Spin‐State Energetics of Fe Complexes from an Optimally Tuned Range‐Separated Hybrid Functional

Abstract

We assess the performance of the optimally tuned range-separated hybrid (OT-RSH) functional approach in predicting the ground-state electronic configuration and spin-state energetics of complexes that can potentially exhibit multiple spin configurations. To that end, we investigate eight iron complexes: four spin-crossover complexes, for which reference data from other approximate density functionals are available, and four smaller complexes, for which reference ab initio data are available. We show that the spin-state energetics are mostly governed by the percentage of short-range exact exchange and are only weakly influenced by the choice of the range-separation parameter. However, the electronic structure, especially the fundamental gap, is much more sensitive to the range-separation parameter. We further find that correct prediction of the ground state in spin-crossover compounds requires a reduction in the amount of short-range exact exchange, likely owing to a larger role of static correlation.

Magnetic configurations of open-shell molecules on metals: The case of CuPc and CoPc on silver

“…The HSE06 functional, which applies exact exchange of 25% in the short range, has been chosen as it has been shown to perform well for phthalocyanine molecules in the gas phase [42,43,70]. In addition, for transition-metal complexes it has been shown that the spin-state energetics depend mainly on the percentage of short-range exchange, while long-range exchange and the exact value of the range-separation parameter are of less importance for this property [57,71]. To check whether this is the case also for the Pc molecules considered here, we performed additional calculations with long-range corrected functionals, elaborated below.…”

“…Our gas-phase calculations are based on two methods: First, using the HSE06 functional, which has been shown to accurately describe the relative alignment of the electronic orbitals in neutral CuPc and CoPc molecules [28,43,95], and which has also been employed in the interface calculations discussed above. Second, we also perform gas-phase calculations of the CuPc and CoPc anions with an OT-RSH functional [71,80,81,96], which uses a system-dependent but nonempirical tuning of the range-separation parameter and can produce highly accurate valence-electron spectra for molecules. In particular, the excellent performance of OT-RSH has been demonstrated for metal Pcs [80,81] and for other metallorganic complexes [71,97].…”

“…Second, we also perform gas-phase calculations of the CuPc and CoPc anions with an OT-RSH functional [71,80,81,96], which uses a system-dependent but nonempirical tuning of the range-separation parameter and can produce highly accurate valence-electron spectra for molecules. In particular, the excellent performance of OT-RSH has been demonstrated for metal Pcs [80,81] and for other metallorganic complexes [71,97].…”

The effect of N-heterocyclic carbene units on the absorption spectra of Fe(ii) complexes: a challenge for theory

“…For the tuned LC-BLYP case, three different optimal pairs of parameters (α, ω) have been considered: no constant exact exchange (α=0.0) as it was predicted by analysis of fractional charges, as well as 15 and 20 % of exchange because those parameters have been proposed in earlier publications. 69,70 Taken CASPT2 as a reference, one can clearly see that the purely local BLYP drastically underestimates the energies of MLCT states but overestimates the MC states. Importantly, the energies of MC states predicted with the tuned LC-BLYP (α=0.0) almost coincide with those of BLYP, supporting the underlying assumption that at short interelectron distances the local BLYP functional dominates in Eq.…”

“…Compound spin-crossover compounds 35,37,68 . In a recent publication 69 , a value of α of 0.10-0.15 has been argued to be optimal for a series of iron spin-crossover compounds after analyzing the adiabatic energy difference between high-and low-spin states based on the comparison with the OPBE reference data 70 implying that the errors inherent to the OPBE functional may also influence this conclusion.…”