Ultrahigh Performance All Solid-State Lithium Sulfur Batteries: Salt

Jul 15, 2018 - With a remarkably higher theoretical energy density compared to lithium-ion batteries (LIBs) and abundance of elemental sulfur, lithium...
0 downloads 0 Views 10MB Size
Download PDF
Article Cite This: J. Am. Chem. Soc. 2018, 140, 9921−9933

pubs.acs.org/JACS

Ultrahigh Performance All Solid-State Lithium Sulfur Batteries: Salt Anion’s Chemistry-Induced Anomalous Synergistic Effect Gebrekidan Gebresilassie Eshetu, Xabier Judez, Chunmei Li, Maria Martinez-Ibañez, Ismael Gracia, Oleksandr Bondarchuk, Javier Carrasco, Lide M. Rodriguez-Martinez, Heng Zhang,* and Michel Armand*

J. Am. Chem. Soc. 2018.140:9921-9933. Downloaded from pubs.acs.org by KAOHSIUNG MEDICAL UNIV on 08/08/18. For personal use only.

Electrical Energy Storage Department, CIC Energigune, Parque Tecnológico de Á lava, Albert Einstein 48, 01510 Miñano, Á lava, Spain S Supporting Information *

ABSTRACT: With a remarkably higher theoretical energy density compared to lithium-ion batteries (LIBs) and abundance of elemental sulfur, lithium sulfur (Li−S) batteries have emerged as one of the most promising alternatives among all the post LIB technologies. In particular, the coupling of solid polymer electrolytes (SPEs) with the cell chemistry of Li−S batteries enables a safe and high-capacity electrochemical energy storage system, due to the better processability and less flammability of SPEs compared to liquid electrolytes. However, the practical deployment of all solid-state Li−S batteries (ASSLSBs) containing SPEs is largely hindered by the low accessibility of active materials and side reactions of soluble polysulfide species, resulting in a poor specific capacity and cyclability. In the present work, an ultrahigh performance of ASSLSBs is obtained via an anomalous synergistic effect between (fluorosulfonyl)(trifluoromethanesulfonyl)imide anions inherited from the design of lithium salts in SPEs and the polysulfide species formed during the cycling. The corresponding Li−S cells deliver high specific/areal capacity (1394 mAh gsulfur−1, 1.2 mAh cm−2), good Coulombic efficiency, and superior rate capability (∼800 mAh gsulfur−1 after 60 cycles). These results imply the importance of the molecular structure of lithium salts in ASSLSBs and pave a way for future development of safe and cost-effective Li−S batteries.

■

electrode, SHE), lowest density (0.53 g cm−3), and lightest nature (6.94 g mol−1), lithium metal (Li°) is regarded as a “holy grail” electrode and accordingly has received extensive attention from both the academic and industry communities.4−8 Among all the Li°-based rechargeable batteries (LMBs), lithium−sulfur (Li−S) technologies are considered to be the most auspicious candidates for the next-generation energy storage systems.9−13 This is attributed to their overwhelming benefits such as high theoretical energy density (2600 Wh kg−1) computed on the basis of Li° anode and cyclo-octa sulfur (S8) cathode, abundance of sulfur resources, low cost, and environmental benignancy.9−13 However, the practical deployment, in light of the above beneficial features, is hampered by several intrinsic problems resulting from the complex Li−S cell chemistry, enlisting Li dendrite/mossy growth and possible cell short circuit, electronically insulating nature of S8 (ca. 10−30 S m−1) and of its lower order reduction products (e.g., Li2S, ca. 10−14 S m−1), soluble long chain polysulfides (PS) and their shuttling between cathode and anode, large volume expansion from S8 to Li2S (ca. 80%), etc.9−13

INTRODUCTION In today’s modern and energy conscious society, the quest of energy storage for usage in large-scale applications such as in the electro mobility (xEVs, EVs/HEVs/PHEVs) and renewable or dispersed energy storage has become compulsory.1,2 Amid existing large spectrum energy storage devices, lithium-ion batteries (LIBs) have dominated as key-enabling technologies in the market. However, the existing practical LIBs fall far below the stringent requirements of the above-mentioned demands where a high-energy density, beyond the current capability of LIBs, is needed. Historically, in comparison to computer industry, which enjoyed a doubling in memory capacity every 18 months as indicated by Moore’s law, the rate of progress in energy storage has been quite lethargic.1,2 In the past 150 years, the practical energy density of commercial batteries has increased only by 6-fold, from the first-generation lead-acid batteries (∼40 Wh kg−1) to the contemporary LIBs (∼240 Wh kg−1).3 Thus, though the evolutionary progress has increased the knowledge of acquisition and the degree of battery technology maturity significantly, a new paradigm shift toward the development of new energy storage devices is urgently needed. With its exceptional high specific capacity (3860 mAh g−1), lowest electrochemical potential (−3.04 V vs standard hydrogen © 2018 American Chemical Society

Received: May 2, 2018 Published: July 15, 2018 9921

DOI: 10.1021/jacs.8b04612 J. Am. Chem. Soc. 2018, 140, 9921−9933

Article

Journal of the American Chemical Society In order to address the above-mentioned dilemmas and improve the safety of Li−S batteries, a considerable amount of work has been devoted to all solid-state Li−S batteries (ASSLSBs).14 The use of solid polymer electrolytes (SPEs) surmounts the safety-induced hazards (enlisting both thermal and chemical threats) arising from the highly flammable/ combustible organic ether-based solvents employed in conventional liquid electrolytes.15 The success of SPEs has been well demonstrated by their preliminary application as electrolytes of solid-state rechargeable LMBs for an electric car, Autolib.16 Moreover, the low density of SPEs, compared to other solid electrolytes like garnets, allows a high gravimetric energy density to be readily achieved. Notwithstanding, the practical energy density and cyclability of SPEs-based ASSLSBs are still far from the expectations.14,17 One of the strategies to overcome such interlinking challenges of ASSLSBs lies in the in-depth understanding of the salt anion chemistry and thereby a proper selection of lithium salts for gaining robust SPEs.18,19 The anion chemistry dictates the nature and quality of both Li° anode and S cathode passivation layers, possible interaction/reaction with PS, wettability of electrolytes (i.e., degree of sulfur utilization), and so on. Lithium bis(trifluoromethanesulfonyl)imide (Li[N(SO2CF3)2], LiTFSI) is among the most widely used salt in the electrolytes for Li−S batteries; however, attempts to use LiTFSI in SPE-based ASSLSBs are grappled with the infinite charging, associated with PS shuttling even in the earliest cycles, though the first discharge capacity is as high as 900 mAh gsulfur−1.20 This is due to the inferior quality of the solid electrolyte interphase (SEI) layer formed on Li° anode.20 Lithium bis(fluorosulfonyl)imide (Li[N(SO2F)2], (LiFSI)), an analogue of LiTFSI, confers improved compatibility with Li° anode which is attributed to the formation of LiF-rich SEI layer, thus resulting in a stable cycling performance. Whereas, LiFSI-based cells have lower initial discharge and areal capacities, ascribed to the poor molten-state wettability of −SO2F compared to −SO2CF3 moiety in LiTFSI and/or possibly due to an irreversible reaction with PS.20,21 This shows that both can hardly be used as single electrolyte salts in ASSLSBs. Thus, a synergistic effect combining the beneficial features of both anions (i.e., FSI− and TFSI−) could be of paramount importance to build ASSLSBs with a robust SEI layer, improved discharge/areal capacity, stable long-term cyclability, high Coulombic/energy efficiency, etc. In this work, lithium (fluorosulfonyl)(trifluoromethanesulfonyl)imide (Li[N(SO2F)(SO2CF3)], LiFTFSI), having both −SO2CF3 and −SO2F functionalities (Figure 1), is proposed as an elegant salt anion to merge the complementary advantages of the TFSI− and FSI− in ASSLSBs. Unprecedentedly, Li−S cells using LiFTFSI/poly(ethylene oxide) (PEO) electrolyte can deliver an extremely high specific discharge capacity of 1394 mAh gsulfur−1 (83.2% of the theoretical capacity), a high areal capacity of 1.2 mAh cm−2 with good Coulombic efficiency, and superior rate capability. The possible mechanism of such a superior enhancement in the cycling performance of LiFTFSI-based ASSLSBs is investigated with the help of experimental and computational studies.

