Flame-retardant polybenzimidazole-based composite separator for enhanced high-temperature performance and fire safety of lithium-ion batteries

Abstract

The non-uniform pore size distribution and high flammability of commercial separators pose significant challenges to the safe application of high-energy-density lithium-ion batteries. In this study, a flame-retardant composite separator (P@HLi) with high thermal stability was successfully developed, which not only suppressed lithium dendrite growth but also improved high-temperature cycling performance of batteries and significantly enhanced their thermal safety. Li//Li symmetric batteries equipped with P@HLi-20 separators demonstrated stable cycling for over 600 h at a low polarization potential (approximately 50 mV), effectively reducing the formation of “dead lithium” and lithium dendrites. The LFP//Li and NCM811//Li cells with P@HLi-20 separators delivered initial discharge specific capacities of 142.0 and 167.9 mAh/g, respectively. Notably, the LFP//Li battery with P@HLi-20 separator showed excellent high-temperature cycling performance, maintaining 98.0% capacity retention and a discharge capacity of 131.1 mAh/g after 100 cycles at 1 C at 90 °C. Furthermore, pouch cells assembled with P@HLi-20 separators exhibited reductions of 52.67% in peak heat release rate (PHRR) and 68.42% in total heat release (THR) compared to those using Celgard separators, demonstrating superior thermal safety. These results confirm that the P@HLi separator offers comprehensive improvements in both electrochemical performance and safety characteristics.

Graphical abstract

Keywords

poly(arylene ether benzimidazole) / fire safety / flame retardant / lithium-ion batteries / high-temperature performance

1 Introduction

The global commitment to carbon-neutral targets has significantly accelerated the development of sustainable energy systems, wherein lithium-ion batteries (LIBs) serve as pivotal components for energy transition [1,2]. This strategic positioning has driven continuous technological advancements and market expansion, particularly in applications requiring high energy density and lightweight design [3,4]. However, the non-uniform pore structure of commercial separators induces inhomogeneous lithium deposition, leading to irreversible lithium dendrite formation during cycling [5–7]. These dendrites not only penetrate separators, causing internal short circuits, but also generate substantial dead lithium, ultimately degrading the energy density and cycle life. Moreover, the limited thermal stability and inherent flammability of conventional polyolefin separators significantly compromise battery safety, as elevated temperatures can trigger thermal runaway hazards [8,9]. Consequently, while higher energy densities are being pursued, safety concerns have emerged as critical bottlenecks hindering further development of LIBs technologies [10].

To enhance safety and electrochemical performance, various strategies have been employed in separators design, primarily including surface-coated separators, nanocomposite separators, and novel polymer-based separators [11]. Surface coating technologies, typically involving ceramic materials or inorganic nanoparticles, improve thermal stability by preventing separator melting or thermal shrinkage under overheating conditions [12,13]. For instance, Liao et al. [14] fabricated microencapsulated ammonium polyphosphate@silicon dioxide (APP@SiO₂)-coated Celgard separators, which enhanced flame retardancy and prevented short circuits through chemical reactions with lithium dendrites. Despite these benefits, surface-coated separators exhibit inherent limitations: prolonged cycling may lead to coating delamination or degradation, while uneven coating distribution or excessive thickness may impair ionic conductivity and charge-discharge efficiency [15]. In previous research, P@AS separators fabricated by incorporating APP@SiO₂ into polyvinyl alcohol (PVA) exhibited both excellent electrochemical and flame-retardant properties [16]. Nevertheless, nanofillers in nanocomposite separators are prone to agglomeration, leading to uneven pore distribution or clogging, reduced lithium-ion transport efficiency, and phase separation due to poor interfacial compatibility between fillers and polymer matrix, thereby triggering phase separation and weakening the mechanical properties.

Recent developments have focused on novel polymer materials, such as polybenzimidazole (PBI) [17], polyimide (PI) [18], poly(ether ether ketone) (PEEK) [19] and polyphenylene sulfide (PPS) [20], which exhibit superior thermal stability and mechanical strength compared to polyolefins. Despite these advantages, polymer-based separators face challenges hindering widespread application. Their complex manufacturing process requires precise control over production parameters and specialized materials, leading to higher production costs compared to conventional counterparts. Achieving uniform pore structures remains difficult due to irregular polymer chain alignment during fabrication, resulting in compromised ionic conductivity and ion transport efficiency. Furthermore, interfacial incompatibility between polymer matrices and inorganic components can reduce mechanical strength and thermal stability, while elevated temperatures may induce shrinkage or decomposition, raising safety concerns. For example, Ren et al. [21] developed a phase-separation-induced poly(m-phenylene isophthalamide) (PMIA) separator with hierarchical pore structures and outstanding dimensional stability. Sun’s group [22] designed a quaternized PBI/PI composite separator (q-PBI@PI), where lithiophilic and anion-affinic functional groups in q-PBI suppressed dendritic growth and enhanced mechanical strength. Yu et al. [23] engineered an ultra-thin heat-resistant separator comprising sea-island structured PPS fibers and glass nanofiber networks, achieving remarkable capacity retention and rate capability.

Poly(arylene ether benzimidazole) (OPBI), as a high-performance polymer, offers several advantages when applied to separators: (1) excellent thermal resistance and flame-retardant properties, with decomposition temperature exceeding 300 °C and minimal deformation under high temperatures, which can significantly improve the safety performance [24,25]; (2) high mechanical strength, attributed to the rigid aromatic backbone and hydrogen bonding between the molecular chains, enabling resistance to lithium dendrites penetration [26]; (3) a controllable pore structure and good ionic conductivity due to the potential formation of homogeneous microporous structure and the presence polar groups [27]; and (4) excellent electrochemical stability over a wide voltage window, enabling suitability for high-voltage environments.

