Conversion of default unit of DOS to states/eV[SOLVED]

Moderators: mverstra, aromero

Conversion of default unit of DOS to states/eV  [SOLVED]

Dear Abinit Users

After the density of states calculation, in the DOS file, the units of DOS is indicated as electrons/Hartree/cell. I would like to change the units to states/eV or electrons/eV, but I am not sure of how to do this.

Can anyone help me to understand the conversion of the default unit of DOS to states/eV.

Any comments on this are highly appreciated.

Regards
Seba
Seba Darshan

Posts: 63
Joined: Mon Feb 17, 2014 6:19 am

Re: Conversion of default unit of DOS to states/eV

Dear Seba,
electrons means "states"
1 Hartree = 27.118 eV

Just multiply the x axis by 27.2118 and divide the y axis by 27.2118

Cheers
Jordan

Posts: 281
Joined: Tue May 07, 2013 9:47 am

Re: Conversion of default unit of DOS to states/eV

Hai

But I have another doubt, the DOS is showing as "electrons/Hartree/cell".

What does "cell" corresponds to? By dividing the Y-axis by 27.118, I can get the DOS in states/eV. Does this "cell" have any significance in the units.

Please shed some light on this.

Regards
Seba
Seba Darshan

Posts: 63
Joined: Mon Feb 17, 2014 6:19 am

Re: Conversion of default unit of DOS to states/eV

Dear all,

I have the same doubt for the conversion of the DOS.

Jordan wrote:Dear Seba,
electrons means "states"
1 Hartree = 27.118 eV

Just multiply the x axis by 27.2118 and divide the y axis by 27.2118

I am not sure that electrons means states. In one state, you can put two electrons :
one with a spin up and the other with a spin down. So 2 electrons means one state.
If the DOS is in electrons/Hartree/cell, I guess, we must divide the results by 2 to
get it in states/Hartree/cell.

Should anyone confirm that ?

Cheers
Hola

Posts: 4
Joined: Thu Mar 29, 2018 10:51 am

Re: Conversion of default unit of DOS to states/eV

cell means per unit cell. You always need to normalize this way as the DOS is extensive otherwise. NB: if you have several formula units you may need to divide by that number as well to compare to other literature

The nsppol coefficient is taken into account. If you have a spin polarized system, you will get 2 DOS, and have to add the two. If you have nspinor 1 each electron gets a band and it's "straightforward". If nsppol=nspinor=1 then there should be a factor of 2 in the occupation, which is folded into the calculation of the DOS. You can check it with a quick example (closed shell) atom in a box: use nsppol 1 and 2 and compare the DOS to the sum of the spin DOS.

M.
Matthieu Verstraete
University of Liege, Belgium
mverstra

Posts: 637
Joined: Wed Aug 19, 2009 12:01 pm