Transfer Hydrogenation of Ketones Catalyzed by Symmetric Imino-N-heterocyclic Carbene Co(III) Complexes.

The synthesis of new moisture-sensitive imine-functionalized N-heterocyclic carbene (NHC) precursor salts [1-(2-[(hydroxyl-benzylidene)-amino]-ethyl)-3-R-3H-imidazole-1-ium bromide; R = methyl (1a), ethyl (1b), and benzyl (1c)] is reported. Subsequent deprotonation of 1a-c and coordination of the in situ generated NHC ligands to CoBr2 led to the isolation of air-stable six-coordinate Co(III) complexes 2a-c, respectively. All the salts and complexes were fully characterized. Single-crystal X-ray analysis of 2a and 2c showed octahedral Co centers hexacoordinated to two NHC carbons, two imine nitrogen atoms, and two phenolate oxygens in the form [C^N^O(Co3+)C^N^O]. The complexes were used in the catalytic transfer hydrogenation (CTH) of a range of ketones in 2-propanol as the solvent and hydrogen donor. Based on a low catalyst concentration of 0.4 mol %, significant conversions in the range of 70-99% were recorded at high turnover frequencies up to 1635 h-1. A mechanism to account for the steps involved in the CTH of cyclohexanone by complex 2a is proposed and supported by data from cyclic voltammetry, low-resolution mass spectrometry, UV, and IR spectroscopic techniques.

Introduction

An imino N-n class="Chemical">heterocyclic carbene (Im-NHC) is a multifunctional and multidentate ligand framework derived with the aim of a synergistic combination of properties of two moieties comprising a Schiff base and NHC in one ligand framework.[1] The idea is to enrich the chemistry of the NHC family of ligands by providing additional binding sites that may lead to the subsequent harnessing of the combined attributes of both the Schiff base and the carbene. The chemistry of Schiff bases has been well explored and documented in the literature, and they are known for their ability to stabilize metal centers in high and low oxidation states.[2] Likewise, the successful isolation and characterization of stable NHCs have led to the emergence of both transition metal-based and transition metal-free NHC catalysts.[3] The NHCs’ strong sigma donor ability has been utilized in various transition metal complex-promoted organic transformations.[4] Therefore, harnessing the potentials of Im-NHC ligands is an advantage in fine-tuning the steric and electronic properties of bound metal centers utilized in homogeneous catalysis. Over the past decade, considerable attention has been devoted to this area as a new frontier of intense research activity for organometallic and coordination chemistry.[5] Besides their robust thermal stability, the other interesting feature of Im-NHC metal complexes is the ease with which the ligand could be varied leading to a theoretically endless possibility for both steric and electronic tuning and hence stabilization of reactive species during catalytic reactions.[6] As compared to monodentate NHC-metal complexes, Im-NHC metal complexes display a much broader range of structural forms and wider scope of reactivity and can overcome decomposition pathways leading to higher stability.[5a,7,8] A recent investigation on Im-NHC Pd and Ni complexes has revealed that the complexes were active for Suzuki coupling reactions, hydroformylation of 1-octane, conjugate addition reactions, and styrene polymerization.[5a,7b,8a,9] Likewise, Im-NHC Ru and Ni complexes are reported to be stable and excellent catalysts for the reduction of aldehydes and ketones.[10] Also, bimetallic and polynuclear transition metal complexes resulting from Im-NHC ligation have been reported,[7,11] and the hybrid nature of the Im-NHC ligand framework has often led to improved actions of the active metal species during catalysis.[12] Hence, the successful utilization of these transition metal-based catalysts has contributed immensely to human development and progress.[3b,13]

More specifically, n class="Chemical">the use of transition metal-based NHC complexes as catalysts in transfer hydrogenation (TH) reactions was pioneered by Nolan’s group.[14] Currently, TH has become the most widely applied and convenient route of transition metal-based hydrogenation reactions.[15] The process is well established, inexpensive, and environmentally friendly as compared to alternative methods of direct hydrogenation using high pressure molecular H2 gas.[16]

Although significant progress has been made regn class="Chemical">arding catalytic TH (CTH) of ketones, it is, however, worth mentioning that catalysis based on heavy metals that include Ru, Rh, Ir, Pt, and Pd still dominate the current literature.[17] Although these have proven to be high yielding catalysts, there is also a growing interest in the catalytic capabilities of first-row transition metal (Fe, Co, Ni, and Cu) complexes. This is due to their low cost, relatively lower toxicity, and the fact that they are abundant and readily available.[18] In addition, recent studies on the application of earth-abundant metals as alternatives to heavy metal-based catalysts in various organic reactions that include the TH of ketones have been reported.[15a,18c,19] Magubane et al. reported the synthesis of Ni(II) and Fe(II) complexes and their application as catalysts in TH of ketones.[20] In another related study, the use of nickel, iron, and cobalt complexes containing chiral aminophosphine ligands as catalysts in TH of ketones was reported and the catalytic systems showed excellent substrate conversions.[21] For almost a decade, we have been working on these low-cost, earth-abundant metals. We have previously reported the utilization of iron and nickel complexes as catalysts in the TH of both aromatic and aliphatic ketones with good to excellent conversions.[10b,18b] In this report, we are expanding the scope to cobalt complexes, and this is because, to the best of our knowledge, there is no report on NHC–Co complexes of this type and their application as catalysts for TH reactions.

Results and Discussion

Synthesis and Characterization

The n class="Chemical">cobalt complexes 2a–c were prepared (Scheme ) by the direct reaction of in situ generated free carbenes and CoBr2, a procedure similar to the one reported by Gorczyński et al.[22] The ligand precursors 1a and 1b were first reported by Zhu et al.,[1] while 1c was reported in our previous article.[11] The complexes (2a–c) were prepared by stirring 1 mmol equiv of the respective ligand precursors (1a–c) with KOBu (3 equiv) in methanol for 15 min and then CoBr2 (0.5 mmol) was slowly added to the in situ generated free carbene, and the resulting mixture was allowed to stir at room temperature for another 4 h. All the afforded complexes were isolated in excellent yields (71–96%). Complexes 2a and 2b appeared as reddish-brown powders, while 2c was isolated as a maroon powder. All the complexes 2a–c were stable in air and soluble in methanol, acetone, dichloromethane, chloroform, and acetonitrile.

Scheme 1

Synthetic Route to the Title Complexes 2a–c

Spectroscopic Analysis

A clear sign of ligand n class="Chemical">coordination is observed in the 1H NMR spectrum of compound 2a, which upon integration showed a total of 14 protons as opposed to the 16 protons in 1a. This corresponds to the loss of the phenolic proton (observed at 12.5 ppm in 1a) and the carbene (C2) proton (observed at 10.2 ppm in 1a). Meanwhile, there is a general upfield shift for all other protons in the compound. For instance, the imine proton which was observed around 8.4 ppm in the ligand precursor shifted to 7.7 ppm in the complex, and the results agree with values reported for a similar compound.[10a] Complexes 2b and 2c did not yield resolved NMR data, perhaps because of paramagnetism exhibited by the cobalt(III) octahedral complexes.[23] Other characterization data, from the high-resolution mass spectrometry (HRMS) of the complexes, are consistent with the calculated values for the relative abundance of cobalt. The results confirm that the complexes were obtained as constituted in Scheme . The Fourier transform infrared (FTIR) spectra of the complexes 2a–c show the disappearance of the phenolic (OH) absorption band around 3400 cm–1, a sign of the coupling of the phenoxy v(OH) bond and the formation of an oxygen-to-metal bond. However, the presence of residual water molecules as indicated by the CHN and single-crystal analysis was responsible for the broad signals observed around 3400 cm–1 in the IR spectra of 2b and 2c. Further evidence of complexation is the slight shift of the v(C=N) stretching frequencies from the region around 1628 cm–1 in the ligand precursors to 1615 cm–1 observed in the spectra of the metal complexes. These trends are all in agreement with reported values for similar compounds in the literature.[24] Characteristic vibrational frequencies of v(O–Co and N–Co) were located around the region of 426 and 486 cm–1, respectively.[25] CHN elemental analysis data are all within the acceptable range (±0.3) between the calculated and measured values.

