Machine Learning Assisted Design of Complex and High Entropy Alloys by Hybrid HiPIMS/Pulsed-DC PVD Process for Low Carbon Energy Applications in Extreme Environments

Abstract

Complex and high entropy alloys are attracting much attention currently thanks to their mechanical and corrosion resistance properties in harsh environments, in particular needed for carbon-free energy applications. However, their elaboration in bulk and in thin film form in a trial-and-error approach is impractical due to their complexity and the cocktail effect. The recent development of artificial intelligence brings a new possibility for their elaboration and adjustment of their properties. Firstly, we present an overview of Materials and data science research. Then we describe how DIADEM - French initiative for Materials and Data science convergence - tackles the development of innovative coatings for carbon-free energy applications (nuclear, high temperature electrolysis, ...) thanks to the development of a nationwide network of synthesis and characterization platforms - the DIADEM discovery hub. We describe in particular DIADEM-2D, an AI-driven Hybrid HiPIMS/Pulsed-DC PVD process using 4 cathodes in confocal combinatorial configuration. We present the high entropy alloy determination using data from the literature for corrosion resistance in molten salt media and nuclear accidental conditions. An element-independent model gathering deposition parameters and coating properties has been implemented allowing the design of protective coatings with a particular composition. We demonstrate the feasibility of this process and its accuracy.