Orbital-resolved magnetic anisotropy energies
Posted: Tue Mar 13, 2018 6:39 pm
Dear VASP users and developers,
I would like to calculate properly some interface and orbital/site-resolved magnetic anisotropy energies (MAE) for a slab geometry, as performed for example in the following papers:
Sci. Rep. 5 18173 (2015) or Phys. Rev. B 88 184423 (2013).
1) In the OUTCAR, I guess that the E_soc values and the following matrices correspond to the site and orbital contributions to the spin-orbit interaction.
How these values are exactly calculated? From the band energies and their contributions?
Is there any projection in the spheres of radius RWIGS? Is this for this reason that by calculating the total MAE by summing the different orbital-resolved MAE I find a value (well) larger than by doing the difference of total energies for two different directions of the magnetization?
If yes, which value seems the most appropriate to get the interface MAE?
2) Am I right by considering that the orbitals in the matrix after E_soc are listed following the same order than in the PROCAR, i.e. py, pz, px and dxy, dyz, dz2, dxz, dx2-y2?
3) I am not really familiar with the procedure used by VASP for NSCF calculations. Maybe could you suggest me a proper reference about it.
Which terms enter in the calculations of the total energy?
In the PRB 93, 224425 (2016), it is written that the MAE is calculated using the force theorem by doing the difference between the total energies. Is it then equivalent to sum the differences of band energies for occupied states (PRB 41 11919 (1990))?
4) In addition to the number of k-points, is there other crucial convergence parameters to take into account?
I thank you in advance for any help you could give me.
Best regards,
R. Arras
I would like to calculate properly some interface and orbital/site-resolved magnetic anisotropy energies (MAE) for a slab geometry, as performed for example in the following papers:
Sci. Rep. 5 18173 (2015) or Phys. Rev. B 88 184423 (2013).
1) In the OUTCAR, I guess that the E_soc values and the following matrices correspond to the site and orbital contributions to the spin-orbit interaction.
How these values are exactly calculated? From the band energies and their contributions?
Is there any projection in the spheres of radius RWIGS? Is this for this reason that by calculating the total MAE by summing the different orbital-resolved MAE I find a value (well) larger than by doing the difference of total energies for two different directions of the magnetization?
If yes, which value seems the most appropriate to get the interface MAE?
2) Am I right by considering that the orbitals in the matrix after E_soc are listed following the same order than in the PROCAR, i.e. py, pz, px and dxy, dyz, dz2, dxz, dx2-y2?
3) I am not really familiar with the procedure used by VASP for NSCF calculations. Maybe could you suggest me a proper reference about it.
Which terms enter in the calculations of the total energy?
In the PRB 93, 224425 (2016), it is written that the MAE is calculated using the force theorem by doing the difference between the total energies. Is it then equivalent to sum the differences of band energies for occupied states (PRB 41 11919 (1990))?
4) In addition to the number of k-points, is there other crucial convergence parameters to take into account?
I thank you in advance for any help you could give me.
Best regards,
R. Arras