Understanding the Charging Mechanism of Lithium-Sulfur Batteries Using Spatially Resolved Operando X-Ray Absorption Spectroscopy

Electrochemical measurements

Operando sulfur K-edge XAS measurements

XAS data analysis

Three standards: S₈, Li₂S_n in DOL-DME, and Li₂S were used to perform analysis of X-ray absorption spectra using the same general approach as described previously.²¹ Li₂S standard was prepared from a commercially available powder (99.98% trace metal basis, Sigma-Aldrich). To minimize the self-absorption experienced by the standard, the sample was both ball-milled to reduce the particle size of Li₂S and diluted with boron nitride to 0.5 wt%. S₈ standard was prepared from commercially available S₈ powders (99.998% trace metal basis, Sigma-Aldrich), which was grinded using mortar and pestle and diluted with boron nitride to 0.5 wt%. Li₂S_n (Li₂S₆ average composition) standard was a solution-based standard synthesized in a DOL-DME solvent using the same commercially available Li₂S and S₈ powders. More specifically, 11.5 mg of Li₂S was mixed 40.1 mg S₈ in 5 ml of DOL-DME solvent and stirred overnight. X-ray absorption near edge structure (XANES) spectra of the three references are presented in Fig. 1. The figure highlights the features, which are subsequently used to distinguish between the three types of sulfur species. S₈ standard has only one peak at 2472 eV and a distinct concave feature at 2475.5 eV. Li₂S has two peaks, with one peak centered at 2473 eV and the other at 2476 eV, and can be identified by a convex shape at 2475.5 eV. Li₂S_x standard has a spectrum that is similar to the spectrum of S₈, but with an additional smaller peak at 2470–2471 eV, which corresponds to a charged terminal sulfur atom found in all polysulfides.²⁸ Although, in theory, it may be possible to distinguish between various polysulfide species by comparing the ratio between the two features at 2470 eV and 2472 eV,²⁸ which correspond to terminal and internal sulfur atoms, such analysis is complicated by the fact that sulfur also has a feature corresponding to an internal atom at 2472 eV and by the possible effects of self-absorption, which cannot be avoided at sulfur concentrations above 30 mM.²⁹ Therefore, in our study, we will not perform quantitative principle component analyses, but will use the three discussed features of S₈, Li₂S, and S_n²⁻ to qualitatively analyze the obtained spectra.

Zoom In Zoom Out Reset image size

Successful spatially resolved XAS measurements during the charging process of Li-S batteries require an operando battery cell that can achieve both a standard electrochemical performance and allow spectroscopic access to the electrodes and the separator. Consequently, prior to using our operando cell at the synchrotron facility, we had focused on the electrochemistry of the cell and confirmed that it could achieve the same electrochemical behavior as a standard T-cell design with a Li₂S cathode (∼2 mg_Li2S/cm²), a lithium metal foil anode, and a glass fiber separator soaked with 80 μl of electrolyte (1 M LiClO₄ and 0.5 M LiNO₃ dissolved in a DOL-DME). In Fig. 2, we plot the galvanostatic charge and discharge curves in both the operando cell and a standard T-cell as a function of time (Fig. 2a) and specific capacity (Fig. 2b). In the experiments, the first charge is performed at a C-rate of 0.1 h⁻¹, while the subsequent discharge and second charge are performed at a rate of 0.2 h⁻¹ (referring to the theoretical capacity of 1165 mAh/g_Li2S). As seen in Fig. 2a, the electrochemical performance of the T-cell and the operando cell are nearly identical. Both cells reach close to 100% of the theoretical capacity during the first charge, then discharge to approximately 75% of the theoretical capacity, and finally, during the last charge, reach 55% of the theoretical capacity. Additionally, in agreement with literature,^8,9,12 the first charge in both cells requires a significantly higher potential than the second charge. To highlight this difference in the required potential and to visualize what fraction of the theoretical capacity was reached during the first and the second charging processes, the same two charge curves are plotted as a function of specific capacity in Fig. 2b, while the discharge curve is omitted for clarity. Fig. 2b clearly demonstrates that the operando cell is able to reproduce the expected galvanostatic charge curves of the Li₂S cathode and can therefore be combined with XAS measurements to identify the intermediates produced during the initial and subsequent charging processes.

Zoom In Zoom Out Reset image size

After appropriately benchmarking the electrochemical performance of the operando cell, we confirmed its spectroscopic spatial resolution capabilities at a synchrotron facility. Fig. 3a compares the raw XANES spectra obtained in the cathode and the separator of the battery at the open circuit voltage (OCV) before starting the initial charge with the raw spectrum of the Li₂S reference. We have chosen to present the raw spectra rather than the spectra with a normalized edge-step of 1, in order to provide a measure of the relative concentration of the sulfur species in the two locations of the battery. Since the battery is resting at OCV, the cathode (a sum of active material and solution species trapped in the pores of the electrode) is expected to contain exclusively Li₂S, while the separator (solution species) is expected to not have any sulfur species, because Li₂S is not soluble in organic solvents such as DOL-DME.³⁰ Inspection of the spectra in Fig. 3 is consistent with these expectations, identifying the species inside the cathode as Li₂S and failing to detect any significant absorption at the sulfur K-edge in the separator. Our results are the first demonstration of XAS measurements with spatial resolution in a direction normal to the electrodes, enabling a spectral distinction between species located in either one of the electrodes versus species contained in the separator of the same cell. After the successful confirmation of the spatial resolution capabilities of the operando cell and identification of the initial species in the cathode and the separator region, the battery was cycled using the same conditions as in the benchmarking experiments, and two new XANES spectra were collected in both locations at an interval of about 130 mAh/g_Li2S. The resulting galvanostatic curves are plotted as a function of time in Fig. 3b. The same charging curves are also replotted as a function of specific capacity and are presented with the associated operando XANES spectra in Figs. 4 and 6.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Fig. 4a presents the initial charge of the Li₂S cathode, demonstrating that close to 100% of the theoretical capacity (1165 mAh/g_Li2S) was achieved using a C-rate of 0.1 h⁻¹ and that the initiation of the charging process required a short-term rise in the potential to a maximum value of 3.4 V (Fig. 4a inset), which, in agreement with literature,^9,11 is dependent on the charging current. Additionally, the figure identifies the approximate location of 9 points (a-i), which correspond to a set of operando spectra collected either in the cathode or in the separator. Inspection of the normalized spectra obtained in the cathode (Fig. 4b), which are progressively shifted up by 0.5 a.u. for clarity, shows that at the very beginning of the charge, the cathode consists of only Li₂S, and that at the very end of the charge, the cathode consists of only S₈. Furthermore, while the spectra are continuously changing throughout the charge, no additional features appear in the 2470–2471 eV region, the signature region of the polysulfide intermediates.²⁸ The normalized spectra from the separator (Fig. 4c) further confirm that no detectable concentration of polysulfide intermediates is generated during the first charge. In particular, the spectra demonstrate that only one type of species, which correspond to a fully oxidized product, S₈, is found in the separator throughout the entire process (b-i). A normalized spectrum is not associated with the OCV period (point a), because after the assembly of the battery, we did not detect any species in the separator (Fig. 3a). To visualize more clearly how the spectra are developing throughout the charge, we plot the normalized intensity related to the Li₂S feature at 2475.3 eV and the normalized intensity of the polysulfide feature at 2470.3 eV for both the cathode and the separator in Fig. 4d. Although Li₂S was present only in the cathode, and polysulfide species were not detected in either location, all four sets of data are included to facilitate comparison and demonstrate the expected normalized intensity in the absence of the species. From Fig. 4d, it is seen that the normalized intensity of the cathode at 2475.3 eV decreases up until 1000 mAh/g_Li2S, indicating a gradual conversion of Li₂S to S₈. After this point, no additional change is detected in the measured spot, despite the fact that the charging process proceeds, indicating that there is heterogeneity in the electrode or that some electrolyte oxidation is occurring.

To extract more information about the intermediates in the separator, Fig. 4e presents the raw XANES spectra, which directly probe the amount of species present, rather than providing information on the relative ratios of components as is the case for the normalized spectra presented above. From the plot, it is evident that the concentration of dissolved S₈ is changing throughout the charge. To visualize these changes, we plot the absorption intensity at incident energy of 2487.3 eV (i.e., the edge height) in the inset of the Fig. 4e. The inset demonstrates that formation of S₈ can be detected as early as after 10% of the charging process, and that the concentration of S₈ is increasing throughout the entire process. Our sensitivity to detecting S₈ in both the separator and the cathode is a significant improvement over previous operando studies based on XRD characterization, which could detect formation of S₈ only after 60% of the charging process,⁸ or not at all.⁹ Furthermore, our results demonstrate that during the first charge occurring at a significant overpotential, polysulfides do not exist in a detectable concentration, and that it is possible to extract the entire theoretical capacity by converting Li₂S to S₈.

To clarify what type of specific processes occur inside the battery based on DOL-DME electrolyte during the first charge, we need to consider how micro-meter sized insulating Li₂S particles could become electrochemically activated. It has been originally proposed that the first step to oxidation of Li₂S is a charge transfer step that leads to a lithium-deficient surface on the surface of large Li₂S particles.^9,12 This mechanism, however, does not provide a sufficient explanation for how an activation of an upper limit of 6% of the volume (10 nm outer layer of a particle with a diameter of 1000 nm, assuming that the whole outer surface of the particle is in contact with carbon) could lead to an extraction of the entire capacity. Therefore, a more likely mechanism involves a redox mediator that diffuses between the conductive carbon surface and the Li₂S particle and thus eliminates the need for direct contact between the two materials.¹⁶ A recent study by Koh et al. has provided support for this mechanism by demonstrating that it was possible to charge Li₂S particles that were electronically completely isolated from the carbon electrode in a cell using a DME electrolyte.³¹ The authors had proposed that the first charge of Li₂S did not occur through a direct electrochemical oxidation of Li₂S but rather through a combination of electrochemical and chemical reactions involving polysulfide impurities that lead to the generation of polysulfide redox mediators.³¹ Although our spatially resolved XAS measurements do not detect polysulfides (sensitivity of XAS is on the order 100 ppm³² or approximately 1 mM of dissolved sulfur (S₁ basis) in DOL-DME) and therefore, rule out polysulfide intermediates as sole redox mediators during the first charge occurring at a significant overpotential, they are consistent with either a combination of S₈ and a trace concentration of polysulfides or electrolyte oxidation fragments acting as such mediators.

To understand which of these two possibilities is occurring inside the battery, we need to consider the electrochemical and chemical reactions that can facilitate conversion of Li₂S to S₈ to an extent equating to close to a 100% theoretical capacity. The possible reaction pathway involving a combination of S₈ and a trace concentration of polysulfides is shown in Equations 1–3. In this proposed mechanism, the electrochemically generated S_8,solid subsequently dissolves and diffuses toward a Li₂S particle. Then, S_8,solution and Li₂S react chemically to produce S_n²⁻ in trace concentrations, and S_n²⁻ converts through a series of chain-growth/disproportionation reactions to a polysulfide species that can be electrochemically oxidized to solid S₈. The produced S_8,solid can then dissolve and again react chemically with Li₂S. Consequently, during the first charge, parts of the micro-meter sized Li₂S particles which have interfacial contact with the conductive carbon support could become electrochemically activated (Equation 4), reflected by the occurrence of a peak potential (see inset of Fig. 4a) during the extraction of the first 10 mAh/g_Li2S of the capacity (∼1% of the theoretical capacity). At the same time, other parts of the micro-meter sized Li₂S particles could become oxidized through a series of chemical reactions (Equations 1–2), leading to production of polysulfides that can be electrochemically oxidized to S₈ (Equation 3), thus, enabling the extraction of the entire theoretical capacity. The fact that no polysulfides were detected by XAS is also consistent with this proposed mechanism, invoking only a trace concentration of polysulfides at any particular point of the charging process.

When considering the possibility of electrolyte facilitating oxidation of Li₂S, it is challenging to come up with a reaction pathway that is consistent with the observed electrochemistry. Unlike direct electrochemical oxidation of Li₂S, electrochemical oxidation of ethers or salts is not expected to depend on the current density and therefore, the short-term rise in the potential at the beginning of the charging process cannot be primarily attributed to electrolyte oxidation. Additionally, if only a small concentration of electrolyte oxidation products is generated during the extraction of the first 10 mAh/g_Li2S, it is unclear how these species would get regenerated during the rest of the charging process, after they are consumed in a redox reaction that leads to oxidation of Li₂S. Alternatively, if the electrolyte species are electrochemically oxidized throughout the entire charging process, then the fact that electrolyte is present in excess in the battery would mean that this process can occur independently of subsequent oxidation of Li₂S, pointing to an unwanted additional side reaction, rather than a mediating process. In this scenario, it is unclear why the charging process would stop and not continue beyond the observed extracted capacity of 1160 mAh/g_Li2S. Taking into account these considerations, we only consider a combination of S₈ and a trace amount of polysulfides as a viable reaction pathway that can lead to a complete conversion of Li₂S to S₈, but note that we cannot rule out that some electrolyte oxidation is also occurring during the charging process.

The discussed charging mechanism involving S₈ and polysulfides is portrayed schematically in Fig. 5. In addition to Equations 1–4, the figure includes three additional processes that are expected to occur inside the battery. In the first additional process, the solution species may diffuse to the anode and react chemically with Li metal. This process explains why it may be difficult to accumulate a considerable concentration of polysulfides during the first charge of Li₂S cathodes and is supported by recent results of Wang et al., who were able to first electrochemically activate Li₂S at 3.8 V and then complete the charging process at a significantly lower potential of 2.6 V in the presence of a ceramic barrier separating solution species from the lithium anode.¹⁹ The other two additional processes correspond to chemical reactions that can occur in the event that a significant concentration of polysulfides is generated inside the battery. In addition to getting electrochemically oxidized at the electrode, these polysulfides can react chemically with solid Li₂S and S₈ and convert more of the total [S] concentration to polysulfides as shown by Equations 5 and 6.

Having clarified the intermediates of the first charge and identified a possible charging pathway in the DOL-DME electrolyte, we now examine the details of the subsequent charge. The second charge began after 75% of the theoretical capacity was extracted during the discharge and reached 55% of the theoretical capacity (73% of the capacity reached during the discharge) at an overpotential that was significantly lower than the overpotential of the first charge. The process is shown in Fig. 6a together with the approximate location of 7 points (a-g), which are associated with a set of operando spectra collected in the cathode, and a set of 6 points (a-f), which are associated with a set of operando spectra collected in the separator. Because the charging process had ended rapidly, it was possible to collect a final spectrum only in the cathode but not in the separator. Inspection of the normalized spectra in Fig. 6b indicates that the initial spectrum in the cathode is already different from the spectrum of the first charging process, due to a polysulfide feature in the 2470–2471 eV region. Furthermore, the presence of the polysulfides in the electrolyte is confirmed by the normalized spectra collected in the separator in Fig. 6c, which contain a charged terminal sulfur feature throughout the entire charging process. The relative changes in the Li₂S and polysulfide components throughout the charge in both the electrode and the separator are visualized in Fig. 6d, while the components of the initial and final spectra of both the second and the first charge are compared in the Appendix. It is determined that the starting composition in the cathode of the second charge is consistent with presence of Li₂S and a polysulfide component, as well as what is likely a small amount of S₈, while, as discussed previously, the cathode of the first charge contains only Li₂S. Furthermore, the figure indicates that the species found in the cathode at the end of both the first and second charge can be completely described by S₈. Using the information about the sulfur species present in the cathode at the beginning and the end of the second charge, we can write the overall equation for the reaction process (Equation 7).

