Regensburg 2022 – wissenschaftliches Programm
Bereiche | Tage | Auswahl | Suche | Aktualisierungen | Downloads | Hilfe
MM: Fachverband Metall- und Materialphysik
MM 29: Data Driven Materials Science: Design of Functional Materials
MM 29.1: Vortrag
Donnerstag, 8. September 2022, 10:15–10:30, H45
Investigations of the Polysulfide Conversion Mechanism via Gaussian Approximation Potentials — •Xu Han1,2, Carsten G. Staacke1, Hendrik H. Heenen1, Xuefei Xu2, and Karsten Reuter1 — 1Fritz-Haber-Institut der MPG, Berlin, Germany — 2Tsinghua University, Beijing, China
Lithium-sulfur (Li-S) batteries have been regarded as promising energy storage systems with ultra-high theoretical energy density. During a charging cycle Li2S is converted to S8 and vice-versa, where intermediate Li polysulfides (LiPS) are formed in a complex reaction mechanism which is still under debate. The theoretical exploration of the involved Li-S chemistry is challenged by an extended reaction network making it intractable for first principles methods. In contrast, machine learning interatomic potentials (MLIPs) which potentially retain predictive accuracy at a fraction of the computational cost are ideally suited for this task.
Here, we establish a training protocol for a Gaussian approximation potential (GAP) to simulate the chemistry of LiPS. Our training is based on a constrained on-the-fly exploration of the LiPS chemical space. In that, we enumerate the connectivity of (poly)cyclic LiPS and explore their stability via global optimization procedures with iteratively refined MLIPs. We use the final, sufficiently accurate MLIP to sample the LiPS phase space and to compute charging/discharging curves which we can directly compare to experimental data. Our MLIP calculations are expected to provide more fundamental insights into the LiPS conversion mechanism in Li-S batteries.