Polydopamine-Assisted Two-Dimensional Molybdenum Disulfide (MoS2)-Modified PES Tight Ultrafiltration Mixed-Matrix Membranes: Enhanced Dye Separation Performance

Tight ultrafiltration (TUF) membranes with high performance have attracted more and more attention in the separation of organic molecules. To improve membrane performance, some methods such as interface polymerization have been applied. However, these approaches have complex operation procedures. In this study, a polydopamine (PDA) modified MoS2 (MoS2@PDA) blending polyethersulfone (PES) membrane with smaller pore size and excellent selectivity was fabricated by a simple phase inversion method. The molecular weight cut-off (MWCO) of as-prepared MoS2@PDA mixed matrix membranes (MMMs) changes, and the effective separation of dye molecules in MoS2@PDA MMMs with different concentrations were obtained. The addition amount of MoS2@PDA increased from 0 to 4.5 wt %, resulting in a series of membranes with the MWCO values of 7402.29, 7007.89, 5803.58, 5589.50, 6632.77, and 6664.55 Da. The MWCO of the membrane M3 (3.0 wt %) was the lowest, the pore size was defined as 2.62 nm, and the pure water flux was 42.0 L m−2 h−1 bar−1. The rejection of Chromotrope 2B (C2B), Reactive Blue 4 (RB4), and Janus Green B (JGB) in aqueous solution with different concentrations of dyes was better than that of unmodified membrane. The separation effect of M3 and M0 on JGB at different pH values was also investigated. The rejection rate of M3 to JGB was higher than M0 at different pH ranges from 3 to 11. The rejection of M3 was 98.17–99.88%. When pH was 11, the rejection of membranes decreased with the extension of separation time. Specifically, at 180 min, the rejection of M0 and M3 dropped to 77.59% and 88.61%, respectively. In addition, the membrane had a very low retention of salt ions, Nacl 1.58%, Na2SO4 10.52%, MgSO4 4.64%, and MgCl2 1.55%, reflecting the potential for separating salts and dyes of MoS2@PDA/PES MMMs.


Introduction
Tight ultrafiltration (TUF) membranes have received increasing attention from water treatment and resource recovery in recent years [1]. Especially in the treatment of dye wastewater, the TUF membrane of mesoporous shows more beneficial effects than nanofiltration (NF) membranes, not only because of the relative lower operating pressure required by the TUF process, but also because of the pore size characteristics of TUF membranes. The sieving effect plays a major role. The dye molecules could be intercepted and the salt ions pass through the membrane pores. Higher purity dye would be obtained. However, due to the combined effect of the sieving effect, dissolution diffusion, and Donan effect [2], traditional nanofiltration (NF) membranes often have a high retention rate of divalent salt ions, which will increase the difficulty of further separation of salt and dye. The TUF membrane is actually a type of ultrafiltration membrane (UF) with a molecular ogy structure, surface characteristics, membrane thermostability, dyes separation, and salts filtration performance of the prepared membranes were investigated.

Synthesis of MoS2@PDA
To prepare MoS2@PDA (Figure 1), 0.5 g of MoS2 sheets were dispersed in 100 mL of 10 mM Tris-buffer (pH ≈ 8.5) by ultrasonic for 4 h to form a suspension solution. Then, 0.3 g DA hydrochloride was added to the suspension solution, stirred rapidly for 30 min, and then changed to low-speed agitation for another 24 h at normal temperature. Subsequently, the mixing solution was centrifuged, and the product was washed three times with DI water and ethanol, and then dried in an oven at 60 °C. This product was labeled as MoS2@PDA.  Table 1 lists the components of different casting solutions. Different contents of MoS2@PDA powders were added into the DMAc solution and sonicated for 30 min to form the MoS2@PDA uniform dispersion solution. After that, PVP and PES were added into the MoS2@PDA solution, and the mixture was stirred at 60 °C until a homogeneous solution was formed. The casting solution was poured on a non-woven fabric on a glass plate. A casting knife gap setting of 200 μm was applied to control the thickness of membranes, and the casting process was conducted under a temperature of 25 °C and humidity of 45 ± 5%. After casting, the plate was immersed into a DI water bath. Finally, the prepared membrane was transferred to fresh pure water for 24 h to remove the residue and then stored in DI water prior to use.   Table 1 lists the components of different casting solutions. Different contents of MoS 2 @PDA powders were added into the DMAc solution and sonicated for 30 min to form the MoS 2 @PDA uniform dispersion solution. After that, PVP and PES were added into the MoS 2 @PDA solution, and the mixture was stirred at 60 • C until a homogeneous solution was formed. The casting solution was poured on a non-woven fabric on a glass plate. A casting knife gap setting of 200 µm was applied to control the thickness of membranes, and the casting process was conducted under a temperature of 25 • C and humidity of 45 ± 5%. After casting, the plate was immersed into a DI water bath. Finally, the prepared membrane was transferred to fresh pure water for 24 h to remove the residue and then stored in DI water prior to use.

