Dataset: Knowledge-driven design of Solid-Electrolyte Interphases on lithium metal via multiscale modelling

Due to its high energy density, lithium metal is a promising electrode for future energy storage. However, its practical capacity, cyclability and safety heavily depend on controlling its reactivity in contact with liquid electrolytes, which leads to the formation of a solid electrolyte interphase (SEI). In particular, there is a lack of fundamental mechanistic understanding of how the electrolyte composition impacts the SEI formation and its governing processes. Here, we present an in-depth model-based analysis of the initial SEI formation on lithium metal in a carbonate-based electrolyte. Thereby we reach for significantly larger length and time scales than comparable molecular dynamic studies. Our multiscale kinetic Monte Carlo/continuum model shows a layered, mostly inorganic SEI consisting of LiF on top of Li2CO3 and Li after 1 µs. Its formation is traced back to a complex interplay of various electrolyte and salt decomposition processes. We further reveal that low local Li+ concentrations result in a more mosaic-like, partly organic SEI and that a faster passivation of the lithium metal surface can be achieved by increasing the salt concentration. Based on this we suggest design strategies for SEI on lithium metal and make an important step towards knowledge-driven SEI engineering.

The published source data file contains the underlying data of all figures of the manuscript and Supplementary Information of the publication “Knowledge-driven design of Solid-Electrolyte Interphases on lithium metal via multiscale modelling”. The source data of each figure is presented in a single sheet named after the corresponding figure. Legends are given if required. Please note that the source data of the kMC boxes is provided as xyz coordinates and that each species number corresponds to a chemical species (cf. legend). Numerical details: The kMC/continuum model was implemented in MATLAB R2021a and differential equations were solved with the ode15s solver. The calculations were performed on an i7-8000 CPU with 16 GB RAM. For detailed information on the modelling approach see the Method section in the manuscript and the Supplementary Information.