Reverse hydrogen spillover onto zeolite-supported metal clusters: An embedded cluster density functional study of models M₆ (M = Rh, Ir, or Au)

A density functional study using isolated cluster models of the zeolite framework (Vayssilov, G. N.; Rösch, N. Phys. Chem. Chem. Phys. 2005, 7, 4019) on supported metal clusters M₆/zeo in a hydroxylated faujasite cage showed that for 12 metals M of the groups 8-11 the hydrogenated state, M₆(3H)/zeo, is energetically preferred over the bare form M ₆/zeo(3H). The former state was obtained as result of reverse hydrogen spillover from zeolite OH groups onto the metal particle. In the present work, we reinvestigated this problem of identifying the favorable chemical state of zeolite-supported metal species for selected M₆ model clusters (M = Rh, Ir, or Au) using an accurate quantum mechanics/molecular mechanics (QM/MM) approach where a QM partition is embedded in an extended zeolite lattice (MM). The embedding method covEPE, an improved variant of the elastic polarizable environment model EPE, adapted for polar-covalent materials, properly accounts for both the mechanical rigidity of the zeolite framework and the electrostatic field due to the infinite lattice. With this method, we discovered the first example where the bare zeolite-supported form M ₆/zeo(3H) is energetically preferred over the hydrogenated from M₆(3H)/zeo: reverse hydrogen spillover from less acidic OH groups onto Au₆ was calculated to be endothermic, on average by 29 kJ/mol per transferred proton. In contrast, reverse spillover from the more acidic hydroxyl groups to Au₆ is exothermic by 47 kJ/ mol per proton. For the other metals, the QM/MM approach predicts reverse hydrogen spillover to be energetically favorable, just as the simple model approach with finite models of the zeolite support. However, the corresponding calculated energy change is strongly reduced (per proton) for Rh₆ from 123 kJ/mol to 73 and 98 kJ/mol and for Ir₆ from 229 kJ/mol to 144 and 160 kJ/mol. We analyzed in detail these differences between the various model approaches.