Characterization of MoS 2 @PDA
The morphological structure of MoS 2 and MoS 2 @PDA was characterized by field emission transmission electron microscopy (FETEM, Tecnai F20-, FEI Corp., Portland, OR, USA). The hydrophilicity of nanomaterials was determined by a contact goniometer by dropping 2.0 µL of water (CA, DSA100, Krüss Company, Ltd., Hamburg, Germany). The zeta potential of MoS 2 and MoS 2 @PDA was measured using zeta potential (Zeta PALS, Malvern Instruments Ltd., Malvern, UK). The pH was adjusted in the range of 3 to 11 by NaOH and HCl solution. Thermogravimetric analyses (TGA, Perkin Elmer) were conducted under nitrogen atmosphere with a flow rate of 60 mL min −1 . The sample was placed in a ceramic crucible; the heating temperature ranged from 40 to 1000 • C at a heating rate of 10 • C min −1 . The crystal structure was tested by X-ray diffraction (XRD, X'Pert Pro, PANalytical, Netherlan) at a scanning rate of 10 • min −1 in the 2θ range of 10-70 • . The chemical structure of MoS 2 and MoS 2 @PDA was analyzed by Fourier transform infrared spectroscopy (FTIR, Nicolet iS10, Thermo Fisher Scientific Inc., Waltham, MA, USA). X-ray photoelectron spectrometry (XPS, ESCALAB 250XI, Thermo Fisher Scientific Inc., Waltham, MA, USA) was used to characterize the surface elemental composition of MoS 2 and MoS 2 @PDA.

Membrane Characterization
To verify the existence of MoS 2 @PDA in the MoS 2 @PDA/PES membranes, field emission scanning electron microscopy (FESEM, HITACHIS-4800, Hitachi Co. Ltd., Tokyo, Japan) was used to observe the structural changes of membrane surface and cross-section at 5.0 kV. Before FESEM observation, the dry membrane sample was sputtered with gold. The surface roughness of membranes was determined with atomic force microscope (AFM, Dimension 3100, Bruker Corp., Santa Barbara, CA, USA). The test area is 5 µm × 5 µm. Average values of roughness for the membrane samples were obtained by measuring three different locations. The surface electrical properties were tested using an electro kinetic analyzer (SurPASS3, Anton Paar, Graz, Austria). The pH was adjusted in the range of 3 to 11 by adding NaOH and HCl solution, and 1 mM KCl solution was used as the electrolyte solution. TGA and XPS were applied for analyzing the characteristics of the as-prepared membranes.

Molecular Weight Cut-Off and Filtration Performance of Membranes
The MWCO of membranes were defined using the different molecular weights of polyethylene glycols (PEGs), which were retained with 90% [22]. The test was performed using a dead-end filtration system (Model 8050, Millipore Corp., Burlington, MA, USA) at 0.1 MPa. First, 50 mL of 1.0 gL −1 PEG solution was used as the feed liquid. The filtration process was completed when ≈20% of feed solution (i.e., 10 mL) was filtrated through the membrane. Both the feed and permeate solution were diluted by 10 times, and then, the concentration of each solution was tested by a total organic carbon analyzer (TOC, TOC-LCSH, Shimadzu, Japan) [23]. The PEG rejection was calculated using Equation (1): where C p and C f are the PEG concentrations of permeate and feed solutions respectively (gL −1 ). It was reported that the mean effective pore size of the membrane equals the Stokes radius (ds) of PEG at 50% rejection, which could be calculated by Equations (2) and ( Membranes 2021, 11, 96 5 of 17 The pore size distribution of the membrane was analyzed by Equation (4).
where r s and d s stand for the Stokes radius and diameter of the PEG, respectively. M PEG is the MWCO of PEG. The geometric mean diameter (µs) can be calculated as ds corresponding to R = 50%, and the geometric standard deviation (σg) can be determined from the ratio of ds at R = 84.13% and R = 50%. By ignoring the dependence of solute separation on the spatial and hydrodynamic interaction between the solute and the pore size, the average effective pore size (µp) and geometric standard deviation (σp) of the membrane can be regarded as the same as the value of µs and σg [25,26]. The water flux was measured by a dead-end filtration cell with a volume capacity of 50 mL. The effective area of the membrane was 13.4 cm 2 . Membrane samples were firstly pre-pressed at 0.15 MPa for 30 min, ensuring that the pure water flux reached a steady state; then, we recorded the permeate weight by an electronic balance with Wedge software at 0.1 MPa. The permeate flux (J) was calculated with Equation (5): where J 0 (L m −2 h −1 ) is the membrane flux, ∆V(L) is the volume of permeated water, A (m 2 ) is the membrane area, and ∆t (h) is the permeation time.
