5
$\begingroup$

I want to study the structural, optical, electronic and transport properties of my material that contains Cu atoms using Quantum ESPRESSO. I tried so many approximations but every time I face an issue that limits my progress in the calculations.

For example to improve the bandgap I tried DFT+U but when I try to calculate optical properties using epsilon it's not working. When I try hybrid functional PBE0 the calculations same to stay forever with PAW pseudo-potentials, so I changed it to Norm-conserving pseudo-potentials, however the calculations are very computationally expensive.

I saw in the internet that some use an approximation called Van der Waals-DFT approximation. Would this be appropriate for modeling a p-type semiconductor?

Improve this question
$\endgroup$
15
  • $\begingroup$ Yes, considering dispersion forces like van DER Waals are computationally expenses. Did you considered moving to other DFT software less computer resources eater like SIESTA? $\endgroup$ Commented Mar 31, 2023 at 13:30
  • 1
    $\begingroup$ I guess Camilla needs help with using the ACE implementation, but I suppose an answer to this question could be "Van der Waals functinals are not likely to help for modeling a p-type semiconductor". I've edited the question to make the system of interest more visible. $\endgroup$ Commented Apr 3, 2023 at 11:39
  • 1
    $\begingroup$ It looks like you just set ace to true in the system card: quantum-espresso.org/Doc/INPUT_PW.html#idm450 $\endgroup$ Commented Apr 3, 2023 at 22:46
  • 1
    $\begingroup$ @PhilHasnip ACE is set to .true. by default. So I guess there is no need to specify it. $\endgroup$ Commented Apr 3, 2023 at 22:49
  • 1
    $\begingroup$ @Camilla if you can show your error messages when installing Yambo, you could ask how to fix them in a question with the "software-assistance" tag. $\endgroup$ Commented Apr 5, 2023 at 13:23

1 Answer 1

Reset to default
8
$\begingroup$

No, van der Waals functionals would not be appropriate for this type of system - at least, not normally.

Most semiconductors of interest are strongly bonded 3D materials, with covalent bonds and perhaps a small polarisation (e.g. GaN). Semi-local exchange correlation methods describe these systems well, on the whole, with the exception of the well-known underestimation of the band-gap. The band-structures are usually modelled well, apart from the band-gap itself, so if you already know what the band-gap is, the simplest approach is just to correct the band-gap using a scissor operator.

Van der Waals functionals are designed to capture the non-local, long-range correlation of weakly-interacting densities, which is the mechanism for the van der Waals interaction. This interaction is important when modelling the energies and forces between atoms which are relatively isolated, for example inter-molecular forces, or the long-range interaction between a molecule and a surface. However, it is a subtle correlation effect with a fairly small interaction energy, and it is easily disrupted at short ranges and in the presence of other, stronger interactions. They could be relevant for bundles of loosely-coupled 1D (e.g. nanotubes) or 2D organic semiconductors (e.g. functionalised graphene), but they are extremely unlikely to be relevant for any conventional 3D semiconductors.

Since you are interested in band-gaps, I will also point out that van der Waals functionals do not address any of the issues relevant to predicting accurate band-gaps; for example, they do not contain the correct derivative discontinuity, they do not correct self-interaction, they do not lead to piecewise-linear energies as a function of band occupancy (Koopman's Theorem), and they do not give good values for the ionisation potential.

Improve this answer
$\endgroup$
1
  • 2
    $\begingroup$ Very well explained. Thank you Phil :) $\endgroup$ Commented Apr 4, 2023 at 9:30

You must log in to answer this question.

Start asking to get answers

Find the answer to your question by asking.

Ask question

Explore related questions

See similar questions with these tags.