Photocatalytic defluoroalkylation and hydrodefluorination of trifluoromethyls using o-phosphinophenolate

Under visible light irradiation, o-phosphinophenolate functions as an easily accessible photoredox catalyst to activate trifluoromethyl groups in trifluoroacetamides, trifluoroacetates, and trifluoromethyl (hetero)arenes to deliver corresponding difluoromethyl radicals. It works in relay with a thiol hydrogen atom transfer (HAT) catalyst to enable selective defluoroalkylation and hydrodefluorination. The reaction allows for the facile synthesis of a broad scope of difluoromethylene-incorporated carbonyl and (hetero)aromatic compounds, which are valuable fluorinated intermediates of interest in the pharmaceutical industry. The ortho-diphenylphosphino substituent, which is believed to facilitate photoinduced electron transfer, plays an essential role in the redox reactivity of phenolate. In addition to trifluoromethyl groups, pentafluoroethyl groups could also be selectively defluoroalkylated. Photoredox catalysis can strongly reduce and cleave unactivated chemical bonds via photoinduced electron transfer. Here the authors use o-phosphinophenolate for photocatalytic C–F activation of a wide range of trifluoromethyl groups in trifluoroacetamides, trifluoroacetates, and trifluoromethyl(hetero)arenes to deliver corresponding difluoromethyl radicals.

P hotoredox catalysis has demonstrated its strong reducing power to cleave unactivated chemical bonds via photoinduced electron transfer [1][2][3][4][5] . Among the various photocatalysis-enabled methods of activating inert bonds, the direct selective C-F activation [6][7][8][9][10][11] of trifluoromethyl groups to deliver corresponding difluoromethyl radicals is an ideal transformation for the synthesis of difluoromethylene-incorporated compounds, which are valuable fluorinated intermediates in the pharmaceutical industry [12][13][14] . The low cost and ready availability of trifluoroacetamide, acetate, and a variety of trifluoromethylated (hetero)aromatics make this transformation appealing. Ingenious photocatalytic methods for selective hydrodefluorination and defluoroalkylation of trifluoromethyl(hetero)arenes have been developed by Jui 15,16 , König 17 , and Gouverneur 18 . While the photocatalysts prevalently used in these transformations are precious metal-based polypyridyl complexes and π-conjugated organic dyes, we conceived that anionic phenolate [19][20][21][22][23] , which has strong reductive potential in its excited state, may work as a suitable catalyst for selective C-F functionalization of trifluoromethyls over a broad scope of substrates. Although simple phenolate failed as a catalyst, probably because of its insufficient excited state lifetime and the poor stability of phenoxy radicals 19,21 , which limited catalyst turnover, we were inspired by the effect of triphenylphosphine that we previously observed in photocatalytic decarboxylative couplings [24][25][26][27] to facilitate photoinduced electron transfer, and we hypothesized that an orthophosphino group easily installed onto phenolate may overcome these problems. We present our catalyst design in Fig. 1a. The installation of ortho-PPh 2 has three benefits. First, the ortho-PPh 2 substituent redshifts the absorption of the ground state anion (PO -) from the ultraviolet to visible light range. Second, phosphine exerts a heavy atom effect 28,29 to facilitate intersystem crossing to access the triplet state and extends its lifetime for efficient photoelectron transfer with substrates. Third, the interaction of phosphine with oxygen radicals is expected not only to facilitate efficient photoelectron transfer in the anionic excited state (*PO -) but also to stabilize the radical double state (PO • ).