The separation performances of the membranes were investigated to filtered dye molecules and salts. The concentration of C2B, RB4, and JGB were measured by Ultravioletvisible (UV-vis) spectrophotometer (Spectra Max M2, Molecular Devices Co., San Jose, CA, USA). The concentrations of NaCl, Na 2 SO 4 , MgSO 4 , and MgCl 2 were measured by a conductivity meter (sensION+EC5, HACH). The permeation flux (J d ) and rejection (R d ) of dye were calculated by Equations (1) and (5), respectively.

Characterization of MoS 2 and MoS 2 @PDA
The morphology of MoS 2 and MoS 2 @PDA was investigated by FETEM ( Figure 2). It can be found that the surface of the MoS 2 was smooth. After modification, MoS 2 @PDA showed many convex coatings in Figure 2d,e. Figure 2b,e show the HRTEM images of MoS 2 and MoS 2 @PDA. It can be observed that the edge structure of MoS 2 @PDA becomes softened compared to the unmodified MoS 2 . In addition, the inset images revealed that both MoS 2 @PDA and MoS 2 consist of the hexagonal lattices and the lattice spacing of 0.27 nm, which was corresponding to the (100) lattice plane MoS 2 [27]. The energy dispersive spectroscopy (EDS) spectrum shown in Figure 2c,f indicated that a new peak N can be detected on MoS 2 @PDA.
The surface hydrophilicity was characterized by the water contact angle (CA) analysis. Average values were obtained by measuring three different locations. As shown in Figure 3, the CA of pristine MoS 2 was 79.6 ± 3.2 • . Impressively, the CA of MoS 2 @PDA was 55.53 ± 1.2 • , which indicated the improved water wettability of MoS 2 by PDA coating. The major reason for this improvement was that dopamine contains a large number of hydrophilic amines and hydroxyl groups, which polymerize under alkaline conditions to form PDA and deposit around MoS 2 , enhancing the surface hydrophilicity and thus reducing the CA [28,29].
When pH = 3, the potential of MoS2@PDA was −27.65 mV, and that of MoS2 was −35.9 mV. However, with the increase of pH, the potential of MoS2@PDA was lower than MoS2. This was due to the negative charge of the PDA phenol group after deprotonation at high pH, which enhanced the negative charge of modified particles [30].   Figure 4b gives the results from TGA and DTG analysis carried out on MoS2 and MoS2@PDA. There are two distinct stages of mass loss in the TGA curve: ≈100 °C and 100-500 °C, which corresponded to the peaks on the DTG curve at 102 °C, 328 °C for MoS2 and 96 °C, 282 °C for MoS2@PDA. In the temperature range of 500-1000 °C, the mass loss of MoS2 changes slowly, and the mass loss of MoS2@PDA was almost a straight line with a slope [31]. The TGA curve of MoS2 showed a weight loss of 2.10% when the temperature increased from 0 to 1000 °C. It was mainly due to the weight loss caused by the water or other impurities absorbed on the MoS2 [32]. The weight loss rate of MoS2@PDA was 6.36%, which was higher than the pristine MoS2, indicating that the PDA was coated on the MoS2 surface. According to the calculation of the TGA curve of unmodified and modified, the PDA deposition on MoS2 sheets was approximately 4.26%. When pH = 3, the potential of MoS2@PDA was −27.65 mV, and that of MoS2 was −35.9 mV. However, with the increase of pH, the potential of MoS2@PDA was lower than MoS2. This was due to the negative charge of the PDA phenol group after deprotonation at high pH, which enhanced the negative charge of modified particles [30].   Figure 4b gives the results from TGA and DTG analysis carried out on MoS2 and MoS2@PDA. There are two distinct stages of mass loss in the TGA curve: ≈100 °C and 100-500 °C, which corresponded to the peaks on the DTG curve at 102 °C, 328 °C for MoS2 and 96 °C, 282 °C for MoS2@PDA. In the temperature range of 500-1000 °C, the mass loss of MoS2 changes slowly, and the mass loss of MoS2@PDA was almost a straight line with a slope [31]. The TGA curve of MoS2 showed a weight loss of 2.10% when the temperature increased from 0 to 1000 °C. It was mainly due to the weight loss caused by the water or other impurities absorbed on the MoS2 [32]. The weight loss rate of MoS2@PDA was 6.36%, which was higher than the pristine MoS2, indicating that the PDA was coated on the MoS2 surface. According to the calculation of the TGA curve of unmodified and modified, the PDA deposition on MoS2 sheets was