why the topmost empty bands are never get optimized correctly

Message

yilmaz · #1 Post by **yilmaz** » Mon Mar 05, 2007 3:54 pm

hi,
i would be thankfull to anyone give me some comments.

problem::
in my band structure calculations, the topmost empty bands (conduction bands) are always zig-zag like, they are never optimized well. but the rest of the bands are quite nice.

the system is a nanotube, (altough i have tested silicon too, but same problem).

ECUT and FFT meshes were quite high.
ISMEAR =0 always
ISTART=0 always
i have followed all the steps in the vasp manual.

1).... got a charge density (ICHARG=2) with a dense k-mesh (1 1 18) over the initially optimized structure.

2).... and used ICHARG=11 with 1 1 30 kpoints to calculte the band energies.

in this case there was 40 filled (80 electrons) and 20 empty bands.
the problem is not related to plotting software, i have checked that.

thanks for any suggestions.
h. yilmaz

#2 Post by **admin** » Mon Mar 05, 2007 4:23 pm

DFT is a ground state theory...

yilmaz · #3 Post by **yilmaz** » Mon Mar 05, 2007 4:34 pm

thanks alot Admin.

so i guess, i am not doing anything wrong. i should not care about the uppermost empty bands.

#4 Post by **admin** » Thu Mar 08, 2007 10:05 am

yes, DFT ONLY optimizes the occupied bands (please have a look at the basic papers of Kohn and Sham, and Hohenberg and Kohn).

tlchan · #5 Post by **tlchan** » Thu Mar 08, 2007 4:29 pm

Is there a way to tell vasp to calculate the unoccupied states exactly as well? For example, the unoccupied states a few eV above the Fermi energy are useful in analysis. Thank you.

#6 Post by **admin** » Fri Mar 09, 2007 8:17 am

no, not within standard DFT. The best approaches to do it are TDDFT or GW. GW will be included in the new vasp release 5.1 (please have a look at out homepage).