Composite Separator with Superior Thermal Shrinkage Resistance for the Safety of Lithium Metal Batteries

ABSTRACT

High‐energy‐density batteries have raised numerous safety concerns, particularly thermal runaway. Separators play an indispensable role in battery safety. Herein, an Al₂O₃‐coated composite separator (LSO‐Al₂O₃@PE) with superior thermal shrinkage resistance is prepared with lithium polysilicate (LSO) as a binder. Different from the common organic binder, the binding force of inorganic LSO binder to Al₂O₃ particles can be further strengthened through the in situ dehydration and condensation at elevated temperatures, significantly enhancing the thermal shrinkage resistance of the composite separator (∼ zero thermal shrinkage at 200°C). The LSO‐Al₂O₃@PE separator also demonstrates thermal shutdown and acid impurity scavenging properties, effectively improving the safety and cycling performance of lithium metal batteries. The Li|LSO‐Al₂O₃@PE|LiFePO₄ cells retains 70% capacity after 1000 cycles, and the Li|LSO‐Al₂O₃@PE|LiNi_0.8Co_0.1Mn_0.1O₂ cells retains 81% capacity after 200 cycles. This work provides a new solution to improve the thermal shrinkage resistance of commercial polyolefin separators.

Al₂O₃ coated composite separator with superior thermal shrinkage resistance is prepared with lithium polysilicate (LSO) as a binder. The LSO binder can strengthen the coating layer at elevated temperatures and endue the composite separator with almost no thermal shrinkage at 200°C. The composite separator also demonstrates thermal shutdown and acid impurity scavenging properties, effectively improving the safety and cycling performance of lithium metal batteries.

1. Introduction

Lithium‐ion batteries (LIBs) are widely used in portable electronics, electric vehicles, and smart grids [1, 2]. To meet the demand for greater convenience, there is an urgent need for LIBs with higher energy density [3]. However, the pursuit of higher energy density has raised growing safety concerns [4, 5, 6]. The separator is an important guarantee for the safety performance of LIBs, yet it can fail under extreme thermal conditions. Separator failure, such as thermal shrinkage or meltdown of conventional polyolefin separators, may trigger internal short circuits, leading to thermal runaway and catastrophic safety risks [7, 8, 9, 10]. Ceramic‐coated separators (CCSs) have emerged as a promising solution to address these safety concerns [11, 12]. By applying inorganic particles such as SiO₂ [13], Al₂O₃ [14], Al(OH)₃ [15], Mg(OH)₂ [16], zeolite [17], etc., onto porous polymeric substrates, CCSs offer enhanced thermal stability and mechanical integrity.

Although CCSs exhibit significantly improved thermal stability, their practical application remains challenged by the limitations of conventional organic binders, such as polyvinylidene fluoride (PVdF), sodium carboxymethyl cellulose (CMC), and styrene butadiene rubber (SBR) [18]. When temperatures exceed the glass transition temperature or melting point of these binders, they soften and lose adhesive strength [19]. This can result in ceramic particle detachment and loss of coating integrity under thermal stress, undermining the safety advantages of the ceramic layer, evident in the increased thermal shrinkage of CCSs at high temperatures [19]. Although high‐temperature‐resistant polymers can partially mitigate separator shrinkage, they do not fundamentally solve the problem [20]. In contrast, inorganic binders generally offer superior thermal stability, making them a viable alternative for enhancing the thermal shrinkage resistance of CCSs.

Lithium polysilicate (LSO) is a water‐soluble inorganic binder that can form a water‐insoluble film upon drying. It exhibits outstanding thermal stability, withstanding temperatures above 1000°C, far surpassing conventional organic binders [21]. Additionally, aqueous LSO shows good permeability, penetrating deep into the substrate's capillary pores to form a dense internal reaction layer. Historically, LSO has been widely used in construction, anti‐corrosion, casting, and related industries [21]. Recently, lithium metasilicate and lithium polysilicate have been successfully employed as a binder for silicon anodes, where their rigid ceramic structure stabilizes the electrode and facilitates charge transport [22, 23, 24]. These properties allow LSO to serve not only as a binder but also as an ion‐conducting component, potentially improving overall battery performance.

In this work, LSO was used as an inorganic binder in Al₂O₃‐coated polyethylene (PE) separators. In contrast to conventional organic binders, LSO undergoes a thermally driven condensation reaction, forming a robust three‐dimensional silicate network that significantly enhances the CCS's structural integrity at high temperatures. This unique feature effectively prevents internal short circuits caused by separator shrinkage under thermal abuse conditions, thereby reducing the risk of battery thermal runaway at its source. Additionally, LSO exhibits multifunctional capabilities in tailoring electrode interfaces by scavenging HF and promoting dendrite‐free lithium deposition. The CCSs fabricated with the LSO binder demonstrate outstanding electrochemical performance in symmetric Li||Li and asymmetric Li||Cu cells, as well as in Li||LiFePO₄ and Li||NCM811 cells. These results highlight LSO as a highly promising alternative binder for next‐generation CCSs, providing an integrated strategy to overcome the safety limitations associated with current lithium‐ion battery technology.

2. Results and Discussion

2.1. LSO‐Al₂O₃@PE Separator Preparation and its Thermal Properties

A composite separator was fabricated via a simple aqueous slurry‐coating strategy (Figure 1a). Al₂O₃ particles (30.0 wt.%) were first homogeneously dispersed in water (∼60.0 wt.%) with the aid of a small amount of thickener and dispersant through planetary ball milling. Subsequently, lithium polysilicate (LSO) binder (9.0 wt.%) and a trace amount of surfactant were introduced, and the mixture was further ball‐milled to obtain a uniform slurry. The resulting slurry was then coated onto one side of the PE separator using a coating rod and dried in a forced‐air oven at 60°C for 24 h, yielding the single‐side coated LSO‐Al₂O₃@PE separator. Scanning electron microscope (SEM) images (Figure 1b) and Energy dispersive X‐ray spectroscopy (EDS) mapping (Figure S1) reveal that the LSO‐Al₂O₃ coating is uniformly distributed on the PE separator. The cross‐sectional SEM image shows that the thickness of the coating layer is about 4.0 µm (Figure 1c). The LSO‐Al₂O₃@PE separator maintains its structural integrity under various deformation conditions, including bending, rolling, and twisting, without any coating delamination (Figure S2).

