In the structure of graphene, each carbon atom has three sp² hybridized orbitals, which form σ bonds with neighboring carbon atoms in the same plane at bond angles of 120°^[32]. This bonding pattern results in a hexagonal arrangement of carbon atoms in the plane, which is the characteristic structure of graphene. Large-area, high-quality graphene films can now be prepared by the CVD method^[33], and can be transformed into graphene porous membranes through various etching techniques, suitable for the precise separation of small-sized substances^[19]. It has also been estimated that the speed at which water molecules pass through graphene pores even exceeds that through aquaporins^[34]. However, these methods remain relatively costly, and the prepared graphene films are more suitable for applications in high-value-added fields such as electronics or optoelectronics, lacking economic competitiveness for use in membrane separation-related applications.

As a derivative of graphene, graphene oxide (GO) has a structure similar to graphene but contains numerous oxygen-containing functional groups, such as carboxyl or hydroxyl groups, at the edges or in the vertical direction of the graphene plane. With Hummer's improvement^[30]on the preparation method of GO, the production cost has been significantly reduced, and the scale and safety of preparation have relatively improved. In the past five years, GO has achieved a certain degree of commercial-scale production. On the other hand, the large number of oxygen-containing functional groups on the surface of GO greatly enhances its hydrophilicity, which is beneficial for improving the water flux and anti-organic fouling performance of water treatment membranes^[35-36]. Graphene has a thickness of 3.4 Å, and after GO nanosheets are stacked, the interlayer spacing (i.e., the space available for mass transfer) is 5.6 ± 1 Å. Theoretically, this spacing allows one or two layers of water molecules to flow within the channel while blocking most ions from passing through^[37]. The selectivity of the interlayer channels can also be adjusted by regulating the interlayer spacing or functionalizing GO^[32,38]. These characteristics make GO an exceptionally advantageous material for membrane fabrication, and related research has become one of the hotspots in experimental or theoretical studies within the field of membrane separation.

Inspired by graphene-based materials, other inorganic two-dimensional materials, including TMDs (such as MoS₂), MXenes, and hexagonal boron nitride (hBN), have also been used in the preparation of separation membranes. These materials differ from graphene-based materials in certain properties (such as hydrophilicity and mechanical rigidity), which can impart distinct performance characteristics to separation membranes, making them emerging materials in the field of membrane separation.

Similar to graphene, TMDs and hBN crystals belong to the hexagonal crystal system and can be obtained as single- or few-layer two-dimensional materials through exfoliation or growth methods. Their in-plane molecular structure is impermeable, but permeability is allowed between adjacent layers. TMDs^[39-40]and hBN^[41]nanosheets prepared via liquid-phase exfoliation can be assembled into laminated membranes. Due to the generally stronger interlayer van der Waals interactions compared to GO, and the cleaner, functional-group-free surface resulting in reduced electrostatic repulsion, these membranes exhibit superior physical stability in water to some extent over GO membranes. Additionally, their smooth surfaces reduce water resistance, thereby increasing water flux. Moreover, each MoS₂molecular layer has three atomic layers in the vertical direction, giving it higher stiffness at the macroscopic level than GO membranes.

MXenes materials are an emerging class of inorganic two-dimensional materials composed of transition metals combined with nitrogen (or carbon)^[42]. Similar to GO, MXenes materials have numerous hydrophilic end groups on their surfaces and exhibit excellent solution processability^[42]. Leveraging the abundant surface termination groups on MXenes sheets, layered MXenes can be used to fabricate gas separation membranes^[43], enabling high-performance separation of H₂and CO₂, with a H₂/CO₂separation factor as high as 160^[44]. In addition, MXenes materials can also impart certain antibacterial properties to separation membranes^[45-46]. Combined with their high water flux resulting from surface hydrophilicity^[45], they demonstrate significant application potential in water treatment membranes. The issue of swelling in two-dimensional material laminated membranes in water can be addressed by inserting Al³⁺between the nanosheets, which strongly interacts with the oxygen-containing functional groups on the MXenes surface, stabilizing the interlayer spacing and inhibiting the swelling of the laminated membrane in water^[47]. After treatment with this strategy, the MXenes membrane exhibits non-swelling stability for up to 400 hours in aqueous solutions, while maintaining a high NaCl rejection rate (approximately 89.5% ~ 99.6%) and a high water flux (approximately 1.1 ~ 8.5 L·m^-2·h^-1)^[47].

In addition to inorganic two-dimensional materials, organic two-dimensional materials can also be used to fabricate high-performance separation membranes. Two-dimensional polymer nanosheets with periodically repeating units on a two-dimensional plane have recently attracted widespread attention in the field of materials science^[48]. Compared with traditional inorganic two-dimensional materials, two-dimensional polymer nanosheets offer advantages such as light weight, good flexibility, and easier functionalization, drawing increasing interest in areas like substance separation and energy storage^[49-50]. More importantly, two-dimensional polymer nanosheets feature uniformly ordered pores, and their pore sizes can be chemically designed and adjusted through molecular engineering. This tunable and ordered pore structure provides shorter permeation pathways and more precise molecular-level sieving. Among these, COFs are the most representative. COFs are constructed by covalently linking organic molecular units into periodic networks^[51], and they are a class of artificial materials designed to mimic natural porous materials, also known as "organic zeolites." Compared to inorganic zeolites, COFs not only retain the properties of porosity and large surface area but also possess unique advantages such as low density, tunable pore size and structure, and ease of customized functionalization^[51]. Additionally, similar to graphite, bulk COFs are also layered materials, with layers regularly stacked through relatively weak intermolecular forces, allowing for the preparation of two-dimensional covalent organic frameworks (2D-COFs) via exfoliation^[52]. The channels inherent in 2D-COFs are absent in graphene and GO, making them an ideal alternative to GO for fabricating separation membranes. Recent studies have also demonstrated that COF-based separation membranes exhibit excellent performance in organic solvent separation^[53], gas separation^[54], and organic nanofiltration^[55].

In addition to the various synthetic materials mentioned earlier, a special class of two-dimensional materials has recently gained attention in the field of separation membranes. Biological cell membranes (and cell walls) serve as barriers that maintain cell shape and regulate the transport of substances in and out of cells. They represent a complex and highly efficient structure evolved and selected by nature over hundreds of millions of years^[56-57]. Cell membranes or cell walls inherently possess certain transport channels for exchanging substances between the inside and outside of cells. Some of these transport channels (such as aquaporins) exhibit both high selectivity and rapid transport rates, which are currently unattainable by synthetic materials^[58-60]. Therefore, constructing biomimetic separation membranes using biological membrane materials could be an effective approach to overcoming the limitations imposed by the trade-off effect.

Zhang et al^[59]first reported in 2021 a strategy that uses fungal cell walls as one of the membrane-forming materials. They prepared microcomposite membranes by vacuum filtration after compounding cell walls with GO in a dispersion. The strong interaction between GO and the cell wall, combined with the anti-swelling behavior of the cell wall, effectively restricts the swelling of the GO layer in solvents, thereby ensuring high selectivity. Meanwhile, the well-developed porosity of the cell wall enhances the flux. This composite membrane demonstrates excellent retention of small molecules and a solvent permeation coefficient as high as 56 L·m^-2·h^-1·bar^-1. Subsequently, Lee's team reported a method for attaching red blood cell membranes containing aquaporins to GO via integrin-mediated molecular recognition, and based on this composite material, they fabricated a biomimetic high-pressure desalination membrane^[61]. The cell membrane provides a relatively stable biomimetic environment for aquaporins, enabling their functional performance. This biomimetic membrane exhibits a NaCl rejection rate of 99.1% and a water permeation coefficient of 7.83 L·m^-2·h^-1·bar^-1when fed with a 1000 ppm NaCl solution, surpassing the performance limits of currently commercialized reverse osmosis membranes. Compared to the more expensive red blood cells, Hu Yunxia's team chose Escherichia coli, which can be cultivated in large quantities through bioengineering methods, as the source for extracting E. coli cell membranes. Using membrane fragments obtained from disrupted E. coli cell membranes as the membrane-forming material, they prepared a nanofiltration membrane via vacuum-assisted filtration^[62]. This membrane has a water permeation coefficient of 52.6 L·m^-2·h^-1·bar^-1, higher than most reported nanofiltration membranes made from two-dimensional materials, and also exhibits high rejection rates for Brilliant Blue R (98.5%, Brilliant Blue R), antibiotics (91.1%, erythromycin), and divalent ions (64.8%, Na₂SO₄). These studies on the application of biomembrane materials in separation membranes offer new insights into overcoming the limitations of synthetic polymer-based membranes.

In general, two-dimensional materials encompass a wide range of materials, including carbon-based materials, inorganic materials, organic materials, and biomaterials. Influenced by various factors such as chemical properties, physical characteristics, and manufacturing costs, their applications in the field of separation membrane fabrication tend to vary. For instance, graphene films produced by CVD are extremely thin (with a thickness of a single atomic layer) and feature small etching pore sizes (typically sub-nanometer), making them suitable for the precise separation of small-sized substances. On the other hand, nanosheet materials like GO and Mxenes, which can be prepared on a relatively large scale, possess numerous hydrophilic functional groups or residues on their surfaces, making them ideal for fabricating separation membranes via dispersion processing for liquid-phase separation and purification. Biomembrane materials, meanwhile, often exhibit excellent biocompatibility, making them well-suited for specialized separation applications such as antibiotic removal or drug purification. In addition to the influence of the materials themselves, different membrane fabrication methods also significantly affect the application areas and performance characteristics of the resulting two-dimensional material-based separation membranes.

Two-dimensional material films (such as graphene films) can be as thin as a single atomic layer, with highly clean and smooth surfaces. The channels formed by these materials allow substances to pass through rapidly, making them the most ideal materials for preparing separation membranes^[63-64]. However, traditional two-dimensional materials such as graphene or h-BN are inherently dense and pore-free, so their films must be artificially perforated to create porous membranes suitable for constructing separation membranes. Currently, the most widely studied and utilized method for preparing large-area two-dimensional materials (such as graphene films) is the CVD method. Leading research institutions and companies in related fields both domestically and internationally have reported strategies for using this method to fabricate high-quality, large-area two-dimensional material films^[33,65-66]. Based on this, methods such as ion irradiation, chemical etching, or plasma etching are employed to create numerous pores within the plane of two-dimensional material films, resulting in pore sizes ranging from 3 to 76 Å and transforming the two-dimensional material films into porous two-dimensional membranes^[12-13].

Fischbein et al.^[67]discovered a method for etching nanopores in graphene using a finely focused electron beam. By controlling the irradiation time, the pore size can be precisely adjusted within a few nanometers without causing additional damage such as deformation to the graphene. Subsequently, methods such as plasma etching^[19]and voltage-pulse etching^[14]have also been employed to fabricate nano- or sub-nanoscale pores in two-dimensional material films. Given that the effective pore size between the carbon atoms in the hexagonal rings of graphene is only 0.064 nm, significantly smaller than the van der Waals diameters of helium and hydrogen molecules (0.28 and 0.314 nm, respectively)^[68],gas molecules can only pass through graphene layers via artificially created pores. Therefore, customized nanoporous graphene membranes can be utilized for precise gas separation.

In this research field, the Bunch team has been dedicated for many years to preparing and studying microporous graphene semi-permeable membranes for gas or liquid separation. They^[69]reported in 2012 that ultraviolet-induced oxidative etching could create pores in graphene membranes, with pore sizes comparable to the kinetic diameter of argon atoms (3.4 Å), laying the foundation for in-depth research into microporous graphene membranes for gas sieving. Subsequently,^[14]they further developed a method using atomic force microscopy probes to apply pulsed voltage, thereby generating micropores in graphene and enriching the techniques available for pore formation on graphene films. Zhao et al.^[19]developed a strategy involving the synergistic effect of oxygen plasma and ozone (O₃) treatment, achieving high-density pore formation (2.1 × 10¹² cm^-2) on graphene films. This membrane demonstrated ultra-high separation factors for several mixed gases, breaking previously reported records, with separation factors for H₂/CH₄ and H₂/C₃H₈ reaching up to 25.1 and 57.8, respectively. In summary, the advantage of thin-film pore-forming strategies lies in the extremely low mass transfer resistance of the resulting separation membranes, fine tunability of pore size, and fewer membrane defects, making them suitable for efficient retention and separation of small-sized substances. However, a drawback of these methods is that both the preparation of high-quality two-dimensional material films and subsequent pore formation are often costly and technically challenging. Moreover, due to the typically poor mechanical properties of two-dimensional material films, the resulting separation membranes can only be used in gas separation systems and struggle to withstand the high operating pressures required in liquid filtration systems, which poses certain obstacles to expanding their commercial applications.