approximately 4.26%. The zeta potential characteristic of particles was shown in Figure 4a. Due to the deposition of zwitterionic PDA, at low pH, the amino group of the PDA was protonated. When pH = 3, the potential of MoS 2 @PDA was −27.65 mV, and that of MoS 2 was −35.9 mV. However, with the increase of pH, the potential of MoS 2 @PDA was lower than MoS 2 . This was due to the negative charge of the PDA phenol group after deprotonation at high pH, which enhanced the negative charge of modified particles [30]. Figure 4b gives the results from TGA and DTG analysis carried out on MoS 2 and MoS 2 @PDA. There are two distinct stages of mass loss in the TGA curve: ≈100 • C and 100-500 • C, which corresponded to the peaks on the DTG curve at 102 • C, 328 • C for MoS 2 and 96 • C, 282 • C for MoS 2 @PDA. In the temperature range of 500-1000 • C, the mass loss of MoS 2 changes slowly, and the mass loss of MoS 2 @PDA was almost a straight line with a slope [31]. The TGA curve of MoS 2 showed a weight loss of 2.10% when the temperature increased from 0 to 1000 • C. It was mainly due to the weight loss caused by the water or other impurities absorbed on the MoS 2 [32]. The weight loss rate of MoS 2 @PDA was 6.36%, which was higher than the pristine MoS 2 , indicating that the PDA was coated on the MoS 2 surface. According to the calculation of the TGA curve of unmodified and modified, the PDA deposition on MoS 2 sheets was approximately 4.26%.
MoS2@ PDA still had characteristic diffraction peaks at (002), (004), (103), (104), (105), and (110), which showed that before and after PDA wrapped, the crystal structure of MoS2 had not changed [35]. However, the existence of PDA affected the crystallinity of the MoS2 crystal [27]. The introduction of carbon would lead to the reduction of the diffraction peak [36]. Due to the deposition of dopamine, the carbon element increased, the purity of MoS2 crystal decreased, and the intensity of the diffraction peak reduced. The FTIR spectra of MoS2 and MoS2@PDA are shown in Figure 4d. Compared to the spectra of MoS2, there was a new peak in the spectra of MoS2@PDA. The peak at around 1498 cm −1 was the C=C skeleton vibration peak of the benzene ring from PDA [37]. The peak at 1630 cm −1 was formed by the oxidation of the dobutamine hydroxyl group to the ketone group C=O [38]. The peak at around 3440 cm −1 can be attributed to the stretching vibration of N-H and OH [32]. Compared to pristine MoS2, the hydrophilicity of MoS2@ PDA was enhanced due to PDA deposition, resulting in an enhanced O-H vibration of H2O [31].
The chemical composition of MoS2@PDA was further investigated by XPS analysis. As shown in Figure 5a, the XPS full energy spectrum investigation indicated that Mo, S, C, N, and O elements were present in the hybrid. It was found that the MoS2 characteristic peaks, such as Mo 3d and S 2p, were reduced after the modification of dopamine. However, the intensity of O1s and C1s were enhanced, and we confirmed the production of dopamine-induced polymerization, which was coated on the MoS2 surface. The  [33,34]. After encapsulation with PDA, there were no new crystallization peaks. The XRD pattern of MoS 2 @ PDA still had characteristic diffraction peaks at (002), (004), (103), (104), (105), and (110), which showed that before and after PDA wrapped, the crystal structure of MoS 2 had not changed [35]. However, the existence of PDA affected the crystallinity of the MoS 2 crystal [27]. The introduction of carbon would lead to the reduction of the diffraction peak [36]. Due to the deposition of dopamine, the carbon element increased, the purity of MoS 2 crystal decreased, and the intensity of the diffraction peak reduced.
The FTIR spectra of MoS 2 and MoS 2 @PDA are shown in Figure 4d. Compared to the spectra of MoS 2 , there was a new peak in the spectra of MoS 2 @PDA. The peak at around 1498 cm −1 was the C=C skeleton vibration peak of the benzene ring from PDA [37]. The peak at 1630 cm −1 was formed by the oxidation of the dobutamine hydroxyl group to the ketone group C=O [38]. The peak at around 3440 cm −1 can be attributed to the stretching vibration of N-H and OH [32]. Compared to pristine MoS 2 , the hydrophilicity of MoS 2 @ PDA was enhanced due to PDA deposition, resulting in an enhanced O-H vibration of H 2 O [31].
