Tissue paper-based composite separator using nano-SiO2 hybrid crosslinked polymer electrolyte as coating layer for lithium ion battery with superior security and cycle stability

In order to develop high power lithium ion batteries (LIBs), urgent requirements including adequate safety, higher current density and superior cyclic stability are proposed for separator. Tissue paper, composed of packed cellulose fibers, possesses lower production cost, easier accessibility, superior wettability together with outstanding thermostability, and is thus a candidate to be the substrate for high performance separator. To address the issue of structural failure usually encountered by single polymer as binder during long term cycling, crosslinked binder was constructed on tissue paper to adhere nano-SiO2 through chemical reactions between poly (vinylidene fluoride-co-hexafluoropropylene) (PVDF-HFP) and hyperbranched polyethyleneimine (PEI) in this work. The effects of crosslinking degree on physical properties and electrochemical performance were studied thoroughly. When the feed ratio of PVDF-HFP and PEI is fixed at 10:1, the crosslinked composite separator displays excellent electrolyte uptake and wettability, superior ionic conductivity, better interfacial compatibility as well as higher Li+ transference number (0.56), thus offering battery with prominent rate capabilities. Besides, this crosslinked composite separator exhibits satisfying dimensional stability even treated at 250 °C, better flame retardancy, enhanced mechanical behavior, wider electrochemical window and outstanding cycle stability. Accordingly, tissue paper-based crosslinked composite separators can meet higher requirements put forward by high power LIBs.


Introduction
Due to high energy density, low self-discharge, long cycle life, no memory effect and so on, lithium ion batteries (LIBs) have not only dominated the computer, communication, and consumer electronics (3C) industries, but also developed into energy storage area and electric vehicles in recent years. This puts forward higher requirements for the components of LIBs including higher current density, superior security as well as sufficient stability Waqas et al. 2019). Traditional polyolefin membranes have been widely used separators, which can isolate the positive and negative electrodes to prevent short circuit while allowing the transport of lithium ion to maintain normal operation of battery. However, some drawbacks cannot be ignored such as lower porosity, poorer electrolyte wettability and worse heat tolerance, which make it difficult for polyolefin membranes to meet the new higher-performance requirements (Yuan et al. 2021).
Fibrous membranes fabricated by nonwoven technology have superior porosity, and can realize their thermoresistance through the choice of materials, thus being considered as the future trend of power battery separator . So far, some types of high performance engineering plastic, such as polyphenylene sulfide , polyimide (Rodriguez et al. 2021) and aramid (Zhu et al. 2019b), have been widely employed to prepare nonwoven fabric. However, there are some shortcomings faced by these synthetic polymers including high cost, complicated manufacture process and poor wettability. Cellulose is the most abundant natural polymer, and possesses good hydrophilicity and prominent thermal stability (Huang 2014;Zhang et al. 2013). Tissue paper, as a common type of cellulose nonwoven product, can be obtained conveniently and cheaply. Accordingly, more attention has been paid to explore the potential applications of tissue paperbased separator in high power LIB (Jia et al. 2020;Wang et al. 2018).
As a result of some inherent defects such as larger pore size, weaker mechanical strength and poorer flame retardancy, tissue paper cannot be directly employed as LIB separators if appropriate modification is not performed . Single polymers such as poly(vinylidene fluoride-co-hexafluoropropylene) (PVDF-HFP) or styrene-butadiene rubber (SBR) have been widely reported as the binder to adhere nanoparticles on substrate (Luo et al. 2018;Zhang et al. 2016). However, this binder would be swollen by electrolyte during long term cycling, so that nanoparticles may gradually shed from the substrates. This would further bring about the occurrence of internal short circuit and even sudden attenuation of cell performance. In our previous study, 3D crosslinked structure has been constructed as a coating layer to decorate PPS nonwoven fabric Zhu et al. 2020). Superior cycle stability and enhanced mechanical strength were realized when compared with traditional design. This encourages us to employ crosslinked polymer as binder to fix nanoparticles on tissue paper, so that fulfilling higher demands raised by energy storage devices. Moreover, it is very worthwhile to study the relationship of its crosslinking ratios with physical structure and electrochemical performance.
