Highly selective nickel-catalyzed gem-difluoropropargylation of unactivated alkylzinc reagents

In spite of the important applications of difluoroalkylated molecules in medicinal chemistry, to date, the reaction of difluoroalkylating reagents with unactivated, aliphatic substrates through a controllable manner remains challenging and has not been reported. Here we describe an efficient nickel-catalyzed cross-coupling of unactivated alkylzinc reagen\ts with gem-difluoropropargyl bromides. The reaction proceeds under mild reaction conditions with high efficiency and excellent regiochemical selectivity. Transformations of the resulting difluoroalkylated alkanes lead to a variety of biologically active molecules, providing a facile route for applications in drug discovery and development. Preliminary mechanistic studies reveal that an alkyl nickel intermediate [Ni(tpy)alkyl] (tpy, terpyridine) is involved in the catalytic cycle.

T he important role of organofluorinated compounds in medicinal chemistry, agrochemistry, and materials science 1, 2 has triggered substantial efforts in developing new and efficient methods for the controllable introduction of fluoroalkyl groups into organic molecules 3,4 . Particularly, the preparation of difluoroalkylated compounds have received much attention over the past few years [5][6][7] due to the unique properties of difluoromethylene group (CF 2 ) that can improve the metabolic stability of biologically active molecules and lead to enhanced acidity of its neighboring group, increased the dipole moments as well as conformational changes 8,9 . To date, the transition-metalcatalyzed difluoroalkylations have emerged as a useful strategy for the construction of Ar-CF 2 R bond and thus to access difluoroalkylated arenes efficiently 6,7,[10][11][12][13][14] . However, it remains challenging to adapt this strategy to the formation of Csp 3 -CF 2 R bond 15 . To the best of our knowledge, the reaction of difluoroalkylating reagents with unactivated, aliphatic substrates catalyzed by transition-metal has not been reported so far. Although some approaches to difluoroalkylated alkanes exist, they mainly rely on the transformation of functional groups, such as carbonyl, olefin, and alkyne [16][17][18][19][20] . We envisioned that if the transitionmetal-catalyzed direct introduction of a CF 2 group into aliphatic chain at any position is successful, it would be a significant advance and provide an effective and general method for siteselective synthesis of difluoroalkylated alkanes.
Accordingly, herein, we demonstrate the feasibility of transition-metal-catalyzed cross-coupling of readily available alkylzinc reagents with gem-difluoropargyl bromides. The use of gem-difluoropropargyl bromides is because of their versatile synthetic utility of the carbon-carbon triple bond and their ready availability 21,22 . Thus, this type of precursor would provide attractive and useful difluoroalkylated structures for applications in medicinal chemistry and materials science. However, efficient methods to such a valuable structure are very limited 15,23 . As part of our ongoing study in transition-metalcatalyzed fluoroalkylations 11,12,[24][25][26] , herein, we describe a highly selective Ni-catalyzed cross-coupling between unactivated alkylzinc reagents and gem-difluoropropargyl bromides [27][28][29][30][31] . This method has several advantages: broad substrate scope, including a variety of unactivated alkylzinc reagents and gem-difluoropropargyl bromides; high efficiency and excellent regiochemical selectivity for the formation of gem-difluoropropargylated products; a catalyst loading of NiCl 2 or Ni(dppf)Cl 2 (dppf, 1,1′-bis(diphenylphosphino)ferrocene) as low as 0.5-5 mol%; and practical for gram-scale production.

Results
Optimization of the Ni-catalyzed cross-coupling. We began our initial studies on the cross-coupling of 3-phenylpropylzinc bromide 1a with (3-bromo-3,3-difluoroprop-1-yn-1-yl)triisopropylsilane 2a in the presence of nickel catalyst with a variety of diamine ligands (Table 1, entries 1-8). However, only low yields of desired product 3a were obtained. To our delight, switching diamine ligand with a triamine ligand (tpy) dramatically improved the yield of 3a and no allenic side product was observed (entry 9). The beneficial effect of tpy on the current nickelcatalyzed process is in sharp contrast to previous work, in which diamine ligands were the best ligands for nickel-catalyzed difluoroalkylations 24,32,33 . We also found that the reaction was insensitive to the nature of nickel catalysts (entries [10][11][12][13][14]. All the tested nickel salts showed good reactivities and Ni(dppf)Cl 2 (2.5 mol%) was the best choice, providing 3a in 94% yield upon isolation (entry 17). Alternatively, the use of 4,4′,4″-tri-tert-butylterpyridine (tpy′) as a ligand also produced 3a in a comparable yield (entry 18), thus providing a complementary ligand of choice for this process. No product was obtained in the absence of nickel catalyst (entry 19) and only 9% yield of 3a was produced without ligand (entry 20), thus demonstrating the essential role of both nickel catalyst and ligand in promotion of the reaction.
