VCA setup

[musanna_galib]

Hello,

I’m trying to run a relaxation using the Virtual Crystal Approximation (VCA) in VASP 6.3.0 for a mixed-anion system with 5 elements, but I’m a bit confused about the correct setup. Here's what I have so far:

Setup:
In the INCAR, I added:

Code: Select all

VCA = 1.0 0.79 0.07 0.07 0.07

In the POSCAR, I started from a LiS₂ structure and wanted to replace the S atoms with a mixture of 4 atoms. Here I just changed S to S_vca:

Code: Select all

Li   S_vca  
8    4

In the POTCAR, I concatenated pseudopotentials for all five elements, assuming that the weights in the VCA tag would take care of the mixing.

However, when I check the OUTCAR, it seems that only Li and S (with weights 1.0 and 0.79) are actually considered. I don’t see anything regarding other elements

My Questions:
Do I need to copy and paste the S atomic coordinates 3 more times in the POSCAR to account for other 3 elements?

I’m planning to use VCA for a migration energy barrier calculation using NEB. Does using VCA make sense in that context?

Is it meaningful to compute DOS or charge density difference with a VCA setup, given that VCA may not accurately capture electronic mixing effects?

I’ve attached the OUTCAR for reference. Any clarification would be appreciated!

Thank you.

[andreas.singraber]

Hello!

Please always provide a full set of all relevant input and output files (see forum posting guidelines). Otherwise it is really difficult for us to find issues with the INCAR tags, your workflow or VASP itself. So far I can only see from the OUTCAR file that VASP does only recognize two distinct elements in your POSCAR (see here how the ion species and numbers should be provided):

Code: Select all

ions per type =               8   4

For those two types the first two entries in your POTCAR file are used, which are Li and S. Hence, it seems that the remaining weights in the VCA tag are probably unused.

All the best,
Andreas Singraber

[musanna_galib]

Hello Andreas,

I added all the setup files here. Can you please take a look at the issues?

My Questions:
Do I need to copy and paste the S atomic coordinates 3 more times in the POSCAR to account for the other 3 elements?

I’m planning to use VCA for a migration energy barrier calculation using NEB. Does using VCA make sense in that context?

Is it meaningful to compute DOS or charge density difference with a VCA setup, given that VCA may not accurately capture electronic mixing effects?

Thank you.

[andreas.singraber]

Hello!

Yes, the coordinates of the four S atoms need to be copied three more times below the existing entries. This can be seen in the example given on the VCA wiki page. There, the Ge and Sn coordinates are also duplicated and then mixed at a ratio 0.99 vs. 0.01. Note that also the atom types and numbers are repeated in the POSCAR section above. I assume that you want to keep the Li atoms without "mixture". Then, their coordinates must appear only once and the corresponding weight should be 1.0. This was already correct in your POSCAR. However, the mixture of S, Se, Te and I atoms we have to create by putting atoms of each type onto the desired coordinates and then set the weights as you already did with the VCA tag. Here is how the POSCAR file would look like:

Code: Select all

Li1.93_S0.79_Se0.07_Te0.07_I0.07
1.0
        5.6703357697         0.0000000000         0.0000000000
        0.0000000000         5.6703357697         0.0000000000
        0.0000000000         0.0000000000         5.6703357697
   Li    S   Se   Te    I
    8    4    4    4    4
Selective dynamics
Cartesian
     4.252751827         1.417583942         1.417583942   T   T   T
     1.417583942         4.252751827         4.252751827   T   T   T
     4.252751827         4.252751827         4.252751827   T   T   T
     1.417583942         1.417583942         1.417583942   T   T   T
     1.417583942         1.417583942         4.252751827   T   T   T
     4.252751827         4.252751827         1.417583942   T   T   T
     1.417583942         4.252751827         1.417583942   T   T   T
     4.252751827         1.417583942         4.252751827   T   T   T
     0.000000000         0.000000000         0.000000000   T   T   T
     0.000000000         2.835167885         2.835167885   T   T   T
     2.835167885         0.000000000         2.835167885   T   T   T
     2.835167885         2.835167885         0.000000000   T   T   T
     0.000000000         0.000000000         0.000000000   T   T   T
     0.000000000         2.835167885         2.835167885   T   T   T
     2.835167885         0.000000000         2.835167885   T   T   T
     2.835167885         2.835167885         0.000000000   T   T   T
     0.000000000         0.000000000         0.000000000   T   T   T
     0.000000000         2.835167885         2.835167885   T   T   T
     2.835167885         0.000000000         2.835167885   T   T   T
     2.835167885         2.835167885         0.000000000   T   T   T
     0.000000000         0.000000000         0.000000000   T   T   T
     0.000000000         2.835167885         2.835167885   T   T   T
     2.835167885         0.000000000         2.835167885   T   T   T
     2.835167885         2.835167885         0.000000000   T   T   T

I quickly tested this and it seems to work as expected in VASP 6.5.1. However, at this point I cannot confirm whether this approach is expedient. I will have to confer with my colleagues before I can answer your remaining questions. I would recommend to carefully study the caveats mentioned on the Wiki page and the available literature on this topic.

All the best,
Andreas Singraber