Amorphous boron nitride (aBN) films possess significant promise for various electronic and coatings applications due to their superior mechanical, chemical, and electronic properties. However, these properties can be willingly or unwillingly tuned by changing the fabrication conditions. A thorough understanding of the relation between fabrication conditions, morphology, and properties of the materials is vital to fully exploit the abilities of aBN films for several applications. This study aims to reveal this relation for aBN films and evaluate their performance for several applications through machine learning-assisted atomistic simulation models and experimental characterisation techniques.
Modelling amorphous materials is quite a challenge due to their complex morphology. Amorphous structures include many different structural possibilities such as different types of bonds, bond lengths, and angles. The methods used for such materials should be able to define the diverse interactions between atoms and allow us to perform calculations with a large number of atoms. Several methods have been used to describe the complex nature of amorphous materials; however, they are either not very accurate in describing the interactions between atoms or computationally too expensive to work with a large number of atoms. Here, machine learning models can act as a bridge between these two realms. We use machine learning models trained on very accurate density functional theory calculations combined with computationally efficient molecular dynamics simulations. This will allow us to achieve a complete understanding of the nature of the material and understand its potential for various electronic and coating applications.
Watch a video about Onurcan’s project:
