Abstract
The way how different Density Functional Approximations (DFAs) are able to predict, in open shell systems, spin density that is the difference between the density of electrons with spin α and that of spin β is interesting to be investigated in order to assess their validity. Here a large panel of functionals were tested on a set composed by seven π-radicals expected to amplify DFA’s errors in modeling electron delocalization and spin polarization effects due to their extended electronic conjugation coupled with their planar structures. Our results show that generally the DFAs performance follow a systematic improvement in going from semilocal to hybrid functionals. More problematic is, instead the case of double hybrid functionals, where the perturbative contribution to correlation damps the positive effect of the presence of a high percent of exact exchange. More interestingly, differences are observed in the spin delocalization and polarization patterns, thus restraining the possibility of applying some of current DFAs to study chemically relevant properties, like molecular magnetism or charge/electron transport.
Funding Information
  • Agence Nationale de la Recherche (ANR-10-LABX 096, ANR-18-IDEX-0001)
  • H2020 European Research Council (648558)
  • Commissariat ? l?Investissement d?Avenir

This publication has 90 references indexed in Scilit: