Degree of localization of the exchange-correlation hole and its influence on the ground-state (structural and magnetic) properties of d metals

The Perdew, Burke and Ernzerhof (PBE) generalized gradient approximation (GGA) is the most popular exchange-correlation energy used in today's ab initio studies. The GGA is tested here in relation to the intrinsic uncertainty in choosing the degree of localization of the exchange-correlation hole (the K-coefficient in the spin-polarized enhancement factor). The proposed and most commonly used value of kappa = 0.804 (best suited for atoms and molecules) works well for some solids but should be modified in many cases in order to predict lattice parameters in good agreement with experiments. The effect on the structural and magnetic properties of 3d, 4d and 5d metals including the structural phase order of Fe is examined using two different state-of-the-art ab initio implementations of density functional theory: the full-potential linearized muffin-tin orbital and full-potential linearized augmented-plane-wave methods. This study gives examples for the case of elemental d metals of the errors associated with these properties when using the PBE-GGA in state-of-the-art ab initio electronic structure studies. (Metals used were Sc, Ti, V, Cr, Co, Fe, Ni, Cu, Zn, Zr, Nb, Mo, Tc, Ru, Rh, Pd, Ag, Cd, Hf, Ta, W, Re, Os, Ir, Pt, and Au.)