[Wien] Fermi-contact hyperfine fields

[pieper]

Thanks a lot for the explanation. So, since my main interest is magnetic 
materials moments on outer shells polarizing core s-electrons will 
remain problematic ... (sigh)

---
Dr. Martin Pieper
Karl-Franzens University
Institute of Physics
Universitätsplatz 5
A-8010 Graz
Austria
Tel.: +43-(0)316-380-8564


Am 21.09.2017 17:09, schrieb Peter Blaha:
> Hi,
> 
> No, Fermi-contact term has not "improved", we are still using the same
> GGA approximation.
> 
> However, in bcc Fe (a ferromagnet) the situation is usually VERY
> different from a NMR calculation.
> 
> In bcc Fe the main contribution comes from the core electrons (3s),
> which respond to the huge 3d moment and the hyperfine field is 30 T
> !!!!. The localized core electrons are not well describe by GGA.
> 
> 
> In NMR for paramagnetic metals, often (not always) the main
> contribution comes from an reoccupation of the valence 4s (or 5s in 4d
> elements, ...) and this valence electrons are "hopefully" well
> described by GGA.
> 
> It is true, however, that sometimes the 3d DOS at EF is large, which
> introduces a 3d moment and also some sizable core response to the HFF.
> However, the fields are only ppm of 100T, so much much smaller.
> (PS: we know for instance that the shift in hcp Sc is completely wrong
> in DFT calculations, because GGA seems to overestimate magnetism).
> 
> Peter
> 
> 
> Am 21.09.2017 um 13:37 schrieb pieper:
>> Dear Wien2k users,
>> 
>> I have a question about the Fermi-contact fields printed at the HFFnnn 
>> labels in the scf file.
>> 
>> Back in 2010 Peter Blaha advised in this mailing list that the 
>> Fermi-contact interaction can be underestimated at least for 3d-metals 
>> like Fe by 10 - 20%
>> [www.mail-archive.com/wien at zeus.theochem.tuwien.ac.at/msg02201.html] 
>> and Pavel Novak indicated that in his opinion it might be difficult to 
>> correct this misbehaviour by some suitable local or semi-local Vxc 
>> [www.mail-archive.com/wien at zeus.theochem.tuwien.ac.at/msg03011.html].
>> 
>> Since this was pre-NMR-package and at the beginning of hybrids I 
>> wonder wether the situation has improved in the meantime? The 
>> NMR-package as I understand it calculates shielding current 
>> distributions so I don't expect it improves on Fermi-contact fields - 
>> is that correct? And is there a way to get improved values from 
>> hybrids that I am unaware of?
>> 
>> Best regards,
>> 
>> Martin Pieper
>> 
>>