| 초록 |
Compared to conventional screening studies, which are highly restricted to the size of database used, inverse design has a great potential to open up a new prospect on development of materials with optimal properties. In this work, by using workflow which integrates machine learning with genetic algorithm, computational design of metal-organic frameworks (MOFs) was conducted aiming for hydrogen storage at cryogenic conditions. Our design workflow resulted in generation of MOFs with superb hydrogen volumetric working capacities where 6277 new structures exceeding current record were found. MOFs whose working capacities exceeding 40.0 g/L (DOE target by 2025) were also found where the highest working capacity obtained from this work was 41.63 g/L, which is higher than any other hypothetical MOFs reported so far. Furthermore, synthesizability of the structures were assessed by comparing relative stability with isomorphic structures. We demonstrate that our methodology used for this work successfully designed MOFs with both high hydrogen capacity and synthesizability and we anticipate this workflow to be widely applied to other design researches of various hypothetical materials. |
