Aerobic alcohol oxidation and oxygen atom transfer reactions catalyzed by a nonheme iron(ii)–α-keto acid complex

An iron(ii)-benzoylformate complex of a monoanionic facial tridentate ligand catalyzes the aerobic oxidation of sulfides to sulfoxides, alkenes to epoxides, and alcohols to the corresponding carbonyl compounds.

Introduction a-Ketoglutarate-dependent enzymes represent the largest subfamily of nonheme iron enzymes and catalyze a myriad of biological oxidation reactions. 1 These enzymes are involved in important biological processes such as the hydroxylation of protein side chains, DNA/RNA repair, biosynthesis of antibiotics, lipid metabolism, biodegradation of toxic compounds, transcription regulation, and oxygen sensing. [1][2][3] Despite the diversity of reactions, all the enzymes in this subfamily have a common '2-His-1-carboxylate facial triad' motif. 3,4 These enzymes require an a-ketoglutarate cofactor for the reductive activation of dioxygen and subsequent generation of ironoxygen oxidant. 3,5 In the reactions, the metal-bound a-ketoglutarate cofactor undergoes oxidative decarboxylation to form carbon dioxide and succinate. On the basis of a large number of crystallographic, 6,7 spectroscopic, [8][9][10] and DFT studies, 10 a common mechanism of a-ketoglutarate-dependent enzymes has been put forward. 1,2,7,11 Initial binding of dioxygen to the iron(II) center of the enzyme leads to the formation of an iron(III)-superoxo species. The nucleophilic superoxide then attacks the electrophilic keto carbon of the coordinated a-KG to form an iron(IV)-alkylperoxo species. The peroxo species undergoes O-O and C-C bond cleavage to generate CO 2 , succinate, and iron(IV)-oxo intermediate. Nonheme iron(IV)-oxo intermediates have been identied as active oxidants in the catalytic cycles of a number of a-keto acid-dependent enzymes. 5,[12][13][14][15][16][17][18] Biomimetic iron(II)-a-keto acid complexes have also provided useful information about the mechanism of a-ketoglutarate-dependent enzymes. [19][20][21][22][23] In situ generated iron(IV)-oxo species, intercepted in the reactions between iron(II)-a-keto acid complexes of polydentate supporting ligands and dioxygen, have been reported to exhibit stoichiometric reactivity toward alkenes, alkanes, suldes and alcohols. 24 Among the reported models, the iron(II)-a-keto acid complexes of tris(pyrazolyl)borate ligands represent the most efficient functional models of the a-ketoglutarate dependent enzymes. Monoanionic tris(pyrazolyl)borates provide a facial N 3 ligand environment at the metal center and mimic the '2-His-1carboxylate facial triad' motif observed in a-ketoglutarate dependent enzymes. 25 Moreover, the steric and electronic properties of these ligands can be easily tuned by placing appropriate substituents on the pyrazole rings. 26 The iron(II)benzoylformate complex [(Tp Me2 )Fe II (BF)] (BF ¼ monoanionic benzoylformate) represents the rst biomimetic complex of a facial N 3 ligand, which reacts with oxygen within 2 minutes to generate an oxidant capable of epoxidizing cyclohexene and cisstilbene. 27 For cis-stilbene, the retention of conguration in the epoxide product indicates the involvement of a metal-based oxidant. The complex, however, cannot oxidize trans-stilbene, suggesting that the oxidant generated is capable of discriminating between the cis and trans isomers of an olen. 27 While the iron(II)-a-keto acid complex of Tp 3tBu,5iPr is unreactive toward dioxygen, 28 those of Tp iPr2 and Tp Ph2 ligands 21,23,29 react with O 2 to exhibit oxidative decarboxylation. Compared to the Tp Me2 system, iron(II)-a-keto acid complexes of Tp iPr2 or Tp Ph2 display slower reactivity toward dioxygen. The accessibility of dioxygen to the iron center depends on the steric bulk engendered by the substituent on pyrazole which in turn decides the reactivity with dioxygen.
