Strategies for Mitigating Dissolution of Solid Electrolyte Interphases in Sodium‐Ion Batteries

Abstract The interfacial reactions in sodium‐ion batteries (SIBs) are not well understood yet. The formation of a stable solid electrolyte interphase (SEI) in SIBs is still challenging due to the higher solubility of the SEI components compared to lithium analogues. This study therefore aims to shed light on the dissolution of SEI influenced by the electrolyte chemistry. By conducting electrochemical tests with extended open circuit pauses, and using surface spectroscopy, we determine the extent of self‐discharge due to SEI dissolution. Instead of using a conventional separator, β‐alumina was used as sodium‐conductive membrane to avoid crosstalk between the working and sodium‐metal counter electrode. The relative capacity loss after a pause of 50 hours in the tested electrolyte systems ranges up to 30 %. The solubility of typical inorganic SEI species like NaF and Na2CO3 was determined. The electrolytes were then saturated by those SEI species in order to oppose ageing due to the dissolution of the SEI.

c) In the cyclic voltammetry (CV) measurements in Figure S1, the cell setup C-BA@Pt was used and cells were cycled from 0.05 V to 2.0 V vs. Na + /Na for 10 cycles and then paused for 15 hours and then cycled again for 10 cycles. The reason why a lower cut-off potential was used in this experiment was the higher resistance of the βalumina separator compared to that for the Solupor® separator.
d) The XPS results in Figure 9, S8, S9, S10 and S11, were obtained with the cell setup C-SP@Cu, including a copper foil (13 mm diameter) electrode and a Solupor® separator. The cells were cycled using CV between 0.1 and 2.0 V vs. Na + /Na for 10 cycles, stopped at 2.0 V vs. Na + /Na and then disassembled prior to the XPS measurements. The intensities were corrected using the core level photoionization cross-sections and the relative XPS intensities were compared for each sample. Although copper foils have a native copper oxide layer which can undergo conversion reactions in Na half cells generating Cu and Na2O, a Na2O peak could not be detected with XPS (see Figure S9). The influence of the CuO and Cu2O conversion reactions can be disregarded since this experiment was only carried out to study the composition of the generated SEI layer. The electrochemical results for the C-SP@Cu cells (see Figure S11) were not included in the electrochemical analysis as their purpose was merely to enable an XPS study of the composition of the SEI layer. e) In the comparison of the Solupor® and β-alumina separators, the same procedure was used as described in a) or b). For comparability, the cell setups C-SP@Pt and C-BA@Pt were used. The pause times, which were arbitrarily selected, only served to show the qualitative differences between the C-SP and C-BA setups. As seen in Figure  S11, there were no significant redox peaks due to the native copper oxide.
General note: All salts and additives were dried under vacuum at 120°C for 24h, and hence had a low water content. Moreover, at low electrolyte potential, the residual water will be depleted due to reduction and reactions with the electrolyte salt or Na metal. The water reduction can be described by following reaction: Thus, the water reduction should give rise to H2 and O 2-, which can further react to other species. Therefore, the water content can affect the results and lead to different SEI components. Possible salt impurities should affect the whole test series and should hence not influence the comparisons of the electrolytes containing the same concentration of a salt. The voltammograms shown in Figure S1 suggest that the water content in the electrolytes can be different. However, this work focuses on SEI dissolution observed in our tested electrolyte systems and how it can be mitigated.

Materials characterizations
XPS. Cu discs (10mm diameter) were cycled (as described in Electrochemical measurements, d) against Na-metal using Solupor® separator. The cells were stopped at 2.0 V vs. Na + /Na and disassembled in an Ar-filled glovebox (<2 ppm H2O and <2 ppm O2). The Cu foils were then cleaned with 0.5 ml dimethyl carbonate (DMC, Sigma Aldrich) and mounted on carbon tape in a glass vial, vacuum sealed in polyethylene-coated aluminium pouch bags and transported to the synchrotron facility. Photoelectron spectroscopy of the material surface and bulk was performed at the I09 beamline at Diamond Light Source (Oxfordshire, UK). The tested electrodes were fixed with carbon tape to an omicron-type copper plate at the end-station at the synchrotron.
Soft photoelectron spectroscopy (SOXPES) was performed by using an excitation energy of 1090 eV at a branch of the beam line with a plane grating monochromator. The measured sample spot was estimated to be ca. 300 µm long and up to 1 mm wide. During the measurements, no charge neutraliser was used. To record the spectra a hemispherical VG Scienta EW4000 analyser, set to a pass energy of 50 eV was utilised. The analysis of the spectra was conducted using Igor Pro. To calibrate the binding energies in the recorded spectra the peak originating from the C-C bonds in the formed SEI set at 285 eV in the C 1s spectra was used.
ICP-OES. The solubilities of NaF, LiF, Na2CO3 and Li2CO3 in PC and EC:DEC were determined with ICP-OES using an Avio 200 instrument. Each one of the abovementioned compounds were added to PC and EC:DEC (1:1) until precipitation. The mixtures were then filtered with a 0.5 µm syringe filter evaporated in an oven at 100°C. Water was added to dilute the residual salt and the concentration of the elements were then measured using ICP-OES. A multi-standard solution (Multi-Element Calibration Standard 3, Perkin Elmer, 5% HNO3) was used to calibrate the instrument. The Li and Na concentrations in the samples were then determined using the Li (670.783 nm) and Na (589.592 nm) lines.

