Multifunctional Z-scheme Bi-MOF/g-C₃N₄ photocatalyst for pharmaceutical degradation, hydrogen evolution, and electricity generation

Abstract

Integrating pollutant degradation with energy production offers a promising pathway for sustainable wastewater treatment. Herein, an advanced semiconductor-based photocatalytic system with dual functionality is developed by constructing a Z-scheme heterojunction composed of a bismuth-based metal–organic framework (Bi-MOF) self-assembled on conductive g-C₃N₄via a one-pot solvothermal method. The resulting Bi-MOF/g-CN photocatalyst exhibits enhanced light harvesting, efficient charge separation, and suppressed carrier recombination owing to the engineered hetero-interface. Under simulated solar irradiation, the optimized Bi-MOF/g-CN-15 demonstrates outstanding photocatalytic performance toward pharmaceutical pollutants, achieving 99.8% degradation of ciprofloxacin (CIP) (k = 6.52 × 10⁻³ min⁻¹) and 91.3% degradation of acetaminophen (ACP) within 150 min (k = 2.45 × 10⁻³ min⁻¹). The photocatalyst retains stable activity over a wide range of pollutant concentrations, solution pH values, interfering ions, and light intensities. The synergistic interaction between Bi-MOF and g-C₃N₄ enables effective pollutant adsorption and rapid interfacial electron transfer, while the narrowed band gap facilitates broad-spectrum light absorption and high redox capability. When employed as a photoanode in a photocatalytic fuel cell, Bi-MOF/g-CN-15 achieves 63% CIP degradation using only 2 mg of catalyst, accompanied by a short-circuit current density of 3.15 mA cm⁻², an open-circuit voltage of 1.68 V, and a power density of 1.8 mW cm⁻². In addition, the photocatalyst exhibits a hydrogen evolution rate of ∼2111 µmol g⁻¹ h⁻¹ under light irradiation. The significantly enhanced hydrogen production and pollutant degradation efficiency highlight the potential of this Z-scheme system as a cost-effective and scalable platform for integrated environmental remediation and solar energy conversion.

---

1. Introduction

With the emerging circular economy, the research and development focus on resource consumption, waste management, and wealth creation, to stabilise the economic growth from the pandemic impact.¹ Concerning waste management, the prevalence of pharmaceutical pollutants (antibiotics and analgesics) in water bodies is at an alarming level, causing illness to aquaculture, animals, and humans.² While penicillin-based antibiotics can be degraded by the environment, fluoroquinolone-based antibiotics like ciprofloxacin resist degradation due to the presence of quinolone cores, fluorine substituents, and piperazine rings, which hinder oxidative and thermal breakdown.³ Conventional wastewater treatment fails to further process the toxic by-products. Recently, a technique called advanced oxidation (ozone and Fenton) has been widely applied in various advanced methods like photocatalysis, electrocatalysis, and piezo-catalysis, to remove antibiotics from wastewater. In these methods, the catalyst produces hydroxyl radicals, which are responsible for breaking down the organic pollutant.⁴ Moreover, photodegradation using a photoactive material (photocatalyst) is emerging due to its effective degradation and the use of a simpler source. Semiconductors with a band gap appropriate to harvest photons convert the photon's energy into chemical energy. Roughly, the electron–hole pair resulting from photoexcited electrons produces reactive species like hydroxyl radicals, superoxide radicals, or photogenerated holes at the photocatalyst surface, which in turn oxidize the organic compounds.

Metal organic frameworks (MOFs), a class of crystalline porous materials, have the advantage of a tunable band gap by tailoring the self-arrangement of transition metal ions and organic ligands, aiming for higher porosity, enlarged specific surface area, reduced density, and regulated pore size.⁵ The pores facilitate the adsorption of pollutants, and the specific surface area creates active sites for catalysis through functional groups or by interfacial contact. Among photocatalytic frameworks, bismuth-based materials exhibit superior photocatalytic activity compared to the conventional TiO₂. The TiO₂'s drawbacks, such as rapid electron–hole recombination and light absorption restricted to the UV region, are tackled by the bismuth's photogenic bandgap and specific electronic configuration, promoting improved charge separation and transport efficiently through the internal electric field. Additionally, the low toxicity, ease of synthesis, and good chemical stability of bismuth render it as suitable metal cluster in the polyhedral network of a metal–organic framework.⁶ However, the wide bandgap of bismuth-based MOFs makes them photoactive predominantly in the ultraviolet region, thereby resulting in charge recombination or poor charge transport, limiting their practical applications. While crystallographic defects and surface engineering have been widely employed to mitigate these limitations, their contribution presents certain physical strains on the catalytic system. For instance, the introduction of oxygen vacancies through doping also leads to the distortion of the lattice, which is a major detriment to the photocatalyst.^7,8 Nguyen et al. introduced oxygen vacancies in Bi-BDC MOF by varying the synthetic approach and obtained an increased photodegradation of RhB due to the crystal's segregated lamellar microstructure and homogeneity.⁹ This expansion of surface area and induced oxygen defects resulted in excellent photocarrier separation. While oxygen vacancies can create additional active sites and reduce recombination, they often introduce structural instability and localized trap states that hinder long-range charge transport and reduce photostability under prolonged irradiation. This intrinsic drawback of defect-induced mid-gap states, which may act as recombination centres, can be precisely overcome by forming heterojunctions, in which the interfacial electric field provides a more efficient, directional pathway for charge-carrier separation and migration without compromising structural integrity.^10,11 Constructing a heterojunction around Bi-MOF with a narrow bandgap semiconductor potentially shortens the valence and conduction band distance, and the presence of internal contact area at the heterojunction with an adjusted band structure minimizes charge recombination. Ling et al. demonstrated that the in situ formation of a Bi/BiVO₄/Bi-MOF heterojunction showed superior ciprofloxacin degradation rates and photostability compared to the defect- or surface-engineered counterparts, owing to the enhanced charge separation and transfer facilitated by the heterojunction.¹² Belousov et al. compared the degradation of MB using doped and heterojunction Bi₂WO₆, namely Bi₂W_0.5Mo_0.5O₆ and 40BWO/CN, respectively. A higher efficiency for dye degradation under LED light is observed for the heterojunction catalyst, at a rate of about 56.6%, while the doped catalyst resulted in 52.9% degradation.¹³ This highlights that while defect and surface engineering are beneficial, the construction of heterojunctions provides a more consistent and effective pathway for improving visible-light-driven photocatalytic performance in Bi-MOF-based systems.

However, attaining carrier separation and enhancing redox capability through the conventional heterojunction method presents considerable challenges. Graphitic carbon nitride (g-C₃N₄), a layered material responsive to visible light, assists the Bi-MOF as a suitable substrate in forming a binary heterojunction by inducing energy band bending in both the semiconductors for a seamless flow of photoexcited electrons.¹⁴ Recently, Hota and coworkers demonstrated enhanced photocatalytic performance through g-CN-based Z-scheme (ZnO QDs/g-C₃N₄) and S-scheme (g-C₃N₄/Fe₂O₃) heterojunctions in photodegradation and H₂ generation, highlighting the versatility of g-CN as an effective substrate for advanced charge separation strategies. Building on this foundation, constructing a heterojunction between Bi-MOF and g-CN leverages complementary electronic properties where Bi-MOF, with the lone-pair active Bi³⁺ (6 s² configuration), acts as Lewis acid sites for O₂ chemisorption and activation, overcoming the g-CN's kinetic limitation. g-CN nanosheets offer superior superoxide radical generation and a large specific surface area for pollutant adsorption. Their direct Z-scheme alignment helps preserve strong redox potentials while suppressing fast charge recombination.^15–17 Moreover, the organic pollutants can act as ‘fuel’, releasing electrons during their degradation by the holes. Thus, the chemical energy from the organic molecule degradation can be converted into electrical energy by the flow of accumulated photo-excited electrons in the conduction band and the electrons from the organic pollutant towards the cathode through the external circuit.¹⁸ Thereby, building a photocatalytic fuel cell can effectively degrade organic pollutants with simultaneous chemical energy recovery. Since the separation of electrons and holes under light irradiation creates a photovoltage, tailoring the photoelectrode with a semiconductor whose E_CB is more negative than E_H+/H2 will also lead to the efficient production of hydrogen.¹⁹ Electrons accumulated on the semiconductor can reduce H⁺ to generate H₂, while holes combine with hydroxyl ions to form the hydroxyl radical, thereby splitting water.²⁰ Thus, the favoured levelling of Fermi levels between the two semiconductors can provide one more strategy of utilising the solar energy to produce renewable energy through water splitting.

Herein, we report the synthesis of a binary composite with a heterojunction between g-C₃N₄ and Bi-MOF via a simple solvothermal method. The composite was optimised with varying g-CN content, and the Bi-MOF/g-CN-15 composite exhibited the highest photocatalytic degradation efficiency for ciprofloxacin and acetaminophen. The optical and electronic nature of the optimised heterojunction catalyst Bi-MOF/g-CN-15 is evaluated with UV-DRS, UPS, PL, and EIS to elucidate the mechanism of photodegradation. To assess the catalyst's stability in real-world applicability, various reaction parameters such as pH, pollutant ppm, light intensity, and the presence of interfering ions were systematically studied. Furthermore, the excellent photocatalytic activity and high efficiency of the optimised catalyst Bi-MOF/g-CN-15 as a multifunctional photocatalyst are explored by employing the composite as a photoanode in a photocatalytic fuel cell (PFC) system, with CIP as fuel to study its redox ability in converting chemical energy to electric energy, and in photocatalytic water splitting to produce hydrogen.

2. Experimental section

2.1. Synthesis of g-C₃N₄

g-C₃N₄ was synthesised using the thermal polymerization method. 20 g of melamine was transferred to an alumina crucible and calcined at 550 °C for 5 h at a heating rate of 5°C min⁻¹. The final, yellow-colored compound of g-C₃N₄ (denoted as g-CN) was collected for further analysis.

