Polyphenylene sulfide quasi-solid-state electrolyte for limited Li metal battery

The urgent need for safe and high energy batteries is pushing the battery studies towards the solid-state direction, and the most central question is finding proper solid-state electrolyte. So far, the recently studied electrolyte systems have obvious advantages and fatal weaknesses, resulting in indecisive plans for industrial production. In this work, we propose a thin and dense lithiated polyphenylene sulfide (PPS)-based solid polymer electrolyte prepared by a solvent-free process in a pilot stage. Moreover, the PPS surface is functionalized to immobilize the anions, increase the Li ion transference number to 0.8-0.9, and widen the electrochemical potential window (>5.1 V). At room temperature, the PPS-based quasi-solid electrolyte (PPS-QSSE) exhibits high intrinsic Li + diffusion coefficient and ionic conductivity (>10 -4 S cm -1 ), excellent thermal stability, and Li + transport rectifying effect, resulting in homogenous Li-plating on Cu at high current density. Based on the limited Li-plated Cu anode or anode-free Cu, high loadings cathode and high voltage, the Li metal batteries with PPS-QSSEs deliver high energy density (>1000 Wh L -1 ) and good cycling at high power (900 W L -1 ) exceeding that of state-of-the-art Li-ion batteries. The results enlighten the mechanism of solid-liquid two phase conduction and promote the solid-state battery towards practicality.


Introduction
With the excessive use of traditional fossil energy sources such as coal and petroleum, environmental pollution and energy shortages put forward serious challenges to modern energy storage methods 1 . Under such circumstances, lithium-ion batteries (LIBs) have become more and more widely used in energy storage due to their advantages of high energy density and cycle life 2 .
However, conventional LIBs usually use a large amount of flammable, explosive, and volatile organic liquid electrolytes, leading to serious safety problems in use 3 . Moreover, the state-of-the-art LIBs are insatiable in the rising demand for higher energy density. Advanced LIBs based on Li metal anodes (LMBs) could provide higher energy density and become a hot research topic 4 . But the "dead" Li and low cycling caused by uncontrolled growth of Li dendrites also limit the development of the LMBs.
By using solid-state electrolytes (SSE) instead of traditional organic liquid electrolytes, solid-state LMBs are considered to be the most promising energy storage method. SSEs can suppress Li dendrites' formation, thereby fundamentally solving the safety and cycling issues of LMBs 5 . To achieve safe, high energy and power density, and better cycle stability, the solid-state electrolytes require a high diffusion rate including bulk and interface between solid-state electrolyte domains in a wide temperature range, thin and flexible electrolyte films, chemical and mechanical stabilities.
However, the solid-state electrolytes reported so far remain formidable challenges to meet all the above requirements. In general, SSEs can be divided into two categories: inorganic glassy or ceramic electrolytes and solid polymer electrolytes (SPEs). For instance, the inorganic solid electrolyte systems, including the garnet-type and perovskite-type ceramic materials and sulfide-type glass, have been extensively studied for the high conductivity at room temperature.
However, due to the brittle nature, poor processing performance, high interface resistance with the electrode, and high sensitivity to moisture for sulfides, they are still not suitable for wide commercial applications 6 . Various organic SPEs, such as the most studied polyethylene oxide (PEO) 7 , the polyacrylonitrile (PAN) 8 , the poly(vinylidene fluoride) (PVDF) 9 , the polymethyl methacrylate (PMMA) 10 , and the polypropylene oxide (PPO) 11 , have good thermal stability and flexibility. Still, their low ionic conductivities restrict the high-power operation of the battery at room temperature. One of the approaches to address above mentioned challenges is to use quasi-solid-state electrolyte (QSSE) by adding the liquid electrolytes (organic solvents or ionic liquid) into the solid electrolytes 12 . The gel polymer solid electrolytes (GPEs) are considered the most promising quasi-solid electrolyte candidate 13 . But the low ion transference number, expensive precursors, flammable and inhomogenous thick film structure of gel-type electrolytes also hinder its potential application 14 . Besides, the amount of liquid electrolyte needs to be minimized for retaining intrinsic safety.
Until 2019, Liu et al. 15 firstly reported that a porous (60-80% porosity) polyphenylene sulfide (PPS) has superior electrolyte wettability, high ionic conductivity, and excellent thermal stability for flame retardant LIBs. Zimmerman et al. 16 patented in 2013 the fabrication of ions-doping PPS-based films through traditional high-temperature extrusion process for the SPEs, which have high ionic conductivity for various ions (Li + , Na + , Mg 2+ , Ca 2+ , Al 3+ , and OH -). But until now, it is still a lack of detailed scientific research on the PPS-based SPEs to verify that such flame retardant material can be commercialized in the battery field. In addition, the traditional polymer film fabrication technologies, such as solvent or high temperature casting processes, always cause high porosity, recrystallization (>150 °C for PPS) 17 , oxidation, and uncontrollable crystallinity 18,19 . Especially, the S elements in PPS start to be oxidized at ~200 °C in air 20 . As a result, the PPS separator with high porosity required a large amount of liquid electrolyte uptake similar to commercial separators, which can not solve the Li dendrites formation for LMBs and scarify the intrinsic safety of solid-state electrolytes due to the presence of a large amount of liquid electrolyte. Besides, Li diffusion needs to be significantly improved.
Recently, we reported a solvent-free process in a pilot stage, combining a high-speed air blowing and rolling process, which can make the free-standing film at relatively low temperature 21 . Hence, in this study, we used this solvent-free process to fabricate a thin (~35 μm) and dense (~9% porosity) lithiated PPS based SPE film (PPS-SPE) with high Li loading of 3 wt%. Tracy of dual-salt LiDFOB/LiBF4 liquid electrolyte (4 μL Ah -1 ) was added to infiltrate the electrodes and PPS-SPEs to obtain the PPS-QSSEs, which showed higher conductivity (>10 -4 S cm -1 ) and diffusion rate in bulk and interface at room temperature. Moreover, the PPS particle surface was functionalized by anions-immobilizing molecules such as tetrachloro-p-benzoquinone (TCBQ), to improve the Li transference number (tLi+=0.8-0.9) and electrochemical potential window (EPW>5.1 V), resulting in homogenous Li plating on Cu at high current density (2 mA cm -2 ). Both experimental and computational studies demonstrate PPS-TCBQ coupling with a thin layer of dual salt as efficient QSSE. The full cells with QSSEs, high loading (2.2-3.5 mAh cm -2 ) cathodes, and limited Li-plating Cu anode showed good cycling, high energy density (> 1000 Wh L -1 ), and power density (>900 W L -1 ) at room temperature.

