Effects of Polysulfide Solubility and Li Ion Transport on Performance of Li–S Batteries Using Sparingly Solvating Electrolytes

Sparingly solvating electrolytes are an emerging class of electrolytes used in Li – S batteries. In this type of electrolytes, polysul ﬁ de dissolution and shuttling can be suppressed, resulting in high Coulombic ef ﬁ ciency and cycle life. To optimize the electrolytes for high energy density cells, effects of polysul ﬁ de solubility and Li ion transport properties on Li – S battery performance were investigated for tetraglyme (G4)-based solvate ionic liquids [Li(G4) x ][TFSA] and a sulfolane (SL)-based concentrated electrolyte [Li(SL) 2 ][TFSA], which are both diluted with a hydro ﬂ uoroether (HFE). The Li 2 S 8 solubility is low (1 mM in atomic S concentration) in [Li(G4) 0.8 ][TFSA] − 4.3HFE and [Li(SL) 2 ][TFSA] − 4.0HFE. Cells with [Li(SL) 2 ][TFSA] − 4.0HFE exhibited better rate capability despite their lower ionic conductivity. The higher transference number ( t Li + ) of [Li(SL) 2 ][TFSA] − 4.0HFE may predominantly contribute to the rate performance, rather than polysul ﬁ de solubility and ionic conductivity. Furthermore, [Li(SL) 2 ][TFSA] − 4.0HFE demonstrated an initial discharge capacity of 1130 mAh g − 1 at a low electrolyte volume to sulfur weight ratio of 4, whereas a typical organic electrolyte failed to achieve such a high capacity owing to limitations of the redox mechanism mediated by dissolved polysul ﬁ des. In addition to the low solubility of polysul ﬁ des, the high t Li + is crucial for achieving high energy density Li – S batteries by reducing the electrolyte amount.

Li-S batteries have attracted attention as a future energy storage system that transcend the limits of Li-ion battery technology. High theoretical energy density (2600 W h kg −1 ) that relies on the high capacity of sulfur (1672 mA h g −1 ), low cost, and natural abundance of sulfur render Li-S batteries a preferred choice. However, practical applications of Li-S batteries are still limited by bottlenecks arising from the chemistry of sulfur active materials, such as the inherent insulating nature of elemental sulfur and its discharge products, large volume change during discharge-charge cycles, and dissolution of lithium polysulfides (Li 2 S m ) into electrolytes. [1][2][3][4] Among the issues above, the dissolution of Li 2 S m causes the polysulfide shuttle effect, resulting in a rapid capacity decay, low Coulombic efficiency, and severe self-discharge. Therefore, many studies have been conducted to manage polysulfide dissolution and shuttling, including seeking efficient sulfur host materials that can trap polysulfides and retard the shuttling effect. [2][3][4][5][6] Additionally, effects of electrolytes has been intensively studied in terms of solvent and additives, or type and concentration of Li salts. 1,7 The passivation of Li metal anode using LiNO 3 additive proved to be effective in withstanding the side reactions between dissolved polysulfides and Li metal anode in a mixed ether-based organic electrolyte comprising dimethoxyethane (DME) and 1,3-dioxolane (DOL). 8 However, dissolution in organic electrolytes remains unresolved, and the passivation effect is transient and imperfect because LiNO 3 is continuously consumed during charge-discharge cycling.
