Computing gold cluster energies with density functional theory: the importance of correlation

Energies calculated with density functional theory depend critically on the choice of the exchange-correlation functional. In this work, we use measured dissociation energies of Aun+ (n = 5-17) clusters as benchmark data to test two very different functionals for calculating total energies in these clusters; the simpler (and fast) PBE and the evolved (and expensive) B2PLYP double-hybrid functionals. PBE consistently gives poor agreement with the experimental results. In contrast, the B2PLYP functional, which implicitly includes electron correlation by performing a perturbative second-order correction, significantly improves the agreement of the calculations, at the cost of much more demanding computations. The better performance of the double-hybrid functional is ascribed to the longer range of the interatomic potential.