▎ 摘　　要

A numerical treatment suitable for the computational investigation of physisorption of molecular hydrogen on carbon nanostructures has not been sufficiently discussed. In this paper it is shown that results used as a reference are actually a product of poorly solved interactions and contaminated estimates with errors which would be of the order of 60%. Moreover, using ab initio molecular orbital theory, under the rigid monomer supermolecular approach, the physisorption energy of molecular hydrogen on graphene was reinvestigated. The graphene surface was modeled as a coronenelike (C24H12) graphene sheet. The basis set superposition error was corrected by means of the counterpoise method. The H-2-H-2 and H-2-benzene interactions were examined, under systematic combinations of basis sets and correlation methods, including the aug-cc-pVQZ basis set and the coupled cluster correlation method with single, double, and noniterative triple excitations, searching for a numerical treatment with a reasonable trade-off between efficiency and accuracy. Asymmetrical modeling strategies, using diffusion augmented basis sets with preference for the adsorbate, were found to be effective. Also local modeling strategies, using more complete basis sets for the nearest atoms to the adsorbate than for the rest of the substrate, were considered. The aug-cc-pVTZ basis set for the adsorbate and for the nearest atoms to the adsorbate and the cc-pVTZ basis set for the rest of the cluster-modeled graphene, at the second-order Moller-Plesset perturbation theory correlation level, was selected as reference treatment. It was found that the physisorption energy of molecular hydrogen on graphene would be of the order of 0.06 eV, which would be 25% less than what has been previously published, though it would be sufficient to permit the storage of hydrogen physisorbed on carbon. To our knowledge this would be the most realistic theoretical estimate of the mentioned energy to date.