Reactivity of a Cobalt(III)-Hydroperoxo Complex in Electrophilic Reactions

Abstract

The reactivity of mononuclear metal-hydroperoxo adducts has fascinated researchers in many areas due to their diverse biological and catalytic processes. In this study, a mononuclear cobalt(III)-peroxo complex bearing a tetradentate macrocyclic ligand, [Co^III(Me₃-TPADP)(O₂)]⁺ (Me₃-TPADP = 3,6,9-trimethyl-3,6,9-triaza-1(2,6)-pyridinacyclodecaphane), was prepared by reacting [Co^II(Me₃-TPADP)(CH₃CN)₂]²⁺ with H₂O₂ in the presence of triethylamine. Upon protonation, the cobalt(III)-peroxo intermediate was converted into a cobalt(III)-hydroperoxo complex, [Co^III(Me₃-TPADP)(O₂H)(CH₃CN)]²⁺. The mononuclear cobalt(III)-peroxo and -hydroperoxo intermediates were characterized by a variety of physicochemical methods. Results of electrospray ionization mass spectrometry clearly show the transformation of the intermediates: the peak at m/z 339.2 assignable to the cobalt(III)-peroxo species disappears with concomitant growth of the peak at m/z 190.7 corresponding to the cobalt(III)-hydroperoxo complex (with bound CH₃CN). Isotope labeling experiments further support the existence of the cobalt(III)-peroxo and -hydroperoxo complexes. In particular, the O-O bond stretching frequency of the cobalt(III)-hydroperoxo complex was determined to be 851 cm⁻¹ for ¹⁶O₂H samples (803 cm⁻¹ for ¹⁸O₂H samples) and its Co-O vibrational energy was observed at 571 cm⁻¹ for ¹⁶O₂H samples (551 cm⁻¹ for ¹⁸O₂H samples; 568 cm⁻¹ for ¹⁶O₂²H samples) by resonance Raman spectroscopy. Reactivity studies performed with the cobalt(III)-peroxo and -hydroperoxo complexes in organic functionalizations reveal that the latter is capable of conducting oxygen atom transfer with an electrophilic character, whereas the former exhibits no oxygen atom transfer reactivity under the same reaction conditions. Alternatively, the cobalt(III)-hydroperoxo complex does not perform hydrogen atom transfer reactions, while analogous low-spin Fe(III)-hydroperoxo complexes are capable of this reactivity. Density function theory calculations indicate that this lack of reactivity is due to the high free energy cost of O-O bond homolysis that would be required to produce the hypothetical Co(IV)-oxo product.

Graphical abstract

A cobalt(III)-hydroperoxo complex was prepared by the protonation of a cobalt(III)-peroxo complex. Reactivity studies reveal that the cobalt(III)-hydroperoxo complex is capable of conducting oxygen atom transfer with an electrophilic character. Alternatively, the cobalt(III)-hydroperoxo complex does not perform hydrogen atom transfer reactions, while analogous low-spin Fe(III)-hydroperoxo complexes are capable of this reactivity. Density function theory calculations indicate that this lack of reactivity is due to the high free energy cost of O-O bond homolysis.

Introduction

Mononuclear metal-oxygen species such as metal-superoxo, -peroxo, -hydroperoxo and -oxo complexes have important roles as key intermediates involved in catalytic oxygenation reactions and biological oxidation reactions.^1–5 Heme and non-heme iron-oxygen intermediates have been invoked in the catalytic cycles of dioxygen activation by metalloenzymes.⁶ Among the iron-oxygen adducts, iron-hydroperoxo species have received considerable attention, since the intermediates have been implicated as reactive species in DNA cleavage activity of bleomycin, a glycopeptide that is effective as an antitumor drug in concert with iron and dioxygen.⁷

In the case of cobalt complexes, a large number of cobalt-O₂ species have been synthesized as models of dioxygen carrier proteins and as oxidants in organic functionalizations.^8,9 In models of biological oxygen carriers, Cavin et al., for example, have extensively examined the oxygen carrier properties of cobalt(II) complexes with SALEN ligands.¹⁰ In addition, the synthetic complexes have potential applications in dioxygen separation and storage.¹¹

Recent advances in the characterization and reactivity of cobalt-O₂ intermediates reveal that the cobalt-superoxo and -peroxo species are active oxidants in electrophilic and nucleophilic reactions.¹² A notable example is the formation of side-on cobalt(III)-peroxo complexes bearing a series of tetraazamacrocyclic ligands.^8a,8e The intermediates have nucleophilic character (e.g., aldehyde deformylation) toward organic substrates. The end-on cobalt(III)-superoxo species has been proposed as reactive species in electrophilic reactions (e.g., hydrogen atom transfer).¹³ Very recently, the reactivity of an end-on cobalt(II)-superoxo intermediate with a redox non-innocent ligand has been investigated in catalytic deformylation reactions.¹⁴

