Cu+ binding energies. Dramatic failure of the G2 method vs. good performance of the B3LYP approach

The Cu+ binding energies of a wide set of neutral systems were calculated using G2-based methods and CCSD(T) formalism. The standard G2 formalism is shown to be quite unsuitable to estimate Cu+ binding energies, with errors greater than 13 kcal/mol. The relaxing of the inner-valence orbitals or the inclusion of all electrons in the correlation space imply a negligibly small improvement. These failures seem to be related to a poor convergence of the MP series for Cu+. Direct CCSD(T) calculations yield binding energies which are not necessarily closer to the experimental values. Quite importantly, the B3LYP method provides reasonably good Cu+ binding energies.