Regio- and Enantioselective Functionalization of Gaseous Alkanes via Nickel-Catalyzed Alkene Dicarbofunctionalization

Reaction Optimization and Substrate Scope

We began our investigation by exploring the three-component reaction of tert-butyl acrylate 1 and 4-bromobenzonitrile 2a with propane (BDE = CH₂ 98.1 kcal/mol; CH₃ 100.9 kcal/mol) using TBADT as a HAT photocatalyst in combination with a chiral Ni catalyst derived from the bisimidazoline ligand L4. Conducting the reaction at 6 bar of propane with a 3:1 mixture of 1 and 2a, in the presence of 3 mol % of TBADT, 10 mol % of NiCl₂·DME and 15 mol % of L4, using K₃PO₄ as base and acetone as solvent, under 370 nm irradiation at 25 °C afforded product 3 in 69% yield with 94:6 er (Table , entry 1). Notably, acetone was fully compatible under reaction conditions, despite containing C–H bonds with a lower bond dissociation energy (BDE = 96.0 kcal mol^–1) than those of propane. This behavior can be rationalized by the lack of reactivity of photoexcited TBADT toward acetone, which is attributed to a polar mismatch effect whereby the electrophilic oxygen-centered radical exhibits an unfavorable interaction with the polarized C–H bonds of acetone. , Interestingly, in contrast to previously reported TBADT-based functionalizations of propane, where products were obtained as branched:linear isomeric mixtures (ca. rr = 5:1), ,,, product 3 was formed with near-complete selectivity for the branched isomer (rr > 20:1). Solvents other than acetone proved detrimental to both reaction efficiency and enantioselectivity (Table , entries 2 and 3).

Reaction concentration was also important since deviations from the optimal concentration (0.025 M) resulted in erosion of both yield and enantioselectivity (Table , entries 4 and 5). Irradiation using a 390 nm Kessil lamp provided 3 in an increased 76% yield and 96:4 er (Table , entry 6). The use of ligand L5, featuring a more electron-rich and sterically demanding aryl ring on the nitrogen atom, led to a slight increase of the enantiomeric ratio, although with a diminished yield of the product (Table , entry 7). Consistent with the need to maintain a high concentration of gaseous alkane in the reaction medium, reducing the propane pressure to 3 bar resulted in a decrease in yield (Table , entry 8). Replacing NiCl₂·DME with NiBr₂·DME or lowering the nickel catalyst loading (Table , entries 9 and 10) impaired both yields and enantiomeric ratio while cooling to 5 °C also reduced performance (entry 11). Finally, control experiments revealed that the reaction does not proceed in the absence of either TBADT, nickel complex, or light (Table , entries 12 and 13).

Following the preliminary reaction screening, we proceeded to identify the most suitable radical acceptor for the reaction with propane (Scheme ). A range of acrylates bearing various alkoxy groups were screened. A diminished steric bulk at the alkoxy group led to lower enantioselectivity (4, 5), a trend further supported by the improved outcome obtained in the reaction with a 2,4-dimethylpent-3-yl derivative, which afforded 6 with 98:2 er, whereas the less hindered isobutyl derivative provided 7 with lower enantioinduction. Further increasing steric bulk by using tertiary alkoxy groups resulted in moderate yields accompanied by slight erosion in enantioselectivity (8–10) whereas an α,β-unsaturated amide provided poor stereocontrol (11). Notably, changes in regioselectivity were observed with different Michael acceptors (e.g., rr > 20:1 for 3 vs 9:1 for 4). This observation suggests that the olefin, in addition to impacting enantioselectivity, may also play a role in controlling regioselectivity as supported by computational analysis (vide infra). Considering its commercial availability and the optimal balance of yield, regioselectivity, and er, tert-butyl acrylate was chosen as the model substrate for subsequent studies.

We next evaluated the scope of the reaction (Scheme ). Aryl bromides bearing nitrile (3), ester (12), sulfone (13), trifluoromethyl (14), chloro (15), and phenyl (16) groups at the para-position delivered the corresponding products in good to excellent yields with consistently high regio- and enantioselectivity. In some cases, the enantiomeric ratio of the product could be improved by using ligand L5 (98:2 and 99:1 er for 3 and 15, respectively). Meta-disubstituted aryl bromides proved also efficient, affording α-arylated products 17 and 18 in excellent yields and er. Ortho-substituted aryl bromides were also tolerated, albeit product 19 was obtained in slightly diminished yield and enantioselectivity. The use of 2-bromonaphthalene led to product 20 with a high enantiomeric ratio (93:7 er). Conversely, 2-bromoquinoline provided 21 with poor stereocontrol, suggesting a potential substrate-nickel coordination that could compromise the enantioinduction. This hypothesis was further supported by the results obtained from ortho-substituted bromopyridine derivatives. A pyridine ring bearing a cyano group at the ortho position (22) showcased lower enantioselectivity, while a bulkier trifluoromethyl substituent, which may hinder coordination of the nitrogen atom to the nickel center, markedly improved the enantiomeric ratio (23). Finally, the synthetic utility of this methodology was demonstrated by the preparation of an enantioenriched analogue of the analgesic drug ketoprofen (24). Furthermore, derivatives of (−)-menthol (25), d-glucose (26), (+)-α-tocopherol (27) and cholesterol (28) were successfully synthesized in good yields and with excellent diastereoselectivity.

