We initiated our studies by optimizing reaction conditions for the envisioned cobalt-catalyzed regioselective three-component coupling of alkenylarene (2a) with bromodifluoroacetate (1a) and phenylzinc pivalate (3a, PhZnOPiv). A cascade cross-coupling reaction was observed in the presence of 10.0 mol % CoBr2 under ligand-free conditions, thus affording the desired aryl-difluoroalkylated product 4 in 83% yield with high regioselectivity (Table 1, entry 1). Among a number of representative chelating ligands, bipyridines have given negative effects, and only trace amount of product was observed (entries 2‒3), tridentate 2,6-bis(N-pyrazolyl)pyridine, 1,10-phenanthrolines, diimine, TMEDA, ME4DACH, as well as dppbz ligands gave poor to high yields, whereas the neocuproine (L5) afforded 4 in 92% (entries 4‒10). Further testing reactions with different solvents verified the crucial importance of MeCN as the reaction medium (entry 11; See SI). Switching from CoBr2 to other representative cobalt salts, such as CoCl2, CoCl2(PPh3)2, led to significant reduced yields (entry 12‒13). In sharp contrast, replacement of CoBr2 by using NiBr2, FeCl2, CrCl2, or CuBr failed to furnish the desired product 4 (entry 14).
Lei[76‒78] demonstrated first that arylzinc reagents prepared by different methods possess very different kinetics in palladium- and nickel-catalyzed oxidative couplings, and further X-ray absorption spectroscopy studies show that changing the halide anion from Cl to Br or I will result in an increase of the Zn‒C bond distance and thereby improve the transmetallation rate.[79] In order to preliminarily reveal the different kinetics between this solid zinc reagent and conventional zinc reagents, a series of control experiments with six different phenylzinc reagents, which prepared by transmetallation of the corresponding phenylmagnesium halides and zinc halides,[80] were also performed under the ligand-free cobalt catalysis (Scheme 2). Interestingly, all of these reactions were almost finished within remarkably short reaction times of only 15 min. It is worth noting that significantly reduced conversions of 4 were observed when using PhZnX (X = Cl, Br or I), Ph2Zn‧2MgCl2 or Ph2Zn‧2Mg(OPiv)Cl instead of PhZnOPiv. Moreover, the results of comparison experiments between Ph2Zn‧2Mg(OPiv)Cl and Ph2Zn‧2MgCl2 show the superiority of the former as well. Hence, these observations highlighted that the presence of M(OPiv)2 (M = Mg or Zn) has made these new organozinc pivalates stand out amongst salt-supported organometallics, thus displaying the distinct advantage of reacting well in our regioselective cobalt-catalyzed difluoroalkylarylation of olefins.
Subsequently, the versatility of this optimized cobalt(II) catalyst was examined in a range of difluoroalkylarylation reactions with various polyfunctionalized arylzinc pivalates 3 (Scheme 3). All arylzinc pivalates were prepared from the corresponding aryl halides by Mg insertion in the presence of LiCl.[81] Although the neocuproine (L5) gave the optimal results in the model reaction, in our efforts to extend the substrate scope of this domino reaction, ligand-free CoBr2 proved to be superior (see the results of products 7, 9, 11). A variety of para- and/or metal- substituted arylzinc pivalates were identified as viable nucleophiles for difluoroalkylarylation with bromodifluoroacetate (1a) and 4-methoxystyrene (2a) to afford the desired products 4‒16 in moderate yields. More sterically hindered 4-chloro-2-methylphenylzinc pivalate was successfully employed, leading to the desired difluoroalkylarylated product 17 in 62% yield. Notably, ferrocenylzinc pivalate, as well as 3-thienylzinc pivalate also smoothly underwent the cobalt-catalyzed cascade cross-coupling, albeit yielding the products 18‒19 in relatively lower yields.
Thereafter, we have explored the substrate scope of the difluoroalkylarylation reaction with a wide range of vinylarenes and bromodifluoroacetate/amides (Scheme 4). Remarkably, alkenylarenes bearing various valuable electrophilic functional groups, such as ether (22), fluoro (23), chloro (24), bromo (25, 35, 37), trifluoromethyl (26), methyloxy (27‒28, 31‒32, 38‒39), cyano (30), acetate (33), esters (36), isobutyl (34) substituents, as well as vinylnaphthalene (29) and unsubstituted styrenes (20‒21), were well tolerated under the reaction conditions and converted to the corresponding difluorinated 1,1-diarylalkanes in moderate to excellent yields (40‒98%), as were also observed when using different bromodifluoroacetamides as the fluorinating reagents. Also, internal alkene with (E)-β-methylstyrene was examined under our cobalt catalysis, but only trace amount of desired product was detected by GC-analysis (See SI). In sharp contrast, coupling of arylzinc pivalate, bromodifluoroacetate with indene gave the desired difluoroalkylarylated product 40 in 55% yield, with high diastereoselctivity (dr > 20:1).
