Mechanism of the CuII-catalyzed benzylic oxygenation of (aryl)(heteroaryl)methanes with oxygen

A mechanistic study of the copper-catalyzed oxidation of the methylene group of aryl(di)azinylmethanes was performed.

Simulations of all EPR spectra were performed with the EasySpin program, a MATLAB toolbox developed for EPR simulations. S3 The HYSCORE data were processed with MATLAB 7.7.0. The time traces were baseline-corrected with a third-order polynomial, apodized with a Hamming window and zero-filled. After a two-dimensional Fourier transformation, the absolute spectra were computed. Spectra recorded with different τ values were added after Fourier transformation to eliminate blind-spot effects.

Additional Results
Chart 1. Mono-and dinuclear copper species present in the reaction Table S1. Relative contributions of the different species detailed in Table 2 to the observed EPR signal of the mononuclear complexes as obtained from simulation.

b) Identification of Cu-(1) ligation in species IV
The involvement of 2-benzylpyridine (1) ligation in the monomeric species IV can be derived from pulsed EPR experiments. Figure S3 shows the 1 H HYSCORE spectrum recorded at 282.8 mT (low field edge of spectra of species IV) for the reaction mixture with low copper content after 1 h in comparison with the spectrum taken of the solution of CuI (0.05 M) and AcOH (0.5 M) in DMSO under the same conditions. We clearly see that addition of 2-benzylpyridine (1) has led to the observation of new cross peaks that are analogous to the signals of the ortho protons of pyridine observed for copper-pyridine complexes. S4 Similar results were obtained at other magnetic field settings (not shown). Davies ENDOR spectra using hard pulses have been shown to suppress the signals of the weakly coupled nuclei (here the 1 H contributions) and thus enhance the contributions of the strongly coupled nuclei. S5 In Figure S4 we make use of this effect to reveal the interaction with the 14 N nuclei for species III and IV. Figure S4a,b shows the Davies ENDOR spectra of the reaction mixture with low CuI concentration (0.05 M) recorded after 10' of reaction at 282.8 mT. In the ENDOR spectrum recorded with soft pulses ( Figure S4a) signals due to the weakly coupled protons and 13 C nuclei (in natural abundance) are visible. These interactions could also be detected by HYSCORE (not shown). There seems to be a signal around 15 MHz overlapping with the 1 H contribution. When hard pulses are used, the 1 H contribution is fully suppressed, leaving a signal at 15-17 MHz agreeing with the interaction with the contribution of at least one 14 N nucleus with hyperfine value of 32 MHz. Similar signals were observed at other magnetic field settings, although overlap with ENDOR lines of the 63/65 Cu interactions are observed at the higher field orientations ( Cu A is in the order of 20-35 MHz at these observer postions (Table 2)). Although the fast electron relaxations (due to the higher Cu II amount) render the recording of ENDOR spectra more difficult at later stages in the reaction, it is clear that the Davies ENDOR spectrum recorded with hard pulses for the reaction mixture after 1 h (where only species IV is contributing) still shows the presence of the 14 N ENDOR signals, confirming that also in species IV, at least one nitrogen base is coordinating to the copper ion.     S6. Experimental (black) and simulated (blue) CW-EPR spectra recorded at 100 K of a solution of CuI (0.05 M) and AcOH (0.5 M) in DMSO with 2-benzoylpyridine (2) heated for 10' at 100 C under O 2 atmosphere. The [CuI]/ [(2)] ratio is 0.1. The species used for simulation are shown in table S2. The spectra were recorded with a microwave power of 0.47 mW and a microwave frequency of 9.73 GHz. Figure S6 shows the CW-EPR spectrum of a solution of CuI (0.05 M) and AcOH (0.5 M) in DMSO with the product 2-benzoylpyridine (2) (0.25 M) that was heated for 10 min at 100 C to mimic the reaction conditions. Simulation of the experiment revealed the presence of four species: IV", III', V and VII (see Table S2). Figure S7 illustrates that a contribution of species V needs to be taken into account to simulate the experimental spectrum, proving that species V may indeed be related to one of the binding modes of the reaction product 2 to copper. The EPR parameters of III' and IV" are similar to the ones of species III and IV found after addition of 1 to the solution (see manuscript). This indicates similar binding modes in both cases (i.e. the direct binding of the pyridine nitrogens of two ligands to the copper ion). The g and hyperfine values of species V seem to indicate a larger ligation of nitrogen ligands (see manuscript). The EPR parameters of VII are in the same order of those found for IV and IV', and may indicate a small change in the ligation mode (2 may bind copper either only via the pyridine nitrogen (as in IV') or may bind via both the carbonyl oxygen and pyridine nitrogen). Species III', IV'' and VII may be present in the actual reaction mixture (spectra recorded after 4h, Figure 6, manuscript), but are hard to distinguish from those of species III and IV.  Figure S7. Enlarged view of the low-field area of the expertimental (black) and simulated (blue) spectra shown in Fig. S6, together with a second simulated spectra (red) without considering species V. The arrows indicate the EPR intensity due to the Cu A z hyperfine splitting of species V. Table S2. Principal g and copper hyperfine values of the mononuclear Cu II complexes observed in the spectra of Figure S6.  Figure S8 shows the difference spectra obtained by subtracting the contribution of the substrate-less mixture from the EPR spectra of the reaction mixture with high copper content in Figure 9. For comparison, the spectra are overlaid with the spectra of mixtures with low copper content. At 10 minutes in the reaction, both spectra are similar. However, as the reaction evolves, the significant evolution in the EPR features found for the low [Cu] sample, is not observed in the difference spectra. In order to understand this evolution, Figure S9 shows the comparison of the spectra of the mixture with low copper content after 10' and 120' in the reaction, with the simulations of the room-temperature EPR spectra of species I-V, assuming a rotational correlation time of 0.4 10 -10 s and line widths of 5 mT. From this comparison, it becomes clear that at early stages in the reaction, mainly species I,II and III can be observed, while the contributions of species IV and V start to dominate the spectra at longer reaction times. In the reaction mixture at high CuI concentration, no evidence of contributions of IV and V are found. In contrast, large amounts of dinuclear copper species can be found at high [Cu] (Figs. S10, S11; see also main text for further explanations).

