Studies on the intermolecular and intramolecular (4+3) cycloadditions of epoxy and aziridinyl enolsilanes

Abstract

The mechanism of intermolecular (4+3) cycloadditions of epoxy enolsilanes has been more fully elucidated, and the incomplete retention of ee observed in endo cycloadducts was found to be a direct outcome of the cycloaddition, not from epimerization or other events after the formation of cycloadducts (Scheme 1). Computations suggested that chiral intermediate 2.9, derived from back attack by triflate on activated epoxide 1.46i, predominated over achiral oxyallyl cation 1.46ii formation, and was the more likely explanation for the observed loss of ee. This was verified experimentally by treating enolsilane (−)-1.46 with 1 equivalent of TESOTf, then adding the dienes after a period of delay (Scheme 2). Inverse enantiomeric cycloadducts of up to −61% ee were formed, showing that chiral triflate 2.9, rather than the achiral oxyallyl cation 1.46ii, was involved in the cycloaddition. The involvement of triflate was further demonstrated by promoting cycloadditions instead with TESB(ArF)4, where in the absence of any triflate, there was no change in the ee of the cycloadducts regardless of the time of addition of diene. The reaction mechanism of the cycloaddition of (−)-1.46 as we now understand it rationalizes that cycloadducts (+)-1.47/(+)-1.48 with complete retention of ee are the result of the direct cycloaddition of activated epoxide 1.46i with dienes under typical reaction conditions (Scheme 3). A slight loss of ee is attributed to the formation of minor amounts of 2.9, which undergoes cycloaddition to form inverse enantiomers (−)-1.47/(−)-1.48. In the absence of dienes, the formation of chiral triflate 2.9 predominates over its achiral regioisomer 2.10, the latter resulting from the attack of triflate on the less electrophilic site of activated epoxide. Cycloaddition with triflate 2.9 produced cycloadducts (−)-1.47/(−)-1.48 with inverse ee’s of up to −61% ee. Attempts to achieve an enantiodivergent synthesis failed to increase the yield and inverse ee of cycloadducts to synthetically useful levels. The optimized conditions include the addition of Et3NMeOTf, which provided cycloadducts in 30% yield and up to −58% ee. Cycloadditions resulting in inverse ee products occurred thus far only for (−)-1.46, and not for other substrates. The activation of epoxy enolsilane by chiral phosphoric acid (R)-2.30 for cycloaddition was also explored (Figure 1); however, the yield was low and no enantioselectivity was found. The intramolecular (4+3) cycloaddition of enol-tethered substrates (Figure 2) was studied. Cycloadditions of these substrates occurred with good yields of up to 95%, but the endo/exo selectivity was low (1:1 to 3:1). The cycloaddition of enolsilane (−)-3.1 afforded cycloadducts with some retention of ee (78% ee), while that of (−)-3.4 generated nearly racemic cycloadducts. Both the diastereoselectivity and optical purity of the cycloadducts were lower than those of cycloadducts derived from epoxide-tethered substrates. The incomplete transfer of chiral integrity from the epoxides suggested that enol-tethered substrates underwent competitive cycloadditions with both activated epoxide and oxyallyl cation intermediates. This was because the rates of reaction of these substrates were slower due to the first bond formation being an SN2′ addition, and because they were more favored to form stabilized oxyallyl cations.