Modelling a structure with fractional elements for Bandgap

[SKM]

Hi Admin
Can we do model fractional stoichiometric models for optimising and finding properties in VASP? Like A12.5 B36.2 C1 ? [The values given here are just random]. My idea is to work on doped hetero structures to find band-gap and band structure, etc. to corroborate experimental results.

1. If we can, then how to do it?
2. What tags are essential in INCAR?

Regards

[merzuk.kaltak]

Fractional stoichiometric models can be studied only with a super-cell.
This means the compound A12.5 B36.2 C1 must be studied with a unit cell that describes A125 B362 C10

[SKM]

Hi Merzuk,
Thanks for a very quick reply. In such case one of my actual model requires A69.8B45.5C1D209.3. That means i have to make Supercell with A698B455C10D2093? That will be a huge cell. Is it really workable to do DFT optimisation and band gap, etc?

Regards.

[ferenc_karsai]

It's definitely computationally very costly even with "bad" parameters.

Please have a look on the tutorials first how to run calculations (especially bandstructure calculations):
https://www.vasp.at/wiki/index.php/Bulk ... -_Tutorial

[alex]

Hi SKM,

maybe you should think more in an idealized way, because these compositions are often the result of a complex optimisation in the lab, which do not necessarily need to be an optimium. Try to find the important properties, e.g. what is so special about your compound?
- make a smaller model to resemble these properties
- even with a smaller model you will have a lot of 'fun' figuring this out

What makes the compound special and what is the root cause? That's the question. And I'll take probably time to find the answer.

Cheers,

alex

[SKM]

Hi Alex,
thanks for the reply.
i also thinking if i can make a small model. This is of course an experimental result based model which i trying to simulate through computations.

but the stoichiometric proportion is important, i cant think any smaller model. however, i did not clearly under stood your reply as in;

what is so special about your compound?
- make a smaller model to resemble these properties
- even with a smaller model you will have a lot of 'fun' figuring this out

can you give some simple example and how can we reduce the model size? that will be interesting.
Regards

[alex]

Hi,

if you state 'This means the compound A12.5 B36.2 C1 must be studied with a unit cell that describes A125 B362 C10', try starting with sth. like A12B35C1 depending on symmetry of the system and have a look, where it goes. ~50 atoms are easily tackled and you will have a chance to observe the effect you are after, because it is not diluted by its embedding in hundreds of atoms... Or what's so special about this little carbon though?

Hth,

alex

[SKM]

Hi Alex
thanks for the response.

the C1 is not meant for Carbon. i am just asking a query using just alphabets ABCD... representing some elements/atoms. but not actual atom names.

As per your logic, and with my needed exact stoichiometric composition of A69.8B45.5C1D209.3 is thus need at least about A70B46C1D209. if you say so, that is what exactly i was trying initially, but due to elemental proportions needed for matching experimental composition, i put the initial query in the forum.
Regards

[alex]

Hi SKM,

I probably got my last posting not clear enough. Soory for that.

If you need A69.8B45.5C1D209.3 try to minimize the effort by keeping the proportions of the main components, e.g. A7B5D21 and add C1 and see how it evolves.

I hope I got it better this time.

Cheers,

alex

[SKM]

Hi Alex,
thanks for the mail. after discussing your suggested proportion with our experimental person, one of the models that i need to do could be say A3B1.5D7.5 then C0.03. Now as per your reply below how do we really add that 4th atom (i.e.C 0.03) in the VASP model? i cannot visualize how to add that fraction or even single atom and where to add?? but if i add single atom the whole purpose of study lost as it doesn't match experimental proportion..

Again if i make the combination as whole numbers again will be a huge atoms. Any idea? two factors holding be back not starting the run (1) huge number of atoms (2) how to account for fractional atoms but with small numbers. [ though it appears back to the square ONE, but i cant help asking same. lol]