Unlocking electrophilic N-aryl intermediates from aryl azides, nitroarenes, and aryl amines in cyclization-migration reactions

Abstract

An Account of our development of reactions to construct N-heterocycles by triggering cyclization-migration tandem reactions from aryl azides, nitroarenes, and aryl amines is described. The reactivity patterns of metal N-aryl nitrenes, nitrosoarenes, N-aryl nitrogen radical anions, and N-aryl nitrenoids are compared.

Graphical Abstract

Introduction

Historically, the development of synthetic transformations involving N-aryl nitrene- or N-aryl nitrenoid reactive intermediates has lagged behind other divalent intermediates because of their penchant to form intractable tars (Scheme 1).¹ The Schuster-, Smolinsky-, and Platz groups studied the reactivity of aryl azides.^2,1a,3,4 They determined that the polymeric products do not originate from singlet N-aryl nitrene ¹1, which has a short lifetime of <10 ps,⁵ but from a subsequent rearrangement product azirine 2⁶ or dehydroazepine 3.^2a,c,d,7 Further complicating matters, an intersystem crossing to the lower energy triplet N-aryl nitrene was found to be favorable to enable an additional decomposition path to afford azoarene by-products.^3e,4a In our opinion, the propensity of N-aryl nitrenes to form these decomposition products limited the design of synthetic transformations that proceeded via N-aryl nitrenes in comparison to other divalent reactive intermediates,⁸ and we anticipated that the reactivity of N-aryl nitrenes might be tamed using transition metal catalysts. In 2006, transition metal-catalyzed generation of metal N-aryl nitrenes from aryl azides was under developed.⁹ Cenini and co-workers demonstrated in 2000 that C–H bond aminations could be achieved using aryl azides as the nitrogen source and cobalt tetraphenylporphyrin as the catalyst.¹⁰ Their investigations demonstrated that controlling the reactivity of the catalytic intermediates was a challenge: for example, exposure of 4-azidonitrobenzene to 1 mol % of Co(TPP) generated a mixture of secondary aryl amine 11 and imine 12 as well as azoarene 13.

Scheme 1.

Generation of N-aryl nitrenes or N-aryl nitrenoids from aryl azides.

Our laboratory took a circuitous route before focusing on studying aryl azides as N-aryl metal nitrene precursors. In 2006, we were interested in developing metal-catalyzed reactions of 2H-azirines to access N-vinyl metal nitrenes for C–H bond functionalization reactions (Scheme 2). We were inspired to study 2H-azirines because of their rich history where C–N bond activation or C–C bond activation could be achieved by modifying the reaction conditions.¹¹ At the time, their reactivity towards transition metals was nascent,¹² and we were curious if the metal could impact whether a nitrene- or a dipole reactive intermediate was formed. 2H-Azirines, however, can be synthesized from vinyl azides,^11a,13 which proved easier to isolate, purify, and handle than the strained 2H-azirines. Because of these merits we were sidetracked into studying the reactivity of vinyl azides towards transition metals. We found that exposure of α-azidoacrylate 14 to 5 mol % of rhodium carboxylate catalyst resulted in the formation of indole at only 60 °C.¹⁴ This reactivity was not unique to β-aryl substituted azidoacrylates, but dienyl azides, such as 17, could be transformed into pyrroles.¹⁵ In contrast to 14, the reactivity of dienyl azides could be unlocked with a variety of transition metal salts, including Zn-, Cu(I)-, Cu(II)-, and perfluorinated Rh₂-complexes, at room temperature.

Scheme 2.

2H-Azirines and vinyl azides as N-alkenyl metal nitrene precursors.

2. Unlocking the reactivity embedded in aryl azides

At the conclusion of our study into the reactivity of vinyl azides towards Rh₂-carboxylate catalysts, we were curious if the aryl azides exhibited similar reactivity (Scheme 3). From our perspective, aryl azides were more ideal as nitrogen-atom sources because a broader range of 2-substituted aryl azides could be constructed from 2-halide- or 2-boronic acid-substituted anilines using well-established transition-metal catalyzed cross-coupling reactions or from 2-halide-substituted nitroarenes through nucleophilic substitution reactions. To test their reactivity, 2-azidostilbene 19 was exposed to 5 mol % of Rh₂(O₂CC₃F₇)₄ at 60 °C and 2-phenylindole 21 was obtained in 98%.¹⁶ Aryl azides proved to be more robust than azidoacrylates to require slightly higher temperatures than 14, and we found that only Rh₂(O₂CC₇H₁₅)₄ or Rh₂(O₂CC₃F₇)₄ to be capable of triggering C–NHAr bond formation. The catalyst loading could be reduced to as little as 2 mol % without attenuating the yield of the amination reaction. This reactivity proved to be broad: carbazoles 24,¹⁷ γ-carbolinium ions 25,¹⁸ α-, β-, or δ-carbolines 26 – 28,¹⁹ benzimidazoles 29,²⁰ benzisoxazoles 30,²¹ or 2H-pyrazoles 31 could be constructed from 2-substituted aryl azides using Rh₂(II)-carboxylate, ruthenium-, or iron salts. Our investigations showed that the reactivity of the 2-substituted aryl azide depended on the identity of the ortho-substituent. While benzisoxazoles, 2H-pyrazoles, and imidazoles could be accessed at 40 °C, 2-substituted indoles and carbazoles required more heat, and electron-deficient pyridinium ions were even more inert requiring 80 °C for efficient reaction for product formation.

Scheme 3.

Aryl azides as N-aryl metal nitrene precursors for N-heterocycle synthesis.

We were delighted to find that intramolecular sp³-C–H bond amination could be achieved using aryl azides. Electron-deficient aryl azides could be transformed into 2-aryl indolines 32 using only 2 mol % of [(cod)Ir(OMe)]₂ at room temperature,²² and a broader range of sp³-C–H bond amination could be accomplished using 5 mol % of Rh₂(esp)₂ if Boc₂O was added and the reaction was heated to 120 °C.²³

Because of the unusual reactivity trends observed in our method development, a series of experiments were performed to gain insight into the mechanism of the C–NAr bond forming step (Scheme 4).²⁴ In contrast to the pyrolysis of 2-azidostilbene 19 where only the E-isomer was smoothly transformed into 2-phenylindole 21,²⁵ exposure of either the E- or the Z-isomer to a substoichiometric quantity of Rh₂(O₂CC₇H₁₅)₄ produced indole in similar yield (95% versus 93%).¹⁶ The equal reactivity of the two isomers was interpreted to indicate that insertion of the rhodium-nitrenoid into the C–H bond was not occurring. Intramolecular competition kinetic isotope experiments produced a secondary kinetic isotope effect of 1.01, to further confirm that insertion in the aryl C–H bond was not occurring. To distinguish between and electrophilic aromatic substitution reaction and electrocyclization for C–NAr bond formation, intramolecular competition experiments were performed using triaryl azides 33 and analyzed using the Hammett equation. While thermolysis produced a linear correlation with σ+-values and a ρ-value of −0.66, a V-shaped curve was obtained when a Rh₂(II)-carboxlate was present. The V-shaped curve that depended on the identity of the Rh₂(II)-carboxylate catalyst with Rh₂(O₂CC₃F₇)₄ exhibiting a shallower V than Rh₂(O₂CC₇H₁₅)₄.

Scheme 4.

Mechanistic experiments that provide insight into the mechanism of C–NAr bond formation.

The results of the mechanistic experiments were interpreted to indicate that C–N bond formation occurred through a 4π-electron-5 atom-electrocyclization (Scheme 5). Coordination of aryl azide 36 with the Rh₂(II)-carboxylate catalyst produces an azide metal complex where the rhodium is coordinated to either the α- or γ-nitrogen. Extrusion of dinitrogen produces rhodium nitrenoid 39, which reacts with the proximal ortho-aryl substituent in an electrocyclization to produce 40 or 41. While disassociation of the rhodium could occur to generate the free N-aryl nitrene, which could react with the aryl substituent, the difference in Hammett correlation between the catalyzed- and non-catalyzed reaction suggests that the Rh₂(II)-catalyst is present for C–NAr bond formation. A [1,5] hydride shift on 41 occurs to furnish the N-heterocycle product and re-generate the catalyst. Electrocyclization, albeit 6π-electron, was posited by Davies and Houk for C–N bond formation in the reductive cyclization of 2-substituted nitrosoarenes, which were formed in situ from the reduction of 2-substituted nitroarenes.²⁶ The role of the dirhodium(II) carboxylate complex in triggering this reaction was investigated by Powers and co-workers, who reported in crystallo structures rhodium coordinated to ortho-biphenyl azide at the α-nitrogen and Rh₂(II)-N-aryl nitrenoids, which rearrange to carbazole upon heating or photolysis.²⁷

Scheme 5.

Mechanistic experiments that provide insight into the mechanism of C–NAr bond formation.

Because the C–H bond was not cleaved in the C–NAr bond forming step, we recognized that our proposed mechanism of carbazole formation could be leveraged to access functionalized N-heterocycles through tandem reactions by replacing the hydrogen at the reaction center with a migrating group (Scheme 6). We found that exposure of β,β-disubstituted 2-nitrostyrenes to a rhodium carboxylate catalyst triggered a tandem reaction of rhodium N-aryl nitrene 44: cyclization produced C3 cation 45, which triggered a migration of one of the β-substituents. While we were not surprised to see a selective [1,2] aryl shift to produce 2,3-disubstituted indoles, such as 46,²⁸ we were surprised to see the exclusive migration of a nitro-group in preference over a hydride to produce 3-nitroindoles 47,²⁹ which is the opposite selectivity observed by Pelkey and Gribble in the thermal reaction.³⁰ Our investigations uncovered several other groups that migrated in preference to hydride including acyl-, benzoyl-, and sulfonyl groups to afford indoles, such as 48 and 49. Our metal-catalyzed results mirrored those of Gairns, Moody, and Rees, who had previously shown that sulfur-groups and acyl-groups readily migrate when 3-azido-2-vinylthiophenes are irradiated.³¹ In contrast, neither carboxylates, nor amides migrated in preference to hydride. Our investigations were able to further distinguish between migrating groups to show in a competition between nitro- and benzoyl groups that only nitro-group migration was observed to provide only 51, and that an amide migrated in preference to aryl group to give solely 52. Our investigations also showed that the electron-rich heteroatoms could be leveraged to achieve selective migration with both ethereal- as well as aminomethylene migration preferred over alkyl migration to afford 2,3-disubstituted indoles, such as 53.³² While no intermolecular crossover was observed in the migration of the aminomethylene substituent, the loss of stereochemical information at the migrating center suggested that an iminium ion was formed as a reactive intermediate.³² The reactivity of fully substituted α,β,β-trisubstituted styryl azides towards Rh₂(II) carboxylates was investigated, and we found that cyclization-migration tandem reactions could be triggered to afford functionalized indoles.³³ When the β-substituent was an acyl group, migration occurred to the nitrogen to afford 1,2-3-trisubstituted indoles, such as 54. In contrast, when the β-substituent was a carboxylate, the Rh₂(II)-catalyzed cyclization-migration tandem reaction afforded 3H-indole 55.³⁴ This preference could be overridden to afford 1,2,3-trisubstituted indoles if the steric environment was increased near the C3 position.³⁵ Oxindoles, such as 56, could also be formed through a migration-cyclization-fragmentation reaction sequence if the β-substituent was an alkyl ether.³⁴ Together, our investigations enable a migratory aptitude scale to be devised where ester- < alkyl- < aryl- < aminomethylene < amide- < H- < sulfonyl- < acyl- << nitro-groups.

Scheme 6.

Exploitation of the mechanism of carbazole formation to trigger cyclization-migration reactions of β,β-disubstituted nitrostyrenes.

We were curious if the ability of N-aryl rhodium nitrenes to engage in cyclization-migration reactions could be exploited to construct larger ring-sized N-heterocycles (Scheme 7).³⁶ We anticipated that replacing the ortho-alkenyl substituent with a strained ring might trigger a ring expansion reaction and were delighted to see that exposure of 2-cyclobutanol-substituted aryl azide 57 to 1 mol % of Rh₂(esp)₂ at 120 °C produced benzazepinone 60. While other rhodium(II) carboxylate complexes could catalyze this process, Du Bois and Espino’s Rh₂(esp)₂ was the most efficient,³⁷ which we attributed to its increased thermal stability relative to other bidentate coordinated carboxylate complexes. The tandem reaction exhibited broad substrate scope, and the migration was stereospecific and selective favoring benzyl- or allyl-group migration over methylene migration, and methine group migration over methylene migration. We interpreted the reactivity trends of the substrates to indicate that the mechanism proceeded via a rhodium N-aryl nitrenoid that triggered a ring expansion of the ortho-cyclobutanol substituent to produce spirocycle 58. Proton-transfer followed by aromatization formed acylium ion 59, which was trapped by the proximal amine to form the seven-membered ring. The hydroxyl group was critical for medium-ring formation, when it was replaced with an ether, sp³-C–H bond amination occurred instead of ring-expansion to form indoline 63.

Scheme 7.

Rh2(II)-Catalyzed formation of benzazepinones from 2-cyclobutanol-substituted aryl azides.

