Report

Dirhodium-catalyzed C-H arene amination using hydroxylamines

See allHide authors and affiliations

Science  09 Sep 2016:
Vol. 353, Issue 6304, pp. 1144-1147
DOI: 10.1126/science.aaf8713

Abstract

Primary and N-alkyl arylamine motifs are key functional groups in pharmaceuticals, agrochemicals, and functional materials, as well as in bioactive natural products. However, there is a dearth of generally applicable methods for the direct replacement of aryl hydrogens with NH2/NH(alkyl) moieties. Here, we present a mild dirhodium-catalyzed C-H amination for conversion of structurally diverse monocyclic and fused aromatics to the corresponding primary and N-alkyl arylamines using NH2/NH(alkyl)-O-(sulfonyl)hydroxylamines as aminating agents; the relatively weak RSO2O-N bond functions as an internal oxidant. The methodology is operationally simple, scalable, and fast at or below ambient temperature, furnishing arylamines in moderate-to-good yields and with good regioselectivity. It can be readily extended to the synthesis of fused N-heterocycles.

Arylamine motifs (1, 2) are prevalent in natural products, pharmaceuticals, agrochemicals, functional materials, and dyestuffs, as well as in many synthetic reagents and catalysts (35). Most traditional methods for their synthesis from a corresponding C(sp2)–H bond involve multistep sequences and/or harsh conditions (68). In many instances, the initial product (nitro, azide, azo, nitroso, chloronitroso, amide/imide/sulfonamide, or imine) requires one or more additional steps before arriving at a free amine (6). The amination of organic anions is a useful and direct alternative (9), but is generally restricted to the introduction of NR1R2 (R1, R2 ≠ H) and is not applicable to base-sensitive substrates. The seminal studies of Ullmann and Goldberg into metal-mediated arene C(sp2)-N bond formation at the beginning of the 20th century were harbingers (10) of more efficient palladium-catalyzed cross-couplings between aromatic halides/sulfonates and ammonia or nitrogen surrogates developed independently by Hartwig (11) and Buchwald (12) in the early 1990s. In turn, a steady succession of technical improvements, up to the present time, has been offered with the aim of mitigating the original stringent Hartwig-Buchwald reaction conditions (1316). All of these procedures, however, are restricted by the need for a prefunctionalized arene.

In more recent years, much attention has been paid to catalytic, direct arene aminations (1719). Many elegant atom- and step-efficient protocols have been introduced, but virtually all require a directing group, while others need an excess of arene, require high temperatures, or generate amides, imides, and sulfonamides instead of free amines (2031). A noteworthy exception is the photoredox-driven process of Nicewicz (32) and colleagues that circumvents many of the preceding limitations.

Extensive density functional theory (DFT) calculations of the dirhodium-catalyzed olefin aziridination reported previously by our laboratories (33) suggested that the reactive rhodium-nitrene complex might exist in a protonated form in the presence of a weakly basic counterion and that this species could follow a different reaction manifold. This was corroborated by analysis of minor by-products from the aziridination of different substrates and the behavior of structurally varied nitrogen donors. After a systematic investigation, proof-of-principle for arene amination versus aziridination was demonstrated (Fig. 1). Treatment of 1-allyl-2-methoxybenzene (o-allylanisole, 1) with O-(2,4-dinitrophenyl)-hydroxylamine (2: DPH) (34) and catalytic Rh2(esp)2 (Du Bois’ catalyst, 6: 5 mol %) in 2,2,2-trifluoroethanol (TFE) at 0°C afforded the corresponding aziridine (3) as the sole product, whereas use of N-methyl-O-tosylhydroxylamine (4a) (35) as the aminating agent under otherwise identical conditions furnished exclusively the arene amination adduct 3-allyl-4-methoxy-N-methylaniline (5). Consequently, a more detailed study of the C-H arene amination was begun and ultimately led to a versatile, regioselective, operationally simple and mild dirhodium-catalyzed direct arene amination suitable for both intermolecular and intramolecular applications. The relatively weak N–O bond of the aminating agents, many of which are commercial or readily prepared (3337), functions as an internal oxidant when cleaved during the catalytic amination process, thus obviating the need for addition of an external oxidant to the reaction mixture. Also, the synchronous release of a stoichiometric amount of arylsulfonic acid during the amination is critical because it protonates the arylamines, which would otherwise inactivate the Rh catalyst. In some instances, acid-sensitive functionality such as epoxides and acetals do not survive.

