Skip to content

Inter-layer potential for hexagonal boron nitride

Evidence and attribution

Authority of statements

Prose below summarizes the publication identified by doi, title, and pdf_path in the front matter. For definitive numerical values and figures, use the peer-reviewed article.

Summary

Stacked hexagonal boron nitride (h-BN) is a prototypical van der Waals layered solid: accurate interlayer energetics control exfoliation, friction, and self-assembly, yet generic Lennard-Jones stacks miss registry-dependent anisotropy and quantitatively wrong binding/sliding can derail mesoscale models. Leven, Azuri, Kronik, and Hod introduce a dedicated interlayer potential (ILP) for h-BN that couples dispersion, anisotropic short-range repulsion, and layer-resolved electrostatics fitted to high-quality DFT references. The resulting model is intended for large-scale molecular dynamics of bilayers, few-layer stacks, and double-walled boron nitride nanotubes (DWBNNTs) where interlayer motion—not intralayer bond rearrangement—is the primary degree of freedom.

Methods

The ILP superposes three physical contributions: (i) dispersion attraction formulated in the spirit of Tkatchenko–Scheffler (TS-vdW) corrections so long-range attraction tracks polarizability and layer separation; (ii) anisotropic Pauli repulsion expressed in a Kolmogorov–Crespi-type registry-dependent form that captures how interlayer overlap depends on in-plane alignment; and (iii) classical monopolar electrostatics between partially charged atoms to represent in-plane ionicity within each BN sheet. Parameters are adjusted against benchmark DFT datasets for interlayer binding curves, in-plane shear/sliding paths on h-BN bilayers, and interlayer telescoping and rotation in DWBNNT pairs spanning multiple relative orientations (see J. Chem. Phys. 140, 104106 and the local extract). Intralayer bonding remains outside this interlayer functional: simulations pair the ILP with an intralayer model appropriate for covalent BN sheets.

1 — MD application (intended use of the ILP)

  • Engine / code: Molecular dynamics with an interlayer potential is the intended downstream use case for large-scale h-BN assemblies (abstract); N/A — specific MD package and timestep not stated on extract p1–2 (JCP article).
  • System targets / composition: bilayer sliding landscapes, few-layer stacks, and double-walled BN nanotube interlayer modes as described in the abstract/introduction (extract); treat these as explicit atomistic targets paired with separate intralayer models.
  • Boundaries / periodicity: Periodic in-plane boundary conditions are standard for bilayer sliding energy maps in this article class; N/A — explicit PBC strings not on extract p1–2.
  • Ensemble: NVT is a common choice for classical MD validation of interlayer models, but N/A — not stated on extract p1–2 (JCP Methods).
  • Timestep / thermostat: N/A — not stated on extract p1–2 (JCP Methods for any validation MD).
  • Duration / stages: Equilibration/production lengths for any reported MD validation are N/A — not on extract p1–2.
  • Barostat / hydrostatic pressure: N/A — pressure control not stated on extract p1–2 (interlayer distance scans are typically energy landscapes).
  • Temperature: N/A — explicit MD temperature set points not stated on extract p1–2.
  • Electric field: N/A — not indicated in the indexed extract opener.
  • Replica / enhanced sampling: N/A — not indicated in the indexed extract opener.

2 — Force-field training (classical interlayer potential, not ReaxFF)

  • Functional form (“parent” models): h-BN ILP combines (i) a Tkatchenko–Scheffler (TS-vdW)-style dispersion attraction, (ii) a Kolmogorov–Crespi-type registry-dependent repulsion, and (iii) classical monopolar electrostatics for in-plane ionicity (extract; J. Chem. Phys. 140, 104106).
  • QM reference / training set: parameters adjusted to advanced DFT benchmarks covering interlayer binding, sliding, and DWBNNT telescoping/rotation across registries (extract); N/A — full functional/basis/k-mesh tables not on extract p1–2.
  • Optimization / software: parameter adjustment workflow as described in the article; N/A — optimizer implementation details not on extract p1–2.
  • External reference data: DFT benchmark energies/landscapes used for fitting and validation (article).

Findings

Across the benchmarks used in the paper, a single interlayer parametrization reproduces binding energies, sliding barriers, and telescoping/rotation energetics for DWBNNT pairs with different crystallographic registry, supporting transferability within the h-BN stacking manifold targeted by the fit. Practically, this enables classical MD of layered BN assemblies—twists, stacks, and nanotube bundles—at sizes and timescales impractical for DFT or full reactive treatments, provided chemistry is limited to non-reactive interlayer motion. The approach is complementary to intralayer ReaxFF or Tersoff-style models: researchers should not expect ILP alone to describe sp³ defect chemistry or nitrogen vacancy formation without augmenting the Hamiltonian.

Limitations

Intralayer bonding uses a separate model; interlayer potential is not a ReaxFF reactive treatment. Users should verify interlayer parameter units and cutoffs against the JCP article when mixing ILP with intralayer engines in LAMMPS input decks.

Citations and evidence anchors

  • DOI 10.1063/1.4867272 (extract citation line).
  • Title and abstract opening (extract page 1).