Covalent organic framework membranes for lithium extraction: facilitated ion transport strategies to enhance selectivity

Abstract

The surging global demand for lithium, driven by the proliferation of electric vehicles and energy storage technologies, has exposed significant limitations in conventional lithium extraction methods, including inefficiency and environmental harm. Covalent organic frameworks (COFs) have emerged as a promising platform to address this challenge and enable more sustainable lithium extraction, owing to their unique advantages such as precisely tunable pore sizes, robust stability, and the ability to incorporate functional binding sites for selective ion transport. This review focuses on structural design and functionalization strategies in COFs to optimize lithium-ion separation, highlighting how pore confinement effects, tailored interlayer stacking arrangements, and strategic functional group modifications can dramatically enhance Li⁺ selectivity over competing ions present in brine solutions. A particular emphasis is placed on the fundamental energy barriers associated with lithium-ion transport. In particular, we discuss how appropriately designed pore environments and lithium-binding functional groups reduce the dehydration energy required for Li⁺ to enter and traverse COF nanochannels, thereby facilitating faster and more selective Li⁺ conduction. We also survey recent advancements in COF-based lithium separation technologies, such as high-performance COF membranes and sorbents for extracting lithium from brines and seawater, evaluating their potential, as well as remaining challenges, for sustainable industrial implementation. This review provides a comprehensive understanding of how advanced COF engineering can enable efficient and selective lithium-ion transport, offering valuable insights for the development of next-generation lithium extraction materials and technologies.

---

Wider impact

The global transition to electrified transportation and renewable energy has triggered an unprecedented surge in lithium-ion battery (LIB) production, with demand projected to far outpace supply in the coming decade. Efficient and selective extraction of lithium from complex brines and seawater is therefore critical for securing sustainable lithium resources. This review provides a mechanistic overview of covalent organic framework (COF)-based membranes as a transformative solution for lithium separation, focusing on pore structure engineering, chemical functionalization, and ion transport energetics. COF membranes offer a unique combination of molecular precision and tunability, enabling them to address key separation challenges that conventional membranes or sorbents cannot. Beyond lithium extraction, the insights developed here have broad implications across membrane science, materials chemistry, and clean energy technologies. As LIB manufacturing scales globally, innovations in membrane-based resource recovery will play an increasingly vital role. The structure–function relationships highlighted in this review lay the foundation for next-generation COF membranes tailored not only for lithium but also for other critical elements, bridging fundamental science with real-world impact in energy storage, water purification, and sustainable materials processing.

1. Introduction

Lithium is one of the most crucial elements in modern society with diverse applications, including pharmaceuticals,^1–3 aerospace^4,5 and nuclear industries,⁶ as well as the well-known energy storage field.⁷ Driven by the significant expansion of the electric vehicle market, the demand for lithium has surged dramatically in recent years (Fig. 1). In 2023, nearly 14 million new electric cars were registered globally, increasing the total number of electric cars on the road to 40 million.⁸ This would drive the global lithium production to reach 964 000 tons of lithium carbonate equivalent (LCE), while the demand was projected to be 989 000 tons LCE by 2040, indicating a huge supply-demand gap.⁸ In contrast, the global reserves of ore-based lithium resources are estimated at 28 million tons, with the mining process causing a significant environmental impact, including water contamination, soil damage, and air deterioration.^9,10 Clearly, it highlights the necessity of developing efficient and sustainable methods for lithium extraction from liquid-based resources. As one of the liquid-based lithium resources, the salt lake brines, particularly those located in regions like the Puna Plateau (Chile-Argentina-Bolivia) and the Qinghai-Tibet Plateau of China, offer substantial lithium reserves with concentrations ranging from dozens to thousands of ppm. Specially, the lithium resources in Qinghai Salt Lake exceed 6 million tons, accounting for over 60% of China's lithium reserves. These brine sources are currently more practical for lithium extraction due to their higher lithium concentrations. As anther liquid-based lithium resources, despite the low concentration around 0.17 ppm, the seawater holds approximately 250 billion tons of lithium ions, presenting a vast and relatively untapped resource.^11,12 Given the depleting traditional sources of lithium and the environmental impacts associated with conventional mining methods, there is an urgent need to develop advanced and efficient technologies for extracting lithium from seawater and salt lake water.

Fig. 1 The global growth trends from 2014 to 2024 in lithium-ion battery (LIB) production volumes (GW h) and the number of scientific publications related to LIBs (thousands).

Recently, many membrane separation technologies have made remarkable progress and exhibited great potential for industrial production. The most commercial ion-separation membranes are the polymer and 2D sheet materials (graphene, LDHs and Mxene), which could provide a high surface area and improved ion transport properties due to their nano-porosity and ultra-thin structures.^13,14 However, considering the internally heterogeneous pores of polymers and the complicated vertical and horizontal transport paths created by stacking 2D materials, both scenarios hinder efficient mass transport and cause an inevitable decrease in permeability, significantly limiting their practical applications. Transport of hydrated ions across atomically thin pores is influenced by electrostatic interactions, coordination with functional groups or charges at pore edges, and steric exclusion of the ion hydration shell. COFs offer several unique advantages for lithium extraction compared to other materials: COFs have highly tunable pore sizes that can be precisely controlled during synthesis.¹⁵ This allows for the selective passage of lithium ions while excluding other ions of similar or larger size.¹⁶ The ordered and rigid structure of COFs provides spatial confinement, enhancing the selectivity of lithium-ion transport. This spatial confinement helps reduce the diffusion path and improve the efficiency of ion separation. COFs are known for their high chemical and thermal stability,¹⁷ making them suitable for use in harsh environments, including high salinity conditions found in seawater and salt lakes.^18–20 COFs can be functionalized with various binding sites that have a high affinity for lithium ions. These binding sites can be tailored to enhance the selectivity and capacity for lithium extraction.²¹

Recent findings have demonstrated that the total energy barrier for ion transport through sub-nanometer pores can be effectively divided into two components: the dehydration energy encountered during partitioning into the pore and the intrapore diffusion energy associated with ion–wall interactions. As reported by Zhou et al.,²² while the former is governed by pore size and hydration shell disruption, the latter is strongly influenced by the electrostatic and steric environment within the channel. Therefore, tailored pore structures and functional sites in COFs allow simultaneous tuning of total energy barrier, enabling precise control over ion-specific transport energetics and selectivity.

Herein, we provide a comprehensive review of COF-based membranes for selective lithium-ion extraction. We focus on how fundamental ion transport mechanisms and pore chemistry can be engineered to achieve high Li⁺ selectivity and throughput. In particular, this review emphasizes the relationship between membrane structure and performance – discussing how pore dimensions, functional group interactions, and charge effects govern ion sieving and transport. Recent advances in both passive diffusion-driven and electro-driven (electrodialysis) COF membranes are examined, highlighting biomimetic design strategies and selectivity enhancement approaches. Overall, this work centers on the design principles and mechanistic understanding of COF-based membranes for lithium separation, aiming to guide the development of next-generation lithium extraction technologies.

2. Ion transport mechanism and selectivity enhancement strategies

2.1 Ion transport theory

Generally, the ion transport was considered to involve the continuously breaking bonding interactions with the host medium (such as hydrogen bonds with water molecules or reversible chemical bonds with chemical groups on the membrane pores), traversing the medium, and subsequently reforming the stable hydration hydrogen bonds in a new aqueous environment.²³ It is well-known that the dehydration of hydrated ions requires overcoming a significant energy barrier. This process is a key part of the ion transport mechanism, especially in confined pore spaces, such as those in COF membranes.²⁴ The energy barrier for ion dehydration is a crucial factor that governs the efficiency and selectivity of ion transport in such membranes. This energy barrier can be thought of as a fundamental reaction behavior, describable by transition state theory, which suggests that the ion transport process in membrane pores can be modeled using the Arrhenius reaction rate equation and the Eyring transition-state theory equation.^25,26
where k, a, k_B and h were the specific rate of the ion transport process, transmission coefficient, Boltzmann constant (1.38 × 10⁻²³ J K⁻¹) and Planck constant (6.626 × 10⁻³⁴ J s), respectively. T was absolute temperature (in K); the R was the universal gas constant (8.314 J mol⁻¹ K⁻¹). And the ΔS^‡ and ΔH^‡ were the entropy and enthalpy of the activation, respectively, which could be determined experimentally from the Arrhenius linear fitting. Then, the experimental activation energy (E_a) could be obtained accompanied with the pre-exponential factor (A).

In the context of membrane filtration, the hydrated ion on the feed side (left) must partially shed its hydration shell to enter the confined pore. It then traverses the pore and rehydrates once on the production side (right) (Fig. 2a). As the ion approaches the pore entrance, it first loses part of its hydration shell (E_a1), which presents a significant energy barrier due to strong ion–water interactions. Once inside the confined pore region, the ion must overcome an additional energy barrier (E_a2) associated with traversing the pore environment, which may involve electrostatic repulsion or specific interactions with the pore–wall. Together, E_a1 and E_a2 compose the overall activation energy (E_a) controlling ion permeation rates through sub-nanometer membrane channels.²⁷ Meanwhile, the pore size plays a significant role in determining both these energy barriers, with the effect on E_a1 being the most pronounced.²⁸ When the pore size is sufficiently small, it directly impacts the extent of ion dehydration, such as the number of water molecules removed from the hydration shell of the ion. As dehydration progresses deeper, the dehydration energy barrier increases exponentially. This occurs because the ion needs to shed more water molecules, which requires higher energy, especially when the pore size restricts the ion's movement and its hydration shell. Ion selectivity in such membrane processes can be understood as the difference in activation energy for different ions.

