Metal Oxide-Modified PES Membranes for Efficient Separation of Oil-in-Water Emulsions and Trace Organic Compounds

Abstract

The efficient removal of emulsified oil and trace organic pollutants via forward osmosis (FO) technology remains challenging due to limited water flux and membrane fouling. In this study, a series of metal oxide-modified PES-based composite FO membranes were fabricated and systematically evaluated to compare the effects of ZnO, Al₂O₃, and CuO nanoparticles on membrane structure and separation performance. The results demonstrated that the membrane modified with 0.04 g of ZnO nanoparticles achieved optimal synergy in terms of hydrophilicity, surface charge, and pore structure. The pure water flux increased from 5.48 L·m⁻²·h⁻¹ for the pristine membrane to 18.5 L·m⁻²·h⁻¹ for the ZnO-modified membrane, exhibiting a 237.5% increase in pure water flux compared to the pristine PES membrane, an oil rejection rate exceeding 97%, and over 95% rejection of typical negatively charged trace organic pollutants such as ibuprofen and tetracycline. Moreover, the ZnO-modified membrane showed excellent antifouling performance and structural stability in various organic solvent systems. This study not only optimized the interfacial chemistry and microstructure of the FO membrane but also enhanced pollutant repellence and the self-cleaning capability through increased hydrophilicity and surface negative charge density. These findings highlight the significant potential of ZnO modification for enhancing the overall performance of FO membranes and provide an effective strategy for developing high-performance, broadly applicable FO membranes for complex water purification.

1. Introduction

In recent years, with the rapid development of industries such as petrochemicals, transportation, mechanical processing, and pharmaceuticals, large quantities of oily wastewater and trace organic pollutants have been continuously discharged into natural water bodies, becoming a major source of global water pollution [1]. Oils in wastewater exist in the form of free oil, emulsified oil, or dissolved oil, with emulsified oil being particularly difficult to remove using conventional physical methods such as sedimentation or flotation due to its small droplet size and high stability [2,3]. Meanwhile, trace organic pollutants frequently detected in aquatic environments, such as pharmaceuticals and antibiotics like ibuprofen and tetracycline, exhibit high biological activity, environmental persistence, and bioaccumulation potential, posing serious risks to aquatic ecosystems and human health, even at low concentrations [4,5].

Membrane separation technology has been widely applied in the treatment of oily wastewater and organic pollutants due to its high separation efficiency, operational simplicity, and relatively low energy consumption [6]. However, conventional pressure-driven membrane processes, such as ultrafiltration (UF), nanofiltration (NF), and reverse osmosis (RO), rely on external hydraulic pressure to drive water transport, which often leads to high energy consumption, severe membrane fouling, frequent cleaning requirements, and low water recovery rates [7,8]. Forward osmosis (FO), a novel membrane process that utilizes osmotic pressure gradients rather than external pressure as the driving force, has attracted increasing attention due to its low fouling propensity, reduced energy demands, and mild operating conditions (typically operated at ambient temperature, 20 ± 1 °C, and without external hydraulic pressure in FO mode) [9]. Nevertheless, when applied to the treatment of emulsified oily wastewater and trace organic pollutants, FO membranes still face challenges such as limited water flux, susceptibility to fouling, and inadequate mechanical strength. These issues are particularly pronounced under harsh conditions involving high viscosity, high salinity, and complex organic compositions [10,11]. Therefore, the development of high-performance, antifouling functional membranes has become a critical research focus in the field of membrane engineering.