■

Figure 1. Comparison of the chemical structures for the LiFTFSI, LiFSI, and LiTFSI salts. corresponding battery-grade lithium salt (such as LiFTFSI (Provisco, Czech Republic), LiTFSI (Solvionic, France), LiFSI (Suzhou Fluolyte, China)) was added. In all the prepared electrolytes, the salt concentration was fixed at the optimized molar ratio of −CH2CH2O− (EO)/Li = 20:1. X-ray Diffraction. Powder X-ray diffraction (PXRD) patterns of the prepared polymer electrolytes were recorded on a Bruker D8 Discover X-ray diffractometer, using λCu‑Ka = 1.54056 Å radiation in the 2θ range from 2° to 80° with a step width of 0.0198°. Thermal Behavior. The phase transitions of the polymer electrolytes were measured on a differential scanning calorimeter (DSC) (Q2000, TA Instruments). A protocol containing two consecutive scans at a cooling and heating rate of 10 °C min−1 in the temperature range from −80 to 150 °C was used. At the two temperature extremes, low (−80 °C) and high (150 °C) ends of each scan, the sample was allowed to stand for 5 min. Thermogravimetric analysis (TGA) was performed on a STA 449 F3 system connected to QMS 403 Aëolos (Netzsch). The samples were heated from room temperature to 600 °C at a heating rate of 10 °C min−1 under argon flow. Ionic Conductivity. The ionic conductivity of polymer electrolytes was measured by AC impedance spectroscopy using a VMP3 potentiostat (Biologic). The frequency ranged from 10−1 to 106 Hz with a signal amplitude of 10 mV. A CR2032 type coin cell using two stainless steel (SS) blocking electrodes (SS | SPEs | SS) was assembled in an argon-filled glovebox and used for the measurement. Anodic Stability. The experiments of the linear sweep voltammogram (LSV) of the polymer electrolytes were performed with a VMP3 potentiostat (Biologic) using a two-electrode cell at 70 °C. Stainless steel (surface area: 0.07 cm2) served as working electrode, and Li disk was used as both counter and reference electrodes. The LSV measurements were carried out between the open circuit potential (OCP) and 6.0 V vs Li/Li+ at a scan rate of 1 mV s−1. Electrochemical Stability of Electrolyte/Li° Electrode. Li° symmetrical coin cells (Li° || Li°) were assembled in an argon-filled glovebox to investigate the electrochemical stability of electrolyte/Li° interphase. The galvanostatic cycling of the Li° symmetric cells was evaluated using a Maccor battery tester (series 4000). The Li° symmetric cells were cycled galvanostatically at a current density of 0.1 mA cm−2, wherein the duration of each half-cycle was 2 h. Surface Morphology and Composition of Li° Deposits. Surface morphologies of the Li° deposits were examined by a field emission Quanta 200 FEG (FEI), operated at 20 kV. The Li° deposits were obtained by the galvanostatic deposition of Li° on Cu substrates using Li | LiX/DME (X = FTFSI, FSI and TFSI) | Cu cells at a current density of 0.1 mA cm−2 for 20 h. The compositions of the surface layer were measured by a Phoibos 150 X-ray photoelectron spectroscopy (XPS) with a nonmonochromatic Mg Kα source (hν = 1253.6 eV). The spectra were recorded with high-resolution scans at low power (100 W,

EXPERIMENTAL SECTION

Preparation of Polymer Electrolyte Membrane. The electrolyte membranes with an average thickness of 50 μm were prepared by a typical solvent casting method, followed by hot-pressing (High Temperature Film Maker Controller, Specac). Briefly, a preweighed amount of PEO was dissolved into acetonitrile, and then the 9922

