Qualitative Difference from VASP for Perovskites

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

Moderator: bguster

Locked
jleehand
Posts: 1
Joined: Fri Feb 08, 2019 6:32 pm

Qualitative Difference from VASP for Perovskites

Post by jleehand » Tue Jul 30, 2019 7:02 pm

I am trying to determine what the lowest energy spin configuration is for a variety of crystal structures of Perovskite Oxides. My research supervisor is also doing some of the ground state calculations in VASP so that we can compare results. We have found, that when using LDA+U, abinit predicts an antiferromagnetic structure whereas VASP predicts a ferromagnetic structure. I think I have all of the settings the same, so we are surprised by the qualitative difference.

I have attached the input files I have used for a particular calculation in abinit and (if anyone has any experience which might help diagnose this) a sample VASP input file for the same structure. I am probably just making a stupid mistake somewhere, so any help is much appreciated.
Attachments
fmpol.in
Structure optimization
(646 Bytes) Downloaded 190 times
fmpol_band.in
Bandstructure calculation
(825 Bytes) Downloaded 205 times
INCAR.in
VASP INCAR
(378 Bytes) Downloaded 232 times

amadon
Posts: 46
Joined: Mon Aug 24, 2009 10:58 am

Re: Qualitative Difference from VASP for Perovskites

Post by amadon » Wed Jul 31, 2019 11:13 am

Dear user,

Could you try LDAUTYPE=1 with VASP ? LDAUTYPE=2 is a simplified approach (it should correspond to use U=U-J and J=0 in ABINIT).
LDAUTYPE=1 should compare well to the ABINIT result.

Best regards
Bernard Amadon
CEA
France

ebousquet
Posts: 469
Joined: Tue Apr 19, 2011 11:13 am
Location: University of Liege, Belgium

Re: Qualitative Difference from VASP for Perovskites

Post by ebousquet » Wed Jul 31, 2019 5:45 pm

Dear jleehand,
On the top of Bernard's comment I would also add that it can depend on the pseudopotential file you are using (the most common are JTH and GBRV tables).
Another point is that you defined a smearing value in Abinit (tsmear=0.005Ha=0.13605eV) while not in Vasp such that in your Vasp calculation you have the default smearing which is not 0.13605 eV. This smearing could be dramatic on the magnetic energy differences in such metallic case.
Best wishes,
Eric

Locked