FIGURE 1.

(a) Schematic illustration of the manufacturing process of LSO‐Al₂O₃@PE. (b) The surface and (c) cross‐sectional SEM images of the LSO‐Al₂O₃@PE separator. (d) Thermal shrinkage and (e) the photos of PE and LSO‐Al₂O₃@PE separators after being heated at different temperatures for 0.5 h. (f) Photos of the PE and LSO‐Al₂O₃@PE separators before and after ignition tests. (g) Comparison of thermal shrinkage resistance performance of LSO‐Al₂O₃@PE separator with recently reported results. (h) Internal resistance of SS||SS cells with PE and LSO‐Al₂O₃@PE separators under different temperatures, the inserted images were optical photos of disassembled cells. (i) Gas permeability of PE and LSO‐Al₂O₃@PE separators at different temperatures. (j) SEM images of the uncoated side of LSO‐Al₂O₃@PE separator after being heated at 140°C for 0.5 h.

To quantitatively evaluate the thermal dimensional stability of the separators, the separators were heated at different temperatures for 0.5 h, and the thermal shrinkage in the machine direction (MD) and transverse direction (TD) can be directly measured from the photographs. As shown in Figure 1d,e, the PE separator begins to shrink obviously at 140°C (MD for 30% and TD for 20%), exhibiting a thermal shrinkage ratio as high as 70% (MD) and 65% (TD) at 200°C. This is attributed to the relaxation of stretched chain segments and the melting of PE crystals during heating, reverting to a curled state. The application of an Al₂O₃ layer on the PE separator can obviously enhance the thermal shrinkage resistance of the Al₂O₃@PE separator. However, when organic binders are used, the thermal stability of the common Al₂O₃@PE separator is still too inferior (> 20% at T > 150°C) as the temperature is higher than the melting temperature of PE [25, 26, 27]. In contrast, the LSO‐Al₂O₃@PE separator maintains almost zero shrinkage in MD and 2.5% in TD at 200°C. To investigate the universality of LSO binder in enhancing the property of thermal shrinkage resistance, it was employed as a binder for various coating materials, including lithium deposition modifiers (e.g., MnCO₃ [28], MnO [29], SnO₂ [30]), flame retardants (e.g., melamine [31], Mg(OH)₂ [16]), and solid‐state electrolytes (e.g., LLZTO [32]). Optical photographs of the fabricated composite separators (Figure S3a) reveal that the coatings using LSO binder are uniform without visible cracks or delamination, demonstrating their broad material compatibility. Thermal shrinkage tests indicate that all composite separators maintain high dimensional stability, exhibiting near‐zero thermal shrinkage at extreme temperatures up to 200°C (Figure S3b). In a word, the use of LSO binder demonstrates significant advantages in enhancing the high‐temperature shrinkage resistance of battery separators, superior to other coated separators reported in the literature (Figure 1g and Table S1). This remarkable performance enhancement may stem from the use of the inorganic LSO binder.

To further evaluate the safety under extreme conditions, ignition tests were performed on three types of separators: bare PE, Al₂O₃@PE using a traditional organic binder, and the LSO‐Al₂O₃@PE (Figure 1f and Figure S4, Videos S1–S3). As expected, PE melted instantly, lost its structural integrity, and shrank completely upon exposure to fire (Video S1). For the Al₂O₃@PE separator, although coated with ceramic particles, the binder softened at extremely high temperatures. Consequently, the ceramic layer severely cracked, curled, and peeled off, failing to maintain the integrity (Video S2). In stark contrast, the LSO‐Al₂O₃@PE separator maintained exceptional structural integrity without catastrophic shrinkage or cracking (Video S3). This is primarily attributed to the non‐flammable and highly thermally stable nature of the inorganic LSO binder.

The thermal shutdown characteristics of LSO‐Al₂O₃@PE separator were evaluated by in situ monitoring internal resistance changes of stainless steel || stainless steel (SS||SS) cells at different temperatures. As shown in Figure 1h, during the programmed temperature rise process, the cell using the PE separator exhibits an internal resistance of 11 Ω at 150°C. In contrast, the internal resistance of the cell using LSO‐Al₂O₃@PE separator abruptly increases to 92 Ω upon reaching 140°C, demonstrating thermal shutdown effect at elevated temperatures. The disassembled cells reveal that the PE separator has shrunk obviously, while the LSO‐Al₂O₃@PE separator can still maintain dimensional integrity. These results demonstrate that the LSO‐Al₂O₃@PE separator shows excellent thermal shutdown ability at high temperatures. Differential scanning calorimetry (DSC) measurements corroborate that the LSO‐Al₂O₃@PE separator exhibits a distinct endothermic peak at approximately 136°C, attributed to the melting of PE (Figure S5). In addition, the results of the permeability test show that air could barely penetrate the LSO‐Al₂O₃@PE separator after being heated above 140 °C (Figure 1i). Further observation of the separator microstructure by SEM reveals that there is no obvious surface morphological change on the coating side (Figure 1b and Figure S6a,b). While on the uncoated side, the porous structure disappears and is replaced by a dense layer (Figure 1j and Figure S6c,d). EDS analysis confirms the substantial presence of silicon (Si) elements in the uncoated side (Figure S7). In summary, it is supposed that the PE melt wets the coating layer and becomes a dense layer, helpful for shutting down the ion channels as well as preventing the internal short circuit caused by shrinkage.

2.2. Mechanism for High‐Temperature Shrinkage Resistance

To investigate the underlying mechanism of LSO significantly enhancing the high‐temperature shrinkage resistance of the separator, the original Al₂O₃ and LSO‐Al₂O₃ particles were first characterized using transmission electron microscopy (TEM). The results show that the original Al₂O₃ particles (Figure 2a and Figure S8a) have distinct particle edges. On the contrary, clear edges are invisible on the LSO‐Al₂O₃ particles, and the surfaces are covered with a thin layer of LSO (Figure 2b and Figure S8b). X‐ray diffraction (XRD) patterns confirm that LSO is in an amorphous state (Figure 2d), suggesting that the Al₂O₃ particles are bonded together by an amorphous LSO binder.

FIGURE 2.

(a, b) TEM image of Al₂O₃ and LSO‐Al₂O₃ particles. (c) The surface SEM images of the LSO@PE separator. (d) XRD patterns of LSO particles, LSO‐Al₂O₃ particles, PE separator, and LSO‐Al₂O₃@PE separator. (e) FTIR spectra of pure LSO, Al₂O₃, and LSO‐Al₂O₃. (f) XPS spectra of O 1s of LSO‐Al₂O₃@PE separator at different temperatures. (g) The binding energies calculated by DFT. (h) The images of LSO‐PE and LSO‐Al₂O₃ structures.

Fourier‐transform infrared spectroscopy (FTIR) was used to characterize the LSO and LSO‐Al₂O₃ samples (Figure 2e). The formation of Si─O band (∼1060 cm⁻¹) in LSO is attributed to condensation reactions between silicon hydroxyl groups [33] (Figure 1a). In contrast, the Si‐O band (∼1070 cm⁻¹) [34, 35, 36] in LSO‐Al₂O₃ exhibits a narrower half‐width and a slight blue shift. This may result from the condensation reaction between Si‐OH groups in LSO and trace Al‐OH groups on the Al₂O₃ surface (Figure 1a), forming Si‐O‐Al bonds at the interface [37]. After heating at 200°C for 0.5 h, the intensity of the ─OH band (∼3480 cm⁻¹) in LSO‐Al₂O₃ significantly decreased, demonstrating that the remaining ─OH groups in LSO undergo deep condensation reactions, further increasing the cross‐linking degree and forming a complete Si─O three‐dimensional network configuration. The attenuated total reflection FTIR analysis on the coating surface can ascertain the further increase in the condensation degree of LSO (Figure S9). While on the uncoated surface, a distinct Si‐O characteristic band appears, confirming that PE melts and wets the inorganic rigid framework of LSO‐Al₂O₃ at high temperature, thereby achieving thermal shutdown property. Moreover, X‐ray photoelectron spectroscopy (XPS) characterization (Figure 2f) reveals that the relative intensity of the ─OH peak (533 eV in the O 1s spectra) diminishes after 200°C heat treatment, while the Si‐O‐Si peak at about 532.4 eV and the Si─O─Al bond peak at 531.2 eV significantly increase [38, 39]. All these results indicate that the LSO is a reactive binder. Compared with the hydrogen bonding and van der Waals force of organic binders [25, 26, 27], the chemical bond between LSO and Al₂O₃ can make the coating layer better integrated as a whole, which is further strengthened during the heating process, leading to superior thermal stability at elevated temperatures.

To explore why the LSO binder can suppress the shrinkage of PE molecular segments during the heating process, LSO@PE composite separator was prepared by coating 3.0 wt.% LSO solution onto PE separator. SEM image reveals that LSO spontaneously forms a continuous 3D network configuration on the surface of the PE separator (Figure 2c and Figure S10). Figure S10d clearly shows that after solidifying on the porous surface of the PE separator, the LSO forms a “hook structure” to achieve mechanical interlocking. Thermal shrinkage test of the LSO@PE separators reveals that although a very thin layer of LSO coating on the surface of the PE separator, the thermal shrinkage performance is significantly enhanced at 200°C (26% vs. 75% for PE, Figure S11), further validating the wetting of PE melt on the surface of the LSO coating layer.

Calculation based on density functional theory (DFT) provides a theoretical basis for the adhesion properties of LSO (Figure 2g,h). Results indicate that LSO exhibits a strong interaction energy with Al₂O₃ (−8.58 eV), confirming its potential for chemical bonding with inorganic ceramics. The binding energy of PE segments with LSO (−1.71 eV) is weaker than that with polybutyl acrylate (PBA, −2.81 eV) (typical component in polyacrylate binder), but comparable with some traditional organic binders, such as polyvinylidene fluoride (PVdF, −1.80 eV), polyacrylic acid (PAA, −1.72 eV) (Figure S12). Electrostatic potential analysis reveals LSO's adhesion potential to the PE separator (Figure S13). The Si─O structure in LSO possesses an electron‐rich cloud with greater electronegativity than the C─F bond in PVdF. When LSO approaches PE, these highly electronegative Si─O groups exert electrostatic attraction toward the positively charged C─H groups on the PE chains. Simultaneously, LSO induces dipole moments in the PE segments, causing electron cloud distortion in the nonpolar C‐H bonds of PE [40, 41]. This enhances induced dipole‐dipole interactions, and the van der Waals force subsequently increases, ultimately achieving efficient adhesion. The above results demonstrate that the three‐dimensional framework of LSO can adhere to the base PE separator, which can promote the adhesion between rigid inorganic LSO‐Al₂O₃ coating layer and the porous substrate, suppressing the shrinkage movement of PE molecular segments at elevated temperatures, and significantly improving the thermal stability of the LSO‐Al₂O₃@PE composite separator.

