English

• Membrane technology has received significant attention as an effective method for water purification, addressing concerns related to water pollution and scarcity [1]. Nanofiltration (NF) membranes, known for their low operating pressure and low maintenance costs [2,3], have found wide applications in desalination, wastewater treatment, and water reuse, due to their capability to separate different valent salts selectively and reject high molecular weight (M_w) organics [4,5]. However, despite their many benefits, conventional thin-film composite (TFC) NF membranes often suffer from a "trade-off" phenomenon in which permeability and selectivity are inversely related [4,6,7]. Membrane synthesis conditions and modifications play essential roles in determining membrane separation performance due to their effect on monomer diffusivity and reaction rate [8]. For example, increasing the piperazine (PIP) concentration from 0.15 wt% to 2.0 wt% leads to an increase of xylose rejection from 57.3% to 78.9%, though the water permeation flux decreases from 5.4 L m^-2 h^-1 bar^-1 to 4.2 L m^-2 h^-1 bar^-1 [9]. Therefore, developing novel NF membranes that maintain selectivity while increasing permeability is crucial.

  One promising strategy to address the permeability-selectivity trade-off is the integration of nanomaterials into the polyamide (PA) layer, forming thin-film nanocomposite (TFN) membranes. Various porous and non-porous nanomaterials, such as silica [10,11], TiO₂ [12,13], carbon nanotubes (CNTs) [14,15], graphene oxide (GO) [16-18], and metal-organic frameworks (MOFs) [19,20], have been employed in the fabrication of TFN membranes. This strategy aims to enhance permeability and selectivity, improving the efficiency of processes like desalination, wastewater treatment, and gas separation [21-23]. The interfacial polymerization (IP) technique, mixing nanomaterials within either the aqueous or organic phase, is commonly employed to create TFN membranes (i.e., mixed matrix membranes). Introducing nanomaterials into the PA matrix forms additional selective nanochannels for water transport, facilitating a faster path of water molecules than through the PA matrix alone [24-26]. However, despite the beneficial properties of nanomaterials in enhancing membrane permeability, their poor compatibility with the PA layer may lead to the formation of nanosized gaps at the interface. These gaps compromise the integrity of the selective layer, thereby significantly reducing salt rejection of the NF membranes. Furthermore, these nanomaterials also profoundly affect the morphology and hydrophilicity of the PA layer, ultimately influencing the separation performance of the TFN membranes [27-29]. Previous studies have demonstrated significant enhancements in water flux while keeping high salt rejection by incorporating nanomaterials into TFN membranes. For example, Rajaeian et al. achieved an increase in pure water flux from 11.2 L m^-2 h^-1 traditional TFC membrane to approximately 20.5 L m^-2 h^-1 by adding 0.05 wt% of modified TiO₂ to TFN membranes without excessively sacrificing the rejection of NaCl [30].

  Nonetheless, nanomaterial aggregation often hinders the successful integration of nanomaterials into TFN membranes, which can deteriorate performance and impede their practical use [24,31,32]. The aggregation of nanomaterials, especially nanosheets, can reduce the water permeance and break the continuity of the PA matrix [24,33,34], leading to non-selective voids and defects that severely decrease salt rejection [28]. To address this challenge, previous studies have primarily focused on modifying nanosheets to enhance their compatibility with the PA layer. While some functionalized and PA-compatible nanosheets have shown promise in alleviating the aggregation impact in TFN membranes [24,27,35], there have been inconsistencies in separation performance [24,31,33], indicating that compatibility between nanosheets and the PA layer may not be the sole factor determining the occurrence of nanosheet aggregation in TFN membranes. Other underlying mechanisms may exist.

  The substrate's characteristics play a crucial role in the structure and separation performance of TFC membranes. Key factors, such as substrate hydrophilicity and pore size, can directly influence the IP process, affecting the PA layer's thickness and tightness, ultimately determining the membrane's permeability and selectivity [36]. It is observed that TFC membranes exhibit improved water permeability when formed on relatively hydrophobic substrates with larger pore sizes [37]. Additionally, Peng et al. highlight that substrates played an essential role in amine monomer storage during the formation of TFC membranes [38]. For TFN membranes, the substrate serves as a reservoir for amine monomers and nanomaterials during the IP reaction. When a nanomaterial-containing monomer solution is spread on the substrate, the distribution and aggregation states of the nanomaterial are influenced by the substrate's characteristics, especially its pore size. The nanomaterial-containing monomer solution can infiltrate the pores in the substrate and form an ultra-thin hydration layer on the substrate's surface [36,38]. The hydration layer may fill the membrane's pores, affecting the distribution of the nanosheets in the membrane, as the size of the pores determines whether and how the nanosheets can enter the substrate. Meanwhile, water molecules in the hydration layer will form hydration with functional groups on the surface of the nanosheet, which may change the surface properties of the nanosheet and thus affect its distribution in the membranes. Notably, when the nanomaterial content within the ultra-thin hydration layer on the substrate's surface exceeds a certain threshold, their dispersibility decreases, causing aggregation. Although this hypothesis suggests the role of substrate pore size in controlling nanomaterial distribution and aggregation, direct evidence that confirms this relationship is currently lacking.

  This study explores the influence of substrate pores' relative size with nanosheets on the distribution and aggregation of nanosheets within TFN membranes. Unlike earlier studies that may have focused on a single type of nanomaterial or a specific membrane configuration, our research systematically varies both the size of nanosheets and the substrate pore sizes. This approach allows for a comprehensive understanding of the interplay between these factors, offering a more nuanced perspective on minimizing aggregation and enhancing membrane performance. MoS₂ nanosheets are ideal for filtration and separation applications, and they have been chosen as a model nanosheet to disperse in an aqueous phase. Their smooth surface, without interfering functional groups, reduces resistance to water molecule transport within the TFN membranes, enhancing water flux compared to other materials with irregularities or functional groups that increase resistance. The hydrophilic nature and excellent dispersity enable uniform dispersion in the aqueous phase during the IP process and prevent nanoparticle aggregation issues that could affect membrane performance. Additionally, MoS₂ nanosheets are suitable for high-resolution qualitative characterization techniques such as TEM, enabling precise analysis of their distribution within the TFN membranes. Moreover, their chemical composition allows accurate quantitative determination after H₂O₂ digestion to maintain the integrity of the PA layer, which is advantageous for post-fabrication analysis and optimizing the membrane's structural properties. The impact of MoS₂ nanosheets on the membrane surface morphology and separation performance was thoroughly investigated, and the occurrence of aggregation was inferred from declines in permeance and rejection. By deliberately varying MoS₂ nanosheet sizes and substrate pore sizes, nanosheets' distribution behavior and aggregation tendencies within TFN membranes can be elucidated. Overall, this investigation bridges the research gap by providing insights into the interplay between nanosheet properties, substrate characteristics, and aggregation phenomena in TFN membranes.

