Problem about the phonon dispersion and PDOS calculation of defective materials

GULP
1

Dear Sir/Madam:

I am trying to calculate the phonon dispersion and phonon density of states of the 2D material WSe2 with vacancy. However, as the input file show, it seems GULP calculate the perfect crystal before the defects are formed and I can not get the phonon dispersion with defect structure. I also try to input the defect configuration with many atoms as cartesian coordination to control the defect percentage, but the picture I got is overlapping and I can not get useful information.

So I want to ask how can I get the PDOS of materials with 1% defects, how to achieve this in GULP. Thank you very much!

Bowen

defect opti conp phonon eigenvectors groupvelocity grueneisen
vectors
3.327070 0.000000 0.000000
-1.663535 2.881327 0.000000
0.000000 0.000000 15.068951
cartesian
W -1.1644760554969078 -0.6722749701720941 5.274132850000007
Se1 0.4990653964547016 -1.6327349126065653 6.969166796030753
Se2 0.4990653964546882 -1.632734912606565 3.57909890396924

centre W
size 6 10
vacancy W

temperature
300.0
species
W
Se

sw2
Se Se 1.265 0.046 60.4319 0 4.05735
W Se 35.783 2.852 32 0 3.49635
W W 6.7 2.019346 29.0197 0 4.63325
sw3
W Se Se 4.157 80.275 0.737 0.737 3.49635 3.49645 4.057735
Se W W 3.52 80.275 2.25 2.25 3.49635 3.49635 4.63325

shrink
3 3 1
dispersion
0.0 0.0 0.0 to 1 1 1

output phonon 3-atom

2

Dear Bowen,
I think the issue here is to appreciate what the defect method is doing. This uses the Mott-Littleton method to study defects at infinite dilution & uses a cluster embedded in a continuum that has the correct dielectric response to the defects. If you want to study defects at finite concentration this is not the way to go and you need to use a supercell where you construct defects at whatever concentration you like. Now to the point about phonon dispersion. The Mott-Littleton method doesn’t have any periodic boundary conditions (as it’s an embedded cluster) and so if there is no PBC then you have no reciprocal space and so phonon dispersion doesn’t exist. Again you need to use a supercell method to get this. Of course defects are rarely perfectly periodic and so phonon dispersion of a defect (which is localised in real space) doesn’t necessarily make much sense for most materials in terms of reality.
Hope that helps,
Julian