2.2. Synthesis of the Bi-MOF/g-CN heterojunction

Scheme 1 displays the detailed synthesis procedure of Bi-MOF/g-CN heterojunction via the solvothermal method. Solution A contains 0.9701 g of Bi(NO)₃.₅H₂O and the desired amount of g-CN was mixed with 10 mL of EtOH under ultrasonication to get uniform phases. Solution B contains 0.4989 g of H₂-BDC (benzene-1,4-dicarboxylic acid) with 40 mL of DMF and 5 mL of AcOH (acetic acid). After stirring, the clear solution B was gradually added to solution A while maintaining continuous stirring for 30 min to attain a homogenous mixture. Subsequently, the entire solution was transferred to a Teflon-lined autoclave, heated at 120 °C for 16 h, and cooled to room temperature. EtOH and DMF were used to wash the obtained product, followed by drying in a hot air oven overnight at 60 °C. The obtained product was labeled as Bi-MOF/gCN-x, with x denoting the amount of g-CN incorporated. Pure Bi-BDC was prepared in a similar manner without the addition of g-CN.

Scheme 1 Schematic representation of Bi-MOF/g-CN synthesis.

2.3. Material characterization

The crystalline properties of the synthesised heterojunction were studied using an X-ray diffractometer (XRD, Rigaku D/max-2500) using Cu Kα radiation (λ = 0.15418 nm) across a range of 8–80° and a scanning rate of 5° min⁻¹. A field emission scanning electron microscope (FE-SEM, JEOL JSM-7000F) assisted by an energy dispersive spectrometer (EDS) was utilised to analyse the heterojunction morphology and energy spectrum. A transmission electron microscope was used to examine microscopic morphology and lattice structure. The optical analysis and band-gap assessment of the synthesised photocatalyst were performed using ultraviolet-visible diffuse reflectance spectroscopy (UV-Vis DRS) on a JASCO V650. Photoluminescence spectra (PL) investigation was conducted utilising a photoluminescence spectrophotometer (PerkinElmer LS50B) with a 350 nm excitation wavelength. The antibiotic concentration was quantified using a UV-visible spectrophotometer (Thermo Scientific Evolution 220 Spectrophotometer). The material's binding energy was determined using X-ray photoelectron spectroscopy (XPS; VGS Thermo K-Alpha) employing Al Kα radiation as the excitation source. The charge transfer dynamics and surface potential were analysed using Kelvin probe force microscopy (Bruker's PeakForce KPFM) with Au as the probe. To investigate the work function, ultraviolet photoelectron spectroscopy (UPS) was performed using a Sigma Probe (Thermo Scientific) instrument.

2.4. Photocatalytic performance

The photocatalytic degradation experiments were conducted utilising a 300 W Xe lamp (PLS-SXE 300 PerfectLight) equipped with an AM 1.5 G filter with a spectral range of 320–780 nm. 30 mg of photocatalyst was mixed with 50 mL of CIP (10 ppm) and the resulting mixture was subjected to constant stirring in the dark for a duration of 30 min to establish the adsorption–desorption equilibrium, prior to the exposure to the irradiation from the Xe lamp. For each 30 min interval of light exposure, 3 mL of the solution was retrieved at regular intervals, using a syringe fitted with a filter tip. The distinct concentrations of the antibiotic solutions at their characteristic wavelengths (CIP: λ = 272 nm) were recorded separately using a UV-Vis spectrophotometer. Similar experimental conditions and measurements were employed to study the photodegradation of acetaminophen (ACP: λ = 242 nm). The real-world applicability is assessed by simulating degradation under varying reaction parameters. The pH of the photosystem is varied from the natural pH of 6.3 for CIP and 6.4 for ACP to pH = 2, 4, 8, and 10. The initial pollutant concentration is also tailored from 1 mg L⁻¹ to 20 mg L⁻¹. The catalyst's recyclability is tested by reusing the same catalyst in a photodegradation study five times in a row. Apart from this, the effect of the presence of interfering ions such as Cl⁻, NO₃⁻, SO₄²⁻, PO₄³⁻, and HCO₃⁻ on the catalyst's reactivity is studied by adding their sodium salts (10 mM). The reactive species in the photodegradation of CIP is determined with 0.2 mmol AgNO₃ (e⁻), 0.5 mL isopropyl alcohol (OH^˙), 1 mL TEOA (h⁺), and ascorbic acid (^˙O₂⁻). The apparent quantum efficiency (AQY) was calibrated by varying the bandpass filters to 400, 450, 500, and 600, using eqn (1),where

No. of incident photons = E_total / E_photo = (P × S × t) / (hc/λ_in) (2)

where n = Moles degraded, N_A = Avogadro’s number (6.023 × 1023), h = Planck's constant, c = Speed of light (3 × 10⁸ ms⁻¹), P = Light irradiation power, S = Irradiation area (15.9 × 10⁻⁴ m²), λ_in = Incident monochromatic intensity, and t = Time.

The ability of the photocatalyst to perform the hydrogen evolution reaction is analysed in a sealed photoreactor fitted with a 300 Xe lamp (PLS-SXE 300 PerfectLight, 320–780 nm spectrum) and a water cooler to maintain 25 °C. About 5 mg of optimised catalyst, dispersed in 100 mL of DI water with 15% TEOA as a sacrificial agent and 20 mg Eosin Y as a photosensitiser, is sonicated for 10 min to ensure uniform distribution. The photoreactor is sealed and purged with argon to remove air, and the light source is positioned 9.5 cm above the solution surface. Continuous Ar bubbling was maintained for 40 min while stirring to deoxygenate the solution, which was then illuminated. After 5 h of visible-light exposure, a 0.5 mL headspace gas sample is withdrawn and analysed using a gas chromatograph (Chromtech GC9800) equipped with an Ar carrier gas and a 5 A molecular sieve column to quantify the evolved H₂ gas.

2.5. Photoelectrochemical performance

The evaluation of the photoelectrochemical properties of the synthesized material is performed using an electrochemical workstation (CHI 7273d) equipped with a 300 W Xenon lamp with an AM 1.5 G filter as the light source. A standard three-electrode system is used with an FTO plate as the working electrode, platinum as the counter electrode, and Ag/AgCl as the reference electrode. The ink for the photocurrent working electrode is prepared by the drop casting method: the ink (10 mg sample + 50 µL Nafion + 100 µL DI water + 150 µL ethanol, followed by 30 min of sonication) was coated on the cleaned FTO plate of working area 1 × 1cm and air dried. The loaded FTO plate is then clipped onto the electrode holder and fixed into the three-electrode cell for electro- and photoelectrochemical studies. The photocatalytic fuel cell (PFC) is built in a single-chamber, membraneless cell, with the as-prepared photoanode and a 2 × 1.5 cm Pt plate as the cathode. A 10 mg L⁻¹ of ciprofloxacin dissolved in 0.5 mol L⁻¹ of Na₂SO₄ acts as the organic fuel and supporting electrolyte of the PFC system. The photo-electric performance of the PFC was evaluated under irradiation with a 300 W Xe lamp with an AM 1.5 G filter, directed onto the surface of the photoanode. Prior to the performance measurement, the system was allowed to reach the electrochemical equilibrium by monitoring the open-circuit potential under dark conditions until obtaining a stable baseline, approximately 30 min. The photocurrent properties of the photoanode are assessed through the amperometric measurement (i–t curve). Concurrently, 3 mL aliquots of the solution are extracted to monitor the concentration of ciprofloxacin (λ = 272 nm) using a UV-Vis spectrometer to estimate the degradation percentage of the antibiotic using the PFC approach.

3. Results and discussion

3.1. Physical characteristics of the prepared photocatalysts

The XRD analysis of Bi-MOF/g-CN reveals the structural modification and interaction between the Bi-MOF and g-CN, along with their crystalline feature and phase purity. Fig. 1a shows the XRD patterns of pure g-CN, Bi-MOF, and Bi-MOF/g-CN composite with different loadings of g-CN. The diffraction peaks of g-CN at 13.3° and 27.3° correspond to the (110) and (002) planes, respectively, which are the in-plane structural packing of tri-s-triazine unit and interlayer stacking of aromatic system, thereby confirming the formation of g-CN.²¹ Bi-MOF has two characteristic peaks at 7.1°, 14.2°, and 24.8°, which are consistent with CCDC No.793572.²² The correlation of Bi-MOF's lattice parameters with the CCDC labels the crystal's structure as monoclinic. The intense (001) plane signifies the layered structure of Bi³⁺ clusters bridged by the benzene dicarboxylic ligand (H₂-BDC), and the (002) plane represents the stacking of Bi³⁺ clusters, confirming the monoclinic crystal system. The Bi–O–Bi bond bridging the Bi³⁺ cluster, forming a 3D structure, is evident from the presence of the (122) plane.^23–25 The preservation of all the characteristic planes of Bi-MOF and g-CN in the Bi-MOF/g-CN composite confirms the effective establishment of in situ growth in the heterojunction composite. However, the gradual decrease in the intensity of g-CN in the composite is due to the grafting of Bi-MOF and dominance of its crystallinity over g-CN, affecting the stacking order of g-CN at the (002) plane. Meanwhile, a significant peak shift was observed in Bi-MOF's (001) peak, gradually shifting towards a lower angle from 7.1° as shown in Fig. 1b. The lower angle shift of the Bi-MOF (001) plane with increasing g-CN content (10 to 25 mg) is primarily attributed to the increased interplanar spacing arising from the lattice strain induced by the g-CN to the matrix of Bi-MOF. Moreover, a strong interfacial interaction and π–π stacking between g-CN and BDC linker at the Bi-MOF/g-CN heterojunction redistributes the electron density over the g-CN, therefore modifying the ionic positions within the Bi-MOF, producing an observable change in lattice parameters.^26,27 The selective (001) peak shift to a lower angle indicates the interlayer expansion of Bi-MOF's basal planes due to g-CN nanosheet intercalation along the c-axis, while the higher index reflections remain intact, preserving the (110), (200), and (122) planes, since the in-plane lattice parameters of Bi–Bi and Bi–O distance in SBUs are unaffected by the c-axis modification by g-CN intercalation.²⁸ Furthermore, the g-CN (002) peak at 27.4° 2θ is diminished in Bi-MOF/g-CN composites due to the exfoliation of bulk g-CN into ultrathin nanosheets during the composite formation. This reduces the g-CN interlayer spacing and crystallite size (0.32 nm) below the Scherrer detection limit (D < 3 nm; β > 1.5°), broadening the (002) plane indistinguishable from the background. However, the intact in-plane structure of g-CN is observed in HRTEM as 0.32 nm lattice fringes, accounting for the (002) plane. Moreover, the anchoring of Bi-MOF nucleating over the g-CN sheets also suppresses the (002) plane reflection in PXRD.^29–31 The decreased peak intensity with increasing g-CN content also reflects the reduced crystallinity. Such a crystalline disorder region is bound to create abundant active sites, along with the nitrogen and oxygen vacancies. These disorder coordination and vacancy sites serve as highly active catalytic sites. These structural and surface modifications enhance the interfacial contact between the Bi-MOF and g-CN, thereby facilitating an efficient charge transfer with suppressed electron–hole recombination.

