Amine Oxidative N-Dealkylation via Cupric Hydroperoxide Cu–OOH Homolytic Cleavage Followed by Site-Specific Fenton Chemistry

Abstract

Copper(II)-hydroperoxide species are significant intermediates in processes such as fuel cells and (bio)chemical oxidations, all involving stepwise reduction of molecular oxygen. We previously reported a Cu^II-OOH species that performs oxidative N-dealkylation on a dibenzylamino group that is appended to the 6-position of a pyridyl donor of a tripodal tetradentate ligand. To obtain insights into the mechanism of this process, reaction kinetics and products were determined employing ligand substrates with various para- substituent dibenzyl pairs (-H,-H; -H,-Cl; -H,-OMe and -Cl,-OMe), or with partially or fully deuterated dibenzyl N-(CH₂Ph)₂ moieties. A series of ligand-copper(II) bis-perchlorate complexes were synthesized, characterized, and the X-ray structures of the -H, -OMe analog was were determined. The corresponding metastable Cu^II-OOH species were generated by addition of H₂O₂/base in acetone at –90 °C. These convert (t_1/2 ~ 53 s) to oxidatively N-dealkylated products, producing para-substituted benzaldehydes. Based on the experimental observations and supporting DFT calculations, a reaction mechanism involving dibenzylamine H-atom abstraction or electron-transfer oxidation by the Cu^II-OOH entity could be ruled out. It is concluded that the chemistry proceeds by rate limiting Cu–O homolytic cleavage of the Cu^II–(OOH) species, followed by site-specific copper Fenton chemistry. As a process of broad interest in copper as well as iron oxidative (bio)chemistries, a detailed computational analysis was performed, indicating that a Cu^IOOH species undergoes O–O homolytic cleavage to yield a hydroxyl radical and Cu^IIOH rather than heterolytic cleavage to yield water and a Cu^II-O^•−.

INTRODUCTION

Mononuclear copper species derived from reactions of ligand-Cu^I with O₂, such as Cu^II-superoxide, Cu^II-hydroperoxide, and copper-oxo {Cu^III=O ↔ Cu^II-O^•−} species,¹ have all been considered as possible reactive intermediates formed during catalytic turnover in certain copper monooxygenases, including peptidylglycine-α-hydroxylating monooxygenase (PHM).² This enzyme possesses a “non-coupled” dicopper active site with two separated (~11 Å) copper ions: Cu_M, with His₂Met ligation where O₂ is activated for reaction with substrate, and Cu_H, an electron-transfer site with His₃ coordination.^2c,2d Recent enzymatic,³ spectroscopic,⁴ and theoretical investigations^4–5 indicate that a Cu^II-superoxide species (from Cu^I/O₂) initially attacks the substrate C-H bond in PHM. However, alternative reaction sequences⁶ have been proposed, including substrate H-atom abstraction coming from a further downstream Cu^II-OOH^2a,7 or Cu^II-O^{•− 8} intermediates. As may be pertinent to PHM, other Cu enzymes,^2d or synthetic and practical oxidation-oxygenation chemistries, it is critical to elucidate fundamental aspects of the formation, structural/spectroscopic, and reactivity characteristics of all these intermediates. Our research program includes such approaches, especially for Cu^II-O₂^{•− 1d,7,9} and Cu^II-⁻OOH^6,10 complexes. Although a few well-characterized Cu^II-OOH complexes have been described,^1a,11 not a great deal is known about the intrinsic scope of their reactivity and reaction mechanism(s).

In a previous study,^10a we showed that a Cu^II-OOH species can perform oxidative N-dealkylation reactions on a pyridyl (PY) lig-and pendant dimethylamine (PY -NMe₂) moiety as substrate, leading to a product 2° amine (PY -NHMe) and formaldehyde deriving from the oxidized methyl group (Scheme 1). This reaction closely resembles the N-dealkylation chemistry occurring in PHM (Scheme 1).

Scheme 1.

For this system, a net H-atom abstraction mechanism was proposed, however, primarily on the basis of the observed kinetic isotope effect, KIE ~ 2,^10a which is low compared to what may be expected. Similar chemistry occurred when a dibenzylamine internal substrate [PY -N(CH₂Ph)₂] was present (Scheme 1).⁶

In this report, we describe experiments designed to more deeply probe the mechanism of these reactions. A combination of experimental results and density functional theory (DFT) calculations have led us to propose a new reaction mechanism that is potentially applicable to many chemical systems and possibly situations where metal-hydroperoxo species form in biology. For the net hydroxylation of the dibenzylamine C–H substrate (i.e., methylene group), a rate-determining homolytic cleavage of the Cu^II–O bond in the Cu^II-hydroperoxide occurs first. This is followed by a site specific copper Fenton chemical reaction between the resulting Cu^I species and a second equivalent of hydrogen peroxide, producing a hydroxyl radical that is responsible for the C–H activation.

