Chiral lithium amides bulk bases

The finding that the use of LDA as bulk base results in non-enantioselective deprotonation indicated that bulk bases which are much less reactive toward the epoxide substrate compared with the chiral lithium amide are needed. But they should be strong enough to regenerate the chiral amide from the amine formed in the epoxide rearrangement. 

In order to further develop the field of enantioselective catalytic deprotonation, it was necessary to develop bulk bases that show low reactivity toward the epoxide but have the ability to regenerate the chiral catalyst. Thus, the bulk bases should show low kinetic basicity toward the substrate, but be thermodynamically and kinetically basic enough to be able to regenerate the chiral lithium amide from the amine produced in the rearrangement. 

Ahlberg and coworkers have found that lithiated 1-methylimidazole (21) and lithiated 1,2-dimethylimidazole (22) work as such bulk bases in the presence of catalytic amounts of a readily accessible homochiral lithium amide 20 (both enantiomers are readily available) (see Section III.C)45,46. These new bulk bases are easily accessible by deprotonation of 1-methylimidazole and 1,2-dimethylimidazole by, e.g., n-BuLi (Scheme 72). Using chiral lithium amide 20 (20 mol%) and bulk base 21 or 22 (200 mol%) in the deprotonation of cyclohexene oxide 1 gave (S)-2 with the same enantiomeric excess (93%) as under stoichiometric conditions (Scheme 15). Apparently, any background reactions of the bulk bases are insignificant. Interestingly, no addition of DBU was needed to obtain the high enantioselectivities under these catalytic conditions. 

Ahlberg and coworkers noted that in some cases the enantioselectivity was increased when running the deprotonations with equimolar amounts of the novel bulk bases and the chiral lithium amide113. This finding initiated a detailed mechanistic investigation using isotopically labeled compounds and multinuclear NMR spectroscopy and kinetics, to elucidate the nature of the reagents and transition states in the deprotonations. They discovered that mixed dimers 23 and 24 are formed in solution from monomers of chiral lithium amide 20 and bulk base 21 and 22, respectively (Scheme 73). 

Deprotonation of 4-f-butyl cyclohexanone 28 with chiral lithium amide 39 (30 mol%) and bulk base 107 (240 mol%) in the presence of HMPA (240 mol%) and DABCO (150 mol%), under external quench conditions, resulted in 79% ee of the silyl enol ether 29 (Scheme 79)121. This stereoselectivity is only slightly lower than that of the stoichiometric reaction (81% ee). 

Apparently there is a need for bulk bases that are kinetically much less basic than LD A but that still are capable of efficiently regenerating the chiral lithium amide. Proton transfers to and from electronegative atoms like nitrogen are usually faster than from and to carbon. Therefore, we have explored carbon-based bases for the present purpose [20-22,24]. Indeed, bases like those displayed in Scheme 10 appear to have the predicted behaviour. 

The results led us to investigate the composition of the reagent solutions by multinuclear NMR spectroscopy. These studies revealed that under these new conditions 5 was no longer present as a homodimer, i.e., a 5 molecule is complexed with another 5 molecule. Instead the results showed new dimers -heterodimers 8 or 9, respectively. A monomer of 5 forms complex with a monomer of a bulk base and these heterodimers are the new reagents rather than homodimers of the chiral lithium amide (Scheme 12) [20-22]. 

In a lithium amide promoted deprotonation, one lithium amide molecule is consumed for each deprotonated epoxide molecule. Since chiral hthium amides are expensive reagents, there is a strong desire to develop less costly synthetic procedures for stereoselective deprotonations. Catalysis has the potential to solve the problem. What are needed are bulk bases capable of regenerating the chiral hthium amide from the chiral diamine produced in the deprotonation reaction. There have been some attempts along this line, e.g., by Asami and co-workers, who used the non-chiral hthium amide LDA as bulk base and the chiral hthium amide 4 as catalyst [9,12,39-41]. However, the stereoselectivity was considerably lower than what had been achieved in absence of the bulk base, i.e., under stoichiometric conditions. Most likely, the decreased stereoselectivity in the presence of bulk LDA is due to competing deprotonation by LDA to yield racemic product alcohol. The situation is illustrated in Scheme 9. 

---