In contrast, there are few examples of cobalt(III)-hydroperoxo intermediates.¹⁵ Recently, the interconversion of cobalt(III)-peroxo and -hydroperoxo species via acid-base reactions has been reported where the cobalt(III)-hydroperoxo complex has shown reactivity in ligand oxidation (Scheme 1).¹⁶ Nevertheless, the reactivity of cobalt(III)-hydroperoxo species has been rarely explored in external substrate oxidation.¹⁷ Herein, we report the synthesis and characterization of a cobalt(III)-peroxo complex bearing a tetradentate macrocyclic ligand, 3,6,9-trimethyl-3,6,9-triaza-1(2,6)-pyridinacyclodecaphane (Me₃-TPADP) (Scheme 2). Upon protonation, the cobalt(III)-peroxo complex, [Co^III(Me₃- TPADP)(O₂)]⁺ (2), was converted into a cobalt(III)-hydroperoxo complex, [Co^III(Me₃-TPADP)(O₂H)(CH₃CN)]²⁺ (3). Complex 3 shows high reactivity in sulfoxidation reactions with electrophilic character, which was confirmed by a Hammett plot. Further, density function theory (DFT) calculations were applied to explain the reactivity difference between 3 and its hypothetical iron analogue in electrophilic reactions.

Scheme 1.

Acid-Base Reaction between the Metal-Peroxo and -Hydroperoxo Complexes.

Scheme 2.

Synthetic Procedures for Mononuclear Cobalt Complexes

Experimental Section

Materials

All chemicals obtained from Aldrich Chemical Co. were the best available purity and used without further purification unless otherwise indicated. Solvents were dried according to published procedures and distilled under Ar prior to use.¹⁸ H₂¹⁸O₂ (95% ¹⁸O-enriched, 2.2% H₂¹⁸O₂ in water) was purchased from ICON Services Inc. (Summit, NJ, USA).

Physical Methods

UV-vis spectra were recorded on a Hewlett Packard 8454 diode array spectrophotometer equipped with a UNISOKU Scientific Instruments for low-temperature experiments or with a circulating water bath. Electrospray ionization mass spectra (ESI-MS) were collected on a Waters (Milford, MA, USA) Acquity SQD quadrupole Mass instrument, by infusing samples directly into the source using a manual method. The spray voltage was set at 2.5 kV and the capillary temperature at 80 °C. Resonance Raman spectra were obtained using a liquid nitrogen cooled CCD detector (CCD-1024×256-OPEN-1LS, HORIBA Jobin Yvon) attached to a 1-m single polychromator (MC-100DG, Ritsu Oyo Kogaku) with a 1200 groovs/mm holographic grating. An excitation wavelength of 355-nm was provided by an Nd:YAG laser (Photonic Solutions, SNV-20F), with 10 mW power at the sample point. All measurements were carried out with a spinning cell (1000 rpm) at −30 °C. Raman shifts were calibrated with indene, and the accuracy of the peak positions of the Raman bands was ±1 cm⁻¹. The effective magnetic moments was determined using the modified ¹H NMR method of Evans at room temperature.¹⁹ A WILMAD^® coaxial insert (sealed capillary) tubes containing the blank acetonitrile-d₃ solvent (with 1.0 % TMS) only was inserted into the normal NMR tubes containing the complexes dissolved in acetonitrile-d₃ (with 0.03 % TMS). The chemical shift of the TMS peak (and/or solvent peak) in the presence of the paramagnetic metal complexes was compared to that of the TMS peak (and/or solvent peak) in the inner coaxial insert tube. The effective magnetic moment was calculated using the equation, μ = 0.0618(ΔνT/2fM)^1/2, where f is the oscillator frequency (MHz) of the superconducting spectrometer, T is the absolute temperature, M is the molar concentration of the metal ion, and Δv is the difference in frequency (Hz) between the two reference signals. CW-EPR spectra were taken at 5 K using an X-band Bruker EMX-plus spectrometer equipped with a dual mode cavity (ER 4116DM). Low temperatures were achieved and controlled using an Oxford Instruments ESR900 liquid He quartz cryostat with an Oxford Instruments ITC503 temperature and gas flow controller. Product analysis was performed on a Thermo Fisher Trace 1310 gas chromatograph (GC) system equipped with a flame ionization detector (FID) and mass spectrometer (GC-MS). ¹H NMR, ¹³C NMR and ³¹P NMR spectra were measured with Bruker AVANCE III-400 spectrometer at CCRF in DGIST.

Generation and Characterization of [Co^III(Me₃-TPADP)(O₂)]⁺ (2)

Treatment of [Co^II(Me₃-TPADP)(CH₃CN)₂](ClO₄)₂ (0.5 mM) with 10 equiv of H₂O₂ in the presence of 2 equiv of triethylamine (TEA) in CH₃CN (2 mL) afforded the formation of a blue solution at low temperature. Physicochemical data, including UV-vis and ESI-MS, were reported in Figure 2. [Co^III(Me₃-TPADP)(¹⁸O₂)]⁺ was prepared by adding 10 equiv of H₂¹⁸O₂ (18 μL, 90% ¹⁸O-enriched, 2.2% H₂¹⁸O₂ in water) to a solution containing [Co^II(Me₃-TPADP)(CH₃CN)₂](ClO₄)₂ (0.5 mM) and 2 equiv of TEA in CH₃CN (2 mL) at 25 °C.

