NiH-catalysed proximal-selective hydroalkylation of unactivated alkenes and the ligand effects on regioselectivity

Alkene hydrocarbonation reactions have been developed to supplement traditional electrophile-nucleophile cross-coupling reactions. The branch-selective hydroalkylation method applied to a broad range of unactivated alkenes remains challenging. Herein, we report a NiH-catalysed proximal-selective hydroalkylation of unactivated alkenes to access β- or γ-branched alkyl carboxylic acids and β-, γ- or δ-branched alkyl amines. A broad range of alkyl iodides and bromides with different functional groups can be installed with excellent regiocontrol and availability for site-selective late-stage functionalization of biorelevant molecules. Under modified reaction conditions with NiCl2(PPh3)2 as the catalyst, migratory hydroalkylation takes place to provide β- (rather than γ-) branched products. The keys to success are the use of aminoquinoline and picolinamide as suitable directing groups and combined experimental and computational studies of ligand effects on the regioselectivity and detailed reaction mechanisms.

Herein, we report a NiH-catalysed proximal-selective hydroalkylation of unactivated alkenes to access βor γ-branched alkyl carboxylic acids and β-, γor δ-branched alkyl amines. Using NiCl 2 (PPh 3 ) 2 as the catalyst, migratory hydroalkylation takes place to provide β-(rather than γ-) branched products (Fig. 1b). A preliminary result with moderate enantioselectivity demonstrates that an asymmetric variant can be realized. The keys to success are the application of aminoquinoline and picolinamide as suitable directing groups and detailed studies of the ligand effects and reaction mechanisms.

Results and discussion
Reaction discovery. We commenced this study with alkene 1a and alkyl iodide 2a as the model substrates (Table 1). Amidedirected proximal-selective hydroalkylation was efficiently  performed using NiBr 2 (diglyme) as the catalyst, MEA as the ligand, DEMS as the hydride source, KF as the base in a DMAc solvent (entry 1), and desired product 4aa was obtained in 90% GC (gas chromatography) yield, 87% isolated yield, and >20:1 r.r. (regioisomeric ratio). NiBr 2 (DME) and Ni(acac) 2 could be used instead of NiBr 2 (diglyme) as catalysts (entries 2-3), but NiCl 2 (PPh 3 ) 2 , which would introduce a phosphine ligand into the reaction system, led to a dramatic negative effect on regioselectivity (entry 4). Oxygen-bearing silanes, such as (MeO) 3 SiH and PMHS (entries 5-6), gave good yields; Ph 2 SiH 2 also performed well (entry 7), but the less electrophilic hydrosilane MeEt 2 SiH was inefficient in this transformation (entry 8). KF turned out to be the best base over the other bases (entries 9-12), such as Na 2 CO 3 , CsF, NaHCO 3 , and K 3 PO 4 (H 2 O), because of its matched ability to activate DEMS to form pentacoordinate hydrosilicate species. We could regulate the reaction activity through the crosscombination of different bases and silanes. This reaction is sensitive to solvent identity (entries 13-17); an excellent yield was obtained in DMF, which is an amide solvent (entry 13), while poor results were obtained using MeCN, DCE, or toluene (entries [14][15][16], and disappointing regioselectivity was obtained when using THF as the solvent (entry 17). Finally, this reaction was completely shut down in the control experiment without a nickel catalyst (entry 18).
