Substrate and Lewis Acid Coordination Promote O–O Bond Cleavage of an Unreactive L₂Cu^II₂(O₂²⁻) Species to Form L₂Cu^III₂(O)₂ Cores with Enhanced Oxidative Reactivity

Abstract

Copper-dependent metalloenzymes are widespread throughout metabolic pathways, coupling the reduction of O₂ with the oxidation of organic substrates. Small-molecule synthetic analogs are useful platforms to generate L/Cu/O₂ species that reproduce the structural, spectroscopic, and reactive properties of some copper-/O₂-dependent enzymes. Landmark studies have shown that the conversion between dicopper(II)-peroxo species (L₂Cu^II₂(O₂²⁻) either side-on peroxo, ^SP, or end-on trans-peroxo, ^TP) and dicopper(III)-bis(μ-oxo) (L₂Cu^III₂(O²⁻)₂:O) can be controlled through ligand design, reaction conditions (temperature, solvent, and counteranion), or substrate coordination. We recently published (J. Am. Chem. Soc. 2012, 134, 8513, DOI: 10.1021/ja300674m) the crystal structure of an unusual ^SP species [(MeAN)₂Cu^II₂(O₂²⁻)]²⁺ (^SP^MeAN, MeAN: N-methyl-N,N-bis[3-(dimethylamino)propyl]amine) that featured an elongated O–O bond but did not lead to O–O cleavage or reactivity toward external substrates. Herein, we report that ^SP^MeAN can be activated to generate O^MeAN and perform the oxidation of external substrates by two complementary strategies: (i) coordination of substituted sodium phenolates to form the substrate-bound O^MeAN-RPhO⁻ species that leads to ortho-hydroxylation in a tyrosinase-like fashion and (ii) addition of stoichiometric amounts (1 or 2 equiv) of Lewis acids (LA’s) to form an unprecedented series of O-type species (O^MeAN-LA) able to oxidize C–H and O–H bonds. Spectroscopic, computational, and mechanistic studies emphasize the unique plasticity of the ^SP^MeAN core, which combines the assembly of exogenous reagents in the primary (phenolates) and secondary (Lewis acids association to the MeAN ligand) coordination spheres with O–O cleavage. These findings are reminiscent of the strategy followed by several metalloproteins and highlight the possible implication of O-type species in copper-/dioxygen-dependent enzymes such as tyrosinase (Ty) and particulate methane monooxygenase (pMMO).

Graphical Abstract

INTRODUCTION

Dicopper centers are a common motif found in the active sites of several O₂-activating metalloproteins that perform an array of physiological functions.¹ Tyrosinase (Ty) is a paradigmatic example of a metalloenzyme with a coupled dicopper active center (Figure 1A, left), which combines the reduction of dioxygen with the ortho-hydroxylation of tyrosine for melanin biosynthesis. Although its catalytic activity has been widely studied during the last several decades, there is still some debate on the identity of the reactive Cu₂/O₂ intermediate that is responsible for substrate oxidation. There is agreement on the first reaction step: Reduced Ty (dicopper(I)) reacts with dioxygen to form a dicopper(II) species with a side-on peroxide moiety (^SP). It is proposed that coordination of tyrosinate (phenolate form, Figure 1A center) to one of the copper(II) centers in oxy-Ty is followed by electrophilic attack of the peroxide to the aromatic ring.^1–3 Alternatively, some authors propose that prior to the electrophilic attack O–O cleavage occurs thus generating a dicopper(III) bis(μ-oxo) species (O) as the active oxidant (Figure 1A, right).^4,5

Figure 1.

Active site and proposed Cu/O₂ reactive species of (A) tyrosinase (Ty) and (B) particulate methane monooxygenase (pMMO).

Another ambiguity exists in copper-dependent particulate methane monooxygenase (pMMO). Detailed structural and spectroscopic studies have led to the proposal that the active site is a dicopper center with only two residues coordinating each Cu, with one Cu intriguingly bound to a N-terminal histidine (Figure 1B, left). Due to the technological implications of methane oxidation,^6,7 elucidation of the identity of the Cu/O₂ active species is of considerable general interest. However, the little information available has led to speculation that oxidation could be carried out by a Cu^II(O₂²⁻)Cu^II (^SP), a Cu^III(O²⁻)₂Cu^III (O), or a mixed-valent Cu^II(O²⁻)₂Cu^III (and possibly protonated) species (Figure 1B, right).^5,8–11

Several research groups, including ours, have extensively explored the O₂ chemistry of Cu(I) model complexes, which has led to the characterization of many distinct types of mononuclear and dinuclear Cu_n/O₂ species (Figure 2, top).^12–16 Especially relevant has been the study of the equilibrium between peroxodicopper(II) (^SP or ^TP) and bis(μ-oxo) dicopper(III) (O) species, which can be controlled by ligand design (including ligand donation to Cu), varying reaction conditions (i.e., temperature, solvent, and/or counter-anion), and substrate coordination (Figure 2, bottom).^12–21 From these studies, several general trends have emerged for controlling this equilibrium. It has been generally established that (i) tridentate ligation (i.e., N₃) favors ^SP formation while use of bidentate ligands (i.e., N₂) usually leads to O species; (ii) weakly coordinating counteranions and polar solvents favor O assemblies; and (iii) coordination of anionic substrates such as phenolates can lead to O–O cleavage prior to substrate oxidation.⁴

Figure 2.

Top: Common LCu/O₂ species derived from the oxygenation of copper(I) complexes. Bottom: Control of the ^SP/O equilibrium by different means.