Figure 2.

(a) UV-vis spectra of [Co^II(Me₃-TPADP)(CH₃CN)₂]²⁺ (1) (black line), [Co^III(Me₃-TPADP)(O₂)]⁺ (2) (red line), and [Co^III(Me₃-TPADP)(O₂H)(CH₃CN)]²⁺ (3) (blue line) in CH₃CN at −30 °C. Inset shows the resonance Raman spectra of 2 prepared with H₂¹⁶O₂ (red line) and H₂¹⁸O₂ (blue line) obtained upon excitation at 355 nm in CH₃CN at −30 °C. (b) ESI-MS of 2 in CH₃CN at −40 °C. Insets show the observed isotope distribution patterns for [Co(Me₃-TPADP)(¹⁶O₂)]⁺ (lower) and [Co(Me₃-TPADP)(¹⁸O₂)]⁺ (upper).

Generation and Characterization of [Co^III(Me₃-TPADP)(O₂H)(CH₃CN)]²⁺ (3)

Treatment of [Co^III(Me₃-TPADP)(O₂)]⁺ (0.5 mM) with 3 equiv of perchloric acid (HClO₄) in CH₃CN (2 mL) at −20 °C afforded a magenta solution. Physicochemical data, including UV-vis, ESI-MS, and resonance Raman, were reported in Figures 2 and 3. [Co^III(Me₃-TPADP)(¹⁸O₂H)(CH₃CN)]²⁺ was prepared by adding 3 equiv of HClO₄ to a solution containing [Co^III(Me₃-TPADP)(¹⁸O₂)]⁺ (0.5 mM) in CH₃CN (2 mL) at −20 °C. [Co^III(Me₃-TPADP)(¹⁶O₂²H)(CH₃CN)]²⁺ was prepared by adding 113.8 μL of D₂O to a solution containing 3 (0.5 mM) in CH₃CN (2 mL) at −20 °C.

Figure 3.

(a) ESI-MS spectra of [Co^III(Me₃-TPADP)(O₂H)(CH₃CN)]²⁺ (3) in CH₃CN at −40 °C, which show the observed isotope distribution patterns for [Co(Me₃-TPADP)(¹⁶O₂H)(CH₃CN)]²⁺ (red line, left), [Co(Me₃-TPADP)(¹⁸O₂H)(CH₃CN)]²⁺ (blue line, left), [Co(Me₃-TPADP)(¹⁶O₂H)(ClO₄)]⁺ (red line, right), [Co(Me₃-TPADP)(¹⁸O₂H)(ClO₄)]⁺ (blue line, right). (b) Resonance Raman spectra of 3 (16 mM) prepared with ¹⁶O (red line) and ¹⁸O (blue line) labeled samples of 2 obtained upon excitation at 355 nm in CH₃CN at −30 °C. Green line shows the spectrum of 3 prepared with a ¹⁶O labeled sample of 2 and HClO₄ diluted in D₂O.

X-ray crystallography

Single crystal of [Co^II(Me₃-TPADP)(CH₃CN)₂](ClO₄)₂ was picked from solutions by a nylon loop (Hampton Research Co.) on a handmade copper plate mounted inside a liquid N₂ Dewar vessel at ca. −40 °C and mounted on a goniometer head in a N₂ cryostream. Data collections were carried out on a Bruker SMART APEX II CCD diffractometer equipped with a monochromator in the Mo Kα (λ = 0.71073 Å) incident beam. The CCD data were integrated and scaled using the Bruker-SAINT software package, and the structure was solved and refined using SHELXTL V 6.12.²⁰ Hydrogen atoms were located in the calculated positions. All non-hydrogen atoms were refined with anisotropic thermal parameters. Crystal data for [Co^II(Me₃-TPADP)(CH₃CN)₂](ClO₄)₂: C₁₈H₃₀Cl₂CoN₆O₈, Monoclinic, P2₁/n, Z = 8, a = 15.7400(3), b = 19.7237(4), c = 16.3257(3) Å, β = 97.6290(10)°, V = 5023.47(17) ų, μ = 0.951 mm⁻¹, ρ_calcd = 1.556 g/cm³, R₁ = 0.0345, wR₂ = 0.0890 for 12456 unique reflections, 641 variables. The crystallographic data for [Co^II(Me₃-TPADP)(CH₃CN)₂](ClO₄)₂ is listed in Table S1, and Table S2 lists the selected bond distances and angles. CCDC-1448782 for [Co^II(Me₃-TPADP)(CH₃CN)₂](ClO₄)₂ contains the supplementary crystallographic data for this paper. These data can be obtained free of charge via www.ccdc.cam.ac.uk/data_request/cif (or from the Cambridge Crystallographic Data Centre, 12, Union Road, Cambridge CB2 1EZ, UK; fax: (+44) 1223-336-033; or deposit@ccdc.cam.ac.uk).

Computational Details

DFT calculations were performed using the Gaussian 09 computational package.²¹ All calculations used the B3LYP functional and the TZVP basis, and solvent effects were included using the polarized continuum model with acetonitrile as the solvent. Geometry optimizations were carried out using the default convergence criteria and were confirmed as minima by the absence of imaginary modes in a frequency calculation. Enthalpies and Gibbs free energies were calculated at 298.15 K. Mayer bond order analysis was performed using QMForge,²² and molecular orbital contours were obtained using LUMO.²³

Reactivity studies

All reactions were run monitoring UV-vis spectral changes of reaction solutions, and rate constants were determined by fitting the changes in absorbance at 523 nm for [Co^III(Me₃-TPADP)(O₂H)(CH₃CN)]²⁺ (3). Reactions were run at least in triplicate, and the data reported represent the average of these reactions. In situ-generated 3 were used in kinetic studies, such as the oxidation of triphenylphosphine (PPh₃) and thioanisole in CH₃CN at −20 and −40 °C. After the completion of reactions, pseudo-first-order fitting of the kinetic data allowed us to determine k_obs values. Products formed in the oxidation of PPh₃ by 3 in CH₃CN at −20 °C were analyzed by ³¹P NMR. Products formed in the oxidation of thioanisole by 3 in CH₃CN at −20 °C were analyzed by injecting the reaction mixture directly into GC and GC-MS. Products were identified by comparing with authentic samples, and product yields were determined by comparison against standard curves prepared with authentic samples.

Results and Discussion

Synthesis and Characterization

Me₃-TPADP was synthesized by a modification of a previously reported procedure [see the Supporting Information (SI), Experimental Section].²⁴ The Me₃-TPADP was characterized by electrospray ionization mass spectrometry (ESI-MS) and ¹H and ¹³C nuclear magnetic resonance (NMR) methods (see the SI, Experimental Section).

Synthetic procedures for Co(II) and Co(III) complexes used are outlined in Scheme 2. The purple starting Co(II) complex, [Co^II(Me₃-TPADP)(CH₃CN)₂](ClO₄)₂ (1-(ClO₄)₂) was synthesized by reacting Co(ClO₄)₂·6H₂O with the Me₃-TPADP ligand in CH₃CN. Single crystals of 1-(ClO₄)₂ contained two crystallographically independent but virtually identical cations in the asymmetric unit (denoted “A” and “B”; see the SI, Table S1 and S2). Complex 1 has a six-coordinated Co(II) ion with four nitrogen atoms of the Me₃-TPADP ligand and two nitrogen atoms of CH₃CN solvent molecules (Figure 1a). The Co(II) geometry is best described as distorted octahedral.

Figure 1.

(a) ORTEP plot of [Co^II(Me₃-TPADP)(CH₃CN)₂]²⁺ (1A) with thermal ellipsoid drawn at the 30 % probability level. Hydrogen atoms are omitted for clarity. Selected bond lengths (Å): Co1-N1 2.0574(14), Co1-N2 2.1999(14), Co1-N3 2.2108(14), Co1-N4 2.2050(14), Co1-N5 2.0514(15), Co1-N6 2.1151(15) (b) X-band EPR spectrum of 1 in frozen CH₃CN at 5 K. Spectroscopic settings: frequency = 9.646 GHz, microwave power = 0.998 mW, modulation frequency = 100 kHz, and modulation amplitude = 1 mT.

The UV-vis spectrum of 1 in CH₃CN shows a broad absorption band at 498 nm (ε = 80 M⁻¹ cm⁻¹) (Figure 2a). The ESI-MS spectrum of 1 exhibits three signals at a mass-to-charge ratio (m/z) of 174.2, 194.7, and 406.2 (see the SI, Figure S1), corresponding to [Co(Me₃-TPADP)(CH₃CN)]²⁺ (calcd m/z 174.1), [Co(Me₃-TPADP)(CH₃CN)₂]²⁺ (calcd m/z 194.6), and [Co(Me₃-TPADP)(ClO₄)]⁺ (calcd m/z 406.1), respectively. At room temperature, 1 exhibits a magnetic moment of 4.2 μ_B in CD₃CN using the ¹H NMR Evans method consistent with the presence of three unpaired electrons,¹⁹ and the X-band electron paramagnetic resonance (EPR) spectrum of 1 in CH₃CN at 5 K shows an axial signal at g = 4.5 and 2.08 (Figure 1b). These results indicate an S = 3/2 ground state for 1.²⁵ In addition, the redox potential of 1 was determined as E_1/2 = 0.16 V (vs Fc⁺/Fc) by cyclic voltammetry (see the SI, Figure S2).²⁶

The cobalt(III)-peroxo complex, [Co^III(Me₃-TPADP)(O₂)]⁺ (2), was prepared by adding 10 equiv of hydrogen peroxide (H₂O₂) to a reaction solution containing 1 in the presence of 2 equiv of triethylamine (TEA) in CH₃CN at 25 °C (Scheme 2), where the color of the solution changed from purple to blue. Complex 2 could not be isolated due to its thermal instability; however it was generated and characterized at low temperature in solution. The UV-vis spectrum of 2 in CH₃CN at −40 °C shows an intense absorption band at 338 nm (ε = 2400 M⁻¹ cm⁻¹) and a weak absorption band at 550 nm (ε = 230 M⁻¹ cm⁻¹) (Figure 2a). The ESI-MS spectrum of 2 exhibits a prominent signal at m/z 339.2 (Figure 2b), whose mass and isotope distribution pattern correspond to [Co(Me₃-TPADP)(O₂)]⁺ (calcd m/z 339.1) (see the SI, Figure S3). When the reaction was carried out with isotopically labeled H₂¹⁸O₂, a mass peak corresponding to [Co(Me₃-TPADP)(¹⁸O₂)]⁺ appeared at m/z 343.2 (calcd m/z 343.1) (Figure 2b, inset). The four mass unit shit upon the substitution of ¹⁶O with ¹⁸O indicates that 2 has two oxygen atoms.

The resonance Raman spectrum of 2 was collected using 355 nm excitation in CH₃CN at −30 °C. 2 prepared with H₂¹⁶O₂ exhibits an isotope-sensitive band at 888 cm⁻¹, which shifts to 842 cm⁻¹ when H₂¹⁸O₂ is used (Figure 2a, inset, see the SI, Figure S4). The observed isotopic shift of its ¹⁶Δ–¹⁸Δ value of 46 cm⁻¹ is in agreement with the calculated value of 51 cm⁻¹ for the O-O harmonic oscillator. This value is comparable to those reported for structurally and spectroscopically characterized side-on cobalt(III)-peroxo complexes, such as [Co^III(12-TMC)(O₂)]⁺ (902 cm⁻¹) and [Co^III(13-TMC)(O₂)]⁺ (902 cm⁻¹).^8e In addition, the Co-O₂ symmetric stretch of 2 was observed at 565 cm⁻¹, which shifts to 546 cm⁻¹ upon ¹⁸O-substitution (¹⁶Δ–¹⁸Δ = 19 cm⁻¹; ¹⁶Δ–¹⁸Δ_(calcd) = 21 cm⁻¹) (Figure 2a, inset, see the SI, Figure S4).

The X-band EPR spectrum of 2 is silent at 4.3 K, suggesting either a low spin (S = 0) or an integer spin (S = 1 or 2) Co(III) d⁶ species. The ¹H NMR spectrum of 2 recorded in CD₃CN at −40 °C exhibits sharp features in the 0 – 10 ppm region (data not shown), indicating that 2 is a low-spin (S = 0) state.

Addition of 3 equiv of perchloric acid (HClO₄) to a solution containing 2 in CH₃CN at −20 °C immediately produced an EPR silent magenta intermediate 3 with electronic absorption bands at ~330 nm (ε = 3300 M⁻¹ cm⁻¹) and 528 nm (ε = 240 M⁻¹ cm⁻¹) (Figure 2a). 3 is thermally unstable. Even at −40 °C, the characteristic absorption bands of 3 decayed over the course of hours (18% decay for 1 h). The decay of 3 obeyed first-order reaction kinetics, and activation parameters, ΔH^‡ = −115.9 kcal mol⁻¹ and ΔS^‡ = −15.7 cal mol⁻¹ K⁻¹, are obtained by the Eyring plot between 283 and 313 K (s). 3 could not be formed via simple oxidation of 1 with excess H₂O₂. The ESI-MS spectrum of the solution at −40 °C suggested the formation of Co(III)-hydroperoxo species, [Co(Me₃-TPADP)(O₂H)(CH₃CN)]²⁺ at m/z 190.7 (calcd m/z 190.6) and [Co(Me₃-TPADP)(O₂H)(ClO₄)]⁺ at m/z 439.2 (calcd m/z 439.1) (Figure 3a, see the SI, Figure S6), together with some unidentified species due to the thermal instability of 3 (see the s). When the reaction was carried out with 2 prepared with isotopically labeled H₂¹⁸O₂, the mass peaks corresponding to 3 shifted to m/z 192.7 for [Co(Me₃-TPADP)(¹⁸O₂H)(CH₃CN)]²⁺ (calcd m/z 192.6) and 443.2 for [Co(Me₃-TPADP)(¹⁸O₂H)(ClO₄)]⁺ (calcd m/z 443.1) (Figure 3a). Intermediate 3 reverted back to 2 on addition of 3 equiv of TEA. The addition of HClO₄ to the resulting solution regenerated 3, suggesting that 2 and 3 can be readily interconverted by the acid-base reaction depicted in Scheme 2. Such chemistry is well known for other metal-based peroxo and hydroperoxo systems.^16,27a,28,29

