Stereoselective cobalt-catalyzed halofluoroalkylation of alkynes

Stereoselective additions of functionalized reagents to unsaturated hydrocarbons are attractive due to the high atom economy, modularity and rapid generation of complexity. We report a stereoselective cobalt-catalyzed (E)-halofluoroalkylation of alkynes/alkenes that under mild conditions (2 mol% cat., 20 °C, acetone/water, 3 h). This reaction operates via a radical chain mechanism involving terminal halogen atom transfer which obviates the need for a stoichiometric sacrificial reductant.


Introduction
Fluorinated hydrocarbons constitute key structural motifs in many bioactive molecules, agrochemicals, and pharmaceuticals due to their high metabolic stability, lipophilicity, and bioavailability compared with the parent compounds. 1 Fluoroalkylation methods of easily accessible precursors have therefore attracted great interest in the past years. 2,3 While many protocols are substitution processes that require highly pre-functionalized starting materials and produce unwanted by-products, direct additions to unsaturated hydrocarbons exhibit higher modularity and atom-efficiency and provide ample opportunities of regioand stereocontrol. The addition of halo-uoroalkanes to alkynes is an especially attractive tool due to the easy availability of the reagents and the great synthetic versatility of the resultant halo-uoroalkenes. Many methods operate via an atom transfer radical addition (ATRA) mechanism in the presence of radical initiators (e.g. BEt 3 , AIBN, Na 2 S 2 O 3 or light) 4 that showed narrow substrate scope and poor selectivity. Mechanistically closely related transition metal-mediated halo-uoroalkylations have been recently reported, but with a narrow focus on iodouoroalkylations and/or moderate stereo-control (Scheme 1). Hu et al. devised an iron-catalyzed addition of peruoroalkyl iodide to alkynes with moderate to good E/Z-selectivities in the presence of Cs 2 CO 3 . The radical reaction with alkyl-substituted alkynes required long reaction times at 60 C and could not convert peruoroalkyl bromides. 5 Besset et al. postulated a different mechanism for the copper-mediated synthesis of diuoromethyl alkenes from BrCF 2 CO 2 Et and alkynes. However, signicantly lower stereoselectivities were obtained and stoichiometric amounts of copper salt were employed. 6 Very recently, Wang and co-workers reported a copper-catalyzed decarboxylative ATRA reaction between ICF 2 CO 2 Et and substituted propiolic acids. 7 Despite the developments of novel iron-and coppercatalyzed procedures, the reactions generally utilize expensive uoroalkyl iodides as starting materials, high catalyst loadings, long reaction times, and high reaction temperatures. An efficient and robust yet highly stereoselective method that operates at mild conditions and low catalyst loadings and that is applicable to various uoroalkyl halides would constitute Scheme 1 Metal-mediated halo-fluoroalkylations of alkynes. a signicant advancement of the current technology and have considerable use in the synthesis of densely functionalized uorinated building blocks. To the best of our knowledge, there are no literature reports of ATRA reaction between alkyl halides and alkynes with low-valent cobalt catalysts. Here, we report the cobalt-catalyzed halouoro-alkylation of alkynes which enables the highly regioselective and stereoselective synthesis of a diverse set of halouoroalkenes under unprecedentedly mild conditions (Scheme 1).
The utility of alkyl halides in cross-couplings and reductive additions has recently been greatly enhanced by the development of low-valent iron group metal catalysts (Fe, Co, Ni) 8-10 that engage in facile alkyl-X activation. The high propensity of late 3d transition metals to undergo single-electron transfer (SET) processes oen results in the intermediacy of carbon-centered radical species. 8g,h,9h,10f The reductive formation of alkyl radicals from alkyl-X electrophiles is thermodynamically favoured when the formal electron septet-carbon is stabilized by heteroatoms, conjugation, hyperconjugation, or inductive effects. Efficient cobalt catalysts have been reported for several reductive coupling reactions between alkyl halides and Michael-type acceptors as well as for Heck-type reactions between alkyl halides and alkenes. 11 These processes mostly follow the same mechanistic scenarios involving (i) initial formation of a lowvalent Co(I) species from a Co(II,III) precursor in the presence of a reductant; (ii) reductive cleavage of the alkyl halide to give an alkyl radical Rc and a Co(II) complex (by SET activation or homolysis of R-Co(III)), (iii) addition of Rc to the olen and formation of an organocobalt species that is subject to disproportionation (Heck-type reaction) or hydrolysis (reductive coupling) to release the Co(III) complex, (iv) regeneration of the Co(I) catalyst with a stoichiometric reductant (Scheme 2, le). We surmised that an ATRA reaction between alkyl halides and alkynes could follow a similar mechanism using low-valent Co catalysts but would require only catalytic amounts of a reductant (Scheme 2, right). Cobalt-catalyzed Heck-type and reductive coupling reactions were realized in the presence of stoichiometric reductants such as Zn, Mn, and Grignard reagents 11 which effect the Co(III) / Co(I) reduction. Cobalt-catalyzed ATRA reactions of alkynes with alkyl halides have not been reported.
We then examined the cobalt-catalyzed halouoroalkylation with different uoroalkyl halides (Scheme 5). Iododiuoroacetate ICF 2 CO 2 Et, peruoroalkyl iodides such as C 4 F 9 I, C 6 F 13 I and C 8 H 17 I, and the peruoroalkyl bromide C 8 F 17 Br were competent electrophiles which afforded the desired adducts in good to excellent yields. The uoroalkyl bromides gave generally better E/ Z selectivities than the iodides. While this trend is in full agreement with the literature, it can now be harnessed at much milder conditions (room temp., 2 mol% catalyst, 3 h). BrCF 2 PO(OEt) 2 , CF 3 I, and CF 2 Br 2 afforded slightly lower yields; the reaction with CF 3 I exhibited low stereocontrol. Reactions of alkyl-substituted alkynes with uoroalkyl iodides gave good yields and moderate E/Z selectivities (3ac-3af). The reaction conditions were also applied to reactions of (cyclo)alkenes with halouoroacetates (3ag-3ak). A method extension to reactions of simple bromoacetates with alkenes gave the desired adducts 3al-3an.

