Fermi energy in the conduction band with some functionnals

[Pierre_Beaujean]

Hello,

We are quantum chemist from the localized basis set world, trying to strengthen our knowledges of plane-wave based code.

When performing some tests with different functionnal on VASP, we observed something quite strange. For the context, the structure is the one of a perovskite, and we take into account the magnetization (with MAGMOM). Using PBE, we observe that there is no gap in the major spin channel, and a large one (>2eV) for the minor spin channel.

However, when switching to PBESol, both gap tends to zero (we checked the convergence of cut-off energy and k-point mesh grid). Looking at the DOS, we observed this:

image hosting for website

So it seems that the Fermi level is just above the conduction band. It is something that is known to happen with some functionnal? Does it makes sense?

Thanks in advance

[jonathan_lahnsteiner2]

Dear Pierre Beaujean,

It is hard for me to tell what is going on in your calculations without any input files. But since you are saying you are studying a perovskite structure I can recommend to take a look at this paper. Otherwise please upload your input files according to the VASP forum guidelines.

All the best Jonathan

[Pierre_Beaujean]

Sorry, the VASP forum was a bit slow yesterday. Here are the inputs and main output:

PBESol_perovskite.zip

[jonathan_lahnsteiner2]

Dear Pierre Beaujean,

You are using are using Gaussian smearing for the DOS computation which is not recommended. The recommended smearing technique for DOS calculations is the tetrahedron method. You can switch on the tetrahedron method by setting ISMEAR=-5.
For further information please take a look at the following vasp wiki pages:
Explanation of ISMEAR:
https://www.vasp.at/wiki/index.php/ISMEAR
Background about smearing techniques:
https://www.vasp.at/wiki/index.php/Smearing_technique
Maybe also this forum post can give more insight:
https://wwww.vasp.at/forum/viewtopic.php?t=19164
And here is a tutorial about computing the density of states for fcc Si:
https://www.vasp.at/wiki/index.php/Fcc_Si_DOS

I hope this helps to resolve your issues.

All the best Jonathan

[Pierre_Beaujean]

Thank you for this answer. However, I have an extra question: since this compound has a metallic nature for the major spin channel, is it fine if I use ISMEAR=-5?

The wiki (https://www.vasp.at/wiki/index.php/Smea ... hod_to_use) says that for metal one should use ISMEAR ≥ 0, but that ISMEAR=-5 should be used anyway for DOS. Is it correct?

Thanks in advance.

[jonathan_lahnsteiner2]

Dear Pierre Beaujean,

You can use tetrahedron method ISMEAR=-5 for DOS calculations also in metals. One should just not use ISMEAR=-5 for optimization (or in general when forces or stresses are needed) in metals because as mentioned on the wiki

The drawback of the tetrahedron method is that it is not variational with respect to the partial occupancies. Therefore the calculated forces and the stress tensor can be wrong by up to 5 to 10% for metals. Only for semiconductors and insulators, the forces are correct because the partial occupancies do not vary and are either zero or one.

Check the page under the header Tetrahedron methods. So yes you can use ISMEAR=-5 for DOS computation in metals and it is the recommended way.

All the best Jonathan