Compared to two-dimensional material thin films, which have higher preparation complexity and costs, two-dimensional material nanosheet powders or dispersions can be produced on a large scale through methods such as liquid-phase exfoliation^[39],CVD^[70],or hydrothermal (solvent) synthesis^[71],significantly reducing both economic and time costs. The technical difficulty of assembling two-dimensional nanosheets into separation membranes is also much lower than that of artificially creating pores in two-dimensional materials. Therefore, the method of preparing separation membranes by stacking or assembling two-dimensional nanosheets has received more extensive research and application in the field of membrane separation. The operational cost and difficulty of the layer-by-layer stacking strategy are considerably lower, often achievable through simple methods such as immersion^[72],vacuum filtration^[73,74],spray coating^[75],or electrodeposition^[76],and the resulting separation membranes have larger areas, making them more suitable for large-scale commercial applications.

Solution filtration is the most common method for preparing two-dimensional material laminated membranes, where nanosheets in a dispersion are deposited onto a porous substrate via pressure-driven filtration^[77]. Hydrophilic two-dimensional nanosheets (such as GO) can disperse well in water or most polar organic solvents, which facilitates the formation of highly aligned layered structures through vacuum filtration. The thickness of the deposited layer can be controlled by adjusting parameters such as applied pressure, dispersion concentration, or dispersion volume^[78]. Most of the previously mentioned two-dimensional nanosheets (GO, TMDs, MXenes, etc.) have been reported to be used in preparing separation membranes using this method^[39-40,79]. Different two-dimensional materials can also be combined to form composite membranes, achieving complementary advantages^[80-81]. For example, COFs can be assembled with GO to fabricate composite membranes^[82]; the potential defects between COF nanosheets are eliminated by flexible GO nanosheets, while the boronate carriers within the COF membrane promote CO₂transport, thereby enhancing separation performance. The CO₂permeance of the composite membrane reaches 164.2 GPU, and the CO₂/CH₄separation factor is as high as 27. Additionally, based on the original vacuum filtration method, a layer-by-layer deposition technique has also been developed^[83]. This method combines multiple layers of two-dimensional nanosheets through interactions such as covalent bonds, hydrogen bonds, or electrostatic adsorption between different materials, significantly enhancing the stability of the deposited layers and reducing the impact of swelling of the two-dimensional nanosheets in solution, thus helping to slow down the performance degradation of the membrane during long-term operation^[84-85]. The thickness of the functional layer can be easily controlled by the number of layer-by-layer cycles^[86]. In addition to conventional filtration deposition methods, other techniques such as spray coating without pressure-driven force^[75], electrodeposition^[76], Langmuir-Blodgett method^[87], and inkjet printing^[88]can also be used for fabricating two-dimensional nanosheet membranes. Since these methods do not rely on external pressure as the driving force for deposition, the resulting two-dimensional material-based separation membranes often have advantages such as thinner functional layers, more ordered nanosheet arrangements, or finely tunable layer numbers, thus broadening the pathways for fabricating two-dimensional material membranes.