Halloysite nanotubes (HNTs) are naturally occurring hollow tubular clay nanotubes formed by the inwardly curling of Kaolin flakes. These materials are abundant, environmentally friendly, and advantageous for the application of separators due to: (1) excellent thermal stability and flame-retardant properties, with decomposition temperatures exceeding 700 °C and structural integrity even after high-temperature calcination [28]; (2) a natural hollow nanotube structure, high specific surface area, and bipolar surfaces, which contribute to electrolyte affinity and adsorption properties [29]. To further enhance the ionic conductivity, HNTs were acidified and lithiated to introduce additional mobile Li⁺ ions into the structure, enhancing lithium storage capacity ion transport pathways. Incorporation of acidified and lithiated HNTs (HNTs-Li) into OPBI separators optimized the pore structure, increased porosity, and synergistically enhanced both ionic conductivity and flame retardancy, thus improving overall separator performance.

Although previous studies explored the use of HNTs [30,31] to enhance the thermal stability of separators or exploited the flame retardancy of OPBI [32,33], the synergistic effects of both materials on ionic conductivity and flame retardancy remained unexplored. The combined use of OPBI and HNTs in this work enabled the simultaneous optimization of ionic conductivity, porosity, and flame retardancy. In this study, HNTs were modified by acidification and lithiation, then integrated into an OPBI matrix via the non-solvent-induced phase separation (NIPS) process [34], resulting in a separator with a three-dimensional interconnected porous structure. This approach addressed the traditional trade-off between porosity and mechanical strength under high loads. Moreover, the P@HLi-20 separator (with 20 wt% of HNTs-Li content) was applied to soft pack batteries to evaluate the safety performance. The intrinsic flame retardancy of OPBI combined with the high thermal stability of HNTs significantly improved the battery safety.

The multifunctional P@HLi (composite separators with HNTs-Li added to OPBI separators) composite separator, fabricated via the NIPS method, demonstrated excellent thermal stability and electrochemical performance. Key innovations in this study include:

(1) Pore regulation and ionic conductivity: The unique tubular structure of HNTs regulated OPBI pore distribution and improved the porosity. Density Functional Theory (DFT) calculations showed strong interactions between polar groups on the HNTs/OPBI and solvent molecules, thus improving the electrolyte affinity of the separators. Additionally, HNTs provided fast Li⁺ transport channels, and HNTs-Li introduced active Li⁺ sites that increase the mobile Li⁺ concentration, achieving a high ionic conductivity of 1.551 mS/cm.

(2) Dendrite suppression: The P@HLi-20 separator exhibited superior dendrite suppression capability. Li//Li cells with P@HLi-20 separator maintain stable over 600 h with low polarization (~50 mV), significantly reducing dead lithium and dendritic formation. High porosity and excellent electrolyte wettability promoted uniform Li⁺ deposition and LiF-enriched solid electrolyte interphase (SEI) formation. The enhanced mechanical strength (8.7 MPa) effectively prevented dendrite penetration and extended cycle life.

(3) High-temperature stability: Both OPBI and HNTs have excellent high-temperature tolerance, and the addition of HNTs acts as a tubular support that improves the structural stability of the separators, resulting in excellent thermal stability of the P@HLi separators. Cells using P@HLi-20 separators maintained a discharge specific capacity of 131.1 mAh/g after 100 cycles at 90 °C, with a 98.0% capacity retention rate. This confirms the excellent high-temperature cycling performance of the P@HLi-20 separator, demonstrating its suitability for high-temperature environments.

(4) Flame-retardant performance: The flame-retardant properties of OPBI enhanced by HNTs-Li, through physical barrier formation and catalytic carbonization, enabled fire self-extinguishing behavior. In thermal runaway tests of soft pack batteries, the peak heat release rate (PHRR) and total heat release rate (THR) of P@HLi-20 were significantly reduced by 52.67% and 68.42%, respectively, validating the enhanced safety performance enabled by the synergistic flame-retardant mechanism.

2 Results and discussion

2.1 Morphology and composition of HNTs and P@HLi composite separator

The synthesis process of the P@HLi composite separator is schematically illustrated in Fig.1(a). HNTs-Li was uniformly dispersed in an N-methyl-2-pyrrolidone (NMP) solution containing OPBI, followed by coating and phase inversion to form a porous composite separator. As shown in Fig.1(b), the scanning electron microscope (SEM) image of HNTs-Li reveals a distinctive tubular morphology with a high aspect ratio (> 50). The Brunner-Emmet-Teller (BET) method was used to analyze the detailed parameters of the pore structure of HNTs. As shown in Fig. S1, the nitrogen adsorption/desorption isotherm results conform to a type IV isotherm, and the gas adsorption capacity of HNTs-Li was higher than that of pristine HNTs, probably due to increased porosity after acidification. The specific surface areas of HNTs-Li and HNTs were calculated as 40.8979 and 19.0615 m²/g, respectively. The pore size distribution indicates the pore diameters of both HNTs-Li and HNTs are mainly centered around 3.5 nm.

The P@Li-20 separator exhibits higher porosity and more uniform pore size distribution compared to the Celgard separator as shown in Fig.1(c) and Fig.1(d). Fig.1(e) and Fig.1(f) show the cross-sectional microstructures of the OPBI (0%HNTs-Li) and the P@HLi-20 separator, respectively, with a magnified view of the cross-section of P@HLi-20 shown in Fig.1(g). P@HLi-20 separator has a higher porosity, a more uniform pore size distribution, and embedded HNTs. Compared to OPBI and P@HLi-10 separators with (containing 10% of HNTs-Li by mass), the increased HNTs content in P@HLi-20 separator facilitates the formation of a three-dimensional network, resulting in a thinner structure with more uniform conductivity and a disordered, highly dispersed high porosity (Fig. S2, Electronic Supplementary Material). Elemental mapping images in Fig.1(h)–Fig.1(k), confirm the uniform distribution of C, O, Al, and Si elements, verifying both the successful fabrication of the P@HLi-20 separator and the uniform dispersion of HNTs.

