Chemoselective catalytic hydrodefluorination of trifluoromethylalkenes towards mono-/gem-di-fluoroalkenes under metal-free conditions

Fluorine-containing moieties show significant effects in improving the properties of functional molecules. Consequently, efficient methods for installing them into target compounds are in great demand, especially those enabled by metal-free catalysis. Here we show a diazaphospholene-catalyzed hydrodefluorination of trifluoromethylalkenes to chemoselectively construct gem-difluoroalkenes and terminal monofluoroalkenes by simple adjustment of the reactant stoichiometry. This metal-free hydrodefluorination features mild reaction conditions, good group compatibility, and almost quantitative yields for both product types. Stoichiometric experiments indicated a stepwise mechanism: hydridic addition to fluoroalkenes and subsequent β-F elimination from hydrophosphination intermediates. Density functional theory calculations disclosed the origin of chemoselectivity, regioselectivity and stereoselectivity, suggesting an electron-donating effect of the alkene-terminal fluorine atom.

In this work, we described a method for metal-free catalytic activation of C-F bonds in trifluoromethylalkenes under mild conditions. gem-Difluoroalkenes and terminal monofluoroalkenes can be chemoselectively produced in almost quantitative yields by simple adjustment of the amount of the terminal reductant PhSiH 3 (Fig. 2c).