Single-Crystal Structural Analysis by X-ray diffraction

The slow evaporation of n class="Chemical">methanolic solutions of the complexes yielded crystals suitable for analysis by single-crystal X-ray diffraction for both 2a and 2c. Details on methods for crystallographic data acquisition, reduction, and refinement are available (see Table SI 1 in the Supporting Information). ORTEP diagrams depicting the molecular structures of compounds 2a and 2c are presented in Figures and 2, respectively. Both structures show hexacoordinated centro-symmetric coordination of the tridentate ligands around each Co(III) center. Compound 2a is composed of a central cobalt ion bound to two moieties of the ligand 1-{2-[(2-hydroxy-benzylidene)-amino]-ethyl}-3-methyl-3H-imidazole-1-ium bromide, while 2c is similarly composed with the ligand being 1-{2-[(2-hydroxy-benzylidene)-amino]-ethyl}-3-benzyl-3H-imidazole-1-ium bromide. Each six-coordinate octahedral geometry is made up of two phenolate oxygen donors, two imine nitrogen donors, and two carbene donors for both 2a and 2c.[18c,26] In 2a, the bond angles between the axial atoms C(4)–Co(1)–N(3) and the equatorial ones C(4)–Co(1)–O(1) range between 83.75(12) and 96.25(12). Similarly, in 2c, the bond angles between C(12)–Co(1)–N(1) and C(12)–Co(1)–O(1) range between 84.58(13) and 95.42(13). The respective bond distances for C(4)–Co(1), N(3)–Co(1), and O(1)–Co(1) in 2a are 1.966(3), 1.931(3), and 1.904(2); likewise, in 2c, the bond distances for Co(1)–C(12), Co(1)–N(1), and Co(1)–O(1) are 1.963(3), 1.934(3), and 1.902(2), respectively. Results for the bond angles and bond distances recorded for both 2a and 2c are comparable to reported values for similar cobalt complexes.[18c,27]

Figure 1

Thermal ellipsoid plot of the asymmetric unit of 2a drawn at the 50% probability level. Hydrogen atoms are omitted for clarity. Selected bond lengths (Å): Co–C(4), 1.966(3); Co–N(3), 1.931(3); Co–O(3), 1.904(2).

Figure 2

Thermal ellipsoid plot of the asymmetric unit of 2c drawn at the 50% probability level. Hydrogen atoms are omitted for clarity. Selected bond lengths (Å): C(12)–Co 1.963(3), N(1)–Co 1.935(3), and O(1)–Co 1.902(2).

Thermal ellipsoid plot of n class="Chemical">the asymmetric unit of 2a drawn at the 50% probability level. Hydrogen atoms are omitted for clarity. Selected bond lengths (Å): Co–C(4), 1.966(3); Co–N(3), 1.931(3); Co–O(3), 1.904(2).

Thermal ellipsoid plot of n class="Chemical">the asymmetric unit of 2c drawn at the 50% probability level. Hydrogen atoms are omitted for clarity. Selected bond lengths (Å): C(12)–Co 1.963(3), N(1)–Co 1.935(3), and O(1)–Co 1.902(2).

CV Data Analysis

A cathodic scan in n class="Chemical">the negative direction records the respective reversible reduction wave peak at circa −1.107, −1.206, and −1.156 V (Figures SI 15, 16, & 17) for 2a, 2b, and 2c which are due to the process (Co3+) + e– → (Co2+). The ease of reduction of the analytes from Co(III) to Co(II) is due to the hemilabile weakly coordinated imine N(s) that allows for accommodation of the added electron density in the antibonding d orbital of the metal.[28] Upon reversal of the scan direction, all the complexes were re-oxidized from Co(II) to Co(III) at potentials of −1.178 V (2a), −1.375 V (2b), and −1.224 V (2c), thus regenerating the six-coordinate (Co3+) complexes. At 100 mV/s, the ΔEp values for the complexes are 0.071 V 2a, 0.169 2b, and 0.068 V 2c (Table ).

