|Home | About | Journals | Submit | Contact Us | Français|
Palladium poprocks: Hold on to your CO2! Enantioselective Pd-catalyzed decarboxylative alkylation of ketone enolates proceeds via η1-σ-allyl Pd-carboxylate complexes by slow loss of CO2.
We recently developed a series of catalytic enantioselective allylic alkylation reactions of cyclic ketone enolates that proceed by decarboxylation of allyl carbonates and β-ketoesters (Scheme 1). These robust reactions proceed in the presence of a wide array of functionality and steric hindrance, in a variety of solvents, and have an unusually high tolerance to water. To gain further experimental insight into this chemistry, we embarked on a mechanistic study that has now resulted in the isolation and full characterization of complex 1, the resting state of a prototypical reaction. In addition to describing the identification of this unusual complex, we discuss its potential implications for Pd-catalyzed decarboxylative asymmetric alkylation reactions and other related transformations.
Initially, we sought to follow the catalytic reaction of a standard β-ketoester substrate, (±)-2, by 31P NMR. Combining (S)-t-BuPHOX (3) and Pd2(dba)3 in a 2.6: 1 ratio at room temperature for 30 min as specified in our standard alkylation procedure, yielded a single new resonance at 18.8 ppm along with the signal for free ligand (3) at −5.95 ppm (Figure 2, A). The addition of β-ketoester (±)-2 resulted in the complete disappearance of the resonance at 18.8 ppm and produced a long-lived resonance at 30.9 ppm (Figure 2, B). As the reaction forming ketone 4 neared completion, the long-lived intermediate slowly reverted to the original resonance at 18.8 ppm (Figure 2, C).
We proceeded to isolate and characterize the complex corresponding to the long-lived resonance at 30.9 ppm and identified it as 1 (Figure 1). Despite the apparent abundance of this complex in solution under the catalytic reaction conditions, 1 proved challenging to isolate due to air sensitivity and thermal instability well below 24 °C both in solution and as a solvent-free solid. Interestingly, impure samples of 1 visibly expel a gas (presumably CO2) in the solid state and effervesce in solvent. After extensive experimentation, crystals of high purity and reasonable stability were isolated and crystallographically characterized as a mixture of diastereomers resulting from the use of racemic β-ketoester 2. Complex 1 is a square planar 16-electron species with a σ-bound, η1-allyl ligand trans to nitrogen and a β-ketocarboxylate ligand trans to phosphorous. By contrast, the analogous Pd(PHOX)allyl·PF6 cationic complex displays η3-π-allyl bonding in the solid state[3,6,7] as do related structures in solution.
The structure of intermediate 1 reveals that it must form after oxidative addition, but prior to decarboxylation. Since 1 is the only observable species during the course of the catalytic reaction, it represents the catalyst resting state and suggests that the rate-determining step for the allylic alkylation of β-ketoester substrates is decarboxylation. This is consistent with our previously reported kinetic data for the overall reaction that shows a first-order dependence on catalyst concentration and an apparent zero-order dependence on the substrate concentration.
The structure of palladium complex 1 is noteworthy. The arrangement of the allyl and carboxylate ligands in intermediate 1 is similar to that of a structure previously calculated by DFT methods for a palladium allyl enolate complex (6). Importantly, enolate 6 was described as the penultimate intermediate prior to the key carbon–carbon bond-forming reductive elimination step in the calculated inner-sphere mechanistic pathway (Figure 3). The high degree of structural similarity between intermediate 1 and complex 6 lends credence to this key calculated structure, and may also represent the best isolable experimental model system for the calculated intermediate.