Recently, we published the crystal structure of the ^SP system [(MeAN)₂Cu^II₂(O₂²⁻)]²⁺ (^SP^MeAN) (MeAN: N-methyl-N,N-bis[3-(dimethylamino)propyl]amine, Figure 3 top).²² Strikingly, despite the unusually long O–O distance (1.54 Å) in this complex, there was no observable ^SPMeAN/O^MeAN equilibrium or reactivity with external substrates. We had attributed this to the weak backbonding from the Cu 3d_xy orbital to the peroxide σ* orbital (Figure 3, middle), which is crucial for O–O cleavage.

Figure 3.

Top: ^SP^MeAN X-ray structure and its geometric and spectroscopic features. Middle: Molecular orbital diagrams of the ^SP^MeAN core that lead to lengthening of the O–O bond without concomitant O–O cleavage. Bottom: Subjects of this report: coordination of phenolates and Lewis acids to ^SP^MeAN imparts tyrosinase-like hydroxylation and pMMO-like C–H oxidation reactivity through O^MeAN.

Herein, we report that the oxidative properties of ^SP^MeAN can be activated by two different approaches (Figure 3, bottom). As previously reported in other systems, coordination of a series of substituted sodium phenolates leads to O–O cleavage to generate observable O^MeAN-RPhO⁻ species that decay to generate the ortho-hydroxylated product.^4,23 A surprising and very new finding is that the ^SPMeAN/O^MeAN equilibrium can also be controlled by addition of various Lewis acids (LA) which are able to trigger the O–O cleavage of ^SP^MeAN species to generate distinctive O^MeAN-LA species with enhanced and tunable oxidative reactivity toward C–H and O–H bonds.

EXPERIMENTAL SECTION

General Considerations

All reagents and solvents were of commercially available quality except as noted. Acetone was distilled from Drierite under argon atmosphere. [(MeAN)Cu^I](BAr^F) (BAr^F: B(C₆F₅)₄⁻) was synthesized as previously described.²⁴ Sodium phenolates were obtained using a synthetic method similar to previous results.²³ All UV–vis measurements were carried out using a Hewlett-Packard 8453 diode array spectrophotometer with a 10 mm quartz cell. The spectrometer was equipped with HP Chemstation software and a Unisoku cryostat for low temperature experiments. ¹H NMR spectra were recorded on a Bruker 400 instrument. Resonance Raman (rR) samples were excited using a Coherent I90C–K Kr⁺ ion laser at 413.1 or 568.2 nm while the sample was immersed in a liquid-nitrogen-cooled (77 K) EPR finger dewar (Wilmad). Power was ~20 mW at the sample for the 413.1 nm line and ~130 mW at 568.2 nm. Data were recorded while rotating the sample to minimize photodecomposition. The spectra were recorded using a Spex 1877 CP triple monochromator with 600, 1200, or 2400 grooves/mm holographic spectrograph grating and detected by an Andor Newton CCD cooled to −80 °C (413.1 nm) or an Andor IDus CCD cooled to −80 °C (568.2 nm). Spectra were calibrated on the energy axis to toluene at room temperature.

[(MeAN)₂Cu^II₂(O₂²⁻)]²⁺ (^SP^MeAN) Generation and Reaction with Sodium Phenolates

A 3 mL aliquot of a [(MeAN)Cu^I](BAr^F) solution (0.2 mM) in acetone was placed in a 10 mm path quartz cell equipped with a stir bar and capped with a rubber septum. After cooling down the cell to −90 °C (acetone), dioxygen was added to generate [(MeAN)₂Cu^II₂(O₂²⁻)]²⁺ (^SP^MeAN) (λ_max: 365 nm ε = 22 mM⁻¹ cm⁻¹). After complete formation, 100 μL of an acetone solution containing excess of the corresponding substrate (substituted phenols, PPh₃, DHA, thioanisole, Me₂Fc, or Fc) was added. No spectral change was observed within 60 min after substrate addition. In contrast, addition of 100 μL of an acetone solution containing the corresponding 4-substituted sodium phenolate (to give a final concentration of 0.2 mM) to ^SP^MeAN led to the immediate formation of the putative [(MeAN)₂Cu^III₂(O²⁻)₂(R-PhO⁻)]⁺, followed by its decay (see Supporting Information). Kinetic analysis was performed by fitting the exponential decay of [(MeAN)₂Cu^III₂(O²⁻)₂(R-PhO⁻)]⁺ at 400 nm. The k_obs obtained are summarized in the Supporting Information. Quantification of the corresponding catechols was performed using a method previously described (see Supporting Information).^25,26

O^MeAN-LA Generation

A 0.2 mM solution of ^SP^MeAN in acetone was prepared as described above at −90 °C. Then, 100 μL of an acetone solution containing the corresponding equivalents of Lewis acid (0.2 mM DMF·CF₃SO₃H, 0.1 mM Sc(CF₃SO₃)₃, or 0.1 mM B(C₆F₅)₃) was added to generate the corresponding O^MeAN-LA species. See Supporting Information for titration experiments.

^SP^MeAN/O^MeAN-LA Reversibility

Formation of O^MeAN-2H⁺ could be reversed by addition of proton sponge (1,8-bis(dimethylamino)-naphthalene) in excess (0.6 mM) leading to full regeneration of ^SP^MeAN (see Supporting Information). O^MeAN-2H⁺ could be formed again by readdition of DMF·CF₃SO₃H (0.6 mM was required for full formation, due to the presence of proton sponge excess in solution). Similarly, formation of O^MeAN-Sc³⁺ could be reversed by addition of 1,10-phenanthroline in excess (0.8 mM), leading only to partial formation (75%) of ^SP^MeAN (see Supporting Information) due to the ability of 1,10-phenanthroline to trap Cu(I) (430 and 455 nm bands, see Figure S9). O^MeAN-Sc³⁺ could be partially formed again by readdition of Sc(CF₃SO₃)₃ (0.2 mM was required for full formation, due to the presence of excess 1,10-phenanthroline in solution). We could also reverse the formation of O^MeAN-2BR₃ by addition of TASF (tris(dimethylamino)sulfonium difluorotrimethylsilicate, a F⁻ source) in excess (0.6 mM), leading to partial formation (60%) of ^SP^MeAN (see Supporting Information). O^MeAN-2BR₃ could be partially formed again by readdition of B(C₆F₅)₃.

