Identification of lithium sites in a model ofLiPO3glass: Effects of the local structure and energy landscape on ionic jump dynamics

Abstract

We perform molecular dynamics simulations to study lithium dynamics in a model of ${\mathrm{LiPO}}_{3}$ glass at temperatures below the glass transition. A straightforward analysis of the ionic trajectories shows that lithium diffusion results from jumps between sites, the positions and properties of which are basically unmodified on the time scale of the lithium ionic relaxation. This allows us a detailed identification and characterization of the sites. The results indicate that the number of lithium sites is only slightly bigger than the number of lithium ions so that the fraction of vacant sites is very limited at every instant. Mapping the ionic trajectories onto sequences of jumps between the sites provides direct access to lithium jump dynamics. For each site, we determine the mean residence time ${\ensuremath{\tau}}_{s}$ of an ion and the probability ${p}_{s}^{b}$ that a jump from this site to another site is followed by a direct backjump. While a broad distribution $G(\mathrm{log}\phantom{\rule{0.3em}{0ex}}{\ensuremath{\tau}}_{s})$ shows that different sites feature very diverse lithium dynamics, high values of ${p}_{s}^{b}$ give direct evidence for correlated back-and-forth jumps. A strong decrease of ${p}_{s}^{b}$ with increasing ${\ensuremath{\tau}}_{s}$ indicates that the backjump probability depends on the dynamical state of an ion. Specifically, we find that correlated back-and-forth jumps are important at short times in the relaxation process, but not on the time scale of the lithium relaxation, where the hopping motion resembles a random walk. We further study how the local glass structure and the local energy landscape affect lithium jump dynamics. We observe substantial effects due to the energy landscape, which are difficult to capture within single-particle approaches. Our results rather imply that lithium migration is governed by the competition of the ions for a small fraction of vacant sites in a disordered energy landscape. Consistently, a statistical analysis shows that a vacancy mechanism dominates the repopulation of the lithium sites.