A compact inorganic layer for robust anode protection in lithium‐sulfur batteries

Funding information Tsinghua University Initiative Scientific Research Program; National Natural Science Foundation of China, Grant/Award Numbers: U1801257, 21825501, 21676160; National Key Research and Development Program, Grant/Award Numbers: 2015CB932500, 2016YFA0202500 Abstract Lithium-sulfur (Li-S) batteries are one of the most promising candidates for high energy density rechargeable batteries beyond current Li-ion batteries. However, severe corrosion of Li metal anode and low Coulombic efficiency (CE) induced by the unremitting shuttle of Li polysulfides immensely hinder the practical applications of Li-S batteries. Herein, a compact inorganic layer (CIL) formed by ex situ reactions between Li anode and ionic liquid emerged as an effective strategy to block Li polysulfides and suppress shuttle effect. A CE of 96.7% was achieved in Li-S batteries with CIL protected Li anode in contrast to 82.4% for bare Li anode while no lithium nitrate was employed. Furthermore, the corrosion of Li during cycling was effectively inhibited. While applied to working batteries, 80.6% of the initial capacity after 100 cycles was retained in Li-S batteries with CIL-protected ultrathin (33 μm) Li anode compared with 58.5% for bare Li anode, further demonstrating the potential of this strategy for practical applications. This study presents a feasible interfacial regulation strategy to protect Li anode with the presence of Li polysulfides and opens avenues for Li anode protection in Li-S batteries under practical conditions.

bottleneck that limits the overall performance of Li-S batteries as the advancement of S cathode design, especially under practical conditions. [13][14][15] The corrosion of Li metal can be attributed to the following two reasons: (a) Li metal is highly reactive toward almost all nonaqueous solvents and Li salts in nature, generating solid electrolyte interphase (SEI) on the surface of Li anode. 16 Moreover, dendritic Li forms and cracks the fragile SEI during Li plating. 17,18 The rupture of SEI aggravates the side reactions between Li and electrolyte. The nonstopping side reactions deteriorate the utilization efficiency of Li anode and cycling performance of batteries. 19 (b) Soluble Li polysulfides shuttle between S cathode and Li anode, causing severe corrosion of Li anode and loss of active materials in Li-S batteries. 20 The continuous corrosion of Li anode and rapid depletion of electrolyte are the main culprits for the failure of Li-S batteries, especially with high sulfur loadings (>4 mg cm −2 ) and slightly excess Li. 21 Therefore, protecting Li anode from the side reactions induced by electrolytes and Li polysulfides is critical for promoting the further development of Li-S batteries under practical conditions. [22][23][24] A key prerequisite for a Li anode protection layer in Li-S batteries is that Li must be protected in the presence of Li polysulfides compared with the battery systems adopting insoluble intercalation cathodes. [25][26][27] Li polysulfides react with Li and involve into SEI constituents, which complexes the interfacial chemistry on Li anodes and invalidates most of the common SEI design strategies, posing great challenges for effective Li protection. Therefore, constructing SEI that concurrently blocks Li polysulfides and suppresses side reactions with Li is imperative. Generally, SEI on Li anode composes of both inorganic and organic species, as it is the result of the decomposition of both Li salts and organic solvents. 16 SEI can be divided into two layers according to dominant components at different depths. [28][29][30] The organic layer is porous and soluble, making it susceptible to polysulfide attacking. The inorganic layer is compact but conductive, which can be tailored to block Li polysulfides and protect Li anode.
An effective strategy for Li anode protection layer in Li-S batteries is to employ SEI-forming additives, such as LiNO 3 , 31-35 lithium polysulfides, 36-38 SOCl 2 , 39 and sulfurcontaining polymers. 40,41 Additives react irreversibly with Li anode to form in situ protective layer, in which the role of inorganic components is realized. [42][43][44][45] However, the sacrificial nature of the film-forming additives cripples their effects during long-term cycling. Constructing a robust ex situ protective layer emerges as another effective way to protect Li anode. 46 For instance, Chen et al prepared a lithium phosphorus oxynitride (LiPON) layer on Li metal anode using plasmaenhanced electron beam evaporation for anode protection in Li-S batteries 47 and demonstrated that such LiPON layer can block the reactions between Li anode and polysulfides. Zhang et al 48 proposed an implantable SEI for Li anode protection by an electrochemical method prior to cycling, which is compatible with S and high-voltage cathodes with enhanced performance. Nevertheless, the cost and process complexity in previous works have to be taken into considerations for practical Li-S batteries. Therefore, a facile but effective method to construct a protection layer on Li anode is strongly requested to block Li polysulfides and improve the performance of Li-S batteries under practical conditions.
In this article, a compact inorganic layer (CIL) was constructed ex situ on Li anode by the spontaneous reactions between Li and ionic liquid (IL) precursor composed of 0.5 M lithium bis(fluorosulfonyl)imide (LiFSI) dissolved in 1-butyl-3-methylimidazolium nitrate ([BMIM][NO 3 ]) IL ( Figures 1A and S1). IL owns great potential in electrochemical applications due to several unique properties such as negligible vapor pressure, wide electrochemical window, and exceptional thermal stability. 49,50 In contrast to routine nonaqueous electrolytes, the IL precursor consists solely of cations and anions and is solvent free. The absence of organic solvents and the chemical stability of [BMIM] cation  3 ] as IL precursor allows the exposure of Li metal to FSI − anions and superconcentrated (5.76 M by calculation) NO -3 anions. NO -3 is known as an excellent SEI precursor widely applied in Li metal batteries 51,52 and the decomposition of FSI − is proven to form LiF-rich SEI with high surface energy, 53,54 rendering a CIL on Li anode. The as-formed CIL is able to block polysulfides, thus enabling a highly stable Li metal anode even under harsh test conditions. A Coulombic efficiency (CE) of 96.7% was achieved in Li-S batteries with CIL-protected Li (CIL-Li) anode in contrast to 82.4% for bare Li anode while no lithium nitrate was employed. Furthermore, 80.6% of initial capacity is retained in Li-S batteries with CIL-protected ultrathin (33 μm) Li anode compared with 58.5% for bare Li anode after 100 cycles.

| RESULTS AND DISCUSSION
The surface morphology of bare Li anode (denoted as bare Li) and CIL-Li metal was characterized using scanning electron microscopy (SEM). Bare Li exhibits a smooth surface, corresponding to the silver color of Li metal plate ( Figure 1B). After dipping Li metal in IL precursor for 24 hours to form CIL, the color of Li metal plate turns dark and the generation of a surface film is clearly observed ( Figure 1C), which indicates the formation of a conformal layer on Li metal. Cross-sectional SEM view reveals that CIL has a thickness of roughly 400 nm ( Figure 1D) and is tightly adhered to the Li metal anode. The composition of CIL was obtained by the in-built energydispersive X-ray spectroscopy in SEM ( Figure 1E). The fluorine (F) mapping indicates that CIL contains a considerable amount of F, and F element is only found in CIL. Nitrogen and oxygen are concentrated at the top CIL even though there are nitrogen and oxygen in both bulk Li metal because Li metal inevitably reacts with air during sample transfer process. Therefore, CIL mainly consists of inorganic fluorides, nitrides, and oxides, but more detailed chemical information needs to be probed.