The resonance Raman spectrum of 3, obtained upon 355 nm excitation in CH₃CN at −30 °C, exhibits an isotope-sensitive band at 851 cm⁻¹ which shifted to 803 cm⁻¹ upon ¹⁸O-substitution (Figure 3b). The observed isotopic shift of its ¹⁶Δ–¹⁸Δ value of 48 cm⁻¹ is in good agreement with the calculated value of 49 cm⁻¹ for the O-O harmonic oscillator. The observed O-O frequency at 851 cm⁻¹ is comparable to that of a low-spin bleomycin-Co(III)-hydroperoxo complex (828 cm⁻¹).^7h The Co-O vibrational frequency of 3 was observed at 571 cm⁻¹, which shifts to 551 cm⁻¹ on ¹⁸O-substitution (¹⁶Δ–¹⁸Δ = 20 cm⁻¹; ¹⁶Δ–¹⁸Δ_(calcd) = 26 cm⁻¹) (Figure 3b). In order to support the formation of a hydroperoxide ligand in 3, we performed additional isotope labeling experiments. Upon ²H-substitution in 3, the 571 cm⁻¹ feature exhibited a 3 cm⁻¹ downshift (Figure 3b). The result is quite similar to those of previously reported Fe(III)-O₂H species.^7g,30 This deuterium isotope effect supports the presence of a hydroperoxide ligand. The ²H-substitution in 3 was further confirmed by ESI-MS (see the SI, Figure S7, inset). The ¹H NMR spectrum of 3 recorded in CD₃CN at −40 °C shows sharp features in the 0 – 10 ppm region (data not shown), indicating that 3 is a low-spin Co(III) d⁶ species. On the basis of the spectroscopic data presented above, intermediate 3 is assigned as a low-spin Co(III)-hydroperoxo complex.

Reactivity

The reactivity of 2 and 3 has been investigated in electrophilic reactions: oxygen atom transfer (i.e., the oxidation of triphenylphosphine (PPh₃) and thioanisole) and hydrogen atom transfer (i.e., the oxidation of xanthene and cyclohexadiene (CHD)) reactions (Scheme 3). Upon addition of the substrates to the solution of 2 in CH₃CN at −20 and −40 °C, the intermediate remained intact and product analysis of the reaction solution did not show oxygenated products (see the SI, Figure S8).

Scheme 3.

Overall reactivity of 2 and 3 in electrophilic reactions (e.g., HAT and OAT)

Unlike 2, addition of PPh₃ to a solution of 3 in CH₃CN at −20 °C shows the disappearance of the characteristic absorption band of 3 with a pseudo-first-order decay (see the SI, Figure S9). Product analysis of the reaction solution revealed the formation of triphenylphosphine oxide in a quantitative yield. Kinetic studies of 3 with thioanisole under the same reaction conditions also exhibit a pseudo-first-order reaction profile (Figure 4a, see the SI, Figure S10), and the first-order rate constants increased proportionally with the substrate concentration (k₂ = 6.7(8) × 10⁻¹ M⁻¹ s⁻¹) (Figure 4b). The rates were dependent on reaction temperature, where a linear Eyring plot was obtained between 233 and 263 K to give the activation parameters of ΔH^‡ = 4.2 kcal mol⁻¹ and ΔS^‡ = −49.4 cal mol⁻¹ K⁻¹ (Figure 4c). The product analysis of the reaction solution of the oxidation of thioanisole by 3 revealed that methyl phenyl sulfoxide was produced with a yield of 75 ± 8%, and the oxygen source of the product was found to be the hydroperoxo ligand of 3 on the basis of an ¹⁸O-labeling experiment performed with ¹⁸O-labeled 3 (see the s). In addition, [Co^III(Me₃-TPADP)(OH)(CH₃CN)]²⁺ was found in the reaction solution as a decomposed product of 3 (see the SI, Figure S12). The FT-IR spectrum of the Co(III)-OH product obtained via precipitation had a peak at 3500 cm⁻¹, which is consistent with an O-H vibration.³¹ The reactivity of 3 was further examined with para-substituted thioanisoles, para-X-Ph-SCH₃ (X = Me, F, H, Cl, Br), to investigate the electronic effect of para-substituents on the oxidation of thioanisoles by 3 (Figure 4d). The Hammet plot of the pseudo-first-order rate constants versus σ_p⁺ gave a ρ value of ⁻3.6(6). The negative ρ value indicates the electrophilic character of 3 in OAT reactions. Product analyses of the final reaction mixture revealed the formation of para-substituted methyl phenyl sulfoxides. It should be noted that there is a proton effect on the reactivity of 3 in the sulfoxidation reaction (see the SI, Figure S13). This result is in sharp contrast with the reactivity of [Fe^III(TMC)(O₂H)]²⁺, where no significant proton effect was observed.^27c The origin of such a proton effect remains to be established in future experiments.

Figure 4.

Reactions of [Co^III(Me₃-TPADP)(O₂H)(CH₃CN)]²⁺ (3) with thioanisole in CH₃CN. (a) UV-vis spectral changes of 3 (0.5 mM) upon addition of 25 equiv of thioanisole at −40 °C. Inset shows the time course of the absorbance at 523 nm. (b) Plot of k_obs against thioanisole concentration to determine a second-order rate constant for 3 at −40 °C. (c) Plot of pseudo-first-order rate constants against 1/T to determine activation parameters for the reaction of 3 (0.5 mM) and 50 equiv of thioanisole. (d) Hammett plot of lnk_rel against σ_p⁺ of para-substituted thioanisoles. The k_rel values were calculated by dividing k_obs of para-X-Ph-SCH₃ (X = Me, F, H, Cl, Br) by k_obs of thioanisole at −40 °C.

