Rational design of a self-cleaning PES/UiO-66-NH2@g-C3N4 mixed-matrix membrane for high-efficiency oil–water separation

Wastewater streams containing oil constitute a critical environmental issue associated with numerous industrial activities, particularly those related to petroleum processing, food industries, and textile manufacturing¹. According to assessments by UNEP, marine ecosystems receive nearly 1.3 billion tons of oil each year, with industrial effluents accounting for a significant share of this input². Such oil-laden discharges introduce toxic substances that can persist in aquatic systems, migrate into drinking water reserves, and progressively accumulate within biological food chains, posing serious risks to ecosystem integrity and human health. In line with this concern, EPA reports have highlighted that oil pollution causes long-lasting ecological degradation, adversely affecting marine organisms and imposing substantial socioeconomic burdens on coastal regions^3,4. The urgency for efficient remediation technologies is further emphasized by evidence that inadequate treatment of oily wastewater leads to freshwater contamination, contributing to annual economic losses estimated at approximately USD 4.6 billion due to water pollution-related impacts^5,6,7.

Conventional treatment methods, including gravity separation, flotation, and chemical coagulation, are often insufficient for removing stable oil-in-water emulsions due to limited separation efficiency^5,8,9. These approaches are typically energy-intensive and fail to meet stringent discharge standards. For example, gravity-based processes can remove only free oil, while emulsified and dissolved oil—accounting for up to 70% of total oil content—remains largely untreated^10,11. By comparison, emerging remediation strategies—including membrane-based separation, electrochemical treatment, and biological approaches—have demonstrated superior removal performance and greater effectiveness in addressing oily wastewater contaminants⁵. Membrane technology has been widely implemented in industrial wastewater treatment, particularly in petroleum refining, petrochemical, and food-processing sectors, due to its high separation efficiency and scalability. It is especially effective for treating stable oil–water emulsions through selective separation mechanisms, enabling efficient removal of dispersed oil and contaminants while consistently producing high-quality effluent suitable for discharge or reuse in industrial applications^12,13.

Despite the advantages of membrane-based separation, achieving a balance between high separation efficiency, sustained permeation flux, and long-term fouling resistance remains a critical challenge. Conventional membranes are inherently prone to fouling, resulting in progressive permeability loss and increased maintenance requirements^14,15. To address these limitations, mixed-matrix membranes (MMMs) incorporating functional nanomaterials have been extensively investigated. Nevertheless, current MMMs still exhibit several fundamental shortcomings, including non-uniform filler dispersion and aggregation, which can generate structural defects and compromise selectivity, as well as weak interfacial compatibility between fillers and polymer matrices, leading to diminished mechanical stability and durability. Moreover, most systems rely predominantly on passive antifouling strategies based on enhanced hydrophilicity, which are insufficient to prevent fouling layer development under continuous operation. Photocatalytic membranes also face challenges related to inadequate integration and stability of catalytic components. Consequently, there is a need for membrane systems that integrate controlled filler distribution, strong interfacial interactions, and active antifouling mechanisms. In this regard, UiO-66-NH₂, a zirconium-based MOF with high structural stability and functional versatility, represents a promising candidate for advanced membrane design^{16,17,18,19,20}.

Polyethersulfone (PES) is widely utilized as a membrane-forming polymer due to its excellent film-forming capability, controllable phase-inversion behavior, and ability to generate asymmetric porous structures with well-defined selective layers. These characteristics enable precise tuning of pore morphology and permeability, making PES particularly suitable for pressure-driven membrane processes such as microfiltration and ultrafiltration^12,21,22,23. In addition to PES, various other polymers, including polysulfone (PSF)²⁴, polyvinylidene fluoride (PVDF)²⁵, polyamide (PA)²⁶, polyethylene (PE)²⁷, and polypropylene (PP)²⁸, are employed in membrane technology for oily wastewater treatment. Each of these materials exhibits distinct physicochemical attributes, such as chemical stability, mechanical durability, and hydrophilicity, which significantly impact their separation efficiency. However, PES membranes often suffer from inherent hydrophobicity and fouling when exposed to oil-contaminated environments, limiting their long-term operational efficiency²⁹. Consequently, the incorporation of functional additives has emerged as an effective strategy to enhance membrane surface properties and separation efficiency³⁰.

Compared with other MOFs such as ZIF-8 and MIL-type frameworks, UiO-66-NH₂ exhibits superior hydrolytic and chemical stability under aqueous and harsh operating conditions, primarily due to the strong Zr–O coordination bonds within its framework. In addition, the presence of –NH₂ functional groups enhances surface hydrophilicity and promotes favorable interfacial interactions with water molecules, which are essential for improving antifouling performance. These attributes make UiO-66-NH₂ particularly suitable for long-term oil–water separation applications compared to MOFs with lower water stability or limited surface functionality. When incorporated into a PES matrix, UiO-66-NH₂ contributes through its high specific surface area and tunable pore architecture, which regulate interfacial interactions and facilitate selective transport. Moreover, the amine functionalities promote hydrogen bonding with water, leading to reduced interfacial free energy, enhanced wettability, and suppressed oil adhesion. Collectively, these properties enable improved separation efficiency, reduced fouling propensity, and enhanced operational stability in membrane-based oily wastewater treatment systems^20,31,32.

Graphitic carbon nitride (g-C₃N₄) imparts visible-light-responsive photocatalytic functionality to the composite membrane, enabling degradation of organic foulants responsible for membrane fouling. Due to its high stability, low toxicity, and efficient reactive oxygen species generation, g-C₃N₄ has attracted considerable attention as a photocatalytic antifouling material^33,34. Its incorporation within polymeric membranes can mitigate fouling accumulation, improve flux stability, and reduce dependence on aggressive chemical cleaning procedures^35,36,37.Recent studies have shown that PES membranes modified with MOFs or photocatalytic nanomaterials can enhance permeability, hydrophilicity, and antifouling performance for oily wastewater treatment^38,39. Membranes incorporating ZIF-8, MIL-type, and UiO-based MOFs have demonstrated improved separation efficiency owing to their porous structures and tunable surface functionalities, while PES/g-C₃N₄ systems have been explored for visible-light-responsive photocatalytic antifouling applications^40,41. However, many reported systems still suffer from filler aggregation, weak filler–polymer interfacial compatibility, limited long-term stability, and insufficient integration between separation and photocatalytic regeneration functions. Consequently, most membranes achieve either enhanced separation performance or photocatalytic activity, but not both simultaneously within a structurally stable framework⁴². Moreover, studies combining UiO-66-NH₂ and g-C₃N₄ within a PES matrix remain scarce, highlighting the need for multifunctional membranes with synergistic separation and antifouling capabilities.

