiation. A mechanistic pathway was proposed in which Mo₃S₄-based clusters act as electron traps, facilitating directional migration of charge carriers and enabling the formation of multiple ROS species through sequential redox reactions.

Taken together, these findings provide a comprehensive understanding of how strategic electronic modification of g-C₃N₄ using transition metal clusters can overcome the intrinsic limitations of the pristine material. This work demonstrates not only the potential of [Mo₃S₄Cl₃(en)₃]Cl cluster-modified g-C₃N₄ as a high-performance, visible-light-responsive photocatalyst, but also sets a foundation for future developments in water treatment technologies targeting persistent organic pollutants. By combining enhanced photoreactivity with improved electronic conductivity and electrocatalytic efficiency, the material offers a robust platform for advanced oxidation processes under real environmental conditions. The combination of enhanced reactivity, structural durability, synergistic charge modulation, and adaptability to real water conditions makes [Mo₃S₄Cl₃(en)₃]Cl cluster-modified g-C₃N₄ a strong candidate for integration into practical wastewater remediation systems.

Author contributions

M. L. G. R.: methodology, investigation, data curation, formal validation, writing – original draft, writing – review & editing, visualization. J. M.-C., A. B. S., M. G.-B., J. B. G. F.: methodology, investigation, data curation, writing – original draft, writing – review & editing. C. R., J. A., R. L., and E. L: methodology, data curation, funding acquisition, validation, resources, writing – original draft, writing – review & editing. M. A.: conceptualization, methodology, investigation, data curation, formal validation, resources, writing – original draft, supervision, project administration, writing – review & editing, visualization, project administration.

Conflicts of interest

There are no conflicts to declare.

Data availability

Data for this article, including characterization and photocatalytic data are available at Zenodo at https://doi.org/10.5281/zenodo.16614608.

Supplementary information (SI): cluster characterization, flat band potentials, polarization curves for HER, electron transfer pathway, and photocatalytic tests. See DOI: https://doi.org/10.1039/d5ta06478c.

Acknowledgements

This study was financed, in part, by the São Paulo Research Foundation (FAPESP), Brasil (#2013/07296-2, #2023/08525-7, #2023/02367-0, #2022/09717-4, #2023/10329-1, 2024/23525-6), the Conselho Nacional Desenvolvimento Científico e Tecnologico (CNPq), and the Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES, #001). The financial support of the Spanish Ministerio de Ciencia e Innovación (PID2022-141089NB-I00), Ministerio de Economía y Competitividad (PID2019-107006GB-C22), Generalitat Valenciana (Grant CIAICO/2021/122), and Universitat Jaume I (UJI-B2021-29 and UJI-B2022-56) is gratefully acknowledged for J. A and R. sL. The Chemistry Institute of São Carlos, University of São Paulo (IQSC-USP).

