Highly crystalline covalent triazine frameworks modified separator for lithium metal batteries

INTRODUCTION

Benefiting from the high theoretical specific capacity (3,860 mAh g^-1, ten times that of graphite) and low redox potential (-3.04 V, vs. standard hydrogen electrode) of lithium metal anodes, lithium metal batteries (LMBs) are considered as one of the most promising high-energy-density secondary batteries^[1-4]. Their main challenge is the irregular lithium metal plating during charging, forming lithium dendrites^[5-7]. Generally, uncontrolled lithium dendrite growth leads to the corrosion of metallic lithium and the consumption of liquid electrolyte, resulting in a low Coulombic efficiency (CE) and poor cycling performance. Further, the lithium dendrite can penetrate the separator, causing a short circuit and explosion of the battery^[8,9]. Its growth is rooted in the slow mass transport of lithium ions (Li⁺). When the diffusion rate of Li⁺ in the electrolyte is slower than that consumed during the electrochemical deposition process, concentration gradients are established, and lithium dendrites tend to grow^[10]. Thus, accelerating Li⁺ transport in the bulk electrolyte is the key to relaxing the concentration gradients and thus achieving homogeneous lithium deposition^[11]. Unfortunately, Li⁺ are usually surrounded by solvent molecules due to their small radius. The bulky solvated Li⁺ have reduced mobility; meanwhile, they always transport more slowly than the anions in electrolytes, resulting in a low Li⁺ transference number ($$ {t_{L i^{+}} } $$) of 0.2-0.4^[12]. Separators are a key component of LMBs. Their surface modification to regulate the transport behavior of ions in electrolytes proves to be an effective strategy to suppress lithium dendrite growth^[13-16].

Among the various coating materials, such as inorganic nanoparticles^[13,17,18] and polymers with ionic groups^[19-21], covalent organic frameworks (COFs) are emerging porous materials with tunable nanoscale pores and tailorable skeletons. They usually contain light elements such as C, H, O, N, and B. Thanks to the nanosized pores and polar functional groups on the pore walls, COFs are expected to greatly promote the selective transport of Li⁺ by size-confinement effect and electrostatic interactions^[22-24]. In addition, their intrinsic insolubility in electrolyte makes them potential candidates for surface modification of separators^[25-27]. However, due to the rapid reaction between monomers and the limited reversibility of polymerization reactions, amorphous or poorly crystalline COFs are usually generated. These amorphous COFs would lead to ion pathways with higher tortuosity due to random stacking, thereby resulting in slower ion mobility and lower diffusion flux^[28,29]. In addition, partial of the active sites (polar functional groups) are buried due to the limited surface area. Therefore, constructing COFs with good Li⁺ transport ability remains a huge challenge^[30]. To solve this problem, studies have been devoted to the structural engineering of COFs to accelerate the transport of Li⁺. Cao et al. propose a sulfonate-rich COF with strong electronegativity and large interlayer spacing, which exhibited $$ {t_{L i^{+}} } $$ of up to 0.88^[31]. Shi et al. fabricate two-dimensional COF nanosheets with different functionalities to improve the battery performance^[32]. Furthermore, exfoliating the bulk COF into nanosheets with few or single layers to expose more active sites can accelerate ion transport^[33].

This work develops a crystallinity engineering strategy for shortened ion diffusion pathways and more accessible active sites in the COF-coated separator, improving ionic conductivity and $$ {t_{L i^{+}} } $$. Covalent triazine frameworks (CTFs) are chosen due to their large number of ordered nanoscale channels constituted by periodically connected triazine groups and versatile synthesis methods^[34,35]. Compared to other COFs with highly reversible covalent linkages such as boroxine, boronate esters, imine and others, the CTFs constructed by the stable triazine linkages exhibit excellent chemical and thermal stability^[35]. Additionally, the rich nitrogen atoms in CTF skeletons are beneficial for improving the electrolyte wettability and desolvation of Li⁺, enhancing Li⁺ transport^[36,37]. To investigate the effect of the crystallinity of CTFs on the electrochemical performance of modified separators, amorphous (AM-CTF) and highly crystalline CTFs (HC-CTF) were synthesized and blade coated on polyethylene (PE) separators. As shown in Scheme 1, an irregular stacking structure is formed in AM-CTF, which elongates the transport pathway and increases the transport resistance of ions through the separator. Meanwhile, the HC-CTF features a more regular structure and higher surface area, exposing more accessible active sites, which is conducive to faster selective transport of Li⁺ and more uniform deposition. Compared to PE and AM-CTF-coated PE separators (AM-CTF-PE), the HC-CTF-coated PE separators (HC-CTF-PE) exhibit the highest $$ {t_{L i^{+}} } $$ of 0.6 and ionic conductivity of 0.67 mS cm^-1. Li//Li symmetric cells assembled with HC-CTF-PE separators deliver cycle stability over 300 h under a current density of 1 mA cm^-2. Capacity retention of 45.4% is achieved for LiFePO₄ (LFP)/Li cells with the HC-CTF-PE separators after 1,000 cycles at 1 C. This work opens up a new avenue to improve the performance of COF-coated separators for high-energy-density lithium batteries.

Scheme 1. Schematic illustration of the transport of ions in LMBs with AM-CTF-PE and HC-CTF-PE separators.

EXPERIMENTAL

RESULTS AND DISCUSSION

CTFs were synthesized by a condensation reaction between 1,4-phthalaldehyde and terephthalamidine dihydrochloride, named CTF-HUST-1, according to reported procedures^[38] [Figure 1A]. This one-pot polymerization was carried out at mild conditions (120 ^oC and ambient pressure at air atmosphere), which is beneficial for large-scale synthesis. By controlling the nucleation and crystal growth rates, the crystallinity of CTFs can be adjusted. When the monomers were fed in one step, the nucleation process is too fast and AM-CTFs are formed. By tuning the feeding rate of 1,4-phthalaldehyde (30 μL min^-1), slow nucleation was achieved and CTFs with high crystallinity were generated^[39]. The successful formation of the triazine frameworks was illustrated by FT-IR spectra. As shown in Figure 1B, the characteristic vibrations of the triazine units appear at 1,523 cm^-1 (C=N) and 1,367 cm^-1 (C-N) for both AM-CTF and HC-CTF, which is accordant with that in the literature^[38]. The structures of AM-CTF and HC-CTF were characterized by solid-state cross-polarization magic angle spinning carbon-13 nuclear magnetic resonance (CP-MAS ¹³C NMR), as shown in Supplementary Figure 1. The results demonstrate that there is essentially no difference between the two structures in solid-state CP-MAS ¹³C NMR spectroscopy. The peaks with chemical shifts at 128 and 138 ppm are attributed to the phenyl carbons, while the peak with chemical shift at 179 ppm is attributed to the carbon in the triazine ring. Powder X-ray diffraction (PXRD) was used to analyze the structural differences between AM-CTF and HC-CTF [Figure 1C]. The PXRD pattern of AM-CTF shows a broad peak at 25.88º, indicating no long-range crystalline order. For the HC-CTF, diffraction peaks at about 7.4º, 12.7º, 14.7º and 26.2º are observed. According to literature, these peaks can be assigned to (100), (110), (200), and (001) facets, respectively, indicating the AA stacking arrangement of crystalline HC-CTF^[39].

Figure 1. (A) Schematic synthesis of AM-CTF and HC-CTF; (B) FT-IR spectra; (C) PXRD patterns of AM-CTF and HC-CTF; (D) FE-SEM image and (E) FE-TEM image of AM-CTF; (F) FE-SEM image; (G)FE-TEM image and (H) high-magnification FE-TEM image of HC-CTF; (I) FFT image from the area of interest in (G).

To further study the CTF structures, FE-SEM, TEM, and FE-TEM were exploited to show their morphology and crystallinity. From FE-SEM, particles with no regular morphology are observed for AM-CTF [Figure 1D]. Meanwhile, the FE-TEM image of this sample shows no lattice fringes [Figure 1E]. Interestingly, as shown in Figure 1F and Supplementary Figure 2, particles with a much smoother surface and a more regular shape are observed for HC-CTF, compared to AM-CTF. FE-TEM images reveal their crystalline characteristic and clear lattice fringes with interplanar lattice spacing of 0.42 nm [Figure 1G and H]. The fast Fourier transform (FFT) pattern [Figure 1I] in Figure 1G illustrates the crystalline structure of the HC-CTF. The results show that CTFs with different crystallinity were successfully synthesized by tuning the feeding rate of reactants.

To evaluate the porosity of AM-CTF and HC-CTF, nitrogen adsorption-desorption isotherms were performed at 77 K [Figure 2A]. The isotherm sorption profile of HC-CTF shows the typical type I characteristics, suggesting the existence of a large number of micropores. Through the nonlocal density functional theory method, the pore size distribution of HC-CTF is close to the reported calculated theoretical values^[39] [Figure 2B]. The surface area calculated by the BET model is 652.18 m² g^-1. However, the BET surface area of AM-CTF drops sharply to 55.01 m² g^-1. The regular and interconnected channels with pores can facilitate a large accessible surface area for nitrogen in the BET measurement. And this morphology is suitable for diffusion of ions. Meanwhile, insignificant microporous structures are observed, suggesting structural defects and a rough texture for AM-CTF [Figure 2B]. The TGA measurement [Figure 2C] reveals that AM-CTF and HC-CTF display outstanding thermal stability under nitrogen atmosphere, and up to 80% of their weight is retained at temperatures of up to 650 °C, implying the formation of a highly thermal stable triazine framework. Notably, AM-CTF shows a relatively slight weight loss around 250 °C, possibly due to the presence of some unreacted terminal aldehyde groups.

Figure 2. (A) N₂ adsorption and desorption isotherms; (B) Pore size distributions and (C) TGA curves of AM-CTF and HC-CTF.

The AM-CTF and HC-CTF were exploited as building blocks for functional layers on commercial PE separators. CTF-modified separators were prepared using the blade coating [Supplementary Figure 3]. This versatile method enables large-scale fabrication of the coated separator [Supplementary Figure 4]. As shown in Supplementary Figure 5A and B, the PE separator has porous and fibrous morphology. For the AM-CTF-modified PE separator (AM-CTF-PE) and the HC-CTF-modified separator (HC-CTF-PE), the CTF particles are uniformly distributed on the surface of the PE separator [Figure 3A and B]; meanwhile, some inevitable voids are observed on the surface of AM-CTF-PE and HC-CTF-PE separators [Supplementary Figure 5C and D]. EDS mapping of N elements shows a uniform distribution of CTFs (inset of Figure 3A and B). Using the same slurry concentration and coating procedures (the distance from the blade to PE surface is 5 μm for the spreading of slurry), the AM-CTF-PE and HC-CTF-PE with a coating layer thickness of about 1.5 μm are obtained, as shown from the cross-sectional SEM images [Figure 3C and D]. In addition, except the vibration peaks of -CH₂- (2,915 and 2,846 cm^-1) for PE, the characteristic vibration peaks of the triazine units (1,523 and 1,367 cm^-1) can be observed in the FT-IR spectra for AM-CTF-PE and HC-CTF-PE separators [Supplementary Figure 6]. These results further confirm the coating of CTF on the PE separator surface. The photographs in Figure 3E exhibit that AM-CTF-PE and HC-CTF-PE separators are flexible enough to maintain their original shape under varying bending conditions. Besides, AM-CTF and HC-CTF layers are firmly attached to the PE separator through blade coating, as evidenced by no stripped and cracked yellow CTF-coated layers after repeated folding and unfolding [Figure 3F]. This property is extremely important for the cycling stability and safety of LMBs.

Figure 3. Top-view SEM images of (A) AM-CTF-PE and (B) HC-CTF-PE separators. Inset is the EDS mapping of N elements; Cross-sectional SEM images of (C) AM-CTF-PE and (D) HC-CTF-PE separators; (E)Photographs of AM-CTF-PE and HC-CTF-PE separators under bending conditions; (F) Photographs of PE, AM-CTF-PE and HC-CTF-PE separators after folding; (G) Contact angle images of PE, AM-CTF-PE and HC-CTF-PE after dropping liquid electrolyte on the surface.