DOI: 10.1021/jacs.8b04612 J. Am. Chem. Soc. 2018, 140, 9921−9933

Article

Journal of the American Chemical Society

Figure 2. Physicochemical and electrochemical properties of LiX/PEO (X = FTFSI, FSI and TFSI) electrolytes. (a) Optical image of LiFTFSI/PEO membrane. (b) XRD patterns and (c) DSC traces of the three electrolytes. (d) TGA traces for the three electrolytes (top) and their neat components (bottom). (e) Arrhenius plots of ionic conductivity and (f) anodic stabilities at 70 °C for the three electrolytes. 20 eV pass energy, and 0.1 eV energy step). The calibration of the binding energy was performed taking into account as reference the graphitic signal at 284.4 eV. The samples were gently rinsed with 1,2dimethoxyethane (DME) and dried thoroughly under vacuum before transferred to the SEM or XPS chamber by a home-designed airtight setup. Electrochemical Reduction Simulation. Biphenyl was dissolved in ultrapure and dry tetrahydrofuran (THF) for about 40 min. Following, grounded Li° was added to the biphenyl/THF solution in a 1:1 mol ratio (biphenyl:Li). The reaction mixture was stirred for about 4 h at room temperature using a special glass-coated magnetic stirring bar. Afterward, a dark blue color, characteristic of the biphenyl radical anion, was observed. For the two electrons extraction, biphenyl: Li of 1:2 by molar was used, resulting in the formation of a deep dark blue biphenyl radical dianion. Reaction of Polysulfide with Anions. The solution of 0.1 M Li2S6/DME had been prepared by mixing Li2S and S with a mole ratio of 1:5 in DME, followed by stirring for 1 week at room temperature.22 The reaction between polysulfide species and the three investigated anions was carried out in DME solution at room temperature. To a stirred solution of 1 M LiX/DME (X = FTFSI, FSI and TFSI), a

predetermined amount of 0.1 M Li2S6/DME solution with a molar ratio of LiX/Li2S6 of 200 was added. The UV−vis measurements were carried out, without dilution, on a Cary 5000 UV−vis spectrophotometer (Varian). DFT calculations using the Becke’s three parameters (B3) exchange functional along with the Lee−Yang−Parr (LYP) nonlocal correlation functional (B3LYP),23,24 as implemented in the numeric atom-centered basis set all-electron code FHI-aims.25,26 The starting geometries of the investigated PSAs were considered with different conformers, in general, by varying either S−N−S−F or S−N−S−C dihedral angles. The initial geometries were constructed using the open-source molecular editor and visualizer Avogadro.27 The universal force field and genetic algorithm search tool as implemented in Avogadro to prescreen dozens of low-energy geometries was used. Then, their atomic structures using a trust radius method enhanced version of the Broyden−Fletcher−Goldfarb−Shanno optimization algorithm28 were fully relaxed. The “tight” settings, including the “tier2” standard basis set in the FHI-aims code for Li, C, N, O, F, and S atoms, were used, and the following thresholds for the convergence criteria were set: 0.01 eV Å−1 for the maximum residual force component per atom in all structural 9923

DOI: 10.1021/jacs.8b04612 J. Am. Chem. Soc. 2018, 140, 9921−9933

Article

Journal of the American Chemical Society

Figure 3. Electrochemical behavior of Li° electrode in the as-prepared electrolytes. Galvanostatic cycling of Li° symmetric cells at 0.1 mA cm−2 (halfcycle time 2 h) using (a) polymer electrolytes at 70 °C and (b) liquid electrolytes at 25 °C. (c) Galvanostatic polarization of Li symmetric cells at 0.1 mA cm−2 for the liquid electrolytes at 25 °C. relaxations, 10−4 electrons for the electron density, and 10−6 eV for the total energy of the system. S Cathode Preparation and Cycling of Li−S Polymer Cell. Composite sulfur cathode was prepared with 40 wt % elemental sulfur (99.5 wt %, Sigma-Aldrich), 15 wt % conductive carbons (Ketjen Black, KJ600, Akzo-Nobel), and 45 wt % LiX/PEO (X = FTFSI, FSI and TFSI) as electrolyte. The S loading was from 0.9 to 1.1 mg cm−2. The procedure has been detailed in our previous work.20,29 Li−S polymer cells were assembled in an argon-filled glovebox using the prepared electrode as cathode, polymer electrolyte membrane as both electrolyte and separator, and Li metal disk (China energy Lithium) as anode. The cells were then cycled galvanostatically at a constant current (CC) mode between 1.6 and 2.8 V at 70 °C using a Maccor battery tester (series 4000).