Results and discussion
Investigation of the reaction conditions. Building on our previous work 59 , we envisioned that diazaphospholenes of super hydricity 60-62 may provide a good chance to realize HDF of trifluoromethylalkenes via an S N 2' path 63 . A preliminary attempt indicated that the reaction of α-trifluoromethyl-styrene 2a with diazaphospholene 1 primarily gave the hydrophosphination intermediate A. Only handful HDF products 3a and 1-F 64 were obtained ( Fig. 3a and Supplementary Figs. 1 and 2). However, the intermediate A was completely converted to 3a and 1-F at an elevated temperature (70°C). This is the rare example of β-F elimination enabled by a non-metal neutral reductant, rather than by the well-established metal systems 65 . When a second equivalent of diazaphospholene 1 was used, the in situ-generated gemdifluoroalkene 3a was further hydrodefluorinated at 70°C to afford the dual-HDF product 4a in an almost quantitative yield after 1 h ( Fig. 3b   Scope of mono-HDF reactions. As expected, mono-HDF of trifluoromethylalkenes 2 in CH 3 CN with a 5 mol% catalyst loading and 0.33 equiv. of PhSiH 3 (i.e., 1 equiv. of Si-H bonds) proceeded smoothly to give gem-difluoroalkenes 3. The high-polarity solvent CH 3 CN favors polar hydride transfer. Other solvents, like toluene and THF, gave the mono-HDF products in <10% yields. The reaction showed a wide substrate scope, as shown in Fig. 4. Generally, reductions performed with either electron-rich or -deficient substrates all occurred facilely to give almost quantitative yields. Substrate 2a furnished gem-difluoroalkene 3a in 99% yield after 12 h at 70°C. Phenyl-substituted 2b and non-substituted 2c were efficiently reduced at lower temperatures. The reactions of substrates with electron-donating groups such as methoxy (2d), methylthio (2e), and dimethylamino (2 f) also worked well, but needed slightly higher reaction temperatures and longer reaction times. Substrates bearing electron-withdrawing groups (2g-2 l) showed high reactivities and gave the corresponding gem-difluoroalkenes 3g-3 l in 91-99% yields. Notably, several functional groups (2j-2 l) that are incompatible with the strong bases used in Wittig, Julia-Kocienski, and Homer-Wadsworth-Emmons reactions, or with strong nucleophiles in S N 2'-type reactions, are well tolerated in our system. The known reduction 67 of aryl ketones by diazaphospholenes was completely depressed by the higher electrophilicity of the trifluoromethyl group in 2j. Substrate 2 m with a susceptible acetal moiety furnished the product 3 m quantitatively (99%) in a prolonged reaction time. The reaction was also applicable to the naphthalene analog 2n. The excellent performance in heterocyclic systems (3o-3r) shows that this reaction can chemoselectively reduce the trifluoromethyl moiety while leaving other unsaturated structures intact 68,69 . For tri-substituted alkenes, only the Z isomer of 2 s is applicable (the E isomer of 2 s did not work, see SI for details), and 3 s is produced in 91% yield. This is probably because of a steric effect in the initial hydride transfer. The low efficiency of the reaction of the endocyclic alkene 2t is probably also ascribable to a steric effect. Aliphatic trifluoromethylalkenes did not work in our conditions due to the low electrophilicity.
Scope of dual-HDF reactions. Because of the super-hydricity of the catalyst 1, the produced gem-difluoroalkenes 2 continued to react with the hydride 1. This explains why addition of a further 0.33 equiv. of PhSiH 3 gave dual-HDF products. Such a result indicates that chemoselective HDF can be achieved by simply adjusting the stoichiometry of the reactants. Accordingly, the preparation of monofluoroalkenes 4 by dual-HDF with 0.7 equiv. of PhSiH 3 (i.e., approximately 2 equiv. of Si-H bonds) was investigated. The results are summarized in Fig. 5. Overall, the reduction showed prominent chemoselectivity for most CF 3containing substrates 2, and monofluoroalkenes 4 were generated quantitatively with good to excellent stereoselectivities, although the slightly elevated temperature (80°C) was necessary for several electron-rich substrates. For examples, substrates 2a-2c gave monofluoroalkenes 4a-4c quantitatively with good E/Z stereoselectivities. Electron-donating groups (2d-2 f) did not significantly depress successive C-F activations, and 4d-4 f were obtained in good to excellent yields (75%-99%). Various electronwithdrawing groups (2g-2 l) were also well tolerated (4g-4 l, 92-99% yields). Notably, the acetal moiety of 2 m was not sensitive to the dual-HDF conditions. Naphthalene 2n gave 4n in an excellent yield and with good diastereoselectivity. The heterocyclic trifluoromethylalkenes 2o-2r were also compatible and gave products 4o-4r in 64-99% yields with moderate to good E/Z stereoselectivities. Similarly to mono-HDF, only the Z isomer of 2 s showed reactivity in dual-HDF. The exocyclic trifluoromethylalkene 2t did not react at all. Note that further increase of the amount of PhSiH 3 could remove the third fluoride from some substrates with electron-withdrawing groups.
Synthetic applications. To show the versatility of the present system, we used its potential for modifying drug molecules. Indometacin is a commonly used drug, which has significant antipyretic, anti-inflammatory, and antirheumatic activities 70 . As shown in Fig. 6, under our reaction conditions the indometacin derivative 5 is effectively transformed into either the mono-HDF product 6 in 85% yield or the dual-HDF product 7 in 30% yield, depending on the amount of the reductant PhSiH 3 .