The chemical composition of MoS 2 @PDA was further investigated by XPS analysis. As shown in Figure 5a, the XPS full energy spectrum investigation indicated that Mo, S, C, N, and O elements were present in the hybrid. It was found that the MoS 2 characteristic peaks, such as Mo 3d and S 2p, were reduced after the modification of dopamine. However, the intensity of O1s and C1s were enhanced, and we confirmed the production of dopamineinduced polymerization, which was coated on the MoS 2 surface. The high-resolution Mo 3d spectra (Figure 5b) had three peaks; 229.34 eV and 232.51 eV related to Mo 4+ 3d 5/2 and Mo 4+ 3d 3/2 , while the peak at 235.59 eV may be derived from the surface oxidation of Mo 6+ [36,39]. In the spectra of the high-resolution S 2p (Figure 5c), there were peaks at 162 eV and 163.3 eV corresponding to the bivalent S 2p 3/2 and S 2p 1/2 (S 2− ) [40]. There were two peaks in N1 (Figure 5d); the one at 399.9 eV was associated with pyrrolic nitrogen in an indole ring, and the one at 395.3 eV related to the pyridinic resulting from the dopamine functionalization [41,42]. In the MoS 2 @PDA (Figure 5e), the binding energy of C1s at 284.5 eV was assigned to C-C/C-H bonds, while that at 285.7 eV was assigned to C-O/C-N, respectively [43]. In the original MoS 2 (Figure 5g), C1s at 284.9 eV and 286.8 eV was related to C-C and C-O, which may derived from precursors left over from the formation of MoS 2 . In the analysis of the O 1s, by comparing Figure 5f,h, there were two new peaks in the modified MoS 2 531.3 eV, corresponding to Mo-O and 533.23 eV related to the chemisorbed oxygen [44,45]. The binding energy 532.44 eV on MoS 2 was related to O-bonding with residual water [46].

Characterization of MoS 2 @PDA/PES TUF Membrane
The surface and the cross-sectional morphologies of membranes are shown in Figure 6. With the increase of MoS 2 @PDA concentration, the influence of MoS 2 @PDA can be noticed. By adding MoS 2 @PDA, the surface was no longer as smooth as the pristine one. It was because during phase inversion, MoS 2 @PDA migrated from the PES matrix to a water bath and appeared on the membrane surface. The surface and the cross-sectional morphologies of membranes are shown in Figure  6. With the increase of MoS2@PDA concentration, the influence of MoS2@PDA can be noticed. By adding MoS2@PDA, the surface was no longer as smooth as the pristine one. It was because during phase inversion, MoS2@PDA migrated from the PES matrix to a water bath and appeared on the membrane surface. The basic membrane in this study was filled with sponge structures, which was achieved by a high additive amount of PVP [47,48]. This single-layered morphology is favorable for the investigation of the influence of particles in the membrane structure. When the concentration of MoS2@PDA was 2.0 wt %, the presence of hydrophilic The basic membrane in this study was filled with sponge structures, which was achieved by a high additive amount of PVP [47,48]. This single-layered morphology is favorable for the investigation of the influence of particles in the membrane structure. When the concentration of MoS 2 @PDA was 2.0 wt %, the presence of hydrophilic MoS 2 @PDA in the casting solution accelerated the solvent-nonsolvent exchange and instantaneous liquid-liquid separation in the casting solution, resulting in the formation of the microvoids structures observed in the M1 membrane [49]. However, the number of macropores was reduced at a higher additive concentration, and the internal structure was more compact, as shown in the cross-section of membranes M3 and M5. This was due to the fact that rheology played a leading role in the increase of the viscosity of the casting solution with the addition of MoS 2 @PDA, which reduced the phase conversion rate and densified the membrane structure [50]. The morphological characteristic of the cross-section was different from that of the MoS 2 /PES TUF membrane; the MoS 2 @PDA/PES TUF membrane displays fewer microvoids, and the supporting layer is dominated by a sponge structure. This provides the possibility of forming smaller MWCO membranes.
The surface roughness of membranes was measured by AFM (Figure 7). Results were obtained by measuring three different points and indicated that the M0 membrane displayed a smooth surface with an average roughness (Ra) value of 1.73 ± 0.18 nm. The addition of MoS 2 @PDA made the membrane roughness rise rapidly. When the additive amount was 3.0 wt %, the roughness of the M3 membrane was 45.63 ± 2.07 nm. With the concentration of MoS 2 @PDA increased to 4.5 wt %, the membrane roughness became 48.8 ± 1.27 nm. The membrane surface became rougher. This was related to the hydrophilic MoS 2 @PDA migration to the direction of coagulation bath in the phase inversion process. The increased of membrane surface roughness was consistent with the observation result of SEM.
