Band level filling has no sharp cut-off at Fermi energy !?

[lahaye]

Hi all,

I have a structure of 14 ions more or less in a plane. The structure has
a mirror plane perpendicular to the structure.

The PROCAR file has all the bands / k-points with their occupancies.
Very low energy levels have an occupancy of 2.000, but approaching
the Fermi energy, the occupancies gradually decrease, which continues
above the Fermi energy.
And also: some levels below the fermi energy have occupancies over
2.000 !?!?!

I'd expect all occupancies to be 2.0000, which abrupty changes to
0.0000 at the fermi energy.

So to me this is very strange, as if there is some kind of thermal energy.
I use as smearing the tetrahedron with Blochl corrections (SIGMA=0.001).

Any idea what is the reason for this?

I use PAW potentials and the INCAR file is like this:

Code: Select all

PREC  = Accurate
EDIFF = 1E-5
LREAL = Auto
NEDOS  = 1001
ISMEAR = -5
SIGMA  = 0.001
LORBIT = 12
NBANDS = 39

KPOINTS file is like this:

Code: Select all

k-points with Monckhorst-Pack
0
Monkhorst-Pack
9 9 9
0 0 0

A printout of energy vs. occupancy of the energies around
the fermi energy is here:
http://surfion.snu.ac.kr/~lahaye/vasp/Fermi

Thanks,
Rob.

[admin]

first of all, please check if
1) your system really has a gap for all k-points
2) you have even/odd NELECT

please mind that
-- the tetrahedron method does not use SIGMA to calculate the weights of the states
-- the Bloechl corrections to the tetrahedron method may lead to occupancies >2 and [ Edited Fri Dec 01 2006, 01:58PM ]

[simly]

Hi,
My system is negatively charged (odd NELECT). This leads to occupancies >2 and <0 in the OUTCAR.
The system should have a bandgap . How do I then interpret the occupancies >2 and <0 found?

Thanks for your help!~