Densely Packed Hydrogen in Porous Hydridic Frameworks Simulated by Hybrid Grand Canonical Monte Carlo and Molecular Dynamics

By integrating machine learning potential with grand canonical Monte Carlo and molecular dynamics, we present an effective method capable of accurately accounting for the dense packing of molecular hydrogen in porous materials. This method quantitatively accounts for the storage capacity of hydrogen and the observation of penta-dihydrogen clusters in recently synthesized porous γ-Mg(BH 4 ) 2 at 25 K [Oh et al. Nature Chemistry 2024, 16, 809]. Equally important, we show that γ-Mg(AlH 4 ) 2, formed by replacing BH 4 with AlH 4, outperforms the hydrogen storage capacity of γ-Mg(BH 4 ) 2 at higher temperature with a hydrogen capacity of 5.9 wt % and 36.3 g/L at 70 K and 100 bar (or 3.3 wt % and 19.8 g/L at 160 K and 100 bar). This methodology is versatile and can be applied to porous materials for capturing other gases, such as CO 2, CO, N 2 O, SH 2 and CH 4 .