Copper/TEMPO Redox Redux: Analysis of PCET Oxidation of TEMPOH by Copper(II) and the Reaction of TEMPO with Copper(I)

Abstract

Copper salts and organic aminoxyls, such as TEMPO (2,2,6,6-tetramethylpiperidine N-oxyl), are versatile catalysts for aerobic alcohol oxidation. Previous reports in the literature contain conflicting proposals concerning the redox interactions that take place between copper(I) and copper(II) salts with the aminoxyl and hydroxylamine species, TEMPO and TEMPOH, respectively. Here, we reinvestigate these reactions in an effort to resolve the conflicting claims in the literature. Under anaerobic conditions, Cu^IIX₂ salts [X= acetate (OAc), trifluoroacetate (TFA), triflate (OTf)] are shown to promote rapid proton-coupled oxidation of TEMPOH to TEMPO: Cu^IIX₂ + TEMPOH → Cu^IX + TEMPO + HX. In the reaction with acetate, however, slow reoxidation of Cu^IOAc occurs. This process requires both TEMPO and HOAc and coincides with reduction of TEMPO to 2,2,6,6-tetramethylpiperidine. Analogous reactivity is not observed with trifluoroacetate and triflate species. Overall, the facility of proton-coupled oxidation of TEMPOH by Cu^II salts suggests that this process could contribute to catalyst regeneration under aerobic oxidation conditions.

Graphical Abstract

Reactions between Cu salts and TEMPO species are featured in a number of catalytic reactions. The results in this study clarify a discrepancy in the literature concerning the favored direction of the equilibrium between Cu^IIX₂/TEMPOH and Cu^IX/TEMPO/HX. Cu^IIX₂ salts are shown to promote rapid proton-coupled oxidation of TEMPOH to TEMPO. Previously reported oxidation of Cu^I by TEMPO is attributed to the presence of adventitious O₂ and, to a lesser extent, to background disproportionation of TEMPOH.

Introduction

Combinations of a copper salt and organic aminoxyl, such as TEMPO (2,2,6,6-tetramethylpiperidine N-oxyl), represent some of the most effective catalyst systems for aerobic oxidation of alcohols.^1–11 The most widely used catalysts feature 2,2’-bipyridine (bpy) as the ancillary ligand. The alcohol oxidation reactions often proceed at room temperature with ambient air as the oxidant, undergo chemoselective oxidation of primary alcohols in the presence of secondary alcohols,^7,8 tolerate diverse functional groups, including sulfides and amines,^7–11 and have been applied to process scale applications in the pharmaceutical industry.^12–14 Mechanistic studies have shown that the Cu^II and aminoxyl engage in a cooperative mechanism in which the two one-electron oxidants promote efficient two-electron oxidation of the substrate.^15–22 Following alcohol oxidation, the reduced catalyst system is regenerated by O₂, completing the overall catalytic cycle depicted in Scheme 1.

Scheme 1.

Simplified Mechanism for Cu/TEMPO-Catalyzed Alcohol Oxidation.

Many details of the catalyst oxidation half-reaction are not well understood. Reactions between O₂ and nitrogen-ligated copper complexes have been the subject of extensive fundamental study.^23–27 Reactive Cu/O₂ species, including superoxo, peroxo, and oxo species, are often invoked in hydrogen-atom transfer (HAT) reactions, and HAT from weak O–H bonds, such as those in TEMPOH and phenols, has been demonstrated in several fundamental studies of reactive Cu/O₂ species.^28–31 These precedents prompted us to propose that analogous reactivity could be involved in the catalyst reoxidation half-reaction of Cu/aminoxyl-catalyzed aerobic alcohol oxidations.⁸ In a complementary effort, we demonstrated electrochemical alcohol oxidation with a (bpy)Cu/TEMPO catalyst system,²¹ and the data from this study implicated a mechanism in which Cu^II mediates oxidation of TEMPOH to TEMPO under anaerobic conditions. In both the aerobic and anaerobic mechanisms, oxidation of the hydroxylamine to an N-oxyl radical is proposed to proceed with reduction of Cu^II to Cu^I (Figure 1A).

Figure 1.

Previously proposed reactions between Cu and TEMPO(H), including (A) oxidation of TEMPOH by Cu/O₂ or Cu^IIX₂ and (B) reduction of TEMPO by Cu^I.

