Dear Antoine; Thanks so much for your eloquent explain. But what is the file you attached?! Regards; Akram --- On Wed, 4/14/10, Antoine Pairet <antoine.pai...@student.uclouvain.be> wrote:
From: Antoine Pairet <antoine.pai...@student.uclouvain.be> Subject: Re: [SIESTA-L] About Band Gap To: siesta-l@uam.es Date: Wednesday, April 14, 2010, 1:13 PM On Wed, 2010-04-14 at 05:30 -0700, Akram Keyhannezhad wrote: > Dear all, > Can someone say me which parameters have affect on band gap energy? > I have a huge underestimate at gap energy calculation. The band gap underestimation is a problem affecting the DFT formalism and is not specific to siesta. If your band gap is underestimated with respect to what is experimentally measured, it is normal. However, if other DFT computations give better results, then you should look at the parameters of your calculation. >From my experience with silicon (GGA), looking at the basis influence: min BC max BV GAP sz -2.7098 -4.2014 1.4916 szp -3.6122 -4.4528 0.8406 dz -2.5853 -3.8765 1.2912 dzp -3.6239 -4.2432 0.6193 While the computed gap seems better without polarization, it is important to note that the band structure is correctly described ONLY when the basis is polarized. In fact, the bandgap is closer to what one expects from DFT as the basis size is increased and polarized. > How I can improve it? * Try with LDA and GGA * You can apply a scissor operator: cutting inside the gap the gap and adding to CB an energy equal to E_f(exp)-E_f(computed). * perform GW computations Hope this helps, Best regards