2.3. Ability to Scavenge Acid Impurity

The trace moisture and high‐temperature operation in batteries lead to the decomposition of LiPF₆ to form hydrofluoric acid (HF) [42]. The produced HF degrades the performance of LIBs by dissolving the solid electrolyte interphase films on the anode (SEI) and cathode (CEI), reacting with transition metal ions (e.g., Co, Mn, Ni) in the cathode, leading to premature capacity loss [42, 43]. The HF scavenging capacity of LSO‐Al₂O₃ materials was measured via Tunable Diode Laser Absorption Spectroscopy (TDLAS) [44]. The result shows that HF concentration is significantly reduced in the electrolyte containing LSO‐Al₂O₃ (Figure 3a). XPS spectra of the filtered LSO‐Al₂O₃ powder show that the Si─O (102.6 eV, Si 2p) completely disappears, replaced by Si‐F (104.3 eV) (Figure 3b), which can be ascertained in F 1s spectra [45, 46]. Figure 3c confirms that the reaction between LSO and HF yields Li₂SiF₆. Density functional theory (DFT) calculations further confirm that LSO‐HF and LSO‐PF₅ exhibit a strong binding energy (Figure S14). The above results indicate that LSO can capture acidic impurities, reducing their harm to SEI and CEI films.

FIGURE 3.

Impurity scavenging mechanism and ability of LSO‐Al₂O₃. (a) The concentration of HF after heating LE (containing 1% water) at 70°C for 2 h without and with LSO‐Al₂O₃ particles, the insert showing the optical photos of LE (left) and LE + LSO‐Al₂O₃ (right) after heat treatment. (b) XPS spectra and (c) XRD patterns of LSO‐Al₂O₃ particles before and after soaking in LE at 70°C for 2 h.

2.4. Physical and Electrochemical Properties

The LSO‐Al₂O₃@PE separator possesses high tensile strength (247 vs. 142 MPa for PE) and high Young's modulus (0.60 vs. 0.53 GPa for PE) (Figure S15). The puncture force has also improved obviously (562.5 vs. 453.2 gf for PE). Furthermore, the LSO‐Al₂O₃@PE separator exhibits superior electrolyte uptakes (∼ 376% vs. 114% for PE) and a lower surface contact angle with ester electrolytes (6.50°vs. 50.4°for PE) (Figure 4a). When the ester electrolyte of the same volume was dropped on the separator surface, it spread almost completely across the LSO‐Al₂O₃@PE surface, while the droplet was still intact on the PE surface (Figure S16), indicating significantly enhanced wettability. The LSO‐Al₂O₃@PE separator exhibits a broader electrochemical stability window (4.51 V vs. 4.32 V for PE) (Figure S17), enabling its application in high‐energy‐density batteries. The ionic conductivity of LSO‐Al₂O₃@PE separator is 0.73 mS cm⁻¹, higher than that of the PE separator (0.66 mS cm⁻¹) (Figure 4b). This improvement can be attributed to the superior electrolyte wettability of LSO‐Al₂O₃@PE separator, which facilitates more efficient electrolyte uptake and ion migration within the separator framework. Furthermore, the Li⁺ transference number ($t_{\mathrm{L}{\mathrm{i}}^{+}}$) was calculated using the Bruce‐Vincent method to quantify cation selectivity. As shown in Figure 4c and Figure S18, the LSO‐Al₂O₃@PE separator exhibits excellent $t_{\mathrm{L}{\mathrm{i}}^{+}}$ (0.69 vs. 0.34 for PE). This benefits from the LSO binder through changing the solvation structure of Li⁺ to increase their migration ability [13, 47]. Tafel plots reveal that the electrode employing LSO‐Al₂O₃@PE exhibits a significantly higher exchange current density (0.28 mA cm⁻²) compared to that using PE separator (0.05 mA cm⁻²) (Figure 4d), indicating that LSO‐Al₂O₃@PE effectively reduces energy barriers of the electrochemical reaction at the lithium anode/separator interface, thereby enhancing charge transfer kinetics. DFT calculations reveal that the binding energy of Li⁺ to some sites of the negatively charged Si‐O framework in LSO is significantly higher than that between Li⁺ and solvent molecules (Figure 4e). When Li⁺ migrates across the coating layer of the composite separator, this coordinating force changes the solvation structure of Li⁺ and promotes the interfacial desolvation process. Meanwhile, the binding energies of Li⁺ with various sites of the LSO framework are different, providing special hopping migration paths for Li⁺, facilitating its transport at the interface. Furthermore, the binding energy between LSO and PF₆ ⁻ (−0.69 eV) is higher than that between PE and PF₆ ⁻ (−0.13 eV), indicating that the LSO‐Al₂O₃@PE separator possesses a strong PF₆ ⁻ trapping capability, thereby slowing down the migration of anions in the electrolyte and significantly increasing the contribution of Li⁺ to the total ionic migration. In summary, the theoretical calculation results align with the determined high $t_{\mathrm{L}{\mathrm{i}}^{+}}$.

FIGURE 4.

(a) Electrolyte uptake and surface contact angles. (b) EIS of SS||SS cells using PE and LSO‐Al₂O₃@PE separators. (c) Polarization curves as well as the initial and steady state impedance spectrum of Li|LSO‐Al₂O₃@PE|Li cell. (d) Tafel curves of Li electrodes when using PE and LSO‐Al₂O₃@PE separators. (e) Binding energy of Li⁺ with solvent molecules and LSO, as well as binding energy of LSO‐PF₆ ⁻ and PE‐PF₆ ⁻. (f) Cyclic voltammetry curves of a Cu||Li cell scanned at 0.1 mV s⁻¹ in the range of 0.0–3.0 V with PE and LSO‐Al₂O₃@PE, respectively. (g) LUMO and HOMO energy levels of the components used in batteries. Li||Cu cells: (h) Voltage profiles; (i, j) SEM images of lithium deposition on Cu foil; (k) Coulombic efficiency (CE) cycled at 0.5 mA cm⁻² and 1.0 mAh cm⁻². (l) Cycling performance of Li||Li symmetric cells using PE, Al₂O₃@PE and LSO‐Al₂O₃@PE at 0.5 mA cm⁻² and 1.0 mAh cm⁻².

