Density-Functional Calculations of Radicals and Diradicals

Open-shell molecules highlight some of the greatest strengths and weaknesses of density functional theory (DFT). When accurate geometries are used, doublet hyperfine couplings are in general well-predicted by spin-polarized DFT calculations. However, for a series of 25 phosphorus-containing radicals, when geometries are optimized at the same level of theory as is used for the prediction of hyperfine couplings, no pure DFT functional does as well as either the UHF or UMP2 levels of theory with identical basis sets. However, hybrid HF/DFT functionals do perform almost as well as MP2. In diradicals, on the other hand, the DFT formalism appears to account well for nondynamical and dynamical correlation effects that are not included in single-determinant Hartree-Fock theory. Singlet-triplet gaps are predicted well for a number of carbenes and nitrenium ions. Finally, a spin-annihilation procedure that involves the construction of a Slater determinant from DFT orbitals is shown to permit the accurate calculation of open-shell singlet energies.