In this study, nano-SiO 2 hybrid crosslinked coating layer was constructed on tissue paper through chemical reactions between PVDF-HFP and hyperbranched polyethyleneimine (PEI). Among them, nano-SiO 2 particles were utilized to improve flame resistance. Hyperbranched PEI owns amine repeating units, abundant internal cavities and ellipsoidal architecture, which can enhance electrolyte uptake through not only H-bonding but also embedding effect, and regulate electrolyte swelling behavior through serious disruption of PVDF-HFP crystallization. Moreover, PEI possesses multiple reactive peripheral groups, which are prone to form firm crosslinked network through their reaction with C-F bonds in PVDF-HFP (Zhao et al. 2009;Zhou et al. 2014). We did a series of tests related to separators including structural identification, surface morphology and pore structure, electrolyte uptake and wettability, thermal analysis and mechanical behavior as well as electrochemical and cell performance. To evaluate the performance of crosslinked composite separator, Celgard 2400 and traditional tissue paper-based composite separator using PVDF-HFP and nano-SiO 2 as a coating layer were together chosen as control groups. Especially, we changed the feed ratios between PVDF-HFP and PEI during membrane preparation, thoroughly discussed their effects on abovementioned properties and established corresponding correlations.
Preparation of crosslinked membrane PVDF-HFP and nano-SiO 2 were firstly dissolved in a solvent mixture of acetone and DMF (4/1, v/v) by ultrasound. Then, PEI was added into this solution at room temperature and stirred for 30 min, through which pre-crosslinked structure would be formed in the coating solution according to previous literatures (Zhou et al. 2014(Zhou et al. , 2015. The final ratio of PVDF-HFP: PEI: SiO 2 : solvent was set as 1: X: 0.2: 10 (X = 0, 0.05, 0.1, 0.2, 0.4). Next, the pretreated tissue paper was respectively dipped into a series of solutions as described above and the excess coating solution was removed by blade after paper fetched out. The coated membrane was dried in air and then placed in vacuum oven at 60 °C for 12 h for the evaporation of residual solvent completely along with subsequence crosslinking reactions between PVDF-HFP and PEI. These membranes were finally hot pressed under the condition of 60 °C, 10 MPa to make the surface smooth. The obtained composite separators with different feed ratios were respectively abbreviated as CS(1:0), CS(20:1), CS(10:1), CS(5:1), CS(5:2) separators. CS(1:0) separator was the traditional composite separator only using PVDF-HFP and nano-SiO 2 as coating layer and chosen as a control group in this study. In order to estimate crosslinked degree, nano-SiO 2 hybrid PVDF-HFP and 3D crosslinked polymer films were prepared firstly by similar method employed in composite separator. Among them, the former was named as F(1:0), the latter were respectively named as F(20:1), F(10:1), F(5:1) as well as F(5:2) with the ratio variation between PVDF-HFP and PEI.

Physical characterization and measurement
Shear viscosity of different coating solutions were measured by Rheometer (MARS III) with a range of shear rate from 10 -1 to 10 3 at 25 °C. Attenuated total reflection Fourier transform infrared (ATR-FTIR) spectra of composite membranes were scanned on infrared spectrometer (BRUKER TENSOR-27 TGA-IR) at a wave number ranging from 600 to 4000 cm −1 . Equal amounts of nano-SiO 2 hybrid PVDF-HFP films and 3D crosslinked polymer films were added into centrifuge tubes containing excess DMF solvent to estimate crosslinked degree. After 72 h shaking, the undissolved crosslinked fragments were obtained by centrifugation at 10,000 r min −1 for 10 min. Then, the crosslinked mass ratio of F films could be calculated according to the weights before and after dissolving. A scanning electron microscope (SEM, JEOL JSM-6510LV) was employed to examine surface morphologies of composite separators after gold sputtering treatment on samples. The pore size and distribution of separator were characterized by a membrane pore size analyzer (BSD-PB, China) according to bubble-point method. Ethanol was used as the infiltration liquid, and the test was carried out by atmospheric pressure infiltration. The porosity of separator was estimated by n-butyl alcohol immersion method and calculated using Eq. 1: where W dry and W wet are respectively the mass of sample before and after immersed in n-butyl alcohol for 1 h, ρ b is the density of n-butyl alcohol, and V p is the volume of dry sample.