Similar to the CTF samples, the BET surface area, pore volume and porosity of the HC-CTF-PE separators are higher than those of the AM-CTF-PE separator [Supplementary Table 1]. The electrolyte wettability of AM-CTF-PE and HC-CTF-PE separators is obviously improved compared with bare PE separators, with contact angles of 46.7°, 41.2° and 65.3°, respectively [Figure 3G]. This improved property of modified separators is mainly due to the CTF layer with nanosized pores and polar skeleton, which feature intrinsic affinity with polar electrolyte. Compared with AM-CTF-PE, the enhanced wettability of HC-CTF-PE separators originates from their larger specific surface area and pore volume. Meanwhile, the EU and ER of the HC-CTF-PE separator are higher than those of the AM-CTF-PE separator [Figure 4A and B]. The well-connected porous structure of HC-CTF can facilitate the penetration and retention of electrolyte, which are beneficial for fast ion transport.

Figure 4. (A) Electrolyte uptake and (B) electrolyte retention of the PE, AM-CTF-PE and HC-CTF-PE separators; (C) EIS plots of the PE, AM-CTF-PE and HC-CTF-PE separators, the inset shows enlarged EIS plots in the high-frequency region; Chronoamperometry profiles of (D) AM-CTF-PE and (E) HC-CTF-PE separators. Inset of (D) is the AC impedance spectra before polarization, and inset of (E) is the AC impedance spectra after polarization; (F) Fitting and evaluation of activation energy using the Nyquist plots.

The electrochemical stability of separators is the basis for the battery safety. As shown in Supplementary Figure 7, the LSV curves show that AM-CTF-PE and HC-CTF-PE separators can stably operate at a voltage range of 2.5-3.9 V, suggesting that these separators are suitable for commonly used batteries. The ion transport behaviors of the CTF-modified separators were further studied using electrochemical measurements. The ionic conductivity of the separators was evaluated by EIS using assembled SS/separator/SS cells. The intersection of the EIS plot and the x-axis is extracted for the conductivity calculation of the separators [Figure 4C]. The calculated ionic conductivities of the HC-CTF-PE, AM-CTF-PE and PE separators are 0.66, 0.46 and 0.33 mS cm^-1, respectively [Supplementary Table 2]. The enhanced electrolyte wettability and larger specific surface area of the HC-CTF-PE separator may contribute to its promoted ion transport ability.

To evaluate the selective transport ability of Li⁺ for different separators, the $$ {t_{L i^{+}} } $$, representing the ratio of Li⁺ mobility to total ionic mobility, was evaluated using the Bruce-Vincent method^[40]. The chronoamperometry profiles of HC-CTF-PE and AM-CTF-PE assembled Li//Li symmetric cells are present in Figure 4D and E. And the potentiostatic polarization process was endowed from the initial to the steady state. From the EIS spectra (inset of Figure 4D and E, Supplementary Figure 8, Supplementary Table 3), the charge transfer resistance (R_ct) of the HC-CTF-PE separator is lower than that of AM-CTF-PE and PE separators both before and after polarization. Accordingly, the $$ {t_{L i^{+}} } $$ of HC-CTF-PE, AM-CTF-PE and PE separators are 0.6, 0.41 and 0.31, respectively. The schematic illustrations of ion transport process in PE, AM-CTF-PE and HC-CTF-PE separators are shown in Supplementary Figure 9. For bare PE separators, the Li⁺ in electrolyte are always in bulky solvated form, which have reduced mobility compared with that of anions^[41]. Thus, the $$ {t_{L i^{+}} } $$ of 0.31 is obtained [Supplementary Figure 8]. For CTF-coated PE separators, the abundant polar triazine units can enhance the affinity of the separator to electrolyte, and the rich polar N atoms with lone pair electrons in CTF layer may interact with the solvated Li⁺ through electrostatic interaction^[42,43], thus reducing the size of the solvated cluster and accelerate their transport^[44]. More importantly, the HC-CTF, which features nanosized pores with the highest specific surface area and pore volume, provides more active sites and ion transport channels for efficient desolvation with sufficient interaction and the mostly enhanced transport of Li⁺.

