Skip to content

Battery interface reactivity workflow

For readers

This protocol is an execution checklist for battery interface reactivity studies that use ReaxFF as the production model while keeping key decisions anchored to DFT-derived references and corpus-grounded validation patterns. It is designed for electrolyte decomposition, interphase chemistry, and interface morphology questions where transferability limits must be made explicit.

Summary

This workflow stages battery-interface modeling into four gates: scope definition, DFT-anchored reference selection, reactive production simulation, and acceptance testing against chemistry-aware observables. In this corpus, the central pattern is to use DFT to constrain training or benchmark targets, then evaluate whether ReaxFF recovers the right qualitative and semi-quantitative trends for the specific interface regime (solid electrolyte transport, liquid-electrolyte decomposition, or coupled interface degradation). The protocol treats cross-domain transfer as a hypothesis that must be tested, not assumed.

Inputs and prerequisites

  • A defined interface question with boundaries (for example, LATP transport window, carbonate reduction near Li, or Si/SEI delithiation-driven morphology change).
  • A parameter set and element scope that already include the target chemistry or can be explicitly extended.
  • Reference dataset plan from DFT and/or published DFT-backed quantities relevant to the target mechanisms (barriers, solvation energetics, structural trends).
  • Simulation engine support for reactive MD and the selected workflow controls (ensemble, timestep, sampling windows, and any reduction-state handling strategy).
  • A validation matrix that distinguishes primary claims (must pass) from exploratory claims (may be provisional).

Procedure

  1. Define the intended inference boundary before setup.
  2. State what this campaign will claim (for example, pathway plausibility, trend direction, or rank ordering) and what it will not claim (absolute lifetime predictions, universal transferability).

  3. Classify target regime as solid-electrolyte transport, liquid-electrolyte decomposition, or mixed interface degradation.

  4. Map chemistry coverage to parameter scope.

  5. List required elements, charge-state behavior expectations, and key reactive motifs.

  6. If a required motif is outside prior training scope, mark as extension-required before production runs.

  7. Build DFT-grounded acceptance targets.

  8. Select a compact set of reference quantities directly tied to the intended claim: representative reaction barriers, solvation motifs, formation or structural energies, and local coordination trends.

  9. Record comparison tolerance as trend-level, rank-level, or value-level to avoid post-hoc metric drift.

  10. Configure initial reactive simulations for stability and realism.

  11. Run short sanity trajectories to detect unphysical bond cascades, charge pathologies, or unstable thermodynamic behavior.

  12. Verify that the selected ensemble, timestep, and temperature window support the targeted interface process.

  13. Execute mechanism-focused production sampling.

  14. For decomposition studies, capture state transitions and local environment descriptors that affect barriers (for example, coordination-dependent behavior).
  15. For solid-interface transport studies, include composition or structural realizations needed to test trend sensitivity.

  16. For degradation studies, track morphology-linked descriptors that can be compared against experimental narratives when available.

  17. Perform ReaxFF-vs-DFT and cross-evidence checks.

  18. Compare targeted ReaxFF observables against the selected DFT-grounded references.
  19. Where available, check consistency with independent experiment-linked trends (for example, conductivity order, morphology directionality, or interphase behavior).

  20. Escalate any contradiction from "minor mismatch" to "model-scope failure" when it changes mechanistic conclusions.

  21. Publish scope-qualified conclusions.

  22. Separate robust findings from hypothesis-level observations.
  23. Document transfer boundaries and unresolved failure points in the same page/report as the results.

Validation checks and acceptance criteria

  • DFT-consistency check passes for the pre-declared reference set at the chosen tolerance level (trend, ranking, or value).
  • No persistent unphysical reaction channel dominates production trajectories in the target window.
  • Claimed mechanism is supported by at least two independent evidence dimensions (for example, reactive trajectory behavior plus DFT benchmark trend, or trajectory trend plus experiment-coupled observation).
  • Sensitivity analysis across key setup levers (composition, local solvation state, or cycle stage) does not invert the headline conclusion unless explicitly reported.
  • Final claim language is scoped to the validated chemistry and phase window.

Failure modes and mitigations

  • Missing chemistry in training scope -> add targeted DFT references and re-fit or restrict claim scope to observed-valid regions.
  • Over-interpreting decomposition pathways from short or unstable trajectories -> add staged equilibration, replicate trajectories, and environment-conditioned analysis.
  • Treating one successful regime as proof of broad transferability -> require explicit out-of-domain holdout checks before reuse.
  • Confusing pathway discovery with predictive ranking -> separate plausibility claims from rate/selectivity claims unless validated against stronger references.
  • Ignoring interface morphology evidence in degradation studies -> include experiment-linked descriptors when corpus anchors provide them.

Variants and when to choose them

  • Solid electrolyte transport variant: use when the objective is composition- and disorder-sensitive ion-transport trends in ceramic frameworks.
  • Liquid electrolyte decomposition variant: use when Li speciation and local solvation structure control reduction/decomposition pathways.
  • Coupled experiment-simulation degradation variant: use when interface morphology evolution is central and microscopy or analogous observations are available.
  • Method-escalation variant (ReaxFF -> higher-fidelity refinement): use when key acceptance checks fail but trajectory-level mechanism hypotheses remain valuable for narrowing a DFT follow-up set.

Evidence anchors