O^MeAN-LA Reactivity toward Weak C–H Bonds and Substituted Phenols

After generation of the corresponding O^MeAN-LA as described above, 100 μL of an acetone solution containing various amounts of the corresponding substrate (weak C–H bonds or substituted phenols) was added, causing the decay of the O^MeAN-LA spectral features. The absorbance changes (λ = 410 nm) were fitted to single exponential decays. In all cases, a linear correlation between the reaction rate (k_obs) and the substrate concentration was found, from which the second-order rate constants were obtained. Quantification of the oxidation products derived from the C–H (BNAH) and O–H oxidation (2,4-^tBu-PhOH) was performed using methods previously described (see Supporting Information for details).²⁷

Resonance Raman Experiments

In a typical experiment, 0.57 mL of a [(MeAN)Cu^I]⁺ (1.0 mM) solution in acetone were placed in a 5 mm NMR tube capped with a septum. After cooling the tube to −90 °C (acetone/N_2(liq) bath), dioxygen was bubbled through the solution mixture to generate [(MeAN)₂Cu₂(O₂²⁻)]²⁺ (^SP^MeAN). To generate the [(MeAN)₂Cu^III₂(O²⁻)₂(2,6-F₂–PhO⁻)]⁺, 50 μL of an acetone solution containing sodium 2,6-F₂-phenolate (3 mM) was added. To generate O^MeAN-2H⁺, 50 μL of an acetone solution containing 2 equiv of DMF·CF₃SO₃H (1 mM) was added. To generate the O^MeAN-Sc³⁺, 50μL of an acetone solution containing 1 equiv of Sc(CF₃SO₃)₃ (0.5 mM) was added. To generate the O^MeAN-2BR_3, 50 μL of an acetone solution containing 2 equiv of B(C₆F₅)₃ (0.5 mM) was added. Samples were frozen either 10 or 300 s after phenolate addition.

DFT Calculations

All calculations were performed with Gaussian 09 (revision D.01)²⁸ using the spin-unrestricted hybrid density functional B3LYP^29–31 and a split basis set (6-311G(d) on Cu, N, and O; 6-31G on all other atoms) on an ultrafine integration grid. All calculations included solvation correction (PCM, acetone solvent). Analytical frequencies were calculated on all geometry-optimized species to verify that the species were true local minima without negative frequencies. Calculations involving 2,6-difluorophenolate included Grimme’s D3 dispersion correction.³² For broken-symmetry singlet states of the side-on peroxos, the Yamaguchi formalism³³ was used to eliminate triplet contamination and obtain a corrected SCF energy (eq 1, where ¹E is the corrected singlet energy, ^BSE and ³E are energies of the broken-symmetry and triplet states, respectively, and 〈S²〉_BS is the calculated spin expectation value obtained from the broken-symmetry wave function).

$${}^{1}E=\frac{2{}^{\text{BS}}E-({\langle S^2\rangle }_{\text{BS}}·{}^{3}E)}{2-{\langle S^2\rangle }_{\text{BS}}}$$ (1)

For time-dependent DFT calculations, the calculated electronic absorption spectra were generated using SWizard³⁴ with 3000 cm⁻¹ line widths and Gaussian bandshapes.

RESULTS AND DISCUSSION

Ortho Hydroxylation of Phenolates

We previously reported²² that ^SP^MeAN was unreactive toward external substrates including 9,10-dihydroanthracene (DHA, BDFE_(C–H) = 76 kcal/mol), 2,4-di-tert-butylphenol (2,4-DTB-PhOH, BDFE_(O–H) = 82 kcal/mol), and thioanisole. We expanded the substrate scope to other substrates even more prone to oxidation by electrophilic metal-peroxo species including 1-benzyl-1,4-dihydronicotinamide (BNAH, BDFE_(C–H) = 71 kcal/mol),²⁷ 2,6-di-tert-butyl-4-methoxyphenol (BDFE_(O–H) = 80 kcal/mol),³⁵ and triphenylphosphine, but no decay of the ^SP^MeAN UV–vis features were observed, even 1 h after addition of the substrate. We speculated that the unreactive nature of ^SP^MeAN was due to the inability of the complex to form an O^MeAN species, which typically oxidizes these types of substrates.^14,36

This has led us to attempt oxidation of other types of substrates. ^SP^MeAN reacts with a series of 4-substituted sodium phenolates (4-MeO, 4-Me, 4-H, 4-Cl, and 4-CN), which generates the corresponding ortho-hydroxylated products in a tyrosinase-like fashion in moderate yields (30–40%). Kinetic studies using the series of 4-R-PhO⁻ were conducted in order to gather mechanistic details for this oxidative process. Addition of the substrate to ^SP^MeAN caused an immediate change of the UV–vis spectrum, followed by a fast first-order decay (see Supporting Information for details). The disappearance of these fleeting species was followed by UV–vis and fit to an exponential decay. When plotting the reaction rates obtained (k_obs) for the varying substrates against the Hammett parameter (σ⁺), a linear correlation was found (ρ = −2.2), suggesting an electrophilic aromatic substitution mechanism (see Figure S2). These findings are analogous to the results found by Stack and co-workers and other research groups,^4,23 pointing toward a mechanism where substrate coordination to ^SP^MeAN triggers O–O cleavage to form a [(MeAN)₂Cu^III₂(O²⁻)₂(R-PhO⁻)]⁺ (O^MeAN-RPhO⁻) intermediate prior to substrate oxidation (Figure 4, top).