semicrystalline nature of the PEO matrix as well as a full solvation of lithium salt in the amorphous region of PEO via the complexion of Li+ cation with the strong electron-donating ethylene oxide units (donor number of 22) and a favorable entropy factor. The obvious endothermic peaks in the DSC traces (Figure 2c) at ca. 65 °C, resulting from the melting transitions of PEO, clearly confirm the existence of crystalline phases in those electrolytes. In addition, both LiFTFSI/PEO and LiTFSI/PEO electrolytes have lower crystallinities (χc) compared to LiFSI/PEO electrolyte (e.g., χc = 54%, 42%, and 59% for LiFTFSI, LiTFSI, and LiFSI/PEO, respectively), due to the sluggish kinetics of crystallization in the presence of −SO2CF3 moieties in both FTFSI− and TFSI−.21 As one of the most important features, thermal stability of SPEs has a pivotal effect on the safety, durability, and long-term stability of ASSLSBs. Individually, the three salts and PEO matrix have decomposition temperatures higher than 200 °C, thus yielding a good thermal tolerance of the corresponding SPEs (Figure 2d). LiFTFSI/PEO with a decomposition temperature at 220 °C (5 wt % mass loss) could sufficiently meet the requirement for its application in LMBs. The temperature dependence of ionic conductivities for the three SPEs shows a notable change at 50−60 °C (Figure 2e), due to the melting of crystallized PEO phases, which is supported by the DSC results where melting transitions around 65 °C are observed. At an operational temperature for Li−S cells, the ionic conductivity of LiFTFSI/PEO electrolyte maintains very close to that of LiTFSI/PEO (e.g., 6.5 × 10−4 S cm−1 (LiFTFSI/PEO) vs 7.0 × 10−4 S cm−1 (LiTFSI/PEO) at 70 °C).

■

RESULTS AND DISCUSSION Physicochemical and Electrochemical Properties of SPEs. Figure 2 displays the fundamental physicochemical and electrochemical properties of LiX/PEO (X = FTFSI, FSI and TFSI) electrolytes at the molar ratio of EO/Li = 20. Thanks to the high molecular weight of PEO matrix, a free-standing and translucent membrane with a thickness of 50 μm could be obtained for the SPEs using the three salts, as exemplified by the optical appearance of LiFTFSI/PEO in Figure 2a. With the increase of salt content from EO/Li = 64 to 16, the LiFTFSI/ PEO membranes become soft and sticky and lose their mechanical strength (Figure S1). The XRD patterns (Figure 2b) of the three electrolytes show only two characteristic peaks, which belong to the crystalline phase of PEO (2θ = 19.3 and 23.7), without any peaks that could be assigned to the corresponding lithium salt. This indicates the 9924

DOI: 10.1021/jacs.8b04612 J. Am. Chem. Soc. 2018, 140, 9921−9933

Article

Journal of the American Chemical Society

Figure 4. SEM images of Li deposited onto Cu substrates at 0.1 mA cm−2 (plating time 20 h) in LiX/DME (X = FTFSI (a, d), FSI (b, d), and TFSI (c, f)) electrolytes at 25 °C.

