Dear Susi, Oleg, ans
Mikael,
Thank you for your responses, which were very
helpful.
I have succeeded to find the broken symmetry solutions with
wavefunction stability checks.
In
short, there was no correlation between ease of finding the broken
symmetry solution and biradical character of the final broken symmetry
solution. There seems to be a correlation with the amount of change in
electron density, at least for the few systems I studied.
I
have a follow-up question: Ess et al. (J. Chem. Phys. A, 2011,115,77)
used fractional spin occupation DFT (FS-DFT) to analyze broken symmetry
solutions without spin contamination. They start from the closed-shell
singlet orbitals and change the occupancies of HOMO and LUMO to 0.5
each, then do a SCF calculation with fixed occupations.
I am trying
to figure out whether this can give the exact solution as claimed. As
the closed-shell orbitals are symmetric, this will not give alpha and
beta electrons different space orbitals. In UHF the HOMO and LUMO form
linear combinations which allow alpha and beta electron to localize at
opposite ends of the molecule. I always considered this as an
essential driving force for symmetry breaking. Can anyone comment on
this?