>>>> What is DZP-DZP for Fe (...) ???? Double zeta plus polarization for both 3d and 4s.
>>>> So your approach is perfectly fine. Unfortunately, the results after that are not. :-( And just the energy differences, since the geometry is almost the same as the PW calculation gives you. For example, the final c/a ratio agrees 99%. And there are other complicated things. How to take into account the deformation caused by the presence of carbon? Not only local deformation, but also the change in the cell shape? >>>> Did you try to do spin-polarized calculations? All the calculations were spin-polarized. >>>> One more thing, as far as I understand, you haven't taken BSSE in your Fe >>>> and graphite vacancy formation energies, and cohesion energies. Thus you >>>> are making the error there as well, so your agreement with literature >>>> might be not that good. I think that is what makes this BSSE correction so controversial. I mean, sometimes you obtain very good results without it, in comparison with both/either experimental and/or BSSE-free plane waves results, for almost everything you want to calculate, but since it has been proven (I guess it has been done by Boys and Bernardi) that there is something ill-defined mathematically (this BSSE stuff), you have to believe all this agreement is merely fortuitous. Sometimes, on the other hand, without the correction, you cannot reach a quantitative agreement at all, neither with experiments nor PW. Of course, I know you cannot apply the correction only when it interests you... Thanks, Roberto Veiga ________________________________ From: Oleksandr Voznyy <[EMAIL PROTECTED]> To: [email protected] Sent: Friday, December 5, 2008 3:56:05 PM Subject: Re: [SIESTA-L] Still energy differences and BSSE corrections What is DZP-DZP for Fe, and DZP-DZ for C???? Yes, BSSE will always reduce the binding energy (cohesion energy, solvation enthalpy, etc.) since without ghosts, the total energy will always be higher than with ghosts. I would say that ghosts around C would be more important than in Fe, since Fe cluster has more degrees of freedom due to many atoms coupled close enough. So your approach is perfectly fine. Did you try to do spin-polarized calculations? (Although, normally they should lower even more the solution enthalpy). One more thing, as far as I understand, you haven't taken BSSE in your Fe and graphite vacancy formation energies, and cohesion energies. Thus you are making the error there as well, so your agreement with literature might be not that good.
