The Catalytic Cross-Coupling of Unactivated Arenes

See allHide authors and affiliations

Science  25 May 2007:
Vol. 316, Issue 5828, pp. 1172-1175
DOI: 10.1126/science.1141956


The industrially important coupling of aromatic compounds has generally required differential prefunctionalization of the arene coupling partners with a halide and an electropositive group. Here we report that palladium, in conjunction with a copper oxidant, can catalyze the cross-coupling of N-acetylindoles and benzenes in high yield and high regioselectivity across a range of indoles without recourse to activating groups. These reactions are completely selective for arene cross-coupling, with no products arising from indole or benzene homo-coupling detected by spectroscopic analysis. This efficient reactivity should be useful in the design of other oxidative arene cross-couplings as well.

The immense scientific and commercial value of biaryl molecules is illustrated by their ubiquity as building blocks in light-emitting diodes, electron transport devices, liquid crystals, and medicinal compounds (1). The structural simplicity of biaryl compounds belies their preparative complexity, and the search for efficient and convergent syntheses has captivated the attention of synthetic chemists for more than a century. Over the past 30 years, biaryl cross-coupling reactions based on carbon fragment preactivation have revolutionized our ability to forge the carbon-carbon biaryl linkage (1, 2). Of these reactions, the most widely accepted and used are the palladium-catalyzed cross-coupling reactions (such as the Suzuki reaction) of aryl halides and aryl organometallics (3). As is common today, these reactions are dependent on preactivation of the two aromatic carbon fragments with halides and electropositive groups, such as boronic acids or stannanes (4). Incorporation of these functional groups can require several synthetic steps, generating waste from reagents, solvents, and purification, and (upon fragment cross-coupling) can produce undesired organometallic by-products. As a means of reducing our dependence on preactivation, increased attention is being focused on direct arylation processes that replace one of the preactivated substrates with the simple arene itself [for a review, see (5)]. Important advances have been made, particularly in the past decade, and more can certainly be anticipated. In stark contrast, the investigation of cross-coupling reactions that are devoid of arene preactivation is rarely considered, and a high-yielding process with simple unactivated arenes has yet to be described [for a recent report that shows the challenge in achieving high selectivity, see (6); for copper-catalyzed and iron-catalyzed reactions between 2-naphthol and 2-naphthylamine, see (7) and (8).

Substantial hurdles impede the conception of a catalytic arene cross-coupling process that does not involve substrate preactivation. In addition to issues of reactivity and regioselectivity, the catalyst must avoid the generation of unwanted arene homo-coupling that would consume the starting material and generate unwanted by-products (912). To meet this demand, the catalyst must be able to react with one arene in the first step of the catalytic cycle and then invert its selectivity in the second step to react exclusively with the other arene (Fig. 1). Achieving such an inversion in reactivity and selectivity is simultaneously the most daunting challenge and the most crucial prerequisite.

Fig. 1.

(A) Methods for the preparation of biaryl molecules. X is a halide or sulfonate. (B) A prototypical arene catalytic cycle [based on palladium(II) catalysis] illustrating the reactivity-selectivity challenges associated with a catalytic oxidative cross-coupling reaction.

Here, we describe the discovery, development, and study of reactions that meet these challenges and validate this long-sought synthetic strategy (Scheme 1). Notably, no products of arene homo-coupling are detected in the crude reaction mixture, indicating that a complete inversion in catalyst selectivity occurs at the crucial arene metallation steps of the catalytic cycle. Furthermore, although several regioisomeric products could be formed by reaction at different aromatic C-H bonds, markedly high regioselectivity is obtained. Although the precise sequence of reaction steps cannot presently be described, the demonstrated dichotomous behavior of palladium in the presence of electron-rich heteroaromatics and simple arenes should be applicable to other arene combinations. Given the value of the products and the efficiency with which they can be prepared by the use of this method, our observations should enable the development of this strategy for the synthesis of industrially and medicinally important biaryl molecules.

Fig. 2.

Mechanisms of arene palladation. The electrophilic aromatic metallation pathway (top) and concerted proton transfer–metallation pathway (bottom) are shown.

Scheme 1.

Our ongoing work in palladium-catalyzed direct arylation led us to believe that the crucial reactivity-selectivity inversion for arene cross-couplings was an achievable goal. It has been shown that palladium(II) complexes can react via an electrophilic aromatic metallation mechanism (SEAr) with good selectivity for electron-rich arenes (Fig. 2) (1315). In 2006, we discovered that a recently described proton transfer–palladation mechanism (16) can exhibit complementary reactivity to the SEAr pathway (17, 18). With simple arenes, this concerted palladation-deprotonation pathway can depend on arene C-H acidity rather than arene nucleophilicity. Important to the current goal, the palladium complexes associated with these two potentially complementary pathways are analogous to the palladium(II) species at step 1 and step 2 of the catalytic cycle in Fig. 1B. We hypothesized that, if the mechanistic duality associated with these two complementary reactivity modes could be accessed within the confines of a single catalytic cycle, the elusive entry point for selective arene cross-coupling could be achieved.

