High Ueff for semiconductor surfaces shows bad convergence
Posted: Tue Aug 22, 2017 9:32 am
I am using a DFT+U scheme for modelling a semiconductor surface, with a high Ueff = 13 eV. Such a high U value is obtained by fitting to bulk experimental band gap of the semiconductor.
My bulk calculations show no problems with such high Ueff in regards to energy convergence. However, when I use this high Ueff (=13 eV) for the semiconductor surface unit cell, the system undergoes arbitrary electronic and ionic jumps. In other words, the ionic and electronic convergence is badly affected due to high Ueff values for a surface, whereas the bulk unit cells do not show any such problem. The same surface calculations behave quite well for Ueff=11 eV.
Has anyone come across any such issue? Is there any way to tune the convergence for such surfaces?
Thanks,
Somil
My bulk calculations show no problems with such high Ueff in regards to energy convergence. However, when I use this high Ueff (=13 eV) for the semiconductor surface unit cell, the system undergoes arbitrary electronic and ionic jumps. In other words, the ionic and electronic convergence is badly affected due to high Ueff values for a surface, whereas the bulk unit cells do not show any such problem. The same surface calculations behave quite well for Ueff=11 eV.
Has anyone come across any such issue? Is there any way to tune the convergence for such surfaces?
Thanks,
Somil