Correlation of Electronic and Geometric Structure in Mononuclear Copper(II) Superoxide Complexes

Abstract

The geometry of mononuclear copper(II) superoxide complexes has been shown to determine their ground state where side-on bonding leads to a singlet ground state and end-on complexes have triplet ground states. In apparent contrast to this trend, the recently synthesized (HIPT₃tren)Cu^II–O₂^•− (1) was proposed to have an end-on geometry and a singlet ground state. However, re-examination of 1 with resonance Raman (rR), magnetic circular dichroism (MCD), and ²H NMR spectroscopy indicates that 1 is in fact an end-on superoxide species with a triplet ground state that results from the single Cu^II–O₂^•− bonding interaction being weaker than the spin pairing energy.

The activation of dioxygen at a single copper(I) site to produce a copper(II) superoxide appears to be essential for the enzymatic function of the non-coupled binuclear copper monooxygenases (PHM, DβM, and TβM)¹ and has been recently proposed for the copper-dependent polysaccharide monooxygenases.² Since isolation of a discrete copper(II) superoxide species in enzymatic systems has been limited to a crystal structure of PHM,^1c insight into the electronic structure and bonding in copper(II) superoxide species has been derived from model complexes.

In these synthetic complexes, two structures have been observed: superoxide coordinated to Cu(II) in a side-on (η²) or end-on (η¹) binding mode. The side-on superoxide species, structurally characterized in a tris(pyrazolyl)borate model complex,³ has an O–O stretching frequency of 1043 cm⁻¹ and a copper(II) X-ray absorption pre-edge feature at ~8979 eV.⁴ Magnetic susceptibility measurements indicated that the side-on superoxide has a singlet ground state.^4a This ground state is a direct result of the two strong Cu–O bonds of the side-on geometry (1.84 Å) causing the HOMO/LUMO splitting to be larger than the spin pairing energy.^4a This results in a doubly occupied HOMO that is a superoxide-based π* orbital that is vertical to the Cu–O₂ plane (π*_v) and an empty Cu d_x2–y2 LUMO that is antibonding with the filled π* orbital that forms a very covalent σ bond with the copper (π*_σ + αd, Figure 1, left).

Figure 1.

Molecular orbital diagram of side-on and end-on copper(II)-superoxide bonding.

An end-on superoxide species, which has been structurally characterized in (TMG₃tren)Cu^II–O₂^•−,⁵ shares similar superoxide spectral features with the side-on complex (ν_O–O of 1120 cm⁻¹).⁶ However, both NMR⁷ and variable-temperature variable-field magnetic circular dichroism (VTVH-MCD)^6b spectroscopies indicate that the end-on superoxide possesses a triplet ground state, in contrast to the singlet ground state of the side-on isomer.^4a For the end-on superoxide, the single Cu–O bond (1.93 Å)⁵ is significantly weaker than the two bonds in the side-on complex and hence the bonding/antibonding interaction between the superoxide π*_σ and the Cu d orbital is unable to overcome the spin pairing energy.^6b This results in a triplet ground state with two singly occupied, orthogonal orbitals: a superoxide, π*_v orbital and a copper, d_z² orbital (that is antibonding with the π*_σ orbital, Figure 1 right).

Recently, Itoh and co-workers synthesized an end-on superoxide adduct 1,⁸ (HIPT₃tren)Cu^II–O₂^•− (HIPT = hexaisopropylterphenyl, Scheme 1) which features a tren ligand platform like (TMG₃tren)Cu^II–O₂^•− (TMG = tetramethylguanidino). Although 1 has similar vibrational features to (TMG₃tren)Cu^II–O₂^•− (ν_O–O = 1095 cm⁻¹ from rR excitation into the O₂^•− → Cu^II charge transfer (CT) at 23,000 cm⁻¹), they proposed that 1 has a singlet ground state due to the observation of chemical shifts between 8-0 ppm in its ¹H-NMR spectrum.⁸ This intriguing result prompted us to probe the electronic structure of 1 in the context of the end-on triplet/side-on singlet correlation described in Figure 1.

Scheme 1.

HIPT₃tren copper(II) superoxo complex 1.

To probe the geometric structure of 1, resonance Raman (rR) spectra were collected on samples prepared with ¹⁶O₂, ¹⁸O₂, and ¹⁶O–¹⁸O mixed isotope dioxygen (^16,18O₂, Figure S1). Two ν_Cu–O were observed in the ^16,18O₂ spectrum, which have the same energy as the ν_Cu–O in the ¹⁶O₂ and ¹⁸O₂ spectra, indicating an asymmetric (i.e., end-on) coordination mode of superoxide.⁹ The ground state and electronic structure of 1 were then probed with VTVH MCD spectroscopy. A 9:1 mixture of n-propanol and acetone was selected as an appropriate solvent since it formed an optical glass without affecting the absorption (Abs) or rR spectra of 1 (Figures S1–S2).

