Radicals are exceptionally reactive intermediates that readily engage with a wide array of organic molecules [
1–
4]. Their defining characteristic lies in their rapid reactivity, often occurring at diffusion-limited rates. While this property enables efficient substrate activation, it also presents challenges in achieving selective control, particularly in the context of enantioselective transformations. Among these, the carbon radical has been established as a crucial intermediate for C–C bond formation, attracting significant attention [
5–
7]. Covalent bond homolysis has been demonstrated as a versatile method for carbon radical generation (
Scheme 1a) [
8–
16]. Homolytic cleavage of a C–C bond by high-temperature thermolysis or ultraviolet irradiation has been employed to generate radical pairs [
8,
9]. Similarly, homolytic cleavage of a carbon-metal bond in alkyl metal compounds may take place, making it prone to generating transient radical species [
10–
15]. Especially when the metals exhibit spin diversity, such as Fe [
11], Co [
12], and Ni [
13,
14] in the early transition metals of the fourth period, the phenomenon becomes particularly noteworthy. The generation of transient radical species
via carbon–metal bond homolysis is highly valuable, as it can be strategically leveraged to facilitate selective and efficient radical-type transformations in conjunction with transition metal catalysis [
16,
17].
Recently, transition metals, including Ni, Cu, Rh, and Pd, catalyzed asymmetric carboxylation with carbon dioxide (CO
2) have attracted much attention as a complementary method to construct important chiral carboxylic acids (
Scheme 1b) [
18–
22]. These transformations usually involve the generation of an optically pure enantiomeric carbon-metal species as the key intermediate for CO
2 transformation and enantioselectivity control. From a theoretical perspective,
π-bond insertion of a prochiral substrate into a metal-hydride/boron bond stereospecifically generates an optically pure enantiomeric carbon-metal species efficiently [
23–
26]. In this case, substrates such as alkenes and alkynes can participate in transition-metal catalyzed asymmetric carboxylation with CO
2 pioneered by Mori [
27], Mikami [
28], Yu [
26,
29–
34], Tang, and Zhou [
35,
36] groups, by constructing alkyl- or alkenyl–metal species that can be directly transformed into various central or axial chiral carboxylic acids (
Scheme 1c).
In addition to the
π-bond insertion strategy which is effective for both alkenes and alkynes in generating carbon-metal species, an alternative approach known as oxidative addition facilitates the formation of carbon-metal species from electrophiles such as alkyl (pseudo)halides [
30,
37–
39]. Nevertheless, oxidative addition of a racemic alkyl (pseudo)halide to a low-valent transition metal usually results in the generation of a racemic carbon-metal species. One potential way to obtain optically pure enantiomeric carbon-metal species may involve the homolysis of the carbon-metal bond. The homolytic process is inherently thermodynamically endergonic, resulting in the formation of a transient radical. The transient nature and inherent reactivity of the radical facilitate the dynamic rotation along the C–C
• bond achieving an isomerization process. After recombination with metal species, another stereo-configurated carbon-metal species can be generated. This means upon generation of racemic carbon-metal species, one enantiomer may convert into another more thermodynamically stable one. We propose that applying this protocol to organo (pseudo)halides could provide chiral carboxylated products by the generation of optically pure enantiomeric carbon–metal species. Herein, we report the theoretical study on enantioconvergent carboxylation with CO
2 based on carbon-metal bond homolysis/recombination model (
Scheme 1d).
To address the above issues, we conducted theoretical investigations and chose Ni-catalyzed enantioconvergent carboxylation of alkyl ammonium salts with CO
2 reported by Li’s research group as the model reaction (
Scheme 2a) [
40]. Their experimental exploration showed that this carboxylation could be achieved in the presence of a Ni catalyst, the chiral ligand Me–SBpy, and manganese (Mn) powder under one atmosphere of CO
2. It was found that the excellent yields and enantioselectivities were almost obtained for two separate reactions whether using (
S)- or (
R)-alkyl ammonium salt
1a, indicating a racemization step before generating a chiral alkyl Ni complex. This study aims to verify the carbon-metal bond homolysis/recombination model for Ni-catalyzed enantioconvergent carboxylation with CO
2, providing direction for the rational design of Ni-catalyzed enantioconvergent carboxylation reactions, which are attractive for synthetic applications.