  MoS₂ nanosheets were synthesized by chemically exfoliating commercial MoS₂ powder (Text S1 in Supporting information), aligning with methodologies from prior research [39]. Different amounts (0.5, 1, 2, 3, and 5 mg) of MoS₂ nanosheets (25 mL) were mixed with 25 mL of 1.0 wt% PIP/H₂O solution to fabricate MoS₂-incorporated TFN membranes via the IP reaction with 0.15 wt% trimethyl chloride (TMC) hexane solution (Fig. 1, more details in Text S1c). The prepared membranes were designated as TFN-0, TFN-0.5, TFN-1, TFN-2, TFN-3, and TFN-5, depending on the amount of MoS₂ added (0, 0.5, 1, 2, 3, and 5 mg, respectively). The scanning electron microscope (SEM) top view and the atomic force microscope (AFM) surface scans of the UP150 substrate revealed a relatively smooth surface (Fig. S1 in Supporting information). As shown in Fig. 2a, the control TFC membrane (TFN-0) demonstrated a nodular pattern on its surface, characteristic of the PA surface formed through the IP reaction between PIP and TMC [40]. However, when incorporating MoS₂, TFN membranes exhibited surfaces featuring a combination of nodule-like and ridge-like heterogeneous formations (Fig. 2a). As the loading of MoS₂ nanosheets increased, the prominence of ridge-like structures in MoS₂-incorporated TFN membranes also increased, suggesting that MoS₂ nanosheets enhanced the surface roughness. It can be inferred that the presence of MoS₂ nanosheets generates an irregular interface for the IP reaction on the substrate, resulting in a distinct morphology of the reaction interface elsewhere [41-43]. This variation likely arises from the non-uniform reaction interface driven by MoS₂, evolving into different ridged nanostructures varying in size and configuration from the initial template during thermal processing. The effect of MoS₂ incorporation on surface roughness was also probed via AFM scanning, detailing surface characteristics (R_a: average roughness; R_q: root mean square roughness; R_max: maximum vertical distance) presented in Table S1 (Supporting information). Fig. 2b and Table Sl depict semi-quantitative AFM analysis revealing a significant increase in the R_max value from 34.3 ± 1.0 nm for the TFN-0 membrane to 45.5–92.4 nm for the MoS₂-incorporated TFN membranes. The AFM micrographs of TFN-3 and TFN-5 membranes presented more obvious ridge-like structures than TFN-0.5, TFN-1, and TFN-2 membranes. This observation may be related to the higher loading and likely aggregation of MoS₂ nanosheets on the substrate surface, creating a more uneven interface for the IP reaction, likely contributing to the enhanced surface roughness observed (Fig. 2b and Table S1).

  Figure 1

  

  Figure 1.  The schematic illustration of MoS₂-incorporated TFN membrane fabrication. (a) Different amounts (0.5, 1, 2, 3, and 5 mg) of MoS₂ nanosheets were mixed with PIP/H₂O solution and then sonicated for 30 min to ensure an even distribution; (b) the mixed solution was applied to the substrate; (c) TMC/hexane solution was added to the MoS₂/PIP-treat substrate; (d) the formation of MoS₂-incorporated PA layer.

  DownLoad: Full-Size Img PowerPoint

  Figure 2

  

  Figure 2.  The morphological characterization of membranes. (a) The SEM and (b) AFM images of the surface of the control TFC membrane and MoS₂-incorporated TFN membranes fabricated on UP150 substrates.

  DownLoad: Full-Size Img PowerPoint

  Based on X-ray photoelectron spectroscopy (XPS) characterization (Text S1d), the O/N ratios in the MoS₂-incorporated TFN membranes were higher than those of the control TFC membrane (Fig. S2a and Table S2 in Supporting information), revealing a reduced cross-linking degree in the former. This decrease in cross-linking degree can be attributed to the hydrophilic interaction between PIP monomers and introduced hydrophilic MoS₂ nanosheets during the IP reaction, which may have limited the diffusion of amine monomer and influenced the synthesis of PA films [44-46]. Additionally, the incorporation of hydrophilic MoS₂ nanosheets also enhanced the membrane's surface hydrophilicity (Fig. S2b in Supporting information). The increased surface roughness of MoS₂-incorporated TFN membranes also decreased the apparent water contact angle [38]. Compared to the control TFC membrane, the MoS₂-incorporated TFN membranes exhibited rougher, more hydrophilic, and less cross-linked PA films, which may affect their separation performance.