EXPERIMENTAL SECTION

General

All materials used were commercially available analytical grade from Sigma-Aldrich, and TCI. Acetone was distilled under an inert atmosphere over anhydrous CaSO₄ and degassed with argon prior to use. Diethyl ether was used after being passed through a 60 cm long column of activated alumina (Innovative Technologies) under argon. Benzaldehyde and triethylamine were distilled under vacuum prior to use. (Ph₃P=O)₂•H₂O₂ was synthesized according to literature protocols, and its identity and purity were verified by ¹H-NMR and Elemental analysis.¹² Synthesis and manipulations of copper salts were performed according to standard Schlenk techniques or in an MBraun glovebox (with O₂ and H₂O levels below 1 ppm). Carbon monoxide (CO) gas was obtained from Airgas and passed through an oxygen and moisture scrubbing column.

Instrumentation

UV-Vis spectra were recorded with an HP Model 8453A diode array spectrophotometer equipped with a liquid nitrogen chilled Unisoku USP-203-A cryostat; kinetic measurements were maintained at −90 °C ± 1%. NMR spectroscopy was performed on Bruker 300 and 400 MHz instruments with spectra calibrated to either internal tetramethylsilane (TMS) standard or to residual protio solvent. EPR measurements were performed on an X-Band Bruker EMX CW EPR controlled with a Bruker ER 041 XG microwave bridge operating at the X-band (~9 GHz). GC was performed on an Agilent 6890 gas chromatograph fitted with a HP-5 (5%-phenyl)-methylpolysiloxane capillary column (30 m ^* 0.32 mm^* 0.25 mm) and equipped with a flame-ionization detector. The GC-FID response factors for benzaldehydes were prepared vs. dodecane as an internal standard. ESI Mass spectra were acquired using a Finnigan LCQDeca ion-trap mass spectrometer equipped with an electrospray ionization source (Thermo Finnigan, San Jose, CA). GC-Mass experiments were carried out and recorded using a Shimadzu GC-17A/GCMS0QP5050 gas chromatograph/mass spectrometer. X-ray diffraction was performed at the X-ray diffraction facility at the Johns Hopkins University. The X-ray intensity data were measured on an Oxford Diffraction Xcal-ibur3 system equipped with a graphite monochromator and an Enhance (Mo) X-ray Source (λ = 0.71073Å) operated at 2 kW power (50 kV, 40 mA) and a CCD detector. The frames were integrated with the Oxford Diffraction CrysAlisRED software package.

Synthesis of Ligands

DB (L^H,H)

BrTMPA (2.1 g, 5.69 mmol) dibenzylamine (6 g, 30.41 mmol) were placed in a high-pressure tube (Ace pressure tube, Aldrich Z18, 106–111) and dissolved in 3 mL toluene/12 mL water with 0.1 g NaOH (Scheme 2). The resulting mixture was stirred at 160 °C for 7 days and then cooled to room temperature. The resulting mixture was extracted with CH₂Cl₂ several times. The combined organic layers were dried over anhydrous MgSO₄, filtered, and the volatile components were removed by rotary evaporation yielding a pale yellow oil. The resulting yellow oil was purified by a column chromatography (silica gel, 100% ethyl acetate, R_f = 0.36). The product fraction was collected and the solvent was removed under reduced pressure to afford the pale yellow oil, DB (1.22 g, 77% yield).^{10a 1}H NMR (400 MHz, CDCl₃): δ 8.59 (d, 2H), 7.63 (d, 2H), 7.54 (t, 2H), 7.37-7.11 (m, 8H), 7.10 (t, 2H), 6.84 (d, 2H), 6.76 (d, 1H), 6.33 (d, 1H), 4.81 (s, 4H, 2CH₂Ph), 3.90 (s, 4H, 2CH₂Py), 3.70 (s, 2H, CH₂Py). ESI-MS, m/z: 508.4 (M + Na⁺), 486.3 (M + H⁺) in MeOH at room temperature.

Scheme 2.

DB-OMe (L^H,OMe)

¹H NMR (400 MHz, CDCl₃): δ 8.60 (d, 2H), 7.63 (d, 2H), 7.54 (t, 2H), 7.35-7.21 (m, 10H), 7.10 (t, 2H), 6.76 (d, 1H), 6.33 (d, 1H), 4.81 (s, 4H, 2CH₂Ph), 3.90 (s, 4H, 2CH₂Py), 3.70 (s, 2H, CH₂Py). ESI-MS, m/z: 538.4 (M + Na⁺), 516.3 (M + H⁺) in MeOH at room temperature.

DB-Cl (L^H,Cl)

¹H NMR (400 MHz, CDCl₃): δ 8.59 (d, 2H), 7.63 (d, 2H), 7.54 (t, 2H), 7.35-7.21 (m, 10H), 7.10 (t, 2H), 6.76 (d, 1H), 6.33 (d, 1H), 4.80 (s, 2H, CH₂Ph), 4.76 (s, 2H, CH₂Ph), 3.89 (s, 4H, 2CH₂Py), 3.70 (s, 2H, CH₂Py). ESI-MS, m/z: 542.2 (M + Na⁺), 520.2 (M + H⁺) in MeOH at room temperature.

