Direct lithium extraction (DLE) using lithium-selective membranes (LSMs) is an emerging technology aimed at efficiently recovering lithium from various sources, such as brines, seawater, and lithium-containing minerals (Foo and Lienhard, 2025; Saleem et al., 2025). LSM technology integrates the advantages of DLE based on sorbents with those of commercialized membrane processes such as selective electrodialysis (S-ED), membrane capacitive deionization, and nanofiltration (NF) (Xu L. et al., 2025; Yao et al., 2025). LSMs are generally designed to permit the passage of Li⁺ ions while rejecting other ions, such as sodium, potassium, magnesium and calcium. This selectivity is achieved due to the properties of the membrane material, which can include polymer, ceramic or composite structures exhibiting specific interactions exclusively with lithium ions.

LSM technology is gaining traction due to its potential to provide a more sustainable and environmentally friendly alternative to traditional lithium extraction methods. The combination of crushing, leaching and reduction, washing, heat treatment, filtration, evaporation, and related steps forms the basis of hydrometallurgical processes used for lithium extraction from brines in the Lithium Triangle countries (Chile, Bolivia, and Argentina), from mineral deposits in Australia, and from secondary resources worldwide (Tabelin et al., 2021; Butylskii et al., 2023). Moreover, in the Lithium Triangle countries and certain regions of China, evaporation ponds, where concentration and solution enrichment occur simultaneously, remain the traditional method (Ezama et al., 2018; King and Dworzanowski, 2021).

Integrated with other membrane-based methods, LSMs are designed to supplant the costly, energy-heavy processes currently used to separate and concentrate lithium from liquid sources, irrespective of the feed solution’s origin (Butylskii et al., 2024). A major advantage of this technology lies in its versatility. It has potential applications in regions where climatic and topographic conditions preclude the use of traditional hydrometallurgical lithium extraction processes (certain lithium-producing areas of Russia, China, the United States, Portugal, and Serbia etc.), including pond evaporation.

However, when analyzing the potential of LSM technology, researchers often focus primarily on its high selectivity (Zhang J. et al., 2024; Abrishami et al., 2025), which is achieved through specific interactions in addition to the size and charge effects of ion exclusion inherent in various types of ion-exchange and filtration membranes. The increase in selectivity indicates the presence of barriers limiting the transport of ions and water, resulting in increased electrical and hydraulic resistance and directly affecting both performance and energy consumption. Thus, analyzing the success of LSM technology solely based on the concept of high selectivity is somewhat flawed.

The literature already contains numerous research and review articles presenting test results for individual LSM groups. However, few studies present a complete classification and compare individual LSM groups with each other. The difficulty lies in the fact that the original studies evaluate the performance of different LSM groups using different parameters. The goal of this paper was to present such a classification and evaluate LSMs from different groups and subgroups using the same parameters (specific flux and selective ion separation coefficient) for performance evaluation.

The explosive growth in the number of LSM types developed by researchers worldwide over the past 5 years highlights the need for a clear and systematic classification. Such classification facilitates cataloging and enables the correlation of the properties of newly developed membranes with those of their «relatives» within the same group or subgroup. Currently, the literature most commonly categorizes LSMs according to their fabrication methods and constituent materials. Among these, the most frequently mentioned are ion-imprinted membranes (IIMs), ion sieve membranes (ISMs), and liquid membranes (LMs) (Figure 1).

LSMs are expected to combine several key advantages, including high lithium-ion selectivity, high specific productivity (ion flux) and extraction efficiency, energy efficiency and continuous operation compared to traditional methods, as well as environmental sustainability through reduced reagent consumption. However, in practice, it is often impossible to achieve all these parameters simultaneously due to their inherently conflicting nature, a phenomenon known as the trade-off effect: increased productivity usually leads to reduced selectivity, while higher extraction efficiency tends to decrease energy efficiency, and vice versa. Consequently, most reports in the literature describe laboratory-scale LSMs that fulfill one or two sustainability criteria but rarely combine all the desirable characteristics.

