Author Topic: Analysis of calculated LSCC U values results  (Read 3443 times)

0 Members and 1 Guest are viewing this topic.

Offline Fan Xiaozheng

  • New QuantumATK user
  • *
  • Posts: 2
  • Country: cn
  • Reputation: 0
    • View Profile
Analysis of calculated LSCC U values results
« on: June 14, 2024, 14:08 »
Dear All,

When calculating the LSCC U values of VS2, the results are as follows (Calculation Converged). What do the corresponding values of the two 3d orbits mean?

+------------------------------------------------------------------------------+
| Calculated LSCC U values (in eV)                                             |
| l = Thomas Fermi screening length (in Angstrom)                              |
+------------------------------------------------------------------------------+
|   0   V    (l=0.29)  3d: 6.02  3d: 1.49                                      |
+------------------------------------------------------------------------------

Thank you for your help.

Offline filipr

  • QuantumATK Staff
  • Heavy QuantumATK user
  • *****
  • Posts: 82
  • Country: dk
  • Reputation: 6
  • QuantumATK developer
    • View Profile
Re: Analysis of calculated LSCC U values results
« Reply #1 on: June 17, 2024, 10:23 »
The two U values are for the two "3d like" orbitals in the basis set. The basis sets used are not exactly the true solutions to the Schrödinger to the isolated atom system - there will be some that resemble the typical occupied orbitals (the normal 3d orbitals) and some that somewhat resemble unbound states: polarization orbitals, and split orbitals, etc. To be honest: the basis sets and how additional orbitals are added are poorly described in the documentation: it is assumed that the user is already familiar with these concepts as it follows same practices as in the quantum chemistry community and many other software packages like Siesta. You can see plots of the basis set radial functions in basis set selection window.

For traditional Hubbard-U you are only interested in the occupied 3d orbital, so only the first U value is relevant. However, in the Nanolab GUI it is currently not possible to only calculate and use U for specific orbitals: one will always select all orbitals with a specific angular momentum. However, the U value of the additional 3d-like orbitals should not matter much in a DFT+U calculation as those orbitals will typically have very low occupation. It is possible to not include these orbitals in the Hubbard-U description in a Python script. You can send your workflow to script and see the basis set definition for how it's actually done and simply remove the selection of the second 3d-like orbital.

Also: I recommend using the Dual Hubbard term as it uses the Mulliken orbital charges for occupations which is a more local projection model than the Onsite and is a better fit for the U values calculated with the LSCC method, as it assumes projection on the actual orbitals (a projection method which we unfortunately currently do not support)

Offline Fan Xiaozheng

  • New QuantumATK user
  • *
  • Posts: 2
  • Country: cn
  • Reputation: 0
    • View Profile
Re: Analysis of calculated LSCC U values results
« Reply #2 on: June 17, 2024, 14:49 »
Thank you!