Table 1

CV Data of Cobalt Complexes 2a–2ca

complex	Epa (V)	Epc (V)	E1/2 (V)	ΔE (Epa – Epc)	process
2a	1.178	1.107	1.444	0.071	Co3+ ⇋ Co2+
2a	0.678	0.396	1.074	0.282	Co2+ ⇋ Co1+
2b	1.375	1.206	1.291	0.169	Co3+ ⇋ Co2+
2c	1.224	1.156	1.190	0.068	Co3+ ⇋ Co2+

Oxidation (anodic) and reduction (cathodic) peak potentials (Epa and Epc, respectively).

Although n class="Chemical">the results are slightly higher than the standard values of 0.058 V for one-electron reversible systems, they are still within the accepted values for a reversible process.[29] Nicholson and Shain reported that in reversible systems, the position of the reverse peak may shift as much as 5 mV.[30] However, in an apparent indication of a simple one-electron process, a second weak reduction process of (Co2+) + e– → (Co1+) was observed with 2a at a much lower potential of circa −0.4 V and oxidation potential of circa −0.7 V.[28] This second wave indicated a quasi-reversible behavior with a ΔEp value around 0.2 V.[29]

Results for both n class="Chemical">the cathodic reductions and the anodic oxidations are summarized in Table .

The production of industrially relevant n class="Chemical">feedstock from simple organic substrates continues to attract the attention of researchers interested in the homogenous hydrogenation of polar functional groups.[31] The process may occur via the use of molecular hydrogen (H2) or the more convenient and safer alternative of using hydrogen transfer methods. Instead of pressurized gas, TH relies on the use of solvents as hydrogen donors of which the most frequently used include isopropanol, ammonia boranes, and formic acid.[16,32] Majority of the Co-catalyzed hydrogenation reactions reported to date are based on complexes stabilized by tri- and tetradentate donor ligand frameworks.[31,33] Some of the leading Co-based catalytic systems include the work by Milstein and co-workers,[34] who reported on the hydrogenation of esters catalyzed by Co lutidine-based pincers with a catalyst loading of 2–4 mol %, at 130 °C and 50 bar H2 gas pressure. The same catalyst was also used to promote nitrile hydrogenation to primary amines.[35] The selectivity of Co-catalyzed hydrogenation of ketones and aldehydes was reported by Kempe and co-workers,[36] while Jones et al. reported on the use of molecular H2 in the hydrogenation of esters.[37] Similarly, Zhang and Hanson reported the application of cobalt amino pincer catalysts in the reduction of ketones, aldehydes, and imines. However, their catalysts showed poor selectivity for the reduction of carbonyls that are in conjugation with olefins.[38] Finally, Shao et al. also utilized ammonia boranes as hydrogen sources in the development of Co catalysts for the TH of nitriles.[39]

