Bonding of NH3, CO, and NO to NiO and Ni-doped MgO:: a problem for density functional theory

Recent experimental results (Hoeft et al 2001 Phys. Rev. Lett. 87 086101) have questioned the capability of current theoretical methods for describing the bonding of NH3, CO, and NO with the NiO(100) surface. We show that these systems do indeed represent a challenge to theory. For different reasons, density functional theory (DFT) fails in describing the bonding of these molecules to the NiO surface. The gradient-corrected functionals which work better for the properties of NH3/NiO and CO/NiO (energies, geometries, vibrations) provide wrong answers for NO/NiO and vice versa. This is not due to the well-known difficulty as regards DFT describing the insulating character of NiO. In fact, exactly the same problem is found for isolated Ni2+ impurities in MgO. A correct description of the bonding of both closed-shell (NH3 and CO) and open-shell (NO) molecules to NixMg1-xO is obtained only after inclusion of dynamical correlation and dispersion forces via wavefunction-based methods. However, even with correlated calculations some uncertainties exist regarding the predicted value of the energy of adsorption of NO on NiO. While CASPT2 calculations reach reasonable agreement with experiment, the results of approximate coupled-cluster calculations (the multi-configuration coupled-electron-pair approach) substantially underestimate the adsorption energy.