References

1. F. Malerba and L. Orsenigo, Bus. Hist., 2015, 57, 664–687.
2. D. G. J. Larsson and C.-F. Flach, Nat. Rev. Microbiol., 2022, 20, 257–269.
3. G. Buttiglieri and T. P. Knepper, Removal of Emerging Contaminants in Wastewater Treatment: Conventional Activated Sludge Treatment, in Emerging Contaminants from Industrial and Municipal Waste, The Handbook of Environmental Chemistry, ed. D. Barceló and M. Petrovic, Springer Berlin Heidelberg, Berlin, Heidelberg, 2008, pp. 1–35.
4. Y. Luo, W. Guo, H. H. Ngo, L. D. Nghiem, F. I. Hai, J. Zhang, S. Liang and X. C. Wang, Sci. Total Environ., 2014, 473–474, 619–641.
5. M. Patel, R. Kumar, K. Kishor, T. Mlsna, C. U. Jr. Pittman and D. Mohan, Chem. Rev., 2019, 119, 3510–3673.
6. V. O. Ajekiigbe, C. E. Agbo, I. J. Ogieuhi, C. S. Anthony, C. S. Onuigbo, T. A. Falayi, O. Z. Oluwapelumi, O. Amusa, G. O. Adeniran, P. O. Ogunleke and I. S. Bakare, Discov. Public Health, 2025, 22, 120.
7. G. Wieërs, S. Absil, I. Maystadt, C. Nicaise, P. Modrie, F.-X. Sibille, L. Melly and J.-M. Dogné, Front. Med., 2024, 11, 1438636.
8. M. A. Jackson, G. E. Schutze and Committee on Infectious Diseases, Pediatrics, 2016, 138, e20162706.
9. M. Sarafraz, A. Shakir, S. Mohsen, H. Zoha, B. Abotaleb, F. Yadolah and M. Khaneghah, Int. J. Environ. Anal. Chem., 2022, 102, 5106–5121.
10. S. Zhou, C. Di Paolo, X. Wu, Y. Shao, T.-B. Seiler and H. Hollert, Environ. Int., 2019, 128, 1–10.
11. A. Jurado, M. Walther and M. S. Díaz-Cruz, Sci. Total Environ., 2019, 663, 285–296.
12. A. K. Al-Buriahi, M. M. Al-shaibani, R. M. S. R. Mohamed, A. A. Al-Gheethi, A. Sharma and N. Ismail, J. Water Proc. Eng., 2022, 47, 102725.
13. D. Zhu and Q. Zhou, Environ. Nanotechnol. Monit. Manag., 2019, 12, 100255.
14. Z. Kuspanov, B. Bakbolat, A. Baimenov, A. Issadykov, M. Yeleuov and C. Daulbayev, Sci. Total Environ., 2023, 885, 163914.
15. J. C. Souza, S. C. S. Lemos, M. Assis, C. H. M. Fernandes, L. K. Ribeiro, Y. Núñez-de la Rosa, M. D. Teodoro, L. Gracia, J. Andrés, L. H. Mascaro and E. Longo, ACS Omega, 2024, 9, 35537–35547.
16. L. K. Ribeiro, M. Assis, A. J. Moreira, C. B. Abreu, R. C. Gebara, G. A. Grasser, H. C. S. Fukushima, R. C. Borra, M. G. G. Melão, E. Longo and L. H. Mascaro, Chemosphere, 2024, 359, 142352.
17. P. Pascariu, C. Gherasim and A. Airinei, Int. J. Mol. Sci., 2023, 24, 9564.
18. G. Dong, Y. Zhang, Q. Pan and J. Qiu, J. Photochem. Photobiol., C, 2014, 20, 33–50.
19. A. R. Ayub, A. Ullah, F. Nawaz, S. Shafiq, A. Abd El-Fattah, H. Li and J. Iqbal, ACS Omega, 2023, 8, 19486–19493.
20. X. Liu, R. Ma, L. Zhuang, B. Hu, J. Chen, X. Liu and X. Wang, Crit. Rev. Environ. Sci. Technol., 2021, 51, 751–790.
21. N. Arif, Y. Ma, M. N. Zafar, M. Humayun, M. Bououdina, S.-Y. Zhang, Q. Zhang, X. Yang, H. Liang and Y.-J. Zeng, Small, 2024, 20, 2308908.
22. W.-J. Ong, L.-L. Tan, Y. H. Ng, S.-T. Yong and S.-P. Chai, Chem. Rev., 2016, 116, 7159–7329.
23. M. Xiao, B. Luo, S. Wang and L. Wang, J. Energy Chem., 2018, 27, 1111–1123.
