The development of catalytic methods for the position-selective functionalization of C—H bonds represents a key challenge in molecular synthesis [1-5]. Despite great advances in ortho-selective C—H functionalizations via directing group chelation [6, 7], only few strategies were developed for challenging para-C-H functionalization of arenes [8-15], such as: (ⅰ) electronic effects [16-18]; (ⅱ) template-assistance [19-22]; (ⅲ) steric effects [23-26]; and (ⅳ) σ-activation [27-29] or Lewis acid activation via coordination with substrates (Scheme 1a) [10, 30-35]. Besides these excellent advances, few reports combining two of these strategies were also disclosed. Nakao's and other groups developed Lewis-acid-enabled highly para-selective functionalization of arenes, combing both Lewis acid activation and steric effects [36-43].

Difluoroalkylated arenes remain privileged moieties for drug discovery and development owing to gem-difluoromethylene group's unique stability, and isosteric properties as an ethereal oxygen atom or a carbonyl group, as well as a lipophilic hydrogen-bond donor [44-47]. Thus, a continued strong demand was required for regioselective C—H difluoroalkylation methods of arenes [8, 9, 48, 49]. Despite electronic effects, electron-rich arenes normally offer both ortho- and para-difluoroalkylated products. Until recently, Zhao and Liang groups realized para-difluoroalkylation of ketoximes and anilides by σ-activation, providing only para-difluoroalkylated arenes (Scheme 1b) [27-29]. Zhao's and other groups reported para-difluoroalkylation of aromatic rings combining steric effects and Lewis acid activation [50-55]. Despite the excellent advances, these catalysts normally need to tune both regioselectivity and redox process for difluoroalkyl radical generation, leading to elevated reaction temperatures, low functional group tolerance, and limited scope. Recently, visible-light-induced Ir-catalyzed difluoroalkylation of aromatic aldehydes and ketones was developed affording both ortho- and para-difluoroalkylation products [56, 57]. Herein, we have devised visible-light-mediated para-selective C–H difluoroalkylation of anilides (Scheme 1c). Combination of steric effects and Lewis acid activation strategies leads to para-difluoroalkylated arenes as sole products. The addition of (C₆H₅O)₂P(O)OH and Ag₂CO₃ properly tunes the redox potential of ruthenium catalyst and leads to mild reaction conditions. The protocol exhibits broad functional group tolerance and allows the late-stage functionalization of complex bioactive molecules.

Despite utilization of tervaleryl, pyrimidyl, and oxazolidinone as directing groups in para-selective C–H difluoromethylation of anilides [27, 28, 52], readily cleavable directing groups, such as tert-butyloxy carbonyl (Boc), were seldom employed for para-difluoromethylation. Inspired by Ackermann's [58] and Greaney's [59] pioneering work on visible light enabled Ru-catalyzed meta-C–H alkylation of arenes, we reacted N-Boc aniline (1a) and ethyl bromodifluoroacetate (2a) in the presence of [RuCl₂(p-cymene)]₂, diphenylphosphoric acid, Na₂CO₃, and Ag₂CO₃ under blue light irradiation, giving 71% yield (Table 1, entry 1). Other ruthenium catalysts afforded lower catalytic efficiency (entry 2). A thorough investigation of additives, silver salts, and temperature revealed the combination of (C₆H₅O)₂P(O)OH, Ag₂CO₃, and 60 ℃ led to the best chemical outcome (entries 3–6 vs. 1), probably because these factors may tune redox potential. Furthermore, the reaction vanished with exclusion of ruthenium catalyst, silver salts, and light, indicating all these factors were crucial for the photocatalytic process (entries 7–9).

With the optimal condition, we explored the scope of this reaction with various anilides (Scheme 2A). Anilides with electron-withdrawing groups afforded lower reactivity than those with electron-donating groups (3b-3d vs. 3e-3f). Medicinal-related CF₃- and F-substituted anilides were viable substrates (3b and 3c). meta-Substituted anilines reduced the reaction efficiency owing to sterically hindered para-position (3g vs. 3e). Primary, secondary, tertiary aliphatic, and aromatic carboxylic anilides were also viable substrates, providing 3h-3l in 64%−72% yield. 1-Phenyl-2-pyrrolidinone, a structure motif widely existing in nature products and medicines, provided 3m in good yield (Scheme 2B). 1, 2, 3, 4-Tetrahydroquinoline, phenothiazine, and carbazoles, important building blocks in organic functional materials and drug discovery, delivered para-difluoroalkylated anilides 3n-3p in 47%−63% yield. Additionally, bromodifluoroacetates and bromodifluoroacetamides were also applicable to the reaction (Scheme 2C, 3q-3t). The reaction of 1a on a 1.0 mmol scale with 2a afforded 3a in 51% yield.

Inspired by the wide functional group compatibility, we next applied this protocol to the late-stage functionalization of amino acid, natural products, and bioactive molecules (Scheme 3A). The hydrophobic amino acid derivatives incorporating Val-OH and Phe-OH gave products 4a and 4b. The polar amino acid derivatives containing ester, acid labile t-butyl ether, and amide groups as side chain were also viable substrates (4c, 4d and 4e). Furthermore, alkyl bromides derived from natural products (menthol and borneol) provided 4f and 4g in 43% and 50% yield, respectively. The robustness of this protocol were demonstrated by late-stage functionalization of atorvastatin derivative, the first totally synthesized HMG-CoA reductase inhibitor developed and marketed as a single enantiomer (Scheme 3B) [60]. Coupling atorvastatin derivative with borneol derived bromide successfully produced conjugation product 4i, which provides a novel manner for late-stage introduction of fluorinated groups and fluorinated drug discovery.

