Developing novel catalytic methods for the site-selective hydrodefluorination (HDF) is of long-standing interest in organic chemistry due to the widespread presence of C-F bonds in agrochemicals, pharmaceuticals, and materials [1-9]. Mono- and di-fluorinated compounds have significant potential in drug discovery research due to their hydrogen bonding abilities, exceptional pharmacokinetic (PK) properties, and enzymatic stability [10-13]. However, late-stage modification of the R-CF₃ group into an R-CHF₂ or R-CH₂F is exceedingly difficult due to the strong C–F bond dissociation energy [2]. Indeed, the reduction and replacement of the first F atom from the CF₃-group requires a strong reducing reagent or catalyst due to the high reduction potential. Traditionally, DAST or Selectfluor was used as a fluorinating agent to incorporate gem-difluoromethylene groups with the generation of a stoichiometric amount of waste [14-16]. Given that the mono-defluorinated compound R-CHF₂ has less reduction potential than the parent molecule R-CF₃, controlling selectivity in HDF to mono-, di-, or tri-defluorination is exceptionally challenging [17].

Selective HDF has numerous advantages, including its complementarity to classical C-F bond formation, the avoidance of multistep synthesis, a more cost-effective process, and the possible use of feedstock chemicals such as trifluoroacetamide or trifluoroacetate [17,18]. There are several methods known for C–F bond formations [2,10,14-16]. However, methods for selective HDF are limited [18-24]. The functionalization of CF₃ groups generally proceeds through heterolytic cleavage of a C–F bond, affording difluoro-substituted carbocation or carbanion intermediates. In contrast, trapping of homolytically cleaved fluoroalkyl radical intermediates to new bond-forming reactions is less explored because of the high bond dissociation energy (BDE) of the C–F bond [17,18]. Recently, several research groups [17,18,21,25-31] disclosed strategies to eliminate an F-atom from CF₃- substituted arenes to generate aryldifluoromethyl radicals via single-electron reduction.

An elegant new methodology has been developed by Wang and Houk group [17] for fluorine elimination of polyfluorinated aliphatic esters and amides via a boryl radical enabled spin-center shift (SCS) [32] strategy (Scheme 1a). Furthermore, the Molander [30] and Glorius [31] groups have independently reported selective defluorination of trifluoroacetate/amide via a HAT process. However, the reported methods are limited to the mono-defluorination and liberate a stoichiometric amount of CO₂, which may hamper the reaction when olefin-trifluoroacetamide was used and lead to hydrocarboxylation as a side reaction.

In recent years, consecutive photoinduced electron transfer (conPET) and multiphoton excitation catalysis have emerged as strategies to overcome the thermodynamic limitations of visible-light photoredox catalysis [33-45]. In this context, conPET photocatalysts, mainly organic dyes, have been explored for dehalogenation reactions (mainly iodo, bromo or chloro arenes) [33,35,37,46]. Similarly, multiphoton excitation Ir-catalysis was applied for reduction and hydrofunctionalization of olefins via a two-photon tandem photoredox process [39]. However, conPET or multiphoton catalysis has not been used for selective defunctionalization, especially for HDF of non-activated CF₃-groups, possibly due to the high redox potential of the photo-excited anionic catalyst, which often leads to poor selectivity. We hypothesized that by employing this concept we may achieve a high selectivity for mono-, di, tri-HDF of trifluoroacetamide motifs. Thus, the successful accomplishment of the proposed concept would not only provide a novel protocol for the sequential HDF of trifluoroacetamide but would represent the first example of a selective defluorination using multiphoton photoredox catalysis and a conPET process (Schemes 1b and c).

At the start of our investigation, trifluoro-N-phenylacetamide was selected as a model substrate (1a) to investigate the selective hydrodefluorination and to form specifically monofluorinated, difluorinated, or fully hydrodefluorinated phenylacetamide. A detailed evaluation of the HDF reaction (complete optimization process given in Supporting information) demonstrated that Ir((dFCF₃ppy₂)₂dtbbpy)PF₆ was the optimal PC in combination with N,N-diisopropylethylamine (DIPEA) as a reductant and hydrogen atom donor to give the desired difluoro- (2a) and monofluoro-(3a) N-phenylacetamide in 70% and 14%, respectively (Table 1, entry 1). It is worth mentioning that the fully defluorinated product 4a was not observed.

