|
|
|
###### Perform SOC calculations {#perform_soc_calculations}
|
|
|
|
|
|
|
|
To include spin-orbit coupling (SOC) into the system of study we take
|
|
|
|
the scalar-relativistic Hamiltonian and add SOC in the Pauli-form \$
|
|
|
|
\\mathcal{H}\_\\mathrm{SO} = \\xi \\, L \\cdot S\$. This is done by the
|
|
|
|
new solver. \_\_Note that the new solver works only for spin-polarized
|
|
|
|
full potential calculations.\_\_ The features of the old solver like for
|
|
|
|
example [DOS](jumu:density_of_states "wikilink"),
|
|
|
|
[QDOS](jumu:qdos "wikilink"), or [decimation](jumu:deci "wikilink") work
|
|
|
|
as usual. The following steps are needed:
|
|
|
|
|
|
|
|
##### Changes in the inputcard {#changes_in_the_inputcard}
|
|
|
|
|
|
|
|
The options //NEWSOSOL// and //fullBZ// have to be set in the inputcard.
|
|
|
|
This tells the program to use the new solver and to forget about all
|
|
|
|
symmetries previously found for the current setup. This is needed since
|
|
|
|
SOC may break some symmetries. In addition some parameters for the
|
|
|
|
newsolver have to be set. These are //NPAN\_LOG//, //NPAN\_EQ//,
|
|
|
|
//NCHEB// and //R\_LOG//. Also do not forget to do spin-polarized
|
|
|
|
calculations for full potential (//NSPIN=2, KSHAPE=2//). If you started
|
|
|
|
with a calculation without spin-polarization, i.e. *NSPIN=1*, you can
|
|
|
|
[adopt your already converged
|
|
|
|
potential](change_nspin_1_to_2 "wikilink"). <code> RUNOPT NEWSOSOL \...
|
|
|
|
\...
|
|
|
|
+\-\-\-\-\-\--+\-\-\-\-\-\--+\-\-\-\-\-\--+\-\-\-\-\-\--+\-\-\-\-\-\--+
|
|
|
|
|
|
|
|
` ***test options*** (2 lines) `
|
|
|
|
|
|
|
|
TESTOPT fullBZ
|
|
|
|
+\-\-\-\-\-\--+\-\-\-\-\-\--+\-\-\-\-\-\--+\-\-\-\-\-\--+\-\-\-\-\-\--+
|
|
|
|
|
|
|
|
\...
|
|
|
|
|
|
|
|
NSPIN=2
|
|
|
|
|
|
|
|
\...
|
|
|
|
|
|
|
|
KSHAPE=2
|
|
|
|
|
|
|
|
\...
|
|
|
|
|
|
|
|
HFIELD= 0.0 VCONST= 0.0d0 NPAN\_LOG= 15 NPAN\_EQ= 5 NCHEB= 12 R\_LOG=
|
|
|
|
1.0d0
|
|
|
|
|
|
|
|
BZDIVIDE= 30 30 30
|
|
|
|
|
|
|
|
\... </code> If the new solver without SOC should be used the test
|
|
|
|
option //NOSOC// has to be added.
|
|
|
|
|
|
|
|
##### additional input files {#additional_input_files}
|
|
|
|
|
|
|
|
Since the new solver does also deal with non-collinear calculations we
|
|
|
|
have to specify the non-collinear angles of each atom in the inputcard
|
|
|
|
in the file [nonco\_angle.dat](jumu:nonco_angle.dat_file "wikilink"). If
|
|
|
|
this file is not given then the angles are fixed to theta=phi=0 (i.e.
|
|
|
|
the z-direction).
|
|
|
|
|
|
|
|
##### Examples
|
|
|
|
|
|
|
|
#### Example 1: bulk Au with and without SOC {#example_1_bulk_au_with_and_without_soc}
|
|
|
|
|
|
|
|
As an example the difference between bulk Au with and without SOC should
|
|
|
|
be calculated. The test system consists of bulk Au (fcc lattice) with a
|
|
|
|
LMAX cutoff set to 2 and a k-mesh of 15x15x15, for the sake of speed of
|
|
|
|
the calculation.
|
|
|
|
`{{:jumu:examples:au_bulk_nosoc_cos.tar.gz|Here}}`{=mediawiki} the input
|
|
|
|
files of both cases with the different inputcards can be found. Using
|
|
|
|
the [QDOS](jumu:qdos "wikilink") the change in the band structure can be
|
|
|
|
investigated.
|
|
|
|
|
|
|
|
#### Example 2: bulk Au with and without SOC {#example_2_bulk_au_with_and_without_soc}
|
|
|
|
|
|
|
|
The second example is the SOC-induced splitting of Au(111) surface
|
|
|
|
states using a half-infinite Au(111) crystal (using the
|
|
|
|
[decimation-technique](jumu:deci "wikilink")) and the
|
|
|
|
[QDOS](jumu:qdos "wikilink") option. \<WRAP center round todo 60%\> Add
|
|
|
|
example files with instructions, also to [examples
|
|
|
|
page](jumu:examples_and_tips "wikilink"). </WRAP> |