The lithium garnet series Li7-xLa3Zr2-xTaxO12 (x = 0-2) has shown great promise as a solid electrolyte material, however the room temperature conductivity is currently too low to find wide commercial success. In order to better understand the mechanisms of ionic diffusion within the crystal, a combined molecular dynamics and quasi-elastic neutron scattering (QENS) study was investigated. Using molecular dynamics simulations, we are able to easily probe atomic scale events that are usually difficult to examine experimentally. Like the local arrangement of lithium on its sublattice, or how the trajectory of lithium ions is affected by the nearest neighbor sites. The QENS experiment directly measures the dynamic structure factor S(Q,ω), which is capable of capturing both the residence time and mean jump distance experimentally allowing us to directly compare experimental and simulated intermediate scattering functions I,(Q,t). Overall, we saw good agreement between the two techniques, both predicting a jump-diffusion model in the form described by Singwi and Sjölander.Three different simulation models were employed in this study, two using classical molecular dynamics (MD), while a third using density functional theory (DFT) based calculations. All three model types are used to first better understand the phase transformation behavior for the end member composition Li7La3Zr2O12, which undergoes a characteristic phase transformation from an ordered tetragonal to disordered cubic phase at 900 K. First DFT methods are used to better understand what role the selection of an electron exchange-correlation functional plays on the accuracy of lattice parameter and phase transformation behavior. In total 14 different functional forms are investigated. Similarly, two different classical MD models, one being a “core-shell” model, where each atomic nucleus is connected by a spring potential to an electron shell that can capture the polarization of species, while the other being a “core-only” model that treats each atom as a point charge, which can be used for larger and faster simulations. The dynamics of the two end member compositions Li5La3Ta2O12 (L5LT) and Li7La3Zr2O12 (L7LZ), were looked at using the core-shell model with respect to properties like the conductivity, self-diffusivity, thermodynamic correction factor, and entropy of configuration. While the core-only model is used to investigate the finite-size effects of atomic simulation, by changing the number of particles within the simulation by using four different crystal sizes for L7LZ. Simulation cells consisting of 1×1×1 (192 atoms), 2×2×2 (1536 atoms), 3×3×3 (5184 atoms), and 4×4×4 unit cells (12288 atoms) were generated in order to find convergent behavior to the properties highlighted above. Having determined a 3 x 3 x 3 simulation provides adequate accuracy, a verity of garnet compositions corresponding to [Li] = 5, 5.5, 6, 6.25, 6.5, 6.75, & 7 were generated to determine the optimal composition for use as an electrolyte material. Our simulations predict that the best performing room temperature composition corresponds to when [Li] = 6.5 corresponding to the maximum lithium concentration that results in a disordered cubic phase at room temperature. Lastly, we look at the role lithium disorder plays in the phase transformation behavior of L7LZ, and the use of excess entropy calculations as a means of determining the performance of an electrolyte material.
Read
