Ensemble RMSD-Driven Metadynamics for Electrolyte-Reduction Pathway Discovery

Solid-electrolyte interphase (SEI) formation in lithium-ion and lithium-metal batteries is governed by highly branched electrolyte-reduction chemistry that is difficult to sample with conventional molecular dynamics.Here we combine RMSD-driven metadynamics with ensembles of independent semiempirical GFN2-xTB trajectories to extract statistically recurrent reduction pathways of representative carbonate electrolytes.A parameter screen on reduced ethylene carbonate (EC) identifies a practical operating window for Gaussian bias parameters and shows that confinement control is required to prevent non-reactive configurational escape.Under these conditions, ensembles of 20 trajectories are aggregated into species-prevalence metrics and consensus reaction networks.EC reduction trajectories recover both direct and indirect routes to ring-opened products and capture competing recombination outcomes relevant to early SEI growth.Extension to linear carbonates dimethyl carbonate (DMC), diethyl carbonate (DEC) and vinylene carbonate (VC) identifies recurrent radicalmediated branching, including alkyl-radical chemistry, VC-VC coupling, and early 1 oligomer-growth motifs accompanied by carbonate rearrangement.The results establish a statistically grounded workflow for exploratory reaction discovery in carbonate electrolytes and provide a practical front end for targeted higher-level DFT refinement of recurrent pathways, barrier heights, and product energetics.