Membranes 2021, 11, x FOR PEER REVIEW 11 of 20 MoS2@PDA in the casting solution accelerated the solvent-nonsolvent exchange and instantaneous liquid-liquid separation in the casting solution, resulting in the formation of the microvoids structures observed in the M1 membrane [49]. However, the number of macropores was reduced at a higher additive concentration, and the internal structure was more compact, as shown in the cross-section of membranes M3 and M5. This was due to the fact that rheology played a leading role in the increase of the viscosity of the casting solution with the addition of MoS2@PDA, which reduced the phase conversion rate and densified the membrane structure [50]. The morphological characteristic of the cross-section was different from that of the MoS2/PES TUF membrane; the MoS2@PDA/PES TUF membrane displays fewer microvoids, and the supporting layer is dominated by a sponge structure. This provides the possibility of forming smaller MWCO membranes. The surface roughness of membranes was measured by AFM (Figure 7). Results were obtained by measuring three different points and indicated that the M0 membrane displayed a smooth surface with an average roughness (Ra) value of 1.73 ± 0.18 nm. The addition of MoS2@PDA made the membrane roughness rise rapidly. When the additive amount was 3.0 wt %, the roughness of the M3 membrane was 45.63 ± 2.07 nm. With the concentration of MoS2@PDA increased to 4.5 wt %, the membrane roughness became 48.8 ± 1.27 nm. The membrane surface became rougher. This was related to the hydrophilic MoS2@PDA migration to the direction of coagulation bath in the phase inversion process. The increased of membrane surface roughness was consistent with the observation result of SEM. The surface charge of membranes is a critical factor to affect the filtration performance of membranes. As shown in Figure 8, the pristine M0 membrane showed negative charges ranging from 3 to 11 as the pH values increased. These negative charges were believed to come from the functional groups of PES (O=S=O) and PVP (O=C-N) [51]. After the introduction of MoS2@PDA, the M3 membrane showed more negative charges than that of the pristine membrane at different pH values, which was caused by the electronegativity of MoS2@PDA.
The TGA analysis was conducted to investigate the thermal stability of the pristine membrane (M0) and the PES/MoS2@PDA membrane (M3). As can be seen from the TGA curves in Figure 8, rapid material losses were observed in both the M0 and M3 membrane as the temperature increased from 400 to 600 °C, which was due to the polymer decomposition of PES [52]. When the analysis temperature reached 1000 °C, the weight loss of the M0 membrane was 81.73%, which was larger than the weight loss of the M3 membrane (75.51%). This indicated that the thermal stability of the membrane was improved after the addition of MoS2@PDA. This was due to the interaction between the MoS2@PDA nanomaterial and PES, which increased the rigidity of the polymer chain and the fracture energy of the polymer chain. The surface charge of membranes is a critical factor to affect the filtration performance of membranes. As shown in Figure 8, the pristine M0 membrane showed negative charges ranging from 3 to 11 as the pH values increased. These negative charges were believed to come from the functional groups of PES (O=S=O) and PVP (O=C-N) [51]. After the introduction of MoS 2 @PDA, the M3 membrane showed more negative charges than that of the pristine membrane at different pH values, which was caused by the electronegativity of MoS 2 @PDA. The TGA analysis was conducted to investigate the thermal stability of the pristine membrane (M0) and the PES/MoS 2 @PDA membrane (M3). As can be seen from the TGA curves in Figure 8, rapid material losses were observed in both the M0 and M3 membrane as the temperature increased from 400 to 600 • C, which was due to the polymer decomposition of PES [52]. When the analysis temperature reached 1000 • C, the weight loss of the M0 membrane was 81.73%, which was larger than the weight loss of the M3 membrane (75.51%). This indicated that the thermal stability of the membrane was improved after the addition of MoS 2 @PDA. This was due to the interaction between the MoS 2 @PDA nanomaterial and PES, which increased the rigidity of the polymer chain and the fracture energy of the polymer chain. The major reason for this trend was that with the addition of nanoparticles, the adsorption between the exposed hydroxyl group on the surface of MoS 2 @PDA and the polymer chain increased the viscosity of the casting solution, resulting in delayed phase separation [53,54]. It was beneficial to create a denser structure in MoS 2 @PDA/PES MMM. However, when the concentration of MoS 2 @PDA further increased, the MWCO increased instead, which was related to the agglomeration of nanoparticles on the membrane structure, and the defects deteriorate the PEG rejection. With the increase of MoS 2 @PDA concentration, the pure water flux of the membrane decreased from M0 53.67 ± 2.73 L m −2 h −1 bar −1 to M5 34.75 ± 1.5 L m −2 h −1 bar −1 . The reduced water flux was directly related to the decreased porosity with the increasing of MoS 2 @PDA in membranes. With the increase of MoS 2 @PDA loading, the porosity decreased gradually from 78.50 ± 0.42% for the M0 membrane to 74.02 ± 0.65% for the M5 membrane (Figure 9b). Moreover, compared to the M0 membrane, the MoS 2 @PDA blended TUF membrane has smaller pore sizes and MWCO values (Figure 9c). The average pore size of the membrane reduced from 2.97 nm for the M0 membrane to 2.62 nm for the M3 membrane. As a result of its smallest MWCO value, the M3 membrane was selected for further evaluation of dye separation performance.  