PNNL

Abstract

We gathered literature data and developed a machine learning model with descriptor optimization to predict the corrosion resistance of a given multi-principal element alloy in an aqueous environment. The model uses two environmental descriptors (pH of the medium and halide concentration), one chemical composition descriptor (atomic % of element with minimum reduction potential), and two atomic descriptors (difference in lattice constant (?a) and average reduction potential). The prediction of corrosion rate is reasonably accurate. This work points to the need for high quality data with metadata for accurate prediction of complex chemical phenomena, like corrosion of metals. The sparse data challenge needs to be remedied through shared databases with high quality data.