  Fig. 3a illustrates the relationship between water permeance and Na₂SO₄ rejection for the control TFC membrane and five MoS₂-incorporated TFN membranes. Overall, the MoS₂-incorporated TFN membranes exhibited enhanced water permeance while maintaining high salt rejection, with all membranes achieving over 95% rejection for Na₂SO₄. The MoS₂-incorporated TFN membranes showed decreased cross-linking degree compared to the control TFC membrane (Table S2 and Fig. S2a), typically resulting in a relatively loose PA layer and reducing the size exclusion effect. However, the decrease of cross-linking degree could increase negatively-charged carboxylic groups derived from hydrolysis of acryl chloride groups, leading to a more negatively-charged membrane surface and then enhancing the rejection of SO₄^2-, finally keeping similar salt rejection with the control TFC membrane. Specifically, TFN-0.5 and TFN-1 membranes showed elevated water permeance relative to the TFN-0 membrane (8.9 L m^-2 h^-1 bar^-1), reaching 14.8 and 17.6 L m^-2 h^-1 bar^-1, respectively. Furthermore, the TFN-2 membrane further increased pure water permeance to 20.5 L m^-2 h^-1 bar^-1, approximately 2.3 times higher than the TFN-0 membrane's permeance rate. However, when MoS₂ loading exceeded optimal levels, water permeance decreased to 17.9 and 14.1 L m^-2 h^-1 bar^-1 for TFN-3 and TFN-5 membranes, respectively. This decrease in permeance of the MoS₂-incorporated TFN membranes beyond the optimal loading is consistent with the parabolic variation observed in other TFN membranes previously reported [28,31,47]. Although MoS₂ nanosheets at lower concentrations have been shown to increase TFN membrane permeability, it is observed that the water permeability decreased when MoS₂ nanosheets were overdosed (i.e., TFN-3 and TFN-5).

  Figure 3

  

  Figure 3.  Performance analysis of the MoS₂-incorporated TFN membranes. (a) A comparison of water permeance and Na₂SO₄ rejection tested with 1000 mg/L of Na₂SO₄ at 5 bar. (b) Comparison of salt rejection for TFN-2 membrane using different salts, namely, Na₂SO₄, MgSO₄, MgCl₂, CaCl₂, or NaCl at 5 bar. (c) Long-term filtration performance of TFN-2 membrane for Na₂SO₄ rejection at 5 bar. (d) A comparison of water permeance and Na₂SO₄ rejection of the optimal MoS₂-incorporated TFN membrane (TFN-2) with other nanomaterials-incorporated TFN membranes is discussed in Table S3.

  DownLoad: Full-Size Img PowerPoint

  The MoS₂-incorporated membrane (TFN-2) exhibited high rejections for Na₂SO₄ and MgSO₄ (Fig. 3b, Fig. S3 and Table S3 in Supporting information) due to the steric hindrance effects and the strong repulsion between the negatively charged membrane surfaces (Fig. S4 in Supporting information) and SO₄²⁻ ions, which was weakened against the monovalent anion Cl⁻, resulting in lower rejections for MgCl₂, CaCl₂, and NaCl. Notably, the MoS₂-incorporated membranes demonstrated excellent selectivity between monovalent and divalent ions, with rejection as low as 29.6% for NaCl. Beyond Donnan repulsion, size exclusion also impacts the salt rejection of TFC membranes [48]. Given the larger hydrated ionic radius of Mg²⁺ (0.428 nm) over Ca²⁺ (0.412 nm) [49], the MoS₂-incorporated membranes could achieve a slightly higher rejection of MgCl₂ than CaCl₂. For water treatment applications, robust membrane stability is crucial, especially for the TFN membranes that face risks such as membrane deformation or nanomaterial leaching. Fig. 3c depicts the durable performance of the MoS₂-incorporated membrane over a two-day continuous filtration trial, maintaining Na₂SO₄ rejection above 97%. Despite a minor decrease in the water permeance due to the concentration polarization, it remained significantly elevated at approximately 20.1 L m^-2 h^-1 bar^-1 compared to the TFC membrane. The steady membrane permeability and salt rejection over the test period demonstrate the TFN membrane's long-term operation stability, ascribed to a balance between concentration polarization effects and the cross-flow feedwater.

  The remarkable enhancement in water permeance observed in MoS₂-incorporated TFN membranes is primarily due to the formation of additional interfacial nanochannels facilitated by adding MoS₂ nanosheets within the PA matrix. These nanochannels expedite the passage of water molecules on the nanosheet surface compared to the PA matrix alone [29]. Notably, the MoS₂-incorporated TFN-2 membrane exhibited even greater water permeance and selective separation factor than other recently reported TFN membranes containing 1D CNT or 3D zeolite particles (Fig. 3d and Table S4 in Supporting information). Such an enhancement in permselectivity is linked to the MoS₂ nanosheets' larger surface area than nanotubes and nanoparticles of the same mass (Fig. S5 in Supporting information). The increased surface area contributed to the formation of more interfacial nanochannels. Additionally, the smooth surface of MoS₂ nanosheets and the inner-layer channels formed between them may facilitate ultrafast water transport within the TFN membrane, thereby enhancing water permeability [50-52]. As such, the MoS₂-incorporated TFN membrane represents a significant advancement with substantial promise for future applications in water treatment.

  The reduction in water permeability of TFN membranes is a common phenomenon typically attributed to the formation of impermeable inorganic aggregates caused by compatibility issues [28,31,47]. However, in the case of MoS₂ nanosheets, nanomaterial aggregation was not observed in the PIP solution even at the highest loading of 5 mg MoS₂ in the TFN-5, as evidenced by the constant particle size of the nanosheets in the PIP solution during 30 min (Fig. S6 in Supporting information). Nevertheless, aggregates were visibly observed on the substrate surface after applying the MoS₂-enriched PIP solution during the IP process (Figs. S7 and S8 in Supporting information). The substrate acts as a depository for amine monomers (PIP) during the IP reaction [38], and for MoS₂ nanosheets. However, the pore size of the substrate determines whether MoS₂ nanosheets can enter the pores or remain distributed on the substrate surface. When the concentration of MoS₂ nanosheets within the ultra-thin hydration layer on the substrate's surface surpasses a critical threshold, the nanosheets may interact more strongly with each other rather than with the surrounding medium. This leads to reduced dispersibility and increased likelihood of aggregation. Aggregation occurs because the nanosheets, at higher concentrations, may experience stronger van der Waals forces or other intermolecular interactions that cause them to clump together. This aggregation can negatively impact the uniformity of the membrane, potentially creating defects or inhomogeneities that compromise membrane permeability and selectivity. Meanwhile, larger substrate pores may allow MoS₂ nanosheets to enter the pores more readily, possibly decreasing their effective concentration on the substrate surface and minimizing aggregation. Therefore, it is conjectured that the aggregation of MoS₂ nanosheets is contingent upon the pore size of the substrate and particle size of the nanosheets.

