2
$\begingroup$

I am using Quantum ESPRESSO to do some DFT calculations. I want to know if the optoelectronic properties obtained by a unit cell would remain the same when the properties are obtained by using a larger supercell.

$\endgroup$
6
  • 2
    $\begingroup$ Do you mean a) Are the answers different when you use open boundary conditions compared to periodic boundary conditions? or b) When using periodic boundary conditions can the the answers depend upon the size of the supercell you use? $\endgroup$
    – Ian Bush
    Jun 8, 2023 at 12:48
  • $\begingroup$ @IanBush the OP confirmed that he is indeed asking (b) $\endgroup$ Jun 8, 2023 at 14:49
  • $\begingroup$ It would be good if you could unpack what you mean by optoelectronic properties in terms of the exact quantities you are interested in. e.g. band structure, dielectric function and absorption spectra, etc. $\endgroup$
    – CW Tan
    Jun 8, 2023 at 15:02
  • 3
    $\begingroup$ As long as the supercell is exactly a periodic reproduction of the original unit cell and the computational parameters are equivalent (be especially careful with the k point sampling) you should get exactly the same, except possibly with a trivial scaling (e.g. the total energy). You are studying the same system, just exploiting symmetry differently. $\endgroup$
    – Ian Bush
    Jun 9, 2023 at 8:11
  • 1
    $\begingroup$ When you are doing calculations, you should always check how the studied properties converge by changing cell size / k point space (the two are interconnected). $\endgroup$
    – Greg
    Jun 9, 2023 at 12:21

1 Answer 1

3
$\begingroup$

This is a partial answer to your question. Here I discuss the effect of the size of supercell on DOS, PDOS, and band structure. Others can add the effect on optical or other electronic properties. As pointed out by Ian Bush in the comment, as long as the supercell is exactly a periodic reproduction of the original unit cell and ... the computational parameters are equivalent, you should get exactly the same, except possibly with a trivial scaling. When I ran a calculation of a carbon chain using 1,2,3,4, and 8 atoms per supercell, I found that the total energy scales almost exactly linearly.

Like the total energy, the DOS also scales linearly (mostly). So for example, if you did your calculation with 2 atoms per supercell and found the DOS as 3 states/eV at some E. Then, you will find the DOS to be 6 states/eV at the same E if you used 4 atoms per supercell. The same thing is true with PDOS.

For the band structure, if you use a larger supercell, then we have a phenomenon called band folding. See this and this questions about band folding and unfolding. Very briefly, for a larger supercell, the band structure will be more densely populated.

$\endgroup$

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .