Improving ofloxacin removal and antifouling performances of polysulfone-based mixed matrix membranes using chitosan-grafted UiO-66-NH₂ MOFs

Abstract

Polymeric membranes are widely used these days for wastewater treatment applications; however, in pressure-driven filtration processes, their intrinsic hydrophobicity often leads to severe fouling and reduced separation efficiency. In this study, a novel chitosan (CTS)-grafted metal-organic framework (UiO-66-CTS MOF) was synthesized and used as a stable filler in a polysulfone (PSF) matrix to enhance the hydrophilicity of resulting mixed matrix membranes (MMMs) and prevent nanoparticle agglomeration. CTS was grafted onto UiO-66-NH₂ via a Schiff-base reaction, utilizing glutaraldehyde (GA) as a linker due to the presence of amino groups in both components. The successful synthesis of MOFs was confirmed by various characterization techniques, with the results demonstrating their effective formation. Separation performances of the prepared membranes were evaluated by pure water flux, pollutant rejection, and antifouling properties. Results revealed that UC-1.25 (containing 1.25% UiO-66-CTS) exhibited the best performance in the evaluation compared with pure PSF and U-1 (containing 1% UiO-66-NH₂) membranes. UC-1.25 showed a significantly enhanced pure water flux of 2.92 L/m².h, compared to 0.39 L/m².h for the pristine PSF and 2.27 L/m².h for U-1. In terms of pollutant rejection, UC-1.25 achieved 94.95% rejection of ofloxacin (OFL) antibiotics, outperforming PSF (71.33%) and U-1 (91.99%). Furthermore, UC-1.25 demonstrated improved antifouling behavior, with a flux recovery ratio (FRR) of 95.14%, whereas PSF and U-1 exhibited FRRs of 61.72% and 91.74%, respectively. These findings confirm that grafting CTS onto the UiO-66-NH₂ MOFs effectively enhances the membrane’s hydrophilicity and separation performance.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-026-46405-0.

Introduction

Water scarcity is a significant issue today, primarily caused by population growth and industrial expansion. Alongside these challenges, the increasing use of antibiotics and their improper disposal have led to their presence in water sources, which raises concerns for both the environment and public health¹. Of these pharmaceutical pollutants, Ofloxacin (OFL)—a widely used antibiotic—is especially persistent and may play a role in the spread of antibiotic resistance. To address this issue, water treatment is an effective solution, and several methods have been employed to purify water for reuse². Among these, membrane separation is distinguished by its superior energy efficiency, simple operation, high design flexibility, space saving, and environmental friendliness^3,4. At present, various membrane types for multiple applications are in progress, including ultrafiltration (UF), nanofiltration (NF), and reverse osmosis (RO) membranes. The selection of these membranes primarily depends on the required degree of filtration and the specific characteristics of the treated water⁵. Moreover, most membranes are made of hydrophobic polymers, resulting in fouling and a reduction in membrane lifespan^6,7.

Mixed matrix membranes (MMMs) are an innovative membrane technology that overcomes the limitations of conventional polymeric membranes⁸. Incorporating fillers into polymer matrices and blending them allows for the utilization of ordered open channels, high porosity, adjustable chemical environments, and enhanced antifouling properties to attain greater separation efficiency and selectivity⁹. Consequently, MMM fabrication has arisen as a viable solution to the issues associated with typical membranes. Diverse nanomaterials, including metal oxides such as zinc oxide (ZnO), titania (TiO₂), iron oxides (Fe₂O₃, Fe₃O₄), zeolite, silica, carbon nanotubes, graphene, graphene oxide (GO), metal–organic frameworks (MOFs), covalent organic frameworks (COFs), etc., have been utilized as fillers in the preparation of MMMs¹⁰. Recently, MOFs have attracted significant attention and have been widely employed in diverse applications, including catalysis, degradation, energy storage, adsorption, and membrane separation¹¹. MOFs are hybrid organic-inorganic crystalline materials characterized by a three-dimensional structure formed by inorganic metal ions or clusters coordinated with organic linkers. Moreover, MOFs have various intrinsic properties, such as extensive surface area, significant porosity, adjustable pore dimensions, designable structure, functionalizability, and favorable compatibility with polymers¹². Consequently, MOF materials have been proposed as alternative nanofillers to enhance membrane separation efficiency^13,14.

Various strategies have been employed to enhance the performance of MOFs as fillers in MMMs, including functionalization through post-synthetic modification, blending with other nanoparticles, in situ growth within the polymer matrix, etc.^15,16. For example, UiO-66 particles activated through Soxhlet extraction were integrated into a polysulfone (PSF) matrix to fabricate MMMs for the removal of both anionic (methyl orange (MO) and methyl red (MR)) and cationic (methylene blue (MB) and malachite green (MG)) dyes. The resulting membranes exhibited enhanced hydrophilicity, antifouling performance, and a dramatic increase in water flux—up to 21,495% compared to the pure PSF membrane⁷. In a study focused on enhancing the performance of polyethersulfone (PES)-based MMM, both MXene and MOF were simultaneously integrated as fillers. The membrane modified with 0.25 wt% UiO-66-NH₂@MXene showed high rejection rates (up to 91%) for various dyes, including MB, MO, crystal violet (CV), methyl blue (MB), acid fuchsin (AF), and Congo red (CR), as well as for salts such as Na₂SO₄ (91%), MgSO₄ (87%), MgCl₂ (79%), and NaCl (62%). It also exhibited markedly enhanced antifouling properties and hydrophilicity¹⁷.

To enhance the compatibility of MOF and polymeric membrane to prevent agglomeration, UiO-66-NH₂ nanoparticles were synthesized via a polyvinylpyrrolidone (PVP)-assisted in situ method, subsequently combined with polyvinyl alcohol (PVA), and integrated into the membrane. The resulting membrane exhibited excellent rejection rates (99.89% to 100%) for dyes such as CR, MB, reactive black 5 (RB5), and direct red 23 (DR23), along with improved antifouling performance and remarkable stability over 50 h¹⁸. In another study, post-modification methods have been utilized, and a melamine-modified UiO-66 MOF was incorporated into PES matrix, resulting in enhanced water permeability (50 L/m².h), high Bovine serum albumin (BSA) rejection (97.8%), and excellent antifouling performance with a 87% fouling recovery ratio¹⁹.

Inspired by these, in the present work we grafted chitosan (CTS) onto UiO-66-NH₂ by a Schiff-base reaction to improve the compatibility between the MOF and PSF membrane, which avoids agglomeration and facilitates superior dispersion of nanoparticles inside the polymeric matrix. The prepared membranes were designed as dense single-layer PSF-MMMs operating under pressure-driven filtration conditions. Furthermore, grafting CTS may enhance the membrane’s hydrophilicity and enhance its rejection and antifouling properties. Importantly, CTS grafting was designed as a multifunctional modification strategy, aiming not only to improve hydrophilicity, but also to enhance MOF–polymer interfacial compatibility, suppress nanoparticle agglomeration, and improve membrane stability during long-term operation. MMMs with different loaded fillers (UiO-66-NH₂ and UiO-66-CTS) were fabricated, and their effectiveness was evaluated to identify the optimal MMM. After identifying the optimal filler content for both UiO-66-NH₂ and UiO-66-CTS, the pure PSF membrane, the UiO-66-NH₂-based MMM, and the UiO-66-CTS-based MMM were compared in terms of performance and further characterized to evaluate the effect of CTS grafting on membrane properties. To assess the rejection performance, OFL, a pharmaceutical contaminant, was utilized as the model pollutant, while BSA was employed to examine the antifouling properties of the prepared membranes.

Experimental section

Materials and chemicals

Polysulfone (PSF) with a molecular weight of 65,000 g/mol was used for membrane fabrication. Zirconium(IV) chloride (ZrCl₄, 99.9%), 2-aminoterephthalic acid (BDC-NH₂, 99%), low molecular weight chitosan (CTS), glutaraldehyde (GA, 50%), dimethylformamide (DMF, 99%), anhydrous methanol (≥ 99.8%), acetone, acetic acid, and hydrochloric acid (HCl, 37%) were used to synthesize both MOFs that were all purchased from Merck and Sigma-Aldrich. Bovine serum albumin (BSA) was used for assessing antifouling properties and acquired from Sigma-Aldrich. Ofloxacin (OFL), as the contaminant, was acquired from Bio Basic.

Synthesis of functionalized MOFs

The pristine UiO-66-NH₂ has been synthesized based on our previous work (provided in the Supplementary Information)². To synthesize the functionalized MOF, 200 mg of synthesized UiO-66-NH₂ was added to 100 mL of anhydrous methanol containing 5% GA and heated at 45 °C for 2 h for the Schiff-base reaction. The resulting UiO-66-GA powders were washed three times with methanol²⁰. Subsequently, 40 mg of CTS was dissolved in 40 mL of 1% acetic acid solution, and the UiO-66-GA was then added to the CTS solution. After one hour of reaction at room temperature, the solution was diluted, washed three times with 1% acetic acid solution and acetone, and then dried at 50 °C to form functionalized MOF that named UiO-66-CTS. Scheme 1 illustrates a simplified synthesis pathway of UiO-66-CTS.

Scheme 1.

Simplified representation of the synthesis of UiO-66-CTS.

Membrane fabrication

The non-solvent-induced phase inversion method was utilized for the manufacturing of MMMs. In order to achieve a uniform dispersion, a specific amount of nanoparticles (UiO-66-NH₂ or UiO-66-CTS) was fully dispersed in DMF as a solvent. Subsequently, the measured amount of pre-formed PSF solution (27 wt%) was gradually introduced to the aforementioned solution to improve interactions between the nanoparticles and PSF chains. The mixture was stirred in a water bath at 70 °C for 18 h to ensure proper dispersion of MOF nanoparticles and achieve a homogeneous cast film solution. After that, the casting solution was finally placed in an oven at 60 °C for 12 h to allow degassing and eliminate any remaining air bubbles. The solution was then spread onto a porous non-woven polyester support using a hand-casting knife with a thickness of 200 μm. To initiate the phase inversion process, the film was instantly immersed in a non-solvent (deionized water) bath at ambient temperature. To avoid solvent saturation, the non-solvent bath was changed after 24 h¹³. The membrane components and sample codes are all listed in Table 1.

Table 1.

The codes of prepared membranes.

Membrane codes	PSF (wt%)	UiO-66-NH2 (wt%)	UiO-66-CTS (wt%)	DMF (wt%)
PSF	27	0	0	73
U-0.25	27	0.25	0	72.75
U-0.5	27	0.5	0	72.5
U-1	27	1	0	72
U-1.25	27	1.25	0	71.75
U-1.25	27	1.5	0	71.5
UC-1	27	0	1	72
UC-1.25	27	0	1.25	71.75
UC-1.5	27	0	1.5	71.5

Characterization methods

Characterization of MOF nanoparticles

Using an X-pert pro diffractometer, an X-ray diffraction (XRD) study was conducted at room temperature to examine the crystalline nature of particles. The X-rays were generated using an accelerating voltage of 40 kV and a current of 40 mA, and the diffractometer was equipped with a copper (Cu) sealed tube. A 2θ range of 5–80° was covered by a step width of 0.05°. Functional groups of nanoparticles were analyzed based on Fourier transform infrared (FTIR) spectroscopy using a BRUKER VECTOR instrument, covering a spectral range of 4000–500 cm^− 1. The N₂ adsorption and desorption isotherms and pore sizes of MOFs were collected using BELSORP-MAX equipment (BELMAX, Japan) at a temperature of 77 K, and samples were degassed at 422 K for a duration of 300 min. Thermogravimetric analysis (TGA, NETZSCH STA 409 PC/PG) was employed to evaluate the thermal stability of the synthesized MOFs from room temperature to 700 °C at a heating rate of 10 °C min^− 1 in the presence of nitrogen. Field emission scanning electron microscopy (FESEM) images of nanoparticles were obtained with the use of a FESEM-MIRA III, TESCAN apparatus with an accelerating voltage of 15 kV. Transmission electron microscopy (TEM) images were captured using a ZEISS EM10C instrument operated at 100 kV (Germany).

Characterization of prepared membranes

Thin films and surface features were analyzed by grazing incidence X-ray diffraction (GIXRD) with Cu Kα radiation utilizing a PANalytical X’Pert PRO MPD diffractometer. To determine the membrane’s functional group, attenuated total reflectance–infrared (ATR-IR) spectroscopy was used (Perkin Elmer), and 400–4000 cm^− 1 spectra were obtained. Atomic force microscopy (AFM), Ara Research (Brisk), was used to determine the roughness of the membrane surface, and the average roughness value of a 3 × 3 µm² area was calculated. A FESEM-MIRA III, TESCAN apparatus with an accelerating voltage of 15 kV was used to evaluate the membranes’ cross section and surface morphology. Prior to analysis, the produced samples were coated with a gold layer using a sputter coater. The surface morphology of the optimum membrane was observed using TEM with a ZEISS EM10C instrument operated at 100 kV (Germany). Differential scanning calorimetry (DSC) analysis was performed to investigate the thermal behavior and glass transition temperature (Tg) of the prepared membranes using a DSC instrument (TA instrument Q100). Samples were heated from 25 °C to 260 °C at a heating rate of 10 °C min^− 1 under a nitrogen atmosphere.

Performance evaluation of membranes

Membrane flux and rejection performance

A laboratory-scale cross-flow filtration module with an effective filtration area of approximately 39 cm² was used for the measurements. Before each test, the membranes were cut to the appropriate size and placed on the module. To achieve steady-state conditions, the mounted membrane was compacted with deionized water at 8 bar for 30 min. After reaching stable conditions, tests were conducted with 500 mL feed at 5 bar for 2 h at ambient temperature, and permeate flow was recorded at predetermined intervals. Flux of permeate was calculated using the following equation²¹:

1

where the variables V, A, and ∆T represent the measured permeate volume (L), effective surface area (m²), and the filtration time (h).

To enable comparison of membrane performance independent of operating pressure, water permeance (P) was also calculated using the following equation:

2

where P is the water permeance (L/m. h. bar), J is the permeate flux (L/m. h), and ΔP is the applied transmembrane pressure (bar).

In addition, the separation and rejection performance of the fabricated membranes was determined by using a 200 mg/L OFL solution, and then, for the optimum membrane, the rejection performance was carried out for various OFL concentrations (50–500 mg/L). Equation 3 is used to calculate the rejection performance²²:

3

where the solute concentrations (mg/L) of the feed and permeate are denoted by C_p and C_f. To evaluate these concentrations, a concentration-absorbance calibration curve was generated using a UV-Vis spectrophotometer (Optizen POP, Mecasys Co., South Korea) at λ_max (OFL) = 330 nm.

Hydrophilicity and porosity experiments

Using the sessile drop approach, the surface hydrophilicity of membranes was characterized using a contact angle measurement device (Dino-Lite Basic AM211 digital microscope; Taiwan). Firstly, the prepared membranes were dried under vacuum, and then the contact angle between designed membranes and water droplets was measured after one minute. To reduce inaccuracies, contact angle measurements were taken at three random locations, and the resulting results represent the average of all measurements.

To assess the impact of nanoparticle addition on membrane porosity, the dry-wet method was employed. Cut membranes of a certain size were submerged completely in deionized water for the entire night. Afterward, excess water was carefully removed to ensure no droplets remained on the membrane surfaces, and the membranes were weighed. The samples were then dried at 60 °C for 24 h, and their dry weights were recorded. The porosity of the membranes was determined by the following equation ²³:

4

where the wet and dry weights (g) of the membrane are denoted by W_w and W_d, respectively. Moreover, d_w, A, and L stand for water density (0.998 g/cm³), membrane surface area (m²), and membrane thickness (µm), respectively.

Furthermore, the average pore size of the membrane was determined using an equation based on the Guerout–Elford–Ferry relationship²⁴:

5

where µ is the water viscosity (8.9 × 10^− 4 Pa s at 25 °C), Q is the permeate volumetric flow rate (m³/s), and ΔP is the operating pressure (Pa).

Membrane antifouling performance test

Evaluating the antifouling capabilities of membranes is essential since they are frequently prone to fouling owing to the accumulation of pollutants. A type of protein known as BSA was used to evaluate the antifouling properties of prepared membranes. The evaluation involved measuring the pure water flux (PWF) of the fresh membranes for 2 h (J_w1), and then the system was fed with 500 mL of BSA solution with a 500 mg/L concentration, and the permeate flux was recorded as J_p after one hour. The contaminated membrane was rinsed for 30 min with deionized water, followed by another PWF test (J_w2) to assess recovery. To assess the degree of fouling resistance in a system, the following parameters were calculated: flux recovery ratio (FRR), total fouling ratio (R_t), reversible fouling ratio (R_r), and irreversible fouling ratio (R_ir), using the following equations, respectively^9,25.

6

7

8

9

To investigate the fouling process of the membranes in greater detail, several resistance parameters were also measured. These include the intrinsic resistance (R_m), which is influenced by the membrane’s chemistry, thickness, and pore size; cake/gel resistance (R_c), caused by the accumulation of a cake layer on the membrane surface; pore resistance (R_f), resulting from pore blockages; and the total filtration resistance (R_t). These parameters were calculated using the following equations^7,26:

10

11

12

13

Adsorption evaluation of prepared membranes

To assess the adsorption capacity of the membranes, 20 mg of membrane material was added to an 80 mL OFL solution with an initial concentration of 50 mg/L. The mixture was stirred for 4 h under standard ambient conditions. At specified time intervals, samples of the solution were withdrawn, and the final concentration was measured using UV-Vis spectroscopy. The adsorption capacity at each time point was calculated using the following equation²⁷:

14

in which C₀ and C_e (mg/L) represent the concentrations of the pollutant before and after adsorption, respectively. m (g) is the adsorbent mass and V (L) represents the volume of solution. Additionally, the pseudo-first-order, pseudo-second-order, and intra-particle diffusion kinetic models were utilized to examine adsorption kinetics and clarify potential rate-controlling mechanisms²⁸. Equations S1-3 represent the linear form of these models.

Results and discussions

Characterization of Functionalized MOFs

The crystallinity of the synthesized nanoparticles was examined using XRD. Figure 1A illustrates the effective synthesis of UiO-66-NH₂, evidenced by the similarity of its principal diffraction peaks to those reported in other studies^29,30. The peaks at 7.52, 8.67, and 25.87° coincide to the (111), (200), and (600) planes of UiO-66-NH₂, respectively. After grafting CTS, the diffraction patterns of the modified nanoparticles closely resemble those of UiO-66-NH₂, without displaying the characteristic diffraction peak of CTS. This could be because UiO-66-NH₂ is the primary component of the composite, while the CTS content is relatively low. This indicates that UiO-66-NH₂ maintained stable under the reaction conditions and its crystallinity did not change³¹.

Fig. 1.

(A) XRD, (B) FTIR, (C) Nitrogen adsorption−desorption isotherms, and (D) TGA and DTG curves of synthesized MOFs.

Figure 1B represents the FTIR spectra of synthesized MOFs. The FTIR spectra of UiO-66-NH₂ produced in the present study align quite well with the previous literature^32,33. The peak in the range of 3200 to 3400 cm^− 1 is indicative of the O–H and N–H stretching vibrations in CTS. The peak at 1660 cm^− 1 is ascribed to the C = O stretching of remaining amide groups. The peaks between 1050 and 1150 cm^− 1 correspond to C–O–C stretching vibrations in the CTS structure^34,35. The FTIR spectrum of UiO-66-CTS was recorded to investigate the grafting between UiO-66-NH₂ and CTS. Alterations in the 3340 cm^− 1 peak can be attributed to two factors. First, the incorporation of CTS into UiO-66-NH₂ may increase the number of available NH₂ groups, enhancing peak intensity. Second, grafting via a Schiff-base reaction may reduce the number of accessible NH₂ groups, leading to a decrease in peak intensity. The increase in the 1253 cm^− 1 peak is associated with the formation of N–C bonds, further confirming the successful synthesis of UiO-66-CTS³⁶.

The N₂ adsorption−desorption isotherm and porous characteristics of the synthesized MOFs are shown in Fig. 1C; Table 2. Both nanoparticles exhibit type I adsorption−desorption isotherms, characterized by a sharp uptake at low relative pressure and a rapid attainment of adsorption saturation, which is typical of microporous materials³⁷. A decrease in the surface area of UiO-66-CTS is observed compared to UiO-66-NH₂, which can be attributed to the partial blocking of MOF pores by CTS chains. Another possible reason is the reduced MOF mass fraction. Moreover, after modification, the average pore size decreases, which may be due to the presence of CTS chains within the pores of MOFs^38,39.

Table 2.

Porous characteristics of UiO-66-NH₂ and UiO-66-CTS based on BET analysis.