To further demonstrate the desolvation effect of CTF layers, the activation energy (E_a) of ion desolvation was evaluated using the Arrhenius equation. Based on temperature-dependent EIS measurements [Supplementary Figure 10], the semicircles of the Nyquist plot in the middle-high frequency region are denoted to the R_ct. According to Arrhenius equation, the E_a can be calculated by linear fitting of ln (R_ct^-1) vs. 1/T:

where A, R, and T represent the frequency factor, ideal gas constant, and absolute temperature, respectively. After fitting, as shown in Figure 4F, the E_a values of HC-CTF-PE and AM-CTF-PE are 59.26 and 60.71 kJ mol^-1, respectively, which are lower than that of pristine PE (63.13 kJ mol^-1), confirming the desolvation effect of CTF coating layers. Based on above results, the CTF coating improves both the $$ {t_{L i^{+}} } $$ and ionic conductivity of the PE separator. According to the Bruce-Vincent method and Sand’s time model^[45,46], it is expected that the HC-CTF separator with the highest $$ {t_{L i^{+}} } $$ and ionic conductivity can delay the lithium dendrite growth and is helpful for stable lithium deposition^[47,48].

To investigate the lithium deposition behavior of CTF-modified separators, the long-term galvanostatic discharging and charging voltage profiles of Li//Li symmetric cells were measured. Figure 5A shows the performance of these cells with PE, AM-CTF and HC-CTF separators at a current density of 1 mA cm^-2 and a fixed capacity of 1 mAh cm^-2. The overpotential of the symmetric cells assembled with PE separators gradually increases as early as 120 h and fluctuates dramatically afterward. This results from the rapid electrolyte consumption and lithium dendrite growth caused by side reactions and polarization phenomena in the cells. The overpotential of symmetric cells with CTF-coated separators is lower than that of PE separators. Moreover, the overpotential of AM-CTF-PE separators gradually increases at about 170 h, while that of HC-CTF-PE separators is stable up to 300 h. From the voltage-time curves at different cycling times [Supplementary Figure 11], it is observed that the Li//Li symmetric cells assembled with HC-CTF-PE separators have the lowest overpotential, followed by AM-CTF-PE separators, while the highest is the PE separator. A higher overpotential suggests a larger nucleation energy barrier for the deposition of metallic lithium. According to literature, it may induce the dendritic lithium growth during plating^[49]. Further, at a current density of 2 mA cm^-2 and capacity of 2 mAh cm^-2, the Li//Li symmetric cells assembled with HC-CTF-PE separators show better cycling stability and lower overpotential. [Supplementary Figure 12]. These results suggest that the CTF layer effectively delays the growth of lithium dendrite. To further study the dendrite suppressing effect of HC-CTF-PE separators, the morphology of the lithium plating onto Cu foils with various separators at a current density of 0.5 mA cm^-2 were monitored. As shown in Figure 5B-D, obviously, abundant dendritic lithium grew on the Cu foil with the PE separator [Figure 5B]. Mossy dendrites are observed on the surface of the Cu foil with the AM-CTF-PE separator [Figure 5C]. When using the HC-CTF-PE separator, a much smoother Cu surface was obtained after cycling [Figure 5D]. Thus, due to the nanosized pores and desolvation effect, the HC-CTF-PE separator can regulate the ion flux and suppress the growth of dendritic lithium. And the electrochemical performance of the separators and assembled cells, including $$ {t_{L i^{+}} } $$, ionic conductivity, and stability of lithium plating, are comparable with that in literature [Supplementary Table 4].

Figure 5. (A) Voltage-time profiles of Li//Li symmetric cells assembled with PE, AM-CTF-PE and HC-CTF-PE separators at 1 mA cm^-2 with a capacity of 1 mAh cm^-2; SEM images of Cu electrodes from Cu//Li cells assembled with (B) PE, (C) AM-CTF-PE and (D) HC-CTF-PE separators after Li plating at a current density of 0.5 mA cm^-2.

