Skip to content

Ab initio molecular dynamics of atomic-scale surface reactions: insights into metal organic chemical vapor deposition of AlN on graphene

Summary

Supercells use ~27 Å vacuum, 6×6 graphene unit cells (72 C atoms), fixed in-plane lattice parameter 2.465 Å, Γ-point sampling, 300 eV plane-wave cutoff, total-energy convergence 10⁻⁵ eV/atom, and 0.1 fs / 1 fs timesteps depending on whether hydrogen dynamics are active. The elevated temperature accelerates rare surface reactions relative to typical MOCVD temperatures yet targets qualitative pathways on pristine graphene.

AlN MOCVD on graphene-related templates couples precursor pyrolysis with nucleation on sp\(^2\) carbon; ab initio MD at strongly elevated temperature accelerates rare events to sample trimethylaluminum, ammonia, and adduct chemistry on a model graphene slab. The reported VASP LDA + DFT-D3 setup with tight timesteps for hydrogen motion targets qualitative reaction sequences ahead of lower-temperature kinetics refinements. Consult the peer-reviewed PDF and any Supporting Information for authoritative tables, figures, and numerical diagnostics behind the summaries above.

Methods

Static QM / DFT setup (VASP). Density-functional ab initio calculations use VASP with LDA (Ceperley–Alder as parametrized by Perdew–Zunger), PAW potentials, Grimme DFT-D3 dispersion, and Gaussian electronic smearing at \(k_BT\) as reported. Numerical settings include a 300 eV plane-wave cutoff, Γ-only Brillouin sampling, and \(10^{-5}\) eV/atom electronic convergence thresholds for the setups summarized on this page.

AIMD application (Born–Oppenheimer trajectories). AIMD explores gas-phase and surface reactions of trimethylaluminum, ammonia, and adduct-derived (CH\(_3\))\(_2\)AlNH\(_2\) on a periodic graphene slab with ~27 Å vacuum separation, using a 6×6 graphene unit cell (72 C atoms) and an in-plane lattice constant 2.465 Å as stated. Temperature is elevated to \(T = 4300\) K in the reported “hot dynamics” protocol to accelerate rare events while staying below the authors’ cited graphene-melting concern. Integration timesteps are 0.1 fs when H motion is active and 1 fs for lighter Al/C adatom diffusion studies. Sampling is finite-temperature Born–Oppenheimer molecular dynamics in the canonical sense used for constant-\(T\) ab initio sampling in VASP (N/A — the exact thermostat implementation string and total trajectory lengths in ps should be copied from the PCCP Methods rather than inferred from the p1–2 extract alone). PBC: three-dimensional periodic boundary conditions for the slab supercell. Ensemble / pressure: constant-volume supercell with controlled temperature (N/A — explicit NPT barostat and external stress control not used in the AIMD description excerpted here). Electric field: N/A — not used as an MD bias in the indexed excerpt. Enhanced sampling: N/A — umbrella / metadynamics / replica exchange not indicated for the AIMD sets described on pages 1–2 of the extract; the study instead relies on high-temperature AIMD to access rare reactive events.

Findings

AIMD highlights precursor fragmentation and surface delivery pathways for Al, C, and mixed Al–C–N hydrogenic species on graphene, with charge-transfer arguments for Al attachment. Carbon adatoms exchange with graphene lattice atoms, enabling through-sheet transport with Arrhenius parameters reported in the abstract (~0.28 ± 0.13 eV barrier, ~2.1 THz prefactor at experimental-referenced conditions). NH3 does not strongly activate pristine graphene in the simulated trajectories, suggesting limited N delivery without defects.

Comparisons and caveats. The abstract explicitly connects simulated rates and activation energies for C exchange/permeation to Arrhenius analysis and experimentally referenced temperature scales, while stressing that AIMD is used as an interpretive tool for MOCVD chemistry rather than a direct room-temperature production kinetic model. Sensitivity: the study’s conclusions depend strongly on the chosen elevated temperature acceleration strategy and on the LDA + DFT-D3 treatment of graphene and precursor chemistry—limitations of LDA for reaction energetics are discussed in the article framing. Corpus honesty: this page is grounded in the checked-in PDF/normalized/extracts/2018sangiovanni-physical-che-ab-initio_p1-2.txt; figures, longer trajectories, and complete parameter tables should be taken from the version-of-record PDF on pdf_path.

Limitations

Very high simulation temperature and short trajectories trade realism of kinetics for sampling; LDA errors for chemistry and graphene electronic structure are acknowledged implicitly by the authors’ validation narrative.

Relevance to group

Provides AIMD reference chemistry for III-nitride growth on carbon supports, adjacent to ReaxFF/MOCVD workflows.

Citations and evidence anchors