Reaction development and optimization. We started screening the C-N coupling by using bis(1,5-cyclooctadiene)nickel(0) (Ni(COD)2) as the nickel(0) source and aryl iodide as the electrophile to form the assumed active aryl-Ni-I complexes (Table S1). After intensive investigation for coupling of aniline 4 with ethyl 4-iodobenzoate 5, the coupling product 6 was obtained in almost quantitative yield when using simple, commercially available 4,4'-di-tert-butyl-2,2'-bipyridine as the ligand, (Ni(COD)2) and 1,8-diazabicyclo[5.4.0]undec-7-ene (DBU) as a base under photo-irradiation with blue light-emitting diodes (LEDs) at 30 ˚C (Table S1, Entry 2). An air-stable NiCl2 salt exhibited comparable reactivity to Ni(COD)2 in tetrahydrofuran (THF) (Entry 1). Other air-stable nickel(II) salts also gave product 6, but in lower yield (Entries 3–5). Subsequently, the NiCl2 salt was used in further optimization studies. In contrast to previous photoredox/nickel dual catalysis3, high polar solvents were not suitable for the coupling process, probably due to their strong coordination ability to the nickel(II) complex29, while a less-polar solvent, toluene, also resulted in lower efficiency than THF (Entries 6–10). The solubility and basicity were considered critical for the deprotonation of anilines on a nickel complex to replace iodine using a suitable base30. Thus, DBU exhibited the best performance among other common organic or inorganic bases (Entries 11–16). The use of proper ligands is key to realizing the visible-light-induced carbon-nitrogen coupling. Ligand screening showed that redox-active bipyridine ligands were required to maintain the reactivity (Entries 17–20). Furthermore, control experiments indicated that visible light, ligand, base, and nickel salt were all essential to achieve the coupling reaction, supporting our assumption that direct photo-excitation of the nickel complex by visible light is required (Entries 21–24). When exposed to air, the process was completely inhibited (Entry 25), indicating that the active nickel catalyst was sensitive to air.
Substrate scope. Substrate scope studies for diverse cross-couplings are summarized in Fig. 2 and Fig. 3. Various nitrogen, oxygen, and sulfur nucleophiles with a broad pKa range successfully afforded coupling products 6-108 from aryl/alkenyl iodides/bromides in good to excellent yields with good functional group tolerance, demonstrating the versatility of the present exogenous photocatalyst-free conditions (Figs. 2 & 3). Arylation of aromatic and alkyl amines gave coupled products (6–12, 30–44) in good to excellent yield, and was scalable to gram scale (30, 95% yield). Both electron-donating and electron-withdrawing anilines reacted smoothly (7, 8). Sterically hindered anilines bearing a substitution at the ortho-position exhibited good reactivity, delivering diarylamines (9, 10). Notably, an aryl chloride unit remained intact, thus affording the corresponding coupled products in high yield (12, 31, 104). Interestingly, 2-aminopyridine coupled twice with aryl iodide (22). Weakly nucleophilic N-reagents such as amides, sulfonamides, sulfoximines, indoles, pyrrole, imine, carbamates, urea, and phosphinic amide smoothly coupled with aryl iodides to afford products (13–29) in high yields. N-Arylation of indoles, pyrrole, and urea to access structures (23, 24, 28, 71, 73) of pharmaceutical interest were realized for the first time under photo-driven nickel catalysis, thereby providing an efficient alternative to copper catalysis31,32.
The protocol was applicable for coupling with diverse O-nucleophiles as well, including alcohols, carboxylic acids12,33, phenol, water, and oxime, to deliver O-arylation products (45–62) in good yield. Electron-rich and electron-poor aromatic or aliphatic carboxylic acids were successfully used in the esterification with aryl iodides (57–60). It is particularly noteworthy that the hydroxylation of aryl halides with water under exogenous photocatalyst-free conditions gave the products (62, 85, 86) in good yields34. Substrates containing terminal alkynyl groups, such as propargylamine and pent-4-yn-1-ol, are rarely used as coupling partners in nickel catalysis, but were applicable in this system (43 and 52). Intramolecular C-O coupling also proceeded smoothly (56). Thiocarboxylic acids coupled with aryl/heteroaryl iodides (63 & 90). As shown in footnote a of Fig. 2, reactions with weak O-nucleophiles proceeded well just by using a nickel(0) source instead of a nickel(II) salt.