With an objective to develop iron-based catalysts for the selective oxidation of organic substrates with O 2 , we have investigated the reactivity of iron(II) complexes supported by a hydrotris(3-phenyl-5-methylpyrazolyl)borate (Tp Ph,Me ) ligand in the presence of different co-substrates as ligands. 30 The coligands provide the necessary electrons for dioxygen reduction to generate a metal-oxygen oxidant which can oxidize organic substrates. As a part of our investigation, we report herein the synthesis, characterization and dioxygen reactivity of a biomimetic iron(II)-a-keto acid complex, [(Tp Ph,Me )Fe II (BF)] (1) (Scheme 1). In a reaction with dioxygen, the iron(II) complex selectively oxidizes alcohols to ketones, alkenes to epoxides, and suldes to sulfoxides using dioxygen as the oxidant. The interception of a high-valent iron-oxo oxidant from O 2 , and catalytic and mechanistic studies of alcohol oxidation and oxo-atom transfer reactions by complex 1, are presented in this article.

Results and discussion
Complex [(Tp Ph,Me )Fe II (BF)] (1) was isolated from the reaction of equimolar amounts of KTp Ph,Me , iron(II) chloride, and sodium benzoylformate (NaBF) in a CH 2 Cl 2 -CH 3 OH solvent mixture (see Experimental section and Scheme 1). The ESI mass spectrum of complex 1 displays a molecular ion peak at m/z ¼ 689.1 with an isotope distribution pattern attributable to {[(Tp Ph,Me ) Fe(BF)] + H} + (Fig. S1, ESI †). In the 1 H NMR spectrum, the complex displays paramagnetically shied resonances of protons as observed with high-spin iron(II) complexes of the ligand (Fig. S2, ESI †). 30 The optical spectrum of the complex in acetonitrile exhibits broad absorption bands at 537 nm (315 M À1 cm À1 ) and 580 nm (300 M À1 cm À1 ) suggesting a chelating bidentate coordination of the a-keto acid (BF) anion. 19 Although the intensities of absorption bands are low in 1, similar absorption features are observed in iron(II)-benzoylformate complexes of tris(pyrazolyl)borate ligands. 21,27,28 The origin of the absorption bands may be attributed to the MLCT transitions from Fe(II)-to-p* orbital of the keto group. 8 The complex was further characterized by single crystal X-ray diffraction studies. The X-ray structure of the neutral complex reveals that the iron center is coordinated by a tridentate monoanionic face-capping Tp Ph,Me ligand and a bidentate monoanionic benzoylformate (Fig. 1). Benzoylformate is coordinated to the iron center via a carboxylate oxygen (O2) and the carbonyl oxygen (O3) with the Fe1-O2 and the Fe1-O3 distances of 1.957(2)Å and 2.262(2)Å, respectively. The M-O distances are comparable to those reported for ve-coordinate iron(II)-a-keto acid complexes of Tp R1,R2 type ligands. 21,28,30 The average iron(II)-N(pyrazole) bond length is typical of high-spin iron(II) complexes of tris(pyrazolyl)borate ligands. The geometry of the ve-coordinate iron complex can be best described as distorted trigonal bipyramidal (s ¼ 0.57) 31 with the equatorial plane being formed by the carboxylate oxygen O2, and the pyrazole nitrogens N6 and N2. The pyrazole nitrogen N4 and the keto oxygen O3 occupy the axial positions with an N4-Fe1-O3 angle of 172.98(7) (Table S1 †). Interestingly, the Fe-O(keto) and the Fe-N4 bonds are signicantly elongated compared to those in complex [(Tp Ph2 )Fe II (BF)]. 21 Complex 1 reacts with dioxygen in acetonitrile over a period of 1.5 h, during which the violet solution turns colorless. In the reaction, the MLCT bands at 537 nm and 580 nm decay following a pseudo-rst order rate (k obs ¼ 5.5 Â 10 À4 s À1 ) to yield an almost featureless optical spectrum, indicating the oxidative decarboxylation of coordinated BF (  at m/z ¼ 661.2 with the isotope distribution pattern calculated for {[(Tp Ph,Me )Fe(OBz)] + H} + (OBz ¼ benzoate). When the reaction is carried out with 18 O 2 , the ion peak is shied two mass units higher to m/z ¼ 663.2 ( Fig. 2 inset). The 1 H NMR spectrum of the nal reaction solution bears resemblance to that of an independently prepared iron(II)-benzoate complex [(Tp Ph,Me )Fe II (OBz)] (2) (see Experimental section, and Scheme S1 and Fig. S2, ESI †). The oxidized solution of 1 is X-band EPR silent at 77 K, further indicating the formation of an iron(II) complex. Time-dependent 1 H NMR spectra aer the removal of the metal ions establish the complete decarboxylation of BF to benzoic acid in about 1.5 h (Fig. S3, ESI †). The conversion of BF to benzoic acid by the iron(II) complex with the incorporation of one oxygen atom from O 2 into benzoate functionally mimics the reaction catalyzed by a-keto acid-dependent oxygenases.
To intercept the iron-oxygen intermediate from 1, external substrates were used as indirect probes (Scheme 2). 29,32,33 Reaction of 1 with O 2 in the presence of thioanisole (10 equiv.) affords 25% thioanisole oxide as the only product (Fig. S4, ESI †). A comparatively higher amount of interception is observed with smaller substrates such as dimethyl sulde (DMS) and dimethyl sulfoxide (DMSO). While DMSO is formed from DMS in about 75% yield, around 50% dimethyl sulfone is formed from DMSO ( Fig. S5 and S6, ESI †). On the other hand, relatively bulky dibenzothiophene could intercept the active oxidant to an extent of only 6%.
The reaction of complex 1 with dioxygen gives the activation parameters of DH s ¼ 27 kJ mol À1 and DS s ¼ À216 J mol À1 K À1 determined from Eyring analysis (Fig. S16, ESI †). The large negative entropy of activation indicates that the rate determining step is associative in nature. Hammett analyses from competitive reactions using equimolar amounts of a parasubstituted thioanisoles and thioanisole with complex 1 and dioxygen reveal a negative r value of À1.00 (AE0.13) supporting an electrophilic oxidant being responsible for sulde oxidation (Fig. 3).
The GC-mass spectrum of thioanisole oxide, which exhibits an ion peak at m/z ¼ 140, is shied two mass unit higher to m/z ¼ 142 (90% 18 O labelled oxygen atoms) when the reaction is carried out with 18 O 2 (Fig. S17, ESI †). The labeling experiment clearly indicates the incorporation of one oxygen atom from molecular oxygen into the product. Mixed labeling experiments on thioanisole and styrene using 16   oxide and styrene epoxide, respectively (Scheme 3, Fig. 4 and S17, ESI †). The reaction of complex 1 with styrene and 18 O 2 shows about 40% incorporation of one labelled oxygen atom into the epoxide product (Fig. 4). The low incorporation of labelled oxygen atoms from 18 O 2 indicates that water present in the solvent may exchange with the oxidant responsible for the epoxidation reaction. When the reaction of 1 with styrene is performed using 16  incorporation is estimated (Fig. S20c, ESI †). These results clearly indicate that the oxygen atom in mandelic acid is derived from a putative iron(IV)-oxo oxidant which exchanges its oxygen atom with water.
Heme and nonheme iron-oxo intermediates have been reported to oxidize alcohols exclusively by hydrogen atom abstraction from the a-CH of alcohol. 43,44 High-valent iron-oxo complexes show very large (>10) KIE values. Intermolecular competitive oxidation of a mixture of PhCH 2 OH and PhCD 2 OH (90% D) by complex 1 gives a kinetic isotope effect (KIE) value of 2.33. A low KIE with a value of around 2.2, however, has been obtained in the manganese-catalyzed oxidation of alcohols, [45][46][47] where high-valent Mn-oxo species have been proposed as active intermediates. Cyclic alcohols such as cyclobutanol or cyclopentanol are frequently used as mechanistic probes to distinguish between a one-electron (open chain aldehyde) versus twoelectron (cyclic ketone) process in alcohol oxidation reactions. 48 Complex 1, upon reaction with cyclobutanol, afforded cyclobutanone as the exclusive product suggesting that alcohol oxidation by complex 1 takes place via a two-electron process (Scheme 5 and Fig. S34, ESI †). The excellent selectivity for alcohol oxidation along with the KIE value suggests that hydroxyl radical is not involved in the oxidation pathway, rather that a high-valent iron-oxo species (I in Scheme 6) is the active oxidant generated from 1. Abstraction of the a-H of the metal bound substrate (II) results in a radical-based intermediate (III) which upon electron transfer rearranges to aldehyde and iron(II)-benzoate species (2). The oxidation of 4-nitrobenzyl alcohol with 18 O 2 by complex 1 reveals no incorporation of labelled oxygen into 4-nitrobenzaldehyde. A mixed labeling experiment with 16 O 2 in H 2 18 O also shows no incorporation of labelled oxygen into the aldehyde product (Fig. S35, ESI †). These results indicate that the oxygen atom in the aldehyde product does not derive from the oxo-iron(IV) unit. Therefore, the aldehyde product is formed by initial hydrogen atom abstraction followed by electron transfer reaction and not by a gem-diol pathway (Scheme 6). 48 For the epoxidation of alkenes, isolation of both the cis-and trans-epoxide from cis-stilbene indicates that the reaction takes place via a step-wise process, not through a concerted pathway. Of note, for the iron(II)-benzoylformate complex of Tp Me2 ligand retention of conguration in the epoxide product was reported. 27 The proposed radical intermediate (IV), formed upon one electron transfer to the iron(IV)oxo, may react with excess oxygen to give more epoxide (for cyclooctene) 49 or allylic oxidation products (for cyclohexene). However, the product distribution does not change in the presence of radical quenchers such as methanol or TEMPO.
These results indicate that the reaction does not proceed via a radical auto-oxidation pathway. It has been reported that the iron(II)-benzoylformate of Tp Ph2 ligand reacted with dioxygen to undergo oxidative decarboxylation concomitant with intra-ligand hydroxylation. 21 On the contrary, complex 1 is quantitatively converted to 2, but no intra-ligand hydroxylation is observed. In the absence of any substrate, the O 2 -derived oxidant from 1 likely decays via some unproductive pathway. But the CT bands of the iron(II)-benzoylformate complex could be regenerated by addition of a solution of benzoylformic acid (HBF) and NEt 3 (1 equiv.) Scheme 5 Products derived from different alcohols used as indirect probes to intercept the active iron-oxygen oxidant from 1.

Scheme 6
Proposed mechanisms for the formation of high-valent iron-oxo oxidant from 1 and the catalytic oxidation of substrates. (Fig. S36, ESI †) to the nal oxidized solution aer the reaction of 1 with O 2 in the presence of excess substrate. This process could be repeated up to 4 cycles. This prompted us to investigate the catalytic reactivity of complex 1 toward alcohols, suldes and alkenes (Table 1 and Experimental section). Acetonitrile was found to be the best solvent for catalytic studies. The reaction with 25 equiv. of NaBF and 50 equiv. of benzyl alcohol affords benzaldehyde with a turnover number (TON) of 14. 4-Nitrobenzyl alcohol is found to be the best substrate with a TON of 16 for the formation of 4-nitrobenzaldehyde and a TON of around 19 for the conversion of BF to benzoate. Almost 50% (TON of 5) of 4-nitrobenzyl alcohol could be selectively converted to 4nitrobenzaldehyde using 10 equiv. each of substrate and BF. The efficiency of the catalyst was studied by using structurally diverse secondary and primary alcohols under the optimized reaction conditions. Almost similar TONs are obtained with cinnamyl alcohol as substrate (Table 1) (Fig. S37, ESI †). The addition of more BF/substrate did not improve the catalytic TON. In case of 4-hydroxybenzyl alcohol, the poor yield of aldehyde can be attributed to the coordination of phenolate oxygen to the iron(II) center which prevents BF from coordinating to the iron center aer the rst cycle.