24. A. Akhundi, A. Zaker Moshfegh, A. Habibi-Yangjeh and M. Sillanpää, ACS ES&T Eng., 2022, 2, 564–585.
25. W.-J. Ong, Front. Mater., 2017, 4, 11.
26. D. Ma, Z. Zhang, Y. Zou, J. Chen and J.-W. Shi, Coord. Chem. Rev., 2024, 500, 215489.
27. D. Van Thuan, T. B. H. Nguyen, T. H. Pham, J. Kim, T. T. Hien Chu, M. V. Nguyen, K. D. Nguyen, W. A. Al-onazi and M. S. Elshikh, Chemosphere, 2022, 308, 136408.
28. K. Hu, R. Li, C. Ye, A. Wang, W. Wei, D. Hu, R. Qiu and K. Yan, J. Clean. Prod., 2020, 253, 120055.
29. R. Liu, X. Han, R. Liu and Y. Sun, Mater. Lett., 2022, 325, 132811.
30. J. Chen, H. Zhu, Q. Ren, S. Chen, Y. Ding, Z. Jin, W. Guo and X. Jia, Appl. Surf. Sci., 2023, 624, 157186.
31. M. Gutiérrez-Blanco, J. Mateu-Campos, M. Oliva, V. S. Safont and R. Llusar, in Inorganic Chemistry in Spain, ed. M. Martínez, M. Ferrer and R. van Eldik, Academic Press, 2024, vol. 84, pp. 317–353.
32. J. Mateu-Campos, M. Gutiérrez-Blanco, E. Guillamón, V. S. Safont, J. Benet-Buchholz, M. Oliva and R. Llusar, Inorg. Chem., 2025, 64, 6552–6560.
33. E. Guillamón, M. Oliva, J. Andrés, R. Llusar, E. Pedrajas, V. S. Safont, A. G. Algarra and M. G. Basallote, ACS Catal., 2021, 11, 608–614.
34. M. Gutiérrez-Blanco, A. G. Algarra, E. Guillamón, M. J. Fernández-Trujillo, M. Oliva, M. G. Basallote, R. Llusar and V. S. Safont, Inorg. Chem., 2024, 63, 1000–1009.
35. V. S. Safont, I. Sorribes, J. Andrés, R. Llusar, M. Oliva and M. R. Ryzhikov, Phys. Chem. Chem. Phys., 2019, 21, 17221–17231.
36. J. Mateu-Campos, E. Guillamón, V. S. Safont, K. Junge, H. Junge, M. Beller and R. Llusar, Dalton Trans., 2024, 53, 4147–4153.
37. M. Gutiérrez-Blanco, C. A. M. Stein, C. Alfonso, E. Guillamón, V. S. Safont, I. Sorribes, H. Junge, M. Beller and R. Llusar, ChemCatChem, 2023, 15, e202300740.
38. E. Guillamón, I. Sorribes, V. S. Safont, A. G. Algarra, M. J. Fernández-Trujillo, E. Pedrajas, R. Llusar and M. G. Basallote, Inorg. Chem., 2022, 61, 16730–16739.
39. F. Gonell, M. Rodenes, S. Martín, M. Boronat, I. Sorribes and A. Corma, Chem. Mater., 2023, 35, 8483–8493.
40. S. Batool, S. P. Nandan, S. N. Myakala, A. Rajagopal, J. S. Schubert, P. Ayala, S. Naghdi, H. Saito, J. Bernardi, C. Streb, A. Cherevan and D. Eder, ACS Catal., 2022, 12, 6641–6650.
41. X. Liu, Y. Jiang, L. Zhang, Y. Zhang, Y. Li, X. Xu, K. Lin and Y. Du, Int. J. Hydrogen Energy, 2020, 45, 21014–21023.
42. P. Adams, J. Bühler, I. Walz, T. Moehl, H. Roithmeyer, O. Blacque, N. Comini, J. T. Diulus, R. Alberto, S. Siol, M. Dimitrievska, Z. Novotny and S. D. Tilley, ACS Energy Lett., 2024, 9, 3828–3834.
43. H. Ou, L. Lin, Y. Zheng, P. Yang, Y. Fang and X. Wang, Adv. Mater., 2017, 29, 1700008.
44. Y. Qin, J. Lu, X. Zhao, X. Lin, Y. Hao, P. Huo, M. Meng and Y. Yan, Chem. Eng. J., 2021, 425, 131844.
45. D. Feng, Y. Cheng, J. He, L. Zheng, D. Shao, W. Wang, W. Wang, F. Lu, H. Dong, H. Liu, R. Zheng and H. Liu, Carbon, 2017, 125, 454–463.
46. S. Yuan, Z. Wang, Z. Qiao, M. Wang, J. Wang and S. Wang, J. Membr. Sci., 2011, 378, 425–437.
47. S. Duval, F. Dumur, L. Guénée, J. Marrot, C. Simonnet-Jégat and E. Cadot, Eur. J. Inorg. Chem., 2013, 2013, 1149–1156.