3. Exploiting the reactivity of nitrosoarenes generated from nitroarenes

Generating electrophilic N-aryl nitrogen reactive intermediates from nitroarenes was reported by Cadogan and co-workers in 1962 (Scheme 8).³⁸ They hypothesized that the mechanism for this transformation involved nitrosoarene 65 and potentially N-aryl nitrene 66 as reactive intermediates,³⁹ which could engage with the proximal aryl substituent via 67 or TS-68 to form the C–NAr bond. In 1967,⁴⁰ Sundberg and co-workers followed this seminal work to show that β,β-disubstituted nitrostyrenes, such as 70, could be transformed into 2,3-disubstituted indoles 73 using phosphite as the superstoichiometric reductant. Sundberg and co-worker’s result suggested that C–NAr bond formation occurred through a cyclization to form 72, which could trigger a [1,2] methyl shift. Catalysis of this reductive cyclization was reported by Watanabe and co-workers in 1994: the combination of a palladium catalyst and tin chloride was shown to enable 2,3-indole formation if a high pressure of CO was used as the stoichiometric reductant.⁴¹ While this reductive cyclization was limited to the formation of indoles or indazoles, Smitrovich and Davies reported that carbazoles could be accessed using 1,10-phenanthroline as the ligand at 140 °C using only 4.76 atm of CO as the stoichiometric reductant.⁴² Calculations by Davies, Houk, and co-workers suggested that C–NAr bond formation occurred through a 6π-electron-5-atom electrocyclization of nitrosoarene 78 to afford 79, which rearranged to produce carbazole 80.²⁶ Against this backdrop, we were curious to compare the reactivity patterns embedded in 2-substituted nitroarenes to 2-substituted aryl azides to determine if selective tandem cyclization-migration reactions could be triggered.

Scheme 8.

Reductive cyclization-amination reactions mediated by phosphite or transition metal catalysts.

We targeted the construction of non-planar N-heterocycles from α,β,β-trisubstituted nitrostyrenes 81 using the combination of a transition metal catalyst and a stoichiometric reductant (Scheme 9).⁴³ We found that an in situ generated palladium(II) phenanthroline complex efficiently catalyzed the desired cyclization-migration tandem reaction using Mo(CO)₆ as the stoichiometric reductant to produce spirocycle 84. Replacing Mo(CO)₆ with 1.5 atmospheres of CO caused the tandem reaction to be short-circuited. While reduction of the nitro-group and cyclization occurred, deprotonation of 83 occurred instead of a [1,2] shift. Mechanistic investigations into the role of the palladium catalyst and Mo(CO)₆ revealed that palladium’s role was to reduce the nitroarene 81 to nitrosoarene. Molybdenum hexacarbonyl appears to have several functions: in addition to serving as a source of CO, our experiments revealed that molybendum also coordinates the nitrosoarene to afford 82—preventing [4+2] trapping with a diene—to potentially activate the nitrosoarene for cyclization with the proximal alkene to produce 83 and prompt the [1,2] shift. Similar migratorial trends were observed in the trisubstituted nitrostyrenes compared to styryl azides. Ring contraction of the ortho-cyclohexenyl group was observed to afford spirocycles such as 84 when the β-substituent was an aryl group—irrespective of its electronic nature. Analogous reactivity to styryl azides was observed when the identity of the β-substituent was changed to a carboxylate (85): instead of ring-contraction, [1,2] carboxylate migration was observed to afford only 3H-indole 86. The difference in migration preference was rationalized by comparing the difference of energy of the partial positive charge formed at the C2 position of the N-heterocycle in transition states TS-87 and TS-88. When R is an aryl group, TS-87 would be lower in energy because the partial positive charge could delocalize throughout the π-system. In contrast, when R is an electron withdrawing group, such as a carboxylate, this transition state is destabilized and ring-contraction via TS-88 becomes favored. Similar migration trends were observed in nitroarenes and aryl azides with acyclic alkenyl ortho-substituents. In both nitrostilbene 89 and stilbene azide 91, phenyl group migration was observed to afford 3H-indole 90 or indole 46.

Scheme 9.

Pd-Catalyzed reductive cyclization using Mo(CO)6 as the terminal reductant.