Recently, the use of sparingly solvating liquid electrolytes has been proposed to address polysulfide dissolution, an alternative to the indirect Li-passivation approach. 7 The thermodynamic suppression of polysulfide dissolution has been accomplished in weakly coordinating electrolyte solutions, such as ionic liquid-based electrolytes [9][10][11] and highly concentrated electrolytes. 12,13 In previous studies, we reported that an equimolar mixture of tetraglyme (G4) and lithium bis(trifluoromethanesulfonyl)amide (LiTFSA), [Li(G4) 1 ][TFSA] classified as solvate ionic liquids (SILs) inhibited polysulfide dissolution significantly. 14 Moreover, the addition of 1,1,2,2-tetrafluoroethyl 2,2,3,3-tetrafluoropropyl ether (HFE) to [Li(G4) 1 ][TFSA] yielded lower polysulfide solubility while enhancing ionic conductivity, resulting in higher capacity and Coulombic efficiency. 15 Kaskel et al. reported that an electrolyte solution consisting of a sulfolane (SL)/HFE mixture containing 1.5 M LiTFSA achieves low polysulfide solubility and enables high discharge capacities and high Coulombic efficiency on pouch cell level with very low electrolyte amount. 16,17 Our recent study revealed that a 2:1 molar ratio mixture of SL and LiTFSA (denoted as [Li(SL) 2 ][TFSA]) and its diluted solution with HFE exhibited low polysulfide solubility comparable to that of [Li(G4) 1 ][TFSA]-based electrolytes, and that Li-S cells using low sulfur loading cathodes (∼1 mg cm −2 ) exhibited better rate capability for SL-based electrolytes. 18 This was interpreted in terms of the unique Li ion hopping/exchange conduction behavior in SL-based electrolytes, where Li ions diffuse the fastest among the diffusive species of the solvent and ions. 19 High sulfur loading and low electrolyte volume have been recognized as essential factors for achieving high energy density in Li-S batteries to rival or exceed the energy density of state-of-theart Li ion batteries at the cell level. 20 For example, previous studies have reported the increase in sulfur loading up to 16 mg cm −2 using three-dimensional current collectors. [21][22][23] However, not many studies have reported high capacity Li-S batteries with high sulfur loading under lean electrolyte conditions. [24][25][26][27] This is owing to an underlying redox mechanism that relies on the dissolution, disproportionation, and mediation of polysulfide species in the typically used DME/DOL organic electrolyte (dissolution-disproportionationprecipitation pathway): electrolyte properties such as viscosity depend strongly on polysulfide dissolution in smaller amounts of the electrolyte, resulting in a more pronounced degradation of battery performance. 28,29 In this context, sparingly solvating z E-mail: mwatanab@ynu.ac.jp = These authors contributed equally to this work. *Electrochemical Society Member. **Electrochemical Society Student Member. electrolytes can separate the volume of an electrolyte from sulfur electrochemical reactions, as the overall reactions primarily proceeds in a solid-solid pathway owing to the significantly low solubility of Li 2 S m . 7 In this study, we determined the polysulfide solubility and Li ion transport properties for [Li(G4) x ][TFSA]-yHFE (x = 0.8, 1, 1.5, y = 4.3, 4.0, 3.4) and [Li(SL) 2 ][TFSA]−4.0HFE and highlighted the importance of high Li ion transport properties for achieving high energy density Li-S batteries; furthermore, we studied their effects on the charge-discharge behavior and rate capability of the Li-S cells with relatively high sulfur loading (∼3 mg cm −2 ) and different electrolyte amounts to obtain sparingly solvating electrolytes that can yield a high capacity of > 1000 mAh g −1 .
Sulfur/carbon (S/C) composites were prepared by the meltdiffusion method. 30 Ketjen black (KB, EC600JD, Lion Corp., Tokyo, Japan) and elemental sulfur (S 8 , Wako Pure Chemical Industries, Ltd., Osaka, Japan) were mixed using an agitating mortar. The mixture was transferred to a vial and heated at 155°C for 6 h to ensure the impregnation of sulfur into the pores of KB. Carboxymethyl cellulose (CMC 2200, Daicel FineChem Ltd., Tokyo, Japan) was used as a binder. An aqueous slurry of the S/C composites and CMC binder was applied into an Al foam sheet (Aluminum-Celmet, Sumitomo Electric, Osaka, Japan) to achieve high sulfur loading (> 3 mg cm −2 ). 31 The composite cathode was dried in an oven at 60°C for 12 h and cut into a circular disk. The mass ratio of the composite cathode was sulfur/carbon/binder = 6:3:1. The composite cathode was compressed to ∼200 μm thickness using a roll press before cell assembly. A typical cathode density was calculated to be 0.15 g cm −3 with the Al foam current collector.