To explore the mechanism, a series of control experiments were carried out. The radical scavenger 2, 2, 6, 6-tetramethylpiperidinooxy (TEMPO) and 1, 1-diphenylethylene completely inhibited the reaction, providing adduct 5 (see Sections 5.1 and 5.2 in Supporting information), indicating a difluoroalkyl radical is involved. Failure of difluoroalkylation with steric hindered 6 suggests that the probable coordination of ruthenium with anilides would trigger the reactions (Scheme 4a) [51]. According to previous investigation, three modes were proposed for ruthenium-substrate coordination (Scheme 4b), such as (ⅰ) internal chelation-assisted cycloruthenation 7 [27-29, 48], (ⅱ) cross-over chelation-assisted cycloruthenation 8 [61], and (ⅲ) ruthenium complex coordination with substrate 9 [31, 50]. D/H exchange was not detected by NMR analysis in deuterium experiments (Scheme 4c) and cross-over H/D exchange experiments (Scheme 4d), indicating the cycloruthenation didn't occur. Thus, coordination mode 9, similar to the Frost's and our observations [31, 50], was reasonable for this catalytic process, (Scheme 4b). Indeed, XPS spectroscopy investigations further supported mode 9. N 1s peaks can be divided into two peaks at 399.5 eV (C—N bond) and 400.8 eV (Ru-N bond), demonstrating the coordination of ruthenium with nitrogen (Fig. S1 in Supporting information) [62]. For XPS spectrum of O 1s, four peaks were identified for the Ru-O = C, P-O, C—O, and Ru-O at binding energies 531.0, 531.8, 532.1, and 533.3 eV, respectively, showing coordination of ruthenium with oxygen of acyl group (Fig. S1) [63, 64]. Furthermore, the absence of either nitrogen or oxygen of amide groups changed the reaction site, confirming that the coordination mode 9 could afford an electron-deficient (hence more active) arene core and is essential to controlling regioselectivity (see Section 5.7 in Supporting information) [31, 51]. Without p-cymene was observed indicating ruthenium precatalyst without decoordination (see Sections 5.8 in Supporting information) [65, 66].

Owing to inexistence of cyclometalation in our difluoroalkylation, cyclic voltammetry studies were conducted to identify the photocatalysts. Cyclic voltammetry studies featured oxidation peak of ruthenium catalysts in 60 ℃ at 1.42 V, ruthenium catalysts with Ag₂CO₃ in 60 ℃ at 1.10 V, ruthenium catalysts with (C₆H₅O)₂P(O)OH in 60 ℃ at 1.43 V, ruthenium catalysts with Ag₂CO₃ and (C₆H₅O)₂P(O)OH in 60 ℃ at 0.45 V, and ruthenium catalysts with Ag₂CO₃, (C₆H₅O)₂P(O)OH and 1a in 60 ℃ at 0.45 V (Fig. 1 and Fig. S2 in Supporting information). However, the reduction capability of ruthenium catalysts with Ag₂CO₃ and(C₆H₅O)₂P(O)OH at room temperature were insufficient (Fig. S3 in Supporting information). Fluorescence quenching experiments on the catalytic system containing [RuCl₂(p-cymene)]₂, Ag₂CO₃ and(C₆H₅O)₂P(O)OH at 60 ℃ with 2a demonstrated a progressive quenching profile in line with the envisaged SET event to generate the difluoroalkyl radical (Fig. S4 in Supporting information). These findings suggest the combination of ruthenium catalysts with (C₆H₅O)₂P(O)OH and Ag₂CO₃ or ruthenium catalysts with (C₆H₅O)₂P(O)OH, Ag₂CO₃ and 1a would be the photocatalytic system, and the appropriate temperature is crucial to adjust its redox capability.

Based on previous literatures [31, 50, 58, 59] and our experiments, a mechanism is proposed for para-selective difluoroalkylation (Scheme 5). [RuCl₂(p-cymene)]₂, C₆H₅O)₂P(O)OH, and Ag₂CO₃ forms ruthenium complex A, which coordinates with 1a to afford B. Meanwhile, A or B is excited by visible light to give A* or B*, which reduces 2a with the formation of radical C. Next, radical C is trapped by B to generate D, which is oxidized by F or G to give complex E. Finally, E releases product 3a and regenerates catalytic species A.

In summary, we have developed a visible-light-mediated ruthenium-catalyzed para-difluoroalkylation of anilides under mild conditions. Given its good efficacy and functional group tolerance, the versatile photo-para-C—H functionalization enabled late-stage functionalization of amino acid, natural products, and bioactive molecules. Overall, our findings demonstrate the unique potential of merging visible-light photoredox catalysis with ruthenium-mediated para-C—H functionalization. Further studies are ongoing in our laboratories and will be reported in due course.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Acknowledgment

The authors gratefully acknowledge support from the University Nursing Program for Young Scholars with Creative Talents in Heilongjiang Province (No. UNPYSCT-2017124).

Supplementary materials

Supplementary material associated with this article can be found, in the online version, at doi:10.1016/j.cclet.2022.06.048.