On 2/6/14 4:00 PM, Hela Ouali wrote:
On 06/02/2014 15:43, Riccardo Rurali wrote:
Dear Siesters,
I am having trouble with Ge in LDA with Siesta.
As is well-known, Ge is a pathological case within DFT-LDA: the gap
shrinks so much that is becomes negative, i.e. Ge is metallic in LDA.
However, I'm having problems in recovering this behavior with Siesta
and I get gaps of 0.4-0.5eV, depending on fine details of the basis set.
I have tried both optimized and automatic DZP basis sets. Short and
long. "Vanilla" pseudos or pseudos with extra tuning (e.g.
core-corrections). Nothing. The gap's still there.
Has any of you experience with treating Ge with Siesta ans trying to
recover to "right" (which is wrong in reality...) LDA behavior?
Any help would be greatly appreciated.
Riccardo
how do you determine your gap ? DOS or bandlines ? and how many k points
do you put ?
I normally look at both, but I tend to trust more the band diagram which
does not depend too critically on the k-point mesh. Anyway I'm using a
12x12x12 grid for Ge primitive cell.
I find difficult to believe, anyway, that I could get rid of a 0.5 eV
gap by increasing the k-point sampling.
Riccardo
--
Riccardo Rurali
Institut de Ciència de Materials de Barcelona (ICMAB)
Consejo Superior de Investigaciones CientÃficas (CSIC)
Campus de Bellaterra
08193 Bellaterra (Barcelona)
Spain
tel.: +34 93 5801853 ext. 347
e-mail: rrur...@icmab.es
http://www.icmab.es/dmmis/leem/
Man, the dope's that there's still hope
