Lithium Surface Restructuring and Dendrite-Like Protrusion Formation in Li-S Electrolytes From Reactive Molecular...

Lithium-metal anodes present remarkable theoretical capacity for advanced batteries. However, their practical application is hindered by interfacial instabilities and dendritic lithium morphologies, which undermine safety and longevity. This study utilizes machine-learned interatomic potentials to conduct molecular dynamics simulations of the interface between a lithium-metal anode and an electrolyte composed of elemental sulfur ( $S_8$ ), propylene carbonate (PC), and lithium Otf (LiOTf). Because the employed neutral MLIP does not explicitly describe electrode potential, electron transfer, or charge-state evolution, the simulations are interpreted as reactive structural models of lithium redistribution, aggregation, and protrusion growth rather than as a complete electrochemical plating/stripping description. Within this scope, the simulations provide atomistic insight into early dendrite-like lithium protrusions across different electrolyte compositions. Species detection, connection-matrix analysis, maximum protrusion height, and local-order analysis are used as complementary structural descriptors of interfacial reorganization. Supporting benchmarks against DFT include RDF validation for the interface model and structural and reaction-energy comparisons for adsorption and reaction in modeled lithium clusters. These benchmarks indicate that the r2SCAN readout gives more consistent structural and energetic agreement for the present lithium surface chemistry, whereas the $\omega$ B97X readout is retained as a sensitivity test for readout dependence. The findings provide an atomic-level view of how $S_8$ , PC, and ${\mathrm{Otf}}^{-}$ correlate with lithium aggregation and morphology changes in S₈/PC/LiOtf electrolytes, while clarifying the present limits of pre-trained MLIPs for mechanistic electrochemical interpretation.