About this Abstract |
| Meeting |
2026 TMS Annual Meeting & Exhibition
|
| Symposium
|
Phase Transformations and Microstructural Evolution
|
| Presentation Title |
Atomistic Insights Into the Graphite-to-Diamond Phase Transformation via Deep Neural Network Potential Molecular Dynamics |
| Author(s) |
Mehrab Lotfpour, Haoran Cui, Yan Wang, Lei Cao |
| On-Site Speaker (Planned) |
Mehrab Lotfpour |
| Abstract Scope |
We present a molecular dynamics (MD) study of the graphite-to-diamond phase transformation using a deep neural network (DNN) potential trained on extensive ab initio data. These high-fidelity potential captures carbon bonding under extreme conditions with quantum accuracy and MD efficiency.
Large-scale simulations on graphite with AB and AA stacking reveal transformation mechanisms under hydrostatic and non-hydrostatic pressures. Hydrostatic pressures above 100 GPa yield polycrystalline cubic diamond with grain boundaries and hexagonal diamond intergrowths. In contrast, non-hydrostatic compression stabilizes pure hexagonal diamond, enabling selective phase control.
Mechanistically, cubic diamond forms via in-plane sliding and graphene layer rearrangement, while hexagonal diamond arises from out-of-plane buckling. These distinct pathways highlight stress-driven control over diamond polymorphs.
Our results demonstrate the power of machine-learned potentials in modeling phase transitions and provide a predictive framework for designing carbon-based materials under extreme environments. |
| Proceedings Inclusion? |
Planned: |