TY - JOUR
T1 - Total-energy all-electron density functional method for bulk solids and surfaces
AU - Weinert, M.
AU - Wimmer, E.
AU - Freeman, Arthur J
PY - 1982
Y1 - 1982
N2 - A new formalism for determining highly accurate total energies of solids within density functional theory is presented in which all necessary terms are easily obtained from the energy-band calculation. A major feature of this all-electron approach is the explicit algebraic cancellation of the nuclear Coulomb singularities in the kinetic and potential energy terms which leads to good numerical stability. As an illustration, the method is implemented in the full-potential linearized augmented-plane-wave method for thin films and applied to monolayers of Cs and graphite. The structural information (lattice parameters, force constants, etc.) for graphite are found to be in very good agreement with experiment on bulk graphite and to be rather insensitive to the quality of the basis. The calculated cohesive energy (relative to a spin-polarized local-density atom), on the other hand, is quite sensitive to the quality of the basis; a limited basis yields results in fortuitous agreement with experiment. The converged result for the cohesive energy is found to be 17% too large compared to experiment, an error which appears to arise from the neglect of correlation with near-lying excited configurations in the local-density atom and not to errors in the condensed system.
AB - A new formalism for determining highly accurate total energies of solids within density functional theory is presented in which all necessary terms are easily obtained from the energy-band calculation. A major feature of this all-electron approach is the explicit algebraic cancellation of the nuclear Coulomb singularities in the kinetic and potential energy terms which leads to good numerical stability. As an illustration, the method is implemented in the full-potential linearized augmented-plane-wave method for thin films and applied to monolayers of Cs and graphite. The structural information (lattice parameters, force constants, etc.) for graphite are found to be in very good agreement with experiment on bulk graphite and to be rather insensitive to the quality of the basis. The calculated cohesive energy (relative to a spin-polarized local-density atom), on the other hand, is quite sensitive to the quality of the basis; a limited basis yields results in fortuitous agreement with experiment. The converged result for the cohesive energy is found to be 17% too large compared to experiment, an error which appears to arise from the neglect of correlation with near-lying excited configurations in the local-density atom and not to errors in the condensed system.
UR - http://www.scopus.com/inward/record.url?scp=26544480072&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=26544480072&partnerID=8YFLogxK
U2 - 10.1103/PhysRevB.26.4571
DO - 10.1103/PhysRevB.26.4571
M3 - Article
AN - SCOPUS:26544480072
VL - 26
SP - 4571
EP - 4578
JO - Physical Review B-Condensed Matter
JF - Physical Review B-Condensed Matter
SN - 1098-0121
IS - 8
ER -