The novelty of the present study lies in the rational integration of PES, UiO-66-NH₂, and g-C₃N₄ to develop a multifunctional composite membrane with enhanced interfacial properties and photocatalytic antifouling capability. This design strategy directly addresses the key limitations of conventional membranes, particularly fouling susceptibility and reduced separation efficiency. The resulting membrane is expected to exhibit high oil rejection, improved flux recovery, and enhanced operational stability, thereby providing a sustainable and advanced platform for next-generation oily wastewater treatment technologies.

Materials

Polyethersulfone (PES, ≥ 98%) was used as the membrane-forming polymer. Zirconium(IV) chloride (ZrCl₄, ≥ 99%) and 2-aminoterephthalic acid (≥ 99%) served as precursors for the synthesis of UiO-66-NH₂. Melamine (≥ 99%) was employed as the nitrogen-rich source for preparing graphitic carbon nitride (g-C₃N₄). N, N-dimethylformamide (DMF, ≥ 99%) and ethanol (≥ 99%) were used as solvents during membrane fabrication and material synthesis. Glutaraldehyde (≥ 95%) was utilised as a crosslinking agent to enhance membrane stability. Deionised water with a resistivity of ≥ 18.2 MΩ·cm was produced in-house and used throughout all experimental procedures.For evaluating photocatalytic activity, methylene blue (MB, ≥ 98%) and rhodamine B (RhB, ≥ 99%) were selected as representative organic dye pollutants. All chemicals were of analytical grade and were obtained from Sigma-Aldrich and used without further purification.

Synthesis methods of (PES/UiO-66-NH₂@g-C₃N₄)composite membrane

Preparation of UiO-66-NH₂MOF

UiO-66-NH₂ was synthesised via a solvothermal approach in which the molar ratio of precursors was carefully controlled. Zirconium(IV) chloride (0.50 g, 2.14 mmol) was first treated under vacuum at 80 °C for 4 h to remove trace moisture. Separately, 2-aminoterephthalic acid (0.40 g, 2.21 mmol) was dissolved in 25 mL of DMF with ultrasonic agitation for 30 min to form a clear ligand solution. The dehydrated zirconium salt was then gradually added to the ligand solution, and the mixture was further sonicated for 15 min to obtain a uniform reaction dispersion. This dispersion was transferred into a sealed Teflon-lined stainless-steel vessel and thermally processed through sequential heating at 100 °C for 1 h followed by prolonged treatment at 120 °C for 24 h to promote MOF framework development. After natural cooling to ambient conditions, the solid product was collected by filtration, washed thoroughly with ethanol (three successive portions of 50 mL) to remove residual impurities, and finally dried under vacuum at 60 °C for 12 h before use⁴³.

Preparation of g-C₃N₄

g-C₃N₄ was synthesised by thermally converting melamine through a solid-state polymerisation pathway. In this process, 1.5 g of melamine was confined within a covered ceramic crucible to minimise volatilisation during heating. The crucible was placed in a muffle furnace and subjected to a programmed temperature increase from ambient conditions to 550 °C at a ramp rate of 5 °C min⁻¹, followed by an isothermal holding period of 2 h to enable structural condensation into a conjugated heptazine framework. After completion of the thermal cycle, the furnace was allowed to cool naturally to room temperature. The resulting solid material was mechanically milled using an agate mortar to reduce particle agglomeration and subsequently sieved through a 200-mesh screen to obtain a uniformly fine powder, thereby improving its dispersibility in the polymer casting solution for membrane fabrication⁴⁴.

Preparation of the PES solution

A homogeneous polymer casting solution was prepared by dissolving 2.0 g of polyethersulfone (PES) in 20 mL of DMF to obtain a polymer concentration of 10 wt%. The mixture was gradually heated to 60 °C under continuous magnetic stirring at 250 rpm and maintained under these conditions for 3 h to ensure complete polymer dissolution and removal of micro-aggregates. The resulting dope solution exhibited a clear, bubble-free, and viscous appearance, indicating successful molecular-level dispersion of PES chains within the solvent matrix. The selected polymer concentration was carefully optimised to balance solution viscosity, membrane formability, and phase-inversion kinetics, thereby providing sufficient mechanical integrity while preserving an interconnected porous structure after coagulation. This composition is particularly suitable for mixed-matrix membrane fabrication, as it facilitates uniform incorporation of inorganic fillers without inducing excessive aggregation or phase separation during casting and subsequent phase inversion¹².

Fabrication, phase inversion, and crosslinking of the composite membrane

The PES/UiO-66-NH₂@g-C₃N₄membrane was fabricated via a controlled phase-inversion route. Briefly, UiO-66-NH₂ (0.50 g) was pre-dispersed in 20 mL of DMF by ultrasonication for 30 min, while g-C₃N₄ (0.30 g) was separately dispersed in 10 mL of DMF and ultrasonicated for 60 min to ensure stable, agglomeration-free suspensions. The preformed UiO-66-NH₂ suspension was slowly incorporated into the PES solution, after which the g-C₃N₄ dispersion was added sequentially. The combined mixture was maintained under continuous magnetic agitation at 60 °C and 300 rpm for 3 h to ensure uniform incorporation of both fillers throughout the polymer matrix. The homogeneous casting solution was then spread onto a thoroughly cleaned glass substrate using a film applicator set to a 200 μm gap, with the process conducted under controlled environmental conditions (20–25 °C, low humidity) to limit atmospheric moisture interference during membrane formation⁴⁵.Following casting, the nascent film was exposed to ambient air for 2 h to allow controlled solvent evaporation before being transferred into 500 mL of deionised water and maintained for 24 h to induce complete phase inversion and extract residual DMF. The resulting membrane was then thoroughly rinsed, dried under vacuum at 40 °C, and subjected to chemical stabilisation by immersion in a 1 wt% aqueous glutaraldehyde solution for 40 min. After crosslinking, the membrane was extensively washed and dried again to improve its mechanical integrity and long-term durability⁴⁶.

