Engineering Peculiar Cathode Electrolyte Interphase toward Sustainable and High‐Rate Li–S Batteries

The lithium–sulfur battery is considered to be one of the most promising rechargeable energy storage systems because of its high theoretical energy density. Unfortunately, the shuttle effect during cycling causes serious loss of sulfur species and corrosion of the lithium metal anode, resulting in severe capacity decay. This work proposes to completely suppress the shuttle effect of lithium polysulfides (LiPSs) without sacrificing the interfacial Li+ transport, through in situ construction of a compact cathode electrolyte interphase (CEI), which is formed of the reaction between vinylene carbonate (VC), bis(trifluoromethane)sulfonimide ions and LiPSs in a self‐limiting manner during the initial discharge process. Hence, the CEI‐confined sulfur cathode in the VC‐based electrolyte with a solid phase conversion mechanism delivers a long‐term cycling stability and high‐rate performance, as well as excellent performance under an extreme climate in a subzero temperature of −20 °C, limited lithium source with a low N/P ratio of 1.1, and even at mechanical mutilation. The present study reveals an appealing approach to tailor the composition and interfacial structure of sulfur cathodes by in situ construction of a robust, self‐healing, and high Li+ conductive CEI from the aspect of electrolyte, and thus completely solve the issue of the shuttle effect.


Introduction
With the booming market of electric vehicles and the need of environment protection, it is urgent to exploit rechargeable energy storage systems with high safety, high energy densities, and high adaptability to various conditions in practical applications, such as high current densities, extreme climate, and low N/P ratios. [1] However, the state-of-the-art lithium-ion batteries (LIBs) are approaching the limits of their theoretical energy densities, especially, the cathode materials. The lithiumsulfur (Li-S) battery is considered as one of the most promising energy storage systems due to its high theoretical specific energy density, friendliness to the environment and low prices of sulfur. [2,3] In spite of this, Li-S batteries still have some challenges in the face of the practical applications. The main problem is the shuttle effect of lithium polysulfides (LiPSs) based on the dissolution-deposition mechanism, [4][5][6] which will lead to the loss of active sulfur species and severe capacity decay (Figure 1). Furthermore, the kinetics of the LiPSs redox reaction is severely restrained under subzero temperature when the LiPSs dissolve and diffuse in the electrolyte, resulting in a poor electrochemical performance. [7] To solve/mitigate this shuttle effect of LiPSs, an efficient way is to coat a nanoshield on the surface of sulfur composites. [8,9] Using polymers as blocking layers is considered as an effective approach to suppress the shuttle effect of LiPSs. [10][11][12][13] One typical example is coating polyethylene glycol (PEG) on the external surface of CMK-3/S composites, resulting in retarding diffusion of LiPSs, that is, minimizing the loss of the active materials, and promoting the cycling stability. [14] Subsequently, various polymers were employed to encapsulate sulfur composites, [9,15,16] especially, conductive polymers. [17,18] Recently, Cui and co-workers studied different conductive polymers for sulfur coating layers, and demonstrated that the poly (3,4-ethylenedioxythiophene) had the best electrochemical performance. [19] Except polymers, other materials, such as silicon oxides, metal oxides/sulfides, and graphene, were also introduced to wrap sulfur composites. [20][21][22][23][24][25] These results indicated that coating layers can block the shuttle effect of LiPSs to some extent, and greatly improve the capacity retention in comparison to their pristine sulfur compositecounterparts. However, the volume expansion is as large as 80% when sulfur is completely converted to Li 2 S, [26] resulting in destruction of coating layers and leakage of LiPSs. Therefore, it is crucial to find an alternative way to thoroughly suppress the shuttle effect, that is, to inhibit the dissolution-deposition of the LiPSs intermediates.
In situ formation of a thin and robust solid-electrolyte interface (SEI) through electrochemical decomposition of the electrolyte is one of key points for graphite electrodes in the commercial LIBs. The SEI layer can prevent the solvent from inserting into the graphite layers and enhance the stability of graphite electrodes. This concept of in situ formation of thin and robust CEI may be employed to completely suppress the shuttle effect of LiPSs during cycling. In general, two kinds of solvents, ethers and esters, are normally used in the electrolytes for Li-S batteries. Most studies are focused on ether-based electrolytes because the LiPSs are stable and easily dissolved in ether solvents, making it impossible to form CEI layers. While ester solvents, such as ethylene carbonate (EC), propylene carbonate (PC), and diethyl carbonate (DEC), can react with LiPSs to generate organic precipitates constantly, blocking the subsequent electrochemical reaction. [27][28][29][30][31] It provides an inspiration that we can use the awkward reaction to in situ form a thin and compact CEI by choosing proper carbonate-based electrolytes, resulting in suppressing the shuttle effect of LiPSs and maintaining high sulfur utilization during cycling. Thus far, a few studies have used the mixture of ether and ester solvents for cyclo-S 8 based sulfur cathodes with high sulfur content (>50 wt%) to suppress the shuttle effect of LiPSs by forming CEI layers. [32][33][34][35][36][37][38][39][40] Interestingly, the redox reaction route of sulfur changes from the dissolution-deposition mechanism to a solid phase conversion mechanism. However, it should be noted that all of them formed organic-dominated CEI layers, leading to relatively poor rate performance compared to sulfur cathode with dissolution-deposition mechanism.
