Skip to content

New empirical approach for the structure and energy of covalent systems

Evidence and attribution

Authority of statements

Prose sections below (Summary, Methods, Findings, etc.) are curated summaries of the publication identified by doi, title, and pdf_path in the front matter above. They are not new primary claims by this wiki.

For definitive numerical values, reaction schemes, and interpretations, use the peer-reviewed article (and optional records under normalized/papers/ when present)—not this page alone.

Summary

This paper introduces empirical interatomic potentials for covalent materials in which bond order depends on local coordination environment—weaker bonds when an atom is over-coordinated, enabling open structures versus close-packed defaults of pair potentials. A detailed silicon potential illustrates the approach: it targets multiple polymorphs with similar cohesive energies—a known challenge—and reports extensive tests of structures, energetics, and suitability for molecular dynamics. The discussion is explicit about limitations of the early silicon parameterization.

Methods

1 — MD application (atomistic dynamics)

The article motivates potentials that are sufficiently global to be used in molecular-dynamics simulations of silicon beyond small harmonic distortions (abstract/Introduction, normalized/extracts/2011tersoff-venue-new-empirical_p1-2.txt).

  • Engine / code: Classical molecular dynamics is explicitly named as a target use case; N/A — no MD program is named on the indexed excerpt pages.
  • System size & composition: Silicon is the primary parameterized element in the excerpted introduction; N/A — supercell sizes for any illustrative MD are not stated on pp. 1–2.
  • Boundaries / periodicity: N/A — not stated on the indexed excerpt pages.
  • Ensemble / timestep / duration / thermostat / barostat / temperature / pressure: N/A — NVT/NPT/NVE production protocols, timestep sizes, trajectory segment lengths, thermostat/barostat algorithms, bath temperature schedules, and pressure control are not stated on the indexed excerpt pages.
  • Electric field: N/A — not stated on the indexed excerpt pages.
  • Replica / enhanced sampling: N/A — not stated on the indexed excerpt pages.

2 — Force-field training

  • Parent FF / elements: A new empirical bond-order-dependent potential form for covalent systems; the excerpt emphasizes two-body + bond-order modulation rather than fixed harmonic expansions about one ground-state geometry (Introduction, extract).
  • QM reference: N/A — the pp. 1–2 excerpt focuses on empirical motivation; any ab initio benchmarks used in fitting are not summarized on these pages—verify pdf_path.
  • Training set / reference data: The excerpt frames the silicon parameterization challenge around many polymorphs with similar cohesive energies (abstract/Introduction, extract).
  • Optimization: N/A — numerical optimizer / objective-function details are not stated on the indexed excerpt pages—verify pdf_path.
  • Reference data used: The excerpt contrasts empirical potentials with quantum-mechanical calculations as a competing route for energies/structures, but does not enumerate the paper’s fitted databases on pp. 1–2—verify pdf_path.

3 — Static QM / DFT-only

N/A — this is an empirical potential methods paper, not a DFT application paper.

Findings

The bond-order construction stabilizes low-coordination covalent motifs characteristic of semiconductors (e.g., tetrahedral silicon) where pairwise models favor dense packing. The fitted silicon potential is reported to capture a wide range of structural and energetic tests in the paper (with explicit discussion of limitations in Section VI of the original article), representing a substantial improvement in transferability versus earlier empirical silicon models of its era for MD-scale simulations.

Historically, this paper is a direct ancestor of Brenner-type refinements and later reactive frameworks: the key move is to let effective bond strength depend on environment, not on fixed harmonic springs.

Limitations

The model remains empirical and can fail outside the classes of configurations used in fitting; later Tersoff revisions and reactive frameworks such as ReaxFF target broader chemistry spaces than this early silicon-focused potential. - Catalog metadata in normalized/papers/ misstates the calendar year; this file uses the journal year 1988.

Relevance to group

Foundational bond-order empirical potential preceding Brenner/ReaxFF lines that saturate the corpus.

Citations and evidence anchors

  • reaxff-family
  • Empirical bond-order models (Tersoff, Brenner, ReaxFF lineage)