Characterization of PES/UiO-66-NH₂@g-C₃N₄membrane

The physicochemical characteristics of the PES/UiO-66-NH2@g-C3N4 composite membrane were systematically examined using a combination of complementary analytical techniques. FTIR analysis (Nicolet iS10) was employed to probe the chemical functionalities and intermolecular interactions over the spectral region of 4000–400 cm⁻¹. The crystalline features and phase integrity of the membrane components were evaluated by XRD using a Bruker D8 Advance diffractometer. Surface and cross-sectional morphologies, as well as the spatial distribution of embedded fillers, were investigated by SEM (JEOL JSM-7100 F). In addition, surface wettability was quantified through static water and oil contact angle measurements conducted with a Krüss DSA100 goniometer to assess hydrophilic and oleophobic characteristics, where hexadecane was used as the model oil phase for oleophobicity evaluation. Collectively, these analyses provide a comprehensive understanding of the structural and surface properties of the composite membrane relevant to wastewater treatment applications.

Filtration experiment

The oil–water separation performance of the PES/UiO-66-NH₂@g-C₃N₄ composite membrane was evaluated using a laboratory-scale pressure-driven filtration system consisting of a feed reservoir, a peristaltic pump for maintaining a constant flow rate, and a membrane filtration cell equipped with the fabricated composite membrane. A model oil–water emulsion was prepared by dispersing the oil phase in deionized water at a concentration of 200 mg L⁻¹ under continuous high-speed stirring. To enhance emulsion stability and minimize rapid phase separation during filtration, sodium dodecyl sulfate (SDS, 0.1 wt%) was employed as an anionic surfactant stabilizing agent. The resulting mixture was subsequently homogenized under controlled mixing conditions prior to each experiment to ensure reproducible emulsion preparation and consistent operating conditions throughout the filtration tests. Although detailed droplet-size distribution analysis was not performed in the present study, all emulsions were prepared using identical procedures to ensure reliable comparison of membrane performance⁴⁷:

WhereV is the volume of permeate collected (in liters), A is the effective area of the membrane (in square meters), and t is the time of filtration (in seconds).

Rejection efficiency (R), which indicates the effectiveness of the membrane in removing oil from the water, was determined using the equation⁴⁷:

where Cp is the concentration of oil in the permeate (in mg/L), and C_f is the concentration of oil in the feed solution (in mg/L). The experiment was performed under varying conditions, such as transmembrane pressure and flow rates, to assess their impact on permeation flux and rejection efficiency. The results provided insights into the membrane’s effectiveness in oil-water separation, reinforcing its potential for wastewater treatment applications.

Photocatalytic self-cleaning evaluation

The visible-light-driven self-cleaning behaviour of the PES/UiO-66-NH₂@g-C₃N₄ composite membrane was examined using MB and RhB as model organic probes. Aqueous dye solutions with initial concentrations of 10–50 mg L⁻¹ were prepared, and membrane specimens (2 × 2 cm) were immersed in 50 mL of the solution. The pH was adjusted to ~ 6.5 using dilute HCl or NaOH to optimise surface charge interactions and favour ROS generation. Prior to irradiation, the system was maintained under dark conditions with continuous stirring for 30 min to establish adsorption equilibrium. Photocatalytic experiments were subsequently performed under visible-light illumination using an LED source (400–700 nm). During irradiation, the membrane specimens were maintained in a stable fixed position under controlled mild stirring conditions to promote homogeneous contaminant dispersion and minimize concentration polarization while reducing excessive mechanical stress, uncontrolled membrane movement, and possible detachment of photocatalytic components from the membrane surface. At fixed time intervals (30 min), 3 mL aliquots were withdrawn, filtered, and analysed by UV–Vis spectroscopy (Shimadzu UV-1800). Photodegradation performance was quantified by monitoring absorbance attenuation at 664 nm for MB and 554 nm for RhB, enabling direct assessment of the membrane’s visible-light-induced self-cleaning efficiency⁴⁸.

Long-term continuous filtration stability test

The long-term continuous filtration stability of the PES/UiO-66-NH₂@g-C₃N₄ composite membrane was evaluated using a laboratory-scale pressure-driven filtration system operated under steady-state conditions. A model oil–water emulsion containing 200 mg L⁻¹ of oil was fed continuously into the filtration system under steady operating conditions, with the transmembrane pressure maintained at 0.5 MPa, the temperature controlled at 25 ± 1 °C, and the feed solution adjusted to near-neutral pH. The membrane was operated continuously for 12 h without any intermediate physical or chemical cleaning in order to realistically simulate practical long-term operation. Permeate samples were collected at predefined time intervals (every 1 h), and the corresponding permeate flux and oil rejection were determined to monitor temporal performance evolution. This testing protocol was designed to evaluate intrinsic fouling resistance, flux stability, and separation robustness under prolonged operation, thereby providing a realistic assessment of membrane durability and applicability for long-term oily wastewater treatment under industrially relevant conditions⁴⁹.

Energy consumption and specific energy demand analysis

The energy consumption associated with the oil–water separation process was evaluated by calculating the specific energy demand (SED) of the membrane filtration system. The SED was determined based on the applied transmembrane pressure, permeate flux, and effective membrane area during steady-state operation. Assuming a pressure-driven filtration process, the specific energy demand was calculated using the relationship SED (kWh m⁻³) = (ΔP × Q⁻¹)/η, where ΔP is the transmembrane pressure (Pa), Q is the volumetric permeate flow rate (m³ s⁻¹), and η represents the overall pump efficiency, which was assumed to be constant and taken from literature-reported values for laboratory-scale systems. The permeate flow rate was derived from experimentally measured permeate flux and membrane area. All calculations were performed under identical operating conditions (ΔP = 0.5 MPa, 25 ± 1 °C) to enable direct comparison between membranes. This evaluation enables quantitative determination of the process energy efficiency and facilitates assessment of the composite membrane’s operational viability for sustained oily wastewater treatment^50,51.

Characterization of PES/UiO-66-NH₂@g-C₃N₄ composite membrane

FTIR analysis

FTIR spectroscopy was employed to elucidate the chemical functionalities and interfacial interactions present in the PES/UiO-66-NH₂@g-C₃N₄ composite membrane, with the corresponding spectra of pristine PES, UiO-66-NH₂, g-C₃N₄, and the composite membrane shown in (Fig. 1). The spectrum of neat PES is characterised by distinct absorption bands at approximately 1240 cm⁻¹, attributed to asymmetric stretching of sulfone (O = S=O) groups, 1150 cm⁻¹ corresponding to C–O–C stretching vibrations, and 1510 cm⁻¹ associated with aromatic C = C skeletal vibrations¹². The UiO-66-NH₂ spectrum displays peaks at 1660 cm⁻¹ (C = O stretching from carboxylate groups), 1250 cm⁻¹(Zr–O stretching), and 3470 cm⁻¹(-NH₂ stretching), confirming the presence of amine-functionalized MOF⁵². The g-C₃N₄ spectrum shows bands at 795 cm⁻¹ (triazine ring breathing), 1230–1640 cm⁻¹ (C–N and C = N stretching), and a broad peak at 3200–3400 cm⁻¹, attributed to N–H stretching⁵³.The FTIR spectrum of the PES/UiO-66-NH₂@g-C₃N₄ composite retains the signature absorption features of each constituent, accompanied by discernible shifts in the bands located near 1660 and 3470 cm⁻¹. These spectral changes signify the establishment of strong interfacial interactions, predominantly hydrogen bonding, among PES, UiO-66-NH₂, and g-C₃N₄. Such interactions play a crucial role in improving membrane stability, oil–water separation efficiency, and resistance to fouling.

XRD analysis

XRD analysis was employed to investigate the structural ordering of pristine PES, UiO-66-NH₂, g-C₃N₄, and the PES/UiO-66-NH₂@g-C₃N₄ composite membrane, as illustrated in (Fig. 2). The diffraction pattern of neat PES is dominated by a broad diffuse band spanning 2θ ≈ 18–22°, indicative of its predominantly amorphous nature and the lack of long-range crystallinity within the polymer matrix⁵⁴. The UiO-66-NH₂ MOF displays sharp and well-defined reflections at 2θ ≈ 7.4°, 8.6°, 12.1°, 14.0°, 17.1°, and 25.7°, characteristic of its highly crystalline Zr-based framework⁵⁵. The XRD profile of g-C₃N₄ is characterised by a low-intensity reflection at 2θ ≈ 12.1°, assigned to the in-plane ordering of tri-s-triazine units, together with a strong diffraction peak at 2θ ≈ 27.4°, which originates from the interlayer stacking of π-conjugated aromatic frameworks⁵⁶. In the composite membrane, the amorphous PES halo is retained, while the characteristic diffraction peaks of UiO-66-NH₂and g-C₃N₄ are present with reduced intensity and slight broadening, indicating their homogeneous dispersion within the polymer matrix without structural degradation. These results confirm successful incorporation and structural integrity of both fillers within the PES membrane⁵⁷.

SEM and EDX analysis

SEM micrographs of the individual components and the fabricated PES/UiO-66-NH₂@g-C₃N₄ composite membrane are presented in (Fig. 3) and (Fig. 4). The UiO-66-NH₂ particles (Fig. 3a) exhibit well-defined polyhedral morphologies with relatively uniform size distribution, confirming successful MOF crystallization²⁰. The g-C₃N₄ (Fig. 3b) displays a characteristic layered and wrinkled sheet-like structure, indicative of exfoliated graphitic domains⁴⁸. The membrane morphology is shown in (Fig. 4). Pristine PES (Fig. 4a) presents a porous sponge-like surface with interconnected pores, typical of phase-inversion membranes⁵⁸.The cross-sectional image (Fig. 4b) reveals an asymmetric structure with a dense selective layer supported by finger-like macrovoids. Surface images (Fig. 4c, d) demonstrate a compact, defect-free layer with uniform incorporation of UiO-66-NH₂ particles and g-C₃N₄ nanosheets within the PES matrix, without noticeable agglomeration. This homogeneous dispersion enhances interfacial stability and contributes to improved antifouling performance⁵⁹. Collectively, uniform filler dispersion and asymmetric structure confirm effective integration, enhancing interfacial stability, permeability, and antifouling performance.

The elemental composition of the fabricated PES/UiO-66-NH₂@g-C₃N₄ composite membrane was further examined using energy-dispersive X-ray (EDX) spectroscopy, as presented in (Fig. 5). The spectrum confirms the presence of the principal elements associated with the membrane constituents, including C, N, O, S, and Zr. Characteristic peaks located at approximately 0.27, 0.39, 0.54, 2.39, and 2.04 keV correspond to C Kα, N Kα, O Kα, S Kα, and Zr Lα transitions, respectively, while weaker peaks near 15.7 and 17.7 keV are attributed to Zr Kα and Zr Kβ emissions. The dominant carbon and nitrogen signals originate mainly from the PES matrix and the nitrogen-rich graphitic framework of g-C₃N₄, whereas the oxygen signal is associated with both the sulfone groups of PES and the oxygen-containing coordination structure of UiO-66-NH₂. The sulfur and zirconium peaks further verify the successful incorporation of PES and UiO-66-NH₂ within the membrane matrix, consistent with the SEM observations and membrane fabrication strategy⁶⁰.

TGA analysis

Thermogravimetric analysis (TGA) was conducted to investigate the thermal stability of pristine PES, UiO-66-NH₂, g-C₃N₄, and the fabricated PES/UiO-66-NH₂@g-C₃N₄ composite membrane, as illustrated in (Fig. 6). All samples exhibited an initial slight weight reduction below 150 °C due to the evaporation of physically adsorbed moisture and residual solvent species. Pristine PES underwent its principal thermal degradation between approximately 420 and 580 °C, retaining nearly 5.12% of its original weight at 800 °C owing to decomposition of the polymer backbone structure. UiO-66-NH₂ displayed superior thermal resistance with gradual multistep decomposition behavior and preserved approximately 40.37% residual weight at high temperature, which is attributed to the formation of thermally stable zirconium-containing residues⁶¹. The g-C₃N₄ nanosheets remained thermally stable up to nearly 500 °C before experiencing rapid structural decomposition⁶². Importantly, the PES/UiO-66-NH₂@g-C₃N₄ composite membrane exhibited enhanced thermal stability compared with pristine PES and retained approximately 20.46% residual weight at 800 °C, confirming successful incorporation of the thermally stable functional components within the membrane matrix⁶³.

Photoluminescence (PL) analysis

Photoluminescence (PL) spectroscopy was performed to investigate the charge-carrier recombination behavior and photocatalytic interaction among g-C₃N₄, UiO-66-NH₂, UiO-66-NH₂@g-C₃N₄, and the fabricated PES/UiO-66-NH₂@g-C₃N₄ composite membrane, as illustrated in (Fig. 7). Pure g-C₃N₄ exhibited the highest PL emission intensity with a characteristic broad emission band centered in the visible region, indicating rapid radiative recombination of photogenerated electron–hole pairs. In comparison, UiO-66-NH₂ displayed relatively lower PL intensity, while the UiO-66-NH₂@g-C₃N₄ composite demonstrated a noticeable reduction in emission intensity relative to pristine g-C₃N₄. This behavior suggests partial suppression of charge-carrier recombination due to interfacial interaction between UiO-66-NH₂ and g-C₃N₄. Furthermore, the PES/UiO-66-NH₂@g-C₃N₄ composite membrane exhibited the lowest PL intensity among the investigated samples, indicating enhanced charge separation behavior after incorporation within the polymer matrix. These findings support the improved visible-light-responsive photocatalytic behavior and antifouling potential of the fabricated composite membrane system^62,64.

Contact angle results

The surface wettability characteristics of the fabricated membranes were systematically evaluated by static water and oil (hexadecane) contact angle measurements, and the results are summarised in Table 1. Pristine PES exhibited a relatively high water contact angle (89.0 ± 1.6°), reflecting its intrinsically hydrophobic surface nature. Incorporation of UiO-66-NH₂ markedly reduced the water contact angle to 68.0 ± 1.3°, owing to the introduction of hydrophilic amine functionalities capable of forming hydrogen bonds with water molecules. Further addition of g-C₃N₄ decreased the water contact angle to 60.0 ± 1.2°, attributable to the presence of nitrogen-rich polar domains. The ternary PES/UiO-66-NH₂@g-C₃N₄ membrane exhibited the highest hydrophilicity (58.0 ± 1.1°), confirming a synergistic surface modification effect. In parallel, the oil contact angle increased substantially from 18.0 ± 1.0° for pristine PES to 94.0 ± 1.5° for the composite membrane, indicating pronounced oleophobicity^12,46.This simultaneous enhancement in hydrophilicity and oleophobicity promotes the formation of a stable hydration layer, suppresses oil adhesion, and underpins the improved antifouling behaviour and sustained permeation performance observed during oil–water separation.

Oil-water separation performance of PES/UiO-66-NH₂@g-C₃N₄ composite membrane

The oil–water separation performance of the PES/UiO-66-NH₂@g-C₃N₄ composite membrane was systematically evaluated under a series of well-defined operational conditions. A single-factor experimental strategy was adopted to independently examine the influence of key operating parameters, thereby enabling an unambiguous assessment of their individual contributions to separation behaviour and ensuring high experimental reproducibility. All filtration experiments were conducted using multiple membrane samples under controlled conditions, with a fixed effective membrane area of 0.15 m², an operating temperature of 25 ± 1 °C, and a near-neutral feed solution pH (≈ 7). Based on preliminary optimisation studies, a transmembrane pressure of 0.5 MPa was selected as the optimal operating condition to balance permeation efficiency and separation selectivity. For rigorous performance benchmarking, the separation results of the composite membrane were systematically compared with those obtained using pristine PES and PES/UiO-66-NH₂ membranes under identical operating conditions.

Effect of time on rejection rate and permeate flux

At a constant transmembrane pressure of 0.5 MPa and a feed oil concentration of 200 mg L⁻¹, the time-dependent separation behaviour clearly demonstrates the superior fouling resistance of the PES/UiO-66-NH₂@g-C₃N₄ composite membrane (Fig. 8). As filtration progressed, the oil rejection increased systematically from 91.0 ± 1.2% at 20 min to 99.5 ± 0.3% after 120 min, indicating the rapid formation of a secondary filtration layer (cake layer) that effectively suppresses oil droplet permeation. In contrast, the blank PES membrane exhibited substantially lower and less stable rejection, increasing only from 62.0 ± 1.5% to 73.5 ± 1.3%, consistent with progressive fouling and deterioration of separation selectivity. Concurrently, the permeate flux of the composite membrane decreased moderately from 436.8 ± 12.4 to 335.2 ± 10.8 L m⁻² h⁻¹ over 120 min due to gradual accumulation of oil droplets and increased hydraulic resistance, whereas pristine PES suffered a sharper decline from 240.0 ± 9.6 to 185.0 ± 8.2 L m⁻² h⁻¹. The comparatively slower flux decay of the composite membrane confirms that the synergistic integration of UiO-66-NH₂ and g-C₃N₄ effectively retards fouling kinetics by enhancing surface hydrophilicity and interfacial stability, thereby sustaining high separation efficiency during prolonged operation^65,66.

Effect of solution dosage on rejection rate and permeate flux

At a constant transmembrane pressure of 0.5 MPa and an oil concentration of 200 mg L⁻¹, the influence of processed feed volume on separation performance was systematically evaluated over a range of 50–350 mL (Fig. 9). The PES/UiO-66-NH₂@g-C₃N₄ membrane exhibited a progressive increase in oil rejection with increasing processed volume, rising from 86.0 ± 1.4% at 50 mL to 99.2 ± 0.4% at 300 mL, followed by a slight decline to 97.2 ± 0.6% at 350 mL. This trend is attributed to the gradual formation and densification of a fouling (cake) layer at the membrane surface, which enhances size-exclusion and contributes to increased separation efficiency at intermediate volumes. The marginal decrease at higher throughput may be associated with local structural heterogeneity or partial destabilization of the deposited layer. In contrast, the pristine PES membrane displayed consistently lower rejection, increasing from 58.5 ± 1.6% to 78.5 ± 1.3%, reflecting its limited resistance to oil permeation⁴⁶.

The permeate flux of the composite membrane increased from 278.0 ± 11.2 to 412.2 ± 14.6 L m⁻² h⁻¹ up to 250 mL, which can be attributed to progressive membrane wetting and activation of effective permeation pathways prior to reaching steady-state conditions. At higher processed volumes, the flux exhibited a slight decline (406.5 ± 13.8 and 384.2 ± 12.5 L m⁻² h⁻¹), consistent with increased hydraulic resistance arising from fouling layer development and partial pore obstruction. By comparison, the pristine PES membrane showed a continuous decrease in flux from 262.0 ± 10.5 to 185.0 ± 9.1 L m⁻² h⁻¹, indicating more severe fouling and limited flux stability. These results confirm that the enhanced performance of the composite membrane is governed by controlled fouling layer formation and improved interfacial stability under increasing volumetric loading⁶⁷.