To solve the above issues, the structure and composition of the in situ formed CEI should be tuned by selecting proper carbonate solvent-based electrolyte, to meet the needs of thickness, compactness and high Li + conductivity. In this study, we are the first to choose a single carbonate ester, vinylene carbonate (VC), as the solvent for Li-S batteries to in situ build a thin, robust, and inorganic-organic hybrid CEI layer between the sulfur composites and the electrolyte, completely blocking the leakage of LiPSs without sacrificing the interfacial Li + transport ( Figure 1). Nuclear magnetic resonance (NMR), transmission electron microscopy (TEM), X-ray photoelectron spectroscopy (XPS), Raman spectra, density functional theory (DFT), molecular dynamic (MD) simulations revealed that VC, as a solvent with low donor number (DN), can fast react with LiPSs in the presence of LiTFSI, to in situ form a uniform, compact, and inorganic-organic hybrid CEI layer. The in situ constructed CEI layer has a unique peculiarity of self-healing after mechanical mutilation. It is like a nanoshield to effectively segregate the sulfur species and electrolyte, and also provides high Li + conductivity. Hence, the CEI-confined sulfur cathode delivers a high specific capacity of 1014 mAh g −1 under 5 A g −1 , a stable specific capacity of 1025 mAh g −1 after 200 cycles at 500 mA g −1 , and 605 mAh g −1 after 600 cycles at 1000 mA g −1 with a 99.9% Coulombic efficiency. Furthermore, it can also stably work at a subzero temperature of −20 °C and a low N/P ratio of 1.1 to satisfy the demands under extreme climate and limited Li resources in the practical application.

Theoretical Understanding of VC-Based Electrolyte
To build a cathode with high sulfur loading for the practical Li-S batteries, the shuttle effect of LiPSs should be completely suppressed. One of the most promising strategies is in situ construction of a CEI layer through the reaction between LiPSs and electrolyte during discharge processes. In order to make this assumption go through successfully, it must meet the following conditions: 1) the reaction should be fast enough to ensure that the consumed sulfur species are as less as possible; 2) in situ formed CEI should be uniform and compact enough to effectively prevent the penetration of electrolyte and isolate the direct contact between electrolyte and sulfur species, as well as adapt to the huge volume change of the sulfur conversion process; 3) the thin and compact CEI layer should be formed in a self-limiting reaction process, and offer a high Li + conductivity for high rate performance.
It is well known that the carbonate solvents can react with LiPSs through a nucleophilic reaction to form insoluble precipitates. [27][28][29][30][31] Figure S1, Supporting Information). It may be attributed to the fact that the cyclic carbonates are easier to be nucleophilicly attracted by the soluble polysulfide intermediates through ring-opening polymerization to form the precipitates. Among the three cyclic carbonates, VC solvent is the fastest one to form a large number of flocculated precipitates when contacting LiPSs. Compared to the molecular structure of EC and PC, the VC molecule has a carboncarbon double bond, which is beneficial to enhance the stability of radicals or anions of the polymer precursor, resulting in a better radical polymerization in VC solvent according to the molecular dynamic simulation ( Figure S2, Supporting Information). Furthermore, all of the carbon atoms in VC molecule show stronger interaction with LiPSs than those of EC and PC molecules, revealing a faster ring-opening polymerization in VC molecule (Figure 2a and Figure S3, Supporting Information). This result can also be explained by the donor number (DN), which is well-known in the Lewis-acid/base theory and always used to measure the electron donating ability, [41][42][43] as shown in Figure 2d and Table S1, Supporting Information. According to the linear relationship between DN and 23 Na NMR shift, the DN of VC is estimated to be 7 ( Figure S4, Supporting Information), which is the smallest among the three carbonate solvents, leading to an easier ring-opening of VC. Interestingly, much more precipitates were observed when added lithium salt, bis(trifluoromethane)sulfonimide lithium (LiTFSI), in the above solution ( Figure S5, Supporting Information). This result indicates that the LiTFSI can accelerate the formation of precipitate and tune the composition of CEI layer, as revealed by DFT calculation (Figure 2b,c). The LiTFSI can decrease the energy of the lowest unoccupied molecular orbital (LUMO) of VC, making it easier for VC molecules to fast react with LiPSs. Thus, VC is quite a suitable solvent for the in situ construction of a compact CEI layer in Li-S batteries ( Figure 1).
