Main

The synergistic interplay of precision synthesis1,2 and functional small-molecule design3 continues to be a major driver of innovation in both disciplines. In the vanguard of enabling technologies, fluorination has a venerable history in tailoring the physiochemical traits of promising active pharmaceutical ingredients (APIs), and the societal impact of Fried’s seminal work on fluorinated steroids is a compelling exemplar4. Function-driven synthesis thus continues to provide a powerful incentive to expand the current methodological arsenal under the auspices of atom and step efficiency5. In particular, the success of the geminal-difluoromethylene group in leading pharmaceuticals has stimulated much interest in the development of main group catalysis-based strategies to facilitate installation from readily available precursors6,7,8.

The prominence of fluorination patterns in contemporary drug discovery9,10,11,12,13,14 disguises the comparative scarcity of naturally occurring organofluorines in marine and terrestrial environments15,16. Although more than 5,000 halogen-containing natural products have been described so far17,18, and fluorine sources are accessible, it is manifest that nature has not been compelled to evolved fluorine biochemistry to any substantial degree19,20. This fluorous juxtaposition between natural and synthetic functional small molecules continues to provide opportunities for the conception and development of new molecular entities with geometries and physiochemistries that are not encountered in biology21,22,23, and it logically follows that this continues to expand the chemical space available for function-driven synthesis. A compelling exemplar is the bond-angle distortion that results from CH2 to CF2 replacement24, which renders the difluoromethylene group a validated bioisostere of oxygen in phosphate mimics25,26. This motif is finding increasing application in the small-molecule drug repertoire, with prominent examples including lubiprostone (Amitiza), tafluprost (Taflotan) and various 5-HT1D agonists (Fig. 1a)27. Motivated by the demand for new fluorinated modules for medicinal chemistry12,14, and cognizant of the emergent importance of alkyne-containing APIs such as efavirenz (Estiva) and levonorgestrel (PlanB One-Step), it was envisaged that a route to homopropargylic difluorides would address a gap in the discovery portfolio: this would provide isosteric surrogates of propargylic ethers and alcohols in which the electronegativity of the fluorine atoms would emulate the non-bonding electron pairs28.

Fig. 1: Development of a catalytic gem-difluorination of enynes.
figure 1

a, Bioisostere design and examples of bioactive molecules containing a CF2 or alkyne moiety. b, Hypervalent iodine-catalysed gem-difluorination of alkenes and the phenonium-ion rearrangement. c, Reaction blueprint to enable catalysis-based fluorinative alkyne-migration. The electron-rich alkyne is envisaged to be a competent proxy for phenyl, enabling the phenonium-ion rearrangement to be replaced by a formal 1,2-alkynyl shift via a stabilized vinyl cation.

Of the many enabling innovations that enable direct, geminal difluoromethylenation of alkenes, hypervalent iodine (I/III) catalysis has proven particularly powerful29,30,31,32,33,34,35,36,37,38,39. However, a precondition of this strategy is the requirement for substrates that undergo skeletal rearrangement to ensure that the desired 1,1-regioselectivity of the fluorination is reached (Fig. 1b)40. This restraint continues to limit the scope of the transformation to styrene derivatives in which a phenonium-ion rearrangement occurs41,42,43,44,45. Although the introduction of heteroatom substituents partially circumvents this limitation35,36,46,47,48, geminal difluorination in the presence of carbon-based groups, in the absence of aryl substituents43, remain conspicuously challenging. To address this, it was envisaged that enynes would be attractive substrates in which the electron-rich alkyne would serve as a phenyl proxy. This would enable the regiochemical paradigm predicated on the phenonium-ion rearrangement to be replaced by a formal 1,2-alkynyl shift via a stabilized vinyl cation (Fig. 1c). Homopropargylic fluorides would also enable direct access to homoallylic and alkyl difluorides, thereby expanding the impact of catalytic difluorinations enabled by I(I)/I(III) catalysis.

A catalytic cycle was conceived based on the in situ generation of an ArIF2 species, via a process of ligand exchange, that would promote an alkene-activation and fluorination sequence. Should the key rearrangement be successful, then the product cation would benefit from fluorine as a stabilizing auxiliary49. This would provide a facile route to homopropargylic difluorides, in which the alkyne handle would facilitate downstream functionalization.

Results and discussion

To validate the working hypothesis delineated in Fig. 1c, enyne S1 was prepared and exposed to catalysis conditions using various inexpensive aryl iodides, oxidants and HF sources (Table 1). Initially, p-TolI was combined with Selectfluor and amine•HF (1:7 ratio) in chloroform at ambient temperature. This enabled the desired homopropargylic fluoride 1 to be generated in 88% yield. Importantly, the vicinal regioisomer was not formed under these conditions, as determined by 19F NMR (<5%). However, in the absence of the catalyst, the vicinal difluoride was formed in 13% yield. A screen of electronically modulated catalysts confirmed the superiority of p-TolI, and revealed the following trend: p-Me > p-H > p-CO2Me > p-OMe. Modifying the amine:HF ratio or the oxidant were found to have a detrimental effect on the reaction outcome (Table 1).