Figure S8.
Comparison between (red) the room-temperature EPR spectra of the copper-catalyzed oxidation of 2-benzylpyridine (1) with 0.05 M CuI recorded at different times after the beginning of the reaction and (green) the room-temperature EPR spectrum obtained by subtracting the spectrum of the substrate-less solution (bottom of Figure 9) from those obtained for the reaction mixture with 0.25 M CuI. The red spectra were scaled by a factor of 2 to allow comparison. Figure S9. Comparison between the experimental room temperature EPR spectra of the copper-catalyzed oxidation of 2-benzylpyridine (1)   The copper-catalyzed oxidation of 2-benzylpyridine (1) was carried out in toluene instead of DMSO under the same experimental conditions at a concentration of 0.05 M CuI. The X-band CW-EPR spectrum of the reaction mixture after 5' was measured at 100 K ( Figure S12). The g and A parameters obtained from simulation are summarized in Table S3. Two species IV''with similar values to species IV and IV' are detected after 5', whilst no DMSO-related species are observed (species III). This result confirms the ligation of DMSO to Cu when the latter solvent is used in the reaction. g) Analysis of the effect of AcOH on the affinity of Cu-N ligation between 1 and catalyst Figure S13. Room-temperature X-band EPR spectra of 0.05 M CuBr 2 in DMSO with increasing 2benzylpyridine (1) Figure S13 illustrates the EPR spectral changes observed when increasing amounts of 1 are added to a 0.05 M CuBr 2 solution in DMSO. The complex spectral changes reflect the gradual replacement of coordinated DMSO by 2-benzylpyridine ligands in the copper complexes. The spectra cannot be described as simple combinations of two components, in line with the expected occurrence of copper complexes with zero up to four 2-benzylpyridine ligands, each with their specific EPR signature. Figure  S14 shows that addition of acetic acid has no effect on the spectral features of these solutions. Hence, for a given [Cu]:[1] ratio the same proportionality of the different copper complexes is found, irrespective of the presence of acetic acid. This implies that acetic acid does not have an effect on the binding affinity of 1 to Cu II in the catalytic reactions under study.