Fig. 1 A marked shift in chemoselectivity arises with different aminating agents.

o-Allylanisole (1) undergoes chemoselective olefin aziridination in the presence of catalytic Rh2(esp)2 (6) and aminating agent DPH (2) in 2,2,2-trifluoroethanol (TFE = CF3CH2OH) to give 3. Changing the aminating agent to TsONHMe (4a) furnishes the corresponding N-Me aniline (5).

Mesitylene (7a) was selected as a model arene to optimize reaction parameters. Use of 2 mol % Rh2(esp)2 and 1.5 equivalents of aminating agent 4a in TFE as solvent smoothly generated N-methylaniline 8a (75%) in just 30 min at 0°C (Fig. 2, entry 1); comparable yields were obtained with 1 and 0.5 mol % catalyst, except that the latter required 1 hour for complete consumption of starting material. Combinations of TFE with CH2Cl2 or MeOH in ratios up to 1:1 were suitable, but aminations run in EtOH or MeOH as the only solvent delivered poor yields (~10 to 20%) and those run in only DMF, CH3CN, THF, toluene, and dioxane failed. Among alternative Rh-catalysts, Rh2(OAc)4 and Rh2(C8H15O2)4 were the next best with moderate yields (~40 to 45%), whereas the performance by Rh2(TFA)4 was poor. Results from other catalysts are summarized in table S1. Generally, reactions were quenched as soon as the substrate was completely consumed to minimize by-product formation. Air and water (5% v/v) were well tolerated, making this methodology operationally convenient.

Fig. 2 Direct intermolecular amination of arenes.

Reactions were conducted on a 0.1- to 0.5-mmole scale at 0.1 M using 2,2,2-trifluoroethanol (TFE = CF3CH2OH) as solvent or using mixtures of TFE and other solvents as indicated.

The successful transfer of a nBuNH- unit to 7a from TsONHnBu (4b), affording 8b in 52% yield (Fig. 2, entry 2), suggested that more complex amino moieties should also be suitable. For the introduction of NH2 into 7a, we used 2,4,6-Me3C6H2S(O)2ONH2 (36) (4c, 2 equiv.) as aminating agent because of its greater stability at room temperature compared with TsONH2 (37). Although all of 4c was consumed, a considerable amount of unreacted 7a was recovered and a modest yield (49%) of 8c was realized (entry 3). By contrast, the simple and expedient generation of the aminating agent in situ from 2,4,6-Me3C6H2S(O)2ONHtBoc (4d, 1.5 equiv.) and CF3CO2H (2 equiv.) proved quite effective and boosted the yield of 8c to 70% (entry 4), although the reaction was much slower than in the case of aminating agent 4c (entry 3) and appeared to be dependent on the rate of tBoc cleavage.