We found that Pd-catalyzed reduction of nitroarenes produced nitrosoarenes that could react with functionality that did not react productively with rhodium N-aryl nitrenes generated from aryl azides (Scheme 10). We found that sp³-C–NHAr bond formation could be achieved using nitroarenes as the nitrogen-atom source by capturing the nitrosoarene intermediate with a proximal nucleophile. Exposure of 92 to 5 mol % of Pd(OAc)₂ and 10 mol % of phenanthroline in the presence of 2 atm of CO produced indoline 94, through the apparent attack of the electrophilic nitrosoarene with in situ generated enol 93.⁴⁴ In contrast to the reactivity of N-aryl metal nitrenes, five-, six-, and even seven-membered ring formation was possible using this reactivity pattern. We were able to leverage the ability of palladium(II) acetate to participate in directed C–H bond activation to achieve the functionalization of aryl C–H bonds to produce amides, such as 97.⁴⁵ This reaction requires no exogenous ligand, and independently synthesized palladium metallocycles could not only be converted to product but were also competent catalysts. Palladium serves dual catalytic roles in this transformation: not only does it reduce the nitroarene to nitrosoarene, but it also catalyzes the activation of the proximal C–H bond.

Scheme 10.

Development of intermolecular nitrosoarene trapping experiments.

Because of the relative harsh conditions required to generate the nitrosoarene reactive intermediate, we focused our attention on the development of milder reactive conditions. We had two goals: replace the palladium catalyst with an earth-abundant first row transition metal and replace the carbon monoxide source with a less toxic reductant (Scheme 11). Working with the High Throughput Experimentation group at Merck, we identified that the combination of Fe(OAc)₂, 4,7-(MeO)₂phen could transform 2-nitrostyrenes 98 into indoles 99 using phenylsilane as the stoichiometric reductant.⁴⁶ We found that the initial hit (using 10 mol % of iron and ligand) identified by the HTE-group could be further optimized to use as little as 0.5 mol % of Fe(OAc)₂ and 0.5 mol % of 4,7-(MeO)₂phen. In contrast to the combination of palladium and Mo(CO)₆, the conditions were impotent to trigger a cyclization-migration reaction to produce 2,3-disubstituted indoles from β,β-disubstituted-2-nitrostyrenes. The inertness of 100 towards reaction conditions suggests that molybdenum (or palladium) is critical to spur the cyclization-migration reaction of the nitrosostyrene intermediate. Iron acetate was not the only earth abundant transition metal salt that could catalyze nitrosoarene formation, we found that intermolecular cross-coupling reactions between nitroarenes or nitroheteroarenes and aryl boronic acids could be realized using copper acetate and phenyl silane as the stoichiometric reductant.⁴⁷ For this reaction, dppe proved to be the best ligand, and the cross-coupling reaction was restricted to the use of boronic acids; trifluoroborate salts and boronic pinacolate esters were inactive coupling partners. The reactivity patterns suggested that copper reduced the nitroarene to nitrosoarene 102, which could be trapped with 2,3-dimethylbutadiene, suggests that coordination to copper is reversible. We interpreted the inertness of boronic esters and trifluoroboronic salts to indicate that the cross-coupling involved the formation of metallocycle 103 where the copper catalyst was coordinated to both the nitrosoarene and the boronic acid.

Scheme 11.

Development of earth abundant metal-catalyzed reactions of nitroarenes.

4. Radical anion N-aryl nitrogen reactive intermediates from nitroarenes

Over the course of our investigations into the reactivity of nitrosoarene reactive intermediates generated from nitroarenes using transition metal catalysts, we were curious if we could access radical N-aryl intermediates from nitroarenes and if they would exhibit similar reactivity patterns (Scheme 12). Zhu and co-workers had shown that the common single electron reductant, TiCl₃, ably reduced α,β,β-trisubstituted nitrostyrenes, such as 105, to afford nitrosostyrenes (e.g. 106) to trigger cyclization-migration tandem reactions to produce complex, functionalized N-heterocycles that could be converted into alkaloid natural products.⁴⁸ The migration selectivity mirrored that observed by our group in the reactivity of 2-azidostyrenes. We were also intrigued by recent reports by the Shi-,⁴⁹ Hayashi-,⁵⁰ Lei-,⁵¹ Charette-,⁵² and Murphy⁵³ groups using tert-butylalkoxide to mediate single-electron transfer reactions of halide-substituted arenes and were delighted to see that exposure of 2-nitrostilbene 111 to two equivalents of sodium tert-butoxide mediated the formation of 2-phenyl-N-hydroxyindole 113.⁵⁴ In contrast to previous reports, which required elevated temperatures and an electron-shuttle additive, such as phenanthroline, nitrostilbenes reacted at room temperature and no additive was required.

Scheme 12.

Development of single electron transfer reactions for the reduction of nitroarenes.

