利用机器学习对金属-有机框架材料进行从材料到系统的宽范围筛选以用于氢气储存

Abstract Hydrogen is pivotal in the transition to sustainable energy systems , playing major roles in power generation and industrial applications. Metal–organic frameworks (MOFs) have emerged as promising mediums for efficient hydrogen storage . However, identifying potential candidates for deployment is challenging due to the vast number of currently available synthesized MOFs. This study integrates molecular simulations , machine learning, and techno-economic analysis to evaluate the performance of MOFs across broad operation conditions for hydrogen storage applications. While previous screenings of MOF databases have predominantly emphasized high hydrogen capacities under cryogenic conditions, this study reveals that optimal temperatures and pressures for cost minimization depend on the raw price of the MOF. Specifically, when MOFs are priced at $15/kg, among the 9720 MOFs tested, 9692 MOFs achieve the lowest cost at temperatures between 170 K and 250 K and a pressure of 150 bar. Under these optimal conditions, 362 MOFs deliver a lower levelized cost of storage than 350 bar compressed gas hydrogen storage. Furthermore, this study reveals key material properties that result in low system cost, such as high surface areas (>3000 m 2 /g), large void fractions (>0.78), and large pore volumes (>1.1 cm 3 /g).