The results from other representative arenes are also summarized in Fig. 2. Despite having only a single aliphatic substituent, cyclopropylbenzene (7d) was well behaved, affording a 1:1 para-/ortho-mixture (8d and 8e) in 20 min at 0°C with no evidence of addition to the strained three-membered ring (entry 5). The directing effect in favor of the para-isomer was more pronounced with anisole (7f) and led to a 16:1 distribution of 8f and 8g (entry 6). This effect completely dominated in veratrole (7h) that gave 8h as the predominant regioisomer (entry 7) and was equally evident in the conversion of 1,3-dimethyoxybenzene (7i) into 8i, although a minor amount (6%) of the ortho-isomer 8j was found despite the increased steric congestion at this site (entry 8). Addition of acetic acid to the reaction solvent for the latter three examples (entries 6 to 8) improved the yields and suppressed the formation of mauve-colored by-products that we assume arose from further oxidation of the aminated adducts. The ortho/para-directing effect of electron-donating substituents is typical for electrophilic aromatic additions; the preference for the para- versus ortho-isomer might reflect the steric demand of the Rh-nitrenoid complex. Consistent with these results, aromatics with only electron-withdrawing substituents (e.g., CF3 and CN) fail to aminate under these reaction conditions.

Aryl bromide 7k (entry 9) and terminal alkyl bromide 7l (entry 10) endured to deliver 8k and a mixture of 8l/8m (3:1), respectively, leaving the halogens available for further manipulation, if desired. The presence of some functional groups, however, partially retard amination; e.g., exposure of benzyl alcohol 7n to the standard amination conditions generated a modest amount of 8n (42%) after 2 hours (entry 11); by contrast, protection as silyl ether 7o resulted in a much improved yield of 8o (68%) (entry 12). Fortunately, a synthetically useful yield of amine could be obtained even in the presence of an electron-withdrawing substituent, e.g., 7p8p (entry 13). The amination process was also extended to fused aromatics: Naphthalene (7q) and 1,4-dimethylnapthalene (7r) gave rise to N-methyl-1-naphthylamine (8q, entry 14) and N-methyl-1,4-dimethyl-2-naphthylamine (8r, entry 15), respectively, and 2-methoxynaphthalene (7s) readily underwent amination to give 8s in 85% yield on a 0.5-mmol scale and in 80% yield on a 5-mmol scale (entry 16). The smooth amination of 2-methoxy-6-bromonaphthalene (7t) into N-methyl-2-methoxy-6-bromo-1-naphthylamine (8t, entry 17) succinctly illustrates the chemo- and regioselectivities of this methodology.

To validate the suitability of this methodology for late-stage functionalization of complex molecules, O-methylestrone (7u) was smoothly converted to a mixture of 8u and 8v (1:7.2, entry 18) in very good combined yield, whereas only one regioisomer 8w (entry 19) was obtained from the parent phenol, estrone (7w), thus demonstrating applicability to substrates containing potentially sensitive benzylic, tertiary, and α-keto hydrogens. The scope was further defined with the amination of methyl naproxen 7x to 8x (entry 20) and morphine derivative 7y to 8y (entry 21). The latter required the addition of p-toluenesulfonic acid (2.5 equiv.) to ensure that the tertiary amine was fully protonated to avoid passification of the catalyst.

The intramolecular version of the amination proved quite facile and represents one of the mildest and most efficient aza-annulations currently available to the practitioner (Fig. 3) (38, 39). The enabling amination functionality was introduced in uniformly good yields via the Mitsunobu inversion under standard conditions from the corresponding alcohols. Acidic cleavage of the tBoc and subsequent in situ amination proceeded at 0°C. Fused aza-annulated bicycles were created in good yields irrespective of additional ring substituents, 10a11a (entry 1), electron-donating, 10b11b (entry 2), or electron-withdrawing functionality, 10c11c (entry 3) and 10d11d (entry 4). Control studies of the reaction of 10b confirmed that there was less than 2% of the ortho-annulation product, 8-methoxy-1,2,3,4-tetrahydroquinoline, based on 13C–nuclear magnetic resonance analysis of the crude reaction product (figs. S5 and S6). Secondary alcohols likewise participated readily in the Mitsunobu and aza-annulation reactions, 9e10e11e (entry 5). This two-step sequence, when conducted using the chiral alcohol 9f, showed no loss of stereochemical integrity (entry 6). Incorporation of a heteroatom into the ring closure, e.g., 9g10g11g (entry 7), did not perturb the chemistry and provided easy access to the dihydrobenzoxazine class of heterocycles. The yield declined somewhat for making the five-membered dihydroindole 11h from alcohol 9h (entry 8), but improved for the seven-membered tetrahydrobenzazepine 11i from 9i (entry 9).

