|Home | About | Journals | Submit | Contact Us | Français|
The use of a rhodium(II) carboxylate catalyst enables the mild and stereoselective formation of carbazoles from biaryl azides. Intramolecular competition experiments of triaryl azides suggested the source of the selectivity. A primary intramolecular kinetic isotope effect was not observed and correlation of the product ratios with Hammett σ+-values produced a plot with two intersecting lines with opposite ρ-values. These data suggest that electronic donation by the biaryl π-system accelerates the formation of rhodium nitrenoid and that C–N bond formation occurs through a 4π-electron-5-atom electrocyclization.
The development of efficient methods for the selective transformation of carbon–hydrogen bonds into carbon–heteroatom bonds remains a goal of synthetic chemistry.1 The direct incorporation of functionality into carbon–hydrogen bonds improves synthetic efficiency2 by minimizing functional group manipulation.1f,3 Dirhodium(II) complexes have proven to be efficient C–H bond amination catalysts by stabilizing the nitrene intermediate.1e,4–9 Our group has used these complexes to catalyze the intramolecular formation of C–N bonds from vinyl- or aryl azides to form indoles, pyrroles, and carbazoles.10 Use of the rhodium catalyst provides a mild and stereoselective alternative to the thermal-or photochemical reaction (Scheme 1).11 In this paper, we report mechanistic studies aimed at determining the role of the catalyst in promoting and controlling this regioselective transformation. Our results enable distinction between a concerted aryl C–H bond insertion, electrophilic aromatic substitution, and electrocyclization of the rhodium nitrenoid as the mechanism for C–N bond formation.
In contrast to the thorough studies on the mechanism of thermal or photolytic nitrene formation from ortho-biaryl azides,11e,12,13 mechanistic studies on metal nitrenoids generated from aryl azides are less common.14 The improved regioselectivity observed when biaryl azide 1 was treated with rhodium(II) octanoate (as compared to thermolysis) reveals that the rhodium catalyst is involved in the C–N bond forming step of the mechanism. The analogous reactivity of the E- or Z-stilbene isomer of 4 towards rhodium(II) octanoate indicated that the functionalization of the C–H bond occurs by a stepwise mechanism (Scheme 2).10b This behavior contrasts with the metal-free pyrolysis of 4, where the yield of 2-phenylindole 5 depended on the stereochemistry of the styryl azide (88% from E-4; 18% from Z-4).15 These results illustrate the differences between a rhodium nitrenoid and an arylnitrene, and suggest that arylnitrenes might not be the best models for rhodium arylnitrenoids.
Instead, we anticipated that the rhodium arylnitrenoid might be more similar to an arylnitrenium ion.16,17 The singlet ground state of an arylnitrenium ion contains extensive delocalization of the positive charge into the adjacent π-system (Scheme 3).17d,18,19 If rhodium nitrenoid 8 behaved similarly to 7, the electronic nature of the aryl substituents would influence the mechanism of N-heterocycle formation through stabilization of the quinoid resonance structure 9. Herein, we report the results of our mechanistic studies, which support this assertion. Our data suggest that electronic donation by the biaryl π-system accelerates rhodium nitrenoid formation from 10 and that C–N bond formation occurs through a 4π-electron-5-atom electrocyclization of 11.
Triaryl azides 15 were chosen to study the rhodium-catalyzed N-heterocycle formation. These substrates allowed the investigation of an intramolecular competition reaction between the reactive intermediate and the C–H bond on either the phenyl- or aryl substituent. This was necessary because the rate-determining step of the mechanism was unknown,20,21 and the intramolecular reaction would allow the ratio of carbazole products obtained from 15 to offer insight into the mechanism of the C–N bond-forming event even if it occurred after the turnover-limiting step. The triaryl azides that were used for the intramolecular competition experiments were synthesized from dibromoaniline 13 by two consecutive Suzuki cross-coupling reactions (Scheme 4).22 Carbazoles 17 were also independently synthesized from carbazole 16 through a Suzuki cross-coupling reaction to verify the product ratio. Triaryl azides 15 would be used first to determine if C–H bond cleavage and C–N bond formation occurred simultaneously and then to investigate the electronic effect of substituents on the ortho-aryl group.
