Paolo Giannozzi wrote: > Stefano Baroni wrote: > > >> On top of that, I would say that there would be little point to use >> k-point sampling to describe a defect, which would require a supercell >> in any case. >> > > yes and no. There are cases in which relatively small supercells > describe well the defect (i.e. the interaction of the defect with its > periodic replica is small) but do not describe well the underlying > crystal structure if one uses the Gama point only. In those cases, > using k-points makes sense. > > Paolo > There are also cases (for example Si interstitials in SiC) in which a defect level in the gap is still dispersed with relatively large supercells but using K-points allows to have relatively well converged total (i.e. formation energy). It has been proposed (Bernardini et al. Eur. Phys. J. B, v38, p437 (2004)) to impose the same occupation occuring at Gamma to the other k-points (where spurious breaking of degeneracies could appear). My impression is that this cumbersome procedure can be dangerous as, I believe, it does not give the DFT ground state (which, as badly converged as it can be with the supercell size, is what we are looking for). I would be happy to have the feeling of others in the list on this point.
Guido -------------- next part -------------- A non-text attachment was scrubbed... Name: guido_roma.vcf Type: text/x-vcard Size: 260 bytes Desc: not available Url : http://www.democritos.it/pipermail/pw_forum/attachments/20090206/f9466b5e/attachment.vcf
