Convergence of different transport coeffcients

AMSET
1

Dear developers,

I am testing convergence of transport coefficient with respect to interpolation factor. While the seebeck, and sigma are well converged, the convergence of mobility is little shaky. This is when I use full elastic tensor.

Surprisingly when i use component of elastic tensor (say xx) and calculate mobility convergence along that direction it convergenes very well.

Is this expected behavior?

Best,
Alex

convergence-test.pdf (23.8 KB)

2

Hi Alex,

This does seem a little unusual.

Iā€™d like to understand if this is an issue converging the underlying band structure or something to do with the acoustic deformation potential scattering in particular. Would you mind try running the calculations with just IMP scattering (i.e., set static_dielectric, and scattering_type = ["IMP"]) and plotting the transport properties again?

Best,
Alex

3

Hi Alex,

Thanks for the response. Please see the attached plot with only IMP scattering.

Best,
AKconvergence-imp.pdf (23.7 KB)

4

Ok, thank you. The doping concentration of IMP looks quite shaky too, but I think that is just because the rates are very sensitive. At the highest doping level, the transport properties are much smoother and the IMP scattering looks converged.

Would you be able to plot a band structure for your material using the amset plot band vasprun.xml --interpolation-factor 10 command. It would be good to check the band structure at multiple interpolation factors to see if the band edges change too much.

I will do a little more digging on my end to see if there is an issue with the ADP scattering or whether the systems just needs to go to larger mesh densities.

5

Hi Alex,

A quick update - the problem was in the interpolation. My scf grid was too coarse to give accurate interpolations. When I ramped up the scf grid the interpolation with higher interpolation-factor converges. and this also improves the convergence of mobility.

Many thanks for the help.

Best,
Alex kent

1 Like
6

Hi Alex,

Happy new year!

Thanks for letting me know. Glad to hear this is not an issue in amset itself.

Best,
Alex