Fourier transform infrared (FTIR) spectroscopy was employed to analyze the chemical composition of HNTs and P@HLi composite separators. As shown in Fig. S3, characteristic absorption bands of HNTs are observed at 3694 cm⁻¹ (O–H stretching vibrations of hydroxyl groups), 1030 cm⁻¹ (symmetric stretching of Si–O–Si bonds), 910 cm⁻¹ (bending vibrations of Al–OH groups), and 540 cm⁻¹ (bending vibrations of Si–O–Al linkages) [35]. The FTIR spectrum of HNTs-Li remains nearly identical to that of pristine HNTs, indicating that lithium incorporation does not significantly alter the original structure. For the OPBI matrix, characteristic peaks appear at 1673 cm⁻¹ (C=O stretching vibrations), 1600 cm⁻¹ (C=C aromatic stretching), 1445 cm⁻¹ (benzimidazole ring vibrations), and 1165 cm⁻¹ (ether linkage vibrations) [36]. The FTIR spectra of P@HLi-10 and P@HLi-20 composites exhibit superimposed features from both components, confirming successful integration of HNTs-Li into the OPBI matrix.

X-ray diffraction (XRD) analysis shows no substantial change in the characteristic peaks of HNTs and HNTs-Li, suggesting the preservation of HNT structure. However, the intensity of HNTs-Li peaks decreases, indicating reduced crystallinity. The (001) diffraction peak shifts from 12.24° to 12.48°, likely due to the smaller ionic radius of Li⁺, which reduces interlayer spacing. This confirms successfully ion exchange modification in the preparation of HNTs-Li.

2.2 Thermal stability, flame retardant properties and mechanism of P@HLi separator

The thermal stability of separators directly determines the battery performance at elevated temperatures. As shown in Fig. S4, the dimensional stability of various separators was assessed by subjecting them to oven treatment at different temperatures for 30 min. The commercial Celgard separator exhibited slight yet discernible shrinkage at 100°C and underwent severe deformation at 150 °C. In contrast, both OPBI and P@HLi separators retained their structural integrity even at 200 °C, owing to the intrinsic thermal resistance of the OPBI polymers and HNTs [37].

Further thermal stability assessments of inorganic particles and separators were conducted via thermogravimetric analysis (TGA) as shown in Fig.2(a). HNTs, as aluminosilicate minerals, exhibits high thermal resistance, undergoing dehydration between 400 and 500 °C and remaining structural integrity up to 800 °C [38]. Accordingly, both HNTs and HNTs-Li display a partial mass loss in 450–500 °C range, with minimal change beyond this point. In comparison, the Celgard separator began decomposing at approximately 350–400 °C under a nitrogen atmosphere. For OPBI and P@HLi separators, minor mass loss observed between 25–250 °C is primarily attributed to the evaporation of residual solvents and moisture from the fabrication process. A pronounced weight loss near 600 °C corresponds to the decomposition of the OPBI polymer. Notably, the P@HLi-20 separator demonstrated significantly enhanced char residue, arising from the inherent high thermal resistance of HNTs and their synergistic interaction with the OPBI matrix.

Differential scanning calorimetry (DSC) results, shown in Fig.2(b), further support these findings. A distinct melting peak was observed for Celgard separators at approximately 165 °C. In contrast, OPBI-based separators exhibited no detectable melting behavior below 200 °C, demonstrating superior thermal stability compared to polyolefin separators.

To assess thermal safety under potential thermal runaway scenarios, micro combustion calorimetry (MCC) tests were employed to evaluate the thermal safety of separators by quantifying thermal runaway conditions, including heat release rate (HRR) and THR. As shown in Fig.2(c), the PHRR of Celgard during combustion reached approximately 1106 W/g, whereas the P@HLi-20 separator exhibited a significantly lower PHRR of 31 W/g. Furthermore, the temperature at which P@HLi-20 reached its PHRR (611 °C) was substantially higher than that of Celgard (474 °C). Similarly, Fig.2(d) demonstrates that the THR of Celgard was 40.5 kJ/g, while the THR of P@HLi-20 was markedly reduced to 2.78 kJ/g. These results indicate that P@HLi separators both reduces the THR and suppresses the HRRs, thereby improving safety in thermal runaway or high temperature conditions.

To further characterize the flame-retardant properties, direct flame exposure tests were conducted (Fig.2(e)). The Celgard separator underwent rapid shrinkage and ignition upon exposure to flame. The OPBI separator melted and combusted but formed a compact char layer. In contrast, the P@HLi-20 separator melted and combusted initially but self-extinguished immediately after flame removal, forming a stable, coherent char structure. This behavior demonstrates that P@HLi-20 possesses intrinsic self-extinguishing capability and superior char-forming properties.

To further investigate the flame-retardant mechanism of P@HLi separators, thermogravimetric-infrared spectroscopy (TG-IR) was employed to analyze the decomposition process and gaseous products at elevated temperatures. As illustrated in Fig.2(f), the commercial Celgard separator (polypropylene-based) exhibited initial decomposition at 408 °C, which aligns with its TGA profile (mass loss onset at 400–420 °C under nitrogen atmosphere). The infrared absorption peaks at 2962 and 2914 cm⁻¹ correspond to the asymmetric and symmetric C−H stretching vibrations of aliphatic hydrocarbon fragments (−CH₃, −CH₂-), respectively, indicating the release of volatile alkane compounds during the thermal degradation of the polyolefin matrix in Celgard, consistent with its intrinsic flammability. At higher temperature (451 °C), additional characteristic peaks appeared at 1460, 1380, 1650, and 890 cm⁻¹ mainly from saturated hydrocarbons caused by the thermal decomposition of polyolefins [39].

The infrared peak intensities reached their maxima in the 451–471 °C range, signifying extensive decomposition during this stage. Fig.2(g) reveals that OPBI began decomposing around 600 °C. The peaks at 3330 and 964 cm⁻¹ correspond to the N–H bond stretching and bending vibration, indicating that the decomposition of OPBI produced NH₃. The peaks at 712 and 2180 cm⁻¹ correspond to the H−C≡N out-of-plane bending vibration and C≡N stretching vibration, respectively, indicating that the decomposition of OPBI may have produced HCN or cyanide-containing small molecules. The peak at 3016 cm⁻¹ corresponds to the C−H asymmetric stretching vibration, which may be attributed to the aromatic hydrocarbons produced by the decomposition of OPBI [24].