Fig. 1 (a) XRD patterns of g-CN, Bi-MOF, and Bi-MOF/g-CN-x (x = 10, 15, and 25), (b) FTIR spectra and (c) Bi 4f, (d) C 1s, (e) N 1s, and (f) O 1s XPS spectra of g-CN, Bi-MOF, and Bi-MOF/g-CN-15.

The elemental composition, binding energy, chemical valence state, and electronic state of Bi-MOF, g-CN, and their composites at their surface are further analysed through XPS. The chemical environment and bonding of the composite are also gauged based on their binding energy shift in the XPS spectra. The Bi 4f spectra of pure Bi-MOF show two characteristic peaks at 164.01 eV and 158.84 eV, attributed to the spin-orbital split of Bi 4f_5/2 and Bi 4f_7/2, respectively, thereby confirming the Bi³⁺ oxidation state.³² The composite Bi-MOF/g-CN-15 in Fig. 1c shows a clear Bi 4f peak shift towards the higher binding energy compared to the pure Bi-MOF, indicating the formation of a heterojunction. This shift illustrates the depletion of electron density over Bi-MOF, indicating transfer of electrons from Bi-MOF to g-CN until their Fermi level attains equilibrium and creates an internal electric field (IEF). This clearly reveals the strong interfacial coupling between the two components of the heterojunction. Fig. 1c shows the O 1s spectra, revealing critical information about the defect formation. Pristine Bi-MOF shows two prominent peaks at 530.73 eV and 529.48 eV, representing the lattice oxygen and Bi–O bonding from the Bi-oxo cluster. Along with the above two O 1s peaks, a new peak at 531.28 eV is observed in the Bi-MOF/g-CN-15 composite. This peak arises from oxygen dissociatively adsorbed at a vacancy site.³³ The appearance of this peak indirectly indicates the presence of oxygen vacancies created during in situ heterojunction formation and exposed upon irradiation.³⁴ These vacancies create localised defect energy states within the bandgap of the composite and act as a recombination centre to trap the photogenerated charge carriers, thereby making the oxygen vacancies one of the highly active catalytic sites on the catalyst's surface. The C 1s spectra of g-CN show major characteristic peaks at 288.28 eV, 286.73 eV, and 285.03 eV associated with the heterocyclic C (N–C=C), C–N–C of the triazine ring, and graphitic C of C–C/C=C as displayed in Fig. 1d.³⁵ The less prominent peak at 289.48 eV is the π–π* satellite peak, one of the characteristic peaks of carbon nitride, representing the delocalised sp² carbon.³⁶ In the composite, the intensities and positions of the three major peaks increase and shift, indicating contributions from carbon in the Bi-MOF, corresponding to the carboxylic C=O and aromatic C=C of the terephthalic ligand at distinct binding energies. The stronger C 1s signal in the composite reflects that the Bi-MOF and g-CN are integrated at the molecular level and are retained throughout the material up to the composite's surface. These heightened intensities are direct evidence that Bi-MOF has successfully grown and covers the g-CN surface. The N 1s (Fig. 1e) spectra of the g-CN are deconvoluted into three peaks at 398.3 eV, 399.8 eV, and 400.8 eV corresponding to the bridging N of triazine's C–N=C, terminal amino group, and tertiary amino N, thereby confirming the nitrogen bonding environment in the aromatic g-CN structure.³⁷ The composite Bi-MOF/g-CN-15 exhibited less intense peaks, indicating the masking of the N environment by the Bi-MOF grown on the g-CN nanosheets. The terminal and bridging N-sites in the g-CN coordinate either with the Bi³⁺ ions or form a hydroxyl bond with the organic linkers, resulting in a modified electronic environment. This modification in the chemical interaction reduces the reactive free nitrogen atoms, further reducing the N 1s signal in the spectra.³⁸ The evidence from the low-angle shifted XRD peak, the altered environment in XPS spectra, and the anchoring of Bi-MOF on g-CN observed in FE-SEM strongly infers the presence of interfacial coupling, electron transfer, and defect formation. Such a synergy of structural and electronic modification collectively enhances the charge separation and strengthens the photocatalytic activity of the composite.

The morphological characteristics of the materials are analysed through FE-SEM. Fig. 2 illustrates the morphological evolution of Bi-MOF/g-CN composites with increasing g-CN content. Fig. 2a shows that the Bi-MOF forms a flower-like microsphere of diameter about 6 µm with numerous nanoflakes on the surface of width ∼500 nm and length ∼200 nm. The introduction of acetic acid and ethanol in the reaction system alters the dimensions of the bismuth cluster by inhibiting the perpendicular growth of the (001) plane along the c-axis. This leads to two-dimensional crystal growth, producing nanoplates with increased surface area and numerous active sites.^39–41Fig. 2b shows the morphology of pure g-CN in a typical layered arrangement of bulk structure. Fig. 2c and d clearly show the morphological change of Bi-MOF with increasing g-CN loading. In the Bi-MOF/g-CN composites, the Bi-MOF microspheres appear to be distributed on the g-CN, either as partially fragmented (Fig. 2c) or aggregated into flakes (Fig. 2d and e). The g-CN serves as a platform for nucleation sites, resulting in Bi-MOF microspheres anchored onto the g-CN. However, the in situ accumulation of Bi-MOF over the g-CN altered the microsphere size of Bi-MOF. The prior anchoring of bismuth ions on the g-CN's surface reduces the agglomeration, thus producing a Bi-MOF microsphere of diameter 4 µm, with well-spread-out petal-like nanoplates on the surface of the microsphere.^42–44 This morphological transformation is controlled by the solution's pH, nucleation kinetics, and Ostwald ripening.⁴⁵ During the synthesis of Bi-MOF, acetic acid introduces an acidic pH (∼1) environment, which provides sufficient H⁺ ions in the medium to prevent the Bi³⁺ hydrolysis and balance the nuclei pH and growth rate of Bi-MOF crystals. Initially, numerous Bi-MOF nuclei aggregate into microspheres, followed by Ostwald ripening, while smaller crystallites dissolve and redeposit onto the larger ones to minimise the surface energy. However, the increasing content of nitrogen-rich surface groups on the g-CN acts as a buffer, further altering the local pH. With fewer available H⁺ ions, the nucleation rate slows, disrupting ripening and leading to an irregularly aggregated morphology in the composite. In addition, the morphology of the composite is consistent with the literature.⁴⁶ Such morphological changes improve interfacial contact between the Bi-MOF and g-CN via numerous defects and active sites, thereby enhancing light absorption through a uniform distribution and rapid transport of photogenerated charge carriers across the well-integrated heterojunction. Furthermore, the elemental mapping of the composite Bi-MOF/g-CN (Fig. 2f) confirms the homogeneous distribution of elements Bi, C, N, and O, highlighting the successful integration of the two components, with Bi-MOF grown in situ on the g-CN surface. Additionally, the composite's flower-like microsphere dispersed on the g-CN is well evident from the images of HR-TEM (Fig. 3a). From the HRTEM image, the g-CN can be observed, associated with the nanoflakes of Bi-MOF. The well-defined lattice fringes found in the HR-TEM imply that the high crystallinity of the Bi-MOF is retained in the composite. Meanwhile, the interplanar spacing (Fig. 3b and c) of 0.32 nm and 0.36 nm corresponds to the (002) and (122) planes of g-CN and Bi-MOF, respectively.^47–49 Moreover, the overlaid nanoflakes of Bi-MOF on the g-CN highly indicate the close contact between the Bi-MOF and g-CN, highlighting the formation of a heterojunction.⁴⁹ The energy-dispersive X-ray (EDX) elemental mapping illustrates the even arrangement of Bi, C, O, and N elements on the Bi-MOF/g-CN-15 composite, as shown in Fig. 3d. The significant morphological change increases interfacial contact between the Bi-MOF and g-CN, thereby creating numerous defects and active sites, and enhancing light absorption through improved scattering and the rapid transport of photogenerated charge carriers across the well-integrated heterojunction.

Fig. 2 FE-SEM analysis for (a) Bi-MOF, (b) g-CN, (c–e) different loading amounts of Bi-MOF/g-CN-X (X = 10, 15, and 25 mg), (f) EDS spectra and (g–k) elemental distribution of Bi-MOF/g-CN-15.

Fig. 3 (a) TEM image, (b) and (c) corresponding HRTEM image, and (d) HAADF and STEM mapping of Bi-MOF/g-CN-15.

The direction of interfacial charge transfer and establishment of an internal electric field in the composite is further confirmed by Kelvin probe force microscopy (KPFM). The work functions of g-CN, Bi-MOF, and Bi-MOF/g-CN-15 were determined from the contact potential difference (V_CPD) using eqn (4).⁵⁰

V_CPD = V_sample − V_tip = (W_tip − W_sample)/e (4)

where V_sample and V_tip are the absolute potentials of the sample and probe's tip, while W_tip and W_sample are their corresponding functions and e is the charge of an electron. On substituting the value of Au probe's WF (5.034 eV), eqn (4) is modified to W_sample = W_tip − V_CPD, and the work functions of g-CN, Bi-MOF, and Bi-MOF/g-CN-15 were calculated to be 4.90, 4.81, and 4.73 eV, respectively. Upon contact, the higher Fermi level of Bi-MOF, indicated by the lower work function, bends until aligning with the Fermi level of g-CN. The interfacial charge transfer from Bi-MOF to g-CN creates a charge disparity, which generates an interfacial electric field (IEF) in the opposite direction. This IEF on the photocatalyst's surface can be quantified through KPFM's linear scanning surface potential spectra (Fig. 4). The 2.5-fold increase in surface potential of Bi-MOF/g-CN-15 compared to pure g-CN and Bi-MOF represents the formation of a surface electric field. The electrons accumulated at the g-CN interface repel the built-in IEF, consequently increasing the surface electrons' potential energy. The elevated localised electron distribution at the interface arises from the high electronegativity of nitrogen compared to carbon in g-CN's triazine unit, which disrupts the electron delocalisation and actively accepts the high electron density from the polarised Bi–O cluster of Bi-MOF.^51,52 Since Bi-MOF is grown over g-CN, the establishment of IEF across the entirety of the composite's surface can be seen in Fig. 4c – potential and height profile, consistent with the interface observed in FESEM and HRTEM. This intense IEF promotes the separation and directional transfer of photogenerated charges under irradiation, thereby reducing the recombination rate during photocatalysis.⁵³

Fig. 4 KPFM surface potential image and the line-scanning profile, along with the height profile of (a) g-CN, (b) Bi-MOF, and (c) Bi-MOF/g-CN-15.

