Abstract: The accurate description of spin-state energetics of transition-metal complexes represents a great challenge for electronic structure ab initio methods.This challenge stems from the lack of error cancellation when computing energy differences using approximate electronic structure methods between spin states exhibiting di erent types and amounts of electronic correlations. In DFT for example, local and semilocal functionals (e.g. LDA and GGA), systematically overstabilise low spin state while Hartree-Fock, which only treat exchange correlations, overstabilises HS. Thus, depending on the system under investigation, a balanced description of the spin energetics can be achieved by adopting a global hybrid with a suitable choice of the amount of exact exchange that may depend on the system under study. In this work, we propose an alternative method which is computationally more efficient compared to the hybrids and does not require a system-dependent parametrization. The premise of our approach is that the calculation of adiabatic energy differences is largely affected by errors in the electronic density. Thus we employ a non-self consistent density-corrected DFT scheme where the corrected density is the DFT+U density with a linear-response self-consistent U [1,2]. The total energy is thus evaluated on the PBE+U density by employing the PBE functional, i.e. without inclusion of the Hubbard term and we name this approach PBE[U]. In my talk I will explain the rationale behind this approach and will show that our results are in excellent agreement with coupled-cluster-corrected multicongurational calculations, i.e. CASPT2/CC, and with experiments [2,3].

 

[1] Mariano, Vlaisavljevich, Poloni J. Chem. Theory Comput., 16, 6755-6762 (2020); 10.1021/acs.jctc.0c00628

[2] Mariano, Vlaisavljevich, Poloni J. Chem. Theory Comput. (2021) in press, 10.1021/acs.jctc.1c00034

[3] Phung, Feldt, Harvey, Pierloot J. Chem. Theory Comput., 14, 2446-2455 (2018); 10.1021/acs.jctc.8b00057.