  To investigate the influence of substrate pore size and particle size on the distribution of the nanosheets, a second type of PES membrane (MF0105, pore size = 100 nm) was selected as substrate in comparison with the UP150 membrane. Surface analysis of these PES substrates through SEM and AFM showed their smooth surfaces with comparable roughness (Figs. S1 and S9 in Supporting information). The MF0105 substrate exhibited higher water permeability and MWCO than the UP150 substrate due to its larger surface pore size (Fig. S10a in Supporting information). Water contact angle measurements showed no significant difference between the UP150 substrate (66.6° ± 1.56°) and the MF0105 substrate (67.8° ± 1.79°) (Fig. S10b in Supporting information). The pore size distribution of the UP150 substrate is predominantly concentrated within the range of 50–55 nm (> 83%, Fig. S10c in Supporting information), while for the MF0105 substrate, it falls within the range of 95–105 nm (> 93%, Fig. S10d in Supporting information). For comparison, smaller MoS₂ nanosheets were also synthesized by increasing the temperature in the chemical exfoliation process. The DLS data (Fig. S11 in Supporting information) showed that the exfoliated MoS₂ nanosheets at 200 ℃ were only tens of nanometers, in sharp contrast to the approximately 200-nm-sized MoS₂ nanosheets exfoliated at 25 ℃. TEM and AFM imaging (Fig. S12 in Supporting information) confirmed that both types of MoS₂ nanosheets exhibited similar flake-like structures with an average thickness of 2.5 nm. Still, different lateral dimensions of approximately 150–400 nm and 20–50 nm for MoS₂ nanosheets exfoliated at 25 ℃ and 200 ℃, respectively. The phase compositions, as determined by Raman and Mo 3d XPS spectra (Fig. S13 in Supporting information), demonstrated the successful preparation of two MoS₂ nanosheets with different sizes but consistent physical-chemical properties.

  To describe the aggregation states of MoS₂ nanosheets and determine the actual loading of Mo in TFN membranes, the MoS₂-incorporated TFN membranes underwent an etching experiment by soaking in 3% H₂O₂ solution at room temperature for one day (details in Text S1f). Fig. 4a illustrates the variations in the separation performance of TFN membranes incorporated with the larger MoS₂ nanosheets (prepared at 25 ℃) prepared on the substrates with smaller pores (UP150, ~50 nm) and larger pores (MF0105, ~100 nm) before and after H₂O₂ etching, which aimed to remove MoS₂ nanosheets or aggregates within the membrane (Text S1f). Initially, the TFN membranes with larger MoS₂ nanosheets on the small-pore-size substrate exhibited enhanced water permeability. However, as the loading of MoS₂ nanosheets increased, the water permeability decreased. After etching, all TFN membranes had an upsurge in water permeance, while Na₂SO₄ rejection of all TFN membranes exhibited varying degrees of decline (Fig. 4a). For instance, the TFN-2 and TFN-5 membranes showed 20.9% and 35.5% decline in Na₂SO₄ rejection, respectively. In comparison, the TFN-2 membrane incorporated with larger MoS₂ nanosheets on the larger pore size substrates (MF0105) still demonstrated the optimal separation performance, recording a water permeance of 24.07 L m^-2 h^-1 bar^-1 and Na₂SO₄ rejection of 97.2%. Only TFN-3 and TFN-5 membranes showed a decline in salt rejection after etching (Fig. 4b).

  Figure 4

  

  Figure 4.  Comparison of the separation performance of MoS₂-incorporated TFN membranes with various-sized nanosheets on different pore-size substrates before (B/F) and after (A/F) H₂O₂ etching. MoS₂-incorporated TFN membranes were prepared on small pore size substrates (UP150) with large (a) and small (b) MoS₂ nanosheets exfoliated at 25 ℃ and 200 ℃, respectively, and on large pore size substrates (MF0105) with large (c) and small (d) MoS₂ nanosheets.

  DownLoad: Full-Size Img PowerPoint

  Regarding TFN membranes incorporated with the smaller MoS₂ nanosheets (exfoliated at 200 ℃), water permeability improved with increased MoS₂ content, though the peak permeability was not achieved (Figs. 4c and d). However, these membranes still showed greater water permeance than those with larger MoS₂ nanosheets. After etching, a slight rise in water permeance was seen across all the membranes. Notably, the salt rejection for membranes incorporated with smaller MoS₂ nanosheets did not decrease after etching and remained at 96.5% (Figs. 4c and d). This suggests that introducing nanocavities, resulting from the etching of dispersed MoS₂ nanosheets within the PA layer, enhanced water permeance without sacrificing the salt rejection capability (Figs. 4c and d) [53,54]. Conversely, TFN membranes incorporated with larger MoS₂ nanosheets showed a decline in rejection after etching (Figs. 4a and b), indicating that the etching of aggregates may have resulted in defects in the PA layer. Interestingly, utilizing substrates with larger pores for fabricating TFN membranes incorporated with the MoS₂ nanosheets appeared to alleviate the impact of aggregation, as demonstrated in TFN-3 and TFN-5 (Fig. 4d). It is worth noting that the etched TFN membranes incorporated with smaller MoS₂ nanosheets (Figs. 4c and d) exhibited enhanced separation performance without the decline of Na₂SO₄ rejection, which can be considered as a potential area for further research.

  To investigate whether each TFN membrane can achieve enhanced separation performance whenever the size of the nanosheets is smaller than the pore size of the substrate membrane, GO nanosheets and graphene oxide quantum dots (GOQDs) were purchased to fabricate TFN membranes. GO nanosheets were about 200 nm, and GOQDs were approximately 10 nm in size (Fig. S14 in Supporting information). GO-incorporated and GOQDs-incorporated TFN membranes were prepared on the substrates with smaller pores (UP150, ~50 nm) and larger pores (MF0105, ~200 nm), and their membrane separation was tested under the same condition as MoS₂-incorporated TFN membranes (Fig. S15 in Supporting information). Consistent with larger-sized MoS₂-incorporated TFN membranes, the GO-incorporated TFN membranes on the substrates with larger pore sizes exhibited higher water permeance than those on the substrate with smaller pore sizes (Figs. S15a and b). As the loading of GO nanosheets increased, the water permeability of GO-incorporated TFN membranes on the substrates with smaller and larger pores increased and then decreased. Consistent with the phenomenon observed in larger-sized MoS₂-incorporated TFN membranes, the reduction in water permeability of the GO-incorporated TFN membranes may be attributed to the aggregation of nanosheets. Similarly, like smaller-sized MoS₂-incorporated TFN membranes, the water permeability of GOQDs-incorporated TFN membranes improved with increased GOQDs content, higher than GO-incorporated TFN membranes (Figs. S15c and d).

