Skip to content

Reactive force field study of Li/C systems for electrical energy storage

Evidence and attribution

Authority of statements

Prose below summarizes the publication identified by doi and pdf_path. This ingest is an ACS proof layout for the same JCTC article as 2015raju-venue-ct-2014-01027v when that sibling exists.

Summary

Graphitic carbon remains the dominant Li-ion anode, yet few simulations simultaneously capture Li energetics, staging, and kinetics in graphite and related nanostructures at sizes and concentrations relevant to experiment. Raju, Ganesh, Kent, and van Duin develop a ReaxFF parametrization for Li–C using van der Waals–corrected DFT as the QM reference, then apply it in grand canonical Monte Carlo (GCMC) studies of Li intercalation in perfect graphite at supercell sizes of order ~1000 atoms. The reported GCMC voltage profile agrees with known experimental and DFT behavior, and the simulations reproduce in-plane Li ordering and interlayer spacing associated with stage I and II staging. Defective graphite with ~1–2% point and topological vacancy content shifts Li/C (capacity) and voltage together with signatures of metallic lithium, connecting to recent lithium plating experiments. Additional model studies on 0D (onion-like) and 1D (nanorod) carbon highlight geometry-dependent pathways: a defective onion-like particle favors fast charge/discharge via surface Li adsorption, whereas a defect-free nanorod requires a critical Li density at edges before intercalation proceeds.

Methods

The authors fit Li–C interactions within ReaxFF using van der Waals–corrected DFT as the quantum reference, then apply the field in grand canonical Monte Carlo (GCMC) studies of Li intercalation in perfect graphite with supercells of order ~1000 atoms (abstract and Sec. II opening). ReaxFF evaluates bonded and nonbonded contributions (bond, valence, torsion, over/under-coordination, lone pair, vdW, Coulomb) with bond orders updated at each MD or minimization step, as summarized in Sec. II.A of the article. GCMC exchanges Li against a reservoir to sample lithiation thermodynamics and staging-related structure. Defective graphite models add point and topological vacancies at 1–2% density to probe how Li/C, voltage, and metallic Li signatures shift relative to the perfect lattice. Additional calculations treat 0D (onion-like) and 1D (nanorod) carbons to compare surface Li adsorption with edge-limited intercalation pathways.

Engine, timestep, thermostat/barostat settings for any energy-relaxation or MD segments paired with GCMC, and full MC sweep statistics, are given in the Computational Methods and SI tables on pdf_path (and the journal-layout sibling 2015raju-venue-ct-2014-01027v)—they are not recoverable from the short indexed excerpt used for this page. System size: ~1000 atoms graphite supercells in the GCMC abstract description, plus larger 0D/1D carbon models as built in the article. Boundaries: bulk graphite and nanostructure cells use 3D periodic supercells as defined there. Ensemble: grand canonical Li exchange for intercalation sampling; barostat / hydrostatic pressure control: N/A for the GCMC-centric lithiation workflow as summarized. Temperature: isothermal temperature setpoints for each reported voltage/staging curve are listed in Methods on pdf_path (not transcribed from the excerpt here). Duration: production GCMC sweep counts and any coupled MD production segment lengths (ps/ns) appear in those same tables. Electric field: N/A. Replica / enhanced sampling: N/A (GCMC, not umbrella or replica exchange).

Force-field training. Parent: reactive C framework extended to Li–C. QM reference: vdW-corrected DFT on graphite-related and defective condensed-carbon configurations (introduction). Training set: equations of state and intercalation motifs spanning perfect and defective graphitic environments (full enumeration in the article). Optimization: ReaxFF refit to those targets using the group’s standard least-squares protocol (software and weights in pdf_path). Validation: GCMC voltage and staging observables compared with experiment and DFT references cited in the abstract.

Static QM / DFT-only production: N/A — periodic DFT supplies training and benchmark data; headline results are ReaxFF + GCMC, not standalone AIMD lithiation trajectories.

Findings

Perfect graphite: GCMC with the new Li–C ReaxFF reproduces a voltage profile consistent with known experimental and DFT results and captures in-plane ordering plus interlayer separations characteristic of stage I and II compounds (abstract).

Defective graphite: As vacancy content increases toward the 1–2% regime modeled, Li/C ratio (capacity) and voltage shift together with behavior linked to metallic lithium, interpreted in the article as a microscopic rationalization for lithium plating trends seen in recent experiments (abstract).

Nanostructured carbons: 0D onion-like models favor fast charge/discharge dominated by surface Li adsorption, whereas a 1D defect-free nanorod requires a critical Li density at edges before intercalation becomes favorable—illustrating how geometry and defects reroute lithiation kinetics within the same force field (abstract).

Corpus honesty: This file is a proof PDF; prefer the journal-layout ingest for pagination-sensitive citations when both exist ([[2015raju-venue-ct-2014-01027v]]).

Limitations

Proof PDFs can show watermarks, color shifts, or figure-resolution differences from the version of record. Electrolyte, SEI, and continuum transport outside the ReaxFF+GCMC stack are outside the paper’s atomistic scope.

Relevance to group

Flagship Li-ion anode reactive modeling led by Raju with van Duin and ORNL coauthors.

Citations and evidence anchors