What is it about?
Silicon nitride is an extensively used material in the automotive, airspace and semiconductor industries. Molecular dynamics simulations needed to study this material are performed with either ab initio methods or empirical potentials. However, the former is slow, while the latter is not always accurate. In this work, we propose a machine learning based alternative which combines high accuracy with low computational costs.
Featured Image
Photo by FLY:D on Unsplash
Why is it important?
Our proposed machine learning interatomic potential enables molecular dynamics simulations for silicon nitride with ab initial level accuracy, while being between 1000 and 10,000 times faster than ab initio methods.
Perspectives
This novel machine learning solution allows to perform molecular dynamics simulations, with exceptionally high accuracy, on system sizes and simulation times previously our of reach by orders of magnitude.
Diego Milardovich
Technische Universitat Wien
Read the Original
This page is a summary of: Machine learning interatomic potential for silicon-nitride (Si3N4) by active learning, The Journal of Chemical Physics, May 2023, American Institute of Physics,
DOI: 10.1063/5.0146753.
You can read the full text:
Contributors
The following have contributed to this page
