Skip to content

Parametrization of a reactive many-body potential for Mo–S systems

Evidence and attribution

Authority of statements

This page summarizes the Phys. Rev. B article by doi. The potential is second-generation REBO-style, not ReaxFF.

Summary

The authors introduce a reactive many-body Mo–S potential using the second-generation reactive empirical bond-order formalism, augmenting the bond-order term with a coordination-dependent analytic function. Fitting uses weighted nonlinear least squares with curated training sets (functions, databases, initial guesses, residual weights). The abstract reports agreement for Mo clusters, 2D Mo, 3D Mo crystals, small S molecules, and binary Mo–S crystals, and demonstrates friction between MoS\(_2\) layers consistent with prior DFT static surface calculations.

Methods

The potential follows the second-generation reactive empirical bond-order (REBO) formalism for Mo–S: a parametrized bond-order factor multiplies pairwise interactions, augmented by an analytic coordination-dependent function on the bond-order term to capture how coordination modulates binding energy. Lennard–Jones terms are included for van der Waals interactions between S–Mo–S trilayers as developed in the paper. Parameters are obtained by weighted nonlinear least squares with explicit control of analytic forms, training databases, initial guesses, and residual weights (the four fitting levers highlighted in the abstract).

1 — MD application (friction showcase). After fitting, the authors illustrate the potential with molecular dynamics of friction between MoS\(_2\) layers (Section III in the PRB article). N/A — MD engine, ensemble, timestep, thermostat/barostat, temperature, normal load, shear rate, system sizes, and PBC details are not stated in normalized/extracts/2009liang-venue-paper_p1-2.txt—use pdf_path for executable protocol text.

2 — Force-field training. Parent / form: second-generation REBO-class Mo–S potential with explicit coordination function on bond order plus LJ interlayer vdW. QM reference: the abstract states agreement with DFT for the authors’ prior static potential-energy surfaces used to contextualize the friction MD; N/A — full DFT code/functional/basis/\(k\)-mesh listing from the short excerpt alone. Training set: Mo clusters, 2D Mo, 3D Mo crystals, small S molecules, and binary Mo–S crystals (abstract). Optimization: weighted nonlinear least squares with the four levers above. Reference data: QM/DFT and structural/energetic targets as described in the paper beyond the excerpt.

Checklist closure (indexed pages). System / composition: MoS\(_2\) sliding surfaces (binary Mo–S composition); per-cell atom counts: N/A — not stated in the short extract. Ensemble: N/A — NVT/NPT/NVE not stated on pp. 1–2. Duration / stages: N/A — production run lengths not stated on pp. 1–2. Pressure: N/A — pressure / normal load for the friction MD not stated in the excerpt.

Findings

Fitting performance. The fitted potential is reported to yield good agreement with structures and energetics for Mo clusters, 2D Mo, 3D Mo crystals, small sulfur molecules, and binary Mo–S crystals in the training set enumerated in the abstract.

Application: MoS\(_2\) friction MD. Shearing MoS\(_2\) layers with the new potential produces friction results described as consistent with the authors’ previous static DFT potential-energy surfaces, motivating reactive MD at scales ab initio MD cannot reach (as claimed in the abstract).

Comparisons. The indexed excerpt emphasizes comparison to the authors’ own prior DFT surfaces for the friction illustration; broader experimental tribology benchmarks are N/A — not summarized on the indexed pages.

Corpus honesty. extraction_quality is partial; numerical friction traces, contact geometry, and sensitivity studies are in pdf_path, not the short extract.

Limitations

REBO differs from ReaxFF in functional form and fitting philosophy; parameters are not transferable between frameworks without retraining. extraction_quality is partial; final parameter tables and benchmarks live in the PDF.

Relevance to group

Methodological neighbor: reactive bond-order MD for TMD tribology—useful contrast case when scoping ReaxFF applications to sulfide/metal-oxide chemistry.

Citations and evidence anchors

  • DOI: 10.1103/PhysRevB.79.245110.
  • PDF: papers/Others/Liang_MoS2_PRB_2009.pdf.
  • Extract: normalized/extracts/2009liang-venue-paper_p1-2.txt.