To enhance the overall performance of forward osmosis (FO) membranes, recent studies have focused on structural and interfacial modifications using inorganic nanomaterials [12,13,14,15]. Among these, metal oxide nanoparticles have demonstrated considerable potential in membrane functionalization due to their excellent dispersibility, surface polarity, and chemical stability [16,17,18]. Specifically, zinc oxide (ZnO) nanoparticles, enriched with highly polar Zn-O bonds on their surface, have been widely investigated for their ability to improve membrane hydrophilicity and surface negative charge density, while also imparting antibacterial and anti-oil-fouling properties [19], For instance, Huang et al. [20] successfully fabricated a ZnO nanostructured coating on glass fiber membranes via chemical bath deposition, resulting in a superoleophobic membrane under water with a water flux exceeding 250 L·m⁻²·h⁻¹ and an oil rejection rate greater than 98%. Aluminum oxide (Al₂O₃) is known for its excellent chemical and thermal stability and is often employed as an inorganic reinforcement material to improve membrane mechanical strength and structural integrity. Hatice et al. [21] prepared polysulfone hollow fiber membranes with varying Al₂O₃ contents and demonstrated that the incorporation of Al₂O₃ significantly enhanced separation efficiency for indigo dye wastewater by increasing membrane hydrophilicity and refining pore structure. Copper oxide (CuO), due to its favorable interfacial polarity and surface energy, has shown promise in modulating membrane wettability and improving fouling resistance [22], Homayun et al. [23] developed PVC/CuO nanocomposite membranes with different CuO loadings and reported that the CuO-modified membranes exhibited increased hydrophilicity, porosity, and water flux, along with improved BSA rejection and antifouling performance. Notably, these membranes demonstrated superior separation stability over two filtration cycles compared to unmodified PVC membranes. The incorporation of such metal oxides not only improves membrane surface wettability and pore morphology but also modulates surface charge distribution, enhancing electrostatic repulsion and molecular sieving effects against contaminants. These synergistic enhancements effectively mitigate membrane fouling and significantly improve separation performance.

Although various metal oxide nanoparticles have demonstrated significant advantages in membrane material modification, their performance enhancement is often constrained by factors such as nanoparticle dispersibility, interfacial compatibility with the polymer matrix, and the precision of doping concentration control [24,25,26]. Excessive nanoparticle loading can lead to aggregation, pore blockage, and abnormal increases in surface roughness, ultimately resulting in a decline in membrane performance [27]. Therefore, the rational selection of metal oxide types and doping strategies is crucial for constructing functional membranes with well-organized microstructures, favorable wettability, and stable interfacial properties. The systematic comparison of the regulatory effects of different metal oxides on membrane structure and separation performance not only helps to elucidate their microscopic mechanisms but also provides important theoretical and practical guidance for the design and optimization of advanced membrane materials aimed at improving separation efficiency and antifouling capability.

In this study, polyethersulfone (PES)-based membranes were modified by incorporating metal oxide nanoparticles, and three types of metal oxide-modified composite forward osmosis (FO) membranes—based on ZnO, Al₂O₃, and CuO—were fabricated. A systematic comparison was conducted to evaluate the effects of different metal oxide species on membrane structure and separation performance. The results revealed that the incorporation of an appropriate amount of metal oxide nanoparticles significantly influenced the phase inversion rate and the morphology of the interfacial layer, thereby affecting membrane performance through three primary mechanisms: (1) pore structure regulation: nanoparticles act as nucleation agents during the non-solvent-induced phase separation process, promoting the formation of micropores and increasing the specific surface area; (2) surface hydrophilicity enhancement: metal oxides such as ZnO and Al₂O₃ possess abundant surface hydroxyl groups (–OH) or polar oxide bonds (e.g., Zn–O, Al–O), which facilitate the formation of hydrogen bonding networks, thereby improving hydrophilicity and promoting water transport across the membrane; (3) surface charge modulation: the incorporation of ZnO and CuO led to an increase in the membrane surface negative charge, as evidenced by the enhanced zeta potential, which strengthened the electrostatic repulsion against positively charged solutes and reduced solute back diffusion. Moreover, the modified membranes maintained excellent separation performance and structural stability during long-term operation and multiple separation cycles, indicating that the addition of metal oxides not only improved the initial water flux and rejection rate but also significantly enhanced the antifouling properties and durability of the membranes.