In addition, we further investigated the cross-coupling of various fluoroalkyl bromides with olefins and arylzinc pivalates (Scheme 5). Firstly, in contrast to bromodifluoroacetate, the bromomonofluoroacetate only gave 34% yield under the standard reaction conditions, and with a poor diastereoselectivity (dr = 1:1) (Scheme 5a). We were also pleased to find that bromodifluoromethylphosphonate smoothly underwent the envisioned cobalt-catalyzed difluoromethylarylation to afford the desired 1,1-diarylalkylphosphonates 42‒45 in 51‒97% yields, and compound 45 was obtained with high diastereoselectivity (dr > 20:1). Besides, the unactivated alkene furnished the desired difluoromethylarylated phosphonate 46 as well, albeit in a modest yield (Scheme 5b). Additionally, using as substrate of α-bromodifluoromethyl substituted benzoxazole proved to be viable with versatile cobalt catalyst and, thereby, provided 47‒48 as the products in 51‒55% yields (Scheme 5c). Remarkably, this cobalt-catalyzed regioselective difluoroalkylarylation reaction was further extended to the decorated difluoroalkyl bromides (Scheme 5d). Functional groups, such as arylsulfonate, ester, were well tolerated under the standard reaction conditions, thus delivering the desired products 49‒52 in good yields and with high diastereoselctivity of 51 (dr > 20:1).
Transformations of unactivated alkenes are acknowledged widely as a challenge in transition metal-catalyzed difunctionalization of alkenes.[49‒56, 69‒71] The reaction conditions previously optimized for the alkenylarenes led to an unsatisfactorily low yield of 53, because significant amounts of a Heck-type coupling product were formed as well. However, we were delighted to found that the transformation of the unactivated alkene difluoroalkylarylation process was significantly improved when using dppbz (L9, 11 mol %) as the ligand, leading to 53 in 54% yield. A number of unactivated alkenes were readily converted into the desired difluoroalkylarylated products 54‒59 in moderate yields.[82] Moreover, various synthetically valuable functional groups, including chloro, ether, and ester remained intact by the cobalt catalyst (Scheme 6a). Beyond that, the possibility of cobalt-catalyzed difluoroalkylation to form an allyl radical, which subsequently underwent 1,3-H-shift and Csp3‒Csp2 cross-couplings with arylzinc pivalates was also investigated (Scheme 6b). Indeed, difluoroalkyl bromide 1k and a quite range of functionalized (hetero)aryl-zinc reagents were realized 1,4-difunctionaliztion of 1,3-dienes with good regioselectivity and diastereoselectivity, thus furnishing 60‒70 in 43‒98% yields, albeit products 69 and 70 were obtained with 1:1 E:Z selectivity and 4:1 regioselectivity, respectively. To our delight, 1,3-octadiene was proven to be suitable substrate as well, giving the product 71 with high diastereoselectivity.
To further illustrate the potential applications of this cobalt-catalyzed regioselective difluoroalkylarylation in late-stage functionalizations of pharmaceutically active molecules, alkenylarenes derivatized from (pre-)drug molecules, such as febuxostat, canagliflozin, as well as indomethacin, were well difluoroalkylarylated with arylzinc pivalates and α-bromodifluorocarbonyl compounds or bromodifluoromethylphosphonate, leading to the corresponding products 72‒77 in 30‒96% yields. These results show the potential utility of this protocol for the discovery of novel bioactive drugs. Importantly, citronellol derivative was readily incorporated into the product 78 with remarkably high regioselectivity and chemoselectivity. Moreover, an unactivated alkene bearing a 4-hydroxycoumarin proved to be viable substrate as well, albeit delivering the phosphonate 79 in a rather modest yield. Finally, we showed that isopropenylzinc pivalate is well suited for the cobalt-catalyzed difluoroalkylalkenylation, although the reaction proceeded with lower yield (Scheme 7).