The Abs and low temperature (5.0 K) MCD spectra of oxygenated samples of (HIPT₃tren)Cu^I require five Gaussian bands to fit both experimental spectra up to and including the O₂^•− → Cu(II) CT transition (band 5) (Figure 2, Table S1). Agreement between the transition energies in both spectra indicates that the MCD intensity in these bands can be assigned to 1. The MCD spectrum of 1 is also similar to that previously obtained for (TMG₃tren)Cu^II–O₂^•−:^6b a derivative-shaped pseudo-A (band 2 and 3), a negative transition to higher energy (band 4), and a positive transition resulting from the O₂^•− → Cu^II CT transition (band 5). However, bands 2–4 are shifted up in energy by > 3,000 cm⁻¹ in 1 relative to (TMG₃tren)Cu^II–O₂^•− indicating HIPT₃tren possesses a weaker ligand field.^6b An additional band (band 0 in pink) is required to fit the lowest energy feature in the MCD spectrum. While the relative intensity of bands 1–5 were constant between multiple samples, the intensity of band 0 was variable (Figure S3) and is assigned as an S = ½ contaminant since its VTVH-MCD isotherms overlay (Figure S4). In contrast, VTVH-MCD isotherms collected on bands 1–4 (Figures 3 and S5) show non-overlapping (nesting) behavior, which results from zero-field splitting (ZFS). This ZFS requires that 1 has an S > ½ ground state. The saturation magnetization curves for bands 1–5 fit to the spin Hamiltonian for an S = 1 system with axial (D) and rhombic (E) zero field parameters of D = +3.0 and 0.22 ≥ E/D ≥ 0.12 (see Supporting Information for details).

Figure 2.

Absorption (183 K, top) and MCD (5.0 K, bottom) spectra of 1. Band 0 (*) indicates an S = ½ contaminant.

Figure 3.

VTVH-MCD isotherms and fit (D = +3.0 and E/D = 0.17) for band 2 collected at 14,040 cm⁻¹.

While the nesting observed in the VTVH-MCD isotherms require 1 to have a paramagnetic ground state (S = 1), reference 8 reported the ground state to be diamagnetic (S = 0) from NMR spectroscopy. Therefore the NMR spectroscopy of 1 was re-examined. The ¹H NMR spectra of oxygen-saturated samples of (HIPT₃tren)Cu^I in acetone were indistinguishable from the original (HIPT₃tren)Cu^I at 183 K (−90 °C) (Figure S7). However, the previously determined equilibrium constant for O₂ binding indicates that 1 is only approximately half oxygenated under these conditions.⁸ Since no new chemical shifts were initially observed in the ¹H NMR spectra, deuterium was incorporated into a single position on each HIPT substituent (see Scheme 1). ²H NMR spectra of oxygen-saturated samples of (d₃-HIPT₃tren)Cu^I show two resonances with approximately equal intensity: one corresponding to (d₃-HIPT₃tren)Cu^I and a new paramagnetically-shifted resonance at −24 ppm (Figure 4), which is assigned to 1. Addition of CO to 1 resulted in a single chemical shift corresponding to (d₃-HIPT₃tren)Cu^ICO, indicating that the resonance at −24 ppm did not arise from an S = 1/2 contaminant. Reanalysis of the ¹H NMR spectra indicated the presence of a broad, weak resonance at −24 ppm (Figure S7–S8). Similarly, density functional theory (DFT) calculations on a truncated model of 1 predict a triplet ground state, with both closed-shell singlet (+23.2 kcal/mol) and broken-symmetry singlet (+13.0 kcal/mol) states at significantly higher energy (Figure S9–S10, Table S3–S4), in agreement with the experimental data presented above.

Figure 4.

²H-NMR spectra of 3 mM d₃-HIPT₃tren derivatives in acetone at 183 K (–90 °C). Cu^I (gold, 7.3 ppm), Cu^II–O₂^•– (1) (green, 7.3 and −24 ppm), Cu^I-CO (black, 7.4 ppm). # indicates d₁₂-SiMe₄ (0 ppm) and * natural abundance deuterium in acetone (2.05 ppm).

In summary, rR spectroscopy with mixed isotope dioxygen confirms the end-on superoxide geometry of 1 while VTVH-MCD and NMR spectroscopy determine that 1 has a triplet ground state. While the Cu–O bond in 1 is stronger than (TMG₃tren)Cu^II–O₂^•−, this single Cu–O bond is unable to overcome the pairing energy required to form a singlet ground state (Figure 1). To date, only a side-on superoxide complex with two Cu–O bonds has the required bond strength to overcome the spin pairing energy resulting in a singlet ground state.^4a These results suggest that the end-on superoxide intermediate proposed in the enzymatic systems^1c should also have a triplet ground state. This prediction awaits experimental evaluation.