Skip to content

A computational framework for guiding the MOCVD-growth of wafer-scale 2D materials

Scope

The CPM model couples CFD, phase-field growth, and reactive MD (including ReaxFF chemistry for WSe\(_2\)) to MOCVD experiments for wafer-scale 2D materials.

Summary

Wafer-scale metal–organic chemical vapor deposition (MOCVD) of 2D materials is sensitive to flow, temperature, and precursor distribution. The authors build a multiscale framework (CPM: continuum fluid and heat transport linked to mesoscale phase-field island growth and nanoscale reactive MD) and benchmark it against detailed MOCVD experiments on WSe\(_2\) as a model TMD, including substrate rotation effects on uniformity. The computational goal is not a single MD trajectory, but a coupled decision tool: reactor CFD narrows feasible window conditions, phase-field island models translate those conditions into morphology metrics, and ReaxFF segments supply surface reaction inputs where empirical rates are uncertain.

Methods

A — ReaxFF / reactive MD (nanoscale chemistry)

  • Precursor–surface reactions for WSe\(_2\)-relevant MOCVD chemistry: ReaxFF segments supply rates or barriers where empirical kinetics are uncertain (full species list and protocols in npj Comput. Mater. / 2022momeni-venue-paper SI).

B — Continuum reactor (CFD)

  • Navier–Stokes, continuity, species transport (convection–diffusion), and energy equation; boundary conditions matched to the experimental furnace geometry (rotation, flow inlets, etc.).

C — Phase-field (mesoscale growth)

  • Island morphology, edge energies, deposition fluxes, edge diffusion, temperature-dependent shapes—fed by CFD precursor fields.

D — Experiments

  • MOCVD WSe\(_2\) wafer maps (coverage, uniformity) benchmark the coupled CPM framework.

Coupling: CFD → phase-field → comparison to experiment; ReaxFF nanoscale segments inform kinetic or reaction inputs for MOCVD WSe\(_2\) (full species/thermal protocols in the article + 2022momeni-venue-paper SI).

1a — ReaxFF / MD slot (for applicable MD blocks)

  • Engine, slab sizes, NVT timesteps, run lengths, long-range QEq/DFT kinetic bridges: N/A in this short wiki narrative — the definitive reactive-MD reactor-side protocol (LAMMPS-style) lives in the version-of-read text + SI; import from [[2022momeni-venue-paper]] for PBC, timestep, and thermostat when reproducing the WSe\(_2\) MOCVD-chemistry tether to the field-level model. Equilibration / production durations (ps / ns in ReaxFF or PF stages) are in those sources — N/A here to quote ns-long trajectory lengths without re-opening the full SI.

1b — Continuum / multiscale (this paper’s emphasis)

  • N/A MD-as-entire paper — the work is not a stand-alone WSe\(_2\) RMD kinetics traj repo; the MOCVD-tethered reactor (CFD), mesoscale phase-field morphology, and 2DCC experiments drive the narrative. NPT /pressure /E-field /rare-event sampling: N/A in the summarized ReaxFF hitch—see 1a (SI).

Findings

Outcomes and comparisons: The 2DCC MOCVD experiments and the CPM (CFD + phase-field + ReaxFF-benchmarked WSe\(_2\) surface kinetics) report how inlet flow, rotation, and local temperature/precursor fields from the reactor model map to WSe\(_2\) coverage and island morphology. Quantitative agreement is in the VOR and SI, not re-tabulated here.

Sensitivity levers: Inlet flow, furnace geometry, and wafer rotation (plus the CFD-mapped T and C fields in the main text). N/A in this short summary: every precursor partial pressure; see VOR and [[2022momeni-venue-paper]] SI.

Limitations

Reactive MD segments remain costly; continuum and phase-field parameters require calibration for each new material stack. The CPM coupling is presented as a closed loop: CFD supplies precursor fluxes and thermal maps that feed phase-field island growth, while nanoscale ReaxFF segments benchmark surface reaction propensities for WSe\(_2\)-relevant precursors under the MOCVD conditions explored experimentally. Substrate rotation enters both flow uniformity and coverage maps, so retrieval users should cite figure panels that tie simulated uniformity to measured wafer maps rather than quoting a single scalar growth rate in isolation.

Relevance to group

Van Duin-group reactive MD is embedded in a multiscale pipeline with 2DCC MOCVD experiments.

Reader notes (navigation)

Citations and evidence anchors