Skip to content

Binding and diffusion of lithium in graphite: quantum Monte Carlo benchmarks and validation of van der Waals density functional methods

Evidence and attribution

Authority of statements

Prose below summarizes the publication identified by doi, title, and pdf_path in the front matter. For definitive numerical values and figures, use the peer-reviewed article.

Summary

Diffusion quantum Monte Carlo (DMC) benchmarks for Li adsorption and diffusion in AA-stacked graphite are compared to several van der Waals–aware DFT approximations. AA graphite is used as a controlled stacking to locate dilute Li sites; QMC lattice constants for pure AA graphite match experiment. Many vdW-corrected DFT recipes that work for AB graphite are shown to struggle for AA graphite; among those tested, vdW-DF2 scores best overall for AA graphite plus Li binding and diffusion, though binding-energy errors remain. Charge-aware vdW corrections (e.g., TS-vdW) are motivated over fixed empirical dispersion when charge transfer matters (abstract; introduction, extract).

Methods

1 — MD application

N/A — this work is DMC and DFT benchmarking of Li in graphite, not production classical or ReaxFF MD.

2 — Force-field training

N/A — not applicable.

3 — Static QM / QMC (Li in AA graphite)

  • QMC (DMC): Diffusion Monte Carlo benchmarks adsorption and diffusion of dilute Li in AA-stacked graphite (symmetry-determined sites; abstract; normalized/extracts/2014ganesh-venue-binding-diffusion_p1-2.txt).
  • Supercells: smaller cell—two graphene layers of 50 C each with optional single Li (\(x=0.06\) in Li\(_x\)C\(_6\)); larger “doubled” cell—200 C with two Li at the same concentration; binding from the larger cell and diffusion barriers from the smaller (extract).
  • Geometry protocol: in-plane C–C fixed at experimental 1.421 Å; interlayer separation scanned to probe vdW treatment (extract).
  • DFT / functional family: comparisons include vdW-DF, vdW-DF2, DFT-D2, and Hirshfeld-partitioned TS-vdW-style charge-aware corrections (abstract; introduction, extract).
  • Dispersion: central question is how vdW is treated across these families (abstract).
  • Basis / k-mesh / pathway details: full plane-wave settings, Brillouin-zone k-mesh choices, and barrier definitions are given in JCTC Methods in pdf_path (not duplicated from the short extract).

Findings

1 — Outcomes and mechanisms

DMC lattice constants for pure AA graphite agree with experiment. AA-stacked graphite is shown to challenge many vdW-inclusive DFT recipes even when those recipes work for conventional AB graphite. Across AA graphite and Li binding and diffusion, vdW-DF2 achieves the highest overall DFT accuracy in their comparison, though binding-energy errors remain. Empirical dispersion (DFT-D) approaches are unreliable unless local charge transfer is accounted for (motivating Hirshfeld-weighted schemes such as TS-vdW). Overall, accurate Li–graphite modeling requires simultaneous treatment of charge transfer and dispersion, favoring self-consistent vdW-inclusive functionals (abstract; extract pages 1–2).

2 — Comparisons

  • DMC vs several vdW-inclusive DFT recipes; DFT rankings vs QMC references (abstract).

3 — Sensitivity

  • Interlayer separation scans couple Li energetics to graphite vdW treatment (extract).

4 — Limitations / outlook

  • AA stacking is a computational construct vs ground-state AB graphite (## Limitations).

5 — Corpus / KB honesty

  • Detailed barrier numbers and DFT settings must be read from pdf_path; this page tracks abstract-level claims only.

Limitations

AA stacking is a computational convenience versus ground-state AB graphite; DMC and DFT comparisons inherit respective cost and functional biases as discussed in the paper.

Citations and evidence anchors

  • DOI 10.1021/ct500617z (article footer in extract).
  • Abstract (extract page 1).