Spin with Siesta

From version 2.31 of c2x, the Siesta support is sufficient for the following example.


SystemLabel FeO
SaveRho  .true.

LatticeConstant  1.0 Ang
%block LatticeVectors
0.000000 4.332000 4.332000
2.166000 0.000000 2.166000
2.166000 2.166000 0.000000
%endblock LatticeVectors

NumberOfAtoms  4
NumberOfSpecies  2
%block ChemicalSpeciesLabel
1  8   O
2  26   Fe
%endblock ChemicalSpeciesLabel

AtomicCoordinatesFormat Fractional
%block AtomicCoordinatesAndAtomicSpecies
 0.250000 0.500000 0.500000 1
 0.750000 0.500000 0.500000 1
 0.000000 0.000000 0.000000 2
 0.500000 0.000000 0.000000 2
%endblock AtomicCoordinatesAndAtomicSpecies

SpinPolarized .true.
%block DMInitSpin
  3  2.000000
  4  -2.000000
%endblock DMInitSpin

MeshCutoff 200 Ry

%block kgrid_Monkhorst_Pack
2 0 0 0.500000
0 3 0 0.000000
0 0 3 0.000000
%endblock kgrid_Monkhorst_Pack

One can then needs to supply suitable pseudopotentials. Both Fe.psf and O.psf can be found amongst the examples distributed with Siesta. With these in the current directory along with the above .fdf file, one can run Siesta as normal:

$ siesta < FeO.fdf > FeO.out


Now one can analyse and visualise the results. In this case we choose to expand the system back to a cubic cell. Note that Siesta's .RHO file does not contain atomic positions, so c2x also reads the corresponding .XV file.

$ c2x -csv -X='(8.664,0,0)(0,8.664,0)(0,0,8.664)' FeO.RHO FeO.xsf
Grid size 48x24x24 spins=2
Recorded data are single precision
Attempting to read FeO.XV 
Also reading FeO.XV
Cell volume 40.647690
natoms      4
Spin components (density): 2
Spin components (bands):   1
First FFT grid     48 24 24
spins=2   spinors=1
Found 3D data for Density
  min=0.0020457 max=6.86912 sum=19045.3 int=28.0001
  (integral is e per cell for charge and spin densities)
Found 3D data for Spin
  min=-3.94505 max=3.94505 sum=6.4837e-14 int=9.53225e-17 int|s|=7.06093
  (integral is e per cell for charge and spin densities)
New cell volume 650.362259 (16 times old)
New FFT grid is 72 72 72
New FFT grid is 72 72 72

The integrated charge of 28 is consistent with the pseudopotentials used (+6 for O, +8 for Fe, two of each). The spin data is almost perfectly antiferromagnetic, and the value of int|s| implies a magnetic moment per Fe atom of 3.53μB, which is consistent with the literature and the other calulations here.

As the resulting picture is indistinguishable from the one produced by QE, that image is repeated below:

FeO and spin density