Direction of magnetic moment (SAXIS, LNONCOLLINEAR)

[kyoyu_kondo]

Dear All,

I have some questions about direction of magnetic moment.

1. If we use the default SAXIS and set ISPIN = 2 and LNONCOLLINEAR = FALSE, is the direction of the magnetic moment of the atom in the z-axis ?

2. Is the result from "ISPIN = 2, LNONCOLLINEAR =FALSE, MAGMOM = 2*3 2*-3" exactly the same as the result from "ISPIN = 2, LNONCOLLINEAR = TRUE, MAGMOM = 0 0 3 0 0 3 0 0 -3 0 0 -3" (SAXIS is default) ?

3. I would like to know the difference between the result from "ISPIN = 2, SAXIS = 1 0 0, LNONCOLLINEAR = FALSE, MAGMOM = 2*3 2*-3" and the result from "ISPIN = 2, LNONCOLLINEAR = TRUE, MAGMOM = 3 0 0 3 0 0 -3 0 0 -3 0 0 (SAXIS is default)".

4. Finally, If the magnetic moment is in the <010> direction and the state is antiferromagnetic, do I need to set up a non-collinear calculation or SAXIS?
The crystal structure is a tetragonal crystal with the directions of a-axis and x-axis, b-axis and y-axis, and c-axis and z-axis coinciding.
(Do I need to change the SAXIS and LNONCOLLINEAR settings from the defaults if I target antiferromagnetic?)

Thanks in advance.

[martin.schlipf]

1. If we use the default SAXIS and set ISPIN = 2 and LNONCOLLINEAR = FALSE, is the direction of the magnetic moment of the atom in the z-axis ?

Without noncollinear magnetism the spin orientation is arbitrary, there is no magnetic anisotropy.

2. Is the result from "ISPIN = 2, LNONCOLLINEAR =FALSE, MAGMOM = 2*3 2*-3" exactly the same as the result from "ISPIN = 2, LNONCOLLINEAR = TRUE, MAGMOM = 0 0 3 0 0 3 0 0 -3 0 0 -3" (SAXIS is default) ?

No, in the case of noncollinear magnetism you get magnetic anisotropy.

3. I would like to know the difference between the result from "ISPIN = 2, SAXIS = 1 0 0, LNONCOLLINEAR = FALSE, MAGMOM = 2*3 2*-3" and the result from "ISPIN = 2, LNONCOLLINEAR = TRUE, MAGMOM = 3 0 0 3 0 0 -3 0 0 -3 0 0 (SAXIS is default)".

SAXIS is only used in combination with LNONCOLLINEAR.

4. Finally, If the magnetic moment is in the <010> direction and the state is antiferromagnetic, do I need to set up a non-collinear calculation or SAXIS?

If the direction of the magnetic moment matters, you need to set LNONCOLLINEAR.

Note that there is also the LSORBIT flag that you want to set for these kind of calculations. We have a brief tutorial here: wiki/index.php/Spin-orbit_coupling_in_a_Ni_monolayer

[kyoyu_kondo]

Thank you for your comments.
The answers are very helpful and I have additional questions.

1. Setting up antiferromagnetic in the collinear calculation means we specify only the direction of the spin is opposite, not the direction of orientation?

2. If spin-orbit coupling is not taken into account ( LSORBIT = .FALSE.), is it not necessary to set SAXIS?

3. If I use the default SAXIS and set MAGMOM = p q r, does this atom have a magnetic moment of p in the x direction, q in the y direction, and r in the z direction?

4. What are the materials for which the direction of the magnetic moment is important?

Best Regards,

[martin.schlipf]

1. Setting up antiferromagnetic in the collinear calculation means we specify only the direction of the spin is opposite, not the direction of orientation?

Yes.

2. If spin-orbit coupling is not taken into account ( LSORBIT = .FALSE.), is it not necessary to set SAXIS?

Yes. Even if you use SOC, you do not need to set it, it will default to the z axis.

3. If I use the default SAXIS and set MAGMOM = p q r, does this atom have a magnetic moment of p in the x direction, q in the y direction, and r in the z direction?

Yes. Note this is only the initial moment. The moment can realign during the scf cycle.

4. What are the materials for which the direction of the magnetic moment is important?

Mostly if you are interested in the magnetic topology (e.g. spin spirals) or magnetic anisotropy.