Results and discussion
Figure S1. Cyclic voltammograms recorded at a scan rate of 0.5 mV/s with the C-BA cell setup using Pt-electrodes (C-BA@Pt) in (a) 1 M NaPF6 in PC, (b) 1 M NaPF6 in EC:PC and (c) 1 M NaPF6 in EC:DEC. The scan was started at the OCV (ca. 2.0 V vs. Na + /Na) and the scan direction is shown in the voltammograms. The cycling range was from 0.05 V to 2.0 V vs. Na + /Na. The cell was cycled for 10 cycles, paused for 15 hours and then cycled again for ten cycles. The low cut-off potential of 0.05 V vs. Na + /Na was used to prevent Na-plating. The large reduction current on the first cycle can be explained by the reduction of solvent and water present in the solvent. The water reduction should occur at ca. 1.1 V vs. Na + /Na. Hence, based on these results, the water contents can be different in each solvent system which can affect the SEI chemistry. Figure S2. Cell setups used in the investigations of the SEI dissolution effect. The species generated at the Na-metal anode could diffuse to the working electrode and thus influence the electrochemical measurements and the composition of the SEI layer on the working electrode. By using a Na-conductive β-alumina separator, this effect can be avoided.      :PC after addition of 10 mg/ml NaF and Na2CO3, and 60 mg/ml Na-metal, respectively. The additives were used to decrease the SEI dissolution rates, as mentioned in the main text. The testing with Na-metal serves the additional purpose of illustrating the effect of the presence of additional species formed at Na-metal electrode in the electrolyte when using the β-alumina separator. The NaPF6-PC results were associated with significant uncertainties, and the effects of the additives were therefore not clear. The use of NaPF6-EC:DEC and -EC:PC resulted in the lowest reduction capacities. For EC:DEC, the presence of NaF and Na-metal resulted in a lower reduction charge, whereas in EC:PC the presence of Na-metal decreased the first reduction charge. After 10 cycles, when the SEI formation was assumed to be essentially completed, the presence of additives did not have any significant effect on the results.

Figure S9.
Relative contents of the different SEI species obtained from the peak integrals of the C 1s, O 1s, Na 2s, F 1s and P 2p spectra. The SEI layer formed in PC had a higher polymeric content than the SEI formed in EC:DEC which can be affected due to different water content in each electrolyte system. Compared to NaPF6-EC:DEC, the SEI layer formed in the NaPF6-PC electrolyte featured higher relative amounts of C-C, C-O, -CF2 species indicating more polymer formation. The SEI layer obtained in the NaPF6-EC:DEC electrolyte was, on the other hand, more inorganic due to the presence of more carbonate containing compounds. The phosphorus and fluorine containing SEI components were similar in PC and EC:DEC which could be explained by products stemming from the NaPF6 salt, such as POF3 and PF5 -. It should, however, be pointed out that the SEI composition could have been affected by species formed at the Na metal anode since the experiments were carried out with a cell containing a Solupor® separator. The results should therefore be carefully evaluated with the electrochemical and ICP-OES data in mind. Figure S11. XPS-spectra of the SEI layer formed after the cycling of a Cu-foil between 0.1 and 2.0 V vs. Na + /Na and stopping the scan at 2.0 V after 10 cycles. (a) C 1s, (b) O 1s, (c) Na 2s before and after NaF addition (10 mg/ml). The Na 2s spectra show a spectral change, before and after NaF addition for NaPF6-PC electrolyte. After the NaF addition, the Na 2s core level was similar to that seen the NaPF6-EC:DEC electrolyte.