DB-Cl,OMe (L^Cl,OMe)

¹H NMR (400 MHz, CDCl₃): δ 8.56 (d, 2H), 7.63 (d, 2H), 7.54 (t, 2H), 7.40-7.10 (m, 9H), 6.82 (d, 2H), 6.75 (d, 1H), 6.32 (d, 1H), 4.75 (s, 4H, 2CH₂Ph), 3.90 (s, 4H, 2CH₂Py), 3.80 (s, 3H), 3.70 (s, 2H, CH₂Py). ESI-MS, m/z: 572.5 (M + Na⁺) in MeOH at room temperature.

Deuterated benzaldehyde

N,N-dimethylformamide-d₇ (1.03 g, 12.85 mmol) was placed in a 2-neck flask in 50 mL THF at −80 °C (Scheme 3). 1.8 M phenyllithium solution (6.5 mL, 11.68 mmol) was slowly added to the solution at −80 °C. After stirring for 1 hour, the mixture solution was warmed up to room temperature. All solvents were removed by a rotary vacuum affording a pale yellow oil. This resulting yellow oil was purified by column chromatography (silica gel, ethyl acetate : hexane = 1 : 6) and this material was used in the next step without further purification (see partially deuterated dibenzyl amine-D2 below).

Scheme 3.

Deuterated benzyl amine

A 250 mL two-necked, round-bottomed flask is charged with lithium aluminum deuteride 98 atom % D (Al-drich), LiAlD₄ (5.1 g 121.66 mmol) (Scheme 3).¹³ The flask is fitted with a condenser and a septum, and is thoroughly purged with a steady stream of Ar. After 5 min, 60 mL of anhydrous diethyl ether is added via a syringe. The homogeneous solution is cooled in an ice bath to 0°C. Benzonitrile (5.0 g, 48.49 mmol) in 10 mL diethyl ether was slowly transferred via a syringe to the mixture over 10 min. After vigorous H₂ gas evolution ceased, the mixture solution was allowed to warm to ambient temperature and then refluxed for the overnight at 80 °C. The solution is cooled to room temperature, diluted with 65 mL of diethyl ether, cooled to 0 °C and quenched by the successive drop-wise addition of 3.8 mL of 10% NaOH solution, and 11.4 mL of water. The colorless precipitate was vacuum filtered through Celite, and the filter cake was washed with diethyl ether (3 × 20 mL). The combined filtrate was concentrated to give a pale yellow oil. This resulting yellow oil was purified by column chromatography (Silica gel, 100% ethyl acetate, R_f =0.38) yielding a pale oil (4.6 g, 87% yield). ¹H NMR (400 MHz, CDCl₃): δ 7.40-7.23 (m, 5H).

Partially Deuterated Dibenzyl Amine-D2

Deuterated benzaldehyde (4.2 g, 48.58 mmol) and benzylamine (4.2 g 58.808 mmol) were placed in 70 mL EtOH in a 250 mL round-bottom flask and then the mixture solution was stirred for 4 hours under Ar to form an imine intermediate (Scheme 3). This intermediate imine can then be isolated and reduced with a suitable reducing agent, deuterated sodium borohydride. Deuterated sodium borohydride 98 atom % D (Aldrich), NaBD₄ (3.3 g, 78.84 mmol) was slowly added to the solution and stirred for 30 min at room temperature. To quench excess NaBD₄, 20 mL MeOH was slowly added at 0°C and then all solvents were removed by rotary evaporation. The resulting crude product was dissolved in 100 mL of CH₂Cl₂, washed two times with a saturated Na₂CO₃ solution. After drying over anhydrous MgSO₄, the solution was filtered and removed by rotary evaporation. The resulting yellow oil was purified by column chromatography (Silica gel, 20% ethyl acetate with hexane, R_f = 0.67) yielding a pale oil (7.5 g, 96% yield). ¹H NMR (400 MHz, CDCl₃): δ 7.40-7.23 (m, 10H), 3.80 (s, 2H). ESI-MS, m/z: 200.1 (L + H⁺) in MeOH at room temperature.