Sheldon and coworkers have separately reported that TEMPO reacts with Cu^IOAc in acetonitrile to generate a Cu^II-TEMPO complex (i.e., with a TEMPO^– ligand), evident from the appearance of a 670 nm band in the UV-visible spectrum of the reaction mixture (Figure 1B).¹⁵ The oxidation of Cu^I by TEMPO has been invoked as a mechanism for catalyst reoxidation in several studies of catalytic aerobic alcohol oxidation.^6,15,32 Although several well-defined Cu^II-aminoxyl complexes have been reported in the literature, all of these precedents formally correspond to Cu^II complexes bearing an “aminoxyl”, rather than an “aminoxide” (e.g., TEMPO⁻), ligand.^33–35 Nonetheless, the reaction in Figure 1B provides an alternative mechanism for Cu^I oxidation in the catalytic reactions.

The reactions presented in Figure 1A and 1B are not directly the microscopic reverse of each other, but they nevertheless appear contradictory. Here, we report a study of reactions between copper (I/II) salts and TEMPO(H) in an effort to resolve this apparent contradiction. We show that a series of Cu^IIX₂ salts (X = OAc, TFA, OTf) mediate efficient proton-coupled oxidation of TEMPOH to afford Cu^IX, TEMPO, and HX. We also observe the oxidation of Cu^I in the presence of TEMPO; however, this reactivity is observed with Cu^IOAc, but not with other Cu^ITFA or Cu^IOTf sources. In addition, this reaction requires the presence of AcOH, it exhibits slow rates, and it occurs with cleavage of the N–O bond in TEMPO, generating 2,2,6,6-tetramethylpiperidine as a by-product. Electrochemical and spectroscopic studies of different Cu salts and TEMPO/TEMPOH under similar reaction conditions provide a foundation for understanding these results, all of which suggest that the simple one-electron oxidation of Cu^I by TEMPO is not a favorable reaction and does not contribute to catalytic alcohol oxidation reactions.

Results and Discussion

Reproduction of previous results that appear contradictory.

We began our study by revisiting the reactions between copper(I/II) salts and TEMPO(H) species that were reported previously by us²¹ and by Sheldon and coworkers¹⁵ (Figure 2). The reaction conditions and reagents are somewhat different in the two studies; however, the original conditions were retained in order to enable direct assessment of the results. The first experiment probed the stoichiometric reaction of (bpy)Cu^II(OTf)₂ with TEMPOH and 2,6-lutidine in MeCN under N₂ by UV-visible spectroscopy (Figure 2A). Addition of TEMPOH led to rapid reduction Cu^II to Cu^I, evident from the disappearance of the Cu^II absorption band at 720 nm and the appearance of a strong absorption feature at 440 nm, attributed to bpy-ligated Cu^IOTf. The second experiment probed the reaction of TEMPO with Cu^IOAc in MeCN (Figure 2B; no bpy ligand was included in this reaction in order to match the originally reported conditions¹⁵). A new absorption feature at 670 nm emerged slowly over the course of 4 h, consistent with the formation of Cu^II under these reaction conditions. The λ_max observed in this spectrum matches that of Cu^IIOAc₂ in MeCN, suggesting that the Cu^II product may not correspond to the original proposed [Cu^II–ONR₂] adduct (see further discussion below).³⁶ These observations validate the previously reported observations.

Figure 2.

UV-visible analysis of reactions between copper(I/II) salts and TEMPO(H) species. (A) Oxidation of TEMPOH by (bpy)Cu^II(OTf)₂ under anaerobic conditions, replicating conditions reported by Badalyan et. al.²¹ Conditions: 2 mM TEMPOH, 2 mM Cu^IIOTf₂, 2 mM 2,2¢-bipyridine, 100 mM 2,6-lutidine, MeCN, 273 K, 50 min. (B) Oxidation of Cu^IOAc by TEMPO, replicating conditions reported by Dijksman et. al.¹⁵ Conditions: 2 mM TEMPO, 2 mM Cu^IOAc, MeCN, 273 K, 4 h. Cu^II concentration for Figure 2B was obtained by constructing a UV-visible calibration curve of Cu^IIOAc₂ in MeCN at 670 nm.

Assessment of reactions between copper (I/II) salts and TEMPO(H) species under uniform conditions.