Effect of oil concentration on rejection rate and permeate flux

The effect of feed oil concentration on separation performance was systematically investigated over a concentration range of 50–350 mg L⁻¹ at a constant transmembrane pressure of 0.5 MPa (Fig. 10). The PES/UiO-66-NH₂@g-C₃N₄ composite membrane exhibited consistently high rejection efficiencies, increasing from 94.5 ± 1.1% at 50 mg L⁻¹ to a maximum of 99.1 ± 0.3% at 200 mg L⁻¹, followed by a marginal decrease to 98.7 ± 0.4% at 350 mg L⁻¹. This behaviour reflects enhanced size-exclusion and interfacial filtration effects at moderate concentrations, with slight attenuation at higher oil loads due to increased pore coverage and fouling layer compaction. In contrast, the blank PES membrane showed markedly lower rejection, rising only from 69.0 ± 1.4% to 76.0 ± 1.2% and subsequently declining at elevated concentrations, indicative of severe fouling and loss of selectivity. Concurrently, the permeate flux of the composite membrane increased from 360.0 ± 12.6 to 435.0 ± 15.2 L m⁻² h⁻¹ up to 250 mg L⁻¹ owing to enhanced driving force, before slightly decreasing to 420.0 ± 14.1 L m⁻² h⁻¹ at 350 mg L⁻¹ due to pore blocking and increased hydraulic resistance associated with higher foulant loading. By comparison, pristine PES exhibited substantially lower and less stable flux values (238.0 ± 9.8–250.0 ± 10.4 L m⁻² h⁻¹), confirming the superior antifouling resilience of the composite membrane under increasing oil loading⁶⁸.

Effect of pump pressure on rejection rate and permeate flux

The influence of pump pressure on separation performance was evaluated over a range of 0.1–0.7 MPa at a fixed oil concentration of 200 mg L⁻¹ and a membrane area of 0.15 m² (Fig. 11). The PES/UiO-66-NH₂@g-C₃N₄ composite membrane exhibited a progressive increase in oil rejection from 88.5 ± 1.3% at 0.1 MPa to 99.4 ± 0.3% at 0.6 MPa, followed by a slight decline to 97.5 ± 0.5% at 0.7 MPa, attributable to pressure-induced pore compaction and reduced effective selectivity at excessive hydraulic loading. In contrast, the blank PES membrane showed substantially lower rejection, increasing only from 60.5 ± 1.6% to 74.6 ± 1.4%, reflecting its higher susceptibility to fouling and pore deformation. Concurrently, the permeate flux of the composite membrane increased from 270.0 ± 11.8 to 392.4 ± 15.6 L m⁻² h⁻¹ up to 0.5 MPa due to enhanced driving force, before slightly decreasing to 375.0 ± 14.2 L m⁻² h⁻¹ at 0.7 MPa as compaction effects became dominant. By comparison, pristine PES exhibited low and weakly pressure-dependent flux (178.0 ± 8.7–170.6 ± 7.9 L m⁻² h⁻¹), confirming the superior pressure tolerance and antifouling robustness of the composite membrane¹².

Photocatalytic self-cleaning performance of PES/UiO-66-NH₂@g-C₃N₄

The visible-light-driven self-cleaning performance of the PES/UiO-66-NH₂@g-C₃N₄ composite membrane was systematically evaluated using methylene blue (MB) and rhodamine B (RhB) as representative organic foulants at initial concentrations of 10, 25, and 50 mg L⁻¹ under near-neutral conditions (pH ≈ 6.5). As shown in (Fig. 12), the composite membrane exhibited pronounced photocatalytic activity with gradual pseudo-first-order-like degradation behaviour under visible-light irradiation. After 120 min of irradiation, MB degradation efficiencies reached 68.5 ± 2.1%, 60.2 ± 1.9%, and 50.8 ± 1.7% for initial dye concentrations of 10, 25, and 50 mg L⁻1, respectively. Comparable trends were observed for RhB, with degradation efficiencies of 65.4 ± 2.0%, 57.1 ± 1.8%, and 48.6 ± 1.6%, respectively. The relatively faster degradation observed during the initial irradiation period followed by a gradual reduction in degradation rate at prolonged irradiation times is consistent with typical heterogeneous photocatalytic behaviour, attributed to progressive depletion of dye molecules and partial occupation of active catalytic sites. This activity is mainly attributed to the visible-light-responsive g-C₃N₄ nanosheets, which generate reactive oxygen species (•OH and O₂•⁻), while UiO-66-NH₂ enhances dye adsorption through its high surface area and amine-functionalized framework. The synergistic adsorption–photocatalysis interaction substantiates the membrane’s effective light-induced self-cleaning capability and antifouling potential during prolonged operation^48,69.

Reusability of PES/UiO-66-NH₂@g-C₃N₄membranes

The reusability and long-term operational stability of the PES/UiO-66-NH₂@g-C₃N₄ composite membrane were systematically evaluated over ten consecutive oil–water separation cycles under optimised operating conditions, using a feed emulsion containing 200 mg L⁻¹ oil at a transmembrane pressure of 0.5 MPa, an operating temperature of 25 ± 1 °C, and a near-neutral pH of 7. Each cycle consisted of a filtration step followed by membrane regeneration through deionised water rinsing and visible-light-driven photocatalytic self-cleaning, exploiting the catalytic activity of the embedded g-C₃N₄ nanosheets. The membrane performance after each cycle was quantified using the flux recovery ratio (FRR), calculated as⁷⁰:

where J_n is the permeate flux after the n-th cleaning cycle, and J₀ is the initial permeate flux before fouling. This parameter is crucial for understanding how effectively the membrane can maintain its separation capacity after each use.

