O₂ Activation and Selective Phenolate ortho Hydroxylation by an Unsymmetric Dicopper μ-η¹:η¹-Peroxido Complex

Understanding the intimate details of O₂ activation at metal sites is of interest because of the relevance of such reactions in biological and technological processes.^[1] Of particular relevance is uncovering basic chemical principles and mechanisms for taming the high oxidizing potential of the O₂ molecule into highly selective oxidative transformations, especially those involving the selective hydroxylation of C–H bonds.

For the particular case of dicopper sites, three basic Cu₂O₂ core structures have been widely described as arising from the interaction of discrete Cu^I complexes and O₂ (Figure 1).^[2] Each specific Cu₂O₂ core determines particular spectroscopic and chemical properties:^[2b,c] whereas end-on trans-Cu^II₂-(μ-η¹:η¹-O₂) species exhibit nucleophilic and basic behavior, side-on Cu^II₂(μ-η²:η²-O₂) and bis-μ-oxido dicopper(III) (Cu^III₂(μ-O)₂) cores show electrophilic character, and can mediate tyrosinase-like phenolate ortho-hydroxylation reactions.^[3] The latter reactivity has never been observed for end-on Cu₂O₂ species; therefore its possible biological relevance has been ignored so far.

Figure 1.

The rich and subtle chemistry exhibited by Cu₂O₂ cores makes unsymmetric options interesting. Actually, Cu₂O₂ species in systems containing distinct copper sites have been rarely observed.^[2b] Herein we describe a novel dicopper complex based on a heptadentate ligand that gives rise to an unsymmetric N₃Cu^IIN₄Cu^II(μ-η¹:η¹-O₂) core, which hitherto exhibits reactivity patterns not observed for symmetric analogues. This nonsymmetric peroxide species shows an exquisite selectivity in its oxygen atom transfer reactivity. It performs the selective intermolecular ortho hydroxylation of a phenolate, but fails to oxidize many common oxophilic substrates.

Reaction of m-Xyl^N3N4 with [Cu^I(CH₃CN)₄X] (X = CF₃SO₃, PF₆, ClO₄) in acetonitrile affords the unsymmetric dinuclear copper(I) complex [Cu^I₂(m-Xyl^N3N4)](X)₂ (1-X; Figure 2). For comparative purposes, [Cu^I₂(m-Xyl^N4N4)]- (ClO₄)₂ (2-ClO₄) was also prepared. Crystallographic characterization of 1-CF₃SO₃ reveals that the copper ion bound to the tridentate arm adopts a highly distorted T-shape geometry, whereas the copper ion bound to the tetradentate arm has a distorted trigonal-pyramidal geometry, with structural parameters nearly superimposable with those of the two tetracoordinated copper ions in 2-ClO₄.^[4]

Figure 2.

Top: Chemical diagram of 1-X (left) and 2-ClO₄ (right). Bottom: Ellipsoid diagrams (30 % probability) of the cationic parts of 1-CF₃SO₃ (left) and 2-ClO₄ (right). H atoms were omitted for clarity.

Acetone solutions of 1-X at −90°C react with O₂ within seconds to form a red-brown species [Cu₂(O₂)(m-Xyl^N3N4)]²⁺ (1-O₂), characterized by a visible band at 478 nm (ε = 7800m⁻¹ cm⁻¹), and a broad shoulder between 575 and 700 nm (Figure 3). The visible spectrum of 1-O₂ is intermediate between those of Itoh’s proposed Cu^II₂(μ-η¹:η²-O₂) species^[5] and reported Cu^II₂(μ-η¹:η¹-O₂) complexes.^[2b,6] UV/Vis monitoring of this reaction shows an isosbestic point at 414 nm indicating the clean transformation of 1-CF₃SO₃ into 1-O₂ without accumulation of any intermediate species. 1-O₂ is stable for hours at −90°C, but it decomposes within seconds when warmed to room temperature.

Figure 3.

Top: UV/Vis spectra for the reaction of 1-CF₃SO₃ with O₂ in acetone at −90°C to form 1-O₂. Bottom: Resonance Raman spectra (λ_ex = 488 nm) of frozen acetone solutions of 1-O₂ from ¹⁶O₂ (A), ¹⁸O₂ (B), and ¹⁸O¹⁶O (C).

