Tuning Low Concentration Electrolytes for High Rate Performance in Lithium-Sulfur Batteries

As we wanted to highlight the impact of electrolytes, we purposely selected S-based cathodes without any protective shells²¹ or oxide additives²² that may improve cathode stability. As such, we selected standard and easy-to-reproduce procedure: S/nanocarbon electrodes are prepared by melt infiltration of S (99.5%, Sigma-Aldrich, USA) into Ketjen Black® (Lion Specialty Chemicals Co. LTD) at 157 °C for 12 h, thus making S/KB composites. The specific surface area of the KB and sulfur-infiltrated KB were analyzed with the Brauner-Emmett-Teller (BET) isotherm method. The excess S was evaporated at 250 °C until the desired S:C ratio (∼50 wt% S) was achieved. The S/KB particles were then mixed with a polyacrylic acid (PAA) binder (450,000 MW, Aldrich, USA) and pure black (Superior Graphite, USA) conductive additives in a 70:20:10 S/KB:PAA:CB weight ratio. To prepare electrodes, the solids were mixed with a solvent (ethanol:water = 1:1 by volume), sonicated for 30 min and stirred in a vial over-night. The produced slurry was then cast on aluminum (Al) foil (MTI) and left to dry in air at room temperature. The active material loading was approximately 0.45 mg_-S cm⁻². The dried electrodes were then punched into ½ inch diameter disks, dried in a vacuum oven at 60 °C for 12 h and then assembled into 2032 coin cells with a Celgard 2500 separator, a 0.75 mm thick Li foil anode (99.9%, Alfa Aessar) and 30 μl of the prepared electrolyte.

All the electrolytes used in this study employed various combinations of LiTFSI (Sigma-Aldrich, USA), LiPF₆ (Aldrich, battery grade), LiNO₃ (Alfa Aesar, 99%) salts in identical (commonly used) ether-based solvent mixture: a 1:1 ratio by volume of dimethoxyethane (DME, 99%, Sigma-Aldrich): dioxolane (DIOX, 99%, Sigma-Aldrich). 1M LiTFSI + 0.1M LiNO₃ and 3M LiTFSI + 0.1M LiNO₃ electrolytes were used as references for standard and high-concentration electrolyte compositions. All the individual salt components were kept at a concentration at or below 0.1M. Table I shows the list of the experimental electrolytes investigated here.

Once the electrolytes were prepared, their resistances and viscosities were tested. Resistance measurements were done by taking an EIS spectrum (Gamry Interface 1000, USA) in Li/Li symmetric coin cells. Viscosity measurements were taken with a RheoSense μViscometer (RheoSense, USA). Contact angle measurements were done by taking a video with a goniometer (Ramé-hart Instrument Co., USA) and processing the frames with ImageJ processing software to determine the contact angle and the time until the electrolyte was fully wetted and soaked into the electrode substrate.

All electrochemical experiments were done using an Arbin system (model BT2X43, USA). For galvanostatic charge-discharge testing, cells were cycled between 1.5 and 2.8 V for two cycles at C/20, followed by one cycle each at C/10, C/5, C/2, 1 C, and 2C rates. For long term tests cells were discharged at C/20 then charged and discharged at C/5 rate for 200 cycles. Each cell was characterized with EIS (Gamry Interface 1000, USA) in the frequency range 1 MHz–0.01 Hz before testing and after long term testing to investigate the change in impedance after cycling.

After testing, the coin cells were disassembled and inspected. The electrodes, separators, and Li foil anodes were visually examined for their color changes and any surface deposits. The color of the remaining electrolyte was also examined as a qualitative indication of the concentration of the polysulfides in the electrolyte. The electrodes were rinsed in DME before analysis with scanning electron microscopy (SEM) using Hitachi 8030 (Japan) and X-ray photoelectron spectroscopy (XPS) using Thermo K-alpha (USA) to investigate their microstructural changes and the CEI composition. The Li foil was washed in DME before SEM and energy dispersive spectroscopy (EDS) analyses on the Hitachi 8230.

For the electrolytes tested, the viscosities expectedly increase with increasing salt concentration. All the low concentration electrolytes have a 0.2M of total salt concentration and show similar viscosities of ∼0.5 mPa·s, regardless of their salt composition (Fig. 1a). Increasing LITFSI salt concentration to 1M tripled the viscosity to ∼1.5 mPa·s and 3M LiTFSI showed the most dramatic increase to 42.5 mPa·s. This increase in viscosity is enough to impact wetting of the separator during coin cell preparation. Figure 1b shows that while the use of higher concentration of 1M gives a slightly lower ohmic cell resistance than the low concentration electrolytes, the overall interfacial resistance with the Li foil was noticeably higher. While low electrolyte resistance is important, low resistance and fast kinetics at the Li anode interface could be more impactful to the overall cell behavior. The 3M LiTFSI gives a higher ohmic resistance (Fig. 1c), but lower interfacial resistance than many other LiTFSI-based electrolytes, while the 0.1M LiPF₆ electrolyte gives the lowest interfacial resistance (Fig. 1d).

Zoom In Zoom Out Reset image size

Contact angle measurements (Fig. 2) showed results for 0.1M LiTFSI, 1M LiTFSI, and 3M LiTFSI samples (A, D, and E). The remaining electrolytes (samples B and C) were too volatile to form a drop. While the higher angle 1M LiTFSI was only slightly higher than the 0.1M LiTFSI, the 3M LiTFSI was more than double the contact angle of both. Samples A, D, and E took 1.5, 3, and 10 seconds respectively to soak into the electrodes, matching with the viscosity data. The lower contact angle and faster time to absorb make these low concentration electrolytes more attractive than their high concentration counterparts from the cell fabrication point of view.

Zoom In Zoom Out Reset image size

The sorption isotherms and the pore size distribution are given in Fig. 3 and clearly indicate that sulfur is filling the pores in the KB samples. The BET specific surface area of the KB is 1303 m ²g⁻¹ and the surface are of the sulfur-carbon composite is 227 m ²g⁻¹.

Zoom In Zoom Out Reset image size

Figure 4 shows the electrochemical performance of the low concentration electrolytes. Of the low concentration electrolytes and 1M LiTFSI, the performance of the low concentration electrolytes across C-rates is similar, with the 0.1M LiTFSI electrolyte on average performing the worst. Figure 4b displays the C/10 charge/discharge cycle for each electrolyte. All low concentration electrolytes exhibit flat plateaus, low hysteresis, and low polarization. Similarly, at 2C (Fig. 4c), the reaction still proceeds along the correct pathway, with higher hysteresis, but no evidence of significant kinetic limitations. A comparison of long-term testing (Fig. 4d) shows the lowest decay in capacity for the mixed LiTFSI/LiPF₆ electrolyte. The 0.1M LiTFSI electrolyte has a higher initial capacity, but is less stable, as is the 0.1M LiPF₆, the combination proves the most stable at this rate.

Zoom In Zoom Out Reset image size

Low concentration electrolyte samples A, B, and C show similar capacity and hysteresis to the 1M LiTFSI electrolyte, indicating that the salt concentration above 0.1–0.2M (to around 1M) may be unnecessary additions to the electrolyte weight and cost. Sample B performs comparably or better than sample D across C-rates (Figs. 5a–5c). Within error, the difference between 1M and the low concentration electrolytes is not significant and thus the lighter, cheaper low-concentration electrolytes can likely replace the 1M LiTFSI standard without sacrificing performance. Even for the relatively thin electrodes tested here, the highly concentrated electrolyte (3M LiTFSI—sample E) enables the highest capacity, it does so with higher polarization, higher resistance, and an inefficient conversion reaction illustrated by the curved plateau. And, as previously discussed, it suffers from undesirably high viscosity and increased cost. In contrast, cheaper and less viscous electrolytes A-C offer much flatter plateaus, significantly lower hysteresis, and capacity similar to or better than the conventional 1M LiTFSI (sample D).

Zoom In Zoom Out Reset image size

Comparing EIS results from before and after long-term testing (Fig. 6) reveals the reduction in impedance for the 0.1M electrolyte to be the greatest. Note that for all the electrolytes shown, the major arc decreased in size during testing, indicating a lower interfacial resistances. Yet, the concentrated 3M electrolyte exhibits the largest impedance and the largest interfacial resistance after cycling.

