Doping system modeling: virtual crystal approximation (VCA) vs supercell method.

Question on input files/tags, interpreting output, etc.

Please check whether the answer to your question is given in the VASP online manual or has been discussed in this forum previously!

Moderators: Global Moderator, Moderator

Locked
Message
Author
hszhao.cn@gmail.com
Jr. Member
Jr. Member
Posts: 99
Joined: Tue Oct 13, 2020 11:32 pm

Doping system modeling: virtual crystal approximation (VCA) vs supercell method.

#1 Post by hszhao.cn@gmail.com » Tue Oct 19, 2021 11:58 am

This is a general problem. Though there are so many papers studying the doping system, but I still not so clear which is a preferable method to construct the input file for a specific doping system in question. I know CASTEP supports the so-called virtual crystal approximation (VCA) method, but I don't know if vasp also has this feature. OTOH, if I construct a doping system, which locations - interstice or the in-situ lattice positions - should be used? It seems that the supercell method is more popular than VCA for doping system modelling, am I right? Finally, are there other feasible methods for modeling doping systems?

Regards,
HZ

martin.schlipf
Global Moderator
Global Moderator
Posts: 455
Joined: Fri Nov 08, 2019 7:18 am

Re: Doping system modeling: virtual crystal approximation (VCA) vs supercell method.

#2 Post by martin.schlipf » Wed Oct 20, 2021 7:18 pm

Depending on the level of doping and the required accuracy there are various ways you can study doped systems with VASP.

The cheapest method applies to systems, where the doping can be approximated by additional charges in the system without structural modification. In that case you can change the number of electrons via the NELECT tag. Alternatively, you can just look at the band structure without setting NELECT and reason about where the electrons/holes would go. Naturally, this method only applies to very low concentration of doping and is a very crude approximation.

You can also use the VCA you mentioned (see the corresponding INCAR tag). This method is more cumbersome than the one above, but a bit more physical and can still be done in the primitive unit cell. However, please note the caveats associated with this method described in the wiki.

Finally, the recommended way to model any realistic doping is to use the supercell method. This actually describes the atomic environment of the dopant including surrounding structural relaxations. Of course this comes at the cost of having to calculate the larger cell. You should be aware that in addition to the increased computational cost, these kind of calculations are also demanding in terms of your input: You will need to provide a set of possible choices how the doping could occur and then carefully analyze their formation energies. There is no general rule which ones are possible; this will depend strongly on both the host system and the dopant.

Locked