Why VASP gives no straight vacuum line for the surface local potential?

[pbsorokin]

Dear colleagues,

I'm calculating local potential (LOCPOT) data in VASP for the semiconductor/metal interface in order to extract work functions (WF), VBM, CBM, etc. It's known that in the case of two different WFs a step will appear when dipole correction is included (e.g. see https://www.researchgate.net/post/Who-c ... k-function)
Indeed, I easily got such a step for the case of Ag/h-BN interface, for example (see Fig. 1) when activate LDIPOL = .TRUE., IDIPOL = 3 (i.e. Z direction).

Q1.JPG

However, for some other semiconductor/metal interfaces when using the same dipole correction I have very strange curve at the vacuum positions that looks like apple cores or oscillations.
Nevertheless, in Fig. 2 one can still see a step despite strange oscillations.

Q2.JPG

However, in Fig. 3 no step is presented at all.

Q3.JPG

So, my two questions are:
1) Why do these strange curves may appear at the vacuum level? (instead of a simple step as in Fig.1)
2) What should I change/control to fix it and find a step

P.S.
I have already played with vacuum size. In Fig.2 and 3 it's twice higher than the structure thickness. No evidences that further increase of vacuum size can help me.

Also, I was tested a bit DIPOL flag (center mass position for dipole correction) and found that VASP can set it automatically without any problem when the vacuum is high enough. But maybe I'm wrong and I have to set up manually DIPOL flag in some trouble cases.

I will highly appreciate any ideas and advices! Thank you.

[chengcheng_xiao1]

1) Why do these strange curves may appear at the vacuum level?
Because you are plotting the total potential using tag `LVTOT`, the total potential includes the exchange-correlation potential which tends to converge very slowly in vacuum.

2) What should I change/control to fix it and find a step?
Use `LVHAR` to exclude the exchange-correlation part.

[pbsorokin]

1) Why do these strange curves may appear at the vacuum level?
Because you are plotting the total potential using tag `LVTOT`, the total potential includes the exchange-correlation potential which tends to converge very slowly in vacuum.

2) What should I change/control to fix it and find a step?
Use `LVHAR` to exclude the exchange-correlation part.

That solved my problem. Thank you a lot!