Zoom In Zoom Out Reset image size

Post-mortem SEM analysis of the tested S-KB cathodes shows very small changes in the microstructures for all samples (Fig. 7). Individual carbon particles are visible without a change in morphology and no thick surface layer (e.g., CEI or re-deposited (poly)sulfides or S), likely related to the very high external surface area of the porous KB particles and relatively small electrode loadings (Fig. 7). This may be beneficial as it should allow conductive carbon of KB to remain being available for charge transfer, but may also explain small differences in the electrochemical performance between cells (Fig. 4). The Li anode (Figs. 8a–8c), on the other hand, showed significant changes after cycling. We can observe distinct new features that exhibit four additional elements segregated together: S, C, F and O. All Li foil samples were mostly bare Li in the center, with the intensity and size of features increasing towards the edge. Figures 8d, 8e show EDS results from the Li foil at the end of a 2C discharge. The 1M electrolyte produced the greatest (normalized) intensity of N, F, and S, as compared with both the low concentration and high concentration electrolytes, indicating the largest S dissolution and the thickest SEI formation. The LiTFSI, LiPF₆ mixed electrolytes has a greater F intensity than the 0.1M LiTFSI electrolyte, which follows from the additional F available from the decomposition of the LiPF₆. Mapping of the mixed low concentration LiTFSI, LiPF₆ electrolyte shows that the features are mostly composed of O, S, and F, with C being more uniform across the surface.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Post-mortem XPS analysis of thin tested cathodes (Fig. 9) indicates decomposition of the LiTFSI salt and electrolyte solvents with the formation of slightly different CEI in the case of all electrolyte compositions.^23–26 For the low concentration electrolytes, CEI species mostly include: CF_n, LiF, N-containing compounds, sulfites, and sulfates (Fig. 10). Overall, even 1/10–1/3 salt concentration in the electrolyte may apparently be enough to reproduce the composition of the protective CEI formed in 1M and 3M electrolyte compositions. Electrolytes containing 0.05–0.1M LiPF₆ show lower intensities of CF₂ than the majority-LiTFSI electrolytes while having a higher ratio of LiF to CF_n. Since LiF can be created by the decomposition of both LiTFSI and LiPF₆,²⁷ the multiple pathways to LiF make it the more dominant F-containing compound on the surface. Interestingly, only the electrolyte comprised of 0.05M LiPF₆ and 0.05M LiTFSI contains the most noticeable amount of Li₃N, a compound very important for favorable electrode protection.²⁸ The low concentration electrolytes have more N–O containing compounds than N-containing organic compounds, relatively, as compared to the 1 and 3M electrolytes. This is likely due to the higher relative proportion of LiNO₃ in the low M electrolytes since all five electrolytes comprised 0.1M LiNO₃. Perhaps with less LiTFSI to break down, there are more N–O containing species leftover to deposit and favorably protect the cathode. Only 0.1M LiTFSI exhibits both CF₂ and CF₃ bonds, while all electrolytes show C–C, C–H, C–O, and C=O bonds consistent with the decomposition and polymerization of the dioxolane solvent.²⁹ Another interesting observation—electrodes from the 1M LiTFSI, 3M LiTFSI and 0.1M LiPF₆ electrolyte cells did not show any Li₂S on the surface, suggesting the incomplete reduction of polysulfides (the electrodes were examined at the end of discharge, thus mostly comprising Li₂S in the bulk). The low concentration LiTFSI samples A and B cells showed mostly colorless electrolytes during the post-mortem analysis, further supporting the minimum content of polysulfides remaining in the bulk of the electrolyte. The lack of yellowing of the low concentration electrolyte also goes against the theory that high concentration electrolytes are necessary to minimize polysulfide dissolution. Low concentrations of LiTFSI showed higher surface intensities of sulfites over sulfates, with the opposite true for high concentrations of LiTFSI. This may indicate a less thorough oxidation of S after the decomposition of the LiTFSI. While increasing electrolyte concentration undesirably increases cell polymerization,²⁷ the low concentration electrolytes are clearly adequate to achieve similar or better rate performance (Figs. 4 and 5) and may have more favorable CEI protection.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size