Scope of the Ni-catalyzed cross-coupling. To ascertain the substrate scope of this method, a wide range of alkylzinc reagents were examined (Table 2). Overall, good-to-high yields of gem-difluoropropargylated alkanes 3 were provided under optimized reaction conditions; even NiCl 2 was used as a catalyst, high product yields were still observed (3a-3c, 3f, 3i, 3n). The reaction exhibits high-functional group tolerance. Substrates 1, bearing important functional groups, such as ether, ester, indole, morpholine, ferrocenyl, and vinyl, all underwent the reactions smoothly (3d-3i, 3n-3q). Importantly, the successful formation of 3j-3l in good yields with intact bromide, chloride, and fluoride provides good synthetic handles for downstream transformations. Most remarkably, an alkylboronic acid ester was also compatible with the reaction conditions (3m), thus highlighting the high chemoselectivity of current nickel-catalyzed process. Moreover, thienyl-containing substrate furnished corresponding product 3r with high efficiency (90% yield). The reaction was readily scalable as demonstrated by the gram-scale synthesis of 3a (72% yield) Table 2 Ni-catalyzed cross-coupling of unactivated alkylzincs 1 with 2a a Bn, benzyl a Reaction conditions (unless otherwise specified): 1 (1.2 equiv in 2 mL DMA), 2a (0.6 mmol, 1.0 equiv), dioxane (4 mL) for 8 h b 5 mol% of NiCl 2 was used c Reaction run in 0.3 mmol scale and 1.5 equiv of 1o was used d 0.5 mol% of Ni(dppf)Cl 2 and 0.5 mol% of tpy were used even when 0.5 mol% of Ni(dppf)Cl 2 was used, thus demonstrating the reliability of current process. In addition to alkylzinc reagents, a variety of gem-difluoropropargylated bromides 2 were investigated (Table 3). Compounds 2 bearing silyl substituents, such as methyldiphenylsilyl and triethylsilyl groups, underwent the reaction smoothly (4a and 4b). Aryl substituted 2 was also applicable to the reaction without observation of allenic side products (4c and 4d). Notably, no defluorination side reaction occurred during the reaction process, in sharp contrast to previous reports, where defluorinations were the major pathway [34][35][36][37] . Remarkably, the alkyl substituted gemdifluoropropargylated bromides 2 were also competent coupling partners and provided corresponding products in good yields (4e-4j), in which the application of this method could efficiently lead to compound 4j, a key intermediate for the preparation of bioactive molecule used for pheromone study 38 . To demonstrate the generality of this approach further, some biologically active molecules containing 2 were also examined. The estrone-derived substrate underwent the reactions smoothly (4k and 4l), even 90% yield of 4k was obtained. The glucose-and prolinecontaining gem-difluoropropargyl bromides were also amenable to the reaction (4m and 4n). Because carbohydrates and amino acids play important roles in life science, it would provide good opportunity to discover some interesting biologically active molecules. Importantly, sulbactam, a drug for the treatment of bacterial infection, derived substrate furnished 4o efficiently, thus featuring the applicability of this protocol in drug discovery and development.
