Tailoring Morphology and Properties of Tight Utrafiltration Membranes by Two-Dimensional Molybdenum Disulfide for Performance Improvement

To enhance the permeation and separation performance of the polyethersulfone (PES) tight ultrafiltration (TUF) membrane, two-dimensional molybdenum disulfide (MoS2) was applied as a modifier in low concentrations. The influence of different concentrations of MoS2 (0, 0.25, 0.50, 1.00, and 1.50 wt%) on TUF membranes was investigated in terms of morphology, mechanical strength properties, permeation, and separation. The results indicate that the blending of MoS2 tailored the microstructure of the membrane and enhanced the mechanical strength property. Moreover, by embedding an appropriate amount of MoS2 into the membrane, the PES/MoS2 membranes showed improvement in permeation and without the sacrifice of the rejection of bovine serum protein (BSA) and humic acid (HA). Compared with the pristine membrane, the modified membrane embedded with 0.5 wt% MoS2 showed a 36.08% increase in the pure water flux, and >99.6% rejections of BSA and HA. This study reveals that two-dimensional MoS2 can be used as an effective additive to improve the performance and properties of TUF membranes for water treatment.


Introduction
To meet the growing demand for clean water, water recovery from wastewater by membrane technologies has been an emerging strategy [1,2]. The ultrafiltration membrane (UF) has become an interesting strategy because of its effectiveness in removing particles, microorganisms, and other organic contaminants [3]. Among them, the tight ultrafiltration (TUF) membrane (a molecular weight cut-off (MWCO) of~1000-10000 Da [4]) has a higher retention performance on organic material. Therefore, using TUF for the removal of natural organic matter (NOM) in water has attracted more and more attention [5][6][7]. However, current TUF membranes face a critical challenge: the trade-off between water permeability and selectivity. In general, conventional polymer-based UF membranes could not break the "upper bound" between the separation factor and membrane permeability [8]. Since the steric effect is the main separation mechanism of the UF membrane [9], and the selectivity of the UF membrane is determined by the pore radius (R) and pore radius distribution (σ), while permeability is mainly determined by pore radius (R), porosity (ε) and the thickness of selectivity layer (δm). The selectivity and permeability are inversely related [10]. Therefore, conventional modification approaches make it difficult to break through the trade-off limitation.
Recently, to enhance the water flux and separation performance of the membranes, polymer matrix membranes are advanced membranes incorporating inorganic materials and the homogeneous casting solution was obtained by heating and stirring. The viscosity of the casting solution containing different concentrations of MoS 2 additives is shown in Table 1. To prepare the membrane, the solution was cast on non-woven fabrics, and then immersed into the deionized water (DI) water bath. Finally, the prepared membrane was immersed in another water bath for 24 h to remove residues and then stored in DI water prior to use. The synthesis and filtration process diagram of PES/MoS 2 membrane is shown in Figure 1. Da,4000 Da,6000 Da,8000 Da,10,000 Da) and HA was acquired from Sigma Aldrich (Sigma-Aldrich, Inc., St. Louis, MO, USA).

Membrane Preparation
Different amounts of dry MoS2 were added to DMAc solution and ultrasonicated for 4 h for well dispersion (QSONICA SONICATORS, Newtown, CT, USA, 500W, 75%). Then, polymer (PES) and membrane pore-forming agents (PVP) were dissolved in the solvent, and the homogeneous casting solution was obtained by heating and stirring. The viscosity of the casting solution containing different concentrations of MoS2 additives is shown in Table 1. To prepare the membrane, the solution was cast on non-woven fabrics, and then immersed into the deionized water (DI) water bath. Finally, the prepared membrane was immersed in another water bath for 24 h to remove residues and then stored in DI water prior to use. The synthesis and filtration process diagram of PES/MoS2 membrane is shown in Figure 1.

Characterization of MoS2 and Membranes
The morphologies of MoS2 were characterized by file emission scanning electron microscopy (FESEM, HITACHIS-4800, Hitachi, Tokyo, Japan) and transmission electron microscopy (TEM, H-7650, Hitachi, Japan). The zeta potential (Zeta PALS, Malvern Instruments Ltd., Malvern, UK) was used to characterize the electronegativity of MoS2.
To verify that the 2D MoS2 sheets were incorporated into the UF membranes successfully, X-ray diffraction (XRD X'Pert, Pro, PANalytical, Almelo, Netherlands) was used to investigate the MoS2 and the prepared membranes in a range of 10-70°. Morphological structures of the membranes were examined by using a FESEM (HITACHIS-4800, Hitachi, Japan) and TEM (H-7650, Hitachi, Japan). The hydrophilicity of the membranes was determined using a contact angle goniometer (Dataphysics OCA20, Dataphysics, Filderstadt, Germany). A water drop with a volume of 5 μL was dropped onto the top surface of each membrane. At least three contact angles at different locations on each sample were recorded to obtain a reliable value. The mechanical properties of prepared membranes were analyzed by measuring tensile stress using tensile testing equipment (LRK-500N, NTS, Tokyo, Japan). The membrane samples were stretched at an elongation rate of 100 mmmin −1 . The membrane was initially fixed by grips

Characterization of MoS 2 and Membranes
The morphologies of MoS 2 were characterized by file emission scanning electron microscopy (FESEM, HITACHIS-4800, Hitachi, Tokyo, Japan) and transmission electron microscopy (TEM, H-7650, Hitachi, Japan). The zeta potential (Zeta PALS, Malvern Instruments Ltd., Malvern, UK) was used to characterize the electronegativity of MoS 2 .
To verify that the 2D MoS 2 sheets were incorporated into the UF membranes successfully, X-ray diffraction (XRD X'Pert, Pro, PANalytical, Almelo, Netherlands) was used to investigate the MoS 2 and the prepared membranes in a range of 10-70 • . Morphological structures of the membranes were examined by using a FESEM (HITACHIS-4800, Hitachi, Japan) and TEM (H-7650, Hitachi, Japan). The hydrophilicity of the membranes was determined using a contact angle goniometer (Dataphysics OCA20, Dataphysics, Filderstadt, Germany). A water drop with a volume of 5 µL was dropped onto the top surface of each membrane. At least three contact angles at different locations on each sample were recorded to obtain a reliable value. The mechanical properties of prepared membranes were analyzed by measuring tensile stress using tensile testing equipment (LRK-500N, NTS, Tokyo, Japan). The membrane samples were stretched at an elongation rate of 100 mmmin −1 . The membrane was initially fixed by grips at a distance of 55 mm, after which the movable crosshead containing the load cell of 500N pulled the membrane until the membrane broke. The concentrations of PEG in the feed solution and permeate were measured by a total organic carbon analyzer (TOC,TOC-LCSH, Shimadzu, Tokyo, Japan) [39]. The PEG rejection of as-prepared membranes was calculated by Equation (1): where C p and C f were the PEG concentrations of permeate and feed solutions, respectively (gL −1 ). The pore diameter of the membrane was equal to the Stokes radius (d s ) of PEG at a 50% rejection, which could be calculated by Equation (2) [40]: The porosity ε (%) was determined by a gravimetric method, as defined by Equation (3) [41]: where ε is the porosity of membranes (%); W w and W d are related to the wet weight and the dry weight of the membrane (g), respectively; D w (0.998 g cm −3 ) is the density of the water; and D p (0.37 g cm −3 ) is the density of polymer. The water flux was tested by a dead-end filtration cell with a volume capacity of 10 mL and the effective area of the membrane was 4.1 cm 2 at room temperature (25 ± 1 • C). The samples were prepressed at 0.15 MPa for 30 min with DI water as the feed solution, ensuring that the flux reaches a steady state, and then the pure water flux of each membrane sample was recorded at 0.1 MPa by monitoring the weight change in permeate with an electronic balance (Sartorius BS224S, Sartorius AG, Goettingen, Germany). The permeate flux (J 0 ) was calculated by Equation (4): where J 0 (Lm −2 h −1 ) was the membrane flux, ∆V (L) was the volume of permeated water during the period of permeation time ∆t (h), and A (m 2 ) was the testing membrane area. The filtration performances of the membranes were investigated by dead-end filtration cell and using BSA and HA as model solutes. The BSA (1.0 gL −1 ) solution in PBS (pH = 7.4), and HA (20 mgL −1 ) in DI water. The concentration of these two chemical substances was measured by ultraviolet-visible (UV-vis) spectrophotometer (Spectra Max M2, Molecular) at 280 nm (BSA) and 254 nm (HA), respectively. The permeation flux and BSA/HA rejections were calculated using Equations (1) and (4), respectively.

Characterization of MoS 2
The morphology of original MoS 2 and MoS 2 after 4 h of sonication was observed by SEM and TEM images. As shown in Figure 2, it can be observed that the MoS 2 flakes with ultrasonic treatment (Figure 2b The morphology of original MoS2 and MoS2 after 4 h of sonication was observed by SEM and TEM images. As shown in Figure 2, it can be observed that the MoS2 flakes with ultrasonic treatment (Figure 2b) have enhanced dispersion compared to the untreated MoS2 flakes (Figure 2a). The original MoS2 comprised a multilayered structure, as shown in Figure 2c. Compared to that, the lamellar number of MoS2 was significantly reduced (Figure 2d). The reduced layer number was further confirmed by TEM images in Figure  2e,f.  [42,43]. The zeta potential of MoS2 is shown in Figure 3b. The zeta potential of MoS2 was -31.5 ± 5.3 mV at pH 3 and gradually reduced to -41.35 ± 2.90 mV at pH 10. The decrease zeta potential of MoS2 was because of the oxidation of Mo. Mo can exist in the form of HMoO4or MoO4 2-in the aqueous solution, making MoS2 negatively charged [44,45].

Characterization of MoS2/PES Membrane °
The microstructure of the fabricated membranes was observed by SEM ( Figure  Compared to the smooth surface of the pure PES membrane M0, there was some M