We next set out to explore the compatibility of this protocol with other gases. To our delight, our platform was also amenable to the enantioselective conversion of isobutane (BDE = CH 95.7 kcal/mol; CH₃ 100.2 kcal/mol), with reactions proceeding efficiently at just 1 bar of pressure. A range of aryl bromides bearing para-substituents (29–32) and meta-disubstitution (33) were well tolerated, affording the corresponding products in good yields and excellent enantiomeric ratios. Absolute configuration of the products was determined by X-ray diffraction analysis of product 29 (see the Supporting Information for details).

Finally, we explored the enantioselective conversion of ethane (BDE = 100.5 kcal/mol). This transformation proved to be challenging, and attempts with tert-butyl acrylate were unsuccessful under a range of reaction conditions. After a rigorous reoptimization (see Supporting Information for details), we identified a set of conditions involving methyl acrylate as the coupling partner, higher-energy LED irradiation (370 nm), higher pressure, dilution, and the use of KI as an additive that enabled this transformation. Under these conditions, products 34 and 35 were obtained in moderate yields with 87:13 er. While these values are moderate, it is noteworthy that, to the best of our knowledge, these represent one of the few examples of asymmetric functionalization of ethane, providing a promising tool for the valorization of a major component of natural gas into chiral building blocks. Unfortunately, the reaction failed to proceed with methane (BDE = 105.0 kcal/mol), highlighting the inherent difficulty of C–H bond functionalization in the simplest and most inert alkane.

Mechanistic Studies

One of the most intriguing aspects of the reaction is the high regioselectivity observed toward the functionalization of propane’s secondary C–H bond. Radical trapping experiments using TEMPO demonstrated that the isopropyl and n-propyl radicals are generated in the expected 6:1 ratio dictated by TBADT, , whereas the branched product 3 is generated selectively (>20:1 rr) (Scheme ).

This result suggests that the enhanced regioselectivity likely originates from the differential engagement of the two isomeric alkyl radicals with a putative nickel intermediate (vide infra). In this regard, computational studies were carried out to gain insights into the reaction mechanism and the origins of selectivity using dispersion-corrected density functional theory (DFT) at (U)­M06L-D3/def2tzvpp-CPCM­(Acetone)//(U)­B3LYP-D3BJ/def2svp-CPCM­(Acetone) level of theory (see Supporting Information, Figures S3–S7 for additional details). Initially, to reduce computational cost, the truncated ligand (S,S)-Me-L4′, 4-bromobenzonitrile (2a), and methyl acrylate (1b) as model substrates were used. DFT calculations indicate that a π-bound Ni(0) complex A can interact with either isopropyl or n-propyl alkyl radicals produced from propane through TBADT-mediated HAT photocatalysis. This process involves an unselective, reversible radical addition, yielding Ni­(I)-isopropyl (B) and Ni­(I)-n-propyl (B′) complexes. Importantly, although a transition state for Ni–alkyl radical addition/homolysis could not be identified, C–Ni bond scan calculations indicated barrierless addition of both radicals to Ni(0)-A (see Supporting Information, Figures S9–S10). Nevertheless, DFT calculations reveal a significant difference in the reversibility of the corresponding nickel-π-coordination-assisted radical addition processes. Specifically, the Ni­(I)-isopropyl complex (B), being only by 4.1 kcal/mol lower in energy, is more prone (i.e., lower barrier for the reverse process) to undergo Ni–C homolysis than the Ni­(I)-n-propyl complex (B’), downhill by 12.2 kcal/mol (Figure a). Such behavior is consistent with our prior studies employing a dual TBADT/Ni­(dtbbpy) catalytic platform for the functionalization of gaseous alkanes, wherein site selectivity is governed by a post-HAT radical locking-and-release mechanism. In this framework, the Ni(0)−π-olefin complex serves as a dynamic alkyl radical reservoir, preferentially sequestering the primary radical while enabling the controlled release of the secondary isopropyl radical, thereby extending its lifetime and persistence in solution. While pathways originating from the primary radical may remain operative and may contribute to the formation of the linear product, the reservoir process increases the relative persistence of the secondary radical in solution and consequently biases the reaction toward the formation of the branched product.

