|Home | About | Journals | Submit | Contact Us | Français|
We report a rapid method for assembling our di-μ-oxo dimanganese catalyst, verified by ESI-MS and EPR, assessing its water oxidation activity by a Clark electrode O2-assay study and its regioselective C–H activation activity by product analysis in catalytic runs.
Synthesis and isolation of suitable high valent catalysts as well as finding a suitable catalyst-substrate combination have presented major challenges in research directed toward biomimetic regioselective C–H bond oxygenation.[1–3] In the absence of molecular recognition, a typical catalyst brings about oxygenation of a substrate at the most reactive site.[4–9] While this approach is of significance, it is of limited use for bringing about reaction at a less reactive site. Chelate control has been used in a number of cases, but this causes activation only at a neighbouring site.[10–12] The use of a molecular recognition unit at a remote location on a ligand to bind the substrate prior to reaction can provide high selectivity. When the target reaction is the oxygenation of a saturated C–H bond, this presents several complications in the design of ligand. As most of the bonds are more oxidation prone than saturated C–H bonds, all the groups in the ligand have to be highly oxidation resistant. Moreover, once such a ligand has been designed and synthesized, the isolation of a high valent metal catalyst bearing the ligand is time-consuming.
The di-μ-oxo dimanganese [MnIII(μ-O)2MnIV] core has been proved to be an exceptionally active oxidation catalyst. But isolation of an intact [LMnIII(μ-O)2MnIVL]X3 (X: anion) complex with a ligand L has proved to be tedious and has only been achieved with limited success. Other than the complex with the molecular recognition ligand, LMR, shown in Scheme 1, high valent di-μ-oxo dimanganese complexes, even those bearing much simpler ligands, were only isolated in low yield.[13, 14] With change of ligand, the solubility, stability, selectivity and activity of [LMn(μ-O)2MnL]3+ change in an unpredictable fashion.
Recent studies by our group indicate that [LMRMnIII(μ-O)2MnIVLMR]3+ can function as an efficient catalyst for oxygenation of saturated C–H bonds with high regio-(>98%) and stereoselectivity (>99%) and with hundreds of turnovers (total turnover >700). Our molecular recognition ligand, LMR, can lead to selective oxygenation at a remote unactivated site. We found that for regioselective catalysis, steric exclusion of unbound substrate by bound substrate is an important requirement. This prevents the unselective reaction of unbound substrate molecules at the active site and puts another constraint on the design.
These difficulties have led us to look for an easier method to screen ligands. In the present study, we have used ligand substitution to prepare the catalyst in situ (Scheme 1). This in-situ catalyst has an activity comparable to the isolated catalyst with the molecular recognition ligand (LMR) suggesting that this strategy can be used as a fast screening method for catalysts that are otherwise difficult to synthesize. Once intial data have been obtained in a screen of this type, the successful system can be isolated and fully characterized for definitive studies.
Ibuprofen, tetrabutyl ammonium oxone (used as primary oxidant), dipy, HPLC grade acetonitrile were bought from Sigma-Aldrich and were used without further purification. [(dipy)2Mn(O)2Mn(dipy)2](ClO4)3 was synthesized according to literature procedure.
1H and 13C NMR spectra were recorded in deuterated solvents (Cambridge Isotope Laboratories) at 25°C on a Bruker 400 or Bruker 500 spectrometer at the Yale Department of Chemistry Instrument Center. Electrospray ionization-mass spectroscopy (ESI-MS) was done on a Waters ZQ LC-MS instrument. Gas chromatography-mass spectroscopy (GC-MS) was done on an Agilent 5973 GC-MS instrument.
To a solution of 7 mM (1 eq) of S1 in 20 ml acetonitrile, 0.03 eq of Cinact and 0.06 eq of LMR or terpy were added. The solution was cooled in an ice bath and 5 ml CH3CN solution of tetrabutyl ammonium Oxone®, TBAO (5 mmol) was added and the solution was stirred at room temperature, 0°C and −20°C. At room temperature, after 6 hours the reaction was quenched by addition of excess aqueous NaHSO3, acidified with 0.1N HCl solution and the products and unreacted substrate were extracted in ether. The ether extract was evaporated under vacuum and the residue was dried in vacuo in the presence of P2O5. The products were analyzed by 1H-NMR and GC-MS.
[(dipy)2Mn(O)2Mn(dipy)2](ClO4)3 (Cinact) (dipy = 2,2′-dipyridyl) has a high valent di-μ-oxo dimanganese site. This complex has been found to be almost inactive catalytically because it lacks exchangeable terminal water ligands, so we do not expect a background oxidation from it. However, the Mn-dipy bonds in the Mn(III/IV) dimer are quite labile, allowing the bidentate dipy to be easily replaced by the tridentate LMR to form the desired active catalyst (Cin-situ) in situ.
To an acetonitrile solution of Cinact, LMR is added Cinact:LMR in 1:2 molar ratio. After 5 minutes, the ESI-MS spectra of this solution gave a peak at m/z = 1042 with isotopic peaks closely matching the simulated isotopic distribution of [(LMR)MnIIIMnIV(O)2(dipy) (ClO4)2]+ showing the formation of the desired high valent dimanganese complex with LMR as a ligand (Figure 1). An EPR spectrum of the same solution gives the 16 line spectrum characteristic of the MnIII(μ-O)2MnIV core (Figure 2). To demonstrate successful formation of an active catalyst, we used oxygen evolution from water oxidation with oxone as primary oxidant as an assay. While Cinact does not evolve O2 from H2O with oxone®, the Cinact+ LMR combination does indeed do so according to oxygen assay studies with Clark electrode (Figure 3). We therefore propose that the Cinact+ LMR combination generates an active catalyst, Cin-situ, in which two dipy ligands have been replaced by LMR.
