Data-driven design of electrocatalysts: principle, progress, and perspective

Abstract

To achieve carbon neutrality, electrocatalysis has the potential to be applied in the technological upgrading of numerous industries. Therefore, the search for high-performance catalysts has become an important topic. To accelerate the discovery of new electrocatalysts, emerging data-driven strategies have been considered promising approaches. In particular, research methods represented by machine learning (ML) have permeated all aspects of electrocatalyst development, including synthesis, characterization, and simulation, which have given rise to numerous new ideas for catalytic-related data generation and analysis. Herein, this review focuses on the systematic construction of a data-driven electrocatalyst design framework. First, we introduce the principles for the basic steps for implementing data-driven electrocatalyst research, including data generation, data preprocessing, and data analysis. Subsequently, the progress of ML methods to design promising electrocatalytic materials (e.g., metals, alloys, and oxides) for numerous typical electrochemical reactions is summarized. Finally, the current challenges and opportunities are outlined for the future of data-driven electrocatalyst design.