Fully Deuterated Dibenzyl Amine-D4

Deuterated benzaldehyde (5.2 g, 47.634 mmol) and deuterated benzylamine (6.3 g, 58.808 mmol) were placed in 70 mL EtOH in a 250 mL round-bottom flask and then the mixture solution was stirred for 4 hours under Ar. Deuter-ated sodium borohydride 98 atom % D (Aldrich), NaBD₄ (2 g, 47.78 mmol) was slowly added to the solution at 0 °C. To quench excess NaBD₄, 20 mL MeOH was added and then all solvents were removed by using reduced pressure. The resulting crude product was dissolved in 100 mL of CH₂Cl₂, washed two times with a saturated Na₂CO₃ solution. After drying over anhydrous MgSO₄, the solution was filtered and removed by rotary evaporation. The resulting yellow oil was purified by column chromatography (Silica gel, 20% ethyl acetate with hexane, R_f = 0.67) yielding a pale oil (7.5 g, 97% yield). ¹H NMR (400 MHz, CDCl₃): δ 7.35-7.26 (m, 10H), 2.18 (s, 1H). ESI-MS, m/z: 201.6 (L) in MeOH at room temperature.

DB-D2 (d₂-L^H,H)

¹H NMR (400 MHz, CDCl₃): δ 8.59 (d, 2H), 7.63 (d, 2H), 7.54 (t, 2H), 7.37-7.11 (m, H), 7.10 (t, 2H), 6.76 (d, 1H), 6.33 (d, 1H), 4.81 (s, 2H, CH2Ph), 3.90 (s, 4H, 2CH2Py), 3.70 (s, 2H, CH2Py). ESI-MS, m/z: 510.5 (L + Na⁺), 488.5 (L + H⁺) in MeOH at room temperature.

DB-D4 (d₄-L^H,H)

ESI-MS, m/z: 490.5 (L + H⁺) in MeOH at room temperature.

Synthesis of Copper (II) Complexes

[(DB)Cu^II(H₂O)(OClO₃)](ClO₄), 1^H,H

The synthesis and characterization of the DB copper complex was recently reported. DB ligand (430 mg, 0.886 mmol) was treated with Cu^II(ClO₄)₂•6H₂O (328 mg, 0.886 mmol) in acetone (20 mL) and was stirred for 10 min at room temperature. The mixture complex was precipitated as a blue solid upon the addition of diethyl ether (120 mL). The supernatant was decanted, and the resulting crystalline solid was washed two times with diethyl ether and dried under the reduced vacuum to afford the blue solid. The blue solid was recrystallized twice from acetone/diethyl ether. After vacuum-drying, the blue crystals weighed 587 mg (87% yield). Elemental analysis: (C₃₅H₃₉Cl₂CuN₅O₁₀) Calculated: C (51.01), H (4.77), N (8.50); found: C (50.95), H (4.66), N (8.42). ESI-MS: m/z= 548.5, corresponding to [(DB)Cu^I]⁺ in acetone at room temperature (Figure S1). EPR spectrum: X-band (ν= 9.186 GHz) spectrometer in acetone at 70 K: g_|| = 2.273, g^⊥ = 2.048, A_|| = 173 G.

[(DB-Cl)Cu^II(CH₃COCH₃)](ClO₄)₂, 1^H,Cl

Elemental analysis: (C₃₅H₃₆Cl₃CuN₅O₉) Calculated: C (50.01), H (4.32), N (8.33); found: C (50.17), H (4.52), N (8.32). ESI-MS: m/z= 582.5, corresponding to [(DB-Cl)Cu^I]⁺ in acetone at room temperature. EPR spectrum: X-band (ν = 9.186 GHz) spectrometer in acetone at 70 K: g_|| = 2.270, g^⊥ = 2.043, A_|| = 172 G.

[(DB-OMe)Cu^II(OClO₃)CH₃COCH₃)](ClO₄), 1^H,OMe

Single crystals of a [(DB-OMe)Cu^II(CH₃COCH₃)]²⁺, 1^H,OMe complex were obtained by vapor diffusion of diethyl ether into a solution of the copper complex in acetone. Elemental analysis: (C₃₆H₃₉Cl₂CuN₅O₁₀) Calculated: C (51.71), H (4.70), N (8.38); found: C (51.68), H (4.86), N (8.37). ESI-MS: m/z = 578.5 corresponding to [(DB-OMe)Cu^I]⁺ in acetone at room temperature. EPR spectrum: X-band (ν = 9.186 GHz) spectrometer in acetone at 77 K: g_|| = 2.277, g^⊥ = 2.049, A_|| = 175G.

[(DB-Cl-OMe)Cu^II(CH₃COCH₃)](ClO₄)₂, 1^Cl,OMe

Elemental analysis: (C₃₆H₃₈Cl₃CuN₅O₁₀) Calculated: C (49.66), H (4.40), N (8.04); found: C (49.35), H (4.45), N (7.79). ESI-MS: m/z = 612.5 corresponding to [(DB-Cl-OMe)Cu^I]⁺ in acetone at room temperature. EPR spectrum: X-band (ν = 9.186 GHz) spectrometer in acetone at 70 K: g_|| = 2.272, g^⊥ = 2.044, A_|| = 176 G.