III.
Methods for in situ IR measurements a) general information Infrared reaction monitoring experiments were performed on a Matrix-MF and a ReactIR spectrometer using respectively a Diamond ATR fiber probe (IN355, Ø 6.3 mm) and a custom made (DST, Ø 6.35 mm) DiComp AgX probe.
b) IR-spectra of the reaction components The initial reaction rate was determined by following the formation of 2-benzoylpyridine (2) over time.
More specifically, the carbonyl C=O stretch of the ketone was chosen as a viable peak for integration due to the fact that it is intense and has no overlap with peaks of other reaction components. Integration was executed for the full area under the curve (integration mode A) between 1653 cm -1 and 1685 cm -1 ( Figure  S12, top left, marked in blue).

c) General synthetic method for in-situ IR measurements
A reaction tube (Ø = 2.5 cm, l = 16.5 cm) was charged with CuI (0.095 g, 0.5 mmol, 10 mol%), 2benzylpyridine (5 mmol), DMSO (10 mL) and acetic acid (0.300 g, 5 mmol). The flask was flushed with O 2 for 1 min, closed with a septum through which the probe was brought as well as a balloon filled with O 2 (the latter pierced through the septum via a needle). The mixture was placed in a preheated oil bath and stirred at 100 °C. At this point the reaction starts and the monitoring software is started.
* Different O 2 /N 2 mixtures were made to determine the reaction order of O 2 . These mixtures were made using a pressure manifold by pressurizing an evacuated steel gas cylinder with pure O 2 to a pressure of X bars followed by adding N 2 to a total pressure of 12 bar. S6 The ratio O 2 /N 2 then equals X/(12-X). * For the experiments with addition of TEMPO, 1 equivalent (5 mmol) of TEMPO was added to the reaction mixture. * To measure the influence of the stirring rate a PTFE coated octagonal magnetic stirring bar (Ø = 5 mm, l = 20 mm) with a pivot ring tube (Ø = 6 mm) was used in combination with an RCT basic stirring plate.  To determine the kinetic isotope effect double deuterated substrate (1-D 2 ) was synthesized by dissolving 1 in 25 mL of D 2 O and adding 0.5 mL of DCl and heating the mixture to 180 °C for 3 hours under microwave irradiation. After the reaction is complete saturated aq. NaHCO 3 (20 mL) is added and the mixture is extracted with diethylether (3 x 20 mL). The organic fractions are collected and dried over MgSO 4 . 1-D 2 was isolated in 94 % yield (± 95 % benzylic deuteration).  ). The singlet of the benzylic protons at 4.16 ppm has almost completely dissapeared and a very small broad triplet, due to mono-deuteration, has appeared at 4.14 ppm. Figure S19. Determination of kinetic isotope effect using in-situ IR spectroscopy. g) Catalyst decomposition?
The possibility of catalyst decomposition was examined by adding another equivalent of substrate after the first reaction was completed (18 hours). The reaction starts up again without any significant loss in reaction rate and evolves to completion. After 42 hours the reaction mixture was worked-up and 92 % (9.2 mmol) of 2-benzoylpyridine (2) was isolated. This indicates that no detectable catalyst degradation is taking place. To determine whether formation of side product 3 has any influence on the catalytical species present in the reaction, the reaction was doped with 20 mol% of alcohol 3 and its effect on the initial rate was examined. Identical initial reaction rates were measured for both reactions. From this can be concluded that alcohol 3 has no influence on the catalytical species present. h) Further proof for existence dimeric catalytic species In addition to the EPR-analysis, in-situ IR was used to support the existence of dimeric catalytical species. The carbonyl stretch vibration of acetic acid located at 1718 cm -1 ( Figure S23) decreases over time. This decrease is caused by the formation of dinuclear acetate bridged Cu II species. This is supported by the S20 fact that the decrease at 1718 cm -1 is accompagnied by an increase of a vibration at 1623 cm -1 . This vibration corresponds to the carbonyl stretch vibration of acetate ions in dinuclear Cu II complexes like Cu(OAc) 2 ·2H 2 O in DMSO ( Figure S24). Integration of the area of the peak at 1718 cm -1 therefore allows us to follow the concentration of free AcOH over time, additionally integration of the area at 1623 cm -1 allows us to monitor the concentration of AcOH caught up in a dinuclear complex. When the initial concentration of CuI was increased more AcOH was used and subsequently more dinuclear complex formed ( Figure S25 and S26).