Coulombic efficiency, poor cyclability, and safety-induced concerns. Hence, a good interfacial stability between Li° electrode and electrolytes is crucial for achieving high performance and safe ASSLSBs.9−14 Figure 3a presents the response of voltage vs time for the Li° | LiX/PEO | Li° symmetrical cells upon continuous galvanostatic cycles. For those three cells with different lithium salts, stable voltage profiles are observed in the first few cycles, due to the formation of SEI layer on the surface of Li° electrode. Both the LiFTFSI- and LiFSI-based Li° symmetric cells demonstrated very stable evolutions of voltage without any erratic values up to ca. 200 h, while LiTFSI-based one fails after being cycled only for 50 h. The same tendency is verified by the galvanostatic cycling of Li° symmetric liquid-based cells, as shown in Figure 3b, where 1,2-dimethoxyethane (DME), a low molecular weight version of PEO, is used as solvent for dissolving the lithium salts and thereby mimics the evolution in the corresponding LiX/PEO electrolytes. The liquid Li° symmetric cells using LiFTFSI and LiFSI salts can be cycled for more than 1400 h, which is 30 times higher than the LiTFSIbased liquid cell. These clearly suggest that the FSO2− functional group can greatly regulate the morphology, composition, and mechanical stability of the SEI layers in both liquid and polymer recipes, thus mitigating the growth of Li° dendrites in the corresponding electrolytes. Interestingly, one may note that LiFTFSI-based Li° symmetric cells in both liquid and polymer electrolytes possess the lowest value of voltage (i.e., over potential) after the first few stabilization cycles, for example, LiFTFSI/PEO (28 mV) < LiFSI/PEO (48 mV) < LiTFSI/PEO (62 mV), which implies that the SEI layer formed on Li° electrode in the LiFTFSI-based electrolytes is less resistive compared to the ones formed in the other two saltbased electrolytes. This can be rationalized by the slightly lower tendency in electrochemical reduction for FTFSI− vs FSI− and

The salt content effect on phase transition and ionic conductivity of the LiFTFSI/PEO electrolytes is systematically studied. The melting transition of PEO turns to be lowered with the addition of LiFTFSI (Figure S2) due to the plasticizing effect of the sulfonamide salts, whereas the optimal ionic conductivity is obtained with EO/Li = 20 and 12 (Figure S3) due to the tradeoff between ion mobility and the number of charge carriers. The sample with EO/Li = 20 is a compromise between the ionic conductivity and mechanical property. The anodic stabilities of the three SPEs are presented in Figure 2f. The linear sweeping voltammetry profiles show relatively low anodic currents of LiFTFSI > LiTFSI (see Table S2 for the comparison of normalized percentage of each F-containing species). That is, LiF is the predominant SEI species in LiFSI, and only trace amount could be detected on the SEI layer originating from LiTFSI-based electrolyte. As reported in literature,32 the differences are ascribed to the bond strength between S−F and C−F. The S−F bond in FSI− is more labile toward reduction compared to C−F in TFSI− or to one side of FTFSI−, and accordingly, LiF is the dominant species. LiFTFSI/DME presents an intermediate tendency in forming LiF, which is in line with its structure where it is endowed half with C−F and half with S−F bond functionality. The electrochemical reduction mechanisms of LiFSI, LiFTFSI and LiTFSI are depicted in Scheme 1. The high and relatively midway amounts of LiF observed for LiFSI and LiFTFSI, respectively, are further verified by the signal at higher binding energy, ca. 690.0 eV, which corresponds either to the trapped pristine salts and/or their incomplete decomposition products. Aiming at substantiating the above Li 1s, C 1s, and F 1s spectral observations and accompanying claims, depth-dependent elemental compositions are given in Figure 5b, and the corresponding XPS survey spectra are also presented in Figure S5. The Li signal depicts an increased concentration value for all three electrolytes with increasing the sputtering depth, due to the removal of the outermost reduction products from salts and solvents. As shown in Figure S6, highly reductive species such as LiF tend to be the pronounced components on Li electrode at

that DME and the above-mentioned salts can be reduced, though their liability toward reduction significantly varies. Figure 5 shows the Li 1s, C 1s, and F 1s spectra harvested from the outermost surface of the SEI layer formed in the three electrolyte systems. The assignments of each peak in the XPS spectra as extracted from literature are summarized in Table S1, and the XPS survey spectra are also presented in Figure S4. The Li 1s binding energy peak at around 57 eV is characteristic of the different Licontaining SEI-building materials such as LiF and could not be that of metallic Li as the binding energy for Li° is found at a lower value (ca. 52.3 eV).32 Armand et al. proposed a twoelectron reduction mechanism of DME leading to the formation of H2CHC−O−CH3, CH3OLi, and LiH (Scheme 1d-1).29 Aurbach and co-workers suggested a two-step two-electron reduction mechanism of DME with CH3OLi and H2CCH2 as a SEI building species.31 According to the authors, the reduction involves first the formation of a Li+ cation stabilized radical anion, which then decomposes to alkoxide, ROLi (CH3OLi), and methoxy ethyl radical. The later in turn undergoes a oneelectron transfer resulting in additional CH3OLi and ethylene (C2H4) (Scheme 1d-2). Hu et al.22 reinforced further the proof for the formation of RO−Li species and oligomers with −OLi end group in DME-based electrolytes on Li° electrode. One of the common features for all tested salts in the C1s photoelectron spectra is the presence of a signal at ∼286.6 eV, attributed to carbon atoms surrounded by one oxygen atom (i.e., C−O), such as CH3OLi generated from the electrochemical reduction of DME (Scheme 1d-1 and d-2). Thus, the spectra confirm the existence of DME-derived contribution to the passivation layer. In the case of LiTFSI and LiFTFSI electrolytes, they present additional broader signals in range of 287.0 to 287.7 eV, which could be ascribed to the presence of R1C−O (H2CHC−O−CH3). Considering the intensity of this peak, the SEI layer formed in LiTFSI/DME can 9928