2. Results

2.1. SEM Image Analysis of the Modified FO Membranes

As shown in Figure 1, the unmodified PES membrane exhibits a relatively smooth surface with irregularly distributed pores of varying sizes, lacking additional surface roughness or particulate features. In contrast, the surface of the ZnO nanoparticle-modified membrane displays pronounced roughness, with nanoparticles of diverse sizes and shapes uniformly distributed across the surface. Higher-magnification images reveal fine pores and micro-cracks within the ZnO layer, which may increase the membrane’s specific surface area, enhance hydrophilicity, and thereby improve the rejection of oil-based substances and trace organic pollutants. The Al₂O₃-modified membrane presents a comparatively smoother surface with uniformly distributed nanoparticles and more consistent pore structures. Although the particles are small and evenly dispersed, the overall surface roughness is lower than that of the ZnO-modified membrane. In the case of the CuO-modified membrane, the surface exhibits irregular particle aggregation, with localized agglomeration and depressions, resulting in a non-uniform surface morphology. Such surface defects may adversely affect the membrane’s overall performance, particularly in the treatment of oily wastewater and organic contaminants.

The cross-sectional SEM images reveal the surface morphology and pore structures of different modified membranes. As shown in Figure 2a, the PES-based membrane exhibits a uniform porous structure with relatively consistent pore-size distribution. Smaller pores are observed in the outer layer, while larger pores are present in the inner layer, forming a typical asymmetric membrane structure that offers good mechanical strength and low permeation resistance. The ZnO-modified membrane, Figure 2b, displays increased surface roughness and a more irregular pore structure compared to the pristine membrane. The uniformly dispersed ZnO nanoparticles contribute to an enlarged surface pore area, enhancing the contact interface between the membrane and the feed solution. This, in turn, improves filtration performance. Furthermore, the presence of ZnO induces the formation of a nanoscale interconnected network both on the surface and within the membrane matrix, which effectively enhances water flux and the rejection of organic pollutants. As shown in Figure 2d, the cross-section of the CuO-modified membrane reveals a less uniform pore structure with localized nanoparticle aggregation. The irregular distribution of pores and uneven particle dispersion lead to a non-uniform membrane morphology, which is inferior to that of the ZnO- and Al₂O₃-modified membranes. This structural heterogeneity may adversely impact the overall separation performance of the CuO-modified membrane.

2.2. FTIR Analysis of Modified FO Membranes

Figure 3 clearly illustrates the changes in key absorption peaks, enabling the evaluation of nanoparticle distribution and uniformity in the interfacial polymerization layer through the analysis of absorption peak areas associated with various functional groups. As shown in Figure 3a, a broad absorption band appears in the range of 3600–3200 cm⁻¹ [28], corresponding to hydroxyl (–OH) groups and potential hydrogen bonding interactions, which are attributed to surface hydrophilicity or the presence of hydroxyl functionalities within the membrane matrix. The peak at 1300 cm⁻¹ is assigned to the symmetric vibrations of aromatic rings [29], while the peak at 1150 cm⁻¹ is related to the stretching vibration of sulfonic acid (S=O) groups [30]. The peak at 1700 cm⁻¹ is attributed to the C=O stretching vibration, which most likely originates from residual solvent (NMP) or polyvinylpyrrolidone (PVP) present in the membrane matrix, as both compounds contain carbonyl groups that absorb in this region [31]. These components contribute to the mechanical strength and chemical stability of the membrane, but the overall hydrophilicity remains limited, which restricts the membrane’s effectiveness in oil–water separation and the removal of trace organic pollutants. In Figure 3b, the ZnO-modified membrane shows characteristic Zn–O stretching vibrations at 420 cm⁻¹ and 530 cm⁻¹ [32], along with an O–Zn–O bending vibration peak at 1380 cm⁻¹ [33], indicating that ZnO nanoparticles have been effectively incorporated into the membrane matrix and have chemically interacted with the base membrane. This interaction significantly improves the hydrophilicity of the membrane and its adsorption affinity for water-soluble contaminants. The Al₂O₃-modified membrane exhibits Al–O stretching vibration peaks at 500 cm⁻¹ and 600 cm⁻¹ [34], confirming the successful incorporation of Al₂O₃ and an improvement in the membrane’s chemical stability. For the CuO-modified membrane, Cu–O stretching vibrations are observed at 470 cm⁻¹ and 600 cm⁻¹, while a Cu–O–C bending vibration appears at 800 cm⁻¹ [35], indicating the interaction between CuO nanoparticles and surface functional groups of the membrane. These results confirm the successful incorporation of CuO nanoparticles onto the polyamide surface, leading to enhanced chemical reactivity and improved structural stability of the membrane.

