I want to study the adsorption energies of many different atoms to some surfaces. For that, I need to calculate the energies of free atoms (adsorbates) in a vacuum.

I understand that I am supposed to put an atom into a large cell to minimize its interactions with its periodic images.

However, I am confused, what kind of cell should I use.

Symmetric or asymmetric?

In the Materials Square site, they advise an asymmetric (orthorhombic) cell in order to avoid partial occupation of orbitals by electrons. VASP wiki also advises low symmetry (here and here).

However, in the Computational Materials Physics course, I learned that it is better to put an atom into an FCC cell because the atom in that cell has the largest number of nearest neighbors which makes their influence on the atom as isotropic as possible and thus minimizes interactions caused by the finite size of the cell.

Which cell is better? A highly symmetric FCC, a slightly distorted orthorhombic, or some other one?

  • $\begingroup$ Nice questions! I also have a question related to 2), that I'd appreciate if anyone can answer: when I learned about the calculation of periodic systems I was told that, e.g. if we want to calculate the adsorption energy of an atom on a surface, then the cell used for the atom should have the same size and shape as the cell used for the surface. I believe the reason is that this helps to remove the spurious interactions of an atom with its periodic image when subtracting energies. My question is, is this better than using a cell whose size is determined by the atom alone? $\endgroup$
    – wzkchem5
    May 19, 2022 at 8:11
  • 2
    $\begingroup$ My 2cents: I learned too to use the same lattice as the whole system. $\endgroup$
    – Camps
    May 19, 2022 at 13:47
  • 2
    $\begingroup$ I think this maybe too many questions to address with one post. You should narrow this down to one question and ask any remaining questions that aren't addressed here in new posts. $\endgroup$
    – Tyberius
    May 19, 2022 at 14:56
  • $\begingroup$ @wzkchem5 This probably better as its own post. If you want to reference the context, you can link to this question when you create your question post. $\endgroup$
    – Tyberius
    May 19, 2022 at 15:49
  • $\begingroup$ When you use a plane-wave code to compute the property of an isolated atom, it is better to have an orthogonal cell (> 20 Å) with different lattice vector, or to explicitly ask for no symmetry. This is because some atoms can adopt weird orbital configuration due to symmetry i.e O, will have 1.33 occupation on px, py and pz. Often computing isolated atoms will lead to problems anyway (depending on the pseudopotentials it might not even converge). It is often better to use half the value of the diatomic or some other references. $\endgroup$
    – Okano
    May 19, 2022 at 19:19

1 Answer 1


In the end, the OP did the following:

"take a unit cell of the bulk of that metal, calculate its energy and then divide it by the number of atoms in the unit cell. "

The question can stay open in case a better answer may come in the future, but the question also doesn't need to remain in the unanswered queue since the OP is not longer working on the project (see the same comment linked above).

  • $\begingroup$ That is the bulk energy, not the atomic energy. $\endgroup$ Feb 9, 2023 at 13:40
  • $\begingroup$ To clarify, the question was "What is the best cell shape to calculate the energy of a free atom with a periodic DFT code (VASP)?". What the answer is saying is that one should not calculate free atoms but atoms that are bonded together in bulk. $\endgroup$ Feb 15, 2023 at 20:18

You must log in to answer this question.

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