Complex 1 acts as a catalyst in OAT reactions for sulde and alkene oxidations ( Table 1). The highest TON is obtained in the epoxidation of styrene (TON ¼ 6). Under similar experimental conditions, cyclooctene shows a TON of 3 whereas cis-2-heptene and 1-octene exhibit no catalytic TON. Similarly, TONs of more than 6 are obtained in the oxidations of DMS and 4-methyl thioanisole (Table 1 and Fig. S38, ESI †). It is important to mention here that iron(II)-benzoate complex [(Tp Ph,Me ) Fe II (OBz)] (2) does not show any oxidation of substrates (thioanisole/cyclooctene/benzyl alcohol/styrene) under similar experimental conditions.
Reactions with complex 1 and varying amounts of BF (1-5 equiv.) in the absence of an external substrate yield only one equivalent of benzoic acid. In none of the reactions is excess BF added to the reaction consumed. These observations rule out the possibility of a parallel path for consumption of BF in the catalytic process. In the presence of an external substrate only, BF is consumed further to generate the active oxidant in the catalytic cycle. However, low TONs are observed with substrates such as alkenes and suldes compared to those for BF decarboxylation. The reason for the relatively low TONs for substrate oxidation with complex 1 was also investigated. The Tp Ph,Me ligand has a tendency to hydrolyze resulting in the formation of free pyrazole (PzH) in solution. The free pyrazole can coordinate to the iron(II)-benzoate center forming a stable complex. The ESI-MS of the solution aer catalytic oxidation shows an ion peak at m/z ¼ 697.1 with the isotope distribution pattern calculated for [(Tp Ph,Me )Fe(PzH)] + (Fig. S39, ESI †). Moreover, Xray quality single crystals of the 5-methyl-3-phenylpyrazole adduct of the iron(II)-benzoate complex were isolated. The X-ray crystal structure (Fig. 5) 21 The hydrogen bonding interaction between the benzoate moiety and the pyrazole makes the complex 2 PzH more stable than 2. Thus, accumulation of 2 PzH during the catalytic reaction results in the deactivation of the catalyst. Hydrolysis of one equivalent of the Tp Ph,Me ligand can produce three equivalents of pyrazole, of which one equivalent gets coordinated to the iron(II)-benzoate complex. Thus, the other two equivalents of pyrazole present in solution might quench the metal-based oxidant without oxidizing the substrate.

Conclusions
In summary, we have isolated and characterized a nonheme iron(II)-benzoylformate complex of a facial tridentate ligand, Tp Ph,Me . The complex exhibits versatile reactivity toward substrates such as suldes, alkenes, and alcohols. An iron(IV)oxo species, generated in situ in the oxidative decarboxylation of coordinated benzoylformate, is postulated as the active oxidant. The iron(II)-benzoylformate complex oxidatively converts phenylacetic acid to mandelic acid and phenoxyacetic acids to the corresponding phenols, mimicking the function of hydroxymandelate synthase (HMS) and 2,4-dichlorophenoxyacetate dioxygenase (TfdA), respectively. Furthermore, the complex performs catalytic oxidation of alcohols to aldehydes, and oxygen atom transfer reactions to olens and suldes using dioxygen as the oxidant.

