We are working on the excitonic properties of cadmium-based Quantum Dots (QD) and QD heterostructures. We have MedeA-VASP and hence we managed to make QD structure. Are QD calculations different from doing bulk calculations using VASP?

How are DFT calculations on Quantum Dots (QDs) done using VASP?

Also, are Gaussian orbitals better compared to plane wave basis sets for QD calculations?

  • 2
    $\begingroup$ +1. Welcome to the site, thanks for asking here, and we hope to see much more of you!!! I can answer the second part: I think you're better off using plane-waves than Gaussians for a cadmium based QD. The integrals will not be easy if you use Gaussians. As for the rest: I think you need to tell us what excitonic property you want to know using DFT, and what you know so far: Have you got an XYZ file for the geometry, to start with? $\endgroup$ – Nike Dattani Jul 20 '20 at 2:23
  • $\begingroup$ Thanks for your response. We have MedeA 3.0, so managed to make QD structure. Want to know how to do calculations on the optical properties of excitons. $\endgroup$ – Suseel Rahul Jul 20 '20 at 2:57
  • $\begingroup$ You will have to tell us what excitonic properties you're interested in calculating using DFT. $\endgroup$ – Nike Dattani Jul 20 '20 at 2:57
  • $\begingroup$ We want to study the absorption properties of excitons in Quantum Dots. $\endgroup$ – Suseel Rahul Jul 22 '20 at 3:47
  • $\begingroup$ Which absorption property? $\endgroup$ – Nike Dattani Jul 22 '20 at 3:48

Calculation of QD using VASP is not that much different from bulk calculations; you just need to ensure that sufficient vacuum is applied in all 3 spatial dimensions. My experience is that 10-15 Angstrom is sufficient, but this needs to be tested for your system and property of interest.

The trends in many optical and electronic properties (such as absorption energy) of QD are reliable using conventional DFT with semi-local functionals such as PBE. It is well known that the first exciton peak decreases in energy as the size of the QD increases, due to the loss of quantum confinment effect.

Another important thing to note is that ligands also play a very important role at controlling the optical properties. They are different from "bare" QD. In general the surface needs to be fully passivated with ligands while ensuring overall stoichiometry.

  • $\begingroup$ What about the kmesh? Is it same as in bulk calculations? $\endgroup$ – Thomas Jul 26 '20 at 9:55
  • 2
    $\begingroup$ Only the Gamma point need to be sampled. $\endgroup$ – liuyun Jul 26 '20 at 10:07
  • $\begingroup$ So a 1x1x1 gamma centered mesh? $\endgroup$ – Thomas Jul 26 '20 at 10:08
  • 2
    $\begingroup$ Yes that is correct. You can also use the vasp_gam binary to speed up the computation. vasp_gam is specially compiled for gamma point k-mesh calculations. $\endgroup$ – liuyun Jul 26 '20 at 10:12
  • 1
    $\begingroup$ Thanks. This will help. What happens if we use more k points? $\endgroup$ – Suseel Rahul Jul 26 '20 at 11:48

It is worthwhile to add to liuyun's answer that even though these calculations are in theory straightforward there are some caveats you have to acknowledge.

  1. Band gap tends to be underestimated by conventional DFT methods. This can be fixed to some degree by relaxing your nanoparticle using functionals such as PBE then using a hybrid functional such as HSE06 as a single point for band structure. This is something that must be benchmarked though since it is possible a relaxation at the HSE06 level will change things.

  2. Computational limits will quickly become a problem with common experimental systems. As a quick check, using ASE-GUI to make a 1.9 nm nanoparticle results in 249 atoms and a massive unit cell. This is where it may be useful to look at trends on smaller systems and/or using a lower level of theory to pre-optimize your particles.

  3. I am not an expert in this sort of calculation but you seem concerned with exciton properties. I do believe you will not get the optical absorption energy due to the exciton binding energy shifting the optical band gap higher than the electronic band gap. You may need time dependent calculations for that, but you can use some codes to analyze the band structure for the effective masses of the electron and hole. This may be enough for you depending on what your goal is.

  4. This is less of an issue and more of a comment, ligands will shift the optical properties but solvents will as well. Using VASPsol to handle solvent interactions may be useful.

You could also consider switching to a different code / approach, I believe GPAW supports TD-DFT calculations in LCAO mode which is a highly parallelizable method. This won't give the same level of accuracy (using default basis sets at least, I do not know about the limit of accuracy as you increase the basis set). This may let you directly find the optical absorption spectra.

  • $\begingroup$ Hi, thanks for your thoughts. Actually we are working with 1nm nanoparticles. Along with PBE, GGA-PBE, and HSE06, we tried GW calculation. With wich, we got excitonic absorption peaks in the bulk calculations. But nanoparticle calculations need large computation time. $\endgroup$ – Suseel Rahul Jul 26 '20 at 16:28
  • $\begingroup$ With such small nanoparticles, this should be feasible. You might look at GPAW though. What kind of computational power do you have access to? $\endgroup$ – Tristan Maxson Jul 26 '20 at 16:30
  • $\begingroup$ Workstation with 24core and 500GB ram. $\endgroup$ – Suseel Rahul Jul 26 '20 at 16:43

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.