[Wien] spin and orbital moment of vanadium

[Hua Wu]

Dear Zhijian,

  Your current solution may not be wrong but a metastable solution.
Since the SOC splitting is about 20 meV for vanadates, LDA+SOC+U
calculations with U of several eVs can easily run into different metastable 
SOC-splitting states. My suggestion is (1) calculate the crystal field level
splitting and then think about the likely SOC ground state. (2) play with 
the density matrix to set the likely ground state and excited states. (3)
compare the LDA+U+SOC total energy for the stable solutions in (2).

In this respect, my recent publication "Spin and orbital states in 
La1.5Sr0.5CoO4 studied by electronic structure calculations" 
http://link.aps.org/doi/10.1103/PhysRevB.80.081105
may be of your interest.

best wishes -- H. Wu 

 
On Wednesday 04 November 2009 01:33, Zhijian WU wrote:
> Dear Novak, Wu hua,and wein3k users,
>
> Thanks a lot for your information. Concerning the valence state of V, we
> just tentatively assigned it as 2+, after careful checking, we found that
> it closes to 3+ (d^2).
>
> The procedure of our calculations is done as Pavel suggested. We first run
> GGA, then GGA+SOC. After that we run GGA+SOC+U. From GGA+SOC calculation,
> the orbital moment of V is very small, close to zero (about 0.0x). But even
> this small orbital moment still paralles with spin moment.After GGA+SOC+U
> calculation, initially, the orbital moment increases slightly with the
> increase of U, then after a certain U value, it has a sudden dramatic
> increase (the dramatic sudden increase has an explanation in a recent paper
> on MnV2O4 (102, 216405,2009, Phy. Rev. Lett.). During this process, the
> spin and orbital moments remain parallel. This is indeed weired because we
> have made many calculations on d orbitals of transition metals (including
> 3d, 4d and 5d transition metals) and such a strange thing never happened
> before.
>
> Any more suggestions?
>
> Thank you very much!!
>
> Regards,
> Zhijian
>
> ======= 2009-11-03 07:37:00 you wrote：=======
>
> >Dear Zhijian WU,
> >
> >I recommend first to run the calculation without U (then L and S should
> >be antiparallel) and only after it converged switch on the correlation.
> >The reason is that LDA+U stabilizes the occupied states. If you switch it
> >on from the beginning there is a danger that incorrect states will be
> >stabilized.
> >
> >Regards
> >Pavel
> >
> >On Tue, 3 Nov 2009, Zhijian WU wrote:
> >> Dear Wien2k users,
> >>
> >> Recently, we calculated (spin-polarised calculations plus spin-orbit
> >> coupling and electron correlation U on vanadium) the spin and orbital
> >> moments of vanadium in a transition metal oxides. The valence state of
> >> vanadium can be assigned as +2, thus leaving the electronic
> >> configuration d^3 for vanadium. From our calculations, it is suprisingly
> >> found that the orbital moment paralells with spin moment, which violates
> >> the Hund's rule (which states that they should be antiparallel for less
> >> than half filled d orbitals). We doubt that something is wrong during
> >> our calculations, but can not figure out where or why.
> >>
> >> Any hint or clue?
> >>
> >> Thanks a lot for your inputs.
> >>
> >> Regards,
> >>
> >> Zhijian
> >>
> >> _______________________________________________
> >> Wien mailing list
> >> Wien at zeus.theochem.tuwien.ac.at
> >> http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien
> >
> >--
> >_______________________________________________
> >Wien mailing list
> >Wien at zeus.theochem.tuwien.ac.at
> >http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien
>
> = = = = = = = = = = = = = = = = = = = =
> _______________________________________________
> Wien mailing list
> Wien at zeus.theochem.tuwien.ac.at
> http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien