Interstitial Diffusion in Li-ion Battery Electrodes and Structural Phase Transitions in Crystalline Solids from First Principles.

Abstract

We perform first-principles investigations of thermally activated phase transitions and diffusion in solids. The atomic scale energy landscapes are evaluated with first-principles total energy calculations for different structural and configurational microstates. Effective Hamiltonians constructed from the total energies are subjected to Monte Carlo simulations to study thermodynamic and kinetic properties of the solids at finite temperatures.

Cubic to tetragonal martensitic phase transitions are investigated beyond the harmonic approximation. As an example, stoichiometric TiH2 is studied where a cubic phase becomes stable at high temperature while ab-initio energy calculations predict the cubic phase to be mechanically unstable with respect to tetragonal distortions at zero Kelvin. An anharmonic Hamiltonian is used to explain the stability of the cubic phase at higher temperature. The importance of anharmonic terms is emphasized and the true nature of the high temperature phase is elucidated beyond the traditional Landau-like explanation.

In Li-ion battery electrodes, phase transitions due to atomic redistribution with changes in Li concentration occur with insertion (removal) of Li-ions during discharge (charge). A comprehensive study of the thermodynamics and the non-dilute Li-diffusion mechanisms in spinel-Li1+xTi2O4 is performed. Two distinct phases are predicted at different lithium compositions. The predicted voltage curve qualitatively matches with experimental observation. The predicted fast diffusion arises from crystallographic features unique to the spinel crystal structure elucidating the crucial role of crystal structure on Li diffusion in intercalation compounds.

Effects of anion and guest species on diffusion are elucidated with Li- and Cu-diffusion in spinel-LixTiS2. We predict strong composition dependence of the diffusion coefficients. A unique feature about spinel-LixTiS2 is that the intermediate site of a Li- hop is coordinated by four Li-sites. This results in di- and triple-vacancy mechanisms at non-dilute concentrations with very different migration barriers. The strong dependence of hop mechanisms on local Li-arrangement is at the origin of large concentration dependence of the diffusion coefficients. This contrasts with spinel-LixTiO2 where the transition states are coordinated only by the end states of the hop, thereby restricting hops to a single vacancy mechanism. Cu ions are predicted to have much slower diffusion rate in TiS2 host compared to Li ions.