A question that should be answered is, can we use the single VC solvent who owns the smallest DN for the electrolyte of Li-S batteries? Here we employed MD simulation to investigate the distribution of average charge and radial distribution function in the VC-based electrolyte. Two other electrolytes, DMEbased (DN = 22) and pyridine-based (DN =30), were chosen as control groups. Figure 2e displays the distribution of average Li charge in total MD process. The value of average Li charge is about 0.80, 0.73, and 0.68 for VC, DME, and pyridine-based electrolytes, respectively. The sequence is in accord with their DN (Figure 2d). This result illustrates that the VC molecules are harder to participate in the solvation of Li ions, that is, there are fewer solvent molecules around the Li ions in the VC-based electrolyte compared to two other electrolytes. It means that the VC molecule has weaker interaction with Li ions, leading to easier transfer of Li + in the VC-based electrolyte as confirmed by the measurement of ionic conductivity in the later part. Furthermore, the radical distribution function from MD simulations exhibits a tendency of longer bond length between Li + and Adv. Energy Mater. 2023, 13, 2300229 the donor atom of the lower-DN solvents, indicating the weaker solvation of Li + in VC solvents (Figure 2f), that is, higher Li + transfer ability in VC system ( Figure S6, Supporting Information). Hence, the electrolyte that employing single VC solvent could be propitious to high performance Li-S batteries.

Structure Characterization
Two difference porous carbon materials were chosen as sulfur hosts, that is, CMK-3 (SBA-15 templated) and MCF-C (mesocellular siliceous foams (MCF) templated, see details in the Supporting Information). Small angle X-ray scattering (SAXS) patterns clearly show that the CMK-3 has a highly ordered mesostructure with a p6mm symmetry, while MCF-C has a disordered mesostructure ( Figure S7, Supporting Information). The morphologies of CMK-3 and MCF-C are rod-like and vesicular structures respectively, with a diameter of 200 nm and length of several micrometers for CMK-3, and 1-2 µm for MCF-C, as revealed by scanning electron microscopy (SEM, Figure S8, Supporting Information). And a various of pores with pore size of 10-30 nm are observed on the surface of MCF-C. The TEM images also show the mesoporous channel arrays with a highly ordered mesostructure of CMK-3 and apparent pore structure of MCF-C ( Figure S9, Supporting Information). In addition, nitrogen sorption isotherms of CMK-3 and MCF-C are performed at 77 K. The Brunauer-Emmett-Teller surface areas and pore volume are calculated to be 1030 m 2 g −1 and 1.35 cm 3 g −1 for CMK-3, 1170 m 2 g −1 and 3.32 cm 3 g −1 for MCF-C ( Figure S10, Supporting Information), respectively.
Sulfur molecules were successfully introduced into the pores of carbon in a quantitative manner through a melting-diffusion way at 350 °C for 6 h and then 155 °C for 6 h, as reflected in the drastic decrease in N 2 uptakes of S-in-CMK-3 and S-in-MCF-C in comparison to the pristine porous carbon ( Figure S9, Supporting Information). The absence of sulfur peaks in both S/C composites confirms the confinement of sulfur within the pores of carbon in their powder X-ray diffraction (PXRD) patterns ( Figure S11, Supporting Information). Additionally, the SEM and TEM images also directly show that there is no sulfur on the surface of both S/C composites ( Figure S8 and S9, Supporting Information). And the exact sulfur loading amounts are analyzed quantitatively by thermal gravimetric analysis (TGA, Figure S12, Supporting Information), 50 and 80 wt% for S-in-CMK-3 and S-in-MCF-C, respectively.

Electrochemical Characterization
aThe electrochemical properties of the S/C composites electrodes were evaluated in coin cell by cyclic voltammetry (CV) and galvanostatic discharge/charge (GDC) test, which were performed within a potential window of 0.5-3.5 V versus Li + /Li under different current densities and electrolytes of 1 and 5 m LiTFSI in VC. Conventional ether-based electrolytes, 1 and 5 m LiTFSI in DME/DOL (1:1, volume ratio), were chosen for control experiments under identical conditions. Figure3a and Figure S13, Supporting Information, display the CV curves of S-in-CMK-3 electrode in both electrolytes at the scan rate of 0.1 mV s −1 . For the ether-based electrolyte, there is a typical dissolution-deposition process of sulfur. However, when the sulfur electrode is operated in the VC-based electrolyte, a small peak of 2.25 V and a large peak of ≈1.50 V appear during the initial reduction process, and the area ratio of those two peaks is 1:16, while in the ether-based electrolyte the ratio is as high as 1:2.4. Interestingly, this small reduction peak disappears and only a pair of reversible redox peaks is observed at subsequent CV curves. The disappeared small reduction peak can be attributed to the in situ formation of CEI layer, that is, CEI-confined sulfur cathode, which can completely block sulfur species leaking out of the sulfur cathode in the subsequent cycles. In other words, only a little sulfur species take part in the construction of the CEI layer. In addition, the redox reaction route of sulfur changes from the dissolution-deposition mechanism to a solid phase conversion mechanism when changing etherbased electrolyte to VC-based electrolyte. Besides, compared to ether-based electrolyte, the Li + participating in the redox of sulfur species must pass through the CEI layer into/from the sulfur-based particles in the VC-based electrolytes. This process produces an extra polarization for the redox of sulfur species, resulting in the choice of wider voltage window of 0.5-3.5 V versus Li + /Li in this study.
The rate capability of the CEI-confined sulfur cathode was tested to evaluate the stability of this 5 m VC-based electrolyte under different current densities, as shown in Figure 3b,c. The corresponding average reversible specific capacities of this electrode is 1924, 1553, 1446, 1332, 1211, and 1014 mAh g −1 at 100, 500, 1000, 2000, 3000, and 5000 mA g −1 , respectively. A stable specific capacity of 1667 mAh g −1 can be reached when the current density is back to 100 mA g −1 . The sulfur utilization is close to 100% even after cycling at large current densities. All the values are much higher than those of the S-in-CMK-3 electrode in ether-based electrolyte at the same current densities (Figure S14, Supporting Information). Especially, a fast capacity decay in the subsequent cycles is clearly observed when the current density returns to 100 mA g −1 in the ether-based electrolyte. It is similar to an open sulfur cathode, that is, continuous loss of the sulfur species during cycling. In addition, only a single plateau for each discharge and charge process is observed at various current densities in the VC-based electrolyte (Figure 3b). We also compare the rate performance of our work with those of the reported cyclo-S 8 based Li-S batteries with solid phase conversion mechanism in Figure 3e and Table S2, Supporting Information, [32][33][34][35][36][37][38] and the CEI-confined sulfur cathode in our work establishes a new benchmark of high-rate performance. This result illustrates that this single VC solvent-based electrolyte can work well at different current densities in Li-S batteries.
To evaluate the cycling stability of the CEI-confined sulfur cathodes, electrochemical performance at various current densities of 500, 1000, and 2000 mA g −1 were measured in the VCbased electrolyte. Figure 3d and Figure  in contrast to only 373 and 297 mAh g −1 for 5 and 1 m etherbased electrolytes. The excellent cycling performance should benefit from the stable electrode/electrolyte interface in the VC-based electrolytes. The GDC curves under different cycles also demonstrate that only a single plateau for each discharge and charge process is observed in VC-based electrolyte, which is in good agreement with the CV results. When the current density increases to 1000 mA g −1 , the reversible specific capacity of the CEI-confined sulfur cathode can reach 605 mAh g −1 after 600 cycles, indicating a great long cycling stability in VCbased electrolyte ( Figure S16, Supporting Information). More importantly, the CEI-confined sulfur cathode exhibits average Coulombic efficiency (CE) of 99.9% at the current density of 1000 mA g −1 , suggesting that the shuttle effect of LiPSs is completely suppressed during the extended cycles. To further increase the current density to 2000 mA g −1 , the initial charge specific capacity is 1250 mAh g −1 and the reversible capacity is 920 mAh g −1 after 150 cycles, corresponding to a capacity reten-tion of 74.0% ( Figure S17, Supporting Information). And there is also just a single plateau for each discharge and charge process even when the current density is up to 2000 mA g −1 after 150 cycles ( Figure S18, Supporting Information). Meanwhile, these results confirm that this VC-based electrolyte can be operated in Li-S batteries with prolonged cycles.
In addition, even though the shuttle effect is thoroughly suppressed in this VC-based electrolyte, the phenomenon of capacity decay is still observed in the prolong cycles ( Figure S16, Supporting Information). In order to identify whether the problem was from the sulfur cathode or the lithium metal anode, we disassembled the cycled battery and refreshed a new lithium metal. Figure 3f exhibits that the reversible specific capacity of CEI-confined sulfur cathode is ≈800 mAh g −1 after 300 cycles, and this value increases to 1100 mAh g −1 when refreshing the anode in the VC-based electrolyte. After 500 cycles, the reversible specific capacity can maintain 880 mAh g −1 , which is much higher than that of the cell Adv. Energy Mater. 2023, 13,   without refreshing anode (658 mAh g −1 at the 500th cycle) in Figure S16, Supporting Information. By contrast, only a little capacity increase is observed when refreshing the lithium anode in ether-based electrolyte, and the capacity fast decays to 185 mAh g −1 after 500 cycles ( Figure S19, Supporting Information). This result illustrates that the lithium metal anode is one of the essential factors of the capacity decay, which is also confirmed by SEM image of cycled lithium metal and the cycling performance of Li||Li symmetrical cell ( Figures S20 and S21, Supporting Information). It may be attributed to the complicated interface or dendritic lithium on the lithium anode after long cycling.

Characterization of Electrolytes
To understand why the single VC solvent-based electrolyte exhibited such an excellent cycling performance, we carried out analysis the electrolytes by various technologies, including Raman spectra, MD calculations, NMR, and Bruce-Vincent-Evans technique. Raman spectra were employed to investigate the coordinated structures of VC and TFSI − at various Li salt concentrations (Figure 4a,b). The Raman spectrum of pure VC in Figure 4a shows a stretching vibration of COC located at 902 cm −1 deriving from free VC molecules, that is, without coordinating to Li + . When LiTFSI was added to 1 m concentration, a new peak appears at 928 cm −1 arising from Li + -solvating VC molecules. In this dilute electrolyte, the solvation structure is as shown in Figure S22, Supporting Information, according to previous reports. [44,45] As the LiTFSI concentration increases, the population of free VC decreases and that of Li + -solvating VC increases ( Figure S23a,b, Supporting Information). Meanwhile, the total peaks' area of free and Li + -solvating VC molecules is decreasing ( Figure S23c 7 Li NMR and d) 19 F NMR of TFSI − in different electrolytes. e) The types of coordination between TFSI − and Li + in various concentrations. Snapshots obtained by MD simulations for f) 1.0 and g) 5.0 m VC-based electrolytes.
existing state is Li + -solvating VC molecules, indicating that most of the VC molecules coordinate to Li + .
