Growth of Ni nanoclusters on irradiated graphene: a molecular dynamics study

Evidence and attribution¶

Authority of statements

Prose below summarizes the PCCP article identified by doi, title, and pdf_path.

Summary¶

ReaxFF MD (LAMMPS) simulates soft deposition of Ni atoms onto defective graphene sheets containing tunable monovacancy concentrations (0.25–1.0% removal levels in the study). Ni is inserted every 5 ps in a 1–2.5 Å “safe” height window with 0.5 fs timestep to avoid impact damage; post-deposition dynamics extend ≥2 ns. The work emphasizes non-equilibrium cluster-size distributions: frequency vs cluster size decreases monotonically, yet most atoms can reside in a few large clusters—behavior captured by a simple cross-section–weighted attachment model and Monte Carlo illustrations—contrasting with hard-landing scenarios that yield more monodisperse islands.

Motivation ties Ni on CVD graphene to catalytic growth and interconnect applications where vacancy density from irradiation or transfer steps can steer nucleation statistics; comparing soft versus hard deposition protocols therefore tests how impact energy sets emergent island-size heterogeneity.

Methods¶

Engine / potential: LAMMPS ReaxFF for C–C, Ni–Ni, Ni–C using Ni/C parameters from Yoon et al. (DFT-trained Ni phases); qualitative discussion of binding/diffusion metrics vs DFT literature in the paper.
Substrate: ~10.2 × 10.3 nm graphene (3936 C), periodic in-plane; random monovacancies to target porosity; conjugate-gradient relaxation; 0.2 ns at 2500 K with zero-pressure barostat to accommodate vacancy strain (per protocol in §2.1).
Deposition: Ni atoms added every 5 ps between 1.0–2.5 Å above the sheet; Nosé–Hoover thermostat for diffusion segments; production runs ≥2.0 ns after deposition completes.
Ensemble: Initial defective-sheet relaxation uses NPT-style zero-pressure control at 2500 K (0.2 ns) as in §2.1; Ni soft-landing and post-deposition evolution use NVT segments with Nosé–Hoover thermal control as described in the PCCP Methods.
Timestep: 0.5 fs during Ni insertion to limit impact damage; longer post-deposit segments follow the article’s production settings (see papers/ReaxFF_others/Valencia-Phys.Chem.Chem.Phys.2018-20-16347.pdf for any rescaled timestep after deposition).
Modeling: Phenomenological rate/Monte Carlo treatment of capture cross-sections vs cluster size to interpret broad, heavy-tailed cluster statistics.
Electric field: N/A — not used.
Replica / enhanced sampling: N/A — not used (standard time-stepped MD plus the analytical/Monte Carlo post-model noted above).

Findings¶

Mobility + pinning: High Ni diffusivity on graphene plus strong vacancy binding yields growth where large clusters accrete wandering atoms efficiently once their cross-section exceeds competing vacancies.
Bimodal statistics: Cluster-count vs size decreases with size, but cumulative weight shows a minority of very large clusters containing a majority fraction of atoms (e.g. up to ~40% of all Ni in one cluster under some T/porosity conditions reported).
Trends: Higher T and lower defect density favor fewer, larger clusters; higher porosity shortens mean free paths and suppresses runaway coarsening into a single giant island.
Corpus honesty: Cluster statistics and DFT comparison numbers should be read from papers/ReaxFF_others/Valencia-Phys.Chem.Chem.Phys.2018-20-16347.pdf (PCCP version of record); this page does not substitute for the tabulated Results.

Limitations¶

ReaxFF Ni–vacancy binding differs quantitatively from some DFT values; out-of-equilibrium deposition statistics are illustrated with single long trajectories rather than thousands of averaged runs.