CCL:G: about electron affinity

["Pierre Archirel" <pierre.archirel:+:u-psud.fr>]

 Sent to CCL by: "Pierre  Archirel" [pierre.archirel~~u-psud.fr]
 Dear colleagues,
 I have a closed shell molecule (rather large: 12 first row atoms) and I want to
 know if this molecule can bind an electron.
 DFT tells me "yes" with positive values of the electron affinity, in
 the range 0.5-1.0 eV according to the functional. I suspect that these values
 are false, or at least largely overestimated.
 I now turn to ab initio methods, but I first see that at the HF level the
 electron is not bound and the SOMO built of the most diffuse gaussians of the
 basis.
 The CCSD(T) and SAC-CI methods (which are affordable in my case) give a negative
 electron affinity... but are these methods reliable? These methods use the HF
 orbitals, can they bind an electron if it is not bound at the HF level?
 Or shall I use CAS methods (very expansive in my case) which optimize orbitals
 at the CI level? Is it obligatory in this case?
 Many thanks in advance,
 Pierre Archirel
 LCP, Universite Paris-Sud, Orsay, France