Contact Angle tester (FM40 Easy Drop, KRUSS) was carried out to measure the static contact Angle of electrolyte droplet (2 μL) on the surface of the membrane. Macroscopically, the capillary absorption height was also measured after immersing one end of the sample in electrolyte about 2 h. In addition, separators were fully immersed in the electrolyte for 2 h, and after excess electrolyte was absorbed by filter paper, electrolyte uptake could be calculated according to Eq. 2: where W wet and W dry are respectively the mass before and after separator immersed in the electrolyte. Thermal dimensional stabilities of membranes were identified by heat treatment at different temperatures from 50 to 250 °C. The Limiting oxygen index (LOI) (HC-2C, China) was performed according to an international standard ASTM D2863 and the sample was cut at 40 mm in width and 100 mm in length. In addition, combustion tests were conducted to characterize flame retardant properties of membranes. Furthermore, thermogravimetric analysis (TGA, TG 209F1) was carried out at a heating rate of 10 °C/min to provide information for thermal stability. Differential scanning calorimetry (DSC, 204 F1 NETZSCH) was employed to analysis thermal properties at the temperature range of 20-300 °C with a heating rate of 10 °C/min. Mechanical behavior was tested with a universal testing machine (Instron-5967, USA) at room temperature. The strain rate of the machine was set at 10 mm/min and the standard sample was cut at 4 mm in width and 15 mm in length.

Electrochemical analysis
Nyquist plots of symmetrical cells (SS/separator/SS, SS = stainless-steel sheets) were gained by scanning electrochemical impedance spectroscopy (EIS) on an electrochemical workstation (Ivium Stat.h) with an AC voltage of 10 mV amplitude in the frequency range of 0.1 to 10 6 Hz. The ionic conductivity (δ) could be calculated with Eq. 3: where d and A respectively represent the thickness and effective testing area of separator, and R is the bulk resistance of separator and obtained from Nyquist plots.
To analyze interfacial resistance between separator and electrode, electrochemical impedance spectroscopy (EIS) of symmetrical cells (Li/separator/ Li) were measured on an electrochemical workstation (Ivium Stat.h) through the same way as above. In addition, we combined chronoamperometry and electrochemical impedance spectra (EIS) of Li/separator/Li cell to estimate lithium ion transference number according to Eq. 4: where in, I 0 and I ss are the initial and steady-state current obtained from chronoamperometry, respectively; R 0 and R ss are the initial and steady-state interfacial resistance acquired from EIS, respectively; ΔV is the step potential difference (10 mV). Moreover, SS/ separator/Li cells were assembled to evaluate electrochemical stability window of separator through linear sweep voltammetry (LSV) measurement with the potential scanned from 2.0 to 6.0 V under a rate of 10 mV/s.

Cell assembly and performance characterization
2016-type coin cells were assembled in an argon filled glove box by sandwiching liquid electrolyte-soaked separators between LiFePO 4 cathodes and lithium foil anodes. LiFePO 4 cathodes were prepared by coating NMP-based mixture of acetylene black, polyvinylidene fluoride and LiFePO 4 (1:1:8 by weight) on aluminum foils in our lab. Cell performance was tested on a LAND battery testing system (CT2001A, Wuhan LAND Electronic Co., Ltd, China) with the cut-off voltage range set from 2.5 to 4.2 V. In the tests of rate capabilities, these cells were charged at 0.2 C rate and discharged at various rates ranging from 0.2 to 2 C and then back to 0.2 C. At each rate the cells were charged/discharged for 10 times. During cycling tests, these cells were cycled at a fixed charge/discharge current density 0.5 C/0.5 C for 100 times at room temperature.