Samples	SBET (m2/g)	Total pore volume (cm3/g)	Average pore diameter (nm)
UiO-66-NH2	693	0.424	2.45
UiO-66-CTS	540	0.316	2.34

To evaluate the thermal stability of the synthesized MOFs, a TGA test was assessed (Fig. 1D). The weight loss of UiO-66-NH₂ occurred in three steps. Firstly, the evaporation of physisorbed water molecules on the nanoparticle surface is responsible for the weight loss at temperatures below 100 °C. The second weight loss at around 200–300 °C may be due to the removal of residual DMF solvent. After this, another weight loss is attributed to the decomposition of the UiO-66-NH₂ framework⁴⁰. The TGA curve of UiO-66-CTS indicates a weight loss at around 250 °C, which might be due to the degradation of the CTS chain⁴¹. The TGA analysis indicates that the final ash residue of UiO-66-NH₂ at temperatures above 600 °C is higher than that of the UiO-66-CTS. This difference is primarily attributed to the incorporation of organic compounds into the modified MOF structure. Therefore, the variation in residual ash contents between these samples can be linked to the extent of grafted organic compounds. Based on the TGA curves, the grafting percentage was calculated using the following equation⁴²:

15

where m^’₇₀₀ and m₇₀₀ denote the residual weights at 700 °C for the UiO-66-CTS and UiO-66-NH₂ respectively. The results indicate that the grafting percentage is approximately 7.5%.

FESEM images of UiO-66-NH₂ and its functionalized form are shown in Fig. 2A–D. The UiO-66-NH₂ particles had an irregular octahedron structure with an average diameter of about 100 to 200 nm that aligned with prior research²⁹. No significant morphological changes were observed after functionalization, reflecting the stability of the MOF structure under post-synthetic conditions⁴³. Moreover, Fig. 2E,F indicate the TEM images of UiO-66-CTS that prove the octahedron morphology of modified MOFs.

Fig. 2.

FESEM images of (A,B) UiO-66-NH₂ and (C,D) UiO-66-CTS. (E,F) TEM images of UiO-66-CTS.

Characterization of prepared membranes

Figure 3A displays the XRD patterns of the prepared membranes. Broad peaks between 15 and 30° confirm the amorphous structure of the PSF⁴⁴. However, the MOF-contained membranes do not show any specific change in their characteristic peaks, which can be because the MOFs are dispersed properly within the polymeric matrix. This can also be due to the low content of nanoparticles^23,45. Figure 3B illustrates the FTIR spectrum of prepared membranes; the PSF spectra exhibit a peak at 1150 cm^− 1, corresponding to the O-S-O group. The peaks at 1244 and 1582 cm^− 1 indicate the existence of C–O–C and C=C aromatic groups, respectively⁶. The presence of aforementioned peaks in MOF-contained membranes shows that the structure of the PSF remained unchanged during membrane fabrication. Furthermore, no alteration or displacement in peaks is detected, indicating that the MOF particles are positioned within the free volume of the polymer chains and don’t chemically interact with PSF chains⁴⁶.

Fig. 3.

(A) XRD patterns and (B) FTIR spectra of the prepared membranes.

Three-dimensional AFM images of the top surface of prepared membranes are shown in Fig. 4. As the UiO-66-NH₂ and UiO-66-CTS are added to the pristine membrane, the surface roughness of MOF-content membranes increases, and UC-1.25 has the roughest surface. The measured average roughness (R_a) values are 6.42, 9.93, and 12.48 nm for PSF, U-1, and UC-1.25, respectively. This is also evident from the images that the valleys and peaks are increased compared with pristine PSF. Typically, higher average surface roughness may contribute to membrane permeability by increasing the effective specific surface area; however, it is not considered the dominant factor governing water transport. In the present study, the enhanced permeability is mainly attributed to the improved hydrophilicity and better dispersion of MOF fillers, which facilitate water transport pathways within the membrane matrix^45,47. Although increased surface roughness is often associated with higher fouling propensity, the hydrophilic nature of the CTS-grafted MOF fillers is expected to mitigate this effect, which is further discussed in the antifouling analysis section.

Fig. 4.

AFM images of (A) PSF, (B) U-1, and (C) UC-1.25 membranes.

The top-surface FESEM images of the prepared membranes are displayed in Fig. 5A–C, which shows how the addition of UiO-66-NH₂ and UiO-66-CTS can affect the membrane morphology. The pristine membrane surface appears smooth and defect-free at the SEM scale, with no visible macro-defects, and the incorporation of MOFs did not significantly alter the membrane’s surface structure. Minimal wrinkles are evident, signifying exceptional compatibility between the PSF and MOFs, consequently demonstrating that the defect-free composite membranes are fabricated⁴⁸. Figure 6D–F depicts top surface TEM images of the UC-1.25 membrane. It is evident that nanoparticles are successfully distributed, providing further proof of the favorable compatibility between the membrane matrix and the modified nanoparticle.

Fig. 5.

Top-surface FESEM images of (A) PSF, (B) U-1, and (C) UC-1.25 membranes; (D–F) top-surface TEM images of UC-1.25 membranes.

Fig. 6.

Cross-sectional FESEM images of (A–C) Pure PSF, (D–F) U-1, and (G–I) UC-1.25 membranes.