Table 1 Reaction optimization

Having identified optimized conditions for the title reaction, the scope and limitations of the transformation were investigated. In the course of this process, reactivity differences were noted in response to subtle changes in the amine:HF ratio. This is in line with early observations related to the impact of trifluoroacetic acid on the reactivity of iodobenzene dichloride50. For that purpose, a gradient of amine:HF ratios was considered starting from 1:4.5 and increasing to 1:7.0 (denoted A–F). For simplicity, only the most effective conditions are indicated in Table 2. Initially, the impact of modifying the capping aryl group was investigated while keeping the alkene substituent constant (R = Me). This enabled a series of gem-difluorides to be generated, and demonstrated functional-group compatibility with electron-withdrawing groups, halogens and small alkyl fragments (110, up to 83%). In the case of product 2, it was possible to rigorously establish the molecular connectivity by single-crystal X-ray diffraction (Table 2; CCDC 2256836). Gratifyingly, the method also proved to be compatible with medicinally relevant heterocycles such as pyridines, quinolines and morpholines (1113, up to 56%). Furthermore, it was possible to replace R = Me with R = CH2X (X = Br and Cl) to create linchpins that could be functionalized at the proximal C(sp3) position (15 and 16, up to 91%). Finally, the compatibility of the method with more complex natural-product-derived scaffolds was validated (1719).

Table 2 Establishing the scope of aryl alkynes

To advance the scope of the transformation beyond aryl-substituted enynes, aliphatic derivatives were then explored with a view to applying the method to target synthesis (Table 3). Simple alkyl and cycloalkyl derivatives were tolerated (20 and 21, up to 63%) and it was possible to introduce functionality in the form of phthalimides (22, 50%) and ethers (23, 71%). Substrates with potentially challenging benzylic/propargylic positions such as 24 were smoothly converted to the desired product. The transformation was found to be chemoselective for the enyne versus cinnamoyl motifs (25, 64%), and alkynoic esters (26, 63%), tosylates (27, 56%) and alcohols (28, 41%) were compatible. Modifying the alkyl substituent was possible (29, 30) and enabled the 1,2,2-trifluoro motif to be generated in a facile manner. The introduction of more complex heterocycles, such as in febuxostat (Adenuric) derivative 32, is an encouraging validation of the method in a drug-discovery setting. With a view to accessing the parent motif derived from the unsubstituted enyne, the triisopropylsilyl (TIPS)-acetylene 33 was prepared in 67% yield.

Table 3 Expanding the scope to alkyl and silyl alkynes

To demonstrate the synthetic utility of this geminal difluorination of enynes, two representative experiments were validated on a 4.0 mmol scale (Fig. 2a), and a series of product derivatization reactions were conducted (Fig. 2b). Initially, alkyne 20 was fully and partially reduced51 to the alkane and alkene products 34 and 35, respectively. To demonstrate the value of the homopropargylic fluorides in heterocycle formation, compound 4 was converted to the quinoxaline 36 through Ru-catalysed oxidation of the alkyne and concomitant condensation with 1,2-phenylendiamine52,53. Desilylation of compound 33 with tetrabutylammonium fluoride (TBAF) furnished the terminal alkyne 37 in 85% yield: this could then be processed further to triazole 38 via a copper-catalysed click reaction54. In situ deprotection of 33 and subsequent Sonogashira cross-coupling proved facile, enabling the electron-rich aryl alkyne 39 to be generated in 64% yield. Because electron-rich enynes undergo uncatalysed side reactions with the Selectfluor55, this approach enables the geminal difluorination products to be generated by an alternative route. Finally, the compatibility of the motif under Suzuki–Miyaura conditions was demonstrated through the generation of compound 40 (93% yield)56.

Fig. 2: Synthetic applications.
figure 2

a, Scale-up experiments. b, Product derivatization. Conditions. (i) Conversion of 20 to form 34: Pd/C (10 mol%), H2, MeOH (0.1 M), r.t., 24 h. (ii) Semi-reduction to generate the Z-alkene 35: TiCl2Cp2 (10 mol%), LiAlH4 (2.0 equiv.), tetrahydrofuran (THF) (0.2 M), r.t., overnight. (iii) Generation of quinoxaline 36: (1) RuCl3 (1 mol%), PhI(OAc)2 (3.0 equiv.), DCM, H2O (4:1, 0.2 M), r.t., 3 h; (2) saccharin (5 mol%), 1,2-phenylendiamine (1.1 equiv.), MeOH (0.2 M), r.t., 12 h. (iv) TIPS deprotection to generate terminal alkyne 37: TBAF (2.0 equiv.), THF (0.4 M), r.t., 2 h. (v) Formation of triazole 38 via a copper-catalysed click reaction: CuTc (10 mol%), 4-acetamidobenzenesulfonyl azide (1.2 equiv.), toluene (0.2 M), r.t., 12 h. (vi) In situ deprotection of 33 and subsequent Sonogashira cross-coupling: Pd(PPh3)2Cl2 (4.5 mol%), CuI (5 mol%), 4-iodoanisole (1.2 equiv.), NEt3 (7.0 equiv.), TBAF (2.0 equiv.), THF (0.2 M), 45 °C, 12 h. (vii) Suzuki-coupling to generate compound 40: Pd(PPh3)4 (10 mol%), 2-furanboronic acid (2.5 equiv.), K2CO3 (2.5 equiv., 2 M in H2O), dimethoxyethane (DME) (0.2 M), 85 °C, overnight. c, Synthesis of CF2-modified palmitic acid. Isolated yields are given. 19F NMR yields are given in parentheses and determined by 19F NMR using ethyl fluoroacetate as an internal standard.