48. A. Ben-Refael, I. Benisti and Y. Paz, Catal. Today, 2020, 340, 97–105.
49. P. Praus, L. Svoboda, M. Ritz, I. Troppová, M. Šihor and K. Kočí, Mater. Chem. Phys., 2017, 193, 438–446.
50. O. P. Nagar, T. Barman, K. Marumoto, Y. Shimoi, K. Matsuishi and N. Chouhan, Int. J. Hydrogen Energy, 2024, 87, 526–538.
51. S. Pareek, M. Sharma, S. Lal and J. K. Quamara, J. Mater. Sci.: Mater. Electron., 2018, 29, 13043–13051.
52. S. C. S. Lemos, T. K. d. L. Rezende, M. Assis, F. d. C. Romeiro, D. A. Peixoto, E. d. O. Gomes, G. M. Jacobsen, M. D. Teodoro, L. Gracia, J. L. Ferrari, E. Longo, J. Andrés and R. C. de Lima, Mater. Res. Bull., 2022, 152, 111849.
53. V. Hasija, P. Singh, S. Thakur, K. Stando, V.-H. Nguyen, Q. Van Le, S. M. Alshehri, T. Ahamad, K. C.-W. Wu and P. Raizada, Appl. Mater. Today, 2022, 29, 101676.
54. I. M. D. Gonzaga, R. Gonçalves, C. H. M. Fernandes, M. Assis, I. F. Teixeira and L. H. Mascaro, Chemosphere, 2024, 352, 141315.
55. C. Rosales-Martínez, M. Assis, C. Castillo-Blas and I. Abánades Lázaro, Chem. Commun., 2024, 60, 8280–8283.
56. B. Choudhury, K. K. Paul, D. Sanyal, A. Hazarika and P. K. Giri, J. Phys. Chem. C, 2018, 122, 9209–9219.
57. M. Rezaei, A. Nezamzadeh-Ejhieh and A. R. Massah, Energy Fuels, 2024, 38, 8406–8436.
58. S. Patnaik, D. P. Sahoo and K. Parida, Carbon, 2021, 172, 682–711.
59. C. Rosales-Martínez, D. López-Alcalá, M. Assis, C. Castillo-Blas, J. J. Baldoví and I. Abánades Lázaro, RSC Adv., 2024, 14, 37984–37992.
60. M. Assis, A. C. M. Rennó, J. Andrés and E. Longo, ACS Omega, 2025, 10, 22323–22346.
61. D. W. deQuilettes, K. Frohna, D. Emin, T. Kirchartz, V. Bulovic, D. S. Ginger and S. D. Stranks, Chem. Rev., 2019, 119, 11007–11019.
62. M. A. R. da Silva, I. F. Silva, Q. Xue, B. T. W. Lo, N. V. Tarakina, B. N. Nunes, P. Adler, S. K. Sahoo, D. W. Bahnemann, N. López-Salas, A. Savateev, C. Ribeiro, T. D. Kühne, M. Antonietti and I. F. Teixeira, Appl. Catal., B, 2022, 304, 120965.
63. V. Giulimondi, M. Agrachev, S. Kuzin, J. M. González-Acosta, A. Ruiz-Ferrando, F. Krumeich, F. Bondino, Y.-T. Chiang, M. Vanni, G. Jeschke, N. López and J. Pérez-Ramírez, Nat. Commun., 2025, 16, 4842.
64. J. B. G. Filho, L. F. G. Noleto, G. D. Iga, F. S. Gonçalves, D. R. Cardoso, I. F. Teixeira and C. Ribeiro, Chem. Eng. J., 2025, 521, 166306.
65. A. Mehtab, S. A. Ali, P. P. Ingole, Y. Mao, S. M. Alshehri and T. Ahmad, ACS Appl. Energy Mater., 2023, 6, 12003–12012.
66. Q. Chen, W. Mo, G. Yang, S. Zhong, H. Lin, J. Chen and S. Bai, Small, 2021, 17, 2102105.
67. C. Li, Z. Yan, G. Wang, T. Yu, Z. Xu and J. Chen, ACS Appl. Nano Mater., 2024, 7, 26672–26685.
68. D. Wang, G. Chen, J. Li, Y. Ren, Y. Zhao, W. Yang and Y. Zhu, Surf. Interfaces, 2025, 69, 106815.
69. M. B. Costa, F. W. S. Lucas, M. Medina and L. H. Mascaro, ACS Appl. Energy Mater., 2020, 3, 9799–9808.
70. M. K. Ghimire, H. Ji, H. Z. Gul, H. Yi, J. Jiang and S. C. Lim, ACS Appl. Mater. Interfaces, 2019, 11, 10068–10073.