PPS-based SPE design and fabrication
In this work, the lithiated PPS powders were prepared on the accounts of the commercial way invented by Edmond and Hill (Philip petroleum Co.) in 1967 22 . The difference was that the reactant of Na2S was replaced by Li2S. And the Li ions were simultaneously introduced into the PPS matrices, as shown in Eq. 1. 1 However, the undesired Cl ions were also inevitably introduced into this system, which can cause corrosion on the current collectors and attack electrode materials 23 . Hence, the Cl ions need to be bound in the SPE, as proposed by recent strategies that the metal-organic frameworks (MOFs) were used to immobilize the anions in order to get single-ion conducting SSEs 24 . In this work, we chose a much cheaper and non-toxic organic compound, tetrachloro-p-benzoquinone (TCBQ), which is commonly used for biomedical analysis to chelate the molecules and ions with lone pair of electrons 25,26 . To investigate the Cl-binding effect of TCBQ, the density functional theory (DFT) simulations were carried out. All the calculation details are given in the Supplementary Methods section. The results are shown in Fig. S1. It is slightly difficult to adsorb the first and second Cl atoms with a single TCBQ molecule. But interestingly, the negative formation energies mean that the further third and fourth complexation reaction will get easier and become spontaneous. Hence, the mixture of lithiated PPS and TCBQ powders was heated at 210 °C with Ar protection determined by differential scanning calorimetry (DSC) and thermogravimetric analysis (TGA) measurements to make the lithiation and complexation reactions happen. Then the powders were washed several times with distilled water to remove the excess LiCl. The Li content in the PPS-SPE film was ~3 wt%, measured by the inductively coupled plasma emission spectrometer (ICP). The XRD patterns (Fig. S2) of the final powder product showed mainly the well-crystallized linear PPS with the space group of Pbcn 27 .   Fig. 1c left. And the minimum Li diffusion energy barrier through the S channel was 0.48 eV calculated by the climbing image nudged elastic band (CI-NEB) method. More interestingly, the Li diffusion pathway on the PPS surface is different from bulk. In this work, the (002) lattice plane was chosen as the exposed PPS crystal surface for that the (002) peak shows the highest intensity in the XRD patterns ( The results indicate that the PPS should be well crystallized and preserve the perfect S channels in the bulk and a large amount of S sites on the surface to gain high Li-ion conductivity. Hence, in this work, the PPS-based SPE was fabricated in the solvent-free and low-temperature way in a pilot stage in order to reserve the property of the lithiated PPS powders utmostly, as shown in Fig. 1c. The whole process consists of the high-speed air blowing process to fiberize the polytetrafluoroethylene (PTFE), the rolling process at 130 °C to make the PPS-based SPE film (for more details, see Methods section and Movie S1). PPS-based SPE film shows a compact and smooth surface. In addition, we can also see a small number of incomplete filling voids between adjacent PPS particles on the surface. Even more small cracks on particle surfaces in nanoscale can be observed in the higher magnification image marked by a square in Fig. S3. These microcracks could be attributed to the fracture among the tufted PPS chains, forming the surface of the exposed chain and facilitating the Li-ion diffusion, as illustrated and calculated in Fig. 1b. The compact and dense structure was further demonstrated by the SPE film's cross-section view, as shown in Fig. 1d. The thickness was measured to be ~35 μm, which is adjacent to the commercial PE or PP based separators. Besides, it is obviously can be seen that the PTFE fibrillated by the high-speed dry air blowing presents a three-dimensional network structure and tightly binds the PPS powders together.