A concise route to the CF2-modified palmitic acid 44 was conceived to validate the method in target synthesis (Fig. 2c). Initially reported by O’Hagan and co-workers in the context of a wider study of the conformational preferences of palmitic acids and nonadecane containing CF2 groups57, this molecule remains a benchmark in difluorination method development58. With the aim of complementing the existing reagent-based approaches, enyne 41 was exposed to the catalytic geminal-difluorination conditions: this furnished the key intermediate 42 in 51% yield. Chemoselective reduction of the alkyne and saponification of the methyl ester enabled the desired compound 44 to be generated in 98% yield over two steps.

Finally, control reactions were performed to establish that deletion of the alkyl substituent was tolerated (Fig. 3). Interest in the difluoromethyl group as a surrogate of primary alcohols28 renders such products appealing in the wider context of molecular design on account of their hydrogen-donor character59. Pleasingly, both the aryl- and alkyl-substituted enynes 45 and 47 could be processed to their respective homopropargyl difluoride products 46 and 48, respectively. Replacing the substituent with an aryl group (49 Ar = p-CF3) was then explored to identify which regioisomer was predominantly formed. The isolation of compound 50 as the sole product of the reaction (40% yield) is consistent with the 1,2-shift out-competing phenonium-ion rearrangement. The skeletal rearrangement that is central to the working hypothesis was supported by deuterium labelling to generate 29-d (56%, 76% D incorporation; Fig. 3c).

Fig. 3: Control experiments.
figure 3

a, Removal of the alkyl group. Enynes 45 and 47 could be converted to 46 and 48, containing the difluoromethyl group, respectively. b, Investigation of regioselectivity. Compound 50 was obtained as the sole product. c, Deuterium labelling experiment. Deuterium atom incorporated exclusively at the propargylic position, unambiguously demonstrating that 1,2-alkynyl migration took place. Isolated yields are given.

Conclusions

The direct, geminal difluorination of alkenes under the auspices of hypervalent iodine catalysis remains a powerful paradigm to expand organofluorine chemical space for contemporary drug discovery. In situ-generated λ3-iodanes regulate regiocontrol by inducing C(sp3)–F bond-forming/rearrangement sequences with exquisite efficiency: the latter step is conditional on substrates that are predisposed to undergo a phenonium-ion rearrangement. To circumvent this limitation, enynes have been validated as competent substrates that deliver the desired 1,1-selectivity, where the phenonium-ion rearrangement can be replaced by a formal 1,2-shift of the alkyne. Computational support for the tentative mechanism outlined in Fig. 1 is available in Supplementary section 1.7. Utilizing the alkyne as a phenyl proxy, it has been possible to achieve the title reaction and deliver homopropargylic difluorides that are highly amenable to downstream functionalization. A broad substrate scope is demonstrated (>30 examples) together with selected derivatization protocols, as well as a short, catalysis-based synthesis of CF2-modified palmitic acid. It is envisaged that this enabling method will find application in the conception of new drug-discovery modules.

Methods

General procedure for 1,1-difluorination of enynes

Unless otherwise stated, a Teflon vial was equipped with a 1-cm stirring bar followed by the addition of enyne (0.2 mmol, 1.0 equiv.), p-iodotoluene (9 mg, 0.04 mmol, 20 mol%) and CHCl3 (0.5 ml). The stated amine:HF mixture was added (0.5 ml) via syringe. After stirring for 1 min, Selectfluor (106 mg, 0.3 mmol, 1.5 equiv.) was added in one portion. The reaction vessel was then sealed with a Teflon screw cap. After stirring (350 r.p.m.) at ambient temperature for 24 h, the reaction mixture was poured into 100 ml of a saturated solution of NaHCO3 (caution! generation of CO2!). The Teflon vial was rinsed with dichloromethane (DCM) and dropped into another flask of saturated aqueous solution of NaHCO3 to guarantee the removal of excess HF. The organics were extracted with DCM (3 × 30 ml), the combined organic layers were dried over Na2SO4, filtered, and the solvent was carefully removed under reduced pressure. An internal standard (ethyl fluoroacetate) was added to the crude residue and the NMR yield was analysed by 19F NMR spectroscopy against the internal standard. The NMR sample was recombined with the crude residue and purification by column chromatography or preparative thin-layer chromatography yielded the desired product.