In this study, n class="Chemical">the Im-NHC cobalt complexes 2a–c were used as catalysts for the TH of a variety of ketones using KOH as an auxiliary base in 2-propanol as the solvent and hydrogen donor (Table ). The selective reduction of acetophenone to the corresponding 1-phenylethanol (Scheme ) was selected at the onset as a model reaction to optimize the reaction conditions. The effects of the catalyst and the base were first tested with a reaction conducted over a period of 8 h without the use of an auxiliary base, which resulted in no conversion to the desired product. However, the reaction with only KOH (20 mol % in 2 mL of 2-propanol, no added catalyst) resulted in 26% conversion to 1-phenylethanol. This observation is in line with literature values for a reaction conducted under similar conditions.[10a] However, to determine the most effective catalyst concentration, reactions were conducted with a gradient of catalyst concentration for compounds 2a–c ranging from 0.1 to 1 mol % (Figure ). The overall results show that there is a linear relationship between the catalyst concentration and substrate conversion up to 0.4 mol %. A further increase in catalyst concentration to 1 mol % resulted in a general decline in activity; hence, all further studies were conducted at 0.4 mol % concentration of the catalyst. Spectroscopic monitoring of the reaction mixture over the course of 6 h did confirm the stability of the metal complex during the course of the CTH (see below). Hence, catalyst decomposition is ruled out as a reason for the decline in activity at higher concentrations. However, several reports in the literature where a similar trend was observed have opined that it is due to the usual agglomeration of the complex in solution, consequently lowering the number of available active species.[20,40] In the absence of any data to the contrary, this is a plausible explanation for the trend observed in Figure . It is also clear from Figure that the catalytic system based on complex 2a is more efficient than either 2b or 2c. For example, the 4 h reaction with 0.1 mol % concentration of 2a yielded 42% conversion whence 2b and 2c yielded 30 and 36% conversions, respectively.[15d,41] The behavior of the catalysts showed that steric hindrance due to the varying wingtip N-substituents is slightly significant in regulating substrate access to the CoII/III centers during the course of the CTH (see the proposed mechanism). Hence, complex 2a bearing the least sterically hindering methyl N-substituent exhibited slightly higher activity than either 2b (ethyl) or 2c (benzyl). In general, the catalytic activities are within a statistically insignificant range which is similar to observations reported for Ni–NHC complexes used for CN coupling reactions.[42] On this basis, we conducted further optimization studies and substrate scope analysis with only complex 2a as the catalyst. The results (Figure ) show that 98 and 95% conversions of cyclohexanone and benzophenone were achieved in 4 and 2 h, respectively, indicating a much better efficiency of the catalyst for these substrates as compared to the 77% recorded with acetophenone in 4 h. The reaction time profile shows that 4 h is the optimum time for substrate conversion. It is important to note that only the substrate acetophenone exhibited a decrease in conversion beyond the 4 h reaction time (Figure ). This may be due to further reaction of the product 1-phenylethanol under the catalytic conditions to yield higher cross-coupling products.[41b] Hence, under the optimized conditions, a variety of ketones were utilized as substrates, and the result is summarized in Table .

Table 2

Scope of the TH of Ketones with 2a as the Catalysta

Reaction conditions: substrates (1 mmol), KOH (20 mol %), catalyst 2a (0.4 mol %), and 2-propanol (2 mL) reflux.

Conversions determined by gas chromatography with flame ionization detector (GC-FID) based on averages of two runs that agree within ±5%.

Turnover number = mole product/mole catalyst.

Turnover frequency (h–1) = mole product/(mole catalyst × time).

Scheme 2

CTH of Ketones

Figure 3

Influence of catalyst concentration on the CTH of acetophenone. Conditions: acetophenone (2.1 mmol); catalysts 2a–c (0.1–1 mol %); (KOH, 20 mol %); 2-propanol (solvent, 2 mL, 82 °C); 4 h. Conversions determined by gas chromatography (GC) as averages of two runs.

Figure 4

Time profile for the CTH of ketones with 0.4 mol % of 2a as the catalyst.

Influence of catalyst concentration on n class="Chemical">the CTH of acetophenone. Conditions: acetophenone (2.1 mmol); catalysts 2a–c (0.1–1 mol %); (KOH, 20 mol %); 2-propanol (solvent, 2 mL, 82 °C); 4 h. Conversions determined by gas chromatography (GC) as averages of two runs.

Time profile for the Cn class="Chemical">TH of ketones with 0.4 mol % of 2a as the catalyst.

Reaction conditions: substrates (1 mmol), n class="Chemical">KOH (20 mol %), catalyst 2a (0.4 mol %), and 2-propanol (2 mL) reflux.

Conversions determined by gas chromatography win class="Chemical">th flame ionization detector (GC-FID) based on averages of two runs that agree within ±5%.