Materials characterization of PPS-based SPEs
The focused ion beam (FIB)-SEM was taken on the cross-section of the PPS-based SPE film. The chemical composition was confirmed by the energy disperse spectroscopy (EDS) element mapping.
As shown in Fig. 1e, the S, Cl, C are homogeneously distributed in the film. However, the enrichment of the O element in the PPS powders interface clearly shows in the EDS mapping (insert in Fig. 1e). The O element was mainly introduced by the TCBQ. Hence, the exact reaction between the PPS and TCBQ was further identified by the X-ray photoelectron spectroscopy (XPS) and solid-state nuclear magnetic resonance (SSNMR). Fig. 1f gives the comparison of O1s deconvolutions derived from the XPS results of four different samples. The pure TCBQ sample exhibits the major O1s peak at 532.5eV, which is attributed to the benzoquinone group. And the pure PPS sample shows a distinct O1s peak at 531.8eV, which could be due to the oxidation of the SSNMR spectrum. It is properly due to that more hydrogenate TCBQ on the PPS surface was exposed after the film-forming process by hot rolling at 130 °C . Fortunately, the Li diffusion on the TCBQ functionalized PPS surface is not too much affected. The CI-NEB calculation result showed that the Li diffusion barrier energy was 0.34 eV (Fig. S4), which was slightly higher than the value for clean PPS surface (Fig. 1b), ensuring the Li high-speed surface diffusion channels for PPS-based SPE.
The value was comparable to the sulfides with the best reported SSE performances at room temperature ((Li2S)7(P2S5)3, 4.58×10 -8 cm 2 s -1 ) 35 , which gives the possibility to achieve a SPE with high Li-ion conductivity (σLi+) using lithiated PPS. However, the measured σLi+ for PPS-SPE was