To our surprise, we found that the identity of the tert-butoxide counterion changed the reaction outcome. When 2-nitrostilbene was exposed to potassium tert-butoxide, 3-hydroxy-3-phenylindolin-2-one 116 was formed instead of N-hydroxyindole 113 (Scheme 12).⁵⁴ Formation of oxindole 116 is striking because both oxygen atoms of the nitro-group apparently transferred to the ortho-alkenyl substituent. While reports of oxygen-atom transfer to pendent acetylenes had been reported,⁵⁵ this reactivity pattern was unknown for olefins. Examination of the scope of the transformation revealed that it tolerated a range of different substituents on either of the nitroarene- or styryl moiety of the substrate. Changing the identity of the β-substituent to an alkyl group, however, rendered the substrate inert. Our first hint into the mechanism of this unusual transformation originated from the observation that the reaction mixture became a deep red color upon addition of tert-butoxide. This color change prompted us to examine the reaction using EPR spectroscopy for radical intermediates, and the spectra was found to depend upon the identity of the counter ion. Reduction using NaOt-Bu exhibited a spectrum with a three-line pattern, which is a common feature of a radical interacting with a 15N-atom. Our collaborators Zadrozny and Sung performed simulations of the spectra and found the best match with radical anion 112, where the sodium ion is bound to both oxygen-atoms of the nitro-group.⁵⁴ In contrast, the spectrum resulting from KOt-Bu was significantly more complex with several additional hyperfine interactions visible. Simulation provided the best match with a 93:7 composite of radical anions 114 and 115, where the potassium ion is not coordinated to the nitroarene.

Additional insight into the mechanism of the N-heterocycle formation came from a series of control experiments (Scheme 13).⁵⁴ Coordination of the counterion as a controlling feature of the reaction outcome was validated by a control experiment: the addition of 15-crown-5 ether to the NaOt-Bu-mediated reduction of 111 changed the reaction outcome to afford only oxindole 116 instead of N-hydroxyindole 113. ¹⁸O-Labeling studies established that oxygen-atom transfer did not come from the tert-butoxide reductant but occurred intermolecularly between nitrostilbenes where the labeled oxygen was incorporated solely at the C2 position of the N-heterocycle. While the exact mechanism of oxindole remains murky, oxygen-atom transfer to the ortho-styryl substituent could occur through intermolecular radical addition to the ortho-alkenyl substituent to produce 119. Intramolecular cyclization-fragmentation could form an epoxide 120, which rearranges to 116. To test if an epoxide is formed as a reactive intermediate, stilbene oxide 120 was exposed to reaction conditions and oxindole 116 was formed.

Scheme 13.

Control experiments to provide insight into the mechanism.

5. Oxidation of aryl amines to access electrophilic N-aryl nitrenoids

While our investigations had established that similar cyclization-[1,2] migration tandem reactions could be teased out of nitroarenes or aryl azides, the reaction conditions required to unlock this reactivity out of either precursor was relatively harsh requiring elevated temperatures and superstoichiometric reductant. We recognized that the requisite reactive N-aryl nitrogen intermediate was formed from either the reduction of nitroarenes or through a redox neutral reaction of azides. The azide substrate, however, was generally synthesized from aniline through an oxidation reaction.⁵⁶ We were curious if we could directly access N-aryl nitrenes or N-aryl nitrenoids through the oxidation of 2-substituted anilines and if they would exhibit the same reactivity as metal N-aryl nitrenes or nitrosoarenes. While the oxidation of phenols is well established as a synthetic method,⁵⁷ the analogous oxidation of anilines is under developed.⁵⁸ Traditionally, the use of amines as nitrogen-atom sources has required a strong N-electron withdrawing group. Du Bois and co-workers established that metal N-aryl amines could be formed from sulfonamides or carbamates, such as 121, and reacted selectively with C–H bonds to form either six- or five-membered N-heterocycles, such as 124.^59,60 Similarly, Antonchick and co-workers reported that the oxidation of 2-biarylamines 125 to form carbazoles was successful only if an N-acetyl group was present.^58b,61 These successes piqued our curiosity if non-activated anilines could be oxidized to create an electrophilic N-aryl nitrenoid or nitrene and how its reactivity would compare to nitrosoarene or N-aryl metal nitrene generated from a nitroarene or an aryl azide respectively.

Towards this end, the reactivity of aniline 133 towards oxidants was examined (Scheme 15).⁶² We found exposure of 133 to iodine(III) oxidants triggered an oxidative cyclization-migration to produce 3H-indole 136. In contrast to our studies of aryl azides and nitroarenes, which required heating above 100 °C, the reaction conditions were significantly milder with smooth conversion to 136 observed below room temperature. The mild conditions resulted in higher yields and an expanded substrate scope. For example, we found that anilines containing an additional ortho-substituent could be transformed in high yield to 3H-indoles such as 136a, and that amine-substituted pyridines reacted smoothly to furnish N-heterocycles 136b and 136c. 4-Pyridone-derived substrates also reacted without concomitant oxidation of the nitrogen-substituent to afford N-heterocycles, such as 136d. While increased substitution and functional group tolerance was exhibited in comparison to the analogous aryl azides or nitrostyrenes, the diastereoselectivity of the process remained modest with a 1:1 mixture of 3H-indole diastereomers 136e obtained.

Scheme 15.

Iodine(III)-mediated oxidative cyclization-migration reactions to access non-planar N-heterocycles.

The migratorial aptitude of non-activated anilines was observed to be the same as that of aryl azides and nitroarenes (Scheme 15).⁶² Ring-contraction to afford spirocyclic N-heterocycles was observed for substrates that bore an aryl- or alkyl β-substituent (e.g. 137). Changing the β-substituent to an electron-deficient group such as a sulfone or carboxylate, however, reversed the reaction outcome to produce N-heterocycles, such as 140, where the electron-poor β-substituent migrated. The identity of the ortho-alkenyl had a significant impact on the migration. When substrates containing an acyclic ortho-substituent were exposed to the oxidative reaction conditions, a [1,2] aryl migration was observed instead of a [1,2] alkyl shift to afford 142. This preference for aryl migration was also observed when the reactivity of larger ortho-cycloalkenyl-substituted substrates were surveyed. Increasing the size of the ortho-substituent from cyclohexenyl- to cyclooctenyl- reversed the reaction outcome: instead of ring-contraction, a [1,2] phenyl shift was observed to produce only 145. Mixtures of products were observed from ortho-cycloheptenyl substituted substrates vide infra. While the origin of these selectivities is still under investigation, our working hypothesis is that they may be caused by conformational changes of the cyclic substituent in the competing migration transition states.

The reactivity of substrates containing strained ortho-cycloalkanol substituents were also investigated to compare to their aryl azide analogs (Scheme 16).⁶² Similar reactivity was observed: exposure of anilines 146 to an iodine(III) oxidant produced benzazepinones 149 through a ring-expansion-fragmentation reaction sequence. There were small differences in reactivity between 2-styryl-substituted anilines 133 and 2-cyclobutanol-substituted 146. While 133 required the addition of 2 equivalents of trifluoroacetic acid for a successful reaction outcome, benzazepinone formation required only the addition of 10 mol % of scandium triflate. Both the Brønsted acid and the Lewis acid catalyst were attributed to assist in the elimination of the iodine trifluoroacetate substituent to enable access to the electrophilic N-aryl iminoiodinane reactive intermediate. Once this intermediate is produced, ring-expansion produces spirocycle 147. Proton transfer followed by fragmentation results in aromatization to produce acylium ion 148, which is trapped by the nucleophilic amine to form benzazepinone 149. A series of substituted cyclobutanol-substituted aryl amines were surveyed, and the migratorial aptitude mirrored that observed in the analogous aryl azides with the more substituted carbon migrating stereospecifically to access 149a – 149d as the only product of the reaction. While no productive reactivity was observed in ortho-pentanol substituted substrates, ring-expansion-fragmentation-cyclization could be triggered from ortho-cyclopropanol-substituted anilines to access dihydroquinolin-2(1H)-ones, such as 149e. Examination of 150 revealed a key difference in the reactivity of the N-aryl nitrenoid intermediate. In contrast to the aryl azide, which upon exposure to a Rh₂(II)-catalyst resulted in sp³-C–H bond amination to produce a cyclobutane-fused indoline (63), submission of 150 to oxidative conditions did not result in any C–H bond amination. The lack of productive C–H bond amination is exciting because it indicates that the N-aryl nitrenoid is not reactive enough at room temperature to engage with the C–H bond. Comparing the harsh conditions required to access the electrophilic N-aryl nitrogen intermediate from an aryl azide or nitroarene, this suggests that catalysis of N-atom transfer from anilines could target the functionalization of the C–H bond as the turnover-limiting step instead of generation of the metal nitrene or metal nitrenoid.

Scheme 16.

Iodine(III)-mediated ring-expansion to afford benzazepinones.

Our first foray into developing catalytic oxidative amination processes was focused on determining if the iodine(III) oxidant could be generated in situ by using an organoiodine catalyst and stoichiometric oxidant (Scheme 17).⁶³ We found that only 1 mol % of PhI was necessary to construct 3H-indoles 136 from non-activated 2-substituted anilines 133 if Selectfluor was used as the stoichiometric oxidant. As before, an excess of trifluoroacetic acid was required. The success of this reaction depended on the identity of the solvent. While trifluoroethanol was identified as the best reaction medium for the stoichiometric reaction, poor yields were obtained when this solvent was used in the catalytic variant. Instead, a 10:1 mixture of hexafluoroisopropanol and water was found to the best. The water co-solvent was critical for reproducibility and could be increased to 50% without attenuating the yield of the reaction. Using these conditions, the scope was wider than that of the stoichiometric reaction. In addition to anilines bearing a 6-substituent or aminopyridines, heteroaromatic amines could be employed as substrates. For example, thiophenes, such as 133f or 133g participated in the oxidative cyclization to produce 136f or 136g. The subsequent migration, however, was not observed for these substrates. Because ring-contraction and [1,2] aryl migration were found to be competitive in ortho-cycloheptenyl substituted anilines 153, the electronic nature of the aryl group was varied to determine its effect on the ratio of N-heterocycles 154 and 155. Our investigations established that the migratorial preference was largely the same between the stoichiometric and catalytic conditions: increasing the electronic nature of the aryl group led to higher percentage of 155 obtained.

Scheme 17.

Development of RI-catalyzed oxidative cyclization-migration reactions to construct 3H-indoles.

The effect of changing the identity of the organoiodine catalyst on the diastereselectivity of the oxidative cyclization-migration reaction was examined (Scheme 18).⁶³ While no stereoselectivity was observed in homoallylic substituted substrates (e.g. 133e), moving the substituent to the allylic position did result in a 79:21 mixture of 3H-indoles 157 and 158. To our surprise, increasing the size of the catalyst reduced the diastereoselectivity to 48:52. Reducing the size of the catalyst to MeI resulted in a slight increase of the selectivity, which could be increased further when the reaction temperature was reduced to 0 °C and trimethylsilyltrifluoroacetate was used in place of trifluoroacetic acid. The dependence of the diastereoselectivity of the process on the structure of the catalyst suggests that RI is present during the stereochemical determining cyclization step of the mechanism. The inverse correlation to the size of RI suggests that elimination of the iodine catalyst is a competing process and that cyclization of the N-aryl nitrene (or nitrenium ion) via TS-160 is not selective. While the details of this mechanism are the subject of ongoing investigations, the effect of the structure of RI on the stereoselectivity is promising because enantioselective catalysis might be achieved through catalyst design.

Scheme 18.

Effect of changing the identity of the organoiodine catalyst on the diastereoselectivity.

6. Conclusion

Over the past 16 years, our group has been focused on the development of new reactions to access electrophilic N-aryl nitrogen reactive intermediates and exploit their reactivity in the construction of C–NAr bonds through cyclization-migration tandem reactions. We have discovered that transition metal catalysts can unlock the reactivity embedded in 2-substituted aryl azides to create functionalized N-heterocycles through the synthesis of multiple bonds. We have shown that similar reactivity patterns can be accessed from nitroarenes using a combination of a metal catalyst and a stoichiometric reductant. We showed that radical intermediates can be generated from nitroarenes using tert-butoxide as the single electron reductant, and that these intermediates also engage in cyclization-migration reactions to furnish 3-substituted oxindoles. The oxidation of aryl amines was also shown to mildly and at low temperature generate electrophilic N-aryl nitrenoids that also engage in similar cyclization-migration reactions. Accessing these reactive N-aryl nitrogen intermediates under mild conditions will enable the discovery of new selective reactions to construct C–NAr bonds.

Scheme 14.

Investigating if N-aryl nitrogen intermediates from anilines exhibit similar reactivity to aryl azides and nitrostyrenes.

Biography

Tom G. Driver is a Professor of Chemistry. He obtained his B.S. in Chemistry from Indiana University, Bloomington in 1999, and his Ph.D. from the University of California, Irvine under the mentorship of K. A. Woerpel in 2004. After an NIH-funded postdoctoral position at Caltech with John E. Bercaw and Jay A. Labinger, he began his independent academic career at the University of Illinois at Chicago in 2006. His group’s research program is centered on the development of new reactions that exploit electrophilic N-aryl nitrogen intermediates for the construction of C–N and C–C bonds and the use of these methods in medicinal chemistry applications. He finds inspiration and serenity racing his bicycles.