This research addresses the long-standing dilemma in the design of solid-state batteries, the low ionic conductivity across solid-solid interfaces and through the solid electrolyte. Understanding the mechanism behind lithium ion diffusion in solid electrolytes can provide significant insights in developing new battery materials. This study focuses on the lithium-oxyhalide anti-perovskites due to their usual non-monotonic conductivity upon increasing the ratio of bromine to chlorine. To explain the variation in conductivity, we simulated Li3OCl1-xBrx with x=0, 0.22, 0.30, and 1. The hypothesis that the Li-Br compared to the Li-Cl bond significantly affects conductivity is tested. First principles DFT-MD was performed on a 3x3x3 Li3OClxBrx supercell using ultrasoft (Vanderbilt) pseudo-potentials as implemented in the Quantum ESPRESSO PWscf package. The size of the supercell (135 atoms) allows for a higher percentage Br substituted doping, comprising of 22% and 30% Br:Cl. The simulations each ran for 10 ps with 1 or 2 Li+ vacancies, as the conduction mechanism requires vacancies. An additional study was performed to determine how temperature (700-900K) effects Li+ ion diffusion in Li3OClxBrx. Maximally Localized Wannier Functions (MLWF) are used to quantify the covalent character of the Li-Cl/Br/O bonds, in particular their dynamic behavior. The diffusion coefficient for each compound was determined using the mean squared displacement. The activation energy was calculated using the diffusion coefficients at 1000, 1500, and 2000K. We identified how the number of Cl and Br bonds and changes in covalency during Li+ jumping affects the diffusion pathway. The bonding characters that were determined from the MLWF analysis show that the bromine has more covalent bonds with Li+ than the chlorine. This strong bond affects the diffusion pathway and diffusivity, in addition to the larger size of the bromine atom. Simulations with two Li+ vacancies revealed correlated motion. Similar to previous studies, the diffusion coefficients were much lower than experiment. In order to simulate significant diffusivity, very high temperatures are required. The change did not correlate with superionic conductance. Thus there is still more that is required to understand the diffusion mechanism in these oxy-halides, especially examining correlated motion using MD simulations with additional defects, including more Li+ vacancies.

Computational experiments on drivers of Li+ diffusion using lithium-oxyhalide anti-perovskites

253rd National Meeting (2017)

The official archive and repository for all ACS Meetings & Expositions abstracts and select posters or presentations.