DOI: 10.1021/jacs.8b04612 J. Am. Chem. Soc. 2018, 140, 9921−9933

Article

Journal of the American Chemical Society

Figure 7. Schematic illustration on the SEI layer formed on Li° electrode in the three salts.

Figure 8. Stability of salt anions against polysulfide species: (a−d) the appearance of 1 M LiX/DME (X = FTFSI, FSI and TFSI) and blank DME solution before and after the addition of PS at room temperature for 60 h. (e) Normalized UV−vis absorption spectra of the PS-added solutions. (f) DFT calculations for the proposed intermediates. Red, yellow, light blue, dark blue, gray, and pink balls in (f) stand for O, S, F, N, C, and Li atoms, respectively.

the inner part of the SEI layers. Moreover, the atomic concentration of Li is found to be higher than that of fluorine

(F 1s), implying that most of the lithium is neither bound to sulfur (S 2p) nor to nitrogen (N 1s) containing SEI species, and 9929

DOI: 10.1021/jacs.8b04612 J. Am. Chem. Soc. 2018, 140, 9921−9933

Article

Journal of the American Chemical Society

Figure 9. (a−c) Discharge/charge profiles of the Li−S cells using LiX/PEO (X = FTFSI, FSI, and (FSI)0.5(TFSI)0.5) electrolytes at 70 °C. (d) Rate capability of those three electrolytes. (e) Cycling stability of the Li−S cells using LiX/PEO (X = FTFSI and FSI) electrolytes. (f) Cycling performance of the LiFTFSI/PEO-based cell at a discharge/charge rate of 0.5/0.5C. (g) Comparison of the cycling performance for ASSLSBs reported in literature and the results of this work. The numbers in squares correspond to the references listed in Table S4, and color code corresponds to type of electrolyte.

this upholds well with the low concentrations of the respective spectra. This once again testifies that the passivation layer in the case of imide salts having a S−F bond is dominated by LiF, as a principal building material. In summary, one can assume that the SEI layer in LiFTFSIbased electrolyte contains an optimum amount of inorganic species (mainly LiF), a prerequisite for a mechanically and electrochemically stable SEI layer. This could explain the improved stripping−plating and most importantly the less resistive behavior of the passivation layer. Mechanistic Understanding of the Anions Chemistry on Li° Electrode. Targeting at simulating the above claimed electrochemical reductions of the salts on Li° surface and thereby substantiate the observed XPS accounts, chemical

simulations of the neat electrolyte salts utilizing biphenyl radical anions as a reducing agent were conducted.32 Addition of a single electron (e−) to a neutral biphenyl molecule generates a radical anion via reduction, with a negative charge and an unpaired e− on it (Figure 6). Further injection of an e− to the biphenyl radical anion results in the formation of biphenyl diradical anion. Biphenyl radical and diradical anions are estimated to have reduction potentials of ca. 0.40 and 0.15 V vs Li/Li+, respectively, and thus, they can be used as powerfulreducing agents. Figure 6 clearly depicts that while LiFSI and LiFTFSI get easily reduced starting at 1.1−1.0 V vs Li/Li+, LiTFSI presented huge resistance and can only get reduced at a very low potential close to that of metallic Li° (