To gain insight into the peroxido binding mode of 1-O₂, resonance Raman spectra of frozen samples of 1-O₂ were obtained. Laser excitation at 488 nm (Figure 3 bottom, insets A and B) gives rise to two resonance-enhanced peaks at 832 and 520 cm⁻¹ [Δ¹⁸O₂-¹⁶O₂ = 45 and 22 cm⁻¹, respectively], characteristic of O–O and Cu–O stretching vibrations of an end-on Cu^II₂(μ-η¹:η¹-O₂) species.^[2b] Excitation profiles indicate that the two vibrations are in resonance with the lower energy band, and no features resulting from a Cu^II₂(μ-η²:η²-O₂) species were observed. Experiments with mixed labeled O₂ (Figure 3 bottom, inset C) showed a ν(O–O) region with three isotopomeric peaks at frequencies that suggest insensitivity of the bound O₂ to the unsymmetrical nature of the ligand.^[7]

In contrast, symmetric complex 2-ClO₄ reacts at −90°C in acetone to form a different metastable purple species, 2-O₂, which is characterized by two intense UV/Vis bands at λ_max = 500 nm (ε = 5000m⁻¹ cm⁻¹) and 635 nm (ε = 3300m⁻¹ cm⁻¹), typical of an end-on trans-Cu^II₂(μ-η¹:η¹:O₂) species.^[2b] The O₂-binding mode was confirmed by the resonance Raman spectra of a frozen 2-O₂ solution, collected with laser excitation at 488 nm (see the Supporting Information), which shows a characteristic resonance-enhanced ν(O–O) band at 826 cm⁻¹[Δ¹⁸O₂-¹⁶O₂=44 cm⁻¹]. The fact that both 1-O₂ and 2-O₂ give rise to ν(O–O) features of nearly the same frequency strongly suggests that the dioxygen moiety is bound in the same fashion in the two complexes. Because of the high energy of the ν(O–O) stretching frequency,^[2b] and because a μ-η¹:η²-O₂ binding mode is unlikely for 2-O₂,^[8] we favor a μ-η¹:η¹-O₂ mode for both O₂ adducts.

The reactivities of 1-O₂ and 2-O₂ with different substrates were explored (Schemes 1 and 2). 1-O₂ and 2-O₂ rapidly and quantitatively react with CF₃CO₂H releasing H₂O₂ (99% yield, see the Supporting Information). Titration experiments reveal that both reactions are complete with 1 equivalent of H⁺ (per Cu), and no other intermediate species are detected when substoichiometric amounts of H⁺ are added. 1-O₂ and 2-O₂ react neither with thioanisole, styrene, triphenylmethane, nor with electron donors such as ferrocene. Thermal decomposition of 1-O₂, by warming up acetone solutions to room temperature, in the presence of large excess of toluene (1000 equiv), did not cause toluene oxidation.^[9] The addition of PPh₃ (10 equiv) to 1-O₂ and 2-O₂ at −90°C induces fast O₂ release (t_1/2 ≈ 5 min) without formation of OPPh₃.^[10] In sum, all the above observations suggest that 1-O₂ and 2-O₂ are not electrophilic oxidants, and typically react like other end-on trans-Cu^II₂(μ-η¹:η¹-O₂) complexes.^[11]

Scheme 1.

Schematic representation of selected reactivity exhibited by 1-O₂ and 2-O₂. (N.R. = no reaction).

Scheme 2.

Reaction of 1-O₂ with the sodium salt of para-substituted phenolate.

However, substantial differences arise when the reactions of 1-O₂ and 2-O₂ with benzaldehydes are studied. 2-O₂ reacts with benzaldehydes to generate the corresponding benzoic acids in quantitative yields. Kinetic analyses of the reactions were performed by using UV/Vis methods to monitor the decay of the spectral features of 2-O₂. The 2-O₂ decay rate can be fitted to single exponential processes, and the measured k_obs values are linearly dependent on substrate concentration (see the Supporting Information). Reaction of 2-O₂ against a series of para-substituted benzaldehydes was studied and the corresponding decay rate constants were extracted by using UV/Vis methods to monitor the reactions. Plotting the decay rate of 2-O₂ against the corresponding Hammett substituent constants (σ⁺) affords a linear correlation which gives ρ value of 1.4 (R² = 0.98, see the Supporting Information), consistent with a nucleophilic oxidizing species that attacks the carbonyl moiety. In contrast, 1-O₂ fails to react with benzaldehyde. Thus we conclude that 1-O₂ is unreactive in oxygen atom transfer reactions to common substrates which are either electrophilic or nucleophilic in nature.

Strikingly, addition of p-Cl-C₆H₄ONa (3 equiv, Scheme 2) to a solution of 1-O₂ at −90°C causes rapid conversion into a short-lived (t_1/2 ≈ 1 min) yellow-brown species 3^Cl (λ_max = 470 nm, ε > 6000m⁻¹ cm⁻¹, see the Supporting Information). The resemblance in the UV/Vis spectral features of 1-O₂ and 3^Cl strongly suggests that the Cu^II₂(μ-η¹:η¹-O₂) core is retained, but the instability of 3^Cl has thus far precluded its Raman characterization. Surprisingly, after complete decomposition of the 3^Cl species, acidic work-up and subsequent HPLC/MS analyses show the formation of p-chlorocatechol in 39 % yield with respect to 1-O₂. Similar addition of p-chlorophenolate to 2-O₂ causes fast bleaching of its spectral features, without accumulation of any intermediate species, and without any sign of phenolate ortho hydroxylation.