Measurements.-The solubility of long-chain polysulfides (nominally Li 2 S 8 ) was measured according to previous studies. 15 Saturated solutions of Li 2 S 8 were prepared by mixing S 8 and Li 2 S at a molar ratio of 7:8. The solutions were stirred at 60°C for 3 d, and then stored for 1 d at room temperature. The supernatant solution was filtered and diluted with 1 M LiTFSA in G4 solution. The dissolved Li 2 S 8 in the solutions were electrochemically oxidized to S 8 in a H-type two-electrode cell with a carbon cloth as a working electrode, lithium foil as a counter electrode, and a lithiumconducting ceramic (LICGC, Ohara) as a separator. The maximum absorption of S 8 oxidized from the polysulfides was recorded using a UV-vis spectrophotometer (UV-2500PC, Shimadzu). The saturated solubility was determined from the absorbance, dilution factor, and calibration curve.
Ionic conductivity (s) was measured by the complex impedance method using VMP3 (Biologic) in the frequency range of 500 kHz-1 Hz with a voltage amplitude of 10 mV using a twoplatinized-platinum-electrode cell (CG-511B, TOA Electronics). Viscosity (η) was determined using a viscometer (SVM 3000, Anton Paar).
The very-low-frequency impedance spectroscopy (VLF-IS) of a Li/Li symmetric cell (2032-type coin cell with a polyethylene separator) was performed to study Li + transference number, as reported by Wohde et al. 32 The VLF-IS measurements were performed using ModuLab XM (Solartron Analytical). Before the measurements, impedance spectra were obtained continuously in an intermediate frequency range from 1 MHz to 0.1 Hz to monitor the change in interfacial impedance. After the interfacial impedance became constant (typically, 24-48 h), the VLF-IS measurement was performed in a frequency range from 1 MHz to 0.1 mHz with an amplitude of 10 mV for measuring the finite diffusion impedance of Li ions in the cells.
For charge-discharge tests, coin cells (2032-type) comprising the S/C composite cathode, microporous polypropylene separator (thickness: 25 μm, porosity: 55%), Li metal foil anode, and electrolyte were assembled in an Ar-filled glovebox. The ratio of the electrolyte volume to sulfur weight is denoted as E/S [μL -electrolyte /mg -sulfur ] according to literature. 33 Galvanostatic charge-discharge tests on the Li-S cells were performed in the voltage range of 1.0-3.3 V for the G4-and SL-based electrolytes. Meanwhile, galvanostatic charge-discharge tests on the Li-S cells were performed in the voltage range of 1.8-3.3 V for the DME/DOL electrolyte at 30°C to avoid the reductive decomposition of LiNO 3 additive at the S/C cathode. The charge-discharge cycle was initiated from the discharge process. The specific capacity of the cell was calculated based on the mass of sulfur. The Coulombic efficiency for each cycle was defined as (discharge capacity)/(charge capacity). Galvanostatic intermittent titration (GITT) measurements were performed with a constant discharging current of 50 μA cm −2 for 1 h with each, and the relaxation time between pulses was set to which the time dependence of the cell voltage was 15.0 mV h −1 .