Aryl iodides with diverse functional groups such as halides, trifluoromethoxy, trifluoromethyl, acyl, formyl, nitriles, esters, and alkynyl groups at the para, or meta-position readily undergo coupling to afford O/N-arylation products (66–70, 73, 75–77, 80, 85, 88). Electron-rich aryl iodides showed similarly high reactivity, giving products (64, 65, 70, 72) in high yields. The protocol was also applicable to sterically hindered ortho-substituted aryl iodides, affording the products in good yield (78 and 79, 85% and 61% respectively). Five- or six-membered heterocycles were well adapted to efficiently generate C-N, C-O and C-S coupled products (74, 81, 82, 87–89, 90). To expand the utility of this protocol, we also investigated the substrate scope of aryl bromides. It is pleasing to find that the aryl bromides also smoothly coupled with N,O-nucleophiles to afford the final products in good yields under slightly modified conditions (91–95). Some alkenyl iodides, which are rarely used as coupling partners in photoredox/nickel dual catalysis, possibly due to their incompatibility with highly reactive photocatalysts, were also applicable under the present protocol, efficiently forging C-N/C-O bonds (96–100). The widespread application of the photo-driven single nickel catalysis was showcased by late-stage modification of pharmaceuticals and natural products (101–108). Interestingly, the triflate group remained intact, affording 108 and 107′, the isomer of the coupled product (107), in 95% yield and 60% yield respectively.
Mechanistic investigations. To gain insight into the supposed NiI/NiIII catalytic cycle of this single nickel catalysis (Fig. 1b), several experiments were run as summarized in Fig. 4 and in supplementary information. Based on Doyle’s proposed NiI/NiIII catalytic cycle, we initially supposed that the activated NiI can be generated via homolysis of NiI-aryl bond. The supposed homolysis of NiI-aryl bond should then generate small amount of dehalogenated arene and corresponding dimerized biphenyl. In Fig. 4a, however, such side adducts were not detected with gas chromatography-mass spectrometry (GC-MS) analysis under our standard conditions using either NiCl2 or Ni(COD)2 and DBU base in THF. In Fig. 4b-(1), competition experiments using ethyl 4-iodobenzoate (1 equiv) and 2-iodotoluene (1 equiv) as competing electrophiles and hexan-1-amine as a nucleophile (0.8 equiv) gave 34 as a sole product in 79% yield. The result indicated that 2-iodotoluene remains intact in the presence of more reactive ethyl 4-iodobenzoate. On the other hand, in Fig. 4b-(2), when using a sub-stoichiometric amount of pre-formed NiII complex 109 as a catalyst, compound 78 was observed in 7% together with 34. Even in this case, neither toluene nor its dimer was detected by GC-MS. In Fig. 4b-(2), we also detected 2-iodotoluene by GC-MS, which is ascribed to reductive elimination from NiIII complexes to form carbon-iodide bond35,36. These results indicated that the active NiI species under our reaction conditions is less likely to be generated via homolysis of Ni-aryl bond. As shown in Fig S5, we assume that the active NiI complexes were generated via homolysis of Ni-I bond of complex 109 under visible light irradiation37, then the second complex 109 was oxidized by iodine radical to form NiIII complex S-II, which underwent facile reductive elimination to deliver 2-iodotoluene or ligand exchange with nucleophiles to deliver coupling product 78 as detected in Fig. 4b-(2). In addition, the formation of coupled product 78 can be obtained from 109 by stoichiometric studies in which the yield of 78 increased with increasing amounts of hexan-1-amine along with some 2-iodotoluene as well, which is consistent with ligand exchange in the presence of excess nucleophile (Fig. S6).
Based on experimental results (Fig. 4(a), (b)), we tentatively propose a NiI/NiIII catalytic cycle depicted in Fig. 4c. The nickel(0) species I is oxidized by aryl iodide to form aryl-NiII-I complex II, which exhibits some absorption in the range of visible-light (Fig. S8 & S9). Then the aryl-NiII-I complex II or its possible related five coordinated complexes with DBU or nucleophiles as the fifth ligand was excited by the visible light and underwent homolysis of NiII-I bond to deliver iodine radical and NiI complex III. Subsequently, iodine radical was rebounded to the second aryl-NiII-I complex II to afford complex IV, followed by ligand exchange and reductive elimination to afford active NiI complex VI and coupling product. This process always requires visible-light irradiation (Fig. S4), probably due to the disproportionation of gradually accumulated NiI complex III and VI to form Ni0 complex I and NiII complex II again. Some other control experiments using pre-formed NiI complex or generated in situ also were run as well25,28; however, we did not observe coupling products (Fig. S3 & Table S2). Therefore, we can not exclude other possibilities, such as Ni0/NiII/*NiII catalytic cycle38,39 (Fig. S7) at the moment.