The rapid integration of renewable energy, driven by the urgent need to address climate change and further intensified by the escalating power demands of artificial intelligence, has made the development of grid-scale energy storage systems a critical technological priority^[1,2]. Consequently, secondary batteries suitable for large-scale applications have garnered increasing attention^[3-5]. After decades of innovation, aqueous zinc-ion batteries (AZIBs) have emerged as a particularly promising candidate, offering a compelling combination of intrinsic safety, economic viability, and high capacity (820 mAh g^-1)^[6,7]. Nevertheless, the reliable deployment of AZIBs faces challenges posed by detrimental dendrite propagation^[8,9], parasitic hydrogen/oxygen evolution^[9-11], dissolution of active materials^[12], and corrosion and passivation of electrodes^[13-15], which critically undermine the cycling stability and necessitate the coordinated innovations of electrodes^[16-20], electrolytes^[21-23] and separators^[24-26]. Compared to optimizing single electrochemical components, engineering the separator is capable of addressing those intertwined issues, given its central role in mediating the interactions between electrolytes and electrodes^[27].

Nanoporous materials such as metal-organic frameworks^[28-30], covalent-organic frameworks^[31], and zeolites^[32-34] offer highly tunable porosity at the nanometric scale, which presents significant potential for regulating the transportation of ions and water molecules, and thus have recently been explored as promising candidates for separators in high-performance AZIBs. Among these, zeolites are particularly notable for their chemical stability, cost-efficiency, and industrial scalability.

Compared to conventional glass-fiber (GF) separators, the zeolite-based membranes can markedly improve the performance across a wide range of current densities, as evidenced by enhanced capacity retention during cycling and superior rate behavior^[35-37]. These performance enhancements arise from several key factors. The ordered micropores in the crystallites facilitate the homogeneous deposition of Zn²⁺ ions on the anode, effectively mitigating the detrimental growth of dendrites^[37]. Concurrently, the inherent water-adsorption capability and size-exclusion properties of these micropores minimize parasitic side reactions at the anode interface^[35,37]. Furthermore, the micropores physically block the shuttling of soluble species generated at the cathode^[32,38,39], such as V³⁺/VO^x+[32,40] or I₃^-[36], which suppresses cathode dissolution^[41], alleviates anode erosion^[42], and mitigates self-discharge^[32]. Notably, these advantages are not limited to their use as bulk separators. Similar protective effects have been achieved by filling zeolites into the large pores in GF separators^[42] or by coating zeolites directly onto zinc anodes^[41]. This versatility further underscores the potential of zeolites to enhance the longevity and reliability of AZIBs.

Zeolites with different topological structures and pore sizes have different application scenarios. In the context of AZIB separators, Zhu and coworkers systematically evaluated nine zeolites and identified an optimal pore size window of 5-7 Å. Within this range, ZSM-5 and beta zeolites demonstrate superior specific capacity and capacity retention over 2,000 cycles^[35]. The transport of guest molecules and ions through zeolite crystals is critically influenced by the diffusion path length along the primary channels^[43]. For optimal performance, the diffusion pathway should generally be minimized, and the crystallographic orientation should be well-aligned to maximize mass transport efficiency. These structural requirements can be fulfilled by assembling ultrathin nanosheets prepared by bottom-up synthesis^[44,45]. While this approach is well-established for ZSM-5, it remains challenging for beta zeolites. Thus, the present work primarily focuses on the former.

ZSM-5 has a topology of MFI and is probably the most extensively utilized zeolite. With dC5 as the structural direct agent, Jeon et al.^[44] synthesized 5 nm thick nanosheets. Since then, considerable efforts have been directed towards fabricating large-area membranes via coating and secondary growth methods^[46,47]. Such membranes have demonstrated marked advantages over conventional zeolite membranes in the separation of xylene and butane isomers^[46,48,49]. More recently, separators based on MFI nanosheets (MFI-NS) have been used for flow batteries^{[39,40,50,51]} and lithium-ion batteries^[52,53]. In these systems, the well-aligned straight channels within the membrane are unequivocally confirmed to enhance charge-carrier transport relative to traditional zeolite membranes^[39,40,52]. However, previous efforts to integrate MFI-NS into battery systems have heavily relied on polymer binders or matrices, which inherently compromise the intrinsic ion-transport and mechanical properties of zeolites.

