Synthesis and characterization of a high-valent heterobimetallic [Cu^III(μ-O)₂Ni^III]²⁺ core involving nucleophilic oxo groups

Abstract

A heterobimetallic CuNi bis(μ-oxo) diamond core is shown to possess nucleophilic oxo groups, and has been demonstrated for the first time as a viable intermediate during the deformylation of fatty aldehydes by cyanobacterial aldehyde decarbonylase.

The study of heterobimetallic dioxygen complexes is interesting with regard to their potential in modeling the structures and reactivities of metalloenzymes containing two different metal ions at their active sites.^[1] Moreover, such investigations may lead to the discovery of novel species exhibiting alternative reactivity patterns with respect to their symmetric counterparts. However, in contrast to the numerous reports of biomimetic synthetic dinuclear complexes involving homometallic dioxygen cores,^[2] reports on mixed-metal [MO₂M′]ⁿ⁺ cores are relatively rare,^[3] presumably because of intrinsic challenges associated with their preparation by the methods usually used to obtain their symmetric analogues. A limited number of heterobimetallic dioxygen complexes has been made in the past, either by the reaction of heterobimetallic M-M′ precursors with dioxygen, or by reaction of a well-defined metal-dioxygen subunit with a reducing heterometal complex.^[3a] The latter strategy has also been used recently by some of us^[3d] to synthesize and structurally characterize the first example of a heterodinuclear peroxo complex, Ni(μ-O₂)K, via one - electron reduction of a room- temperature stable nickel-superoxo complex [L₁Ni^II(O₂)] (1; L₁=[HC(CMeNC₆H₃(ⁱPr)₂)₂])^[4] with elemental potassium. Furthermore, replacement of the potassium ion in the [Ni(μ-O₂)K]ⁿ⁺ core by Zn²⁺ or Fe⁺ ions initiated peroxido O-O bond scission and subsequent H -atom abstraction to generate the heterobimetallic bis(μ-hydroxo) Ni-Zn^[3d] or alkoxo hydroxido Ni-Fe^[3e] complexes, respectively. Theoretical studies^[3d,e] suggested the involvement of heterobimetallic Ni(μ-O)₂M′ (M′ = Zn, Fe) cores in the H-atom abstraction process; their electronic structures and reactivities were predicted to be strongly dependent on the nature of the second metal ion. Unfortunately, these intermediates could not be directly detected by spectroscopic methods and hence the predictions of the theoretical calculations could not be experimentally verified.

We now demonstrate the low temperature trapping of the proposed heterobimetallic bis(μ-oxo) intermediate during the one-electron reduction of 1 using a Cu(I) triamine complex, [Cu(MeAN)](BF₄) [2- BF₄; MeAN = N,N,N′,N′,N′- pentamethyl-dipropylenetriamine],^[5] as a reductant. We report the synthesis of the [L₁Ni^III(μ-O)₂Cu^III(MeAN)]⁺ complex 3, in nearly quantitative yield, and its characterization with optical, electron paramagnetic resonance (EPR), resonance Raman (rRaman) and X-ray absorption spectroscopies (XAS), as well as electrospray ionization mass spectrometry (ESI-MS), reactivity studies, and density functional theory (DFT) calculations. Interestingly, the oxo groups of 3 act as nucleophiles, in sharp contrast to the electrophilic oxo groups of the well characterized homometallic [Ni₂(μ-O)₂]²⁺ and [Cu₂(μ-O)₂]²⁺ analogues. In fact, complex 3 represents the only example of a high-valent^[6] bis(μ-oxo)dimetal core involving nucleophilic oxo groups that can perform deformylation of aldehydes. Until this report, only metal - bound peroxides^[7] were believed to be sufficiently nucleophilic to attack aldehydes leading to production of formate and oxidized coproducts. This study, therefore, underlines the importance of subtle electronic changes in the reactivity of the biologically relevant metal-dioxygen intermediates.