Fig. 2 (a) Schematic illustration of single-ion transport across a sub-nanometer membrane pore, involving partial dehydration (E_a1), pore traversal (E_a2), and subsequent rehydration. (b) Schematic comparison of two different ions (ion1 in blue vs. ion2 in red), demonstrating how variations in hydration energy and ionic radius result in distinct activation energy barriers (E_a) and selective ion permeation. (c) Schematic representation of facilitated ion transport via pore functionalization: selective binding sites on the pore walls compensate for dehydration energy, reducing the overall activation barrier and enabling enhanced transport of target ions.

It is well-known that the dehydration of hydrated ions requires overcoming a significant energy barrier, which is a key step in ion transport through confined pores.^24,29,30 Ion selectivity in such membrane processes can be understood as differences in activation energy for different ions. When comparing different ions (ion1 in blue and ion2 in red as shown in Fig. 2b), variations in their hydration energetics, hydration shell stability and ionic radii lead to create distinct activation energies for transmembrane transport. The ion species with a lower overall activation barrier (E_a) will have a thermodynamic advantage in terms of faster transport, leading to more efficient ion separation. This selectivity can arise from variations in hydration energy (leading to distinct E_a1) or differing ion–pore interactions (shaping E_a2). In this way, membranes with engineered pores can selectively favor ions that require less energy to transport, achieving higher selectivity. Furthermore, the complex multi-segment pore structures in membranes could be conceptualized as a series of elementary transition state processes. Strategically engineered pore structures could enhance the difference in E_a among various ion species, thereby significantly improving ion selectivity by favoring the migration of specific ions over others.

However, the above size-sieving effect in the channel structure is particularly useful for ions with notably different hydrated radii, it becomes less effective when comparing monovalent ions (especially those in the same group as Li⁺) whose hydrated diameters are very similar. In such cases, selectively reducing the dehydration barrier through strategic functionalization of the channel interior is crucial. By introducing specialized binding sites on the pore walls, one can overcome this limitation and enhance the selective transport of a target ion. In this scenario, as shown in Fig. 2c, targeted functional groups (e.g., crown ethers or acidic sites) are incorporated onto the pore walls to reduce the dehydration penalty for a specific ion (ion1), thereby lowering its overall activation energy and enhancing facilitated transport.³¹ By forming favorable coordination complexes or electrostatic interactions with the partially dehydrated ion, these functional sites compensate for the energetic cost of water removal, effectively driving a “thermodynamic shortcut” through the pore. As a result, the desired ion experiences a significantly diminished barrier to passage, enabling rapid migration while simultaneously impeding ions that lack comparable binding affinities. This approach underpins many advanced designs in COF membranes, where judiciously chosen ligands create strong ion-specific interactions and elevate separation performance.

Therefore, the architecture and chemical environment of COF have a fundamental influence on ion transport behavior, primarily by modulating dehydration energy barriers and ion–pore interactions that govern selective permeability. Leveraging such structural and chemical tuning approaches enables researchers to lower the transport barrier for target ions, thereby achieving precise control over ion selectivity. This strategic adjustment of pore size, interlayer stacking, and functional groups effectively bridges the theoretical insights on ion transport energetics with practical membrane design improvements.

2.2 Strategies for advancing ion selectivity

2.2.1. Pore sieving. The design of pore structures directly governs the forces exerted on hydrated ions traversing the membrane, representing a primary physical parameter for controlling ion–membrane interactions. Extensive experimental and theoretical studies have precisely determined the hydrated ionic diameters for a variety of ions (e.g., Li⁺: 7.64 Å, Na⁺: 7.16 Å, K⁺: 6.62 Å, Mg²⁺: 8.56 Å, Ca²⁺: 8.24 Å).^15,32,33 When the internal pore dimensions of COFs are tailored to closely match these diameters, the pore walls interact with the hydration shells of the ions, thereby imposing an energy barrier that is directly related to the degree of dehydration required for ion transport. This raises an intriguing challenge: how to precisely modulate the activation energy for ion transfer via deliberate pore engineering.

Pore engineering in the COFs could be understood as two aspects: (i) pore size regulation and (ii) tunable stacking mode. Given the naturally nanometer-sized pores in COFs, narrowing these pores to the sub-nanometer scale was critical and also the most vital step for enhancing ion separation performance in COF membranes. Current research related to the strategies focuses on regulating the pore size, with gradually flourishing studies exploring modifications of the stacking mode in the COFs.

2.2.1.1. Pore size regulation. The confinement effect has been validated in the catalytic chemistry,³⁴ electrochemistry,³⁵ and biochemistry³⁶ to effectively regulate the free energy of guest molecule diffusing within the pores, which has also been extensively applied in the design of various separation materials including the COFs. Precise pore size regulation in COF membranes leverages fundamental size-exclusion principles to control ion transport. When a pore's diameter is comparable to the hydrated radius of an ion, the ion's solvation shell cannot pass intact – it must partially shed water molecules, creating an energy barrier for transit. The magnitude of this barrier (i.e. the activation energy for ion transport) depends sensitively on how closely the pore size matches the ion's hydrated diameter. By tuning pore dimensions, one can therefore modulate dehydration requirements and thus discriminate between ions: a slightly smaller pore forces a strongly hydrated ion to shed more of its water, impeding its passage relative to less-hydrated ions. COFs offer a versatile platform for such control because their pore sizes can be systematically adjusted by design. Theoretical and experimental studies have long shown that spatial confinement can alter the free energy landscape of guest molecules diffusing in nanopores.

In practice, the pore apertures in COFs are dictated by the lengths of organic linker units, which means selecting different building blocks (or adding pendant groups) yields pores ranging from ultramicroporous (∼0.4 nm) to mesoporous (∼10 nm) scales (Fig. 3).³⁷ This tunability is a core advantage of COF membranes, allowing engineers to tailor pore sizes to target specific ions and separation goals.

Fig. 3 Schematic depiction of the condensation reaction between boronic acids and HHTP for COF formation with tunability of pore dimensions (pore sizes from 0.4 nm to 10 nm) via organic linker design.³⁷ Copyright 2009, American Chemical Society.

It was noteworthy that due to diameter of a single hydrated lithium ion being 7.64 Å, the sub-nanometer COFs was identified as the most appropriate candidate with the thermodynamically-advantageous confinement effects for achieving effective lithium-ion extraction,^38–40 as the pore exerts just enough steric pressure on the hydration shell to raise the transport energy barrier for larger competing ions (such as Mg²⁺ with ∼8.6 Å hydration). Moreover, this approach has been validated by incorporating flexible side chains or functional groups into COF structures to fine-tune pore openings at the sub-nanometer level. As a result, COF membranes with appropriately narrowed pores (∼1 nm or below) have achieved enhanced Li⁺ selectivity via effective pore sieving, demonstrating how fundamental pore–solvation thermodynamics can be harnessed for practical ion separations.

2.2.1.2. Tunable stacking mode. Beyond controlling intrinsic pore size, an emerging strategy in COF membranes is to engineer the stacking mode of 2D COF layers to create oriented sub-nanometer pore structures, thereby modulating the structure of one-dimensional channels that form between layers.^41,42 Fundamentally, 2D COF sheets are designed to stack through π–π interactions in either an eclipsed (AA) mode or an offset (staggered) mode, and this stacking sequence profoundly influences the pore architecture in the through-thickness direction.⁴³

In a conventional AA stacking, each layer's pores align perfectly on top of one another, creating straight, continuous channels. By contrast, introducing a relative offset between adjacent layers (e.g. an AB or ABC stacking pattern) causes the pores of one layer to partially overlap or stagger with those of the next. This steric misalignment effectively divides the transport pathway into narrower constrictions or multi-segment pore sections at the layer interfaces. The result is a more complex pore geometry: instead of a uniform cylinder, the ion encounters a series of alternating cavities and necks as it moves through the membrane. Such structural modulation can be achieved by strategic design, for instance, using different linker combinations or bulky substituents to force an AB or ABC stacking sequence (Fig. 4).

Fig. 4 (a) Schematic illustration of the construction of free-standing COF membranes via controlled monomer assembly to form the AA and AB stacking with different pore size.⁴¹ Copyright 2020, Springer Nature. (b) Diagram showing the impact of different stacking modes (AA, AB, and ABC types) on the formation of oriented one-dimensional channels within the COF structure.⁴⁴ Copyright 2013, American Chemical Society.

