Impact of\nLithium Nonstoichiometry on Ionic Diffusion\nin Tetragonal Garnet-Type Li<sub>7</sub>La<sub>3</sub>Zr<sub>2</sub>O<sub>12</sub>

Understanding ion transport mechanisms on the atomistic\nscale in\nsolid-state electrolytes is crucial for the development of all-solid-state\nbatteries. Li<sub>7</sub>La<sub>3</sub>Zr<sub>2</sub>O<sub>12</sub> (LLZO) is a promising oxide solid electrolyte material, whose phase\ntransition behavior and ion transport mechanisms have attracted significant\nresearch attention. Previous studies have primarily focused on ion\ntransport in the cubic phase (intrinsic high-temperature phase or\ndoped variants). In contrast, the tetragonal phase of LLZO, despite\nits close relationship with the cubic phase, has received less attention\ndue to its relatively low ionic conductivity and high computational\ncost. A few recent computational studies have shown significant discrepancies\nin conductivity and activation energy between calculated and experimental\nvalues. Therefore, the unclear ion transport mechanisms in the tetragonal\nphase of LLZO are critical to understanding and designing oxide solid\nelectrolytes. In this study, we employ state-of-the-art machine-learning-based\nneuroevolution potential molecular dynamics simulations to investigate\nthe effects of lithium nonstoichiometry on the ionic conductivity\nand phase stability of LLZO. We demonstrate that small deviations\nfrom stoichiometry, particularly lithium deficiency, dramatically\nreduce the activation energy for Li<sup>+</sup> diffusion in tetragonal\nLLZO from 1.227 to 0.425 eV, increasing room-temperature ionic conductivity\nby 10 orders of magnitude. The slight lithium nonstoichiometry, which\ncommonly occurs during high-temperature synthesis, has a significant\neffect on ion transport in the tetragonal phase. Our findings highlight\nthe crucial role of lithium nonstoichiometry and defect chemistry\nin enhancing LLZO performance and provide insights for the rational\ndesign of high-performance solid electrolytes through defect engineering.