A multi-objective optimization-driven screening approach for maximizing hydrogen storage capacities in MOFs

The inherent tradeoff between gravimetric and volumetric hydrogen storage capacities in metal-organic frameworks (MOFs) limits their commercial viability. While benchmarked MOFs like MOF-5, IRMOF-20, and PCN-610 perform well at 77 K, maintaining their performance at elevated temperatures (298 K) remains challenging. To address this, a multi-objective particle swarm optimization framework was developed to identify promising MOFs for hydrogen storage. The optimization was guided by predictions from the bootstrapped-random forest trees. This optimization yielded 152 theoretical MOF feature combinations, which were matched with 733,792 existing structures. A nearest neighbor search identified 43 promising MOFs, with Zn-based MOF-2087 emerging as the global best, exhibiting consistent hydrogen storage performance across temperatures. Grand Canonical Monte Carlo simulations confirmed its high hydrogen uptake (5.3 wt% and 7.4 gH 2 L − 1 at 298 K). Molecular dynamics simulations further revealed C-clusters and metal sites as key adsorption centers, supporting the enhanced hydrogen storage behavior of MOF-2087. These findings highlight MOF-2087 as a computationally promising MOF for hydrogen storage up to 298 K and demonstrate the effectiveness of the optimization-driven screening strategy.