U_MINUS_J values smaller than other software

[mdsimula...@gmail.com]

Hi,

I'm using DFT+U to model transition metal oxides in the GAPW method and I've noticed that the U_MINUS_J values I get tend to be smaller than those values reported in literature for software such as VASP, Quantum Espresso and/or abinit.  Anyone know why?

Thanks,

Frank

[Krack Matthias (PSI)]

Hi

 

U values are not transferable between methods/codes.

 

M.

--
You received this message because you are subscribed to the Google Groups "cp2k" group.
To unsubscribe from this group and stop receiving emails from it, send an email to cp2k+uns...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/cp2k/768143a7-738b-44eb-ab6d-68ce31338847n%40googlegroups.com.

[mdsimula...@gmail.com]

Hi Matthias,

Thanks and I understand they are not transferable but I have noticed that CP2K's U values tend to be significantly smaller than other method/codes, by at least half.  For instance, the U value for CuO in VASP and Quantum Espresso is around 7 eV, while for CP2K, I'm finding it to be around 2 eV.

What makes this large difference?

I'm trying to understand the differences so I can better match literature properties (spin moment, band gab, cell structure).

Thanks,

Frank

[Krack Matthias (PSI)]

Hi Frank

 

Have a look at this paper in which the differences in U values are analyzed and discussed. Codes using localized basis sets seem to need often much smaller U values compared to the U values usually employed for PW codes.

 

Matthias

To view this discussion on the web visit https://groups.google.com/d/msgid/cp2k/c912d74b-8f7c-4841-812b-fd073361badfn%40googlegroups.com.