Altering the stacking mode directly impacts key material properties. Porosity and aperture size can be reduced when layers are staggered, since the offset creates smaller interlayer openings than the intrinsic pore diameter. Notably, a recent study demonstrated that converting a COF from AA to AB stacking shrank its effective pore size from 1.1 nm down to ∼0.6 nm, greatly amplifying its molecular sieving capability.⁴¹ Staggered stacking may also influence stability, for example, changing how layers interact can affect mechanical integrity and framework flexibility, though in some cases offset arrangements can enhance framework interlocking.

Most importantly, a tunable stacking configuration introduces additional ion transport mechanisms beyond straight size-exclusion. The staggered pores impose sequential dehydration/rehydration steps: an ion must navigate through multiple constricted junctions between layer segments, each acting as an energy barrier that favors smaller or less-hydrated ions. In other words, the ion experiences a multi-step filtration – a small ion can percolate through the zigzag channel, whereas a larger ion might be halted at one of the narrow necks. This interlayer transport aspect (ions moving from one layer's pore into the next layer's offset pore) can enhance selectivity by exacerbating differences in transport energy barriers between ions.

However, it is crucial to recognize that excessive offset (or overly constricted interlayer necks) will slow down overall ion flux. Thus, just as with pore size reduction, there is an inherent trade-off: while staggered stacking designs can dramatically improve ionic selectivity, they may also increase path tortuosity and resistance, reducing permeability. Optimizing the stacking mode therefore means finding a balance: achieving enough misalignment to create discriminating sub-nanometer pore throats, but not so much that the membrane's throughput is compromised.

2.2.2. Pore functionalization. Generally, the Li⁺ ions in aqueous solution carry tightly bound hydration shells.⁴⁵ In bulk water, Li⁺ typically coordinates with about four water molecules in its primary shell (often in a near-tetrahedral arrangement), and it can maintain a secondary shell due to its strong electrostatic field.^29,46–48 These inner-shell water molecules are highly oriented and polarized around Li⁺, making the ion's solvation very stable. To enter a sub-nanometer COF pore, however, a Li⁺ ion must shed or rearrange some of these waters (i.e. undergo partial dehydration). Stripping away the primary hydration shell exacts a substantial energy penalty, whereas removing outer-shell waters is comparatively easier. Notably, different ions have markedly different dehydration energetics depending on their charge density (charge-to-size ratio). For instance, a small, highly charged ion like Mg²⁺ holds its waters even more tightly than Li⁺, whereas a larger monovalent ion (Na⁺, K⁺) binds waters less strongly.⁴⁹ These disparities in dehydration energy create an opportunity for selectivity: a membrane environment that demands partial dehydration will impede ions unevenly, potentially blocking those with higher dehydration barriers while allowing others to pass. Fundamentally, achieving Li⁺ selectivity in COF membranes hinges on exploiting these differences by tuning the pore environment to favor Li⁺ desolvation and transport over that of competing cations. Fortunately, the easy regulation of COF structure can provide a variety of chemical group modification possibilities (Fig. 5).

Fig. 5 (a) Schematic illustration for the surface engineering of COFs through the combination of condensation reaction and click chemistry. (b) Systematically designed COF structure featuring tailored pore environments.⁵⁰ Copyright 2011, Springer Nature.

2.2.2.1. Neutral polar functionalization. One effective strategy is decorating COF pores with neutral polar chains such as oligoether (poly(ethylene oxide)-like) segments. Sun et al.⁵¹ demonstrated this principle by systematically varying oligoether chain lengths in a COF membrane and examining Li⁺vs. Mg²⁺ transport. These ether chains are lithiophilic – their oxygen atoms bear lone pairs that can coordinate with Li⁺ much like water's oxygen does, albeit in a confined geometry. When a Li⁺ ion enters a pore lined with, for example, ethylene oxide units, it can form ion–dipole bonds with the ether oxygens. This replaces some of the water–Li⁺ interactions by ether–Li⁺ interactions, effectively stabilizing the partially dehydrated Li⁺ in the pore. The result is a significant reduction in the dehydration energy penalty for Li⁺, since the ether functional groups act as an internal solvation shell. Importantly, the flexibility and length of the tethered ether chains determine how well they can wrap around and coordinate the cation. It highlights how neutral polar functional groups can be fine-tuned to preferentially solvate Li⁺ in situ, thereby lowering the activation energy of Li⁺ transport and boosting its permeation rate relative to other ions.
2.2.2.2. Charged polar functionalization. An alternative and complementary functionalization strategy is to embed charged polar functional groups (such as sulfonic acid or carboxylic acid moieties) along the pore walls. When deprotonated (as they typically are in neutral or basic aqueous solutions), these groups present fixed negative charges (e.g. –SO₃⁻) that serve as strong Coulombic attraction sites for cations. A negatively charged site in the pore can electrostatically bind a Li⁺ ion, effectively anchoring it and replacing part of its hydration environment with a charged coordination site. This ion–ion interaction is stronger and more specific than the neutral ether–Li⁺ dipole interaction, and it can robustly stabilize a dehydrated or partially dehydrated Li⁺. Jiang et al.⁵² illustrated this approach by integrating acidic groups into COF membrane channels to tune the surface chemistry and pore size simultaneously. The introduction of –SO₃⁻ groups in their COF created densely populated binding sites that Li⁺ ions must coordinate with while traversing the channel. In other words, a sulfonate-lined pore wall functions like an ion-exchange medium: as Li⁺ approaches, it displaces some of its hydrate waters and attaches to the fixed sulfonate site, which compensates the lost water-binding energy with a favorable Li⁺–SO₃⁻ interaction. By lowering the free energy of Li⁺ within the pore, these acidic functional groups reduce the activation energy needed for Li⁺ translocation.

The above principles demonstrate how electrostatic interactions, coordination chemistry, and hydration dynamics work in concert to govern Li⁺ selectivity in functionalized COF pores. By starting from a fundamental understanding of ion dehydration (i.e. the energy cost for removing hydration shells) and the specific binding preferences of Li⁺, one can rationally design pore functional groups that create a selective pathway for Li⁺ transport. In summary, the pore functionalization offers a powerful handle to manipulate ion transport energetics: by lowering the effective energy barrier for Li⁺ (through tailored solvation environments) while keeping barriers high for other ions, COF-based membranes can attain high lithium-ion selectivity for efficient extraction. This theory-driven, molecularly informed approach to membrane design is key to developing next-generation Li⁺ separation technologies.

3. Applications for lithium extraction

3.1. Overview of the lithium resources

Lithium is a critical element driving the global transition to low-carbon energy systems.⁵³ Prior to 2020, China's identified lithium reserves accounted for approximately 6–7% of the global total.⁵⁴ However, accelerated exploration efforts in regions such as Qinghai, Sichuan, and Xinjiang have expanded China's resource base to over 30 million tons, raising its global share above 16% and positioning it as the world's second-largest holder of lithium resources. This rapid growth in domestically identified lithium underscores the nation's resolve to secure its own supply.

Within China, lithium exists in both mineral and brine forms,⁵⁵ with salt lake brines contributing about 75–80% of the national reserves.^56,57 Notably, Qinghai Province alone accounts for approximately 60% of the total, equating to 15.9 million tons of lithium carbonate equivalent (LCE). Despite this abundance, exploiting lithium from these salt lakes has historically been challenging due to low lithium concentrations and high levels of Mg²⁺ and other impurities. In many brines, magnesium concentrations often exceed those of lithium by more than twenty-fold,⁵⁸ and the similarity in ionic size between Li⁺ and Mg²⁺ further complicates their separation.⁵⁹ As a result, China's lithium resources have been deemed lower in quality compared to those in regions such as Chile or Australia. By 2020, only about 1.5% of the lithium in Qinghai's salt lakes had been extracted, leading to a heavy reliance on imports, approximately 74% of China's lithium needs in 2020 were met by foreign sources, with 80–90% of raw lithium imports originating from Australia. This import dependence represents a strategic vulnerability for China's rapidly growing battery and electric vehicle sectors.

Recognizing the strategic significance of lithium, China has ramped up efforts to develop domestic extraction capabilities, with Qinghai at the forefront. Although the brine quality is generally lower than in regions such as South America, technical innovations have enabled substantial improvements. These include advancements in selective separation technologies such as adsorption, solvent extraction, precipitation, and increasingly, membrane separation methods. Membrane-based lithium extraction is gaining attention due to its modularity, energy efficiency, and reduced environmental footprint.

3.2 Conventional lithium extraction technologies

A variety of methods can be employed to extract lithium from salt lake brines, including chemical precipitation, high-temperature calcination (roasting), solvent extraction, adsorption using specialized sorbents, and membrane-based separation (Fig. 6). In practice, the most widely applied techniques for lithium brine extraction today are solvent extraction, ion adsorption, and emerging membrane separation processes. Each approach has distinct mechanisms and advantages, as well as challenges when applied to China's salt lake brines.

Fig. 6 The conceptual diagram of conventional lithium extraction technologies: (a) adsorption, (b) solvent extraction, (c) nanofiltration, (d) electrodialysis and (e) electrochemical adsorption.

