Skip to content

Development of a Transferable ReaxFF Parameter Set for Carbon- and Silicon-Based Solid Systems

A H/C/N/O/Si Reaxff parameter set is fitted to first-principles data using simulated annealing to support tribochemistry of carbon- and silicon-based solid lubricants with H₂O, H₂, and O₂. Oxidation MD of SiC and Si₃N₄ in O₂ is compared to experiments on high-temperature oxidation.

Summary

The authors unify H/C/N/O/Si chemistry for graphite/DLC/SiC/Si₃N₄/SiO₂-class environments under tribo-relevant molecules. Simulated annealing Reaxff optimization on DFT training data is followed by static reaction energy tests and ~100 ps NVT oxidation MD at 1200 K with 0.25 fs in LASKYO (in-house) on SiC and Si₃N₄ with O₂ (Section 3.2). The DFT database (Section 2) covers molecules, clusters, and condensed phases so the parameter set can bridge gas-phase oxidants and covalent ceramics in one framework**.

Methods

1 — MD application. The authors run reactive molecular dynamics in LASKYO (in-house integrator); 0.25 fs time step; NVT; ~100 ps O₂ atmosphere SiC and Si₃N₄ cells at 1200 K; O–Si/O–N bond counts vs time; N/A — full thermostat label if the wiki omits it—confirm PDF; N/ANPT; N/AE-field; N/Ametadynamics. PBC supercells for gas+solid as in Section 3.2. Normal hydrostatic pressure N/ANVT constant-volume oxidation cells (not NPT pressurization in excerpt).

2 — Force-field training. H/C/N/O/Si Reaxff; DFT reference on clusters, molecules, and condensed phases (SiO₂, Si₃N₄, Si, diamond); simulated annealing global search of parameters (Section 2). Validation on H₂, H₂O, H₂O₂, O₂ reaction energies on diamond(111), Si(111), SiC(111), Si₃N₄(100), SiO₂(001) (Figures 5–9). Reference data: DFT; MD oxidation vs experiments (500–1800 K trends).

3 — Static QM. The DFT level and k-sampling are in Section 2; N/A for a separate k-point paragraph in this short wiki notesee pdf_path.

4 — Review or non-simulation. N/A

Findings

Outcomes and mechanisms. The H/C/N/O/Si set reproduces qualitatively correct environmental molecule + surface chemistry in the static and MD tests selected. SiC and Si₃N₄ O₂ MD tracks experimental high-T oxidation behavior (thicknesses/ rates as cited).

Comparisons and sensitivity. MD+kinetics vs experiments; T (1200 K MD window vs wider expt range).

Authored limitations and outlook. Gas oxidation is an intermediate test; tribo with shear/third-body is hardersee ## Limitations.

Corpus honesty. JPCC VOR tables favor pdf_path.

Limitations

Simulated annealing fits can underperform on states outside the training manifold; tribointerfaces with shear and third-body particles add complexity beyond gas oxidation tests.

Relevance to group

Independent Reaxff parameterization line for SiC/Si₃N₄/carbon tribology—useful cross-reference for group Si/O/C models.

Citations and evidence anchors

  • DOI: 10.1021/acs.jpcc.0c01645