Results and discussion
Crosslinked structure fabricated on tissue paper Figure 1(a) describes the procedure of nano-SiO 2 hybrid crosslinked coating layer constructed on tissue paper. Firstly, PVDF-HFP, hyperbranched PEI and nanoparticle SiO 2 were well mixed in solvent with a series of different ratios at room temperature at which (4) t Li + = I ss ΔV − I 0 I ss R 0 I 0 ΔV − I 0 I ss R ss time pre-crosslinking structure would be produced. Then, a certain size tissue paper was dipped into the pre-crosslinking coating solution. The final composite separators with different ratios were obtained after vacuum drying at 60 °C. It should be pointed out that crosslinking reactions occurred between -NH 2 groups in PEI and VDF segments in PVDF-HFP. As reported in previous literatures (Taguet et al. 2006;Zhou et al. 2014Zhou et al. , 2015, VDF segments could take off HF molecules through autocatalytic effect of basic amines to form -CF = CH-bonds. Afterwards, due to electron withdrawing effect of F atoms in -CF = CH-groups and nucleophilicity of amines, Michael addition reactions would take place between them to create new bonds. Rheometer was employed to detect the precrosslinked structure in coating solution through the variation of viscosity. As shown in Fig. S1, compared with the initial mixed solution, the mixture of PVDF-HFP and PEI at any feed ratio displayed increased viscosity after the reaction proceeded for 30 min, demonstrating the formation of pre-crosslinked structure. Moreover, the viscosity of mixture grew higher with the increase of PEI content, implying the fabrication of more compact crosslinked network. ATR-FTIR spectra were carried out to detect the generation of C = N bonds in membrane and shown in Fig. 1(b). Though similar absorption peaks appear in the two composite separators with/without crosslinked structure, there is still a characteristic absorption peak at 1642 cm −1 distinguished for separator CS(10:1), corresponding to C = N stretching vibration. This indicates the successful occurrence of crosslinking reactions between PEI and PVDF-HFP. In addition, another two different bands centered at 1565 cm −1 and 1476 cm −1 can be discovered in CS(10:1) curve, which are respectively assigned to the N-H bending vibration of unreacted primary amine groups and the N-H deformative vibration of secondary amine groups in PEI Chuang et al. 2021). To clarify the variation of crosslinked degree with feed ratio, soaking experiment was employed in excess DMF solvent . As shown in Fig. S2, after 72 h shaking, nano-SiO 2 hybrid PVDF-HFP film was turned to be clear and transparent, while nano-SiO 2 hybrid 3D crosslinked films still had undissolved crosslinked fragments. Moreover, this fragment mass ratio gradually rose as the feed ratio grew, indicating higher crosslinked degree existed in with the increase of PEI content. The existence of crosslinked network will produce a steady reticular structure to endure external force. As displayed in subsequent data from Fig. 3(d), significant enhancement of tensile strength can be observed after PEI added into the composite separator, which further provides supporting evidence for the occurrence of crosslinking reactions in this system (Shin et al. 2021).
Morphology and pore structure The separator sandwiched between two electrodes not only plays an important role in physical isolation, but also requires appropriate pores to allow rapid transmission of lithium ions in the battery (Yang et al. 2021). SEM is usually employed to detect information about surface morphology and pore structure. As seen in Fig. 1(c) and (d), tissue paper is made from a disorderly stack of fibers, which results in apertures ranging from a few microns to dozens of microns and distributed non-uniformly. To avoid short-circuit phenomenon and uneven lithium deposition, it is imperative to give a modification for tissue paper. Accordingly, nano-SiO 2 hybrid crosslinked polymer coating was introduced on the surface of tissue substrate and shown in Fig. 1(g-n). Compared with CS(1:0) separator only using nano-SiO 2 hybrid PVDF-HFP as coating layer (Fig. 1(e-f)), the crosslinked composite separators are observed to give more highly interconnected honey comb-like structures, which will provide unobstructed access for lithium ion transmission ) and simultaneously be advantage to prevent the penetration of lithium dendrite (Lopez et al. 2019). Notably, the pore size is exhibited to increase step by step as crosslinked degree raised. The size distributions of all composite separators were carried out by bubble-point method and shown in Fig. 1(o). An enlarged trend for pore size can be found with the increase of PEI content, consisting with the results of SEM images. Besides, in comparison to CS(1:0) separator (1.14 μm), basically the same or larger hole diameter is developed in CS(5:1) and CS(5:2) separator, while lower pore size is generated in CS(10:1) (0.65 μm) and CS(20:1) (0.57 μm) separator, which level is more suitable for battery separator when inhibiting micro short-circuit (Lee et al. 2014). The structure of porous membranes fabricated using the solvent volatilization method is a phase separation process and not affected by a monotonous factor (Chen et al. 2016;Zhang et al. 2019).
The separation rate and state will be influenced by the viscosity of crosslinked coating solution and the content of hydrophilic PEI, thus different sized pore structures developed with the variation of crosslinked degree.
The capacity of separator to accommodate electrolyte has a close relationship with porosity (Zhang et al. 2018). As shown in Table 1, with the increase of crosslinking agent PEI, the porosity of crosslinked composite separator gradually decreases, giving an opposite trend for pore size. This result demonstrates sparse distribution of pore structure incubated in deepened crosslinked system, which can also be identified from SEM images. When fixing the ratio of PVDF-HFP and PEI at 20:1 or 10:1, the porosity of about 70% can be obtained, which is basically similar with that of traditional composite separator CS(1:0), but much higher than that of Celgard 2400 (30%). This demonstrates higher porosity and smaller pore size can simultaneously achieve through controlling the crosslinked degree, which are the prerequisite for absorbing more electrolytes and preventing the penetration of lithium dendrite (Guan et al. 2020).
