About the calculation of magnetic anistropies energy

[okuno]

Dear Vasp Users.

I want to get the magnetic anistropies energy and have the questions
on the method of calculation.

First we calculate collinear magnetic state with

MAGMON=1 -1 1 (ferri magnet and 3 atom for example)

and then I want to compare the total energy of system where
the magnetic moment is directed [001] and [100] .

At the second run, following the manual,
I set [001] directed magnetic system,

LSORBIT=.TRUE.
SAXIS=0 0 1
ICHRG=11
NBAND=2*number of NBAND of 1st run.
(and No(!) MAGMOM term)

for [100] case,

SAXIS=1 0 0
and other terms are same as [001] case.

I think 1st run the magnetic moment is directed [001] and
the 2nd run will gives the magnetic moment (1 -1 1) for 1, 2 , 3rd
atom, (of course the magnitude of magnetic moment will change, I
concern only the sign of magnetic moment)
but SAXIS=1 0 0 case, does Vasp yield the (1 -1 1)
magnetic moment which are parallel along [100] direction ?

And in the manual, the calculation of magnetic anistropies energy are difference of 2nd non-self consistent run, but I think the
true magnetic anistoropies energy is the difference of energies
self consistent run with spin-orbit interaction.

So is it more accurate that we compare the energy with
the runs with [001] directed case
MAGMOM= 0 0 1 0 0 -1 0 0 1 (3 * atom number)
SAXIS=0 0 1

and
MAGMOM = 0 0 1 0 0 -1 0 0 1 (3 * atom number) same as [001] case
SAXIS=1 0 0

?

[admin]

if you do self-consistent runs, (ie ICHARG is not 11), your system will most probably relax back into the magnetic ground state configuration (unless there is a significant barrier), therefore you might not obtain the total energy of the artificial magetic structure.
concerning MAGMOM:
please do give it, and just change SAXIS, as suggested in the manual:
in your case. please set
MAGMOM=0 0 1 0 0 -1 0 0 1
and SAXIS = 1 0 0 (for magnetization along x) and
SAXIS = 0 0 1 for magnetization along z

[okuno]

Dear Administrator.
Thank you for your reply.

Sorry, I have one question.

For the 1st run, the ground state is [001] magnetic direction, and second run restarting from the 1st run,
I think the initial state of magnetic direction is [001]
even if we set SAXIS= 1 0 0 .
How the magnetic direction rotates in Vasp ?
Or if we set MAGMON in the 2nd run,
automatically the magnetic direction is set to MAGMON and SAXIS direction by　priority?

And If we fixed the magnetic direction with I_CONSTRAINED_M and M_CONSTR, and run scf calculation with SAXIS=100, and SAXIS=001 and compare
the energy. Do we get more accurete MAE energy than
non-scf calculation ?

[admin]

the convention to give MAGMOM in VASP is that MAGMOM along z (i.e. 0 0 1) defines a moment lying parallel to the spin quantization axis, as given by SAXIS: MAGMOM defines a relative orientation with respect to SAXIS, not an absolute orientation in the coordinate system. Therefore it is sufficient to rotate the quantization axis and keep MAGMOM parallel to it (001).
If you use the option to constrain a moment to a certain direction, the energy contribution of the external field which has to be established to stabilize the magnetization in this (non-groundstate) direction is added to the total energies. The charge densities (Hamiltonian) are the fully updated under this constraining field (ie. ICHARG is not kept fixed), whereas the first approach is single run to show the energy difference between 2 magnetic configurations at zero external field.

[okuno]

Dear Administrator.
Thank you for reply.

Then, Vasp's OUTCAR file

magnetization 0.00 0.00 40.00 (in OUTCAR)
with
MAGMOM=0.0 0.0 1.0 0.0 0.0 -1.0 0.0 0.0 1.0
SAXIS = 0 0 1 (in INCAR)

and magnetization 0.00 0.00 40.00 (in OUTCAR)
with
MAGMOM=0.0 0.0 1.0 0.0 0.0 -1.0 0.0 0.0 1.0
SAXIS = 1 0 0 (in INCAR)

are different magnetization direction ?

[vasp16888]

[quote author=0.0 0.0 1.0 0.0 0.0 -1.0 0.0 0.0 1.0
SAXIS = 0 0 1 (in INCAR)

and magnetization 0.00 0.00 40.00 (in OUTCAR)
with
MAGMOM=0.0 0.0 1.0 0.0 0.0 -1.0 0.0 0.0 1.0
SAXIS = 1 0 0 (in INCAR)

are different magnetization direction ?



[/quote]</span>