Stereoselective Alkylation of Chiral Titanium(IV) Enolates with tert-Butyl Peresters

Here, we present a new stereoselective alkylation of titanium(IV) enolates of chiral N-acyl oxazolidinones with tert-butyl peresters from Cα-branched aliphatic carboxylic acids, which proceeds through the decarboxylation of the peresters and the subsequent formation of alkyl radicals to produce the alkylated adducts with an excellent diastereoselectivity. Theoretical calculations account for the observed reactivity and the outstanding stereocontrol. Importantly, the resultant compounds can be easily converted into ligands for asymmetric and catalytic transformations.

* sı Supporting Information ABSTRACT: Here, we present a new stereoselective alkylation of titanium(IV) enolates of chiral N-acyl oxazolidinones with tert-butyl peresters from Cα-branched aliphatic carboxylic acids, which proceeds through the decarboxylation of the peresters and the subsequent formation of alkyl radicals to produce the alkylated adducts with an excellent diastereoselectivity. Theoretical calculations account for the observed reactivity and the outstanding stereocontrol. Importantly, the resultant compounds can be easily converted into ligands for asymmetric and catalytic transformations.
T he need for more efficient and broad scope methods for the stereoselective construction of chiral molecular architectures is an endless source of inspiration for the development of new carbon−carbon bond-forming reactions. 1 In this context and despite the advances reported in the last decades, the α-alkylation of carbonyl compounds still remains as a challenging objective. 2 It is certainly true that successful methods based on the alkylation of metal enolates and enamines are widespread, but they are usually restricted to a privileged set of alkylating agents, namely sterically unhindered and active alkyl halides or sulfonates, able to react through an S N 2-like mechanism. 3−5 Alternative methods based on an S N 1like mechanism have been also reported, but they mostly require stabilized carbenium or oxocarbenium intermediates. 6−8 As a result, the chemo-and stereoselective introduction of any secondary or tertiary alkyl groups continues to be an unresolved issue. 9 Radical chemistry may offer an appealing way to achieve such an objective. Indeed, the tremendous success of the SOMO activation mode concept coined by MacMillan in the context of the direct and asymmetric alkylation of aldehydes illustrates the synthetic potential of the radical approach. 10 Inspired by these ideas and considering the biradical character of the titanium(IV) enolates, 11 we envisaged that they might undergo highly stereoselective alkylations provided that the required radical intermediates were generated in the reaction mixture. The feasibility of such an approach was clearly demonstrated in the alkylation of chiral N-acyl oxazolidinones with diacyl peroxides (Scheme 1). 12−14 Unfortunately, diacyl peroxides from α-branched aliphatic carboxylic acids are difficult to manipulate, which made the reaction with tertiary alkyl groups particularly elusive. In the search for more stable carboxylic acid derivatives to enable the introduction of secondary and tertiary alkyl groups we focused our attention on redox-active esters (Scheme 1). 15 Widely used phthalimidederived esters 16 containing an O−N bond proved to be unreactive, but peresters containing an O−O bond turned out to be much more satisfactory. 17 Herein, we describe the chemo-and stereoselective Cα alkylation of titanium(IV) enolates from chiral N-acyl oxazolidinones with tert-butyl peresters from branched aliphatic carboxylic acids, which permits the stereocontrolled introduction of secondary and tertiary alkyl groups with moderate to high yields (Scheme 1). Importantly, this method gives a straightforward access to enantiomerically pure intermediates that can be employed as precursors for ligands in catalytic and asymmetric synthesis. 18 Taking advantage of our experience, we were pleased to observe that the titanium(IV) enolate of (S) 4-benzyl-5,5dimethyl-N-propanoyl-1,3-oxazolidin-2-one (1 in Table 1) reacted with the tert-butyl perester from 1-adamantanecarboxylic acid (a in Table 1) under mild conditions similar to those employed for the alkylation with diacyl peroxides. 12 Indeed, the alkylated adduct 1a was isolated with a high yield and an excellent diastereoselectivity (74% and dr 97:3, see Table 1) through the simple stirring of a mixture of the titanium(IV) enolate of 1 with 1.5 equiv of a in 1,2-dichloroethane for 1.5 h at room temperature. Slight variations of such conditions also gave the desired adduct 1a but in lower yields ( Table 1).
