The demand for eco-friendly energy resources that can replace conventional fossil fuels has increased exponentially as global climate change has accelerated. Lithium-ion batteries (LIBs), which have been successfully commercialized as energy conversion/storage systems, are now expanding from small- to large-scale devices [1]. Specifically, LIBs have been established as the preferred power source for electric vehicles (EVs), resulting in an exponential increase in their application market, accompanied by a high average annual growth rate. As the LIB market diversifies, the characteristics required for advanced LIBs are also evolving; for instance, reliable driving ranges for a single charge are necessary for EVs, which are predominantly influenced by the energy density of LIBs [2]. Note that the energy density of LIBs is directly proportional to the capacity and voltage of the electrode materials; consequently, extensive research has been conducted to discover advanced electrode materials that can provide high specific capacity and high voltage [3,4]. Numerous investigations have focused on advanced anode materials to explore those offering higher reversible capacities than commercial graphite anode materials, such as alloy reaction-based pure Si [5] and SiOx [6]. However, critical limitations exist regarding the direct use of pure Si and SiOx-based anode materials owing to severe volume changes during cycling. Therefore, considerable attention has recently been paid to particular attempts based on the use of graphite/Si composites to increase the specific capacity of anode materials by incorporating a higher amount of Si [7,8].

Numerous attempts have been undergone over the past several decades to employ Li metal as an anode material [9]. Notably, Li metal possesses the lowest electrochemical potential among anode candidates for LIBs, making it well recognized for its potential to enhance the energy density of LIBs [10]. Furthermore, Li metal can provide a high capacity of 3,860 mA h g^–1, which is approximately 10 times higher than that of conventional graphite anode materials [11]. Nevertheless, there are significant limitations in utilizing Li metal as an anode, and owing to its low electrochemical potential for Li deposition and dissolution; it is crucial to avoid continuous electrolyte decomposition at the Li interface during cycling [12]. The decomposition of the electrolyte is inevitable because of the low electrochemical potential; once electrolyte decomposition occurs, the Li interface becomes nonuniform, which accelerates dendritic Li growth around localized Li seeds appearing at the Li interface [13]. As dendritic Li grows rapidly, it can easily penetrate the poly(olefin)-based separator placed between the electrodes, triggering the thermal runaway of LIBs via internal short circuits [14]. This strongly suggests that the suppression of dendritic Li growth is essential to ensure stable cycling retention and safety characteristics of LIBs; therefore, inhibiting dendritic Li growth is the most challenging task for advanced batteries employing Li metal. For instance, controlling the anode structure by employing 1) a porous current collector [15], 2) porous graphene/graphite networks [16], and 3) conductive nanostructure scaffolds [17] has been developed to inhibit substantial volume changes in Li anodes. Furthermore, functional additives that make robust solid-electrolyte interphases (SEI) regulate the undesired surface reactions, such as tris (2,2,2-trifluoroethyl) borate (TTFEB) [18], tris(pentafluorophenyl)borane (TPFPB) [19], 2-fluoropyridine [20], fluoroethylene carbonate (FEC) [21], and vinylene carbonate (VC) [22]. Controlling the interfacial morphologies between the Li anode and the separator is also an effective way to inhibit dendritic Li growth by coating with Si₃N₄ [23], AlN [24], and SiO₂ [25] because such coating materials can prevent uneven growth of Li at the interface.

