We opted for 2-hydroxymethylpyridine as the electron acceptor and glycine derivatives as the electron donor for our preliminary investigation. Firstly, the 2-hydroxymethylpyridine was esterified using 3,5-bis(trifluoromethyl)benzoic chloride, while the glycine derivative was converted into a methyl ester.
We were thrilled to discover that pyridin-2-ylmethyl-3,5-bis(trifluoromethyl)benzoate (1a) and N-phenylmethyl glycinate (2a) were successfully activated by the photocatalyst Ir(ppy)3, and the resulting radicals underwent cross-coupling with a 49% yield. This initial finding prompted us to further investigate the aryl groups on the glycinate substrate. The presence of strong electron-withdrawing groups, such as the trifluoromethyl group on the benzene ring, was found to have a negative impact on the reaction. On the other hand, electron-donating groups like methoxy led to a slightly lower yield, while a fluoride atom in the para-position displayed the highest yield (Table 1, entry 1-4). Furthermore, we tested a range of alcohol-activating acyl groups under the photoredox conditions. It was determined that 3,5-bis(trifluoromethyl)benzoate (1a) was the most effective single electron accepting group for this deoxygenative cross-coupling, with other benzoates showing comparatively lower yields or no coupling product at all in the case of electron-donating benzoates (Table 1, entry 5-7, for additional details, please refer to the supporting information). Solvents were also evaluated, and the results revealed that dichloromethane performed the best compared to the others (Table 1, entry 8-9). Control experiments were also conducted, and the results demonstrated a slight decrease in yield without the involvement of phosphoric acid (Table 1, entry 8). This suggests that the reaction may proceed in a self-catalytic manner, as the benzoic acid by-product forms during the reaction. Phosphoric acid likely plays a role in the initial stage of the coupling reaction. Furthermore, the reaction produced a yield of 19% in the absence of the iridium catalyst, indicating that these two substrates may form a weak electron-donor-acceptor (EDA) complex under the reaction conditions. Lastly, no product was observed when the reaction was conducted in the dark, indicating the light emission was essential to the single electron transfer process.
With the optimal conditions in hand, we embarked on an exploration of the range of substrates for this reaction. Firstly, we reexamined N-(para-methoxyphenyl) (PMP) substituted glycinate using the optimum conditions, resulting in a yield of 67% (1). This is of great significance for the practicality of this protocol, as the PMP group is readily removed under oxidation conditions. Subsequently, we examined the substitution of groups with varying electronic properties and positions on the pyridine ring. The results indicated that electron-donating groups, such as methoxy (3), as well as electron-withdrawing groups like fluoride and bromide (4, 5), provided moderate yields for the 3-substituted pyridines. Similar results were obtained for the 4-substituted pyridines, with no notable impact of the electronic properties of the substituents on the reaction efficiency (6-9). Likewise, the 5-substituted pyridines offered moderate to good yields without any apparent electronic effect (10-16). We also investigated substitution groups at the 6-position of the pyridine, and the results revealed an intriguing steric effect on this reaction. The yield decreased as the atomic radius increased from fluoride to bromide (17-19). This phenomenon suggests that hydrogen bond interactions play a crucial role in the success of this reaction, as steric bulky groups are not conducive to its formation. Multi-substituted pyridines also yielded satisfactory results with moderate to good yields (22-24). It is worth noting that all the halogen atoms were compatible in this reaction, allowing for further modification of the coupling product. Additionally, the substrates derived from 4-pyridin-methanol and 3-pyridin-methanol exhibited yields of 51% and 72% respectively (25, 26). Considering the challenges encountered in synthesizing unnatural amino acids containing nitrogen heterocycles using traditional methods, we decided to shift our focus towards exploring alternative aza-heterocyclic substrates. To our delight, a range of aza-heterocycles performed well under the reaction conditions. The substrate derived from 2-pyrimidine methanol achieved a yield of 63% (27), other aza-heterocycles such as isoquinoline (28), oxazole (29), thiazole (30, 31), imidazole (32, 34), and pyrazole (33) all exhibited moderate yields. Furthermore, a series of secondary alcohol derivatives were tested under optimal conditions, resulting in moderate to good yields, albeit without significant diastereoselectivity (35-40). This approach also demonstrated remarkable capability in forming quaternary all-carbon centers through cross-coupling reactions between tertiary alcohols and glycine derivatives(41-46), which are typically a significant challenge in transition-metal-catalyzed cross-coupling reactions.
Having demonstrated a broad scope for the heterobenzylic alcohol substrates, our interest turned to expanding the scope of potential coupling partners. A series of a-amino ketones were found to work well under the reaction conditions. For instance, α-aminoacetophenone yielded 71% (47), while both electron-donating and electron-withdrawing substituents on the benzene ring resulted in moderate to good yields (47-52). Furthermore, various heteroaromatic amino ketones also exhibited good to moderate yields in this reaction (53-56). Aliphatic amino ketones were tested as well, all of which provided good yields (57-59). Notably, benzyl amine only proved effective in this reaction when an electron-withdrawing group was present on the benzene ring, resulting in a moderate yield (60). To showcase the versatility of this method in constructing complex functional molecules, we examined a series of peptides containing glycine residues. To our delight, all of these substrates, including dipeptides and tripeptides, performed well and yielded good to excellent results (61-64). Additionally, we attached a range of chiral auxiliaries to the glycine substrates in order to explore the stereoselective potential of this reaction. Ultimately, we discovered that chiral 2,5-diphenyl pyrrolidine exhibited the best diastereoselectivity (65-69). Furthermore, this method proved to be highly convenient for the late-stage modification of the beta-adrenergic receptor blocker Pirifibrate (71) and the synthesis of histone deacetylase inhibitor (74)45.
To gain further insight into the mechanism underlying this coupling reaction, a series of control experiments were conducted. Initially, when benzylic alcohol was used as the substrate, no product was observed (Figure 2a). This observation suggests that the presence of a nitrogen atom on the substrate is essential for the success of the reaction. We speculate that the nitrogen atom not only stabilizes the newly formed radical, but also provides a lone pair electron for hydrogen bond interaction. Additionally, we observed a significant decreasing effect in reaction yield as the atomic radius increases for the 6-substituted pyridine methanols (Figure 2b), which could be attributed to the challenge in forming hydrogen bonds when the substituted group becomes more sterically bulky. An effort was also made to capture the radical intermediate by adding TEMPO to the reaction conditions. The heterobenzylic alcohol was converted into the TEMPO-trapped product with a yield of 10% (Figure 2c, 75), indicating the formation of a pyridyl methylene radical in the reaction. Furthermore, Minisci-type side product 76 and homo-coupling product 77 were also isolated in yields of 5% and 8% respectively (Figure 2d), suggesting that the glycine substrate also underwent a radical process. In order to explore the potential formation of an electron-donor-acceptor (EDA) complex in this reaction46-48, we conducted optical absorption spectra tests on substrates 1a, 2c, and their combination. The results revealed that the combination of 1a and 2c did not display a significant red-shift (Figure 2e), indicating that the EDA complex is not strong enough to effectively drive this reaction. Therefore, we propose that the single electron transfer (SET) process in this reaction occurs through the utilization of an iridium photocatalyst as the mediator for electron transfer. Based on these findings, we have put forward a proposed reaction pathway illustrated in Figure 2f. Substrate 1a accepts one electron from the reductive iridium species, leading to the formation of a pyridyl methylene radical intermediate (78). Concurrently, substrate 2c donates one electron to the oxidative iridium species, resulting in a radical at the α-position of the amine (79). The cross-coupling of 78 with 79, facilitated by hydrogen bond interaction, ultimately results in the formation of the C(sp3)−C(sp3) bond.