mirror of https://gitlab.com/QEF/q-e.git
35 lines
1.6 KiB
Plaintext
35 lines
1.6 KiB
Plaintext
This example shows how to compute the magnetic anisotropy energy (MAE) with the
|
|
"Force Theorem" method (Phys. Rev. B 90, 205409 (2014). The system is a 3-layer
|
|
Co slab and two magnetic configurations, parallel and perpendicular to the slab
|
|
surface, are considered.
|
|
|
|
The procedure is as follows:
|
|
1) run SCF calculation without SOC to get charge and spin moment
|
|
densities.
|
|
2) Copy them to two folders to be used for two SOC calculations
|
|
for different magnetic moment orientations - parallel and perpendicular
|
|
to the slab surface.
|
|
3) for each orientation run NSCF calculation with SOC, in the input
|
|
file one should specify in this case lforcet = .true. standing for the option
|
|
"Force Theorem".
|
|
|
|
In each case one will have already in output file the total band energy.
|
|
Their difference, eband_par - eband_perp will give the total Magnetic
|
|
anisotropy energy, MAE, in the Force theorem approximation.
|
|
|
|
To get its local decomposition over different atomic orbitals we should run
|
|
4) projwfc.x calculation for each orientation which will decompose
|
|
each band energy over atomic orbitals using PDOS. One will get two files,
|
|
eband_par.dat and eband_per.dat, containing this decomposition
|
|
for each magnetic orientation.
|
|
|
|
Taking the above difference orbital by orbital we obtain orbital-resolved MAE.
|
|
|
|
There are two new parameters to specify: lforcet and ef_0 (in projwfc.x run).
|
|
The last one should be the Fermi energy of one of NSCF calculations with SOC
|
|
and will be substracted from all the eigenvalues to produce correct
|
|
local decomposition of MAE.
|
|
|
|
Note that in order to get reliable results for the magnetic anisotropy,
|
|
you need a much denser grid of k-points than what used in this example.
|