Results and Discussion
The electrolytes were prepared with different compositions of solvents (G4 or SL), LiTFSA, and HFE, denoted as  various polysulfide species of Li 2 S m (2 ⩽ m ⩽ 8) as well as the intended Li 2 S 8 because of the disproportionation reactions in the solutions, the solubility was found to be the highest at this composition among the mixtures of S 8 :Li 2 S forming Li 2 S m (2 ⩽ m ⩽ 8). 11 The solubility in the DME/DOL electrolyte is shown in Fig. 1 for comparison. The solubility in the DME/DOL electrolyte (6600 mM_S) is significantly high, i.e., in orders of magnitude higher than the other electrolytes. A similar level of solubility has been reported for an ether-based organic electrolyte, i.e., 0.98 M LiTFSA in G4 (6050 mM_S). 34 In previous studies, for the [Li(G4) 1 ][TFSA] SIL electrolyte, the percentage of G4 molecules that remains uncoordinated was estimated as only a few mol%, 14 and the scarcity of the uncoordinated G4 was suggested to be responsible for the low solubility of Li 2 S 8 (61 mM_S). 15 Our recent study indicated a similar scenario for [Li(SL) 2 ][TFSA] (52 mM_S). 18 With the addition of nonpolar HFE to these concentrated electrolytes, the solubility further diminishes, while the Li + coordination structure remains unchanged. 15 Here, the relatively high S loading (∼ 3 mg cm −2 ) electrode was used for the charge-discharge tests measured at slow charge (50 μA cm −2 ,1/100 C) and discharge (100 μA cm −2 , 1/50 C) rates under electrolyte-rich conditions of E/S = 10. The charge-discharge curves of all the cells showed a typical two-step discharge behavior, which has been assigned to the conversion of S 8 to longer polysulfides (higher voltage plateau) followed by a further reduction of longer polysulfides to lower-order polysulfides such as Li 2 S 2 and Li 2 S (lower voltage plateau). 3 The initial discharge capacity of the cell with the DME/DOL electrolyte nearly reached the theoretical limit (∼1500 mAh g −1 ), while the cells with the sparingly solvating electrolytes delivered lower initial capacities of 1100 and 1200 mAh g  Fig. 3 to demonstrate the effects of electrolyte composition on battery performance. The initial discharge capacity exceeded 1000 mAh g −1 at x = 0.8 and 1.5 under the same charge-discharge conditions shown in Fig. 2. However, the second charge capacity became higher than the initial discharge capacity at x = 1.5, suggesting that polysulfide shuttling was not retarded in this electrolyte with the Li 2 S 8 solubility of 760 mM_S. However, at the intermediate x = 1, the second plateau reaction disappeared in the discharge curve and the capacity declined markedly to 270 mAh g −1 .
It appears that the difference in polysulfide solubility between x = 0.8 and 1.5 (Fig. 1) (Fig. 4). In the GITT measurement, open-circuit voltage and discharge potential with current flow can be measured alternately to discriminate the thermodynamic and kinetic potentials during charge and discharge processes. In the GITT profiles, the cells with  37 These results suggest that the electrochemical reduction of sulfur proceeds similarly in these electrolytes, and this range of difference in the polysulfide solubility did not affect the thermodynamic reaction pathway (through the non-negligible dissolution of polysulfides) for the S/C electrode. However, the cells with [Li(G4) 0. 8

][TFSA]−4.3HFE and [Li(SL) 2 ][TFSA]−4.0HFE
showed a thermodynamic potential lower than 2.3 V for the first plateau, which was attributable to the formation of less stabilized reduction products of Li 2 S m in the sparingly solvating electrolytes. 34 Moreover, the voltage difference between the first and second thermodynamic plateau voltages was smaller for these cells. These thermodynamic voltage profiles were similar to the reported one for an acetonitrile (ACN)-based solvate electrolyte mixed with HFE (1:1 by volume, [Li(ACN) 2 ][TFSA]-HFE), in which the second thermodynamic plateau voltage was comparable to or even higher than the first plateau voltage (∼2.2 V). 37 Based on operando X-ray diffraction and X-ray absorption spectroscopy results for the unique discharge behavior in [Li(ACN) 2 ][TFSA]-HFE electrolytes, Nazar and Balasubramanian et al. proposed a complicated quasi-solid-state reaction pathway including two electrochemical reactions, 1/2S 8 + 2Li + + 2e − ⇄ Li 2 S 4 (first plateau) and 1/2Li 2 S 6 + Li + + e − ⇄ Li 2 S 3 (second plateau) combined with redox mediation and disproportionation reactions that consumed S 8 and longer polysulfides to form Li 2 S. Although operando measurements were not performed for the present systems, the discharge reaction was expected to proceed in a quasi-solid-state reaction pathway, given that   electrolyte from that of the G4-based electrolytes indicating a capacity of ∼200 mAh g −1 at the first plateau. The cause has not been elucidated. Further investigations on operando measurements and/or kinetic modeling are required to elucidate the difference in the first plateau capacity between these two sparingly solvating electrolytes.
