Constructing and regulating nanochannels in two-dimensional-material-based membranes for specified separation applications

Synthesis of 2D nanosheets

2D nanosheets serve as the foundational materials for constructing nanochannels from 2D materials. The synthesis of 2D nanosheets can be generally classified as “top-down” and “bottom-up” strategies, as shown in Figures 2 and 3.

Figure 2. Schematics of the main “top-down” synthesis strategies of 2D nanosheets. (A) Micromechanical cleavage exfoliation^[171]. Copyright 2012, Elsevier Ltd. (B) Ball milling shear-force exfoliation. (C) Sonication exfoliation^[172]. Copyright 2018, Elsevier B.V. (D) Electrochemical (E-Chem) exfoliation^[38]. Copyright 2018, Springer Nature. (E) Ion/molecule intercalation^[173]. Copyright 2017, John Wiley & Sons, Inc. (F) Thermal oxidation etching^[47]. Copyright 2021, John Wiley & Sons, Inc.

Figure 3. Schematics of the main “bottom-up” synthesis strategies of 2D nanosheets. (A) Chemical vapor deposition^[174]. Copyright 2017, Springer Nature. (B) Physical vapor deposition^[47]. Copyright 2021, John Wiley & Sons, Inc. (C) Wet-chemical synthesis^[175]. Copyright 2021, American Chemical Society.

The “top-down” strategy involves the exfoliation of 2D nanosheets from bulk-layered materials by breaking weak interlayer van der Waals, π-π stacking, and/or hydrogen bonding interactions [Figure 2]. Novoselov et al. pioneeringly achieved a single layer of graphite using the micromechanical cleavage method [Figure 2A], which enables the preparation of a variety of 2D nanosheets with a clean surface and high quality on a macroscopic scale^[33,34]. However, it is worth noting that this method has the drawback of low efficiency. Paton et al. took the lead in conducting shear force-assisted^[35] [Figure 2B] and sonication-assisted^[36] [Figure 2C] exfoliations in liquids, where bulk graphene and TMDs crystals can be efficiently dispersed in common solvents to obtain large quantities of mono- and few-layer nanosheets. Another representative shear force-assisted and sonication-assisted mixed work is demonstrated by Peng et al., who developed a soft-physical process of preparing 2D MOF nanosheets with large lateral sizes by low-energy wet ball milling coupled with ultrasonication in methanol/propanol mixtures^[37]. Electrochemical (E-Chem) exfoliation [Figure 2D] is a promising bulk method for producing graphene from graphite, where an applied voltage drives ionic species to intercalate into graphite, forming gaseous species that expand and exfoliate individual graphene sheets. Achee et al. reported a method for sustained graphene output within a permeable and expandable containment system, indicating both high yield (65%) and extraordinarily large lateral size ( > 30 μm) of graphene^[38].

The use of specific ions or molecules intercalation [Figure 2E] to weaken interactions within the layers by exchanging or reacting can further enhance the efficiency of liquid-phase exfoliations, which have been widely employed for preparing various types of 2D nanosheets such as GO, MXene, TMDs, and LDHs^[39-42]. Notably, these intrinsically nonporous 2D nanosheets still require a subsequent perforation process to fabricate nanochannels (as discussed in the next section). For example, intrinsically porous 2D nanosheets can be exfoliated from their related layered materials with high purity via melt blending with a polymer matrix, as demonstrated by Varoon et al.^[43]. And the synthesis of g-C₃N₄ nanosheets can be achieved by thermal oxidation [Figure 2F] with long-time heating and etching, which was shown by Ren et al., to produce a series of g-C₃N₄ nanosheets with a thinner layer thickness, larger BET surface area, and higher graphic nitrogen ratio^[44]. They proposed that the higher activity of the g-C₃N₄ from long-time thermal oxidative etching might be ascribed to the enlarged specific surface, pore volume, and higher graphic nitrogen ratio with the loose and soft laminar morphology^[44].

However, it should be noted that the “top-down” strategy for exfoliation is limited by the availability of layered materials, which may result in incidental structural deterioration and morphological damage during the exfoliation process. Such issues may hinder the obtainment of ideal 2D nanosheets.

The “bottom-up” strategy [Figure 3] involves synthesizing 2D nanosheets through chemical reactions from specific precursors under given conditions, which is theoretically feasible for all types of 2D nanosheets. Chemical vapor deposition (CVD) [Figure 3A] growth and physical vapor deposition (PVD) [Figure 3B] growth are reliable routes for synthesizing highly crystalline 2D nanosheets such as graphene, MXene, h-BN, TMDs, and others, with large-area uniformity^[45-47]. However, transferring the crack-free as-prepared 2D nanosheets from the substrate remains challenging^[48-50]. As an alternative, wet-chemical synthesis [Figure 3C] is widely employed due to its good controllability, reproducibility, and scalability, especially for preparing those crystalline porous materials made from predesigned skeletons such as zeolites, MOFs, and COFs. Jeon et al. successfully produced high-aspect-ratio MFI-type zeolite nanosheets using a nanocrystal-seeded growth method that was triggered by a single rotational intergrowth^[51]. These highly oriented MFI nanosheets with straight vertical micropores allow the secondary growth of thin and defect-free MFI membranes with extraordinary performance for separating xylene isomers^[51]. Makiura et al. reported a procedure for the rational modular assembly of MOF nanosheets with perfect orientation on a solid substrate by integrating the layer-by-layer growth and the Langmuir-Blodgett methods^[52]. Similarly, Rodenas et al.^[53] presented a new approach to produce self-standing intact MOF nanosheets that relied on the diffusion-mediated modulation of the MOF growth kinetics. The synthesis medium involves three vertically arranged liquid layers, where a topmost solution of cations and a bottom solution of linker precursors diffuse into the intermediate solvent layer, causing a slow supply of the MOF nutrients to form MOF nanosheets in a highly diluted medium. On the other hand, COF nanosheets can either grow on various substrates under solvothermal conditions or at solid-liquid/liquid-liquid/air-liquid interfaces via interfacial polymerizations^[54,55]. An excellent representative work was done by Kandambeth et al., who successfully demonstrated the fabrication of various flexible, continuous, and defect-free COFs by casting and baking the mixed solution containing organic linkers and co-reagents^[56]. In addition, the “bottom-up” approach can also be used to directly fabricate porous graphene with a well-defined pore structure by selecting appropriate rigid molecular building blocks as monomers for the organic synthesis^[57].