EPR analysis was used to identify the presence of unpaired electrons and paramagnetic species generated under light irradiation. Fig. S1 illustrates the distinct EPR signals of the samples g-CN, Bi-MOF, and Bi-MOF/g-CN-15 under irradiation, which evidently photoinduce the unpaired electrons associated with reactive species and surface defects. However, under dark conditions, Bi-MOF shows no EPR signal due to the diamagnetic nature of the Bi³⁺ centre. The accumulation of denser unpaired electrons under illumination, evident by the strong signal in Bi-MOF/g-CN-15, clearly indicates interfacial electronic coupling between Bi-MOF and g-CN, resulting in efficient charge separation. Furthermore, EPR confirms oxygen vacancies at g = 2.004 (Fig. S1), which are characteristic of electrons trapped at O_v. XPS O 1s deconvolution at 531.2 eV correlates with the EPR intensity, thereby supporting the presence of oxygen vacancies.

3.2. Optical properties of the prepared photocatalyst

The photo-absorption characteristics and electronic structure of the composites were investigated via UV-DRS and PL spectroscopy, to evaluate the absorption band, bandgap engineering, and charge carrier dynamics. The absorption edges determined from the UV-DRS spectra (Fig. S2a) reveal a systematic redshift with increasing g-CN loading. Pure g-CN exhibited active absorption of both ultraviolet and visible light, with a 470 nm absorption edge, corresponding to a band gap of ∼2.61 eV, typical of its π–π* electronic transitions in the conjugated heptazine units.⁵⁴ In contrast, Bi-MOF shows strong absorption of ultraviolet light with an absorption edge of 363 nm (∼3.4 eV), consistent with the wide bandgap-semiconducting nature derived from the Bi-MOF clusters and aromatic terephthalate linkers. Compared with the material, Bi-MOF/g-CN-15 exhibited a red shift, extending the absorption band into the visible-light region, shifting progressively from 387.89 nm (Bi-MOF/g-CN-10) to 395.82 nm (Bi-MOF/g-CN-15) to 407.31 nm (Bi-MOF/g-CN-25) with increasing g-CN content. The π-conjugated g-CN acts as a photosensitiser and extends the absorption edge of Bi-MOF through charge transfer interaction between N-rich g-CN and the Bi-oxo clusters. This interaction indicates the narrowed optical bandgap with enhanced light absorption due to synergistic effects.⁵⁵ The band gap (E_g) of g-CN, Bi-MOF, and Bi-MOF/g-CN-15 is calculated from the Tauc plot using the Kubelka–Munk formula (eqn (5)),

αhν = A(hν − E_g)^n/2 (5)

where α is the absorption coefficient, h is Planck's constant, ν is the light's frequency, and A is the proportionality constant. The term n is the transition type, and here g-CN and Bi-MOF are direct semiconductors, n = 1. Using eqn (5), the band gap values of g-CN, Bi-MOF, and Bi-MOF/g-CN-15 are found to be 2.67, 3.40, and 3.05 eV, respectively, as shown in Fig. S2b.

The photocatalytic activity is highly influenced by the efficient separation and transfer of the generated photocarriers. The recombination rate of the photoexcited charges was evaluated using PL spectra with a 380 nm excitation wavelength and is shown in Fig. S2c. Pristine g-CN exhibits an intense emission centred near 450–470 nm, arising from the band-to-band radiative recombination of photogenerated charge carriers. The PL emission intensity decreases dramatically across all Bi-MOF/g-CN composites, indicating optimal interfacial charge transfer and minimal radiative recombination. This suppressed PL results from the construction of a built-in electric field across the heterojunction, driving the photogenerated electron from the conduction band of g-CN to Bi-MOF, while holes transport in the opposite direction, confirmed by the XPS electron transfer direction, and indicates the increasing lifetime of the charges.^56,57 The quenching of PL, therefore, provides direct evidence of enhanced charge separation, defect-assisted carrier migration, and suppression of recombination. These optical studies provide a quantitative confirmation that g-CN predominantly enhances the optical properties of the pristine Bi-MOF through bandgap narrowing, extended light absorption, efficient charge separation, and reduced carrier recombination.⁵⁸

3.3. Photocatalytic pollutant removal and H₂ production

The photocatalytic efficiency of the prepared materials was evaluated towards the degradation of two common antibiotic pollutants under light irradiation. Fig. 5a shows the time-dependent UV-Vis spectra of CIP, revealing the progressive decline in the characteristic absorption peak at 272 nm. Prior to the illumination, an adsorption–desorption equilibrium was achieved by treating the solution in the dark for 30 min, where almost 36% of CIP is adsorbed onto the composite Bi-MOF/gCN-15 at equilibrium. Upon light exposure for 90 min, an additional 63% of CIP is photodegraded. g-CN and Bi-MOF exhibited a decent degradation of CIP of about 81% and 75%, respectively, while the composites Bi-MOF/g-CN-10, Bi-MOF/g-CN-15, and Bi-MOF/g-CN-20 demonstrated high photocatalytic performance, achieving about 97%, 99%, and 97% degradation efficiency, respectively. Fig. 5b and c show the superior degradation kinetics of the Bi-MOF/g-CN composite compared with pure Bi-MOF and g-CN. A near-complete degradation equilibrium (99.5%) is achieved under 90 min of irradiation with a rate constant of 5.32 × 10⁻³ min⁻¹ while complete degradation (99.8%) occurred by 150 min, with a higher rate constant of 6.52 × 10⁻³ min⁻¹. The high activity of Bi-MOF/g-CN-15 is attributed to the synergistic properties of the high specific surface area and high porosity of Bi-MOF and the light responsiveness induced by the g-CN through narrowing the bandgap. Moreover, the heterojunction formed between Bi-MOF and g-CN in the composite promoted photocarrier separation, effectively suppressing recombination and enhancing charge mobility. The trend observed in photocatalytic efficiency with varying g-CN content suggests the requirement for an optimal ratio for better heterojunction formation. An improved photo-efficiency with a sufficient ratio of g-CN corresponds to the even distribution of Bi-MOF over the g-CN with exposed active catalytic sites, providing a platform for charge transportation with increased lifetime. As the g-CN ratio exceeds 15 mg, the adsorption of CIP onto the photocatalyst is reduced due to the agglomerated surface of the composite with Bi-MOF. As the g-CN weight ratio ranged from 15 to 25 mg, a nearly 2% decrease in activity was observed. This could be understood from FE-SEM analysis, wherein the composite Bi-MOF/g-CN-15 is found to retain the flower-like microsphere structure of Bi-MOF in the composite, while the Bi-MOF/g-CN-25's microsphere is ruptured, having the nanoplates of Bi-MOF distributed on the g-CN. This minimizes the active surface area of the composite, thereby reducing degradation efficiency with the increasing g-CN ratio. The linear kinetic plot of ln(C/C₀) vs. time reveals the reaction kinetics for CIP degradation as pseudo-first order, suggesting that the concentration of the photocatalyst remained unchanged, marking its stability in degrading the pollutant (lowering the CIP concentration).⁵⁹ The higher rate constant value for Bi-MOF/g-CN-15 further confirms the efficiency of the photocatalyst in degrading ciprofloxacin, almost 4 times and 5 times faster than g-CN and Bi-MOF, respectively. The optimised photocatalyst Bi-MOF/g-CN-15 also established substantial photocatalytic activity against other classes of pharmaceutical pollutants, including acetaminophen (an antipyretic agent). This highlights the broad-spectrum efficacy of the optimised photocatalyst in treating pharmaceutical water pollutants.

Fig. 5 (a) Time-bound UV-Vis absorption of CIP with Bi-MOF.g-CN-15, (b) CIP concentration vs. time, (c) rate constant and CIP degradation percentage values, (d) effect of water medium, (e) pH values, (f) interfering ions, for CIP removal with Bi-MOF/g-CN-15, (g) UV-Vis absorption spectra of CIP with Bi-MOF.g-CN-15, (h) CIP concentration vs. time, and (i) rate constant and CIP degradation percentage {CIP/ACP = 10 ppm, 50 mL; interference ion-10 mM}.

The operational stability and practical applicability of the optimised Bi-MOF/g-CN-15 photocatalyst are evaluated with a series of stress-conditions and control experiments to elucidate its catalytic behaviour under varying conditions, as shown in Fig. 5d–f. To simulate the fluctuations present in natural water bodies or industrial effluents, the reaction system is systematically altered by varying the concentration of the pollutant, adjusting the acidity and alkalinity (pH), and introducing co-existing ions as contaminants that interfere with the photocatalyst. The initial concentration of ciprofloxacin in 50 mL of reaction medium is varied (1, 5, 10, 15, and 20 mg L⁻¹), and the removal percentage remained almost the same. A visible reduction in the degradation percentage is observed at 20 ppm, probably due to the saturation of pores and inadequate active species to break down the pollutant.⁶⁰ In addition, the photocatalytic removal of CIP was evaluated in different real-water matrices such as tap, river, and seawater, as shown in Fig. 5d, which are the major discharge pathways for domestic and industrial effluents. A noticeable decrease in degradation efficiency was observed relative to deionised water, primarily due to the presence of naturally occurring organic matter and inorganic ions, such as humic and oxalic acids, which can act as radical scavengers, thereby slowing CIP degradation.^61,62 Nevertheless, the performance remained relatively high, with degradation efficiencies of 82.76% (tap water), 83.81% (river water), and 79.5% (sea water), demonstrating the strong environmental adaptability of Bi-MOF/g-CN-15. Similarly, Fig. 5e shows the influence of solution pH on CIP removal, revealing high photocatalytic efficiency across a wide pH range. Under neutral water conditions (pH = 6.5), the photocatalyst exhibited the highest degrading efficiency (up to 99%), whereas a slight decrease in efficiency was observed when the reaction medium was alkaline or acidic. At lower pH = 2, degradation decreases to 91.3%, and at higher pH (8–10), the degradation diminishes linearly to 80.8% (pH 10). This trend arises from several interacting factors: around pH 6, the surface charge of the composite Bi-MOF/g-CN-15 and the protonation state of CIP (a fluoroquinolone with pK_a values of 6.1 and 8.2) are optimally matched, promoting the strong electrostatic adsorption and effective oxidative degradation. Under strong acidic conditions, excess H⁺ may compete with CIP at the adsorption site and inhibit the formation of deprotonated CIP. Ciprofloxacin primarily exists in its anionic form under alkaline conditions (pKa = 8.9), and so deprotonated CIP electrostatically repels the negatively charged photocatalyst's surface.⁶³ Hence, at pH = 10, the absorption of CIP onto the catalyst decreases, resulting in a sharp decline in degradation efficiency of approximately 80%. However, the decline is relatively modest, which explains the photocatalyst's ability to operate over a wide pH range, from acidic to alkaline, without compromising its activity.

