I want to build amorphous $\ce{Al_2O_3}$ with the AIMD (ab initio molecular dynamics) method implemented in VASP. I'm thinking of this process in three steps:

  1. Build initial supercell
  2. Melt
  3. Quench

I have questions about all these steps, but here I will just address the first.

Supercell size

The conventional cell of $\ce{Al_2O_3}$ (taken from the materials project website) contains 15 atoms, as shown in the following figure.

Al2O3 conventional cell

To do AIMD calculations, we must have enough atoms in the simulated cell. This means that I need to build a supercell. For the current system, is a $3 \times 3 \times 1$ (135 atoms) supercell OK? To obtain a reasonable amorphous structure, how big should the melt supercell be?

  • $\begingroup$ +1. But I just changed the title slightly. There's a few disadvantages of using MathJax in titles, such as disqualifying the question from appearing on the Hot Network Questions list: mattermodeling.meta.stackexchange.com/a/190/5 $\endgroup$ Oct 30, 2020 at 19:38
  • $\begingroup$ How "amorphous" your structure is will depend on the cell size. Do you intend to make a surface out of it? I think this question is answerable as is, but some slightly more detailed goals might help. If you need a surface, it may affect things differently. $\endgroup$ Oct 31, 2020 at 4:06
  • 1
    $\begingroup$ @TristanMaxson:researchgate.net/publication/… Please take a look at this paper. The amorphous Fe2O3 is built from AIMD. $\endgroup$
    – Jack
    Oct 31, 2020 at 12:54
  • $\begingroup$ I made some edits to the question to bump it back up, but part of the reason it may not be getting answers is it's fairly broad. It would probably work better as 3 separate questions that link to each other. $\endgroup$
    – Tyberius
    Dec 7, 2021 at 1:34
  • $\begingroup$ @Tyberius Thanks. $\endgroup$
    – Jack
    Dec 7, 2021 at 2:49


Browse other questions tagged .