To further prove the prominent advantages of the HC-CTF-PE separator, the electrochemical performance of LFP//Li cells was investigated. The stability of the battery with different separators was evaluated by cycling performance tests. As shown in Figure 6A, the capacity of PE separator-assembled cells is only 23.4 mAh g^-1 after 300 cycles at 1 C with the LFP loading of 1.58 mg cm^-2, which is considered as the poor cycling stability with a capacity retention rate of 16.8% of the first cycle capacity. Notably, the cells assembled with AM-CTF-PE and HC-CTF-PE separators exhibit excellent cycling stability, whose capacity retentions after 1,000 cycles maintain 41.8% and 45.4%, respectively. Meanwhile, in Figure 6B, the 100th circle discharge/charge voltage profiles at 1 C exhibit that the HC-CTF-PE separator has the lowest overpotential (0.17 V). However, the AM-CTF-PE and PE separators demonstrate relatively larger overpotential (0.25 and 0.43 V) and strong battery polarization. This may be induced by the irreversible loss of lithium metal and electrolyte. In addition, with increasing coating layer thicknesses, the pathway of ion transport across the separators is prolonged, resulting in the deterioration of battery performance [Supplementary Figure 13]. As shown in the following FT-IR spectra, the characteristic vibrations of the triazine units at 1,523 cm^-1 (C=N) and 1,367 cm^-1 (C-N) can clearly be observed in the separators after 100 cycles at 1 C of LFP/Li cells, suggesting the chemical stability of the CTF coatings [Supplementary Figure 14A]. In addition, from the EDS mapping of N elements of separators, their uniform dispersion indicates the structural integrity of CTF coating layers after cycling [Supplementary Figure 14B-E]. The surface morphology of CTF side of the cycled separators differs from the original ones due to deposited lithium during the cycling process. Further, to meet the requirements of practical application, higher LFP loading is necessary to increase the energy density of batteries. With commercial cathodes (LFP loading of 10.52 mg cm^-2), as shown in Figure 6C, the capacity of the HC-CTF-PE separator maintains 99.5 mAh g^-1 after 120 cycles at 1 C, much higher than that of the AM-CTF-PE (85.22 mAh g^-1) and the PE (77.6 mAh g^-1).

To further prove the prominent advantages of the HC-CTF-PE separator, the electrochemical performance of LFP//Li cells was investigated. The stability of the battery with different separators was evaluated by cycling performance tests. As shown in Figure 6A, the capacity of PE separator-assembled cells is only 23.4 mAh g^-1 after 300 cycles at 1 C with the LFP loading of 1.58 mg cm^-2, which is considered as the poor cycling stability with a capacity retention rate of 16.8% of the first cycle capacity. Notably, the cells assembled with AM-CTF-PE and HC-CTF-PE separators exhibit excellent cycling stability, whose capacity retentions after 1,000 cycles maintain 41.8% and 45.4%, respectively. Meanwhile, in Figure 6B, the 100th circle discharge/charge voltage profiles at 1 C exhibit that the HC-CTF-PE separator has the lowest overpotential (0.17 V). However, the AM-CTF-PE and PE separators demonstrate relatively larger overpotential (0.25 and 0.43 V) and strong battery polarization. This may be induced by the irreversible loss of lithium metal and electrolyte. In addition, with increasing coating layer thicknesses, the pathway of ion transport across the separators is prolonged, resulting in the deterioration of battery performance [Supplementary Figure 13]. As shown in the following FT-IR spectra, the characteristic vibrations of the triazine units at 1,523 cm^-1 (C=N) and 1,367 cm^-1 (C-N) can clearly be observed in the separators after 100 cycles at 1 C of LFP/Li cells, suggesting the chemical stability of the CTF coatings [Supplementary Figure 14A]. In addition, from the EDS mapping of N elements of separators, their uniform dispersion indicates the structural integrity of CTF coating layers after cycling [Supplementary Figure 14B-E]. The surface morphology of CTF side of the cycled separators differs from the original ones due to deposited lithium during the cycling process. Further, to meet the requirements of practical application, higher LFP loading is necessary to increase the energy density of batteries. With commercial cathodes (LFP loading of 10.52 mg cm^-2), as shown in Figure 6C, the capacity of the HC-CTF-PE separator maintains 99.5 mAh g^-1 after 120 cycles at 1 C, much higher than that of the AM-CTF-PE (85.22 mAh g^-1) and the PE (77.6 mAh g^-1).