The impact of inorganic ions on the photocatalytic degradation of CIP was examined to assess catalyst stability and selectivity under real conditions, as shown in Fig. 5f. The photocatalytic efficiency followed the order: DI water (99.5%) > Cl⁻ (95.8%) > NO₃⁻ (94.1%) > SO₄²⁻ (87.8%) > PO₄³⁻ (29.2%) > HCO₃⁻ (16.4%). A noticeable decrease in performance was observed for phosphate (PO₄³⁻) and bicarbonate (HCO₃⁻) ions. This inhibitory effect of the inorganic ion is probably due to the following reasons: (i) the pH of the reaction medium is elevated on administering the ions, (ii) the ions compete with the active species (^˙O₂⁻ and h⁺), and (iii) the ions get adsorbed on the catalyst's surface thereby blocking the pollutant and suppressing the generation and reactivity of ^˙O₂⁻ and h⁺ according to eqn (6) and (7):⁶⁴

PO₄⁻³ + h + → PO₄²⁻ (phosphate radical, competes with pollutant oxidation) (6)

HCO₃⁻ + ^˙O₂⁻ → CO₃²⁻ + HO₂⁻ + O₂ (increases pH and scavenges the superoxide) (7)

Photolysis experiments performed under light irradiation without the photocatalyst achieved only 22% CIP degradation, whereas dark adsorption experiments using the photocatalyst without light resulted in 48% CIP removal, as shown in Fig. S3a. These results indicate that neither light nor the catalyst alone is sufficient for effective pollutant degradation. Light supplies the required photon energy for charge generation but lacks catalytic activity, whereas the catalyst provides active sites and an appropriate band structure for charge separation but requires an external force to excite the photon and initiate charge generation. However, the integration of the photocatalyst with visible-light irradiation drives photocatalytic degradation, achieving up to 99% CIP removal. Radical trapping studies further clarified the involvement of active species in the photocatalytic degradation, as displayed in Fig. S3b. Ascorbic acid (˙O₂⁻), KI (h⁺), IPA (˙OH), and AgNO₃ (e⁻) are added as common quenchers to generate their respective active species. The scavenger results revealed that h⁺ and ˙O₂⁻ are the primary reactive species responsible for pollutant removal. Upon ˙O₂⁻ and h⁺ scavenging, the CIP degradation efficiency was heavily decreased. This suppression clearly indicates that direct hole-mediated oxidation and indirect superoxide-based attack are parallel mechanisms in photodegradation. The inhibition caused by isopropanol was comparatively smaller, indicating that the ˙OH radical plays a secondary yet supportive role in the degradation pathway. This mechanistic insight confirms that the Bi-MOF/g-CN heterojunction preferentially generates and utilizes charge carriers (holes and electrons for superoxide formation) rather than hydroxyl radicals, thereby validating the Z-scheme band alignment. The long-term durability of the photocatalyst in the reaction medium is assessed through cycling degradation tests, and the apparent quantum yield (AQY) can be used to quantify the degradation efficiency under varying irradiation. The photo-to-pollutant conversion efficiency across the solar spectrum was analysed through the monochromatic light photolysis experiment, conducted using bandpass optical filters at characteristic visible wavelengths. The light absorption capacity and quantum efficiency of the Bi-MOF/g-CN-15 composite for CIP degradation at different monochromatic excitation wavelengths were analysed as shown in Fig. S3c. The apparent quantum efficiencies were calculated to be 1.2% (AM 1.5G), 3.9% (BP 400), 1.3% (BP 450), 1.5% (BP 500), and 1.1% (BP 600). The maximum AQY at 400 nm demonstrates that the Bi-MOF/g-CN-15 heterojunction exhibits the most efficient charge separation and reactive oxygen species (ROS) generation in the near-UV/blue region. The recyclability profile shown in Fig. S3d reveals the excellent stability of the heterojunction, where CIP degradation gradually diminishes from 99.5% in the initial cycle to 83.7% in the fifth cycle, primarily due to minimal loss of the catalyst in the recovery process, partial surface fouling by adsorbed intermediates, and slight deactivation of oxygen-vacancy sites upon repeated irradiation.⁶⁵ Nevertheless, retaining >80% efficiency after five cycles highlights the structural resilience of the Bi-MOF microspheres and the durability of the interfacial electron-transfer channels. The XRD peaks of the photocatalyst before and after recycling are compared to elucidate the crystallinity. The intensities of the characteristic peaks were nearly unchanged, with no shifts (Fig. S4), indicating preservation of the crystal structure. This confirms the photocatalyst's stability and reusability, with only minor deviations in overall activity.

The photodegradation of acetaminophen was also examined to further evaluate the versatility of Bi-MOF/g-CN composites, as shown in Fig. 5g and h. The distinctive absorption band of ACP at 242 nm decreased steadily with increasing irradiation time and completely disappeared within 150 min for the Bi-MOF/g-CN-15 composite, as shown in Fig. 5g. The photocatalytic degradation of acetaminophen (ACP), expressed as the variation of C_t/C₀ with irradiation time, is illustrated in Fig. 5g and h. Pristine g-CN and Bi-MOF exhibit relatively low degradation efficiencies of 68% and 26%, respectively, owing to limited visible-light absorption and inefficient charge separation, while the composite Bi-MOF/g-CN-15 showed the highest rate constant of 2.45 × 10⁻³ min⁻¹, remarking the effect of heterojunction with optimal g-C₃N₄ loading. Upon constructing the Bi-MOF/g-CN heterojunction, the degradation performance increased remarkably. Among all the prepared composites, Bi-MOF/g-CN-15 exhibited the highest photocatalytic activity, achieving 91.3% ACP degradation, highlighting the strong synergistic interaction between Bi-MOF and g-CN that enables efficient charge transfer, enhanced reactive oxygen species (ROS) generation, and improved optical absorption. However, further increasing the g-C₃N₄ content to 25 mg led to reduced activity, consistent with FE-SEM and XRD observations. This deterioration is attributed to excessive g-CN coating, which disrupts the Bi-MOF microsphere structure, masks active sites, and weakens light catalyst interaction. Control studies under ‘light only’ and ‘catalyst only’ conditions showed negligible ACP removal, confirming that ACP is highly stable and that both visible light and the heterojunction catalyst are required for effective degradation. Additionally, the recyclability analysis (Fig. S5a) demonstrated that Bi-MOF/g-CN-15 exhibits excellent photostability, retaining high degradation efficiency across five successive cycles, thereby verifying its durability and suitability for real-world wastewater treatment. The ACP removal efficiency under different pH conditions (2, 4, 6, and 8) was studied (Fig. S5b). A clear enhancement in degradation is observed as the pH increases from highly acidic to neutral values. The removal efficiency increases from 72% at pH 2 to 89.9% at pH 4, reaching a maximum of 91.3% at pH 6. This improved activity under acidic and near-neutral conditions can be attributed to ACP being present mainly in its neutral (non-ionized) molecular form, which adsorbs more effectively on the catalyst surface and is more easily oxidized by the dominant reactive species, holes (h⁺) and superoxide radicals (˙O₂⁻). However, when the medium becomes alkaline (pH 8), the degradation efficiency drops sharply to 15%. This sharp decline occurs because ACP undergoes deportation at higher pH and converts into its negatively charged species (ACP⁻). The negatively charged ACP⁻ interacts poorly with the catalyst surface and experiences strong repulsion, reducing the adsorption and limiting its accessibility to h⁺ oxidation.⁶⁶ In addition, excess OH⁻ ions in alkaline solution compete with dissolved oxygen for photogenerated electrons, decreasing the formation of ˙O₂⁻ radicals. OH⁻ can also consume photogenerated holes, further suppressing the key oxidative pathways. Since h⁺ and ˙O₂⁻ are the primary reactive species in ACP degradation, their inhibition at alkaline pH results in extremely low removal efficiency as observed at pH = 8. Fig. S4c presents the recycling experiment with the optimized Bi-MOF-g-CN-15, which demonstrates stable performance over 5 cycles.

The photocatalytic hydrogen evolution performance of the prepared photocatalysts was evaluated under visible-light irradiation in the presence of Eosin Y (EY) as a photosensitizer and TEOA as a sacrificial electron donor. Fig. S6 shows the H₂ production of g-CN, Bi-MOF, Bi-MOF/g-CN-15, and Eosin Y. The EY photosensitizer alone shows a H₂ evolution rate of 219 µmol g⁻¹ h⁻¹, indicating that the dye can absorb light and generate photoelectrons and holes; however, rapid charge recombination and the absence of an efficient catalytic site lower the H₂ production rate compared to the presence of the photocatalyst. Upon the introduction of the photocatalyst, the H₂ production rate significantly increases. Pure g-CN shows a H₂ evolution rate of 736.82 µmol g⁻¹ h⁻¹ which is attributed to its suitable conduction band position for proton reduction and its ability to accept the electron from EY. Bi-MOF also exhibits an average H₂ production rate of 684.82 µmol g⁻¹ h⁻¹ suggesting that Bi-MOF provides accessible metal centres and porous channels for catalytic reactions. After the formation of heterojunction between g-CN and Bi-MOF, a significant increase in the H₂ production rate of about 2111.03 µmol g⁻¹ h⁻¹ is observed, which is nearly 2.9 times higher than g-CN, 3.1 times higher than Bi-MOF and 9.6 times higher than EY. This rapid improvement clearly indicates the strong synergistic interaction between Bi-MOF and g-CN. The IEF formed in the composite promotes efficient charge separation and migration, meanwhile suppressing the electron hole recombination and enabling more photogenerated electrons that participate in proton reduction. The optimized Bi-MOF/g-CN-15 shows a sufficient interfacial junction while retaining sufficient light absorption and accessible active site, resulting in high HER performance.