Experimental
All chemicals and reagents were purchased from commercial sources and were used without further purication. Solvents were distilled and dried before use. Preparation and handling of air-sensitive materials were carried out under an inert atmosphere in a glove box. Air-sensitive complexes were prepared and stored in an inert atmosphere. The ligand KTp Ph,Me was synthesized according to the protocol reported in the literature. 58 Fourier transform infrared spectroscopy on KBr pellets was performed on a Shimadzu FT-IR 8400S instrument. Elemental analyses were performed on a Perkin Elmer 2400 series II CHN analyzer. Electro-spray ionization mass spectra were recorded using a Waters QTOF Micro YA263. 1 H NMR spectra were measured at room temperature using a Bruker DPX-500 spectrometer. Solution electronic spectra (single and time-dependent) were measured on an Agilent 8453 diode array spectrophotometer. GC-MS measurements were carried out using a Perkin Elmer Clarus 600 using an Elite 5 MS (30 m Â 0.25 mm Â 0.25 mm) column with a maximum temperature of 300 C. Labeling experiments were carried out using 18   a methanolic solution (1 mL) of sodium benzoylformate (0.086 g, 0.5 mmol) was added. The resulting violet solution was allowed to stir for 8 h and then dried. The residue obtained was dissolved in dichloromethane (10 mL) and ltered. The ltrate was further dried, re-dissolved in acetonitrile and ltered. The ltrate was then evaporated to dryness. To the residue, 0.5 mL of dichloromethane and 5 mL of hexane were added and then stirred for 1 h during which time a violet precipitate settled down. The violet solid was then collected by ltration. X-ray quality single-crystals were obtained by slow evaporation from the solution of the complex in dichloromethane and hexane mixture. Yield: 0.27 g (80%). Anal. cald for C 38  [(Tp Ph,Me )Fe II (OBz)] (2). A suspension of FeCl 2 (0.064 g, 0.5 mmol) and Tp Ph,Me (0.260 g, 0.5 mmol) in 10 mL of dichloromethane was vigorously stirred for 20 minutes. To the resulting mixture a methanolic solution (1 mL) of sodium benzoate (0.072 g, 0.5 mmol) was added. The colorless solution was allowed to stir for 8 h and then dried. The residue obtained was dissolved in dichloromethane (10 mL) and ltered. The ltrate was then dried and washed twice with 5 mL of hexane to obtain a pale white solid. Anal. cald for C 37

Reactivity with dioxygen
A solution of the complex (0.02 mmol) in 10 mL of dioxygen saturated CH 3 CN was allowed to stir at room temperature. Aer the reaction, the solvent was removed under reduced pressure and the residue was treated with 10 mL of 3 M HCl. The organic products were extracted with diethyl ether and the organic phases were washed with brine solution. The combined organic part was dried over Na 2 SO 4 and the solvent was removed to dryness. The organic products were analyzed by 1 H NMR spectroscopy or by GC-MS. Quantication of benzoic acid was done by comparing the peak area associated with two ortho protons of benzoic acid (d 8.09-8.11 ppm) with one proton (d 6.612 ppm) of 2,4-di-tert-butylphenol used as an internal standard.

Interception studies with external substrates
Complex 1 (0.02 mmol) was dissolved in 10 mL of dry acetonitrile. To the solution the desired amount of external substrate was added. The solution was then saturated with dioxygen by purging dioxygen for 5 min and the reaction was allowed to continue for 1.5 h at room temperature under a dioxygen atmosphere. The reaction solution was then dried and the residue was treated with 3.0 M HCl solution (10 mL). The organic products were extracted with diethyl ether (3 Â 20 mL) and washed with brine solution (2 Â 20 mL). The combined organic phase was dried over Na 2 SO 4 and the solvent was removed under high vacuum. The organic products were then analyzed using 1 H NMR spectroscopy or GC-MS. The reaction with adamantane was carried out in benzene instead of acetonitrile. All experiments were performed in triplicate and the average values are reported in each case.
The products were quantied by comparing the peak area associated with two ortho-protons of benzoic acid (d 8.09-8.11 ppm) with the peak area for the protons of the oxidized substrate. In certain cases, where the two ortho protons of benzoic acid could not be integrated, 2,4-di-tert-butylphenol (0.02 mmol) was used as an internal standard.
