DFT+U in CP2K 2024.3

[Sachin Ramesh]

Hello everyone, 

As part of an ongoing study, I am investigating how the electronic energy levels of the MgO(001) surface change when different Hubbard U corrections are applied to Mg and O orbitals (both occupied and unoccupied states).

In CP2K (2024.3), when a U_eff value of 1.5 eV is applied to the Mg-3s and Mg-2s orbitals, the calculated:

* HOMO (VBM) value, and
* HOMO–LUMO (band-gap) value

turn out to be identical.

However, performing similar calculations in Quantum ESPRESSO gives different values for the VBM and the band gap. I would like to understand whether this behavior arises from:
1. differences in the DFT+U implementations/projector definitions between CP2K and Quantum ESPRESSO, or
2. a possible issue in my CP2K setup or interpretation of the results.

I have attached the corresponding CP2K input and output files.

Any insights would be greatly appreciated.

Sachin 

[Krack, Matthias]

Hi Sachin

For QS, the U(eff) value for the Mg s orbitals is applied to all its s orbitals, i.e. to the 2s and 3s concurrently, which explains your results. The N value is ignored (as described) unless it is a PWDFT run using the SIRIUS library.

The orbital occupations calculated with CP2K usually differ from the ones obtained with PW codes and thus the appropriate U(eff) values are also different.

HTH

Matthias

--
You received this message because you are subscribed to the Google Groups "cp2k" group.
To unsubscribe from this group and stop receiving emails from it, send an email to cp2k+uns...@googlegroups.com.
To view this discussion visit https://groups.google.com/d/msgid/cp2k/8d90223a-1862-4801-b6b0-e371b862c32dn%40googlegroups.com.