Figure 4.

Top: Postulated reaction mechanism of ^SP^MeAN toward sodium phenolates to generate the corresponding ortho-hydroxylated products through O^MeAN-RPhO⁻. Bottom left: Core structures of DFT-optimized ^SP^MeAN and O^MeAN-RPhO⁻ with relevant metrical parameters, calculated Cu–O normal modes, TD-DFT-calculated LMCT energies, and energy of peroxo isomers relative to the corresponding bis(μ-oxo) isomers. Bottom right: Absorption spectra of ^SP^MeAN (brown) and intermediate O^MeAN−F₂PhO⁻ (blue). Inset: Resonance Raman spectra of ^SP^MeAN (λ = 413.1 nm) and O^MeAN−F₂PhO⁻ (λ = 568.2 nm) with ¹⁶O₂ (blue), ¹⁸O₂ (brown), and the ¹⁶O₂-¹⁸O₂ difference spectra (green).

In a recent contribution, Serrano-Plana et al. reported that the MeAN-inspired [(m-XYL^MeAN)₂Cu^III₂(O²⁻)₂] ²⁺ complex was also able to perform the ortho-hydroxylation/defluorination of 2-fluoro-substituted phenolates in a selective fashion.²⁶ Using similar reaction conditions, we found that the ^SP^MeAN oxidized 2,6-F₂-phenolate to the corresponding 3-F-catechol in moderate yield (30%). Interestingly, coordination of the substrate to ^SP^MeAN (Figure 4 bottom right, brown spectrum) was found to form a phenolate-bound O-type species [(MeAN)₂Cu^III₂-(O²⁻)₂(F₂PhO⁻)]⁺ (O^MeAN-F₂PhO⁻, Figure 4, blue spectrum) with spectroscopic characteristics similar to other phenolate-coordinated O species.²³

Due to the slower reaction rate found for the fluorinated substrate, we were able to cold-trap and characterize O^MeAN-F₂PhO⁻ by rR spectroscopy (Figure 4, bottom right inset). Upon laser excitation at 568.2 nm, three main isotope-sensitive modes were observed at 663 (Δ (¹⁸O₂) = −23 cm⁻¹), 604 (Δ (¹⁸O₂) = −13 cm⁻¹), and 590 cm⁻¹ (Δ (¹⁸O₂) = −31 cm⁻¹), which are assigned as ν_Cu–O vibrations of the Cu₂O₂ core of O^MeAN-F₂PhO^−.37 The characteristic ν_O–O = 722 cm⁻¹ mode of ^SP^MeAN (Figure 4, bottom left inset) was not observed after phenolate addition, demonstrating full conversion to O^MeAN-F₂PhO⁻. One additional mode with very small ^16/18O₂ sensitivity (513 cm⁻¹, Δ (¹⁸O₂) = −3 cm⁻¹) is also observed in O^MeAN-F₂PhO⁻, which is likely a predominantly Cu–N or Cu–O_phenolate mode that gains minor ^16/18O₂ isotope sensitivity through mixing of a core Cu–O mode. The number of observed Cu–O modes in rR can be a useful reporter of the local symmetry of the Cu₂O₂ core, since only symmetric modes (i.e., with A symmetry) are rR active.³⁸ A maximally symmetric Cu₂O₂ bis(μ-oxo) of D_2h symmetry has only one of four Cu–O stretching modes symmetry allowed (A₁), while descent in symmetry to C_2h or C_2v results in two allowed modes (A_g), and further descent to C_s results in all four modes being symmetry allowed (A′). The observation of three Cu–O modes in O^MeAN-F₂PhO⁻ thus suggests local pseudo-C_s symmetry, which is consistent with coordination of phenolate to one Cu.

DFT calculations were used to support the assignment of O^MeAN-F₂PhO⁻ and investigate the role of phenolate in modulating the ^SP/O equilibrium. Two different isomers of the resulting O^MeAN-F₂PhO⁻ adduct were optimized (Figure 4, bottom left): one with the phenolate coordinated to one of the copper(III) centers in an axial position (O^MeAN-F₂PhO⁻ (ax)), which was 10 kcal/mol higher in energy than the other with phenolate coordinated in an equatorial position (O^MeAN-F₂PhO⁻ (eq)). Only the equatorial phenolate model correctly reproduced the observed direction of the ^SP/O equilibrium, with the bis(μ-oxo) favored over the side-on peroxo isomer by 7.7 kcal/mol (for the axial phenolate, ^SP was favored by 3.1 kcal/mol). Furthermore, calculated normal modes for O^MeAN-F₂PhO⁻ (eq) better matched the experimental rR data than O^MeAN-F₂PhO⁻ (ax) (See Supporting Information for details); thus, the observed species is most consistent with possessing a O^MeAN-F₂PhO⁻ (eq) structure.