Turning to the vibration mode of TFSI − (SO symmetrical stretching) in Figure 4b, a deconvolution analysis shows that the Raman band consists of three peaks at 1126, 1131, and 1137 cm −1 , arising from free TFSI − , contacted-ion-pair (CIP, TFSI − coordinating to one Li + ), and aggregate (AGG, TFSI − coordinating to two or more Li + ), respectively, which is described in Figure S22, Supporting Information. When the concentration increases, the number of TFSI − increases, and the total peaks' area of free, CIP and AGG TFSI − is increasing ( Figure S23f, Supporting Information), and the ratios of AGG/free and CIP/ free increase ( Figure S23d,e, Supporting Information), which is also confirmed by the result of MD calculation (Figure 4e-g). At concentration of 5 m, the majority of TFSI − exists as AGGs and CIPs ( Figure S22b, Supporting Information). It means that most of anions participate in the solvation of Li + , which is beneficial to the formation of an inorganic-organic hybrid structure CEI with robustness and high Li + conductivity.
In addition, the NMR measurements on both VC-and etherbased electrolytes are to study the chemical environment of Li and F (Figure 4c,d). Compared to ether-based electrolyte, the 7 Li peak shifts downfield in VC-based electrolyte, which is indica-tive of weaker Li + -VC interaction, resulting less electron density around Li + . It indicates that less VC molecules are around each Li + , [46,47] which can move faster in this electrolyte. This result is confirmed by the Li + transference number through the Bruce-Vincent-Evans technique, [48,49] as indicated in Figure S24, Supporting Information. The value of Li + transference number is as high as 0.79 in this VC-based electrolyte, while only 0.59 is measured in the ether-based electrolyte, as reflected in the excellent rate performance in the former section (Figure 3b,c). It illustrates that the Li + transfer rate can be easily tuned by using a solvent with low DN. Additionally, 19 F NMR of TFSI − shows the same trends (Figure 4d), that is, a downfield shift in VC-based electrolyte, indicating stronger Li + -TFSI − interaction and weaker solvation ability of VC than that of DME/DOL.

Characterization of Interface
Furthermore, we carried out TEM and XPS to investigate the electrode/electrolyte interface.    (Figure 5f). It indicates that much more LiPSs and VC molecules react to form a relative incompact and thick CEI layer in the dilute VC-based electrolyte, as reflected in the cycling performance (Figure 3d). At 5 m concentration, a uniform and smooth layer, whose thickness is ≈10 nm, is completely covered on the S-in-CMK-3 particles. This result confirms that the CEI layer is first formed during the initial discharge process, that is, the small peak during the initial CV reduction process is related to the formation of CEI layer (Figure 3a). Therefore, it may offer an optimization scheme to characterize the CEI layer through a simple CV technology. We tested CV of S-in-CMK-3 cathodes in various VC-based electrolytes during the initial cycle ( Figure S25, Supporting Information). As expected, with the increasing concentration, the area of the small reduction peak (≈2.25 V) decreases gradually, that is, less amount of LiPSs is involved in the reaction to form CEI layer when the concentration increases. Notably, a linear relationship between the area of this small reduction peak and the concentration is observed ( Figure S25, Supporting Information), which is related to the thickness of CEI layer. It is a direct evidence that the TFSI − anions participate in the forming process of the CEI layer. This result illustrates that the concentration will affect the structure of CEI layer, supported by the TEM analysis ( Figure 5). After 5 cycles in 5 m VC-based electrolyte, the thickness of the CEI layer is almost as same as that after the initial full discharge (Figure 5e and Figure S26, Supporting Information), further illustrating that the CEI layer has been formed during the first discharge process. The CEI layer is a compact and robust layer, like a nanoshield, to block the leaking of LiPSs, which is further verified through the fact that there is no any color change in DME solvent when the discharged 400 mAh g −1 cathode is soaked into it ( Figure S27, Supporting Information). The TEM and CV results reveal that TFSI − anion is a key point to form a thin and compact CEI layer, and it can quantitatively tune the CEI structure by controlling the concentration of electrolyte.
To elucidate the chemical components of the CEI layer, XPS was used to investigate the chemical bonding environments of those species. Figure 5g,h display the deconvolution of C 1s and F 1s spectra of the cathode after the initial discharge process in 5 m VC-based electrolyte, corresponding to the TEM sample in Figure 5e. Except the reference peak at 284.8 eV, the C 1s spectrum shows four peaks located at 286.3, 288.5, 290.0, and 293.0 eV, corresponding to CO, ROCO 2 Li, Li 2 CO 3 , and CF respectively, which are attributed to the decomposition of VC solvent and TFSI − . Compared to the C 1s spectrum of the TEM sample in Figure 5a, much more organic components in the CEI layer are observed in that sample ( Figure S28, Supporting Information). It indicates that much VC molecules participate into forming CEI layer, which is confirmed by the CV result in Figure S25a, Supporting Information, and consistent with previous reports with DME/DOL/VC as solvents. [34,35] This result illustrates that the components and mechanism of CEI layer on sulfur cathodes can be tuned by the concentration of lithium salt.