Transformations of compounds 3 and 4. The utility of this approach can also be highlighted by the transformations of resulting gem-difluoroalkylated alkanes. As shown in Fig. 1a, deprotection of 3a with TBAF, followed by [3 + 2] cyclization 39 afforded difluoroalkylated pyrrole 7 in high yield. Sonogashira reaction of 5 with heteroaryl iodides proceeded smoothly, offering a complementary method to the approach illustrated in Table 3. The CF 2 -containing alkane 10 can also be readily prepared through hydrogenation of carbon-carbon triple bond without defluorination, providing a facile route for site-selective introduction of a CF 2 group into an aliphatic chain (Fig. 1b). Notably, compounds 3 and 4 can serve as key intermediates for the synthesis of biologically active molecules. For example, compound 11, a probe for the study of hydrophobic interaction in pheromone reception 38 , was readily prepared by selective Table 3 Ni-catalyzed cross-coupling of 1 with 2 a TES triethylsilyl, TBS terbutyldimethylsilyl a Reaction conditions (unless otherwise specified): 1 (1.5 equiv in 1 mL DMA), 2 (0.3 mmol, 1.0 equiv), dioxane (2 mL) for 8 h reduction of compound 4f (Fig. 1c). In light of a series of pheromone analogs containing Z-difluoroalkylated alkenes 38,40 , this transformation is highly relevant to the synthesis of such a kind of structures. Furthermore, compound 12 derived from gem-difluoroalkylated alkane 3i can also functionalize as a key intermediate for the synthesis of analog of omega-3 fatty acids 13 for therapeutic use (Fig. 1d) 41    metabolites. Importantly, the copper-free click reaction 42 of terminal alkyne 12 with the dansyl fluorophore provided [3 + 2] cycloaddition products 15 with high efficiency (86%, as a mixture of regioisomers) in phosphate-buffered saline (PBS, pH 7.4) at 37 o C (Fig. 1d). Yet, no reaction occurred by using its nonfluorinated counterpart (for details, see Supplementary Methods). Thus, these results clearly demonstrate that the presence of CF 2 group is critical for the high reactivity of 12 in the copper-free click reaction. This reactivity is not only because the lowest unoccupied molecular orbital (LUMO) of the difluoroalkylated alkyne can be lowered by the CF 2 group 43 , but also because the stability of the transition state can be increased through hyperconjugative interactions between the carbon-carbon triple bond π system and vicinal σ* C-F acceptors 44 45,46 , was prepared efficiently through the current nickel-catalyzed process, thus demonstrating the utility of this approach further (Fig. 1e).

Discussion
To gain mechanistic insights into the current reaction, radicalinhibition experiment was conducted. We found that the reaction was almost terminated (giving 3i in a yield of 2%) by the electron transfer inhibitor (1,4-dinitrobenzene, 1.0 equiv) 47 that was added to the reaction mixture of 1i and 2a under standard reaction conditions (Fig. 2a). Furthermore, a ring-expanded product 20 (10% yield) was afforded when α-cyclopropylstyrene 19, which is known to participate in radical rearrangements 48 , was treated with 2a under standard reaction conditions (Fig. 2b). This result confirms the possibility of forming gem-difluoropropargyl radical intermediate during the nickel-catalyzed process. As a direct evidence, the electron paramagnetic resonance (EPR) study of reaction of 2a with spin-trapping agent phenyl tert-butyl nitrone 49 showed a spin adduct of the trapped gemdifluoropropargyl radical 21 (a N = 14.10 G, a H β = 2.01 G) in the EPR sextet spectrum (Fig. 2c and Supplementary Fig. 168), suggesting that the difluoroalkyl group is substituted at the β-carbon atom of nitroxide so that the a N and a H β values are decreased by the strong electron-withdrawing effect of CF 2 group through reducing the interaction of the electron spin with β-H atom as well as the electron pair on the nitrogen atom (usually, a N values are 14-16.5 G for the non-fluorinated dialkyl nitroxides) 49. Moreover, EPR spectrum also showed the EPR signal of spin adduct 21 at the beginning of the reaction ( Supplementary  Fig. 169). Thus, a combination of the above results demonstrates that a gem-difluoropropargyl radical generated via a nickelinitiated single-electron transfer pathway is involved in the reaction.
We also prepared Ni(I) complex [(tpy)Ni I -I] 50, 51 (A1) and Ni (II) complex [(tpy)Ni II Cl 2 ] (D1) 52 as the nickel species to add in the reaction of 1a with 2a, and found that both complexes catalyzed the reaction to afford compound 3a with high yields (Fig. 3a, b), thus demonstrating that the ligand dppf in the nickel catalyst Ni(dppf)Cl 2 just play a trivial role to promote the reaction. We propose that dppf in this precatalyst benefits the formation of an active nickel species that can facilitate the catalytic cycle. To further investigate the detailed mechanism, we prepared methyl nickel complexes [(tpy)Ni-Me] (B1) 53 and [I-Ni(tpy′)Me] (C1) 54 as the catalysts in the mechanistic studies. In C1 tpy′ is used instead of tpy because [I-Ni(tpy)Me] (C1′) is prone to decomposition to [(tpy)NiI] 51 , and because tpy′ as the ligand showed similar reactivity as tpy in the current cross-coupling (Table 1, entry 18). In the presence of B1 catalyst, the reaction could provide compound 3a in high yield (Fig. 3c), as well as a methylated product 22 in a yield in the range of 3-17%, increasing as raising amount of B1 from 5 to 20 mol% (Fig. 3c). On the contrary, the use of C1 (10 mol%) as a catalyst led to only 1% yield of compound 22, albeit formation of 3a in a 73% yield (Fig. 3d). These findings suggest that compared with complex C1, complex B1 is more favorable for the formation of 22. In addition, cross-coupling product 23 was not generated between alkylzinc 1a and C1, and even the stoichiometric reaction of C1 with 1a only afforded 23 in 5% yield (Fig. 3g). Both results indicate that a simple transmetalation of C1 by 1a is unlikely, ruling out a Ni(0/II) redox cycle. On the basis of the above analysis, we envisioned that a Ni(I/III) catalytic cycle is involved in the current reaction and alkyl nickel complex [(tpy)Ni-alkyl] (B) may be the key intermediate. To further support this possible pathway, a stoichiometric reaction of B1 with 2a (1.0 equiv) was conducted and provided 22 in 51% yield (Fig. 3e). However, only 18% yield of 22 was obtained when C1 was used (Fig. 3f). Thus, the reaction via a key Ni(I) intermediate [(tpy)Ni-alkyl] (B) generated by a transmetalation-first pathway between alkyzinc and nickel(I) complex [(tpy)Ni-X] (A) is reasonable.