When the process shown in the Equation 7 is compared against the process shown in Equation 4, it becomes evident that the presence of polysulfides is important to lowering the charging potential during the second charge, as has been previously suggested in the literature.^9,19,33 When interpreted in the context of the mechanism in Fig. 5, the addition of polysulfides directly introduces a concentration of solution species that can undergo chain-growth and disproportionation reactions (ii.b) as well as diffuse to the electrode and either get electrochemically oxidized to S₈ (ii.c) or react chemically with Li₂S (ii.a) and S₈ (iii.b) to produce more polysulfides. The electrochemically formed S₈ can then regenerate polysulfide species by either dissolving into the electrolyte (iii.a) and subsequently reacting chemically with Li₂S (ii.a) or by reacting chemically with polysulfides (iii.b). This process can occur throughout the entire charging process until all of Li₂S is consumed. Consequently, the same charging mechanism shown in Fig. 5 is consistent with both the first and the second charge in the DOL-DME electrolyte.

Zoom In Zoom Out Reset image size

It has been previously suggested by several research groups that a higher order polysulfide, such as S₆²⁻ or S₈²⁻, and not S₈ is the initial oxidation product in the DOL-DME electrolyte.^9,19,22 In this pathway, lower order polysulfide (S_n²⁻) could first be electrochemically oxidized to a higher order polysulfide, (i.e., S₈²⁻, Equation 8), which could then react chemically with Li₂S and regenerate S_n²⁻(Equation 9), until all the lower order polysulfides and Li₂S are converted to a higher order polysulfide. At this point, the second step of oxidation could occur at a higher potential, leading to electrochemical oxidation of a higher order polysulfide to S₈ (Equation 10).

To determine whether S₈ or a higher order polysulfide is forming when polysulfides are electrochemically oxidized at a potential below 2.5 V during the second charge, we carefully examine the polysulfide feature. In the cathode this feature first increases in intensity, then remains constant for almost the entire charge, and decreases only in the last spectrum (Fig. 6d). Although these changes in the normalized intensity of the charged terminal sulfur in the polysulfide intermediate could be related to electrochemical and chemical processes occurring inside the battery, they can also be influenced by the diffusion of the polysulfides from the bulk electrolyte. In particular, at the end of both discharge and charge processes the concentration of polysulfides near the electrode surface would get smaller, as they would get rapidly reduced or oxidized. These rapid changes can explain both the initial rise in the polysulfide component in the cathode during the beginning of the charging process (the depleted polysulfides get replenished from the bulk of electrolyte) and the subsequent elimination of the polysulfide signal at the end of the process. Consequently, only the polysulfides in the separator (bulk electrolyte) could clarify whether a higher order polysulfide or S₈ is forming throughout the charge.