As illustrated in (Fig. 13), the composite membrane exhibited a gradual yet controlled decline in separation performance with increasing cycle number. The oil rejection efficiency decreased moderately from 99.1 ± 0.4% during the first cycle to 92.8 ± 0.6% after the tenth cycle, indicating sustained separation selectivity despite progressive foulant accumulation at the membrane interface. Simultaneously, the FRR decreased from 97.8 ± 1.1% to 91.1 ± 1.3%, suggesting the occurrence of partial irreversible fouling while still demonstrating effective membrane regeneration following each cleaning step. Importantly, no abrupt deterioration in either rejection efficiency or permeation behavior was observed throughout the repeated operational cycles, confirming the structural robustness and interfacial stability of the composite membrane under prolonged filtration conditions. The relatively limited performance decline further indicates that the incorporation of UiO-66-NH₂ and g-C₃N₄ effectively suppresses severe pore blockage and mitigates fouling accumulation through combined hydrophilic and photocatalytic antifouling effects. These findings demonstrate the strong potential of the PES/UiO-66-NH₂@g-C₃N₄ membrane for repeated oily wastewater treatment applications requiring stable long-term operational durability and efficient membrane regeneration⁴⁵.

Long-term continuous filtration performance and operational stability

The long-term continuous filtration performance of the PES/UiO-66-NH₂@g-C₃N₄ composite membrane was evaluated during uninterrupted operation over 12 h under steady-state conditions to assess durability and fouling resistance. During continuous filtration of a 200 mg L⁻¹ oil–water emulsion at 0.5 MPa, the membrane exhibited a gradual and controlled decline in permeate flux, decreasing from 358.4 ± 11.5 L m⁻² h⁻¹ at the initial stage to 321.2 ± 10.3 L m⁻² h⁻¹ after 12 h, corresponding to a flux retention of approximately 90.1%. In parallel, the oil rejection remained high throughout the test, showing a modest reduction from 98.4 ± 0.5% to 93.2 ± 0.7%, indicating sustained separation selectivity despite prolonged exposure to oily feed without intermediate cleaning. The absence of sudden flux collapse or rejection failure suggests effective suppression of fouling accumulation, which can be attributed to the synergistic contribution of enhanced surface hydrophilicity, oleophobicity, and the presence of UiO-66-NH₂and g-C₃N₄ within the membrane matrix. These results confirm stable long-term operation and highlight the membrane’s suitability for continuous oily wastewater treatment under industrially relevant conditions⁴⁹.

Energy consumption and process efficiency analysis

The energy efficiency of the pressure-driven oil–water separation process was assessed through the specific energy demand (SED) calculated under steady-state operating conditions. At a constant transmembrane pressure of 0.5 MPa, the PES/UiO-66-NH2@g-C₃N₄ composite membrane exhibited a relatively low SED, with values decreasing from 0.82 kWh m⁻³ during the initial operation to 0.91 kWh m⁻³ after extended filtration, reflecting the moderate flux attenuation observed under continuous operation. Despite the gradual decline in permeate flux over time, the increase in energy demand remained limited, indicating that fouling-induced hydraulic resistance did not lead to excessive energy penalties. In comparison, the pristine PES membrane required consistently higher energy input (≈ 1.15–1.25 kWh m⁻³) under identical conditions, primarily due to its lower permeate flux and poorer fouling resistance. The reduced energy demand of the composite membrane can be attributed to its enhanced surface hydrophilicity and oleophobicity, which mitigate oil adhesion and pore blockage, thereby maintaining higher permeability at a given pressure⁵⁰. These results demonstrate that the incorporation of UiO-66-NH₂ and g-C₃N₄ not only improves separation performance and stability but also contributes to a more energy-efficient filtration process, reinforcing the practical viability of the composite membrane for long-term oily wastewater treatment.

Wettability-controlled separation and fouling mechanism

The oil–water separation performance and fouling resistance of the PES/UiO-66-NH₂@g-C₃N₄ composite membrane are governed primarily by surface wettability and interfacial transport phenomena. The membrane exhibits enhanced hydrophilicity and oleophobicity due to the incorporation of UiO-66-NH₂, whose –NH₂ functionalities promote hydrogen-bonding interactions with water molecules. This facilitates the formation of a stable hydration layer at the membrane–liquid interface, which lowers interfacial free energy and inhibits oil adhesion. Consequently, water permeation is favored while oil droplets experience repulsive interfacial interactions, leading to effective separation at the early stage of filtration. This wettability-controlled mechanism establishes the initial selectivity of the membrane without invoking adsorption-based interpretations, and is consistent with established models of interfacial energy-driven separation in hydrophilic membranes^71,72.

As filtration progresses, the separation behavior transitions to a regime dominated by fouling layer evolution and transport resistance. The continuous deposition of oil droplets at the membrane surface results in the formation of a dynamic cake layer, which acts as a secondary selective barrier. The progressive densification of this layer enhances rejection through size-exclusion and steric hindrance while simultaneously increasing hydraulic resistance, leading to flux decline. The presence of UiO-66-NH₂ contributes to controlled foulant deposition by promoting surface-level accumulation and limiting deep penetration into membrane pores, thereby mitigating internal pore blocking. Accordingly, the observed increase in rejection with filtration time and feed volume is consistently attributed to cake layer formation, while flux behavior is governed by resistance buildup and partial pore obstruction. This unified interpretation eliminates inconsistencies associated with adsorption-based explanations and aligns with classical membrane fouling models^73,74.

The incorporation of g-C₃N₄ introduces a visible-light-responsive photocatalytic antifouling mechanism associated with the generation of reactive oxygen species (ROS) under irradiation conditions. Upon visible-light exposure, g-C₃N₄ can generate electron–hole pairs that participate in surface redox reactions, producing ROS such as hydroxyl radicals (•OH) and superoxide anions (O₂•⁻). These reactive species contribute to the oxidative degradation of organic foulants accumulated within the cake layer, thereby reducing foulant adhesion propensity and suppressing fouling layer compaction. In the present study, the photocatalytic contribution is interpreted as a synergistic adsorption–photocatalytic effect arising from the combined presence of UiO-66-NH₂ and g-C₃N₄ rather than definitive evidence of a tightly coupled heterojunction-mediated electron-transfer mechanism. Importantly, ROS generation remains localized near photocatalytic domains within the polymer matrix, minimizing undesirable interactions with the UiO-66-NH₂ phase. Furthermore, the –NH₂ groups are structurally anchored within the robust Zr–O framework, contributing to the structural stability of the membrane under the applied operating conditions. Collectively, this photocatalytic process enables partial in situ regeneration of the membrane surface and contributes to sustained antifouling performance^75,76.