Kinetic analysis indicates that the decay of 3^Cl is a first-order process. The analogous species 3^X (X = F, Me, H, and OMe) were generated by the addition of 3 equivalents of p-X-C₆H₄ONa to 1-O₂ at −90°C in acetone, and their corresponding UV/Vis decay rates were fitted to a single exponential function by nonlinear regression methods. Product analysis after 3^X (X = F, Me, H, and OMe) decomposition reveals that the corresponding catechol is formed in 34 %, 34 %, 36 %, and 14% yields, respectively. A Hammett plot (log (k_obs) for 3^X versus σ⁺) affords a linear correlation which gives a ρ value of −0.6 (R² = 0.98, see the Supporting Information), consistent with an electrophilic oxidizing species that attacks the aromatic ring in the rate-determining step of the reactions. In line with this reactivity, no catechol product was formed when electron-poor phenolates (X = CN, NO₂, and CO₂Me) were used as substrates. In contrast, substrate hydroxylation does not appear to be only determined by the electron-releasing nature of the substrate because the electron-rich, sterically more demanding 2,4-di-tert-butyl-catecholate was neither hydroxylated nor oxidized to the corresponding diphenol coupled product. We conclude that hydroxylation occurs exclusively for non-electron-poor, sterically unhindered phenolate substrates. Furthermore 1-O₂ differs from any other Cu^II₂(μ-η¹:η¹-O₂) intermediate in its capacity to carry out electrophilic arene hydroxylation.^[2c] We propose that the difference in the reactivity of 1-O₂ and any previously reported end-on Cu^II₂(μ-η¹:η¹-O₂) species (including 2-O₂) stems from the possibility that phenolate can initially bind at the N₃Cu site, as proposed in a symmetric m-xylyl-bridged bis-tridentate Cu^II₂(μ-η²:η²-O₂) system.^[3b] Indeed, in the present example, the substrate binding event can be understood as playing a selective peroxide-activation role, since 1-O₂ by itself lacks oxygen atom transfer reactivity.

The selective oxygen atom transfer reactivity exhibited by 1-O₂ was additionally substantiated by DFT computational methods.^[12] The computed structure of 1-O₂ (see the Supporting Information) reveals a Cu^II₂(μ-η¹:η¹-O₂) complex with a Cu⋯Cu distance of 4.31 Å and structural parameters in good agreement with a crystallographically characterized example.[13] We have also found that the Cu^III₂(μ-O)₂ isomer is 36.8 kcal mol⁻¹ higher in energy.[12] In addition, attempts to perform geometry optimizations on side-on Cu^II₂(μ-η²:η²-O₂) isomeric cores proved unsuccessful.[14] Therefore, consistent with the experimental observations, the end-on Cu^II₂(μ-η¹:η¹-O₂) is the most stable species.[14-16] Phenolate binding to 1-O₂ retains the Cu^II₂(μ-η¹:η¹-O₂) core as the most stable isomer (in agreement with our formulation of 3^X based on its UV/Vis spectrum), and causes an elongation of the Cu⋯Cu distance up to 4.50 Å. Interestingly, the phenolate π system is adjacent to the peroxide oxygen atom bound to the other Cu in 3^Me (Figure 4), offering a plausible pathway for a σ* electrophilic attack of the peroxide moiety on the aromatic ring.^[17] Most remarkably, the computed activation barrier for this reaction is only 14.7 kcalmol⁻¹, and no intermediates regarding the isomerization to side-on Cu^II₂(μ-η²:η²-O₂) or Cu^III₂(μ-O)₂ cores are found along this attack, thereby strongly suggesting that the trans end-on peroxido core is a competent species for executing the aromatic C–H hydroxylation event.^[12]

Figure 4.

Stationary points along the reaction pathway of the ortho hydroxylation of para-methyl phenolate into 4-methyl catechol from 3^Me.

In conclusion, our study of O₂ activation at a novel asymmetric dicopper complex 1-O₂ has hitherto uncovered reactivity patterns thus far not observed for symmetric analogues. 1-O₂ is basically unreactive in oxygen atom transfer reactions. However, it has an available coordination site that selectively binds phenolate and mediates its ortho hydroxylation, therefore functionally mimicking tyrosinase through a unique pathway. Furthermore, coordination of the phenolate substrate turns on the unprecedented electrophilic reactivity of the asymmetric end-on trans-peroxido core. The combined experimental and computational evidence indicate that the ortho hydroxylation of a phenolate by a Cu₂O₂ species can occur by adjacent binding of phenolate and O₂ at a common N₃Cu site without requiring the peroxido to be side-on bound, thus offering a conceptually new understanding of O₂ activation at dicopper sites.

Experimental Section

See the Supporting Information for the full experimental details for the synthesis, spectroscopic, and crystallographic characterization of 1-X (X = CF₃SO₃, PF₆, ClO₄) and 2-ClO₄, the experimental procedures for the generation, characterization, and reactivity studies of 1-O₂ and 2-O₂, and the computational details on the DFT calculations.