IV. Involvement of radical intermediates?
To determine whether or not radical intermediates are formed during the catalytic cycle several control experiments were performed. First the effect of addition of a few known radical inhibitors on the rate of the reaction was examined. TEMPO and 1,1-diphenylethylene were selected as radical inhibitors and their addition did not slow down the reaction neither could any adducts be detected. In addition to these two liquid radical inhibitors activated carbon was added which is known to act as a radical scavenger.
The addition of activated carbon did not influence the conversion of the reaction after 4 hours ( Figure  S27). This reaction could not be monitored with in-situ IR spectroscopy due to the tendency of activated carbon to stick to the probe. In addition to these kinetic experiments no organic radical could be detected via EPR-spectroscopy. Any sufficiently stable organic radical (such as a benzylic radical), even in very low quantities, should be detectable by EPR. Finally a radical clock type molecule (2-{[2-(prop-2-en-1-yl)phenyl]methyl}pyridine (26)) was synthesized and oxidized under our standard conditions. If a benzylic radical would be formed it immediately cyclizes in a 5-exo-trig fashion. This reaction is expected to be very fast and outcompetes the intermolecular trapping by O 2 . This ring-closed product could however not be detected and the expected ketone [2-(prop-2-en-1-yl)phenyl](pyridin-2-yl)methanone (27) was isolated in 54% yield. Based on the combination of these experiments we can exclude involvement of radical intermediates. Figure S27. The effect of adding 1 eq. of activated carbon at the start of the reaction on the isolated yield after 4 hours. Reaction conditions: 1 (0.5 mmol), Activated Carbon (1 eq.), CuI (10 mol %), AcOH (1 eq.), O 2 (balloon), DMSO (1 mL), 100 °C.

V. Gas-Uptake measurements
A home-made three-necked round bottomed flask (total volume = 366 mL) with one neck equipped with a valve was attached to a manometer to monitor the pressure. The second neck was fitted with a thermometer to monitor the temperature of the headspace. The remaining one was connected to a three-way valve connected to a vacuum pump and an oxygen bottle. The flask was charged with 2benzylpyridine (0.846 g, 5.00 mmol), CuI (0.095 g, 0.50 mmol) and DMSO (10 mL). The reaction mixture was brought in an oil bath set at 100C under stirring, until the temperature of the headspace remained constant. Next the headspace was flushed for 1 min with O 2 and acetic acid (0.286 mL, 5.00 mmol) was added via a syringe. The valve was closed and the initial pressure was measured (P i ). After 4 hours of reaction the final pressure was noted (P e ) and the flask was removed from the oil bath and opened. The reaction mixture was allowed to reach room temperature and the general work-up was performed. An 1 H-NMR yield of the reaction product using 1,3,5trimethoxybenzene as the internal standard was determined giving the number of moles of reaction product formed (n end ). The average of three experiments was taken to determine the stoichiometric ratio.

VI. Derivation of the rate laws
The rate of formation of 2 equals: It is clear that at low concentrations of 1, the equation can be simplified to (with

Which is a first order relation for [(1)]
At high concentrations of 1, the equation can be simplified to: Which is zeroth order in [ (1)  Kinetic model for inhibition in function of catalyst concentration: Note: If the substrate is N-ligated it acts as a ligand (1), in C-ligation it acts as an anion (A). Other anions (AcO -, I -) and ligands (DMSO, (2)) are not directly involved in the mechanism and are left out of this model for simplicity. (1), Cu II (1) 2 and Cu II 2 (1) 2 are considered to be in fast equilibrium.

S26
1 , 2 , , − , 3 , 4 , , The steady-state approximation can be made for the change of [A] over time The change over time of [Cu(1) 2 A]: To solve this differential equation we set [Cu (1) At t=0 : [Cu(1) 2 A]=0, the equilibrium between 1 and A has already been set at t=0.
When t is small (in the beginning of the reaction) we can approximate by using a Maclaurin expansion: The change over time of Cu 2 (1) 2 A is equal to: In the same manner we can derive for Cu 2 (1) 2 A: The rate of formation of 2 equals:

VII. Synthetic experimental procedures and NMR data
The following section contains the compounds never before synthesized by this oxidation protocol. Characterization data of the remaining compounds (2,7) can be found in our communication. S7

