[Wien] Magnetocrystalline anisotropy

[Jaroslav Hamrle]

Dear Xavier,

your problem somewhat resembles me my problem I had when calculating 
magnetic linear dochroism (MLD) on bcc Fe. The similarity is that we 
both want to see small changes in electronic structure when rotating 
magnetic field direction.

What help me:
1) run fine convergence criteria, such as runsp_lapw -p -cc 0.000001 -ec 
0.000001


2) as suggested Prof. Blaha, it was important to increase k-mesh (in my 
case up to 100), and apply fine BZ integration (TEMP or TEMPS) with 
small value as 0.001, not default TETRA.
for example change case.in2 by using command
sed '3s/^................/TEMP    0.001   /' $file.in2 > $file.in2_TEMPnew
more here"
https://www.mail-archive.com/wien@zeus.theochem.tuwien.ac.at/msg16815.html
https://www.mail-archive.com/wien@zeus.theochem.tuwien.ac.at/msg16844.html


3) for some magnetization direction, I had problem with either wrong, 
either suspicious values of local rotation matrix,
https://www.mail-archive.com/wien@zeus.theochem.tuwien.ac.at/msg16894.html

Problem was that for some external magnetization directions, the local 
direction of magnetization was not [001].
In one case (external M along [-111]), the local magnetization direction 
was [-0.94281 0   -0.33333] which I think is not correct, and MLD was 
wrong too.
In some case, local magnetization direction was along x or along y, 
which I dont know if it is correct. On one hand, eigenenergies agreed 
perfectly, but anyway I saw small change in MLD in those cases.

But as a blind suggestion for you, try if local rotation matrices are 
correct. Namely try if
mag_glob*R = mag_loc
where mag_glob is your (external i.e. in global coordinates) 
magnetization direction, R is local rotation matrix for each atom (can 
be found in case.struct or case.outsymso) and mag_loc is local 
magnetization direction, which in my (maybe naive and wrong) 
understanding should be [001].

Hoping it helps
With my best regards

Jaroslav


On 09/01/18 09:44, Xavier Rocquefelte wrote:
> Dear Colleagues
>
> I recently obtained a surprising result concerning the calculation of 
> the magnetocrystalline anisotropy energy (MAE) of SeCuO3.
>
> This compound has a monoclinic symmetry (SG. P21/n) and is known to be 
> antiferromagnetically ordered at low temperature.
>
> Here I provide the results obtained for two magnetic orders, named FM 
> and AFM1 (see attached document) :
>
> https://filesender.renater.fr/?s=download&token=1da93a22-9592-3a7e-ba2e-1533fcae45d2 
>
>
> These calculations have been done using WIEN2k_17, GGA = PBE, RKMAX = 
> 6, kmesh = 5 4 4 and in P1 symmetry. The results are the same using 
> RKMAX = 7.
>
> The AFM1 order is the more stable one, as expected.
>
> However, as shown in the document the MAE of AFM1 order is not 
> symmetric, which is not expected. In contrast the MAE for FM order is 
> symmetric.
>
> Based on the recent discussion "zigzag potential", it seems to me that 
> the AFM1 MAE should be symmetric, because the magnetic moment is a 
> pseudo-vector. Is it possible that the present problem is related to 
> the fact that in the present implementation of the spin-orbit coupling 
> we neglect the off-diagonal terms? Do you have any idea about the 
> problem we are facing? Does someone observe such unusual MAE for other 
> systems?
>
> Best Regards
>
> Xavier
>
> _______________________________________________
> Wien mailing list
> Wien at zeus.theochem.tuwien.ac.at
> http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien
> SEARCH the MAILING-LIST at: 
> http://www.mail-archive.com/wien@zeus.theochem.tuwien.ac.at/index.html


-- 
------------------------------------------------------------------
Mgr. Jaroslav Hamrle, Ph.D.
Institute of Physics, room F232
Faculty of Mathematics and Physics
Charles University
Ke Karlovu 5
121 16 Prague
Czech Republic

tel: +420-95155 1340
email: hamrle at karlov.mff.cuni.cz
------------------------------------------------------------------