Chiral Brønsted Acid‐Catalyzed Asymmetric Synthesis of N‐Aryl‐cis‐aziridine Carboxylate Esters

Abstract We report a multi‐component asymmetric Brønsted acid‐catalyzed aza‐Darzens reaction which is not limited to specific aromatic or heterocyclic aldehydes. Incorporating alkyl diazoacetates and, important for high ee's, ortho‐tert‐butoxyaniline our optimized reaction (i.e. solvent, temperature and catalyst study) affords excellent yields (61–98 %) and mostly >90 % optically active cis‐aziridines. (+)‐Chloramphenicol was generated in 4 steps from commercial starting materials. A tentative mechanism is outlined.

Such is the versatility of organocatalysis and its ability to mediate ap lethora of diverse reaction types [1] it is,n ow,a n indispensable "tool" in the synthetic chemists "toolbox". [2] Indeed, improving atom-and reaction-efficiencyi sakey driver to developing new reactions and protocols;i nt his context organocatalysis has demonstrated its importance by efficiently mediating many different convergent reactions or multi-component syntheses.T he work here supports these aspects by generating structure and function-diverse motifs via fewer synthetic, isolation and purification steps.
Activating the C = Nb ond of an imine with aB INOL phosphoric acid [5] lowers its LUMO energy and generates an imminium ion pair that can, but not always will, react with an ucleophile.S eminal work by Akiyama et al. established chiral BINOL phosphoric acids [pK a % 13 (CH 3 CN) [6] ]a ctivate aldimines (derived from, specifically,arylglyoxals and panisidine) and react with ethyl diazoacetate (EDA) affording cis-aziridines in 92-97 % ee. [7] Similarly,o ther Brønsted acids [8] and pyridinium triflate activate ad iverse array of imines,i ncluding for example,2 -pyridyl derived 5,e nabling the presumed imminium ion-pair (not shown) to react with EDAa nd afford cis-rac-aziridine (8,8 3% yield) (Scheme 2). [9] With these racemic studies complete our focus shifted to developing asubstrate enhanced and diverse, multi-component asymmetric aza-Darzens reaction. Inspired by the work of Akiyama et al. [7] and the Mannich reaction reported by Yamanaka et al. we considered the inclusion of 5 may generate ac onstrained hydrogen-bonded and activated complex similar to 11;w ew ere drawn to the use of 5 to generate 11 due to similarities in the chiral non-racemic rigid environment proposed by Yamanaka (using a N-(2-hydroxy-phenyl)imine starting material). [10] Screening chiral nonracemic BINOL and VA NOL phosphoric acids,a sw ell as aH -QUIN-BAM triflate salt [11] we were disappointed no reactions were observed. We attribute the failure using 5,a s well as other alternative imines,t ot he low pK a so ft he Brønsted acids and their inability to generate as ufficiently "activated" form of 10 or 11.
Encouraged by the results with 6,s terically encumbered tert-butyl ester 7 was investigated. Ag radual increase in ee was observed but, overall, the levels of stereoinduction were, generally,i nferior.S o, N-benzyl 5 was substituted for ar otationally less flexible N-4-(methoxyphenyl) or N-PMP group.
As olvent and temperature study using 1mol %o f21 established chloroform at À60 8 8Cwas the optimum combination for transforming 2-(tert-butoxyphenyl)-24 into cis-27 with an excellent 98 % ee and 95 %y ield. Probing the catalytic activity of 21 at 0.5 and 0.25 mol %loadings the reaction times increased to 48 and 62 hours.I nb oth examples cis-27 was afforded in very similar 87 %/86 % ee and 98 %/95 %y ield, respectively.
Supporting protonation, not hydrogen-bond activation, [17] Houk et al. described am echanism and origins of catalysis DFT and experimental study in which as imilarly N-protonated, to 60,r eactive hydrazonium-phosphoramide [18] anion (not shown) was formed from aB INOL N-triflylphosphoramide and ahydrazone.Activation of 31 is crucial;the widely accepted aza-Darzens mechanism [19] invokes attack of adiazo nucleophile (i.e. 7)onanimminium cation (i.e. 60)generating an a-diazonium b-amino ester (i.e. 61,s ee Path A). The importance of the latter,f rom ar eaction kinetics and enantioselectivity point of view has been established by the reluctance of these intermediates to undergo aretro-Mannich reaction. [20] Generating, presumed, kinetic product 61 with excellent enantioselectivity is possible only if 7,w ith its heterotopic faces that is, 7 Re and 7 Si ,e fficiently discriminates between the Si and Re faces of optically active 60.P ath A outlines how anti-diazonium intermediate 62 (Scheme 6) forms when the sterically encumbered heterotopic 7 Re face approaches the Si face of imine 60 minimising the steric interactions between the intramolecularly hydrogen bonded bulky ortho-(tert-butoxy)phenyl imminium and the tert-butyl ester on 7 Re .Although we have no direct evidence ( 1 H-NMR) for the backbone rigidifying hydrogen bond in 60 similar intramolecular hydrogen bonds in ortho-substituted Schiff basesare known. [21] Newman projection 62 affords adetailed depiction of the minimized steric interactions between the tert-butyl ester and ortho-tert-butylphenyl ether. An intramolecular S N 2c yclization (release of N 2 )b etween the antiperiplanar amine and diazonium groups affords cis-41.Path B proceeds via ion-pair 60,h owever approach of 7 Si onto the imine Si face is,n ow,i nhibited by the two sterically bulky groups.Thus,formation of a-diazonium b-amino ester 63 and trans-41 is disfavoured. Thecrude 1 H-NMRs of our reactions afforded no evidence of trans-41 or a-diazonium b-amino ester 63.