Skip to content

Reaction pathways in atomistic models of thin film growth

Evidence and attribution

Authority of statements

Prose below summarizes the publication identified by doi, title, and pdf_path in the front matter. For barriers, rates, and case-study parameters, rely on the JCP article and any full text under papers/—local extraction is partial (first pages only).

Summary

Thin-film growth involves diffusive and concerted atomic moves on or near a substrate; at experimental deposition rates, brute-force molecular dynamics cannot reach the relevant timescales because of the Arrhenius separation between MD timesteps (\(\sim 10^{-15}\) s) and growth events. This perspective introduces adaptive kinetic Monte Carlo (AKMC): from each local minimum, transition rates to neighboring states are estimated, events (including deposition) are sampled probabilistically, and the clock advances by \(\Delta t = -\ln u / \sum_i R_i\). Adaptations for frequent low-barrier moves are discussed. Worked examples include Ag on Ag(100) and ZnO growth using a ReaxFF parametrization for ZnO with LAMMPS-based energy evaluations, illustrating how AKMC catalogs complex, non-intuitive pathways (exchange, concerted hops, cluster moves) for metals and metallic oxides.

Methods

1 — MD application (where MD appears vs AKMC). The perspective’s primary method is adaptive kinetic Monte Carlo (AKMC): from each local minimum, escape pathways are explored using harmonic transition-state theory-style treatments and energy/force calls (the ZnO growth illustration uses LAMMPS-evaluated ReaxFF energies). Conventional MD appears mainly as a cross-check where Ag on Ag(100) dynamics overlap AKMC predictions at 300 K on accessible MD timescales, using periodic boundary conditions for the surface slab models described in the perspective. N/A — single unified MD production protocol (timestep, multi-ns trajectory, thermostat damping): the article is not a benchmark NVT/NPT study of one slab; it instead contrasts AKMC with MD, TAD, hyperdynamics, and parallel replica for rare-event surface kinetics. N/A — barostat / macroscopic pressure control in AKMC framing: the focus is event catalogs and rates, not constant-pressure MD of a bulk reservoir.

2 — Force-field training / fitting. Worked ZnO example: uses a published ReaxFF parameterization for ZnO (cited) as the interatomic model inside AKMC moves—no new QM refit is performed in this perspective.

3 — Static QM / DFT. N/A — standalone DFT production protocol: DFT enters as literature context for barriers and materials systems, not as the paper’s own ab initio workflow.

4 — Review / non-simulation framing. JCP perspective on rare-event algorithms for thin-film growth; the Methods substance is literature scope + algorithmic comparison, per AGENTS.md block 4.

Findings

AKMC exposes rare-event pathways that are impractical to sample with long MD at experimental fluxes. The Ag and ZnO illustrations show qualitative consistency with MD where timescales overlap, while extending to regimes relevant to deposition rates; the discussion stresses the need to enumerate pathways up to barriers set by the competition between diffusion and arrival kinetics (order \(\sim 0.6\) eV for a representative surface area at 300 K and \(\sim 10\) monolayers per second, as stated in the paper).

Limitations

  • Summary here draws on a partial text extract; barrier values, supercell sizes, and AKMC parameters for each figure should be taken from the full PDF.
  • AKMC accuracy depends on completeness of the event catalog and on the quality of the interatomic model (EAM/EMT vs ReaxFF for oxides); transferability to new chemistries is not automatic.
  • Work is methodological and illustrative rather than a single-material benchmark against experiment.

Relevance to group

Not authored by the van Duin group. It is still useful corpus context: ReaxFF is used as the reactive engine in the ZnO growth example alongside AKMC, showing how reactive force fields plug into rare-event surface kinetics workflows that complement reactive MD studies elsewhere in the knowledge base.

Citations and evidence anchors

  • DOI: 10.1063/1.4986402
  • Text-aligned pointers: normalized/extracts/2017lloyd-venue-paper_p1-2.txt (introduction and AKMC formulation).

Reader notes (navigation)

  • Force-field context: reaxff-family — ZnO growth example uses a ReaxFF parametrization cited in the article.
  • Methods / rare events: complements standard reactive MD when experimental time scales are out of reach for MD alone.