X-ray photoelectron microscopy (XPS) with Ar + sputtering was employed to investigate the chemical compositions of CIL along its depth ( Figure 2). Compared with bare Li, an increasing N, O, and F content and low C content indicate the inorganic nature of CIL-Li ( Figure S2). There is only one peak with a binding energy of 534.0 eV at 0 minute for O 1s spectra (Figure 2A), representing the N O or O S bond in the NO -3 and FSI − anions. As the surface etching continues, a 530.9 eV peak for Li 2 O emerges as the reduction product of anions at 0.5 minute and further increases till 2.0 minutes. F 1s spectra follow the same trend ( Figure 2B), exhibiting a F-S peak (688.6 eV) at the surface and strengthening LiF (684.6 eV) peak in the inner part as a result of F S reduction by Li metal. The high mechanical modulus and surface energy of LiF were proven to suppress Li dendrites in previous publications. 55-57 N 1s spectra exhibits four peaks at the surface due to its strong oxidizing nature ( Figure 2C), namely, NO -3 (407.0 eV), N SO 2 F ð Þ -2 (401.6 eV), and their reduction products NO -2 (403.3 eV) and Li 3 N (399.0 eV). NOS -2 (400.1 eV) emerges after 1.0 minute of etching. 58,59 As etching continues, all N species converts to Li 3 N which has an ultrahigh ionic conductivity of 10 −3 to 10 −4 S cm −1 , thus enabling fast Li-ion transport through CIL. 60 Based on these results, CIL is mainly composed of Li 2 O, LiF, and Li 3 N as inorganics and the possible reaction mechanisms of the decomposition of IL precursor in contact with Li metal are also proposed: In order to test the long-term cycling stability of CIL-Li, Li-Li symmetrical cells are assembled using different electrolytes ( Figure 3A). In carbonate-based electrolytes, Li metal anode tends to form massive dendrites and is notoriously unstable ( Figure S3a,b). The voltage hysteresis of 1s, B, F 1s, and, C, N 1s spectra at different etching times. CIL, compact inorganic layer; XPS, X-ray photoelectron microscopy bare Li is relatively stable in the first 80 hours, then sharply increases to 0.6 V followed by short circuit at a current density of 1 mA cm −2 and a capacity of 1 mAh cm −2 . This can be attributed to the dendrite penetration through the separator. However, a stable operation of 160 hours with only slightly increased polarization is observed and dendrites are clearly suppressed with a protective CIL ( Figures S3c,d and S4a). The CIL-Li exhibits a more remarkable performance in ether-based electrolyte under a large current density of 5 mA cm −2 , maintaining a stable polarization of~120 mV for more than 470 hours ( Figure S4b). In contrast, bare Li exhibits an everincreasing polarization until the cell open circuits due to the accumulation of dead Li. The excellent long-term stability of CIL-Li anode is the premise for stable operation of Li-S batteries.
The surface chemistry of CIL-Li after cycling was particularly investigated to gain insights into the reason of its excellent stability. For Li anodes cycled in carbonate electrolyte ( Figure S5), XPS spectra of C 1s from the surface down to 10.0 minutes etching depth are presented in Figure S6a,b.
There are various C-containing species derived from the decomposition of solvents on bare Li surface. This can be attributed to the break, rebuild, and accumulation of SEI as carbonate solvents continuously decompose. By sharp contrast, C 1s signal becomes undetectable after 3.0 minutes of etching for CIL-Li, indicating that solvent decomposition is strongly restrained and CIL effectively protects fresh Li from electrolyte corrosion. For Li anodes cycled in ether electrolyte, the XPS spectra of F 1s of bare Li show Li F and C F bond while no obvious N 1s signal can be detected ( Figure S7a,b). The formation of C F bond can be attributed to the solvent decomposition, indicating that the routine SEI cannot effectively prevent solvent penetration and subsequent side reactions. By sharp contrast, XPS spectra of CIL-Li after cycling exhibit solely LiF and Li 3 N signals without any organic species, indicating that CIL successfully suppresses the solvent corrosion of Li anode and the composition of CIL maintains unchanged ( Figure S7c,d).The inhibition of the side reactions between Li and electrolytes suggests that CIL is compact, chemically, and electrochemically stable in nature, which is essential for blocking Li polysulfides in working Li-S batteries.
A protective layer for Li anode must ensure adequate Li ion diffusion through itself while blocking shuttle species. To further understand the ion transport features of the CIL, a symmetrical cell composed of bare Li as the reference and counter electrode, CIL-Li as the working electrode was assembled and a series of electrochemical tests were conducted. Electrochemical impedance spectroscopy (EIS) before and after cycling were measured ( Figure 3B) and the corresponding Nyquist plots were fitted using the equivalent circuit shown in Figure S8. The initial SEI resistance (R SEI ) of CIL-Li is comparable to bare Li, but the R SEI resistance for bare Li sharply increases from 25.8 to 89.6 Ω after cycling, while CIL-Li only slightly increases from 23.7 to 38.5 Ω. Tafel plots obtained from the cyclic voltammetry (CV) test ( Figures 3C and S9) confirm that the exchange current density of CIL-Li (0.304 mA cm −2 ) is more than two times as bare Li (0.132 mA cm −2 ), which is in accordance with the greater current response observed in the CV test. The reduction of interfacial resistance and high exchange current density signify that CIL possesses a higher Li-ion conductivity compared to routine SEI.