The experimental procedure was next applied to a number of tert-butyl peresters from Cα branched carboxylic acids. 19 The introduction of tertiary alkyl groups proved to be possible in variable yields and heavily dependent on their structure but with outstanding stereocontrol since a single diastereomer (dr ≥97:3) of the alkylated adducts 1a−d was observed in all cases. Indeed, the results summarized in Scheme 2 show that they range from excellent for 1a (74%) to low for 1d in which perester d contains a tert-butyl-like chain possessing an aryl ether (23%). Importantly, peresters b−d react slowly compared to a, and we have occasionally observed the formation of the carboxylic acid derived from the reaction of the titanium enolate from 1 with the carbon dioxide released in the perester decarboxylation. Therefore, slow kinetics allow undesired side reactions to emerge and reduce the overall yield. Remarkably, the stereochemical outcome of the alkylation was firmly established through X-ray analysis of tert-butyl alkylated adduct 1c.
The reaction with secondary alkyl groups proved to be much more successful. As summarized in Scheme 2, the reaction with tert-butyl peresters e−j with Cα acyclic and cyclic aliphatic chains also gave the corresponding adducts 1e−j as a single diastereomer (dr ≥97:3) in good to high yields. Finally, it is worth pointing out the lack of stereocontrol of the βstereocenter in the alkylation with perester f, so adduct 1f was isolated as a 2:1 mixture of two diastereomers (Scheme 2).
Having established the scope of the alkylating agent, we next examined the influence of the N-acyl group on the outcome of the alkylation with a. The reaction turned out to be sensitive to steric hindrance but at the same time chemoselective. Indeed, a variety of functional groups as double or triple bonds, esters, or phenyl rings may be embedded in the acyl chain and produce the corresponding alkylated adducts in yields up to 67% (Scheme 3). Importantly, protected α-hydroxy and α-amino acyl derivatives (α-OTBS and α-pyrrole, 8 and 9, respectively, in Scheme 3) proved to be successful platforms from which the alkylated adducts 8a and 9a were obtained in high yields in a multigram scale, which demonstrates the robustness of the method and represents a straightforward way to get access to enantiomerically pure α-hydroxy and α-amino acids.
At this point, we carried out a comprehensive theoretical study to unveil the origin of the observed reactivity and selectivity. As for the reaction with diacyl peroxides, 12 DFT calculations 20 of the alkylation of 1 with perester a indicated that it also may proceed through an electron transfer from I Scheme 1. Stereoselective Decarboxylative Alkylation of Titanium(IV) Enolates from Chiral N-Acyloxazolidinones  such a fragmentation may produce up to four different species shown in Scheme 4. These species may be in an equilibrium favoring the carboxylate anion III, more stable than the radical counterpart IV. However, IV is a very unstable intermediate and undergoes a spontaneous decarboxylation to the corresponding tertiary radical V in an almost barrierless step (ΔΔG ⧧ < 5 kcal mol −1 ); importantly, a parallel decomposition of anion III is precluded by kinetic and thermodynamic reasons (ΔG°≈ 50 kcal mol −1 ). Thus, a Curtin−Hammett model may account for the formation of the adamantly radical V, which combines with the highly reactive Ti(IV) radical II to lead to the alkylated product 1a after C−C bond formation and decoordination of the titanium. Given the short distance at which the reagents must approach for the occurrence of the electron transfer, and due to the bulkiness of I and a, a good diastereoselectivity was ensured. Thereby, a remarkable minimum energy difference of at least 5.0 kcal·mol −1 corresponding to a dr >99:1 was calculated for the approach of distinct conformations of the perester to both π-faces of the enolate, with C−O distances ranging from 1.8 to 4 Å. This effect can be visualized in the 3D-representation shown in Scheme 4 of the approach between I and a, where the bulky adamantyl perester and the benzyl directing group are located at opposite faces of the enolate. Therefore, such a proposal accounts for both the observed reactivity and stereoselectivity since carbon-centered alkyl radicals are involved in the alkylation, whose stereochemical outcome hinges on the approach of the entire perester to the less sterically shielded Si π-face of the enolate.
Eventually, the easy access to α-adamantyl alkylated adducts 1a−9a led us to explore their conversion into enantiomerically pure building blocks and derivatives that might be employed as ligands for chiral catalysts. The results matched our expectations (Scheme 5). Indeed, reductive removal of the chiral auxiliary from 1a with LiBH 4 gave the corresponding alcohol 10 in 64% yield. Furthermore, 1,2-dihydroxy and 2-amino-1hydroxy derivatives 11 and 12, respectively, were synthesized from adducts 8a and 9a in a similar way, which represents a straightforward approach to such interesting ligands. 21 In summary, we have developed a highly stereoselective alkylation of titanium(IV) enolates of a variety of chiral N-acyl oxazolidinones with tert-butyl peresters from Cα branched aliphatic acids under experimentally mild conditions. The resultant alkylated adducts are isolated in moderate to high