Zerina Mehmedovic1 Vanessa Wei1 Andrew Grieder1 Nicole Adelstein1

1, San Francisco State Univ, San Francisco, California, United States

Introduction: Our simulations are designed to answer the long-standing question in solid-state batteries: how to increase Li-ion conductivity in solid electrolytes through engineering the structure and chemistry of the anion sub-lattice. Understanding the mechanism behind lithium ion diffusion in promising solid electrolytes, such as lithium-oxyhalide anti-perovskites, can provide design principles for new battery materials. Our ab-initio molecular dynamics (AIMD) simulations of Li3OCl with Li vacancy concentrations between 0.5 and 2% reveal the origin of complex interactions between vacancies and new insights into the effect of correlated motion on the diffusion mechanism. Previous simulations of Li3OCl compute conductivities that are much lower than measured experimentally [1] and our simulations show that finite size effects are part of the problem.
Methods: Ab-initio molecular dynamics were performed on a 4x4x4 supercells using VASP. The size of the supercell (320 atoms) allows for realistic vacancy concentrations and reducing finite size effects. The supercells had at least one and up to four Li+<font size="1"> </font>vacancies. Maximally Localized Wannier Functions were used to calculate the dynamic polarization of anions and their effect on Li+ diffusion, which is a novel technique that we developed for analyzing solid electrolytes [2]. As is standard, the diffusion coefficient for each compound was determined using the mean squared displacement. The `binding energy’ of Li+ vacancies is calculated and lattice screening is separated into the strain and electronic effects.
Results: We discovered new collective diffusion mechanisms in Li3OCl and that finite size effects plague most AIMD simulations. For example, simulating two Li+ vacancies (0.5% concentration) increases diffusion because the negatively charged vacancies "push" each other along. The largest vacancy concentration we simulated was 2%, which gave the highest diffusion due to disordering of the Li-sublattice, rather than a "push" mechanism. Vacancies can interact up to 7 Å from each other, with strain effects dominating 5 Å or closer. Vacancies communicate through the lattice, especially through polarization of anions. The dynamics of the anion polarization can be quantified easily in a nudged-elastic-band calculation and is still important even in AIMD. Thus, diffusion may be controlled by changing the polarizability of the anions, such as creating Br and F alloys with Li3OCl.

[1] Zhang, Y., Zhao, Y., & Chen, C. (2013). Ab initio study of the stabilities of and mechanism of superionic transport in lithium-rich anti-perovskite. Phys. Rev. B Physical Review B, 87 (13).
[2] Adelstein, N. and Wood, B. (2016) Li+ conductivity in a superionic electrolyte driven by dynamically frustrated bond disorder. Journal of Materials Chemistry 28 (20) 7218-7231.