Regarding DOS calculation

[chandan_kumarvishwakarma1]

Dear users
I am trying to reproduce the DOS and PDOS plots of
La3Se4 as reported in the following
paper,
LInk of the paper: https://pubs.acs.org/doi/10.1021/acs.chemmater.7b04767
In my calculation, I am not getting the electronic contribution
because of the virtual orbital of La i.e. f orbital of La which is present in
the reported paper. I plotted the PDOS and DOS using p4v.
Here I am attaching the Inputs (POSCAR, KPOINTS, INCAR, POTCAR) and output
(DOSCAR and OUTCAR) in a .zip file. I am also attaching the jpg file of the plot
(screenshot of p4v).
How can I get the contribution of the La f-orbital?
Please help me out!

---
Chandan

[henrique_miranda]

You did not attach the POSCAR file.
Furthermore, since you are using p4v I assume you are reading the vasprun.xml file. Could you attach it as well?
Did you modify the POTCAR files?

You have both ISPIN = 2 and LNONCOLLINEAR = .TRUE. set in the input file.
If you are performing a noncollinear calculation ISPIN is set to 1 (you can check in the OUTCAR file).

[chandan_kumarvishwakarma1]

Dear henrique_miranda
I am attaching the POSCAR file for the La3Se4 structure, apart from this
I am unable to attach the vasprun.xml file as the file size is huge.
I have also implemented your suggestion by taking ISPIN = 1, but it doesn't
effect much.
Actually my doubt is regarding the DOS contribution of La-f orbital, as
you can see the total DOS contribution is much higher than the PDOS contribution.
I have also checked for the contribution of all the orbitals.

Regarding this please help me out!

---Chandan

[chengcheng_xiao1]

VASP has changed it's orbital notation since version 5.4.4, specifically:

Code: Select all

v5.4.4        v5.4.1
x2-y2    <-->   dx2
fy3x2    <-->    f1
fxyz     <-->    f2
fyz2     <-->    f3
fz3      <-->    f4
fxz2     <-->    f5
fzx2     <-->    f6
fx3      <-->    f7

p4vasp still only recognizes the old notation. I've submitted a Github that fixes this issue. Give it a try if you like.

The compilation process of p4vasp is at its current stage - janky. IMHO I think it needs a full rehaul for some dependencies are now deemed deprecated. If you cannot compile it but still really want to use p4vasp, maybe try to replace these orbital notations in `vasprun.xml` yourself or give my python script a try.

Alternatively, if p4vasp is too much of a hassle, try pyband!

[chengcheng_xiao1]

Also, regarding your question about the total DOS being is much higher than the PDOS contribution, keep in mind that a projecting on to a set of atomic-centered orbitals does not necessarily mean these orbitals form a complete basis set. Similar explanation can be found here.

The projection routines are in `sphpro.F`. Basically, if you use LORBIT<10, VASP will construct a projector using spherical harmonics and atomic-centered Bessel functions with a cutoff of RWIGS and then do the projections (If I'm not mistaken). In this method, RWIGS should be carefully chosen.

If you use LORBIT >10, the projections are calculated directly using the PAW projectors (with a "normalize factor" of PP%QTOT(LP,L)).