Triaryl azide 15a–d5 was used to determine if the reaction between the rhodium nitrenoid and the ortho-aryl C–H bond was concerted (Scheme 5).23,24 If C–N bond formation and C–H bond cleavage occurred simultaneously (via 20),24 a primary kinetic isotope effect would result from preferential reaction of the reactive intermediate with the unlabeled phenyl substituent.25,26 Exposure of 15a-d5 to rhodium(II) perfluorobutyrate at 70 °C, however, gave a kinetic isotope effect of 1.01. A similar isotope effect was observed in 21-d1.10a These experiments show that C–H bond cleavage occurs after the product-determining step of the reaction.27,28 Together with the equal reactivity of Z- and E-4, these results confirm that a concerted insertion of the nitrenoid into the ortho-aryl C–H bond is not occurring to form either carbazole- or indole-products.
A series of triaryl azides with electron-donating or electron-withdrawing aryl substituents were screened as substrates to determine the electronic dependence of the C–H bond functionalization (Table 1). 4-Substituted aryl groups were chosen to simplify product analysis by providing only one regioisomer. The results of these experiments are given in Table 1 and show that C–N bond formation occurs preferentially with the more electron-rich aryl group.
The product ratios from the reaction of 15 were analyzed using the Hammett equation to determine if an electrophilic aromatic substitution produced the C–N bond (Scheme 6).29 In this mechanism, nucleophilic attack of the pendant aryl- or phenyl group onto the electrophilic rhodium nitrenoid 22 would form arenium ion 23 or 24. If this mechanism occurs, a linear correlation of the product ratios from triaryl azides 15 with Hammett σm-values should be observed because the C–N bond is forming meta to the R group.30,31 Figure 1 clearly shows that the relationship between σm-values and the product ratios from 15 is not linear. We interpret this result to mean that C–N bond formation does not occur by electrophilic aromatic substitution.32 This conclusion is supported by the reactivity patterns of aryl- and vinyl azides (40, 42, and 43), which will be discussed below (eq 2 and Scheme 9).
Alternatively, the aryl substituents could affect the formation of rhodium nitrenoid by controlling the amount of electron density in the π-system (Scheme 7). Overlap of the π-system with σ*N–N2 best occurs when the N2-leaving group is oriented orthogonal to the triaryl group (31). If all three arenes are planar, a destabilizing steric interaction occurs between the rhodium catalyst and an ortho-aryl hydrogen (32). As a result, the more electron-deficient aryl group would rotate out of the plane to form 27 or 28. Expulsion of N2 then occurs to form N-rhodiumimine 29 or 30.33,34 If the electronic nature of the R group affects the rate of nitrenoid formation, linear correlation of the product ratios from 15 with Hammett σ+-values (or σp-values) would be expected because the reaction is occurring para to the R substituent. Because electron density is flowing away from the arene, a ρ-value less than zero should be observed.
