Discovery of an α-Amino C–H Arylation Reaction Using the Strategy of Accelerated Serendipity

See allHide authors and affiliations

Science  25 Nov 2011:
Vol. 334, Issue 6059, pp. 1114-1117
DOI: 10.1126/science.1213920


Serendipity has long been a welcome yet elusive phenomenon in the advancement of chemistry. We sought to exploit serendipity as a means of rapidly identifying unanticipated chemical transformations. By using a high-throughput, automated workflow and evaluating a large number of random reactions, we have discovered a photoredox-catalyzed C–H arylation reaction for the construction of benzylic amines, an important structural motif within pharmaceutical compounds that is not readily accessed via simple substrates. The mechanism directly couples tertiary amines with cyanoaromatics by using mild and operationally trivial conditions.

Accidental or serendipitous discoveries have led to important breakthroughs in the chemical sciences. With regard to bond-forming reactions, such fundamental synthetic transformations as Friedel-Crafts, Wittig olefination, and Brown hydroboration reactions were found when the objectives of the initial experiments were not in accord with the observed outcomes (1).

Recently, we questioned whether serendipity could be forced or simulated to occur on a predictable basis in the realm of reaction discovery, thereby providing a reliable platform to access valuable transformations or unexpected pathways. Herein, we describe the successful execution of these ideals and describe a fundamentally distinct C–H functionalization-arylation reaction that we expect will be of broad use to practitioners of chemical synthesis and, in particular, medicinal chemistry.

Assuming that serendipity is governed by probability (and thereafter manageable by statistics), performing a large number of random chemical reactions must increase the chances of realizing a serendipitous outcome. However, the volume of reactions required to achieve serendipity in a repetitive fashion is likely unsuitable for traditional laboratory protocols that use singular experiments. Indeed, several combinatorial strategies have previously been used to identify singular chemical reactions (211); however, the use of substrate-tagging methods or large collections of substrate mixtures does not emulate the representative constituents of a traditional chemical reaction. On this basis, we posited that an automated, high-throughput method of reaction setup and execution, along with a rapid gas chromatography–mass spectrometry (GC-MS) assay using National Institute of Standards and Technology (NIST) mass spectral library software, might allow about 1000 random transformations to be performed and analyzed on a daily basis (by one experimentalist) (Fig. 1). Although we recognized that it is presently impossible to calculate the minimum number of experiments that must be performed to achieve “chance discoveries” on a regular basis, we presumed that 1000 daily experiments would be a substantial starting point.

Fig. 1

Approach to reaction discovery without preconceived design via the concept of accelerated serendipity. R indicates a generic organic substituent; X and Y, heteroatoms.

Substrate pools were created of molecules containing common functional groups that would be considered nonreactive or benign in one-to-one combinations (see figs. S2 and S3 for substrate pools). As a critical design element, we identified that such a process must be free from any preconceived bias as to which chemical reaction is discovered or the mechanism by which it may happen. As such, we sought to minimize substrate combinations that are likely to participate in established reaction pathways. A Chemspeed (ChemSpeed Technologies, Basel, Switzerland) robotic system was used to arrange all of the pairwise combinations into 96-well plates (for example, a pool of 19 substrates equates to 171 different combinations) before a catalyst system was added (i.e., catalyst, ligand, additive, solvent, etc.). This screening platform was then used in a repetitive fashion to examine a range of catalyst systems. Our analysis method to detect potential coupling products had its basis in a GC-MS assay where peaks of substantial intensity and molecular weight were an indication of possible coupling between the two reactants, with the NIST mass spectral library providing an indication of possible structure. A critical evaluation stage was used to determine whether a reaction is mechanistically unanticipated and might furthermore show potential as an interesting or useful chemical transformation. Initially, transition metal complexes derived from Pd, Ru, Au, Fe, etc., were examined. New catalysts were identified for three existing transformations: AuCl3 catalyzed indole alkylation with styrene, FeCl3 catalyzed alkyne dimerization, and Ru3(CO)12 catalyzed styrene hydroesterification with MeOH, where Me is a methyl group (fig. S8) (12). However, in order to exploit the full potential of serendipity and discover unanticipated chemical reactions, we recognized that implementing this discovery process in a relatively uncharted area of synthetic methodology might statistically aid in the pursuit of this goal (Fig. 2A). In this context, photoredox catalysis was selected as a target area, given that it is a relatively young and emerging field in organic synthesis that has recently delivered a variety of powerful bond-forming processes (13).

Fig. 2

Discovery of a new photoredox amine C–H arylation reaction. Ph phenyl group; DMF, N,N-dimethylformamide; DMA, N,N-dimethylacetamide.

Substrate pools were exposed to a series of inorganic photoredox catalysts in the presence of a household 26-W fluorescent lamp (all performed on a Chemspeed robotic platform). The use of GC-MS analysis revealed a reaction hit for the combination of N,N-dimethylaniline and 1,4-dicyanobenzene (1,4-DCB) with Ir(ppy)2(dtbbpy)PF6 (where dtbbpy is 4,4′-di-t-butyl-2,2′-bipyridine and ppy indicates 2-phenylpyridine) as a photocatalyst (Fig. 2B). Investigation into the reaction mixture revealed the formation of an unusual α-amino cyanobenzene coupling product, 2, that formed in 11% yield (14). In evaluating the observed outcome, we found that the discovery was, indeed, a new example of a photoredox catalyzed process. Photolytic methods for the α-functionalization of amines typically require the use of high-energy light (and therefore the availability of specialized equipment for reaction setup) (1519). Additionally, we recognized the value of the benzylic amine product 2. More specifically, α-aryl amines are a prominent structural class found among medicinal agents, with 8 of the 100 top-selling pharmaceuticals containing this motif (and a vast array of others being simple derivatives thereof) (20). Given the potential utility of such a transform, optimization was undertaken, and improvements in efficiency were made with changes to the solvent, base, and most notably the photocatalyst, with the commercial Ir(ppy)3 system providing the desired α-arylation adduct 2 in 85% yield. With this highly efficient process in hand, our attention turned to the full exploitation of this α-amine C–H arylation reaction as a mild and operationally trivial means of forming benzylic amines (2126).

Structural scope exploration began with five-membered pyrrolidine, six-membered piperidine, morpholine, N-Boc (where Boc is tert-butoxycarbonyl) piperazine, and seven-membered azepane rings, all of which provided excellent results (Fig. 3, entries 1 to 5). Acyclic amines also functioned efficiently in this protocol (entry 6). Variation of the N-aryl group was tolerated, including methyl and halogen substituents as well as N-naphthyl–substituted amines (entries 7 to 10). A p-methoxyphenyl (PMP) substituent, which serves as a well-established protecting group for the nitrogen atom (27), can also be used (entry 11). Moreover, we have developed an alternative, nonaryl protecting group in the form of dimethoxy butane (DMB) that can be successfully used in the photoredox arylation (entry 12) and easily removed thereafter by using acidic conditions [supporting online material (SOM)] (28). Privileged cyclic amine structures from medicinal chemistry were also subjected to this arylation reaction, such as protected indolines and tetrahydroquinolines, and there is a distinct preference for arylation at the ring position as opposed to the acyclic benzylic site (entries 13 to 15). However, the situation is altered by using PMP-protected tetrahydroisoquinoline with arylation at the cyclic benzylic position as the only detectable outcome (entry 16).

Fig. 3

Photoredox C–H arylation: amine scope. R, generic alkyl or aryl substituent. Data reported as entry, % isolated yield, regiomeric ratio (r.r., determined by 1H nuclear magnetic resonance analysis). 2.5 to 3.0 equivalents of amine used (see fig. S6 for stoichiometry effects).

The scope of the aryl ring component in this α-amine arylation protocol has also been studied (Fig. 4). We found that benzonitriles substituted with esters, amides, phosphonate esters, and electron-deficient tetrazoles are suitable substrates (entries 1 to 4). The cyano group also proved effective as a coupling handle in the challenging steric environment of 2,6-disubstituted aryl rings to generate bis-ortho-substituted products (entry 5). Furthermore, the site specificity of the coupling process can be further exploited through the use of 1,2-dicyanobenzene, efficiently forming the corresponding ortho-substituted isomer (entry 6). In recognizing the electron-poor nature of the arene nucleus as an essential feature for reactivity, our attention turned to electron-deficient heteroaromatics as coupling partners. These moieties are among the most widespread constituents of pharmaceutical compounds (29). Our initial results in this area appear to be extremely encouraging. Cyano-substituted pyridines smoothly undergo coupling with the amine substrate, albeit with lower efficiency in the ortho case (entries 7 to 9). Azaindoles, an important biological isostere for indoles (29), are also amenable to this arylation strategy (entry 10).