[(d₂-L^H,H)Cu^II(H₂O)](ClO₄)₂

ESI-MS: m/z = 550.5, corresponding to [d₂–(L^H,H)Cu^I]⁺ in acetone at room temperature (Figure S2). EPR spectrum: X-band (ν = 9.186 GHz) spectrometer in acetone at 70 K: g_|| = 2.270, g^⊥ = 2.043, A_|| = 175 G (Figure S3).

[(d₄-L^H,H)Cu^II(H₂O)](ClO₄)₂

ESI-MS: ESI-MS: m/z = 552.5, corresponding to [(d₄–L^H,H)Cu^I]⁺ in acetone at room temperature (Figure S2). EPR spectrum: X-band (ν = 9.186 GHz) spectrometer in acetone at 70 K: g_|| = 2.270, g^⊥ = 2.043, A_|| = 172 G (Figure S3).

DFT Calculations

Calculations on the copper hydroperoxo complex supported by L^H,H were performed with the B3LYP functional and a TZVP basis set on the Cu-OOH fragment and the coordinating nitrogen atoms and SV on all other atoms as implemented in Gaussian 09.¹⁴ A TZVP basis set was utilized for an additional equivalent of H₂O₂ and on the nitrogen atom of NEt₃ (SV on all of the other atoms). Optimizations were performed in acetone with the Polarizable Continuum Model on an untrafine integration grid. Spin contamination was accounted for as described in the Supporting Information. Analytical frequency calculations were performed on all stationary points and thermal corrections to the Gibbs free energy were determined at −90 °C. Transition states were connect to the corresponding products and reactants by optimizing from the transition state geometry, slightly distorted along the computed imaginary frequency or with an intrinsic reaction coordinate.¹¹ Calculations on the [(TMG₃tren)Cu^II-OOH]⁺ system (TMG₃tren = (1,1,1-tris[2-[N²-(1,1,3,3-tetramethyl-guanidino)]-ethyl]amine))) were performed as described previously¹⁵ with Gaussian 03.¹⁶

RESULTS AND DISCUSSION

Mononuclear copper(II) complexes [(L^X1,X2)Cu^II(Y)](ClO₄)₂ (1^X1,X2) (Y = H₂O, acetone) were generated, isolated and purified by reacting the various ligands (Scheme 4; X = Cl, H or OMe) with Cu(ClO₄)₂•6H₂O in acetone, precipitating solid products with Et₂O, and recrystallizing.¹⁷ An X-ray crystallographically derived structural diagram for 1^H,OMe is shown in Figure 1 along with spectroscopic data.¹⁷

Scheme 4.

Figure 1.

(a) X-ray crystal structure of [(L^H,OMe)Cu^II(OClO₃)-(CH₃COCH₃)]⁺ (1^H,OMe) revealing Cu(II) distorted octahedral coordination with three equatorial N and one O_(acetone) donors (Cu–ligand = 1.984 Å_(ave.)) and two elongated axial ligands, one pyridyl group (Cu–Npy = 2.698(2) Å) and a perchlorate O-atom (Cu–O = 2.492(2) Å).¹⁷ (b) UV-vis spectrum of 1^H,OMe (λ_max = 630 nm) and Cu^II-OOH complex 2^H,OMe. (c) EPR spectrum and simulated plot (red) of 1^H,OMe (2 mM, g_|| = 2.205, A_|| = 206 G, g^⊥ = 2.003) and (d) EPR spectrum and simulated plot (red) of 2^H,OMe (2 mM, g_|| = 2.227, A_|| = 205 G, g^⊥ = 2.044); X band (ν = 9.186 GHz) in acetone at 77 K.

Cupric hydroperoxo complexes 2^X1,X2 are obtained at −90 °C in acetone by reacting Cu^II precursors 1^X1,X2 with 10 equiv H₂O₂ or X•H₂O₂, X = urea, (Ph₃P=O)₂ in the presence of Et₃N (2 equiv) under Ar (Scheme 4). Under these conditions, a bright green product, [(L^H,H)Cu^II-OOH]⁺ (2^H,H), forms, possessing a LMCT band at 384 nm (ε = 2,200 M⁻¹cm⁻¹) with a d-d band at 685 nm (ε = 150 M⁻¹cm⁻¹); similar UV-Vis features are obtained for the other [(L^X1,X2)Cu^II–OOH]⁺ complexes including 2^H,OMe (Figure 1).¹⁷ The g and A values obtained from EPR spectroscopy on the [(L^X1,X2)Cu^II-OOH]⁺ (2^X1,X2) series are all similar, but distinguishable from the precursor copper(II) complexes [(L^X1,X2)Cu^II(Y)]²⁺ (1^X1,X2) (see Figure 1). Integration of the EPR intensity of 1^X1,X2 and 2^X1,X2 indicates that 2^X1,X2 is formed with ~80 % yield under these conditions.

