2022irg1-5

2022 IRG-1: Machine Learning Prediction of Interfacial Segregation in Complex Concentrated Alloys

In this work, the grain boundary co-segregation behavior in a NbMoTaW refractory complex concentrated alloy was investigated with atomistic simulations and machine learning tools.

• A random solid solution was simulated with atomistic simulations and local atomic environments were converted into vector descriptions, which were fed into an artificial neural network to predict the segregation behavior.

• Preliminary results from a deep neural network model offer good prediction power.

• This method will allow researchers to predict the partitioning of elements at different internal features in complex concentrated alloys without the need to run expensive atomistic simulations.

TJ Rupert, D Aksoy, D Apelian (University of California, Irvine)
SP Ong, J Luo (University of California, San Diego)