UC Berkeley

Lithium-sulfur (Li-S) batteries have been considered as an attractive alternative to current Li-ion batteries due to their large theoretical capacity (1672 mA-h/g) and theoretical energy density (2600 Wh/kg) while having a low cost, an abundance of the material, and relatively non-toxic properties. However, the low cyclability and significant capacity fading during the first several cycles prevent Li-S rechargeable batteries from being commercialized. During discharge, elemental sulfur is reduced to the final product Li2S through a series of soluble intermediate species, lithium polysulfides (Li2Sx, 2  x  8). Lithium polysulfides dissolved into the electrolyte in the separator can no longer participate in redox reductions, resulting in a loss of active materials, as well as a “shuttling effect” that causes capacity fading and low coulombic efficiency. Despite the fact that decades of research have attempted to solve this, the problem is still not resolved due to a lack of fundamental understanding of the system. This includes how lithium polysulfides are produced during discharge interactions with other components in the cell and the reaction mechanisms (the electrochemical and chemical processes) during cycling. The objective of this dissertation is to provide a fundamental understanding of lithium polysulfides produced during discharge of a Li-S cell. This is an essential piece of knowledge when designing and identifying the issues associated with Li-S batteries.

To begin, the morphology, thermal properties, and ionic conductivity of an ether-based nanostructured block copolymer containing lithium polysulfides were investigated. Previous work has shown that nanostructured block copolymer electrolytes containing an ion-conducting block and modulus-strengthening block has the potential of enabling solid-state lithium metal rechargeable batteries. This is of particular interest for a lithium-sulfur battery to fully explore its high energy density and capacity. Understanding the thermal and electrochemical properties of these block copolymer electrolytes containing lithium polysulfides is essential for evaluating their potential use in Li-S batteries. A systematic study of polystyrene-b-poly(ethylene oxide) (SEO) block copolymer mixed with Li2Sx with an average x value of 4 and 8 was conducted. Small angle X-ray scattering, differential scanning calorimetry, and ac impedance spectroscopy were used to measure the morphology, thermal properties, and ionic conductivities of all samples. The ionic conductivity of SEO/Li2Sx mixtures were compared with those of poly(ethylene oxide) (PEO) mixed with Li2Sx to quantify the effect of nanostructuring on ion transport. The conductivities of both SEO and PEO samples containing polysulfides with a longer average chain length higher than the same polymer containing polysulfides with a shorter average chain length at all salt concentrations, indicating that dissociation of long-chain polysulfides occurs more readily than short-chain polysulfides. Normalized conductivity was used to quantify the effect of morphology on ion transport. The results showed that SEO suppressed the migration of polysulfides relative to PEO. However, this suppression is inadequate for practical applications. In other words, cathode architectures that prevent polysulfides from entering the electrolyte are necessary for enabling Li-S batteries with block copolymer electrolytes. Nevertheless, the results obtained in this study are important as they enable quantification of polysulfide migration in Li-S batteries with imperfect polysulfide encapsulation, a limitation that applies to all known Li-S batteries.

Next, UV-vis spectroscopy with radiation wavelength in the range 200 - 800 nm was used to study different polysulfides in ether. Ex-situ UV-vis spectra were measured for chemically synthesized lithium polysulfides in TEGDME, Li_2 S_(x_mix ) | TEGDME solutions for xmix values of 4, 6, 8, and 10 and sulfur concentrations of 10, 50, and 100 mM. The peaks are generally more resolved at lower concentrations than at higher concentrations for all xmix values, suggesting a concentration dependence of spectra shape. The peak at 617 nm was used to confirm the existence of S3•- radical anion, which supports the argument that polysulfide radical anions are stable in ether-based electrolytes, and may play an important role in Li-S reaction mechanism. Using in-situ UV-vis method was discussed and challenges for Li-S reaction mechanism study were evaluated. A new fluorinated-ether based electrolyte was explored. Its low polysulfide solubility makes it a good candidate to be used in in-situ Li-S reaction studies because UV-vis radiations do not have a large penetration path through high concentration of polysulfide-containing materials. However, the main challenge in using UV-vis spectroscopy to study Li-S reaction mechanism is the ambiguity in peak assignments arised both from a lack of spectra standards for different polysulfides. It is difficult to experimentally obtain polysulfide spectra standards because polysulfides cannot be separated.

The need for optical spectra standards for lithium polysulfides motivated a computational project to simulate optical spectra for different polysulfides solvated in ether theoretically. Configurations of a pure lithium polysulfide species can be obtained using computational methods, which circumvents the issue related to obtaining experimental spectrum for a pure polysulfide. Calculating optical spectra requires the calculation of both ground state and various excited states of solvated lithium polysulfides and this is not trivial work. The main goal was to find out the complexity necessary to compute reliable optical spectra for solvated lithium polysulfides using Time-Dependent Density Functional Theory. The configurations of lithium polysulfides solvated in diglyme were obtained using first-principles molecular dynamics simulations in a previous work. Gaussian calculations revealed that solvent played an important role in the calculated spectra and that explicit solvent molecules were needed to capture the local solvent-solution interactions. The results calculated with Gaussian approximations were compared to those calculated with plane-wave approximations and the two methods were comparable at their most optimized state. For a large system such as lithium polysulfides with explicit ether solvents, plane-wave calculations are efficient at achieving numerical convergence. However, a high level of functional to approximate the exchange-correlation function such as cam-b3lyp or higher is needed to calculate physically representative optical spectra for solvated lithium polysulfides and the scientific community currently lacks the computational power to do these calculations.

X-ray Absorption (XAS) has the power of being elemental specific and detecting both amorphous and crystalline sulfur-containing species. The recent simulation by Pascal et al.1 also provided a reliable set of spectral standards for species analysis. With the foundation of previous work, an in operando XAS study of a solid-state Li-S cell was conducted where all sulfur-containing species through the entire depth of a Li-S cell were detected. Li2S8 was used as active material inside cathode instead of S8 to provide better contact between the active materials and the solid electrolyte. In operando XAS spectra were taken before and throughout the charge-discharge cycle. Inefficiency in the initial charge revealed that lithium polysulfide dissolved into the separator layer reacted with lithium metal at the anode. The relationship between the average discharge polysulfide chain length inside the cathode, xavg,cathode, and the number of electrons passed per S atom, ne, at different stages of discharge was evaluated. During the first voltage plateau, while a small amount of S8 was converted to Li2S8, the major electrochemical reaction was the reduction of Li2S8 to Li2S6 (about 75%). During the transition region between the two plateaus, Li2S8 continued to be reduced to Li2S6 while almost half of Li2S6 was reduced to Li2S4. Evidence of the formation of Li2S was observed from the beginning of the second voltage plateau, which supports the argument that chemical disproportionation reactions play an important role in the formation of Li2S. The challenge of using XAS to study Li-S reaction mechanism is the similarity in the peak locations for different lithium polysulfides which makes it difficult to distinguish between different polysulfides. A spectroscopy with simple distinctive peaks for different polysulfides and well-established spectra standards would be better for Li-S reaction mechanism study.