Table 1 summarizes the results from recent studies proposing and testing various structures and compositions of ion-imprinted, ion-sieve, and liquid membranes. It should be noted that these results were obtained under diverse experimental conditions. The reported systems differ in driving forces (concentration, pressure, or electric field gradients), feed solution compositions (from ternary electrolytes containing lithium and magnesium salts to multicomponent solutions with monovalent ions such as Li⁺, K⁺, Na⁺ to divalent ions such as Ca²⁺, Mg²⁺, Ni²⁺, Mn²⁺, and Co²⁺), and in the design of test modules (ranging from simple laboratory H-type cells to multi-compartment electrodialysis cells with several repeating cell pairs).

Among the various types of ion-imprinted membranes, composite IIMs are likely the most widely used in lithium extraction. They are characterized by ease of fabrication and broad variability in component composition. Moreover, 12-crown-4 ether (12C4E) is most frequently employed as a chelating agent in such membranes (Sun et al., 2017; Cui et al., 2019; Li L. et al., 2021; Yang et al., 2022). The cavity size of 12C4E closely matches the ionic radius of lithium, enabling the formation of specific recognition sites through complexation with oxygen atoms. However, the structure remains permeable to smaller sodium and potassium ions. In addition, 12C4E can also interact with magnesium and calcium ions, forming sandwich structure. As a result, membranes of this type allow the transfer of not only lithium but also other coexisting ions (Yang et al., 2022), which leads to selective ion separation (or specific permselectivity) coefficient, S L i + / C n + , often significantly lower than 1 (Table 1).

The selective ion separation coefficient (Equation 1) determines the ability of the membrane to preferentially transport the target ion over coexisting ions in the feed solution and is defined as follows: S L i + / C n + = j L i + / j C n + c L i + f / c C n + f = Δ c L i + p / Δ c C n + p c L i + f / c C n + f (1) where j L i + and j C n + are the flux densities of Li⁺ and coexisting Cⁿ⁺ cations through the membrane, respectively; c L i + f and c C n + f are the concentrations of coexisting cations in the feed solution; Δ c L i + p and Δ c C n + p are the changes in concentrations of these ions in the concentrate (product) compartment over time.

This implies that, based on the physical meaning of the parameter S L i + / C n + , IIMs are actually «selective» toward coexisting cations rather than lithium itself. Since the chelating agents incorporated in their composition exhibit specific affinity for lithium, the performance of IIMs is typically evaluated not only in terms of the selectivity coefficient but also by parameters such as diffusion permeability, sorption coefficients and adsorption capacity, that is, by characteristics obtained under equilibrium conditions or driven solely by concentration gradients. Luo et al. (2020) noted that for membrane applications in electric field–driven processes, it is essential to evaluate them under appropriate operating conditions. Moreover, both the specified current modes (Jiang et al., 2019; Zabolotsky et al., 2020; Golubenko and Yaroslavtsev, 2021) and hydraulic regimes (Zhang et al., 2021) can significantly influence membrane performance. Due to the coupled effects of concentration polarization and phenomena occurring at elevated (overlimiting) current densities, membranes can exhibit specific permselectivity only at current densities below the limiting value (Gorobchenko et al., 2022). Diffusion permeability, sorption coefficients, and adsorption capacities indeed play a major role at low current densities and when the proportion of the target component is high compared to coexisting components (20%–60%). This can then be used to calculate the selective ion separation coefficient (Luo et al., 2020). Therefore, while these parameters are important, specific permselectivity remains the key indicator, as it enables a direct assessment of the performance of different membrane systems and facilitates comparative analysis of the results independent of external factors.

Another parameter important for practical applications is the specific flux of the competing cations (Equation 2), j _k, mol/(m²×h): j C n + = V s d c C n + p d t (2) where V is the volume of feed or permeate solution, s is the membrane surface area, t is time of an experiment.

