STRUCTURE AND DYNAMICS OF HYDROGEN SORPTION IN MESOPOROUS MCM-41

Adsorption isotherms taken at temperatures ranging from 20 to 77 K sho w a large pore volume and surface area of 980 m(2) g(-1) for the physi cal adsorption of molecular hydrogen on MCM-41. The adsorbed hydrogen behaves more like a solid than a liquid and isosteric heats of adsorpt ion reveal a heterogenous surface. The evaporation behaviour of the ad sorbed hydrogen indicates that the hexagonally packed tubes in MCM-41 may be effectively interconnected into a single void space. Neutron in elastic scattering shows that molecular hydrogen exists in two sites i n the pores of MCM-41. We designate as surface states those with a var iety of weakly hindered rotational excitations centred at 11.8(2) meV and as bulk states, at high doping of H-2, those which have narrow rot ational excitations at 14.7(3) meV. These excitations are hardly chang ed from those of the free crystal in energy, energy width or momentum transfer dependence. Each type accounts for ca. 50% of the total pore volume. Strong hydrogen recoil scattering is also observed at momentum transfers above ca. 3 Angstrom(-1). Neutron diffraction from the fill ed sample at 1.9 K shows a single peak at 3.1 Angstrom, characteristic of the H-2-H-2 correlations in bulk hydrogen. This peak is much weake r for the surface adsorbed species. The concentration dependence of th is peak also shows a 1:1 division of void space between surface-adsorb ed and bulk-like species. In addition we observe a separate peak at 14 .38(3) Angstrom, the hexagonal (21) of MCM-41 involving the silica fra mework, whose diffracted intensity changes dramatically with hydrogen doping, in a way inconsistent with a smooth walled, hexagonally packed mesopore. These data, together with previous X-ray diffraction data, provide information on silica density, absorbing surface and free volu me projected onto the basal plane. The detail and complexity of the pr ojection were unexpected. The data agree well with a model for which t he silica in the walls is loosely interwoven, and occupies only 40% of the wall volume. The 35% silica surrounding the empty 7 Angstrom hole as a 12.7 Angstrom lining has a projected density of ca. 90% that of the walls. Either highly divided, 'hairy' walls or a structure like a highly defective variant of the MCM-48 structure would fit the data. T he projection of smooth-walled but highly twisted channels onto a hori zontal plane could give a density distribution equivalent to the highl y porous silicate walls in straight channels.