[Wien] Calulating optical property with spin-orbit coupling

[Moon, Chang-Youn]

Well, I guess I've just fixed the problem by recovering a few lines
which were commented out originally in lapw2.F,  describing that in the
case of including SO, just artificially increase the number of electrons
twice (elecn=elecn*2.0d0, around line 207 in lapw2.F) then estimate the
Fermi level, and then recover the number of electrons to the original
value. After modifying the code that way, lapw2 estimates the Fermi
level correctly with -SO flag, and the resultant optical spectra also
seems correct. Do you have any comment about this?

 

Chang-Youn 

 

  _____  

From: wien-bounces at zeus.theochem.tuwien.ac.at
[mailto:wien-bounces at zeus.theochem.tuwien.ac.at] On Behalf Of Moon,
Chang-Youn
Sent: Tuesday, September 27, 2005 3:30 PM
To: wien at zeus.theochem.tuwien.ac.at
Subject: [Wien] Calulating optical property with spin-orbit coupling

 

Dear users,

 

I'm trying to calculate the optical property of an semiconductor
material, and I'm following the user's guide for non-spin polarized
calculation with spin-orbit (SO)  interaction (page 127). When I didn't
include the SO switch, the resultant optical spectra ( the imaginary
part of dielectric matrix) seemed to be reasonable, but when I did
include the SO, the spectra seemed to be very strange. I guess it
resulted from the erroneous estimation of Fermi level in executing lapw2
with -fermi and -so switch to generate weight files. When SO is
included, the number of band becomes doubled because in this case the
spin-up and spin-down states merge. Then, the number of occupied band
should be also doubled, but I found that the lapw2 execution with -fermi
and -so switch also just gave me the same number of occupied band with
the case without -so switch,  and hence giving erroneously low value of
Fermi level. I tried to fix this problem by goring through the source
codes, but I couldn't. Do you think this is definitely a bug of the
program, or I did something wrong? I attach the first few lines (for one
k-point) of 'case.weight' files with (case.weight1) and without
(case.weight2) -SO flag in executing lapw2 with -fermi flag. You'll see
the evident error in determining the Fermi level in the latter case
comparing to the former one.  I'm using wien2k version 3 released on
18/01/2003.

 

Thanks,

 

Chang-Youn Moon

National Renewable Energy Laboratory

 

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://zeus.theochem.tuwien.ac.at/pipermail/wien/attachments/20050928/6085f83e/attachment.html