In this study, we attempt to incorporate an artificial solid-electrolyte interphase (SEI) into a polyethylene (PE) separator, which not only regulates the formation of dendritic Li but also simultaneously decreases electrolyte decomposition at the Li interface (Fig. 1a). Although PE separators are typically used in conventional LIBs, they are susceptible to internal short circuits upon the formation of Li dendrites, owing to their weak mechanical properties [26]. To address the poor mechanical properties of the PE separator, size-controlled oyster shells (OS), primarily composed of Ca elements that provide high mechanical strength to the substrate [27], were incorporated into the PE separator through a simple coating process to enhance its mechanical rigidity. OS are generally considered marine wastes that pose environmental risks because they can only be disposed of by landfilling instead of incinerating [28]. Nevertheless, numerous attempts have been made to utilize OS in various industries because they primarily comprise calcium carbonate (CaCO₃), which offers excellent mechanical properties [29]. Notably, preliminary studies have reported that a separator reinforced with mechanically improved materials having a shear modulus of over 7 GPa can inhibit the formation of Li dendrites [30]. This indicates that the OS-coated PE separator could regulate the growth of Li dendrites. Furthermore, it is anticipated that when the OS-coated side of the PE separator faces the Li metal anode, the interface acts as a physical SEI layer during cycling. This can reduce the contact between the electrolyte and the anode, leading to stable cycling behavior by suppressing electrolyte decomposition. Considering these strategies, the OS was incorporated on the Li metal side of the PE separator using an appropriate binder solution, and their physical, chemical, and electrochemical behaviors were systematically elucidated to assess their potential as separators for Li metal-based batteries.

The OS-coated PE separator was prepared as follows. Oyster shells, collected from Tongyeong, Korea (34°47'18'' N, 128°23'4'' E), were washed with de-ionized water to remove impurities and milled using a planetary ball mill (PM 100, Retsch). The particle size of OS was measured by a laser diffraction particle size analyzer (LDPSA, Bettersizer S, Bettersize), confirming that the final particle size used in this work was approximately 750 nm. After the milling process, 0.1 g of poly(vinylidene fluoride) (PVDF, Solvay) and 0.1 g of milled OS were dispersed in 3.23 mL of N-methyl pyrrolidone (NMP, Ashland), and it was coated on the PE separator by blade casting method. The OS-coated PE separator was dried at 60°C for 3 h and subsequently further dried at 60°C for 12 h in a vacuum oven. The average loading amount of the OS was 0.2 mg cm^–2, and the average thickness of the coating layer was approximately 2.0 μm.

The surface and cross-sectional images of the separators were obtained using scanning electron microscopy (SEM; JSM IT-800F, JEOL), and their chemical compositions were characterized using X-ray photoelectron spectroscopy (XPS; K-Alpha, Thermo Fisher Scientific). The tensile test of separators was evaluated by a tensile tester (JSV H1000, Coretech) using 1.5 cm × 5.0 cm dimensioned separators and the Gurley number of separators was measured by Gurley densitometer (G-B3C, Toyoseiki). For the shrinkage test, the separators (19 pi) were stored in an oven for 30 min, and the changes in dimensions were monitored with increasing temperature. The electrolyte compatibilities of the separators were measured as follows: 0.1 mL of electrolyte was dropped onto the separator, and images were taken after 30 s. The contact angles of the electrolyte-wetted separators were analyzed using a contact angle goniometer (Phoenix-MT, S.E.O.), and images were taken after 3 s. To quantify the ratio of the electrolyte uptake to retention, each separator was immersed in the electrolyte for 10 min (W_wet) and then dried for 10 min (W_dry). Electrolyte uptake/retention rates were calculated by the following equation: (W_wet – W_dry) / W_dry × 100% and (W_wet – W_dry) / W_wet × 100%, respectively. The ionic conductivities of the separators were measured using a potentiostat (VSP, Biologic). For estimating the electrochemical reactivities, linear sweep voltammetry (LSV) was used for each separator: the cells were fabricated with stainless steel (as working electrode), Li metal (as counter and reference electrodes), the electrolyte (ethylene carbonate (EC):ethyl methyl carbonate (EMC) = 1:2 v% + 1 M LiPF₆, Donghwa Electrolyte) and they were scanned from open circuit potential to 0.0 V (vs. Li/Li⁺) (cathodic polarization) or 5.0 V (vs. Li/Li⁺) (anodic polarization) with a scan rate of 0.1 mV s^–1.