Li anode with CIL protection was further evaluated in working Li-S batteries. The tests were designed to be carried out under two harsh conditions: ultrathin Li foil as anodes and LiNO 3 -free electrolytes. In conventional Li-S batteries test, thick Li metal plates (typically >500 μm) with sufficient Li reservoir are applied and batteries can achieve long cycles even if Li is continuously consumed. The cycling data of such batteries mainly depend on cathode properties and the true electrochemical performance of Li anode is concealed. Therefore, ultrathin Li metal anode (33 μm, 6.6 mAh cm −2 ) was used in all Li-S tests in this work, aiming to evaluate the Once removing LiNO 3 in ether-based electrolytes, Li-S batteries display much deteriorated performance. Therefore, the effect of CIL can be examined without any interference in a LiNO 3 -free Li-S battery. By employing LiNO 3 -free  (Figure 4A), bare Li is incapable of stopping polysulfide shuttle, exhibiting an initial CE of merely 82.2% and typical overcharge ( Figure 4B). The cell fails after 15 cycles along with the depletion of Li anode, which originates from the severe side reactions between Li and polysulfides. CIL-Li demonstrates a high initial CE of 96.7% and an average CE of 94.8% in the following 50 cycles, showing much smaller polarization and minor capacity decay. It can be calculated that an average 0.44 mAh cm −2 of Li is consumed during each cycle for bare Li, taking up as much as ca. 55% of the total cell capacity. For CIL-Li, only an average 0.132 mAh cm −2 Li is consumed irreversibly through each cycle, suggesting that CIL protects Li anode from the corrosion of polysulfides and significantly enhance the anode reversibility. Moreover, CIL-Li anode exhibits superior cycling performance compared to unprotected Li anodes with the presence of LiNO 3 . The introduction of LiNO 3 effectively improves the cycle performance of bare Li at a demanding rate of 2 C, but rapid cell fade still occurs after 30 cycles and the capacity retention by the end of 100 cycles is merely 58.5% with an average CE of 98.6% (Figures 4C and S10). This can be attributed to the breakage of LiNO 3 -induced SEI under large current densities and subsequent polysulfide corrosion. As comparison, CIL-Li retains 80.6% of its initial capacity after 100 cycles and exhibits a high average CE of 98.8%. The superior performances of Li-S batteries under harsh conditions including ultrathin anode, LiNO 3 -free electrolytes, and high-rate cycling firmly verify the high efficacy of CIL in robust Li metal anode protection against Li polysulfides.
The excellent electrochemical performance of Li-S batteries with CIL-Li anode can be attributed to Li polysulfides blocking, which was carefully investigated using quantitative electrochemical method and surface chemistry analysis. The shuttle currents at different potentials were directly probed using constant potential amperometry. A distinct current peak occurs at 2.36 V for bare Li, signifying the strongest polysulfide shuttle induced by the most soluble Li 2 S 6 species at this potential. As comparison, CIL-Li shows no current peak at all potentials, and its shuttle current density at 2.36 V is reduced by six times compared to bare Li anode. The appreciable decrease of shuttle current indicates the suppression of polysulfide shuttle and subsequent inhibition of the side reactions between Li anode and polysulfides, which was clearly proved by the surface chemistry analysis of Li anodes after cycling. XPS of Li anode was carried out after five cycles in LiNO 3 -free Li-S batteries (Figures 4E, F and S11). According to Li 1s and S 2p spectra, bare Li is covered by a series of S-containing species including SO x and LiS x while CIL-Li only exhibits much weaker LiS x peaks. The reduction of SO x species for CIL-Li can be attributed to both the prevention of lithium bis(trifluoromethanesulfonyl)imide (LiTFSI) decomposition and scarcity of LiS x species because it may evolve SO x species through oxidation, and the reduction of LiS x species indicates that less LiS x is generated from the reactions between Li and polysulfides. It can be concluded that CIL effectively blocks the majority of diffusing polysulfides and restrains the side reactions, thus leading to much improved Li-S battery performance.