  The TEM cross-sectional views and HAADF images (Fig. 5 and Fig. S16 in Supporting information) displayed the independent existence of the substrate and the PA layer in TFN-5 membranes prepared with large and small MoS₂ nanosheets on smaller pore-sized (UP150) and larger pore-sized (MF0105) substrates. Based on the corresponding elemental map, the TFN-5 membrane prepared with large MoS₂ nanosheets exhibited the accumulation of Mo on the substrate (green part in Fig. 5a and Fig. S16a), implying the stacking or aggregation of MoS₂ nanosheets. However, the elemental map of Fig. 5b and Fig. S16b showed no dense distribution of Mo in the TFN-5 membrane prepared with small MoS₂ nanosheets, indicating their uniform distribution and minimal loading within the membrane. The actual amount of MoS₂ integrated into the TFN membrane was determined by measuring the Mo⁴⁺ concentration after H₂O₂ etching (Fig. S17 in Supporting information). TFN membranes with smaller MoS₂ nanosheets demonstrated a steady rise in the actual content of MoS₂, with those prepared on the substrates with larger pores (MF0105, ~200 nm) having slightly higher MoS₂ loading compared to those on the substrates with smaller pores (UP150, ~50 nm). This could be attributed to the smaller MoS₂ nanosheets being well dispersed in the hydration layer and pores of UP150 and MF0105 substrates (Fig. 5b and Fig. S16b). Compared to the UP150 substrate, the MF0105 substrate's larger pores acted as a more excellent reservoir for amine monomers and MoS₂ nanosheets (Fig. S16a). A significant increase in Mo concentration was observed for the TFN membranes incorporated with larger MoS₂ nanosheets, particularly between TFN-3 and TFN-5 (Fig. 5a, Figs. S16a and S17). The irregular growth indicated a change in the distribution of MoS₂ nanosheets, possibly due to MoS₂ aggregation [55]. Remarkably, despite the similar Mo loading observed for TFN membranes prepared on substrates with larger or smaller pores, such as TFN-2 and TFN-3 membranes (Fig. S17), they exhibited distinct impacts on the separation performance, particularly on rejection after etching (Fig. 4). This finding highlights the significant influence of substrate pore size and nanosheet size on aggregation, which helps address the decrease of water permeance in TFN membranes with high loading nanosheets.

  Figure 5

  

  Figure 5.  TEM cross-sectional micrographs and high angle angular dark field-scanning transmission electron microscopy (HAADF-STEM) images of TFN-5 membranes prepared with (a) large and (b) small MoS₂ nanosheets on smaller pore-sized substrates (UP150, ~50 nm), respectively. In HAADF-STEM images, molybdenum is shown in green.

  DownLoad: Full-Size Img PowerPoint

  To investigate the dispersion of MoS₂ nanosheets based on the membrane's pore size and the nanosheets' dimensions, mixed solutions of MoS₂ nanosheets and PIP solution with varying concentrations were deposited onto a water-impermeable plastic plate, UP150 substrate and MF0105 substrate (Text S1f). With increasing MoS₂ concentrations from top to bottom (Fig. 6), the color of the dots left on the plastic plate surfaces and substrates gradually deepened. Fig. 6a utilizes the water-impermeable plastic plate as a baseline for color intensity comparison. The intensity of the color served as an indication of the amount of MoS₂ nanosheets remaining on the surfaces. On the two PES substrates, the dots on the left side, representing the presence of large MoS₂ nanosheets, appeared darker than those on the right side (Figs. 6b and c). This suggests that the amount of the smaller MoS₂ nanosheets (exfoliated at 200 ℃) retained on UP150 and MF0105 substrates was significantly lower than that of the larger MoS₂ nanosheets (exfoliated at 25 ℃). This observation supports the speculation that a substantial portion of the small MoS₂ nanosheets exfoliated at 200 ℃ entered the membrane pores, especially for the substrate with a pore size of 100 nm (MF0105).

  Figure 6

  

  Figure 6.  Distribution of large and small MoS₂ nanosheets on different substrate surfaces: (a) Impermeable plastic plate, (b) UP150 substrate (small pore size), and (c) MF0105 substrate (large pore size). The left side of each image represents 10 µL droplets of larger MoS₂ nanosheets, while the right side shows 10 µL droplets of smaller MoS₂ nanosheets. From top to bottom, the concentrations of MoS₂ nanosheet in the droplets were 0.06 g/L, 0.1 g/L, and 0.2 g/L, respectively.

  DownLoad: Full-Size Img PowerPoint

  The influence of MoS₂ nanosheet sizes on their distribution location was further elucidated through cross-section micrographs and the corresponding element mapping on the UP150 substrate with smaller pores. In Fig. S18a (Supporting information), a clear boundary between larger MoS₂ nanosheets and UP150 substrate can be observed. Notably, the S and Mo mapping revealed a higher concentration of S and Mo at the top of the cross-section micrographs. Conversely, no distinct boundary is visible in the cross-section micrographs of smaller MoS₂ nanosheets on the UP150 substrate (Fig. S18b in Supporting information). Additionally, no significant enrichment of S and Mo elements at any position indicates that the smaller MoS₂ nanosheets predominantly entered the membrane pores. In contrast, the larger MoS₂ nanosheets, having an approximate size of ~200 nm and exceeding the ~50 nm pore size of the substrate, were retained on the membrane's exterior. When the substrate pores are more prominent, nanosheets more easily enter the pores, potentially decreasing the effective concentration of nanosheets on the substrate surface and contributing to reduced aggregation.

  Drawing from the separation performance observed in MoS₂-incorporated TFN membranes (Fig. 4), the actual Mo loading (Fig. 5), and the characterization of MoS₂ nanosheets on the different substrates (Fig. 6, Figs. S18 and S19 in Supporting information), a conceptual representation of the underlying mechanism is proposed in Fig. 6. When MoS₂ nanosheets are incorporated on substrates with a pore size smaller than the nanosheets, aggregation becomes more likely within the TFN membranes (Fig. 7 and Fig. S18a). The nanosheets cannot enter the substrate pores and fail to disperse effectively in the substrate's thin hydration layer, forming impermeable aggregates that reduce water permeance with higher concentrations of MoS₂ nanosheets. Moreover, etching away these aggregates disrupts the PA layer's continuity, leading to a decrease in salt rejection (Figs. 4a and b). In contrast, using a substrate with pore sizes exceeding that of the MoS₂ nanosheets' dimensions helps prevent aggregation (Fig. 7). The larger substrate pores allowed the smaller MoS₂ nanosheets to enter and disperse within the membrane, preventing significant aggregation. After etching, the continuity of the PA layer remained intact, and numerous nanocavities appeared within the PA layer. The nanocavities created by etching improved water permeability while preserving salt rejection (Figs. 4c and d). Overall, the substrate's pore size and the MoS₂ nanosheets' particle size play crucial roles in controlling aggregation and optimizing the separation performance of TFN membranes.

  Figure 7

  

  Figure 7.  The schematic illustration of the mechanisms of the MoS₂-incorporated TFN membranes before and after etching.

  DownLoad: Full-Size Img PowerPoint

  Various nanomaterials, ranging in sizes from tens of nanometers to a few micrometers, have been used to fabricate TFN membranes, including modified SiO₂ nanoparticles (100 nm) [56], MIL-101 (Cr) nanoparticles (47 nm) [57], aluminosilicate single-walled nanotubes (2.7–3.1 nm × 1.01–1.17 nm × 150 nm) [58], functionalized multi-walled carbon nanotubes (5 nm × 1.3–2.0 nm × 50 µm) [59], and GO nanosheets (500 nm–2 µm) [60]. These nanomaterials have been incorporated into PES or polysulfone UF or MF membranes with pore sizes of 100–220 nm [30,56,59]. However, the pore sizes of common UF and MF membranes are often smaller than the particle size of nanomaterials, which can lead to aggregation issues. Nanomaterial aggregation is a common issue in TFN membranes and negatively impacts separation performance. While previous studies have focused on enhancing the compatibility of nanomaterials with PA through nanomaterial modification [24,27,35], less attention has been given to the role of substrate pore size. This study highlights the significant influence of substrate pore size on the distribution and aggregation of MoS₂ nanosheets within TFN membranes. Notably, using substrates with larger pore sizes than the particle size of MoS₂ nanosheets can effectively mitigate aggregation issues. Therefore, when preparing TFN membranes by IP reaction, it is crucial to consider both nanosheet sizes and substrate pore sizes carefully. Optimizing substrate pore size and nanosheet size can effectively overcome the permeability-selectivity trade-off in composite membranes. These optimizations ultimately lead to improved membrane performance and reduced energy consumption during water treatment and desalination.

  This research emphasizes the pivotal influence of substrate pore size on controlling the distribution and aggregation of nanosheets within TFN membranes. By incorporating MoS₂ nanosheets with different sizes on substrates with varying pore sizes, it was observed that substrates with pore sizes smaller than the MoS₂ nanosheets tend to promote aggregation, leading to decreased water flux and salt rejection in TFN membranes. Conversely, using a substrate with pore sizes larger than MoS₂ nanosheet dimensions can effectively mitigate aggregation issues. Meanwhile, it increases the adequate loading capacity of MoS₂ nanosheets and thus enhances the water permeance and salt rejection of TFN membranes. Our study provides a detailed analysis of how different MoS₂ nanosheet sizes and substrate pore sizes influence aggregation tendencies, offering practical guidelines for reducing aggregation in future membrane designs. This focus on aggregation mechanisms adds a novel dimension to the research, distinguishing it from previous studies. It proposes that this approach is widely applicable and provides direction for developing novel high-performance TFN membranes, extendable to various nanosheets. By carefully selecting the appropriate combination of nanosheets and substrates, researchers can maximize the operative loading mass of nanosheets, make full use of their interfacial channels, and improve the overall performance and stability of TFN membranes for different separation applications. Moreover, further optimization of substrate hydrophilicity is significant for achieving excellent membrane separation performance, especially in terms of how the hydration layer and the polar groups on the hydrophilic substrate affect the aggregation behavior of the nanomaterials on the substrate. Future studies must also investigate a broader range of nanomaterials and substrates to establish general principles and recommendations for engineering TFN membranes with superior separation performance.

  Declaration of competing interest

  The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

  CRediT authorship contribution statement

  Siyu Cao: Writing – review & editing, Writing – original draft, Methodology, Investigation, Data curation. Yufei Shu: Methodology, Investigation, Data curation. Li Wang: Investigation, Data curation. Qi Han: Investigation, Data curation. Meng Zhang: Methodology, Investigation, Data curation. Mengxia Wang: Methodology, Investigation, Data curation. How Yong Ng: Writing – review & editing, Supervision, Conceptualization. Zhongying Wang: Writing – review & editing, Supervision, Conceptualization.

  Acknowledgments

  This work was financially supported by the National Natural Science Foundation of China (No. 22076075), Guangdong Provincial Key Laboratory of Soil and Groundwater Pollution Control (No. 2023B1212060002), the Key Program of Fundamental Research from the Shenzhen Science and Technology Innovation Commission (No. JCYJ20220818100218039) and the General Program of Fundamental Research from the Shenzhen Science and Technology Innovation Commission (No. JCY20230807092500001). The authors acknowledge the assistance of SUSTech Core Research Facilities.

  Supplementary materials

  Supplementary material associated with this article can be found, in the online version, at doi:10.1016/j.cclet.2024.110793.

