[DFTB-Plus-User] Self-interaction in DFTB3

[Luca Babetto]

Dear all,

I need to carry out energy difference calculations between states differing by the number of electrons in the same system (i.e., redox potential calculations). I am aware that in DFTB methods it is necessary to explicitly account for the electron self-interaction error. For example, in this paper (https://doi.org/10.1021/acs.jpca.0c05052) Grimme's group determined an (empirical) self-interaction energy of 4.846 V for the electron in the GFN2-xTB Hamiltonian. I could not find any such paper for self-interaction energies for the SCC-DFTB Hamiltonian implemented in DFTB+. I'm specifically interested in the SCC-DFTB3 Hamiltonian with 3OB-3-1 Slater-Koster parameters.

Is anyone aware of papers treating this aspect? If not, is there a way of "measuring" the self-interaction energy for DFTB methods, maybe comparing them to high-level calculations such as CCSD(T), or maybe even Hartree-Fock, since in theory it should not suffer from self-interaction errors?

Thank you kindly for your responses

Luca
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.zfn.uni-bremen.de/pipermail/dftb-plus-user/attachments/20220923/c8163dae/attachment.htm>