Intrigued by the high regioselectivity and efficacy of our cobalt-catalyzed difluoroalkylarylation, a series of intermolecular competition experiments were performed (Scheme 8). A competition experiment between bromodifluoroacetate (1a) and 2-bromo-2-methylpropanoate showed that BrCF2CO2Et reacted much faster than these α-bromocarbonyl compounds. These findings can be rationalized in terms of a prioritized direct halogen atom abstraction from difluoroalkyl bromides via single electron transfer from a cobalt catalyst (Scheme 8a).[83] Intermolecular competition experiments with different alkenylarenes, and arylzinc pivalates revealed electron-rich styrenes and electron-deficient arylzinc pivalates to be slightly reactive substrates (Scheme 8b and 8c). These results suggested that vinylarenes and arylzinc reagents might not be involved in the rate-determine step.[56]
Beyond that, radical-clock experiment with substrate 83 bearing a radical clock cyclopropane moiety, the ring-opened difluoroalkylarylated product 84 was generated in 11% yield. Similarly, both three- and two-component coupling products were observed when using N,N-diallyl-2-bromo-2,2-difluoroacetamide (85) as a radical probe under the standard reaction conditions, the cyclized products 86 (dr = 2:1) and 87 were generated in 17% and 34% yields, respectively. Moreover, a difluoroalkylated benzylic radical homocoupling dimer 88 was detected by GC as well. With these findings, we propose this cobalt-catalyzed difluoroalkylarylation involves a single-electron-transfer (SET) process (Scheme 9a).
According to the earlier mechanistic studies for cobalt-catalyzed cross-coupling reactions with using organomagnesium reagents, an in situ low-valent Co(0) was proposed as the catalytically active species.[52, 72‒73, 84‒85] On the other hand, a mechanism involving Co(I)/(III) couple was also proposed for many cobalt-catalyzed cross-couplings.[37‒39, 83] Therefore, we performed experiments of CoBr2 (1.0 equiv) with excess of ArZnOPiv under typical reaction conditions for 30 min. These reactions furnished the corresponding homo-products of 89a and 89b in near 0.5 equiv ratio to that of CoBr2, respectively. These findings support the formation of a Co(I)-species based on the stoichiometry shown in scheme 9b. In this context, the well-defined Co(I)-complex, such as CoCl(PPh3)3 was proved to be active for the desired difluoroalkylarylated process, yielding product 4 in 66%, while Co2(CO)8 gave a poor yield (Scheme 9c). Further experiments to examine the catalytic activity of the in situ generated low-valent cobalt(I) species were performed. A mixture of vinylarene 2a (0.25 mmol) and CoBr2 (0.025 mmol) was treated with 2.0 equiv of 3,4-(methylenedioxy)phenylzinc pivalate (0.05 mmol) at 23 °C for 30 min to generate the proposed Co(I)-species, followed by addition of bromodifluoroacetate 1a (0.3 mmol) and another 0.5 mmol of phenylzinc pivalate. The difluoroalkylarylated product 4 was isolated in 57% yield as the sole product, while the product 11 was obtained in 79% yield when exchanging the order of the two arylzinc reagents (Scheme 9d). These findings are consistent with the in situ generated low-valent cobalt(I)-species might be the active catalyst for the current three-component cross-coupling reaction. A series of EPR spin-trapping experiments show the existence of C-centered radicals trapped by DMPO (g=2.0066, AN = 13.9 G, AH = 19.3 G), which was considered to be •CF2R.[86] These results strongly supported the single electron transfer progress for the activation of BrCF2R was only promoted by the in situ formed Co(I)-species (Scheme 9e).
Based on the above experimental findings, along with previous mechanistic insights,[37‒39, 74, 83] a mechanism for this regioselective cobalt-catalyzed difluoroalkylarylation of alkenes has been proposed as shown in Scheme 10. The reduction of the precatalyst CoBr2 with arylzinc pivalates forms the catalytically active Co(I)-species (A), which reduces difluoroalkyl bromides (1) by SET and generates difluoroalkyl radical B, then followed by a facile radical addition of B into olefins (2) to afford a secondary alkyl radical species, along with subsequent rapid trapping with LnCo(II)XBr (X = Br) into intermediate C, which undergoes transmetalation with ArZnOPiv (3) to lead to the organocobalt(III) species D. Subsequent reductive elimination finally delivers the difluoroalkylarylated product and regenerate the active cobalt(I)-catalyst (path a). In addition, another possible pathway is that transmetallation of arylzinc pivalates could also occurred after the initial reduction step, thus in situ forming the LnCo(I)X (X = Ar) species as the catalyst to promote the SET process. Radical addition and reductive elimination give rise to the desired products and regenerate the active Co(I)-species (path b).
We were also pleased to find that this cobalt-catalyzed difluoroalkylarylation can be easily scaled up to gram level. Under the optimized reaction conditions, the difluoroalkylarylated product 90 was afforded with high efficacy (65% yield, Scheme 10a). Finally, we further demonstrated the synthetic potential of this cobalt-catalyzed difluoroalkyarylation strategy through the late-stage modification of the obtained difluoroalkylarylated products. For example, the resulting N-morpholino amide 90 can be readily converted into various ketones by treating with Grignard reagents, thus furnishing the products 92a‒b in moderate yields. Moreover, the reduction of the ester group of substrate 4 by using NaBH4 provides the corresponding alcohol 93, which readily undergoes various derivatization (Scheme 11b).