The method I found, which seems to be the most commonly used for simulating surfaces, involves constructing a slab and doing a structural optimization by freezing a few of its lowermost layers.

What confuses me is, how thick should such layers be? Also how many layers should be frozen?

  • 4
    $\begingroup$ +1 The answer is probably material-dependent, so you may need to do a convergence test to decide. $\endgroup$
    – ProfM
    Jun 19, 2021 at 9:19
  • $\begingroup$ @ProfM Thank you for the response! I was thinking of using using Valence band energy or Fermi energy for the convergence test because there would be surface states generated as the actual surface forms. But that would not be material dependent, would it? Also, what else would we take into account especially when we don't have much data on the material? $\endgroup$ Jun 19, 2021 at 13:53
  • $\begingroup$ +1. I think @ProfM's answer is correct. For example you are asking how many layers should be frozen, and the correct answer is probably for you to do the calculation with 2 layers frozen, then 3, then 4, etc. until the result ha converged. Your follow-up question asks whether you should use the "valence band energy" or the "Fermi energy" for the convergence test, and this I think ought to be a new question. You also ask what other things should be taken into account, and again, that's not the question you asked originally, and if you ask too many questions in one post it becomes hard to answer $\endgroup$ Jun 28, 2021 at 0:20
  • $\begingroup$ Hi @AshiqueLal! Have you managed to make any progress on this? I ask since it's been almost 6 months now! $\endgroup$ Dec 15, 2021 at 7:31
  • $\begingroup$ This question appears to be abandoned. It can be reopened if OP addresses questions/suggestions in the comments or anyone else wants to provide an answer. $\endgroup$
    – Tyberius
    Jan 13, 2022 at 2:06


Browse other questions tagged .