About this Abstract |
| Meeting |
2027 TMS Annual Meeting & Exhibition
|
| Symposium
|
2D Materials – Preparation, Properties, Modeling & Applications
|
| Presentation Title |
MoS₂ Epitaxial Growth Simulations via Ultra-Fast Machine-Learned Interatomic Potential |
| Author(s) |
Emir Bilgili, Nicholas Taormina, Richard Hennig, Simon R. Phillpot, Youping Chen |
| On-Site Speaker (Planned) |
Emir Bilgili |
| Abstract Scope |
A machine-learned interatomic potential (MLIP) for multilayer MoS₂ was developed using the Ultra-Fast Force Field (UF3) framework, trained on an extensive density functional theory (DFT) dataset. The potential reproduces key properties in strong agreement with DFT, including lattice constants, interlayer binding energies, phonon spectra, and the highly anisotropic elastic tensor across 1H, 2H, and 3R phases. Defect formation energies correlate strongly with DFT (R˛=0.91) across ten defective monolayers, and the difference between the energies of zigzag and armchair edges is reproduced within 5% of DFT. Molecular-dynamics simulations reproduce two experimentally observed features of MoS₂ epitaxial growth: (1) formation of van der Waals gaps between successive epilayers and (2) zigzag-edge-terminated triangular domain formation. With speed comparable to empirical potentials, we demonstrate a near-mesoscale homoepitaxial growth simulation (lateral dimensions ≥0.1 μm, ≥20 million atoms, ≥100 ns), capturing nucleation, growth, and coalescence of triangular domains into a well-defined epitaxial layer at atomic resolution. |
| Proceedings Inclusion? |
Planned: |
| Keywords |
Machine Learning, Thin Films and Interfaces, Modeling and Simulation |