Fig. 4

Photoredox C–H arylation: arene and heteroarene scope. Ar, generic aryl group; X, CH or heteroatom; EWG, electron-withdrawing functional group. Data reported as entry, % isolated yield. 3.0 equivalents of amine used (see fig. S6 for stoichiometry effects). *Run with 5 mol % catalyst. †Leaving group is a chloride; photocatalyst is Ir(ppy)2(dtbbpy)PF6. ‡Reaction time is 72 hours.

We have also identified that five-membered heterocycles are suitable substrates. For instance, a triazole was found to undergo coupling in moderate yield (entry 11). Furthermore, for certain classes of five-membered heteroaromatics, a simple chloride can function equivalently to CN as a suitable leaving group. This enables arylation of amines with chloro-substituted caffeine, benzoxazole, benzothiazole, and N-Boc benzimidazole (entries 12 to 15). Lastly, the direct derivatization of pharmaceutical agents has been demonstrated by using linezolid, an antibiotic that undergoes direct heteroarylation in 58% yield (entry 16). This result further demonstrates the capacity of druglike molecules to readily participate in this C–H functionalization reaction.

Our proposed mechanistic explanation for the C–H arylation process is described in Fig. 5. Triscyclometalated Ir(III) complexes, such as Ir(ppy)3 (3), are reversibly promoted to their excited state form [*Ir(ppy)3 (4)] upon absorption of a photon from the 26-W light source (30). *IrIII(ppy)3 (4) is a powerful reductant [oxidation potential (E1/2ox) = –1.73 V versus saturated calomel electrode (SCE) in CH3CN] (30, 31) and, upon encountering 1,4-DCB (E1/2red = –1.61 V versus SCE in CH3CN) (32), could donate an electron to form the corresponding arene radical anion 5 (3335). The resultant IrIV(ppy)3 (6) is a strong oxidant (E1/2red = +0.77 V versus SCE in CH3CN) (30, 31) and would be capable of undergoing a single-electron transfer event with amine 7, generating amine radical cation 8, as well as re-forming IrIII(ppy)3 (3) and thereby completing the photoredox cycle. The C–H bonds adjacent to the nitrogen atom in 8 are weakened by about 40 kcal/mol and so could undergo deprotonation by NaOAc (where OAc is an acetoxy group) to give α-amino radical 9 (19, 36). A radical-radical coupling reaction could then unite intermediates 5 and 9, representing the key bond-forming step (1519, 3743). Elimination of CN from 10 would then form the aromatized benzylic amine product 11.

Fig. 5

Photoredox C–H arylation: proposed mechanistic pathway. R, generic alkyl or aryl substituent; SET, single-electron transfer.

In summary, the concept of accelerated serendipity has been successfully executed, resulting in the discovery of a photoredox amine C–H arylation reaction. Requiring only commercially available materials, mild conditions, and operationally trivial reaction protocols, we anticipate this carbon-carbon bond-forming protocol will be widely used in the synthesis of benzylic and heterobenzylic amines.

Supporting Online Material

Materials and Methods

SOM Text

Figs. S1 to S10

References (4446)

References and Notes

  1. Other examples of reaction discovery using a combinatorial approach have been reported (311).
  2. After completion of this work, an alternative strategy for reaction discovery was described (11).
  3. These reactions were discovered within 3500 experiments involving transition metal catalysis.
  4. α-Amine product 2 was discovered within 1000 photoredox experiments.
  5. As based on the 2008 figures compiled by the Njardarson group (
  6. Other notable examples of catalytic amine α-arylation have been reported (2226).
  7. Supported by luminescence quenching studies; see figs. S9 and S10.
  8. Radical anion 5 could be persistent under the reaction conditions; see (35) for evidence of stability. We thank a reviewer for emphasizing this point.
  9. For electron-transfer mediated processes using organic photooxidants, see (39, 40).
  10. Other types of radical aromatic substitution reactions are known to occur; see (42, 43).
  11. Acknowledgments: A.M. thanks the Marie-Curie Actions for an International Outgoing Fellowship. Financial support was provided by the NIH General Medical Sciences (NIHGMS) (grant NIHGMS R01 01 GM093213-01) and gifts from Merck, Amgen, Abbott, and Bristol-Myers Squibb. We would also like to thank J. C. Conrad for assistance.
View Abstract

Navigate This Article