On the basis of literature [
40,
41] and our proposed model, we proposed the mechanistic cycle for the Ni-catalyzed enantioconvergent carboxylation of alkyl ammonium salts with CO
2, as shown in
Scheme 2b. In this process, Ni(0) complex is considered to the active catalytic species. Oxidative addition of the racemic ammonium salt
1a to Ni(0) complex
A generates benzylic Ni(Ⅱ) intermediates
B and
B’ as two enantiomers. Then, C–Ni bond homolysis-recombination occurs, facilitating the formation of a thermodynamically stable enantiomer. The resulting optically pure enantiomeric Ni(Ⅱ) complex could be reduced by Mn(0) to generate Ni(Ⅰ) species
C. This is followed by the insertion of CO
2 into the C–Ni(Ⅰ) bond, leading to the formation of Ni(Ⅰ) carboxylate intermediate
D. Further reduction then regenerates the Ni(0) species
A and releases the carboxylated product.
All structures in this study were optimized using the Gaussian 09 package [
42], employing the B3LYP-D3(BJ) [
43,
44] functional in the
N,N-dimethylformamide (DMF) solvent. The relative effective core potential (ECP) in combination with the LanL2DZ [
45–
47] basis set was employed for Ni and Mn atoms, whereas the 6–31G(d) basis set [
48] was assigned to all remaining atoms. Thermodynamic energy corrections were obtained at 298 K. Frequency analysis was conducted to confirm that the optimized geometries corresponded to a minimum (with no imaginary frequencies) or a transition state (with exactly one imaginary frequency). The solvent effect was evaluated using the self-consistent reaction field (SCRF) method and the SMD model [
49]. Intrinsic reaction coordinate (IRC) [
50] calculations were performed to verify the transition state (TS) associated with the corresponding minimum on the potential energy surface. Single-point energy calculations were carried out at the B3LYP-D3(BJ)/[SDD, 6–311+G(d,p)](SMD, DMF) level of theory [
51,
52]. Optimized molecular structures were visualized with CYLview (version 2.0) [
53]. To gain further insight into the electron properties of key species in the reaction, natural bond orbital (NBO) analysis was carried out at the same level of theory as the structural optimization [
54,
55]. The extended transition state-natural orbitals for chemical valence (ETS-NOCV) [
56–
59] analysis was performed with Multiwfn 3.8 (dev) software [
60], and the corresponding results were visualized using VMD software [
61].
A detailed density functional theory (DFT) calculation was performed to investigate the mechanism of the Ni-catalyzed enantioconvergent carboxylation, which was initiated with oxidative addition. As shown in
Fig. 1, the Ni(0) complex
1 generated by coordination with (
S)-ammonium salt
1a was considered as the zero point of the free energy profiles. Owing to the features of the Ni(0) complex and the alkyl (pseudo)halides [
62–
64], two possible pathways for the oxidative addition of C(benzyl)–N bond were proposed and investigated, including nucleophilic substitution-type C–N oxidative addition (pathway a) and single-electron transfer (pathway b) [
65]. We failed to locate three membered-ring oxidative addition transition state, likely due to steric hindrance. First, we calculated the activation free energy for the nucleophilic substitution-type C(benzyl)–N bond oxidative addition with the Ni(0) complex to deliver benzyl-Ni(Ⅱ) complex
4 (pathway a). This nucleophilic substitution step
via transition state
3-ts, featuring a triplet state at the Ni center, has an energy barrier of 5.8 kcal/mol. Geometry analysis shows that the length of C2–N, C11–Ni, C12–Ni bonds in
3-ts are 1.946, 2.588, and 2.144 Å, respectively, indicating a coordination of Ni center to naphthalene group. Pathway b involves a radicophilic Ni(0) center [
63], which undergoes single-electron transfer to form a benzyl radical through the breaking of the C(benzyl)–N bond. The relative free energy of
6-ts is calculated as 7.4 kcal/mol. Geometry analysis shows that the length of C2–N bond in
6-ts is 1.947 Å, and the spin population of Ni atom and the total benzyl group is −0.907 and 0.562, respectively. The relative free energy of transition state
3-ts is 7.6 kcal/mol lower than that of transition state
6-ts. Therefore, nucleophilic substitution-type oxidative addition
via 3-ts is the dominant pathway according to our computational results, being exoergic by 35.0 kcal/mol for the generation of (
R)-benzylic Ni(Ⅱ) intermediate
4 (Figs. S2 in Supporting information).