Wettability and electrolyte uptake Wettability can reflect adsorption and expansion rate of electrolyte on separator (Lv et al. 2021). Figure 2(a) gives immersion-height photograph at the time of separator immersed in liquid electrolyte 120 min. It was found that with the increase of hyperbranched PEI content, electrolyte immersion-height for the crosslinked composite separator raised firstly and then reduced. When the feed ratio of PVDF-HFP and PEI was set at 10:1, highest immersionheight achieved with the value of 3.2 cm. As a comparison, only 0.3 cm and 2.2 cm immersion-height were respectively obtained for Celgard 2400 and Fig. 1 a Fabrication procedure of crosslinked coating layer constructed on tissue paper and crosslinking reaction process thereof; b ATR-FTIR spectra of CS(10:1) and CS(1:0) separator (inset: an enlarged graph at the range of 1750-1450 cm −1 ); SEM images of (c and d) tissue paper, (e and f) CS(1:0) separator, (g and h) CS(20:1) separator, (i and j) CS(10:1) separator, (k and l) CS(5:1) separator and (m and n) CS(5:2) separator with different magnification; And o Pore size and distribution of CS(5:2), CS(5:1), CS(10:1), CS(20:1) and CS(1:0) separator ◂ the control CS(1:0) separator. This behavior demonstrates superior wettability of CS(10:1) separator. Moreover, contact angles of electrolyte on separators were measured and shown in Fig. 2(b). Compared with Celgard 2400 and CS(1:0) separator, the lowest contact angle was obtained for CS(10:1) separator no matter at initial or after contacted 20 s. As the crosslinking degree increased, there were complete opposite trends observed between contact angle and immersion-height, indicating coincident wettability results. These results can be attributed to hydrophilicity of tissue paper substrate and strong polarity provided by abundant amines in hyperbranched PEI, which can improve electrolyte affinity with separator and be helpful for electrolyte adsorption (Sheng et al. 2021). On the other hand, the incorporation of crosslinking agent at an appropriate content can regulate the formation of interconnected pore structure but produce negligible effect on porosity, which can make the adsorbed electrolyte infiltrate and diffuse in separator quickly (Sheng et al. 2021). However, deeply crosslinking can hinder the penetration of electrolyte and result in attenuation of wettability, which is mainly related to the reduction of porosity as displayed in Table 1. Good wettability of CS(10:1) separator will be beneficial to lithium ion migration between electrodes especially for high power LIB and reduce standing time after cell assembly (Sheng et al. 2021).
Separator serves as electrolyte reservoir to ensure number of lithium ions transferred during battery operation, thus electrolyte uptake is an important parameter to decide electrochemical performance ). According to data from Table 1, compared with Celgard 2400, the control composite separator CS(1:0) can accommodate more electrolyte. This behavior has a close relationship with its higher porosity and better hydrophilicity to electrolyte caused by the employment of polar coating layer (PVDF-HFP and SiO 2 ) and substrate (Yuan et al. 2021). With the addition of crosslinking agent PEI, the capacity to absorb electrolyte begins to increase  gradually. Specially, the crosslinked composite separator CS (10:1) gives the highest electrolyte uptake with the value of 254%. This cannot separate from good affinity between strong polarity PEI and electrolyte. In addition, the application of hyperbranched PEI forming crosslinking structure can reduce the crystallinity of polymer coating PVDF-HFP, which will be advantageous to liquid electrolyte swell into coating layer (Zhou et al. 2015). These two factors and basically unchanged porosity endow CS(10:1) separator with promoted electrolyte uptake. As PEI content continued to increase, the electrolyte uptake of the crosslinked composite separator begins to decrease mainly due to reduced porosity displayed in Table 1. Higher electrolyte uptake of CS(10:1) separator will accommodate more Li + to participate in charge transfer, thus producing lower internal impedance (Yuan et al. 2021).
