AI-driven advances in metal–organic frameworks: from data to design and applications

Abstract

Metal-organic frameworks (MOFs) are a versatile class of porous materials with unprecedented structural tunability, surface area, and application potential in areas such as gas storage, carbon capture, and biomedicine. However, their immense chemical design space poses significant challenges for conventional discovery and optimization methods. Recent advances in artificial intelligence (AI) and machine learning (ML) have introduced transformative capabilities to this field, enabling accurate property prediction, automated structure generation, and synthesis planning at scale. This review provides a comprehensive overview of AI-driven strategies for accelerating MOF research. It discusses key databases, deep learning architectures, generative models, and hybrid AI-simulation frameworks that have reshaped the design and screening of high-performance MOFs. Techniques such as graph neural networks and AL have enabled breakthroughs in structure-property prediction, while integration with robotics is advancing autonomous laboratories. Despite these advancements, challenges remain in data quality, model interpretability, and experimental validation. Future directions include physics-informed ML models, standardized data protocols, and deeper integration of AI with chemical robotics. By highlighting both opportunities and current limitations, this review aims to provide a roadmap for the next generation of AI-accelerated MOF innovation.