Figure 6displays cross-section FESEM images of fabricated membranes. Pure PSF membrane shows a dense and spongy structure with teardrop-like macrovoids. Adding nanoparticles leads to a small increase in the macrovoid size and the number of teardrop-like macrovoids. The increase in macrovoid size is ascribed to the hydrophilic characteristics of the nanomaterials, resulting in a greater influx of non-solvent (water) during the phase inversion process of membrane fabrication. Additionally, the accelerated demixing rate contributes to an increased number of macrovoids, which are homogeneously distributed throughout the membrane cross-section⁶. It can be observed that all prepared membranes exhibit a porous sublayer; however, this porosity does not extend to the surface. Instead, a very thin and dense top layer is present, which can significantly enhance the membrane performance⁴⁹. These observations are consistent with the previously discussed surface FESEM images, further confirming the existence of a dense and defect-free surface layer. As it is shown in Fig. 6, the number of macrovoids in the UC-1.25 membrane is greater than that of the U-1 membrane, which can show the effect of grafted CTS on increasing hydrophilicity. Accordingly, it can be expected that the membranes containing functionalized MOFs showed better separation performance compared to those containing pristine MOFs. The EDX mapping images (Figures S2-4) of the membrane cross-sections confirm the presence of nanoparticles within the membrane matrix. The elemental distribution indicates that the nanoparticles are homogeneously dispersed throughout the structure of the membranes, with no signs of agglomeration.

DSC analysis was conducted to evaluate the thermal behavior of the prepared membranes and to investigate the interfacial compatibility between the polymer matrix and incorporated fillers. DSC provides valuable information about the mobility, rigidity, and structural changes of polymer chains. For amorphous polymers such as PSF, the glass transition temperature (Tg) is a key parameter used to describe polymer chain flexibility and membrane structure. The DSC thermograms show that the pristine PSF membrane exhibits a Tg at 174 °C, which is in good agreement with previously reported Tg values for polysulfone membranes, typically ranging between 180 and 190 °C depending on measurement conditions and polymer grade⁵⁰. After incorporation of UiO-66-CTS nanoparticles, the Tg values of the MMMs increased compared to the pristine PSF membrane (181 °C). The increase in Tg indicates restricted polymer chain mobility due to strong interfacial interactions between the PSF matrix and the incorporated nanoparticles. This behavior can be attributed to the presence of functional groups in UiO-66-CTS, which enhance interfacial adhesion and limit segmental motion of polymer chains, resulting in increased chain rigidity. The observed Tg shift suggests improved compatibility and stronger interfacial interaction between the filler and polymer matrix, confirming effective dispersion of the nanoparticles within the membrane structure. Overall, the DSC results demonstrate enhanced polymer–filler interaction and improved structural stability of the modified membranes⁵¹.

Permeability, hydrophilicity, and porosity of prepared membranes

Figure 7A illustrates the PWF data comparing the impacts of integrating UiO-66-NH₂ and UiO-66-CTS into pure PSF. The PWF values for all modified membranes exceed the PWF of the pristine PSF, which was 0.398 L/m².h. The incorporation of nanoparticles into the pristine membrane may generate water channels and enhance membrane hydrophilicity. Enhanced surface hydrophilicity can adsorb water molecules and facilitate their passage through interfacial channels, hence improving water permeability. To enable pressure-independent comparison of membrane performance, water permeance was also calculated (Fig. 7A). The trend in permeance was consistent with the PWF results, confirming the enhanced water transport properties of the modified membranes. At comparable filler concentrations, UC membranes exhibit superior PWF compared to U membranes, likely attributable to their more hydrophilic characteristics (Fig. 7C), as CTS is abundant in hydroxyl and amine functional groups, recognized for their strong interactions with water molecules. These interactions may result in the development of a hydration layer on the membrane surface, enhancing water uptake and flux. Increased filler loadings may lead to further enhancement in PWF due to nanoparticle agglomeration inside the MMM, potentially creating more water channels and greatly improving water permeability^13,19. However, although such interactions can improve water flux, the filler content must be carefully controlled, as excessive nanoparticle clustering may negatively affect the membrane’s structural stability or rejection rate.

Fig. 7.

(A) Pure water flux and permeance, (B) total porosity and mean pore size, (C) water contact angle, and (D) rejection performance of the fabricated membranes.

Figure 7B represents the porosity and pore radius of prepared membranes. Data for the pure PSF are found to be lower than those of samples with varying amounts of nanoparticles, suggesting that porosity increases as nanoparticle proportion rises. These findings are supported by FESEM images, which show an increased in the number of macrovoids. The formation of a more porous membrane structure may be attributed to the hydrophilic nature of the nanoparticles, which accelerates the solvent–non-solvent exchange during the phase inversion process, thereby promoting the development of macrovoids and enhancing overall membrane porosity^52,53. To further illustrate the microstructural pore distribution of the optimum membrane (UC-1.25), cross-sectional FESEM images were analyzed using Image J, and the resulting pore size histogram is provided in Figure S5. The histogram confirms the presence of a larger fraction of transport-relevant pores in the MOF-containing membrane, consistent with the calculated effective pore radius. Figure 7C represents the water contact angle of the fabricated membranes, measured using the sessile drop method. The pristine membrane exhibited a contact angle of 75.97°, confirming its hydrophobic nature. This relatively high contact angle indicates the intrinsic hydrophobic properties of pure PSF, a known limitation in water treatment applications where surface wettability is crucial. The incorporation of fillers continuously decreases the contact angle, signifying an enhancement in the hydrophilicity of the prepared membranes. Membranes containing UiO-66-CTS, with the same filler content as those with UiO-66-NH₂, displayed a lower contact angle, highlighting the enhanced hydrophilicity of UiO-66-CTS due to the grafted CTS. This reduction in contact angle can be attributed to the presence of many polar functional groups in the ligands of the MOF and CTS structures. Additionally, the increase in membrane roughness, as confirmed by the AFM test, may also contribute to this decrease in contact angle^53,54.