General oxidation procedure:
Two 10 mL microwave vials were subsequently charged with CuI (0.0095 g, 0.05 mmol, 10 mol%), substrate (0.5 mmol), solvent (1 mL) and acetic acid (0.030 g, 0.5 mmol). The vials were flushed with O 2 for 1 min, capped with an aluminum crimp cap/septum and finally stirred at 100 °C for 24 h in an oil bath with a balloon filled with O 2 through the septum. After cooling down to room temperature, the content of the vials was transferred into a separation funnel and the vials were rinsed with dichloromethane (20 mL). Sodium bicarbonate solution (10 mL, sat.) was added and the organic phase was separated. The water phase was extracted twice with dichloromethane (10 mL). The combined organic fractions were washed with brine (20 mL), dried over MgSO 4 and filtered over a pad of Celite ® . The solvent was removed under reduced pressure and the resulting residue was purified via column chromatography with an automated chromatography system using a Silica Flash Cartridge applying a heptane-ethyl acetate gradient (from 100 % heptane to 100 % ethyl acetate in 25 minutes, 25 mL/min). For 1 H-NMR screening of the reaction mixture with an internal standard, a known quantity of 1,3,5trimethoxybenzene was added to the mixture obtained after work-up but before chromatography and everything was carefully dissolved in CDCl 3 . 1 H-NMR is then used to calculate the yield and conversion of the reaction. Alternative work-up procedure: After the reaction is completed and the vials have cooled to room temperature they can be rinsed with dichloromethane (20 mL) and filtered over Celite ® . The solvent can then be removed under reduced pressure and the residue can be further purified by column chromatograpy. This alternative work-up was used in the purification of compound 2 explaining the discrepancy between the earlier reported yield. S7 6-Formylnicotinonitrile (13): The general oxidation procedure was followed using 5-cyano-2-methylpyridine (12) (0.059 g, 0.5 mmol). 6-Formylnicotinonitrile (13) was isolated as a yellow solid in 27 % yield (17 mg) (60 % (38 mg) at 120 °C reaction temperature).  C-NMR (CDCl 3 , 100 MHz) δ C : 191.8, 154.2, 152.8, 140.8, 121.1, 115.8, 113.7

VIII. DFT calculation of equilibrium constants and bond dissociation energies a) Equilibrium constants
To calculate the desired equilibrium constants, the geometries of the different conformations of both the imine-and the enamine-form were optimized at the B3PW91/aug-cc-pVDZ level of theory. Due to the rather large dipole moment of the molecules and the difference in dipole moment of the imine and the respective enamine tautomer (table S5, the self-consistent reaction field (scrf) model was used to account for solvent-solute interactions with DMSO. The Gibbs free energies of all conformers were calculated and the most stable imine and respective enamine conformer were used to calculate the equilibrium constant K eq and pK eq as follows: with the temperature T=298.15 K and ideal gas constant R= 8.31 Jmol -1 K -1 . For all calculations the Gaussian09 program was used. S11

S33
Compounds 28-35 are not included in the manuscript text since they have been previously published. S7 Interesting to note however is the case of compound 34 which has two benzylic positions viable for oxidation. Tautomerization from the double benzylic position provides a pK eq value of 11.7 similar to 2benzylpyridine (1). Tautomerization from the 6-methyl in 34 provides a pK eq of 15.6 which is similar to 2methylpyridine (10). However when the double benzylic position of 34 is oxidized this creates an electron withdrawing effect causing the pK eq value of the 6-methyl to decrease to 14.7 (see 35) making it viable for oxidation as exemplified in the original communication.
The enthalpy of formation of the lowest energy conformers were used. S12 All calculations are performed using the Gaussian G09 rev D.01 program, using the B3PW91/aug-cc-pVDZ level of theory with the scrf model to take solvent-solute interactions with DMSO into account. S11 No correlation between the homolytical BDE of the substrates in Table S6 and their reactivity could be found. For instance, while 2-and 4-benzylpyridine (1 and 6) are reactive, diphenylmethane (8) and 3-benzylpyridine (4) are not. Their bond dissociation energies are however within a margin of error similar (see Table S6 entries 1-4). Furthermore when we tried to oxidize diphenylmethane (8) under our standard conditions in the presence of a pyridine resembling 2-benzylpyridine, which can act as a ligand, no oxidation product was seen and the starting material was recovered in 87 % ( Figure S28).