Too many bands in the result

[tomic]

I was doing some spin-polarized calculations on the on the orthorhombic phase of CaFe2As2. Density of states came out correctly, but the bandstructure features additional bands, not appearing in the bandstructure obtained with the Wien2k code.

I have also tried to simplify things by doing the non-spin-polarized calculations on the tetragonal high temperature phase of the CaFe2As2 and the same problem repeated.

Can anyone help me to mitigate this problem since I am new to VASP and DFT altogether and I ran out of ideas.

Thanks in advance

[forsdan]

You are quite vague about where you see the additional bands. Anyhow, the most probably reasons to more bands are that you either have more valence electrons in your pseudopotentials, more empty bands or looking at different symmtery directions (where degeneracies are lifted). Different pseudopotentials may also lead to lifted degeneracies. Neither is specificly connected to VASP or DFT.

Cheers,
/Dan

[tomic]

Thanks for the quick response

I have used the same k-path as in Wien2k: K-Z-Gamma-X-P-N

What is baffling to me is that on the X-P-N section, bandstructure output from VASP coincides with the output of Wien2k but on K-Z-Gamma-X section I see additional bands alongside bands I got with Wien2k also.

I have also plotted only the bands which cross the fermi level. I have ten bands, which is OK since I have 2 nonequivalent iron atoms in the unit cell and their d-levels are the ones which cross the fermi level. Since iron atoms are symmetry related I should only see 5 bands due to degeneracy which is the case on the X-P-N section but not elsewhere.
<span class='smallblacktext'>[ Edited Wed Aug 18 2010, 03:11PM ]</span>

[tomic]

OK, what is apparently happening is the mixing of bands at kz=2pi and kz=0

Here's the bandstructure:

Bandstructure plot

Bandstructure section K-Z-Gamma-X has mirror symmetry around it's median line and in the reciprocal lattice Gamma-X and K-Z points have mirror symmetry around kz=pi plane:

K = (0.5, 0.5, 1)
Z = (0, 0, 1)
Gamma = (0, 0, 0)
X = (0.5, 0.5, 0)

Spurious bands appearing in one half of the K-Z-Gamma-X section are mirror images of the correct bands in the other half. If I would remove those mirror images, my bandstructure would coincide with the Wien2k result.

I cannot figure out why this is happening. Any help is, again, very welcomed.

Thanks in advance [/link] [link=hyperlink url]
<span class='smallblacktext'>[ Edited Wed Aug 18 2010, 03:47PM ]</span>

[aron]

WIEN2K uses the crystal symmetry to constuct the primative unit cell. In VASP, you have to input the primative cell yourself, hence, why you are seeing band folding to the gamma point.

[tomic]

So, how does one deal with this problem? Can some simple trick with VASP be used to remedy this, or external code that does symmetry analysis is needed?

[pkroll]

What aron was pointing out was: if you are using a structure with space group I4/mmm for your CaFe2As2, you have to do the band structure using the primitive unit cell.
You yourself have to provide the positions and the cell parameters. Check a standard book on crystallography – or use some common sense. It's easier than you think.
Once you understand the transformations necessary, you know how to handle the input-matrix in VASP (line 3-5) and can reorientate your unit cell according to your liking (e.g. to have only particular orbitals contributing to bands along a defined direction in the band structure).