Properties of the one-electron oxidized copper(II) salen-type complexes: relationship between electronic structures and reactivities

Syntheses and redox behavior of Cu(II) complexes

The N₂O₂-donor ligands (Fig. 2) reacted with Cu(II) ion to give the corresponding Cu(II) salen-type complexes as crystals. The salen-type complexes of asymmetric phenolates were also prepared by a step-by-step method by use of the monohydrochloride salt of cyclohexyldiamine as shown in Scheme 1 [14].

Scheme 1

Synthetic route to asymmetric salen-type complexes [26].

The cyclic and differential pulse voltammetry (CV and DPV) results of Cu(II)-salen type complexes showed two reversible redox waves, whose potentials are similar to each other within the 0.25 V range as listed in Table 1. Among the symmetric complexes (R1 = R2), those bearing methoxy groups are oxidized at lower potentials than those bearing i-PrS groups. This order matches the electron-donating properties expected from Hammett σ⁺ constants of these substituents [29]. For the nonsymmetric complexes (R1 ≠ R2), the two redox couples observed can be correlated to the sequential oxidation of the two phenolates: The more electron-rich phenolate is oxidized at a lower potential and the less electron-rich phenolate at a higher potential [26]. These results can be seen as two phenolate-based oxidation processes. On the other hand, the potentials of a 6-membered chelate complex, Cu(1,3-salcn), were only 0.03 V lower than those of the 5-membered chelate Cu(1,2-salcn). These redox waves were found to be one-electron processes from the current analysis for each wave. This result suggests that the one-electron oxidized complexes can be generated by adding an equimolar amount of chemical oxidants such as AgSbF₆ [30]. Indeed, the addition of an equimolar amount of AgSbF₆ (E = 0.65 V vs. Fc/Fc⁺ in CH₂Cl₂) to the salen complexes in CH₂Cl₂ caused a color change, indicating formation of oxidized species.

Solid-state characterization of the oxidized Cu(II) complexes

Crystal structures

When the solutions of some of oxidized complexes prepared by AgSbF₆ oxidation were kept at low temperature in CH₂Cl₂/toluene, crystals of the one-electron oxidized complexes could be isolated. The X-ray crystal structure analysis gave the structures of [Cu(1,2-salcn)]SbF₆, [Cu(1,2-MeO₂-salcn)]SbF₆, and [Cu(1,3-salcn)]SbF₆ (Fig. 3) [23–25]. All complexes exhibit a pseudo square-planar geometry formed by two phenolato oxygen atoms and two imino nitrogen atoms. The Cu coordination sphere geometry of [Cu(1,2-salcn)]⁺ is symmetric with contraction in comparison with neutral complex; on average the Cu–O and Cu–N bond lengths decrease by 0.04 and 0.03 Å, respectively, upon oxidation [23]. On the other hand, the coordination geometries in [Cu(1,2-MeO₂-salcn)]⁺ and [Cu(1,3-salcn)]⁺ are asymmetric, indicating that one salicylideneimino moiety becomes a weaker donor in these complexeswith a longer bond to the Cu atom [24, 25]. The C–O bond length of the weakly donating phenolate moiety is shortened, while such a shortening of the C–O bond in [Cu(1,2-salcn)]⁺ could not be detected. This quinoid bonding pattern of [Cu(1,2-MeO₂-salcn)]⁺ and [Cu(1,3-salcn)]⁺ are consistent with radical localization to one side of the molecule, in agreement with the bonding pattern observed for other localized phenoxyl radical complexes, whereas the solid state structure of [Cu(1,2-salcn)]⁺ suggests that it is a phenolate complex [31]. In addition, the SbF₆^– counterion was located close to the copper ion in [Cu(1,2-salcn)]⁺, while a close contact with the quinoid moiety was observed in [Cu(1,2-MeO₂-salcn)]⁺ and [Cu(1,3-salcn)]⁺. These results support that [Cu(1,2-salcn)]⁺is assigned to the Cu(III)-phenolate species, while [Cu(1,2-MeO₂-salcn)]⁺ and [Cu(1,3-salcn)]⁺ are assigned to the Cu(II)-phenoxyl radical species.

Fig. 3

Structures of Cu(III)-phenolate complex, [Cu(1,2-salcn)]SbF₆, and Cu(II)-phenoxyl radical complexes, [Cu(1,3-salcn)]SbF₆ and [Cu(1,2-MeO₂-salcn)]SbF₆.