Collectively, the membrane performance is governed by a coupled mechanism involving wettability-controlled separation, fouling layer formation, and photocatalytic mitigation of surface foulants. While interfacial interactions assist in regulating oil droplet behavior at the membrane surface, the dominant factors controlling performance are the evolution of the cake layer and the associated increase in hydraulic resistance, which influences flux decline. The integration of UiO-66-NH₂ and g-C₃N₄ enhances surface properties and enables partial self-cleaning, thereby reducing fouling accumulation and stabilizing long-term separation efficiency. This combined mechanism accounts for the improved antifouling performance, sustained permeability, and high oil rejection observed during continuous operation^12,46.

Comparison with existing membranes

To evaluate the performance of the PES/UiO-66-NH₂@g-C₃N₄ composite membrane, a comparative analysis was conducted against various membranes reported in the literature. (Table 2)summarizes the oil rejection efficiencies of different membrane systems, including zeolite@polyethersulfone/cellulose acetate, PES/SiO₂-f-MWCNTs, polyphenylsulfone (PPSU)-TiO₂, and polysulfone/PEG-based membranes. The results indicate that the PES/UiO-66-NH2@g-C₃N₄membrane exhibits the highest oil rejection efficiency (99.2%), surpassing conventional polymeric and ceramic membranes. This superior performance is attributed to the synergistic effects of UiO-66-NH₂ and g-C₃N₄, which enhance hydrophilicity, fouling resistance, and oil-water separation efficiency. The findings highlight the potential of the proposed composite membrane as an advanced material for sustainable wastewater treatment applications.

Limitations and future directions

While the PES/UiO-66-NH2@g-C3N4 composite membrane demonstrated high oil–water separation efficiency, enhanced antifouling behavior, improved thermal stability, and measurable visible-light-driven photocatalytic activity, several limitations should be acknowledged to ensure balanced interpretation of the obtained results. First, the photocatalytic functionality was evaluated using model organic dyes (methylene blue and rhodamine B), which do not fully reproduce the physicochemical complexity and interfacial behavior of real oily foulants. Therefore, the present findings provide evidence of photocatalytic antifouling behavior rather than direct confirmation of oil-specific self-cleaning capability. In addition, direct quantification of photocatalytic oil degradation using TOC analysis, residual oil measurements, or chromatographic identification of degradation intermediates was not performed. Furthermore, although high flux recovery was achieved during cyclic operation, comparative regeneration studies under illuminated and dark conditions were not conducted. Surface characterization before and after fouling/cleaning cycles was also not performed to directly verify foulant removal from the membrane interface.

Another limitation of the present study is that detailed oil droplet-size distribution analysis of the prepared oil–water emulsion was not conducted. Although the emulsion was stabilized using sodium dodecyl sulfate (SDS) under controlled homogenization conditions to ensure reproducible filtration experiments, quantitative droplet-size characterization using techniques such as dynamic light scattering (DLS) or optical microscopy would provide deeper insight into the relationship between emulsion stability, droplet characteristics, and membrane rejection performance. In addition, direct mechanical characterization of the membrane, including tensile strength and elongation analysis, was not performed, although the membrane exhibited stable operation during continuous filtration without observable structural degradation. Furthermore, although PL analysis suggested partial suppression of charge-carrier recombination and possible interfacial interaction between UiO-66-NH2 and g-C3N4, advanced electrochemical and band-structure analyses, including electrochemical impedance spectroscopy (EIS), transient photocurrent measurements, Mott–Schottky analysis, XPS valence-band studies, BET surface area measurements, pore-size distribution analysis, and zeta potential characterization were not conducted. Therefore, definitive confirmation of heterojunction formation, interfacial electron-transfer pathways, and detailed surface-interaction behavior remains beyond the scope of the present study. Accordingly, the mechanistic discussion throughout the manuscript has been carefully moderated to avoid overinterpretation beyond the currently available characterization data.

Future investigations should therefore focus on (i) evaluating photocatalytic regeneration under controlled illuminated and dark conditions, (ii) performing pre- and post-filtration surface characterization to directly confirm foulant removal, (iii) conducting quantitative oil degradation studies using TOC and residual oil analysis, (iv) determining oil droplet-size distribution and emulsion stability characteristics, (v) evaluating membrane mechanical properties through tensile testing, (vi) performing advanced electrochemical, surface, and band-structure analyses to comprehensively clarify the interfacial charge-transfer behavior between UiO-66-NH2 and g-C3N4, including BET surface area analysis, zeta potential characterization, pore-size distribution measurements, EIS, transient photocurrent response, Mott–Schottky analysis, and XPS valence-band investigations, and (vii) validating membrane performance using real industrial oily wastewater matrices. Such investigations would provide a more comprehensive understanding of membrane stability, photocatalytic antifouling behavior, interfacial interactions, and long-term operational performance under realistic treatment conditions.

A multifunctional PES/UiO-66-NH₂@g-C₃N₄ mixed-matrix membrane was successfully developed to overcome key limitations in oily wastewater treatment. The synergistic incorporation of UiO-66-NH₂ and g-C₃N₄ imparted pronounced hydrophilicity and oleophobicity (water contact angle 58 ± 1.2°, oil contact angle 94 ± 1.5°), enabling efficient oil–water separation. Under optimized conditions (0.5 MPa, 25 ± 1 °C), the membrane achieved an oil rejection of 99.5 ± 0.3% with a permeate flux of 345.2 ± 10.8 L m⁻² h⁻¹ after 120 min, alongside suppressed fouling kinetics across time-, dosage-, concentration-, and pressure-dependent tests. Visible-light-driven self-cleaning was confirmed by photocatalytic degradation of 68.5 ± 2.1% methylene blue and 65.4 ± 1.9% rhodamine B within 120 min. Reusability over ten cycles maintained a 91.1 ± 1.3% flux recovery ratio, while oil rejection decreased moderately from 99.1 ± 0.4% to 92.8 ± 0.6%. Long-term continuous filtration over 12 h retained ~ 90% of the initial flux with oil rejection above 93.2 ± 0.7%, and a reduced specific energy demand of 0.82–0.91 kWh m⁻³. Future work should target scale-up, testing with real industrial effluents, solar-driven self-cleaning optimization, and techno-economic assessment to advance practical deployment.