I am trying to calculate the d orbital energy of the Cu single atom, like in this page, I have tried to use the Eigenband energies minus the vacuum energy level(about 0.007eV).

 k-point     1 :       0.0000    0.0000    0.0000
  band No.  band energies     occupation 
      1      -6.8877      1.00000
      2      -6.8877      1.00000
      3      -6.8872      1.00000
      4      -6.8863      1.00000
      5      -6.8863      1.00000
      6      -5.9796      1.00000
      7      -0.9533      0.00000
      8      -0.8847      0.00000
      9      -0.5722      0.00000
     10      -0.1787      0.00000
 Fermi energy:        -5.7513822398

 spin component 2

 k-point     1 :       0.0000    0.0000    0.0000
  band No.  band energies     occupation 
      1      -6.6023      1.00000
      2      -6.6010      1.00000
      3      -6.6010      1.00000
      4      -6.6010      1.00000
      5      -6.6010      1.00000
      6      -5.2707      0.00000
      7      -0.4024      0.00000
      8      -0.2912      0.00000
      9       0.0570      0.00000
     10       0.0751      0.00000

The result is far from the experient result. My question is : how should I calculate the d orbital energy using VASP?

  • $\begingroup$ Are you sure the method you used was appropriate for a transition metal species? If you are planning on using DFT (you've used it as a tag, so I assume that's what you're using), as opposed to an ab initio method, you need to make sure you pick the right functional and basis set. $\endgroup$ Nov 4, 2023 at 7:25
  • $\begingroup$ In VASP, I have to use a plane-wave basis set. I already tried to use GGA or HSE06, but the results aren't consistent with the experiment. I am wondering if the pseudopotential used in VASP isn't suited for calculating the exact orbital energy of the isolated atoms. $\endgroup$
    – Jack
    Nov 4, 2023 at 7:47
  • $\begingroup$ I know this technically doesn’t answer the question, but my suggestion would be to use an ab initio program like ORCA, Gaussian or NWChem. That way you’ll have access to far more accurate methods that way. Unless this is a proof-of-concept calculation…? I looked up VASP, though—could you see if MP2 works? $\endgroup$ Nov 4, 2023 at 16:51
  • $\begingroup$ @isolatedmatrix. I could find data using DFT to calculate the orbital energy, but the data is far from the experimental data. e.g. for the O atom, the 2p orbital energy is -0.338381 Hartree, which is absolutely not consistent with the photoelectron spectroscopy result. Here are the reference results calculated by NIST. nist.gov/pml/… $\endgroup$
    – Jack
    Nov 5, 2023 at 8:38

1 Answer 1


I will suppose that you have done the following as your approach looks correct, you shoud have performed an (SCF) calculation and include a sufficient number of bands to accurately capture the d orbital energies. Make sure to converge the calculation properly, including the k-point sampling. Once the SCF calculation is complete, you can analyze the band structure output. This output usually includes the energy levels and occupations for each band at different k-points in the Brillouin zone. Identify the d orbital bands: Look for the bands associated with the d orbitals of the Cu atom. I think in your provided output, the d orbital bands are not visible due to the incomplete band structure. Typically, the d orbitals appear at higher energies, so you may need to examine bands beyond the ones shown. Determine the d orbital energy: Once you have identified the d orbital bands, you can calculate their energies. The energy of a band is given relative to the Fermi energy. In your case, subtracting the vacuum energy level (0.007 eV) from the band energies should give you an approximation of the d orbital energies. Keep in mind that the accuracy of your calculation may depend on various factors, such as the exchange-correlation functional used, the basis set, and the convergence criteria. It's important to ensure that your VASP calculation is appropriately converged and includes all relevant factors for an accurate determination of the d orbital energies.

  • $\begingroup$ The model is a single atom in a box, since there have been some bands unoccupied in the result, I don't think the band structure is incomplete, I have tried to set NBAND=30, and the result is almost the same. $\endgroup$
    – Jack
    Nov 4, 2023 at 1:10

You must log in to answer this question.

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