In addition, there are two peaks located at 684.6 and 688.7 eV in F 1s spectrum (Figure 5h), which correspond to LiF and C-F respectively, verifying that the TFSI − anion participates in CEI formation process. Combined with the result of Raman spectra (Figure 4b), the inorganic-organic hybrid structure of CEI is attributed to the fact that the LiPSs can fast react with the coordinated VC in the structure of CIPs and AGGs ( Figure S22b, Supporting Information), while the TFSI − in the coordination of CIPs and AGGs can also attend the formation of CEI layer, leading to a much more compact CEI in a high-concentrated electrolyte. The thin and compact CEI layer with inorganicorganic hybrid structure suppresses the dissolution and diffusion of LiPSs, supported by the CV analysis in Figure 3a. This result is further confirmed through a visible experiment, that is, dropping Li 2 S 6 solution into pure VC solvent and VC-based electrolyte which contains LiTFSI salt ( Figure S5, Supporting Information). In addition, the inorganic-organic hybrid structure of this CEI layer is beneficial to the high Li + -conductivity and stability. [50,51] Its high Li + -conductivity is strongly reflected by the excellent rate capability (Figure 3c) and CV results with different scanning rates ( Figure S29, Supporting Information). The difference in the composition of CEI layer may be one of the main factors of the difference in the cycling performance ( Figure 3d).

Evaluation the Stability of Interface
To further evaluate the stability of interface at different conditions, we tested the cycling performance of Li-S batteries under high sulfur loading, a wider voltage window, a subzero temperature, and replacing Li foil by lithiated graphite. First, the cycling performance and GDC curves of S-in-MCF-C cathode with 80 wt% sulfur loading are displayed in Figure 6a and Figure S30, Supporting Information. It should be noted that the areal mass loading of sulfur is ≈2.0 mg cm −2 for the S-in-MCF-C cathodes, resulting in an obvious polarization in the first ten cycles. After 145 cycles, the reversible specific capacity can reach 1010 mAh g −1 .
Meanwhile, there are only a discharge plateau and a single charge plateau in the GDC curves. This result indicates that the CEI-confined sulfur cathode with a high sulfur loading can also run well in this VC-based electrolyte. It should be noted that 80 wt% is almost the highest sulfur loading in the solid phase conversion mechanism (Table S3, Supporting Information). Second, when expanding the voltage window to 0.5-4.0 V ( Figure S31, Supporting Information), the CEI-confined sulfur cathode can maintain a reversible specific capacity of 1052 mAh g −1 after 200 cycles (Figure 6b and Figure S32, Supporting Information), indicating that VC-based electrolyte and the CEI layer are stable under such a high voltage.
Besides, the subzero temperature performance of this VCbased electrolyte was assessed at −20 °C, as shown in Figure 6c and Figure S33, Supporting Information. The initial charge specific capacity of CEI-confined sulfur cathode is 868 mAh g −1 , and this value decreases to 580 mAh g −1 after 100 cycles. And a single plateau for each discharge and charge process is observed ( Figure S33, Supporting Information). In contrast, a serious overcharge phenomenon is observed in the etherbased electrolyte under such a low temperature ( Figure S34, Supporting Information). This phenomenon is attributed to the restrained kinetics of the LiPSs redox reaction under the subzero temperature when the LiPSs dissolve and diffuse in the electrolyte. [7] For the VC-based electrolyte, the dissolution of LiPSs is completely suppressed by the in situ formed CEI layer, resulting in a superior low-temperature performance. This result also confirms that the CEI layer is stable under at −20 °C. And it is worth noting that this is the first example of the cyclo-S 8 based sulfur cathode operated in the solid phase conversion mechanism under such a subzero temperature.
In addition, the negative/positive (N/P) ratio is an important parameter in battery design. It will significantly influence the energy density, cycle life and safety. To precisely control the quantity of lithium, we tried to employ a lithiated graphite as lithium source to replace Li metal in this VC-based electrolyte. The capacity of lithiated graphite were precisely controlled through the electrochemical intercalated method. As we know, the capacity of anode is normally 8 to 20% higher than that of cathode in commercial LIBs. Here, we assembled a Li-S battery with a lithiated graphite (2.03 mAh, Figure S35a, Supporting Information) and a CEI-confined sulfur cathode (1.84 mAh, Figure S35b, Supporting Information), in which N/P ratio was as low as 1.1. Figure 6d and Figure S36, Supporting Information, display the cycling performance of this Li-S batteries. The reversible capacity is 800 mAh g −1 after 50 cycles, indicating that the VC-based electrolyte can fit the Li-S batteries with such a low N/P ratio. The excellent performance can be attributed to the stable CEI layer. It is worthwhile to note that this low N/P ratio is the lowest value in reported Li-S batteries, and the limited lithium source greatly improves the energy density of Li-S batteries.
In order to further assess the stability of CEI layer, we tested the self-discharge behavior by running coin cells after 20 cycles, and then resting for 2 days at room temperature ( Figure S37, Supporting Information). It can be clearly observed that the cell with ether-based electrolyte undergoes a capacity loss of 19.4% and discharge plateau decrease of 184 mV. By contrast, both of those phenomena are not observed in the cell with VCbased electrolyte. Moreover, when extending the resting time to 5 and 8 days, the cell with VC-based electrolyte still exhibits a performance without any capacity loss ( Figure S38, Supporting Information). Furthermore, a CEI-confined sulfur cathode was discharged to 400 mAh g −1 at the 21st cycle, rested for 2 days, and then continued to discharge to 0.5 V. The total capacity of both discharge steps in 21st cycle is 1297 mAh g −1 , which is almost the same as the values of the 20th and 22nd discharge capacity (Figure 6e and Figure S39, Supporting Information). This result confirms the stability of the CEI layer, which can completely suppress the leakage of sulfur species. More interestingly, the CEI layer has a unique peculiarity, that is, selfhealing. No self-discharge and only one discharge plateau are observed when the CEI layer is damaged with a mechanical method, such as being pierced through a syringe needle, as shown in Figure 6f and Figure S40, Supporting Information. It indicates that the damaged CEI layer can self-heal quickly and block the leakage of LiPSs promptly, which is consistent with the results of MD simulations ( Figure 2). Besides, to further test the feature of self-healing of the formed CEI layer, we measured the electrochemical performance of the CEI-confined sulfur cathode by galvanostatic discharge/charge in ether-based electrolyte without VC solvent, as shown in Figure S41, Supporting Information. There is an additional discharge plateau of 2.3 V versus Li + /Li after several cycles, which means that the CEI is partly destructed by the expansion in the lithiation of sulfur. These results reveal this CEI layer with advantages of stability, robustness, and self-healing.

Conclusion
In summary, we performed a surface-controlled CEI growth on the sulfur cathode. By employing single VC as the solvent for Li-S batteries, the low DN solvent can fast react with LiPSs in the presence of LiTFSI through nucleophilic and radical polymerization, leading to the in situ formation of a robust, thin, and selfhealing CEI and completely suppress the shuttle effect of LiPSs. The sulfur cathode with this self-healing nanoshield exhibits excellent cycling stability, rate capability, and nearly 100% Coulombic efficiency. Furthermore, the sulfur cathode in VC-based electrolyte with the solid phase conversion mechanism worked well under high voltage (4 V), subzero temperature (−20 °C), high sulfur loading (80 wt%), and low N/P ratio (1.1), even at mechanical mutilation. This work has shed a new light on how surface engineering can be used to tailor the CEI structure of sulfur cathode and completely suppress the sulfur leakage during cycling. Our methodology provides a new perspective of designing electrolytes for high performance Li-S batteries.

Experimental Section
Preparation of Surface Casting Carbon (MCF-C): In a typical synthesis of MCF-C, 20.0 g P123, 650 mL deionized water and 100 mL concentrated HCl (37 wt%) were successively put into a 1000 mL wide-mouth bottle and then stirred for 3 h at 38 °C in water bath. Then 20.0 g mesitylene and 0.23 g ammonium fluoride were added to the above solution. After stirring for 1 h, 44.0 g TEOS was added into the solution and continuously stirred for 24 h at 38 °C. The solution was transferred into a hydrothermal reactor and heated at 110 °C for 24 h, then filtered and dried at 50 °C to obtain P123@MCF. Second, 8.0 g P123@MCF, 120 mL concentrated HNO 3 (65 wt%), and 40 mL hydrogen peroxide aqueous solution (35 wt%) were added into a 1000 mL round-bottom flask successively and heated at 80 °C for 3 h. And then the obtained turbid liquid was dried at 50 °C to obtain MCF after being filtered and washed for three times using ethanol. [52] Third, 25 mg oxalic acid dihydrate dissolved into a mixed solution (FA/ TMB) of 5 mL furfuryl alcohol (FA) and 5 mL trimethylbenzene (TMB), then dropped the mixed solution onto MCF, and polymerized to obtain PFA@MCF by heating. Then, the PFA@MCF composite was carbonized in a tube furnace at 850 °C for 4 h under Ar atmosphere. After cooling down to room temperature, the material was washed with diluted hydrofluoric acid to remove the silica template, and freeze-dried to obtain final product MCF-C. Furfuryl alcohol and P123 were supplied by Sigma-Aldrich. All other reagents were purchased from Sinopharm Chemical Reagent Co. Ltd.
