Hello Alberto, Thanks for your answer. Comments below...
On Thu, Mar 11, 2010 at 4:12 PM, Alberto Garcia <alber...@icmab.es> wrote: > Nicolas, > > There are two possible reasons for the differences. > > 1. If the system is an insulator, the placement by Siesta of the Fermi > level is rather arbitrary: you can count on it being in the gap, but > the actual > location will depend on how many states you start with in the > bisection algorithm used. Mprop uses a different algorithm. In this > case the physics > does not depend on the location, as long as it is in the gap. > My system is not an insulator : it's functionnalized graphene with a shifted band crossing at the Dirac point (somewhere on the path between K' and \Gamma). So it is still semi-metallic. Or does the mere fact I am calculating with SIESTA the wave-functions (.WFSX input to be used with mprop) at the \Gamma point only interferes somehow in the algorithm? > > 2. In general, mprop will work with a certain (gaussian) smearing. > Siesta would have used typically a Fermi-Dirac smearing, according to > the electronic temperature specified. If you use a more or less > equivalent smearing in mprop, you should get similar values for the > Fermi level. > If point 1 does not concern me, it should be related to this point. But it does not fix the issue : - In the .EIG file of the SIESTA run, my electronic temperature is 350 K (which corresponds roughly to 0.03 eV). I obtain a Fermi energy energy of -4.7 eV. - In the .stt file after mprop post-processing with the same 0.03 eV smearing, I obtain a Fermi energy value of -6.4 eV. The discrepancy between the two values is still considerable. Any other comments/ideas? Thanks in advance, Nicolas > Alberto > > > On Thu, Mar 11, 2010 at 9:52 AM, Nicolas Leconte > <lecontenico...@gmail.com> wrote: > > Hello, > > Quick question. Is there a particular reason the Fermi energy value > obtained > > by a regular SIESTA calculation in the *.EIG file is different in > absolute > > value from the one found in the .stt file after a COOP analysis with the > > mprop executable in the Util directory? > > Thanks in advance, > > Nicolas >