Solvent extraction exploits differences in solubility or chemical affinity to transfer lithium ions from the aqueous brine into an organic phase.⁶⁰ Typically, an organic extractant (such as an organophosphate, crown ether, or certain alcohol/ketone compounds) selectively complexes with Li⁺, drawing it out of the brine and into the organic solvent. The lithium is then recovered by “back-extraction” into a fresh aqueous solution, from which lithium carbonate or other salts can be produced. Solvent extraction can achieve high selectivity and lithium recovery efficiency, and it has been successfully deployed at industrial scale. The main drawbacks lie in the cost and handling of large volumes of organic chemicals and the risk of solvent loss or residues causing environmental pollution.

Adsorption methods are another key route, using solid materials (lithium-selective adsorbents) to capture Li⁺ from brines.⁶¹ In these processes, brine is passed through a column or bed of the adsorbent, which contains active sites that bind lithium ions. After the adsorption step, a dilute acid wash is used to elute the concentrated lithium, regenerating the adsorbent for reuse. Several types of lithium ion-sieves and adsorbents have been developed, commonly classified by their active material: aluminum-based (e.g. lithium aluminate),^62–64 manganese-based (e.g. spinel lithium manganese oxide),^65–72 and titanium-based (e.g. lithium titanium oxide)^73–76 adsorbents. Adsorption is attractive because it can be highly selective for lithium over magnesium and other cations, and it operates at ambient conditions with relatively low chemical usage. However, adsorbents can be expensive, and their performance may degrade after repeated cycles (for instance, manganese-based sieves can suffer capacity loss due to gradual manganese dissolution during acid elution). Despite these challenges, adsorption-based lithium extraction has seen pilot and commercial implementation in China, taking advantage of its strong Li⁺ selectivity in high-magnesium brines.

In recent years, membrane separation technologies have gained increasing attention as a promising alternative or complement to conventional lithium extraction methods. Membrane-based processes offer several potential advantages: they often require less energy and chemical input than evaporation or chemical precipitation, can be modular and scaled, and can achieve continuous separation with minimal environmental impact. Two membrane processes in particular – nanofiltration (NF) and electrodialysis (ED) – have been widely studied and applied for lithium recovery from salt lakes. These processes leverage semi-permeable membranes to selectively separate lithium ions from other ions in the brine.

NF is a pressure-driven membrane process that sits between ultrafiltration and reverse osmosis in terms of pore size.⁷⁷ NF membranes are typically thin-film composites that carry charge and have nanometer-scale pores. They excel at selectively rejecting multivalent ions while allowing most monovalent ions to pass, a phenomenon governed by the combination of Donnan electrostatic exclusion (charge repulsion) and steric hindrance (size-based sieving). Compared to traditional evaporation ponds, NF is much faster and can achieve a high degree of lithium purification. However, off-the-shelf NF membranes often exhibit only moderate selectivity between Li⁺ and Mg²⁺. In fact, the Li/Mg separation factor for typical commercial NF membranes is usually less than 20. This means multiple passes or membrane stages are required to reach battery-grade lithium purity. Overall, NF offers a compelling combination of efficiency and selectivity for lithium extraction, especially as membrane materials continue to improve.

ED is another important membrane-based technique for lithium recovery. In an ED system, the Li⁺ migrate under the electric field toward the cathode and pass through cation-exchange membranes, while anions (like Cl⁻) move toward the anode through anion-selective membranes.⁷⁸ A key feature of ED is the use of ion-exchange membranes (IEMs) that can be tailored to preferentially allow certain ions to traverse. The benefits of ED include its high throughput and the ability to produce a concentrated lithium solution without extensive chemical additions, and it could be operated continuously, making it attractive for large-scale brine processing. Moreover, because the driving force is electrical, ED can be tuned by adjusting voltage or current, and it can potentially be powered by renewable electricity for a more sustainable process. On the downside, ED systems tend to be energy-intensive when treating very saline solutions, and they also face issues of membrane fouling and scaling, as other ions or impurities in brines can clog or degrade the ion-exchange membranes over time. Nonetheless, ongoing improvements in membrane materials and cell design are mitigating these issues. In summary, ED is a powerful technique to selectively extract lithium electrically, offering high separation efficiency even in brines with challenging compositions, though careful management of energy and membrane durability is required.

In summary, while conventional methods such as solvent extraction and adsorption offer relatively simple operation, they often suffer from poor selectivity, low reusability, and limited adaptability to complex brine compositions. Compared to NF and commercial polymer membranes, COFs provide unique advantages including rigid crystalline structures, highly tunable and uniform pore size distributions, and modular functionalization through reticular chemistry. Furthermore, compared to MOF-based membranes, COFs generally exhibit higher chemical stability in aqueous and alkaline environments and reduced risk of metal ion leaching. Several COF membranes have demonstrated excellent Li⁺/Mg²⁺ selectivity with high fluxes under mild conditions, which surpasses many conventional polymeric membranes and adsorbents. These features collectively make COF-based membranes particularly promising candidates for selective lithium extraction, especially under challenging brine conditions.

3.3. Progress of COF membrane based diffusion dialysis method

Biological membranes have long served as inspiration for artificial membranes, owing to their exceptional ion selectivity and permeability. In diffusion dialysis (concentration-driven) processes, COF membranes are being designed to mimic such biological ion channels and exploit fundamental separation mechanisms like size exclusion, specific ion coordination, and electrostatic gating (Fig. 7). For example, Sun et al.⁷⁹ developed a biomimetic COF membrane with sub-nanometer channels by introducing lithium-affinitive oligoether side chains onto a 2D COF grown on a polyacrylonitrile (PAN) substrate. These oligoether functionalities act as coordination sites that reversibly bind Li⁺, creating a pathway of “hopping” sites similar to natural ion channels. Lithium ions rapidly and reversibly coordinate with the ether oxygen atoms, allowing Li⁺ to hop from one binding site to the next and thus diffuse quickly through the COF nanochannels. Measurements of ionic transport (via reversal potential analysis) revealed a permeability order of Li⁺ > K⁺ > Na⁺ > Ca²⁺ > Mg²⁺, with a Li⁺/Mg²⁺ selectivity factor of ∼64. This biomimetic design illustrates how specific ion–pore interactions can be harnessed to achieve both high permeability and selectivity for Li⁺.

Fig. 7 (a) Diagrammatic sketch of ion channels in nature. (b) Schematic illustration of the construction of lithium channels using a 2D COF as a designer platform by implanting lithiophilic functionalities. The Li ion transfer was enhanced, while other ions were obstructed, allowing for high selectivity as well as permeability.⁷⁹ Copyright 2021, Elsevier.

Jiang et al.⁵² took a similar approach by functionalizing COF pore walls with acidic groups to create a “confined cascade separation” system for cation differentiation (Fig. 8a). In their work, three COF membranes (TpPa–PO₃H₂, TpPa–SO₃H, and TpPa–CO₂H) were prepared by post-synthetic modification with phosphonic, sulfonic, or carboxylic acid groups. These functional groups drastically alter the pore environment by forming hydrogen-bonded hydrated acid clusters inside the 1D COF channels. The hydrated acidic domains act as constricted “gates”: water molecules strongly associate with –SO₃H and –PO₃H₂ groups (more so than with –CO₂H), creating voluminous hydration shells that partially block the pore. This effectively reduces the local pore diameter in those regions. The interplay between steric exclusion in the acidic segments and fast transport in the non-acidic segments yields a unique cascade separation mechanism. In mixed-ion diffusion tests, the optimally spaced TpPa–PO₃H₂ COF showed effective sieving of Li⁺ in the presence of K⁺, with a practical K⁺/Li⁺ selectivity around 4.2 (and ideal selectivity ∼13.7 for K⁺/Li⁺). Liang et al.⁸⁰ reported a custom nanochannel engineering strategy to address both selectivity and fouling in COF membranes (Fig. 8b). They synthesized positively charged, lithium-affinitive COF nanosheets (DhaT–GCl COF) and modified their pore walls with oligoether (OE) chains via solvent exchange. These OE-functionalized COF nanosheets were then vacuum-filtered to form a stacked COF membrane (COFM) with well-aligned 1D channels (aided by strong π–π stacking between nanosheets). The resultant membrane had narrowly tuned pores decorated with lithium-binding sites and overall positively charged walls. This combination led to dual separation mechanisms: size exclusion (due to the angstrom-level pore diameter that resists large hydrated ions) and Donnan exclusion (due to the positively charged pores repelling multivalent cations). In practice, the OE-COF membrane exhibited a Li⁺/Mg²⁺ separation factor of 30.2, with a high water flux of 32.1 L m⁻² h⁻¹ bar⁻¹ in simulated brine. The positively charged channels also inhibited the deposition of scaling salts, endowing the membrane with excellent antifouling properties. This work shows that by tailoring the nanochannel width and surface charge, one can create COF membranes that simultaneously achieve high selectivity, high permeability, and resistance to fouling – all crucial for efficient lithium extraction.

Fig. 8 (a) Schematic illustration of hydrated cations through TpPa-SO₃H COFs channels.⁵² Copyright 2022, Springer Nature. (b) Schematic illustration of COFNs optimization and COFMs construction.⁸⁰ Copyright 2023, Elsevier.

It was well-known that the biological membranes possess smart nanochannels that can regulate ion transport behavior in response to external stimulus, offering high ionic selectivity. Other possibilities include photo-responsive or pH-responsive moieties grafted into COF channels to mimic this process. Liang et al.⁸¹ developed a smart COF membrane (COF-Azo) decorated with photo-responsive azobenzene derivatives (Fig. 9). Upon exposure to light of different wavelengths, the photo-responsive molecules grafted on the COF inner walls undergo isomerization, which modulates the effective pore size and channel characteristics. The electron-rich nanoscopic channels exhibit switching behavior and demonstrate excellent selectivity for monovalent/monovalent ions and monovalent/divalent ions (with ideal selectivities of 17.9 for K⁺/Li⁺ and 24.9 for Li⁺/Mg²⁺). The separation performance of COF membranes modified with azobenzene derivatives featuring different acidic groups (COF-Azo, COF-AzoCO₂H, and COF-AzoSO₃H) was compared. After multiple light-switching cycles, the COF membrane retains a K⁺/Li⁺ selectivity of 6.1 and a Li⁺/Mg²⁺ selectivity of 20.7 in a mixed salt solution, demonstrating substantial potential in lithium extraction and providing a novel pathway for rapid and precise ion separation.

Fig. 9 Schematic illustration of photo-responsive COF membrane under UV or visible light irradiation.⁸¹ Copyright 2024, Elsevier.

Extremely narrow COF channels can also exploit ion dehydration and binding effects to achieve high selectivity. Xu et al.³⁸ demonstrated this using an ultrathin (∼20 nm) COF membrane (TpBDMe₂) with densely packed hydrogen-bond donor sites lining ∼1.4 nm pores (Fig. 10). Under concentration-driven diffusion, monovalent cations permeated in the order K⁺ > Na⁺ > Li⁺ ≫ divalent cations (e.g., Ca²⁺, Mg²⁺). While basic size sieving plays a role (smaller hydrated ions generally diffuse faster), the chemistry of the pore surface proved equally important. The pore walls of TpBDMe₂ present –NH groups capable of hydrogen bonding with passing cations’ hydration shells. It was found that, for Mg²⁺ (with its higher charge and stronger hydration), hydrogen-bonding interactions with the COF's –NH sites dominate the transport resistance, rather than electrostatic attraction or repulsion. As a result, the TpBDMe₂ membrane achieved exceptionally high selectivity, with separation factors on the order of K⁺/Mg²⁺ ∼ 765, Na⁺/Mg²⁺ ∼ 680, and Li⁺/Mg²⁺ ∼ 217. Stimuli-responsive modifications offer another route to adjust pore size and selectivity in situ. Li et al.⁸² demonstrated a pH-responsive COF membrane by leveraging dopamine, a pH-sensitive molecule, to reversibly tune the effective pore diameter. They utilized a COF (TpPa-COOH) which has intrinsic 1D in-plane pores (∼1.25 nm) and interlayer spacing (∼0.35 nm). The COF nanosheets bear –COOH groups with negative ζ-potential in neutral water, while dopamine (DA) is protonated to DA–NH₃⁺ in acidic solution. Under acidic conditions (low pH), single dopamine molecules (positively charged) can insert themselves into the COF's 1D channels, effectively acting as molecular plugs that partition the pores. The polymerized dopamine instead deposits between COF layers, expanding the interlayer channel from ∼0.33 nm to ∼0.45 nm via electrostatic and hydrogen-bonding interactions. Thus, by simply toggling the pH, the membrane could switch between a low-flux, high-selectivity state and a high-flux state. These studys highlights the synergy between pore surface chemistry and ion hydration: by tuning pH or functional groups, one can modulate electrostatic attraction and hydrogen-bonding within the nano channel, dramatically altering ion transport rates. Such insights are foundational for designing COF membranes that differentiate ions not just by size, but by their affinity to the pore environment.

Fig. 10 Schematic of the fabrication of TpBDMe₂ membranes by an interfacial growth using the condensation reaction between Tp and BDMe₂ monomers with the assistance of co-reagent PTSA.³⁸ Copyright 2021, Wiley-VCH.

Three-dimensional COFs (3D COFs) are also gaining attention in diffusion-driven ion separation due to their inherently smaller pores and interconnected frameworks. Unlike 2D layered COFs, 3D COFs form a continuous 3D network of pores (typically 0.5–1.5 nm in size) which can improve size-selective sieving while providing multiple pathways for ion transport. Wang et al.⁸³ grew a 3D hydroxyl-functionalized COF (3D-OH-COF) as a membrane on a flexible cross-linked polyimide (CPI) substrate. Then through a ring-opening reaction, the –OH groups were converted to –COOH, yielding a 3D-COOH-COF membrane with even smaller pore apertures and charged pore walls. Thanks to the tighter pore size and the added electrostatic repulsion from –COO⁻ groups, the 3D-COOH-COF showed drastically improved selectivity for monovalent over multivalent cations (K⁺/Cu²⁺ = 766, Na⁺/Cu²⁺ = 634 and Li⁺/Cu²⁺ = 490). This study demonstrates that 3D COFs, with their entangled pore networks and functionalizable linkers, are promising platforms for precise ion sieving. By choosing appropriate substrate supports and leveraging the rich chemistry of COF linkers.

In summary, the diffusion dialysis advances in COF membranes highlight several key mechanistic strategies: bioinspired ion channels for facilitated Li⁺ transport, pore size modulation (either static or stimuli-responsive) for size-based selectivity, functional group incorporation (acidic, ionic, or ligand groups) for specific ion binding or charge exclusion, and composite designs that integrate COFs with polymers to overcome practical limitations. Each strategy leverages fundamental principles of ion transport – such as differences in hydration shell energy, ionic charge, and affinity to functional sites – to preferentially enrich lithium ions over competing cations. These studies collectively establish a theoretical foundation for how COF structure and chemistry influence ion sieving, guiding the development of next-generation lithium-selective membranes.

3.4. Progress of COF membrane based electrodialysis method

Electrodialysis (ED) uses an applied electric field to drive ions through ion-selective membranes, offering another avenue to exploit COF membranes for lithium extraction. In ED-based systems, ionic conductivity and permselectivity of the membrane become paramount, and COF membranes have been investigated for their ability to maintain high Li⁺ throughput under an electrical driving force while resisting multivalent ions. A notable example is provided by Sun et al.⁸⁴ synthesized COF membranes on PAN supports via interfacial polymerization and then grafted oligoether chains of varying lengths into the COF channels (Fig. 11a). The membrane's selectivity was found to depend strongly on these side chains: by adjusting the number of ethylene oxide units, the Li⁺ over Mg²⁺ transport could be preferentially facilitated. Mechanistically, the oligoether chains act as multiple binding and hopping sites for Li⁺, effectively forming a continuous “ion-conduction pathway” that shuttles Li⁺ quickly across the membrane. Under an electric field, the COF-EO2/PAN membrane achieved an exceptionally high Li⁺/Mg²⁺ selectivity factor of ∼1352. While experimental evidence showed the promise of oligoether-grafted COFs, the underlying separation mechanism was further elucidated by Zhang et al.,⁸⁵ through non-equilibrium molecular dynamics (NEMD) simulations (Fig. 11b). Using models of COF-EOX membranes derived from Sun's experiments, Zhang et al. analyzed the ion trajectories, ion-pore interactions, potentials of mean force (PMFs), and dehydration behaviors during Li⁺/Mg²⁺ transport. They also predicted that other functional groups could enhance Li⁺ selectivity: for instance, COFs functionalized with –OH or epoxy groups (COF-4OH, COF-4Epoxy) were simulated and later shown experimentally to achieve high Li⁺/Mg²⁺ selectivity and near-complete Mg²⁺ rejection. These findings highlight the importance of considering both pore size effects and interaction energetics (binding sites, dehydration energy) in designing COF membranes for ED.

Fig. 11 (a) The Chemical structures and conceptual diagram of engineering the solvation ability for COF-EOx.⁸⁴ Copyright 2024, NAS. (b) Atomic structures of three COFs with different oxygen-containing side chains (COF-4EO, COF-4OH and COF-4Epoxy) with their 2D density distribution maps of Li ions.⁸⁵ Copyright 2024, Elsevier.

