Skip to content

Development and initial applications of an e-ReaxFF description of Ag nanoclusters

First e-ReaxFF parametrization for silver with explicit classical electrons, reproducing 2D vs 3D isomer transitions for Ag\(_{N\le 20}\) clusters against DFT/CCSD(T) references, with demonstrations on Ag-halide redox and plasmon-like electron dynamics.

Summary

Small Ag clusters favor 2D configurations at certain sizes and transition to 3D with only a few added atoms—a pattern standard EAM misses. The authors augment ReaxFF with explicit electrons (e-ReaxFF) and a bond-order-dependent dihedral penalty that disfavors premature 3D branching for small clusters while allowing 3D at larger sizes. After training on QM energy ordering and barriers, NVT MD in the standalone ReaxFF code explores halide redox and electron motion phenomena not accessible to fixed-charge models.

The work is a methodological bridge between classical metallic potentials and quantum accuracy for sub-nanometer silver, where delocalized electronic effects matter.

Methods

  • QM references: DFT and CCSD(T) data for Ag\(_N\) isomers and selected reactions (functionals/basis sets per article).
  • e-ReaxFF form: Explicit electron particles paired with Ag\(^+\) cores; four-body dihedral term controlling 2D/3D competition; parameters fit to reproduce 2D–3D transition near Ag\(_5\)–Ag\(_7\) sizes as reported.
  • MD (application / validation): Molecular dynamics in the standalone ReaxFF code; NVT; Nosé–Hoover thermostat; timestep 0.250 fs to resolve fast electron motion (Methods). System size (Ag cores + explicit electrons): on the order of 10–200 atoms in the cluster examples (\(N \le 20\) in the main training corpus, with extensions in the article). Boundaries / PBC: open-boundary, isolated Ag\(_N\) cluster supercell (non-bulk, PBC: N/A in the free-cluster sense of the code’s treatment). Duration (eq./prod.): N/A to transcribe all ps segments here; see J. Chem. Phys. Temperature (thermostat setpoints): 300 K (and other values in application sections, per the paper). Barostat: N/A; hydrostatic pressure: N/A; external electric field: N/A; umbrella / replica / metadynamics: N/A.

Training emphasizes relative isomer energies and barriers because cluster databases are sparse compared to bulk silver benchmarks.

Findings

  • New potential tracks the 2D lowest-energy structures for small \(N\) and the onset of 3D stability by Ag\(_7\), consistent with QM references.
  • EAM comparisons highlight failures for planar vs nonplanar energetics in the sub-20 atom regime where e-ReaxFF agrees more closely with QM ordering.
  • Case studies show e-ReaxFF capturing oxidation-state changes in silver halide-like reactions and enabling plasmon-relevant electron dynamics in MD (as illustrated in the paper’s application sections).

Demonstrating explicit electron dynamics in MD differentiates this line from prior ReaxFF metals limited to formal oxidation states embedded in bond orders alone.

The J. Chem. Phys. article documents the full training corpus, standalone integrator details, and additional cluster sizes beyond Ag\(_{N\le 20}\) for readers extending the parametrization to surfaces or salts.

Explicit-electron workflows remain more expensive than fixed-charge ReaxFF; scale tests should consult the reported timings in the article.

Limitations

e-ReaxFF adds cost and parameter complexity; transfer to bulk Ag surfaces and alloys requires separate validation.

Relevance to group

Foundational van Duin-group e-ReaxFF extension to metallic silver, cited as the first e-ReaxFF metal parametrization in the article.

Citations and evidence anchors