Spin-state energetics of metallocenes: How do best wave function and density functional theory results compare with the experimental data?

Abstract

We benchmark the accuracy of quantum-chemical methods, including wave function theory methods [coupled cluster theory at the CCSD(T) level, multicongurational perturbation-theory (CASPT2, NEVPT2) and internally-contracted multireference conguration interaction (MRCI)] and...

Effects of 2,6‐Dichlorophenyl Substituents on the Coordination Chemistry of Pyridine Dipyrrolide Iron Complexes

“…A further complication is that the doubleshell effect has not been captured by our CAS(10,8) reference. The inclusion of the 4d (often called 3d’) shell in the zeroth‐order wavefunction is important to account for radial electron correlation, a phenomenon which is not handled by perturbation theory but has been shown to improve the accuracy of electronic structures and spin‐state energetics [30,32–36] . Inclusion of these high‐energy orbitals has also proven critical to obtaining more accurate electronic descriptions for other systems with M=E (M=Mn or Fe; E=O or N) multiple bonds [35,36] .…”

Zeolites at the Molecular Level: What Can Be Learned from Molecular Modeling

“…This conclusion might appear unexpected in view of the presumed multireference character of TMIs (and especially those containing the nitrosyl ligand). However, the benchmarking results for several different TMI complexes, where quantitative experimental data of spin-state energetics are available, have so far confirmed the high accuracy of the CCSD(T) method in predicting spin-state splittings [111,112]. This is parallel to other known examples in which the CCSD(T) method was able to correctly describe the structure and energetics of transition metal systems [121], for instance the bond energies in transition metal diatomics [130].…”

“…However, certain energetic properties of TMI sites with potential relevance in catalysis, such as metal-ligand bond energies and spin-state splittings, are known to be computationally much more problematic [105][106][107][108]. For such challenging electronic properties, not only are the computed results strongly dependent on the choice of functional (with discrepancies of~50 kJ mol −1 not being rare), but the optimal choice of functional is also system dependent (in particular, it is dependent on the TMI oxidation state and its coordination environment, including ligand field strength or binding mode), so that finding functionals that perform universally well for different systems is currently problematic [109][110][111][112]. This also means that attention is required when extrapolating the conclusions of benchmark studies in the literature to other systems, not included in testing set [100].…”

“…However, in lowsymmetry environments, such as in zeolite frameworks, electronic degeneracy is usually lifted, meaning that the above problem is of little practical relevance. Moreover, even in the presence of electronic degeneracy, a single-determinant method can still represent (at least qualitatively correctly) a single component of the degenerate state [112]. Even more complicated electronic structures, such as antiferromagnetically coupled multi-center TMI complexes, can be described reasonably well using spin-polarized (unrestricted) DFT, leading to broken-symmetry (BS) solutions (see Refs.…”

“…This is parallel to other known examples in which the CCSD(T) method was able to correctly describe the structure and energetics of transition metal systems [121], for instance the bond energies in transition metal diatomics [130]. More benchmarking is needed for the problem of spin-state splittings to establish definite conclusions, but for relatively simple mononuclear TMI species, where suitable reference data can be derived from the experimental data, the performance of CCSD(T) is appealing, especially when contrasted with the much greater errors obtained not only with DFT methods, but even with such "respected" methods as CASPT2, NEVPT2 and multireference configuration interaction when using the standard active space [111,112]. The systematic bias of CASPT2 in favor of higher spin states is partly mitigated [111,112] by the CASPT2/CC approach of Phung et al, i.e., the CASPT2 description of valence correlation combined with the CCSD(T) description of metal outer-core correlation [131].…”