Page 1 of 1

SOC calculation for non-magnetic material

Posted: Tue Nov 13, 2018 1:48 pm
by liuhuimeiwoo
Dear VASP users,
I am trying to calculate a material Ca2N, which is non-magnetic, I think.
However, where I take spin-orbit coupling (SOC) into consideration by setting LSORBIT=.TRUE., the OUTCAR file shows the results have a very small magnetization like:

number of electron 9.0000009 magnetization 0.0009660 0.0013043 0.0013053

Though it is very small, it may break the time reversal symmetry.
We know that at Gamma point in non-magnetic materials, the eigenvalues should be doubly degenerate.
However, the EIGENVAL file shows that the eigenvalues of Gamma point like this:

7 -3.864897
8 -3.863837
9 -3.227240
10 -3.225686
11 -2.158353
12 -2.155542
which has a small deviation bwtween (7 and 8); (9 and 10); (11 and 12).

Is there a standard way to avoid the above problems?

PS,
I have set the PREC=accurate but the above problems still exist.
I know by setting NUPDOWN=0, one can make the total magnetization to be zero, but it does not force all the magnetic moments to be zero.
And for some other materials, we do not know it is non-magnetic or not before calculation, thus we cannot use the setting NUPDOWN=0.

Re: SOC calculation for non-magnetic material

Posted: Thu Sep 14, 2023 11:23 pm
by hongyang_ma
Hi,

I was wondering have you figured out a way to handle this issue?

Regards,
Hongyang

Re: SOC calculation for non-magnetic material

Posted: Fri Sep 15, 2023 7:40 am
by merzuk.kaltak
Dear hongyang_ma,

without proper input and OUTCAR files, it is impossible for us to figure out what is going on.
If you are interested, you may post INCAR, POSCAR, KPOINTS, POTCAR, OUTCAR and stdout files so that we can help you.

Re: SOC calculation for non-magnetic material

Posted: Sun Sep 17, 2023 5:44 am
by hongyang_ma
Hi,

My question is similar to what he mentioned: When using NUPDOWN and ISPIN = 2, we can fix the number of electronic spins. However, when we consider SOC (Spin-Orbit Coupling), is it possible to fix each of the x, y, and z magnetic components to specific, separate values (e.g., 1, 1, 0 for x, y, and z, specifically)?

Regards,
Hongyang

Re: SOC calculation for non-magnetic material

Posted: Mon Sep 18, 2023 6:46 am
by merzuk.kaltak
Dear Hongyang,

in non-collinear calculations (vasp_ncl) you can set the quantization axis and initial magnetic moment in an arbitrary direction using SAXIS in combination with MAGMOM.
Note, in this case MAGMOM contains 3 times more entries than in collinear calculations.