Lithium ion solvation and diffusion in bulk organic electrolytes from first-principles and classical reactive molecular dynamics
Evidence and attribution¶
Authority of statements
Prose below summarizes the J. Phys. Chem. B article identified by doi and pdf_path. Quantitative diffusion and coordination values should be checked against the full Methods/Results in the PDF.
Summary¶
Born–Oppenheimer AIMD (VASP, PAW–PBE, Γ-only, 450 eV, NVT Nosé–Hoover, 0.5 fs, 330 K) characterizes Li⁺/PF₆⁻ solvation and diffusion in EC, EMC, and 3:7 EC/EMC at dilute ~0.23 M salt conditions. ReaxFF (LAMMPS) extends selected electrolytes to ~1 ns and ~6400 atoms to probe finite-size and sampling effects versus DFT reference. The abstract emphasizes tetrahedral Li⁺ first shells involving carbonyl and/or ether oxygens (sometimes PF₆⁻ participation), larger diffusivity in linear EMC than cyclic EC, and weaker, poorly defined PF₆⁻ first shells associated with higher anion mobility—motivating electrolyte design heuristics (abstract, extract).
Methods¶
Static QM / AIMD (primary)¶
Born–Oppenheimer AIMD in VASP with PAW potentials and the PBE GGA; Γ-only Brillouin sampling; 450 eV plane-wave cutoff (as in the article’s Methods). Born–Oppenheimer propagation uses NVT Nosé–Hoover chains (~1000 cm⁻¹ frequency, ~32 fs period), Δt = 0.5 fs, 330 K, with ~5–7.5 ps equilibration followed by ~30 ps production trajectories for EC, EMC, and 3:7 EC/EMC electrolytes at dilute ~0.23 M LiPF₆-like stoichiometry in three-dimensional periodic boundary conditions (PBC) cubic supercells (lattice vectors in pdf_path). Analysis: radial distribution functions, coordination numbers, residence times, and Li⁺/PF₆⁻ diffusion coefficients from VACF and MSD treatments with 5–15 ps averaging windows and 50 fs time origins as reported.
MD application (ReaxFF validation / scaling)¶
LAMMPS ReaxFF runs use NVT with a Nosé–Hoover chain thermostat (three chains), Δt = 0.25 fs, ~1333 cm⁻¹ chain frequency; an illustrative EC cell (630 EC + 10 LiPF₆ formula units) uses ~125 ps equilibration plus ~1 ns production at 330 K to probe finite-size and sampling effects relative to AIMD, likewise under 3D PBC with box vectors reported in the article.
Force-field training¶
N/A — this work uses a published organic electrolyte ReaxFF parametrization for comparison to AIMD; it is not primarily a new ReaxFF fitting paper.
Bulk AIMD and ReaxFF validation trajectories are constant-volume (NVT); barostat / applied pressure, electric fields, and umbrella-style enhanced sampling are not part of the protocols summarized above.
Findings¶
Solvation: Li⁺ is coordinated in roughly tetrahedral fashion by carbonyl and/or ether oxygens of the carbonates and sometimes by PF₆⁻, with preferred motifs depending on EC vs EMC vs mixture. Diffusion: Li⁺ diffusivity is somewhat larger in EMC than in EC in the authors’ calculations; magnitude tracks strength of Li⁺ solvation in their analysis. PF₆⁻ shows higher diffusivity when its first solvation shell is weakly bound / poorly defined. Model comparison: ReaxFF reproduces many Li⁺–O coordination features versus AIMD but can differ in ether-oxygen angular/PCF detail; reported diffusion coefficients for ReaxFF vs AIMD agree within roughly 40–50% in the cases highlighted—used to argue for ReaxFF as a scalable complement while noting DFT finite-size limits.
Limitations¶
Dilute salt vs commercial concentrations; PBE electrolyte chemistry; ReaxFF organics parametrization limits for subtle EC/EMC ether conformations.
Relevance to group¶
Adri C. T. van Duin co-authorship; couples AIMD electrolyte benchmarks with ReaxFF scaling arguments.
Citations and evidence anchors¶
DOI 10.1021/jp508184f.