2.3. Roughness Analysis of Modified FO Membranes

A detailed atomic force microscopy (AFM) analysis was conducted to investigate the surface characteristics of the composite forward osmosis (FO) membranes, with a focus on roughness variations induced by the incorporation of metal oxide nanoparticles and their potential impact on membrane performance [36]. The AFM images shown in Figure 4a–d reveal the surface microstructures of four different membranes. These complex surface morphologies can influence water-molecule adsorption and diffusion pathways, thereby indirectly affecting membrane permeability, rejection efficiency, and antifouling properties. Within a scanning area of 10 μm × 10 μm, surface roughness parameters—including arithmetic average roughness (Ra), root mean square roughness (Rq), and maximum height roughness (Rz)—were quantified, as summarized in

Table 1. Surface roughness parameters and structural characteristics of the modified membranes.

Membrane Material	Ra (±SD)/nm	Rg (±SD)/nm	Rz (±SD)/nm	Surface Morphology
PES	43.6 ($\pm 0.4$)	56.2 ($\pm 0.6$)	346 ($\pm 3.1$)	Ridge-valley Type
ZnO-PES	83.7 ($\pm 0.5$)	106 ($\pm 0.8$)	733 ($\pm 3.5$)	Ridge-valley Type
Al2O3-PES	52.7 ($\pm 0.8$)	71 ($\pm 0.9$)	532 ($\pm 2.8$)	Ridge-valley Type
CuO-PES	28.0 ($\pm 0.6$)	37.8 ($\pm 1.1$)	426 ($\pm 2.6$)	Ridge-valley Type

.

Figure 4a shows a relatively smooth surface on the pristine PES membrane, with uniformly distributed pores of consistent size. However, the low surface roughness and lack of additional microstructural features result in a limited contact area between the membrane and fluid, thereby restricting its filtration efficiency. In contrast, the ZnO-modified membrane in Figure 4b exhibits significantly increased surface roughness, with a maximum height variation reaching 383.5 nm. Prominent surface undulations and nanoparticle distributions are observed, leading to a more complex and irregular structure. Such surface features likely contribute to a larger effective surface area and improved hydrophilicity, enhancing the membrane’s ability to retain oil droplets and organic pollutants. As shown in Figure 4c, the Al₂O₃-modified membrane displays a relatively smooth morphology with minor height fluctuations and uniform pore distribution. Although its surface roughness remains lower than that of the ZnO-modified membrane, it still shows a moderate improvement compared to the pristine PES membrane. This structure may be more suitable for applications requiring stable filtration performance, although its efficiency may be somewhat limited when dealing with oily contaminants. Figure 4d illustrates the CuO-modified membrane, which features a more irregular surface morphology with evident nanoparticle agglomeration and pronounced height variations. These characteristics lead to non-uniform pore structures and may adversely affect long-term membrane stability. In particular, when treating high concentrations of oil and organic matter, the filtration performance of the CuO-modified membrane is significantly reduced.

2.4. Zeta Potential Analysis of Modified FO Membranes

As shown in

Table 2. Zeta potential of modified membrane surfaces.

Membrane Type	Zeta Potentials (±SD)/mV
PES-based Membrane	−12.9 ($\pm 0.1$)
ZnO-PES Membrane	−15.7 ($\pm 0.1$)
Al2O3-PES Membrane	−10.1 ($\pm 0.1$)
CuO-PES Membrane	−14.9 ($\pm 1.1$)

