Development of Graph Neural Network Interatomic Potential to Investigate Diamond Oxidation, Graphitization, and Wear

 

氏名:Enriquez John Isaac Guinto

所属:Graduate School of Engineering, Osaka University

概要:Diamond has been attracting attention as a material for advanced technologies. Oxidation and graphitization are among the most important processes that occur on diamond materials during growth, device fabrication and operation. Quantum mechanical-level simulations of these phenomena will help explain experimental observations, predict material properties, and gain intuition that will enable the engineering and design of novel technologies. With the advent of machine learning interatomic potentials trained with quantum simulation data, simulations of physically realistic systems with quantum-level accuracy have been made possible. In this work, we develop machine learning interatomic potential based on graph neural network model to investigate diamond oxidation, graphitization, and wear. Our results show that both oxidation and graphitization exhibit surface facet dependence, suggesting that developing surface facet-specific processing and fabrication of diamond devices will be beneficial. We apply the insights that we learned from these simulations to propose a method of graphitization suppression of the stepped C(111) surface by saturating the dangling bonds with O, OH, and H species, all of which effectively prevents sp3-sp2 rehybridization. Finally, we present a novel method of epitaxial graphene self-assembly on the C(111) surface through selective graphitization of the non-terminated surface atoms. During the course of this project, we developed computational tools that we used in the database and interatomic potential construction and analysis of molecular dynamics simulation results. In particular, using graph message passing, we construct a vector representation of chemical environment which we then used to estimate the reliability of the interatomic potential. In addition, we develop a method to analyze the thermal degradation rate and characterize diamond phase transformed atoms. Finally, we develop a saddle point search approach based on the extrema of the chemical environment descriptors.

 

論文掲載,発表実績:
(学術雑誌掲載論文)

  • Enriquez, J.I.G., Yamasaki, T., Michiuchi, M., Inagaki, K., Geshi, M., Hamada, I., Morikawa, Y., “Origin of the Surface Facet Dependence in the Oxidative Etching of the Diamond (111) and (100) Surfaces from First-Principles Calculations” The Journal of Physical Chemistry (Accepted 25 March 2024)
  • Enriquez, J.I.G., Halim, H. H., Yamasaki, T., Michiuchi, M., Inagaki, K., Geshi, M., Hamada, I., Morikawa, Y., “Origin of the Surface Facet Dependence in the Thermal Degradation of the Diamond (111) and (100) Surfaces in Vacuum Investigated by Machine Learning Molecular Dynamics Simulations”, Under Review Preprint: http://dx.doi.org/10.2139/ssrn.4731456

 




Posted : 2024年03月31日