3.4. Photoelectric properties and PFC performance of the catalyst

The photoelectrochemical and photocatalytic fuel cell performances of g-CN, Bi-MOF, and Bi-MOF/g-CN-15 are shown in Fig. 6 and Table 1. The LSV curve (Fig. 6a) shows that the Bi-MOF/g-CN-15 composite delivers a markedly higher photocurrent density compared to pristine g-CN and Bi-MOF, indicating enhanced photo-induced charge separation and interfacial charge transfer due to the formation of an efficient heterojunction. Furthermore, the separation rate of the photocarriers throughout the photocatalytic measurement is studied using the transient photocurrent (i–t) curve. The composite Bi-MOF/g-CN-15 (Fig. 6b) displayed the highest photocurrent density, approximately 3.2 and 5.3 times greater than that of pure g-CN and Bi-MOF, respectively. Additionally, the uniform photocurrent intensity of the composite throughout the reaction indicates the material's stability in retaining charge separation under continuous generation. Electrochemical impedance spectroscopy (EIS) is used to examine the charge-transfer properties of the photocarriers. As shown in Fig. 6c, the smallest semicircle arc in the Nyquist plot for the composite Bi-MOF/g-CN-15 indicates lower charge resistance, indicating faster charge transport across the heterojunction. Thus, the abundance of separated charges with lower resistance would significantly enhance the redox ability of the composite under irradiation. The Bode phase plots (Fig. 6d) show a phase maximum shifted toward lower frequencies for Bi-MOF/g-CN-15, indicating a longer charge-carrier lifetime and reduced recombination. This photoelectrochemical performance is evaluated by employing the optimised composite Bi-MOF/g-CN-15 as the photoanode in the PFC system with CIP as the fuel.

Fig. 6 (a) LSV, (b) photocurrent measurement, (c) EIS, (d) Bode plot for Bi-MOF, g-CN, and Bi-MOF/g-CN-15, (e) CIP concentration vs. time during the wastewater treatment, and (f) J–V & power density curve across various photoanodes.

Table 1 Obtained photocatalytic fuel cell results for synthesised photoanodes

Materials	V oc (V)	J sc (mA cm^−2)	P max (mW cm^−2)	FF	Degradation
g-CN	1.47	1.55	1.13	0.496	44%
Bi-MOF	1.46	2.05	1.09	0.364	41%
Bi-MOF/g-CN-10	1.36	2.95	1.70	0.424	48%
Bi-MOF/g-CN-15	1.59	3.15	1.81	0.361	63%
Bi-MOF/g-CN-25	1.68	3.15	1.59	0.300	56%

---

To assess the dual functionality of the Bi-MOF/g-CN system in wastewater treatment and simultaneous power generation, a self-driven photocatalytic fuel cell (PFC) was constructed using CIP (10 ppm) as the model pollutant. The degradation behaviour of the PFC-integrated photoanodes is presented in Fig. 6e, where Bi-MOF/g-CN-15 demonstrates the most rapid decline in the CIP concentration, arising from its superior charge-separation ability and enhanced interfacial charge transfer. Under light irradiation, photogenerated electron–hole pairs in the semiconductor layers become spatially separated and generate an internal photovoltage, driving pollutant degradation and electricity generation simultaneously. The optimized Bi-MOF/g-CN-15 composite deposited on FTO served as the photoanode, with a Pt plate as the photocathode and 1 M Na₂SO₄ as the electrolyte. CIP acts as the organic fuel, undergoing oxidation by photogenerated holes, consistent with typical wastewater compositions.

The composite Bi-MOF/g-CN-15 degrades approximately 63% of CIP within 60 min at an approximate photocatalyst loading of 2 mg, while simultaneously converting chemical energy from pollutant breakdown into electrical energy. An open-circuit voltage of 1.68 V and a short-circuit current density of 3.15 mA cm⁻² with a maximum power density of 1.81 mW cm⁻² are produced by the PFC system. The highest electrical output is achieved by the composite Bi-MOF/g-CN-15, compared with the other composites and the pristine material, indicating the ordered dispersion of Bi-MOF microspheres on stacked g-CN, thereby enabling effective heterojunction formation. Through the PFC study, the synergistic properties of Bi-MOF/g-CN-15 are demonstrated, yielding the best photoelectrocatalytic performance for the effective conversion of chemical energy into electrical energy. The current–voltage (J–V) and power-density characteristics of the PFC (Fig. 6f) reveal clear performance enhancement for the heterojunction electrodes. The Bi-MOF/g-CN-15 photoanode delivers an open-circuit voltage (V_oc) of 1.59 V, a short-circuit current density (J_sc) of 3.15 mA cm⁻², and a maximum power density (P_max) of 1.81 mW cm⁻² significantly surpassing pristine g-CN (V_oc = 1.47 V; P_max = 1.13 mW cm⁻²) and Bi-MOF (V_oc = 1.46 V; P_max = 1.09 mW cm⁻²). This substantial internal bias originates from the favourable band alignment: electrons transfer from Bi-MOF to g-CN, creating a strong built-in electric field across the interface. Consequently, electrons flow efficiently from the photoanode to the Pt photocathode through the external circuit, while holes accumulate at the photoanode surface to oxidize the pollutant. This Z-scheme electron flow dramatically improves both charge separation and CIP degradation, fully consistent with the PEC trends of the composite.

In terms of pollutant removal within the PFC, Bi-MOF/g-CN-15 again exhibits superior degradation efficiency (63%) compared with g-CN (44%) and Bi-MOF (41%), confirming that the rapid extraction of electrons into the external circuit enhances hole availability for oxidation. The enhanced generation and migration of charge carriers is also supported by the increased short-circuit current and power density. The incident photon-to-current efficiency (IPCE) calculated from the monochromatic bandpass further confirms the efficient utilization of photogenerated carriers. To quantify the energy conversion capability, the chemical-to-electricity conversion efficiency (η) was calculated using eqn (8):

The calculated chemical-to-electricity conversion efficiencies (η) further support this trend: g-CN (1.13%), Bi-MOF (1.09%), Bi-MOF/g-CN-10 (1.70%), Bi-MOF/g-CN-15 (1.81%), and Bi-MOF/g-CN-25 (1.59%). The slight decrease observed for Bi-MOF/g-CN-25 arises from excessive g-CN coverage, which partially blocks Bi-MOF active sites, lowers light absorption and electron mobility, and ultimately reduces P_max and photocatalytic efficiency. Overall, the strong correlation between PFC output parameters (V_oc, J_sc, and P_max) and CIP degradation demonstrates that Bi-MOF/g-CN-15 possesses the most efficient charge separation and fastest interfacial electron transfer, enabling it to function simultaneously as an efficient photoanode for power generation and a high-performance catalyst for wastewater treatment.

3.5. Photodegradation, fuel cell, and the H₂ evolution mechanism

To elucidate the charge transfer behaviour and photocatalytic mechanism, the electronic structure and band potentials of the photocatalyst are determined using the Valence band spectra and UPS spectra and are shown in Fig. 7. From the UV-absorption spectra, the band gap is found to be 2.61, 3.40, and 3.05 eV for g-CN, Bi-MOF, and the composite Bi-MOF/g-CN-15. The optimisation of the band gap in the composite is due to the band bending for seamless transfer of electrons from the valence band to the conduction band through the interface with minimised recombination. This results in the accumulation of strong electrons available for efficient oxidation of the pollutant. The direction of the charge flow can be determined through the work function from the width of UPS spectra, i.e., UPS's cut off and Fermi region, as shown in Fig. 7a for g-CN, Bi-MOF, and Bi-MOF/g-CN. The work function is determined by deducting the cut-off energy of the secondary electron from the incident energy (He: 21.2 eV) used in the instrument, using eqn (9):

Φ = hν − (width of UPS) (9)

Fig. 7 (a) UPS spectra, (b) XPS valence band spectra of g-CN, Bi-MOF, and Bi-MOF/g-CN-15, and (c) photocatalytic mechanism for antibiotic degradation.

and the values are 4.87 and 4.45 eV for g-CN and Bi-MOF, respectively. These values translate to 0.43 and 0.01 eV (vs. NHE), which are the positions of E_f. This creates an internal electric field at the interface with the bending of the band, enabling efficient separation and transfer of charges in the composite.⁶⁷ Furthermore, the position of the valence band is calculated from the XPS Valence band spectra. The gap between the Fermi (E_f) and actual valence band was determined to be 1.94 eV for g-CN, 1.79 eV for Bi-MOF, and 1.70 eV for Bi-MOF/g-CN as displayed in Fig. 7b. Based on these values, the E_VB values are estimated to be 2.42 eV and 1.8 eV (vs. NHE) for g-CN and Bi-MOF, respectively. Combined with the optical band gap from the absorption spectra and the equation E_CB = E_VB − E_g, the value of E_CB is calculated to be −0.19 eV and −1.6 eV (vs. NHE) for g-CN and Bi-MOF, respectively. This positioning of valence and conduction bands results in a staggered band alignment with bending. Since the Fermi level of Bi-MOF is more negative than that of g-CN (Φ_g-CN > Φ_Bi-MOF), ideally, the electrons flow from Bi-MOF to g-CN until their Fermi levels are completely aligned. This creates a negatively charged state on g-CN and a positive hole-rich state on Bi-MOF. This was also confirmed by Bi 4f XPS, which showed a higher binding energy for the composite relative to Bi-MOF, indicating electron transfer from Bi-MOF to g-CN. Therefore, in this case, a Z-scheme pathway becomes more apparent due to the accumulation of opposite charges at the two interfaces, which enabled band bending, creating an internal electric field from Bi-MOF to g-CN. Under irradiation, the photogenerated CB electrons of g-CN combine with the VB holes in Bi-MOF, forming a Z-type electron flow pathway, preserving highly reducing electrons at the Bi-MOF surface and highly oxidizing holes at the g-CN VB. This extends the lifetime of the photo-excited electrons in the CB of Bi-MOF and photo-excited holes in the VB of g-CN, rendering them available for redox reactions in photocatalysis.⁶⁸

