Machine learning-assisted development of gas separation membranes: A review

Abstract

Gas separation membranes have been a hot topic of research in recent decades due to their low costs, high energy efficiency and wide range of applications. Machine learning provide a fast way to design gas separation membranes with required performance. This review systematically describes the process of machine learning-assisted gas separation membrane development. In addition, the experimental data on CO₂/CH₄, CO₂/N₂ and O₂/N₂ separation performance were summarized to provide basis for future work on machine learning-assisted design of gas separation membrane for carbon dioxide capture, and natural gas purification as well as oxygen or nitrogen enrichment. Moreover, we discuss the classical materials that make up gas separation membranes, including MOFs, polymers and COFs, and analyze the strengths and weaknesses of the different materials. Finally, we discuss the challenges in the development of machine learning method for next-generation gas separation membranes.