With ren class="Chemical">ference to acetophenone (entry 1) as the model substrate, entries 2–11 show the influence of structural and electronic variations on the reactivity of the C=O bond of the various ketones (entries 2–12). A wide variation in catalyst efficiency (turnover frequency, TOF) was observed up to a maximum of 1635 h–1 for 1-phenyl-2-butanone (entry 8).

The general trend is n class="Chemical">that acetophenone derivatives bearing mild deactivating halogen groups para to the C=O interacted better with the catalyst resulting in better efficiencies (entries 2–4), while the opposite effect was observed for electron-donating groups para to C=O (entries 5 and 6). These results agree with the work of Sortais and co-workers on the reduction of aryl-alkyl ketones, catalyzed by manganese complexes bearing chiral diamine ligands.[43] The results are also comparable with data from a similar work based on ruthenium complexes with 10 mol % of a base refluxed in 2-propanol for 45–180 min from which conversions between 90 and 97% were reported.[44] However, the least conversion of 17% recorded for 2-methyl-3-hexanone over 20 h (entry 12) is clearly due to steric hindrance by the neighboring flexible sp3 carbons that limit accessibility and coordination to the C=O bond. In general, this is the observation and explanation for the poor conversions of linear ketones in CTH.[18b]

Mechanistic Study of the CTH of Ketones by Complex 2a

A proposed mechanism for the n class="Chemical">CTH of ketones reported in this study is presented in Scheme . Based on the experimental data, it is proposed that the reduction of the ketones to alcohols proceeded via one-electron transfer processes. There are numerous proposals in the literature, but the route that involves the formation of a metal alkoxide 6 is the most reported and accepted pathway.[45]

Scheme 3

Proposed Mechanism for the CTH of Cyclohexanone with Complex 2a as the Catalyst

Hence, the mechanism is proposed to begin win class="Chemical">th the conversion of the precatalyst 2a to isopropanol-coordinated 3 which was possible because of the hemilabile imine N-donor. Transformation of 3 into a metal hydride allows for the insertion of the carbonyl to form 4 which releases isopropanone to yield compound 5. The metal alkoxide 6 is then formed from 5 and after ligand exchange with the solvent 2-propanol, the product is released. To complete the cycle, the solvent (which also serves as the hydrogen donor) isopropanol is coordinated to regenerate 3. It is worth noting that 3–6 were monitored and observed by liquid chromatography–mass spectrometry (LC–MS) (See the Supporting Information for details).

To shed more light on the structural changes n class="Chemical">that the catalyst undergoes during the CTH that led to the proposed mechanism (Scheme ), the catalytic process was monitored using UV–vis and IR spectroscopic techniques. The UV–vis spectrum of complex 2a dissolved in 2-propanol (Figure standard) displays two strong absorption maxima at 250 and 450 nm which are assigned to the Co–O and Co–N metal-to-ligand charge transfer bands, respectively.[28] A similar pattern was also observed with related imino Ni(II) and imino Cu(II) complexes.[46] At the beginning of the reaction, with the substrate and oxidant added, no changes in the pattern of the absorption spectrum were observed (Figure , 0 h). However, decreased intensity of the low energy band at 450 nm which was accompanied by the appearance of a broad band at 320 nm (Figure , 1 h) is attributed to an overlap of the n−π* transition because of the C=O (substrate) and C=N (ligand) absorption bands.[47] We propose this to be an indication of hemilability of the imino (Co–NC) donor during the catalytic cycle and clear sign that the phenoxy (Co–O) donor remained unchanged (absorption maxima at 250 nm) throughout the course of the reaction.[48]

Figure 5

UV–vis results showing the stability of 2a during a 6 h reaction time at 82 °C.

Furthermore, n class="Chemical">the reaction was closely monitored using infrared spectroscopy. A setup similar to that described for the UV–vis study was utilized, and the results are presented in Figure . In support of the UV-results, the IR spectra also revealed that both Co–N and Co–O absorption bands remained unchanged in the region 486 and 426 cm–1, respectively.[25,49] This observation confirms the structural stability of complex 2a during the course of the reaction. Therefore, the combination of the cyclic voltammetry (CV), UV–vis, FTIR, and low-resolution MS (LRMS) results suggest a one-electron-based mechanism as the pathway for the CTH processes.