  
  
• 1. [1]

     D.M. Warsinger, S. Chakraborty, E.W. Tow, et al., Prog. Polym. Sci. 81 (2018) 209–237.

  2. [2]

     Y.K. Ong, T.S. Chung, Environ. Sci. Technol. 48 (2014) 13933–13940. doi: 10.1021/es503258s

  3. [3]

     Y. Du, Y. Lv, W.Z. Qiu, et al., Chem. Commun. 52 (2016) 8589–8592.

  4. [4]

     H.B. Park, J. Kamcev, L.M. Robeson, et al., Science 356 (2017) eaab0530.

  5. [5]

     B.E. Logan, M. Elimelech, Nature 488 (2012) 313–319. doi: 10.1038/nature11477

  6. [6]

     R. Zhang, J. Tian, S. Gao, et al., J. Mater. Chem. A 8 (2020) 8831–8847. doi: 10.1039/d0ta02510k

  7. [7]

     H. Wu, B. Tang, P. Wu, J. Phys. Chem. C 114 (2010) 16395–16400. doi: 10.1021/jp107280m

  8. [8]

     Z. Yang, H. Guo, C.Y. Tang, J. Membr. Sci. 590 (2019) 117297.

  9. [9]

     Y. Liu, Y. Zhao, X. Wang, et al., J. Membr. Sci. 582 (2019) 274–283.

  10. [10]

      H. Abadikhah, E.N. Kalali, S. Behzadi, et al., Polymer 154 (2018) 200–209.

  11. [11]

      S. Islam, K. Touati, S. Rahaman, ACS EST Engin. 1 (2021) 467–477.

  12. [12]

      M. Ghanbari, D. Emadzadeh, W.J. Lau, et al., Desalination 371 (2015) 104–114.

  13. [13]

      T. Li, Y. Xiao, D. Guo, et al., J. Colloid Interface Sci. 572 (2020) 114–121. doi: 10.3390/pr8010114

  14. [14]

      S.M. Xue, Z.L. Xu, Y.J. Tang, et al., ACS Appl. Mater. Interfaces 8 (2016) 19135–19144. doi: 10.1021/acsami.6b05545

  15. [15]

      H. Wu, H. Sun, W. Hong, et al., ACS Appl. Mater. Interfaces 9 (2017) 32255–32263. doi: 10.1021/acsami.7b09680

  16. [16]

      S. Bano, A. Mahmood, S.-J. Kim, et al., J. Mater. Chem. A 3 (2015) 2065–2071.

  17. [17]

      W. Zhao, H. Liu, N. Meng, et al., J. Membr. Sci. 565 (2018) 380–389.

  18. [18]

      Y. Li, J. Jiao, Q. Wu, et al., Chin. Chem. Lett. 33 (2022) 5001–5012.

  19. [19]

      H. Liu, M. Zhang, H. Zhao, et al., RSC Adv. 10 (2020) 4045–4057. doi: 10.1039/c9ra09672h

  20. [20]

      H. Liu, B. Li, P. Zhao, et al., Chin. Chem. Lett. 34 (2023) 108369.

  21. [21]

      S.Y. Fang, P. Zhang, J.L. Gong, et al., Chem. Eng. J. 385 (2020) 123400.

  22. [22]

      D. Emadzadeh, W.J. Lau, M. Rahbari-Sisakht, et al., Desalination 368 (2015) 106–113.

  23. [23]

      Z. Zhang, J. Hu, S. Liu, et al., Chin. Chem. Lett. 32 (2021) 2882–2886.

  24. [24]

      S. Abdikheibari, W. Lei, L.F. Dumée, et al., J. Mater. Chem. A 6 (2018) 12066–12081. doi: 10.1039/c8ta03446j

  25. [25]

      S. Yang, Q. Jiang, K. Zhang, J. Membr. Sci. 604 (2020) 118052.

  26. [26]

      M.Q. Ma, C. Zhang, C.Y. Zhu, et al., J. Membr. Sci. 591 (2019) 117316.

  27. [27]

      X. Wu, R.W. Field, J.J. Wu, et al., J. Membr. Sci. 540 (2017) 251–260.

  28. [28]

      Q. Xie, W. Shao, S. Zhang, et al., RSC Adv. 7 (2017) 54898–54910.

  29. [29]

      Z. Yang, H. Guo, Z. Yao, et al., Environ. Sci. Technol. 53 (2019) 5301–5308. doi: 10.1021/acs.est.9b00473

  30. [30]

      B. Rajaeian, A. Rahimpour, M.O. Tade, et al., Desalination 313 (2013) 176–188.

  31. [31]

      S. Yang, K. Zhang, J. Membr. Sci. 595 (2020) 117526.

  32. [32]

      Y. Qin, H. Liu, Y. Liu, et al., J. Membr. Sci. 604 (2020) 118064.

  33. [33]

      M.E.A. Ali, L. Wang, X. Wang, et al., Desalination 386 (2016) 67–76.

  34. [34]

      H.R. Chae, C.H. Lee, P.K. Park, et al., J. Membr. Sci. 525 (2017) 99–106.

  35. [35]

      K. Zarshenas, H. Dou, S. Habibpour, et al., ACS Appl. Mater. Interface 14 (2022) 1838–1849. doi: 10.1021/acsami.1c16229

  36. [36]

      R. Xu, F. Gao, Y. Wu, et al., Sep. Purif. Technol. 281 (2022) 119884.

  37. [37]

      A.K. Ghosh, E.M.V. Hoek, J. Membr. Sci. 336 (2009) 140148.

