In 2003, with hydrogen economies on every policy agenda, Rosi and colleagues reported that MOF-5 adsorbed hydrogen, 4.5 weight percent at 78 K, and, remarkably, around 1 percent at room temperature and modest pressure. Inelastic neutron scattering identified distinct binding sites on the zinc cluster and the linker, giving the field its first molecular-level picture of where H2 actually sits in a framework.
The impact was immediate and enormous. Hydrogen storage became the defining MOF application of the decade, hundreds of frameworks were screened, and the paper's site-specific binding analysis established neutron methods as a standard tool. The core physical insight, that both the inorganic cluster and the organic face contribute adsorption sites, and that the cluster sites bind more strongly, has held up completely.
The honest part of this review: the room-temperature uptake was subsequently revised sharply downward, including by the original group, once measurement artifacts in early volumetric setups were understood. Cryogenic capacities stood; ambient ones did not. This episode is worth teaching in itself, because it shows a healthy correction loop operating in public, and it explains why modern gas-sorption papers obsess over activation procedures, buoyancy corrections and sample purity.
What remains is a landmark that defined a research program and its central physical obstacle in the same stroke: physisorption of hydrogen is simply too weak at 298 K, roughly 5 kJ/mol on typical surfaces, against the 15 to 25 kJ/mol needed for practical ambient storage. Twenty years of open-metal-site engineering descends directly from that gap, first quantified here.