Perforation on 2D nanosheets

Unlike intrinsically porous 2D nanosheets, the perfect monolayer nonporous 2D nanosheets are almost impermeable, requiring the artificial drilling of out-of-plane nanoscale holes to serve as nanochannels. Continuous experimental endeavors have been made to develop various perforation techniques to realize this goal, including physical and chemical methods such as focused electron beam, bombardment/focused ion beam, oxygen plasma etching, ultraviolet-induced oxidative etching, and chemical etching [Figure 4]. For example, Fischbein et al. showed that graphene nanosheets could be controllably nano-sculpted using the focused electron beam ablation technique with few nanometer precisions [Figure 4A]^[58]. Similarly, Koenig et al. utilized ultraviolet-induced oxidative etching [Figure 4B] to create angstrom-sized pores in the pristine graphene membrane, which were used as molecular sieves and exhibited selective gas transport capabilities^[59].

Figure 4. Schematics of perforation on 2D materials. (A) Low-dose e-beam patterning of ZIF-L with pyridine as an etchant^[176]. Copyright 2021, American Chemical Society. (B) UV irradiation etching of MoS₂/WS₂ nanosheets^[177]. Copyright 2016, Royal Society of Chemistry. (C) Focused ion beam etching process on GO^[178]. Copyright 2019, Elsevier Ltd. (D) Plasma etching process^[179]. Copyright 2020, MDPI. (E) The synthesis of porous Si/C composite nanosheets by chemical etching^[180]. Copyright 2019, American Chemical Society.

The focused ion beam perforation method is another effective technique [Figure 4C], which was successfully employed by Celebi et al. to produce narrowly-distributed pore sizes ranging from < 10 nm to 1 μm in free-standing graphene^[60]. Later, Russo et al. created pore nucleation sites on graphene using an argon ion beam followed by edge-selective electron recoil sputtering, yielding graphene nanopores with radii as small as 3 Å, all without the use of focused beams^[61]. However, the effective areas of these nanoporous graphenes are limited to the micrometer scale. To overcome this limitation, O’Hern et al. introduced isolated and reactive defects into the graphene lattice through ion bombardment, which were then enlarged by oxidative etching to produce permeable pores with diameters of 0.40 ± 0.24 nm and densities exceeding 10¹² cm^-2[62]. In another approach, Surwade et al. employed an oxygen plasma etching process [Figure 4D] to produce nanopores with tunable diameters on a graphene monolayer, which exhibited rapid water transport and excellent salt rejection^[15]. Despite the success of these methods in achieving tailored pore sizes and densities, they can be relatively expensive due to the use of specialized instruments.

Generally, physical etching is an effective method to prepare arrays of monodisperse nanopores with a precise and tunable pore size distribution, whereas chemical etching [Figure 4E] is relatively easy and low-cost to produce nanopores in the graphene plane on a large scale. KOH, HNO₃, and others are commonly used to prepare porous graphene. Zhu et al. reported simple activations with KOH to generate nanoscale pores on the exfoliated GO nanosheets^[63]. The result in highly conductive, free-standing, and flexible porous GO paper possessed a very high specific surface area with excellent electrical conductivity and yielded outstanding performance, making it ideal for high-power energy storage^[61,63,64]. Zhao et al. introduced carbon vacancy pores into graphene nanosheets using a facile solution method that the HNO₃ reacted with the coordinatively unsaturated carbon atoms to partial detachment and removal of carbon atoms from the nanosheet^[65]. Wang et al. prepared porous graphene nanosheets by refluxing reduced GO (rGO) nanosheets in a concentrated HNO₃ solution^[66]. The diameters of nanopores can be readily modulated from several to hundreds of nanometers by varying the acid treatment time^[66]. Besides, H₂O₂^[67] and metal oxides^[68] can also be applied as chemical reagents to etch graphene.

In addition to graphene, other porous 2D nanosheets, such as h-BN^[69], MoS₂^[70,71], and g-C₃N₄^[19], can be obtained through artificial perforation approaches. The atomically thin 2D nanosheets hosting either intrinsical or artificial nanopores impart outstanding molecular transporting and sieving capabilities, making them promising materials for the direct fabrication of selectively permeable membranes with abundant nanofluidic channels.

Nanochannel membrane assembly strategies

In-plane parallel capillaries^[72] are another type of nanochannels that can be constructed by assembling isolated atomic planes of 2D nanosheets into van der Waals heterostructures made layer by layer in a precisely chosen sequence, namely van der Waals assembly [Figure 5A]. A typical stacking procedure starts by isolating micrometer-sized 2D nanosheets on top of a thin supporting film as one brick for the Lego wall, which are then put face-down onto a chosen target with the supporting film removed or dissolved. This process is repeated until the desired stack is assembled. While covalent solid bonds provide in-plane stability of 2D nanosheets, relatively weak, van der Waals-like forces are sufficient to keep the stack together. Radha has made outstanding contributions to this technique, having fabricated narrow and smooth capillaries through van der Waals assembly, with atomically flat graphene sheets at the top and bottom separated by spacers with a precisely controlled number of layers^[73]. They found that the water transport through the channels created by this method was characterized by an exceptionally fast flow that can be attributed to high capillary pressures and large slip lengths and can be associated with the structural ordering degree of nanoconfined water. They also investigated hydrated ion transport through ultimately narrow slits with dimensions approaching the size of small ions and water molecules^[74]. The ions with hydrated diameters larger than the slit size can still permeate through by distortions of their hydration shells. The mobility of ions under angstrom-scale confinement showed a notable dependence on the electric charges inside channels or at their entries. Using the same method, 2D channels made from graphene, MXene, and h-BN allowed helium gas flow that is orders of magnitude faster than expected from theory, whereas similar 2D channels made from MoS₂ exhibited much slower permeation that remains well described by Knudsen diffusion^[75]. It demonstrated the ballistic molecular transport effect that surface scattering could be either diffuse or specular, dependent on the fine details of the atomic landscape of the surface. The van der Waals assembly technique opens up an avenue to making capillaries with channel sizes tunable to angstrom precision and controllable transport properties through a wide choice of available 2D materials as channel walls. These series of 2D nanochannels in the forms of in-plane parallel capillaries provide an ideal experimental platform for offering a new understanding of many excellent nanofluidic observations, such as the unexpectedly high transport of thermal protons^[76,77],the qualitatively different coordination of square ice^[78], the anomalously low dielectric constant of confined water^[79], and the transistor-like electrohydrodynamic effect of hydrated ions^[80].

Figure 5. Schematics of the main assembly methods for 2D-material-based nanochannel membranes. (A) Building van der Waals heterostructures^[72]. Copyright 2013, Springer Nature Limited. (B) Pressure/Vacuum filtration. (C) Spin-coating. (D) Spray coating^[181]. Copyright 2011, Royal Society of Chemistry. (E) Rob casting^[181]. Copyright 2011, Royal Society of Chemistry.

