[QE-users] Correct procedure: CIF to slab with convergence tests

[Léon Luntadila Lufungula]

Dear QE users,

I want to construct an anatase (101) slab starting from the experimental CIF file of anatase (obtained from the American Mineralogist Crystal Structure Database) but I am unsure of the correct procedure with the associated convergence tests of the parameters of the QE calculation. I am basing my procedure on (sparsely documented) work performed by my predecessor with some adjustments based on stuff I read and learned online (including from the MaX School on Advanced Materials and Molecular Modelling with QE organized by the QE developers).

The proposed procedure is as follows:


  1.  Use the cif2cell software package to create a template QE input of the anatase bulk from the anatase CIF
  2.  Run convergence tests on this bulk structure (i.e. perform several SCF calculations on the structure while varying the given parameter):
     *   Ecutwfc
     *   Ecutrho (I'm using PAW pseudopotentials, so the standard ecutwfc x 4 is not sufficient)
     *   K-points
     *   Cell parameters
  3.  Optimize structure with optimized parameters to get final bulk structure
  4.  Convert QE output to CIF
  5.  Use the cif2cell software package to create QE input for a (101) slab created from the optimized bulk structure CIF
  6.  Freeze bottom layer in QE input
  7.  Run convergence tests on slab structure:
     *   Vacuum width
     *   Layer thickness

My specific questions for this procedure are the following:


  1.  Steps 2a-c: My predecessor stated that convergence is achieved once the total energy difference between two consecutive calculations is smaller than 1 kcal/mol, but during the MaX School they did convergence tests and chose converged parameters when the total energy differed by much less than 1 kcal/mol (they said 1 meV, so around 0.023 kcal/mol). Is 1 kcal/mol enough, or should I go lower?
  2.  Step 2d: My predecessor did a vc-relax run to optimize the cell parameters, but in the MaX School I saw they optimized alat instead of performing a vc-relax run. Which of the two is correct?
  3.  Is the order of the convergence tests on the bulk structure right, or should I perform them in another order?
  4.  Step 7a: I'm intending to investigate adsorbates (not adatoms, but small molecules) on the surface, so should I do the convergence tests of the vacuum width on a clean surface or on a surface with a model adsorbate to be certain the actual vacuum is large enough to avoid interactions between the adsorbate and the adjacent slab?
  5.  Step 7b: I'm uncertain how to perform the convergence test for this parameter. I have read one option where you look at the difference in total energy between two consecutive structures and state convergence once this difference is constant within chemical accuracy (1 kcal/mol), but I'm not certain if this is the correct way.

I just want to make sure I do this correctly, so I can start performing calculations without having to worry that my data is inaccurate. Hopefully someone can help me find the correct procedure.

Thanks in advance and kind regards,

Leon Luntadila Lufungula
PhD student at Structural Chemistry Group
University of Antwerp

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.quantum-espresso.org/pipermail/users/attachments/20221212/5c6ffa75/attachment.html>