Low-Temperature Transport in Li-Ion Battery EC/EMC/FEC Electrolytes: Molecular Dynamics and Machine-Learning Modeling
Low-temperature operation imposes severe limitations on lithium-ion transport in battery electrolytes, yet the coupled effects of solvent composition and fluorinated additives in the cold-temperature regime remain insufficiently resolved. Here, we combine classical molecular dynamics (MD) and machine learning (ML) to investigate 1 M LiPF$_6$ electrolytes containing ethylene carbonate (EC), ethyl methyl carbonate (EMC), and EC/EMC (3:7), with 0-10 mol% fluoroethylene carbonate (FEC), from 298 to 233 K. MD simulations quantify Li$^+$ self-diffusion, Nernst-Einstein (NE) and Green-Kubo (GK) conductivities and local coordination, while Gaussian-process surrogates model conductivity across composition and temperature. Cooling produces a pronounced transport penalty, particularly in EMC-containing electrolytes, whose GK conductivity decreases by more than 98% at 233 K, compared with approximately 90% in EC-rich systems. Li$^+$ self-diffusion activation energies are 0.49-0.54 eV for EMC-containing systems and 0.27-0.29 eV for EC-based systems. Within the EC family, 0-2 mol% FEC gives comparable cold-temperature transport, whereas 5-10 mol% FEC shows lower conductivity retention at the coldest simulated temperature. The analyzed coordination channels remain solvent dominated, while direct Li$^+$-FEC coordination is not quantified in the present RDF set.The Gaussian-process surrogates achieve composition-disjoint cross-validated RMSE values of 0.56 and 0.57 mS cm$^{-1}$ for NE and GK conductivity. Within the simulated liquid-state trajectories, temperature and host-solvent composition dominate the bulk-transport response, with FEC acting as a secondary modifier.