The assembly of uniform 2D layered films from well-dispersed 2D nanosheets provides an ideal platform for constructing densely packed nanochannels for nanofluidic transport due to the large aspect ratio with atomic thickness and micron lateral dimension. Furthermore, the spaces, including in-plane slits (or defects) and plane-to-plane interlayer galleries within the 2D laminates, create numerous channels for nanofluidic transport. Among various feasible strategies, solution-assisted assembly offers a general, facile, and scalable way for high-throughput constructing densely packed nanochannels from 2D materials. The internal forces, such as electrostatic and van der Waals attractive interactions that are existed inside the 2D laminate, together with external forces, such as compressive, centrifugal, and shear forces that are applied outside the 2D laminate, are largely responsible for tuning the 2D building blocks into the ordered 2D laminar membranes^[81]. Predominantly used solution-assisted assembly methods include pressure/vacuum filtrating, coating, and layer-by-layer assembling.

The pressure/vacuum filtrating method [Figure 5B] is the most widely adopted to stack 2D laminar membranes. The thicknesses of the membrane can be straightforwardly controlled by altering the loading amount of 2D nanosheets via pressure control. For instance, Dikin et al. first reported the preparation of a free-standing GO membrane with macroscopic flexibility and stiffness by vacuum filtration of colloidal dispersions of individual GO nanosheets^[82]. The X-ray spectrum of a typical GO membrane showed a peak corresponding to a layer-to-layer distance (d-spacing) of about 0.83 nm, which can be attributed to an approximately one molecule-thick layer of water that is presumably hydrogen-bonded between the interlocked GO nanosheets within the laminates. Preparation parameters such as driving force, deposition rate, and substrate effect play critical roles in determining the structural formation of 2D laminar membranes. For example, Tsou et al. induced the assembly of GO membranes with different laminar microstructures ranging from highly ordered to highly random utilizing pressure, vacuum, and evaporation filtrating techniques^[83].

It is crucial to mention that due to the extremely thin feature, 2D laminates are usually supported by a porous substrate to form practical membranes. The differences in substrate properties also lead to a distinct assembly structure of resulting 2D laminar membranes with uneven performance. An example can be found in the study carried out by Zhang et al., who investigated the effect of substrate on GO membrane formation and separation^[84]. The surface morphology and chemical structure of substrates induced the assembly of GO and determined its adhesion. Furthermore, the bulk pore structure of substrates dominated the whole transport resistance of the GO membrane.

Besides filtrating, various coating methods, such as spin-coating, spray-coating, and casting, have been reported to assemble 2D laminar membranes. During the spin-coating process [Figure 5C], a solution containing 2D materials to be deposited is spread out uniformly over the substrate under centrifugal force, forming the ultrathin and laminar membranes. Kim et al. demonstrated two different spin-coating methods to prepare GO membranes on polymeric substrates^[85]. When contacting the substrate surface to the air-liquid interface of the GO solution followed by spin-coating, the repulsive edge-to-edge electrostatic interactions lead to an island-like assembly of GO nanosheets, resulting in a relatively heterogeneous GO stacking structure. In contrast, when contacting the substrate surface to GO solution only during spin-coating, the face-to-face attractive capillary forces created by the spin-coating overcome the repulsive interactions between GO edges, leading to a considerably dense GO laminar deposition. Chi et al. found that a speed balance between deposition and solvent (water) evaporation is crucial to obtain smooth GO membranes with uniformly aligned GO nanosheets via spin coating^[86]. Faster deposition may lead to overflow of the solution, while more rapid evaporation will cause uneven distribution of nanosheets. By matching up deposition and evaporation speeds, they obtained a uniform ultrathin GO membrane with a thickness of 20 nm that grants high gas fluxes for efficient gas separation. For spray-coating [Figure 5D], Guan et al. carefully controlled spraying times and evaporating rates to achieve facile structure manipulation of GO membrane from disordered-to-ordered and porous-to-compact^[87]. Ibrahim et al. further verified that using dilute concentration GO suspensions in spray-coating can help minimize the edge-to-edge interactions and reduce extrinsic wrinkles formation^[88]. Casting [Figure 5E] also belongs to coating methods that allow the continuous production of large-scale membranes. Akbari et al. utilized the flow properties of a nematic GO fluid in developing an industrially adaptable process to produce GO membranes with large in-plane order and stacking periodicity by shear-induced alignment of liquid crystals of GO^[89]. Zhong et al. showcased a universal, scalable, efficient continuous centrifugal casting method to produce highly aligned and compact 2D laminar membranes with impressive performances^[90]. Fluid mechanics analyses indicated that the simultaneous generation of shear force and centrifugal force during the continuous centrifugal casting process could be responsible for the alignment and compaction of 2D nanosheets, respectively.

Layer-by-layer assembly is a flexible construction process involving alternate deposition of different materials, often using one or more of the previously described methods. Such a flexible layer-by-layer construction process enables precise control over the thickness of the selective layer by varying the number of deposition cycles and is beneficial for introducing various interlayer stabilizing forces, including covalent, electrostatic, or hydrogen bindings between adjacent 2D nanosheets, thereby producing membranes with robust and ordered laminar structures^[81]. Hu et al. reported a 1,3,5-benzenetricarbonyl trichloride cross-linked GO membrane made via layer-by-layer deposition^[91]. The covalent cross-linking enhanced the stability of the stacked GO nanosheets, overcoming their inherent dispensability in the water environment while fine-tuning their charges, functionality, and spacing^[91]. In addition, layer-by-layer assembly has been used to create stable GO membranes through the electrostatic bonding of negatively charged GO nanosheets and positively charged polyallylamine hydrochloride^[92]. These methods provide a gateway to rationally design the charge properties and other functionalities of interlayer nanochannels of 2D laminar membranes by carefully choosing the intercalated molecules, polyelectrolytes, or nanomaterials. For instance, Song et al. developed both positively charged polyallylamine hydrochloride@GO and negatively charged polystyrene sulfonate@GO, which they alternately deposited on polycarbonate substrates by the layer-by-layer assembly, producing polyelectrolyte intercalated GO membranes with tunable charge-gating ion exclusion effects^[93]. Apart from a single interaction, Zhao et al. fabricated GO-based hybrid membranes via layer-by-layer self-assembly of gelatin molecules and GO nanosheets driven by multiple interactions, including electrostatic attraction, hydrogen bond, and hydrophobic interaction^[94]. Electrostatic attractions were formed between ionized carboxyl groups on GO and protonated amino groups on gelatin, hydrogen bonds between various polar groups on gelatin and GO, and hydrophobic interactions between the hydrophobic carbon backbone of GO and hydrophobic amino acid side chains on gelatin. These multiple interactions may contribute to the formation of highly ordered 2D laminar membranes with finely tuned 2D nanochannels for nanofluidic transport.

