The evaluation of HER catalysts. To better understand the relationship of structure and redox property, we initially synthesized a series of Co (II)-salen48–49 HER catalysts cat 1-cat 7 (Fig. 3a). The novel structure of pentacoordinated catalyst was unambiguously identified by the single crystal diffraction analysis of cat 1 (Fig. 3b),50 and it shows that cobalt center is shielded by a roof-like nitrogen coordination site, which would significantly enhance its stability. Redox property of the catalysts was next explored to support the above speculation. We concluded all of the redox peaks of catalysts arising from the process of CoIII/CoII and CoII/CoI (Fig. 3c). As expected, pentacoordinated catalysts (cat 1-cat 3, 1.98–1.92 V) have uniformly larger redox potential gap than that of the conventional catalysts cat 4-cat 7 (1.14–1.34 V). This result verifies that pentacoordinated Co-salen catalysts are more stable under oxidation and reduction compared with the tetracoordinated counterparts. Noteworthy, CF3 group (cat 1) was found to significantly improve the oxidation potential of CoII/CoIII with the most positive peak at -0.157 V (vs. Fc/Fc+). This redox property endows cat 1 good compatibility with the aforementioned cocatalyst TEMPO. Moreover, we also compared the anodic oxidation of cat 1 (-0.157 V) with that of TEMPO (0.195 V); the narrow potential difference51 made it possible using cat 1 to facilitate the oxidation TEMPO. Further cyclic voltammetry experiment mixing cat 1 with TEMPO confirms the above conclusion by detecting obvious increase of the anodic peak of cat 1 (Fig. 3d). Nevertheless, other cobalt catalysts failed to produce the same results when mixing with TEMPO. Taken together, cat 1 showed good compatibility with the H−T catalyst TEMPO. The synergistic effect of the catalyst combination was demonstrated by accelerating the catalytic cycle of TEMPO at lower anodic potential52, thereby obviating undesired overoxidation.
Reaction optimization. Encouraged by the redox property study, we next investigated the cross-coupling reaction between benzyl alcohol (1a) and allylbenzene (2a) with various catalyst combinations. We screened catalysts using Cs2CO3 as base additive, nBu4NClO4 as electrolyte, DMF as solvent, graphite felt and copper plate as anode and cathode, respectively. As shown in the Table 1, a suitable catalyst combination is crucial for the reaction efficiency in terms of yields (Entry 1–10). The optimal catalyst combination was identified with cat 1 as HER catalyst and TEMPO as H-T catalyst, and the desired C-C coupling product 3a was obtained in 85% yield with exclusive C-C coupling and isomerizing-selectivity. By comparing the yields arising from other catalyst combinations, we concluded that HER catalyst with larger potential gap and H-T catalyst with lower oxidation peak would benefit the reaction performance. The necessity of both catalysts was also supported by the control experiments removing either of the catalysts (Entry 11–13). We also highlighted the superiority of this electrochemical protocol when compared to conventional CDC conditions (Entry 14, see Table S3 for details).
Exploration of Scope. With the optimal reaction conditions in hand, a wide range of substrates bearing weakly acidic C(sp3)-H were investigated to couple with benzyl alcohol 1a (Table 2). Initially, we examined a series of toluene derivatives, and the desired C-C coupling products (3b-3i) were obtained with moderate yield. Notably, bioactive amide and sulfonamide substrates (3e, 3h-3i) bearing reactive α C-H were well-tolerated. Moreover, diaryl ketone was also amenable to give the corresponding product 3f, which conventionally preferred to proceed a Pinacol coupling. To our delight, mixture of β-methylstyrene isomers can be directly used as substrate to afford corresponding products with high E/Z selectivity (3j, 3k). Subsequently, various allylbenzene derivatives were tested in the electrochemical protocol. Substrates with multiple double bonds (3l) or different substitution patterns (3m-3p) are all well-tolerated to deliver the desired products with exclusive isomerizing-selectivity and E/Z selectivity. The site-selectivity of the electrochemical approach was also demonstrated by the case of 3q, and more acidic C-H is favored in the reaction. This cross-coupling reaction also enabled a late-stage functionalization for the natural product Magnolol and Eugenol derived substrates (3r-3s). Specifically, both of the allyl groups underwent isomerization to give a mono-coupling product 3r. Additionally, we investigated the electronic effect (3t-3ac) and position of substituents (3ad-3af), fused ring (3ag-3ah) and heterocycle (3ai). Uniformly good yields were observed for the substrates, although thienyl group led to diminished E/Z selectivity. It is noteworthy that radical sensitive groups such as cyclopropyl (3o, 3w) and ortho-vinyl (3af) were untouched during the electrochemical transformation, ruling out the possible radical pathway. Finally, toluene, hexene and 4-phenyl-1-butene proved to be failed substrates in the reaction due to their less acidic C-H bonds.