Characterization of MoS 2 /PES Membrane •
The microstructure of the fabricated membranes was observed by SEM ( Figure 4). Compared to the smooth surface of the pure PES membrane M0, there was some MoS 2 exposed on the surface of PES/MoS 2 membranes. With an increase in the concentration of MoS 2 in casting solutions, more MoS 2 was observed on the membrane surface. This result was associated with the influence of MoS 2 on the membrane fabrication process. In the phase inversion process, the MoS 2 migrated from the PES matrix to the water bath, which reduced the interfacial energy between the casting solution and the water bath. As a result, the content of MoS 2 locally increased on the skin layer of PES membranes and more MoS 2 can be observed on the surface of PES membranes [46]. It was observed that the M0 membrane was full of sponge-like structures. The introduction of MoS 2 led to more finger-like pore structures in the membrane, as confirmed by the comparison between SEM images of the M0 membrane and the other PES/MoS 2 membranes in Figure 4. Moreover, with an increase in the MoS 2 concentration from 0 (M0) to 0.50 (M2) wt%, the finger-like structures became longer and wider. Additionally, the microvoids were closer to the skin layer. The major reason for this result was due to the addition of MoS 2 which accelerated the instantaneous exchange of solvent and non-solvent in the casting solution, leading to the formation of microvoids in the membranes [47,48]. In addition, MoS 2 sheets migrated to the skin layer of the membrane when the solvent and non-solvent exchange happened, which resulted in the formation of finger-like structures in the membrane interior.
The cross-sectional structures were further investigated by TEM, as shown in Figure 5. This indicates that the MoS 2 dispersed evenly in the PES matrix in the PES/MoS 2 membrane. With an increase in the MoS 2 concentration, the MoS 2 brought various porous structures to the PES matrix. This was due to the formation of cavities between 2D sheets and polymers. Interestingly, some cavities appeared in the dense skin layer of the PES/MoS 2 membranes. By using Image J software, the thickness of the dense skin layer in each membrane was measured. This indicates that the skin layer thickness increased with the increased concentration of MoS 2 in membranes, as M0 (3.46 µm) < M2 (3.83 µm) < M4 (4.80 µm). This may be related to the increased viscosity of the casting solution by adding MoS 2 , as shown in Table 1. To investigate the effect of the MoS 2 concentration on the hydrophilicity of membrane surface, the water contact angles (CAs) of the membranes were measured. It was found that the contact angle of the M0 membrane was approximately 47.9 • (Figure 7a). With the concentration of the MoS 2 increasing, the CA values of the TUF membranes increased gradually, which can be ascribed to the hydrophobic nature of the MoS 2 [53].
The mechanical strength properties of the PES and PES/MoS 2 membranes were expressed in terms of tensile strength and elongation at break. The result in Figure 7 indicates that the addition of MoS 2 improved the tensile strength of the modified membranes which was in line with other studies on MoS 2 [25]. For example, the M0 membrane was found to have a tensile strength of 36.09 MPa at elongation at a break of 7.98%. Compared to that, the M4 membrane had a tensile strength of 48.59 MPa at a break of 10.97%. In the mixed-matrix membranes, MoS 2 could cross-link with the polymeric chains and increase the rigidity of polymeric chains. Due to this, the mechanical strength of mixed-matrix membranes was enhanced. However, the tensile strength did not show a gradual increase with increasing MoS 2 content for membranes M1-M4. When the concentration of MoS 2 was 0.50 wt%, the membrane M2 showed lower tensile strength than that of the membrane M1, which was due to the longer and wider finger-like structures in the membrane M2. The porous structures in the M2 membranes reduced the tensile strength. The cross-sectional structures were further investigated by TEM, as shown in Figure  5. This indicates that the MoS2 dispersed evenly in the PES matrix in the PES/MoS2 mem- tures to the PES matrix. This was due to the formation of cavities between 2D sheets and polymers. Interestingly, some cavities appeared in the dense skin layer of the PES/ MoS2 membranes. By using Image J software, the thickness of the dense skin layer in each membrane was measured. This indicates that the skin layer thickness increased with the increased concentration of MoS2 in membranes, as M0 (3.46 μm) < M2 (3.83 μm) < M4 (4.80 μm). This may be related to the increased viscosity of the casting solution by adding MoS2, as shown in Table 1.     To investigate the effect of the MoS2 concentration on the hydrophilicity of membrane surface, the water contact angles (CAs) of the membranes were measured. It was found that the contact angle of the M0 membrane was approximately 47.9° (Figure 7a). With the concentration of the MoS2 increasing, the CA values of the TUF membranes increased gradually, which can be ascribed to the hydrophobic nature of the MoS2 [53].  The MWCOs of as-prepared membranes were measured, and the results are shown in Figure 8. The fabricated membranes exhibited an MWCO range between 7.80 and 9.78 kDa, which confirmed that these membranes were TUF membranes. Compared to the pristine PES membrane with 7.83 kDa of MWCO, the MWCO increased to 9.78 kDa for the membrane M1 and 9.67 kDa for the membrane M2, but then dropped to 7.83 kDa for the membrane M3 and 7.80 kDa for the membrane M4. The corresponding pore radius of the TUF membranes is shown in Table 2. The results indicate that the pore sizes of the membranes increased as the concentration of MoS 2 in the casting solutions was 0 to 0.50 wt%, while higher contents of nanofillers (1.00-1.50 wt%) may lead to the MoS 2 having a strong interaction with PES and DMAc, which reflected an increase in the viscosity of the casting solution. These effects led to a delay in PES precipitation, a decrease in the diffusion rate of water in the PES matrix, and a delay in liquid-liquid stratification, which inhibited the synthesis of macropores [54,55] and consequently reduced the membrane pore size. To investigate the effect of the MoS2 concentration on the hydrophilicity of memb surface, the water contact angles (CAs) of the membranes were measured. It was fo that the contact angle of the M0 membrane was approximately 47.9° (Figure 7a). With concentration of the MoS2 increasing, the CA values of the TUF membranes increa gradually, which can be ascribed to the hydrophobic nature of the MoS2 [53]. 0.50 wt%, the membrane M2 showed lower tensile strength than that of the mem M1, which was due to the longer and wider finger-like structures in the membran The porous structures in the M2 membranes reduced the tensile strength.