The lithium deposition/stripping efficiency was evaluated using Li||Cu cells. In the cyclic voltammetry (CV) curve of Li|PE|Cu cell, the reduction peak at near 1.30 V is attributed to the reduction reaction of nitrate ions (Figure 4f). In contrast, a new reduction peak appears at 1.62 V in the Li|LSO‐Al₂O₃@PE|Cu cell, which could originate from the reduction reaction of polysilicate ions. The reduction peak almost disappears after the 3rd cycle, indicating that the reaction was irreversible. The initial discharge curve shows that the Li|LSO‐Al₂O₃@PE|Cu cell consumes more lithium (0.113 vs 0.018 mAh for PE), and the nucleation overpotential is significantly lower (26.1 vs. 91.5 mV for PE) (Figure 4h). This may be attributed to the participation of LSO in the reduction reaction to generate lithophilic species (Li_xSiO_y), which is beneficial for reducing polarization and promoting more uniform lithium deposition. DFT calculations further confirm that LSO has the lowest the lowest empty molecular orbital (LUMO) energy level than the other electrolyte components, suggesting that it is preferentially reduced and participates in the formation of the solid electrolyte interface (SEI) on the lithium metal anode(LMA) (Figure 4g). LSO has lower the highest occupied molecular orbital (HOMO) energy level, indicating high oxidation stability. The Li deposition morphology on Cu foil was further characterized using SEM. Distinct loose Li dendrites with large cracks are observed on the Cu foil from Li|PE|Cu cell (Figure 4i). In contrast, a more compact and uniform Li deposition is seen on that from Li|LSO‐Al₂O₃@PE|Cu cell (Figure 4j).

The Li deposition/stripping efficiency of different separators was evaluated using Li||Cu cells. As shown in Figure 4k, Coulombic efficiency (CE) of Li|PE|Cu cell declines obviously after 60 cycles. Notably, the CE of the Li||Cu cell with an Al₂O₃@PE separator using traditional organic binders declines obviously after 100 cycles. In contrast, Li|LSO‐Al₂O₃@PE|Cu exhibits stable cycling performance up to 190 cycles with an average CE of 98.7% and maintains high CE (97.14%) even after 200 cycles, suggesting a more stable SEI film. Long‐term Li plating/stripping behaviors were further evaluated in Li||Li cells (Figure 4l). Li|PE|Li cells show obvious overpotential hysterisis after 300 h, indicating that the internal resistance surges due to unstable SEI formation and the accumulation of Li dendrites and “dead Li”. In contrast, the Li|Al₂O₃@PE|Li cell exhibited a prolonged cycling lifespan due to the presence of the Al₂O₃ coating, showing overpotential hysterisis after 550 h. On the contrary, although there is a larger initial overpotential, the Li|LSO‐Al₂O₃@PE|Li cells maintain a constant and stable overpotential (< 40 mV) for over 1000 h, indicating that the LSO can stabilize the interface of the Li metal anode. Collectively, these results indicate that the unique advantages of LSO in stabilizing the Li anode surface by facilitating more uniform Li⁺ flux and suppressing dendritic Li deposition.

2.5. Battery Performance

Li||LiNi_0.8Co_0.1Mn_0.1O₂ (NCM811) cells were assembled with LSO‐Al₂O₃@PE and compared with those using PE separators. Figure 5a shows that the NCM811 cells have higher C‐rate capacities when using the LSO‐Al₂O₃@PE separator, which can be attributed to its better wettability (Figure 4a) and improved charge transfer kinetics (Figure 4d). During long‐term cycling, the cycle performance of the Li|LSO‐Al₂O₃@PE|NCM811 cell is also significantly superior to that of the cell using PE separator (Figure 5b). The Li|LSO‐Al₂O₃@PE|NCM811 cell retains 81% capacity after 200 cycles and can be stably cycled for 600 cycles with a capacity retention of 60% and an average capacity decay of 0.117 mAh g⁻¹ per cycle. The capacity of the Li|PE|NCM811 cell experiences an obvious capacity decay after 178 cycles. The capacity retention is only 37% after 600 cycles, with an average capacity decay of 0.188 mAh g⁻¹ per cycle. The results indicate that the LSO‐Al₂O₃@PE composite separator can significantly improve the cycling performance of Li||NCM811 cells. To investigate the universality of this separator, Li||LiFePO₄ (LFP) cells were also assembled. Figure 5c,d shows that the Li|LSO‐Al₂O₃@PE|LFP cells show much better C‐rate and long‐term cycling performances than the Li|PE|LFP cell. The Li|LSO‐Al₂O₃@PE|LFP cell can be stably cycled for more than 1000 cycles with an average CE of 99.94% and capacity retention of ∼70%, significantly superior to other coating separators reported in the literature (Figure 5e and Table S2), demonstrating an obvious advantage in improving the cycling performance of LMBs.

FIGURE 5.

(a) Rate performance and (b) long‐term cycling performance of Li||NCM811 batteries. (c) Rate performance and (d) long‐term cycling performance of Li||LFP batteries. (e) Comparison of cycling performance in this work with recently reported results on Li||LFP.

Electrochemical impedance spectroscopy (EIS) was employed to monitor the evolution of internal resistance of the Li||NCM811 cells (Figure 6a,b). EIS fitting results reveal that the interfacial resistance (R _SEI) and charge transfer resistance (R _ct) of the Li|LSO‐Al₂O₃@PE|NCM811 cell show smaller fluctuation (Figure 6c and Table S3), indicating more stable SEI films and better reaction kinetics, which align with previous analysis. The EIS data were further analyzed using the distribution of relaxation time (DRT) method (Figure 6d,e) [48, 49]. Generally, the peak at 10⁻⁶–10⁻⁵ s can be assigned to the contact resistance, whereas the peak at ∼10⁻⁴ s can be attributed to the Li⁺ ion transport across the SEI layer [50]. Figure 6e shows much smaller resistance fluctuation at ∼10⁻⁴ s, confirming more stable interface layers on the electrode surfaces. The stabilization of interfacial layers is beneficial for uniform Li deposition and suppressing the dissolution of the transition metal. The Li deposition morphology on the Li anode from Li||NCM811 batteries after 50 cycles further confirms that the LSO‐Al₂O₃@PE separator can promote a smooth and uniform Li deposition with a thinner thickness (Figure 6g,h). Inductively coupled plasma‐atomic emission spectroscopy (ICP‐AES) tests reveal that, compared with the batteries using PE separators, the concentrations of transition metal ions in the electrolyte decrease three times from those using LSO‐Al₂O₃@PE separators, further verifying the advantage of LSO in scavenging HF and suppressing the dissolution of transition metal ions (Figure 6f).