Magnetic properties

The temperature-dependent magnetic susceptibility of Cu(Salcn) complexes displayed a response typical for a simple d⁹ Cu(II) paramagnet, and the data were fit to a Curie–Weiss law expression (C = 0.405, θ = –0.554 K). The data for a powdered crystalline sample of [Cu(1,2-salcn)]SbF₆ show that this sample is essentially diamagnetic (μ_eff = 0.3 μ_B) in the temperature range 5–300 K. They were fit with a ca. 4 % impurity mainly due to the neutral Cu(II) complex as estimated from the low-temperature data, which is consistent with a small amount of decomposition in the solid sample. These results suggest that [Cu(1,2-salcn)]SbF₆ in the solid state has a diamagnetic electronic ground state (S = 0) with no thermally accessible triplet state at 300 K [23]. The electronic structure of [Cu(1,2-salcn)]SbF₆ was also supported by XAFS measurement. The pre-edge of the solid sample of [Cu(1,2-salcn)]SbF₆ was found to be more than 1 eV higher than that of neutral Cu(1,2-salcn), clearly indicating that the solid sample of the [Cu(1,2-salcn)]SbF₆ has a metal-centered oxidation state.

The methoxy-substituted complexes, [Cu(1,2-MeO₂-salcn)]⁺ and [Cu(1,2-t-Bu,MeO-salcn)]⁺, are EPR silent, since large zero-field splitting (ZFS) parameters are thought to render the one-electron oxidized species EPR silent at X-band frequencies [25, 26]. On the other hand, the EPR spectra for oxidized complexes [Cu(1,2-i-PrS₂-salcn)]⁺ and [Cu(1,2-t-Bu,i-PrS-salcn)]⁺ exhibit markedly different behavior. One-electron oxidation does not result in a loss of EPR intensity, or rather, the signals broaden considerably, and the spin integration remains relatively constant at that of their parent neutral forms [26]. Thus, despite the similarities of the EPR spectra of the species before oxidation, significant substituent-dependent differences exist between the spectra of their one-electron oxidized species.

The Q-band spectrum of [Cu(1,3-salcn)]SbF₆ at 12 K clearly consists of a two-line pattern centered at g = 2.06, as well as a half-field ΔM_S = ±2 transition at g = 4 (Fig. 4a) [24]. This signature is characteristic of a spin triplet system having ZFS parameters smaller than the Q-band energy quantum (~1.1 cm^–1). The spectrum could be satisfactorily simulated by using the ZFS parameters |D| = 0.470 ± 0.035 cm^–1, E/D = 0.06 ± 0.02, and g_iso = 2.06. These parameters were additionally used for simulation of the X-band spectrum, and a reasonable agreement with the experimental data has been attained. From these EPR results, the triplet spin state of a Cu(II)-phenoxyl salen complex could be evidenced experimentally. The ZFS parameters reported here are the first measured experimentally for Cu(II)-phenoxyl salen complexes.

Fig. 4

(a) Q-band EPR spectrum of a frozen 0.02 M CH₂Cl₂ solution of [Cu(1,3-salcn)]SbF₆: The solid black line is the experimental spectrum, and the dotted red line is a simulation of the triplet system using the parameters given in the text. The asterisk denotes a paramagnetic mononuclear copper (S = 1/2) impurity present in the sample; (b) temperature-dependent magnetic susceptibility of [Cu(1,3-salcn)]SbF₆. Red dots are actual measurements, and the black line is a fitting based on the Bleaney–Bower’s equation.

The temperature dependent magnetic susceptibility of [Cu(1,3-salcn)]SbF₆ in the solid state exhibited a decrease in the susceptibility as the temperature was lowered (Fig. 4b). This fitting is contradictory to the result of EPR experiments. The best-fit parameters are J/k_B = 4.1 cm^–1 and θ = –3.4 cm^–1 using the values from the results of EPR measurements, g = 2.06 and D/k_B = 0.47 cm^–1. These results indicate that the intramolecular interaction between the Cu and radical unpaired electron spins (J/k_B) is ferromagnetic, while the intermolecular interaction (θ) is weakly antiferromagnetic. This intramolecular coupling value matches with the EPR experiments. Thus, complex [Cu(1,3-salcn)]⁺ can be assigned to a Cu(II)-phenoxyl radical species with a S = 1 ground state exhibiting a weak ferromagnetic interaction between the Cu(II) center and the ligand radical [32, 33].

