I am studying atomic slabs which are periodic in 2 dimensions. Normally in the third dimension, I add a vacuum layer and use dipole correction to mimic proper boundary conditions. Recently I have discovered a property 'assume_isolated' with a parameter '2D' in Quantum Espresso implemented in the way described in Phys. Rev. B, 96(7), 75448 (2017) In the paper the authors apply this formalism to the graphene which is atomically thin. Is this formalism applicable for slabs with a thickness of about 3 nm? When I applied it I saw strange oscillations in the potential for the vacuum layer where I expected to see flat lines. It seems like as there are some atoms in the vacuum layer appeared (image charges?).

Below are two examples with a silicon slab.

  1. With a vacuum layer and dipole correction: Vacuum layer and dipole correction
  2. With assume_isolated with the parameter 2D: With assume_isolated + 2D
  • 1
    $\begingroup$ I have an example in literature where someone has done this. arxiv.org/pdf/1407.2698.pdf , sorry I cannot be more help than this though. $\endgroup$ – Tristan Maxson Oct 12 '20 at 0:48
  • 1
    $\begingroup$ Just adding a comment apart from my answer: what do the other properties of your system look like when compared between the two methods? Bands, DOS, etc. It may be worth sending your question to the QE mailing list as well (there are a few questions there about '2D', but not exactly what you are asking): lists.quantum-espresso.org/mailman/listinfo/users $\endgroup$ – Kevin J. M. Oct 12 '20 at 1:53

Technically I'm not answering your question of whether this method is applicable to slabs of ~3 nm thickness, but wanted to give you an answer that could be helpful anyway. I would not generally use '2D' for slabs, although I have not tested it before. The parameter of interest here is assume_isolated = 'esm'. In my own experience, with ESM vs. a standard dipole correction the energies and potentials come out identical, assuming you converged your vacuum thickness and other calculation parameters.

As a side note, switching between a dipole correction or ESM can be a strategy to try when a slab is giving you SCF convergence problems.

Edit: Also, I think you should be able to use '2D' for slabs. There may be something odd going on here.


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.