FIGURE 6.

EIS profiles of (a) Li|PE|NCM811 and (b) Li|LSO‐Al₂O₃@PE|NCM811; (c) EIS fitting results. The distribution of relaxation time of (d) Li|PE|NCM811 and (e) Li|LSO‐Al₂O₃@PE|NCM811. (f) ICP‐AES results of transition‐metal dissolution in Li||NCM811 batteries after 50 cycles. The surface and cross‐sectional SEM images of Li metal anodes from (g) Li|PE|NCM811 and (h) Li|LSO‐Al₂O₃@PE|NCM811 batteries after 50 cycles. XPS spectra of Li metal anodes after 50 cycles: (i) C 1s, (j) F 1s, (k) Si 2p, and (l) Li 1s spectra.

XPS was applied to characterize the composition of the SEI and CEI films on the electrode surfaces of Li||NCM811 cells. On the surface of the Li metal anode, the characteristic peak near 290.6 eV in the C 1s spectrum is attributed to Li₂CO₃/ROCO₂Li, primarily originating from the decomposition of ester‐based electrolytes (Figure 6i) [51, 52]. The peak area proportion of Li₂CO₃/ROCO₂Li from LSO‐Al₂O₃@PE cell (6.5%) is lower than that from the PE cell (11.8%). This indicates a relative reduction in solvent decomposition products when using the LSO‐Al₂O₃@PE separator. In the F 1s spectrum, the peaks at 684.4 and 687.5 eV correspond to LiF and Li_xPO_yF_z, respectively (Figure 6j) [53, 54]. The SEI film of LSO‐Al₂O₃@PE cells shows a significantly higher LiF/Li_xPO_yF_z ratio. This may be attributed to the fact that LSO, as a strong Lewis base, efficiently captures the acid impurities at the source. The reduction of acid impurities is beneficial for cutting off their continuous destruction to the interface, changing the way of electrolyte decomposition, and promoting the formation of a more stable LiF‐rich SEI film. In addition, Si‐O‐Li species (101.4 eV, Si 2p) (Figure 6k) and the peak at 55.3 eV (Li 1s) (Figure 6l) can validate the presence of Li_xSiO_y in the SEI film [45], which can serve as lithophilic sites to reduce the nucleation overpotential and facilitate uniform lithium deposition. The inorganic‐rich SEI film has better stability, which can stabilize the Li anode interface and promote more uniform Li deposition, aligning with the SEM morphologies in Figure 6h.

On the NCM811 cathode surface, the peak area proportion of Li₂CO₃/ROCO₂Li in LSO‐Al₂O₃@PE cell (6.6%) is lower than that in the PE cell (9.3%) (Figure S19a), suggesting less solvent decomposition. In the F 1s spectrum, the area ratio of LiF/Li_xPO_yF_z is significantly higher on the cathode from LSO‐Al₂O₃@PE cell (Figure S19b). The acid scavenging capability of LSO‐Al₂O₃ coating layer may change the way of electrolyte decomposition on the cathode to form LiF‐rich CEI film. In the Ni 2p spectrum, the peaks at 854.8 and 856.8 eV correspond to Ni²⁺ and Ni³⁺, respectively (Figure S19c) [55, 56]. The Ni^{2 +}/Ni^{3 +} ratio serves as a sensitive probe for Li⁺/Ni^{2 +} antisite defect formation, where higher ratios reflect more severe structural degradation [57]. The Ni²⁺/ Ni^{3 +} ratio on the cathode from the LSO‐Al₂O₃@PE cell (33%) is lower than that from the PE cell (43%), indicating that LSO successfully suppresses the Li⁺/Ni^{2 +} cation mixing. This may benefit from the LSO capturing acidic impurities in the electrolyte, thereby reducing the dissolution of transition metal ions and inhibiting further structural degradation.

3. Conclusion

In summary, this work has developed a triple‐functional LSO‐Al₂O₃@PE separator through a simple coating process, combining high‐temperature resistance, thermal shutdown, and acid impurity scavenging capability. The Si‐OH groups in the aqueous LSO binder undergo self‐cross‐linking as well as cross‐linking with Al‐OH groups on the Al₂O₃ particle surface during the drying process, forming Si─O─Si and Si─O─Al bonds through dehydration condensation reactions, achieving strong adhesion. The dehydration condensation can proceed at elevated temperatures to further improve the thermal shrinkage resistance properties of the composite separator. The LSO‐Al₂O₃@PE separator exhibits significantly enhanced thermal stability (near‐zero thermal shrinkage at 200°C) and thermal shutdown capability (shutdown at 140°C). This prevents the risk of internal short circuits caused by separator shrinkage under extreme conditions, substantially reduces thermal runaway risks, and enhances battery safety. Furthermore, DFT calculations and experimental results indicate that LSO‐Al₂O₃@PE also possesses HF scavenging capability. The reduction of acid impurities changes the way of electrolyte decomposition, and promotes the formation of more stable SEI and CEI films to improve the battery performance. The Li||Li symmetric cells and Li||Cu cells using LSO‐Al₂O₃@PE separator demonstrate outstanding long‐term cycle performance and high CE. Moreover, Li|LSO‐Al₂O₃@PE|LiFePO₄ cell retains 70% capacity after 1000 cycles. Compared with those employing PE separators, Li|LSO‐Al₂O₃@PE|NCM811 cell demonstrates a 23% higher capacity retention rate after 600 cycles. This straightforward, low‐cost separator preparation strategy offers novel insights for enhancing the safety performance of high‐energy‐density LIBs.