To further demonstrate the generality of the electrochemical cross-coupling reaction, we next tested a broad range of alcohols (Table 3). In general, variations on the electronic property (3aj-3bf) and substitution pattern (3bg-3bp) of benzyl alcohols are well tolerated. Electron-deficient substrates (3ar-3ax) with higher oxidation potential proved to be less efficient compared with the electron-rich alcohols (3aj-3aq). The functional-group tolerance of this approach was highlighted by the cases of 3ay-3bd, which commonly cannot survive under conventional CDC conditions. Furthermore, site-selectivity favoring less-hindered benzylic C-H was observed in the substrates containing two hydroxyl groups (3be-3bf). Other aromatic (3bq-3bv), aliphatic (3bw-3ca), secondary (3cb-3cc) and bioactive molecule (amylcinnamyl alcohol, Adapalene) derived alcohols (3cd-3ce) were also found to be suitable substrates to afford corresponding products with synthetically useful yields. Alcohols bearing cyclopropyl group gave the desired products 3bw-3bx. None of the ring expansion byproduct was detected, further confirming an ionic pathway.
Applications in Synthesis. Having examined the reaction generality, we turned our attention to probe the utility of our electrochemical protocol with gram-scale reaction and derivatization of products (Fig. 4). Gratifyingly, scaling up the model reaction to gram-scale afforded product 3a in a satisfactory yield (70%) even with lower catalyst loading (Fig. 4a). The byproduct, that is, the hydrogen gas was successfully collected using a balloon and verified with GC. This result suggests that protocol not only provides a route for organic transformation but also enables an avenue for fuels. Using a simple solar cell as electricity supply largely maintained the reaction efficiency (Fig. 4b). Under acidic conditions, homoallylic alcohol 3l proceeded a facile dehydration to give a conjugated light-emitting molecule 4 (Fig. 4c), which displayed prominent luminescent properties both in solution and solid state with maximum emission peak at 493 nm. Additionally, bioactive tetrahydrofuran product 5 was also accessed in high yield via NIS initiated iodocyclization.
Mechanism investigation. To gain insight into the reaction mechanism, a series of cyclic voltametric experiment and control experiments were conducted (Fig. 5). First, the electrochemical properties of 1a and 2a were investigated (Fig. 5a-5b). As shown in Fig. 2A, 1a and 2a seem to be redox inert within the window of -2.5-2.5 V (vs. Fc/Fc+). After enlarging the range from − 0.5 to -2.5 V, a couple of small waves (10− 6 ~ 10− 5 A) were detected at -2.04 V, which were attributed to the hydrogen atom absorption and desorption process in HER (see Fig. S20 for GC detection of H2). Moreover, their intensity increases with the concentration of 2a. These observations illustrate that both of substrates are relatively unactive in the absence of catalysts. Second, catalytic role of cat 1 and TEMPO was explored (Fig. 5c-5d). Treatment of cat 1 with excessive 2a led to significant increases of both cathodic peaks of 2a and CoII/CoI; this observation verifies the catalytic role of cat 1 in the HER of 2a. On the other hand, mixing TEMPO with benzyl alcohol 1a also resulted in an obvious catalytic current suggesting the catalysis of TEMPO in the oxidation of 1a. Third, we conducted control experiments to elucidate the necessity of both electrode reactions (Fig. 5e). In a divided cell, no desired coupling product 3a was observed (Eq (a)), while benzaldehyde and allylbenzene carbanion (see Fig. S26 for its UV-vis spectra) were detected in anodic and cathodic chamber, respectively. Further replacing TEMPO with stoichiometric oxidant 2,2,6,6-tetramethyl-1-oxopiperidinium tetrafluoroborate only afforded benzaldehyde (Eq (b)). These results exclude the mechanism which only relies on anodic oxidation. Kinetic isotope effect (KIE) study was also investigated with intramolecular competition and parallel experiment,53 and it reveals that the oxidation of 1a is the rate-determining step (Eq (c)-(d)).
On the basis of the above experimental observations and related mechanism reports,37,44–46 a plausible reaction mechanism was proposed (Fig. 5f). Under anodic oxidation, pentacoordinated CoII-salen (cat 1) is first oxidized to CoIII species, which could subsequently initiate the oxidation of TEMPO via single electron transfer (SET). The in situ generated reactive species A abstracts a hydride from 1a to deliver cation C and regenerates a protonated TEMPO (B). Over the cathode surface, the dissociation of proton from weakly acidic C(sp3)-H is promoted by the CoI species arising from cat 1, and isomerized carbanion D is simultaneously generated along with hydrogen byproduct. Direct reaction between C and D affords the dehydrogenative product 3a.