Dye removal experiments were conducted by using three kinds of dyes solution (C2B, RB4, and JGB) with different concentrations (10, 50, and 100 mgL −1 ). Figure 10 shows the retention performance of different dyes. Figure 10a,c,d shows the separation effect of M3 on 50 mgL −1 dye, displaying the UV-visible spectra of the dye solution before and after the filtration. The obvious difference of colors of the feed solution and permeate solution indicated the high dye rejection performance of the M3 membranes. Moreover, compared with the pristine M0 membrane, the M3 membrane showed better dye retention effects to those dyes. The RB4 and JGB selectivity performances of the M3 membrane were stable even when the dye concentration increased from 10 to 100 mgL −1 . However, for the C2B solution, as the dye concentration increased, the membrane retention rate fell from 94.17% to 78.43%. This may be due to the increase of impurities with the increase of dye concentration. The addition of these ions enhanced electrostatic shielding, thus reducing the rejection of the dye [55]. The zeta potential characterization of the membrane is shown in Figure 8a. With t increase of pH, the negative charge on the membrane surface was enhanced, and the z potential of the M3 membrane was always lower than that of the M0 membrane. T lower zeta potential for the M3 membrane was more conducive to the adsorption of p itively charged dye JGB. Due to the concentration polarization on the membrane surfa the dye retention rate would increase with the increase of pH (Figure 11a) [56]. T membrane's interception rate of JGB decreased with the extension of filtration time in longer period of filtration ( Figure 11b). As the separation time was extended to 180 m The zeta potential characterization of the membrane is shown in Figure 8a. With the increase of pH, the negative charge on the membrane surface was enhanced, and the zeta potential of the M3 membrane was always lower than that of the M0 membrane. The lower zeta potential for the M3 membrane was more conducive to the adsorption of positively charged dye JGB. Due to the concentration polarization on the membrane surface, the dye retention rate would increase with the increase of pH (Figure 11a) [56]. The membrane's interception rate of JGB decreased with the extension of filtration time in a longer period of filtration ( Figure 11b). As the separation time was extended to 180 min, the JGB rejection of the M3 membrane was 88.61%, while the JGB rejection of the M0 membrane fell to 77.59%, indicating the good stability of the M3 membrane. Figure 11c shows the water flux of the M0 and M3 membranes during the long filtration test. Compared to the M0 membrane, the M3 membrane showed a low water flux reduction rate, which further confirmed the stable dye filtration performance of the M3 membrane. In future work, we will select more types of pollutants with different sizes and involve membrane cleaning cycles to investigate the stability and the service life of the membrane. the JGB rejection of the M3 membrane was 88.61%, while the JGB rejection of the M0 membrane fell to 77.59%, indicating the good stability of the M3 membrane. Figure 11c shows the water flux of the M0 and M3 membranes during the long filtration test. Compared to the M0 membrane, the M3 membrane showed a low water flux reduction rate, which further confirmed the stable dye filtration performance of the M3 membrane. In future work, we will select more types of pollutants with different sizes and involve membrane cleaning cycles to investigate the stability and the service life of the membrane. The high salinity in the textile wastewater affects the retention of dyes by the membrane. Passing the salt through the membrane is a possible strategy to separate dye and salt in one step. This will save costs for subsequent recovery of dye and salt [57]. The filtration performance of the pristine membrane and MoS2@PDA-modified membrane was further evaluated with different salts (shown in Figure 12). As expected, the rejection of salts was inefficient, showing 0.76%, 10.64%, 4.31%, and 1.99% for NaCl, Na2SO4, MgSO4, and MgCl2, respectively. Accordingly, because M0 and M3 were negatively charged, the low salt rejection could be ascribed to the membrane screening effect. The hydration radius of salt ions was much smaller than the membrane pore size (Na + : 0.36nm, Mg 2+ : 0.43 nm, SO4 2− : 0.38 nm, Cl − : 0.33 nm) [58]. The small-sized ions could be allowed to pass through the membrane pores easily [59]. Figure 13 compared the water flux of the M3 membrane and other TUF membranes with similar MWCO in previous studies. The result indicated that the M3 membrane The high salinity in the textile wastewater affects the retention of dyes by the membrane. Passing the salt through the membrane is a possible strategy to separate dye and salt in one step. This will save costs for subsequent recovery of dye and salt [57]. The filtration performance of the pristine membrane and MoS 2 @PDA-modified membrane was further evaluated with different salts (shown in Figure 12). As expected, the rejection of salts was inefficient, showing 0.76%, 10.64%, 4.31%, and 1.99% for NaCl, Na 2 SO 4 , MgSO 4 , and MgCl 2 , respectively. Accordingly, because M0 and M3 were negatively charged, the low salt rejection could be ascribed to the membrane screening effect. The hydration radius of salt ions was much smaller than the membrane pore size (Na + : 0.36 nm, Mg 2+ : 0.43 nm, SO 4 2− : 0.38 nm, Cl − : 0.33 nm) [58]. The small-sized ions could be allowed to pass through the membrane pores easily [59].