Although this parameter is as universal as the selective ion separation coefficient, it is not as widely used. Its definition and expression vary across different membrane-based methods. For example, in baromembrane concentration and separation processes (reverse osmosis and NF), which are widely used in water treatment processes (Jones et al., 2019), performance is typically evaluated by the membrane permeate flux, j _v, L/(m²×h). During NF treatment, lithium and other monovalent ions enter the permeate along with water molecules, whereas multivalent ions are predominantly retained in the retentate.

To develop integrated lithium extraction technologies and critically evaluate the progress of membrane-based methods, including those employing LSMs, it is essential to compare their performance using standardized parameters. Previously, we were the first to compile, calculate (when data were not explicitly provided in the original studies), and analyze the results of a wide range of membrane-based methods (ED, S-ED, BMED, EL, DD, CID, MCDI, NF, RO, etc.) for lithium extraction from both primary and secondary sources (Butylskii et al., 2023). The comparison was based on selective ion separation coefficients and specific ion flux. The analysis primarily focused on lab-made monovalent cation-selective and lithium-selective membranes. Although these membranes often demonstrate high selective ion separation coefficients, their technology readiness level remains too low for commercialization due to complex and costly fabrication procedures and testing performed only under simplified conditions using artificial brines. Therefore, a separate assessment was carried out for membranes and membrane technologies tested in either artificial or real multicomponent brines, as well as under near-industrial conditions (large effective membrane area, high total dissolved solids, integration with other processing stages, etc.) (Butylskii et al., 2024). This analysis highlighted the high potential of S-ED and NF methods for lithium extraction from mixtures containing divalent ions, as well as capacitive deionization, membrane capacitive deionization and electrobaromembrane methods for reagent-free separation of lithium from monovalent sodium and potassium ions.

Therefore, only comprehensive, multifactorial testing of the developed materials can provide a definitive conclusion regarding their applicability in lithium extraction technologies. Nevertheless, for preliminary assessments, the selective ion separation coefficient and specific flux values are sufficient.

In Ref. (Yang et al., 2022), a composite IIMs TiO₂/PVDF-12C4E was proposed. The membrane was fabricated using a PVDF matrix with the addition of TiO₂ to enhance overall hydrophilicity, while surface modification with 12-crown-4 ether imparted lithium-ion selectivity. The selective permeation ability of created membrane was measured in multicomponent solution with the same concentration of Li⁺, Na⁺, K⁺, Mg²⁺, Ca²⁺ ions. The authors found that selective permeation ability varied in the following order: Li⁺ < Mg²⁺ < K⁺ < Ca²⁺ < Na⁺, which aligns with the transport mechanism previously described for IIMs and the relative hydrated ion sizes. It is noteworthy that the membrane exhibited close permeation ability for Li⁺ and Mg²⁺ ions, which can be attributed to their comparable hydrated ionic radii. According to our calculations, the membrane demonstrates a lithium flux of 0.03 mol/(m²×h) and a specific permselectivity S L i + / M g 2 + = 0.25 .

Similar results were reported in Ref (Cui et al., 2019) for a CA/CS-12C4E hybrid membrane composed of cellulose acetate and chitosan, further modified with TiO₂ and 12-crown-4 ether. However, in a mixed solution, the permeation ability differed and followed the order: Li⁺ < Na⁺ < K⁺ < Mg²⁺ < Ca²⁺. This deviation can be attributed to structural differences in the substrate membrane and variations in the loading level of the chelating agent. Under identical feed solution concentrations and operating solely under a concentration gradient, the lithium flux through the CA/CS-12C4E membrane, according to our estimates is lower (0.014 mol/(m²×h)), compared with the TiO₂/PVDF-12C4E membrane (Table 1).