4. Experimental Section

4.1. Materials

Lithium polysilicate (LSO) solution and carboxymethylcellulose sodium (CMC) were purchased from Aldrich Chemical Industry. α‐Al₂O₃ was purchased from Macklin. PE separators (12µm) were kindly provided by Cangzhou Mingzhu Co., Ltd (China). Liquids ethers electrolytes (LS‐002) were purchased from DuoDuo Chemical Tech Co., Ltd (Suzhou, China). Ether electrolyte was composed of 1.0 M bis(trifluoromethane sulfonyl)imide (LiTFSI) in 1,3‐dioxolane (DOL)/1,2‐dimethoxyethane (DME) binary solvent (v/v = 1:1) with 1.0 wt% LiNO₃. Liquids ester electrolyte (KLD‐1230C) was purchased from Kelude Co., Ltd (Dongguan, China). All materials and solvents were used without further purification.

4.2. Preparation of the LSO‐Al₂O₃@PE Separator

The initial slurry was prepared by planetary ball‐milling the mixture of 30.0 wt.% Al₂O₃ particles with 0.2 wt.% carboxymethylcellulose sodium and 0.5 wt.% sodium polyacrylate in 58.3 wt.% deionized water at 300 rpm for 10 h. And then, 9 wt.% LSO solution (binder) and 2.0 wt.% surfactants were added to the slurry at 150 rpm for 2 h. The resulting slurry was uniformly single‐sided coated onto the PE separator using a coating rod with a thickness of 4 µm. Finally, the obtained separators were dried at 60 °C for 24 h to obtain the LSO‐Al₂O₃@PE separator, which was used by cutting into discs with a diameter of 19 mm.

4.3. Fabrication of Cathodes

LiFePO₄ cathodes were prepared by mixing 85 wt.% active material, 5 wt.% conductive graphite, 5 wt.% acetylene black and 5 wt.% PVDF binder with N‐methyl pyrrolidinone (NMP). The slurry was homogenized using a planetary ball mill and then coated onto carbon‐coated aluminum foil using a doctor blade. The coated electrodes were dried at 65°C in an air‐circulating oven for 2 h, followed by vacuum dried at 120°C for 24 h, and then cut into discs with a diameter of 10 mm. NCM811 cathodes were prepared with the same procedures in a dry room (dew point: < −40°C) using NCM811 as the active materials. The areal density of active material was about 6 mg cm⁻².

4.4. Characterization

Thermal shrinkage ratio test: The separators were heated at different temperatures for 0.5 h, and the thermal shrinkage in the machine direction (MD) and transverse direction (TD) can be directly measured from the photographs. The thermal shrinkage ratio was calculated according to the following equation:

$$Thermalshrinkagerate=\frac{S_0-S}{S_0}\times 100\%$$

where S₀ (cm²) and S (cm²) are the areas of the separator before and after thermal treatment at various temperatures for 0.5 h, respectively.

Internal resistance test: The PE and the LSO‐Al₂O₃@PE separator were assembled into stainless steel || stainless steel (SS||SS) cells. The electrolyte was 20 µL of 1 m LiTFSI in tetraethylene glycol dimethyl ether (TEGDME). The assembled batteries were placed in an oven and subjected to a programmed temperature rise: the test temperature range was 100°C–150°C. After reaching the target temperature, it was held at this temperature for 0.5 h, and then the internal resistance was tested using an internal resistance tester.

HF concentration test: In a vial, 5.0 mL ester‐based electrolyte (1.0 m LiPF₆ in a mixture of EC/ DMC / EMC (v /v /v  = 1:1:1)), 1.0 wt.% (0.05g) H₂O, and 0.3 g LSO‐Al₂O₃ were sealed and heated at 70°C for 2 h. Subsequently, a syringe was used to extract 5.0 mL of volatile gas, and the concentration of hydrogen fluoride was measured via Tunable Diode Laser Absorption Spectroscopy (TDLAS). The control sample without LSO‐Al₂O₃ was prepared and tested with the same procedure.

The morphology was characterized by field‐emission scanning electron microscopy (SEM, SU‐8010, Hitachi) and transmission electron microscopy (TEM) (JEOL JEM‐F200). X‐ray diffraction (XRD) data were obtained using a D/MAX‐3C X‐ray diffraction meter with Cu Kα (λ = 1.5406 Å) radiation. Fourier‐transform infrared spectroscopy (FTIR) (Thermo Fisher Nicolet iS50 ATR) was performed in the wavenumber range of 4000–400 cm⁻¹. X‐ray photoelectron spectroscopy (XPS) (EscaLab 250Xi spectrometer) was employed for elemental and valence analysis. Mechanical properties were measured using a universal material testing machine (Instron 3366) at an elongation speed of 25 mm min⁻¹. The puncture force was tested with a C610H Auto Tensile Tester (Labthink, China). Contact angle measurements were conducted using a SZ‐CAMB3 machine. The gas permeability time of the separators was tested using a homemade gas permeability tester (1 inch², 40 mm H₂O column pressure, 100 ml air). Electrolyte uptake was obtained by measuring the weight of the samples before and after liquid electrolyte soaking. Differential scanning calorimetry (DSC) (PerkinElmer DSC8000) was conducted under a N₂ atmosphere at a heating rate of 10°C min⁻¹.