The hydroperoxo-copper(II) complexes [(L^X,X)Cu^II–OOH]⁺ (2^X1,X2) (Scheme 5) are metastable and in all cases, biomimetic oxidative N-dealkylation occurs, similar to the enzyme PHM (Scheme 1).² The rate of the reaction was monitored by the loss in the ~ 384 nm band, that decayed in a first-order process at similar rates for all of the complexes in Scheme 5, k_decay = 4.17 ± 0.14 × 10⁻² s⁻¹ for [(L^H,H)Cu^II-OOH]⁺ (0.3 mM).¹⁷ These results were initially surprising since the rate of benzylic H-atom abstraction has been shown to depend on the para-substituent in other systems.¹⁸ Hence, further studies were undertaken to test whether various para-benzyl (i.e., aryl) substituents in the hydroperoxo complexes 2^X1,X2 would lead to different yields of aldehyde products.

Scheme 5.

The overall and relative yields of benzaldehyde (Scheme 5) were quantified by GC analysis from samples taken directly from reaction solutions at −80 °C after 30 minutes. Complementary yields of secondary amines, i.e., the initial ligand with a lost benzylic arm, were obtained following reaction solution workup using Na₂EDTA_(aq) or NH₄OH_(aq) to demetallate the copper, followed by extraction of organics into CH₂Cl₂, and subsequent product analysis by NMR spectroscopy and ESI-MS (Scheme 6). The remainder of the ligand was unreacted L^X1,X2, indicating good mass balance for these reactions.¹⁷

Scheme 6.

These results indicate that there is no discrimination for attack of the reactive species at the benzylic methylene substrate position, whether the aryl ring is electron rich (X = OMe) vs. electron poor (X = Cl) since equal yields (average of 3 trials) of substituted benzyaldehydes are obtained. Even for the case of 2^Cl,OMe, equal amounts of p-chloro and p-methoxy benzyldehydes are obtained, Scheme 5. The lack of a substituent effect strongly implies that attack by whatever oxidant is involved does not include substrate methylene group H-atom transfer (HAT) or hydride abstraction. Cu^II-OOH electron-transfer oxidation of the amine is also unlikely, vide infra.

Further insight comes from comparing the reactivity of the parent ligand (L^H,H with its –N(CH₂Ph)₂ substrate,) to a ligand with fully deuterated benzylic C–D bonds (d₄-L^H,H with –N(CD₂Ph)₂) (Scheme 7). Kinetic interrogation of the decomposition rates of [(L^H,H)Cu^II–OOH]⁺ vs. [(d₄–L^H,H))Cu^II–OOH]⁺ reveals that the KIE = k_H/k_D = 1.0 ± 0.1. We also examined products found from competitive internal ligand oxidation chemistry with the copper hydroperoxy complex of d₂–L^H,H with –N(CD₂Ph)(CH₂Ph) (Scheme 8). Again the KIE = 1.0 ± 0.1, based on the observation of equal yields of benzaldehyde and deuterated benzaldehyde.

Scheme 7.

Scheme 8.

These results point to the fact that in the rate-determining step (rds) the Cu^II–OOH moiety decays/transforms to something more reactive, which leads to the oxidative N-dealkylation chemistry observed. This hypothesis is in agreement with past computational studies^4–5,19 that have indicated that H-atom abstraction by a Cu^II-OOH species is an unfavorable process. However, Poater and Cavallo have suggested that the distal oxygen of the Cu^II–OOH species, supported by the related tripodal tetradentate N₄ TMG₃tren ligand, directly abstracts a H-atom from the guanidinium methyl group (ΔG^‡ = 22 kcal/mol).¹⁵ Given these disagreements, we computed the barrier for H-atom abstraction by the distal oxygen in 2^H,H (Scheme 9, top). We calculated a large barrier for this mechanism (ΔG^‡ = 34.0 kcal/mol) and would expect to observe a large, normal KIE on the decay of the Cu^II–OOH species, however this is inconsistent with the experimental results.

Scheme 9.

To determine the large discrepancy between the calculated barriers in these two systems, we reexamined the reported¹⁵ calculations on the TMG₃tren system. We find that the 22 kcal/mol transition state for the H-atom abstraction by the Cu^IIOOH does not connect on an intrinsic reaction coordinate (Figure S19–S20) to the Cu^II-OOH intermediate. Rather, this transition state is best described as the reaction of a hydroxyl radical with the guanidinium H-atom (Scheme 10, second step). To form this hydroxyl radical, the Cu^II–OOH must first undergo homolytic cleavage of the O–O bond, a process with a barrier of ΔG^‡ = 30.3 kcal in the TMG₃tren system.¹⁷ Computing the barrier of O–O bond homolysis in the present system is complicated by the presence of an open, axial coordination site that leads to coordination of the hydroxyl radical produced (Scheme 9, middle). The quartet spin state is predicted to be lower in energy than the doublet state and the energy of the intersystem crossing (ΔE_ISC) is ~33 kcal/mol, computed from the O–O potential energy surface (Figure S15). This barrier is inconsistent with the experimental rate (1.32 × 10⁻² sec⁻¹, see above) that predicts ΔG^‡ ≈ 12 kcal/mol at −90 °C. Hence these calculations predict that both mechanisms, the direct H-atom abstraction by the distal oxygen (Scheme 9, top) and the homolytic O–O bond cleavage (Scheme 9, middle) are inconsistent with the experimental findings.