Thermal analysis, flame retarding property and mechanical behavior To cope with heat generated during the operation of high power LIBs, separator must have sufficient thermal stability and flame retardancy, so that providing superior safety when battery working in a variety of possible normal cases, or not causing more serious explosions when sudden short-circuit reaction occurred (Babiker et al. 2021). Figure 3(a) gives photographs of separators endured heat treatment at different temperatures. We can find that Celgard 2400 separator suffered from significant dimensional shrinkage above 200 °C until completely fused. However, the original size can be maintained for the prepared tissue-based composite separator in the whole temperature range. This is mainly related to cellulose and lignin, main components of tissue paper, as well as SiO 2 nanoparticle, which have excellent thermal stability and can provide a stable supporting skeleton for composite separators (Wang et al.  . Compared with PVDF-HFP as binder in coating layer, the introduction of crosslinked structure will also contribute to the sufficient thermal stability of composite separators. From DSC curves displayed in Fig. 3(c), we can find there is a melting peak of PVDF-HFP at about 150 °C for the control CS(1:0) membrane (Zhang et al. 2021b), while no melting signal is distinguished for CS(10:1) separator with crosslinked structure. This demonstrates that the construction of crosslinked network can restrict the melting of PVDF-HFP, thus boosting heat resistance of the overall structure again (Zhu et al. 2019a). However, the color is observed to get darker gradually for the crosslinked composite separator with the increase of PEI content and heating temperature from Fig. 3(a). This phenomenon can be explained as further deepened crosslinking reactions and oxidation of unreacted amine groups in PEI (Zhao et al. 2009). TGA curves were also measured and shown in Fig.  S3. Compared with separator CS(1:0) without SiO 2 , the incorporation of crosslinked network is observed to give composite separator with higher solid residue when temperature increased above to 600 °C. Especially, the addition of SiO 2 further raises this value and offers CS (10:1) separator with best thermal stability. By comparison, Celgard 2400 separator leaves no solid when temperature rises to 500 °C. Flame retardancy was tested by combustion experiment and given in Fig. 3(b). An important difference can be discovered when separators are approached with fire. Celgard 2400 burned and shrank rapidly resulting from inherent thermal sensitivity and inflammable nature of polyolefin. Pure tissue paper also burns rapidly, while the composite separators with and without a crosslinked structure exhibit a self-extinguishing phenomenon. Limiting oxygen index (LOI) data were tested and shown in Fig. S4, LOI value is observed to increase in the order of tissue paper < CS (1:0) without SiO 2 < CS (10:1) without SiO 2 < CS (10:1). This result demonstrates that the addition of nano-SiO 2 , the existence of fluorine element in coating layer and the fabrication of crosslinked network have synergetic effect in improving flame retardancy of CS(10:1) separators (Zhang et al. 2021a). This behavior will be helpful for the improvement of battery safety. Separator should achieve certain mechanical properties to meet winding operation in the production process and resist the growth of lithium dendrite during battery cycling (Wu et al. 2019). The stress-strain curves are displayed in Fig. 3(d). Pure tissue paper had a very low tensile strength of 2.9 MPa. After PVDF-HFP and nano-SiO 2 coated on tissue paper, enhanced tensile strength (7.5 MPa) is obtained for CS (1:0) separator. Especially, when PEI was also introduced in coating layer, tensile strength increased substantially and the maximum value of about 9.9 MPa was achieved for CS (10:1) separator. This can be ascribed to the formation of 3D crosslinked network between PVDF-HFP and PEI, which is hard to deform when exposed to external force (Chuang et al. 2021). With the continuous increase of crosslinked degree, Young's modulus begins to increase for CS (5:1) and CS (5:2) separators, while tensile strength gives an opposite trend. We attribute it to the likelihood of excess free PEI existing in these highly crosslinked systems, which will have a weakening effect on mechanical behavior. Clearly, tensile strength can reach the maximum improvement when the coating layer is applied with the optimal degree of crosslinking. This is a very satisfactory result when compared with other coating layers reported in the previous literature , but still lower than that of Celgard 2400 due to its special uniaxial tension process during fabrication ). Besides, we placed sample CS (10:1) without SiO 2 at 250 °C for 0.5 h and measured the corresponding mechanical performance. As shown in Fig. S5, compared with the original sample, treated sample exhibits enhanced tensile strength and higher Young's modulus. This demonstrates continuously deepened crosslinked reaction occurred in sample, which further strengthens the ability of resisting deformation. Accordingly, it can be concluded that the crosslinked composite separator is stable enough to withstand high temperature.
