Machine learning and molecular dynamics based models to predict the temperature dependent elastic properties of silver nanowires

Abstract

Metallic nanowires are now extensively used in several nanoscale devices and applications. To further enhance their efficient usage, the estimation and prediction of thermal and mechanical properties of these nanowires is very important. Performing experimental studies on the objects of such a small dimension is quite challenging. Molecular dynamics simulation technique can easily simulate and perform virtual experimentation on the objects of nanoscale dimensions. In the present work, silver nanowires of known dimension simulated and a uniaxial stress has been implemented using the Molecular dynamics approach. The stress-strain data generated by MD simulation, has been utilized to train, test and validate different machine learning models. These machine-learning models offer a reasonably good amount of predictability of the tensile characteristics of the silver nanowire at any temperature.

Keywords:

• Molecular dynamics
• machine learning
• nanowire
• ultimate tensile strength
• tensile properties

Disclosure statement

No potential conflict of interest was reported by the author(s).