Problems with vacanies and NEB when calculating low symmetry crystals

[postmons]

Hi,

I am calculating the oxygen migration in a usually highly symmetric oxide crystal. I use GGA+U and the NEB method to calculate the migration energy. In its high symmetry there are six nearest oxygen neighbors and all migration paths are equal. If I now change the symmetry of my cell, those migration paths should still be equivalent. But I get different migration barriers anyway and I am pretty sure, it is not a real NEB problem, but my start and end postions - meaning removal of oxygen atoms on different lattice sides - are not equal.

So far I tried different approaches:
-include/remove Gamma-point in k-Point mesh
-increase/decrease k-point mesh density
-make a more or less symmetric supercell
-use bigger supercell
-switch symmetry on/off for calculation using the ISYM tag
-raise/lower convergence limit
-use GGA instead of GGA+U
-relax the ideal supercell to a higher convergence level

but none of it worked. Any ideas? I run out of ideas.

[alex]

Hi,

maybe the release of symmetry constraints allows a different occupation of bands?

Cheers,

alex

[postmons]

Hi alex,

thank you for your answer. I tried the release of symmetry through the ISIF and the ISYM tag. Unfortunatly one of it worked.
Do you have anymore ideas?