In addition to simply using two-dimensional materials for membrane fabrication, researchers in the field of membrane separation have recently developed a strategy that combines the advantages of two-dimensional materials with those of traditional polymer membranes—incorporating high-performance inorganic fillers into polymers to create composite membranes^[89]. Compared to conventional nano-fillers, two-dimensional nanosheet materials are thin and do not increase the thickness of the functional layer. They can also be easily functionalized to enhance their affinity with the base membrane, making them one of the most promising filler materials. By providing additional rapid mass transfer channels without compromising selectivity, these materials can significantly improve the permeability of polymer membranes. The preparation of such composite membranes generally falls into two strategies: (1) dispersing two-dimensional material nanosheets in a polymer casting solution, which are then directly mixed into the polymer membrane during coating and film formation, creating mixed-matrix membranes (MMMs) that combine the inherent advantages of polymer membranes with the superior properties of the fillers. The Eddaoudi team^[90]dispersed MOF nanosheets in a CHCl₃solution of polyimide, and subsequently obtained a polyimide mixed-matrix membrane containing MOF nanosheets through casting and solvent evaporation. This membrane exhibited an exceptionally high separation factor of 60.3 for total acid gases (H₂S and CO₂) versus CH₄in natural gas, demonstrating outstanding performance in separating hydrogen sulfide and carbon dioxide from natural gas. (2) During interfacial polymerization for preparing polyamide or polyester membranes, two-dimensional material nanosheets are dispersed in the aqueous phase solution. As interfacial polymerization occurs between the aqueous and organic phases, the nanosheets become incorporated into the formed polyamide or polyester layer, creating thin-film nanocomposite membranes (TFN membranes). The Wei Zhong team^[91]prepared TFN membranes by adding COFs to the polysulfonamide layer via interfacial polymerization. The introduction of COF nanosheets helped enhance the water permeability and rejection of inorganic salts in the membrane. The water permeance coefficient and magnesium chloride rejection rate of the nanofiltration membrane increased by 149.6% and 4.6%, respectively, reaching 15.1 L·m^-2·h^-1·bar^-1and 93.3%.

An important issue that needs to be addressed for such composite membranes is the compatibility between the polymer and the filler. Poor adhesion between the filler and the polymer often leads to non-selective interfacial voids^[92]. Therefore, achieving a high filler loading while ensuring a defect-free polymer-filler interface is crucial for preparing high-performance composite membranes^[93]. For instance, GO, the most common nanosheet material, exhibits poor compatibility with polymer matrices and tends to disperse unevenly in composite membranes^[94]. However, graphene oxide can be easily functionalized by connecting it to other desired functional molecules through oxygen-containing functional groups, allowing polymers or hydrophilic functional groups and various other components to be grafted onto the surface of graphene oxide, thereby enhancing its compatibility within the polymer matrix. For example, Keneiloe et al.^[93] reported a polyimide membrane incorporating β-cyclodextrin-functionalized graphene oxide (β-CD-f-GO). Compared to the unmodified polyamide membrane, the water permeation flux and antibacterial activity of the membrane containing β-CD-f-GO increased by approximately 38% and 45.9%, respectively, without compromising salt rejection. In addition to functionalizing the nanosheets, other post-treatment strategies, such as annealing, can further eliminate interfacial defects between the nanosheets and the polymer matrix, positively impacting the performance of the final MMMs membranes.

Separation membranes prepared by layer-by-layer assembly using two-dimensional nanosheets as materials have mass transfer channels primarily consisting of interlayer channels, namely the interlayer or in-plane gaps formed by the stacking of nanosheets^[98]. The Mi team^[98]characterized the interlayer spacing of GO membranes in aqueous environments using quartz crystal microbalance and X-ray diffraction, finding that the interlayer spacing of GO can vary between 0.76 and 6.2 nm. This variation in interlayer spacing of two-dimensional nanosheets provides possibilities for regulating the performance of stacked membranes. By adjusting the membrane-forming pressure^[62]or introducing intercalating agents^[21], the interlayer spacing can be controlled to regulate channel size (i.e., membrane pore size), thereby tuning the selectivity and permeability of the membrane. For example, Yang et al.^[27]prepared a poly(4-vinylphenylboronic acid)-grafted MXene membrane, where diols anchored between layers can switch from anionic to cationic states through the formation and breaking of covalent bonds. The interlayer spacing of MXene can be adjusted between 1.49 and 1.60 nm, affecting the membrane's permeation selectivity, and consequently, the water permeance coefficient of the membrane varies between 12.3×10³and 55.8×10³L·m^-2·h^-1·bar^-1. However, a drawback of interlayer mass transfer channels is their tortuous mass transfer paths, resulting in a high tortuosity factor and significant mass transfer resistance, which reduces the membrane's permeability^[99]. Therefore, it is necessary to enhance the membrane's permeability by increasing the number of channels or introducing in-plane mass transfer channels.