We then investigated how the enantiomeric carbon-metal species was formed. Our calculations using (
S)-ammonium salt
1a ligated with Ni(0) species demonstrate the generation of benzyl Ni(Ⅱ) complex
4 via configuration reversal nucleophilic substitution-type oxidative addition transition state
3-ts. In experimental observations, TEMPO-benzyl adduct and benzyl radical homocoupling side product were detected in the reaction system mechanistic studies, indicating the formation of a benzyl radical. We proposed that the homolysis of C2–Ni bond in
4 could form benzyl radical and Ni(Ⅰ) species. This process
via aryl-ring coordinated C–Ni bond homolysis transition state
10-ts requires an energy barrier of 8.0 kcal/mol, leading to the formation of Ni(Ⅰ) species
11. Geometry analysis shows that the lengths of C2–Ni, C11–Ni and C12–Ni bonds in
11 are 3.091, 2.575 and 2.153 Å, respectively, indicating an
η2-coordination of Ni center (Fig. S4c in Supporting information). Moreover, spin population analysis of
11 shows spin densities of 0.522, −0.162, 0.252 and 1.217 on the C2, C11, C12 and Ni atoms, respectively, indicating the radical character of benzyl fragment. Furthermore, we located the transition states
12-ts associated with the C–C bond rotation in intermediate
11. The energy barrier
via 12-ts is 16.5 kcal/mol, indicating that the process is kinetically feasible (Fig. S7 in Supporting information). These results mean C–C
• bond rotation can take place, establishing a pre-equilibrium between the
13 and
11 species [
66]. The endergonic process makes it difficult for the transient radical to be released, thus enabling rapid recombination and aromatization between the benzyl radical and Ni(Ⅰ) center. This step generates benzyl Ni(Ⅱ) complex
15 via aryl-ring coordinated recombination transition state
14-ts forming C–Ni bond, which has an energy barrier of 0.6 kcal/mol. Both geometry and spin population analysis reveal the radical character of benzyl fragment in
11 and
13, indicating the possible homolysis and radical recombination. Since
15 is 1.7 kcal/mol more stable than
4, this makes the process of C2–Ni bond homolysis-recombination thermodynamically favorable and benzyl Ni(Ⅱ) complex
15 is regarded as the major one in the reaction system (
Fig. 2). This benzyl radical involved transformation can also be supported by the concept of radical buffering because of the high stability of benzyl radical [
67]. By comparing the H···H distances between the substrates and ligands of intermediates
4 and
15 (
Fig. 3), we observed that the distance of H12···H24 in
4 is 2.502 Å, whereas in
15, the distance of H21···H24 is 2.941 Å. Additionally, the distance of H25···H26 in
4 is 2.231 Å, while in
15, the distance of H22···H26 is 2.713 Å. These results suggest the H–H repulsion between the substrate and chiral ligand in
4 and
15 may be responsible for stability difference.
Ni(Ⅱ) complex
15 can undergo one-electron reduction by Mn(0) (
Fig. 4). The Gibbs free energy change (∆
G) is −7.9 kcal/mol, indicating that one-electron reduction readily occurs [
68].
We also investigated the cleavage process of C2–Ni bond in Ni(Ⅰ) species
16. This process proceeds
via a 1,3-Ni migration transition state
17-ts with an activation free energy of 13.7 kcal/mol, leading to the formation of Ni(Ⅰ) species
18. Geometry analysis shows that the lengths of C2–Ni, C2–C11 and C12–Ni bonds in
18 are 3.573, 1.370 and 2.062 Å, respectively, indicating an
η1-bonding of Ni center (Fig. S5c in Supporting information). Moreover, spin population analysis of
18 shows spin densities of −0.007, 0.003, 0.000 and 0.963 on the C2, C11, C12 and Ni atoms, respectively, which suggests a negotiable radical characteristics of benzyl fragment. These computational results show a significant double bond characteristic of C2–C11 bond in
18, thus hampering the rotation of C2–C11 bond. As a result, the corresponding transition state
19-ts requires an energy barrier of 43.5 kcal/mol, suggesting an unlikely rotation process. Therefore, the stero-isomerization process involving the cleavage of C2–Ni bond in Ni(Ⅰ) species
16 followed by bond rotation would not occur (
Fig. 5).
