Investigation of reaction conditions. We started with our investigations by using alkynyl sulfide 1a and commercially available TMSCl as the model substrates. Extensive examination of the reaction parameters revealed that the combination of 10 mol% of CoI2 as the catalyst and 15 mol% of L8 (tpy) as ligand with zinc as the reductant in DMF at room temperature was the optimal conditions to deliver the desired product 3a in 85% isolated yield (Table 1, entry 1, see Supporting Information Tables S1-S5 for details). Different Co(II) precursors were tested, CoCl2, CoBr2 and Co(acac)2 gave the product 3a in high yields but with less efficiency (entries 2–3). Meanwhile, the cross-selectivity decreased dramatically and a certain amount of 1,3-diyne was obtained when NiI2 or CrCl2 was utilized to instead of CoI2 as the catalyst (entry 4).20,24,27 The alkynyl sulfide 1a was fully recovered in the presence of iron salt precatalyst (entry 5).26 Solvent screening demonstrated that acetonitrile (MeCN) or tetrahydrofuran (THF) shut down the reaction completely (entry 6). The replacement of Zn to Mn resulted no conversion of this reaction (entry 7). When 1 equiv of ZnI2 was added together with Mn, the reaction generated the product 3a in 67% yield (entry 8). This indicated that Zn not only acted as reductant, but also the in-situ formed zinc salt as a thiophilic reagent to decrease the inhibition of the leaving sulfide part to catalyst.40 Control experiments showed that the stoichiometric reductant that Zn not only acted as reductant, but also the in-situ formed zinc salt as a thiophilic reagent to decrease the inhibition of the leaving sulfide part to catalyst.40 Control experiments showed that the stoichiometric reductant zinc, the cobalt catalyst and ligand were all essential for this transformation (entry 9). The effect of ligand indicated that bidentate ligands such as bipyridine and phenanthroline derivatives though were not as good as terpyridine L8, but still gave the desired product in reduced yields. Conversely, the sterically hindered ligand L7 failed to give the desired product 3a.
Substrate scope. With the optimized reaction conditions in hand, we explored the scope of alkynyl sulfides for this cobalt-catalyzed XEC with TMSCl or PhMe2SiCl, and the results were summarized in Fig. 2. Alkynyl sulfides with different electronic properties or substitution patterns (3a-t) were tolerated. Firstly, alkynyl sulfides with diverse electron-donating groups at the para-position of the aryl moiety, including ether (3a-c), -OAc (3f), -OCF3 (3g), -OTIPS (3h) and -NHAc(3j), were converted into the corresponding products in good yields (41–93%). The method also demonstrated high Csp-S chemo-selectivity as reactions with an alkynyl sulfide containing an aryl sulfide moiety to give the product 3i in 91% yield.41–43 Additionally, the electron-withdrawing groups, such as -F (3k-l), -CN (3m), -CO2Me (3n), -CONMe2 (3o) and -CO2H (3p) were all compatible with the method and delivered the desired products in moderate to excellent yields (55–93%). Although the presence of a terminal alkene or alkyne was challenging to the silylation procedure,44,45 the corresponding products 3q and 3r were successfully obtained in 75% and 88% yield, respectively. Moreover, a free alcohol (3s) or amine (3t), which could react directly with chlorosilanes, was well tolerated under the standard conditions. When alkyl or alkenyl substituted alkynyl sulfides were selected as substrates, the alkynylsilylation reactions occurred as expected to afford the corresponding products in excellent yields (3u-y). The reactions of alkynyl sulfides containing polyaromatic or heteroaromatic ring proceeded smoothly to produce alkynylsilanes (3z-aj) in moderate to good yields, including fluorene (3z), naphthalene (3aa) thiophene (3ab, 3ac), pyridine (3ad), quinoline (3ae), benzofuran (3af), indole (3ag), benzothiophene (3ah), carbazole (3ai) and dibenzofuran (3aj). Given the broad substrate scope and good functionality tolerance of our strategy, we further explored this transformation to for late-stage functionalization with more complex molecules. Alkynyl sulfides derived from Donepezil (3ak), Desloratadine (3al), Citronellol (3am), DL-Menthol (3an), Gemfibrozil (3ao), and Cholesterol (3ap), were all suitable substrates, providing the corresponding products in 41–87% yields.
Next, we explored the reactivity of chlorosilanes by varying the substituents on silicon for these transformations (Fig. 3). A series of chlorodimethyl(alkyl)silanes with varied chain lengths and steric properties reacted with alkynyl sulfide 1a, giving the corresponding alkynylsilylation products in moderate to high yields. For instance, chlorodimethyl(ethyl) silane (4a), chlorodimethyl(butyl) silane (4b), chlorodimethyl(3-phenylpropyl) silane (4c) and chlorodimethyl-(butanenitrile) silane (4d) furnished the desired products in 83–96% yields. Chlorosilanes bearing steric substituents, such as diaryl (4f), isopropyl (4g), cyclohexanyl (4h) and TMS (4i), all worked well for this cobalt-catalyzed system, affording the corresponding products with good efficiency. The presence of allylic and vinyl functionalities on the chlorosilanes was well-tolerated, offering opportunities for further diversification of the resulting products (4j-l). It is worth noting that several chlorotrialkylsilanes with larger steric hindrance were also tested, yielding the corresponding products in the range of 69–71% (4m-n).