It is worth noting that 3 is not capable of performing HAT reactions (data not shown). However, the reactivity of [Fe^III(TMC)(O₂H)]²⁺ in both OAT and HAT reactions has previously been reported.^27b One possible explanation for the different reactivity of metal(III)-hydroperoxo species is the inability of cobalt to access its high-valent state. DFT calculations for thermodynamics of O-O homolysis for 3 relative to a low-spin Fe(III) analogue provide significant insight into their difference in reactivity (vide supra).

Density Functional Theory Studies

In a previous study evaluating the HAT reactivity of the low-spin Fe(III)-hydroperoxo complex [Fe^III(N₄Py)(O₂H)]²⁺, it was found that the barrier for HAT was mostly due to homolysis of the O-O bond.^27d In the Me₃-TPADP ligand system in this study, the Fe(III)-hydroperoxo complex is too unstable to be isolated for experimental evaluation. To calibrate this ligand set to the [Fe^III(N₄Py)(O₂H)]²⁺ results for correlation to 3, DFT calculations were performed on the hypothetical [Fe^III(Me₃-TPADP)(O₂H)(CH₃CN)]²⁺ complex to determine the thermodynamics of O-O bond homolysis relative to the results for the low-spin [Fe^III(N₄Py)(O₂H)]²⁺ complex, which were supported by experiment.^7g,27d The calculated ΔH of O-O homolysis for the Me₃-TPADP on the S = 1/2 surface is 23.2 kcal mol⁻¹, compared to 23.1 kcal mol⁻¹ previously reported for N₄Py, and the ΔG for Me₃-TPADP was found to be 13.0 kcal mol⁻¹, compared to 12.7 kcal mol⁻¹ for N₄Py. In the Me₃-TPADP case, as in the N₄Py case, the final products are an S = 1 Fe^IV=O and an hydroxyl radical. Given the good agreement between the present set of calculations and those previously reported,^7g these calculations were extended to the Co(III)-hydroperoxo complex 3.

The geometry of 3 was optimized on the S = 0, S = 1, and S = 2 surfaces. The relative energies of these structures are given on the left of Figure 5. The lowest-energy structure for 3 was found to be S = 0, consistent with the ¹H NMR data (vide supra); thus, we focus on the S = 0 calculation. Relevant geometric and vibrational parameters calculated for 3, as well as those for its iron analogue, are presented in Table 1. The structures are similar, with the main difference being the calculated M-O bond length, which is 1.88 Å for 3 and 1.80 Å for the hypothetical iron analogue. This difference arises due to the additional electron in the Co(III) relative to the low-spin Fe(III) complex, which occupies a metal d orbital π antibonding with the hydroperoxide. The calculated ν(Co-(O₂H)) is 555 cm⁻¹, in agreement with the experimental value of 571 cm⁻¹ and 66 cm⁻¹ lower in energy than the corresponding calculated stretch in the iron analogue, consistent with the longer Co-O bond length. The calculated ν(O-O) is 928 cm⁻¹, which overestimates the experimental value of 851 cm⁻¹ by 77 cm⁻¹. A similar disagreement between calculation and experiment was observed in the previous study on the low-spin [Fe^III(N₄Py)(O₂H)]²⁺ complex,^7g where an equivalent DFT calculation overestimated the experimental ν(O-O) by 98 cm⁻¹. This was attributed to the calculations underestimating the donation from hydroperoxy σ and π bonding orbitals to the metal. Given the similarity between the present DFT calculations for 3 and the previous experimentally-calibrated calculations for [Fe^III(N₄Py)(O₂H)]²⁺, these calculations for 3 were used to evaluate the thermodynamics of its O-O bond homolysis. The enthalpies and Gibbs free energies of the O-O homolysis products of 3 on the S = 0, S = 1, and S = 2 surfaces are presented on the right-hand side of Figure 5. The S = 2 products are uphill by ΔG = 39.3 kcal mol⁻¹ and are discounted. The S = 1 and S = 0 products are at 21.7 and 27.3 kcal mol⁻¹ in Gibbs free energy and correspond to S = 3/2 Co^IV=O and S = 1/2 Co^IV=O complexes, respectively, antiferromagnetically aligned to the S = 1/2 hydroxyl radical. Figure 5 also presents the energetics of O-O homolysis for the iron analogue of 3, which is ~9 kcal mol⁻¹ more favorable than the lowest energy O-O homolysis pathway for 3. Note that this pathway would further require a crossover from the S = 0 to the S = 1 surface.

Figure 5.