For cycling the Li||Li symmetric cells, the cells were assembled with Li electrodes, an electrolyte, and each separator. The cells were precycled for 3 cycles with a current density of 0.2 mA cm^-2 for 1 h, and a current density was subsequently increased to 1.0 mAcm⁻². After cycling the Li||Li symmetric cells, the surface states of the Li electrodes were analyzed by SEM (JSM IT-200, JEOL). For the electrochemical performance of separators, the LiMn_0.6Fe_0.4PO₄ (LMFP64, Skyland) cathodes were prepared as follows. 1.6 g of the LMFP64 cathode material, 0.2 g of PVDF, and 0.2 g of a carbon conducting agent (Super C65, C-NERGY) were well dispersed in 4.8 mL of NMP and mixed for 1 h. The cathode slurries were coated onto an Al current collector and dried in a vacuum oven at 120°C for 12 h. The coin cells were assembled with the LMFP64 cathode, Li anode, electrolyte, and each separator, and were cycled at 1.0 C for 100 cycles.

The surface of the PE separators was analyzed by SEM, as shown in Fig. 1b and 1c. In contrast to the standard PE separator, the surface of the OS-coated PE separator was completely modified because new OSbased layers were deposited onto the PE separator. In further cross-sectional images of OS-coated PE (Fig. 1d), a Ca-based new layer was observed in EDS analysis with a thickness of 2.0 um. Note that the main component of the OS was Ca [31], indicating that the OS coating material was well incorporated onto the surface of the PE separator. The XPS analysis reveals that the upper layer of the OS-coated PE separator is composed of calcium elements [32], appearing at 345.3 eV (2p_3/2 of CaO), 348.8 eV (2p_1/2 of CaO) [33], 346.0 eV (2p_3/2 of CaCO₃) [34], and 349.6 eV (2p_1/2 of CaCO₃) [35] that are well harmonized with its EDS result (Fig. 1e). From these results, it can be concluded that the OS materials were well incorporated into the PE separator by the blade-casting method.

To estimate the changes in mechanical properties, a tensile test was performed, as shown in Fig. 2a. The standard PE and OS-coated PE separators showed similar tensile profiles; the OS-coated PE separator exhibited a slightly higher ultimate strength (138.0 MPa) than the PE separator (134.8 MPa). There were meaningful differences between the separators in terms of their Gurley numbers (Fig. 2b). Although it took 201.4 s to penetrate 100 cc of air from side to side, the OS-coated PE required a longer time (479.8 s). Notably, when the OS materials were incorporated into the PE separator, they clogged the pores of the PE separator, leading to a higher Gurley number than that of the PE separator. In the shrinkage tests (Fig. 2c), an improvement in the thermal stability of the OS-coated PE was observed. In detail, all the PE separators (with or without OS coating materials) declared identical dimensional areas when the temperature was increased to 135°C. However, when the temperature increased to 150°C, the standard PE shrank; only 61.9% of the dimensions remained at 150°C. Conversely, the OS-coated PE pronounced stable thermal shrinkage behaviors: 100.0% of dimensional are still maintained at 150°C. This implies that the incorporation of the OS coating material can increase the thermal stability of the PE separator, even at temperatures higher than the melting temperature of the PE separator.

The physical and electrochemical properties of the PE separators were further elucidated, as shown in Fig. 3. In the wettability tests (Fig. 3a), all the PE separators were well wetted with the hydrophilic conventional electrolytes; specifically, the electrolyte was immediately immersed in the OS-coated PE, whereas a longer wetting time was required for full wetting in the case of the standard PE. For the contact angle measurement, 50.26° was recorded for the standard PE. By contrast, 21.86° (a narrower angle than that of standard PE) was measured in the OS-coated PE. This means that the hydrophobic PE surface became more hydrophilic by incorporating the OS-based coating layers (mainly composed of CaCO₃), thereby leading to proper wettability against conventional electrolytes. These results agree well with the quantification results of the electrolyte uptake/retention behaviors (Fig. 3b). In the electrolyte uptake tests, only 142.3% of the electrolyte remained after the standard PE was immersed in the electrolyte. By contrast, the OS-coated PE exhibited an uptake ratio of 229.2%, which was nearly 1.6 times higher than that of the standard PE. Although the remaining electrolyte gradually decreased when the electrolyte-immersed PE separators were dried at ambient temperature, the OS-coated PE still displayed a higher retention ratio (69.6%) than standard PE (58.7%). This indicates that the incorporated OS-based layers favorably interact with the hydrophilic electrolytes, thereby increasing the wetting performance. It is supported by an analysis of the ionic conductivities of each separator (Fig. 3c): although the PE separator only showed ionic conductivity of 0.68 mS cm^–1, the OS-coated PE separator afforded increased ionic conductivity (1.02 mS cm^–1) because of its improved wetting performance against the conventional electrolytes. In the LSV tests (Fig. 3d), the standard PE showed stable reductive/oxidative behavior in the typical potential range of LIBs. In the OS-coated PE, the undesirable electrochemical reduction and oxidation behaviors were negligible in the identical potential ranges, implying that the OS-coated PE can be applied in the typical potential ranges of LIBs without side effects.