As shown in Fig.2(h), the P@HLi-20 separator also decomposes at about 600 °C, and the small molecules produced by the decomposition of P@HLi-20 separator are mainly NH₃, HCN or small molecules containing cyanide, aromatic hydrocarbons, etc., as evidenced by the characteristic peaks of infrared absorption. Importantly, HNTs maintains its structural integrity throughout due to its excellent thermal stability, with no obvious decomposition products formed.

These results indicate that the flame-retardant mechanism of P@HLi-20 mainly arises from both condensed-phase and gas-phase flame retardancy. In the condensed phase, OPBI and HNTs can synergistically form a dense carbon layer barrier that inhibit heat and oxygen transfer during the combustion process. In the gas phase, nitrogen-containing gases such as NH₃ released from the decomposition of OPBI can inhibit free radical formation and reduce the concentration of flammable gases, thus effectively quenching the combustion.

2.3 Physical parameters of the separator and DFT calculations

Contact angle measurements were employed to evaluate the electrolyte affinity of separators. As shown in Fig.3(a)–Fig.3(c), the initial contact angles for Celgard, OPBI, and P@HLi-20 were 68.1°, 45.9°, and 29.1° respectively. Over time, the contact angle of Celgard remained nearly unchanged (Fig. S5), indicating its poor electrolyte affinity. In contrast, the contact angle of the OPBI separator decreased to 28.3° within 60 s, while that of P@HLi-20 rapidly diminished to nearly 0° within just 2 s. These results demonstrate that increasing HNTs-Li content significantly enhances the electrolyte affinity, as also reflected by the contact angle (32.5°) of P@HLi-10. These findings suggest that the strong polar N–H groups in the imidazole rings of the OPBI backbone, the tubular structure and surface-abundant hydroxyl groups of HNTs, along with the porous architecture constructed via the NIPS method, synergistically contribute to improving the electrolyte affinity of the separators [40].

The porosity of separators directly influences their electrolyte infiltration efficiency. As demonstrated in Fig.3(d) and Table S1, the Celgard separator exhibits a porosity of 49.20%, whereas the OPBI separator shows significantly higher porosity (67.42%) due to the pore-forming nature of the NIPS fabrication method [41]. Notably, the P@HLi-20 separator achieves an exceptional porosity of 90.9%, attributed to the incorporation of HNTs-Li. This enhancement arises from two synergistic effects: (1) the inherent tubular architecture of HNTs introduces additional void spaces, and (2) the homogeneous dispersion of HNTs within the OPBI matrix reduces intermolecular interaction forces, thereby mitigating polymer chain contraction and facilitating the formation of a loosely packed porous network.

Electrolyte uptake capacities of different separators are compared in Fig.3(e). The Celgard separator displays limited uptake (73.45%), primarily owing to its intrinsic hydrophobicity and absence of electrolyte-philic functional groups. In stark contrast, both OPBI (335.94%) and P@HLi-20 (550.32%) demonstrates remarkable electrolyte absorption. This superior performance originates from multiple mechanisms: (1) abundant polar groups in OPBI establish hydrogen bonding and electrostatic interactions with electrolyte solvent molecules, (2) the microporous structure of HNTs-Li creates supplementary electrolyte reservoirs, and (3) HNTs-Li reinforcement concurrently improves the mechanical robustness and structural integrity of the composite, ensuring sustained electrolyte retention capability.

To investigate the interaction between the separator and the electrolyte, the Electrostatic Surface Potential (ESP) distribution of the separator and solvent molecules was calculated using DFT, as shown in Fig.3(g). It can be observed that both OPBI and HNTs exhibit negatively charged polar groups, capable of strong interactions with solvent molecules such as EC (ethylene carbonate) and DMC (dimethyl carbonate). In contrast, PP (polypropylene) lacks such groups, resulting in weaker interactions with the solvent molecules. These findings also validate the superior electrolyte affinity and wettability of P@HLi. The high porosity and excellent electrolyte absorption rate of P@HLi-20 are primarily due to the strong dipole-dipole interactions between the polar groups in OPBI and HNTs and the carbonate-based electrolytes. Furthermore, the tubular structure of HNTs provides high porosity, while the mesoporous structure offers significant capillary forces that help maintain liquid phase retention. Additionally, HNTs enhance the dimensional stability of the separator, preventing structural collapse during the electrolyte saturation.

The ionic conductivity is directly related to the ionic conductivity of the separator in the electrolyte. The Nyquist impedance spectrum (Fig. S6) test revealed that the impedance value of the separator showed a trend of decreasing and then increasing with the increase of HNTs-Li content. The impedance value of the Celgard separator was 7.48 Ω, and the impedance values of the OPBI separator, P@HLi-5 (the additions of HNTs-Li to the OPBI separator at a percent mass ratio of 5%), P@HLi-10, and P@HLi-20 separators were gradually decreasing, which were 2.98, 2.23, 1.84, and 0.932 Ω, respectively, while the impedance value of P@HLi-30 (the additions of HNTs-Li to the OPBI separator at a percent mass ratio of 30%) separator was 3.73 Ω. This indicates that the low additive level has limited enhancement of the conductivity of the separator, whereas excessive loading destroys the three-dimensional structure of the separator and reduces the conductivity. The calculated ionic conductivities (Fig.3 (f)) of the Celgard, the OPBI, the P@HLi-10, and the P@HLi-20 separator are 0.167, 0.661, 0.9456, and 1.551 mS/cm, respectively. Compared to the Celgard and OPBI separators, the ionic conductivity of the P@HLi-20 separator is significantly improved, which is primarily attributed to the addition of HNTs-Li [42]. The hollow tubular structure of HNTs serves as a rapid transport channel for lithium ions, and the lithium-induced active sites further increase the concentration of mobile lithium ions in the separator, providing more channels for ion conduction [43].

