Machine learning-driven computational screening of covalent organic frameworks for gas separation applications

Covalent Organic Frameworks (COFs) represent a rapidly advancing class of crystalline porous materials with exceptional potential for gas adsorption applications, attributed to their modular structures and versatile functional properties. The integration of machine learning (ML) techniques and computational material science is advancing COF research. It offers powerful tools to enhance the prediction and screening of COFs for gas capture and separation. This review provides a comprehensive overview of the application of ML in COF research, emphasizing its role in predicting adsorption performance and enabling large-scale virtual screening. Key ML methodologies are analysed, focusing on their use in evaluating adsorption efficiency and uncovering structure-performance correlations. Additionally, this review addresses critical challenges in leveraging ML for COF studies, such as the need for high-quality datasets, the selection of meaningful COF descriptors, and the exploration of generative models for the discovery of novel COFs. The discussion highlights the importance of model interpretability and the integration of ML with high-throughput computational screening frameworks. By showcasing the transformative potential of ML in COF research, this review outlines a clear pathway for the accelerated discovery and design of efficient COFs for sustainable gas adsorption-based separation applications.