Reaction optimization. In comparison with various hydroaminoalkylations of alkenes, the difficulty of hydroaminoalkylation of alkynes was ascribed to the following possible reasons: (1) strong basicity and nucleophilicity of alkyl or aryl amines could coordinate to metal centers, resulting ni either deactivation of transition metals or undesired side reactions such as hydroaminations; (2) weak acidity of N−H bonds cannot effectively undergo protonolysis of metallacycle intermediates. Thereby, the selection of proper N-protecting groups to increase the acidity of amines could be critical to the reaction efficiency, because more acidic amine would significantly reduce its coordination with metal centers and other side reactions such as hydroamination. However, to accommodate acidic N−H bonds, sensitive early transition metal complexes should be replace by late transition metals such as Pd, Ru and Ni, because they could have better compatibility with protonic substrates and solvents.
Following this hypothesis, we conducted an extensive survey on N-protecting groups, transition metals, ligands and other reaction parameters. Ultimately, triisopropylsulfonyl (TPS) was identified as the superior N-protecting group and Ni/IPr/PCy3 was identified as the optimal catalyst. With their combination, hydroaminoalkylation of alkyne 2a with N-TPS amine 1a smoothly proceeded under mild conditions (80 °C), providing the corresponding allylic amines 3a in nearly quantitative yield (Fig. 2, entry 1).
Control experiments showed that the alteration of TPS resulted into significantly diminished yields (entries 2 − 8). For example, the replacement of isopropyl groups (TPS) by methyl groups (TMS) gave only 9% yield (entry 2). Common p-tolylsulfonyl (Ts) further decreased the yield to a trace amount (entry 3). The combination of NHC (IPr) and phosphine (PCy3) ligands also proved critical to the reaction (entries 9 − 17). The absence of IPr.HCl completely inhibited the reaction (entry 9), whereas the reaction still gave 3a in 14% yield without the addition of PCy3 (entry 13), demonstrating the vital role of IPr and the promoting effect of PCy3. In fact, a yield of 68% was detected with IPr alone at an elevated temperature (110 °C) but with poor reproducibility (see the Supporting Information for details). Other carbenes and phosphines were less effective (entries 10 − 12 and 14 − 17). Without Ni(cod)2 or with other nickel species, the reaction did not work (entries 18 − 20).
Scope of amines and alkynes. Under the optimized conditions, various N-TPS amines were then examined (Fig. 3). Results showed that the reaction tolerated a broad range of functional groups on the phenyl ring of N-benzyl amines, including simple alkyl (Me, 3b − 3d), electron-donating groups (alkoxy, 3e, and 3f), and electron-withdrawing groups (OCF3, F, Cl, CF3, CN, and CO2Me, 3 g − 3n), providing the corresponding allylic amines in 62 − 94% yields. In addition, the position of substituents did not have a strong influence on the reaction yield (3b − 3d and 3 h − 3j). Notably, both 1-naphthyl (3o) and heteroaryl (3p) instead of the phenyl of 1a also worked well, affording both 86% yields. When the phenyl was replaced by the alkenyl, a decreased yield was obtained (45%, 3q) in the presence of 10 mol% of the catalyst at 110 oC. Notably, various N-alkyl amines were still compatible with the reaction (3r − 3u, 41 − 54% yields), but requiring harsher conditions (130 oC and 20 mol% catalyst) and a Ts protecting group.
Next, a broad range of alkynes were investigated under the standard conditions (Fig. 4). Various diaryl alkynes bearing alkyls (4a − 4d) and electron-donating groups (4e and 4f) on the phenyl rings were well compatible with the current reaction, providing the corresponding products in 79 − 92% yields. Notably, 2-tolylalkyne gave a 1:1 mixture of E:Z isomers (4c), probably because the significant steric hindrance on the aryl ring forced the alkene to isomerize.
In contrast, electron-deficient groups such as OCF3 (4 g), F (4 h), and CF3 (4i) on the phenyl ring led to slightly lower yields even at a higher temperature. In addition, both dialkyl alkynes (4j and 4 k) and alkyl aryl alkynes (4 l − 4p) were well tolerated, providing both good yields and good to excellent regioselectivities. For example, 1-phenylpropyne gave 8.1:1 regioisomeric ratio (4 l), and the change of methyl to ethyl significantly increased the ratio to 20:1 (4 m). Bulkier alkyls (4n − 4p) or silyl (4q) led to a single regioisomer. However, non-symmetrical dialkyl alkyne (4r) cannot afford good regioselectivity probably owing to low differentiation between isopropyl and methyl groups.
Reaction Utility and mechanistic discussion. To demonstrate the utility of the reaction, a gram-scale reaction of the model substrates was conducted under the standard conditions, affording the desired product 3a in 88% yield, without significant loss of the yield (Fig. 5a). In addition, the formed allylic amine 3a can act as a versatile synthetic intermediate to participate into various transformations. For example, hydrogenation followed by typical deprotection protocol of the sulfonyl group provided compound 5 in 68% yield. And direct oxidation of 3a resulted into a synthetically useful α-amino ketone 6 in 90% yield.
To gain insights into the reaction mechanism, some mechanistic experiments were carried out. Deuterium labeling experiments revealed that 100% allylic deuterium and 94% olefinic deuterium existed in product d-3a, suggesting that one benzylic hydrogen atom was transferred to the olefinic position (Fig. 5b). In addition, deuterated Z-stilbene was obtained, indicating that a part of alkynes were reduced during the reaction process. Crossover experiments between d-1a and 1e suggested that the allylic and olefinic hydrogens may originate from different amide molecules (Fig. 5c), excluding an oxidative addition pathway. The observed kinetic isotopic effect (kH/kD = 2.7 in the intermolecular competitive reaction and kH/kD = 2.2 in parallel reactions, Fig. 5d) implied that the cleavage of the benzylic C − H bond could be involved in the rate-determining step. Notably, in case of dimethylamino benzylic amide 1v as the substrate, the imine 1v′ was detected (Fig. 5e). Moreover, the competitive reaction between amide and the imine showed that both of them gave the corresponding products in comparable yields (see Scheme S7). These results suggested that an imine intermediate could be involved in the catalytic cycle.
In addition, the stoichiometric reaction of a five-membered nickelacylce43−45 and amide 1a with or without IPr afforded the desired product 3b in 68% and 9% yields, respectively, suggesting that both the nickelacycle and IPr were critical to the reaction (Scheme S10). Based on these mechanistic experiments and previous literature reports,46−50 a possible reaction mechanism was proposed (Fig. 6). At the induction stage, the nickel-catalyzed transfer hydrogenation of alkyne 2a with sulfonamide 1a furnishes Z-stilbene and imine 1a′. Then, 1a′, 2a, and the nickel catalyst undergo an oxidative cyclometallation to generate nickelacycle B, which is subsequently protonated by 1a. The resulting intermediate C then proceeds through a direct intramolecular hydrogen transfer to give Ni − product complex D. Finally, catalyst transfer between D and 2a occurs, releasing product 3a and completing the catalytic cycle.
To further shed light on each individual elementary step of the catalytic reaction, we performed DFT calculations on the model reaction of N-benzylbenzenesulfonamide and diphenylethyne in the presence of a simplified Ni/NHC catalyst (see Figure S1 for details). At the induction stage, two critical steps were a ligand-to-ligand hydrogen transfer (LLHT) via TS1 with an activation Gibbs energy of 14.3 kcal/mol and an intramolecular hydrogen transfer via TS2 with an overall activation Gibbs energy of 18.1 kcal/mol. At the product-formation stage, the turnover-limiting transition state is predicted to be the hydrogen transfer transition state TS5 with an overall activation Gibbs energy of 24.8 kcal/mol, which is accordance with the observed kinetic isotopic effect (Fig. 5d). Notably, an activation Gibbs energy for the transition state of the oxidative cyclometallation (TS3) is 23.7 kcal/mol, which is a little lower than that of TS5. DFT calculations with TPS-protected substrate 1a and IPr ligand indicated that the overall activation Gibbs energy is 22.4 kcal/mol. Replacement of TPS by Ts leads to a higher overall activation Gibbs energy of 24.9 kcal/mol. These results suggested that, as compared with the Ts group, a ca. 30-fold acceleration effect of the TPS group would be expected at 80° C, which nicely reproduced the experimentally observed superior performance of the TPS protecting group (Fig. 2, entry 1 vs entry 3). In addition, DFT calculations also suggested that the presence of PCy3 could not reduce the overall activation Gibbs energy of the [Ni(NHC)]-catalyzed reaction (see Fig S3), and instead, PCy3 may act as an auxiliary ligand to facilitate the generation of the catalytic species and/or to inhibit catalyst deactivation.