The MWCOs of as-prepared membranes were measured, and the results are s in Figure 8. The fabricated membranes exhibited an MWCO range between 7.80 an kDa, which confirmed that these membranes were TUF membranes. Compared pristine PES membrane with 7.83 kDa of MWCO, the MWCO increased to 9.78 kD the membrane M1 and 9.67 kDa for the membrane M2, but then dropped to 7.83 kD the membrane M3 and 7.80 kDa for the membrane M4. The corresponding pore rad the TUF membranes is shown in Table 2. The results indicate that the pore sizes membranes increased as the concentration of MoS2 in the casting solutions was 0 t wt%, while higher contents of nanofillers (1.00-1.50 wt%) may lead to the MoS2 hav strong interaction with PES and DMAc, which reflected an increase in the viscosity casting solution. These effects led to a delay in PES precipitation, a decrease in the sion rate of water in the PES matrix, and a delay in liquid-liquid stratification, whi hibited the synthesis of macropores [54,55] and consequently reduced the membrane size.  Table 2 also shows the pure water flux of the prepared TUF membranes. The indicates that the pure water flux of membranes initially improved with the incr content of MoS2 when the concentration was 0 to 0.50 wt%. The pure water flux of th membrane without MoS2 was found to be 72.37 Lm −2 h −1 , and it increased to 91.86 L for the M1 membrane. The water flux permeability of UF membranes was determin properties including the porosity, pore size, and thickness of the membranes' skin Compared to the M0 membrane, the M1 membrane has similar porosity but a MWCO and mean pore radius. Therefore, the improved permeability in the M1 brane was potentially dominated by the influence of the large pore size. Similar r could also be found when comparing the performance of the M2 and M3 memb Compared to the M0 membrane, the M2 membrane exhibited both improved porosit pore size, which resulted in 1.36 times higher water flux (98.48 Lm −2 h −1 ) than that M0 membrane. However, a further increase in the MoS2 concentration increased the  Table 2 also shows the pure water flux of the prepared TUF membranes. The result indicates that the pure water flux of membranes initially improved with the increased content of MoS 2 when the concentration was 0 to 0.50 wt%. The pure water flux of the M0 membrane without MoS 2 was found to be 72.37 Lm −2 h −1 , and it increased to 91.86 Lm −2 h −1 for the M1 membrane. The water flux permeability of UF membranes was determined by properties including the porosity, pore size, and thickness of the membranes' skin layer. Compared to the M0 membrane, the M1 membrane has similar porosity but a larger MWCO and mean pore radius. Therefore, the improved permeability in the M1 membrane was potentially dominated by the influence of the large pore size. Similar results could also be found when comparing the performance of the M2 and M3 membranes. Compared to the M0 membrane, the M2 membrane exhibited both improved porosity and pore size, which resulted in 1.36 times higher water flux (98.48 Lm −2 h −1 ) than that of the M0 membrane. However, a further increase in the MoS 2 concentration increased the membrane surface hydrophobicity, increased the thickness of the skin layer, and the reduced the membrane pore size, which consequently decreased the water flux of the M4 membrane to 80.18 Lm −2 h −1 . Figure 9 shows the filtration performance of as-prepared TUF membranes by using BSA solution as the feed solution. The result indicates that all as-prepared PES/MoS 2 membranes have higher BSA rejections than the pristine membrane (M0), and all of them were over 99.50%. This was related to the fact that the negative charge brought by MoS 2 increased the negative charge on the membrane surface. The permeation flux of the BSA solution for the M0 membrane was 62.37 Lm −2 h −1 . The water flux increased to 68.29 Lm −2 h −1 for the M1 membrane, and the rejection increased to 99.85%. As a function of the MoS 2 concentration, the permeation flux decreased gradually. An initial increase in the permeation flux was attributed to the pore size change, while the following decrease was due to the increased hydrophobicity of the membranes [56].  Figure 9 shows the filtration performance of as-prepared TUF membranes by BSA solution as the feed solution. The result indicates that all as-prepared PES/MoS2 branes have higher BSA rejections than the pristine membrane (M0), and all of them over 99.50%. This was related to the fact that the negative charge brought by Mo creased the negative charge on the membrane surface. The permeation flux of th solution for the M0 membrane was 62.37 Lm −2 h −1 . The water flux increased to 68.29 L for the M1 membrane, and the rejection increased to 99.85%. As a function of the concentration, the permeation flux decreased gradually. An initial increase in the p ation flux was attributed to the pore size change, while the following decrease was d the increased hydrophobicity of the membranes [56]. The filtration performance of prepared TUF membranes was further investigat using HA aqueous solution as the feed solution. This result shows in Figure 10, ind that all the membranes have excellent separation performance for HA with the rej rate between 99.21 and 99.60%. Moreover, the permeation flux of the HA solution f M1 membrane (0.25 wt%) was 77.24 Lm −2 h −1 , which was 33.4% higher than that of th membrane (57.90 Lm −2 h −1 ). This increase was primarily dependent on the change in size. However, it was also found that the effective pore sizes of the M1 and M3 memb were almost the same and the permeation was changed. This phenomenon may be c by an increase in the surface hydrophobicity of the membrane due to an increase content of MoS2. HA adsorption onto a PES membrane has been confirmed in oth search and mainly correlated to the hydrophobic groups of molecules [57]. The eff adsorption reduced the permeability of the membrane. The results show that furth The filtration performance of prepared TUF membranes was further investigated by using HA aqueous solution as the feed solution. This result shows in Figure 10, indicates that all the membranes have excellent separation performance for HA with the rejection rate between 99.21 and 99.60%. Moreover, the permeation flux of the HA solution for the M1 membrane (0.25 wt%) was 77.24 Lm −2 h −1 , which was 33.4% higher than that of the M0 membrane (57.90 Lm −2 h −1 ). This increase was primarily dependent on the change in pore size. However, it was also found that the effective pore sizes of the M1 and M3 membranes were almost the same and the permeation was changed. This phenomenon may be caused by an increase in the surface hydrophobicity of the membrane due to an increase in the content of MoS 2 . HA adsorption onto a PES membrane has been confirmed in other research and mainly correlated to the hydrophobic groups of molecules [57]. The effect of adsorption reduced the permeability of the membrane. The results show that further research on enhanced hydrophilicity of 2D MoS 2 would improve the antifouling performance of the mixed-matrix membrane. Table 3 compares the HA separation performance of the PES/MoS 2 membrane with some other membranes in the previous literature [23,[58][59][60][61][62], and indicates that the PES/MoS 2 membrane has better HA selectivity compared to UF membranes and improved water permeability compared to NF membranes with similar selectivity properties. mance of the mixed-matrix membrane. Table 3 compares the HA separation performance of the PES/MoS2 membrane with some other membranes in the previous literature [23,[58][59][60][61][62], and indicates that the PES/ MoS2 membrane has better HA selectivity compared to UF membranes and improved water permeability compared to NF membranes with similar selectivity properties.