In-plane mass transfer channels exist in stacked membranes prepared using intrinsically crystalline porous two-dimensional nanosheets or in porous membranes fabricated by creating pores in two-dimensional material films. The molecular structure^[101](or crystal structure^[102]) of crystalline porous two-dimensional materials (such as COFs^[100]or two-dimensional vermiculite^{[22] [101]}contains intrinsic pores with diameters ranging from 9 to 32 Å^[101], which can be used for the transport of gases, water, or other molecules^[103]. For two-dimensional material films that do not inherently contain pores, artificial pores can be created on the film surface through methods such as plasma etching, forming in-plane mass transfer channels^[104]. The in-plane mass transfer channels formed by these two methods typically run perpendicular to the membrane surface, allowing substances to pass through the membrane layer along the shortest path, significantly reducing the tortuosity factor of the membrane pores and enhancing the permeation rate of substances, thereby giving the membrane the advantage of high flux^[97]. However, the greater the number of in-plane mass transfer channels, the higher the membrane porosity, which may lead to a decrease in membrane structural strength^[8,104]. Therefore, maintaining high structural strength while increasing the number of in-plane mass transfer channels has become one of the critical issues that must be addressed to improve the performance of two-dimensional material-based separation membranes.

For two-dimensional material-based separation membranes, especially those composed of layered materials, a rational mass transfer channel structure is crucial for achieving high permeation selectivity^[106]. In separation membranes constructed by stacking nanoporous sheet materials such as GO, permeant molecules flow in a tortuous manner, first passing through in-plane gaps and then through nanochannels between adjacent layers. Due to the irregularity of in-plane gaps, interlayer nanochannels become the primary pathway controlling separation efficiency^[107]. Early studies mainly focused on enhancing the regularity of these channels by simply adjusting the stacking efficiency, aiming to create single-layer-separated channels and improve the precision of size sieving^[108-110]. However, layered material-based membranes inherently have high mass transfer resistance due to their tortuous transport paths. Increasing the total amount of interlayer channels would further reduce the permeability of permeants, and membrane performance would still be limited by the permeation-selectivity trade-off.

In contrast, a more widely used approach to regulate the interlayer spacing of layered materials is to insert organic small molecules^[21,111],carbon nanotubes^[112],or nanoparticles^[106,113]as intercalants. These intercalants can increase or decrease the interlayer spacing through physical or chemical interactions with the layered materials, thereby adjusting the membrane's permeation selectivity. For example, Wei-Qun Shi's team selected cucurbituril as a pillar unit inserted between Ti₃C₂T _xMXene nanosheets, constructing a two-dimensional layered membrane and adjusting the interlayer spacing^[114]. Cucurbituril forms stable and strong interactions with the hydroxyl functional groups on the MXene nanosheet surface through multiple binding sites, anchoring itself onto the nanosheet surface. After inserting cucurbituril pillars, the MXene nanosheet interlayer spacing expanded to 22 Å, an increase of 8.4 Å compared to the original Ti₃C₂T _xmembrane's interlayer spacing (≈13.6 Å). The composite membrane exhibited excellent nanofiltration performance (a water permeance of 69 L·m^-2·h^-1·bar^-1and a methylene blue rejection rate of 93.6%), and could be reused at least 30 times without significant degradation. Yang et al.^[115]achieved a reduction in GO interlayer spacing through intercalation, selecting amino-functionalized polyhedral oligomeric silsesquioxane (POSS-NH₂) as a multifunctional intercalant for GO membranes. Through POSS-NH₂crosslinking, the interlayer channel size of the GO membrane decreased from 9.1 Å to 4.7 Å, enhancing its selectivity for CO₂/CH₄. After optimization, the membrane achieved a CO₂/CH₄separation factor of 74.5, surpassing most previously reported two-dimensional material-based membranes in CO₂/CH₄separation performance. Although considerable work has already been done in adjusting the interlayer spacing of layered material membranes, in some studies, the size of the spacer molecules may be too large (larger than the size of the rejected molecules), causing the membrane channels to expand into an inappropriate size range. While this significantly increases the permeance, it weakens the size-segregation capability. Therefore, optimizing the selection of intercalants based on structure and properties to enhance the compatibility among layered materials, intercalants, and separation targets, or optimizing the interlayer microstructure (e.g., increasing channels perpendicular to the layer direction to shorten mass transfer paths)^[116]will be a key focus for researchers in the near future.