Fig. 3 Direct intramolecular cyclization (aza-annulation) of arenes.

Cyclizations were conducted at 0.1 M using 2,2,2-trifluoroethanol (TFE = CF3CH2OH) as solvent on a 0.1- to 0.3-mmol scale unless otherwise indicated.

As a beginning toward gaining insight into the mechanism of the amination, a 1:1 mixture of naphthalene (7q) and 7q-d8 was treated with a limited amount of amination reagent (4a, 0.5 equiv.) under otherwise standard reaction conditions. Samples were taken and quenched at 10, 20, 30, and 40 min. Analysis via selected ion monitoring–liquid chromatography–mass spectrometry revealed that the product ratios remained constant at ~1:1, a ratio inconsistent with an organometallic C-H activation pathway, which would typically manifest ~3:1 or higher ratios (40, 41). Based on DFT calculations, we previously suggested that aziridination of alkenes involves the dirhodium-nitrenoid intermediate B shown in Fig. 4 that arises from overall NH transfer from the DPH-aminating reagent to the dirhodium catalyst (33). In contrast, reaction of O-tosylhydroxylamine reagents with the dirhodium catalyst favor intermediate A because TsO is weakly basic and the equilibrium with intermediate B lies far to the left. The chemoselectivity might be explained by the more electrophilic nature of intermediate A versus nitrenoid B. This preliminary hypothesis is consistent with the observation that moderate-to-strong bases such as K2CO3, Et3N, and pyridine completely inhibit amination, but not aziridination. Moreover, addition of TsOH (1.5 equiv.) to the reaction of 1 with 2,4-DNPONHMe (12) produced only the arene amination adduct 5 and no aziridine. As an additional control, it was shown that the presence of 2,4-DNP-OH (1.5 equiv.) did not alter the reaction manifold in favor of aziridination when 4a was used as the aminating reagent and only 5 was observed.

Fig. 4 Proposed intermediates leading to amination versus aziridination and control study of arene amination versus benzylic C-H insertion.

It was also instructive to compare our methodology with the intermolecular Rh-catalyzed amination procedure of Du Bois to gain a perspective on their respective complementary chemoselectivities (Fig. 4) (42). Both have similar efficiency using p-ethylanisole (13), but the Du Bois procedure leads to benzylic C-H insertion only, whereas our methodology gives arene amination exclusively, providing 15 and 16 in a combined 67% yield.

The influence of ligands and counterions on the reactivity of organometallics is well precedented (43, 44). However, examples of such dramatic bifurcation of the reaction manifold are rare and warrant closer study to understand the energetics and full synthetic potential of this metalloid-nitrogen umpolung for direct arene aminations.

Supplementary Materials

www.sciencemag.org/content/353/6304/1144/suppl/DC1

Materials and Methods

Figs. S1 to S4

Table S1

References (4561)

References and Notes

Acknowledgments: J.R.F. thanks NIH (grant HL034300, HL111392, DK038226) and the Robert A. Welch Foundation (grant I-0011) for funding. L.K. gratefully acknowledges the generous financial support of Rice University, NIH (grant R01 GM-114609-01), NSF (CAREER:SusChEM CHE-1455335), the Robert A. Welch Foundation (grant C-1764), American Chemical Society Petroleum Research Fund (grant 51707-DNI1), Amgen (2014 Young Investigators’ Award to L.K.), and Biotage (2015 Young Principal Investigator Award). A provisional patent application (patent no. 62/360,859) has been submitted and assigned jointly to University of Texas Southwestern and Rice University.
View Abstract

Navigate This Article