The crystal structure of intermediate 1 is not only the first example in the CSD consisting of a palladium species with a β-ketocarboxylate ligand, but it is also the first X-ray structure of any transition metal complex with a σ-bound allyl cis to a carboxylate ligand. Given the enormous number of transition metal-catalyzed reactions involving allylic acetates and carbonates, this structure may have important mechanistic implications not only for the immediately related decarboxylative allylic alkylation reactions,[1,10] but also for palladium catalyzed allylic oxidations,[11,12] palladium catalyzed 1,4-diacetoxylation,[11,13] and late transition metal-catalyzed decarboxylative reactions in general. With this in mind, we synthesized and crystallographically characterized the neutral η1-allyl(PHOX)Pd(OAc) (7, Figure 4). Similar to carboxylate 1, complex 7 is a 4-coordinate Pd(II) square planar species bearing a trans relationship of the phosphine and acetate. Despite its simplicity and central nature to many catalytic pathways, this canonical oxidative addition adduct of allyl acetate and LnPd(0) has not been previously characterized. In light of the availability of η1-allyl complexes 1 and 7, it may be reasonable to consider analogous neutral intermediates in a variety of studies.
With the catalyst resting state of the catalytic cycle identified, we studied the general reactivity of complex 1. After 40 min in anhydrous THF at 24 °C, samples of intermediate 1 convert to the allylic alkylation product 4, in 89% ee, which is consistent with the enantioselectivity observed in catalytic reactions conducted under our standard conditions (Scheme 2). Allowing this thermal decomposition to occur in the presence of free dba ligand generates complex 5 (Scheme 2). Phosphorous-31 NMR indicated that this complex (5) was identical to the phosphorus-containing compound previously observed in the 31P NMR study described above (Figure 2, A and C). Additionally, complex 5 can be formed independently by mixing Pd2(dba)3 with (S)-t-BuPHOX (3), suggesting that complex 5 is the initial adduct in our, as well as other related, allylic alkylation systems,[10a,b,e] and is the predominant palladium-containing species in the absence of substrate.
We proceeded to isolate and characterize complex 5, which was assigned by X-ray analysis to be monomeric Pd(PHOX)(dba) (Figure 5). Isolated 5 is a competent catalyst for the asymmetric alkylation reaction providing yields and enantiomeric induction analogous to that of our previously published procedures.[1,6]
A mechanistic picture for the transformation (±)-2 → (S)-4 begins to emerge with these results in hand (Scheme 3). Initial formation of a Pd(PHOX)(dba) (5) precedes substrate coordination and oxidative addition to form carboxylate 1, the resting state of the catalytic cycle. Turnover-limiting decarboxylation produces enolate 6, which undergoes rapid C–C bond forming and enantiodetermining reductive elimination to form (S)-4 and a palladium species capable of engaging further substrate to continue the cycle. Alternatively, upon consumption of substrate (±)-2, complexation with dba produces complex 5 again.
In summary, we have experimentally detailed the mechanism of the early stages of our enantioselective allylic alkylation reaction. We have isolated and characterized all of the 31P NMR-observable intermediates from the catalytic alkylation reaction of β-ketoester (±)-2, including cis-η1-σ-allyl intermediate 1. We believe this unique structure (1) and the acetate counterpart (7) are of potential mechanistic significance to a broad range of reactions. Experimental mechanistic investigations into the decarboxylation, bond-forming, and stereoselective steps of our allylic alkylation reaction are currently underway.
**The authors wish to thank NIH-NIGMS (R01 GM 080269-01), Abbott Laboratories, Amgen, Merck, Bristol-Myers Squibb, Boehringer Ingelheim, the Fannie and John Hertz Foundation (predoctoral fellowship to DCB), the Gordon and Betty Moore Foundation, and Caltech for financial support. Mr. Lawrence Henling and Dr. Michael Day are gratefully acknowledged for X-ray crystallographic structural determination. The Bruker KAPPA APEXII X-ray diffractometer was purchased via an NSF CRIF:MU award to the California Institute of Technology, CHE-0639094. Professors J. E. Bercaw, R. H. Grubbs, S. E. Reisman, W. A. Goddard III, and Dr. J. A. Keith are acknowledged for helpful discussions. Dr. David Vander Velde and Dr. Scott Ross are thanked for helpful assistance related to NMR.