Crown ethers, a class of macrocyclic compounds known for their strong binding affinity toward alkali metal monovalent cations, have been widely investigated in ion recognition and separation applications.^86,87 Their ability to selectively coordinate with monovalent cations arises from the specific size-matching effect of the ether oxygen atoms, which stabilize the hydrated monovalent cations while excluding divalent species.⁸⁸ Zhu et al.⁸⁹ tackled the permeability/selectivity trade-off by embedding a lithium-chelating macrocycle into a porous matrix. They synthesized a mechanically interlocked porous organic framework (a POF, conceptually similar to COFs) containing 24-crown-8 ether rings (denoted “Crown-POF”). In the Crown-POF, each cage-like unit offers five coordination sites (four ether oxygens and one tertiary amine nitrogen) arranged optimally for Li⁺ capture. The Crown-POF provided specific Li⁺ transport channels: Li⁺ ions can hop from one crown ether site to the next along the framework, akin to a facilitated transport mechanism. It was showed that the Li⁺ conduction far exceeding that of Na⁺, K⁺, or Mg²⁺ (e.g., 40.3 nS cm⁻¹ for Li⁺ vs only 5.6 nS cm⁻¹ for Mg²⁺). The crown ethers essentially act as molecular sieves that prefer Li⁺ – they temporarily bind Li⁺, carry it through the framework, and release it, whereas larger or non-fitting cations are not carried along as efficiently.⁹⁰ Pushing this idea further, Fang et al.⁹¹ embedded 12-crown-4 (12C4) ether rings directly into a 3D COF's backbone as ion-responsive gating units (Fig. 12). Two novel 3D COFs (JUC-590 and JUC-591) were synthesized by polymerizing tetrahedral linkers that bore 12C4 groups on their periphery. The crown ethers are thus integral to the COF's channel walls, effectively lining the pores with Li⁺ “lock sites.” The Li⁺-activated COF membrane exhibited outstanding selectivity – essentially, it remained in a low-conductivity state for other ions, but for Li⁺ it switched to a high-conductivity state (with a gating ratio up to 23.6).

Fig. 12 Strategy for preparing 3D functionalized JUC-590 and JUC-591 COFs.⁹¹ Copyright 2022, Wiley-VCH.

Lastly, a hybrid membrane strategy by Zhang et al.⁹² combined the strengths of metal–organic frameworks (MOFs) and COFs in a single membrane to achieve synergistic Li⁺/Mg²⁺ separation (Fig. 13). They fabricated a bilayer membrane consisting of a MOF layer (ZIF-8) and a COF layer (TpPa-SO₃H) on a nylon support. ZIF-8 is a nanoporous MOF known for its hydrophobic pores and positive framework charge, which tends to repel cations (especially multivalent ones) via the Donnan effect. TpPa-SO₃H is a COF with sulfonic acid groups that provide strong cation-coordination sites (and a slight negative charge) within its pores. The MOF and COF layers were found to form strong interfacial bonds, ensuring the membrane remained structurally stable and there were no voids or delamination (critical for maintaining selectivity). The ZIF-8/TpPa-SO₃H bilayer achieved a Li⁺/Mg²⁺ separation factor of 501 – an impressively high value for a practical membrane – along with excellent chemical and mechanical stability in salt solutions. This hybrid approach exemplifies the combination of multiple mechanistic barriers: one layer enforces a charge-based coarse separation, and the second layer adds a finer, chemical affinity-based discrimination. It also opens avenues for creative MOF/COF hybrids, leveraging the rich chemistry of both classes of porous crystals.

Fig. 13 Schematic for the synthesis and separation mechanism of ZIF-8/TpPa-SO₃H/nylon membrane.⁹² Copyright 2024, Wiley-VCH.

The progress in COF-based electrodialysis mem branes reinforces many concepts seen in diffusion-driven systems, while also highlighting unique considerations under electric fields. Mechanistically, achieving high Li⁺ selectivity in ED involves tailoring energy barriers and charge effects so that Li⁺ carries most of the current while competing cations are left behind. Approaches like functional side chains (to create favorable conduction pathways for Li⁺), charge density optimization, extreme pore size tuning, and even multi-layer or dynamic gating systems have all been demonstrated.

After summarizing the key characteristics of reported COF membranes for lithium extraction (Table 1), it was collectively shown that by grounding membrane design in fundamental principles – ion sorption and diffusion energetics, dehydration kinetics, and electrostatic interactions – COF membranes can be rationally engineered to approach the selective transport performance, yet with the chemical robustness and processability needed for industrial lithium extraction. Each work contributes to a deeper mechanistic understanding, which is crucial for guiding future innovations in lithium-selective membrane technology.

Table 1 Summary of reported COF membranes for lithium extraction

Method	Membrane	Selectivity (Li^+/Mg^2+)	Li^+ permeability (mmol h^−1 m^−2)/water flux (L m^−2 h^−1 bar^−1)	Ref.
Electrodialysis	TpPa-PO3H2	13.7 (K^+/Li^+)	13–77 (Li^+)	52
	dCOF-3	18.7 (K^+/Li^+)	276 (Li^+)	93
	COF-4EO-PAN	64	6.8–230 (Li^+)	79
	TpBDMe2	217	100–200 (Li^+)	38
	COF-EO2/PAN	1352	41 (Li^+)	51
	COF-EB1BD1/PAN	505	350 (Li^+)	94
	Crown-POF/PDMS	19.8	402.84 (Li^+)	89
	CC3	280	510 (Li^+)	95
Diffusion dialysis	PA-PIP/Fe-COF/PES	73.0	16.1 (water)	96
	cCOF-PA	49.1	10.1 (water)	97
	EBCOF@PEI-2	30.7	10.22 (water)	98
	TpTGCl	21.3	19.6 (water)	99
	OE2-COFM	30.2	32.1 (water)	80
	COF-AzoSO3H	36.6	10.1 (water)	81
	COF-4EO	15	200 (water)	85
	3D-COOH-COF	490 (Li^+/Cu^2+)	—	83
	JUC-590	23.6	—	91
	PCOF	61.6	—	100
	TGCl-DHA	80	58 (Li^+)	101
	TpHZ-D	38.71 (Li^+/K^+)	141.62 (Li^+)	102
		7.2 (Li^+/Na^+)
	TpHZ-T	17.02 (Li^+/K^+)	123 (Li^+)
		5.23 (Li^+/Na^+)
	COF-300	36	123 (Li^+)	16
	TpEBr	48	100 (Li^+)	87
	TAT-TP-P	823	1908.81 (Li^+)	21

---

4. Conclusions

In conclusion, the COF-based membranes represent a transformative platform for ion separation, combining molecular sieving precision with chemical tunability. Recent advances, underpinned by theoretical insights into ion transport, demonstrate that rational design can achieve high lithium selectivity. Addressing challenges in scalability, stability, and permeability/selectivity optimization will further enhance their viability. COF membranes hold immense promise for lithium extraction from brines, aiding energy storage and sustainable resource recovery. Their ability to finely tune pore environments allows for unprecedented ion separation control, with broader applications in clean water production and energy conversion. These innovations will be crucial for tackling global resource purification challenges.

However, there are several key challenges remain before their full potential can be realized in practical applications:

(1) Permeability/selectivity trade-off: achieving high selectivity often comes at the cost of reduced ion flux, limiting large-scale applicability. Strategies such as ultra-microporous architectures and high-density functional groups improve Li⁺/Mg²⁺ selectivity but hinder throughput. Conversely, increasing flux via larger pores or thinner films reduces discrimination between ions. Recent advances, including 3D COF networks, hierarchical membrane designs, and hybrid structures, have shown potential to break this trade-off. Future research should focus on integrating multi-layered architectures or external stimuli (e.g., electric fields) to enhance selective ion transport while maintaining high permeability, ensuring COF membranes are both efficient and practical for lithium extraction.

(2) Long-term stability: COF membranes must withstand prolonged exposure to harsh brine conditions, where extreme salinity, pH fluctuations, and scaling/fouling reduce performance. Issues such as pore clogging, chemical degradation of linkages, and mechanical brittleness remain concerns. Promising developments include composite membranes that resist degradation and COF-polymer hybrids with improved mechanical flexibility. However, extended testing in real-world brine environments is essential. Future advancements should focus on cross-linking strategies, polymer reinforcement, and antifouling coatings to enhance durability. Ensuring stability under fluctuating conditions (high Mg²⁺/Li⁺ ratios, sulfate/bicarbonate content, and temperature swings) will be critical for industrial-scale lithium recovery.

(3) Fabrication complexity and scalability: scaling COF membranes from laboratory-scale synthesis to industrial production remains a significant challenge. Current fabrication methods (such as solvothermal synthesis and interfacial polymerization) are complex, time-intensive, and difficult to control at large scales. Maintaining crystallinity, orientation, and defect-free morphology in large-area membranes is particularly challenging. Recent efforts in mixed-matrix membrane and COF–polymer composites offer potential solutions but require optimization to prevent filler aggregation and non-selective voids. Future directions should explore roll-to-roll processing, continuous flow synthesis, and additive manufacturing to streamline production. Overcoming these fabrication hurdles will be essential for widespread adoption of COF membranes in lithium extraction.

Looking ahead, future lithium-ion extraction technologies will focus on reconciling the selectivity–permeability trade-off through hierarchical membrane architectures and hybrid COF systems. Responsive and bioinspired pore environments may offer dynamic control over ion transport, while coupling with external fields (e.g., electric or pressure-driven) could further enhance extraction efficiency. Scalable and environmentally friendly fabrication methods will also be essential for real-world application. Ultimately, continued progress will rely on interdisciplinary efforts bridging materials design, membrane engineering, and system integration to meet the rising demand for sustainable energy and resource recovery.