The different reagents in the two reactions, including the presence/absence of bpy as a neutral ligand, the presence/absence of base, and the use of OAc/OTf as an anionic ligand, could influence the reaction outcome. We therefore tested the reaction of TEMPOH with three different Cu^IIX₂ sources (X = OAc, TFA, OTf) in the absence of bpy and base (Figure 3). The reactions proceed rapidly in all cases and lead to reduction of Cu^II, evident by disappearance of the absorption band between 670–760 nm. No further reaction is apparent following oxidation of TEMPOH by Cu^IITFA₂ and Cu^IIOTf₂; however, the reduction of Cu^IIOAc₂ to Cu^IOAc by TEMPOH is followed by a slow reappearance of Cu^II, evident by reappearance of the broad absorption feature at 670 nm. The latter result notwithstanding, these data show that TEMPOH undergoes rapid oxidation by Cu^IIX₂ salts, even in the absence of the 2,2¢-bipyridine and 2,6-lutidine additives present in Figure 3A.

Figure 3.

Analysis of reactions between copper (I/II) salts and TEMPO(H) species with different anions (X = OAc, TFA, OTf). (A) Reduction of Cu^IIX₂ species observed upon addition of TEMPOH. Single wavelength data obtained at 670 nm (OAc, red), 715 (TFA, blue), and 760 (OTf, black). Conditions: 2 mM TEMPOH, 2 mM Cu^IIX₂, MeCN, 273 K, 0.5 h. For second-order fits to the kinetic data and estimated rate constants, see Figure S5 in the Supporting Information. (B) Optical changes observed at 670, 715, and 760 nm upon addition of TEMPO to solutions of Cu^IOAc, Cu^ITFA and Cu^IOTf respectively, in MeCN. Oxidation to Cu^II is only observed with Cu^IOAc. Conditions: 2 mM TEMPO, 2 mM Cu^IX, MeCN, 273 K, 4 h. Cu^II concentrations for Figure 3A and 3B were obtained by constructing UV-visible calibration curves of the corresponding Cu^IIX₂ salts in MeCN at their respective λ_max. Lines reflect smooth fits of the data, simply to guide the eye.

Slow regeneration of Cu^II in the reaction of TEMPOH with Cu^IIOAc₂ in Figure 3A resembles the appearance of Cu^II in the reaction of Cu^IOAc and TEMPO in Figure 2B. This similarity prompted us to investigate reactions of different Cu^IX salts (X = OAc, TFA, OTf) with TEMPO (Figure 3B). TEMPO was added to a solution of the Cu^IX species in MeCN, and the reaction was monitored by UV-visible spectroscopy at the λ_max wavelength corresponding to the d-d transition of the Cu^II species (670–760 nm). Only in the case of Cu^IOAc was formation of Cu^II observed from the reaction (Figure 3B; cf. red trace for OAc versus the blue and black traces for TFA and OTf).

Several experiments were done to further probe the (re)generation of Cu^II observed from reactions conducted in the presence of acetate. The reduction of Cu^IIOAc₂ by TEMPOH was monitored over a longer time period (Figure 4, black trace) under the original experimental conditions, and a 91% yield of Cu^II was obtained after 8 h. Recognizing that adventitious O₂ could contribute to background oxidation of Cu^I, efforts were made to ensure rigorous exclusion of oxygen from the reaction mixture (see sections VI–VIII in the Supporting Information for details). Even under these conditions, slow formation of Cu^II was observed over 8 h, but in much smaller amounts (Figure 4, red trace). An analogous experiment was conducted starting from Cu^IOAc and TEMPO and 1 equiv of HOAc, mimicking the product mixture expected from the reaction of Cu^IIOAc₂ and TEMPOH. The resulting time course (Figure 4, blue trace) is very similar and corresponds to a 15% yield of Cu^IIOAc₂. Repetition of the latter experiment, but in the absence of HOAc, resulted in negligible formation of Cu^II (Figure 4, green trace).

Figure 4.

Single wavelength (l = 670 nm) UV-visible time courses monitoring the [Cu^II] from reactions between copper(I/II) acetate and TEMPO(•/H). Conditions: 2 mM Cu^IIOAc₂ or Cu^IOAc, 2 mM TEMPO or TEMPOH, ±2 mM HOAc, 4 mL MeCN, N₂ or 15 mtorr static vacuum, 273 K, 8 h. Cu^II concentrations for Figure 4 were obtained by constructing a UV-visible calibration curve of Cu^IIOAc₂ in MeCN at 670 nm. Lines reflect smooth fits of the data, simply to guide the eye.