Conclusion
In this work, 2D MoS2 was blended into the PES matrix to enhance the performance and properties of membranes. A series of PES/MoS2 membranes were fabricated and their morphologies and separation performance were investigated. The results indicate that the mechanical strength property of membranes was improved by the addition of MoS2. SEM images proved the presence of MoS2 in the mixed-matrix membrane, and the finger-like macrovoid appeared. Further, the dispersed morphology was given by TEM. These changes enhanced the permeability of both pure water and the BSA/HA solution. By optimizing the concentration of MoS2, the PES/MoS2 membrane with 0.50 wt% MoS2 exhibited a 36.08% increase in the water flux compared to the pristine TUF membrane, and

Conclusions
In this work, 2D MoS 2 was blended into the PES matrix to enhance the performance and properties of membranes. A series of PES/MoS 2 membranes were fabricated and their morphologies and separation performance were investigated. The results indicate that the mechanical strength property of membranes was improved by the addition of MoS 2 . SEM images proved the presence of MoS 2 in the mixed-matrix membrane, and the finger-like macrovoid appeared. Further, the dispersed morphology was given by TEM. These changes enhanced the permeability of both pure water and the BSA/HA solution. By optimizing the concentration of MoS 2 , the PES/MoS 2 membrane with 0.50 wt% MoS 2 exhibited a 36.08% increase in the water flux compared to the pristine TUF membrane, and excellent rejection properties to BSA (99.85%) and HA (99.60%). This study exhibited the feasibility of applying low concentrations of MoS 2 to tune the structure and properties of TUF membranes, which could be an effective strategy to fabricate high-performance TUF membranes.

Data Availability Statement:
The data presented in this study are available on request from the corresponding author.