Chiral Catalysis

Much of organic chemistry is concerned with the generation of chiral natural and non-natural compounds. The synthesis of these molecules often requires the introduction of at least one chiral center, and this can be achieved by asymmetric transformations. What is even more valuable, are reagents that can produce a specific functional group array that can be used to direct the introduction of additional chiral centers into a molecule. An example of this type of functional group is the chiral allylic alcohol. Not only is this functionality an important end product, it also serves as a substrate for many important reactions including asymmetric epoxidations, cuprate S_N2´ displacements, Claisen, Wittig, and sigmatropic rearrangements, and Diels-Alder reactions.

Because the tools to perform such transmutations are not always available, the synthesis of many interesting and relevant molecules either cannot be realized or is highly inefficient. Therefore, the invention of selective, chiral reagents has become one of the most important facets of synthetic chemistry. This project proposes the design, synthesis, and evaluation of just such a catalyst. In order to create a catalyst that will recognize and react with only one type of functional group array, it is necessary to explore all facets of the catalyst-substrate interaction.

Catalyst design requires an intimate knowledge of the mechanism of interaction between the proposed catalyst and the substrate. The proposed catalyst 10 (above) will be a palladium(0) species with a chiral bidentate phosphorus ligand (not unlike DIPHOS), where the hydride transfer reagent will reside. The hydride source is an adduct of a pyridine derivative with borane. While many reagents utilize steric bulk to effect enantioselective reduction, the key interaction for this catalyst will be a palladium(0) h²-olefin complex with the enone. Once the chiral palladium species has docked in the appropriate orientation, selective 1,2-hydride transfer from the ligand to one face of the enone will be facile.

Issues to be addressed in predicting the selectivity of catalyst 10 with the enone substrate include facial and conformational selectivity, strength of complex formation, and hydride transfer orientation. First, there are several h-complexes between the palladium species and an a,b-unsaturated ketone that can be imagined: one h⁴- and two h²-complexes (either through the olefin or the carbonyl group). Only structures corresponding to the h²-olefin complex were found from a search of the Cambridge Crystal Structure Data Bank. The lowest energy interaction for h-complexation occurs when the Pd, P, and olefin C atoms all lie in the same plane. This allows for the two components for binding to be optimized: the first is the olefin p orbital overlap with an s-type acceptor orbital on the metal; the second is a back-bond resulting from overlap between the filled Pd d orbital and the p* orbital of the olefin. The two possible low energy orientations catalyst 10 can have with a conformationally locked s-cis enone are depicted above. The complex on the left will lead to reduced product, because only this orientation has the carbonyl carbon and borane hydrogen within bonding distance. The other complex will undoubtedly form, but it will be a non-productive event and will subsequently dissociate. As the catalyst contains an axis of symmetry, there are two identical complexes that will lead to the desired reduction reaction.

Conformationally flexible a,b-unsaturated ketones allow for an additional interaction to occur; this is the result of s-cis to s-trans isomerization of the enone. Both of these interactions can now lead to product formation (inset, right). Because these complexes present opposite faces to the catalyst, the formation of a racemic mixture of allylic alcohols in the ratio dictated by the equilibrium constant for isomerization will result.

Enantioselectivity could be restored if the formation of one of the complexes pictured above was favored over that of the other. This preference could be manifested by the addition of steric bulk on the phenyl group adjacent to the pyridine complex; the conformation of the enone which minimizes steric interactions in the complex should be favored. The inset shows a derivative catalyst 11 (which contains a 2,6-disubstitution pattern of isopropyl groups on the phenyl rings) complexed to an s-cis and s-trans enone. Even when the R-group in the figure is as sterically undemanding as a methyl group, the complex most likely to form would be that shown on the right, where the enone has adopted the s-trans conformation. The bulky substituents serve to decrease the interaction energy for the s-cis enone in the Figure; this drives the equilibrium to the right, and reestablishes facial selectivity. The synthesis of catalyst 11 will be pursued after the complete synthesis of catalyst 10.

The strength of the catalyst-substrate interaction is an important issue; if the complex is not formed, then the desired reaction will not occur. Formation constants for h²-olefin complexes have been measured. One system (involving an a,b-unsaturated ketone) obtained a value of -3.22 kcal/mol at 33^oC for the complex free energy of association. At the lower temperature of this experiment (-78^oC) this number would be expected to increase by a factor of two or more. This indicates that the reactive complex will indeed form. The product of the reaction, the allylic alcohol, could also complex with the catalyst; however, this interaction should be quite weak (h-complexes between low valent metals and electron-rich double bonds are notoriously weak), and, therefore, not compete extensively for the catalyst. Inspection of molecular models indicates that catalyst 10 can adopt an almost perfect Bürgi-Dunitz trajectory between the borane hydrogen and carbonyl carbon when bound to the substrate. This transition state should have a reduced activation energy causing a concomitant acceleration of the reduction reaction. To make the system truly catalytic (10 mol percent reagent or less), then the catalyst must be regenerated. This can be done by using an excess of catecholborane, which will readily reduce the borane complex of the reagent, but will not react substantially with the substrate. All of these issues are addressed by the catalyst-enone complex illustrated above and below; this structure is the lowest energy conformation as determined by molecular modeling. Several parameters, namely, bond lengths and angles for which force field parameters were not available, have been fixed. Table 1 contains some key structural data that were determined by crystal structures of related complexes. These values were used to assure proper placement of unparameterized atoms after minimization of the various complexes was completed. Since the enone is constrained to the s-cis conformation, only the complex orientation given can react. The distance between the borane hydrogen and carbonyl carbon is about 1.2 Å, and the angle formed by borane hydrogen-carbonyl carbon-carbonyl oxygen is approximately 100^o.

The last two entries in Table 1 refer to the optimum parameters for a Lewis acid interaction with the carbonyl oxygen of the enone. In order for hydride transfer to occur, a Lewis acid, like trimethyl borane or boron trifluoride, must be present in solution to activate the carbon-oxygen double bond. The figure above (and to the right of the Table) shows the complex formed between the catalyst 10 and 2-cyclopentenone as predicted by Spartan Pro and other molecular modeling tools.

The synthesis of catalyst 10 will be highly efficient. Initially, the asymmetric phosphorus moiety will be synthesized and resolved (compound 5a, Scheme 1). Treatment of dichloro-phenylphosphine with methanol and pyridine will produce dimethylphenylphosphonite. Reaction of 2 with methyl iodide will produce methyl methylphenylphosphinate (3). The action of 3 with phosphorus pentachloride will produce methylphenylphosphinyl chloride (4). Finally, treatment of 4 with (-)-menthol will produce the diastereomeric menthyl methylphenylphosphinates (5a and 5b). The diastereomers are separated by differential recrystallization from hexane to isolate 5a. Initially, only diastereomer 5a will be synthesized and resolved and carried throughout the synthesis to afford one enantiomer of the final catalyst 10.

Scheme 1.

With resolved compound 5a in hand, the chiral auxiliary can now be removed (Scheme 2). Lithium-halogen exchange will effect the transformation of 5a into (R)-methylphenylpyridylphosphine oxide (6) with inversion of chirality at phosphorus. Copper-mediated coupling of two equivalents of compound 6 will result in the production of (R,R)-1,2-bis(phenylpyridylphosphinoyl)ethane (7). Reduction of the phosphine oxides with thrichlorosilane will produce (R,R)-1,2-bis(phenylpyridylphosphinyl)ethane (8). Reaction of PdCl₂ with 8 and subsequent reduction with n-butyl lithium will produce the solvated palladium complex 9. Finally, the solvated, functional catalyst 10 is made by the reaction of diborane in THF with complex 9.

Scheme 2.

As described above, catalyst 10 promises to be a highly selective chiral reducing reagent for a,b-unsaturated ketones. Because a possible mechanism of action has been proposed and explored, highly predictable absolute stereochemistry should be expected. The catalyst will be synthetically available in either enantiomeric form, offering a distinct advantage over many previously reported catalysts. The catalyst can also be recovered after use. Some potential limitations can be anticipated with this system. Low valent palladium is notoriously reactive toward oxidative insertion. Therefore, organic species with aryl halogen bonds will not be viable substrates. Additionally, since inorganic acids can react with the palladium(0) species, these reagents should be avoided.

Both enantiomeric forms of 10 will be synthesized and characterized; however, initially, only the one enatiomeric form will be produced to test the hypothesis. The determination of the absolute configuration of each enantiomer will be resolved by x-ray analysis of the complex formed between 10 and (-)-dimenthyl maleate. Once the absolute stereochemistry of the catalyst has been determined, this information will be used to predict the enantioselectivity anticipated when any s-cis or s-trans constrained enone is used as the substrate. The selective reducing ability of catalyst 10 will be obtained for endocyclic and exocyclic enones containing various ring sizes and substitution patterns. Acyclic isomerizable enones will be subsequently tested to determine the conformational discrimination of this catalyst. If no enantiomeric preference is exhibited, then a second catalyst 11, with steric bulk on the neighboring phenyl rings of the ligand, will be synthesized to potentially restore enantioselectivity. Catalyst 11 will also be tested on rigid enones to see what affect, if any, steric bulk has on catalyst reactivity.

The active catalyst 10 will be freshly prepared for each reaction. A typical procedure will be the following: one equivalent each of the bidentate phosphorus ligand 8 and PdCl₂ are dissolved in dry, distilled THF under an inert atmosphere. The solution is chilled to -78^oC and stirred to allow for ligation to occur. The resulting Pd(II) complex is reduced with n-butyl lithium. The active catalyst is formed by the addition of a diborane-THF solution to the Pd(0) complex. For stoichiometric reduction, an equimolar quantity of the enone and trimethyl borane in a THF solution are added to the solution containing the reagent 10. For catalytic reaction, the procedure is changed by adding catecholborane to the catalyst solution before addition of the substrate. After reduction is complete, the reaction mixture is poured over ice water and acidified with HCl. Simple extraction of the acidic aqueous solution, followed by purification techniques yields the enantiomerically pure allylic alcohol. The aqueous layer, which contains the phosphorus ligand oxidized to the phosphine oxide, is concentrated by lyophilization. The resulting solid is suspended in CH₃CN, reduced with HSiCl₃ (as shown in Scheme 2), and filtered to remove inorganic solids to yield the regenerated chiral ligand 8.