A bug with Rh?

Posted on : February 02, 2018 (Fri) 23:05:07

by Vitaliy

The problem appeared with half-Heusler phase ZrNiSn where Ni is substituted by Rh, the Fermi level should be shifted to the valence band, while it goes deeper into the conduction band. For higher concentrations (>=5%) it looks fine. It doesn't depend on the type of potential, nmag/mag, etc. Moreover if you change Zr for Hf - the picture will be the same. I have tried the same system in SPRKKR code and it is fine with low Rh concentrations - Fermi level in the valence band. Something wrong with Rh specie. Here is the inp file:
c---------------------------------
log data/ZrNiRhSn_003 fcc 11.570861678004535147392290249433,,,,,,
0.010 1.1 srals gga91 mag 2nd
update 2 3000 0.010
4
Ni 2 0 0 2 28 97.0
45 3.00
Zr 1 0 0 2 40 100
Sn 1 0 0 2 50 100
Vac 1 0 0 0 0 100
4
0.0 0.0 0.0 Ni
0.25 0.25 0.25 Sn
0.5 0.5 0.5 Vac
0.75 0.75 0.75 Zr
c---------------------------------
log data/ZrNiRhSn_003 fcc 11.570861678004535147392290249433,,,,,,
0.001 1.1 srals gga91 mag 2nd
update 9 3000 0.010
4
Ni 2 0 0 2 28 97.0
45 3.00
Zr 1 0 0 2 40 100
Sn 1 0 0 2 50 100
Vac 1 0 0 0 0 100
4
0.0 0.0 0.0 Ni
0.25 0.25 0.25 Sn
0.5 0.5 0.5 Vac
0.75 0.75 0.75 Zr
c---------------------------------
dos data/ZrNiRhSn_003 fcc 11.570861678004535147392290249433,,,,,,
0.001 1.1 srals gga91 mag 2nd
update 9 3000 0.010
4
Ni 2 0 0 2 28 97.0
45 3.00
Zr 1 0 0 2 40 100
Sn 1 0 0 2 50 100
Vac 1 0 0 0 0 100
4
0.0 0.0 0.0 Ni
0.25 0.25 0.25 Sn
0.5 0.5 0.5 Vac
0.75 0.75 0.75 Zr

[Re:04] A bug with Rh?

Posted on : February 04, 2018 (Sun) 23:27:59

by Vitaliy

Thank you for reply. Yes, always delete pot file before starting new calculation with changed settings and I'm a happy AkaiKKR user since 2006! But back to the case : I have found a sweet spot with energy window - 1.6 Ry to got a quick convergence across a batch of concentrations of Rh atoms. It also depends on the parameters used for specx compilation.
Hope it will help somebody.

[Re:03] A bug with Rh?

Posted on : February 03, 2018 (Sat) 07:36:22

by Administrator

Did you discard the potential file? I first scratched the potential data and recalculated. The results was quite reasonable.

KKR administrator

[Re:02] A bug with Rh?

Posted on : February 03, 2018 (Sat) 00:25:20

by Vitaliy

Yeah, that was the first thing I thought about, but even with ewidth 1.5 it didn't work. Ok I will try with higher value. Thank you for your time.

[Re:01] A bug with Rh?

Posted on : February 03, 2018 (Sat) 00:15:42

by Administrator

Please use ewidth=1.2 or even a larger value such as 2.0, and you will get proper results. The reason you have gotten a strange result is that during iteration, an ionized state of Rh was stabilized and stuck there. To recover the charge neutrality, the Fermi level was obliged to shift up.

KKR administrator