Substrate scope. With appropriate reaction conditions in hand, we sought to examine the substrate scope of this amide-directed nickel-catalysed hydroalkylation. As summarized in Fig. 2, a wide range of alkyl halides and alkenes was examined and provided satisfying coupling yields (40-88%) and regioselectivities (4.0:1~>20:1) in almost all cases. Both alkyl iodides 2a-2l, 2n, 2q-2s and bromides 3a, 3m, 3o, 3p, and 3t were suitable substrates, and 0.5 equiv NaI was added for the in situ generation of the corresponding alkyl iodides from alkyl bromides. The β-branched alkyl halides (2b and 2c) delivered products in slightly diminished yields and equally good regioselectivities. Concerning alkenes, both terminal (1a) and internal alkenes with differentiated steric resistances (1b-1g, 1i-1l) could be used. The Z/E configuration of the internal alkene would not affect the coupling yield or regioselectivity (4ba). In addition, the trisubstituted alkene (1g), which is difficult for nickel-hydride insertion, could be transformed in a moderate yield. An interesting example was 4la. The alkene substrate 1l with a potentially competitive activated benzylic reaction site delivered the desired amide-directed γalkylation product with high regioselectivity, indicating that the aminoquinoline auxiliary displayed stronger directing ability compared to an aromatic ring. In another example, the derivative of but-3-enoic acid (1m) was selectively converted to the βbranched product 4ma following proximal hydroalkylation selectivity. Finally, picolinamide as an auxiliary also produced gratifying proximal hydroalkylation results (β-alkylation product 4na, γ-alkylation product 4oa, and δ-alkylation product 4pa) under standard conditions. However, the moderate regioisomeric ratios (4.0:1~4.8:1) need further reaction condition optimization. A variety of functional groups were well accommodated, such as trifluoromethyl (4ad), ester (4ae), ether (4af, 4ag and 4ea), and cyano (4ah, and 4ai) groups. Under mild reaction conditions,    many sensitive functional groups could be tolerated, including base-sensitive ketones (4aj and 4ak) and acid-sensitive tertbutoxycarbonyl (4al) and propylidene (4ka) protecting groups. Phosphonate (4am) and internal alkenyl groups (4an) were compatible during the transformation. Heterocyclic compounds such as thiophene (4ao), indole (4ap), tetrahydropyran (4ia), and furan (4ja) posed no problems. Electrophilic aryl chloride (4aq) and aryl bromide (4ar), as well as nucleophilic arylboronate (4as) survived, providing the possibility for further transformation through transition-metal-catalysed cross-coupling reactions. Finally, this reaction could be conducted in the presence of an unprotected hydroxyl group (4at), demonstrating its high degree of compatibility with sensitive functional groups. We examined several secondary and tertiary electrophiles (see Supplementary  Table 5 and Supplementary Fig. 20 for more details). For the reaction using iodocyclohexane as an electrophile, the hydroalkylation product was delivered in a 23% yield (2.4:1 r.r.) along with the alkene protonation byproduct in a 38% yield; slight modification using (MeO) 3 SiH as hydride source gave out a 35% total yield (1.9:1 r.r.). For the reaction using tertiary electrophiles, such as 2-iodo-2-methylpropane, 2-bromo-2-methylpropane, and 1bromoadamantane, the desired hydroalkylation products were not observed. In the example of trisubstituted alkene 4ga, a secondary-primary C(sp 3 )-C(sp 3 ) bond formed with a moderate yield. However, the attempted construction of a tertiary-primary C(sp 3 )-C(sp 3 ) bond failed in the examination of 1,1-disubstituted alkene and tetrasubstituted alkene (see Supplementary Fig. 21 for more details). Therefore, this reaction might be sensitive to the steric hindrance of both alkyl halides and alkene substrates.
Synthetic applications. We explored the usefulness of this hydroalkylation reaction as an efficient method for the late-stage modification of biorelevant molecules (Fig. 3a). Alkyl halides derived from drug molecules and natural products, such as indomethacin, lithocholic acid, methylumbelliferone, and D-galactopyranoside, underwent the hydroalkylation process smoothly with high chemoselectivity and good yields. The tolerance of aryl chloride, acetal groups, and unprotected alcohol hydroxyl groups highlights the excellent compatibility with sensitive functional groups. In a scale-up reaction (Fig. 3b), we achieved alkene hydroalkylation product 4aa with a satisfactory 78% isolated yield and 12:1 regioselectivity. Under mild nickel-catalysed conditions, the directing amide group was transformed into a methyl ester group to deliver the desired methyl carboxylate 5a in a 70% isolated yield. After hydrolysis of 4aa in hydrochloric acid solution, alkyl carboxylic acid 5b was obtained in a 73% isolated yield with 96% aminoquinoline recovered. The gram-scale reaction and successful removal of the directing group highlight the practicality of this nickel-catalysed alkyl-alkyl coupling reaction.