OFL rejection performance

The effect of nanoparticles concentration on membrane performance was assessed for OFL solutions (Fig. 7D). The flux of the OFL solution follows the same trend as PWF, indicating the formation of channels due to the addition of fillers, and the increased hydrophilicity results in better flux for UC membranes compared to U membranes. PSF shows a relatively low rejection rate (71.33%). However, the addition of nanoparticles significantly increases this rate, likely due to the enhanced hydrophilicity of the prepared membranes (as confirmed by contact angle tests), which can form a hydrated layer on the membrane surface and improve the rejection rate⁵⁵. Moreover, the high surface area of MOFs allows them to act as adsorbents and further increase the rejection rate⁵⁶. It is indicated that both U and UC membranes have an optimum filler content that provides the highest rejection rate. U-1 and UC-1.25 membranes are the optimum ones, with rejection rates of 91.99% and 94.95%, respectively. At MOF contents higher than these concentrations, the rejection rate decreases. This might be due to the agglomeration of nanoparticles and the formation of non-selective pathways that allow pollutants to pass through the membrane⁵⁷.

The impact of pollutant concentration on the performance of the optimum membrane (UC-1.25) was examined (Fig. 8A). As the concentration increases, the rejection rate shows a slight increase, while the flux slightly decreases. An increase in pollutant concentration reinforces the driving force for separation, resulting in improved rejection efficiency as the membrane exhibits improved contaminant rejection. Higher pollutant concentrations increase the probability of membrane fouling, since OFL molecules accumulate on the membrane surface or within its pores, hindering water flow. Moreover, the raised viscosity of higher concentrated solutions diminishes the mobility of water molecules, resulting in an additional decrease in flux⁵⁸.

Fig. 8.

(A) The impact of OFL concentration on the performance of the optimum membrane (UC-1.25) and (B) adsorption capacity of the prepared membranes.

Another factor assessed in evaluating the performance of the membranes is the adsorption capacity of the prepared membranes, and the results are depicted in Fig. 8B. Modified membranes have a better adsorption capacity compared to pure PSF; this outcome is the opposite of what is expected because, by increasing the hydrophilicity, a hydration layer covers the surface of the membranes, and it is expected that the adsorption capacity decreases⁵⁹. The increased adsorption can be attributed to the large surface area of the incorporated MOFs, which creates more active adsorption sites that can overcome the increased hydrophilicity of the membranes⁷. The better adsorption capacity of the U-1 compared with the UC-1.25 membrane can be attributed to the partial blocking of MOF pores by CTS chains, which reduces the total surface area and accessible active sites³⁹. Moreover, Figure S6 and Table S1 demonstrate that the pseudo-second-order model had the highest R² values and closely aligned with the experimental adsorption capabilities, suggesting that chemisorption dominates the process for prepared membranes^60,61. The better adsorption capacity of the U-1 compared with the UC-1.25 membrane can be attributed to the partial blocking of MOF pores by CTS chains, which reduces the total surface area and accessible active sites.

Moreover, the effect of pH was investigated on the optimum membrane to understand whether hydrophilicity is the predominant phenomenon or if other factors, such as electrostatic interactions, are also effective. The initial pH of the unadjusted solution was approximately 8.2. The pKa values for OFL are approximately 5 and 8⁶², suggesting that the initial solution carries a negative charge. Based on previous literature, PSF exhibits a zeta potential near zero at acidic pHs, and its charge becomes more negative as the pH increases⁶³. Furthermore, Fig. 9B illustrates the zeta potential of UiO-66-CTS; its charge starts positive at low pHs and continuously decreases with increasing pH, eventually reaching negative values.

Fig. 9.

(A) The impact of pH on the performance of the optimum membrane (UC-1.25) and (B) zeta potential of UiO-66-CTS as a function of pH.

As shown in Fig. 9A, maximum rejection occurs at pH > 8. This is because, at these pH values, PSF, UiO-66-CTS, and the feed all possess negative charges, indicating electrostatic repulsion, which leads to the highest rejection. These results suggest that electrostatic repulsion is present in all membranes. Therefore, the increased rejection observed for UC-1.25 compared to PSF and U-1 is attributed to enhanced hydrophilicity.

Antifouling performance of the optimum membranes

The antifouling performance of the membrane is regarded as a crucial determinant for the practical implementation of the membrane process. To assess the dynamic antifouling behavior of the prepared membranes, a circulating filtration test was carried out under steady-pressure conditions (Fig. 10A). The PWF and BSA flux of modified membranes exceed that of pure PSF, as previously noted. In the dynamic antifouling test for all membranes, the flux remained comparatively steady over 2 h of pure water filtering. After switching to the BSA solution, the flux has sharply decreased and stayed relatively constant during filtration. The sharp decrease can be attributed to the deposition of BSA on the surface of membranes, and stable flux can be due to reaching an equilibrium of adsorption and desorption of BSA molecules⁶⁴. After washing the membrane, a PWF test was carried out, and flux was recovered in the prepared membranes. Modified membranes showed a much better recovery in their PWF compared to pure PSF (Fig. 10B), reaching FRR values of 61.72%, 91.74%, and 95.14% for PSF, U-1, and UC-1.25, respectively. The exceptional antifouling performance of the UC-1.25 can be attributed to the enhanced hydrophilicity gained from incorporating the highly hydrophilic UiO-66-CTS into the pristine PSF⁶⁵. Furthermore, at the test pH (~ 7.0), both the membrane surface and the BSA molecules (isoelectric point ≈ 4.7) carry a negative charge, leading to a strong electrostatic repulsion that further prevents protein deposition⁶⁶. This discovery aligns with the previously discussed water contact angle data (Fig. 7C). The FRR is compared with that of several previously reported fillers used in PSF membranes for BSA filtration, as summarized in Table 3.