The product ratios from 15 were plotted against Hammett σ+-values29c,35 to test this mechanistic hypothesis (Figure 2).36 Both rhodium perfluorobutyrate and rhodium octanoate gave plots that contained two intersecting lines (Figure 2). As predicted by our hypothesis, a negative ρ-value was obtained from triaryl azides 15b – 15f. Triaryl azides with stronger electron-withdrawing groups (15f – 15h), however, diverged from this trend to provide a line with a positive ρ-value. The V-shaped Hammett plot suggests a change in mechanism (or rate-determining step) because the two lines exhibit opposite ρ-values.37 We did not observe this V-shape for the metal-free thermolysis of triaryl azides 15, which gave a single line with a negative ρ-value (–0.66; Figure 3).31,38,39
The V-shape of the Hammett plot suggests that at least two mechanisms are occurring in the rhodium-catalyzed formation of carbazoles.37 The negative ρ-value observed for triaryl azides 15b – 15g reveals that electron density is leaving the aryl π-system in the product-determining step. This result is best accommodated by a mechanism where the R substituent assists in the extrusion of N2 from the initial metal-azide complex to form quinoid 35 (Scheme 8). In support of the potential quinoidal structure of 35, time-resolved infared spectroscopy of related N-aryl-N-methyl nitrenium ions revealed that the frequency of the C=C bond stretch increased in proportion to the electron-donating ability of the aryl-substituent.40 Falvey and co-workers interpreted these results as evidence that arylnitrenium ions can be described as 4-iminocyclohexa-2,5-dienyl cations.40
Electronic donation by the biaryl π-system to form the rhodium nitrenoid influences the mechanism of C–N bond formation (Scheme 8). The planar nature of quinoid 35 enables a 4π-electron-5-atom electrocyclization to form the C–N bond.28,41,42 This pericyclic reaction can be visualized easier by examining 36, the resonance form of 35, which clearly contains a contiguous, planar array of π-orbitals.43 Upon formation of 38, a 1,5-hydride shift then provides carbazole 39.44–46
Aryl azide 40, which contains a methylene spacer in between the two arenes, was examined to test the requirement for a contiguous array of π-orbitals (eq 2). Exposure of 40 to reaction conditions did not produce dihydroacridine 41. Instead, >95% of azide 40 was recovered. The lack of reactivity of 40 agrees with our proposed electrocyclization mechanism and is inconsistent with an electrophilic aromatic substitution mechanism, which would not require a contiguous π-system.
The electrocyclization mechanism also offers an explanation for the unusual reactivity trends that we observed with methoxy-substituted azidoacrylates (Scheme 9). In azide 42, the para-methoxy-substituent is positioned to donate electron density to accelerate the formation of the nitrenoid species to result in the efficient formation of indole 44. In contrast, placing the methoxy group at the 3-position of vinyl azide causes it to act as an inductive electron-withdrawing group and slow nitrenoid formation from 45: only 22% of azide 45 was converted into indole 47 with 5 mol % of catalyst. The remainder of the mixture was unreacted vinyl azide 45. If an electrophilic aromatic substitution mechanism occurred to form indole, azide 45 would be expected to react faster because the methoxy group is positioned to donate electron density to attack the electron-deficient rhodium nitrenoid (48).
The V-shape of the Hammett plot indicates that a change in mechanism occurs for substrates bearing strongly electron-withdrawing groups (σ+ ≥ 0.6). The positive ρ-value for this portion of the plot reveals that the aryl group accepts electron density in the product-determining step. Changing the identity of the product-determining step to the electrocyclization or hydride shift appears unlikely, however, because it would require a nucleophilic rhodium(II) nitrenoid species47,48,49 and a reversible N2-extrusion step.21 Alternatively, a different mechanism could be operating for these substrates. Coordination of the rhodium carboxylate complex to the nitro- or sulfone group might cause the catalytic system to deviate from the thermal behavior of these substrates (Scheme 10).50
Triaryl azide 15i (X = NMe3) was used to investigate if the rhodium carboxylate were functioning as a Lewis acid through σ-coordination to the X substituent (eq 3).50,51 If this phenomenon was occurring, the product ratio from 15i would not correlate linearly with substrates 15f – 15h because coordination to the trimethylammonium substituent is not possible. The product ratio from 15i, however, does correlate linearly with the azides 15f – 15h using Hammett σp-values. This relationship reveals that σ-coordination of the R substituent to the rhodium(II) carboxylate is not occurring in the mechanism for substrates bearing strongly electron-withdrawing groups (Figure 4).
The mechanism of N-heterocycle formation from aryl- and vinyl azides was examined through intramolecular competition experiments. The lack of a primary intramolecular kinetic isotope effect suggests that the C–H(D) bond cleavage does not occur in the product-determining step. Correlation of the product ratios obtained from a series of substituted triaryl azides with the Hammett equation generated plots with two intersecting lines. The best linear correlation was obtained when the methoxy substituent was considered to be an electron-donating group. We interpreted these results as evidence for two different mechanisms. Electron-rich substrates appear to react through a mechanism in which electronic donation by an aryl R substituent assists in the formation of a planar arylnitrenoid that undergoes a 4π-electron-5-atom electrocyclization to form the new C–N bond. In agreement with our mechanistic hypothesis, substrates that lack a contiguous π-system are unreactive. While our experimental results prevent definitive conclusions for the mechanism operating in electron-deficient substrates, our data does suggest that the rhodium carboxylate does not σ-coordinate to the R group. We believe that these studies provide new models for azide reactivity toward rhodium carboxylates, and that the mechanistic insight gained herein will guide future methodology development that builds on our understanding of the interactions of metal complexes and azides.
