Unrealistically high ADP scattering

I’m calculating scattering in a family of oxide materials, and I’m consistently getting very high ADP scattering leading to calculated mobilities ~8 orders of magnitude lower than experiment. Does anyone know of some basic “sanity checks” I can do with my deformation potentials and elastic constants to see if there is a problem with those calculations? Or alternately, if there are any classes of materials where ADP scattering should not be considered?
My deformation tensors are
[[ 0.67 2.65 0.03]
[ 2.65 0.67 0.03]
[ 0.03 0.03 1.19]]
for VBM and
[[ 1.92 0.05 0.05]
[ 0.05 1.92 0.05]
[ 0.05 0.05 1.92]]
for CBM, while my elastic 6x6 matrix (including ionic relaxation contribution) in GPa is
[193.95973, 136.1056, 136.19056, -0.0000, -0.0000, 0.0000],
[136.19056, 193.95973, 136.19056, -0.0000, 0.0000, 0.0000],
[136.19056, 136.19056, 193.95973, -0.0000, -0.0000, 0.0000],
[-0.0000, -0.0000, -0.0000, 83.82644, 0.0000, 0.0000],
[-0.0000, 0.0000, -0.0000, 0.0000, 83.82644, 0.0000],
[0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 83.82644]



1 Like

I never figured out a solution to this problem, other than turning off ADP scattering to give somewhat reasonable results.

Hi @Steven_Hartman.

Sorry for the delay getting back to you. This is quite odd. I can only assume that it is due to extremely large deformation potentials away from the CBM/VBM. One way to check would be to use the alternative definition for the deformation potentials and just specify two values (e.g., as done in this example: amset/settings.yaml at 714bfd5039378718e6283aeb4c3bc106e3f5b9fa · hackingmaterials/amset · GitHub)

If that reduces the deformation potential scattering then it is definitely an issue with the deformation.h5 file. In that case, would you be willing to share your deformation files (vaspruns for bulk and deformations) and I can do some tests.


Thanks for the response.
After some further investigation I think I just found my mistake; I must have taken the uniform band structure calculation as my undeformed calculation. This calculation had no ICORELEVEL tag, which led to extremely high deformation potentials. But then when I went to make the original forum post, I presumably used the correct undeformed calculation to print out the deformation potentials with “amset deform read”, but didn’t realize I was doing it differently? At least that’s my best guess, I’m getting decent results now for at least one system.