Hello
I have question regarding the output of projwfc (the Lowdin charge
part). I've computed some case where iron is involved. For this I've
used pseudopotential (pp) file with semicore states in valence:
from UPF file:
Valence configuration:
nl pn l occ Rcut Rcut US E pseu
3S 1 0 2.00 1.100 1.300 -6.910119
4S 2 0 2.00 0.800 1.400 -0.388933
3P 2 1 6.00 1.000 1.300 -4.413015
4P 3 1 0.00 1.000 1.600 -0.097407
3D 3 2 6.00 1.400 2.000 -0.551558
Generation configuration:
3S 1 0 2.00 1.100 1.300 -6.910160
4S 2 0 2.00 0.800 1.400 -0.388938
3P 2 1 6.00 1.000 1.300 -4.413054
4P 3 1 0.00 1.000 1.600 -0.097411
3D 3 2 6.00 1.400 2.000 -0.551588
3D 3 2 0.00 1.400 2.000 -0.400000
so the valence configuration is 3s2 3p6 3d6 4s2 4p0 and the number of
valence electrons is 16.
The output of Lowdin charge is
Atom # Fe: total charge = 16.3479, s = 2.2797, p = 7.1296, d =
6.9386,
spin up = 8.2041, s = 1.1314,
spin up = 8.2041, p = 3.5414, pz= 1.1805, px=
1.1805, py= 1.1805,
spin up = 8.2041, d = 3.5313, dz2= 0.7002,
dxz= 0.7103, dyz= 0.7103, dx2-y2= 0.7002, dxy= 0.7103,
spin down = 8.1438, s = 1.1484,
spin down = 8.1438, p = 3.5882, pz= 1.1961, px=
1.1961, py= 1.1961,
spin down = 8.1438, d = 3.4073, dz2= 0.5199,
dxz= 0.7892, dyz= 0.7892, dx2-y2= 0.5199, dxy= 0.7892,
polarization = 0.0603, s = -0.0170, p = -0.0467, d
= 0.1240,
The questions are:
1. Why p shell is filled by 7 electrons? Is the output summing both p
shells found in pseudopotential file (so total charge is the sum of 3p
and 4p)?
2. Why therefore the s shell is not summed also? (3s2 + 4s2 should give
total s =4). Am I missing something?
3. Or in more general way: is the Lowdin output of projwfc giving sums
over each shell of the same type (if more present in PP file)?
The total charge for Fe in this case is also little bit strange (it
looks like iron gained electron in 4p despite the loss of 4s ones?)
Well, I know that Lowdin charge analysis is not too precise, but I'm
hoping to get some qualitative understanding of the above output.
Also, I believe that my calculation is well prepared and converged cause
results are repeatable.
Thanks in advance!
Konrad Gruszka
_______________________________________________
Pw_forum mailing list
[email protected]
http://pwscf.org/mailman/listinfo/pw_forum