TD-DFT calculations of O^MeAN-F₂PhO⁻ (eq) also reproduced the absorption spectrum of the observed bis(μ-oxo) species. While both O^MeAN-F₂PhO⁻ isomers showed typical spectra for a O-type species (two intense absorptions centered around 300 and 400 nm), only the spectrum calculated for the O^MeAN-F₂PhO⁻ (eq) isomer had the intense lower energy bands (450–550 nm), which are assigned as phenolate-to-copper charge transfer transitions based on orbital analysis. This arises from the orientation of the phenolate in the Cu₂O₂ plane, which provides good overlap between the donor (phenolate π) and acceptor (Cu d_x²_–y²) orbitals and thus intense LMCT character.

As previously reported by Stack and co-workers,³⁷ we propose that initial coordination of the phenolate occurs at one of the available axial positions (O^MeAN-F₂PhO⁻ (ax)) and then rapidly isomerizes to the spectroscopically observed equatorial form (O^MeAN-F₂PhO⁻ (eq)). The substrate is then electrophilically attacked by the Cu^III₂(O²⁻)₂ core. In the axial form, DFT calculations suggest there is a weak Cu^III···F interaction (d(Cu···F) = 3.02 Å) worth ~2 kcal/mol as compared to that of the phenolate-bound complex without the ortho F (calculated using 2,4-F₂-phenolate, see Supporting Information). However, this weak Cu–F interaction is not sufficiently strong to prevent isomerization to the equatorially bound species (O^MeAN-F₂PhO⁻ (eq), 10 kcal/mol more stable) which would then lead to ortho-hydroxylation and defluorination.

O–O Cleavage Promoted by Lewis Acids

Due to the unreactive character of ^SP^MeAN core, we wondered if addition of Lewis acids could boost its oxidative power, as observed with iron or manganese oxo (or peroxo) complexes.^39–42 Strikingly, addition of Sc(CF₃SO₃)₃ to ^SP^MeAN (λ_max = 365 nm; ε = 22 mM⁻¹ cm⁻¹) led to the formation of a new species with UV–vis features (Figure 5) that are characteristic of O-type complexes (λ_max = 412 nm; ε = 24 mM⁻¹ cm⁻¹). Titration experiments established that only 1 equiv of Sc³⁺ was required for full formation of the adduct, thus formulated as O^MeAN-Sc³⁺ (Figure 5, top). Surprisingly, this reactivity is not exclusive for Sc³⁺: Addition of nonmetallic Lewis acids such as B(C₆F₅)₃ or DMF·CF₃SO₃H (Brønsted acid) also led to O–O cleavage and formation of O-type species. In these latter cases, titration experiments suggested that 2 equiv of B(C₆F₅)₃ or 2 equiv of DMF·CF₃SO₃H were necessary to form the corresponding adducts, thus formulating the products as the species O^MeAN-2BR₃ (λ_max = 414 nm; ε = 24 mM⁻¹ cm⁻¹) and O^MeAN-2H⁺ (λ_max = 410 nm; ε = 21 mM⁻¹ cm⁻¹), respectively.

Figure 5.

Top: Generation of the different O^MeAN-LA complexes. Middle: Spectroscopic characterization of the O^MeAN-LA series by UV–vis (inset shows titration experiments) and rR at 413.1 nm excitation (¹⁶O₂ in blue, ¹⁸O₂ in red, and Cu(I) background in black). Mode assignments are indicated, and instrument artifacts are marked with (*). Bottom: Structures of DFT-optimized O^MeAN-LA species with relevant metrical parameters, calculated Cu–O normal modes, TD-DFT-calculated LMCT energies, and energy of peroxo isomers relative to the corresponding bis(μ-oxo) isomers.

We made use of rR spectroscopy to characterize this family of O^MeAN-LA species. Laser excitation (λ_exc. = 413 nm) of samples of O^MeAN-Sc³⁺, O^MeAN-2BR₃ and O^MeAN-2H⁺ prepared with ¹⁶O₂ and ¹⁸O₂ enhanced three isotope-sensitive modes at 514–602 cm⁻¹ (ν₂, ν₃, and ν₄ in Figure 5 middle) characteristic of Cu–O core modes of O-type species, and similar to O^MeAN-F₂PhO⁻. Combination bands (ν₂ + ν₄ at ~1120 cm⁻¹ and ν₁ + ν₄ at ~740 cm⁻¹, where ν₁ is a low-energy isotope-insensitive mode at ~140 cm⁻¹) and an overtone (2ν₄ at ~1210 cm⁻¹) are also observed.^43,44 Surprisingly, the rR spectra of the different O^MeAN-LA were almost superimposable, suggesting that the bis(μ-oxo) dicopper(III) cores had very similar structures. This indicates that the Lewis acids are not interacting directly with the oxygen atoms of the Cu₂O₂ core but rather coordinating to the axial N donors of the ligand. DFT calculations support this hypothesis: We found that in the computationally optimized structures for all three O^MeAN-LA,^43,45 the Lewis acid was bound to the N ligands of MeAN (Figure 5, bottom). Moreover, our calculations were able to reproduce the stoichiometry found in our titration experiments: While the Sc³⁺ ion was able to coordinate one N from each of the MeAN ligands, two molecules of borane or acid were required for O–O cleavage. The stabilization energy calculated upon binding of the Lewis acid was found to be very similar for all three O^MeAN-LA complexes (between 3 and 6 kcal/mol, see Supporting Information). The optimized geometries for the O^MeAN-LA species are very similar in all the cases, where the copper(III) ions are bound to two of the N donors of the MeAN ligand and two oxygen atoms. Similarly, the calculated UV–vis and rR features were also nearly identical for all adducts, corroborating our experimental findings (see Supporting Information).