The band edge position determines the formation of reactive oxygen species, which are the primary oxidizing agents in photocatalytic pollutant degradation. The high-energy electrons of the conduction band readily undergo redox reactions with the species present on the photocatalyst's surface. Since the conduction band of Bi-MOF is more negative than the potential of O₂/˙O₂⁻ (−0.33 eV), the abundant photo-excited electrons react with the dissolved O₂ and produce superoxide radicals, while the rest of the electrons are transferred to the external circuit, producing electricity. Meanwhile, the holes remaining in the VB of g-CN (+2.42 eV) are well within sufficient oxidative power to oxidize surface water to the hydroxide radical ˙OH/H₂O (+1.99 eV). Therefore, accumulated photogenerated holes in the valence band and the superoxide radicals generated at the conduction band attack the CIP, leading to its decomposition into smaller non-toxic organic molecules.⁶⁹ These reactive species serve as the primary oxidative species, attacking the π-conjugated band and aromatic ring in the pollutant molecule to yield smaller oxygenated intermediates. The superoxide radical selectively attacks the electron-rich aromatic rings and C=C through nucleophilic addition, initiating ring oxidation with low activation barriers. In addition, the photoinduced hole directly oxidizes adsorbed organic molecules, further accelerating degradation.⁷⁰ Furthermore, ESR spectra confirm the generation of ˙O₂⁻ under the light irradiation, with the Bi-MOF/g-CN-15 composite showing the highest signal intensity (Fig. S7). The scavenger test demonstrated that ˙O₂⁻ and h⁺ are the main reactive species in CIP degradation. The formation and attack of reactive species onto the piperazine moiety of ciprofloxacin⁷¹ induces ring opening, followed by N-dealkylation and decarbonylation, as expressed using eqn (10)–(13),

Bi-MOF/g-CN + hv → Bi-MOF (h⁺, e⁻)/g-CN (h⁺, e⁻) (10)

Bi-MOF (h⁺, e⁻)/g-CN (h⁺, e⁻) → Bi-MOF (e⁻) + g-CN (h⁺) (11)

Bi-MOF (e⁻) + O₂ →˙O₂⁻ (12)

h⁺/˙O₂⁻ + ciprofloxacin → ring opening → small organic molecules → CO₂ + H₂O + F⁻ + NO₃⁻ (13)

The morphological changes maximise the surface area, creating numerous pores and defect sites for pollutant adsorption, whereas the Z-scheme band alignment enables electron accumulation on Bi-MOF and hole accumulation at g-CN.^71–73 This charge transfer process facilitates continuous generation of ROS, leading to rapid and complete degradation of the pollutant into an environmentally benign product while simultaneously generating electrical power in the photoelectrochemical fuel cell configuration.

In the PFC configuration, the Bi-MOF/g-CN composite acts as a photoanode that simultaneously degrades organic pollutants and generates electrical energy. During the light irradiation, the heterojunction generates an electron–hole pair with efficient spatial separation due to the Z-scheme band alignment. The organic pollutant adsorbed on the photoanode surface acts as the fuel source. When oxidized by photogenerated holes, they provide electrons to the photoanode, which subsequently flow through the external circuit toward the cathode, generating photocurrent and power output. An internal bias is established by the Fermi-level difference between the photoanode and cathode, driving spontaneous electron flow through the external circuit without requiring an externally applied voltage. This internal bias is the fundamental driving force enabling self-power generation. In the photocatalytic HER, the sensitizer EY plays an important role in overall photocatalytic activity because, under light irradiation, EY acts as an efficient photosensitizer and initiates the HER. Therefore, in the absence of EY, no H₂ is produced. During the light irradiation, EY is excited from the ground state to the singlet excited state (EY1*), which undergoes intersystem crossing to generate the long-lived triplet excited state (EY3*). In the presence of a sacrificial electron donor, EY3* is reductively quenched to form the strong reducing radical species EY˙⁻. The generated electrons are injected into the conduction band of Bi-MOF, owing to the favorable energetic alignment between EY˙⁻ and Bi-MOF. Subsequently, driven by the Z-scheme configuration, electrons diffuse from the conduction band of g-CN to the valence band of Bi-MOF. This directional electron migration results in electron accumulation at the Bi-MOF side of the heterojunction, while positive charges are retained at the g-CN interface. Such spatial separation of charge carriers effectively suppresses electron–hole recombination and prolongs carrier lifetime. The accumulated electrons on Bi-MOF act as active reduction centres for proton reduction, leading to efficient H₂ evolution, while the oxidized sacrificial agent consumes the holes from g-CN. Meanwhile, EY is regenerated and continuously participates in the sensitization cycle. Compared with pristine g-CN and Bi-MOF, the Bi-MOF/g-CN-15 heterostructure exhibits significantly enhanced interfacial charge separation and electron utilization efficiency, which accounts for its markedly superior hydrogen evolution performance.⁷⁴

4. Conclusion

In summary, a Z-scheme heterojunction photocatalyst was successfully constructed with a MOF and conductive polymer as a structural template, growing 2D/2D microspheres as Bi-MOF/g-CN-15. The crystalline and porous nature of the Bi-MOF was retained in the composite, providing active adsorption sites for the catalytic reaction. The stacked morphology of g-CN reduces the charge recombination, creating active centres for the photocatalytic and redox reactions. Under simulated sunlight, 99% of the ciprofloxacin removal percentage was achieved for the composite Bi-MOF/g-CN-15 under 90 min. The composite retained its degrading ability with a higher/lower pollutant concentration (1–20 ppm) and initial pH (2–10), even with most of the interfering ions. A higher rate of the photocatalytic HER is observed in the composite, producing 2111.03 µmol g⁻¹ h⁻¹ of H₂. This highlights the strong overpotential offered by the composite for H₂ evolution. The composite also exhibited redox ability by converting the chemical energy produced from CIP degradation to electricity. The composite exhibits excellent electrical conductivity, as evidenced by a short-circuit current of 3.15 mA cm⁻² and a power density of 1.8 mW cm⁻². The stability and recyclability rate of the composite in degrading the pollutant is found to be 84%, with the crystallinity and structural integrity of the composite still preserved. The mechanism of charge transfer for the photodegradation of the pollutant is estimated as a Z-scheme with the built-in electrical field, which regulates the recombination and directs the photocarriers towards reactive species production. This concurrent photoactivity and redox ability make the Bi-MOF/g-CN-15 composite suitable for real-time simultaneous water treatment and energy production. In the future, an effective PFC system can be built using this framework to selectively degrade a broad spectrum of pharmaceutical pollutants in a single chamber.

Author contributions

Sri Vanaja: conceptualization, data curation, investigation, methodology, software, writing – original draft. Esakkinaveen Dhanaraman: conceptualization, data curation, investigation, writing – original draft. Atul Verma: data curation, formal analysis, investigation. Sathish Kumar: formal analysis, investigation. Tanay Kundu: investigation, supervision, writing – review & editing. Yen-Pei Fu: funding acquisition, investigation, resources, supervision, writing – review & editing.

Conflicts of interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Data availability

The data supporting this article have been included as part of the supplementary information (SI). Supplementary information: characterisation details and experimental data. See DOI: https://doi.org/10.1039/d6ta00414h.

Acknowledgements

The authors express their gratitude to the National Science and Technology Council, Taiwan, for financial support under the contract numbers NSTC 112-2221-E-259-002-MY3 and NSTC 114-2221-E-259-010. Authors Sri Vanaja and Tanay Kundu sincerely thank the MOE of Taiwan for supporting this study through the TEEP Internship. Additionally, the authors appreciate the instrumentation centre at National Tsing Hua University for providing access to EPR analysis facilities. The authors also thank Prof. H.W. Yen and Dr P.H. Chiu for their assistance with the instrumentation centre's TEM/STEM/EDS experiments at National Taiwan University. Special thanks to Mr Cheng Han-Tsai for coordinating instrument bookings and Mr Sathya G S for helping with PFC electrochemical experiments.