Membranes 2021, 11, x FOR PEER REVIEW 16 of 20 manufactured in this study had a higher water flux than other TUF membranes with similar MWCO values. All these further revealed that the MoS2@PDA-modified TUF has a high feasibility to be applied for dye separation.

Conclusions
This study developed a one-step phase inversion approach to fabricate MoS2@PDA-modified TUF membranes. Improvements of MoS2@PDA on the properties and filtration performance were investigated by varied membrane characterization and performance evaluation. The blending of MoS2@PDA into PES membranes results in a denser membrane structure and enhanced thermal stability and electronegativity. In addition, TUF membranes with different MWCO were obtained by controlling the loading of MoS2@PDA. The membrane with 3.0 wt % MoS2@PDA had a lowest MWCO (5589 Da), showed a good pure water permeation flux as 42.0 Lm −2 h −1 bar −1 , and had a high rejection of dye JBG up to 99.88% at pH 5. The salt ions removal rate of membrane M3 was mostly lower than 10%, indicating that the membrane provided free salt ions permeation. The  Figure 13 compared the water flux of the M3 membrane and other TUF membranes with similar MWCO in previous studies. The result indicated that the M3 membrane manufactured in this study had a higher water flux than other TUF membranes with similar MWCO values. All these further revealed that the MoS 2 @PDA-modified TUF has a high feasibility to be applied for dye separation.
Membranes 2021, 11, x FOR PEER REVIEW 16 of 20 manufactured in this study had a higher water flux than other TUF membranes with similar MWCO values. All these further revealed that the MoS2@PDA-modified TUF has a high feasibility to be applied for dye separation.

Conclusions
This study developed a one-step phase inversion approach to fabricate MoS2@PDA-modified TUF membranes. Improvements of MoS2@PDA on the properties and filtration performance were investigated by varied membrane characterization and performance evaluation. The blending of MoS2@PDA into PES membranes results in a denser membrane structure and enhanced thermal stability and electronegativity. In addition, TUF membranes with different MWCO were obtained by controlling the loading of MoS2@PDA. The membrane with 3.0 wt % MoS2@PDA had a lowest MWCO (5589 Da), showed a good pure water permeation flux as 42.0 Lm −2 h −1 bar −1 , and had a high rejection of dye JBG up to 99.88% at pH 5. The salt ions removal rate of membrane M3 was mostly lower than 10%, indicating that the membrane provided free salt ions permeation. The  [60][61][62], commercial TUF membranes reported in the literature [63,64], ceramic TUF membranes reported in the literature [65,66] and a MoS 2 @PDA/PES mixed matrix TUF membrane in this work are compared.

Conclusions
This study developed a one-step phase inversion approach to fabricate MoS 2 @PDAmodified TUF membranes. Improvements of MoS 2 @PDA on the properties and filtration performance were investigated by varied membrane characterization and performance evaluation. The blending of MoS 2 @PDA into PES membranes results in a denser membrane structure and enhanced thermal stability and electronegativity. In addition, TUF membranes with different MWCO were obtained by controlling the loading of MoS 2 @PDA. The membrane with 3.0 wt % MoS 2 @PDA had a lowest MWCO (5589 Da), showed a good pure water permeation flux as 42.0 Lm −2 h −1 bar −1 , and had a high rejection of dye JBG up to 99.88% at pH 5. The salt ions removal rate of membrane M3 was mostly lower than 10%, indicating that the membrane provided free salt ions permeation. The MoS 2 @PDA/PES membrane prepared in this work has the potential to efficiently separate and recover dye and salt.