Big difference for Fermi energy between Abinit and VASP

External optional components of Abinit (BigDFT, Wannier90...) and codes using/providing data from/to Abinit (AtomPAW, ONCVPSP, EXC, DP, Yambo...)

Moderators: ebousquet, bxu

Locked
onion2440
Posts: 34
Joined: Sat Sep 05, 2015 10:04 am

Big difference for Fermi energy between Abinit and VASP

Post by onion2440 » Fri Aug 09, 2019 4:11 pm

Dear every one,
Now I want to reproduce the electronic band structure of Nb. I calculated it with Abinit (NC potential) and VASP (PAW) code, respectively. However, there is a big difference for Fermi energy between them, as showing below.
I want know how to reset the Fermi level in Abinit?
By the way, the band structure data calculated with Abinit is extracted directly from the tffield_xo_DS2_EBANDS.agr file. Since "# Energies are in eV. Zero set to efermi, previously it was at: 9.21 [eV]", I don't modify the Fermi energy again.

Best,
Yang
Attachments
band-abinit.jpg
band-vasp.jpg

onion2440
Posts: 34
Joined: Sat Sep 05, 2015 10:04 am

Re: Big difference for Fermi energy between Abinit and VASP

Post by onion2440 » Sat Aug 10, 2019 3:59 am

I'm sorry I got a mistake in calculate band structure with VASP.
But I still want to know how to modify the Fermi energy in Abinit, or some thing (such as tsmear, etc) that affect the Fermi energy.

Best wishes,
Yang

NPike
Posts: 34
Joined: Fri Mar 18, 2016 1:15 pm

Re: Big difference for Fermi energy between Abinit and VASP

Post by NPike » Sun Oct 27, 2019 9:25 pm

Hello Yang,

The Fermi energy is determined by "counting" the number of electrons in the unit cell. You can think of this as integration of the density of states from negative infinity to some energy Ef such that the result of the integral is the number of electrons in the unit cell.

Thus, the exact value of the Fermi energy that you get is a function of all the approximations you made to determine the energy of the system. So, one should not expect the exact same number for a the Fermi energy from a VASP and ABINIT calculation due to differences in approximation methods (your choice of pseudopotential being one of them).

However, to change the value of the Fermi energy, you can add or remove charge from the system using the input variable charge. See https://docs.abinit.org/variables/gstate/#charge for more information.

-Nick
Nicholas A. Pike
Research Scientist

Locked