In Ref. (Yu et al., 2020), a PDMS-PDA-calix [4]arene membrane was developed and tested using a different chelating agent (calix [4]arene). The enhanced lithium retention was attributed to its interaction with ester groups. However, based on the authors’ data, the lithium flux through this IIM is approximately 0.5 mol/(m²×h), while the selectivity toward coexisting singly charged Na⁺ and K⁺ ions is nearly an order of magnitude lower than that achieved with membranes employing 12-crown-4 ether as the chelating agent (Table 1).

The performance of lithium extraction using IIMs can be improved by applying an electric current or pressure as an additional driving force alongside the concentration gradient. In Ref. (Li L. et al., 2021), a polysulfone-type membrane, Sf-sg-AB12C4, with 12-crown-4 ether, was synthesized and tested under a constant potential drop of 2 V. Unfortunately, the authors did not report specific flux values, and the available quantitative data are insufficient to perform an independent calculation. Nevertheless, it is worth noting that the membrane demonstrates significantly higher selective ion separation coefficient.

Thus, the analysis shows that although composite IIMs can exhibit high selective ion separation coefficients with respect to coexisting ions, they typically demonstrate low specific lithium fluxes. However, for this membrane type, low fluxes imply low lithium loss due to its removal with competing ions. This is because the target lithium product remains in the feed solution compartment rather than passing through the membrane, unlike in most other membrane processes.

The ISMs, conversely, are designed for the direct extraction of lithium by selectively transporting only Li⁺ through the membrane, while coexisting ions are effectively excluded (Xu et al., 2024). Membranes of this type have been tested in both pressure-driven (Han et al., 2024), and electric-field-driven (Ounissi et al., 2022) processes.

MOFs-based ISMs deserve particular attention due to the rapidly increasing number of studies on their synthesis. In Ref. (Lu J. et al., 2021), a 3D UiO-66-COOH membrane was fabricated using an amino-functionalized PET-NC film whose pores were filled with carboxyl-functionalized UiO-66 crystals. This approach enabled the formation of subnanometer pores within a polymer nanochannel, resulting in a membrane that, according to our calculations, exhibits an exceptional selective ion separation coefficient for lithium over magnesium of 136.5 under an applied electric field. In Ref. (Han et al., 2024), a 2:1 [BMIM][AlCl4]-MOF TFN nanofiltration membrane was prepared on a PVDF film modified with MIL-101 grafted with different ionic liquids (ILs). Similar to other MOFs-based ISMs used in electrodialysis, the mechanism of lithium selectivity arises from a combination of the sieving effect and electrostatic interactions. In this membrane architecture, ionic liquids are employed to enhance electrostatic exclusion, thereby improving selectivity. According to the authors (Han et al., 2024), the membrane exhibited lithium selectivity in the presence of Mn²⁺, Co²⁺, and Ni²⁺, with values ranging from 9 to 10 (Table 1).

Although MOFs-based ISMs frequently achieve high selectivity, the specific lithium flux through them is often low at practically reasonable lithium concentrations in the feed solution (up to 0.5 g/L). Another limiting factor for the practical implementation of MOFs-based ISMs technology is the complexity of MOF fabrication and their associated high production costs.

Reviews (Abrishami et al., 2024; Xu T. et al., 2025) summarize both the achievements and the limitations of MOFs-based ISMs, enabling an assessment of their prospects for integration into lithium extraction technologies. The authors highlight a significant gap between laboratory-scale samples and the large-area MOF-based ISMs required for industrial ion separation. However, MOF-based membranes for gas separation are being actively developed in parallel, with reports of the fabrication of ZIF-8 flat membranes with areas of up to 2,400 сm², and even up to 4,800 сm² for spiral-wound membrane module (Yu et al., 2023).

A potential solution to the challenge of rapid scale-up is the use of less expensive materials for creating ISMs. MXene-based ISMs have recently emerged as a promising candidate in this regard. Deng et al. (Deng et al., 2021) reported the fabrication of a Ti₃C₂T_x membrane with a large area of up to 575 cm². Although MXenes are two-dimensional materials, unlike MOFs, the interlayer gaps in their structures provide comparably effective pathways for lithium-ion separation. Lithium selectivity relative to Na⁺ and K⁺ can reach values of around 10, while selectivity relative to Mg²⁺ reaches 25, with a lithium flux of 0.08 mol/(m²×h) (Lu Z. et al., 2021).