Fig. 10.

(A) Flux variations over time during the antifouling evaluation process, (B) FRR values, (C) fouling resistance ratio, and (D) fouling resistance parameters of the prepared membranes.

Table 3.

Comparison of FRR of PSF-based MMMs incorporating various fillers for BSA filtration.

Filler	FRR (%)	Reference
Zn: Al2O3	98.4	67
Amino-functionalized TiO2 nanotube	90	68
Dodecylamine-functionalized GO	89.1	69
UiO-66	95.62	7
UiO-66-NH2	91.74	This work
UiO-66-CTS	95.14	This work

Following the fouling experiment with the BSA solution, the reversible and irreversible fouling resistance of the prepared membranes were calculated to further evaluate the impact of membrane modification. Figure 10C illustrates a considerable reduction in R_t for U-1 and UC-1.25 in comparison to PSF, suggesting that the modified membranes exhibit less fouling resistance. Furthermore, in PSF, R_ir is the predominant source of resistance, indicating that most contaminants on the membrane surface cannot be easily eliminated. However, in modified membranes, particularly UC-1.25, R_ir constitutes a smaller portion of the fouling resistance, suggesting that a hydration layer arises on the membrane surface, preventing cake layer formation. Consequently, most pollutants on the membrane surface can be eliminated by simple rinsing with pure water⁶⁵. These observations suggest that UiO-66-CTS has improved the antifouling properties of the PSF membranes, making them more suitable for reuse.

Moreover, four further fouling resistance parameters were calculated to better analyze the antifouling properties of the prepared membranes, as depicted in Fig. 10D. Accordingly, it was found that all resistance parameters were decreased after adding UiO-66-NH₂ and UiO-66-CTS. R_t, which is the sum of other parameters, decreased significantly, and the other three parameters decreased as well, which shows the modification has a positive effect on all cake/gel, pore, and intrinsic resistance, which can be attributed to the hydrophilic nature of nanoparticles and grafted CTS^7,59. Based on the results and discussions presented, the UiO-66-CTS particles are a viable candidate for enhancing the antifouling performance and pollutant removal capabilities of PSF membranes.

Long-term stability and sustainability of OFL rejection

The long-term stability of the optimized UC-1.25 membrane was evaluated to assess the sustainability of OFL rejection and the structural integrity of the incorporated UiO-66-CTS nanoparticles. A continuous 48 h filtration test demonstrated that the membrane maintained stable OFL rejection throughout the operation (Figure S7A), indicating that the separation mechanism is not dominated by reversible adsorption but is primarily governed by stable size- and charge-based interactions. ICP analysis of the feed solution after the 48 h test confirmed negligible leaching of metal ions (0.008 ppm), further verifying the strong interfacial stability of the MOF within the PSF matrix. In addition, XRD patterns of the UC-1.25 membrane after one year of immersion in water revealed no significant changes compared to the fresh membrane (Figure S7B), confirming the long-term structural stability of the composite. These results collectively demonstrate that the UC-1.25 membrane provides sustainable OFL rejection during extended operation, while maintaining both structural integrity and filler stability.

Conclusion

In this study, a novel nanoparticle (UiO-66-CTS) was successfully synthesized, as confirmed by characterization techniques, and effectively incorporated into a PSF matrix to enhance the performance of the polymeric membrane. CTS grafting significantly improved the dispersion of MOFs within the matrix and increased the hydrophilicity of the resulting membranes, reducing the water contact angle from 75.97° for pure PSF to 57.06° for the UC-1.25 membrane. Among the fabricated membranes, UC-1.25 exhibited the most promising results, achieving a pure water flux of 2.92 L/m².h, which was approximately 7.5 times higher than that of the pristine PSF membrane (0.39 L/m².h). Furthermore, UC-1.25 achieved rejection rates of over 94.95% for OFL and demonstrated a flux recovery ratio (FRR) of 95.14%, indicating excellent antifouling performance. These results demonstrate the superior of UiO-66-CTS as an efficient filler in MMMs. Overall, the findings confirm that this modified MOF is a strong candidate for practical use in wastewater treatment, offering a viable and scalable solution for real-world membrane applications.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

Amirhosein Ghorbani conducted the experiments, analyzed the data, and wrote the manuscript.Seyyed Abbas Mousavi and Hossein Molavi supervised the project and provided critical feedback throughout the study.All authors reviewed and approved the final manuscript.

Funding

This work was financially supported by the Iran National Science Foundation (INSF) with Grant No: 4033873. The authors gratefully appreciate their support.

Data availability

All data relevant to this study are included in the manuscript and supporting information file.

Declarations

Competing interests

The authors declare no competing interests.