Hi!,
How is the voronoi tesselation performed?
I have a perfect (unrelaxed) bcc structure with 2 different kinds of atoms and I get different Voronoi volume values for them.
Thanks, and all the best
Voronoi partition
Moderator: moderators
Re: Voronoi partition
That is surprising - the voronoi calculation trivially associates every point with the nearest atom. If the atoms are different, the integrated charges will be different but the volumes should be the same. If they are off by a small amount, that could be due to the discretization of the CHGCAR file. You can check this to see if the difference becomes smaller with a finer grid.
-
- Posts: 5
- Joined: Tue Sep 15, 2020 11:20 am
Re: Voronoi partition
Thanks! I've continued with the testing. Indeed is due to the grid.
It seems a fine grid is needed to get converged volume values.
When I used a coarse grid I got the two different kind of atoms with distinctively different volumes, and all of the atoms of the same kind with the same volume. Which led me think of some kind of radical Voronoi tesselation. Increasing the grid leads to closer volume values of the two atom types. Using a fine grid I got convergence up to one decimal place.
It seems a fine grid is needed to get converged volume values.
When I used a coarse grid I got the two different kind of atoms with distinctively different volumes, and all of the atoms of the same kind with the same volume. Which led me think of some kind of radical Voronoi tesselation. Increasing the grid leads to closer volume values of the two atom types. Using a fine grid I got convergence up to one decimal place.