In this work, we exploit the self-assembly capability of MFI-NS to fabricate binder-free separators with well-aligned straight channels for AZIBs. It is demonstrated that this strategy effectively suppresses dendrite growth and mitigates parasitic side reactions, substantially extending the cycling life of AZIBs. Specifically, the V₂O₅||MFI-NS||Zn full cell delivers a specific capacity of 420 mAh g^-1 at a current density of 1 A g^-1 and retains 71% of its initial capacity after 1,200 cycles, demonstrating exceptional cycling stability. These merits underscore the strong potential of MFI-NS for improving the safety and service life of next-generation AZIBs.

Novel MFI-NS separators with highly oriented straight channels were fabricated following the procedure illustrated in Figure 1A. Briefly, MFI-NS with high aspect ratios were first synthesized via a bottom-up approach. The MFI-NS membrane was then prepared by vacuum filtration, followed by calcination to remove the SDA and promote inter-nanosheet condensation.

The XRD pattern of the as-synthesized MFI-NS displays the characteristic diffraction peaks of ZSM-5, such as the (101), (020), (501), (051), and (303) reflections^[39]. Beyond these typical features, three additional low-angle reflections below 8° are observed, which can be ascribed to the formation of a superlattice structure resulting from interactions among template molecules^[54]. In bulky ZSM-5, the intensity of (020) reflection is generally lower than that of (101) reflection. The markedly reversed relative intensity in Figure 1B suggests a preferential orientation of [010]. Consistent with previous reports^[44], the nanosheets consist of 50-100 nm thick small cores and ~5 nm thick peripheral regions that laterally extend 1-3 μm [Figure 1C and D, Supplementary Figure 1]. The SAED pattern of MFI-NS is unambiguously indexed with the [010] zone-axis reflections of ZSM-5 [Supplementary Figure 2], indicating that the longitudinal direction of the nanosheets aligns with the b-axis, which agrees well with the XRD pattern.

The MFI-NS membrane was prepared by vacuum filtration, which results in the lamellar stacking of nanosheets [Supplementary Figure 3A]. The thickness of the membrane is readily controlled by adjusting the concentration of the MFI-NS suspension [Supplementary Figure 3B-D]. To balance mechanical robustness and cost, a membrane of approximately 0.14 mm in thickness was selected for microstructural and electrochemical characterizations. In the as-assembled membrane, the SDA occupies the micropores, leaving only larger interparticle voids exceeding 14 Å (MFI-NS-mem in Figure 2A and Supplementary Table 1), which impairs the otherwise homogeneous transportation of Zn²⁺. Furthermore, while self-assembly renders the membrane structurally tailorable, it lacks sufficient mechanical strength to resist potential dendrite growth. Therefore, the membrane was calcined at 550 °C to remove the SDA and produce a self-supporting separator. Under 0.2 mm min^-1 tension, the obtained MFI-NS separator is two orders of magnitude stronger than that of the GF separator [Supplementary Figure 4], which can offer remarkable mechanical robustness against dendrite penetration.

After calcination, the characteristic ~0.6 nm micropores of the MFI framework are restored [Figure 2A, Supplementary Table 1]. Concurrently, the weak low-angle peaks below 8° in the XRD pattern [Figure 1B] completely disappear [Figure 2B], confirming the removal of organic templates and the sintering of adjacent nanosheets. The sintering proceeds via the dehydration of adjacent Si-OH groups in the MFI-NS^[55,56], as evidenced by the diminishing of the Si-OH peak in the FT-IR spectra [Supplementary Figure 5]. The calcined MFI-NS separator (MFI-NS-mem-cal in Figure 2A and Supplementary Table 1) exhibits a specific surface area of 383 m² g^-1, comparable to that of the calcined nanosheets (MFI-NS-cal in Figure 2A and Supplementary Table 1), indicating that calcination and concurrent sintering do not noticeably reduce the specific surface area. The restoration of micropores, together with the retention of high surface area, ensures efficient Zn²⁺ transport, as shown in later sections.

Notably, the preferred [010]-orientation is further enhanced after calcination [Figure 2B]. The I₍₁₀₁₎/I₍₀₂₀₎ intensity ratio decreases from 0.60 to 0.30, which is substantially lower than that of commercial ZSM-5 (1.49)^[57]. Consistently, the surface morphology exhibits well-aligned, face-on stacking of lamellar MFI-NS [Figure 2C and D]. The enhanced pore alignment is expected to further improve Zn²⁺ transport efficiency, approaching that of a single crystal oriented along the b-axis.

