Partial occupancy of atom in a crystal structure

[tasdiquearman]

Hello all. Can anyone help me with the DFT calculations for a structure having partial occupancies? Out of 5 elements in my compound, 2 are S and P, and they are supposed to have an occupancy of 0.67 and 0.33, respectively. The total number of atoms is 108, out of which 18 atoms are S and P when added together. To maintain the occupancy ratio, I took 12 S atoms and 6 P atoms to create the structure and carried on with the calculations. Is this approach correct? Because while calculating the bandgap, I get different values when S and P are randomly positioned. Also, how can I use the VCA approach to do the calculations? What is the procedure of the VCA approach?

[ahampel]

Hi,

thank you contacting us via the official VASP forum.

I assume you mean by partial occupancies that the sites have partial occupancies as in randomly occupied by S or P in a ration of 2/3 to 1/3? I think I am not too surprised that the band structure changes for certain atomic site configurations. Typically you would need to create a few of these structures and obtain a final electronic structure by averaging for the resulting DOS. Only if the different structures are symmetry equivalent they will result in the same DOS, i.e. bandgap.

See for example these forum posts:
https://wwww.vasp.at/forum/viewtopic.php?t=13214
https://www.vasp.at/forum/viewtopic.php?t=16848
and maybe also this one in relation to VCA:
https://vasp.at/forum/viewtopic.php?t=19224
https://vasp.at/forum/viewtopic.php?t=19846

For help on VCA, did you check: https://www.vasp.at/wiki/index.php/VCA ? This would then not require a super cell but instead you assign the partial occupancy in the INCAR VCA tag. Let me know if this helps you.

Best,
Alex

[tasdiquearman]

Thank you for the valuable insights. I will do a few runs to get an average bandgap as it changes for different atomic configurations. However, after completing the calculations using the VCA method, the system is shown to be metallic(i.e., band gap = 0), which should not be the case. I followed the below steps for the VCA method. Kindly verify if I did it correctly.
1. Created the POSCAR by duplicating P and S atomic coordinates, i.e., writing them twice in the POSCAR file. This results in the POSCAR having 18 coordinates for S and 18 coordinates for P. Other elements have coordinates as earlier.
2. Included the VCA tag in the INCAR file, i.e., VCA = 1 1 0.67 0.33 1, for five elements.
3. The above two steps are for ionic relaxation.
4. Ionic relaxation was followed by SCF calculation, where I removed the VCA tag from the INCAR and changed the POSCAR to having 12 coordinates of S and 6 coordinates of P (to maintain the ratio of 2/3 & 1/3, respectively).
5. Then, I did the DOS and band structure calculations.

Kindly tell me what I did wrong. Or is there anything to do with the POTCAR files I am using for P and S? I am using PAW_PBE S 06Sep2000 for S and PAW_PBE P 06Sep2000 for P.
As I understand, the VCA method can be helpful with doped systems having partial occupancies.

Thank you.