Cyclization of a carbon-centered radical derived from oxaziridine cleavage

Treatment of an oxaziridine with low-valent iron or copper salts generates a carbon-centered radical able to cyclize onto an appended olefin


Introduction
Oxaziridines undergo a wide variety of chemical transformations including rearrangement and atom transfer reactions. 1Oxaziridines have long been known to react stoichiometrically with low-valent metals such as Fe(II) via single-electron transfer (SET) pathways, affording nitrogen radical/oxygen-anion (or hydroxyl if the reaction is run in an acidic medium) pairs that undergo different secondary reaction pathways depending on the structure of the oxaziridine. 2Many of these products were proposed to result from β-scission of the nitrogen radical to afford a carboncentered radical (Scheme 1).2a The existence of R 2 • has generally been assumed based on the isolation of products consistent with traditional radical processes such as hydrogen atom abstraction, disproportionation, dimerization, 2c and the ability to initiate redox alkene polymerization.2b Addition reactions of R 2 • to electron-poor heterocycles, 2e,f NO, 2d and an Fe(III)-complexed acetylacetate ligand 2i have also been cited as evidence for C-radical formation in these processes.

Scheme 1
The observation of 5-exo-trig cyclizations is considered strong evidence for the intermediacy of radicals 3 and a powerful synthetic tool besides. 4In this article, we disclose the deliberate generation of a carbon-centered radical via oxaziridine SET and β-scission using catalytic quantities of a low-valent metal salt under neutral conditions and the cyclization of the resulting radical onto an appended olefin.

Results and Discussion
We have previously examined the SET reactions of 3-aryl oxaziridines using catalytic amounts of a soluble Cu(I) salt in refluxing THF (eq 1). 5 Following formation of a putative nitrogen radical/oxygen-anion pair, a complex series of steps led to either pyrroline (1, eq 1) or azidirine (not shown); both of these products were proposed to result from nitrogen radical addition to the olefin followed by migration of the phenyl (N-CH(CH 3 )Ph) group or rearrangement.5a Subsequent attempts to extend this chemistry to oxaziridines bearing a carbomethoxy substituent at C-3 failed, leading instead to the isolation of amide (2, eq 1).5b The formation of amide is consistent with the preferential β-scission route as shown in Scheme 1 and suggests that the formation of pyrroline is contingent on the presence of a C-3 substituent unable to afford a stable radical upon homolysis.

ARKAT
To test this, a series of oxaziridines 3-7 bearing different alkyl substituents at C-3 were prepared and subjected to treatment with 5 mol% of [Cu(PPh 3 )Cl] 4 in refluxing THF (eq 1 and Table 1).Throughout this study, oxaziridines were prepared by condensing a ketone with an amine in toluene at reflux and treating the crude solution of imine thus made with m-CPBA at -78 °C. 1,5Except for entries 1, 3, and 4, the oxaziridines in Table 1 were obtained and used as mixtures of stereoisomers.The relative amounts of pyrroline 1 vs. amide (usually 2) depended on the nature of the C-3 substituent.In particular, the oxaziridines bearing a radical-stabilizing alkyl group at this position furnished the amide exclusively (entries 3-5), whereas when R = Ph, none of this product was obtained (entry 1).These data are consistent with a mechanism shown in Scheme 1, with only R groups able to afford a reasonably stable radical undergoing βcleavage.
Oxaziridine 4 derived from a methyl ketone (entry 2) was about midway between these extremes; in fact, N-acetyl-α-methylbenzylamine 9, and not amide 2, was the β-scission product observed in this case because loss of the 3-butenyl radical should be favored relative to Me•.The pyrroline 8 and amide 9 obtained in this case could not be separated and, hence, 8 was isolated as the pivaloyl amide 10 after treating the reaction mixture with pivaloyl chloride.The observed variations in ratios of 1 and 2 could not be explained based solely on the relative rates of βfragmentation for the R groups (collected in Table 1).Additionally, no definitive information regarding the nature of the departing group or its fate could be inferred from these experiments (no attempt was made to isolate the non-amide products of these reactions).

ARKAT
We considered the possibility of generating radicals that could be utilized in further radical processes.Accordingly, oxaziridine 12 (see Scheme 2 for structures) was synthesized in the expectation that it would undergo β-scission efficiently due to the production of a benzylic radical in so doing.An achiral amine was used to minimize complications in product analysis resulting from stereoisomer formation.Indeed, treatment with a variety of Cu(I) and Fe(II) salts in refluxing THF led smoothly to the formation of products 13a, 13b, and 14 resulting from cyclization of the carbon-centered radical.As seen in Table 2, all of the soluble catalysts used gave comparable results in terms of yield and product ratio.a Determined by 500 MHz 1 H-NMR integration.
Our proposed mechanism is depicted in Scheme 2. Cyclopentanes 13a and 13b result from addition of the benzylic radical to the olefin in a 5-exo-trig manner followed by hydrogen atom abstraction from a molecule of THF. 7 In the similar cyclization of 1-phenylhexen-5-yl radical investigated by Walling, the trans-cyclopentane 13b was obtained as the only product. 8In our case, 13b was the major product in accord with the Beckwith-Houk model. 9The decreased diastereoselectivity may be related to the changes in solvent polarity or the possibility that the cyclization is occurring inside the solvent cage.The production of 14 is attributed to the oxidative addition of the initially formed methylene radical into the phenyl group; such additions are commonly observed 10 and appeared in the mechanism originally proposed for the formation of 1. 5 It seems reasonable that Cu(I) is regenerated via SET from the amide anion to the Cu(II) formed in the first step (shown in Scheme 2 as an ion pair for book-keeping purposes).Two aspects of these reactions are noteworthy.First, the observation of cyclization products from 12 constitutes good evidence that the initially formed nitrogen-centered radical actually undergoes β-scission to eject a carbon-centered radical.Second, the use of catalytic amounts of soluble Fe(II) and Cu(I) salts for initiating oxaziridine SET under neutral conditions is quite mild and synthetically attractive relative to the original techniques used for such reactions. 2Although further studies will be necessary to establish the generality of this process, the recognition that oxaziridines are readily accessible from the corresponding ketones suggests that the overall sequence of 11 → 12 → radical may constitute an interesting approach to a decarbonylative method for carbon radical generation under mild, catalytic conditions.

ARKAT Experimental Section
General Procedures.THF was freshly distilled from sodium-benzophenone and degassed by multiple freeze-thaw cycles at <0.5 mm Hg. 1 H and 13 C NMR spectra were recorded on a 500 MHz NMR spectrometer (at 500 and 125.5 MHz, respectively) or on a 300 MHz machine (at 300 and 75.6 MHz, respectively).All NMR samples were dissolved in deuteriochloroform, and chemical shifts are given in parts per million (ppm) relative to the internal standard tetramethylsilane (TMS).Optical rotations were measured on a Perkin-Elmer 241 polarimeter, and concentrations are reported in g/100 mL.Elemental analyses were obtained in-house.Unless otherwise noted, starting materials were obtained from commercial suppliers and used without further purification.[Cu(CH 3 CN) 4 ]PF 6 , 11 [Cu(bpy)(PPh 3 )]Cl, 12 and [Cu(PPh 3 )Cl] 4 13 were prepared according to literature procedures.General procedure for preparation of oxaziridines.To a solution of the starting ketone (1.0 equiv) in toluene was added 1.2-2.0equiv of α-methylbenzylamine (α MBA), freshly crushed 5Å molecular sieves, and 5 mol% of p-toluenesulfonic acid monohydrate.The reaction mixture was then refluxed in a three-necked round-bottom flask equipped with a water-cooled condenser and drying tube for ca.48 h.While the imine solution was cooling, a flame-dried round bottom flask kept under N 2 atmosphere was charged with m-CPBA (1.2-2.0 equiv) and methylene chloride to completely dissolve the m-CPBA.The flask was cooled to -78 °C and the imine solution was then transferred to the m-CPBA solution via a wide-bore cannula.The reaction was followed by TLC and was generally complete within one hour, at which time, it was quenched with saturated Na 2 S 2 O 3 and allowed to warm to room temperature.The layers were separated, and the organic layer was washed with saturated NaHCO 3 , brine, and dried over Na 2 SO 4 .Concentration followed by column chromatography or medium-pressure liquid chromatography (MPLC) on silica gel with the indicated solvent system provided the oxaziridine.
General procedure for Cu(I) or Fe(II) promoted cyclization (Table 2, Entry 4).The substrate concentration was fixed at 0.05 M, and 5 mol% of catalyst was used in all cases.The following procedure is representative.A mixture of [Cu(CH 3 CN) 4 ]PF 6 11 (12 mg, 0.032 mmol, 0.05 equiv) and degassed THF (110 mL) in a Schlenk-type reaction vessel was refluxed for 30 min under Ar atmosphere.To the resulting solution was transferred dropwise a solution of 2diphenylmethyl-3-phenyl-3-(1'-phenyl-5'-hexen-1'-yl)oxaziridine 12 (246 mg, 0.642 mmol) in degassed THF (15 mL plus 5 mL rinse) via a cannula over a 5 min period.Refluxing was continued for an additional 20 h.After cooling to room temperature, the solvent was carefully evaporated.Column chromatography of the residue using pentane afforded 78 mg (76%) of a colorless oil whose composition was assigned as a 20:70:10 mixture of cis and trans-1-phenyl-2methylcyclopentane 13a,b and 1,2,3,3a,8,8a-hexahydrocyclopent[a]indene 14 by 1 H-NMR integration.This mixture was chromatographically (TLC) inseparable, and the products thus obtained were identical in all respects with the authentic samples prepared as reported below.

Table 2 .
Results of cyclization experiments using 12.