In a dry 500 mL round bottom flask, phenylboronic acid (5.00 g, 43.0 mmol, 1.3 equiv), K2CO3 (21.50 g, 156 mmol, 4.0 equiv), and Pd(PPh3)4 (2.00 g, 1.73 mmol, 0.1 equiv) were then dissolved in 150 mL of toluene, 100 mL of H2O, and 50 mL of EtOH. 2,6-Dibromo-4-methylaniline (10.48 g, 39.8 mmol, 1.0 equiv) was added, and the resulting mixture was heated to 95 °C for 16 hours. After cooling, the biphasic solution was diluted with 100 mL of saturated aqueous NH4Cl and 100 mL of CH2Cl2 and separated. The aqueous phase was extracted with an additional 2 × 100 mL of CH2Cl2, and the combined organic phases were washed 1 × 100 mL of water and 1 × 100 mL of saturated aqueous NaHCO3. The organic phase was dried over Na2SO4 and filtered. The filtrate was concentrated in vacuo to afford a brown oil. Purification by MPLC (15:85 benzene:hexanes) afforded 2-bromo-4-methyl-6-phenylaniline as a white powder (4.90 g, 47%), mp 58 °C, Rf = 0.31 (15:85 benzene:hexanes, visualized by 254 nm UV light). 1H NMR (CDCl3, 500 MHz): δ 7.50-7.45 (m, 4H), 7.42-7.38 (m, 1H), 7.31 (s, 1H), 6.93 (s, 1H), 4.04 (s, 2H), 2.30 (s, 3H); 13C NMR (CDCl3, 125 MHz): 139.3 (C), 139.0 (C), 132.1 (CH), 130.4 (CH), 129.1 (CH), 128.9 (CH), 128.7 (C), 128.5 (C), 127.7 (CH), 109.9 (C), 20.2 (CH3); IR (thin film): 3477, 3385, 3045, 3028, 2924, 1619, 1593, 1473, 1441, 1302, 1234, 1073, 1058, 861, 638, 575 cm−1. HRMS (EI) m / z calculated for C13H12NBr (M+): 261.0153, found: 261.0154.
To a dry 100 mL round bottom flask equipped with a stir bar were added 2-bromo-4-methyl-6-phenylaniline (0.500 g, 1.90 mmol, 1.0 equiv), 4-chlorophenylboronic acid (0.343 g, 2.09 mmol, 1.3 equiv), K2CO3 (1.050 g, 7.63 mmol, 4.0 equiv), and Pd(PPh3)4 (0.110 g, 0.950 mmol, 0.1 equiv). Toluene (30 mL), 20 mL of H2O, and 10 mL of EtOH were added and the resulting mixture was heated to 95 °C for 16 hours. After cooling, the biphasic solution was diluted with 100 mL of saturated aqueous NH4Cl and 100 mL of CH2Cl2 and separated. The organic phase was washed 1 × 100 mL of water and 1 × 100 mL of saturated aqueous NaHCO3. The organic phase was dried over Na2SO4 and filtered. The filtrate was concentrated in vacuo to afford a brown oil. Purification by MPLC (0:10:90 – 10:10:80 EtOAc:benzene:hexanes) afforded 2-(4-chlorophenyl)-4-methyl-6-phenylaniline as a white powder (0.463 g, 83%), mp 101 °C, Rf = 0.56 (10:10:80 benzene:EtOAc:hexanes, visualized by 254 nm or 365 nm UV light). 1H NMR (CDCl3, 500 MHz): δ 7.55-7.45 (m, 8H), 7.42-7.39 (m, 1H), 7.03 (d, J = 1.6 Hz, 1H), 6.98 (d, J = 1.6 Hz, 1H), 3.67 (br s, 2H), 2.36 (s, 3H); 13C NMR (CDCl3, 125 MHz): δ 139.7 (C), 138.4 (C), 138.2 (C), 133.2 (C), 130.8 (CH), 130.7 (CH), 130.3 (CH), 129.4 (CH), 129.0 (CH), 128.9 (CH), 128.4 (C), 127.6 (C), 127.4 (CH), 126.9 (C), 20.5 (CH3); IR (thin film): 3455, 3386, 3053, 2985, 2923, 2859, 2305, 1615, 1599, 1492, 1466, 1438, 1392, 1265, 1092, 1015, 896, 872, 837, 749, 704 cm−1. HRMS (EI) m / z calculated for C19H16NCl (M+): 293.0971, found: 293.0973.
