We report on the results of density-functional-theory based calculations of the vacancy formation energies in metals using the revised Tao–Perdew–Staroverov–Scuseria (revTPSS) functional (Pedrew et al., 2009), which is a self-consistent semilocal meta-generalized gradient approximation functional. The motivation for this work is to determine if the improved accuracy of surface energies for revTPSS compared to local and generalized gradient approximation functionals also leads to improved vacancy formation energies since vacancies can be viewed as internal surfaces. In addition, we report on the lattice constants, cohesive energies and bulk moduli predicted by revTPSS. By comparing the vacancy formation energies and bulk properties, the performance of revTPSS is assessed against four functionals: the local spin density approximation (LSDA), Perdew, Burke and Ernzerhof (PBE), Perdew–Wang-91 (PW91), and PBE for solids (PBEsol). Using an automated computational approach, we calculate the vacancy formation energies and the macroscopic properties of 34 metal systems for the five functionals. For macroscopic properties (lattice constants, cohesive energies and bulk modulus), we find the results for revTPSS typically lie between LDA and PBE with a mean absolute percentage deviation of 1.1% and 12.1% from the experimental data for lattice constants and cohesive energies respectively. Further, it is found that revTPSS predicts higher vacancy formation energies when compared to the four other functionals surveyed. We have observed the following order for the functionals with respect to the computed vacancy formation energies, Efxc:EfrevTPSS>EfPBEsol∼EfLDA>EfPBE>EfPW91. We also consider the effects of a surface-energy error correction that has been proposed for standard LDA and GGA functionals. This correction increases the vacancy formation energies of LDA, PBE and PW91 functionals. The revTPSS computed VFEs are greater than the surface-energy-corrected PBE VFEs by a mean relative difference of 14.8%.
Read full abstract