As revealed by the GITT profiles, a prominent voltage decrease at the end of the first discharge plateau was resulted from a purely kinetic contribution. It has been suggested that the voltage "dip" between the first and second plateaus is associated with the overvoltage required for the nucleation, initial growth, and precipitation of insoluble solid products during discharge. Chiang et al. demonstrated that the polarization at the dip was more pronounced in a smaller amount of DME/DOL electrolyte, where Li 2 S m dissolution is more limited. 38  The lithium transference number (t Li + ) was determined using the VLF-IS of a Li/Li symmetric cell as listed in Table I, and Nyquist plots of the impedance spectra and their analysis results are shown in Fig. S1 and Table SI (available online at Table I. We confirmed that [Li(G4) 1 ][TFSA] demonstrated a low value of t Li + = 0.03 (Table SI), which is consistent with the t Li + value (0.026) reported by Wohde et al. 32 The significantly low t Li + for [Li(G4) 1 ][TFSA] has been interpreted as a result of multiple factors, including the scarcity of the uncoordinated G4 molecules, constraint of momentum conservation for the longlived [Li(G4) 1 ] + cation and TFSA anion, and resulting anticorrelated motions of the ions under the anion blocking condition in the polarized Li/Li symmetric cell.
As predicted in a MD simulation study, 39 the presence of uncoordinated solvents (G4 in this case) in [Li(G4) 1 [TFSA]−4.0HFE, respectively. This was likely owing to the efficient Li ion transport via a dominant Li ion hopping/exchange mechanism in the SL-based concentrated electrolytes, as reported in our previous studies. 19 The presence of HFE may fragment the unique chain-like Li ion coordination structure by SL and TFSA to smaller clusters, leading to a reduced contribution of Li ion hopping to overall Li ion conduction. This may be responsible for the reduction of t Li+ in the presence of HFE. 18 Finally, the significantly low t Li + value can be considered as a factor responsible for the large overvoltage at the dip in [Li(G4) 1 ][TFSA]−4.0HFE. Upon discharge, Li ions deplete more immediately at the S/C cathode surface, and steeper Li salt concentration gradients would form in the electrolyte with lower t Li + . In such a situation, the discharge reaction is more susceptible to the diffusion-limited process of Li ions. Studies on kinetic modeling for Li-S cells have shown that the discharge overvoltage is governed by the mass transfer of reactants in addition to the passivation of the cathode surface by Li 2 S and/or Li 2 S 2 . 43,44 Compared with the first reduction process, the second reduction process (associated with nucleation and growth of lower-order polysulfides) is affected more by the mass transfer of Li ions because more Li ions are required to form Li 2 S and/or Li 2 S 2 from higher-order polysulfides. Owing to the diffusion-limited process, the formation of Li 2 S (that causes 80% volume expansion) can be localized at the porous cathode surface closer to the electrolyte, which may passivate the electrode or further inhibit Li ion transport into the porous electrode. In addition, the charge transfer kinetics of the second reduction process may deteriorate when a large concentration gradient of Li salt is developed after the first reduction process. Consequently, severe concentration polarization may be a possible factor for the observed overvoltage at the dip in [Li(G4) 1 ][TFSA]−4.0HFE.
The importance of t Li + in the Li-S battery is highlighted in the rate capability tests. Figure 5 shows the charge and discharge curves of the cells at different discharge C rates. The cell with the DME/ DOL electrolyte showed good rate performance for discharge capacities exceeding 1100 mAh g −1 up to 1/25 C. However, the second plateau disappeared at current densities higher than 1/12 C (400 μA cm −2 ). For the sparingly solvating systems, the cells with  (Fig. 3). For [Li(SL) 2 ][TFSA]− 4.0HFE with a lower σ (0.7 mS cm −1 ) but higher t Li + (0.48), the second plateau did not disappear even at 1/6 C (800 μA cm −2 ), albeit a larger overvoltage and lower discharge capacity observed at a higher C rate. A good cycle stability of the cell with  S2). Therefore, we believe that the Li ion transport property, characterized herein by t Li + , dominates the rate performance of Li-S batteries rather than other electrolyte-derived parameters, such as polysulfide solubility, ionic conductivity, and viscosity.