In a 20 mL scintillation vial, 2-(4-methoxyphenyl)-4-methyl-6-phenylaniline (0.076 g, 0.253 mmol, 1.0 equiv) was dissolved in 3 mL of HOAc and chilled in an ice bath. NaNO2 (0.026 g, 0.367 mmol, 1.45 equiv) was added slowly, and the resulting mixture was stirred at 0 °C for one hour. NaN3 (0.393 g, 0.393 mmol, 1.55 equiv) was then added slowly, and the resulting mixture was warmed to ambient temperature, and stirred for 30 minutes. The solution was diluted with 20 mL of water and 20 mL of CH2Cl2 and basified by the slow addition of K2CO3 until bubbling ceased. The phases were separated and the aqueous phase was extracted with an additional 2 × 20 mL of CH2Cl2. The combined organic phases were washed 1 × 20 mL of water and 1 × 20 mL of brine. The resulting organic phase and dried over Na2SO4 and filtered. The filtrate was concentrated in vacuo to afford an oil. Purification by MPLC (100% hexanes) afforded azide 15b as a faint yellow oil (0.080 g, 97%), Rf = 0.55 (10:10:80 benzene:EtOAc:hexanes, visualized by 254 nm UV light). lH NMR (CDCl3, 500 MHz): δ 7.53-7.52 (m, 2H), 7.49-7.45 (m, 4H), 7.42-7.39 (m, 1H), 7.13 (m, 2H), 7.03–7.00 (m, 2H), 3.88 (s, 3H), 2.42 (s, 3H); 13C NMR (CDCl3, 125 MHz): δ 159.2 (C), 138.7 (C), 136.4 (C), 136.1 (C), 135.2 (C), 132.1 (C), 131.1 (CH), 131.0 (C), 130.8 (CH), 130.5 (CH), 129.4 (CH), 128.4 (CH), 127.6 (CH), 113.9 (CH), 55.4 (CH3), 20.9 (CH3); IR (thin film): 2122, 2093, 1513, 1455, 1422, 1290, 1265, 1179, 1033, 896 cm−1. HRMS (EI) m / z calculated for C20H17ON3 (M+): 315.13717, found: 315.13679.
We are grateful to the National Institutes of Health NIGMS (R01GM084945), Petroleum Research Fund administered by the American Chemical Society (46850-G1), and the University of Illinois at Chicago for their generous support. We thank Professors Martin E. Newcomb (UIC), Laura L. Anderson (UIC), Douglass Taber (U Delaware), Hélène Lebel (U Montreal) for helpful discussions. We also thank Dr. Dan McElheny (UIC) for assistance with NMR spectroscopy, and Mr. Furong Sun (UIUC) for mass spectrometry data. The 70-VSE mass spectrometer (UIUC) was purchased in part with a grant from the Division of Research Resources, National Institutes of Health (RR 04648).
BRIEFS. The mechanism of dirhodium(II)-catalyzed C—H bond amination was examined using intramolecular competition reactions of triaryl azides.
Supporting Information Available: Complete experimental procedures, spectroscopic and analytical data for the products (PDF) is available free of charge via the Internet at pubs.acs.org.