The mechanical properties of the separator are crucial for battery safety. The mechanical properties of different separators are shown in Fig. S7. The biaxially oriented Celgard separator exhibits a tensile strength of 7.6 MPa, owing to its crystalline polypropylene microstructure. The rigid aromatic backbone of OPBI provides higher mechanical strength at 12.1 MPa, while the introduction of HNTs-Li enhances the ductility of the separator to some extent. Specifically, P@HLi-5 maintains excellent mechanical properties at 10.68 MPa, while P@HLi-10 and P@HLi-20 retain an appropriate tensile strength around 8.5 ± 0.3 MPa, meeting the operational requirements of LIBs. However, excessive addition of HNTs-Li (P@HLi-30) leads to a sharp decrease in strength to 4.5 MPa, compromising the integrity of the separator. This establishes P@HLi-20 as the optimal formulation, achieving a balance between mechanical and electrochemical performance.

2.4 Electrochemical performance studies

Due to the excellent ionic conductivity of the P@HLi-20 separator, its rate performance was tested to evaluate the battery’s fast charge-discharge capability. As shown in Fig.4(a), tests were conducted at current densities of 0.1 C, 0.2 C, 0.5 C, 1 C, 2 C, 4 C, and 8 C. The corresponding discharge specific capacities of cells assembled with different separators at different multipliers are shown in Table S2. The results indicate that the discharge capacity of the battery with the P@HLi-20 separator is significantly higher than that with Celgard, demonstrating its excellent rate performance.

Fig.4(b) and Fig.4(c), and S8 show the voltage-discharge capacity profiles of the batteries with different separators. As expected, discharge capacity decreases with increasing current density. Notably, the cell using Celgard has a faster capacity decay and a larger charge‒discharge voltage difference compared to those using OPBI and P@HLi-20. In contrast, cells with OPBI and P@HLi-20 maintain higher capacities and smaller differential pressure, further confirming the excellent multiplier performance of the P@HLi-20 separator. This is mainly due to the excellent electrolyte affinity and high ionic conductivity of the P@HLi-20 separator, which allows ions to migrate faster within the cell, resulting in higher charge‒discharge rates and reduced concentration polarization, thereby improving rate performance.

Li symmetric cell tests, shown in Fig.4(d) and S9 further support these findings. At an area capacity of 1 mAh/cm², the battery with the P@HLi-20 separator has a much lower polarization potential compared to the battery with the Celgard separator. After approximately 250 cycles, the cell employing the Celgard separator short-circuited, primarily due to its non-uniform pore structure and poor mechanical integrity. In contrast, the cell with the P@HLi-20 separator cycled stably for over 600 h at a lower polarization potential, demonstrating its excellent cycling performance. This result suggests that the high ionic conductivity, excellent electrolyte affinity, and uniform high-porosity structure of the P@HLi-20 separator contribute to reducing the internal resistance of the battery, alleviating concentration polarization, and minimizing the formation of “dead lithium” and lithium dendrites [44].

The electrochemical stability of the separators was assessed via linear sweep voltammetry (LSV) as depicted in Fig.4(g). The OPBI separator demonstrated exceptional electrochemical stability due to the robust electrochemical stability and oxidation resistance of its benzimidazole units. Both OPBI and P@HLi separators exhibited oxidation potentials exceeding 4.5 V, significantly higher than that of Celgard separators [45], fulfilling the requirements for applications in LFP//Li and NCM811//Li cells.

The cycling performance of LFP//Li cells with different separators is shown in Fig.4(e). Cells with Celgard separators displayed an initial discharge specific capacity of 128.3 mAh/g, while those employing OPBI and P@HLi-20 separators exhibited higher initial capacities of 142.8 and 142.0 mAh/g, respectively, indicating a notable improvement. After 300 cycles, the P@HLi-20-based cell retained a discharge capacity of 125.9 mAh/g, significantly surpassing the Celgard-based cell.

As shown in Fig.4(f), for NCM//Li cells, the Celgard separator delivered an initial discharge capacity of 160.7 mAh/g, whereas the P@HLi-20 separator achieved 167.9 mAh/g. After 200 cycles, the P@HLi-20-based cell retained a discharge capacity of 152.4 mAh/g, significantly higher than that of the Celgard-based cell. The P@HLi-20 separator demonstrated enhanced discharge capacity and robust cycling stability, thereby exhibiting exceptional cycle performance.

Thanks to the exceptional thermal stability and anti-shrinkage properties of OPBI, the P@HLi-20 separator demonstrates promising potential as a flame-retardant separator for applications in extreme high-temperature environments [46]. As illustrated in Fig.4(h) and Fig.4(i), the P@HLi-20 separator and Celgard separator were tested in LFP//Li batteries under 1 C cycling at 60 and 90 °C for 100 cycles to evaluate their high-temperature cycling stability. At 60 °C, the initial discharge specific capacity of the Celgard-equipped cell was 103.8 mAh/g at 1 C, whereas the P@HLi-20-equipped cell achieved a significantly higher initial capacity of 133.2 mAh/g under identical conditions. Similarly, at 90 °C, the Celgard-based cell suffered rapid capacity decay after a few cycles, while the P@HLi-20-based cell maintained a discharge specific capacity of 131.1 mAh/g with 98.0% capacity retention after 100 cycles. The superior high-temperature cycling performance of the P@HLi-20-equipped cell, attributed to its exceptional thermal stability, underscores the potential of P@HLi-20 as a next-generation polymer composite separator for LIBs operating under extreme thermal conditions.