To be practically employed in AZIBs, the MFI-NS separator was wetted with the electrolyte, specifically a 3 M ZnSO₄ solution. The contact angle is 29°, indicating that the MFI-NS separator exhibits good wettability to the electrolyte [Supplementary Figure 6]. When 50 µL of the electrolyte was dropped onto the surface, the separator turned fully transparent, confirming sufficient electrolyte infiltration [Figure 3A]. The distribution of the electrolyte critically affects the electrochemical performance and was examined via high-angle annular dark-field scanning transmission electron microscopy (HAADF-STEM) combined with energy-dispersive X-ray spectroscopy elemental mapping. As demonstrated in Figure 3B-F, there are no obvious elemental segregations, suggesting that the electrolyte is homogeneously distributed.

Linear sweep voltammetry was performed to evaluate the corrosion of the Zn anode. As shown in Figure 4A, the anode paired with the MFI-NS separator exhibits a negligible hydrogen evolution reaction current compared to that paired with GF, indicating effective suppression of hydrogen evolution. The Tafel plot in Figure 4B further reveals that while the Zn anode with the MFI-NS separator displays a corrosion potential comparable to that with GF, its corrosion current (0.1160 mA) is an order of magnitude lower than that of GF (1.9331 mA). This substantial reduction suggests superior corrosion resistance enabled by the MFI-NS separator. Collectively, these results demonstrate that the MFI-NS separator effectively mitigates parasitic side reactions at the Zn anode, thereby likely extending the cycle life of AZIBs.

In addition to side reactions, Zn²⁺ transport is a critical factor influencing the performance of AZIBs and was evaluated by EIS [Supplementary Figures 7 and 8]. The calculated Zn²⁺ transference number (t_Zn²⁺) for the MFI-NS separator reaches 0.333, approximately three times that of the GF separator [Figure 4C]. The elevated t_Zn²⁺ reduces concentration polarization and enables more uniform ion transport, which could further suppress Zn dendrite formation^[35]. Notably, the t_Zn²⁺ for MFI-NS separator is ~50% lower than that measured in MFI/polymer composites^[35]. To exclude the effect of polymers and make a fair comparison between the separators with and without well-aligned straight channels, separators made by binder-free MFI granular particles (MFI-GP) were tested. The I₍₁₀₁₎/I₍₀₂₀₎ ratio in the XRD pattern of MFI-GP is 1.19 [Supplementary Figure 9], close to that for randomly dispersed MFI powder (1.49) and significantly larger than that for MFI-NS separators (0.30), confirming the absence of preferential channel alignment in the MFI-GP separator. The t_Zn²⁺ for the MFI-GP separator is 0.091, significantly lower than that of MFI-NS. In addition, from the impedance spectra in Supplementary Figure 10, the ionic conductivity for MFI-NS and MFI-GP is estimated to be 0.336 and 0.207 mS cm^-1, respectively. The remarkably higher t_Zn²⁺ and ionic conductivity for MFI-NS compared to MFI-GP suggest that the well-aligned straight channels in MFI-NS facilitate the transport of Zn ions.

To assess the efficacy of MFI-NS separators in suppressing dendrite formation on the Zn anode, long-term galvanostatic Zn plating/stripping cycling was performed using Zn||Zn symmetric cells. At a current density of 1 mA cm^-2 and an areal capacity of 0.5 mAh cm^-2, the cell with the MFI-NS separator maintains stable cycling for over 12,000 min, whereas the cell with the GF separator shortcircuits after only approximately 7,500 min [Figure 4D]. When the areal capacity is increased to 1 mAh cm^-2, the MFI-NS cell sustains stable cycling for over 9,000 min, which is four times longer than the GF cell [Supplementary Figure 11].

The origin of the enhanced performance was further investigated by disassembling the symmetric cells after 12,000 min of cycling and examining the Zn electrode surfaces. The Zn electrode paired with the MFI-NS separator exhibits a significantly smoother surface than that paired with GF [Supplementary Figure 12]. Further SEM examinations reveal that the anode from the MFI-NS cell maintains a smooth and dense surface with no evidence of dendrite formation [Figure 4E]. In contrast, the anode from the GF cell displays random zinc protrusions across the bare zinc foil, which is indicative of uncontrollable dendrite growth and undesirable side reactions [Figure 4F and G]. Collectively, these results demonstrate that the MFI-NS separator effectively regulates Zn deposition on the electrode surface, thereby enabling efficient suppression of dendrites.

