Predicting the preferred morphology of hexagonal boron nitride domain structure on nickel from ReaxFF-based molecular dynamics simulations
Summary¶
ReaxFF MD in LAMMPS is used to simulate hBN island growth on Ni(111) by sequential B/N deposition at controlled B : N feed ratios, comparing emergent island shapes (triangular versus hexagonal, and edge terminations) with DFT-based expectations. Adaptive biasing force (ABF) calculations in LAMMPS/Colvars support the interpretation in terms of nitrogen chemical potential. The work targets MBE-like, hydrogen-free conditions where edge polarity and metal–adsorbate chemistry, rather than a single global thermodynamic minimum, can steer domain morphology of a 2D insulator on a transition-metal substrate.
Methods¶
MD setup (Ni/B/N ReaxFF). LAMMPS with the PuReMD/Aktulga-style ReaxFF integration as cited; 5-layer FCC Ni(111) slab (12×12 surface cell, 720 Ni), periodic xy, ~90 Å vacuum; bottom Ni layer harmonically restrained (10 kcal/mol/Ų). B and N atoms deposited sequentially onto the bare surface at 0.25 ps interval with minimum initial separation 1.90 Å to avoid premature B–N pairing.
Integration. 1500 K, Nosé–Hoover thermostat; velocity Verlet; ≥10 ns trajectories; Δt = 0.25 fs.
Electronic structure reference. VASP PAW PBE Γ-only, 350 eV cutoff, 0.02 eV/Å force tolerance for DFT benchmarks cited in the paper.
Free energy. Adaptive biasing force (ABF) via Colvars in LAMMPS for selected coordinates (see article).
Barostat / target pressure: N/A — deposition and production runs use constant-volume NVT-style control at 1500 K as stated; no NPT servocontrol called out for the main growth protocol. Electric field: N/A — not used. Replica / enhanced sampling: ABF (above) is the only enhanced sampling noted; umbrella / metadynamics / replica exchange: N/A unless cited elsewhere in the article.
Findings¶
Domain morphology depends strongly on B : N ratio and N availability: B-rich feeds yield fragmented / small hBN patches; near-stoichiometric feeds produce triangular islands with B-terminated zigzag edges; N-rich feeds grow larger domains and drive transitions toward hexagonal shapes with N-terminated zigzag edges at high N excess—trends aligned with DFT-based expectations discussed in the text. Comparisons to DFT edge energetics and chemical potential arguments (including ABF-supported interpretation) appear in the article for the Ni(111) MBE-like setup. Sensitivity to deposition timing, B:N stoichiometry, and temperature (1500 K in the MD protocol) is central to the morphology map. Limitations and future items the authors flag (e.g. hydrogen-containing CVD) are summarized under ## Limitations and in the Nanoscale Discussion; corpus honesty: numbers not repeated here should be taken from the VOR PDF, not this note alone.
Limitations¶
Hydrogen-free gas-phase picture matches MBE-like conditions; CVD with hydrogen is not fully replicated.
ABF/Colvars free-energy estimates depend on collective variable choice and sampling duration; confirm barrier heights against independent DFT NEB where disagreement matters for mechanism claims.
Edge energetics on Ni can shift with adatom coverage and temperature beyond the deposition protocols explored in any single figure panel—scan conditions in the full text before generalizing morphology rules.
Nickel substrates are common catalyst supports for hBN CVD; translating these MBE-like simulation insights to H₂-rich CVD requires explicit hydrogen chemistry not fully captured here (discussion caveat).
Relevance to group¶
Ni/B/N ReaxFF application to 2D hBN on Ni, co-authored by van Duin.
Citations and evidence anchors¶
DOI: 10.1039/C8NR10291K