Assessing a solids-biased density-gradient functional for actinide metals

Recent developments of new electron exchange and correlation functionals within density-functional theory include a solids-biased modification of the popular Perdew-Burke-Ernzerhof (PBE) functional and is referred to as PBEsol. The latter is claimed to remove a bias toward free-atom energies in the former and is therefore better suited for equilibrium properties of densely packed solids and surfaces. We show that PBEsol drastically worsens the equilibrium properties of the actinide metals compared to PBE and produces results closer to that of the local density approximation. The PBEsol atomic volume of delta-Pu is 12% and 14% smaller than PBE and experimental values, respectively. Also, iron is predicted to have the incorrect ground-state phase within PBEsol. These results illustrate the difficulties and limitations in improving the gradient approximations of the electron exchange and correlation functional in a general fashion even when the application is restricted to solids. We comment on the possibility of formulating a unique functional without these limitations that is applicable to solids as well as to finite-sized systems such as atoms and molecules.