Characterization of the oxidized complexes in solution

The electronic spectrum of Cu(1,2-salcn) is typical of a Cu(II) d⁹ complex with an intense CT transition at 26 500 cm^–1 (ε = 11 600 M^–1 cm^–1) and a weak d-d transition at 17 600 cm^–1 (ε = 600 M^–1 cm^–1) . As reported previously, oxidation to [Cu(1,2-salcn)]⁺ at room temperature results in the appearance of two new bands, an intense band at 18 600 cm^–1 (ε = 6500 M^–1 cm^–1) and a low-energy transition at 5800 cm^–1 (ε = 2900 M^–1cm^–1) (Fig. 5a). Interestingly, the 18 600 cm^–1 band exhibits large intensity changes with temperature; from 298 to 190 K, this band approximately doubles in intensity and shifts slightly to higher energy. This change is fully reversible and indicates that [Cu(1,2-salcn)]⁺ is involved in a temperature-dependent equilibrium in solution. The intensity of the 557-nm absorption band for [Cu(1,2-salcn)]⁺ is independent of concentration at both 295 and 203 K, which excludes dimerization as a possible reason for the observed temperature-dependent changes in band intensity. The ¹H NMR spectrum of [Cu(1,2-salcn)]⁺ in CD₂Cl₂ at 298 K displayed broad signals over a wide spectral range, indicating the presence of a substantial amount of a paramagnetic species. Solution susceptibility measurements by the Evans method indicate a χ_mT value of 0.442 cm³mol^–1 K (1.1 ± 0.1 unpaired electrons; spin-only calculation) at room temperature. An interpretation of this bulk susceptibility is the presence of approximately equimolar amounts of a ferromagnetically coupled Cu(II)-phenoxyl radical species and a diamagnetic counterpart Cu(III)-phenolate [23].

Fig. 5

UV–vis spectra of complexes Cu(1,2-salcn) and Cu(1,3-salcn) in CH₂Cl₂ before and after oxidation: (a) complex Cu(1,2-salcn), (b) complex Cu(1,3-salcn). Black line: neutral complexes, red line: one-electron oxidized complexes.

On the other hand, the UV–vis–NIR absorption spectrum of [Cu(1,3-salcn)]⁺ in CH₂Cl₂ is different from that of [Cu(1,2-salcn)]⁺ (Fig. 5b). The NIR bands of [Cu(1,3-salcn)]⁺ are of significantly lower intensity compared to the NIR transitions for the Cu(III) ground-state complex [Cu(1,2-salcn)]⁺ and fully delocalized ligand radical complexes [23, 34]. From the low intense NIR bands, complex [Cu(1,3-salcn)]⁺ can be assigned to the class II localization system based on the Robin and Day classification [35]. Further, the resonance Raman spectrum of [Cu(1,3-salcn)]⁺ in CH₂Cl₂ showed the typical phenoxyl radical ν_7a band at 1492 cm^–1 [24]. Therefore, the absorption spectral features of [Cu(1,3-salcn)]⁺ clearly suggest that complex [Cu(1,3-salcn)]⁺ is a localized Cu(II)-phenoxyl radical species in solution.

The majority of Cu-salen phenoxyl radical complexes with methoxy and isopropylthio substitution display Class II-like behavior, regardless of the phenolate substitution pattern. Recent reports on the band-shape analysis of a series of nonsymmetric Cu-salen phenoxyl radical complexes indicate that the electron coupling coefficient values of H_AB, whose average is at 2100 ± 200 cm^–1, remain rather constant despite the different functional groups and the wide range of IVCT ν_max values [14, 26]. From the relationship between the redox potential and the IVCT band of the oxidized asymmetrical salcn complexes, the absorption spectral feature of the active form of GO can be explained. The active site of GO has two different phenolate donors, one of which is the Tyr 272 residue having an o-alkylsulfanyl moiety. Assuming the difference of the oxidation potentials between the two phenolates (ca. 600 mV ≈ 4800 cm^–1) and the sum of their reorganizational energies, the LLCT band at ca. 13 600 cm^–1 is predicted for the active form of GO. This analysis is admittedly very approximate but very well within the envelope of the broad visible band of the enzyme [26].