Interfacial compatibility between separator and electrode is another parameter to influence ohmic polarization degree and Li + storage capacity . Figure 4(b) gives interfacial impedances of simulate batteries based on Celgard 2400 or a series of composite separators. The distance between intercepts of Nyquist plots on the real axis is considered as interfacial resistance, which includes solid electrolyte interface (SEI) film impedance located at high frequency region and charger transfer impedance placed in medium or low frequency region (Xu et al.  . The distance between semicircle intercepts can be found to increase stepwise with a sequence of CS (10:1) < CS (20:1) < CS (5:1) < CS (5:2) < CS (1:0) < Celgard2400, indicating better interfacial compatibility existed between crosslinked composite separator and electrode. Especially when the feed ratio regulated at 10:1, most favorable interfacial compatibility is reached. This can be ascribed to strong binding force produced by crosslinked network to liquid electrolyte, which will reduce free electrolyte and minimize its side reactions occurred on lithium electrode, thus cutting down SEI film impedance (Lv et al. 2021). For CS (10:1) separator, more electrolyte is entrapped by crosslinked network so that thinner SEI film and lower impedance obtained. On the other hand, superior electrolyte absorption behavior of CS (10:1) separator can facilitate the formation of a gel layer with adhesive property at the separator/electrode interface, thus leading to closer interfacial contact and faster Li + transfer ). These two factors together bring about the interfacial optimization between CS (10:1) separator and electrode.
Separator after absorbing electrolyte should maintain stability within a wide voltage range to offer a higher operating potential for LIB or withstand overcharge/overdischarge (Kim et al. 2020). Electrochemical windows based on SS/separator/Li cells were detected by linear sweep voltammetry (LSV) method and the corresponding results were depicted in Fig. 4(c). There is a sudden current increase observed in LSV curve when voltage increased to a certain value, which indicates the occurrence of oxidation decomposition reaction in this system (Kim et al. 2020). Accordingly, initial decomposition voltages can be found to increase in the order of Celgard2400 < CS (5:2) < CS (5:1) < CS (1:0) < CS (20:1) < CS (10:1), indicating better anodic stability for CS (10:1) separator. This behavior can be attributed to improved electrolyte storage capability and enhanced binding force between composite separator and carbonate electrolyte by the construction of crosslinked network, which can alleviate the decomposition of free solvent molecules on the anode, thus being inclined to provide broader electrochemical window for LIBs (Liang et al. 2021). While crosslinking degree increased, electrolyte storage capability begins to get weaker, thus resulting to the existence of more free electrolyte in cell and the earlier appearance of oxide decomposition. Lithium ion transference number is an important parameter in investigating concentration polarization phenomenon (Zahn et al. 2016). In this study, Li + transference numbers (t Li+ ) were estimated using Eq. 4 by the combination of chronoamperometry and EIS of Li/separator/Li cells as shown in Fig. 4( di). An increasing trend is presented for t Li+ with the order of Celgard2400 < CS (5:2) < CS (5:1) < CS (1:0) < CS (20:1) < CS (10:1), indicating that appropriate crosslinking degree can improve lithium ion transference number. This is mainly related to physical binding produced by crosslinked network to larger ions PF 6 − , which will restrict mobility of anion and give lithium ions more chance to take part in charge transfer (Suriyakumar and Stephan 2020). Meanwhile, nano-SiO 2 in the coating layer of composite separator can also serve as Lewis acid centers to trap electrolyte anions (Liu et al. 2017;Zhou et al. 2016). These two factors finally give a superior t Li+ for CS(10:1) separator compared with Celgard 2400 and traditional composite separator. However, as crosslinking degree continues to increase, there is an obvious decline observed for t Li+ . This can be interpreted as its less effective fixing to PF 6 − due to the enlargement of aperture and the reduction of porosity as discussed above, which will directly result in free movement of anions and cut down charge transfer duty of Li + . The higher t Li+ for CS (10:1) separator will provide timely replenishment of Li + during electrode reaction and be benefit to weaken concentration polarization degree (Xia et al. 2021). According to detailed investigations and analysis, in the following study we will choose CS (10:1) separator to fabricate battery due to its outstanding physical properties and electrochemical performance.

