Accelerating the prediction of stacking fault energy by combining ab initio calculations and machine learning

Stacking fault energies (SFEs) are key parameters to understand the deformation mechanisms in metals and alloys, and prior knowledge of SFEs from ab initio calculations is crucial for alloy design. Machine learning (ML) algorithms used in the present work show a $\ensuremath{\sim}$ 80 times acceleration of generalized stacking fault energy predictions, which are otherwise computationally very expensive to get directly from density functional theory calculations, particularly for alloys. The origin of the features used for training the ML algorithms lies in the physics-based Friedel model, and the present work uncovers the connection between the physics of $d$ electrons and the deformation behavior of transition metals and alloys. Predictions based on the ML model agree with the experimental data. Our model can be helpful in accelerated alloy design by providing a fast method of screening materials in terms of stacking fault energies.