To investigate the role of Lewis acids in stabilizing the O^MeAN isomer relative to ^SP^MeAN, we optimized a structure of ^SP^MeAN with both axial N ligand arms uncoordinated (Figure 6A, right). Although removing the axial ligands is calculated to cost ΔG = +8 kcal/mol, the “arm off” isomer favors an O configuration by ΔG = −3.5 kcal/mol, which is similar to the preference for O in the Lewis acid adducts (–3.5 to −6.7 kcal/mol stabilization of O relative to that of ^SP, Figure S23), and inverse from that of the “arm on” isomer, which favors the ^SP configuration by ΔG = −2.2 kcal/mol. The “arm off” peroxo model featured a longer O–O bond (1.504 vs 1.488 Å) and shorter Cu–O (average 1.93 vs 1.98 Å) and Cu–N bonds (1.99 vs 2.03 Å) compared to the “arm on” structure (Figure 6A, top). The “arm off” structure thus allows better orbital overlap with the equatorial N ligands, which is also consistent with the less pyramidalized Cu (3° deviation from the Cu₂O₂ plane vs 27°, Figure 6A, center). Therefore, by sequestering the axial N, addition of Lewis acid allows stronger equatorial ligand donation, which in turn leads to a greater degree of O₂ activation (Figure S26) that results in formation of a low energy O^MeAN species via O–O cleavage.⁴³

Figure 6.

(A) DFT-optimized structures of ^SP^MeAN with the axial amine ligand arm in an (left) “on” binding position as observed experimentally by crystallography, and (right) “off” binding position. Top: core geometry and bond lengths (Å). Bottom: core structure viewed perpendicular to the Cu₂O₂ plane, showing the smaller deviation of the equatorial N ligands out of the Cu₂O₂ plane in the “arm off” isomer. (B) Role of Lewis acids in O–O bond formation at the active center of Photosystem II (left), the reversible O–O cleavage shown in the ^SP^MeAN/O^MeAN− LA series (middle) and O–O cleavage in nonheme iron(III)-peroxo reduction/oxidation (right).

Surprisingly, the O–O bond cleavage could be reversed by addition of Lewis acid chelating reagents. For example, 1,10-phenanthroline was able to capture Sc³⁺ from O^MeAN-Sc³⁺ to recover 75% of ^SP^MeAN, and 1,8-bis(dimethylamino)-naphthalene (proton sponge) was added to O^MeAN-2H⁺ to fully regenerate ^SP^MeAN (see Supporting Information for details).⁴⁶ Recently, Que and co-workers have studied the effect of Sc³⁺ addition to the mononuclear nonheme iron(II) complex, which when combined with dioxygen led to the formation of a metastable iron(III)-peroxo-Sc³⁺ adduct that then irreversibly transformed to a corresponding high-valent iron(IV)-oxo complex (Figure 6, right).⁴⁷ Nam and co-workers expanded that reactivity to other Lewis acids, where depending on the nature of the Lewis acid identity, the iron(III)-peroxo-LA adduct could be reduced to liberate dioxygen (LA: Ca²⁺, Sr²⁺) or led to O–O cleavage (LA: Zn²⁺, Lu³⁺, Y³⁺, Sc³⁺).⁴⁸ To the best of our knowledge, the reactivity reported herein is the first example of Lewis acid mediated reversible O–O bond cleavage/formation, which is reminiscent of the chemistry occurring in water oxidation process catalyzed by the Mn₄Ca²⁺ cluster in photosystem II (Figure 6, left).

Reactivity of O^MeAN-LA toward C–H Bonds

At this point, we decided to test the reactivity of O^MeAN-LA species toward external substrates, alongside the analogous parent ^SP^MeAN complex. Surprisingly, we found that both ^SP^MeAN and O^MeAN-LA were unable to oxidize substrates that are prone to accept O atoms, such as triphenylphosphine and thioanisole.¹⁴ The first difference in reactivity was found when substrate one-electron (1e⁻) reductants were added. While both ^SP^MeAN and O^MeAN-LA reacted with strong 1e⁻ reducing reagents (Me₁₀Fc or Me₈Fc, E⁰ < – 0.4 V vs Fc^+/0), only the O^MeAN-LA series was found to be able to oxidize the more difficult substrate, Me₂Fc (E⁰ = −0.12 V vs Fc^+/0). Based on this bracketing analysis, we can conclude that the O^MeAN-LA species possess higher reduction potentials than ^SP^MeAN, that is, they are stronger one-electron oxidants.

Another main difference in reactivity was found in reactions with substrates with weak C–H bonds. As previously reported, ^SP^MeAN was not able to oxidize dihydroanthracene (DHA; BDE: 78 kcal/mol).²² Here, we find that the O^MeAN-LA species are also not able to abstract a H atom (H^•) from DHA. However, when 1-benzyl-1,4-dihydronicotinamide (BNAH, BDE: 71 kcal/mol) was used, the ^SP^MeAN core was unreactive, but the series of O^MeAN-LA complexes reacted to generate the corresponding BNA⁺ product in 75–99% yield (Figure 7A, see Supporting Information for product quantification). Kinetic interrogation of this reaction showed pseudo-first-order decay behavior with respect to [BNAH] (see Supporting Information). Second-order rate constants were obtained (Table 1), with the O^MeAN-2BR₃ reacting slightly faster than the O^MeAN-2H⁺ and O^MeAN-Sc³⁺ cores (Table 1). The same set of experiments was carried out using the deuterated analogue BNAD, which led to the determination of primary kinetic isotope effects of 10.3 for O^MeAN-2BR₃, 8.5 for O^MeAN-2H⁺, and 6.1 for O^MeAN-Sc³⁺ (temperature: −90 °C). These values are consistent with C–H bond cleavage during the rate-determining step (r.d.s.).

Figure 7.

C–H oxidations performed by O^MeAN-LA cores: (A) kinetic analysis of the reaction (see Supporting Information for further details); (B) substrates used and proposed mechanism for HAT + e⁻ transfer.

Table 1.

Kinetic Parameters of C–H Oxidation Reactions by O^MeAN-LA

substrate	BDE (H•) (kcal/mol)	BDE (H−) (kcal/mol)	kOMeAN32H+ (M−1 s−1)	kOMeAN3Sc3+ (M−1 s−1)	kOMeAN32BR3 (M−1 s−1)	notes
BNAH	70.7	64.2	66 ± 1	58 ± 1	138 ± 2	yield: 80–99%
BNAD			7.8 ± 0.1 (KIE: 8.5)	5.7 ± 0.1 (KIE: 10.3)	22.7 ± 0.3 (KIE: 6.1)
BzImH	73.4	49.5	12.1 ± 0.2a	6.0 ± 0.1b	19.1 ± 0.2c	kBNAH/BzImH: 5.5, 9.7, 7.2a,b,c
AcrH2	73.0	81.1	too slow	too slow	too slow	less than 5% decay after 1 h
DHA	76		no reaction	no reaction	no reaction	no decay (t = 1 h, −90 °C)

^ak^OMeAN−2H+.

^bk^OMeAN−Sc3+.

^ck^OMeAN−2BR3.

Two mechanistic scenarios are possible: (i) stepwise H^• transfer (HAT) during the r.d.s. followed by e⁻ transfer or (ii) a one-step hydride (H⁻) transfer during the r.d.s. We previously reported that a copper(I)-dioxygen adduct, formally a cupric-superoxide species, also preferentially effected the energetically more difficult H^•-abstraction in preference to hydride abstraction chemistry.²⁷ In order to distinguish between these two possibilities, the reaction between 1,3-dimethyl-2,3-dihydrobenzimidazole (BzImH) and the O^MeAN-LA series was tested. BzImH is a better hydride donor (49.5 kcal/mol heterolytic C–H BDE, vs 64.2 kcal/mol for BNAH) but a poorer H^• donor (73.4 kcal/mol homolytic C–H BDE vs 70.7 kcal/mol for BNAH). For all the O^MeAN-LA complexes, slower reaction rates for BzImH were observed (5–10 times slower, see Table 1), consistent with HAT during the r.d.s. (Figure 7) rather than concerted hydride transfer.

The reactivity described herein is consistent with previous reports that suggest that O-type species rather than ^SP carry out the intra- or intermolecular oxidation of C–H bonds.^14,20,21 Although the difference in the reaction rates found for the different O^MeAN-LA species was modest (the O^MeAN-2BR3 adduct was 2–3 times faster for all the substrates tested), it is worth mentioning that subtle changes in the stereoelectronic properties of the O^MeAN-LA cores, tuned by the use of different Lewis acids, could lead to differential and useful oxidation kinetics/chemistries. Our findings are in accordance with the data reported by Goldberg and co-workers, where the Mn^IV(O):B(C₆F₅)₃ adduct ([(TBP₈Cz^•+)Mn^IV(O):B(C₆F₅)₃]) (TBP₈Cz = octakis(p-tert-butylphenyl)corrolazinato) was found to perform the oxidation of C–H and O–H bonds via HAT faster than the corresponding Mn^IV(O):Zn²⁺ adduct.^40,46

Reactivity of O^MeAN-LA toward O–H Bonds

We have also explored the reactivity of the ^SP^MeAN and O^MeAN-LA toward phenols. As previously reported, ^SP^MeAN was found to be unreactive toward 2,4-^tBu₂-phenol. As shown in Table 2, we tested the reactivity of ^SPMeAN/O^MeAN-LA with a series of 4-substituted phenols with different O–H bond strengths and varying one-electron redox potentials (E⁰). The ^SP^MeAN complex was unreactive toward all of the phenols tested. However, the family of O^MeAN-LA cores was found to be reactive in the 1e⁻/1H⁺ oxidation of phenols with BDEs lower than 84.6 kcal/mol and E⁰ values lower than 1.12 V (Table 2, see Supporting Information for product quantification). The oxidation of 2,4-di-tert-butylphenol (2,4-^tBu-PhOH) by the O^MeAN-LA cores led to the formation of the C–C coupling product (see Supporting Information for details on its detection and quantification).

Table 2.

Kinetic Parameters of O–H Oxidation Reactions by O^MeAN-LA

substrate	BDE (H•) (kcal/mol)	E0 (V vs Fc+/0)	kOMeAN32H+ (M−1 s−1)	kOMeAN3Sc3+ (M−1 s−1)	kOMeAN32BR3 (M−1 s−1)	notes
4-MeO-PhOH	80.9	1.01	64 ± 1	53 ± 1	2.9 ± 0.2	a
4-MeO-PhOD		1.01	52 ± 1 (KIE: 1.25)	42 ± 1 (KIE: 1.19)	2.4 ± 0.1 (KIE: 1.17)	a
2,4-tBu2-PhOH	81.9	1.04	1.8 ± 0.2	1.1 ± 0.1	0.12 ± 0.01	yield: 60–70%
4-PhO-PhOH		1.07	0.51 ± 0.02	0.44 ± 0.02	0.046 ± 0.001	a
4-tBu-PhOH		1.10	0.025 ± 0.001	0.023 ± 0.001	0.0027 ± 0.0003	a
4-Me-PhOH	84.6	1.12	0.035 ± 0.002	0.027 ± 0.001	0.0024 ± 0.0002	a
4-Cl-PhOH	85.7	1.21	no reaction	no reaction	no reaction	no decay (t = 1 h)

^aProduct not quantified (see text and Supporting Information).

