Hydrogen energy provides an approach to addressing the energy crisis and global climate change. However, additional energy consumption and carbon emission in the conventional storage process of H2 tackle the hydrogen economy’s prosperity. Metal-organic frameworks (MOFs), new type materials with exciting structures and properties, represent a blueprint for realizing large-scale applications of hydrogen energy by lowering energy consumption and cost of facilities. Traditional hydrogen storage MOFs have stepped to an advanced level, and the discovery inevitably slows down. Materials are fundamental to low-cost hydrogen storage, and the screening and design of H2 storage MOFs are crucial for the hydrogen economy. We aim to propose a novel paradigm of hydrogen storage MOFs material design that combines machine learning and first principles calculation, such as density functional theory (DFT). By constructing an active learning framework and using DFT calculation results as training data, a self-improving model that can screen existing material databases and guide experiments design is obtained. The prediction model’s performance is examined in conventional ways (root mean square error, coefficient of determination, etc.) and will be further tested in practical considerations (test the performance of MOFs guided by the model).
Keywords hydrogen energy, MOF, machine learning, density functional theory