VCA in alloy

[kaynat_kalvi]

Hi

I have read the VCA topic on vasp wiki and all the discussion in vasp forum and I also did Ge-Sn example and it works fine. My confusion is about doing doping in one such alloy, let suppose I have GeSn, 10 atoms, and they have 0.7 and 0.3 occupancy, respectively. So VCA = 0.7 0.3, now this system is GeSn is an alloy, and Now I want to do doping of a similar atom e.g adding another Ge atom in this system. How should I write the POSCAR and VCA tag now?
In Poscar, would it be Ge Ge Sn, 1 10 10 and followed by 0 0 0 being the coordinate of pure Ge and the 10 Ge coordinates and then repeating these 10 for Sn?
and in INCAR, VCA=1 0.7 0.3 ?

thanks!

[michael_wolloch]

Dear Kaynat Kalvi,

This method would give you a working VASP calculation, and you would end up with a single Ge atom surrounded by virtual GeSn mixed atoms.
However, I am quite sceptical about how physical this setup would be. The VCA works best for bulk properties of well-mixed random alloys, and, as the VASP wiki states, also then only with major caveats and not in all cases.

Please provide more context about what quantity you are trying to calculate, which VASP version you are using, and other possibly relevant information. Then it is much easier for me to asses if the VCA can achieve what you want and possibly help you achieve your goals by other means.

Thank you, and all the best,
Michael

[kaynat_kalvi]

Hi,

Thanks for your response. There is this paper, (https://www.sciencedirect.com/science/a ... 0124005122), which I want to reproduce. Here, they calculated the electronic band structure calculations of a 128 atom supercell of unstrained Si0.6Ge0.4 containing a Ge [110] split self-interstitial (which is one extra atom in SiGe supercell).

There, they uses ABINIT, but I am using Vasp5.4. PBE functionals, Please help me here!

[michael_wolloch]

Dear Kaynat Kalvi,

Indeed, VASP's VCA feature should be closely related to the "alchemical mixing" of ABINIT. You should be able to do a similar computation in VASP 5.4 using PBE or LDA pseudopotentials for Ge and Si.

If you have any trouble getting your calculation to run, please attach the input and relevant output files of your calculation in a .zip file, according to the posting guidelines.

Thanks, Michael

[kaynat_kalvi]

Hi,

I had multiple tries using 128 and 64 atom SiGe but got bunch of errors(internal error in VASP: SETUP_PRIMITIVE_CELL, Could not associate all atoms to
their respective cells. try to change SYMPREC in the INCAR file, or better improve precision of the PoSCAR file
POSCAR, INCAR and KPOINTS ok, starting setup
FFT: planning ...
WAVECAR not read
scaLAPACK: Routine ZPOTRF ZTRTRI failed! 2 1 1
LAPACK: Routine ZTRTRI failed!
scaLAPACK: Routine ZPOTRF ZTRTRI failed! 2 1 1
Abort(1) on node 102 (rank 102 in comm 0): application called MPI_Abort(MPI_COMM_WORLD, 1) - process 102
scaLAPACK: Routine ZPOTRF ZTRTRI failed! 2 1 1
LAPACK: Routine ZTRTRI failed!
Abort(1) on node 43 (rank 43 in comm 0): application called MPI_Abort(MPI_COMM_WORLD, 1) - process 43
scaLAPACK: Routine ZPOTRF ZTRTRI failed! 2 1 1
LAPACK: Routine ZTRTRI failed!)
I tried varying SYMPREC,
attaching the file here, please let me know why am I getting such errors.
Thanks a million!

[michael_wolloch]

Dear Kaynat Kalvi,

The supercell in your POSCAR file is not set up correctly. You try to cram 64 atoms into the unit cell capable of holding 2.
I would advise using ASE, or pymatgen to create a supercell. Attached is a small Python script using ASE that I used (I got it from ChatGPT, so it might not be the most elegant, but it works).

Starting from a primitive cell based on your POSCAR with just Ge (ASE gives nicer results if you are not using VCA double atoms here):

Code: Select all

cubic Ge
5.659
    0.0   0.5   0.5
     0.5   0.0   0.5
     0.5   0.5   0.0
Ge 
 2
Direct
0.0000 0.0000 0.0000
0.2500 0.2500 0.2500

Use the script like this for e.g. a 3x3x3 supercell with 54 atoms:

Code: Select all

python write_supercell_ase.py POSCAR POSCAR_3x3x3.vasp 3 3 3

Now you just copy the Ge positions to the Si duplicate atoms and change the comment line at the top, the symbol, and the number of atoms lines and you receive the POSCAR_3x3x3.vasp file I also attached.

It is a good idea to always check your structures with a visualization tool before you start your calculation. You can use e.g., ASE, or VESTA

Let me know if the calculation still gives you trouble with a proper POSCAR file,

Cheers, Michael

[kaynat_kalvi]

Hi Chris, I was doing a silly thing, thanks for pointing it out, it works for 64 atom cell, but then i tried a 128 atom cell and the relaxation and scf calculations are done successfully, when I tried to do band structure cal from a CHGCAR file (gotten form an scf cal), the calculations are done successfully but I think while writing the results there happen some issue which i am not been able to figure out. Please see the attached file and let me know where did i go wrong.
thanks