, the zeta potential measurements of the different modified membranes reveal changes in surface-charge characteristics. The pristine PES membrane exhibits a zeta potential of −12.9 mV, indicating a slightly negative surface charge, which is likely attributed to the presence of sulfonic groups (S=O) in the membrane matrix that possess strong electronegativity. This suggests that, while the PES membrane surface is relatively stable, its low hydrophilicity results in weak interactions with water molecules. The zeta potential of the ZnO-modified membrane increases significantly to −15.7 mV. The incorporation of ZnO nanoparticles enhances the negative surface charge due to their polar nature and the ability to form hydrogen bonds with water molecules, thereby improving the membrane’s hydrophilicity. This increased surface negativity not only improves membrane stability but also contributes to enhanced antifouling performance, particularly by promoting electrostatic repulsion against negatively charged contaminants. For the Al₂O₃-modified membrane, the zeta potential decreases to −10.1 mV, slightly less negative than that of the pristine PES membrane. Given the relatively high isoelectric point of Al₂O₃, its surface tends to be positively charged under neutral or slightly acidic conditions, thus the introduction of Al₂O₃ nanoparticles reduces the overall negative charge on the membrane surface. The CuO-modified membrane shows a zeta potential of −14.9 mV. CuO surfaces tend to acquire negative charges in aqueous environments, and their incorporation increases the surface negativity of the membrane. However, due to the potential leaching of Cu²⁺ ions and the associated ionic aggregation, the enhancement in surface charge is less pronounced compared to that achieved with ZnO modification.

2.5. Wettability of Modified Membranes

Figure 5 presents the water contact angles of different membrane materials, which were used to assess their surface wettability. The pristine PES membrane exhibits the highest water contact angle, indicating a relatively hydrophobic surface. Water droplets form large contact angles on the membrane surface due to the lack of polar functional groups on PES, resulting in weak interactions between water molecules and the membrane, and thus poor hydrophilicity. In contrast, the water contact angle of the ZnO-modified membrane significantly decreases to 66°, demonstrating a substantial enhancement in hydrophilicity. This improvement is attributed to the strong polarity of ZnO nanoparticles, which can form hydrogen bonds with water molecules and which can increase the membrane’s affinity toward water. Additionally, the incorporation of ZnO increases surface roughness, promoting a favorable Cassie–Baxter wetting state that further reduces the contact angle. The Al₂O₃-modified membrane exhibits a water contact angle of approximately 80°, indicating moderate improvement in wettability compared to the pristine PES membrane, though less pronounced than that of the ZnO-modified membrane. This can be attributed to the presence of polar O–Al–O bonds on the Al₂O₃ surface, which contribute to hydrophilicity. However, the interaction between Al₂O₃ and water is weaker than that of ZnO, resulting in a smaller reduction in contact angle. The CuO-modified membrane shows a slightly lower water contact angle than the pristine membrane, suggesting a marginal improvement in hydrophilicity. Nevertheless, the effect is less significant, likely due to uneven particle distribution and suboptimal surface roughness, which limit the enhancement of water affinity.

2.6. Effect of Metal Oxide Content on Membrane Separation Performance

As shown in Figure 6a, with increasing ZnO content, the water flux of the membrane reaches a peak value of 19 L·m⁻²·h⁻¹ at a loading of 0.04 g, while the reverse solute flux drops to its minimum. This indicates that the ZnO nanoparticles are uniformly dispersed within the membrane, resulting in an optimized structure. However, when the loading exceeds 0.06 g, particle agglomeration and pore-blocking effects become prominent, leading to a decline in water flux and an increase in reverse solute flux. In Figure 6b, the Al₂O₃-modified membrane exhibits a broader tolerance to nanoparticle loading. Even at a content of 0.08 g, the membrane maintains relatively high water flux and low reverse solute flux, suggesting that Al₂O₃ has a milder effect on the structural regulation of the membrane. In contrast, the CuO-modified membranes shown in Figure 6c demonstrate limited performance enhancement at loadings of 0.02–0.04 g. As the CuO content increases, the water flux decreases significantly and the reverse solute flux rises sharply. This is mainly due to the strong tendency of CuO nanoparticles to aggregate, resulting in membrane densification and reduced pore connectivity. Separation selectivity, defined as the ratio of reverse solute flux to water flux ($J_S$/$J_W$), is an important parameter for evaluating FO membrane separation performance and structural integrity. A lower $J_S$/$J_W$ value indicates higher separation efficiency at a given water flux. Figure 6d provides a more intuitive comparison of selectivity across different concentrations. The optimal selectivity is achieved at nanoparticle loadings of 0.04 g ZnO, 0.08 g Al₂O₃, and 0.04 g CuO, respectively, indicating the most favorable balance between water permeability and solute rejection under these conditions.