Figure 6

IR Spectra showing the stability of 2a during a 6 h reaction time at 82 °C.

Conclusions

In summary, n class="Chemical">three new Co–NHC complexes, 2a, 2b, and 2c, were synthesized and used as efficient catalysts for the TH of a variety of ketones. The complexes were fully characterized using both analytical and spectroscopic techniques. Single-crystal X-ray diffraction (XRD) analysis of compounds 2a and 2c determined the geometry of the complexes, with each complex having a hexacoordinated Co(III) at the center of an octahedral array of ligands. The results for the catalytic studies confirm that the Co–NHC complexes have good potential for use as catalysts in the TH of ketones. The results showed that the complex bearing a small N–Me wingtip substituent is relatively more active compared to those with bulkier N-substituents. On-stream spectroscopic analyses confirmed catalyst stability during the course of the CTH reaction; hence, the observed drop in catalytic activity at higher complex concentrations is due to agglomeration of active species in solution.

Experimental Section

General Information

All reactions except where mentioned otherwise were performed using standn class="Chemical">ard Schleck techniques under an inert atmosphere. All solvents were dried and purified using standard procedures before use. Glassware was washed and dried in an oven at 120 °C. 1H and 13C NMR spectra were measured on a Bruker AVANCE-III 400 MHz spectrometer at ambient temperature with tetramethylsilane (at 0.00 ppm) as an internal standard. All chemical shifts are quoted in δ (ppm) and coupling constants in hertz (Hz). Abbreviations used for the multiplicity of the NMR signals are s = singlet and m = multiplet. Infrared spectra were recorded on a PerkinElmer universal attenuated total reflection (ATR) spectrum 100 FT-IR spectrometers. Mass spectrometry and elemental analysis were recorded on a Waters Micromass LCT Premier TOF MS–ES+ and ThermoScientific Flash2000 Elemental Analyser, respectively. Thin-layer chromatography was carried out on Macherey-Nagel POLYGRAM SIL/G/UV254 precoated plates. Melting points (mp) were recorded using an Electrothermal 9100 melting point apparatus. All other chemicals were purchased from Sigma-Aldrich and used without further purification.

All the n class="Chemical">three complexes 2a–c were synthesized following the same general procedure described below for the formation of 2a (Section ).

Bis-2-([2-(3-methyl-2,3-dihydro-imidazole-1-yl)-ethylimino]-methyl)phenoxy Cobalt(III) Bromide 2a

Into a clean Schleck tube containing 1a (1 mmol, 0.31 g) was added n class="Chemical">methanol (10 mL) followed by KOBu (3 equiv). The mixture was allowed to stir at room temperature for 15 min. Thereafter, CoBr2 (0.5 mmol, 0.11 g) was slowly added to the mixture and allowed to stir for another 4 h at room temperature. After the completion of the reaction, the mixture was filtered over a bed of Celite and all volatiles were removed under reduced pressure, and the residues were washed 3 times with diethyl ether to afford the complex as a reddish-brown air-stable powder. Yield: 0.21 g, 71%, mp 182 °C (decomposed), IR (ATR cm–1): 3389, (OH, H2O), 3077, (CH, sp3), 1616, (C=N), 1537, (C=C): HRMS (ESI): [M+ – Br–] calcd for C26H28 CoN6O2, 515.1606; found, 515.1613. 1H NMR (400 MHz, CDCl3): 3.66 (5H, m, CH3, CH2–N), 3.95 (2H, t, N–CH2), 6.46 (1H, s, imi), 7.05 (2H, m, Ar), 7.26 (3H, m, Ar), 7.70 (1H, s, N=CH−). CHN Anal. Calcd for [C26H28CoN6O2]·Br, 1.3H2O; C, 50.47; H, 4.98; N, 13.58. Found: C, 50.33; H, 5.10; N, 13.84.

