Is it necessary to perform a new geometric relaxation with vdW corrections enabled for each implementation (DFT-D2, DFT-D3, etc.) when comparing their effects on the adsorption of small molecules on a surface? Or can I use the initially relaxed structure (without vdW corrections) and simply incorporate the vdW corrections during the SCF calculation?


1 Answer 1


Using DFT, DFT-D2, DFT-D3, and DFT-4 will all give you slightly different energies at each geometry, and therefore you will get slightly different geometries if you do a geometry optimization with each of those methods.

Whether or not these differences in geometries matter to you will depend on many factors, such as the computational cost of the geometry optimization calculations (we can't know this without knowing your system), the required accuracy, and the DFT functionals used, among other factors.

I recommend to strike the best possible balance between computational cost and computational accuracy that you can achieve for the geometry optimization (when choosing whether to use DFT, DFT-D2, DFT-D3, DFT-D4, etc.), then to do single point energy calculations at the relaxed geometry with all of those methods. If you are seeing drastic changes in the single-point energy, then perhaps the optimal geometries will also be quite different for each of those methods. If you are seeing barely any difference in the single-point energies, then depending on what the final goals are for your project, perhaps it's okay for you to move on with your project, at the relaxed geometry that you found in your first attempt.

  • $\begingroup$ Thank you for answer, I would like to know whether I should compare the change in total energy of the system or the change in the property I am interested in, for instance the adsorption in my case? $\endgroup$
    – Phill
    Jan 21 at 4:49
  • 1
    $\begingroup$ @Phill That's bordering on a new question (deserving of a new post that explains what you want in more detail, and gives people an opportunity to answer the question with more space than the comment box allows), but if it's feasible in terms of computational cost, it would be better to compare the actual property in which you are interested, rather than just the energy. My answer was about geometry optimization though, which almost always optimizes the energy and nothing else. I also mentioned checking the energy, because this tells you how good the geometry optimization (of energy) worked! $\endgroup$ Jan 21 at 23:45

You must log in to answer this question.

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