Combination of equimolar amounts of 1 and 2-BF₄ at −90 °C in CH₂Cl₂ leads to the immediate generation of a metastable intermediate 3 (t_1/2 = 900 sec at −50 °C) with absorption maxima λ_max (ε_max; M⁻¹cm⁻¹) centered at 300 (20000), 370 (13500), and 895 (5000) nm (Fig. 1). The electrospray mass spectrum (Fig. S1) of 3 exhibits the most prominent peak at a mass-charge-ratio (m/z) of 771.38, with a mass and isotope distribution pattern corresponding to [L₁Ni(O)₂Cu(MeAN)]⁺. When the reaction is carried out with ¹⁸O enriched 1, a mass peak corresponding to [L₁Ni(¹⁸O)₂Cu(MeAN)]⁺ appears at m/z of 775.38, thereby revealing the presence of two oxygen atoms in 3. Although, we are currently unable to assign the low-energy band in the absorption spectrum of 3, the similarity of the features at λ_max = 300–400 nm to ones typical of homometallic [Ni₂(μ-O)₂]²⁺ and [Cu₂(μ-O)₂]²⁺ cores^[2] may suggest an analogous formation of a bis μ-oxo core in 3. Indeed, the rRaman (Fig. 1) spectrum of 3 using 413 nm laser excitation in resonance with the 370 nm transition supports the view of a [Ni(μ-O)₂Cu]²⁺ structure. The band at 625 cm⁻¹ of 3 [ν(3-¹⁶O-¹⁶O)] shifts down to 595 cm⁻¹ [ν(3-¹⁸O-¹⁸O)] in the corresponding ¹⁸O₂-substituted complex 3. Furthermore, single oxygen-isotope labeling experiments^[8] show an additional intermediate band at 615 cm⁻¹ corresponding to the vibration of the partially labelled species [ν(3-¹⁶O-¹⁸O)]. In analogy to the well-studied [M₂(μ-O)₂]²⁺ (M= Fe, Ni, Cu) systems, for which similar rRaman signatures and oxygen isotope effects were reported,^[9] we assign the rRaman feature at 625 cm⁻¹ to a tetra-atomic vibration (Fig. S2) mode of the [Ni(μ-O)₂Cu]²⁺ core. This assignment is also supported by DFT calculations (see below). Furthermore, the fact that the frequency of ν(3-¹⁶O-¹⁸O) is not halfway between ν(3-¹⁶O-¹⁶O) and ν(3-¹⁸O-¹⁸O), and the lack of isotope-sensitive bands in the 700 – 1200 cm⁻¹ range, strongly argue against the presence of any peroxo or superoxo species.

Figure 1.

Top: Absorption spectra of 2–BF₄ (solid trace), 1 (dash–dotted trace), 3 (dotted trace) in CH₂Cl₂ at −90 °C. Inset: X–band EPR spectrum of 3 in CH₂Cl₂ (0.47 mM) at 10 K (frequency 9.470251 GHz; power 0.05 mW; modulation 0.5 mT). Bottom: Calculated (DFT, see SI for details) and experimental (CH₂Cl₂; −90 °C; 413 nm excitation) rRaman spectra of 3–¹⁶O₂ (solid trace), 3–¹⁸O₂ (dotted trace), and 3–¹⁶O¹⁸O (dash-dotted trace). Bands originating from the solvent are marked by asterisks.

X-ray absorption spectroscopic studies are performed at Cu and Ni K-edges in order to directly probe the metal oxidation states in 3. Figure 2A shows a comparison of the Cu K-edge spectrum of 3 with its Cu(I) precursor and the previously reported [(AN)Cu^II(NHTs)] [4, AN = 3,3′-iminobis(N,N - dimethyl- propylamine; Ts= tosyl]^[10] complex. A progressive blue shift in the edge energy from 2-BF₄ to 4 to 3 is in accord with a stepwise increase in the oxidation state of copper in the series. Most importantly, a 2 eV blueshift, from 8978 to 8980 eV, and a broadening of the pre-edge are observed from 4 to 3, which strongly suggest^[11] the presence of a Cu(III) ion in 3. In addition, the Ni K-edge data (Fig. 2B), which displays a subtle blueshift and reduced intensity for the pre-edge, as well as a broader blue - shift of the overall edge in 3 relative to its nickel(II) -superoxide precursor, are consistent with a nickel-based oxidation from 1 to 3.^[12] The X-band EPR spectrum (Fig. 1) of 3 is also found to be in good agreement with the [L₁Ni^III(μ - O)₂Cu^III(MeAN)]⁺ assignment and is dominated by a typical S=1/2 Ni(III) rhombic signal (g_x = 2.01, g_y = 2.14, g_z = 2.41) that lacks the hyperfine splitting characteristic of Cu(II),^[13] and corresponds, by integration, to 90 % of the amount of 2-BF₄ added. The absence of hyperfine splitting renders copper in the diamagnetic 3d⁸ low-spin state, which moreover reveals planar coordination.

Figure 2.

A: Normalized Cu K–edge XANES spectra for 2–BF₄ (dotted line), 3 (dash-dotted line) and 4 (solid line). The inset depicts an expansion of the pre–edge region for 3 and 4. B: Normalized Ni Kedge XANES spectra for 1 (solid line) and 3 (dash-dotted line). The inset shows an expansion of the pre–edge region. C. Fourier–transformed Cu K–edge EXAFS spectra of 2–BF₄ and 3 (Experiment: dotted line; simulation: bold line). D. Fourier–transformed Ni K–edge EXAFS spectra of 1 and 3. For the EXAFS data on a wavevector scale before calculation of the Fourier transform see Fig. S3.

