Coupling Molecular Dynamics and Machine Learning for Predicting Aluminum Segregation to Magnesium Grain Boundaries
Abstract
Engineering magnesium alloys for Army applications requires an understanding of corrosion behavior, including how grain boundaries (GBs) or second phases/intermetallics affect corrosion at the atomic scale. In magnesiumaluminum alloys, precipitation of the Mg17Al12 phase at GBs can have important implications for mechanical and corrosion behavior, but first, single atom segregation to GBs is required. The objective of this study is to quantify the energetics of segregation of aluminum to a dataset of different symmetric tilt grain boundaries (STGBs) for magnesium. In this work, aluminum atoms were iterativelyplaced at various atomic sites within 20 Angstrom of the GB center for a dataset of 30 h0001i STGBs. Results show how GB structure affects the energetics and length scales of aluminum segregation as well as the dependence of segregation energetics on the local atomic environment, which was used to form a surrogate (machine learning) model for segregation energetics. The ability to compute grain boundary physical properties of interest using machine learning techniques can have broad implications for the area of grain boundary science and engineering.
Document Details
- Document Type
- Technical Report
- Publication Date
- Sep 01, 2018
- Accession Number
- AD1059513
Entities
People
- Kaushik Joshi
- Mark Tschopp
- Santanu Chaudhuri
Organizations
- United States Army Research Laboratory