The separator is expected to be representative of the bulk concentration of solution based sulfur species in the battery, because the reported diffusion coefficients of polysulfides and sulfur range from 1.2·10⁻⁵ to 2·10⁻⁷ cm²/s,^34–36 which corresponds to a diffusion time through the entire cathode and the separator (340 μm) of less than 1 h, even when considering the smallest reported diffusion coefficient in the calculation. To directly illustrate that the species in the separator are associated with electrochemical processes occurring in the cathode, Fig. 7 presents the discharge curve and the associated intermediates forming in the separator during the discharging process. The presented spectra are characterized both by changes in the composition of the intermediates and in their concentration. Specifically, it is seen that as the discharge transitions from a high to a low voltage plateau, the concentration of the solution based species (Fig. 7b) as well as the relative fraction of the charged terminal sulfur atoms (Fig. 7c) significantly increase. These changes correspond to reduction of S_8,solid to polysulfides and to shortening of the polysulfide chain length. Consequently, our data confirm that characterization of the separator provides information about the solution phase products that are forming in the cathode, and indicate that if a higher order polysulfide and not S₈ was the main electrochemical oxidation product during the second charge, then its formation could be tracked by observing the changes in the separator species.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Considering formation of a higher order polysulfide, it is expected that this process would lead to a decrease in the relative ratio between the charged sulfur and internal sulfur atoms and/or a significant increase in the concentration of the solution based species. The decrease in the ratio of the two peaks would correspond to the formation of S₆²⁻ or S₈²⁻ from lower order polysulfides (reduction in the number of charged terminal sulfur atoms), while the change in the concentration of the solution based species would correspond to formation of polysulfides from solid Li₂S. Because the discharge process in between the first and second charge resulted in 75% conversion of the capacity stored in S₈ to Li₂S, the starting concentration of polysulfides can be estimated to correspond to approximately 30% of the total sulfur species. Consequently, the formation of a higher order polysulfide instead of solid S₈ should lead to a conversion of the remaining 70% of the species to S₆²⁻ or S₈²⁻ and a considerable increase in the concentration of the solution based species. Furthermore, the solubility of polysulfides is not expected to limit this process, as dissolution of the entire 2.0 mg of Li₂S in the added electrolyte (80 μl) could only lead to 0.07 M concentration of S₈²⁻ (0.09 M concentration of S₆²⁻).

The already presented Fig. 6c rules out the gradual decrease in the peak ratios, demonstrating that the peaks are characterized by a small increase and not decrease in the ratio (the feature at 2472 eV decreases slightly, while the feature at 2470 eV does not change) between spectrum-b and spectrum-f. These small changes could only correspond to a light shortening and not lengthening of the polysulfide chain. To understand how the concentration of polysulfides changes throughout the charge, Fig. 6e compares the raw spectra, in which the edge jump is proportional to the concentration of sulfur atom in all of its possible forms (both S_8,solution and polysulfides). From the figure, it is evident that the concentration increases only slightly throughout the charging process. As discussed previously, insufficient diffusion time cannot play a role in the lack of significant changes in the concentration, because the charging process lasted 3 h, while the time that was needed to diffuse through the electrode and the separator can be calculated to be less than one hour. Consequently, as illustrated in Fig. 8, our measurements are consistent with the charging pathway in which the main electrochemical product is a solid S₈ and not a solution based S₈²⁻ and indicate that the first and the second charging processes occur via the same mechanism shown in Figure 5.

We note that Yamin and co-workers have previously speculated about a similar type of charging mechanism.^34,37 More specifically, the authors had hypothesized that the anodic cyclic voltammetry (CV) features of various polysulfides dissolved in tetrahydrofuran (THF) solvent could only be explained, if all the polysulfides were to chemically convert to the same intermediate, which would then be electrochemically oxidized at the electrode surface.^34,37 They based their reasoning on the fact that the value of the slope of the peak anodic current versus the scan rate plotted on a log scale had a value of significantly lower than 0.5, while the potential at which the sole anodic peak occurred was independent of the used polysulfide chain length and concentration. Because S₈ is the most stable oxidation product, it was concluded by the authors that S₈ and not a polysulfide was the most likely electrochemical product.^34,38 Due to similarities in the cyclic voltammetry features of Li-S batteries based on THF^19,34,37 and DOL-DME^17,35,39 solvents, it is likely that Li-S batteries based on these solvents share the same mechanism of oxidation.

The developed understanding of the charging mechanism in the DOL-DME based electrolyte can be applied to the recent experimental reports that identify three main strategies of reducing the overpotential of the first charge: introduction of polysulfides into the electrolyte,^9,19,33 reduction in the charging rate,^9,16 and reduction in Li₂S particle size.^9,11,12 In the case of the introduction of polysulfides into the electrolyte, the charging mechanism has the possibility to occur via diffusion of the polysulfides to the electrode structure, if the added concentration of polysulfides is sufficient to sustain the charge at the prescribed rate without depleting the concentration of polysulfides to a trace level. Consequently, the process does not require pathway-(i) in Fig. 5, corresponding to an electrochemical activation of Li₂S, and can occur at a normal overpotential using pathway-(ii). In the case of the slower charging rate, S_8,solid, which is being produced electrochemically, has more time to dissolve into the DOL-DME electrolyte throughout the charging process (Fig. 5: iii.a) and react chemically with Li₂S to produce polysulfides (Fig. 5: ii.a). As a result, if the charging rate is slow enough to generate a considerable concentration of the polysulfide species and the generated species are not lost to reactions with the lithium anode, then the first charge will be able to proceed at the same overpotential as the second charge. In the last strategy, which uses modified cathodes with smaller Li₂S particles, the smaller particle size leads to a significantly larger surface area of Li₂S, which facilitates the chemical reaction between S₈ and Li₂S (Fig. 5: ii.a). The increase in the chemical reaction promotes the charging pathway-(ii). The above analysis demonstrates how the effect of changes in experimental conditions, cathode structure, and electrolyte components can be understood from a mechanistic perspective and corroborates the charging mechanism proposed in our study. Future use of the identified mechanism will help inform rational design of novel cathodes and electrolytes and facilitate development of improved models of Li-S battery operation.

Our work highlights the advantages of spatially resolved XAS characterization, which can detect solution phase polysulfides and amorphous as well as crystalline solids, while also differentiating between the species forming in the electrode and the separator. Using this technique, we are able to obtain information about intermediates present in the cathode and the separator of an operating Li-S battery during the first and second charge of Li₂S and gain insight into the mechanism of Li₂S oxidation in the DOL-DME electrolyte solvent. Specifically, our results demonstrate that even though the first charging process, which occurs at a large overpotential, leads to the extraction of the entire theoretical capacity, corresponding to a complete conversion of Li₂S to S₈, no significant concentration of polysulfide intermediates can be detected throughout the process. Furthermore, analysis of the second charge, which occurs at a small overpotential, shows that the polysulfide species remain at a relatively constant concentration and composition, which is consistent with solid S₈ forming during the entire process. Finally, our results indicate that oxidation of Li₂S particles requires a chemical step, and that during the second charge, this chemical step is facilitated by the presence of a significant concentration of polysulfide intermediates. The identified pathways are expected to assist in reaching higher areal capacities in Li-S batteries by guiding both the development of models of Li-S battery operation and the design of improved cathode structures and electrolyte components.

The described XAS measurements were carried out at the Stanford Synchrotron Radiation Lightsource, a Directorate of SLAC National Accelerator Laboratory and an Office of Science User Facility operated for the U.S. Department of Energy Office of Science by Stanford University. Use of the Stanford Synchrotron Radiation Lightsource, SLAC National Accelerator Laboratory, is supported by the U.S. Department of Energy, Office of Science, Office of Basic Energy Sciences under Contract No. DE-AC02-76SF00515. The authors also thank Diamond Light Source, Didcot, UK for access to beamline I18 (SP 8734), K. Ignatyev and T. Geraki for support during the experiments at DLS, S. M. Webb and C. Roach for the support during the experiments at SLAC, as well as O. Gröger, A. Siebel, M. Wetjen, A. Berger, and A. Eberle for helpful discussions. Y. Gorlin gratefully acknowledges the support of the Alexander von Humboldt Postdoctoral Fellowship and Carl Friedrich von Siemens Fellowship Supplement. Portions of the research leading to these results have received funding from the German Ministry of Education and Research (BMBF) under agreement No. 03ET6045D.

See Figure A1.

Zoom In Zoom Out Reset image size