In summary, the structural properties of nanochannels in nanochannel films based on 2D materials, including layer thickness, stacking order, defects and grain boundaries, surface functionalization, embedding of additives or polymers, and nanosheet alignment and orientation, largely affect the overall filtration performance of the membrane. Controlling these factors can tune the interlayer spacing, pore size distribution, surface interactions, and flow dynamics within nanochannels, leading to membranes with higher selectivity, permeability, and fouling resistance. The choice of synthesis and assembly method depends on the desired material, application, scalability, and desired quality of 2D nanosheets. Each method has its advantages and limitations. Further studies to understand and optimize these structural properties will facilitate the development of more efficient and tailored nanochannel membranes for various separation applications.

Characterization methods

Characterization techniques are essential for understanding the properties of 2D materials. Various techniques allow for a thorough examination of these materials. XRD is used to determine the crystal structure and orientation of 2D materials. It involves directing X-rays onto the material and measuring the resulting diffraction pattern. XRD can provide information about lattice parameters, crystal symmetry, and the presence of specific crystalline phases. In addition, the layer spacing is calculated using XRD data according to Bragg’s law. Optical spectroscopy techniques such as UV-Vis absorption spectroscopy and photoluminescence spectroscopy are used to study the optical properties of 2D materials. They can reveal information about the material band gap, exciton properties, and light-matter interactions. X-ray photoelectron spectroscopy (XPS) is used to analyze the chemical composition and electronic states of 2D materials. It involves irradiating the surface of a material with X-rays and measuring the energy of the emitted electrons. XPS can provide information about elemental composition, chemical bonding, and the presence of impurities or functional groups. Raman spectroscopy is used to analyze the vibrational modes of 2D materials. It involves shining a laser on the material and measuring the scattering spectrum. Raman spectroscopy provides insight into the crystal structure, composition, and strain of material and identifies different types of 2D materials. Atomic Force Microscopy (AFM) is a powerful technique for imaging the morphology and surface properties of 2D materials. It uses a sharp probe to scan the entire surface of a material, detecting force changes between the probe and the sample. AFM can provide information about the height, roughness, and mechanical properties of a material at high spatial resolution. Scanning electron microscopy (SEM) provides high-resolution images of the surface/sectional morphology of 2D materials. It uses a focused electron beam to scan the surface of a material, generating detailed images that reveal features such as the size, shape, and arrangement of thin sheets or layers of material. Transmission electron microscopy (TEM) is used to study the internal structure and atomic-level features of 2D materials. It involves the transmission of an electron beam through a thin sample to image the atomic structure and defects of the material. TEM can provide information about crystal structure, grain boundaries, stacking sequences, and even individual atomic arrangements.

Among other things, these characterization techniques allow researchers to gain a comprehensive understanding of the structural, morphological, chemical, and optical properties of 2D materials. By combining multiple techniques, a complete characterization of 2D materials can be achieved, facilitating their optimization and utilization in a variety of applications. The development of advanced characterization techniques, such as cryoelectron microscopy and in-situ Raman, will provide researchers with more reliable means and opportunities to explore the deeper mechanisms of structure and separation and purification of 2D-material-based membranes^[95-98].

Regulating size of nanochannels

In 2D-material-based nanochannel membranes, nanofluidic transport occurs in interlayer capillaries and wrinkles formed by adjacent 2D nanosheets and inter-edge gaps and intrinsic pores formed on planar 2D nanosheets interconnect with each other to create numerous nanochannels. Firstly, the channel size [Figure 6], one of the most prominent characteristics of nanochannels, firmly decides the entry of nanofluids. Joshi et al. found that the interlayer spacing of a GO membrane was ~ 0.9 nm^[16], allowing any ion or molecule with a hydrated radius of 0.45 nm or less to enter the nanochannels and permeate at a speed order of magnitude faster than would occur through simple diffusion, while all species larger than this are sieved out. Such sharp size cutoff determined by the interlayer spacing has significant implications on a myriad of occasions. By adjusting the channel size through diverse methods, such as confinement, reduction, cross-linking, and intercalation, a broad range of different-sized nanochannels can be designed to sieve target ions and molecules from the bulk solution precisely.

Figure 6. Regulating the size of nanochannels. (A) Physical confinement. (B) The stable, optimized geometries of Na⁺-(H₂O)₆@GO clusters from density functional theory computation^[101]. Copyright 2017, Springer Nature. (C) The fabrication procedure of free-standing rGO membranes using reduction control^[104]. Copyright 2015, John Wiley & Sons, Inc. (D) Original GO membrane (left) and the composite GO-framework prepared by p-phenylenediamine cross-linking GO membrane (right)^[105]. Copyright 2014, American Chemical Society. (E) Graphene/carbon nanotube composite membranes are prepared by assembling refluxed GO (rGO) and multi-walled carbon nanotubes^[120]. Copyright 2015, American Chemical Society.

Due to the inevitable swelling effect when immersed, hydration would increase the size of nanochannels and deprive their sieving capability. Achieving a small channel size for the 2D laminates immersed in solvents has been a challenging task. Abraham et al. described a physical confinement method [Figure 6A] to control the interlayer spacing from ~ 9.8 Å to ~ 6.4 Å to obtain accurate and tunable ion sieving^[99]. GO laminates were stored at different relative humidities to yield controllable interlayer spacing attributed to incorporating water molecules into various sites between GO nanosheets. Subsequently, epoxy was used to encapsulate stacked GO laminates for preparing physically confined GO membranes since the epoxy mechanically restricts laminate swelling upon water exposure. As ions and water permeate along the GO nanochannel, the permeation rate for Na⁺ and K⁺ showed an exponential dependence, decreasing by two orders of magnitude as interlayer spacing decreased from 9.8 Å to 7.4 Å. However, the water permeation rate showed only a slight variation, reducing by a factor of ~ 2 within the same range of interlayer spacing. The former is related to the partial clogging of graphene capillaries, and the latter is attributed to a low barrier and a large slip length for water in graphene capillaries. Similarly, Li et al. reported an external pressure regulation method for controlling the interlayer spacing of GO laminates against swelling^[100].

Apart from channel size regulation based on external forces, cationic control is a representative method of channel size regulation based on internal forces. Chen et al. demonstrated cationic control of the interlayer spacing of GO membranes with angstrom precision using K⁺, Na⁺, Li⁺, Ca²⁺, or Mg²⁺ ions [Figure 6B]^[101]. The confinement of interlayer spacing is mainly due to the interaction between hydrated cations and aromatic rings (cation-π interactions) on the GO nanosheet and the interaction between hydrated cations and the oxygenated groups on the GO nanosheet. As a result, the interlayer spacing controlled by one type of cation can efficiently and selectively exclude other cations with larger hydrated volumes. Especially the K⁺-controlled GO membranes can even reject K⁺ itself owing to the comparable interaction energy between K⁺ and GO nanosheets concerning the dehydration energy of K⁺.