References

1. T. Ibn-Mohammed, K. B. Mustapha, J. Godsell, Z. Adamu, K. A. Babatunde, D. D. Akintade, A. Acquaye, H. Fujii, M. M. Ndiaye, F. A. Yamoah and S. C. L. Koh, Resour. Conserv. Recycl., 2021, 164, 105169.
2. C. Bhagat, M. Kumar, V. K. Tyagi and P. K. Mohapatra, npj Clean Water, 2020, 3, 42.
3. Z. Maghsodian, A. M. Sanati, T. Mashifana, M. Sillanpää, S. Feng, T. Nhat and B. Ramavandi, Antibiotics, 2022, 11, 1461.
4. A. Saravanan, V. C. Deivayanai, P. S. Kumar, G. Rangasamy, R. V. Hemavathy, T. Harshana, N. Gayathri and K. Alagumalai, Chemosphere, 2022, 308, 136524.
5. H. Furukawa, K. E. Cordova, M. O’Keeffe and O. M. Yaghi, Science, 2013, 341, 1230444.
6. E. Dhanaraman, A. Verma and Y. P. Fu, ChemNanoMat, 2025, 11, e202500282.
7. Z. Wang, M. Xiao, J. You, G. Liu and L. Wang, Acc. Mater. Res., 2022, 3, 1127–1136.
8. Y. Zhang, H. Li and D. Yin, Nanomaterials, 2025, 15, 331.
9. V. H. Nguyen, T. D. Nguyen and T. Van Nguyen, Top. Catal., 2020, 63, 1109–1120.
10. A. Balapure, J. Ray Dutta and R. Ganesan, RSC Appl. Interfaces, 2024, 1, 43–69.
11. L. Zhang, S. Hou, T. Wang, S. Liu, X. Gao, C. Wang and G. Wang, Small, 2022, 18, 2202252.
12. M. Ling, F. Ma, H. Zheng, Z. Yang, Y. Wu, R. Ji, Y. Yu and L. Li, J. Alloys Compd., 2025, 1010, 177288.
13. A. S. Belousov, A. A. Parkhacheva, E. V. Suleimanov, D. G. Fukina, A. V. Koryagin, I. Shafiq, O. V. Krasheninnikova and V. V. Kuzmichev, Catal. Commun., 2023, 180, 106705.
14. V. Hasija, V.-H. Nguyen, A. Kumar, P. Raizada, V. Krishnan, A. A. P. Khan, P. Singh, E. Lichtfouse, C. Wang and P. Thi Huong, J. Hazard. Mater., 2021, 413, 125324.
15. S. Padhiari, M. Tripathy, S. Panigrahi, P. Aparajita, U. Sahoo, M. Mishra, S. M. Patra and G. Hota, Sep. Purif. Technol., 2026, 385, 136390.
16. U. Sahoo, S. Pattnayak, S. Choudhury, P. Aparajita, S. Das and G. Hota, Opt. Mater., 2025, 160, 116778.
17. U. Sahoo, S. Pattnayak, P. Aparajita, K. S. Nag and G. Hota, J. Indian Chem. Soc., 2025, 102, 102282.
18. S. Tang, N. Li, D. Yuan, J. Tang, X. Li, C. Zhang and Y. Rao, Chemosphere, 2019, 234, 658–667.
19. Z. Wu, G. Zhao, Y. Zhang, J. Liu, Y. Zhang and H. Shi, J. Mater. Chem. A, 2015, 3, 3416–3424.
20. Q. Zeng, S. Chang, A. Beyhaqi, S. Lian, H. Xu, J. Xie, F. Guo, M. Wang and C. Hu, J. Hazard. Mater., 2020, 394, 121425.
21. E. Dhanaraman, A. Verma, P. Anand and Y. P. Fu, J. Environ. Chem. Eng., 2023, 11, 111323.
22. A. C. Wibowo, M. D. Smith and H.-C. zur Loye, CrystEngComm, 2011, 13, 426–429.
23. C. Cao, D. Ma, J. Jia, Q. Xu, X. Wu and Q. Zhu, Adv. Mater., 2021, 33, 2008631.
24. A. Thirumurugan and A. K. Cheetham, Eur. J. Inorg. Chem., 2010, 2010, 3823–3828.
25. X. Zhang, S. Jiang, L. X. Sun, Y. H. Xing and F. Y. Bai, J. Mol. Struct., 2022, 1270, 133895.
26. C. Zhang, Y. Wu, D. Li and H.-L. Jiang, Chem. Sci., 2025, 16, 13149–13172.
27. H. Xin, J. Zhang, Z. Lin, W. Li and F. Ma, RSC Adv., 2025, 15, 28870–28880.
28. M. B. Nguyen, L. H. T. Nguyen, M. T. Le, N. Q. Tran, N. H. T. Tran, P. H. Tran, A. T. T. Pham, L. D. Tran and T. L. H. Doan, J. Ind. Eng. Chem., 2024, 134, 217–230.
29. D. A. Giannakoudakis and T. J. Bandosz, Molecules, 2019, 24, 4529.
30. R. Fagan, D. McCormack, S. Hinder and S. Pillai, Materials, 2016, 9, 286.
31. Y.-J. Yuan, Z. Shen, S. Wu, Y. Su, L. Pei, Z. Ji, M. Ding, W. Bai, Y. Chen, Z.-T. Yu and Z. Zou, Appl. Catal., B, 2019, 246, 120–128.
32. E. Dhanaraman, A. Verma, P. H. Chen, N. D. Chen, Y. Siddiqui and Y. P. Fu, Sol. RRL, 2024, 8, 2300981.
33. H. Idriss, Surf. Sci., 2021, 712, 121894.
34. L. Zhang, F. Xie, J. Liu, Z. Sun, X. Zhang, Y. Wang, Y. Wang, R. Li and C. Fan, Chem. Eng. J., 2022, 450, 138066.
35. L. Zhang, X. Zhou, S. Liu, H. Liu, S. Zhu, Y. Mao, Q. Yang, S. Zhu, C. Zhang, T. Wang and C. Wang, J. Clean. Prod., 2023, 425, 138912.
36. A. Mitra, P. Howli, D. Sen, B. Das and K. K. Chattopadhyay, Nanoscale, 2016, 8, 19099–19109.
37. E. Dhanaraman, A. S. Mary, A. Verma, P. Murugan and Y.-P. Fu, J. Environ. Chem. Eng., 2025, 13, 117848.
38. C. Xu, X. Liu, D. Li, Z. Chen, J. Yang, J. Huang and H. Pan, ACS Appl. Mater. Interfaces, 2021, 13, 20114–20124.
39. T. Van Huynh, V. T. T. Nguyen and O. T. K. Nguyen, Mater. Res. Bull., 2025, 186, 113342.
40. K. Wu, Z. Sun and J. Cui, Cryst. Growth Des., 2012, 12, 2864–2871.
41. K. Sakata, S. Kashiyama, G. Matsuo, S. Uemura, N. Kimizuka and M. Kunitake, ChemNanoMat, 2015, 1, 259–263.
42. W. Gu, Q. Li, H. Zhu and L. Zou, J. Colloid Interface Sci., 2022, 606, 1998–2010.
43. C. Wu, J. Mu, X. Wang, P. Ho, M. Zhou, C. Yang, C. Leung, Y. Huang and W. Ho, Small, 2025, 21, 2502116.
44. L. Zhang, H. Wang, Z. Chen, P. K. Wong and J. Liu, Appl. Catal., B, 2011, 106, 1–13.
45. M. Xu, R. Zhang, J. He, Z. Li, Y. Liu, D. Chen, L. Peng, Y. Shu, L. Jiang and G. Zhao, J. Water Proc. Eng., 2024, 66, 106036.
46. S.-M. Zhou, D.-K. Ma, P. Cai, W. Chen and S.-M. Huang, Mater. Res. Bull., 2014, 60, 64–71.
47. L. Cao, J. Huang, X. Wu, B. Ma, Q. Xu, Y. Zhong, Y. Wu, M. Sun and L. Yu, Nanoscale, 2023, 15, 19522–19532.
48. A. Lei, G. Yang, K. Pan, W. Liu, C. Deng, M. Fang, X. Zhang, C. Tan, H. Dong and F. Lai, J. Power Sources, 2025, 625, 235744.
49. J. Wang, Z. Liu, L. Wang, L. Li, Q. Liu, Z. Liu, Z. Cao, Y. Zhang, L. Cheng and J. Yang, J. Alloys Compd., 2024, 1002, 175424.
50. Z. Meng, J. Zhang, H. Long, H. García, L. Zhang, B. Zhu and J. Yu, Angew. Chem., Int. Ed., 2025, 64, e202505456.
51. N. Liu, T. Li, Z. Zhao, J. Liu, X. Luo, X. Yuan, K. Luo, J. He, D. Yu and Y. Zhao, ACS Omega, 2020, 5, 12557–12567.
52. H. S. Moon, K. Hsiao, M. Wu, Y. Yun, Y. Hsu and K. Yong, Adv. Mater., 2023, 35, 2200172.
53. M. Gao, J. Wang, Q. Chu, M. Meng, M. Xu, Y. Gong, Z. Zhang, Y. Cui, X. Wang, Y. Feng and C. Sun, J. Colloid Interface Sci., 2025, 698, 138099.
54. S. Chandrappa, S. J. Galbao, A. Furube and D. H. K. Murthy, ACS Appl. Nano Mater., 2023, 6, 19551–19572.
55. B. Hu, M. Hu, Q. Guo, K. Wang and X. Wang, Appl. Catal., B, 2019, 253, 77–87.
56. M. Delnavaz, S. Amiri and S. Najari, Ecotoxicol. Environ. Saf., 2025, 295, 118175.
57. Q. Sun, S. Huang, Z. Li, D. Su and J. Sun, J. Environ. Chem. Eng., 2023, 11, 109065.
58. K. M. Alam, P. Kumar, P. Kar, U. K. Thakur, S. Zeng, K. Cui and K. Shankar, Nanoscale Adv., 2019, 1, 1460–1471.
59. H. D. Tran, D. Q. Nguyen, P. T. Do and U. N. P. Tran, RSC Adv., 2023, 13, 16915–16925.
60. Y. Cao, X. Yuan, H. Chen, H. Wang, Y. Chen, J. Chen, H. Huang, Y. Mou, Z. Shangguan and X. Li, Chem. Eng. J., 2023, 456, 140971.
61. Y.-W. Li, S.-Z. Li, M. Zhao, L.-Y. Liu, Z.-F. Zhang and W.-L. Ma, Sci. Total Environ., 2023, 896, 165316.
62. D. E. Santiago, J. Araña, O. González-Díaz, M. E. Alemán-Dominguez, A. C. Acosta-Dacal, C. Fernandez-Rodríguez, J. Pérez-Peña and J. M. Doña-Rodríguez, Appl. Catal., B, 2014, 156–157, 284–292.
63. C. A. Igwegbe, S. N. Oba, C. O. Aniagor, A. G. Adeniyi and J. O. Ighalo, J. Ind. Eng. Chem., 2021, 93, 57–77.
64. X. Gao, Q. Guo, G. Tang, W. Peng, Y. Luo and D. He, J. Water Reuse Desalin., 2019, 9, 301–309.
65. N. Huang, S. Liang, X. Liang, H. Chai, Y. Sun, L. Lyu and W. Tan, J. Phys. Chem. Solids, 2026, 208, 113091.
66. A. Verma, E. Dhanaraman, Y.-S. Peng and Y.-P. Fu, Chem. Eng. J., 2025, 505, 159171.
67. B. Ma, Y. Zha, R. Xu, J. Li, Y. Guo, J. Liu, S. Wang, B. Yan, Y. Lan, B. Zhao and H. Xie, J. Water Proc. Eng., 2024, 60, 105134.
68. H. Yang, Mater. Res. Bull., 2021, 142, 111406.
69. N. Liu, N. Lu, Y. Su, P. Wang and X. Quan, Sep. Purif. Technol., 2019, 211, 782–789.
70. Y. Zhou, C. Huang, L. Yang, R. Zhang, Y. Yin, C. Liu, K. Sun, S. Yao, N. Geng, Y. Luo, T. Yang, B. Zou and T. Liu, Energy Mater., 2025, 5, 500010.
71. K. Wang, S. Zeng, G. Li, Y. Dong, Q. Wang, L. Zhang, Z. Ren and P. Wang, Appl. Catal., B, 2024, 343, 123565.
72. X. Du, D. Jiang, Q. Liu, N. Hao and K. Wang, Anal. Chem., 2019, 91, 1728–1732.
73. M. S. Sharifiyan, A. Fattah-alhosseini and M. Karbasi, Appl. Surf. Sci. Adv., 2023, 18, 100541.
74. H. Lim, I. Roh, J. Chung, J. Lee, J. Woo Song and T. Yu, J. Ind. Eng. Chem., 2023, 123, 81–87.

---