Efforts were undertaken to identify the oxidant that accounts for the slow oxidation of Cu^IOAc. A small amount of 2,2,6,6-tetramethylpiperidine (TMPH, 6%) was identified by GC-MS from the anaerobic reactions between Cu^IIOAc₂/TEMPOH and Cu^IOAc/TEMPO/HOAc. Recognizing that the conversion of TEMPO to TMPH/H₂O is a 3 e^–/3 H⁺ redox process (see further discussion below), this reaction accounts for the oxidation equivalents needed for formation of the observed 15% yield of Cu^IIOAc₂ observed in Figure 4 (blue trace).

Analysis of Cu^IIX₂ species in solution.

The influence of the anion (OAc, TFA, OTf) and role of HOAc on (slow) regeneration of Cu^II (cf. Figures 3 and 4) prompted us to obtain cyclic voltammograms (CVs) of the different Cu salts and TEMPO under conditions relevant to the reactions (Figure 5A). The TEMPO/TEMPO^– and TEMPO/TEMPOH redox couples are electrochemically irreversible, consistent with previous reports in the literature;^37–39 however, approximate mid-point potentials (E_mp) of –1.5 V and –0.84 V vs. Fc^0/+ were estimated from CVs obtained in the absence⁴⁰ and presence of a buffered mixture of HOAc/Bu₄NOAc (Figure 5A, red and blue traces). The different Cu^IIX₂ salts exhibit quasi-reversible CV features, with values that differ by > 1 V, depending on the identity of the anionic ligand: E_mp(Cu^II/I) = –0.61 V (OAc) and 0.64 V (OTf) vs. Fc^0/+. The CV of Cu^IITFA₂ exhibits two ill-defined quasi-reversible redox features at 0.010 and 0.48 V. Addition of 5 equiv of trifluoroacetic acid (TFAH) to this solution, however, leads to the appearance of a single quasi-reversible wave at 0.48 V.

Figure 5.

(A) Cyclic voltammograms of TEMPO in the presence and absence of HOAc/nBu₄NOAc buffer and of Cu^IIX₂ salts. CV conditions (TEMPO): 2 mM TEMPO, ±20 mM HOAc, ±20 mM nBu₄NOAc, 0.1 M Bu₄NPF₆, MeCN, 200 mV/s scan rate. CV Conditions (Cu^IIX₂): 2 mM CuX₂, ±10 mM TFAH, 0.1 M Bu₄NPF₆, MeCN, 100 mV/s scan rate, glassy carbon working electrode, Pt wire counter electrode. (B) EPR spectra of Cu^II salts. 10 mM Cu^IIX₂, ±50 mM TFAH, MeCN, 150 K. (C) Proposed equilibrium between the EPR-silent [Cu^IITFA₂]₂ paddlewheel dimer and EPR-active monomeric Cu^IITFA₂, promoted by TFAH.

The three Cu^IIX₂ salts were also analyzed by electron paramagnetic resonance (EPR) spectroscopy (see Figure 5B for spectra of frozen MeCN solutions). Cu^IIOAc₂ exhibits negligible EPR activity, consistent with it being present in solution as the EPR-silent [Cu(OAc)₂]₂ paddlewheel dimer.^41,42 Cu^IIOTf₂ an axial EPR signal consistent with a monomer Cu^II species solvated by MeCN.⁴³ Cu^IITFA₂ exhibits an EPR signal with only 67% of the signal intensity of Cu^IIOTf₂; however, the intensity increases to 96% when 5 equiv of TFAH is present.

The above CV and EPR data suggest that two different Cu^IITFA₂-derived species are present in solution, consistent with an equilibrium mixture of EPR-silent [Cu(TFA)₂]₂ dimer and the EPR-active monomer (Figure 5C). This observation is consistent with the presence of a single electroactive species present in the CV obtained from this mixture (Figure 5B). Consistent with this interpretation, X-ray crystal structures of both monomeric Cu^IITFA₂(H₂O)₄ and paddlewheel dimeric [Cu^IITFA₂]₂•2MeCN have previously been obtained by recrystallization of Cu^IITFA₂ from H₂O and MeCN, respectively.⁴⁴ We postulate that TFAH favors monomeric Cu^II via hydrogen bonding to the Cu-bound trifluoroacetate ligands.

Mechanistic analysis and discussion.

Cu^IIX₂ sources with all three anionic ligands, X = OAc, TFA, OTf, promote rapid oxidation of TEMPOH to TEMPO, as shown in Figure 3A. The data presented also show, however, that Cu^IOAc undergoes oxidation to Cu^IIOAc₂ in the presence of TEMPO (Figure 4), resembling the observations reported previously.¹⁵ The reaction is complicated by the facile oxidation of Cu^I by O₂, even in trace quantities, (Figure 4), but oxidation of Cu^IOAc by TEMPO via cleavage of the N–O bond to form TMPH can also account for slow generation of Cu^II under these conditions.

The effect of bpy on the Cu^II/I redox potential was analyzed in a previous mechanistic study of Cu/TEMPO catalyzed alcohol oxidation,¹⁸ but the contribution of the anionic ligand the Cu^II/I redox potential in bpy-free catalyst systems has not characterized.^5,15,45 The strong influence of the anionic ligands on the Cu^II/I redox potentials exceeds what might be expected on the basis of the anion basicity alone. This deviation is attributed to the ability of acetate (and, to a lesser extent, trifluoroacetate) to stabilize Cu^II via formation of the carboxylate-bridged dicopper paddlewheel complex. In the case of acetate, the Cu^II/I redox potential approaches that of the TEMPO/TEMPOH redox potential (Figure 5A). The two different Cu^II/I redox potentials observed with TFA, which are assigned to dimeric and monomeric structures and which differ by nearly 400 mV, highlight the stabilizing effect of dimer formation.

The anionic ligand effects on the Cu^II/I redox potentials rationalize why Cu^IOAc is the only Cu^IX species that undergoes oxidation by TEMPO under anaerobic conditions (cf. Figures 3 and 4). Single-electron transfer from Cu^I to TEMPO is thermodynamically unfavorable, and the lack of reactivity observed in the absence of HOAc is consistent with this conclusion (cf. Figure 4, green trace.). Proton-coupled reduction of TEMPO to TEMPOH by Cu^IOAc (cf. Figure 6A) is also unfavorable, but this reaction is only slightly uphill and should be thermally accessible: estimated ΔE_mp (TEMPO/TEMPOH – Cu^II/I) ≈ 200–300 mV or 4.5–7 kcal/mol. The TEMPOH generated in this manner can undergo disproportionation to TMPH and oxoammonium (TEMPO⁺) (Figure 6B), as observed previously.^15,46–50 The involvement of this process is supported by the detection of 6% TMPH and 15% Cu^IIOAc₂ following oxidation of Cu^IOAc in the presence of TEMPO and HOAc (cf. Figure 4 and associated text). The TEMPO⁺ generated from this step can then oxidize another equivalent of Cu^IOAc (Figure 6C).⁵¹ The net reaction arising from these steps, shown in Figures 6A–6C, shows that 3 equiv of Cu^IOAc are oxidized for each equiv of TEMPO converted to TMPH.

Figure 6.

Elementary steps proposed to explain the oxidation of Cu^IOAc by TEMPO in the presence of HOAc. (A) Redox equilibrium between Cu^IOAc/HOAc/TEMPO and Cu^IIOAc₂, TEMPOH. (B) TEMPOH disproportionation to TMPH and TEMPO⁺. (C) TEMPO⁺ mediated oxidation of Cu^IOAc.

Conclusion.

The results described here demonstrate that a series of different Cu^IIX₂ sources (X= OAc, TFA, and OTf) mediate efficient proton-coupled oxidation of TEMPOH to afford the aminoxyl radical. The reported oxidation of Cu^IOAc by TEMPO has been reproduced; however, this reactivity is attributed the presence of trace O₂ in the reaction mixture or a slower oxidation process that involves cleavage of the N–O bond of TEMPO. Cu^IIX₂-mediated PCET oxidation of TEMPOH provides the basis for regeneration the TEMPO co-catalyst during anaerobic (electrochemical) alcohol oxidation, and the facility of this process suggests it could also contribute under the aerobic oxidation conditions, without requiring the involvement of reactive Cu/O₂ species.⁵² The identity of the anionic (X) ligand has a significant influence on the Cu^II/I redox potential, but rapid oxidation of TEMPOH is observed in all cases, even with the low-potential Cu^IIOAc₂ species. These results are rationalized by the involvement of proton transfer in the oxidation process, whereby the beneficial effect of more basic anionic ligands offsets the lower Cu^II/I redox potentials.