Author contributions

Da Lei: investigation, software, writing – original draft & editing. Yongjie Zhu: formal analysis, investigation. Lan-Lan Lou: supervision, writing – review & editing. Zhong Liu: conceptualization, supervision, writing – review & editing.

Data availability

No primary research results, software or code have been included and no new data were generated or analysed as part of this review.

Conflicts of interest

There are no conflicts to declare.

Acknowledgements

We gratefully acknowledge financial support from the National Natural Science Foundation of China (U20A20141, 22304185, U23A20119), Qinghai Provincial Department of Science and Technology (2024-GX-120), CAS Project for Young Scientists in Basic Research (YSBR-039), Strategic Priority Research Program of the Chinese Academy of Sciences, Grant No. XDB 1130301, and Qinghai Province “High-end Innovative Talents Plan”.

Notes and references

1. R. Byck, The Lancet, 1976, 308, 72–73.
2. M. Tomkiewicz and H. Steinberg, Nature, 1974, 252, 227–229.
3. G. Wu and M. Huang, Chem. Rev., 2006, 106, 2596–2616.
4. M. Mayor, C. Lovis and N. C. Santos, Nature, 2014, 513, 328–335.
5. W. Xu, N. Birbilis, G. Sha, Y. Wang, J. E. Daniels, Y. Xiao and M. Ferry, Nat. Mater., 2015, 14, 1229–1235.
6. K. Cheng, J. Li, F. Jiang, Q. Xiao, Z. Zou, J. Qin and W. Jing, Energy, 2024, 313, 133777.
7. J.-M. Tarascon and M. Armand, Nature, 2001, 414, 359–367.
8. Global EV outlook 2024, (IEA), 2024.
9. Mineral Commodity Summaries 2024, (USGS), 2024.
10. A. H. Asif, C. Li, H. Lim and H. Sun, Ind. Eng. Chem. Res., 2024, 63, 2073–2086.
11. X. He, S. Kaur and R. Kostecki, Joule, 2020, 4, 1357–1358.
12. S. Yang, F. Zhang, H. Ding, P. He and H. Zhou, Joule, 2018, 2, 1648–1651.
13. J. Luo, R. Qiao and B. Ding, Matter, 2024, 7, 3351–3389.
14. F. Wang, Z. Zhang, I. Shakir, C. Yu and Y. Xu, Adv. Sci., 2022, 9, 2103814.
15. S. Yuan, X. Li, J. Zhu, G. Zhang, P. Van Puyvelde and B. Van der Bruggen, Chem. Soc. Rev., 2019, 48, 2665–2681.
16. T. Wu, Y. Qian, Z. Zhu, W. Yu, L. Zhang, J. Liu, X. Shen, X. Zhou, T. Qian and C. Yan, Adv. Mater., 2025, 2415509.
17. S.-Y. Ding and W. Wang, Chem. Soc. Rev., 2013, 42, 548–568.
18. S. Hou, W. Ji, J. Chen, Y. Teng, L. Wen and L. Jiang, Angew. Chem., Int. Ed., 2021, 60, 9925–9930.
19. Q.-W. Meng, S. Wu, M. Liu, Q. Guo, W. Xian, X. Zuo, S. Wang, H. Yin, S. Ma and Q. Sun, Sci. Adv., 2023, 9, eadh0207.
20. S. Yin, J. Li, Z. Lai, Q.-W. Meng, W. Xian, Z. Dai, S. Wang, L. Zhang, Y. Xiong and S. Ma, Nat. Commun., 2024, 15, 8137.
21. W. Meng, S. Chen, M. Wu, F. Gao, Y. Hou, X. Zhan, W. Hu, L. Liang and Q. Zhang, Angew. Chem., Int. Ed., 2025, e202422423.
22. X. Zhou, Z. Wang, R. Epsztein, C. Zhan, W. Li, J. D. Fortner, T. A. Pham, J.-H. Kim and M. Elimelech, Sci. Adv., 2020, 6, eabd9045.
23. G. M. Geise, D. R. Paul and B. D. Freeman, Prog. Polym. Sci., 2014, 39, 1–42.
24. R. Epsztein, E. Shaulsky, M. Qin and M. Elimelech, J. Membr. Sci., 2019, 580, 316–326.
25. B. J. Zwolinski, H. Eyring and C. E. Reese, J. Ind. Eng. Chem., 1949, 53, 1426–1453.
26. H. Eyring, Trans. Faraday Soc., 1938, 34, 41–48.
27. N. Lakshminarayanaiah, Chem. Rev., 1965, 65, 491–565.
28. P. Läuger, Angew. Chem., Int. Ed. Engl., 1969, 8, 42–54.
29. S. Sahu, M. Di Ventra and M. Zwolak, Nano Lett., 2017, 17, 4719–4724.
30. R. Tan, A. Wang, R. Malpass-Evans, R. Williams, E. W. Zhao, T. Liu, C. Ye, X. Zhou, B. P. Darwich, Z. Fan, L. Turcani, E. Jackson, L. Chen, S. Y. Chong, T. Li, K. E. Jelfs, A. I. Cooper, N. P. Brandon, C. P. Grey, N. B. McKeown and Q. Song, Nat. Mater., 2020, 19, 195–202.
31. T. Xu, B. Wu, W. Li, Y. Li, Y. Zhu, F. Sheng, Q. Li, L. Ge, X. Li and H. Wang, Sci. Adv., 2024, 10, eadn0944.
32. Z. Li, Y. Zheng, T. Gu, X. Meng, H. Wang, K. Xu, L. Cheng, R. Kasher, R. Zhang and Z. Jiang, J. Membr. Sci., 2023, 675, 121551.
33. Z. Mu, Y. Zhu, B. Li, A. Dong, B. Wang and X. Feng, J. Am. Chem. Soc., 2022, 144, 5145–5154.
34. J. Wu, Y. Xie, Y. Li, M. Jin, L. Liu, G. Pan, C. Wang and F. Li, Coord. Chem. Rev., 2023, 497, 215437.
35. L. Wang, M. Peng, J. Chen, T. Hu, K. Yuan and Y. Chen, Adv. Mater., 2022, 34, 2203744.
36. X. Heng, F. Shan, H. Yang, J. Hu, R. Feng, W. Tian, G. Chen and H. Chen, Adv. Healthcare Mater., 2023, 12, 2301536.
37. H. Furukawa and O. M. Yaghi, J. Am. Chem. Soc., 2009, 131, 8875–8883.
38. F. Sheng, B. Wu, X. Li, T. Xu, M. A. Shehzad, X. Wang, L. Ge, H. Wang and T. Xu, Adv. Mater., 2021, 33, 2104404.
39. C. Jiang, M. Tang, S. Zhu, J. Zhang, Y. Wu, Y. Chen, C. Xia, C. Wang and W. Hu, Angew. Chem., Int. Ed., 2018, 57, 16072–16076.
40. Z. Li, J. Fan, L. Wang, X. Yang, L. Guo, H. Chen, D. Gong, G. Yang, Q. Xu, S. Zou and G. Zeng, J. Membr. Sci., 2024, 699, 122645.
41. Y. Li, Q. Wu, X. Guo, M. Zhang, B. Chen, G. Wei, X. Li, X. Li, S. Li and L. Ma, Nat. Commun., 2020, 11, 599.
42. Z. Wang, Y. Zhang, T. Wang, E. Lin, T. Wang, Y. Chen, P. Cheng and Z. Zhang, Small, 2023, 19, 2303684.
43. C. Zhang, B.-H. Wu, M.-Q. Ma, Z. Wang and Z.-K. Xu, Chem. Soc. Rev., 2019, 48, 3811–3841.
44. X. Wu, X. Han, Y. Liu, Y. Liu and Y. Cui, J. Am. Chem. Soc., 2018, 140, 16124–16133.
45. I. Persson, Dalton Trans., 2024, 53, 15517–15538.
46. C. Lu, C. Hu, C. L. Ritt, X. Hua, J. Sun, H. Xia, Y. Liu, D.-W. Li, B. Ma, M. Elimelech and J. Qu, J. Am. Chem. Soc., 2021, 143, 14242–14252.
47. K. H. Michaelian and M. Moskovits, Nature, 1978, 273, 135–136.
48. D. Paschek and R. Ludwig, Angew. Chem., Int. Ed., 2011, 50, 352–353.
49. E. A. Gomaa, M. A. Tahoon and A. Negm, J. Mol. Liq., 2017, 241, 595–602.
50. A. Nagai, Z. Guo, X. Feng, S. Jin, X. Chen, X. Ding and D. Jiang, Nat. Commun., 2011, 2, 536.