Reduction is another effective method to control the small channel size for the solvated 2D laminates. These 2D nanosheets possess abundant functional oxygenated groups that open the interlayer spacing between adjacent nanosheets and allow solvent molecules to intercalate into 2D laminates. Partially removing these oxygenated functional groups through various thermal or chemical reductions can enhance interlayer π-π interactions, thereby producing narrowed and stabilized nanochannels. Qiu et al. reported that hydrothermal treatment could readily control the reduction of GO nanosheets in water^[102]. Meanwhile, Han et al. obtained an rGO dispersion by base-refluxing to generate moderate holes and create a large number of permeation “gates”, ultimately increasing the permeability of the rGO membrane^[103]. However, the pre-treatment of reduction may lead to the worse dispersion of nanosheets that is not conducive to uniformly assembling them into well-stacked laminates. Liu et al. prepared a GO membrane beforehand and placed it above a hydrogen iodide (HI) solution [Figure 6C]^[104]. The HI steam acts as a reducing agent, which reduces GO and triggers the initial delamination between the rGO membrane and substrate. Fumagalli et al. found that GO laminates reduced using thermal, HI, and VC reductions are highly impermeable to strong chemicals and salt solutions, owing to a high degree of graphitization of the laminates, causing nanochannels collapse during the reduction process^[79].

The enlarged channel size can be achieved by cross-linking or intercalating cross-linkers, such as specific molecules, polyelectrolytes, or nanomaterials, to reinforce transport efficiency. The most commonly used cross-linkers for GO membranes are diamines^[105], including ethylenediamine (EDA), butylenediamine (BDA), and phenylenediamine (PDA), which can chemically bond with adjacent GO nanosheets through condensation and nucleophilic addition reactions [Figure 6D]. The spatial configuration of amine monomers primarily determines the channel size of cross-linked GO membranes in a dry state, whereas the bonding strength of amine monomers within the GO laminate likely determines their channel size in a wet state^[106]. Other molecules holding characteristic terminal groups, such as dicarboxylic acids^[107], tannic acid^[108], borate^[109], isophorone diisocyanate^[110], porphyrins^[111], and polybenzimidazole^[112], have also been proved to cross-link 2D-material membranes to address 2D channel size control chemically. Yang et al. demonstrated the preparation of thiourea covalently linked GO membrane where thiourea bridged GO laminates periodically through the reactions of amino and thiocarbonyl groups with the functional groups of GO, leading to mechanically stable and structurally well-defined 2D channels^[113]. These molecule cross-linkers usually yield covalent bonds through a limited number of active sites. In contrast, polyelectrolytes possess long polymer chains with abundant functional groups, which can serve as polymer cross-linkers to be incorporated into 2D laminates through sufficient active sites to produce a dense composite structure. The commonly used polymer cross-linkers are polyvinyl amine (PEI)^[114,115], polyvinyl alcohol (PVA)^[116], polyethylene glycol (PEG)^[117], and so on. Ran et al. utilized imidazolium-functionalized brominated poly (2,6-dimethyl-1,4-phenylene oxide) (Im-PPO) and sulfonated poly (2,6-dimethyl-1,4-phenylene oxide) (S-PPO) to connect neighboring GO nanosheets via non-covalent π-π, electrostatic, and hydrogen bonding interactions^[118]. These cross-linking strategies may pave the way to access highly stable and efficient transport of 2D lamellar membranes.

Besides cross-linking, intercalating nanomaterials of specific sizes is another means to significantly improve the permeability of 2D-material membranes, as it effectively creates large amounts of broadened pathways for nanofluidic transport. Carbon-based nanomaterials, metal oxide nanoparticles, MOFs, and COFs are among the representative intercalated nanomaterials. By embedding carbon nanodots of controllable sizes, Wang et al. were able to tune the permeability of GO membranes^[119]. Han et al. expanded the interlayer spacing of GO membranes by intercalating multi-walled carbon nanotubes (CNTs) [Figure 6E]^[120]. Goh et al. further proposed that intercalating of CNTs of different diameters can effectively inhibit the restacking (or aggregation) of GO nanosheets and thus create plenty of nanochannels for water transport while retaining the molecular sieving capability of the ensuing membranes. Moreover, CNTs can also act as anchors to interact and interconnect with the adjacent GO nanosheets due to the perfect compatibility of carbon-based materials to reinforce the membrane stability^[121]. Intercalation of metal oxide nanoparticles is another general and scalable approach to expand 2D nanochannels significantly. Zhang et al. prepared nanoparticles@GO membranes using a simple in-situ solvothermal synthesis method, where size- and density-controllable nanoparticles were uniformly grown on GO nanosheets through coordination^[122]. Compared to the membrane produced by filtering GO solutions mixed directly with nanoparticles, the in-situ fabricated nanoparticles@GO membrane uniformly introduced nanoparticles within GO laminates without damaging their initial ordered stacking structures and hence exhibited ultrahigh water permeance and excellent rejections for various solutes in water, as well as good stability under high pressure and cross-flow operation. This general concept of intercalation can also be demonstrated by several kinds of nanoparticles such as TiO₂^[123], silica^[124], and hydroxy sodalite nanocrystals^[125], endowing nanoparticles intercalated GO membranes with additional multifunctions. Compared with nonporous metal oxide nanoparticles, intrinsically porous nanocrystals are more beneficial for fast nanofluidic transport when they are intercalated into 2D nanochannels. By incorporating MOFs^[126,127], COFs^[128,129], and biomimetic water channels^[130] with extra internal pathways through sub-nano-sized apertures as the microporous fillers into 2D laminates, both the size and the number of nanochannels are increased, leading to greatly enhanced nanofluidic transport performance. Li et al. described a route for fabricating MOFs channeled graphene composite membranes with molecular sieving properties using in-situ crystallization. A series of MOFs, including ZIF-8, ZIF-7, CuBTC, and MIL-100, were showcased to be impregnated into interlayers of GO laminates, which firmly anchored and bolstered up GO laminates by coordination bonds to form a highly porous architecture with uniform nanochannels^[127].