Extended X-ray absorption fine structure (EXAFS) analysis reveals further structural details (Figs. 2C; 2D and S3; Table S1). For 1 a shell of 4 N/O scatterers at 1.84 Å is obtained from Ni EXAFS, in good agreement with the molecular structure determined previously by X-ray crystallography (Table S2).^[4] On the other hand, a fit to the data for 3 requires two subshells of N/O scatterers at 1.79 Å (assigned to the two Ni^III-O units) and 1.88 Å (assigned to the two Ni^III-N units), consistent with the structure derived from DFT calculations (Table S2). In addition, Ni-EXAFS of 3 shows an additional peak at 2.81 Å corresponding to the Cu scatterer, thereby strongly supporting the presence of a heterodinuclear Ni···Cu center in 3. Further support for including a Ni···Cu distance at 2.81 Å comes from Cu EXAFS, which also detects the Ni-scatterer at 2.81 Å in 3, but not in the mononuclear Cu(I) precursor. Furthermore, the best fit of the Cu EXAFS of 3 can be obtained with two subshells, each containing two O/N scatterers at 1.80 Å and 2.00 Å, thereby supporting a four coordinate Cu center in 3. It is important to note that a fit with two O/N scatterers at 1.88 Å and three O/N scatterers at 2.00 Å gave a high Debye-Waller factor, which together with the DFT calculations (see below) preclude the presence of a five-coordinate Cu center in 3 (Table S1).

DFT calculations (see SI for details) on 3 yield a minimum energy for a C₁ structure (Scheme 1; Fig. S4; Table S3) with a four coordinate planar geometry at both the copper and the nickel centers, and geometrical parameters in excellent agreement with that obtained from XAS studies (Table S2). In the calculated structure MeAN acts as a bidentate ligand, with one of the terminal tertiary amine nitrogens remaining out of the coordination sphere of Cu. The ground state is calculated to be S=1/2, consistent with the EPR studies. The unpaired spin is distributed such that the nickel center has 0.7 spin, whilst the copper and oxygen centers carry an average spin density of only 0.01 and 0.06, respectively (Table S4). These results reveal the presence of Ni(III) (S=1/2; 3d⁷)and Cu(III) (S=0; 3d⁸) centers in 3, thereby, supporting a [L₁Ni^III(μ-O)₂Cu^III(MeAN)]⁺ assignment. Vibrational analysis on the DFT-optimized structure of 3 in the S=1/2 ground state predicts a strong [Ni^III(μ-O)₂Cu^III]²⁺ core vibrational mode at 640 cm⁻¹ (Figs. S5–S6; Table S5). This frequency and in particular the calculated ¹⁸O/¹⁶O isotopic shifts are in excellent agreement with the experimental data (Fig. 1). Thus, the spectroscopic results altogether validate the calculated molecular and electronic structures of 3.

Scheme 1.

Formation, DFT calculated structure, and products of the reactions of 3 with different substrates.