Figure 6. (A) Long-term cycling stability of LFP//Li cells at 1 C with the LiFePO₄ loading of 1.58 mg cm^-2; (B) Charge-discharge curves at 1 C for 100th revolution with the LiFePO₄ loading of 1.58 mg cm^-2; (C) Long-term cycling stability at 1 C with an increased LiFePO₄ loading of 10.52 mg cm^-2; (D) Cycling performance at 3 C with the LiFePO₄ loading of 1.58 mg cm^-2; (E) Rate performance of PE, AM-CTF-PE and HC-CTF-PE separators assembled cells at various current densities of 0.5-5 C; (F) Charge-discharge curves of HC-CTF-PE separator assembled cells at various current densities of 0.5-5 C; (G) The potential gaps of PE, AM-CTF-PE and HC-CTF-PE separators assembled cells at various rates, respectively.

The CE is further studied to evaluate the cell performance during repeated cycling. As shown in Figure 6D, the CEs of both HC-CTF-PE and AM-CTF-PE separators are around 95% over 300 cycles at 3 C. The HC-CTF-PE also has the highest capacity retention (49.5% after 300 cycles). However, the CE of PE separators is consistently unstable, with a low value of 90%. To clearly illustrate this trend, the enlarged graph of the CE for the first 100 cycles further shows the superior cycling stability of the HC-CTF separator [Supplementary Figure 15]. The lower and unstable CE of the pristine PE separator at 3 C suggests the Li⁺ loss during each cycle. It is worth noting that the CE of the HC-CTF-PE separator-assembled cells slightly decreases after cycling. The decrease of the CE at the later stage of the battery may be due to soft short circuit inside the battery^[50].

The rate performance of cells assembled with different separators was evaluated at current rates ranging from 0.5 to 5 C [Figure 6E]. The HC-CTF-PE separator can deliver the average specific capacity of 146, 136, 118, 102, 85 and 69 mAh g^-1 at 0.5, 1, 2, 3, 4 and 5 C, respectively. By contrast, the AM-CTF-PE and PE separators show much lower capacities of 56 and 28 mAh g^-1 at 5 C. It suggests a faster Li⁺ transport kinetic in the HC-CTF-PE separator, especially at the high rate. The charge/discharge voltage curves of the corresponding HC-CTF-PE, AM-CTF-PE, and PE separators in Figure 6F, Supplementary Figures 16 and 17 clearly show that the CTF-coated separators have a clear discharge plateau, while the PE separator discharge platform contracts obviously. And the discharge capacity is significantly reduced due to the limited Li⁺ transport rate of the PE separator. In addition, the HC-CTF-PE separator has the lowest overpotential at various charge and discharge rates [Figure 6G], indicating the weakest polarization and fastest Li⁺ transport^[51].

CONCLUSIONS

In summary, The AM-CTF and HC-CTF were successfully synthesized and coated on the surface of commercial PE separators. It can be found that compared with AM-CTF, the well-connected porous structure of HC-CTF shortens ion diffusion pathways, and a larger specific surface area provides more accessible active sites, enhancing ionic conductivity and $$ {t_{L i^{+}} } $$. As a result, the Li//Li symmetric cells assembled with HC-CTF-PE separators exhibit excellent cycling stability for as long as 300 h with low polarization and negligible voltage fluctuation. The LFP/Li full cells based on HC-CTF-PE separators exhibit satisfying lifespan, high-capacity retention (45.4% at 1 C for 1,000 cycles) and CE. Overall, this research offers a new strategy for improving the performance of COF-coated separators for high-energy-density lithium batteries.