Quantication of thioanisole oxide was done by comparing the peak area of the three protons -CH 3 (d 2.76 ppm) with two ortho-protons of benzoic acid. The products derived from dihydroanthracene (DHA), uorene, and adamantane were analyzed by GC-MS and quantied using calibration curves obtained for authentic compounds. The aldehyde products were quantied by comparing the peak area associated with one proton of aldehyde hydrogen (d 9.5-10.1 ppm) with either two ortho protons of benzoic acid or by one proton (d 6.612 ppm) of 2,4-di-tert-butylphenol used as an internal standard. Phenol and 2,4-dichlorophenol were quantied by comparing the peak area associated with two protons of phenol (d 6.82 ppm) and one proton of 2,4-dichlorophenol (d 6.99 ppm) with two ortho protons of benzoic acid (d 8.10 ppm).

Interception with dimethyl sulfoxide and dimethyl sulde
Complex 1 (0.02 mmol) was dissolved in dry acetonitrile (10 mL). To the solution dimethyl sulfoxide (10 equiv., 0.2 mmol) or dimethyl sulde (100 equiv., 2 mmol) was added. Dry dioxygen gas was bubbled through the solution for 5 min and the solution was stirred at room temperature under an oxygen atmosphere for 1.5 h. The solvent was then removed from the reaction mixture and distilled benzene (2 mL) was added to dissolve the residue. A slight excess of sodium dithionite (0.04 mmol) was then added to the benzene solution followed by the addition of D 2 O (1 mL) and the resulting solution was stirred for 15 min. To the solution was added 1,10phenanthroline monohydrate (0.06 mmol) and stirred for an additional 30 min. The D 2 O layer was then collected and analyzed by 1 H NMR spectroscopy. 1

Interception with alkenes, alkanes and alcohols
Complex 1 (0.02 mmol) was dissolved in 2 mL of dioxygen saturated dry acetonitrile. To the solution external reagents, 100 equivalents (2 mmol) of alkenes or 10 equivalents of alcohols (0.2 mmol), were added. Dioxygen was purged through the solution for 5 min and the reaction solution was allowed to stir at room temperature for 1.5 h. The solution was then passed through a 15 cm silica column (60-120 mesh size) using dichloromethane/diethyl ether as the eluent. The combined organic phase was then analyzed by GC-mass spectrometry. The reaction with ethylbenzene was performed in benzene instead of acetonitrile. Quantication of the oxidized products was carried out by GC-mass spectrometry using standard calibration curves obtained with authentic compounds and naphthalene as an internal standard.

Kinetic isotope effect
Complex 1 (0.02 mmol) was dissolved in 2 mL of acetonitrile. To the solution was added a 1 : 1 mixture of PhCD 2 OH (0.2 mmol, 90% D) and PhCH 2 OH (0.2 mmol). Dioxygen gas was then purged through the solution for 5 min and the reaction solution was allowed to stir at room temperature for 1.5 h. The solution was then passed through a 15 cm silica column (60-120 mesh size) using dichloromethane/diethyl ether as the eluent. The combined organic phase was then analyzed by GC-mass spectrometry. The percentage of benzaldehyde PhCHO/PhCDO formed was calculated using the relative intensities of the mass peaks in GC-MS.

Catalytic experiments
The catalytic experiments were carried out using 0.02 mmol of the complex either in benzene or in acetonitrile under the different conditions mentioned in Table 1. The reactions were carried out for about 8 h.
NMR data of the oxidized products. X-ray crystallographic data collection, renement and solution of the structures X-ray single-crystal data for 1 and 2 PzH were collected at 120 K using Mo K a (l ¼ 0.7107Å) radiation on a SMART-APEX diffractometer equipped with CCD area detector. Data collection, data reduction, structure solution and renement were carried out using the APEX II soware package. 59 The structure was solved by a direct method and subsequent Fourier analyses and rened by the full-matrix least-squares method based on F 2 with all observed reections. 60 The non-hydrogen atoms were treated anisotropically. Routine SQUEEZE 61 was applied to the intensity data of complex 2 PzH to take into account disordered solvent molecules.
Crystal data of 1: MF ¼ C 38