Immobilization of anions
Li transference number (tLi+) is another very important factor for the SSE of LIB. A large transference number is essential for reducing concentration polarization of electrolytes during charge-discharge steps, and thus producing higher power density and homogenous Li diffusion/plating/stripping 40 . It is highly desirable that the transference number of lithium ions approaches 1 in an electrolyte system. The measured tLi+ of the dual-salt LiDFOB/LiBF4 electrolyte through PE separator using the Bruce-Vincent-Evans method 41 was ~0.45, as shown in Fig. 4a. For the PPS-QSSEs with the dual-salt electrolyte, it was found that the tLi+ increased with the TCBQ content and reached up to 0.8-0.9 when the TCBQ was higher than 2.5%. And poor tLi+ results were obtained for the ones without TCBQ, even lower than the liquid electrolyte (Fig. 2f). As we illustrated above, the TCBQ was used to chelate the Cl ions. The DFT simulations demonstrated that the -F bonds on the BF4 and DFOB also prefer to dipole adsorb on the C sites of the TCBQ molecule, and the calculated formation energies were as low as -1.80 and -0.31 eV, respectively, as shown in Fig. 2g. The charge difference densities before and after the adsorption were also calculated to quantify the charge redistribution and dipole moments of the interaction. Obviously, both of the BF4 and DFOB show strong dipole adsorption on the TCBQ. In addition, the interaction between BF4 and TCBQ was much stronger. The 19 F SSNMR spectra of the lithiated PPS powders with and without TCBQ after mixing and drying with the dual-salt electrolyte were given in Fig. 2h.
For the sample without TCBQ, two main peaks at -148 and -153 ppm correspond to the BF4and DFOBrespectively, as same as the dual-salt mixture and the reported results by Weber et al. 42 .
Distinct chemical environments for the BF4and DFOBwere presented for the sample powders with TCBQ. The chemical shift was nearly 10 ppm, which could be due to that the BF4and DFOBions are chelated and adsorbed on the TCBQ via diople adsorption. The 19 F SSNMR spectrum of fresh PPS-SPE was mainly attributable to the PTFE. After cycling in LMBs, an additional peak was found at -163 ppm with the same peak position for the infiltrated sample with TCBQ.
Another surprising finding was that the electrochemical potential window (EPW) was widened by adding the TCBQ in the PPS-QSSE with the dual-salt electrolyte.  Fig. S10), and X-ray photoelectron spectroscopy (XPS) (Fig. S10), which were demonstrated the Cl corrosion on the Al current collector. The corrosion side reaction was eased by adding the TCBQ in the PPS-QSSE (Fig. 2I, Fig. S9, and S10). The widest EPW of 5.  Li-plating surface with large mossy-like grains (~8 μm) was gained, as shown in Fig. S13c. And increasing the current density to 2 mA cm -2 , the Li-plating structure got even denser with little pores (Fig. 2c), and the grains were refined to ~1 μm (Fig. S13d). Variations of the Li-plating structures were even more pronounced by investigating the cross-section of the Li-plating layers. Typical mossy-like Li grains growing on Cu foil show a thick Li layer (~48 μm) with high porosity for the sample prepared by PE separator (Fig. 3d). It is still possible to find some protruding dendritic Li as g marked by a circular in Fig. 3d. For the two samples lithiated through PPS-QSSEs, the refined Li grains formed the compact layers ( Fig. 3e and f). Especially, for the sample lithiated at high current density of 2 mA cm -2 , the deposited Li layer was so dense and showed a thickness of ~15 μm, which was adjacent to the theoretical value of 9.7 μm (for 2 mAh cm -2 ) 48 . As we illustrated above, the Li-ions transport in the PPS-QSSE following three fixed paths, the S channels in the bulk of the PPS crystal, the PPS particle surface along the chains, and the gaps infiltrated by the liquid phase.