An important challenge for achieving high energy density is to decrease the electrolyte amount in the cell. It has been reported that the discharge capacity and cycle life decline with decreasing E/S value. 28,45,46 This has been rationalized by the decreased conductivity and poor electrodeposition of Li 2 S, 38 associated with the higher concentration of dissolved polysulfides at lower E/S. We elucidated the effects of E/S on the discharge capacity for the sparingly solvating electrolytes and compared the result with that for the common DME/DOL system. In Fig. 6a, the fraction of potentially soluble sulfur species in the cathode was estimated as a function of E/S, based on the saturation solubility (Fig. 1). After the first reduction process at the initial discharge in the DME/DOL, 100% of sulfur species in the cathode can be solubilized when E/S is larger than ∼5, and the polysulfide concentration increases with decreasing E/S. The concentration of polysulfides reaches the solubility limit at E/S ∼4.74, and a further reduction in E/S would lead to the precipitation of Li 2 S m . As shown in Fig. 6b, the initial discharge capacity decreased linearly with decreasing E/S to 5, likely owing to the gradual decrease in the transport property of the DME/DOL electrolyte. At E/S = 4, the cells only delivered the significantly low capacity featured by the severe overvoltage and disappearance of the second discharge plateau (Fig. S3a), and this may originate from the limited mass transport of polysulfides in a saturated condition. 38 These results agree well with the reported critical E/S value of 4-5 for the DME/DOL electrolyte. 38,[46][47][48] For the sparingly solvating electrolytes, the amount of potentially soluble sulfur species in the cathode is minute in this E/S range (e.g., 0.032% at E/S = 10). Hence, Li-S battery performance is expected to be less dominated by E/S. A high capacity exceeding 1130 mAh g −1 was achieved with [Li(SL) 2 ][TFSA]−4.0HFE at E/S = 4, although the capacity deteriorated to 46 mAh g −1 at E/S = 3 presumably because of insufficient ionic conduction pathway (Figs. 6b and S3b). Nevertheless, sparingly solvating [Li(SL) 2 ][TFSA]−4.0HFE with high t Li + value enables the decoupling of E/S from the dissolution/ precipitation-controlled redox mechanism for the sulfur cathode and is considered as more efficient than typical DME/DOL electrolytes in this regard. As the amount of unutilized electrolyte in the dead volume of the coin cell is non-negligible, we believe that the critical E/S can be reduced further for [Li(SL) 2 ][TFSA]−4.0HFE using a rational electrode and cell structure design. Although the high density of the electrolyte (Table I) may not be beneficial in the reduction of the electrolyte weight, a study using this highly Li-ion conductive, sparingly solvating electrolyte is currently in progress to achieve E/S < 3 (as demonstrated with pouch cells by Kaskel et al.) 16 and a gravimetric energy density of > 300 Wh kg −1 . [TFSA]−4.0HFE was compared with that with DME/ DOL electrolytes; subsequently, they were discussed in terms of difference in polysulfide solubility and Li ion transport properties. Reversible high capacity exceeding 1000 mAh g −1 and high Coulombic efficiency of > 95% were confirmed for cells with sparingly solvating electrolytes, whereas cells with the DME/DOL  3HFE and DME/DOL emphasized the importance of t Li + for improving sulfur utilization. Hence, Li ion transport properties would contribute more significantly to the high-capacity sulfur cathodes, where more Li ions were involved in the electrochemical reaction compared with typical lithium-ion battery cathode materials. The sparingly solvating [Li(SL) 2 ][TFSA]−4.0HFE with high t Li + contributed to the high capacity of 1130 mAh g −1 even at E/S = 4 and allowed the decoupling of E/S from the dissolution/ precipitation pathway of the sulfur cathode. In addition to the low polysulfide solubility, the fast Li ion conduction, characterized herein by a high t Li + , was discovered to be a crucial parameter for developing more efficient electrolytes for practical Li-S batteries with high energy density.