I'll try to do that. But I think I have to use the same EnergyShift for both the molecule and the isolated atom, sure?
If you don't use optimized basis set, the more extended the basis functions, the better. So I would use the "long orbitals" for both the molecule and the isolated atom. On the other hand, if you were to calculate, say, an O interstitial in some bulk material, you should probably do well enough using the extended basis set when you calculate the atom alone. The idea, as far as I understand it, is that when the O atom is inside the bulk host he has the basis functions of all the surrounding atoms available to self-consistently converge the charge, while with the isolated atom you do constrain it into a too short orbitals.
Hence, coming back to your question, I'd use the extended basis for both O and O2.
Let us know if it works! :-) Riccardo -- Riccardo Rurali Escola Tècnica Superior d'Enginyeria (ETSE) Universitat Autònoma de Barcelona (UAB) Campus de la UAB 08193 Bellaterra (Barcelona) Spain tel.: +34 93 581 3531 fax.: +34 93 581 2600 e-mail: [EMAIL PROTECTED] Man, the dope's that there's still hope
