A guided review of machine learning in the design and application for pore nanoarchitectonics of carbon materials

Abstract

Porous carbon materials have demonstrated significant potential in areas such as carbon capture, gas separation, energy storage, and catalysis, improving energy efficiency and aiding in reducing carbon emissions. With the advancement of global environmental policies, developing efficient and sustainable materials is critical to addressing energy and environmental challenges. However, traditional trial-and-error approaches are often costly and inefficient. Recently, the rapid development of artificial intelligence and machine learning (ML) has introduced data-driven methods to materials science, significantly improving the efficiency of new material development. This review summarizes the application of ML in porous carbon materials, outlining key learning processes and commonly used algorithms, and highlights the latest advancements of ML in porous carbon synthesis and applications, such as carbon capture, energy storage, and supercapacitors. Specifically, it discusses the impact of essential features, such as pore shape, surface area, and pore volume, on different applications, identifies research gaps for non-biomass precursors like coal and tar pitch, and proposes future research directions. This review aims to serve as a resource for ML applications in the field of porous carbon materials, promoting the efficient development and broad application of novel porous materials.