Validation of density-functional versus density-functional+U approaches for oxide ultrathin films.

A comparison between available experimental information and the predictions of density-functional and density-functional+U approaches is presented for oxide ultrathin films grown on single-crystal metal surfaces. Prototypical examples of monolayer phases of an ionic oxide (ZnO), a late transition metal oxide (NiO), and an early transition metal oxide (TiO(2)) are considered. The aim is to validate the theoretical approaches, focusing on the prediction of structural features and the reproduction of scanning tunneling microscopy images, rationalized in terms of the local density of states of the systems. It is found that it is possible to reasonably estimate the optimal lattice constant of ultrathin supported films and that the inclusion of the Hubbard U term appreciably improves the accuracy of theoretical predictions, especially in the case of nonpolar ultrathin phases of a transition metal oxide. Moreover, the optimal value of U for the oxide layer at the interface with the metal support is found to differ from that appropriate for the bulk oxide, as a consequence of the intermixing of oxide and support electronic states and screening effects.