Preparation of S/C composites: The degassed MCF-C and sulfur (Sinopharm Chemical Reagent Co. Ltd) were mixed uniformly in a mortar at a weight ratio of 1:4. Then the mixture was sealed in a Pyrex tube under vacuum and heated at 350 °C for 6 h where sulfur was fully evaporated and got into the pores of MCF-C. Then it was cooled to 155 °C and kept at this temperature for 6 h. In this process the sulfur was condensed in the pores of MCF-C. After cooling down to the room temperature, the S-in-MCF-C composite was obtained. Similarly, the S-in-CMK-3 composite, which mixed the sulfur and CMK-3 (XFNANO, Nanjing) at a weight ratio of 1:1, was also synthesized in the same way.
Preparation of sulfur cathodes and electrolytes: The sulfur cathodes were prepared by mixing S-in-MCF-C (or S-in-CMK-3), Super P, and carboxymethylcellulose sodium with a mass ratio of 70:20:10 to form a slurry. The slurry was coated onto aluminum foil and dried at 55 °C for 12 h under vacuum. The sulfur mass loading ranged from 1.9 to 2.2 and 1.2 to 1.5 mg cm −2 for S-in-MCF-C and S-in-CMK-3 electrodes, respectively. Electrolytes were prepared by dissolving LiTFSI in VC or DME/DOL (1:1, v/v). The molarities were calculated based on the moles of lithium salt and the volumes of solvents. All of the LiTFSI and solvents werepurchased from DoDoChem (Suzhou).
Characterization: PXRD and SAXS were used to investigate the structure of porous carbon and S/C composites. PXRD was performed on a Rigaku SmartLab diffractometer with filtered Cu Kα radiation (λ = 1.5405 Å). SAXS was performed on a Rigaku215 NanoPix instrument equipped with a rotating Cu anode, using a confocal MaxFlux216 mirror under the power of 40 kV, 30 mA. The morphology of porous carbon and C/S composites were characterized by using a field emission scanning electron microscope (FEI Verios-460 and Tescan MIRA) and TEM (FEI Talos F200X). TGA was carried out under a N 2 atmosphere on a DTG-60 (Shimadzu) instrument at a heating rate of 10 °C min −1 to detect the sulfur loading in the composites. N 2 absorbtion/desorption experiments were carried out on an ipore 400 instrument (PhysiChem Instruments Ltd) to track the surface area and porosity changes before and after sulfur impregnation. XPS analyses were performed on an ECSALAB 250Xi highperformance electron spectrometer (Thermo Fisher Company) to study the surficial information of cycled sulfur cathodes. 23 Na NMR (Bruker Avance III HD 400 MHz) was used to estimate DN of VC by measuring 0.01 m solution of NaTFSI in a range of aprotic solvents. 7 Li NMR and 19 F NMR of the electrolytes were also performed on a Bruker Avance III HD 400 MHz NMR and a Bruker AVANCE NEO 800 MHz NMR, respectively, to identify the ion binding environment. The chemical shifts of 7 Li NMR were referenced to 1 m LiCl in D 2 O at 0 ppm and the chemical shifts of 19 F NMR were referenced to 0.1 m 4-fluoronitrobenzene in CDCl 3 at −102 ppm. Each external standard solution was injected into a capillary tube, which was sealed and inserted into an NMR tube containing electrolytes inside an argon glovebox. Raman spectroscopies of VC, LiTFSI, and VC-based electrolyte were performed on a DXR2 instrument (ThermoFisher) with a laser wavelength of 532 nm and laser power of 5 mW.
Electrochemical Measurement: Electrochemical experiments were carried out in CR2016-type coin cells, which were assembled in an argon-filled glove-box using Li foil (or lithiated graphite) as the counter electrode and Celgard 2400 as a separator. The electrolyte-to-sulfur ratio is 20 µL per milligram sulfur. The cells were galvanostatically dischargedcharged on a battery test system (SLAN BT100, Wuhan) between 0.5 and 3.5 V at room temperature. For the low temperature test, the cells were run at a subzero of −20 °C. CV, electrochemical impedance spectroscopy, and Bruce-Vincent-Evans measurements were conducted on a VMP3 potentiostat/galvanostat station (Bio-logic Science Instruments).
Theoretical Calculation: All MD simulations in this work were implemented by using large-scale atomic/molecular massively parallel simulator (LAMMPS) [53] with reactive force field (ReaxFF), [54] in which the ReaxFF used here was developed by refs. [55] and [56], describing successfully the molecules considered in this work. In order to reveal that evolution of systems at room temperature, the NVT ensemble generated by the Nose-Hoover thermostat was used in MD simulations, where the temperature was set as 300 K. Besides, time step and steps of simulations were set 0.25 fs and 4×10 5 , respectively, which was enough to bring the systems in this work into equilibrium.
The DFT calculations of spin-polarization in this article were performed by the Vienna ab initio simulation package [57,58] with the generalized gradient approximation of Perdew-Burke-Ernzerhof. [59] A plane-wave energy cut-off of 520 eV was used in the expansion of the electronic wave function and the convergence threshold was 10 −5 eV in energy and 0.01 eV Å −1 in force for electronic relaxation and ion relaxation, respectively. Density of K point was set 0.06 Å −1 with gamma method for all the calculations.

Supporting Information
Supporting Information is available from the Wiley Online Library or from the author.