4.5. Batteries Assembly and Electrochemical Measurements

CR2025 coin cells were assembled in an Ar‐filled glove box (O₂ < 0.1 ppm, H₂O < 0.1 ppm). Li foils (diameter: 16 mm) were used for all cells. 60 µL ether electrolyte were used in the Li||Li and Li||Cu cells. 60 µL ester electrolyte was used in the Li||LFP and Li||NCM811 full cells. For the cells using LSO‐Al₂O₃@PE separator, the coating side faced the cathode.

Li||LiFePO₄ cells were activated at 0.1 C and then cycled at 0.5 C between 2.5 and 4.0 V (1 C = 170 mAh g⁻¹). Li||NCM811 cells were activated at 0.1 C and then cycled at 0.5 C between 2.8 and 4.3 V (1 C = 200 mAh g⁻¹). All galvanostatic charge/discharge tests were performed using the CT2001A cell test instrument (LAND Electronic Co. Ltd) at 30°C.

Electrochemical measurements were conducted on an electrochemical workstation (CHI 660E, ChenHua Instruments Co., China). SS||SS cells were assembled to determine the ionic conductivity. EIS data were collected in the frequency range of 10⁶ Hz to 0.1 Hz with an amplitude of 10 mV. Tafel curves were conducted in the voltage range of −0.15 ∼ 0.15 V with a scan rate of 0.1 mV s⁻¹. The Li⁺ transference number ($t_{\mathrm{L}{\mathrm{i}}^{+}}$) of different separators was determined by conducting EIS test, chronoamperometry test, and EIS test. The Li⁺ transference number was calculated using the formula:

$$t_{\mathrm{L}{\mathrm{i}}^{+}}=\frac{I_s\left(\mathrm{\Delta }V-I_0{R}_0\right)}{I_0\left(\mathrm{\Delta }V-I_S{R}_S\right)}$$

where I ₀ and I _s are the initial current and steady‐state current, respectively; R ₀ and R _s are the initial resistance and steady‐state resistance, respectively; ΔV is the applied polarization voltage (10 mV).

The electrochemical stabilities of the separators were evaluated through linear sweep voltammetry (LSV) with lithium metal and stainless steel electrodes (Li|separator|SS cells) on a Gamry electrochemical workstation within the voltage range of 2.0 and 6.0 V (vs Li/Li⁺) at a scan rate of 0.1 mV s⁻¹.

Inductively coupled plasma‐atomic emission spectroscopy (ICP‐AES) was used to measure the concentration of transition metal ions in Li||NCM811 batteries after 50 cycles. After cycled, Li||NCM811 batteries were disassembled, and all the accessories in the battery, except the NCM811 electrode sheet, were immersed in about 3.0 mL of DME. After they were soaked for 24 h, the supernatant was taken for ICP‐AES analysis.

4.6. Theoretical Calculation and Simulation

Molecular orbital energy calculations were performed utilizing the Gaussian 09 software package, using the B3LYP functional method and the 6–31+G(d, p) basis set.

Binding energy calculations were conducted utilizing Materials Studio 2020. The calculations in the study utilized the density functional theory (DFT) method, specifically the Perdew‐BurkeErnzerhof (PBE) exchange‐correlation function within the generalized gradient approximation (GGA). Molecular geometries were optimized using the DMol3 module. Parameters set for the calculations included a maximum energy of 10⁻⁵ Ha, a force of 0.002 Ha Å⁻¹, a displacement of 0.005 Å, and a self‐consistent field (SCF) convergence criterion of 10⁻⁶. The orbital cut‐off distance was set to 5.1 Å.

The binding energies (ΔE _b) of LSO and HF or PF₅ were calculated using the following equation:

$$\mathrm{\Delta }{E}_{\mathrm{b}}=E_{\mathrm{total}}-E_{\mathrm{LSO}}-E_{\mathrm{HF}\mathrm{or}\mathrm{P}{\mathrm{F}}_5}$$

where E _total, E _LSO, $E_{\mathrm{HF}\mathrm{or}\mathrm{P}{\mathrm{F}}_5}$ are the energies of the total system, LSO monomer, HF, and PF₅, respectively.

Additionally, the binding energies of Li⁺ or ${\mathrm{PF}}_6^{-}$ and the solvent molecules or LSO were calculated using the following equation.

$$\mathrm{\Delta }{E}_{\mathrm{b}}=E_{\mathrm{total}}-E_{\mathrm{solvent}\mathrm{molecules}\mathrm{or}\mathrm{LSO}}-E_{\mathrm{L}{\mathrm{i}}^{+}\mathrm{or}{\mathrm{PF}}_6^{-}}$$

where E _total, E _{solvent molecules or LSO}, and $E_{\mathrm{L}{\mathrm{i}}^{+}\mathrm{or}{\mathrm{PF}}_6^{-}}$ are the energies of the total system, the solvent molecules or LSO monomer, and Li⁺ ion or ${\mathrm{PF}}_6^{-}$ anion, respectively. In these calculations, all energy terms were obtained from single‐point energy calculations.

The binding energies between different binders and PE separators or Al₂O₃ were performed using the Forcite package in Materials Studio software. The size of the amorphous cells was 20 × 20 × 30 ų. Production runs were performed in an NVT ensemble and COMPASSII force field, and the temperature was controlled using a Nosé thermostat with a target temperature of 298 K. The corresponding binding energy could be calculated using the following Equation:

$$\mathrm{\Delta }{E}_{\mathrm{b}}=E_{\mathrm{total}}-E_{\mathrm{A}}-E_{\mathrm{B}}$$

where E _total, E _A, E _B are the energies of the total system, different binders (e.g., LSO, PVdF, PAA, PBA), and PE separators or Al₂O₃, respectively.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Supporting File 1: smll73915‐sup‐0001‐SuppMat.docx.

Supporting File 2: smll73915‐sup‐0002‐VideoS1.mp4.

Supporting File 3: smll73915‐sup‐0003‐VideoS2.mp4.

Supporting File 4: smll73915‐sup‐0004‐VideoS3.mp4.

Data Availability Statement

The data that support the findings of this study are available in the supplementary material of this article.