Li metal battery with limited Li-plating Cu anode
To further characterize the practicality of the PPS-QSSE (with 5% TCBQ) at a relatively high current density and room temperature comparing to other solid-state battery works, the LiFePO4 thick solvent-free electrodes for Li-plating were removed and replaced by HVLCO or presented a high discharge capacity of 183 mAh g -1 at 0.1 C (1 C=160 mA g -1 ) and maintained 80 % of the initial capacity (145 mAh g -1 ) at a high current density of 1 C (2.2 mA cm -2 ), confirming that the PPS-QSSE has good high-voltage stability and acceptable Li-ion conductivity at room temperature. Fig. S15 gives the GDC curves of NCM523-Li-Cu cells with PPS-QSSE, which also suggests acceptable rate capability at room temperature, comparing to other solid-state and liquid LMBs works 51,52 , as listed in Table S1 and S2. In addition to rate performance, cycle life (the cycle with 80% of the initial capacity) is also a critical parameter for evaluating LIBs. Fig. 4b presents the comparison of cycle performance between PE separator with liquid electrolyte and PPS-QSSE with and without TCBQ by using 150μm Li metal anodes and HVLCO cathodes. At 0.5 C charge and 1 C discharge current, the cell with commercial PE separator failed after 73 cycles (Fig. 4c), which was mainly caused by the uncontrollable growth of Li dendrites at high current density coinciding with the SEM and Li-Cu cell tests (Fig. 3). However, the two samples assembled with PPS-QSSEs exhibited better cycling life. Especially, PPS-QSSE with 5 wt% TCBQ showed the best cyclic stability, which cycled with a discharge capacity of over 150 mAh g -1 , and presented 19.1% capacity loss after 200 cycles.
Moreover, high coulombic efficiency was maintained at ~99.4%. However, PPS-QSSE without TCBQ showed much lower capacity and unstable coulombic efficiency. As discussed above, the unstable cycling mainly resulted from the Clcorrosion side reaction, lacking the anions-immobilizing effect of TCBQ. The cycled HVLCO cathodes were examined by SEM (Fig.   S10) and XPS (Fig. S10d). Distinct Cl2s (270.1 eV) and Cl2p (200.3 eV) peaks were presented in the XPS spectra, and the corrosion damage of the Al current collector was evident for the sample used PPS-QSSE without TCBQ, which were barely detected and observed for the other two samples using PPS-QSSE with TCBQ and PE separator. Cu anode at the highest current density of 2 mA cm -2 for 1h presented the best cycle performance.
This result was extraordinary consistent with the dense and homogenous Li-plating layer with refined grains Li-plated at 2 mA cm -2 in Fig. 3, which could give the best electrical connection and promoted the further reversibility of Li-striping and the cycling stability. Hence, the Li-plating current of 2 mA cm -2 was fixed, the Li-plating time was gradually lengthened to stepwise increase the Li loadings (N value), in order to achieve the best cycling performance, as shown in Fig. 4c.
The 0.5 C charge and 1 C discharge currents at 25 °C were chosen to perform the cycling in order to meet the vast majority of practical battery power requirements, which were nearly the highest cycling current among the reported solid-state batteries with limited Li anodes and lean electrolyte, as listed in Table S1. For the commercial PE separators, it was prone to short-circuit when plating at high current, a low Li-plating current of 0.2 mA cm -2 for 10 h was used for the repeatable result. As shown in Fig. 4c, the cycling performance of the cell with PE was very poor (< 10 cycles) at 1 C. In comparison, all the cells with PPS-QSSE presented high coulomb efficiency of over 99% and better cycle reversibility, especially when using 6 mAh cm -2 Li-plated Cu, the cycling life can be extended to 130 cycles (80% of their initial capacities).
Moreover, the TCBQ functionalized PPS-QSSE also showed excellent stability in the high-voltage 5V LiNi0.5Mn1.5O4 (LNMO) system, as indicated in Fig. 2I. Here, 5 wt% Li2C2O4 in LNMO cathode was used for the Li-compensating and plating on Cu. Due to the deliquescent character of Li2C2O4, the solve-free electrode process 21 was used to fabricated the free-standing LNMO film ( Fig. 4d and Movie S5) and final LNMO cathode (P=1.5 mAh cm -2 ). Fig. 4c presents the GDC curves of the LNMO-Cu anode-free cell (N=0 mAh cm -2 ) cycling at 0.2 C (1 C=120 mA g -1 ). The long initial charge curve indicates the decomposition of the Li2C2O4, forming extra Li-plating on Cu, which ensures the initial 5 stable cycles. Then, the degradation happened. The cycling performance was improved by using a Li-plated Cu anode (N=1 mAh cm -2 ), as shown in Fig. 4d. And even better cycling can be achieved by using the Li metal anode (no fading at 0.5 C for 400 cycles) and a graphite anode. The detailed electrochemical profiles are given in Fig. S17.
It should be noted here that both the N/P ratio (R value) and cycling life (C value) are important factors for the practical use of Li metal batteries (LMBs) 53 . Hence, the C/R value has been used for the comparison among different LMB systems 47 . In our PPS-QSSE work, the best C/R value of 96 was achieved by using the HVLCO cathode with high loadings of 3.5 mAh cm -2 and 2 mAh cm -2 Li loadings anode (N/P=0.57), which was more outstanding than most of recently reported LMBs works, especially at high charge/discharge current density, as listed in Table S1 and S2. Based on the calculation model proposed by Xu et al. 47 and Niu et al. 53 , assuming that an 1 Ah pouch cell adopts the same cell parameters and the 1 Ah pouch cell with PPS-QSSE demo fabricated by Ampreus (Wuxi) Co., Ltd. (Fig. S18), the LMB with PPS-QSSE and HVLCO could achieve a high energy density of 1049 Wh L -1 (319 Wh kg -1 ) and power density of ~900 W L -1 , for rough comparison (Fig. 4f). The detailed energy density calculation was given in Table S3. Besides, the NCM523-Li-Cu presented better cycle performance at 0.5C/1C charge/discharge cycling currents, an initial discharge capacity of 138.9 mAh g −1 with a capacity retention of 80% after 90 cycles, as shown in Fig. 6f. However, the C/R value was 45, and the energy density was estimated to be 683 Wh L -1 (268 Wh kg -1 ). The anode-free LNMO-Cu could achieve 602 Wh L -1 (213 Wh kg -1 ), as listed in Table S3. We made a coffee-bag cell for the HVLCO-Li-Cu system and powered a fan.
Interestingly, the cell still regular worked after the cutting in the middle, as shown in Movie S6.