Reactivity of the oxidized complexes with benzyl alcohol

Some of the salen-type Cu(II) complexes have been reported to have a GO-like catalytic reactivity for oxidation of primary alcohols, while the mechanism of the oxidation is still unclear [36–39]. On the other hand, Stack et al. have reported a detailed reaction mechanism of the one-electron oxidized salen-type complexes including [Cu(1,2-salcn)]⁺ [40, 41]. However, the electronic structure of [Cu(1,2-salcn)]⁺ was assigned simply to “a Cu(II)-phenoxyl radical”, which is slightly different from that discussed above. Recently, Thomas et al. have reported that phenoxyl radical complex [Cu(1,2-MeO₂-salcn)]⁺ has a high reactivity for oxidation of benzyl alcohol, but the ionic strength of the solution was quite different due to generation of the oxidized species by electrolysis [25]. Therefore, we studied the detailed reaction mechanism of [Cu(1,3-salcn)]⁺ and compared it with that of [Cu(1,2-salcn)]⁺, in order to discuss the relationship between electronic structure and reactivity under the same conditions.

CH₂Cl₂ solutions of the oxidized complexes [Cu(1,2-salcn)]⁺ and [Cu(1,3-salcn)]⁺are relatively stable at 293 K. Addition of excess benzyl alcohol to the solutionscaused a color change within a few hours. GC-MS and absorption spectral changes revealed that benzaldehyde was formed in 50 % yield relative to the initial amount of the oxidized complexes. This reaction was also observed under a nitrogen atmosphere, which indicates that the presence of air did not affect the benzyl alcohol oxidation. The UV–vis absorption spectral change showed that the spectra of the final products are different from those of Cu(1,2-salcn) and Cu(1,3-salcn). However, addition of one equivalent of Et₃N to the solutions of the final products afforded spectra which are similar to those of 1 and 2, indicating that the final product of this reaction can be assigned to the proton adducts of the Cu(II) complexes, [Cu(H1,2-salcn)]⁺ and [Cu(H1,3-salcn)]⁺. The overall stoichiometry of the reaction observed is expressed as follows [24]:

In the presence of a large excess of benzyl alcohol, conversion of [Cu(1,2-salcn)]⁺ and [Cu(1,3-salcn)]⁺ can be regarded as a first-order process, suggesting that only one molecule of [Cu(1,2-salcn)]⁺ and [Cu(1,3-salcn)]⁺, respectively,is involved in the rate-determining step. The first-order decay constant (k_obs) of complex [Cu(1,3-salcn)]⁺ was not changed by varying the concentration of the complex (k_obs = (3.61–4.06) × 10^–3 s^–1 for 0.1–0.3 mM solution of [Cu(1,3-salcn)]⁺ at [C₆H₅CH₂OH] = 1.0 M), while the rate constant depended on the benzyl alcohol concentration. A plot of k_obs against the concentration of benzyl alcohol exhibited a linear correlation, i.e., k_obs = k₂[C₆H₅CH₂OH], giving the second-order rate constant k₂ = (3.89 ± 0.05) × 10^–3 M^–1s^–1 at 293 K for the rate-determining process. We could not determine the association constants of complexes [Cu(1,2-salcn)]⁺ and [Cu(1,3-salcn)]⁺ with benzyl alcohol, due to the linear correlation. The reactivity characteristics of [Cu(1,3-salcn)]⁺ are very similar to the reaction of complex [Cu(1,2-salcn)]⁺ with benzyl alcohol, although the second-order rate constant of [Cu(1,3-salcn)]⁺ is ca. 4 times larger than that of [Cu(1,2-salcn)]⁺ (k₂ = 1.02 × 10^–3 M^–1s^–1 at 293 K was obtained at [C₆H₅CH₂OH] = 1.0 M). From the kinetic analyses, the primary alcohol oxidation by phenoxyl radical complex [Cu(1,3-salcn)]⁺ likely occurs by a mechanism similar to that of [Cu(1,2-salcn)]⁺ but with a faster reaction rate.

In order to determine whether C–H bond cleavage is involved in the rate-determining step, we carried out kinetic analyses using α-benzyl-d₂ alcohol. The kinetic isotope effect (KIE), k₂(H)/k₂(D), for complex [Cu(1,2-salcn)]⁺ was estimated to be 19 at 298 K [40, 41], and the KIE value of [Cu(1,3-salcn)]⁺ was estimated to be 13 at 293 K [24], which indicates that the C–H bond scission is included in the rate-determining step. On the other hand, the reaction rate for C₆H₅CH₂OD (k₂ = 3.6 × 10^–3 M^–1s^–1 obtained at [C₆H₅CH₂OD] = 1.0 M) was very similar to that of C₆H₅CH₂OH, indicating that the benzyl alcohol oxidation proceeds without the O–H hydrogen abstraction in the rate-determining step.