2.7. Rejection Performance of Modified Membranes for Trace Organic Compounds

Figure 7 illustrates the water flux and rejection rates of various modified membranes when treating ibuprofen (Figure 7a) and tetracycline (Figure 7b), highlighting the effects of metal oxide modification on the separation performance for trace organic compounds. Compared to the pristine PES membrane, all three metal oxide-modified membranes exhibit simultaneous improvements in both water flux and rejection. Among them, the ZnO-modified membrane demonstrates the best overall performance, achieving a water flux of approximately 18.5 L·m⁻²·h⁻¹ and a rejection rate exceeding 95%. This enhancement is primarily attributed to the excellent dispersibility of ZnO nanoparticles and the presence of highly polar Zn-O bonds, which improve the hydrophilicity and surface negative charge of the membrane. These characteristics contribute to the formation of well-organized hydrophilic channels and promote strong electrostatic repulsion against the negatively charged ibuprofen and tetracycline molecules, thereby facilitating water transport while effectively inhibiting solute permeation. The Al₂O₃-modified membrane also exhibits improved separation performance due to the incorporation of polar Al-O groups, which enhance surface hydrophilicity and lead to increased water flux and rejection. However, its electrostatic repulsion effect is weaker than that of ZnO due to its lower electronegativity. In contrast, the CuO-modified membrane performs reasonably well in ibuprofen removal but shows reduced rejection efficiency for tetracycline. This is likely due to nanoparticle aggregation and the resulting non-uniform pore structure, which limits the membrane’s ability to intercept larger, more complex organic molecules. In summary, ZnO-modified membranes achieve superior separation performance in the removal of both model trace organic pollutants by simultaneously optimizing surface chemistry and pore structure, demonstrating strong potential for the development of high-efficiency forward osmosis membranes.

2.8. Oil–Water Separation Performance and Antifouling Properties of Modified Membranes

As shown in Figure 8a, the ZnO-modified membrane exhibits the best performance among all samples, achieving a water flux of 17 L·m⁻²·h⁻¹ and an oil rejection rate exceeding 97%, significantly outperforming both the Al₂O₃- and CuO-modified membranes as well as the unmodified PES membrane. This superior performance can be attributed to the structural and interfacial regulation imparted by ZnO nanoparticles. On one hand, the ZnO surface is rich in polar Zn-O bonds and hydroxyl functional groups, which markedly enhance the hydrophilicity of the membrane. This allows for the rapid formation of a stable hydration layer on the membrane surface under water, effectively preventing oil droplets from directly contacting the membrane and thus minimizing fouling. On the other hand, ZnO also increases the surface negative potential of the membrane, reinforcing electrostatic repulsion against weakly negatively charged oil droplets and further reducing their adsorption and adhesion to the surface. In addition, the uniform dispersion of ZnO nanoparticles within the membrane matrix contributes to the formation of a honeycomb-like nanostructured channel network, which not only facilitates rapid water transport but also provides spatial sieving effects for the oil phase. This synergistic mechanism enables the membrane to achieve both high flux and high separation selectivity.

Figure 8b shows that the ZnO-PES membrane maintains an oil rejection rate above 95% in various organic solvent environments (including n-hexane, toluene, and petroleum ether), demonstrating excellent universality and interfacial stability. The superior antifouling performance of the membrane is further validated in Figure 8c,d. After continuous operation for 120 min and over 10 separation cycles, only slight declines in water flux and rejection rate were observed, reflecting outstanding operational stability. Between each separation cycle in Figure 8d, the membrane was cleaned by flushing with deionized water for 10 min, without the use of any chemical cleaning agents. This physical cleaning step effectively removed residual oil droplets and foulants, ensuring the consistency of each cycle test. This antifouling behavior is attributed not only to the enhanced surface wettability and electrostatic repulsion, but also to the “low-adhesion energy surface” provided by ZnO nanoparticles. This surface property allows oil droplets to be easily removed by water flow even after brief contact, thereby reducing fouling accumulation. In addition, ZnO improves the mechanical strength and interfacial adhesion of the polyamide selective layer, helping to prevent pore collapse or delamination during long-term operation. These synergistic effects together ensure the reliable performance and excellent durability of the membrane in complex oil–water separation environments.

