3
$\begingroup$

I am currently conducting a DFT investigation into the magnetism of 2D materials by introducing defects into the structure. Is there a minimum distance I should maintain between two defects in introducing defects into the 2D structure?

$\endgroup$
2
  • $\begingroup$ You should have an advisor introducing you to these things. What is your background, that you would be new even to quantum mechanics? $\endgroup$ Mar 26 at 5:28
  • 1
    $\begingroup$ It depends on what you are trying to model. Are you trying to model some kind of doping, i.e., very very low concentration of that defect? or some kind of 'alloy'-like structure where the defect is much more abundant? Also, can your defect carry/trap charge? $\endgroup$ Mar 26 at 7:22

1 Answer 1

4
$\begingroup$

It depends on the system and the goal.

If the goal is to model truly isolated defects (i.e. not clustered together), then you have to use a system large enough that there are no spurious effects of defect-defect interaction. You have to converge the properties you are interested in wrt. to system size.

Note that the converged system size for isolated defects depends on the nature of the material. E.g. if the material is metallic, the E-field produced by a charged defect is screened very well and the field falls of quickly with distance. In an insulator, the opposite is true and the E-field from a defect falls of slowly, so you need a large supercell to model an isolated defect.

In many cases, point defects arent really isolated and cluster together (e.g. Wadsley defects in TiO2), so you don't necessarily want to keep the defects separated. If you are starting from scratch, you have to sample all possible configurations of the defects and find the most energetically favorable one(s). Note, even for clusters of defects, you still have to converge wrt. to supercell size.

Simple combinatorics will show you that you have a lot of calculations to do! I suggest you read papers on the material you want to model. The best case scenario in general is to model defects that were found in experiments (nature picks the lowest energy configuration for you!). Since you are looking at a 2D material, this seems unlikely since not many 2D materials actual exist in the lab. The second best case, is to find papers where other people have done the hard work finding the most likely defects. If you cant find these, find comparable materials and see what defects were studied in those systems.

$\endgroup$
2
  • 1
    $\begingroup$ +1 but I'd like to add one thing: for material with a band gap, we can use charge correction approach. QE has Makov-Payne or Martyna-Tuckerman correction or they can sometimes be even calculated by hand provided the material's dielectric constant. That means we don't have to use very large supercell but some computationally efficiently tractable supercell with these correction terms should suffice. I have tried a 128 atoms supercell and a 64 atoms supercell and even though their energy per atom differ, after applying the correction term, the difference (almost completely) vanishes. $\endgroup$ Mar 27 at 9:48
  • 1
    $\begingroup$ Excellent! Thanks for bonus info. I'll look into it next time I need it! $\endgroup$ Mar 27 at 16:29

You must log in to answer this question.

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