Khan et al. created a hybrid laminar membrane by assembling COFs and GO nanosheets^[128]. The incorporation of COFs nanosheets provides a large number of pores that shorten the transport pathway while retaining the interlayer distance. Similarly, Sui et al. also intercalated rigid 2D COFs into GO laminates to realize a robust GO/COF laminar membrane^[129]. The atomically thin 2D COFs with pores serve as a nano spacer to increase the interlayer spacing between GO nanosheets and provide direct transfer channels, thereby reducing water transfer resistance. On the other hand, the COFs enhance the self-supporting capacity of GO networks on a substrate with large pores. As a result, this strategy led to a significant increase in the water permeance of the optimized GO/COF laminate membrane compared to the pristine GO membrane without compromising its rejection rates to organic dyes. Mao et al. embedded imidazole-ureido bola-amphiphile-imidazole compound as biomimetic imidazole-quartet water channels into assembled GO laminates to enhance water transport selectivity over butanol benefiting from the hydrophilic water-preferential nanochannels along the imidazole-ureido molecular scaffolds^[130].

As the GO membrane is a showpiece that provides flexible platforms for developing versatile novel 2D-material membranes, the methods mentioned above for regulating channel size are adaptable for other 2D-material membranes. For example, Wang et al. related the performance of MoS₂ membranes to the size of their nanochannels in different hydration states^[131]. They found that the water-impermeable behavior of the dry MoS₂ nanochannel, which has a 0.62 nm interlayer spacing, is caused by the irreversible nanosheet restacking during a drying process. In comparison, the fully hydrated MoS₂ membrane possesses a stable 1.2 nm interlayer spacing, leading to high water permeability and moderate-to-high ionic and molecular rejection. Meanwhile, the MoS₂ nanochannel has a much stronger van der Waals attraction force than the GO nanochannel, which prevents the interlayer spacing from increasing, thereby ensuring the aqueous stability of MoS₂ membranes. Inspired by the structural characteristics of GO membranes, where oxidized zones act as spacers to provide a relatively large interlayer distance to accommodate water molecules, Ran et al. incorporated appropriate acid spacers between g-C₃N₄ interlayers to enlarge the width of the 2D channels^[132]. The intercalation molecules successfully break up the tightly stacked structure of g-C₃N₄ laminates. Accordingly, the modified g-C₃N₄ membranes give rise to two orders of magnitude higher water permeance without sacrificing the separation efficiency. In a similar route, Wang et al. devised a method to partially exfoliate dg-C₃N₄ nanosheets with artificial nanopores and unstripped fragments as self-supporting spacers and assemble them into 2D laminar membranes for water purification^[19]. The artificial nanopores and the spacers between the partially exfoliated g-C₃N₄ nanosheets provide numerous nanochannels for nanofluidic transport with ultralow friction while effectively retaining larger molecules. Wang et al. also demonstrated a strategy for stabilizing the Ti₃C₂T_x laminar architecture by alginate hydrogel pillars formed between the adjacent nanosheets^[133]. After pillared by different alginate hydrogel pillars, the nanochannel diameters are effectively fixed at ~ 7.4 Å, and the resulting membrane exhibited significantly enhanced the ion-sieving property with distinct ions permeation cutoff depending on the multivalent cations cross-linked with alginate molecules.

Regulating the length of nanochannels

Another physical factor influencing nanofluidic transport through 2D-material-based nanochannel membranes is channel length, which largely depends on the membrane thickness and porosity. Nanoporous 2D materials of single- or few-atom thickness are the ultimate building blocks for constructing ultrathin membranes with minimal resistance to maximize permeance.

As a straightforward way to shorten the channel length, many theoretical and experimental studies have demonstrated the ultrafast permeation performance of ultrathin 2D-material membranes. For example, Cohen-Tanugi et al. predicted that ultrathin nanoporous graphene membrane could have water permeability several orders of magnitude higher than conventional membranes thanks to the chemical functionalization, which may have a valuable role to play in water purification [Figure 7A]^[134]. Han et al. fabricated ultrathin (~ 22-53 nm) graphene nanofiltration membranes on microporous substrates for efficient water purification^[103]. Liu et al. prepared free-standing ultrathin rGO membranes with thickness down to ~ 20 nm by HI vapor and water-assisted delamination^[104]. Yang et al. reported highly laminated GO membranes of only several layers in thickness (~ 8 nm), exhibiting outstanding sieving properties accompanied by ultrafast solvent permeation^[135]. Li et al. described a reproducible facile filtration method to produce ultrathin GO membranes down to 1.8 nm in thickness, which exhibited superior gas separation performance^[136]. Furthermore, single-layer 2D-material membranes for practical use have also been attempted. For example, Heiranian et al. showed that a single-layer nanoporous MoS₂ effectively allowed water transport at a high rate associated with permeation coefficients, energy barriers, water density, and velocity distributions in the pores [Figure 7B]^[137].

Figure 7. Regulating length of nanochannels. (A) side view of the computational system of ultrathin graphene membrane in the desalination process^[134]. Copyright 2012, American Chemical Society. (B) The simulation box consists of a single-layer MoS₂ sheet (molybdenum in blue and sulfur in yellow), water (transparent blue), ions (in red and green), and a graphene sheet (in gray)^[137]. Copyright 2015, Springer Nature. (C) The fabrication process of density pores on graphene-nanomesh/carbon-nanotube hybrid membranes using O₂ plasma^[138]. Copyright 2019, Science Publishing Group.

Although the ultrathin 2D-material membranes exhibit exceptional permeation performances, the limited mechanical strength of these membranes over large areas remains a hindrance to their widespread use. To overcome this limitation, Yang et al. reported the production of an atomically thin nanoporous membrane with a single-layer graphene nanomesh (GNM) supported by an interwoven network of single-walled carbon nanotubes (SWNT)^[138].The monolayer GNM featuring high-density subnanometer pores [Figure 7C] allows efficient transport of water molecules with minimum resistance while effectively blocking solute ions or molecules to enable size-selective separation. The mechanically strong, interconnected SWNT network, acting as the microscopic framework, separates the GNM into microsized islands, thus ensuring the structural integrity of the atomically thin GNM. The resulting large-area, ultrathin GNM/SWNT hybrid membrane showed high water permeance and excellent size selectivity combined with excellent anti-fouling characteristics, making it highly attractive for energy-efficient and robust water treatment.

