The evaluation of four high-level composite methods based on the modification of Gaussian-3 (G3) theory for radicals and 18 exchange-correlation density functionals, including modern long-range and dispersion-corrected functionals, in the modelization of singlet diradicals has been performed in this work. Structural parameters and properties such as singlet-triplet gaps, electron affinities, ionization potentials, dipole moments, enthalpies of formation, and bond dissociation energies have been calculated in a set of six well-characterized singlet diradicals, and benchmarked against experimental data and wavefunction-based CASSCF/CASPT2 calculations. The complexity of the open-shell singlet ground state is revealed in the difficulties to properly represent the diradical character reported by some DFT functionals, specially those that do not comprise a certain amount of Hartree-Fock exchange in their formulation. We find that STGs, EAs, dipole moments, and thermochemical properties are, in general, satisfactorily calculated, while for IPs larger deviations with respect to the experiments are found in all cases. The best overall performance is accounted for by hybrid functionals, including some of the long-range corrected functionals, but also pure functionals, comprising the kinetic energy density in their formulation, are found to be competent. Composite methods perform satisfactorily, especially G3(MP2)-RAD and G3X(MP2)-RAD, which calculate singlet-triplet gaps and electron affinities more accurately. On the other hand, G3-RAD and G3X-RAD provide better ionization potentials. This study emphasizes that the use of recently developed functionals, within the broken symmetry approximation, is an appropriate tool for the simulation of organic singlet diradicals, with similar accuracy compared to more expensive composite methods. Nevertheless, suitable selection of the methodology is still crucial for the accomplishment of accurate results.
Download full-text PDF |
Source |
---|---|
http://dx.doi.org/10.1007/s00894-016-2950-z | DOI Listing |
Enter search terms and have AI summaries delivered each week - change queries or unsubscribe any time!