2.5 Mechanism of lithium dendrite suppression by P@HLi separator

As shown in Fig.5(a) and Fig.5(c), the surface morphology of lithium metal after cycling 100 h at an area capacity of 1 mAh/cm² in Li//Li symmetric cells with different separators was examined. The battery using the Celgard separator showed significant lithium dendrite formation on the lithium metal surface after cycling, with prominent cracks. In contrast, the cell using the OPBI separator exhibited a more loosely distributed lithium anode, with some larger but less prominent lithium dendrites in certain regions. The battery using the P@HLi-20 separator displayed a uniformly distributed and densely packed lithium deposition on the surface of the lithium metal, with no apparent lithium dendrites formed. Similarly, as shown in Fig. S10, the lithium metal surface of the battery with the Celgard separator exhibited a large amount of black deposits, which corresponds to irreversible “dead lithium” and lithium dendrites. In contrast, the lithium metal surface of the battery with the P@HLi-20 separator was relatively smooth, with only a small amount of irreversible “dead lithium” present. This indicates that the P@HLi-20 separator promotes the stable formation of the SEI layer on the lithium metal surface, effectively suppressing the growth of lithium dendrites [47].

Additionally, as shown in Fig. S11, SEM analysis of the separators from Li//Li symmetric cells after cycling 100 h at an area capacity of 1 mAh/cm² was performed. The results show that the surface of the Celgard separator was severely covered by lithium metal, with uneven lithium dendrites formed, which could easily lead to battery short-circuiting. The OPBI separator exhibited aggregated lithium dendrites attached to its surface, while the P@HLi-20 separator displayed only a small amount of lithium metal adhered to the surface, in a flat configuration, with no significant lithium dendrites formed.

X-ray photoelectron spectroscopy (XPS) analysis of the anode from Li//Li symmetric cells with different separators after cycling 100 h at an area capacity of 1 mAh/cm² was conducted, as presented in Fig.5(d)–Fig.5(i). Comparative analysis of the C 1s, F 1s, and Li 1s spectra revealed that for the battery using the P@HLi-20 separator, the content of Li₂CO₃ increased, the content of C=O species decreased, the Li–F content increased, and the Li–O content decreased. This indicates that the SEI layer formed in the battery with the P@HLi-20 separator is more stable, which facilitates Li⁺ transport, thereby suppressing lithium dendrite growth and enhancing the cycling stability of the separator.

Electrochemical impedance spectroscopy (EIS) of the Li//Li symmetric cells with different separators before and after cycling was tested, as shown in Fig.5(j). The interfacial resistance and charge transfer resistance of the battery with the P@HLi-20 separator were significantly lower compared to the battery with the Celgard separator. Moreover, after 200 h of cycling, the increase in battery impedance for the P@HLi-20 separator remained minimal. This suggests that the SEI formed on the lithium metal surface of the Celgard-based battery is unstable and cannot support long-term cycling of the battery.

Based on the above, it can be concluded that the mechanism by which the P@HLi-20 separator suppresses lithium dendrite growth lies in its homogeneous porous structure, excellent electrolyte affinity, and high ionic conductivity. These properties facilitate uniform lithium deposition and inhibit the growth of lithium dendrites. Moreover, the enhanced mechanical strength of the P@HLi-20 separator inhibits further dendrite growth, preventing separator penetration and short-circuiting. More importantly, the P@HLi-20 separator promotes the formation of a stable SEI layer, thereby improving cycling stability [44].

2.6 Fire safety analysis of LFP//Gr pouch batteries

To evaluate battery fire safety, different separators were assembled in pouch-type batteries, which were charged to 100% state of charge (SOC). After charging, the batteries were heated using a heating plate. The thermal runaway behavior of the pouch cells was analyzed using techniques such as cone calorimetry, infrared thermography, and digital cameras to assess the fire safety of the battery. As shown in Fig.6(a), a blank heating experiment was conducted on the aluminum plastic film, and it can be seen that the temperature of the aluminum plastic film increased rapidly under the heating plate, and intense combustion occurred after approximately 180 s. It reached a maximum temperature of 528 °C at about 185 s, and completely consumed by 210 s.

As shown in Fig.6(b) and 6(c), thermal runaway tests were conducted using pouch cells of Celgard and P@HLi-20, respectively. The pouch battery using Celgard experienced significant swelling and emitted a large amount of white smoke by 81 s, which was caused by the evaporation of the low-boiling electrolyte solvent, which propped up the soft pack as the pressure increased. Subsequently, combustion was initiated at about 180 s, with the temperature continued to rise, peaking at 631 °C at 200 s, and burn out completely at 264 s. The pouch cell using P@HLi-20 was heated up to 110 s before significant swelling and constant outgassing occurred, followed by ignition and combustion of the cell at about 180 s, reaching a maximum temperature of 528 °C at 185 s and extinguishing at 240 s. In comparison, the maximum temperature and time of combustion of the pouch battery with P@HLi-20 were significantly lower than those of the pouch battery with Celgard, while the time of thermal runaway was the same, mainly because the aluminum lamination of the soft pack burned when it was heated up for about 180 s, making the battery undergo thermal runaway.

Fig.6(d)–6(f) present cone calorimetry data, showing the soft pack with Celgard separator reaches a PHRR of 287.84 kW/m² at 215 s, while the soft pack with P@HLi-20 had a markedly lower PHRR of 136.23 kW/m² at 188 s, a significant reduction of 52.67%. Similarly, the THR of the soft pack with P@HLi-20 was 3.59 MJ/m², a 68.42% reduction compared to the 11.37 MJ/m² observed for the soft pack with the Celgard separator.

As shown in Fig.6(g), the autopsy of the batteries after thermal runaway reveals that the P@HLi-20 separator still retained its structural integrity, while the Celgard was obviously damaged, which indicates that the P@HLi-20 separator has excellent flame retardancy and is able to carbonize to further impede the transfer of heat when thermal runaway occurs, preventing the battery from short-circuiting and further thermal runaway.

These results demonstrate that replacing the Celgard separator with the flame-retardant P@HLi-20 separator can significantly reduce the maximum temperature, combustion time, PHRR and THR during thermal runaway. These improvements indicate that P@HLi-20 effectively suppresses heat release and flame spread, thereby substantially enhancing the fire safety of lithium-ion batteries.

3 Conclusions

In summary, this study successfully developed a multifunctional composite separator, P@HLi, with excellent thermal stability and electrochemical performance, by incorporating acidified and HNTs-Li into OPBI through a nonsolvent-induced phase separation method. The resulting composite separator improves both the electrochemical and safety performance of the battery. Specifically, by introducing 20 wt% HNTs-Li, the P@HLi separator exhibits self-extinguishing behavior upon ignition, forming a dense carbon layer during combustion. The decomposition of nitrogen gas generated in the process inhibits the formation of free radicals, thus suppressing combustion. DFT calculations show that the strong interaction between the P@HLi separator and solvent molecules enhances the separator’s electrolyte affinity. The lithiation of halloysite nanotubes introduces active lithium sites, significantly improving the separator’s ionic conductivity to 1.551 mS/cm. Additionally, the P@HLi separator demonstrates excellent rate capability and cycling performance. Electrochemical tests show that batteries with P@HLi-20 separators exhibit initial discharge capacities of 142.0 mAh/g for LFP//Li and 167.9 mAh/g for NCM811//Li, with their Li//Li batteries stably cycling stably for over 600 h at a low polarization (around 50 mV). Remarkably, the P@HLi-20 separator exhibits excellent high temperature cycling performance even in high temperature environments due to their excellent thermal stability. Batteries with P@HLi-20 separators not only cycle stably at 60 °C, but also maintain 98.0% capacity after 100 cycles at 90 °C. Furthermore, pouch cells with P@HLi-20 separators show a reduction of 52.67% in PHHR and 68.42% in THR compared to Celgard separators, significantly enhancing the fire safety of the batteries. Therefore, this work provides a simple method for fabricating high ionic conductivity intrinsically flame-retardant polymer separators that not only exhibit excellent flame-retardant properties but also improve the electrochemical performance, providing valuable insights and guidance for the research and development of advanced polymer-based intrinsically flame-retardant separators.

References

[1] Zhang C, Yang Y, Liu X. . Mobile energy storage technologies for boosting carbon neutrality. Innovation, 2023, 4(6): 100518

[2] Held M, Tuchschmid M, Zennegg M. . Thermal runaway and fire of electric vehicle lithium-ion battery and contamination of infrastructure facility. Renewable & Sustainable Energy Reviews, 2022, 165: 112474

[3] Liu B, Zhang J G, Xu W. Advancing lithium metal batteries. Joule, 2018, 2(5): 833–845

[4] Armand M, Tarascon J M. Building better batteries. Nature, 2008, 451(7179): 652–657

[5] Orendorff C J, Lambert T N, Chavez C A. . Polyester separators for lithium-ion cells: Improving thermal stability and abuse tolerance. Advanced Energy Materials, 2013, 3(3): 314–320

[6] Heidari A A, Mahdavi H. Recent development of polyolefin-based microporous separators for li-ion batteries: A review. Chemical Record, 2020, 20(6): 570–595

[7] Dai X, Zhang X, Wen J. . Research progress on high-temperature resistant polymer separators for lithium-ion batteries. Energy Storage Materials, 2022, 51: 638–659

[8] Chen X, Zhang R, Zhao R. . A “dendrite-eating” separator for high-areal-capacity lithium-metal batteries. Energy Storage Materials, 2020, 31: 181–186

[9] Lee H, Yanilmaz M, Toprakci O. . A review of recent developments in membrane separators for rechargeable lithium-ion batteries. Energy & Environmental Science, 2014, 7(12): 3857–3886

[10] Cheng X B, Zhang R, Zhao C Z. . Toward safe lithium metal anode in rechargeable batteries: A review. Chemical Reviews, 2017, 117(15): 10403–10473

[11] Su M, Huang G, Wang S. . High safety separators for rechargeable lithium batteries. Science China. Chemistry, 2021, 64(7): 1131–1156

[12] Tang W, Zhao T, Wang K. . Dendrite-free lithium metal batteries enabled by coordination chemistry in polymer-ceramic modified separators. Advanced Functional Materials, 2024, 34(18): 2314045

[13] Rahman M M, Mateti S, Cai Q. . High temperature and high rate lithium-ion batteries with boron nitride nanotubes coated polypropylene separators. Energy Storage Materials, 2019, 19: 352–359

[14] Liao C, Wang W, Han L F. . A flame retardant sandwiched separator coated with ammonium polyphosphate wrapped by SiO₂ on commercial polyolefin for high performance safety lithium metal batteries. Applied Materials Today, 2020, 21: 100793

[15] Zhou K, Yan H, Tang Q. . Chemically cross-linked poly(vinyl alcohol) and halloysite nanotubes as composite separator for stable Zn-organic battery. Journal of Power Sources, 2024, 591: 233854

[16] Liao C, Li W, Han L. . Microcapsule modification strategy empowering separator multifunctionality to enhance safety of lithium-metal batteries. Small, 2024, 20(43): 2404470

[17] Deng Y, Hussain A, Raza W. . Morphological modulation of the PBI membrane and performance optimization for Li-metal battery. Chemical Engineering Journal, 2023, 474: 145800

[18] Zhang Q, Chen L, Li X. . Robust, high-temperature-resistant polyimide separators with vertically aligned uniform nanochannels for high-performance lithium-ion batteries. ACS Nano, 2024, 18(46): 32162–32174

[19] Zeng X, Chen Y, Nie H. . Advanced poly (ether ether ketone) separator for lithium metal battery. Small, 2025, 21(13): 2411626

[20] Zhou H, Yu C, Gao H. . Polyphenylene sulfide-based solid-state separator for limited Li metal battery. Small, 2021, 17(51): 2104365

[21] Ren J, Gao L, He H. . Heat-resistant PMIA separator with highly interconnected pore structure for thermally stable and high energy lithium-ion batteries. Journal of Energy Chemistry, 2025, 104: 716–725