  38. [38]

      L.E. Peng, Z. Yao, Z. Yang, et al., Environ. Sci. Technol. 54 (2020) 6978–6986. doi: 10.1021/acs.est.0c01427

  39. [39]

      S. Cao, A. Deshmukh, L. Wang, et al., Environ. Sci. Technol. 56 (2022) 8807–8818. doi: 10.1021/acs.est.2c00551

  40. [40]

      R. Dai, Z. Yang, Z. Qiu, et al., J. Membr. Sci. 662 (2022) 120966.

  41. [41]

      C. Jiang, X. Ma, L. Zhang, et al., J. Membr. Sci. 653 (2022) 120522.

  42. [42]

      S. Zhao, L. Li, M. Wang, et al., Sep. Purif. Technol. 258 (2021) 118029.

  43. [43]

      K. Chen, F. Li, T. Wei, et al., J. Membr. Sci. 684 (2023) 121882.

  44. [44]

      S. Karan, Z. Jiang, A.G. Livingston, Science 348 (2015) 1357–1351.

  45. [45]

      X. Song, B. Gan, S. Qi, et al., Env. Sci. Technol. 54 (2020) 3559–3569. doi: 10.1021/acs.est.9b05892

  46. [46]

      J. Wang, R. Xu, F. Yang, et al., J. Membr. Sci. 556 (2018) 374–383.

  47. [47]

      H.J. Kim, M.Y. Lim, K.H. Jung, et al., J. Mater. Chem. A 3 (2015) 6798–6809.

  48. [48]

      J. Schaep, B. Van der Bruggen, C. Vandecasteele, et al., Sep. Purif. Technol. 14 (1998) 155–162.

  49. [49]

      B. Mi, Science 343 (2014) 740–742. doi: 10.1126/science.1250247

  50. [50]

      G.S. Lai, W.J. Lau, S.R. Gray, et al., J. Mater. Chem. A 4 (2016) 4134–4144.

  51. [51]

      Z. Wang, Q. Tu, S. Zheng, et al., Nano Lett. 17 (2017) 7289–7298. doi: 10.1021/acs.nanolett.7b02804

  52. [52]

      K. Yuan, Y. Liu, H. Feng, et al., Chin. Chem. Lett. 35 (2024) 109022.

  53. [53]

      Z. Yang, X. Huang, X. Ma, et al., J. Membr. Sci. 570–571 (2019) 314–321.

  54. [54]

      K.P. Schlichting, D. Poulikakos, ACS Appl. Mater. Interfaces 12 (2020) 36468–36477. doi: 10.1021/acsami.0c07277

  55. [55]

      B. Liu, Q. Han, L. Li, et al., Environ. Sci. Technol. 55 (2021) 16379–16389. doi: 10.1021/acs.est.1c03576

  56. [56]

      H. Wu, B. Tang, P. Wu, J. Membr. Sci. 428 (2013) 341–348.

  57. [57]

      S. Sorribas, P. Gorgojo, C. Téllez, et al., J. Am. Chem. Soc. 135 (2013) 15201–15208. doi: 10.1021/ja407665w

  58. [58]

      G.N.B. Baroña, J. Lim, M. Choi, et al., Desalination 325 (2013) 138–147.

  59. [59]

      M. Amini, J. Membr. Sci. 435 (2013) 233–241.

  60. [60]

      J. Yin, G. Zhu, B. Deng, Desalination 379 (2016) 93–101.

• 

  Figure 1  The schematic illustration of MoS₂-incorporated TFN membrane fabrication. (a) Different amounts (0.5, 1, 2, 3, and 5 mg) of MoS₂ nanosheets were mixed with PIP/H₂O solution and then sonicated for 30 min to ensure an even distribution; (b) the mixed solution was applied to the substrate; (c) TMC/hexane solution was added to the MoS₂/PIP-treat substrate; (d) the formation of MoS₂-incorporated PA layer.

  下载: 全尺寸图片 幻灯片
  

  Figure 2  The morphological characterization of membranes. (a) The SEM and (b) AFM images of the surface of the control TFC membrane and MoS₂-incorporated TFN membranes fabricated on UP150 substrates.

  下载: 全尺寸图片 幻灯片
  

  Figure 3  Performance analysis of the MoS₂-incorporated TFN membranes. (a) A comparison of water permeance and Na₂SO₄ rejection tested with 1000 mg/L of Na₂SO₄ at 5 bar. (b) Comparison of salt rejection for TFN-2 membrane using different salts, namely, Na₂SO₄, MgSO₄, MgCl₂, CaCl₂, or NaCl at 5 bar. (c) Long-term filtration performance of TFN-2 membrane for Na₂SO₄ rejection at 5 bar. (d) A comparison of water permeance and Na₂SO₄ rejection of the optimal MoS₂-incorporated TFN membrane (TFN-2) with other nanomaterials-incorporated TFN membranes is discussed in Table S3.

  下载: 全尺寸图片 幻灯片
  

  Figure 4  Comparison of the separation performance of MoS₂-incorporated TFN membranes with various-sized nanosheets on different pore-size substrates before (B/F) and after (A/F) H₂O₂ etching. MoS₂-incorporated TFN membranes were prepared on small pore size substrates (UP150) with large (a) and small (b) MoS₂ nanosheets exfoliated at 25 ℃ and 200 ℃, respectively, and on large pore size substrates (MF0105) with large (c) and small (d) MoS₂ nanosheets.

  下载: 全尺寸图片 幻灯片
  

  Figure 5  TEM cross-sectional micrographs and high angle angular dark field-scanning transmission electron microscopy (HAADF-STEM) images of TFN-5 membranes prepared with (a) large and (b) small MoS₂ nanosheets on smaller pore-sized substrates (UP150, ~50 nm), respectively. In HAADF-STEM images, molybdenum is shown in green.

  下载: 全尺寸图片 幻灯片
  

  Figure 6  Distribution of large and small MoS₂ nanosheets on different substrate surfaces: (a) Impermeable plastic plate, (b) UP150 substrate (small pore size), and (c) MF0105 substrate (large pore size). The left side of each image represents 10 µL droplets of larger MoS₂ nanosheets, while the right side shows 10 µL droplets of smaller MoS₂ nanosheets. From top to bottom, the concentrations of MoS₂ nanosheet in the droplets were 0.06 g/L, 0.1 g/L, and 0.2 g/L, respectively.

  下载: 全尺寸图片 幻灯片
  

  Figure 7  The schematic illustration of the mechanisms of the MoS₂-incorporated TFN membranes before and after etching.

  下载: 全尺寸图片 幻灯片
•