The Eyring plots of complexes [Cu(1,2-salcn)]⁺ and [Cu(1,3-salcn)]⁺ showed linear fits with similar slopes. The activation parameters of [Cu(1,2-salcn)]⁺ were determined to be ΔH^‡ = 10.7 ± 0.1 kcal mol^–1 and ΔS^‡ = –35.6 ± 0.3 cal mol^–1K^–1, and those of [Cu(1,3-salcn)]⁺ were estimated to be ΔH^‡ = 10.2 ± 0.1 kcal mol^–1 and ΔS^‡ = –34.8 ± 0.5 cal mol^–1K^–1 (Table 2). The difference of the activation energy (ΔE_a) between [Cu(1,2-salcn)]⁺ and [Cu(1,3-salcn)]⁺ is estimated to be 0.6 ± 0.2 kcal mol^–1 at 293 K. This value is in good agreement with the C–H activation energy difference estimated from the KIE values (Δ(ΔE_a^KIE) = 0.4 kcal mol^–1) [42]. Thus, the activation parameter difference between [Cu(1,2-salcn)]⁺ and [Cu(1,3-salcn)]⁺ is relatively small, suggesting a similar reaction mechanism for [Cu(1,2-salcn)]⁺ and [Cu(1,3-salcn)]⁺, as well as similar geometries for the substrate association.

The increased reaction rate for [Cu(1,3-salcn)]⁺ in comparison to [Cu(1,2-salcn)]⁺ can be attributed to the different electronic structure, i.e., a Cu(II)-radical for [Cu(1,3-salcn)]⁺ and a Cu(III) ground state for [Cu(1,2-salcn)]⁺ which is in equilibrium with the Cu(II)-ligand radical in solution at room temperature. The constant k₂ for the reaction of [Cu(1,2-salcn)]⁺ corresponds to the rate constant of a simple bimolecular process between Cu(II)-radical and C₆H₅CH₂OH (k₂*) in the rate-determining step (k₂ = k₂*), while in the reaction of [Cu(1,2-salcn)]⁺, k₂ should involve an equilibrium between Cu(III) and Cu(II)-radical species (K_pre= [Cu(II)-radical]/[Cu(III)]) as expressed by the following equation [42]:

Since complex [Cu(1,2-salcn)]⁺ has been reported to exist in ca. 1:1 equilibrium between Cu(III)-phenolate and Cu(II)-phenoxyl radical species, K_pre is estimated to be ≈1, and hence k₂ for the reaction of [Cu(1,2-salcn)]⁺ is calculated to be k₂ ≈ k₂*/2. If the Cu(III) species does not react with benzyl alcohol, the k₂ value for the reaction of [Cu(1,3-salcn)]⁺ is at least double the value for [Cu(1,2-salcn)]⁺ on the assumption that k₂* is the same for both systems [23].

From these results and discussion, the reaction mechanisms for the benzyl alcohol oxidation are summarized as shown in Scheme 1. The reaction mechanisms of the two [Cu(Salcn)]⁺ species are very similar. The rate-determining step of these complexes involves the hydrogen atom abstraction from the α-position of benzyl alcohol. Complex [Cu(1,3-salcn)]⁺ has a slightly faster reaction rate in comparison to [Cu(1,2-salcn)]⁺ (Scheme 2). The slower reaction rate of [Cu(1,2-salcn)]⁺ is mainly due to the existence of the equilibrium between the Cu(III)-phenolate and Cu(II)-phenoxyl radical species. In addition, it is plausible that the higher reactivity of the Cu(II)-radical of [Cu(1,3-salcn)]⁺ toward C₆H₅CH₂OH may be due to the increased conformational flexibility of the 6-membered chelate ring of [Cu(1,3-salcn)]⁺ in comparison to the 5-membered chelate 1,2-cyclohexanediamine backbone of [Cu(1,2-salcn)]⁺.

Scheme 2

Proposed mechanisms for benzyl alcohol oxidation: (a) simple hydrogen abstraction by a localized phenoxyl radical such as complex [Cu(1,3-salcn)]⁺; (b) reaction scheme for the equilibrium between less reactive Cu(III)-phenolate and active Cu(II)-phenoxyl radical species in the reaction of complex [Cu(1,2-salcn)]⁺.