Computational Design of Metal–Organic Frameworks with Unprecedented High Hydrogen Working Capacity and High Synthesizability

Compared to conventional computational screening studies
that are
limited by the size of database, inverse design has a great potential
to facilitate identifying new materials with optimal properties. In
this work, we integrate machine learning with genetic algorithm to
computationally design metal–organic frameworks (MOFs) for
hydrogen storage applications at cryogenic conditions. As such, we
identified 6277 MOFs that exceed the current record (37.2 g/L of NPF-200)
at operating conditions between 5 and 100 bar at 77 K. MOFs, whose
working capacities exceed 40.0 g/L (system-based 2025 DOE target)
were also identified, where the highest working capacity obtained
from this work was 41.6 g/L, which is higher than any other hypothetical
MOFs reported thus far. Furthermore, synthesizability of the top performing
structures was assessed by comparing relative stability with their
polymorphic structures while taking into account the possibility of
interpenetration. We demonstrate that our methodology can successfully
design MOFs with both high hydrogen capacity and synthesizability
and we anticipate our workflow can be widely applied to various other
materials and applications.