Discussion
Compared to other SSEs systems, there are no serious deficiencies for the PPS-QSSEs, which would be easily translated for industrial roll-to-roll production, not only for the SPEs fabrication but also for the pouch cells production. The only drawback was the insulated gap between two adjacent PPS particles. Hence, future research to improve the performance of PPS-QSSEs based LIBs should focus on: (1) from an engineering point of view, decreasing the porosity of the PPS-SPEs to achieve better Li-ion flux connection (for instance, exhaust-gas disposal of powders after the PTFE fibration process and before the rolling process); (2) surface modification of the PPS particles making the conglutination more readily during the rolling; (3) developing more compatible second phase materials, not only liquid phase but also solid phase, to achieve more effective Li-ion bridge and wider working temperature range; (4) combining with the solvent-free electrodes to decrease the electrode/electrolyte interphase resistance and make the all-solid-state battery.

Conclusions
We have demonstrated a PPS-based SPE with a thin and dense film structure, high intrinsic Li + diffusion coefficient of 1.92×10 -8 cm 2 s -1 at room temperature, and excellent thermal stability fabricated by a solvent-free process in a pilot stage, combining a high-speed air blowing and rolling process. The high conductivity (2.2×10 -4 S cm -1 ) at room temperature was achieved for PPS-QSSEs with a small amount of liquid electrolyte. The Li-ion conduction is mainly due to three paths: the S channels in the bulk of the PPS crystal, the PPS particle surface along the chains, and the gaps infiltrated by the liquid phase. Moreover, the TCBQ with anions-immobilizing effect was introduced to functionalize the PPS particles surface, and the high Li-ion transference number (0.8-0.9) and wide EPW (>5.1 V) were gained due to the strong dipole adsorption between the TCBQ and anions. These outstanding parameters of PPS-QSSEs ensured the homogeneous and dendrites-free Li-plating on Cu at high pre-lithiation current density. Based on the pre-lithiated Cu anode, high loadings HVLCO and NCM523 cathode, and PPS-QSSEs, we demonstrated the fabrication of an LMB with high energy density. Our study paves a new PPS solution for the SPEs, which could push the solid-state battery towards practicality.