The use of LMs for lithium extraction also shows significant promise. Membranes of this type are characterized by high specific permselectivity (Table 1). Zante et al. (2020) reported the development and testing of PVDF-based FSSLMs employing heptafluoro-dimethyloctanedione (HFDOD) and tri-n-octylphosphine oxide (TOPO) as lithium-selective ligands. The selectivity toward Na⁺ reached 400, while the lithium flux in a concentration-gradient-driven system reached 0.052 mol/(m²×h). Another similarly successful strategy is the use of BLMs. The sandwiched liquid-membrane developed by Zhao et al. (2020) incorporates a lithium-selective agent in the internal space between two ion-exchange membranes. The use of ionic liquids such as tri-n-butyl phosphate (TBP), which are well established in solvent extraction of lithium, enables high selectivity in the electric-field-driven process while simultaneously achieving enhanced lithium fluxes.

Similar to other membrane types, LMs also face surface area limitations that affect their performance and hinder broader adoption. In Ref. (Dong et al., 2025), 14C4PI(TBP-NaNTf₂) HFSLMs designed as membrane contactors were fabricated and tested. Membrane contactors are devices equipped with polymer partitions that enable direct phase contact (gas–liquid or liquid–liquid) across an interface without dispersing one phase into the other. Their key advantage is efficient mass transfer between phases (Drioli et al., 2011). To maximize the effective use of the interface, membrane contactor configurations with the highest membrane surface area–to–volume ratio, namely, tubular or hollow-fiber membranes, are typically preferred. The 14C4PI(TBP- NaNTf₂) membrane is made of 14-crown-4 ether functionalized polyimide, the pores of which were filled with TBP and sodium bis (trifluoromethylsulfonyl)imide as a selective agent. With an area of 1.4 × 10⁻³ m², in Li⁺ and Mg²⁺ separation tests, the membrane demonstrated a lithium selectivity of 15.8 at a lithium flux of 0.18 mol/(m²×h) (Table 1).

Thus, this review demonstrates that, although numerous types of lithium-selective membranes and a wide range of implementation strategies have been reported in the scientific literature, moving to high levels of availability requires significant experimental development. Only a small fraction of laboratory prototypes has undergone detailed, long-term testing in multicomponent solutions for lithium extraction from natural brines. The complexity of fabrication methods, the need for continuous monitoring of membrane stability, the scarcity of comprehensive performance evaluations, and the inconsistent reporting of productivity, selectivity, and efficiency metrics (which vary depending on precursors, membrane type, and intended application) collectively hinder the implementation of LSM technologies for direct lithium extraction.

Direct lithium extraction using lithium-selective membranes represents a promising advancement in the field of lithium recovery. By leveraging the unique properties of specialized membranes, this method offers a more efficient, sustainable, and environmentally friendly approach to meet the growing demand for lithium. However, the present review highlights that the concept still faces a number of critical challenges. Despite achieving high selective ion separation coefficients, most LSMs exhibit relatively low specific lithium fluxes, necessitating productivity enhancement either through substantial increases in membrane surface area or by intensifying the driving force via pressure or electric field application rather than relying solely on concentration gradients. At the same time, the requirement for large membrane areas remains a major bottleneck for most LSMs. The cost of the precursors required for their fabrication and the overall complexity of the production technology continue to limit both productivity and industrial scalability. Furthermore, the productivity, selectivity, and efficiency of different types of LSMs are often expressed in non-uniform parameters, which complicates their comparison with other membrane types as well as with other processes that must be integrated into a complete extraction system. It is expected that continued research and development, aimed at improving LSM design and optimizing lithium extraction processes, will enhance the overall feasibility and efficiency of DLE systems.