First-Principles Molecular Dynamics of Non-Arrhenius Li+ Diffusion in Solid Electrolytes for Batteries

Abstract

Our simulations address a long-standing problem in the design of solid-state batteries – the low ionic conductivity across solid-solid interfaces and through the solid electrolyte. Significant insights in developing new battery materials can be achieved through understanding of Li+ ion diffusion, using first-principles molecular dynamics simulations. Previous research from our group has determined the mechanism for the lithium ion diffusion pathway through the Li3InBr6 crystal1. Our results expanded on these findings by simulating diffusion in Li3InCl6 and Li3InBr6-xClx. In particular, we focus on determining the effects of high temperature and electronic structure, in order to explain variation in the Li+ conductivity in these lithium halides. We explore the hypothesis that covalent-like bonds between Li+ and the halide anions help drive conductivity through correlated motion of the Li+ ions. Our novel approach dynamically tracks covalent bonds, especially correlated bond breaking and forming. Thus, the electronic structure must be simulated and to do this we use ab-initio DFT simulations, within the Quantum Espresso implementation. In our previous research, we noted that the conductivity of Li3InBr6 is non-Arrhenius at high temperatures. The diffusion mechanism does not change with temperature, so the effect was hypothesized to be from changes in the electronic structure, such as the weakening of bonds. The first step of our investigation was to determine the simulated melting temperature of Li3InBr6, as our simulations at 900K may be near the simulated melting temperature, which could explain the weakening of bonds. The melting temperature is simulated starting with an amorphous, 3 x 3 x 3 supercell containing 520 atoms. Then using the two-phase approach (TPA) with 1040 atoms, we determined the melting temperature. Second, we simulated Li+ diffusion along with the breaking and forming of covalent-like bonds using first principle dynamics and Wannier analysis. Exchange of Br anions with Cl anions changes the nature of covalent type bonds, but does not affect ionic bonds. To further explore nature of the interaction between the Li+ and the anion sub-lattice, the Br concentration in Li3InBr6-xClx was varied and bond breaking and forming is tracked. We established that lithium ion diffusion, near the melting temperature, is impeded compared to lower temperatures owing to a degradation of covalent-like bonds. The Li+ diffusion mechanism, from octahedral to tetrahedral back to octahedral sites, is propelled by the fluctuations of neighboring covalent-like bonds. Decreasing the Br concentration in Li3InBr6-xClx shows a non-monotonic change in conductivity, which is unexplainable in a purely ionic view of Li+ diffusion. The disorder caused by different bonds strength between Li+ and Cl- versus Br- leads to a relatively high conductivity when the ratio of Cl to Br is 1:1 but a marked decrease with other ratios2. Our novel first-principles study of the dynamic electronic structure is the only way to understand these unusual trends in conductivity and potentially predict conductivity in many solid state electrolytes. 1) Adelstein, N. and Wood, B. Li+ conductivity in a superionic solid electrolyte driven by dynamically frustrated bond disorder. Journal of Materials Chemistry (2016) submitted. 2. Tomita, Y. Substitution effect of ionic conductivity in lithium ion conductor, Li3InBr6-xClx. Solid State Ionics. (2008) 179, 867-870.