Cobalt-catalyzed branched selective hydroallylation of terminal alkynes

Here, we reported a cobalt-hydride-catalyzed Markovnikov-type hydroallylation of terminal alkynes with allylic electrophile to access valuable and branched skipped dienes (1,4-dienes) with good regioselectivity. This operationally simple protocol exhibits excellent functional group tolerance and exceptional substrate scope. The reactions could be carried out in gram-scale with TON (turn over number) up to 1160, and the products could be easily derivatized. The preliminary mechanism of electrophilic allylation of α-selective cobalt alkenyl intermediate was proposed based on deuterium labeling experiment and kinetic studies.

Here, we reported a cobalt-hydride-catalyzed Markovnikov-type hydroallylation of terminal alkynes with allylic electrophile to access valuable and branched skipped dienes (1,4-dienes) with good regioselectivity. This operationally simple protocol exhibits excellent functional group tolerance and exceptional substrate scope. The reactions could be carried out in gram-scale with TON (turn over number) up to 1160, and the products could be easily derivatized. The preliminary mechanism of electrophilic allylation of α-selective cobalt alkenyl intermediate was proposed based on deuterium labeling experiment and kinetic studies.
In this work, we report cobalt-hydride-catalyzed branched selective hydroallylation of terminal alkynes with allylic bromides as electrophiles to access branched terminal skipped dienes with good regioselectivity and excellent functional group tolerance (Fig. 1d).

Reaction optimization
We performed the study by using 4-ethynylanisole 1a as a model substrate, 2-methylallylbromide 2a as an allylic electrophile, and PMHS as a hydride source ( Table 1). The Co(OAc) 2 was used as a catalyst, N-(2-(4,5-dihydrooxazol-2-yl)phenyl)quinoline-2-carboxamide (L1) and LiOtBu were used as the ligand and base, respectively. The reaction was performed in a solution of tetrahydrofuran (THF) at 50°C for 24 h to afford electrophilic hydroallylation product in 42% yield with 83/17 b/l (entry 1). A significant increase in the regioselectivity was observed when a larger gem-dimethyl group was used on the oxazoline moiety, which gave rise to skipped diene in 70% yield with >95/5 rr (ratio of regioselectivity; entries 1-3). However, when the size of the substituent was further increased, the regioselectivity slightly decreased (entries 4-6) which implied that the steric hindrance of the substituents on the oxazoline moiety might affect the yield and regioselectivity. Changing the steric effect on pyridine moiety, the selectivity of the reaction decreased slightly (entries 7-8). Using various hydrosilane, such as PhSiH 3 , (EtO) 3 SiH, and Ph 2 MeSiH led to poor yield and selectivity (entries 9-11). Using Ni(OAc) 2 instead of Co(OAc) 2 , a poor yield of hydroallylation was observed (entry 12). Additionally, Cu(OAc) 2 could not promote this transformation. However, the Sonogashira coupling reaction of allyl bromides with terminal alkynes could be promoted under mild conditions (entry 13) (We should thank one of the reviewers for the suggestion of using nickel or copper catalyst to performing the control experiments.). Using CoBr 2 instead of Co(OAc) 2 , this transformation could process smoothly (entry 14). The model reaction could be completed in 20 min (entry 15). The (L3-H)•CoBr complex reported in our previous studies 46 could also be used as an efficient catalyst (entry 16). The standard conditions were identified as 1.0 mmol of terminal alkyne, 0.50 mmol of allylic electrophile, 5 mol% of (L3-H) •CoBr, 0.75 mmol of LiOtBu, and 0.75 mmol of PMHS in a solution of THF (1 mL) at 50°C.

Substrate scope
Compared to other olefins, 2,4-disubstituted skipped dienes are difficult to obtain through classic Wittig reaction, due to the ketoenol tautomerism of 1,3-diketone. Thus, with the optimized conditions in hand, we mainly examined the substrate scope of 2,4disubstituted skipped dienes (Fig. 2). However, due to the presence of Lewis acid metal catalysts and corresponding bases, the inevitable side reaction of terminal alkynes self-polymerization and dehalogenation of allylic bromides would reduce the yield of allylation products. Respectively, 3b and 3c could be obtained on gram scale. The volatile dienes (3d, 3af) could also be obtained via distillation on gram-scale. Various allylic bromide (2d-2i) could be tolerated in this transformation to deliver skipped diene with moderate yield and excellent regioselectivity. Z−1,2-disubstituted bromide (2j) was also investigated. Interestingly, S N 2′-type skipped alkene (3j) was obtained as a major product in this transformation. The significant regioselectivitity differences between 3 f and 3j might owe to the different E/Z stereoconfigurations and steric hindrance of allyl bromides between 2 f and 2j. Besides allylic bromides, different allylic electrophilic reagents such as allyl iodide and allyl phosphate, could also be tolerated in this transformation with slightly decreased yield and regioselectivity (3b, 3d). Benzyl electrophiles, which exhibit similar properties with allylic reagent, could also be transformed smoothly on gram-scale with excellent regioselectivity 47,48 . The 1,4bis(bromomethyl)benzene (2 l) could also react with terminal alkynes smoothly to obtained the 1,1-disubstituted alkenes with excellent regioselectivity. For broader synthetic interests, a variety of functional groups on phenyl rings were investigated. Methyl, trifluoromethyl, fluorine, bromine, chlorine, protected alcohol, and ester could be well tolerated to afford the skipped dienes (3m-3v) in moderate yields with good to excellent regioselectivity (55-76% yield, 92/8 ->95/5 b/l). The alkynes containing heterocycles, such as pyridine 1w and thiophene 1x, could also be tolerated to deliver branched terminal skipped dienes in 48-50% yield. Conjugated enyne (1 y), silyl alkyne (1z), and cyclopropyl acetylene (1aa) could undergo this hydroallylation reaction smoothly. Simple terminal alkynes (1ab-1ae) were also amenable to this transformation to deliver the corresponding product in 55-67% yields with 91/9 to >95/5 rr. Additionally, terminal alkynes contained in bioactive molecules were investigated. Naproxen, menthol, and geraniol derivative (1ag-1ai) could be employed to deliver corresponding products in 45-69% yield.

Mechanistic studies
To elucidate the C(sp 2 )-C(sp 3 ) bond forming process in this transformation, control experiments were conducted. 3(E)-deuterated allylic bromide 10 was prepared to distinguish substitution at the 1-and 3-positions of the electrophile. This reaction was performed to give the  deuterated skipped dienes 11 in 42% yield with 0.17 D at terminal carbon and 0.69 D at sp 3 carbon (Fig. 4a). This result indicated that substitution occurred through the S N 2′-like process accompanied with the partial S N 2 process via attack of the postulated cobalt species at the 1,3position of the allylic bromides. To elucidate the hydrometallation process, the hydroallylation of deuterium labeling phenylacetylene 12 was performed in 30 min to give the deuterated skipped alkene in 59% NMR yield with 0.40 D at C(sp 2 ) 1(E)-position and 0.40 D at C(sp 2 ) 1(Z)-position (Fig. 4b). Due to the presence of strong bases, deuterium atoms might loss during this process. This result combined with our previous mechanistic studies indicated that the E/Z ratio of deuterated product might owe to the relatively fast Crabtree-Ojima-type isomerization 42,46 . The reaction could undergo smoothly in the presence of 1,1-diphenylethylene or butylated hydroxytoluene (BHT), which might rule out the radical reaction pathway (Fig. 4c). The hydroallylation product (3 g, h, j, k) showed that the configuration of    the allylic bromides would affect the S N 2 and S N 2′ selectivity of this transformation. Additionally, the π-allyl pathway could not be exclusively ruled out. Quantitative kinetic studies were also performed to determine the roles of alkyne, allylic bromide, hydrosilane, and (L3-H)•CoBr complex. Kinetic studies on alkyne showed that with a zero-order rate dependence on alkyne, however, as the concentrations of alkyne increased, the initial rates (k in ) of the reaction decreased (Fig. 5a). This result demonstrated excessive alkyne might be a ligand to coordinate with cobalt catalyst leading to the reduction of the rate of reaction. Measurements of the initial rates (k in ) of the reaction with different concentrations of allylic bromide and (L3-H)•CoBr complex showed a corresponding rise in the rates of the reactions. Plots of k in versus the concentrations of allylic bromide and (L3-H)•CoBr complex (Fig. 5b, c) gave two linear curves (slope = 1.19 × 10 −4 Ms −1 ; 6.79 × 10 −3 Ms −1 ), which suggested a first-order rate dependence on allylic bromide and (L3-H) •CoBr complex. Similar kinetic studies on PMHS showed no change in k in within a certain concentrations range (Fig. 5d), indicating a zeroorder rate dependence on hydrosilane. These quantitative kinetic studies suggests that the nucleophilic substitution of cobalt(II) alkenyl intermediate with allylic bromide could be the turnover-limiting step.
Based on the experimental studies and previously reported literatures 35,42,46,[57][58][59] , a possible mechanism is shown in Fig. 6. The cobalt hydride species C was obtained from the reaction of active intermediate (L3-H)•CoBr with LiOtBu and hydrosilane. The alkyne coordination with species C followed by the insertion of terminal alkyne into the cobalt hydride bond delivering majorly α-selective cobalt-alkenyl intermediate E. The quick isomerization balance between E and cobalt carbene zwitterion F led to the E/Z ratio variation based on the deuterium labeling experiment. The following S N 2′ and S N 2-like process of E with allylic electrophile generates species to deliver the corresponding skipped dienes. The hydrosilane and alkoxide might be likely responsible for the observed regioselectivity increase during the catalysis process.

Discussion
In summary, we reported an efficient cobalt-hydride catalyzed branched selective electrophilic hydroallylation of terminal alkynes with allylic electrophile to access terminal skipped dienes with good regioselectivity and functional group tolerance under mild conditions. The reaction could be carried out on gram scale, and the TON was up to 1160. The primary mechanism of electrophilic allylation of α-selective cobalt alkenyl intermediate was proposed based on deuterium labeling experiment and kinetic studies. Various metal-hydridecatalyzed regioselective hydrofunctionalization of terminal alkynes will be further explored in our laboratory.

Materials
For NMR spectra of compounds in this manuscript, see Supplementary  General procedure for hydroallylation of terminal alkynes A 25 mL Schlenk flask equipped with a magnetic stirrer and a flanging rubber plug was dried with flame under vacuum. When cooled to ambient temperature, it was vacuumed and flushed with N 2 . This degassed procedure was repeated for three times. Then (L3-H)•CoBr (0.025 mmol, 5 mol%), THF (1.0 mL, 0.5 M), PMHS (0.75 mmol, 1.5 equiv.), terminal alkynes (1.0 mmol, 2 equiv.), allylic bromides (0.5 mmol, 1.0 equiv.), and LiOtBu (0.75 mmol, 1.5 equiv.) were added sequentially. The reaction was run at 50°C for 30 min to 4 h. Then the resulting solution was quenched with 10 mL of PE and filtered through a pad of silica gel, washed with PE/ EtOAc (5/1) (3 × 20 mL). The combined filtrate was concentrated under vacuum and the ratio of b/l was monitored by 1 H NMR analysis. The mixture was purified by flash column chromatography to give the corresponding product. Article https://doi.org/10.1038/s41467-022-32291-3