Li ion diffusion barrier
The migration barriers of Li ion diffusion in the PPS bulk or surface were calculated using the climbing-image nudge elastic band (CI-NEB) method of Transition State Tools for VASP (VTST) 59,60 . In the CI-NEB process, we first made structural relaxation of the initial and final states, and then interpolated five reaction coordinates between the initial state and final state for searching of the minimum energy path (MEP).
The DFT-D3 correction proposed by Grimme was used for the long-range dispersion, because of its excellent description of van der Waals interactions 61 . The convergence criteria of the NEB calculations for energy and force were set to be 10 −6 eV and 0.05 eV Å −1 , respectively.

Anions-binding effect of TCBQ
The interaction mechanisms of Cl atoms, BF4 or DFOB molecule with TCBQ were demonstrated using first-principles DFT calculations. Spin-polarized total-energy calculations and structure relaxations were performed using the VASP with PAW potential method 54,55 .   The 2A in the denominator accounts for the fact that there are two surface areas of interest from the slab. All the atoms in the slab were allowed to fully relax during the geometric optimization. The bulk energy of PPS crystal was calculated as -73.23 eV unit -1 of PPS. The surface area of (220) and (120) slab was 115.12 and 211.90 Å 2 , respectively.

Fabrication of the PPS-SPE
The The PTFE fibers form a net-like binding structure on the surface of PPS particles getting cotton candy-like powders. The candy-like powders were collected by a gas filter and then hot-rolled at 130 °C to form a free-standing film using a horizontal-type roller (Movie S1). The free-standing film was collected by a winding system. The thermal stability of PPS-SPE was tested by differential scanning calorimetry (DSC) and thermogravimetric analysis (TGA) measurements with temperature range of 30-500 °C at a heating rate of 10 °C min −1 under Ar atmosphere.

Material characterization
The liquid electrolyte contact angle on the PPS-SPE surface was measured by Dataphysics DCAT21.

Cell assembly and Li ion conductivity measurement
Electrochemical studies were all performed using CR2032 coin cells, unless some of high voltage were used for assembling the stainless steel (SS)/SS symmetric cells. The SS|SS cells were kept rest for overnight before the testing to ensure the homogenous penetration of the liquid phase. The CLiBF4 value can be calculated as following: 7 where the film porosities are 9% and 42% for PPS-QSSE and PE separator, respectively.
For the test of temperature-dependent σLi+ from -30 to 70 °C , the σLi+ was examined by EIS and calculated according to the Equation (3)   8 where L is thickness of PPS-QSSE (cm), R is total resistance (Ω), and S is area of stainless steel disc (cm 2 ).
Charge/discharge analysis was performed galvanostatically with an 8-channel battery analyzer (Neware, BTS-5V6A) at room temperature (T=25 °C ). For the HVLCO-Li-Cu cell, a constant-voltage charge at 4.4 V with limitation to 0.02 C was added following the galvanostatic charge procedure. For the NCM523-Li-Cu cell, the constant-voltage procedure was operated at 4.35 V.

Li transference number
The transference number tLi+ can be estimated from the Bruce-Vincent-Evans (BVE) equation 41 , which is evaluated by potentiostatic polarization and EIS using a Li|Li symmetric cell: 9 Where I0 and Iss are the initial and steady state currents before and after the potentiostatic polarization, respectively, ΔV is the potentiostatic polarization of 10 mV, R0 and RSS are the initial and steady state resistance, respectively.

Li-plating
The Li-plating on Cu foil was performed by using gavanostatic charge, and the LiFePO4 (LFP) cathodes were used as the Li source and positive electrode, as illustrated in Figure S9. Solvent-free electrode process was adoped to fabricating the thick LiFePO4 electrode with high-loadings up to 10 mAh cm -2 , as we reported recently 21 . The Li-plating currents from 0.25 to 2 mA cm -2 were used in this work. After the Li-plating procedure, the cell was opened up in the glove box and the LFP cathode was replaced by the HVLCO or NCM523 cathode. Trace of liquid dual-salt electrolyte was added for making the new cell. However, the total liquid dual-salt LiDFOB/LiBF4 electrolyte was controlled at 4 μL mAh -1 considering the Li-plating process.