Design and fabrication of high entropy ceramic coatings based on machine learning

Xiaoqian Xu, Tao Guo, Xiaolu Pang*

Beijing Advanced Innovation Center for Materials Genome Engineering, Beijing 100083, China

EXTENDED ABSTRACT: Surface coating is an effective method to improve material performance and durability.High hardness, good adhesion, friction resistance and good oxidation resistance are the most important properties.The high entropy ceramic coating based on the idea of multiple high entropy alloy is expected to become a new generation of superhard coating materials.The properties of coatings are affected by many factors, such as chemical composition, crystal structure, microstructure and preparation process. The research and development of new coating systems is faced with a complex search space, and the traditional "trial and error" method is costly, time-consuming and inefficient.To solve this problem, based on the concept of material genetic engineering, this study proposed a method of fast screening hard coating system.The research content mainly focuses on the following two aspects: 1) Learning the data of high entropy nitride coating in existing literature, establishing a machine learning prediction model to predict coating hardness through chemical composition and preparation process.Based on this model and combined with the active learning strategy, a new high entropy nitride coating system (AlCrNbTaTi) N with high hardness was successfully designed through several iterations.The influence of different elements on coating hardness was explained by the model.2) The high entropy nitrode coatings with continuous distribution of components were prepared by using magnetron three-target co-sputtering technology on a self-designed multi-station sample table.This study proves that the combination of machine learning and high throughput experimental method can effectively accelerate the composition optimization design of multi-component materials.