Mechanistic investigations. Density functional theory (DFT) calculations were used to gain mechanistic insights into the outstanding catalytic performance of diazaphospholene 1 in HDF. The calculations were performed at the (SMD)-M06-2X/6-311 + +G(2df,2p)//(SMD)-M06-2X/6-31 + G(d) level of theory 71,72 with trifluoromethylalkene 2c as the template substrate. The results are shown in Fig. 7. During the first HDF, hydride transfer from diazaphospholene 1 to 2c proceeds via TS1, with a Gibbs activation barrier of 18.9 kcal mol −1 , to generate the hydrophosphination intermediate A, in line with our room-temperature reaction conditions for initial hydrophosphination (Fig. 3a). Exothermic cis-β-F elimination from intermediate A furnishes the mono-HDF product 3c via TS2, with a 20.7 kcal mol −1 barrier. This suggests a need for elevated temperatures.
The possible paths for the second HDF are more complicated because the hydride can be transferred to either the C1 or C2 site of 3c, to produce intermediates B or B', respectively (Fig. 8a). Intuitively, the C2 site is preferred, because of the strong electronwithdrawing ability of the fluorine atom. However, our experimental and DFT results uniformly led to good regioselectivity for the hydride transfer to the C1 site, which proceeds via TS3 with an energy barrier about 15 kcal mol −1 lower than the transfer to the C2 site via TS3′. This result is primarily attributed to the following two factors: (1) the stabilizing effect of the aromatic ring on the incipient benzyl carbanion during hydride addition at the C1 site, and (2) the repulsive interaction between the fluorine lone pairs of electrons and π-electrons, which makes the C1 site relatively electron deficient (Fig. 8b). Our results suggest that the alkene terminal fluorine atom has an electron-donating effect rather than the conventional electron-withdrawing effect. The NPA (natural population analysis) analysis also supported the regioselectivity in the second hydride transfer process.
The energy profile for the second HDF is given in Fig. 9. As shown, hydride transfer from 1 to 3c via TS3 has a higher Gibbs activation barrier (24.0 kcal mol −1 ) than that of TS1 in the first HDF (18.9 kcal mol −1 ). This 5.1 kcal mol −1 difference guarantees excellent chemoselectivity for mono-and dual-HDF. A second cis-β-F elimination from B can give two isomers of the monofluoroalkene 4c. Based on the difference between TS4 and TS5 (0.9 kcal mol −1 ), the E-isomer of the monofluoroalkene, E-4c, is preferentially formed. This stereoselectivity can be explained by using Newman projections of the conformers involved in cis-β-F elimination (Fig. 10a). The preferred conformer, which results in the E isomer being the major product, clearly diminishes electronic repulsion between the fluorine atom and the aromatic ring. Indeed, the stereoselectivity (E/Z = 82/18) calculated from the Gibbs activation energies for both β-F elimination steps is in good agreement with the experimental result (E/Z = 85/15).
A plausible mechanism for this HDF process is outlined in Fig. 10b. First, hydride transfer from the catalyst 1 to trifluoromethylalkenes 2 furnishes the hydrophosphination intermediate A. Subsequent β-F elimination gives the mono-HDF products 3 and 1-F. After complete consumption of 2, the excess PhSiH 3 regenerates catalyst 1, which renders a second HDF to yield monofluoroalkenes 4.
In summary, we have developed a method for diazaphospholene-catalyzedchemoselective C-F bond activation of trifluoromethylalkenes, which enables the convenient construction  NMR spectroscopic studies showed a hydrophosphination intermediate, which subsequently underwent β-F elimination at elevated temperatures. This metal-free strategy is applicable to a broad range of trifluoromethylalkenes. It shows good functional group tolerance and gives almost quantitative yields of both mono-and di-hydrodefluorinated products. DFT calculations suggested that the good chemoselectivity between mono-and dual-HDF stems from differences in the substrate electrophilicities, and the regioselectivity for hydride transfer to gem-difluoroalkenes is partly attributed to the electron-donating ability of the alkene terminal fluorine atoms. Other diazaphospholene-catalyzed HDF reactions are currently being investigated in our laboratory.

Methods
General information. Catalyst 1 has been synthesized and characterized in our previous work 59,73,74 . Trifluoromethyl alkenes 2 were synthesized according to references (see Supplementary Information for details). Other reagents and solvent were purchased from J&K or TCI Chemicals and used without further purification unless specified otherwise. Acetonitrile was purchased from J&K Chemical (99.9%, Extra dry, water <10 ppm, J&K seal) and degassed and distilled by standard methods. Reaction temperature refers to the temperature of an aluminum heating block or a silicon oil bath, which was controlled by an electronic temperature modulator from IKA.

ARTICLE
Reactions. All hydrodefluorination reactions were carried out in dry glass wares under an argon atmosphere using Schlenk technique throughout the reaction procedures.