The oxo groups in high-valent [M₂(μ-O)₂]ⁿ⁺ (M= Fe, Co, Ni, Cu) cores act as strong electrophiles and are known to be typically unreactive towards other electrophiles.^[2] Consistent with these previous observations, addition of cyclohexane carboxaldehyde (CCA) or 2- phenylpropionaldehyde (2- PPA) to [(AN)₂Cu₂(μ -O)₂]^{2+ [5,14]} or [(L₁)₂Ni₂(μ - O)₂],^[15] which are the homometallic analogues of 3, at −90 °C do not change the UV-Vis spectra. In contrast, similar reactions with 3 result in instantaneous bleaching and disappearance of the bands at 370 and 895 nm in the UV-vis absorption spectrum (Fig. S7). Product analysis of the reaction mixture reveals the formation of the deformylated^[16] products in 65 –90% yields (Scheme 1; Table S6). The reactivity of 3 is further investigated using benzoyl chloride at −50 °C, in order to confirm the nucleophilic property of 3. A pseudo-first order decay of the absorption feature at 895 nm and the subsequent formation of benzoic acid (Fig. S8) were observed upon adding benzoyl chloride to a preformed solution of 3 at −50 °C. The rate constant increases proportionally with the substrate concentration, affording a second-order rate constant, k₂, of 0.11 M ⁻¹s ⁻¹ at −50 °C. Additional mechanistic studies with para-substituted benzoyl chloride (para-Y -C₇H₄OCl; Y = OMe, Me, H, CN) reveal a good linear correlation of the rate to the σ_P⁺ values of the para-substituents. A positive Hammett ρ value of 2.5 (Fig. S9) is obtained, which further supports the nucleophilic character of the [L₁Ni^III(μ-O)₂Cu^III(MeAN)]⁺ core in the oxidation of benzoyl chloride. To the best of our knowledge, 3 represents the only example of a high-valent^[6] bis(μ-oxo) dimetal complex involving nucleophilic oxo groups. To investigate in how far the dangling amine group assists reactivity or is even responsible for the observed nucleophilicity, experiments with N,N,N′,N′-tetramethylethylenediamine (TMEDA), as a co-ligand for Cu instead of MeAN were envisaged. Unfortunately, unlike the 2-BF₄ Cu^I(TMEDA) complex has been reported to be unstable,²¹ so that it had to be generated in-situ prior to the reaction with 1. Preliminary experiments with the resulting complex L₁Ni^III(μ-O)₂Cu^III(TMEDA)]⁺ (5, Scheme S1),²² structurally similar to 3, but lacking any dangling amine group, suggest that the latter is not the origin of the nucleophilic behavior of the oxido ligands within 3: 5 reacts with benzoyl chloride at −90 °C in acetone at a rate much faster than 3 (Fig. S12). In this context it is also worth mentioning that [(AN)₂Cu₂(μ-O)₂]²⁺ may also be expected to have dangling or at least loose amine arms and is an electrophilic oxidant.

Complex 3 also performs hydrogen atom abstraction from the C - H (or O - H) bonds of 1,3-cyclohexadiene (CHD), 9,10 -dihydroanthracene (DHA), 2, 4-di-tert-butylphenol (2,4-DTBP), and 2,6-di-tert-butylphenol (2,6-DTBP); the second order rate constants derived from these studies in CH₂Cl₂ solution are listed in Table S6. Surprisingly, the rates for reactions with DHA and 2,6-DTBP are found to be significantly lower than those determined for CHD and 2,4- DTBP, respectively. Since DHA/CHD and 2,4-DTBP/2,6-DTBP have comparable C-H and O-H bond dissociation energies,^[17] the large rate differences observed for 3 suggest that the β-diketiminate ligand impedes access of the bulkier DHA or 2,6-DTBP to the [L₁Ni^III(μ-O)₂Cu^III(MeAN)]⁺ core. Such sterically-derived mitigation of reactivity has been observed previously for other β-diketiminate -ligated systems.^[18] No reaction of 3 with triphenyl phosphine has been observed.

A number of enzymes^{[19, 20]} use a bimetallic active site to bind and activate dioxygen to form reactive species involving superoxide, peroxide, and metal oxide cores, which are then responsible for carrying out a variety of electrophilic and nucleophilic oxidation reactions. While nucleophilic reactions are typically carried out by superoxides or peroxides, electrophilic reactions are attributed to metal oxides. For example, the [Cu₂(μ-O)₂]²⁺ and [Fe₂(μ-O)₂]⁴⁺ cores have been considered as active oxidants in the hydroxylation of phenol and methane by tyrosinase and soluble methane monooxygenase, respectively.^[2,19] On the other hand, a homo- or heterodinuclear peroxo core^[20] has been proposed as the active species responsible for the deformylation of fatty aldehydes by cyanobacterial aldehyde decarbonylase (AD); the metal centers in the dimetal-cofactor of AD have yet to be identified. In this work, we have now reported the synthesis and spectroscopic characterization of a high-valent heterodinuclear [Ni^III(μ-O)₂Cu^III]²⁺ core, which can act as nucleophile to attack (among other electrophiles) aldehydes leading to production of formate and oxidized coproducts. The nucleophilic property of 3, which is in sharp contrast to the electrophilicity of oxo groups in the homodinuclear analogues containing [Cu₂(μ-O)₂]²⁺ or [Ni₂(μ-O)₂]²⁺ cores, may be related to a reduced covalency of the metal-oxygen bonds owing to the mismatch of the bonding orbitals in the asymmetric [Ni^III(μ - O)₂Cu^III]²⁺ core. Further experimental and theoretical studies are, however, warranted to understand the origin of the nucleophilic behavior of 3. These studies are ongoing in our team.

In conclusion, we have reported the synthesis and spectroscopic characterization of a mixed-metal Ni-Cu bis (μ-oxo) complex 3 that is potent in the deformylation of aldehydes. Whether or not, a similar asymmetric bis(μ–oxo) core also acts as an active species in the catalytic cycle of cyanobacterial AD is now a particularly intriguing question.