| About this Abstract |
| Meeting |
MS&T22: Materials Science & Technology
|
| Symposium
|
Mesoscale Phenomena in Functional Polycrystals and Their Nanostructures
|
| Presentation Title |
D-21: Machine-learned Large-scale Model for Layered Amorphous Graphene: A Study of Its Structure and Thermodynamics |
| Author(s) |
Chinonso Ephraim Ugwumadu, Rajendra Thapa, Kishor Nepal, David Drabold, Jason Trembly |
| On-Site Speaker (Planned) |
Chinonso Ephraim Ugwumadu |
| Abstract Scope |
The microscopic thermodynamical properties and atomistic structure of layered amorphous graphene (LAG) using Machine learning (ML) based interatomic potential (GAP) is discussed herein. ML-based potentials are opening new frontiers to understanding atomistic large-scale phenomena with near-first-principle properties. Fast empirical interatomic potentials like EDIP and REBO make large-scale molecular dynamics simulations possible but remain empirical and overestimate sp3 bonds in amorphous Carbon(aC). On the other hand, tight-binding schemes, including ab-initio methods like DFT give more accurate predictions but becomes computationally expensive with system size; limited to a few hundred atoms. The GAP ML interatomic potential for aC provides new insights into the transition of aC to LAG at canonical and isobaric-isothermal ensemble for large system sizes (~ 1000 atoms) while still reproducing DFT obtained results at small system size (~ 160 atoms). The thermal conductivity of LAG is also discussed with results being a direct consequence of large system scaling. |