Bis-2-([2-(3-ethyl-2,3-dihydro-imidazole-1-yl)-ethylimino]-methyl)phenoxy Cobalt(III) Bromide 2b

Reddish-brown air-stable powder. Yield: 0.30 g, 96%, mp 177 °C (den class="Chemical">composed), IR (ATR cm–1): 3389, (OH, H2O), 3156, (CH, sp3), 1596, (C=N), 1450, (C=C): HRMS (ESI): [M+ – Br–] calcd for C28H32CoN6O2, 543.1919; found, 543.1906. CHN Anal. Calcd for [C28H32CoN6O2]·Br: 0.3H2O; C, 53.94; H, 5.17; N, 13.48. Found: C, 54.23; H, 4.90; N, 13.19.

Bis-2-([2-(3-benzyl-2,3-dihydro-imidazole-1-yl)-ethylimino]-methyl)phenoxy Cobalt(III) Bromide 2c

Maroon powder. Yield: 0.32 g, 86%, mp 120 °C, (den class="Chemical">composed). IR (ATR cm–1): 3360, (OH, H2O), 2970, (CH, sp3), 1615, (C=N), 1447, (C=C), HRMS (ESI): [M+ – Br–] calcd for C38H36CoN6O2, 667.2232; found, 667.2224. CHN Anal. Calcd for [C38H36CoN6O2]·Br: 0.7H2O; C, 60.04; H, 4.96; N, 11.06. Found: C, 60.21; H, 5.11; N, 11.21.

Procedure for the Catalytic Studies

Samples for catalytic studies were prepared as follows: a substrate (n class="Chemical">ketone) was placed into a clean Schleck tube fitted with a reflux condenser and a stir bar, followed by the addition of the cobalt complex (2a–c, 0.4 mol %) and KOH (20 mol % in 2 mL of 2-propanol). The mixture was then refluxed at 82 °C for 4 h. Conversions were monitored using GC-FID. An aliquot was taken at a predetermined time and passed through a pad of cotton wool and then injected (1 μL) into the GC equipped with a DB5 wax polyethylene column (30 m × 0.25 mm). A comparison of the observed retention times with those of standards purchased from Sigma-Aldrich identified the products. Percentage conversions were calculated from the respective peak areas.

Procedure for CV

Cyclic voltammograms were measured in acetonitrile, using a Metrohm 797 potentiostat and n class="Chemical">three-electrode system, consisting of glassy carbon as a working electrode, a platinum wire as the reference electrode, and a Ag/AgCl system as a counter electrode. A 0.1 M solution of NBu4PF6 was used as a supporting electrolyte. A blank solution of acetonitrile and the solution of the electrolyte were run for a background check. All solutions were purged and maintained under an inert atmosphere of N2 gas during the experiment. A scan rate of 100, 250, and 500 mV/s was measured in a potential window of 0.3 to −1.7 V.[50]

Procedure for Structural Determination by Single-Crystal XRD Data Analysis

Quality single crystals of compounds 2a and 2c suitable for XRD analysis were selected and attached to a Mitegen loop and centered in n class="Chemical">the X-ray beam by the aid of a video camera. Intensity data were collected on a Bruker APEXII diffractometer with Mo Kα radiation (λ = 0.71073 Å) equipped with an Oxford Cryostream low-temperature apparatus operating at 100(1) K. The initial cell matrix was determined from three series of scans consisting of twelve frames collected at intervals of 0.5° in a 6° range with the exposure time of 10 s per frame. Each of the three series of scans was collected at different starting angles, and the APEX2[51] program suite was used to index the reflections and refined using SAINT.[52] Data reduction was performed using SAINT software, and the scaling and absorption corrections were applied using the SADABS[53] multiscan technique. The structures were solved by direct methods using SHELXS.[54] Non-hydrogen atoms were first refined isotropically and then by anisotropic refinement with full-matrix least-squares based on F2 using SHELXL.[54] All hydrogens were positioned geometrically, allowed to ride on their parent atoms and refined isotropically.