Thermodynamics of O-O homolysis for 3 on the S = 0 (violet line), S = 1 (red line), and S = 2 (magenta line) surfaces, as well as for the Fe analogue on the S = 1/2 (green line) surface. Energies are given as enthalpies, with Gibbs free energy at 298.15 K in parentheses. Homolysis of the O-O bond is 8.7 kcal mol⁻¹ more unfavorable for 3 relative to the iron complex. Insets show the geometry optimized structures of 3 (upper) and [Co^IV(Me₃-TPADP)(O)(CH₃CN)]²⁺ (lower) (black, C; blue, N; cyan, H; red, O; pink, Co).

Table 1.

Calculated Geometric and Vibrational Parameters for 3 and Its S=1/2 Fe^III-O₂H Analogue.

Complex	3	S=1/2 FeIII-O2H
M-O (Å)	1.88	1.80
O-O (Å)	1.44	1.44
M-O-O (°)	117	117
M-Lequatorial, ave (Å)	2.03	2.05
ν(M-(O2H)) (cm−1)	555	621
ν(O-O) (cm−1)	928	888

To understand the difference in O-O homolysis energy for 3 relative to its iron analogue, the electronic structures of the M^IV=O reaction products were analyzed, and are summarized in Figure S14 in the SI. Geometric parameters, as well as Mayer bond orders, for the products are presented in s. An S = 1 Fe^IV=O has one σ-bond between Fe and O from d_z2 interacting with the oxo p_z orbital, and one net Fe-O π bond from d_xz and d_yz interacting with the oxo p_x and p_y, respectively. In going from an S = 1 Fe^IV=O to an S = 1/2 Co^IV=O complex, a M-O π bonding interaction is lost through addition of an electron to the d_xz orbital, which has a π* interaction with the O p_x orbital (see the SI, Figure S14, bottom to upper-left), leading to a weaker Co-O relative to Fe-O bond. This additional π* interaction is reflected both in its longer bond length (1.79 vs 1.64 Å) and lower Mayer bond order (1.14 vs 1.39). This weaker Co-O bond destabilizes the S = 1/2 Co^IV=O complex relative to the S = 1 Fe^IV=O. The homolysis to form this Co^IV=O product is 14.3 kcal mol⁻¹ higher in energy than the comparable homolysis reaction for Fe (27.3 vs. 13.0 kcal mol⁻¹). For the lower energy S = 3/2 Co^IV=O case, the extra electron relative to Fe^IV=O S=1 is instead added to the d_x2-y2 orbital (see the s, bottom to upper-right), which is σ antibonding to the equatorial ligands. In this case, the Co-O bond is not perturbed by the additional electron, and is in fact stronger than the Fe-O bond, as reflected in its shorter bond length (1.62 Å vs 1.64 Å for Fe^IV=O) and higher Mayer bond order (1.85 vs. 1.39). This is due to the greater Z_eff on Co^IV relative to Fe^IV, which stabilizes its d manifold and allows stronger mixing with the oxygen p orbitals. However, the equatorial Co-L bonds are greatly weakened relative to the iron complex (see the SI, Table S3; the average M-L_equatorial bond increases to 2.17 from 2.07 Å and the total Mayer bond order for these bonds decreases to 1.74 from 2.16). This overcomes the stronger Co-O bonding interaction and results in an S = 3/2 Co^IV=O that is less stable than the S = 1 Fe^IV=O by 9 kcal mol⁻¹. Note that the Co^III-O₂H is also less stable than the Fe^III-O₂H due to weakened M-O₂H bonding arising from the additional dπ* electron, but this involves loss of a relatively weak π bonding interaction (vide supra; also see differences in calculated M-O bond lengths and stretching frequencies in Table 1). Thus, 3 is much less reactive than equivalent low-spin Fe(III)-hydroperoxo complexes due to the higher barrier for O-O bond homolysis resulting from the additional electron in a equatorial sigma antibonding orbital of the Co^IV=O S = 3/2 product.

Conclusions

The metal-hydroperoxo intermediates in organic functionalizations are of current interest in enzymatic processes, pharmaceutical research and industrial catalysis. We have synthesized mononuclear cobalt(III)-peroxo (2) and -hydroperoxo (3) complexes bearing a common macrocyclic Me₃-TPADP ligand, where 3 was prepared by protonation of 2. The consecutive interconversion of 3 to 2 by addition of a base supports acid-base chemistry. The intermediates were characterized with a variety of physicochemical methods. Although the UV-vis spectra of 2 and 3 are not very different, ESI-MS spectra of 2 clearly exhibit the formation of the cobalt(III)-peroxo adduct at low temperature and resonance Raman spectra of 3 show an O-O stretching vibration at 851 cm⁻¹ for ¹⁶O samples (803 cm⁻¹ for ¹⁸O samples), which is assignable to that of hydroperoxo species. The reactivities of 2 and 3 were compared in electrophilic reactions; 3 is capable of conducting OAT reactions, whereas 2 exhibits no OAT reactivity. Alternatively, 3 does not perform HAT, whereas low-spin Fe(III)-hydroperoxo complexes do. DFT calculations show that this is due to the high O-O bond homolysis energy relative to the iron analogue. This energy difference is due to the additional electron in an antibonding orbital that would destabilize a high-valent Co-oxo product.