Mechanistic studies
Further attention was devoted to the study of the reaction mechanism. In addition to the initial optimization reactions (Table 1), key mechanistic experiments were conducted. The model reaction between 1a and 2a was completely inhibited in the presence of 2,2,6,6-tetramethyl-1-piperidinyloxy (TEMPO). The TEMPO-CF 2 CO 2 Et adduct was observed by mass spectrometry (Scheme 7, eqn (1)). The TEMPO-CF 2 CO 2 Et adduct was not detected when treating 2a with equimolar Zn which supports the notion that the SET reduction of the alkyl halide is induced by the cobalt catalyst (Scheme 7, eqn (2)). Upon employment of cyclo-propylacetylene (4), the vinylcyclopropane product was formed in 11% yield while ring-opening to the 7bromohepta-3,4-dienoate (56%) was the major pathway. This is in full agreement with the intermediacy of an internal vinyl radical formed by radical addition of EtO 2 CCF 2 c to the alkyne (Scheme 7, eqn (3)). Identical rates and yields were observed in reactions where 1a and 2a were successively added to the catalyst solution. The reverse order of addition (2a, then 1a) gave an identical result. Importantly, no product was formed when prior to the addition of 1a and 2athe catalyst suspension (CoBr 2 , dppbz, Zn) was ltered (to remove residual zinc) or when the supernatant solution was decanted into a new reaction vessel (eqn (4)). These experiments suggest that the initially formed Co(I) species alone cannot catalyze the reaction but requires the presence of zinc, at least for the rst turnover of the catalytic mechanism. Zn is employed only in catalytic amounts (5 mol%, i.e. 2.5 equiv. per Co)! The addition of sodium iodide and sodium bromide, respectively, shed light on the nature of the operating halogen transfer. 1a and 2a reacted with added NaI Scheme 3 Cobalt-catalyzed bromo-carboxydifluoromethylation of alkynes. Standard conditions: 1 (0.3 mmol), 2a (0.45 mmol), CoBr 2 (2 mol%), dppbz (2 mol%) and Zn (5 mol%), 0.6 mL acetone/H 2 O, 20 C, 3 h under N 2 . Isolated yields are given; E/Z ratios in parentheses (by 19 F NMR). [a] 20 mol% Zn. [b] 40 mol% Zn, 8 h.