Enhancing the long-term operational stability of two-dimensional material-based separation membranes is also a critical issue that must be addressed to promote their commercial application. In solution environments, layered material stacked membranes may undergo microstructural changes such as swelling, leading to an expansion of interlayer channels and a significant reduction in selectivity^[98,105-117]. For example, when a dry GO membrane is immersed in water, its interlayer spacing increases sharply from approximately 0.8 nm to 6–7 nm^[98]. The primary reason for this phenomenon is the relatively low van der Waals forces between graphene oxide nanosheets, which struggle to balance the repulsive forces generated between layers upon immersion in water. These forces mainly consist of two components: one is the electrostatic repulsion arising from the same charge on GO sheets, and the other is hydrogen bonding formed between GO and water. Under the influence of these repulsive forces, the increased distance between GO nanosheets may eventually lead to cracking of the membrane. Strategies to address this issue generally involve introducing intermediate substances between layers or between layers and the substrate membrane, leveraging strong interactions between these intermediates and the two-dimensional materials to counteract the effects of swelling^[59,61], or enhancing structural stability through chemical cross-linking^[47,111]. For instance, the team led by Yunxia Hu^[118]synthesized a bisphenol molecule—5,5′-diamine-2,2′-biphenol (BIPOL-NH₂)—which features a twisted structure with two rigid benzene rings connected together, and used it as an interlayer spacer for GO to fabricate GO stacked membranes^[21]. The hydroxyl and amino groups on the BIPOL-NH₂molecule can anchor themselves between layers by reacting with hydrophilic oxygen-containing groups on GO nanosheets, stabilizing the interlayer spacing at 1.13–1.16 nm. The embedding of BIPOL-NH₂molecules within the interlayer space creates an anti-swelling microenvironment, thereby inhibiting membrane swelling. After being immersed in solutions with pH values ranging from 2 to 11 for 30 days, no obvious disintegration was observed in the BIPOL-GO membrane. Long-term tests also demonstrated that this membrane could maintain stable NaCl/Congo red separation efficiency (~separation factor ~100) during continuous filtration for 72 hours. In addition to introducing intermediate substances, another strategy is to select two-dimensional materials with weaker repulsive forces but stronger attractive forces between layers as alternatives to GO. For example, at the same interlayer spacing, the van der Waals forces between Ti₃C₂T _xnanosheets are about one order of magnitude higher than those of GO, while the electrostatic repulsion is approximately two orders of magnitude lower. Consequently, Ti₃C₂T _xmembranes exhibit high stability in aqueous solutions without any modification, and can still relatively maintain their structural integrity after being immersed in water for one month^[119].

In addition, membrane fouling is also a significant issue that affects the long-term operational stability of separation membranes in liquid separation systems. As filtration time extends, retained substances or other contaminants in the feed liquid (such as calcium and magnesium ions, small organic molecules, colloids, or microorganisms) may gradually accumulate on the membrane surface, forming a filter cake layer, which leads to pore blockage and a decrease in the membrane's permeability coefficient^[120]. One characteristic of nanosheet materials such as GO and MXene is the abundance of functional groups or residues on their surfaces, allowing their charge and hydrophilicity to be modified through chemical treatments^[94]. This modification can alter the hydrophilicity of the prepared membranes or significantly change their surface morphology and roughness, thereby enhancing their anti-fouling performance. For instance, Ge et al.^[121]prepared polydopamine (PDA)-modified graphitic carbon nitride (g-C₃N₄) and used it as an aqueous-phase additive to fabricate g-C₃N₄/PDA-doped TFN membranes via interfacial polymerization. During the interfacial polymerization process, g-C₃N₄/PDA slowed down the diffusion of m-phenylenediamine in the aqueous phase, resulting in a smaller leaf-like or nodular surface morphology and reduced surface roughness, thus improving the membrane's anti-fouling performance. After being fouled by model pollutants such as bovine serum albumin, sodium humate, and silica, the flux recovery rate achieved by water washing reached as high as 98.5%, indicating excellent anti-fouling performance against organic, inorganic, and mixed contaminants.