3. Materials and Methods

3.1. Materials

Polyvinylpyrrolidone (PVP10), sodium hydroxide, toluene, petroleum ether, n-hexadecane, n-hexane, hydrochloric acid, ibuprofen, tetracycline, sodium chloride, nano zinc oxide, nano aluminum oxide, and nano copper oxide were purchased from Shanghai Macklin Biochemical Co., Ltd. (Shanghai, China). Polyethersulfone (PES) was obtained from Shanghai Yuan Ye Biological Technology Co., Ltd. (Shanghai, China). N-methylpyrrolidone (NMP) was supplied by Sinopharm Chemical Reagent Co., Ltd. (Shanghai, China). Tween80 was purchased from Shanghai Yongchuan Biotechnology Co., Ltd. (Shanghai, China).

3.2. Preparation of Modified Membranes

A specified amount of PES and PVP was added to an appropriate volume of NMP (as listed in

Table 3. Composition of casting solutions for different membranes.

Membrane Type	PES (g)	PVP (g)	NMP (g)	ZnO (g)	Al2O3 (g)	CuO (g)
PES	18	4	78
ZnO-PES	18	4	78	0.02
				0.04
				0.06
				0.08
				0.1
Al2O3-PES	18	4	78		0.02
					0.04
					0.06
					0.08
					0.1
CuO-PES	18	4	78			0.02
						0.04
						0.06
						0.08
						0.1

) and was stirred at 70 °C for 6 h until a uniform and transparent polymer solution was obtained. Subsequently, metal oxide nanoparticles were added, and the solution was further stirred for 4 h, followed by probe-type ultrasonication at a power of 300 W for 60 min to ensure the uniform dispersion of nanoparticles in the casting solution. The resulting solution was allowed to stand overnight at room temperature to remove any entrapped air bubbles. An appropriate amount of the degassed casting solution was then poured onto a clean glass plate and cast into a film with a thickness of 15 μm using a casting knife. The freshly cast film was immediately immersed in deionized water to initiate phase inversion through solvent–nonsolvent exchange, thereby forming a porous membrane structure. After complete phase separation, the membrane detached spontaneously from the glass surface. The formed membrane was then removed from the water bath, thoroughly rinsed several times with deionized water to remove residual solvents, and stored in deionized water for further use.

3.3. Preparation of Emulsified Oil

A total of 18 mL of n-hexadecane and 980 mL of deionized water was added to a beaker, followed by the addition of 2 mL of Tween 80 as a surfactant. The mixture was then ultrasonicated for 60 min to obtain a stable n-hexadecane oil-in-water emulsion, which remained stable for several hours at room temperature. The preparation methods for other emulsions, including toluene, n-hexane, and petroleum ether, were the same as described above.

3.4. Characterization

Scanning electron microscopy (SEM; Regulus 8600, Hitachi High-Technologies Corporation, Tokyo, Japan) was used to characterize the pore structure and distribution of the membranes. The SEM system was equipped with a Schottky field emission electron gun operated at an accelerating voltage of 5.00 kV. Fourier transform infrared spectroscopy (FTIR; Nicolet iS50 FT-IR, Thermo Fisher Scientific, Waltham, MA, USA) was employed for the qualitative and quantitative analysis of functional groups and chemical bonds present in the membrane samples. Atomic force microscopy (AFM; Dimension Edge, Bruker Corporation, Billerica, MA, USA) was used to observe the membrane surface morphology, structure, and physical features at the nanoscale. The wettability of the membrane surface was assessed using a contact angle goniometer (Theta Flex, Biolin Scientific, Gothenburg, Sweden), and membrane hydrophilicity was quantified by measuring the water contact angle (WCA).

3.5. Performance Testing

A bench-scale cross-flow forward osmosis (FO) system was employed to evaluate membrane performance, with the membrane fixed in the center of the testing cell, providing an effective area of 30 cm², as illustrated in Figure 9. All of the tests were conducted in FO mode with the active layer facing the feed solution. Both sides of the membrane were circulated at a constant flow rate of 250 mL/min (corresponding to a linear velocity of approximately 0.17 m s⁻¹). Water flux was determined in real time by measuring the weight change of the draw solution using a digital balance, while the conductivity of the feed solution was monitored using a conductivity probe linked to a computer. Unless otherwise specified, a 2 mol L⁻¹ NaCl solution was used as the draw solution, and deionized water was used as the feed. For trace organic compound rejection tests (e.g., ibuprofen and tetracycline), the feed solution contained 20 mg L⁻¹ of the target organic compound and 0.5 g L⁻¹ of NaCl to simulate typical background salinity, with the pH adjusted to 7.0 ± 0.1 using 0.1 mol L⁻¹ HCl or NaOH. All of the experiments were performed at an ambient temperature of 20 ± 1 °C. The pure water flux ($J_W$) was calculated using Equation (1):

$$J_W={\displaystyle \frac{\Delta m}{\rho \Delta tA}}$$

(1)

In the equation, $∆m$ is the mass increase of the draw solution within time $∆t$ in grams (g); $∆t$is the time, in hours (h); ρ is the density of water in grams per cubic centimeter (g·cm⁻³); and A is the effective membrane area in square meters (m²).

The rejection rate R can be calculated using Equations (2) and (3):

$$R=1-{\displaystyle \frac{{\rho }_P}{{\rho }_f}}$$

(2)

$${\rho }_P={\displaystyle \frac{V_t\ast C_t-V_0\ast C_0}{V_t-V_0}}$$

(3)

In the equation, ${\rho }_P$and ${\rho }_f$ refer to the solute concentrations in the draw solution and feed solution, respectively. $C_0$ and $C_t$ denote the solute concentrations in the draw solution at the initial time and after a duration of t, respectively. $V_0$ and $V_t$ represent the volumes of the draw solution at the initial time and after a duration of t, respectively. The concentration of emulsified oil was measured using an ultraviolet-visible spectrophotometer (Jinghua, 722N, Shanghai, China).

The reverse solute flux $J_S$ is calculated using Equation (4):

$$J_S={\displaystyle \frac{\Delta (C_t\ast V_t)}{\Delta t\ast A}}$$

(4)

In the equation, $\Delta \left(C_t*V_t\right)$represents the mass difference of salts in the feed solution over time; $\Delta t$ is calculated through the conductivity of the feed in grams (g); A is the effective area of the membrane in square meters (m²); and $\Delta t$ is the time in hours (h).

4. Conclusions

In this study, a series of metal oxide-modified PES-based composite forward osmosis (FO) membranes were successfully fabricated and systematically evaluated for their overall performance in oil–water separation and trace organic pollutant removal. The incorporation of ZnO, Al₂O₃, and CuO nanoparticles into the casting solution effectively regulated the membranes’ hydrophilicity, surface charge density, and pore structure. Among the modified membranes, the ZnO-modified membrane with a doping amount of 0.04 g exhibited the most outstanding performance, achieving a water flux of 18.5 L·m⁻²·h⁻¹, an oil rejection rate above 97%, and rejection rates exceeding 95% for trace organic pollutants such as ibuprofen and tetracycline. These performance enhancements were primarily attributed to the ZnO-induced significantly hydrophilic interface, underwater superoleophobic surface, and enhanced electrostatic repulsion from the negatively charged membrane surface, which effectively suppressed contaminant adsorption and pore blockage. Additionally, the ZnO-modified membrane maintained stable and efficient separation in various organic phases (e.g., petroleum ether and n-hexane), and demonstrated excellent antifouling performance and interfacial structural stability during 120 min of continuous operation and over 10 separation cycles. The ZnO-PES composite membrane developed in this study not only shows high efficiency for oil–water separation but also exhibits significant potential for the removal of trace organic pollutants, offering an effective strategy and technical foundation for the development of durable and broadly applicable FO membranes.