Control experiments revealed that light, PC, and the reductant are essential for this transformation (Table 1, entries 2–4). In the absence of water, the selectivity was significantly reduced, and 2a was obtained in 35% NMR yield (Table 1, entry 5). Changing the reductant from DIPEA to Hantzsch ester failed to give the desired product (Table 1, entry 6). Previously reported photocatalytic procedures required sodium formate (HCOONa) and thiol as HAT substrates for the HDF [30,31]. Hence, we examined HCOONa and HCOONa/cyclohexane thiol as reductants instead of DIPEA. However, no product was observed (Table 1, entries 7 and 8). Interesting, our newly developed system operates with only DIPEA as a reductant for HDF. It is known that tertiary amines will be oxidized by the PC to form an amino radical cation that enables back electron transfer to deactivate the PC radical anion, which is one of the significant side reactions in this catalytic system [47]. Noteworthy, the tandem photoredox catalyst Ir((ppy)₂dtbbpy)PF₆ that was used for olefin reduction reaction in the previous report [39], and other Ir-PCs and organic dyes gave comparatively lower yield and selectivity than Ir((dFCF₃ppy₂)₂dtbbpy)PF₆ (PC-Ⅰ) (Table 1, entries 9–13, Table S5 in Supporting information).

After successfully optimizing the reaction condition, the applicability of this protocol was tested with different trifluoroacetamide substrates for mono-defluorination, giving pharmaceutically important gem-difluoromethylene analogues in good to excellent yield (Scheme 2). More challenging electron-neutral/donating trifluoroacetamide derivatives underwent this protocol to give gem-difluoromethylene acetamides in up to 76% yield (2a-2g). Interestingly, the developed protocol is also suitable for sterically crowded substrates (2h-2j). In addition, electron-deficient and heteroaromatic compounds containing CF₃- group were selectively mono-defluorinated to give desired compounds in good to excellent yields. Functional groups such as chloro (Cl), cyano (CN), S-Me, benzyl, and ester were tolerated (2i, 2j, 2m-2p). Synthetically important drug-like molecules could be applied and gave the desired product in 58%−64% yields with good selectivity (2u-2w). In the case of electron-withdrawing and complex molecules, 2 equiv. of base are sufficient for the conversion, while sodium acetate and acetic acid were added for the improvement of selectivity (Table S2 in Supporting information).

The scope of fluoroacetamide was also investigated with slightly different conditions. Interestingly, organic dye 3DPA2FIPN (PC-Ⅱ) was a suitable catalyst for dihydrodefluorination. The optimal combination with DMSO:MeCN as a solvent at 390 nm gave more than 70% selectivity (detailed optimization see Tables S3 and S4 in Supporting information). Likewise, both electron-donating and electron-withdrawing groups containing trifluoroacetamides underwent dihydrodefluorination smoothly and furnished the desired products in good to high yields (Scheme 3).

This protocol also has good functional group compatibility (3b-3r). Indeed, late-stage functionalization of complex molecules, especially in a selective fashion, is a challenging and key component in drug-discovery research. In this context, compounds 3s-3u were synthesized in 62%−70% yield.

Next, we explored the complete hydrodefluorination of the trifluoroacetamide derivatives. In this section, 3DPA2FBN (PC-Ⅲ) or Ir-PC-Ⅰ was used depending on the electronic nature of the substrates. Similarly, wavelength, solvent, and reaction time play a crucial role in complete defluorination (Tables S3 and S4). Some sensitive functional groups were unaffected by the reaction (Scheme 4, 4f–4r). Bioactive compounds such as steroid derivatives gave desired product in 61% yield (4s-4v).