Regulating morphology of nanochannels

The morphology of the channel in ultrathin 2D-material membranes plays a crucial role in determining the distance traveled by nanofluids. Inter-edge gaps [Figure 8A] and intrinsic pores [Figure 8B] are crucial elements that influence this distance^[135,136]. Ibrahim et al. proposed using nanofluidic pathways in laminar GO membranes, where permeation occurs through pinholes within GO flakes and capillaries between them^[139]. Utilizing small and porous nanosheets to assemble membranes can efficiently introduce more inter-edge gaps and intrinsic pores, enabling nanofluids to take shortcuts to perform ultrafast transport. A few works have developed simple methods for the lateral size fractionation of 2D nanosheets using techniques such as sonication^[140], filtrations^[141], differential centrifugation^[142], and controlled directional freezing^[143]. It is found that when the dimension of nanosheets within the membrane is changed from microsize to nanosize, the amount of nanofluidic pathways formed within the GO membrane is increased significantly, resulting in the enhancement of trans-membrane transportation in the case of nanosized 2D-material membranes^[144]. Nie et al. exploited this concept of lateral dimension control to engineer shorter and less tortuous transport pathways for solvent molecules, leading to the development of small-flake GO membranes that achieved ultrafast selective molecular transport^[145]. The methanol permeance in these membranes reached up to 2.9-fold higher than its large-flake GO counterpart, with high selectivity towards organic dyes^[145]. Section 2.2 has described various perforation techniques to produce nanopores on 2D nanosheets^{[15,19,58-71]}. The creation of nanopores can provide a greater number of transport nanochannels for nanofluids and sharply reduce the average transport distances. Ying et al. reported GO membranes with introduced in-plane mesopores by a reoxidation process, demonstrating 2-3 folds enhancement of water permeance as that of the pristine GO membranes^[146]. Li et al. also proposed thermally reduced nanoporous GO membranes through a combination of mild H₂O₂ oxidation and moderate thermal treatment, increasing the water permeability by 26 times^[147]. Those intrinsically porous 2D materials are more attractive as building blocks to assemble 2D laminar membranes. Their uniform nanopores consisting of angstrom-sized windows and nanometer-sized cavities can provide numerous cross-layer shortcuts and ensure precise size exclusions. Peng et al. reported ultra-permeable and super selective molecular sieve membranes made of 2D MOF nanosheets, which achieved H₂ permeance of up to several thousand GPU with H₂/CO₂ selectivity greater than 200^[37,148]. Shinde et al. demonstrated ultrathin 2D COF membranes with a well-defined ordered porous structure^[149]. These membranes displayed remarkable permeabilities for polar and nonpolar organic solvents, which were approximately 100 times higher than the amorphous polymer membranes^[149].

Figure 8. Regulating morphology of nanochannels: inter-edge gaps (A), intrinsic pores (B), and nanowrinkles (C).

Besides the tortuous transport pathways through internal nanochannels of 2D laminar membranes extending in horizontal and vertical directions, nanowrinkled morphologies [Figure 8C] also exist in 2D laminar membranes, whose effects in transmembrane transport should not be neglected. Xi et al. proposed an rGO membrane with 2D nanochannels uniformly confined with a space size of ~ 8 Å^[150]. The mild reduction avoids the hydrothermal corrugation of GO nanosheets and thus enables the creation of highly parallel 2D nanochannels for precise sieving of mono-/multi-valent metal ions^[150]. Li et al. also demonstrated a mild-thermal annealing for preparing rGO membrane for nanofiltration since the mild reduction condition might favor the formation of a more ordered and better-controlled transport nanochannel^[151]. Contrary to these efforts to eliminate nanowrinkles, Saraswat et al. hypothesized that the imperfect stacking in the 2D laminates could lead to voids, wrinkles, and disordered microstructures that could provide alternative non-ideal transport pathways for nanofluids, resulting in a higher effective permeance^[152]. Kang et al. further revealed the roles of nanowrinkles in mass transport across GO membranes^[153]. They found nanowrinkles by themselves serve as fast transporting ways while their connection with narrow interlayer channels can form a selective network^[153]. Huang et al. developed a nanostrand-channeled GO membrane, whose permeance offered a 10-fold enhancement without sacrificing the rejection rate compared with that of pristine GO membrane, attributing to the generation of more nanofluidic channel networks in the membrane^[154]. The nanostrand-channeled concept is also extendable to other 2D laminar membranes, such as MXene^[17] and WS₂ membranes^[155], for increased permeability. These findings corroborate that nanowrinkles serve as fast tracks for nanofluids to enhance membrane permeability. Considering nanowrinkles broadly existing within flexible 2D nanosheets, they are expected to show critical transport­controling effects in nanofluidic transport in 2D-material membranes.

Regulating surface physicochemical properties of nanochannels

The 2D-material nanosheets have abundant functional groups. During the filtration process, the 2D material-based membrane is prone to swelling effect, resulting in the collapse of nanochannels and loss of sieving performance. Targeted modification of 2D-material surfaces is, therefore, significant, and a number of methods have been developed for modifying the chemical properties of membrane surfaces [Figure 9]. One commonly used strategy is the Chemical intercalation of cross-linking agents (e.g., ions, small molecules, or macromolecules), which enhances the interlayer force and improves the mechanical stability of 2D nanosheet membranes. At the same time, this strategy modulates the affinity of water/ions to the nanosheet surfaces, thereby regulating their mobility as they pass through the channels.

Figure 9. Regulating surface physicochemical properties of nanochannels. (A) Intercalating metal ions. [a] A snapshot of the ab initio molecular dynamics simulation at 27 ps. [b] The interaction energy between hydrated cations and graphene oxide sheets [labeled cation-(H₂O)₆@GO] and the hydration energy of the cation [labeled cation-(H₂O)₆]. [c–e] The most stable optimized geometries of cation-(H₂O)₆@GO clusters from density functional theory computation, where the cations are Na⁺, K⁺, and Li^+[101]. Copyright 2017, Springer Nature. (B) A GO polymer cross-linked network composite membrane for forward osmosis desalination^[158]. Copyright 2017, The Royal Society of Chemistry. (C) Restacked Ti₃C₂T_x-based membrane by introducing reduced GO as the spacer^[160]. Copyright 2019, Springer Nature. (D) Oxidation process from graphite to graphene oxide^[161]. Copyright 2021, MDPI.

Modifying the surface of 2D materials with a small number of metal ions, such as Mg²⁺ and Ca²⁺, is considered an effective strategy to improve the membrane performance significantly. This enhancement is mainly due to the electrostatic bonding between metal ions and functional groups on the surface of 2D materials^[156]. To further understand this mechanism, Chen et al. modified the GO membrane with hydrated Na⁺ ions and found, through DFT calculations, that the hydrated ions were adsorbed through hydrogen bonds in the region where the oxidized groups and aromatic rings coexisted, effectively adjusting and fixing the interlayer distance^[101]. The interlayer distance is proportional to the hydration radius of the corresponding Li⁺ > Na⁺ > K⁺ class ions [Figure 9A]. Ding et al. effectively reduced the d-spacing to ~ 1.5 nm. They effectively suppressed the swelling effect by intercalating Al³⁺ into the Ti₃C₂T_x MXene film combined with abundant surface termini (such as ¼O, eOH, and eF functional groups) within the nanosheets^[157]. Cross-linking 2D materials by organic molecules is another effective strategy to tune and fix the interlayer distance in layered films. Epoxy, a commonly used chemical raw material, is employed to modify the surface of 2D-material nanosheets. This strategy successfully prevented the swelling of GO and produced nearly 97% NaCl repulsion^[99]. Hung et al. cross-linked GO with EDA monomer to make the nucleophilic substitution condensation reaction between the oxygen-containing groups of amine and GO to form a CeN covalent bond, which effectively suppressed the swelling effect^[105]. In-situ polymerization on the surface of 2D materials to generate twins is also a commonly used strategy^[72-75]. Kim et al. using N,N’-methylenebisacrylamide (MBA) as the cross-linking agent, N-isopropylacrylamide (NIPAM) as the monomer, and then ammonium persulfate (APS) as the initiator, successfully synthesized the GO@polymer twins^[158]. The prepared membranes have excellent anti-swelling properties and desalination ability^[158,159] [Figure 9B].