Mechanistic studies. We carried out a series of control and mechanistic experiments to elucidate the reaction mechanism (Fig. 4). In the radical clock experiment (Fig. 4a) using (bromomethyl)cyclopropane (3y) as an electrophile, coupling product 4ay originating from the ring-opening of the cyclopropylmethyl radical was detected. When 6-iodohex-1-ene (2z) was subjected to the standard conditions, we observed that the expected ring-cyclized product 4az originated from the cyclization of the 5-hexenyl radical. These data support the radical nature of this reductive hydroalkylation reaction. In the isotope labelling experiment (Fig. 4b) with Ph 2 SiD 2 (97% deuterium content) under standard conditions, deuterium was incorporated exclusively at the terminal position with a 95% deuterium content, indicating that hydrogen is provided by hydrosilane via the generation of NiH species. In addition, we experimented with δ,δ-dideuterated alkene d 2 -1a. The δ-position deuterium atoms in the dideuterated substrate remained intact, and no H/D exchange was observed along the alkyl chain. Next, we examined the influence of nitrogen-containing ligands on the coupling yield and regioselectivity (Fig. 4c). This reaction could be conducted in the absence of ethanolamine ligand or by using ethylenediamine (L1), butylamine (L2), or aniline (L3) as the ligand. However, diminished 5.3:1~7.8:1 regioisomeric ratios demonstrated the significance of ethanolamine (MEA) in regulating the regioselectivity. Moreover, bidentate bipyridine (L4) and tridentate terpyridine (L5) ligands showed a conspicuous decline in coupling efficiency and regioselectivity. This could be attributed to the excessive occupation of the coordination space on the nickel centre. In a preliminary experiment using chiral diamine ligand L6 (Fig. 4d), this proximal-selective hydroalkylation reaction could be realized with moderate enantioselectivity (see Supplementary Tables 2 and 3 Fig. 2, 0.2 mmol scales. Standard conditions B: determination of regioisomeric ratio was consistent with standard conditions A, 0.1 mmol scales. Standard conditions C: total yield for the mixture of all regioisomers. r.r. refers to the regioisomeric ratio, that of the major product to the sum of all other isomers. In most cases, a mixture of βand γ-selective products was obtained; we could hardly observe other isomers. * 1 H NMR yield with dibromomethane as an internal standard. † GC yield with triphenylmethane as an internal standard. ¶ Isolated yield in parentheses. using Tang and Xu's method 69 . Inspired by the precedents of Koh and others 15,67,68 , we focused our attention on investigating the ligand effect of phosphine ligands (Fig. 4e). Both PPh 3 and PMe 3 altered the regioselectivity, and migratory β-alkylated amide (6aa) was isolated as the main product. After performing systematic screening of the influence of other reaction parameters, we determined that NiCl 2 (PPh 3 ) 2 as the catalyst, (MeO) 3 SiH as the hydride source, KF as the base, and in DMAc solvent provided a moderate 50% GC yield and 42% isolated yield with an excellent 19:1 βalkylation selectivity (6aa). Representative substrates from the different dimensions of the directing group (6ra), chain length (6qa), and an internal alkene with considerable steric hindrance (6ca) smoothly underwent the desired migratory hydroalkylation, demonstrating the generality and portability of the phosphine liganddominated β-alkylation selectivity (Fig. 4f).