The next step is the carboxylation of
16 with CO
2 for the formation of C–C bond, as shown in the
Fig. 6a. In this process, two pathways, including stereoretentive inner-sphere and stereoinvertive outer-sphere pathway were considered [
69]. In inner-sphere pathway, approaching of CO
2 generates an
η2-coordinated intermediate
22 with a Ni···CO
2 distance of 1.971 Å and the Wiberg bond order of 0.486, indicating a chelation-assisted CO
2 insertion process. ETS-NOCV analysis reveals a significant back-donation interaction between the
dx2-y2 orbital of Ni atom and the unoccupied
π* orbital of CO
2 (
E = −77.9 kcal/mol), effectively stabilizing the Ni–CO
2 complex and contributing to the activation of CO
2 as well (
Fig. 6b). As a result, the C5–O bond in
22 is significantly weakened, accompanied by a decrease in O–C–O bond angle from 180° in free CO
2 to 139°. One C=O bond of CO
2 which coordinates with the Ni center becomes somewhat longer to 1.249 Å than that of free CO
2 (1.169 Å). This activation effect also enhances the condensed local softness (s-) [
70] of the C5 atom in CO
2, facilitating subsequent C–C bond formation. Thereafter, carboxylation occurs
via a four-membered ring transition state
23-ts, generating Ni(Ⅰ) carboxylate species
24 with the energy barrier of 8.2 kcal/mol. We also investigated the carboxylation proceeds in an outer-sphere pathway. In this case, a molecular complex
25 might be formed, in which CO
2 has no significant interaction with metal center, suggesting a nucleophilic attack on CO
2 by the backside of the C2 atom in
16. A slight structural deformation is observed in CO
2 fragment, with the O–C–O bond angle changing to 177° and two C=O bonds of CO
2 (1.169 Å, 1.170 Å) being similar to that of free CO
2 (1.169 Å). It means CO
2 was less activated, resulting in the carboxylation
via an outer-sphere linear transition state
26-ts has a higher activation free energy (12.4 kcal/mol) than that
via inner-sphere transition state
23-ts. Therefore, the outer-sphere pathway can be excluded.
Based on the results above, a comprehensive pathway for the Ni-catalyzed enantioconvergent carboxylation with CO
2 was proposed, with the corresponding energy profiles shown in
Fig. 7. The Ni(0)-L catalyst ligated with racemic
1a, generates
1/
28. The following nucleophilic substitution-type oxidative addition takes place
via transition state
3-ts/
30-ts. After releasing NMe
3, benzyl Ni(Ⅱ) complex
4/
15 was thermodynamically favorably formed. Since
15 is 1.7 kcal/mol more stable than
4, this makes the process of C(benzyl)–Ni bond homolysis-recombination thermodynamically favorable, enabling the transformation of
4 into
15 as the major one in the reaction system. Once the benzyl Ni(Ⅱ) species
15 is generated, it can be easily reduced to low-valent Ni(Ⅰ) species
16 in the presence of excess Mn powder. Subsequent CO
2 insertion into the Ni(Ⅰ)–C(benzyl) bond of
16 affords the (
S)-isomeric product, which is consistent with experimental observation. This step
via inner-sphere transition state
23-ts requires an activation free energy of 8.2 kcal/mol. After carboxylation, reduction, dissociation of the carboxylate (
1), and the coordination of the substrate
1a occur to regenerate the Ni(0) intermediate
1.
In summary, a carbon–metal bond homolysis/recombination model for Ni-catalyzed enantioconvergent carboxylation of benzyl ammonium salt with CO2 was proposed and validated. The key step of C(benzyl)–N bond activation by Ni(0) complex is stereoinvertive nucleophilic substitution-type oxidative addition, resulting in the formation of benzyl Ni(Ⅱ) complex. Then the C–Ni(Ⅱ) bond homolysis, rotation and recombination might occur giving a more thermodynamically stable stereoisomeric benzyl Ni(Ⅱ) complex. Subsequent reduction and CO2 insertion through stereoretentive inner-sphere pathway give the chiral carboxylated product. This model achieves catalytic carboxylation of benzyl (pseudo)halides with CO2 in high enantioselectivity. The comprehensive mechanistic investigation establishes fundamental principles governing the enantioconvergent catalytic carboxylation of benzyl (pseudo)halides with CO2, highlighting the potential utility of carbon–metal bond homolysis-recombination protocols in asymmetric reductive cross-coupling reactions. Future experimental and computational studies will further expand the scope and explore the generality of this mechanism.
PDF
Full
55
32
关闭全屏
No. 86 Xueyuan South Road, Haidian District, Beijing
010-62662699(Operation and Technical Support)
qkjq@cast.org.cn