The potential interaction of functional groups on the surfaces of different 2D materials makes the superimposition of different 2D materials a possible strategy for modifying the surface chemistry of materials [Figure 9C]. Xie et al. reported a Ti₃C₂T_x MXene-based membrane whose microstructure was optimized by inserting rGO between the layers^[160]. The surface of the membrane was progressively hydroxylated to increase the accessibility of Ti₃C₂T_x, thus improving the wettability of the film and enhancing the adsorption and reduction of heavy metal ions.

Oxidation and reduction processes are commonly used to modify and prepare 2D materials, so surface properties are often modified during the preparation process. For example, using a modified Hummers method (i.e., oxidation procedures), Alkhouzaam et al.^[161] prepared a GO membrane with more active surface properties than a graphene membrane, which facilitates its functionalization to meet the request [Figure 9D].

The properties of nanochannel membranes based on 2D materials, including swelling resistance and water permeability and retention, are closely related to the transport control effects of nanochannels and the interleaving of surface properties. These factors play a key role in determining the overall performance and functionality of the membranes. Therefore, it becomes imperative to thoroughly evaluate and consider the effects of various modification methods on the nanochannels. This careful evaluation enables researchers to select and optimize the most suitable 2D-material-based membranes for specific applications. By understanding and exploiting the complex relationship between nanochannels and surface properties, researchers can unlock the full potential of these membranes in areas as diverse as water purification to energy storage.

Liquid molecular separation

Membranes are highly susceptible to swelling effects when applied in liquid environments (i.e., liquid molecular separations, including liquid solvent-molecular solute separations and liquid-liquid separations), leading to the collapse of nanochannels, reduced selectivity, and shortened membrane lifespan. Therefore, it is imperative to improve the swelling resistance of membranes in liquid environments compared to membranes in other conditions.

A series of anti-swelling MXene nanochannel membranes were synthesized by Xing et al. using the interaction forces generated by introducing negatively charged polymers into positively charged MXene membranes^[162]. And by tuning the ratio of polymer and MXene, different sizes of nanochannels can be obtained for separating different hydrated molecules [Figure 10A]. Xing et al. reported a separation membrane that can adjust the size of nanochannels by operating pressure, which can retain 99.9% of hydrated dye molecules under a small operating pressure and 51.8% of the hydrated ions under a large operating pressure^[14]. Liquid-liquid separation plays a pivotal role in chemical production. Qi et al. reported a GO-based nanochannel membrane with an interlayer channel size between water and methanol molecules, which can effectively separate methanol-water mixtures with a high separation efficiency of 99.89 wt.% [Figure 10B]^[163]. For a better comparison of related membranes, a summary of liquid molecular separation membranes is shown in Table 2.

Gas separation

Unlike other separations, gas separation is more sensitive to adsorption and desorption on membrane surfaces, requiring smaller nanochannels than other separation systems used throughout many industries, such as methane reforming, CO₂ capture, and hydrogen purification. Therefore, Controlling the nanochannel size and modifying the surface is essential.

The preparation of high-performance hydrogen purification membranes has become critical for constraining hydrogen energy to replace fossil dyes in catalytic hydrogen production processes. Ding et al. prepared a two-position separation membrane with a nanochannel size ~ 0.35 nm, which possesses H₂ permeability of > 2,200 Barrer and H₂/CO₂ selectivity of > 160, demonstrating excellent eventual commercialization potential [Figure 10C]^[164]. Carbon neutrality has become a hot topic in recent years. The separation of CO₂/N₂ is a prerequisite for CO₂ capture, leading to the successful synthesis of organic matter from CO₂. Zhou et al. cross-linked piperazine with GO, resulting in a membrane with a high affinity for CO₂, thus significantly improving the separation efficiency of CO₂/N₂ [Figure 10D]^[165]. The separation of hydrogen isotopes is vital for medical diagnosis and treatment. Lozada-Hidalgo et al. reported that graphene monolayers and BN membranes could separate hydrogen ion isotopes with a separation factor of about 10^[166]. For a better comparison of related membranes, a summary of gas separation membranes is shown in Table 3.

Ion sieving

Recently, researchers have paid increasing attention to high-performance ion sieve membranes in addition to traditional separation methods. Membranes with various properties are required to meet the different needs of ion sieve membranes for applications such as water desalination, microcurrent, hydrogen production, and energy storage.

In seawater desalination, stringent channel dimensions are necessary, with nanochannels often needing to be sub-nanometers in size to achieve ultra-high desalination efficiency. Chen et al. reported a graphene desalination membrane with sub-nanopores that achieved 99.99% NaCl rejection with an ultrafast water flux combined with evaporation methods [Figure 10E]^[167]. In energy storage applications, ion sieve membranes are often used as diaphragms, where excellent ionic conductivity and electrical insulation are required. Ghazi et al. synthesize a MoS₂/celgard separator with outstanding lithium ion passage and polysulfide retention capacity, which can effectively inhibit the shuttle effect in lithium-sulfur batteries, significantly improving the battery performance^[168]. Yang et al. prepared a metal-organic framework as a multifunctional ionic sieve membrane for aqueous zinc–iodide batteries, greatly extending the battery lifespan^[169]. The recovery of rare metals has become increasingly popular due to the rising popularity of electric vehicles and electronic products. Efficient screening of metal ions is the key to recovering rare metals. Xu et al. developed a special lithium ion sieving UiO-67/AAO membrane, which achieved an ultra-high Li⁺ permeability of 27.01 mol m^-2 h^-1 and a Li⁺/Mg²⁺ selectivity of up to 159.4 [Figure 10F]^[170]. For a better comparison of related membranes, a summary of ion sieving membranes is shown in Table 4.

Important factors to consider when selecting a 2D nanochannel membrane for separation are the object to be separated, expected performance, ideal structure, cost-effectiveness, and environmental impact. In various application scenarios, choosing the appropriate membrane fabrication and modification process is the precondition for the membrane to ensure the ideal performance. Taking these factors into account can lead to the successful development of efficient and environmentally friendly separation processes.