Accelerated discovery of magnesium intermetallic compounds with sluggish corrosion cathodic reactions through active learning and DFT calculations

Yaowei Wang,Qingli Tang,Xinchen Xu,Paul Weng,Tao Ying,Yao Yang,Xiaoqin Zeng,Hong Zhu

DOI: https://doi.org/10.1016/j.actamat.2023.119063

IF: 9.4

2023-01-01

Acta Materialia

Abstract:Magnesium (Mg) alloys can potentially be widely applied in transportation, aerospace and biomedical fields due to the light weight and biocompatibility. However, they are usually subjected to serious galvanic corrosions due to high chemical activity. In this work, active learning is employed to discover the intermetallic compounds which can suppress the corrosion cathodic reaction of Mg alloys. The hydrogen adsorption energy, which is a descriptor for the rate of the cathodic hydrogen evolution reaction (HER), is predicted by machine learning models using the geometric and chemical features of the H adatom's Voronoi neighbors. After five active learning iterations, the prediction error of the H adsorption energy for the strong/weak adsorption configuration is 0.196 eV (MAE) with the training set size less than 1% unknown data set. Furthermore, we find that the surfaces with strong H adsorption transfer more electrons to H adatoms than the weak H adsorption surfaces. Finally, the ability of the binary Mg intermetallics to inhibit the HER is ranked according to their surface stabilities and predicted H adsorption energies. This work suggests the binary Mg intermetallics that could greatly suppress the corrosion cathodic reaction through active learning and density functional theory (DFT) simulations, which is expected to accelerate the design of corrosion-resistant Mg alloys.