An extensive investigation of reaction conditions with a range of substrates, palladium catalysts, and additives led to the establishment of the conditions described in Table 1. A survey of electron-rich arenes, in conjunction with benzene as the second coupling partner, revealed that indoles exhibited promising reactivity. The indole nitrogen substituent also dramatically influences the reaction. In initial screens, the free N-H indole did not react, whereas N-methylindole produced self-dimerization predominantly. In contrast, the use of N-acetylindole gave more promising results, which was selected for further catalyst development studies. Optimal catalytic reactivity was achieved with a palladium trifluoroacetate (TFA) catalyst in combination with catalytic quantities of 3-nitropyridine and cesium pivalate (2,2-dimethylpropionate). Although the addition of these last two additives is not crucial to achieve catalytic turnover, superior turnover numbers and reproducibility are associated with their use. We believe that the pyridine additive may be acting to stabilize the palladium(0) before re-oxidation, preventing or slowing the formation of palladium black, which precipitates from the reaction mixture (19). The beneficial impact of the catalytic quantity of cesium pivalate is less clear, but it may interact with the Pd(TFA)2 to generate palladium pivalate early in the reaction. The optimal solvent for the reaction was discovered to be pivalic acid, and a screen of stoichiometric oxidants revealed that copper(II) acetate [Cu(OAc)2] could provide efficient catalytic turnover. The combination of these efforts led to the establishment of optimized conditions involving the treatment of N-acetylindole with an excess of benzene (∼30 equivalents) with 2 to 10 mole percent Pd(TFA)2, 2 to 10 mol % 3-nitropyridine, 40 mol % cesium pivalate, and 3 equivalents Cu (OAc)2 in pivalic acid (2,2-dimethylpropionic acid) under thermal or microwave heating from 110° to 140°C (20).

Table 1.

Development of a catalytic indole-benzene cross-coupling reaction. The products 1, 2, and 3 correspond to those illustrated in Scheme 1, in which R=R′=H. Pd(TFA)2 and (if relevant) Cu(OAc)2, 3-nitropyridine, cesium pivalate (CsOPiv), and/or N-acetylindole were added to a Schlenk tube or microwave vessel, which was followed by the addition of benzene (∼30 equivalents), pivalic acid, and heating according to the indicated method. Oxidant (equivalent), additive (mol %), and Pd (mol %) values were calculated relative to N-acetylindole. Unless otherwise indicated, the values for percent conversion (% conv.), 1:2:3 ratio, and percent yield 1 were determined by GC-MS. The asterisk denotes isolated yield. nd, not determined.

EntryMol % PdOxidant (equiv.)Additive (mol %)Heating methodT (°C)Time (h)% Conv.1:2:3% Yield 1
1 100 None None Oil bath 110 24 75 4.4:1:2.6 55
2 10 Cu(OAc)2 CsOPiv (40) Oil bath 110 24 67 27:1:0.3 64
3 0 Cu(OAc)2 3-Nitropyridine (10) CsOpiv (40) Oil bath 110 24 0 nd 0
4 10 Cu(OAc)2 3-Nitropyridine (10) CsOPiv (40) Microwave 140 5 100 8.9:1:0.3 87*
5 5 Cu(OAc)2 3-Nitropyridine (5) CsOPiv (40) Microwave 140 5 92 13.8:1:0.3 84
6 2 Cu(OAc)2 3-Nitropyridine (2) CsOPiv (40) Microwave 140 5 66 27:1:0 63

A drawback of the thermal heating protocol was the prolonged reaction time (typically 48 hours) required to achieve high conversions with 10 mol % palladium. Notably, a change to microwave heating at 140°C provides a 92% conversion with 5 mol % Pd(TFA)2 in less than 5 hours with a 13.8:1:0.3 ratio of the 1:2:3 isomers and an 84% gas chromatography–mass spectroscopy (GC-MS) yield of the C3 isomer 1 (entry 5 in Table 1). This acceleration is also accompanied by slight drop in C3:C2 selectivity; however, an improvement in C3:C2 regioselectivity occurs with decreased catalyst loadings. For example, with 2 mol % palladium, a 27:1 C3:C2 regioisomeric ratio is obtained with 66% conversion of N-acetylindole (33 turnovers of the palladium catalyst) (entry 6 in Table 1). Under these conditions, the reaction is completely selective for arene cross-coupling, and no compounds arising from indole or benzene homo-coupling are detected by crude proton nuclear magnetic resonance spectroscopy and GC-MS analysis. This finding indicates that the crucial reactivity-selectivity inversion described in Fig. 1 can occur with high precision and fidelity.

Additional examples of reactions with substituted indoles and benzenes are included in Table 2. Thermal heating was used in reactions with chloro-substituted indoles (entries 3 and 4) because small amounts of hydrodechlorination were observed under microwave heating, which hampered product isolation.

Table 2.

Scope of the palladium-catalyzed indole-benzene cross-coupling. Pd(TFA)2 (indicated amount), Cu(OAc)2 (3 equivalents), 3-nitropyridine (1 equivalent to Pd), CsOPiv (40 mol %), and the N-acetylindole were added to a microwave vessel. The arene (∼30 equivalents) and pivalic acid were then added, which was followed by microwave heating. Percent conversion and the 1:2:3 ratio values were determined by GC-MS. Values in the percent yield 1 column denote isolated yield. The asterisks denote that samples were heated thermally in a Schlenk tube. nd, not determined.

Insufficient data exist at present to allow a detailed mechanistic discussion. Although superior reactivity is observed for indoles bearing electrondonating groups, no clear trends have yet emerged with respect to the benzene component. This relative reactivity is also observed in competition studies (see the supporting online material for further details). Nonetheless, these results clearly demonstrate that the dichotomous catalytic behavior required at each of the two metallation steps can be achieved. This knowledge should prompt the investigation and development of a broad range of other palladium-catalyzed oxidative cross-coupling reactions with different substrates.

Supporting Online Material

Materials and Methods

Table S1


References and Notes

View Abstract

Stay Connected to Science

Navigate This Article