Molecular models of a Mn(μ-O)2Mn complex with LMR as the ligand were constructed by importing crystal structure parameters of ligand LMR and the Mn(μ-O)2Mn core followed by energy minimization (MM2, CAChe 5). The crystal structure of LMR shows a ~32° angle between the -COOH and the plane of the imide group. Thus, an sp3 carbon center is required α to the -COOH group to bring the remote C–H bond within ~4–5 Å distance from the metal, a distance necessary for successful attack to take place by the high-valent metal site (Figure 4).[2, 15, 16]
Ibuprofen [2-(4-isobutyl-phenyl)-propionic acid] (S1) was selected as the most suitable substrate. Ibuprofen is a rigid substrate with at least two alternate sites of attack: the benzylic C–H bonds remote from the –COOH group and the –COOH group itself. But, with molecular recognition, we expected selective oxidation at the remote benzylic C–H bond with Cin-situ as the catalyst, as shown in Figure 4, giving P1 as the major product, any initial –CH(OH) intermediate being rapidly oxidized to the ketone. Oxidation at the –COOH site is expected to give P1′ via oxidative decarboxylation under reaction and workup conditions (Scheme 2).[17–21] Without the molecular recognition both P1 and P1′ are expected to form in comparable amount.
With -COOH…HOOC- H-bonding as the source of the recognition, it was necessary to use a non-protic solvent. Acetonitrile is not only non-protic but also relatively oxidation resistant. For a typical run, the ratio of substrate:Cinact:LMR:TBAO was set at 100:3:6:500 and the reaction was carried out at room temperature, 0 °C and −20 °C (See Experimental section).
Three control experiments were run under identical conditions i) with added terpy (instead of LMR) to solutions of Cinact (terpy: [2,2';6',2'']terpyridine), ii) with used Cinact alone and iii) with the manganese catalyst omitted.
Products of ibuprofen oxidation were analyzed by 1H NMR (CDCl3) (Figure 5) and ESI-MS. As the carboxylic acids were found to be inappropriate for GC-MS, their methyl esters were prepared by addition of trimethylsilyldiazomethane in a methanol-benzene (2:7) solution of the residue. Product P1 was isolated as a pure compound by reversed phase HPLC using a C8 column. 1H NMR (Figure 6) and ESI-MS (Figure 7) were used to identify the products unequivocally by comparison with the 1H NMR of P1 and GC-MS of both P1 and P1′, which are available in the literature.[15, 17, 23]
We found that in the absence of manganese catalyst no oxidation takes place. With Cin-situ, P1 is always formed in a higher percentage than in any of the control experiments. This is consistent with our molecular model with ibuprofen H-bonded to the proposed active complex (Figure 4). On lowering of temperature, the selectivity was enhanced, presumably due to the entropic factor with ~96.5% selectivity for P1 at −20°C (Table 1). Turnovers of 20–27 were recorded with Cin-situ catalyst combination indicating the reversibility in substrate-catalyst binding. In our control experiments with ‘Cinact+terpy’, where the ligand -COOH group is absent, P1′ was always obtained in appreciable amounts and the P1:P1′ ratio remained practically unchanged at all temperatures.
We attribute the high regioselectivity in oxidation of ibuprofen by Cin-situ to molecular recognition via H-bonding to the -COOH group in LMR, the ligand in Cin-situ. To test of the effect of H-bonding, experiments were performed by adding acetic acid to the catalytic solutions to disrupt the potential H-bonding leading to molecular recognition. With 4 equivs. acetic acid in the medium for each equiv. of substrate in the Cin-situ system, the P1:P1′ ratio fell from ~7:1 to ~3:1 on adding acetic acid. Control experiments with ‘Cinact+terpy’ catalytic system gave the same 3:1 product mixture under identical conditions (Table 1). This is consistent with the proposed role of COOH•••HOOC- H-bonding for high regioselectivity in the oxygenation of ibuprofen.
We were concerned about the potential degradation of P1′ leading to an apparent lower yield of P1′ versus P1, with the Cin-situ catalytic system. A time course study was carried out to test this point. The Cin-situ catalytic system at 0°C takes 24 hours to reach completion. We, therefore, began our product analyses at 0.5 hours followed by a series of measurements over 10 hours. The P1:P1′ ratio remained essentially constant throughout. A similar study with the control in-situ catalyst (‘Cinact+terpy’) also showed a constant P1:P1′ ratio throughout. This suggests that the high P1:P1′ ratio with the Cin-situ catalytic system is, indeed, due to molecular recognition.
We report a rapid screening method for testing ligands with a di-μ-oxo dimanganese catalyst. The activity of the catalysts formed by displacement of dipy by terpy or LMR was confirmed by a Clark electrode O2-assay study (water oxidation activity) and the ligand displacement reaction itself was detected by an ESI-MS study. This indicated the formation of a new complex, Cin-situ, by replacement of dipy by LMR. With Cin-situ, oxidation of ibuprofen (S1) occurs with high selectivity (~97% at −20°C) at the benzylic C–H bond remote from the -COOH group, with 24 turnovers. This selectivity is attributed to anchoring and orientation of the substrate (S1) via -COOH•••HOOC- H-bonding between substrate -COOH group and the –COOH group of LMR. Loss of selectivity after disruption of the H-bonding to the substrate with AcOH supports the proposed origin of the selectivity. The isolated di-μ-oxo dimanganese catalyst with LMR as the ligand gave higher selectivity (>98%) and much higher turnovers (>500) so it is necessary to isolate the complexes to gain maximal activity. Nonetheless, displacement studies may be useful in comparing a range of ligands for catalysis.
This work was supported by National Science Foundation (CHE-0614403) and NIH (GM32715)