51. J. Yuan, Z. Liu, Y. Wu, J. Han, X. Tang, C. Li, W. Chen, X. Yi, J. Zhou, R. Krishna, G. Sastre and A. Zheng, Proc. Natl. Acad. Sci. U. S. A., 2021, 118, e2102097118.
52. H. Wang, Y. Zhai, Y. Li, Y. Cao, B. Shi, R. Li, Z. Zhu, H. Jiang, Z. Guo, M. Wang, L. Chen, Y. Liu, K.-G. Zhou, F. Pan and Z. Jiang, Nat. Commun., 2022, 13, 7123.
53. S. Yang, Y. Wang, H. Pan, P. He and H. Zhou, Nature, 2024, 636, 309–321.
54. H. Li, T. Zhu, X. Chen, H. Liu and G. He, Front. Environ. Sci., 2022, 10, 938534.
55. S. Jing, H. Su, Y. Xiong, T. Liu, K. Zhu and L. Zhang, Acta Geol. Sin., 2020, 94, 1757–1773.
56. K. Matsuoka, S. Hasegawa, T. Yuma and Y. Goto, J. Mol. Liq., 2019, 294, 111663.
57. J. Li, T. Zou, X. Liu, D. Wang and X. Ding, Acta Geol. Sin., 2015, 89, 652–670.
58. Q. Peng, R. Wang, Z. Zhao, S. Lin, Y. Liu, D. Dong, Z. Wang, Y. He, Y. Zhu, J. Jin and L. Jiang, Nat. Commun., 2024, 15, 2505.
59. S. Xu, J. Song, Q. Bi, Q. Chen, W.-M. Zhang, Z. Qian, L. Zhang, S. Xu, N. Tang and T. He, J. Membr. Sci., 2021, 635, 119441.
60. T. Kanagasundaram, O. Murphy, M. N. Haji and J. J. Wilson, Coord. Chem. Rev., 2024, 509, 215727.
61. M. R. Mojid, K. J. Lee and J. You, Sustainable Mater. Technol., 2024, 40, e00923.
62. Y. Deng, G. Chai and Y. Zhang, J. Ind. Eng. Chem., 2025 DOI:10.1016/j.jiec.2025.01.014.
63. H. S. Kim, Y. J. Suh, D.-W. Kim, H. Yoo, H.-J. Hong and I.-S. Park, Chem. Eng. J., 2025, 505, 159220.
64. X. Xu, Y. Chen, P. Wan, K. Gasem, K. Wang, T. He, H. Adidharma and M. Fan, Prog. Mater. Sci., 2016, 84, 276–313.
65. C. Li, H. Zhan, Y. Qiao, C. Shi, Z. Qian and Z. Liu, Desalination, 2025, 593, 118216.
66. H. Zhan, Y. Qiao, Z. Qian, J. Li, Z. Wu and Z. Liu, Sep. Purif. Technol., 2023, 305, 122485.
67. F. Qian, M. Guo, Z. Qian, B. Zhao, J. Li, Z. Wu and Z. Liu, Sep. Purif. Technol., 2021, 264, 118433.
68. H. Zhan, Y. Qiao, Z. Qian, B. Lv, Z. Wu and Z. Liu, Chem. Eng. J., 2024, 497, 154859.
69. H. Zhan, C. Liu, Y. Qiao, Z. Qian, B. Lv and Z. Liu, Sep. Purif. Technol., 2025, 355, 129631.
70. Y. Qiao, Y. Ran, Z. Qian, B. Zhao, C. Li and Z. Liu, Chem. Eng. J., 2024, 497, 155009.
71. F. Qian, B. Zhao, M. Guo, Z. Wu, W. Zhou and Z. Liu, Sep. Purif. Technol., 2021, 256, 117583.
72. H. Zhan, Z. Qian, Y. Qiao, B. Lv, R. Liu, H. Chen and Z. Liu, ACS Nano, 2024, 18, 31204–31214.
73. Y. Qiao, X. Dai, B. Zhao, Z. Qian, Z. Wu and Z. Liu, Sep. Purif. Technol., 2024, 351, 128060.
74. B. Zhao, Z. Qian, Y. Qiao, J. Li, Z. Wu and Z. Liu, Chem. Eng. J., 2023, 451, 138870.
75. B. Zhao, Z. Qian, M. Guo, Z. Wu and Z. Liu, Chem. Eng. J., 2021, 414, 128729.
76. B. Zhao, Y. Qiao, Z. Qian, W. Wei, J. Li, Z. Wu and Z. Liu, J. Mater. Chem. A, 2023, 11, 7094–7104.
77. X. Wen, P. Ma, C. Zhu, Q. He and X. Deng, Sep. Purif. Technol., 2006, 49, 230–236.
78. C. Jiang, B. Chen, Z. Xu, X. Li, Y. Wang, L. Ge and T. Xu, AlChE J., 2022, 68, e17710.
79. S. Bing, W. Xian, S. Chen, Y. Song, L. Hou, X. Liu, S. Ma, Q. Sun and L. Zhang, Matter, 2021, 4, 2027–2038.
80. L. Ren, J. Chen, J. Han, J. Liang and H. Wu, Chem. Eng. J., 2023, 462, 142112.
81. L. Ren, J. Chen, J. Han, J. Liang and H. Wu, Chem. Eng. J., 2024, 482, 148907.
82. Z.-H. Wu, M. Wang, X.-Y. Liu, J. Wu, M.-M. Liu, Y. Liu, Y. Cao, X.-Q. Wu, B. Wang and D.-S. Li, Adv. Funct. Mater., 2025, 35, 2416228.
83. X. Shi, Z. Zhang, S. Fang, J. Wang, Y. Zhang and Y. Wang, Nano Lett., 2021, 21, 8355–8362.
84. Q.-W. Meng, X. Zhu, W. Xian, S. Wang, Z. Zhang, L. Zheng, Z. Dai, H. Yin, S. Ma and Q. Sun, Proc. Natl. Acad. Sci. U. S. A., 2024, 121, e2316716121.
85. J. Li, Q. Xie, C. Chen, L. Wang and L. Zhang, J. Membr. Sci., 2025, 717, 123577.
86. X. Liu, W. Lin, K. B. Al Mohawes and N. M. Khashab, Small Struct., 2024, 2400496,  DOI:10.1002/sstr.202400496.
87. R. Wang, L. Ding, J. Xue, H. Wu, C. Cai, Z. Qiao, J. Caro and H. Wang, Small Methods, 2024, 9, 2401111.
88. S. An, Q. Xu, Z. Ni, J. Hu, C. Peng, L. Zhai, Y. Guo and H. Liu, Angew. Chem., Int. Ed., 2021, 60, 9959–9963.
89. X. Ruan, C. Zhang, Y. Zhu, F. Cai, Y. Yang, J. Feng, X. Ma, Y. Zheng, H. Li, Y. Yuan and G. Zhu, Angew. Chem., Int. Ed., 2023, 62, e202216549.
90. N. Li, F. Chen, J. Shen, H. Zhang, T. Wang, R. Ye, T. Li, T. P. Loh, Y. Y. Yang and H. Zeng, J. Am. Chem. Soc., 2020, 142, 21082–21090.
91. X. Yu, C. Li, J. Chang, Y. Wang, W. Xia, J. Suo, X. Guan, V. Valtchev, Y. Yan, S. Qiu and Q. Fang, Angew. Chem., Int. Ed., 2022, 61, e202200820.
92. Y. Ma, L. Liu, R. Yu, M. Ma, N. Gao, L. Sun, N. Zhang and G. Zhu, Small, 2024, 21, 2409570.
93. Z.-H. Wu, M. Wang, X.-Y. Liu, J. Wu, M.-M. Liu, Y. Liu, Y. Cao, X.-Q. Wu, B. Wang and D.-S. Li, Adv. Funct. Mater., 2025, 35, 2416228.
94. L. Hou, W. Xian, S. Bing, Y. Song, Q. Sun, L. Zhang and S. Ma, Adv. Funct. Mater., 2021, 31, 2009970.
95. T. Xu, B. Wu, L. Hou, Y. Zhu, F. Sheng, Z. Zhao, Y. Dong, J. Liu, B. Ye, X. Li, L. Ge, H. Wang and T. Xu, J. Am. Chem. Soc., 2022, 144, 10220–10229.
96. Y.-X. Fang, Y.-F. Lin, Z.-L. Xu, D. Pandaya and S.-J. Xu, Desalination, 2024, 588, 117977.
97. G. Wang, T. Wu, J. Zhao, B. Shi, T. Gu, R. Zhang, X. Wang, R. Kasher, Y. Su and Z. Jiang, J. Membr. Sci., 2023, 684, 121863.
98. B. Hu, H. Deng, Y. Zheng, Z. Zhang, T. Wu, Z. Liu, B. Jia, H. Lin, R. Zhang and Z. Jiang, J. Membr. Sci., 2025, 715, 123489.
99. Z. Si, Z. Zhang, C. Yin, T. Ju, M. Wei, J. Huang and Y. Wang, J. Membr. Sci., 2023, 680, 121759.
100. S. Xu, H. Lin, G. Li, Q. Han, J. Wang and F. Liu, Adv. Sci., 2024, 11, 2405539.
101. X. Liu, W. Lin, K. B. Al Mohawes and N. M. Khashab, Small Struct., 2024, 2400496.
102. Q. Sun, Z. Song, J. Du, A. Yao, L. Liu, W. He, S. U. Hassan, J. Guan and J. Liu, ACS Nano, 2024, 18, 27065–27076.

---

Footnote

† These authors contributed equally.

---