Centrosymmetric Crystal

[Tzou M.]

Dear TB2J group members,

I would like some help in explaining my results on Exchange Interactions. I am studying the centrosymmetric crystal L10 FePt for non-collinear magnetic ordering with spin-orbit coupling. 

The isotropic part of Jexch confirms the tendency of the system towards ferromagnetic ordering. What is interesting is the existence of strong DMI interactions in a system (composition-uniform, no crystal inversion symmetry breaking) that should not exist. How can I explain the existence of these interactions? What also strikes me is the behavior of Jyy of the Jani tensor. Since the structure consists of successive Fe and Pt atomic planes along the c-axis, I would expect the Jyy to behave similarly to the Jxx component. Is it possible that a numerical error has been made when calculating or merging the files?

The command I use to run wann2j: wann2J.py --np 10 --elements Fe Pt --spinor --prefix_spinor rel_L10_FePt_(mx or my or mz) --groupby orbital --efermi 19.9433 --kmesh 10 10 10

Best regards,

Michail

[Xu He]

Hi, 

It is likely that these are from the noise in the Wannier function Hamiltonian. It is quite tricky to get the Wannier function with good symmetry in non-collinear spin calculations with SOC. From the wout file, it seem the Wannier functions are localized but the symmetry isn't well preserved.  I would suggest trying the DFT codes with LCAO basis (SIESTA or OpenMX) so that Wannierization is not necessary, to see what happens if the symmetry is preserved. 

Best regards, 

HeXu

--
You received this message because you are subscribed to the Google Groups "TB2J" group.
To unsubscribe from this group and stop receiving emails from it, send an email to tb2j+uns...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/tb2j/45a2e04c-4cca-4f68-a3ca-4411ceb29499n%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[Tzou M.]

What exactly are you looking for in the Wannier output files and do you see that the symmetry is not well preserved?  Can you show me so I can understand  better?

In fact, I used the flag nosym=true (QE) when I ran the scf and nscf calculations, to avoid reducing the k-points.

I am aware of a module "open_grid.x" which allows the scf and nscf to be run with the flag nosym=false and letter writes the Kohn-Sham orbitals for the full k-point grid. Do you think this would help? 

[Xu He]

Hi, 

I was simply looking at the Wannier spread of some orbitals and see if they are close. For example, if a dxy and dyz should be symmetric, the spread shoud also be the same. 

Using symmetric wavefunction probably helps but I think the most difficult part is from the disentanglement of the band in the Wannierization procedure. It would be interesting to see the result if you can try with different approaches to reduce the noise. 

Best regards, 

HeXu

[Tzou M.]

Dear HeXu,
Thank you for your reply. I understand what you are referring to. I will definitely try to reduce the noise and let you and the group know.
Best regards,
Michail