Scheme 10.

We alternatively hypothesized that the observed products could result from rate-limiting Cu–OOH homolysis to give a Cu^I complex and a hydroperoxyl radical (•OOH). Such a reaction has been described,²⁰ and is consistent with the previously reported one-electron reduction of a Cu^II complex by hydrogen peroxide.^11c This reaction is also favored by poorly donating ligands which stabilize the lower-valent Cu^I oxidation state. A reaction coordinate for the concerted Cu^II–O homolysis coupled to H-atom abstraction by the proximal oxygen yields an unfavorable activation barrier (ΔG^‡ = 27.3 kcal/mol) and is inconsistent with an isotope effect of 1 (Scheme 9, bottom). However, the calculated barrier for the homolytic cleavage of the Cu^II–OOH bond is equivalent to the bond dissociation energy (Figure S16) and energetically accessible (ΔG = 14.8 kcal/mol, Scheme 11). Hence, DFT calculations suggest that the most favorable decay pathway of the Cu^II–OOH intermediate produces a Cu^I complex and a hydroperoxyl radical.

Scheme 11.

The proposed homolytic cleavage of the Cu^II–OOH bond is also supported by the following observation. Bubbling carbon monoxide (CO) into a solution of [(L^H,H)Cu^II-OOH]⁺ (2^H,H) at −90 °C causes the solution to change from light green (due to 2^H,H) to colorless (that being typical of such ligand-Cu^I-CO complexes)²¹ and a much lower yield of benzaldehyde (~19 % compared to the usual ≥ 50 %) is produced, an observation that may be explained by CO trapping the lower-valent Cu^I generated from Cu^II–OOH homolysis.

After the Cu^II–OOH bond has been cleaved, we hypothesized that the dibenzylamine containing C–H bond could be activated via a number of potential mechanisms to lead to the observed products (Scheme 11). Initially, we considered C–H abstraction by the free hydroperoxyl radical, but the computed barrier for this process is unfavorable (ΔG^‡ = 31.7 kcal/mol, Scheme 11, middle). These calculations suggest that this radical is not sufficiently reactive to activate the substrate C–H bond, likely due to the formation of a relatively weak O–H bond (88 kcal/mol) compared to the strong O–H bond of water (119 kcal/mol).²² Another possibility involves the oxidation of the tertiary amine via electron-transfer to the •OOH species to produce an ammonium radical cation (Scheme 11, bottom). However, DFT calculations suggest that the most favorable one electron oxidation of the Cu^I species gives the Cu^II complex, with the Cu^I ammonium radical cation being significantly higher in energy.¹⁷ These computational results are consistent with experimental cyclic voltammetry data that observe a single, copper based redox feature at −280 mV vs Ag⁺/Ag° for all of the 1^X1,X2 cupric complexes (Figure S5).

This led us to consider disproportionation of the hydroperoxyl radical (Scheme 11, top) and that the observed C–H activation could be due to a subsequent reaction of the Cu^I complex. DFT calculations suggest that this disproportionation reaction is favorable (ΔG = −1.0 kcal/mol), in agreement with rapid disproportionation rates in water and organic solvents.²³ We hypothesized that the Cu^I complex could either react with an additional equivalent of hydrogen peroxide in a Fenton reaction or could react with O₂ produced from the disproportionation of either the hydroperoxyl radical or hydrogen peroxide. Experimentally, the reaction of the dibenzylamine containing parent ligand L^H,H, [(L^H,H)Cu^I]⁺ (3)⁶ with dioxygen led to the formation of the bis-3-oxo species [{(L^H,H)Cu^III}₂(μ-O²⁻)₂]²⁺, however no benzaldehyde was detected.⁶ However, the addition of 5 or 10 equiv H₂O₂ to (3) in acetone at −80 °C leads to the formation of 35–45 % benzaldehyde¹⁷ indicating that the Cu^I intermediate was competent for amine oxidative N-dealkylation under the experimental conditions.

These results suggest that the reduced metal complex, formed after the homolysis of the Cu^II–OOH bond, undergoes a Fenton reaction with the pool of excess H₂O₂ (Scheme 12) present in solution, that was added for the initial synthesis/generation of the parent [(ligand)-Cu^II-OOH]⁺ complex (vide supra). Electron transfer rates and calculations from Marcus theory disfavor an outer sphere mechanism due to the unfavorable one electron reduction of hydrogen peroxide.²⁴ Instead, the proposed Fenton reaction would proceed via an inner sphere mechanism involving a Cu^I–OOH species. Presumably, this species is formed from a subsequent equivalent of hydrogen peroxide and involves a proton transfer to the excess triethylamine present in solution. DFT calculations suggest that the protonated triethylamine produced by this reaction prefers to form an explicit hydrogen bond with the proximal oxygen atom of the Cu^I–OOH species (Scheme 12), a process that is close to thermoneutral (ΔG = 1.1 kcal/mol). We also considered the formation of a Cu^I–OOH species where the protonated triethanol amine is hydrogen bonded to the distal oxygen (Scheme 12), however this species is predicted to be higher in energy (ΔG = 6.9 kcal/mol).

Scheme 12.

In iron catalyzed Fenton chemistry, the oxidant responsible for C–H activation is believed to be either an iron^IV=O or a hydroxyl radical resulting from the homolysis or heterolysis of the O–O bond depending on the conditions of the reaction.^24–25 In the present system we computed the two analogous pathways, the formation of either a Cu^II–OH and •OH or a Cu^II–O^•− and H₂O (Scheme 12). The formation of a hydroxyl radical initially proceeds via a rapid proton transfer to the proximal oxygen (Scheme S10) to form a Cu^I(H₂O₂) complex hydrogen bonded to NEt₃ that is followed by homolysis of the O–O bond (ΔE^‡ = 2.4 kcal/mol and ΔG^‡ = 9.4 kcal/mol, Scheme 12 top). In contrast, the formation of a copper(II)-oxyl proceeds via a proton transfer to the distal oxygen and an intersystem crossing from the singlet Cu^IOOH species to the triplet Cu^II–O^•− species (Scheme 12 bottom). The barrier for this process can be approximated from the 2-D potential energy surface (Figure S17–S18) and has a much larger barrier (ΔE_ISC^‡ ≈ 12.9 kcal/mol) than the formation of a hydroxyl radical.

These results indicate that the formation of a hydroxyl radical is energetically more favorable than the formation of a Cu^II–O^•−. The experimentally observed products would then result from a subsequent attack of the hydroxyl radical at the benzylic position of the ligand substrate producing a carbon radical that can then react with the Cu^II-OH moiety giving a ligand alcohol and copper(I) complex (Scheme 13). Under the strong overall oxidative conditions, the copper(I) ends up as copper(II) in the final product mixture (Scheme 13), as observed by EPR spectroscopy.

Scheme 13.

The possible presence of a hydroxyl radical under these conditions was experimentally evaluated with the oxygen-radical trap DMPO (5,5-dimethyl-pyrroline N-oxide).^26,27 Addition of excess DMPO to the hydroperoxo solution (of 2^H,H) at −90 °C, dramatically decreases the yield of benzaldehyde (to ~20 %) indicating that a radical species is responsible for the observed N-dealkylation products (Schemes 6 & 12). Also, EPR spectra (Figure S22) of the DMPO derivative radical are consistent with DMPO–OH,²⁷ the hydroxyl radical adduct of DMPO (see the Supporting Information for details). This result indicates that a hydroxyl radical is responsible for the observed C–H activation and is consistent with the site-specific Fenton reaction described here.

Overall, we propose a mechanism where the Cu^II-OOH species decays by rate determining Cu^II–OOH homolysis with a calculated barrier of ΔG^‡ = 14.3 kcal/mol, in agreement with the observed rate of Cu^II–OOH decay (4.17 × 10⁻² s⁻¹). In a subsequent Fenton reaction, the Cu^I complex produced reacts with a second equivalent of hydrogen peroxide, which proceeds with a lower barrier (ΔG^‡ = 9.4 kcal/mol) (Scheme 12). Additionally, the absence of a product isotope effect in d₂-L^H,H is consistent with the lack of an isotope effect observed in the reaction of a hydroxyl radical (generated from pulse radiolysis) with cyclohexane.²⁸

These finding also lead us to warn researchers in copper or iron oxidative chemistries concerning the use of excess hydrogen peroxide (plus base) to generate M-(⁻OOH) complexes. Such procedures are very common,^{6,10,11b,11c,19b,29} but as seen here, the presence of additional H₂O₂ may complicate matters and allow for unintentional reaction pathways or mechanisms.

CONCLUSION

A mononuclear copper(II) hydroperoxide species possessing an appended dibenzylamine moiety is observed to be competent to perform the biomimetic (Scheme 1) oxidative N-dealkylation reaction to form benzaldehyde. A combination of experimental observations and DFT calculations indicates that the Cu^II-OOH species itself is not capable of performing C–H activation. This has led us to hypothesize and evaluate a new mechanism to account for the observed products involving rate-limiting Cu^II–OOH homolysis. The Cu^I species that is produced from this reaction further undergoes a Fenton reaction with an additional equivalent of H₂O₂. Experimental observations and DFT calculations favor homolytic O–O bond cleavage to form Cu^II–OH and •OH where the hydroxyl radical is ultimately responsible for the C–H activation.