The effect of the OS coating material on the Li anode was evaluated using Li||Li symmetric cells (Fig. 4a). When the cell was assembled with the PE separator, significant degradation occurred after 1,000 h, which was attributed to increasing polarization. By contrast, the OS-coated PE afforded stable cycling behavior after 1,000 h. In the quantification of the potential hysteresis (Fig. 4b), a gradual increase in polarization was observed in the cell with the PE separator, indicating that the undesirable surface reactions were not regulated in the presence of the PE separator. Otherwise, relatively delayed polarization behavior was observed in the presence of the OS-coated PE separator. The recovered Li anodes indicated that the OS-coating approach efficiently regulated the undesired reactions at the interface of the Li anodes (Fig. 4c, d). The topview SEM image of the Li anode without OS-coating coating reveals the irregular growth of Li at the interfaces. By contrast, the recovered Li anode in the presence of the OS-coating layer showed that such irregular Li growth was alleviated because the OS-coating layer induced a uniform Li⁺ flux upon cycling. In the XPS analyses, considerable differential behaviors were observed in each separator (Fig. 4e, f). In all cycled separators, decomposed adducts corresponded to Li_xPF_y [36,37], LiF [38], and Li_xPO_yF_z [36,39] were commonly found; however, more abundant decomposed adducts appeared in the none-treated PE separator. In detail, Li_xPF_y/Li_xPO_yF_z and LiF occupied a prominent ratio in F 1s and P 2p spectra (4.89% and 38.24% in F 1s, respectively) while the portion of such decomposed adducts was markedly reduced in the OS-coated PE separator after cycling (1.00% and 1.66%, respectively). It indicates that the parasitic reaction appeared less in the OS-coated PE separator, which is well harmonized with its SEM analyses after cycling. To confirm the applicability of the proposed OS-coated separator with Li anode, cells composed of a Li anode and an LMFP64 cathode and their cycling behaviors were investigated (Fig. 4g, h). At the initial capacity, the cells exhibited similar discharge-specific capacities (PE: 150.3 mA h g^–1 and OS-coated PE: 150.5 mA h g^–1). However, as the number of cycles increased, cycling retention was rapidly fading in the absence of the OS-coated PE; only 66.9% of retention remained after 100 cycles. By contrast, the cell with the OS-coated PE showed a retention of over 83.8%, revealing that the OS-coated PE contributes to the increase in cycling retention by suppressing the unfavorable reactions at the Li anode.

OS, which is mainly composed of CaCO₃ with a high shear modulus, was proposed as an efficient coating material for PE separators that can suppress fast-growing dendritic Li in LMBs. The OS-coated PE was prepared using a PVDF binder solution via the blade-casting method. The thickness of OS layers was controlled at 2.0 μm, and further SEM analysis indicated that the OS was uniformly distributed on the PE separator. In the tensile test, the OS-coated PE delivered comparable stress–strain curves comparable to those of the nonmodified PE. In the shrinkage test, it was observed that the thermal stability of the OS-coated PE remarkably increased compared to that of the unmodified PE, as the coated OS inhibited the drastic shrinkage behavior of the PE substrate. Moreover, the wettability of the OS-coated PE against conventional electrolytes was improved by incorporating hydrophilic OS materials into the hydrophobic PE separator, leading to increased electrolyte uptake/retention ratios and ionic conductivities. The OS-coated PE was stable in the typical potential ranges of LMBs, accompanied by increased cycling retention of the Li||Li and Li||LMFP64 cells, because undesired surface reactions were alleviated by the incorporated OS layers.