Cell performance
Lithium ion batteries assembling with Celgard 2400, CS (1:0) or CS (10:1) separator were respectively charged at a constant current rate 0.2 C and discharged at different current rates ranging from 0.2 to 2 C and then back to 0.2 C. As shown in Fig. 5(a-c), discharge capacities of the three systems decrease gradually with the increase of discharge C rate. This can be attributed to aggravated polarization phenomenon at a higher current density including ohmic, concentration and electrochemical polarization, which will cause the reduction of discharge voltage and shorten discharge time to less than theoretical value, thus leading to attenuated discharge capacity . When discharge rate varied from 2 C back to 0.2 C, compared with other two systems, the battery using CS (10:1) separator presents higher coincidence whether in discharge voltage or capacity, indicating its better reversibility in cell performance. Figure 5(d) summarizes rate capabilities of the three systems. Superior discharge capacities are observed for CS (10:1) sample in all of the C-rate. Especially, when current rate increased to 2 C, a substantial increase was observed in the capacity divergence between them. As discussed above, CS (10:1) separator has superior ionic conductivity, excellent interface compatibility between electrode and separator as well as higher Li + transference number, which will lead to lower ohmic resistance and concentration polarization existing in battery, thus giving enhanced discharge performance. With the increase of discharge current rate, the internal polarization would be magnified, thus the discharge capacity discrepancy became more evident (Valverde et al. 2020). This outstanding cell performance at elevated rate will provide possibility for the application of CS (10:1) separators in high power LIBs.
The battery must maintain a high discharge capacity even after multiple cycles, which is a premise of application (Xia et al. 2021). Figure 5(e) gives the variation of discharge capacity and coulombic efficiency with cycle numbers, and Fig. S6(a-c) shows the corresponding charge-discharge curves for selected cycle numbers. We can find that all the three systems show nearly 100% coulombic efficiencies. Moreover, in the initial cycles, system CS (10:1) reveals slight higher discharge capacities (about 153 mAh g −1 ) than system CS (1:0) (about 149 mAh g −1 ), but a larger capacity divergence appeared between system Celgard 2400 (about 138 mAh g −1 ) and CS (10:1). This phenomenon is consistent with rate capabilities mentioned above. As cycling tests proceed, CS (1:0) system displays a stepwise decay phenomenon in discharge capacity. Especially a sudden drop occurs along with discharge capacity decreases to 90 mAh g −1 when cycle number surpassed 65. Finally, LIB using CS (10:1) separator gives a discharge capacity retention about 98% after 100 cycles, much higher than that of Celgard 2400 (about 84%) and CS(1:0) system (about 64%). This can be attributed to the construction of 3D crosslinked coating layer on tissue paper through hyperbranched PEI and polymer matrix PVDF-HFP, which can not only fix electrolyte firmly in separator but also provide a steady coating structure to endure long term swelling of electrolyte. These behaviors will not only ensure sufficient electrolyte for lithium ion transport but also maintain efficient physical isolation existed between electrodes during the whole cycling, thus bringing about stable cycling performance for system CS (10:1) (Xia et al. 2021). On the contrary, separator CS (1:0) only employs PVDF-HFP as binder, which allows easy loss of structure after being swollen by electrolyte and further leads to the shedding of nano-SiO 2 from substrate. This results in a micro-short circuit in the last stage and the attenuation of capacity (Jia et al. 2020). Also, the cycle performance of these LIBs is compared with various separators from the literature in Table S1. The results show that our crosslinked composite separator provides superior cycle performance for LIBs.

Conclusions
In summary, a series of composite separators was successfully fabricated through constructing nano-SiO 2 hybrid coating layer with different crosslinking ratios on tissue paper. When the ratio of PVDF-HFP and PEI were set at 10:1, superior wettability, enhanced ionic conductivity, depressed interfacial impedance and higher lithium ion transference number were obtained for the composite separator, which finally endowed battery with better rate capabilities. This can be attributed to smaller pore size but basically unaffected porosity, improved hydrophilicity and the formulation of good interconnected pore structures along with the addition of certain hyperbranched PEI. Moreover, the crosslinked coating layer was investigated to bring about the improvement of mechanical behavior, the broadening of electrochemical window and unchanged cycle performance for the composite separator, demonstrating its effect on overall structure stability when suffered from various types of external stimulations. Finally, the composite separator displayed sufficient thermal resistance and obvious flame retardant phenomenon, which had a close relationship with the employment of tissue substrate, nano-SiO 2 and crosslinked fluorinated polymer network in this system. Therefore, this designed composite separator will pave a way for the development of higher power LIBs.