Scheme 4 Reactions with alkyl-substituted terminal alkynes.
(1.5 equiv.) to give the iodo adduct 3u as major product (3a : 3u ¼ 1 : 20, Scheme 7, eqn (6)). With NaBr added, the reaction between 1a and 2b gave 3a and 3u in a 1 : 6 ratio (Scheme 7, eqn (7)). Control experiments documented that no EtO 2 CCF 2 I was converted into EtO 2 CCF 2 Br using NaBr as additive; only minimal amounts of EtO 2 CCF 2 Br (<2%) were converted to EtO 2 CCF 2 I with NaI as additive under the same conditions. A similar outcome was observed when the standard reaction was performed with 50 mol% CoBr 2 /dppbz (3a : 3u ¼ 1 : 7, Scheme 7, eqn (8)). These experiments document that the halogen atom X in the product does not originate from the electrophilic R F X via a direct radical chain transfer but is transferred from the cobalt catalyst. This is a ne but important distinction from previously reported ATRA reactions that all involved halogen transfer from R F X to the vinyl radical. This has great implications for catalyst design and reaction development as the thermodynamics and kinetics of the halogen atom transfer step are no longer depending on the nature of the employed substrates but can be nely tuned through the stereoelectronic properties of the catalyst. We further believe that halogen atom transfer to a vinyl radical intermediate (rather than a vinyl cation) is operative: (i) the addition of water (as a nucleophile) resulted in no product bearing oxo functions; (ii) the presence of methyl acrylate as a radical acceptor led to the formation of the heptene-1,7-dioate via radical insertion of the acrylate (eqn (9)). A cationic intermediate would not add to this Michael acceptor. Catalyst formation and substrate additions were monitored by 31 P NMR and 1 H NMR spectroscopy (Fig. 1). The reduction of the (NMR silent) CoBr 2 /dppbz mixture with Zn resulted in a Co(I) species with a 31 P resonance at 75.2 ppm. The 1 H NMR spectrum of this low-spin Co(I) complex gave signals 7-8 ppm.
No changes were observed in 31 P and 1 H NMR spectra when phenyl acetylene (1a) was added to Co(I) which suggests the absence of signicant alkyne-catalyst coordination. On the other hand, complete disappearance of the 31 P (75.2 ppm) and 1 H (7-8 ppm) signals was observed upon addition of EtO 2 -CCF 2 Br (2a). This is a direct consequence of the reductive activation of 2a which leads to a paramagnetic Co(II,III) species and the carbon-centered radical. 12 These results are consistent with the UV-vis spectra (Fig. 2). Reduction of Co(II) with Zn (and removal of residual Zn) resulted in an intense absorption of the Co(I) complex at 428 nm (green curve). Addition of 1a to this solution gave no change of the absorption in this region (blue curve), whereas the addition of 2a to Co(I) led to immediate colour change and the appearance of two weak bands at 412 and 451 nm (yellow curve).
The standard reaction between 1a and 2a went to completion within 90 min (with 2 mol% catalyst) and 12 min (4 mol% catalyst), respectively. Analysis of the initial rates (0.5-8 min, 1-4 mol% catalyst) displayed a near-2 nd order behavior of the catalyst concentration. We postulate the following reaction mechanism (Scheme 8). Complexation of dppbz with CoBr 2 leads to the formation of [Co II (dppbz) 2 Br] + as observed by the so and inert mass spectrometric technique for sensitive organometallics LIFDI-MS (liquid injection eld desorption ionization mass spectrometry). Reduction of the Co(II) complex with equimolar Zn generates the catalytically active [Co I (dppbz) 2 2 Br] 12h-j which can undergo another reductive activation of R F X to give [R F Co III (dppbz) 2 Br] + . The catalytic amounts of Zn present in the reaction dictate that another mechanism operates from the 2 nd turnover on, most likely an ATRA reaction involving halogen atom transfer from the cobalt complex [R F Co III (dppbz) 2 Br] + . Accordingly, the addition of R F c to the alkyne results in the formation of a vinyl radical intermediate which undergoes rapid halogen atom abstraction from [Co III (dppbz) 2 Br] + to form the catalytically active Co(I) complex and R F c. 15 The high E-selectivity of the radical addition is  a direct consequence of the steric hindrance by the R F group in the vinyl radical. 16 The higher E/Z stereoselectivity of the bromoalkylation over the iodoalkylation reactions can be explained by the shorter Co-Br bond (vs. Co-I) in the key catalytic Co(III) species which effects an enhanced facial differentiation of the vinyl radical. The facile operation of this halogen atom transfer step with the intermediate vinyl radical is the key to the realization of an overall process that is catalytic in both metals, Co and Zn.

Conclusions
We have developed a convenient cobalt-catalyzed halouoroalkylation that exhibits wide substrate scope including terminal and internal alkynes, alkenes, and various uoroalkyl and alkyl bromides and iodides. The protocol enables the highly regio-selective and stereoselective synthesis of densely functionalized halogenated (E)-alkenes under very mild reaction conditions (2 mol% catalyst, 5 mol% Zn, acetone/water, 20 C, 3 h). Contrary to literature reports, mechanistic studies documented for the rst time that the halogen atom transfer is a cobalt-mediated process. The R F Co III X complex is the key catalytic intermediate which generates the free R F c radical and mediates the halogen atom transfer to the terminal vinyl radical. This mechanistic deviation from substrate control to catalyst control may provide the basis for the development of related halogen atom transfer reactions through catalyst design. Further, this ATRA reaction operates without a stoichiometric reductant for the regeneration of the low-valent Co(I) catalyst.
The high functional group tolerance and mild reaction conditions make this protocol highly attractive in the context of complex molecule synthesis with potential utility for medicinal chemistry endeavours.

Conflicts of interest
There are no conicts to declare.