The blocking of Li polysulfides and the inhibition of Li corrosion induced by CIL were further visualized by SEM. After five cycles in LiNO 3 -free electrolyte, dead Li starts to accumulate ( Figure 5A) and the surface clearly demonstrates two distinct patterns on bare Li metal anode caused by polysulfide attacking ( Figure 5B-D): the mossy dendritic morphology and large numbers of pinholes and cracks. On the opposite, dead Li is reduced significantly and a compact, intact protective layer maintains on the top surface for CIL-Li ( Figure 5E-H). The surface evolution of bare Li and CIL-Li can be depicted by schematics ( Figure 5I,J). During cycling, the surface erosion of bare Li is caused by Li polysulfides, which destroys the original flat surface, undermines the uniformity of SEI, and creates lots of pinholes. Therefore, the formation and growth of Li dendrites prevail, causing rapid accumulation of dead Li and battery fade. CIL-Li anode blocks the majority of the polysulfides and suppresses the corrosion of Li, maintaining the original surface and SEI for uniform Li deposition and finally ensuring Li-S batteries with exceedingly long lifespan.

| CONCLUSIONS
In conclusion, a CIL constructed by employing a feasible pretreatment method was proposed to protect Li anode in Li-S batteries, in particular with ultrathin Li anode. The CIL is capable of blocking Li polysulfide shuttling and suppressing the corrosion of Li due to the abundant inorganic components and compact structure. Therefore, Li-S batteries with CIL-Li anode achieve a CE up to 96.7% compared to 82.4% for bare Li with no LiNO 3 additives. Moreover, the corrosion of Li during cycling is effectively relieved, and 80.6% of the initial capacity after 100 cycles retains for Li-S batteries with CILprotected ultrathin (33 μm) Li anodes in contrast to 58.5% for bare Li anode. This study demonstrates an inspiring route to protect Li anode with the presence of Li polysulfides, which opens avenues for high energy, ultra-stable Li-S batteries under practical conditions.

| Structure characterizations
The morphologies of the Li metal anodes were characterized by SEM (JSM 7401F, JEOL Ltd., Tokyo, Japan) operated at 3.0 kV, 10 μA. XPS spectra were obtained using PHI Quantera SXM (ULVAC-PHI, Inc., Kanagawa, Japan) with an AlK α radiation (pass energy 55.0 eV) at a pressure lower than 10 −7 Torr. All the Li anode samples were washed three times using pure DME to clean the surface residual Li salts and organic solvents. During transferring process before any characterization, all samples were protected in Ar-filled containers to avoid direct contact with air and moisture.

| Electrochemical measurements
The cathode of Li-S batteries in this article was prepared by blade-coating method. The S cathode was composed of S, conductive carbon nanotubes (CNTs), and poly(vinylidene fluoride) (PVDF) binder. Sulfur and CNTs were firstly mixed in a batch with a weight ratio of 7:3. The obtained composite was then mixed with the PVDF binder with a weight ratio of 9:1 in N-methyl-2-pyrrolidone under sonication and stirring to form a homogeneous suspension. The suspension was then coated on an Al current collector and dried at 60 C for 12 hours. All electrodes were dried under vacuum before use and punched into round pieces with a diameter of 13 mm. The mass loading of the S cathode is 1.0 mg cm −2 . Two-electrode cell configuration using standard 2032 coin-type cells was applied in all the electrochemical measurements and assembled in an Ar-filled glove box with O 2 and H 2 O content below 0.1 ppm. The CV measurements were performed within the range of −0.3 to 0.3 V at a scanning rate of 1 mV s −1 . The EIS measurements were performed in frequencies ranging from 10 5 to 1 Hz with a voltage amplitude of 5 mV. CV and EIS tests were conducted using Solartron 1470E electrochemical workstation (Solartron Analytical). The Li-S cells were cycled in galvanostatic mode within a voltage range 1.7-2.7 V, respectively. The Li-S batteries were all first activated at 0.2 C for one cycle prior to the following cycling at more demanding rates. Battery cycling tests were all conducted using LAND multichannel battery cycler (Wuhan LAND Electronics Co., Ltd.).