Results and discussion
Studies of properties of PO catalysts and reaction parameters. Our mechanistic hypothesis depicted in Fig. 1c was inspired by previous examples of photoredox/HAT synergistic catalysis 15,16,18 . The photoexcited POcatalyst (*PO -) reduces the trifluoromethyl substrate (I) to deliver difluoromethyl radicals (II). II can be reduced by a hydrogen donor to produce a hydrodefluorination product or be intercepted by an alkene to generate a new alkyl radical (III). The alkyl radical (III) can be reduced by a polarity-reversal thiol HAT catalyst 34 to produce a defluoroalkylation product. The thiol catalyst regenerates through hydrogen abstraction from formate (BDE of formate C-H: 88 kcal/mol) to deliver CO 2 • -, which is a strong reductant (E 1/2 CO 2 /CO 2 • -= − 2.2 V vs. SCE), to reduce PO • and complete the POredox cycle. Guided by the hypothesis, we first focused on the selective defluoroalkylation of trifluoroacetamide with alkenes, a transformation that has not been successfully developed using photocatalysis. We discovered that a catalytic amount of PO in combination with 1-adamatanethiol (1-AdSH) as the HAT catalyst in the presence of formate catalyzed monoselective defluoroalkylation of N-phenyltrifluoroacetamide with alkenes under irradiation with a 427 nm LED (Kessil®, emission FWHM of~20 nm) 35 . A similar transformation was reported by Houk and Wang et al. under thermal conditions using a stoichiometric amount of DMAP-BH 3 via a spincenter shift strategy 36 . Figure 2 shows the key reaction parameters. Under optimal conditions (Fig. 2a), defluoroalkylation product (3) was obtained in 90% yield determined by 1 H-NMR, along with the generation of 4 in 5 % yield. The reaction mixture appeared as a light yellow homogeneous solution, suggesting its promising applications in flow photosynthesis 37,38 . UV-Vis absorption spectra of catalysts, substrates, and reaction mixtures were measured to determine the light absorbing species (Fig. 2b). Both substrates (1 and 2) absorb light only in the ultraviolet range (< 325 nm). PO1 in its phenol form had an absorption onset at~370 nm (purple line in Fig. 2b). Upon mixing with Cs 2 CO 3 , the deprotonated anion of PO1 (PO1 -) exhibited redshifted absorption into the visible light range with an absorption onset at~425 nm. Comparison of the absorption spectra of PO1and tert-butylphenolate showed that the ortho-PPh 2 substituent significantly redshifted the absorption curve by~70 nm (see Supplementary Information for details). PO1absorbs visible light with an onset at~425 nm and shows an emission maximum at 540 nm. The reduction potential of excited PO1 -(*PO1 -) is estimated to be -2.89 V vs. SCE, a value sufficient to reduce a broad scope of trifluoromethyl aromatic and carbonyl compounds  In the presence of 1, the absorption spectrum of PO1did not show a bathochromic shift, showing that an electron-donor-acceptor complex 39-41 between 1 and PO1was not formed (blue line). The absorption curve of the reaction mixture (red line) was identical to that of PO1 -, indicating that PO1was the light absorbing species in the reaction mixture. Figure 1c shows the PO catalysts with different structures. 2-Methyl (PO2)-, 2,4-di-tertbutyl (PO3)-, and 2,4-dimethyl (PO4)-substituted o-phosphinophenols all showed performances comparable to that of PO1. o-Phosphinophenol (PO5) without a p-substituent showed reduced catalytic efficacy. p-Phosphinophenol (PO6) gave 3 in only 15% yield, which suggests the essential role of intramolecular P-O interactions in catalytic efficacy. A bulky PO catalyst with a dicyclohexylphosphine substituent (PO7) was ineffective. Regarding to stability, PO1 is bench-stable white powder and can be stored under ambient air for months without apparent decomposition and oxidation. Table 1 summaries key reaction parameters. Using cyclohexanethiol as the HAT catalyst reduced the yield (entry 1, Table 1). Using 10 mol % 1-AdSH resulted in decreased yield (entry 2). PO1 (2 mol %) still catalyzed the reaction in 52% yield (entry 3). Increasing the amount of alkene to 2.0 equivalents suppressed hydrodefluorination and increased the defluoroalkylation yield to 96%. Replacing formate with triethylsilane gave 2 in 30% yield (entry 5), indicating that formate is not essential for C-F activation. LEDs with an emission peak at 440 nm (emission range from 415 nm to 470 nm) also promoted the reaction (entry 6), but LEDs with emission peaks at 456 