[22] Sun G, Cui J, Zhang Q. . Polybenzimidazolium-reinforced polyimide separators to inhibit dendrites for high-security lithium-ion batteries. Science China. Chemistry, 2025, 68: 1869–1870

[23] Yu Y, Jia G, Zhao L. . Flexible and heat-resistant polyphenylene sulfide ultrafine fiber hybrid separators for high-safety lithium-ion batteries. Chemical Engineering Journal, 2023, 452: 139112

[24] Li D, Shi D, Xia Y. . Superior thermally stable and nonflammable porous polybenzimidazole membrane with high wettability for high-power lithium-ion batteries. ACS Applied Materials & Interfaces, 2017, 9(10): 8742–8750

[25] Zhang G, Xian D, Hussain S. . Enhancing lithium metal battery safety and performance: a thermally stable poly(aryl ether benzimidazole) separator with 2D-functionalized boron nitride for 3000 hours lithium plating/stripping. Journal of Materials Chemistry. A, Materials for Energy and Sustainability, 2025, 13(6): 4486–4495

[26] Min Y, Guo L, Wei G. . Enhancing the safety and cyclic performance of lithium-ion batteries using heat resistant and wettable separator based on covalent organic framework and polybenzimidazole. Chemical Engineering Journal, 2022, 443: 136480

[27] Liu X, Zhang B, Wu Y. . The effects of polybenzimidazole nanofiber separator on the safety and performance of lithium-ion batteries: Characterization and analysis from the perspective of mechanism. Journal of Power Sources, 2020, 475: 228624

[28] Sengel S B, Sahiner M, Aktas N. . Halloysite-carboxymethyl cellulose cryogel composite from natural sources. Applied Clay Science, 2017, 140: 66–74

[29] Liu S, Han Q, He C. . Ion-sieving separator functionalized by natural mineral coating toward ultrastable Zn metal anodes. ACS Nano, 2024, 18(37): 25880–25892

[30] Wang W, Yuen A C Y, Yuan Y. . Nano architectured halloysite nanotubes enable advanced composite separator for safe lithium metal batteries. Chemical Engineering Journal, 2023, 451: 138496

[31] Zhang Y, Xiao Y, Zou L. . Halloysite nanotube coated-separator enhances the safety and electrochemical performance of lithium-ion battery. Journal of Power Sources, 2025, 630: 236155

[32] Min Y, Liu X, Guo L. . Construction of diversified ion channels in lithium-ion battery separator using polybenzimidazole and ion-modified metal-organic framework. ACS Applied Energy Materials, 2022, 5(7): 9131–9140

[33] Hussain S, Hussain M B, Zhou Q. . Polypyrrole as the MOFs/polymer interfacial binder applied in mixed matrix porous separator for high temperature safe lithium-ion batteries. Chemical Engineering Journal, 2024, 502: 158149

[34] Duan Z, Zhao Q, Wang S. . Halloysite nanotubes: Natural, environmental-friendly and low-cost nanomaterials for high-performance humidity sensor. Sensors and Actuators. B, Chemical, 2020, 317: 128204

[35] Ni J, Hu M, Liu D. . Synthesis and properties of highly branched polybenzimidazoles as proton exchange membranes for high-temperature fuel cells. Journal of Materials Chemistry. C, Materials for Optical and Electronic Devices, 2016, 4(21): 4814–4821

[36] Guo Y, Feng B, Wang Y. . A thermally managed separator for lithium metal batteries operating safely above 100 °C. Nano Energy, 2025, 133: 110472

[37] Bi Q Q, Li Y M, He L. . Bio-derived modified halloysite nanotubes as eco-friendly flame retardants to endow epoxy with high thermal stability, mechanical performance and flame retardancy. Chemical Engineering Journal, 2024, 500: 157438

[38] Nie S, Zhang M, Yuan S. . Thermal and flame retardant properties of novel intumescent flame retardant low-density polyethylene (LDPE) composites. Journal of Thermal Analysis and Calorimetry, 2012, 109(2): 999–1004

[39] Park B K, Ahn Y K, Kwon Y K. . Developing eco-friendly ceramic composite separator with competitive electrochemical properties using water-based polymer binder for lithium-ion batteries. International Journal of Energy Research, 2022, 46(2): 1398–1406

[40] Xu P, Chen H, Zhou X. . Gel polymer electrolyte based on PVDF-HFP matrix composited with rGO-PEG-NH₂ for high-performance lithium ion battery. Journal of Membrane Science, 2021, 617: 118660

[41] Liu H, Tao R, Guo C. . Lithiated halloysite nanotube/cross-linked network polymer composite artificial solid electrolyte interface layer for high-performance lithium metal batteries. Chemical Engineering Journal, 2022, 429: 132239

[42] Yu G, Cui Y, Lin S. . Ultrathin composite separator based on lithiated COF nanosheet for high stability lithium metal batteries. Advanced Functional Materials, 2024, 34(24): 2314935

[43] Ahn Y K, Kwon Y K, Kim K J. Surface-modified polyethylene separator with hydrophilic property for enhancing the electrochemical performance of lithium-ion battery. International Journal of Energy Research, 2020, 44(8): 6651–6659

[44] Ma C, Zhang J, Xu M. . Cross-linked branching nanohybrid polymer electrolyte with monodispersed TiO₂ nanoparticles for high performance lithium-ion batteries. Journal of Power Sources, 2016, 317: 103–111

[45] Sun G, Guo J, Niu H. . The design of a multifunctional separator regulating the lithium ion flux for advanced lithium-ion batteries. RSC Advances, 2019, 9(68): 40084–40091

[46] Hao Z, Zhao Q, Tang J. . Functional separators towards the suppression of lithium dendrites for rechargeable high-energy batteries. Materials Horizons, 2021, 8(1): 12–32

[47] Tan J, Matz J, Dong P. . A growing appreciation for the role of LiF in the solid electrolyte interphase. Advanced Energy Materials, 2021, 11(16): 2100046

RIGHTS & PERMISSIONS

Higher Education Press 2025