Kinetic analyses of these reactions (Figure 8A) led to pseudo-first-order behavior with varying phenol concentrations and gave linear second-order plots allowing for the calculation of second-order rate constants, which are listed in Table 2. Overall, the trends found in phenol (O–H) oxidation are reversed from BNAH/BzImH (C–H) oxidation; the O^MeAN-2H⁺ and O^MeAN-Sc³⁺ complexes exhibit higher reaction rates (10–20 times) than the corresponding O^MeAN-2BR₃ complex. When the reaction rates found were plotted against the 1e⁻reduction potentials of the phenols ((RT/F)·ln(k) vs E⁰), a linear correlation was obtained (Figure 8A, right).^35,49,50

Figure 8.

O–H oxidation performed by O^MeAN-LA cores: (A) kinetic analysis; (B) possible reaction mechanisms and comparison of KIEs (see Figure S18) and reduction potential effects to literature examples.

According to Marcus theory, the slope value obtained in this correlation can be used to distinguish between pure H^• transfer (HAT: slope = 0.0) from electron transfer (ET) or proton-coupled electron transfer (PCET: slope = −0.5 if ΔG⁰_et < 0; slope = −1.0 if ΔG⁰_et > 0).⁴⁹ The low KIE’s observed (1.1–1.2, see Figure S18) along with the slope values (around −1.0) suggest a mechanistic scenario where endergonic electron transfer (ET) is followed by fast proton transfer (PT). This proposal is in agreement with the 1e⁻ reactivity observed with substituted Fc (see Supporting Information), where the O^MeAN-LA complexes were only reduced by donors with E⁰ equal or lower than −0.12 V (e.g., Me₂Fc), a value substantially lower (by 1.0–1.5 V) than the phenol 1e⁻ oxidation potential. (The O^MeAN-LA complexes were not reduced by Fc, E⁰ = 0.0 V vs Fc^+/0.) An alternative mechanism with rate-limiting PT followed by 1e⁻ oxidation of the phenolate is not consistent with the large negative slope of the rate/potential correlation or the small primary H/D isotope effect.

The mechanistic study of the 1e⁻/1H⁺ oxidation of phenols by LCu/O₂ derived species was first reported by Itoh and coworkers, where ^SP and O species were found to oxidize 4-substituted phenols via proton-coupled electron transfer (PCET) with somewhat smaller slopes (−0.72 and −0.71, respectively) and similar KIEs (1.2–1.6 and 1.2–1.5, respectively), where the smaller slope indicates less endergonic ET such that ET and PT are kinetically similar.⁴⁹ Similarly, Costas and co-workers reported that end-on peroxo species are also able to oxidize phenols via PCET (slope = −0.49, KIE: 1.5–2.0), where the slope of ~ −0.5 is consistent with exergonic ET.⁵¹ Very recently, we reported that a mononuclear end-on copper(II)-superoxide intermediate (^ES) oxidizes phenols via HAT (slope: −0.29, KIE: 9.0).³⁵ The results reported herein (stepwise ET-PT, slope = −1.0, KIE: 1.1–1.3) are unprecedented for the reactivity shown by Cu/O₂-derived species, in particular the ability to perform substrate oxidation that involves endergonic ET.⁵² Furthermore, compared to ^SP^MeAN, which is unreactive toward phenols, the enhanced oxidative reactivity of O^MeAN-LA suggests that one important consequence of Lewis acid binding is to raise the reduction potential of the Cu₂O₂ core to allow heightened oxidative reactivity. Thus, Lewis acid coordination is a strategy that can not only effect the core structure (^SP/O equilibrium) but also may tune the reduction potential of the Cu₂O₂ core to facilitate substrate oxidation.

CONCLUSIONS

In this paper, we showed that the unreactive Cu₂O₂ core of ^SP^MeAN could be activated by substrate binding (i.e., sodium 4-R-phenolates) and by Lewis acid addition triggering O–O cleavage to form O^MeAN cores. First, phenolate binding to the copper ion (primary coordination sphere) led to formation of a fleeting phenolate-bound dicopper(III) species (O^MeAN-RPhO⁻) which was characterized by UV–vis and rR spectroscopies. TD-DFT calculations provided evidence that sequential phenolate binding as an axial ligand and isomerization to the equatorial form occurs before the r.d.s., where the O core electrophilically attacks the phenyl ring.

Second, we demonstrated that addition of Lewis acids to ^SP^MeAN also led to O–O cleavage (O^MeAN-LA formation) with coordination of the different Lewis acids occurring at the secondary coordination sphere via binding to what was the axial N of the MeAN ligand. Importantly, the oxidative properties of the O^MeAN-LA species toward external substrates were influenced by the Lewis acid identity. Mechanistic studies provided evidence that C–H oxidation occurred via hydrogen atom transfer, with the borane adduct O^MeAN-2BR₃ reacting slightly faster than did the O^MeAN-2H⁺and O^MeAN-Sc³⁺ cores. Conversely, phenol oxidation was found to happen via stepwise electron-transfer proton-transfer (ET-PT), with the O^MeAN-2H⁺ and O^MeAN-Sc³⁺ cores reacting much faster than the O^MeAN-2BR₃ analogue.

These findings can inspire new routes to develop oxidation systems based on first row metals and dioxygen by utilizing Lewis acids to fine-tune the stereoelectronic properties of the reactive intermediates, leading potentially to novel selectivity patterns. Moreover, these results emphasize the importance of understanding the influence of the secondary as well as the primary coordination spheres in order to explain the speciation (type of Cu/O₂ species) and reactivity of metalloenzymes such as pMMO.