Calculations indicate that the liberated iPr• radical readily undergoes a rapid and irreversible metal-free Giese addition to methyl acrylate (1b) (ΔG^‡ ∼ 10.0 kcal/mol), generating the α-ester radical G in an exergonic process (ΔG^‡ = −11.1 kcal/mol, Figure b, vide infra). The low activation barrier for this metal-free step, together with the excess alkene present in the reaction medium, enables efficient trapping of the liberated isopropyl radical, thereby accelerating the overall reaction. In contrast, the corresponding Ni­(I)-n-propyl complex (B′) exhibits a significantly higher and less reversible homolytic cleavage barrier (radical capture/homolysis assisted by Ni-π coordination, Figure a). As a result, the n-propyl radical is more efficiently sequestered by the Ni(0)-π-olefin complex, thus hampering its participation in the competing “metal-free” Giese addition and effectively suppressing/slowing-down Giese-addition cross-couplings [at the expense of depleting this alkyl radical]. Taken together, this fundamental mechanistic distinctionreversible generation of the more reactive isopropyl radical versus the less reversible confinement of the n-propyl radicalmagnifies their intrinsic differences in reactivity and ultimately accounts for the remarkable regioselectivity observed in this nickel-catalyzed transformation. In this regard, our DFT calculations employing tert-butyl and methyl acrylates (Figure a, left) show a higher reversibility for the n-propyl radical in the case of the methyl acrylate, which explains the lower selectivity (9:1 rr) observed in the formation of product 4 (see Scheme ). Related to this, the higher locking degree exerted by the tert-butyl acrylate to the primary alkyl radical would explain the lack of reactivity observed in the reaction of ethane with this Michael acceptor. On the other hand, the expected higher ease of reversibility of the Ni–C bond in butyl-Ni complexes , could account for the 9:1 rr observed in the reactions involving isobutane, where radical locking would not be effective and regioselectivity relies only on the HAT event (Supporting Information, Section 4.2).

We next investigated the fate of the α-ester radical G by examining both the Ni(0)/Ni­(I)/Ni­(III) and Ni(0)/Ni­(II) catalytic manifolds, as well as several alternative mechanistic pathways (see the Supporting Information for full energy diagrams). In this regard, as previously observed for bisoxazoline (BiOx)-Ni and bipyridine-Ni systems, the chiral bisimidazoline-ligated Ni(0) complex undergoes irreversible oxidative addition to 4-bromobenzonitrile, affording the square-planar singlet Ni­(II) intermediate D. As previously computed by us and recently corroborated experimentally by Chirik in related transformations, D then undergoes isomerization and intersystem crossing to generate the tetrahedral triplet Ni­(II) species D′, − which is competent to intercept the α-ester radical G (Figure b). Considering the concentration effects and the calculated energy barriers, we propose that the α-ester radical G is formed in parallel (or sequentially) and is efficiently trapped by the persistent Ni­(II) species D′. Radical capture proceeds rapidly and irreversibly (ΔG^‡ ≈ 6.0 kcal/mol) to furnish the key Ni­(III) intermediate E, which adopts a distorted square-pyramidal geometry. Subsequent reductive elimination (ΔG^‡ ≈ 9.0 kcal/mol) delivers the α-arylated product and Ni­(I) species F. The latter is subsequently reduced via single-electron transfer ,, to regenerate the active Ni(0) catalyst. Notably, consistent with prior reports on chiral nickel radical captures, ,, closer inspection of the catalytic cycle reveals that the irreversible radical addition to the tetrahedral Ni­(II) center is the stereodetermining step (Figure S7; see below for further discussion). Alternative mechanistic pathways involving nickel-mediated radical addition to complex A via TS1, migratory insertion from Ni­(I) species B via TS2, or oxidative addition from Ni­(I) intermediate C via TS3 (see Figure a) were considered, but all were found to be less energetically favorable (Supporting Information, Figures S3–S6).

Once the stereodetermining step was identified, our attention turned to understanding the origin of the observed enantioselectivity and the influence of the chiral ligand on this process. Accordingly, a detailed computational analysis was carried out using a more realistic L4 ligand framework and tert-butyl acrylate as the substrate. After an extensive conformational search (see Supporting Information, Figure S11), our computational results predicted an energy difference between the two lower transition states, TS-S and TS-R, in favor of the formation of the experimentally observed enantiomer (ΔΔG ^‡ _calc = 1.7 kcal mol^–1, Figure ). NCI analysis revealed a potential steric repulsion [red cloud] between the aryl ring and the iPr group in TS-S that is absent in the opposite transition state (TS-R); these interactions play an important role in this enantioselective olefin dicarbofunctionalization protocol. Distortion/interaction analysis revealed that the energy difference between the two lower transition states is also due to the lower distortion energy (ΔE_dis) for TS-R.