Fermi level and Fermi energy differ from other softwares  [SOLVED]

Total energy, geometry optimization, DFT+U, spin....

Moderator: bguster

Locked
antonio
Posts: 52
Joined: Tue Apr 23, 2013 6:16 pm

Fermi level and Fermi energy differ from other softwares

Post by antonio » Wed Jan 15, 2014 6:00 pm

Dear all,

I need some clarification about the definition of Fermi energy and Fermi level in abinit. To my knowledge, the Fermi energy or HOMO is the last occupied level (Highest Occupied Molecular Orbital), while the Fermi level is the hypotetical level that would have an occupancy of 1/2 at thermodynamic equilibrium. Now, according to this post

http://forum.abinit.org/viewtopic.php?f=9&t=2184&p=6936&hilit=fermi+energy#p6936

the "Fermi energy" I get in the header of the DOS_AT* files is the amount of energy I have to use as a shift to have the Fermi level at zero. However, if I look for the "Fermi energy" in the output, I always get a value that is different from that I find in the header. My question is: is the "Fermi (or HOMO) energy (hartree)" value I get in the output the same as the Fermi level? Or it is just the energy of the highest occupied molecular orbital and the value of the Fermi level is elsewhere?

My question comes from the following incongruence. To check my results in a particular case, I calculated the DOS with abinit and VASP using the same settings and basis set. In abinit, I get the "Fermi energy"=0 in the header of the DOS* files, so, according to the above mentioned post, I don't have to translate my DOS, despite in the output, at the line "Fermi (or HOMO) energy (hartree)" I get a value different from zero; in VASP, I get the Fermi level at exactly the same energy value that abinit reports in the line "Fermi (or HOMO) energy (hartree)". My problem here is that if I don't translate the abinit DOS (as per the DOS* header files), I get a metallic state, while if I translate them I get the same results as in VASP, that is an insulating state as expected.

Thanks a lot in advance for your precious help.

Best regards

Antonio

User avatar
gmatteo
Posts: 291
Joined: Sun Aug 16, 2009 5:40 pm

Re: Fermi level and Fermi energy differ from other softwares  [SOLVED]

Post by gmatteo » Sat Jan 18, 2014 12:51 am

Use the Fermi energy reported by Abinit at the end of the GS SCF cycle.
The fermi energy cannot be computed during the NSCF band structure calculation
and therefore abinit sets it to zero.
This explains why you have Fermi energy=0 in the header of the DOS file.

antonio
Posts: 52
Joined: Tue Apr 23, 2013 6:16 pm

Re: Fermi level and Fermi energy differ from other softwares

Post by antonio » Wed Jan 22, 2014 5:13 pm

OK. Thanks a lot for your reply.

Best

Antonio

Locked