Essential components within photoredox catalysis, encompassing processes like multiphoton excitation and electrophotocatalysis, have been proposed to involve photosensitizers reduced by a single electron. Nonetheless, the study of these diminished chromophores has encountered certain limitations, which has hindered a thorough investigation into the complicated mechanisms of the electron transfer processes they encompass. In this particular scenario, Knowles group [42] has recently disseminated research outcomes pertaining to iridium photocatalysts that have been singularly reduced and can be isolated. The disclosed spectroscopic characteristics of these catalysts, emphasizing reduction centered on ligands, provide compelling support for our hypothesis. Based on the previous reports [34,39,40,42] and the proposed hypothesis, we conducted quenching studies with Ir-PC and organic dyes. The Ir-photocatalyst (PC-Ⅰ) is exhibiting an excited state redox potential of E_1/2(Ir^Ⅲ*/Ir^Ⅱ) = +1.21 V vs. SCE, which can potentially oxidize the DIPEA (E_1/2 = +0.78) vs. SCE (MeCN) [48] to its radical amino cation and to form the reduced photocatalyst (E_1/2(Ir^Ⅱ/Ir^Ⅲ) = −1.37) (Scheme 1c, Ⅱ), as supported by fluorescence quenching experiments (Scheme 5A). Thermodynamically, the resulting photo-excited catalyst (Scheme 1c, Ⅲ) cannot reduce the trifluoroacetamide (1a) via SET because of the reduction potential of the substrate (E_1/2 = −2.26 V vs. SCE), which is almost a 1 V difference with the photocatalyst, a result further supported by emission quenching studies (Scheme 5B). The substrate (1a) does not interact with the excited Ir^Ⅲ* (PC-1*), while DIPEA quenched the excited state PC-1* (Schemes 5A and B). The involvement of the second excited state of [PC-2˙⁻*] was demonstrated by emission quenching studies, where PC-Ⅰ with 500 mmol/L of DIPEA was irradiated at 440 nm light. The resulting slow quenching rate of substrate 1a possibly arises from the poor solubility of trifluoroacetamide (Scheme 5C). However, the new emission spectra shift was observed from 470 nm to 522 nm, indicating the formation of [PC-2˙⁻*] (Scheme 5D). Further, to support the high reduction potential of the in situ formed [PC-2˙⁻*], 1,1-diphenylethylene (E_1/2 = −2.25 V vs. SCE) [39] was tested as a quencher. The resulting Stern-Volmer plot was linear, confirming the interaction between the [PC-2˙⁻*] and 1,1-diphenylethylene (Fig. S10 in Supporting information). Like Ir-PC (PC-1), the organic dyes 3DPA2FBN and 3DPAFIPN were quenched with DIPEA, confirming the excited state PC-1* interacted with DIPEA and also yielding linear Stern-Volmer plots. In addition, the emission spectra of both organic dyes also shifted towards the lower wavelength region (Figs. S13 and S16 in Supporting information); this might be the reason that di- and tri- defluorination occurred at 390 nm.

The bond dissociation energy (BDE) of an α-C-H bond of DIPEA [48] radical cation is 90.7 kcal/mol which is relatively low compared to that of the C-F bond (BDE = 118.1 kcal/mol) in trifluoroacetamide [49]. Hence, halogen atom abstraction is not likely to operate in the mechanistic cycle, which is further supported by density functional theory (DFT) calculation (Scheme 5E). The calculation showed that the fluorine atom abstraction by α-amino radical is kinetically (31.6 kcal/mol energy barrier) and thermodynamically (7.5 kcal/mol less stable) unfavored. Next, we examined the source of hydrogen atoms (DIPEA, H₂O, and MeCN) for HDF. The reaction was conducted using model substrate 1a under condition A with the addition of D₂O instead of H₂O. The product 2a was obtained in 59% isolated yield along with 15% mono deuterium incorporated 3a (Fig. S2). A similar tendency was noted in the presence of D₂O and CD₃CN. Therefore, we conclude that the major hydrogen atom comes from DIPEA which also acts as an effective hydrogen atom donor for the formation of PC-2 via reduction and semisaturation of the ancillary dtb-bpy ligand.

In conclusion, we have developed a novel and straightforward methodology for the site-selective hydrodefluorination of trifluoroacetamides. Applying this process, a wide range of molecules have been selectively defluorinated to give mono-, di-fluoro substituted acetamides and simple acetamide derivatives. This protocol also functions for late-stage functionalization of drug derivatives. Mechanistic and spectroscopic studies demonstrate that the multiphoton tandem photoredox catalyst reaches a high reduction potential under irradiation, allowing it to carry out selective transformations in a controlled manner.

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.

Jiaqi Jia: Methodology, Writing – original draft, Writing – review & editing, Conceptualization, Data curation, Formal analysis, Investigation. Kathiravan Murugesan: Writing – original draft, Writing – review & editing. Chen Zhu: Investigation, Writing – original draft, Writing – review & editing. Huifeng Yue: Investigation. Shao-Chi Lee: Investigation. Magnus Rueping: Writing – review & editing, Funding acquisition, Project administration, Resources, Supervision.

We thank E. Dempsey (Colorado School of Mines) for the assistance with MALDI measurement, and B. Hohman (University of Colorado) for the help with DLS test. S. H. thanks the partial financial support by the Summer Graduate School Fellowship from University of Colorado Boulder.

Supplementary material associated with this article can be found, in the online version, at doi:10.1016/j.cclet.2024.109866.