To further exhibit the excellent chemo-selectivity of alkynyl sulfides as alkynyl electrophiles for these reactions, we conducted the XEC with alkynyl sulfides containing various electrophilic and nucleophilic reaction sites and the results were summarized in Fig. 4. These studies demonstrated that this Co-catalyzed silylation of alkynyl sulfides offered orthogonal selectivity towards other well established cross coupling methods. Aryl electrophiles, including chloride (3aq), bromides (3ar-as), iodide (3at), tosylate (3au), and triflates (3av, 3aw), were all tolerated under the standard conditions. In addition, alkynyl sulfides containing additional electrophilic functional groups were converted to alkynylsilanes in good yields (3ax-bc), which provided the chance to further modify these molecules. An alkynyl sulfide containing an alkyl chloride moiety was selectively silylated at the Csp-S site (3bd). Alkynylsilane containing a boronic ester group (3be) was constructed in 85% yield for this Co-catalyzed XEC.
Alkynylsilanes are synthetic useful intermediates, which have been widely used in synthetic chemistry.46–54 To show the synthetic utility of this protocol, we conducted a series of further transformation of the desired alkynylsilane. First, a gram-scale reaction of alkynyl sulfide 1ay with a reduced catalyst loading (5 mol%) to generate alkynylsilane 3ay in 71% yield. Tetrabutylammonium fluoride catalyzed the addition of 3ay to trifluoromethyl ketone, producing the CF3-substituted tertiary propargylic alcohol 5a in moderate yield (Fig. 5-a). A Cu-catalyzed three-component coupling reaction of 3ay, o-hydroxybenzaldehyde and amine, formed the corresponding benzofuran 5b in 76% yield, involving an intramolecular 5-exo-dig cyclization (Fig. 5-b). A copper-catalyzed three-component coupling reaction of alkynylsilane 3ay, aldehyde and amine generated the propargylic amine 5c in 84% yield (Fig. 5-c). A ZnCl2-catalyzed Diels-Alder/retro-Diels-Alder reaction between electron-deficient 2-pyrone and alkynylsilane 3ay enabled the synthesis of arylsilane 5d in 73% yield (Fig. 5-d). The gold-catalyzed reaction of o-alkynylbenzaldehydes with alkynylsilane 3ay gave naphthylsilane 5e in 49% yield (Fig. 5-e). And the Cobalt(I)–diphosphine catalyzed dehydrogenative annulation between alkynylsilane 3ay and salicylaldehyde afforded the corresponding 2-aryl-3-silylchromone 5f in 69% yield (Fig. 5-f). Moreover, a Co-catalyzed regioselective [3 + 2] annulation of ortho-functionalized arylboronic acid with 3ay gave the corresponding cyclized products 5g and 5h in 81% and 65% yield respectively (Fig. 5g-h).
To elucidate the underlying mechanism of these reactions, a set of experiments were conducted as shown in Fig. 6. Firstly, to reveal whether a radical process was involved in the activation of R3Si-Cl, several radical acceptors (3.0 equiv) were added to the standard reaction conditions and 4e was obtained in 69–82% yields with no sign of the expected side product 7 from radical trapping (Fig. 6a). A radical clock experiment of chlorosilane 2o was performed, and the directly silylated product 4o was obtained in 87% yield with no cyclized derivative dectected (Fig. 6b). These results suggest that Si-Cl is not activated through a radical process. Subsequently, alkynyl sulfide 1a was transformed to the homodimer (diyne) product 1a-1 in 89% yield in the absence of chlorosilane, which indicated that oxidative addition of alkynyl sulfide with low valent Co to form Csp-Co species was possible. Moreover, in the absence of alkynyl sulfide, chlorosilane 2l failed to dimerize. In addition, adding 3 equivalents of water to the reaction system under standard conditions, the corresponding terminal alkyne 1a-2 was generated in 72% yield. These results further indicated that the low valant cobalt catalyst might activate alkynyl sulfide firstly and generate the Csp-Co species (Fig. 6c).
On the basis of these experimental results and previous reports,20,21,23 a catalytic cycle was proposed in Fig. 6d. A low-valent Con species was generated in-situ via reduction of Co(II) precursor with zinc. Then, the alkynyl sulfide 1 reacted with Con via oxidative addition to give Csp-Con+2 Int. A, which was subsequently reduced by Zn to provide a more nucleophilic alkynyl Con+1 Int. B.55,56 The SN2 oxidative addition of chlorosilanes to Int. B57,58, possibly through a five-coordinate cobalt intermediate C,21,59,60 to deliver the complex D. Reductive elimination of complex D would afford the desired product with generation of Con+1, which was further reduced to Con by zinc to enter next catalytic cycle.