On the basis of above results 55, 56 , a plausible reaction mechanism involving a Ni(I/III) catalytic cycle is proposed (Fig. 3h). The comproportionation of in situ-generated Ni(0) from the reaction of Ni(II) with alkylzinc reagent and the remaining Ni(II) species generates the Ni(I) complex [(tpy)Ni I -X] (A) 53 , which subsequently produces alkyl nickel complex [TpyNi-Alkyl] (B) through transmetalation with alkylzinc. Because such a terpyridine-based nickel(I) complex B is actually a nickel(II) complex [(tpy −1 )Ni II -alkyl] 53 , a one-electron, ligand-based redox pathway would be occurred in the reaction of B with gemdifluoropropargyl bromide 2 to generate C and difluoroalkyl radical 53 . Recombination of C with difluoroalkyl radical produces nickel(III) complex [Alkyl(Tpy)Ni III (R f )X] (E, R f = CF 2 CCR 2 ), which subsequently undergoes reductive elimination to deliver gem-difluoropropargylated alkane and regenerate A simultaneously.
In conclusion, we have demonstrated a Ni-catalyzed difluoroalkylation of unactivated aliphatic substrates. The reaction proceeds under mild reaction conditions and allows a variety of alkylzinc reagents and gem-difluoropropargyl bromides with high efficiency and excellent chemoselectivity. Importantly, most of the resulting gem-difluoropropargylated alkanes are unknown and can serve as versatile building blocks for the organic synthesis and medicinal chemistry. Preliminary mechanistic studies reveal that the transmetalation between Ni(I) and alkylzinc to generate alkyl nickel complex [alkylNi(tpy)] initiates the reaction, and a Ni (I/III) catalytic cycle is involved in the reaction. This mechanism is different from that of recently reported nickel-catalyzed Negishi arylation of proparylic bromides, which has been suggested through a bimetallic pathway 56 . Because of the synthetic utility of carbon-carbon triple bond and the unique properties of CF 2 , which can significantly improve the metabolic stability of biologically active molecules, this reaction will have applications in life and materials science as well as prompt the research area in fluoroalkylation of unactivated, aliphatic compounds from lowcost fluoroalkyl halides. Further studies to other derivative reactions are now in progress in our laboratory.

Methods
General procedure for nickel-catalyzed cross-coupling. To a 25 mL of Schlenk tube was added Ni(dppf)Cl 2 (2.5 mol%) and tpy (2.5 mol%). The tube was evacuated and backfilled with Ar for three times, then gem-difluoropropargyl bromide 2a (0.6 mmol, 1.0 equiv) and 1,4-dioxane (4 mL) were added. The resulting mixture was stirred at room temperature for 5 min, alkylzinc reagent 1 (0.72 mmol, 1.2 equiv in 2 mL of DMA) was then added dropwise over a period of 5 min. The tube was screw capped and put into a preheated oil bath (40 o C). After stirring for 8 h, the reaction mixture was cooled to room temperature and diluted with EtOAc (50 mL). The resulting mixture was filtered with a pad of cellite. The filtrate was washed with water (15 mL), the organic layer was dried over Na 2 SO 4 , filtered, and concentrated. The residue was purified with silica gel chromatography to give the pure product.
Data availability. The authors declare that all the data supporting the findings of this study are available within the paper and its supplementary information files.