DFT calculations. Based on previous literature and our experiments, we proposed two possible catalytic cycles for the proximal-selectivity hydroalkylation of unactivated alkenes (Fig. 5a) 31,[43][44][45]66,68,70 . In the left catalytic cycle, Ni(I)X (A) was generated as an initial catalyst under the reductive conditions. Subsequently, Ni(I)X (A) was associated with an alkene to form a chelated Ni(I) species (B) which reacted with silane to generate a crucial Ni(I)-H species (C). Then, the Ni(I)-H is inserted into the C-C double bond via a six-or sevenmembered-ring mode, which controls alkylation position to offer Ni(I) species D-1 or D-2. The resulting D-1 abstracted halogen from an alkyl halide to form Ni(II) and an alkyl radical. The alkyl radical was trapped to give Ni(III) species (F), which underwent rapid reductive elimination to obtain the target product and regenerate catalyst A. Another possible catalytic cycle is shown on the right. Initially, Ni(I)X (A) reacted with an alkyl halide through halide abstraction to afford Ni(II) (B') and an alkyl radical. Then, silane reacted with B' to generate the Ni(II)-H (C'), which chelated with an alkene to give Ni(II) species (D'). D' inserted into an alkene and captured the alkyl radical, followed by reductive elimination to obtain the target product (D' → E' → F). Preliminary DFT calculations indicate that the migration insertion of the Ni(II)-H bond into the terminal alkene group generating branched products via a sixmembered-ring mode (TS1 in Fig. 5b, barrier: 7.3 kcal/mol) is remarkably more facile than that of the seven-membered-ringmode generating linear products (TS2, barrier: 16.2 kcal/mol). This result correlates with the preferential hydrogen atom incorporation on the terminal carbon atom observed experimentally. By contrast, the linear migration insertion process on Ni(I) system is excluded due to its high energy demands (TS4, barrier: 25.4 kcal/mol). However, Ni(I)-catalysed sixmembered-ring migration insertion generating branched products is also possible under experimental conditions (TS3, barrier: 15.3 kcal/mol). Optimized 3D structures of TS1, TS2, TS3 and TS4 are shown in Fig. 5c. Totally, the energies for migration insertion steps in Ni(II)-H pathway are lower than that in Ni(I)-H pathway, but the possibility of Ni(I)-H pathway could not be ruled out at this stage.
In summary, we describe a nickel-catalysed proximal-selective hydroalkylation of unactivated alkenes to provide a retrosynthetic method for valuable βor γ-branched alkyl carboxylic acid and β-, γor δ-branched alkyl amine building blocks. The sensible employment of aminoquinoline and picolinamide as suitable directing groups enables the hydroalkylation of both unactivated terminal and internal alkenes. A broad range of alkyl iodides and bromides with different functional groups can be installed with excellent regiocontrol to be available for site-selective late-stage functionalization of biorelevant molecules. Under modified reaction conditions using NiCl 2 (PPh 3 ) 2 as the catalyst, migratory βselective hydroalkylation can also be realized. Mechanistic studies and DFT calculations illuminate the ligand effects on regioselectivity and the detailed reaction mechanism. Further efforts on the catalytic asymmetric variant of this transformation are ongoing in our laboratory.

Methods
General procedure for proximal-selective hydroalkylation of unactivated alkenes. In air, a 10 mL Schlenk tube equipped with a stir bar was charged with NiBr 2 (diglyme) (0.02 mmol, 10 mol%), KF (0.6 mmol, 3.0 equiv), alkene (0.2 mmol, 1.0 equiv), and alkyl halide (0.4 mmol, 2.0 equiv) (0.5 equiv NaI was added when alkyl bromide was used). The Schlenk tube was evacuated and filled with argon (three cycles). To these solids, a solution of ethanolamine (0.024 mmol, 12 mol%) in anhydrous DMAc (1.0 mL) was added under an argon atmosphere. Then, DEMS (0.6 mmol, 3.0 equiv) (if the alkene or alkyl halide was liquid, it was also added at this time) was added under an argon atmosphere. The mixture was stirred at 25°C or 40°C for 12 h, diluted with H 2 O followed by extraction with ethyl acetate, dried with anhydrous Na 2 SO 4 and concentrated in vacuo. The residue was purified by column chromatography to produce the target product.
General procedure for migratory hydroalkylation of unactivated alkenes. In air, a 10 mL Schlenk tube equipped with a stir bar was charged with NiCl 2 (PPh 3 ) 2 (0.02 mmol, 10 mol%), KF (0.6 mmol, 3.0 equiv), alkene (0.4 mmol, 2.0 equiv), and alkyl halide (0.2 mmol, 1.0 equiv). The Schlenk tube was evacuated and filled with argon (three cycles). To these solids, anhydrous DMAc (1.0 mL) was added under an argon atmosphere. Then, (MeO) 3 SiH (0.6 mmol, 3.0 equiv) (if the alkene or alkyl halide was liquid, it was also added at this time) was added under an argon atmosphere. The mixture was stirred at 100°C for 12 h, diluted with H 2 O followed by extraction with ethyl acetate, dried with anhydrous Na 2 SO 4 and concentrated in vacuo. The residue was purified by column chromatography to produce the target product.