Practical applicability of the MFI-NS separator was further assessed in full cells with V₂O₅ as the cathode and Zn foil as the anode. The CV curves [Figure 5A] exhibit multiple electrochemical redox peaks consistent with conventional V₂O₅||Zn systems^[35,58,59]. Notably, in the first cycle, the CV curve exhibits three redox pairs. The dominant pair at 1.34/0.99 V is likely caused by trapped H₂O molecules acting as interlayer pillars that stabilize the structure during Zn²⁺ (de)intercalation^[59]; this pair gradually diminishes in subsequent cycles. In contrast, the pairs at 0.82/0.43 and 1.08/0.90 V, corresponding to the stepwise intercalation and deintercalation of Zn²⁺ ions, are relatively weak in the first cycle, but become dominant and shift to 0.93/0.46 and 1.17/0.77 V after the first two cycles, respectively. Specifically, the redox pair at 0.93/0.46 V is attributed to the V⁴⁺ ↔ V³⁺ transition, whereas the pair at 1.17/0.77 V corresponds to the V⁵⁺ ↔ V⁴⁺ transition^[60,61]. Consistently, galvanostatic charge-discharge profiles display two distinct pseudo-plateaus during both charging and discharging [Figure 5B], corresponding to the changes in vanadium valence state^[35,61].

The V₂O₅||MFI-NS||Zn full cell exhibits outstanding rate performance compared with the GF-based counterpart [Figure 5C and D]. At current densities of 0.5C, 1C, 2C, 4C, 8C, and 10C (1C = 589 mA g^-1), the MFI-NS cell delivers specific capacities of 432, 427, 418, 396, 335, and 298 mAh g^-1, respectively [Figure 5C and D]. Upon returning to 0.5C from 10C, the capacity recovers to 426 mAh g^-1, corresponding to a recovery rate of 98.6%. Compared with previously reported composite separators consisting of MFI zeolites and polymer binders^[35], the MFI-NS membrane demonstrates slightly superior rate capability and capacity retention [Supplementary Table 2]. Long-term cycling at 0.5 A g^-1 [Figure 5E] and 1 A g^-1 [Figure 5F] further confirms the high stability of V₂O₅||MFI-NS||Zn. Specifically, after 600 cycles at 0.5 A g^-1, the capacity retention remains at 61% (226 mAh g^-1), and after 1,200 cycles at 1 A g^-1, a retention of 71% (300 mAh g^-1) is achieved. The full cell also exhibits remarkable performance under both much lower (0.1 A g^-1) and higher (5 A g^-1) current densities [Supplementary Figure 13], demonstrating robust cycling across a wide range of operating conditions. As V and S were detected on the Zn electrode after 400 cycles [Supplementary Figure 14], the performance degradation during long-term cycling is likely caused by cathode dissolution and anode passivation.

The exceptional electrochemical performance of V₂O₅||MFI-NS||Zn is attributed to the suppression of side reactions at the anode, enhanced Zn²⁺ transport, and inhibition of dendrite growth enabled by the uniform micropores of the MFI-NS separator. Expectedly, with the straight channels randomly oriented, the MFI-GP-based cell achieves only around 220 mAh g^-1 initially and retains 44 mAh g^-1 after 1,200 cycles [Figure 5F], which corroborates that the highly oriented straight channels are crucial for the superior electrochemical performance of the MFI-NS separators.

In this work, we have developed binder-free, highly [010]-oriented MFI-NS separators that simultaneously address multiple degradation mechanisms in AZIBs. The self-assembled lamellar separators are characterized by well-aligned straight channels perpendicular to the electrode surface, delivering a Zn²⁺ transference number 266% higher and an ionic conductivity 62% higher than those of separators comprising randomly oriented MFI particles, which thereby enables rapid and efficient Zn²⁺ transport. Concurrently, the uniform micropores in MFI zeolites promote homogeneous zinc deposition and regulate water transportation, effectively retarding dendrite formation and suppressing hydrogen evolution, respectively. The binder-free strategy further provides exceptional mechanical robustness, with tensile strength two orders of magnitude higher than that of commercial GF separators. These synergistic benefits endow MFI-NS separators with outstanding electrochemical performance. The V₂O₅||MFI-NS||Zn full cell delivers a specific capacity of 420 mAh g^-1 at 1 A g^-1 and retains 71% of its initial capacity after 1200 cycles, substantially outperforming the cells with separators comprising glass fibers or randomly oriented MFI particles.