In addition, regarding the issue of low mechanical strength and susceptibility to physical damage in two-dimensional material membranes, studies have found that carbon nanotube networks can serve as a microscopic support framework. While enhancing the mechanical stability of two-dimensional material porous membranes, they can also effectively limit the further spread of mechanical damage, enabling the separation membranes to maintain their separation performance during long-term operation^[8]. Similar research has also employed fiber structures with strong interlayer adhesion as supporting layers for large-area porous graphene membranes, thereby improving the overall mechanical strength of the composite membranes^[104]. Furthermore, Yunxia Hu's team^[122]developed a method of inserting tannic acid nanowire networks between GO layers to enhance the anti-compaction performance of GO membranes. These nanowire networks provide support for the GO, resulting in a linear increase in membrane flux with increasing pressure. When the pressure was increased from 1 bar to 4 bar, the membrane flux rose from 50.2 L·m^-2·h^-1to 120.6 L·m^-2·h^-1. Through these studies, significant progress has been made in improving the stability of two-dimensional material-based separation membranes. However, in commercial applications such as water treatment or organic nanofiltration, cross-flow filtration is often used^[78], which places even higher demands on the stability of separation membranes. Therefore, the long-term operational stability of two-dimensional material-based separation membranes in such practical applications still requires further improvement.

The structure of two-dimensional material-based separation membranes differs significantly from that of traditional polymer-based separation membranes. The unique interlayer mass transfer channels in two-dimensional materials are not present in polymer-based membranes, resulting in differences in their pervaporation mechanisms compared to conventional separation membranes. Determining the true structure of the two-dimensional interlayer channels is crucial for studying the pervaporation mechanism of two-dimensional material-based separation membranes and enhancing their permeation selectivity. In early studies, the two-dimensional interlayer channels were often simplified as being formed by ideally stacked, flat nanosheets arranged in a regular and parallel manner. However, during actual membrane fabrication, the assembly of nanosheets can be disturbed by factors such as pressure, vibration, or dispersion fluid flow, leading to phenomena like bending or folding in the deposited layers^[123-124]. These nanoscale wrinkles accumulate with each layer, further affecting the membrane's permeation or separation performance. Irregular wrinkles within the membrane provide additional pathways for cross-membrane diffusion, promoting substance permeation; however, the defects they introduce may reduce separation efficiency^[123,125]. To improve separation capabilities, mechanical stirring of the nanosheet dispersion during deposition can be employed to apply continuous shear stress, flattening the nanosheets and significantly reducing wrinkles and nanoscale defects in the layered structure. This results in a well-aligned membrane structure and enhanced separation performance^[123].

Whether it is the two-dimensional interlayer channels formed by stacking layers or the pores inherent in certain two-dimensional materials, the dimensions of these mass transfer channels can be as small as nanometers or even subnanometers. The transport behavior of substances within these channels differs to some extent from that observed in traditional polymer separation membranes^[126-127]. When the channel diameter is less than 10 nm, the confined liquid within the channel exhibits non-continuous fluid behavior due to confinement effects^[128]. The Geim team discovered that sub-micron-thick membranes made from GO allow ultra-fast water permeation without any permeation of other liquids, vapors, or gases, attributing this phenomenon to the low-friction flow of monolayer water in two-dimensional capillaries^[129]. Furthermore, the Geim team and its collaborators have assembled sub-nanometer artificial channels using single-layer graphene or molybdenum disulfide to investigate the transport behavior of substances within two-dimensional material channels^[64,130-131]. They found that when the hydrated ion diameter exceeds the channel size, ions can still pass through the channel at different rates by squeezing or flattening the hydration layer^[64]. Regarding gas molecule permeation, studies have shown that when different gases attempt to pass through angstrom-scale pores in graphene membranes, if the mean free path of the gas molecules is greater than the pore diameter, the gas molecules will be blocked and unable to pass through^[69]. These novel transport and cutoff behaviors exhibited by two-dimensional material functional layers require further research. A deeper understanding of their mechanisms will help researchers optimize the design of two-dimensional material-based separation membranes, further enhancing their performance and facilitating their transition into commercially viable, practical separation membranes.