71. H. Zhao, Z. Zhang, Y. Zhang, C. Yin, Z. Li, W. Zhou, M. Ma, D. Zhang, P. He, J. Wu, P. Zheng and Y. Ling, Int. J. Hydrogen Energy, 2025, 145, 681–691.
72. D. Ma, C. Zhi, J. Chen, X. Zhang, K. Song, J. Li, Y. Chen, Y. Zhang, Y. Zhang, Y. Gui, Y. Cheng and J.-W. Shi, ACS Mater. Lett., 2024, 6, 2497–2505.
73. Y. Zhang, D. Ma, J. Li, Y. Chen, L. Shi, X. Feng and J.-W. Shi, J. Mater. Chem. A, 2025, 13, 10431–10450.
74. M. de Jesus Silva Chaves, G. de Oliveira Lima, M. de Assis, C. de Jesus Silva Mendonça, I. M. Pinatti, A. F. Gouveia, I. L. Viana Rosa, E. Longo, M. A. P. Almeida and T. C. Rodrigues dos Santos Franco, J. Solid State Chem., 2019, 274, 270–279.
75. K. Hu, R. Li, C. Ye, A. Wang, W. Wei, D. Hu, R. Qiu and K. Yan, J. Clean. Prod., 2020, 253, 120055.
76. B. Tu, R. Che, F. Wang, Y. Li, J. Li and J. Qiu, Sep. Purif. Technol., 2024, 345, 127323.
77. M. Sarafraz, M. Sadeghi, A. Yazdanbakhsh, M. M. Amini, M. Sadani and A. Eslami, Process Saf. Environ. Prot., 2020, 137, 261–272.
78. Z. Jia, T. Li, Z. Zheng, J. Zhang, J. Liu, R. Li, Y. Wang, X. Zhang, Y. Wang and C. Fan, Chem. Eng. J., 2020, 380, 122422.
79. P. Rajiv, M. Nezamaddin, M. Gordon and D. Balarak, Int. J. Environ. Anal. Chem., 2023, 103, 2193–2207.
80. F. Chen, Q. Yang, Y. Wang, F. Yao, Y. Ma, X. Huang, X. Li, D. Wang, G. Zeng and H. Yu, Chem. Eng. J., 2018, 348, 157–170.
81. X.-J. Wen, C.-G. Niu, L. Zhang, C. Liang, H. Guo and G.-M. Zeng, J. Catal., 2018, 358, 141–154.
82. Y. Zhu, J. Xu and M. Chen, Sep. Purif. Technol., 2022, 282, 120004.
83. I. Mukherjee, V. Cilamkoti and R. K. Dutta, ACS Appl. Nano Mater., 2021, 4, 7686–7697.
84. J. Fu, K. Liu, K. Jiang, H. Li, P. An, W. Li, N. Zhang, H. Li, X. Xu, H. Zhou, D. Tang, X. Wang, X. Qiu and M. Liu, Adv. Sci., 2019, 18, 1900796.
85. L. Shen, Z. Xing, J. Zou, Z. Li, X. Wu, Y. Zhang, Q. Zhu, S. Yang and W. Zhou, Sci. Rep., 2017, 7, 41978.
86. H. Li, C. Wang, W. Ding, L. Wu, J. Wang, T. Wei, J. Hu, C. Wu, M. Chen, H. Zhang, T. Lin and W. Liao, J. Mater. Sci.: Mater. Electron., 2022, 33, 15369–15384.
87. Q. Wu, A. Abraham, L. Wang, X. Tong, E. S. Takeuchi, K. J. Takeuchi and A. C. Marschilok, J. Electrochem. Soc., 2020, 167, 050513.
88. G. A. Grasser, L. K. Ribeiro, E. Longo and M. Assis, Catal. Today, 2025, 444, 114999.
89. L. O. Libero, L. K. Ribeiro, L. I. Granone, M. S. Churio, J. C. Souza, V. R. Mastelaro, J. Andrés, E. Longo, L. H. Mascaro and M. Assis, ChemCatChem, 2023, 15, e202300421.
90. M. Assis, A. F. Gouveia, L. K. Ribeiro, M. A. Ponce, M. S. Churio, O. N. Oliveira, L. H. Mascaro, E. Longo, R. Llusar, E. Guillamón and J. Andrés, Appl. Catal., A, 2023, 119038.
91. J. G. Cuadra, S. Molina-Prados, G. Mínguez-Vega, L. A. Abderrahim, J. Colombari, J. B. Carda, N. P. F. Gonçalves, R. M. Novais and J. A. Labrincha, Appl. Surf. Sci. Adv., 2025, 27, 100751.
92. M. Jiménez-Salcedo, M. Monge and M. T. Tena, Chemosphere, 2020, 247, 125910.
93. X. Hu, X. Hu, Q. Peng, L. Zhou, X. Tan, L. Jiang, C. Tang, H. Wang, S. Liu, Y. Wang and Z. Ning, Chem. Eng. J., 2020, 380, 122366.

---