[QE-users] How to handle slab calculations

[Kazume NISHIDATE]

Hi, Léon;

> I don’t quite understand why you have to do a dipole correction
> when adsorbing a molecule on the surface, is this just due to the
> fact that you induce a dipole upon binding and the PBC enlarge this
> effect?

Modern DFT codes assume the PBC for the systems. If you put a dipole
bar (a molecule) in the system, something like (+)=(-), then the (+) pole will
directory interact with the (-) pole over the PBC. It causes
artificial effects on its energy, and therefore it brings about a
wrong result on the structural relaxations too.

> what I understand, I cannot use Makov-payne as it is “ONLY FOR CUBIC
> SYSTEMS” and my cell is monoclinic (a=b≠c, α=ß=90°≠γ), esm is
> also not an option as it “Requires cell with a_3 lattice vector along

Alternately, you can put the same molecules on the othe side of the
slab. In this case, the dipole effect from the one side will be
eliminated by the dipole effect from the other side if you set correctly
the molecules.

> Are they useless without applying a dipole and/or coulomb correction?

You can check the effectiveness of the dipole correction by plotting
the potential energies averaged over x-y plane.
 
Dr. C. Wolf gave a detailed explanation on the issue.

https://christoph-wolf.at/tag/dipfield/


> 2022/09/10 21:27、Léon Luntadila Lufungula <Luntadilatiti at hotmail.com>のメール:
> 
> Dear QE users,
>  
> I have been performing calculations on surface-modified TiO2 for some time now, as I took over a research project from a previous PhD student who quit his PhD. The project was already running for some time so I based my calculations on his input files as I started to learn how to use QE. Now I’m reading a lot of stuff online about performing QE calculations and I start to notice that my input files might be wrong...
>  
> I am performing calculations on a 3-layer anatase (101) slab with a 20 Å vacuum width above it and adsorbing different adsorbates onto the surface. I just found out that apparently when you are adsorbing molecules onto a slab you should use a dipole correction (dipfield = .TRUE.), which I did not do... I also read that for slab calculations it is recommended to use the assume_isolated keyword to cut off the Coulomb interaction in the z-direction and that the dipfield and assume_isolated keyword should not be used simultaneously.
>  
> I tried reading the papers on dipfield and the different assume_isolated options but I’m still pretty new to QE and periodic calculations so I found them hard to understand completely... Therefore, I would be most grateful if somebody could elaborate on the following questions:
>  
> 	• Is it correct that you should use either dipfield or assume_isolated and not both at the same time? If so, which one is the best one to use?
> 	• I don’t quite understand why you have to do a dipole correction when adsorbing a molecule on the surface, is this just due to the fact that you induce a dipole upon binding and the PBC enlarge this effect?
> 	• If assume_isolated is the way to go, which option should I use? From what I understand, I cannot use Makov-payne as it is “ONLY FOR CUBIC SYSTEMS” and my cell is monoclinic (a=b≠c, α=ß=90°≠γ), esm is also not an option as it “Requires cell with a_3 lattice vector along z, normal to the xy plane, with the slab centered around z=0.” which is not the case for my system. So this leaves 2D and martyna-tuckerman as possible options, but I can’t seem to find a straight answer which is best or what the differences are.
> 	• What about the calculations I have already performed? Are they useless without applying a dipole and/or coulomb correction?
>  
> Thanks in advance!
>  
> Léon Luntadila Lufungula
> Master student
> University of Antwerp
>  
>  



西館数芽
Kazume NISHIDATE Ph.D