nm and 467 nm, which did not overlap with the absorption of PO1 -, were ineffective (entries 7 and 8). PPh 3 and 4-tert-butylphenol used alone were both ineffective (entries 9 and 10), while 10% of 3 was generated by using a mixture of them (entry 11), suggesting certain role of P-O interactions in efficient photoelectron transfer. Control experiments showed that PO1, 1-AdSH, and light were all essential parameters (entries 12, 13, 15). In the absence of formate salt, product 3 was detected in 18% yield (entry 14), suggesting the role of formate in catalyst turnover (ref. Fig. 1c). The cation moiety of formate salt affects not only solubility but also reactivity, because the alkali metal cations act as counter cations of both formate and generated fluoride salt, that may affect the rates of HAT and defluorination. Hence, different formate salts (Li, Na, K, Cs) were tested (see Supplementary  Table 1 in SI page 6 for details). The quantum yield of 3 was estimated to be 4.4 according to the literature 42,43 , which suggested that CO 2 •generated after HAT may activate -CF 3 substrates (e.g., reduction potential of 1,3-bistrifluoromethylbenzene, E red 1/2 = -2.07 V vs. SCE; reduction potential of 1, E red p/2 = -2.11 V vs. SCE) in relay with the thiol HAT catalyst to deliver 3 (pale dashed arrow in Fig. 1c) 44,45 . Exposure to air completely killed the catalytic reactivity (entry 16) resulting recovery of starting materials, as air can quench excited triplet state of PO1and oxidize thiol.
Scope of the reactions. Figure 3 illustrates the scope of the defluoroalkylation of trifluoroacetamides. The reaction can be easily scaled up to the gram scale under batch conditions using Kessil LEDs and a Schlenk flask (3). For trifluoroacetamides possessing electron-neutral and electron-rich N-aryl substituents, monodefluoroalkylation products accompanied by a small amount of hydrodefluorination byproduct (5-10%) were obtained (5,6,7). For trifluoroacetamides with strong electron-deficient Naryl substituents, products of didefluoroalkylation were obtained as major products (8,9), and monodefluoroalkylation products were observed only in trace amounts (<3%). The recovered amide staring materials accounted for moderate yields (8,9). For N-3-   (10), a monodefluoroalkylation product was obtained in 68% yield. An N-aryl substituent is essential for defluoroalkylation (13,14), and the reactivity is not applicable to tertiary amides (12).
In the absence of an alkene, hydrodefluorination products were generated (Fig. 5). Thiol as a polarity reversal catalyst 34 is essential for a high yield of hydrodefluorination. Cesium formate used in 1.2 equivalents is critical to ensure high monoselectivity (4,(36)(37)(38)(39). The N-phenyltrifluoroacetamide with a para-ester substituent underwent thorough hydrodefluorination to generate acetamide (40). The same reaction conditions are also applicable to the selective hydrodefluorination and defluoroalkylation of pentafluoropropionamide (Eqs. (1) and (2) in Fig. 6), suggesting the further application of PO catalysts in the selective C-F functionalization of polyfluorinated compounds (41, 43) 49-51 .
In summary, o-phosphinophenolates were developed as efficient photocatalysts for the selective C-F activation of a wide range of trifluoromethyl groups in trifluoroacetamides, trifluoroacetates, and trifluoromethyl (hetero) arenes to deliver corresponding difluoromethyl radicals for defluoroalkylation with alkenes or hydrodefluorination to prepare valuable difluoromethylene-incorporated carbonyl and aromatic compounds. Similar reactivity is also applicable for selective functionalization of pentafluoroethyl groups. In these reactions, o-phosphinophenolate works synergistically with an alkyl thiol HAT catalyst 52 . This work offers practical methods of synthesizing valuable geminal difluoro-substituted carbonyl and    were placed in a transparent Schlenk tube equipped with a stirring bar. The tube was evacuated and filled with argon (three times). To the mixture, anhydrous DMA (2 mL) were added via a gastight syringe under argon atmosphere. The reaction mixture was stirred under irradiation with violet LEDs (Kessil® 427 nm) in a HepatoChem photoreactor at room temperature for 24 h. The mixture was quenched with brine and extracted with ethyl acetate (3 × 10 mL). The organic layers were combined and concentrated under vacuo. The product was purified by flash column chromatography on silica gel.