error in bond dissociation energy of gas phase molecule

[Jun]

Thanks for the comments.

> I remember a similar case where the production of a spin polarized
> intermediate resulted in a "wrong" activation energy. The problem
> was related (at that time) to the carbon pseudopotential
> that had bad transferability from non-spin-polarized to
> spin-polarized situations. This could be a similar situation.

If it is the case, the problem seems serious. Is it possible that the
problem is less serious in condensed phases?

> BTW check the literature for the really bad behavior of all
> non SIE corrected functionals for this type of reaction.

We are comparing the computed bond dissociation energies for the same
functional, other than with experiments. Or, am I misunderstanding
anything? B3LYP gives rather satisfactory value, while BLYP value is
too large. Our gpw calculations always give values about 0.5 eV lower
than gapw, gaussian and literature results.

Jun