Mechanisms of nickel-catalyzed reductive cross-coupling reactions

INTRODUCTION

Nickel-catalyzed RCC reaction, also known as cross-electrophile coupling (XEC), offers a powerful strategy for C−C bond formation, which has attracted considerable research interest^[1-8]. This protocol selectively joins two commercially available electrophiles under reducing conditions with a wide substrate scope of both coupling partners [Scheme 1A and B]. Typically, bidentate and tridentate nitrogen ligands, such as bipyridine (bpy), bioxazoline (biOx), biimidazole (biIm), bisoxazoline (box), terpyridine (terpy), and pyridine-biscarboxamidine (PyBCam), and bidentate phosphine ligands, such as Xantphos, BINAP, and so on, were usually used in this system [Scheme 1C]. In general, when nitrogen ligands are used, the reaction tends to proceed via a radical pathway, while a closed-shell reaction generally occurs with phosphine ligands. There are various mild reducing conditions available, including metal reductants such as Mn and Zn^[9,10] and organic reductants such as B₂pin₂ and hydrazine^[11,12]. In addition, photoredox catalysis^[13-17] and electrochemistry^[18-23] can also be utilized [Scheme 1D]. This strategy enables the construction of a range of C(sp²)−C(sp²)^[24-26], C(sp²)−C(sp³)^[27-31], and C(sp³)−C(sp³)^[32-34] bonds with high levels of cross-selectivity and stereoselectivity. For example, Kim et al. discovered the first nickel-catalyzed RCC of aryl chlorides with primary alkyl chlorides, utilizing a small amount of iodide or bromide in conjunction with the pyridine-2,6-bis(N-cyanocarboxamidine) (PyBCamCN) ligand and Zn reductants^[28].

Scheme 1. Nickel-catalyzed reductive cross-coupling reactions. (A) Experimental studies; (B) The scope of substrates; (C) Some common ligands; (D) Reducing conditions; (E) Several mechanistic questions.

With the continuous development of RCC reactions, some mechanistic problems have sparked intense research interest of chemists [Scheme 1E]: (1) The mechanism of RCC deviates from traditional cross-coupling (XC) reactions that typically proceed through oxidative addition (OA)/transmetalation (TM)/reductive elimination (RE) process^[35,36]; (2) Nickel intermediates likely possess various oxidation states ranging from Ni⁰ to Ni^IV; (3) Reaction appears to undergo both two-electron and single-electron redox processes; (4) The radical forms are involved in cage-escaped radicals or solvent caged radicals; (5) The origin of cross-selectivity over homo-selectivity involved in reactions. Until now, considerable achievements have been made in both experimental and computational mechanistic studies to answer these mechanistic problems^[37-44]. In this review, we focus primarily on the mechanistic aspects of RCC reactions, with an emphasis on experimental studies, such as stoichiometric, competitive, radical, kinetics, kinetic isotope effect (KIE) experiments, and so on. However, we also take into account theoretical research where appropriate. The aim of this review is to enhance the comprehension of the mechanism and selectivity of these reactions, thereby aiding chemists in designing novel catalytic systems. In the following sections, we will provide an integrated discussion of these mechanisms, including their fundamental steps, catalytic systems, detailed mechanistic studies, and the origin of cross-selectivity.

THE INVESTIGATION OF MECHANISM I

Scheme 2 depicts the scenarios of Mechanism I, which consists of five elementary steps: OA, radical addition, RE, radical formation, and reduction. This mechanism essentially proceeds through a radical chain process and involves a key Ni⁰/Ni^II/Ni^III catalytic cycle. It is worth emphasizing that radicals are generated at one nickel center, and the resulting cage-escaped radicals subsequently combine with another nickel species prior to the OA step. Additionally, when a photoredox catalyst is employed, the formation of radicals may exhibit variability and typically be independent of the catalytic cycle of nickel. A detailed discussion on this will be presented later. Finally, it should be noted that in the product release process, no reducing agents are involved. Alternatively, there is another possibility that radicals may directly combine with Ni⁰ to form Ni^I[45,46] intermediates. Xu et al. has computationally demonstrated that even though the addition of radicals to Ni⁰ is energetically more favorable than the OA of Ni⁰ to substrates aryl bromides, OA still occurs due to the low concentration of radicals and the abundance of aryl bromides^[47]. Hence, it will not be further elaborated here.

Scheme 2. The catalytic cycle of Mechanism I.

Mechanism I predominantly occurs in the Ni⁰-catalyzed reductive couplings between C(sp²) and C(sp³) electrophiles in the presence of pyridine-type nitrogen ligands and Zn or Mn reductants. The substrates for this mechanism mainly consist of unactivated simple halogenated hydrocarbons. As shown in Figure 1A, Biswas et al. conducted a series of experimental mechanistic investigations on the Ni⁰/L₁ (L₁ = 4,4'-di-tert-butyl-2,2'-bipyridine (dtbpy))-catalyzed RCC of iodobenzene 1a with iodooctadecane 1b, using Mn as a reductant^[48]. Compared to the previously published reactions^[49,50], several small modifications were made to these reaction conditions.

Figure 1. Experimental mechanistic investigations of Mechanism I. (A) Model reaction used in experimental mechanistic investigations; (B) Competition reaction studies between 1a and 1b; (C) Stoichiometric studies of IM1; (D) Radical clock experiments; (E) Radical lifetime studies. DMA: N,N-Dimethylformamide; TMSCl: Trimethylchlorosilane; ND: not detected; Conversion with respect to the amount of L₁Ni⁰(cod).

According to Mechanism I, the reaction begins with the initial OA to give Ni^II intermediate. To determine whether aryl or alkyl Ni^II intermediate was formed initially, competition reaction studies between 1a and 2a were performed [Figure 1B]. The results showed that iodobenzene reacts with Ni⁰ species 4.7 times faster than iodooctadecane, which preferentially leads to Ni^II aryl intermediate. Wotal et al. also isolated Ni^II acyl intermediates and discovered that they can react with a series of carbon electrophiles^[51]. In addition, the similar aryl nickel complex [(dtbpy)Ni^II(o-tolyl)I] was synthesized by Sheng et al.^[52]. Subsequently, stoichiometric experiment studies between isolated and characterized Ni^II aryl intermediates IM1 and 1b were conducted to further verify that IM1 acts as the initial key intermediate of the catalytic cycle [Figure 1C]. Significantly, cross-product 4c can be formed through this process without the Mn reductant, which not only confirms that Ni^II aryl species is the active intermediate of the reaction but also proves the release of the product does not require reducing agents. Moreover, radical clock experiments implied that the alkyl radical would be formed in the system [Figure 1D]. Furthermore, the findings from radical lifetime studies reveal a positive linear correlation between the 7c/8c ratio and the catalyst concentration [Ni], suggesting the involvement of a radical chain process [Figure 1E]. Additionally, they carried out organometallic experiments and excluded the possibility of an organozinc reagent acting as an intermediate. Based on all the experimental mechanistic investigations, it has been confirmed that this type of transformation [Figure 1A], which is restricted to the coupling of C(sp²) electrophiles with C(sp³) electrophiles, predominantly proceeds through a radical chain pathway.

More importantly, Biswas et al. also elucidated the origin of the cross-selectivity over homo-selectivity^[48] [Figure 2]. They found that iodobenzene PhI reacts with Ni⁰ species faster than iodooctadecane C₈H₁₇I to afford Ni^II aryl intermediate, while L₁Ni^II species may react with C₈H₁₇I rather than PhI, resulting in alkyl radicals. This sequential activation mode accounts for the experimentally observed cross-selectivity. Typically, due to the lower stability of aryl radicals compared to alkyl radicals, C₈H₁₇I is more prone to generate radicals by reacting with Ni⁰ species. In contrast, PhI is more likely to undergo OA with Ni⁰ species, which can be attributed to the favorable π-metal interactions between the substrate and the metal^[53]. Several theoretical calculations have confirmed that C(sp²) electrophiles undergo OA more rapidly than C(sp³) electrophiles. For instance, Ren et al. demonstrated that the activation energy barrier of aryl halides is 4.4 kcal/mol lower than alkyl halides^[54]. Similarly, Kumar et al. obtained identical results^[55].

Figure 2. The origin of the cross-selectivity.

It is gratifying to note that Ren et al. performed detailed density functional theory (DFT) calculations on NiBr₂/4,4'-di-methyl-2,2'-bipyridine-catalyzed RCC of aryl bromide and alkyl bromide^[54]. They found that radical addition is the rate-limiting step with the energy barrier of 10.4 kcal/mol, suggesting that the radical chain mechanism is a feasible process. In addition, Wang et al. disclosed this mechanism still operates when aryl iodides and tertiary alkyl halides are used^[56]. These two research groups also independently calculated an alternative Mechanism II (discussed below), and both mechanisms are energetically feasible, making it difficult to distinguish between them based solely on calculations. Recently, Ji et al. carried out experimental mechanistic investigations on the Ni(1,2-Dimethoxyethane (DME))Br₂/box-catalyzed enantioselective coupling of acid chlorides with α-bromobenzoates, indicating that this system also follows a radical chain mechanism^[57]. It is worth noting that some of the recently developed Ni/photoredox^[58-62] and Ni/electrochemistry^[63,64] dual catalytic RCCs also involve a radical chain process. Despite the changes in reduction conditions, the essence of the Ni-involved catalytic cycle remains unchanged, primarily influencing the formation pathway of the radicals.

Although Weix proposed a possible mechanism and suggested that radicals may be generated with the assistance of Ni^IX species, no conclusive evidence has been provided to support this claim. Building on previous studies^[65-69], there are four different pathways that can be taken, including single electron transfer, either outer-sphere or inner-sphere, two-electron OA, and concerted halogen-atom abstraction, as shown in Figure 3.

Figure 3. Several pathways for radical formation.

Thankfully, Lin et al. used electrochemical methods and DFT calculations to investigate the activation modes of radical formation in the (bpy)Ni-catalyzed system^[69]. They ruled out electron transfer and two-electron OA and revealed radical formation occurs through a halogen atom abstraction process via transition state TS1, with the energy barrier of 7.4 kcal/mol [Figure 4]. Note that, unlike the study of Weix, the radical is assisted by Ni^I aryl intermediates IM4 rather than Ni^IX in this process. Additionally, Diccianni et al. further confirmed that alkyl radicals are generated in a similar manner in the (Xantphos)Ni-catalyzed system with the aid of kinetic studies and DFT calculations^[65].

Figure 4. Concerted halogen atom abstraction process. Selected bond distances are given in Å.

Of particular note, photoredox catalysts can also aid in the generation of radicals. In a study by Wang et al., they reported a Ni/photoredox-catalyzed enantioconvergent RCC between α-bromophosphates and aryl iodides^[60] [Figure 5A]. To determine whether the radicals were generated by the photoredox catalyst or the nickel, they conducted comparison experiments using substrates 3a and 5b under conditions with photoredox and Ni catalysts, respectively [Figure 5B]. The results showed that the photoredox catalyst was responsible for generating the desired product 10c, suggesting that the radicals are formed by the photoredox catalyst in this system. The process of free radical formation is illustrated in Figure 5C. Initially, the photoredox catalyst 4CzIPN is excited by light, and the resulting excited 4CzIPN^* subsequently undergoes reductive quenching with Hantzsch ester (HEH), leading to 4CzIPN^−• with a strong reducing capacity [4CzIPN/4CzIPN^−• = -1.21 vs. saturated calomel electrode (SCE)]. Eventually, the single-electron transfer process occurs between 4CzIPN^−• and 5b to release radicals and regenerate photoredox catalyst 4CzIPN. Moreover, they also confirmed that the radical addition step is the enantioselectivity-determining step through DFT calculations. Similar results were also reported in the work by Guo et al. for the Ni/photoredox-catalyzed enantioselective three-component carboarylation of alkenes with tertiary and secondary alkyltrifluoroborates and aryl bromides^[70]. It is important to note that the photoredox catalysts assisted generation of radicals occurs independently of the Ni catalyst. Consequently, this catalytic cycle involves a Ni⁰/Ni^II/Ni^III/Ni^I/Ni⁰ sequence, in which the reduction occurs directly from Ni^I to Ni⁰, rather than the Ni^I→Ni^II→Ni⁰ pathway proposed by Weix, where Ni^I reacts with alkyl halides to generate radicals and Ni^II, followed by the reduction of Ni^II to Ni⁰.

Figure 5. Photocatalyst-assisted radical formation. (A) Model reaction of Ni/photoredox-catalyzed enantioconvergent reductive cross-couplings; (B) Competition experiments between 3a and 5b; (C) Photoredox catalyst-mediated radical formation process. SCE: saturated calomel electrode.

THE INVESTIGATION OF MECHANISM II

As shown in Scheme 3, unlike the radical chain process, Mechanism II proceeds through successive OA, reduction, OA, RE, and reduction, and it features a key Ni⁰/Ni^II/Ni^I/Ni^III process. Notably, the second OA can occur via a concert two-electron (Ni^I→Ni^III, black line) or stepwise single-electron process (Ni^I→Ni^II→Ni^III, blue line). When C(sp²) electrophiles are used, the reaction generally undergoes a two-electron OA process. On the other hand, the reaction can generate radicals through a single electron process when the C(sp³) electrophiles are present. The main differences between Mechanism II and Mechanism I can be summarized as follows: (1) Mechanism II involves twice OA and reduction and requires a reduction step before the release of the product, while Mechanism I only experiences once OA and reduction in the whole catalytic cycle; (2) Mechanism I necessarily involves the radicals, and the formation of radicals occurs prior to OA, whereas Mechanism II may or may not involve radicals. Even if radicals are involved, they are of the solvent-caged type and undergo a radical rebound process during the reaction while being cage-escaped radicals in Mechanism I.

Scheme 3. The catalytic cycle of Mechanism II.

Mechanism II is mainly observed in reductive coupling systems involving C(sp²)-C(sp²) electrophiles or C(sp²)-C(sp³) electrophiles, which are catalyzed by nickel/pyridine-type ligands in the presence of metal reductants. Generally, when C(sp³) electrophiles are involved, the system may generate radical intermediates, making it difficult to distinguish from Mechanism I merely from a computational point of view. In such cases, corresponding experimental mechanistic studies are needed to differentiate between them. As early as 2014, Jiang et al. investigated the mechanisms on (dtbpy)Ni/Zn-catalyzed RCC reaction of aryl halides from a theoretical calculation perspective^[71]. They confirmed the feasibility of the Ni⁰/Ni^II/Ni^I/Ni^III/Ni^I cycle and identified the second OA is the rate-determining step.

Later, Liu et al. reported a similar reaction of aryl halides with vinyl bromides [Figure 6A] and further explored the mechanism of this reaction by a combination of experimental and DFT calculations, offering a more in-depth understanding^[72]. Firstly, they conducted a control experiment using TEMPO (2,2,6,6-Tetramethylpiperidin-1-oxyl) to demonstrate that the reaction does not involve a radical chain process and that the participation of vinyl radicals is less likely. Next, a possible mechanism was determined through calculations, as shown in Figure 6B. The results revealed that the Ni⁰ catalyst initially undergoes an OA process with vinyl bromides through the concerted two-electron, three-member transition state TS2, resulting in the formation of the vinyl Ni^II intermediate IM7. This intermediate is subsequently reduced by Zn to form the vinyl Ni^I intermediate IM8. The second OA involving bromobenzene then occurs via the transition state TS3, generating the Ni^III intermediate IM10. This process has an activation energy of 14.2 kcal/mol and constitutes the rate-determining step of the reaction.

Figure 6. Computational mechanistic studies. (A) Model reaction used for DFT calculations; (B) Energy profile for oxidative addition process. Gibbs free energies are given in kcal/mol. Computations at the B3LYP-D3(BJ)/SDD for Ni, 6-311G(d,p) for Br, 6-31G(d,p) level. Selected bond distances are given in Å.

Thereafter, Long et al. reported a similar mechanism for a reductive coupling of 2-Haloanilines^[73]. They also discovered that the energy required for single-electron halide abstraction is less favorable than that of the two-electron OA process by 15.4 kcal/mol. This further implies that the formation of aryl radicals in this system is quite challenging. Recently, Long et al. developed a (bpy)Ni/Mn-catalyzed reductive coupling of 2-halophenol derivatives^[74]. They successfully synthesized the Ni^II aryl intermediate through the reaction of Ni⁰ and aryl halides. Moreover, this intermediate was found to efficiently catalyze the reaction, leading to the desired coupling products. This observation suggests that the OA of Ni⁰ to aryl halides takes place in the catalytic cycle, and the resulting Ni^II aryl intermediate might be a key active species. All of the aforementioned findings indicate that the Ni⁰/Ni^II/Ni^I/Ni^III process is possible for Ni/(pyridine-type ligands)-catalyzed reductive couplings involving C(sp²) and C(sp²) electrophiles using Zn or Mn reductants.

According to Mechanism II, when involving C(sp³) electrophiles, a stepwise single-electron OA process can occur. Ren et al. revealed that the stepwise single-electron OA process between the Ni^I aryl intermediate IM11 and CyBr 7b is completed through the successive halogen atom abstraction transition state TS4 and the radical addition transition state TS5^[54] [Figure 7]. This process forms the Ni^III intermediate IM12, requiring relatively low activation energy barriers of 5.0 and 10.4 kcal/mol, respectively. A similar result was obtained for nickel-catalyzed reductive XC of activated primary amines with aryl halides, as reported by Yue et al.^[75].

Figure 7. Stepwise single-electron oxidative addition process. Computations at the B3LYP/6-31g* level. Selected bond distances are given in Å.

THE INVESTIGATION OF MECHANISM III

In general, a fundamental question in these catalytic systems is identifying the nature of the active catalyst at the outset of the reaction. Mechanism III is distinct from the previously discussed two mechanisms, as it is characterized by the absence of Ni⁰ throughout the entire process, with Ni^I species serving as the active catalyst. As illustrated in Scheme 4, starting from the active catalyst LNi^IX, the reaction can proceed through two distinct OA pathways. The first possibility involves a bimolecular OA to form LNi^IIX(R¹), while the second involves a single-molecule process that yields LNi^III(X)₂(R¹). These two intermediates are then reduced to LNi^I(R¹), followed by halogen atom abstraction to form R² radical and LNi^IIX(R¹), which, upon radical rebound, produce LNi^III(X)(R¹)(R²). Finally, RE occurs, leading to the formation of the product and regeneration of the active catalyst. Generally speaking, in the reductive coupling system involving C(sp²) and C(sp³) electrophiles, catalyzed by Ni catalyst and pyridine-type nitrogen ligands, it has been observed that when the substrates change to alkyl halides featuring more complex structures, such as alkenes and heteroatoms, the reaction mechanism tends to shift towards Mechanism III.

Scheme 4. The catalytic cycle of Mechanism III.

This mechanism was corroborated through various experimental mechanistic investigations, including kinetic, spectroscopic, and organometallic studies, conducted by Lin et al.^[76]. They used the model reaction depicted in Figure 8A, where aryl bromide 5a and alkyl bromide 8b can be effectively cross-coupled to generate product 12c in the presence of Zn and NiBr₂·DME/1,10-phenanthroline. Firstly, they carried out substrate probe experiments to demonstrate the presence of radical intermediates in the system. Subsequently, the kinetic studies were performed, and the reaction order was obtained, as illustrated in Figure 8B. The zero-order dependence on substrates 5a and 8b, together with the first-order dependence on [Ni], and the observation that the reaction rate increases along with increasing agitation rate and Zn loading. Integrating these results suggests that the reduction of Ni by Zn is the rate-determining step of the reaction. After that, they identified Ni^II intermediates IM13 and IM14 as the catalyst resting state through EPR, ¹H NMR, and UV-visible spectroscopy analysis [Figure 8C]. In contrast to the previous system where Ni⁰ served as the reducing species, the resting state IM13 can only be reduced by Zn to Ni^I species IM15, which was confirmed by comparing its cyclic voltammetry (CV) and EPR data with those of isolated and X-ray characterized (phen^*)Ni^IBr IM16 (phen^* = 2, 9-di-sec-butyl-phenanthroline) [Figure 8D]. Therefore, Ni^IBr IM15 serves as the starting point for the reaction. Several reports have also demonstrated that the Ni^IX complex serves as the initial active catalyst in nickel-catalyzed reductive coupling reactions^[77-80].

Figure 8. Experimental mechanistic investigations of Mechanism III. (A) Model reaction used for mechanistic investigations; (B) Kinetic studies of model reaction; (C) Spectroscopic studies; (D) Reduction of Ni^II by Zn; (E) Competition experiments between 5a and 8b; (F) Stoichiometric experiments; (G) Control experiments. ^a and ^b: refer to conversion and yields, respectively; DMA: N, N-dimethylacetamide, DME: 1,2-Dimethoxyethane.

Subsequently, competition experiments were conducted to explore the electrophile activation of substrates [Figure 8E]. The reaction rates of 5a and 8b with IM15 were 1.2 and 0.42 mM/min, respectively, indicating that the OA of Ni^IBr with PhBr is faster than its reaction with haloalkanes to generate radicals. They speculate that the OA process takes precedence over radical formation. These findings contradict the results proposed by the radical chain mechanism, which suggests that Ni^IBr preferentially activates alkyl halides to generate radicals, occurring prior to OA. It is important to note that Ting et al. further demonstrated that aryl halides undergo OA with Ni^I halide complexes through quantitative experiments, in which they obtained structurally characterized bipyridine-ligated Ni^III aryl complexes^[81]. In addition, Till et al. discovered that (dtbpy)Ni^IBr can be generated through pulse radiolysis, and it is capable of undergoing OA with aryl halides^[82]. Then, Breitenfeld et al. performed stoichiometric experiments to demonstrate that 6ao-TolBr reacts with IM15 via bimolecular OA, leading to IM13 and IM17^[83,84] [Figure 8F]. EPR spectroscopy reveals that IM17 can also only be reduced by Zn to form Ni^I aryl species. Similar structures of (IPr)Ni^I alkyl and aryl complexes have been characterized by Laskowski et al., and they confirmed that Ni^I alkyl species can activate alkyl halides via a radical process^[85]. In order to identify how Ni^II aryl intermediate interacts with substrate 8b, an array of control experiments was carried out [Figure 8G] . Stoichiometric experiments have shown that the efficiency of the reaction is low in the absence of Zn, with only 30% of 8b being converted (entry 1). However, with the addition of Zn, the conversion of 8b increases to 100% (entry 2), suggesting that Zn plays a crucial role in promoting the reaction. This finding contrasts with the results reported by Weix, where the reaction could efficiently yield the desired product even without Zn. Moreover, the formation of 14c infers the existence of the radical intermediate that undergoes cyclization and dimerization, while the cross-coupled product 15c is obtained through a radical addition and reduction elimination process. They also found that when the substituent Ar is o-Tol, both 15c and 14c can be obtained (entry 2). When the substituent is Mes, only dimer 14c can be formed (entry 3). This suggests that the larger steric hindrance prevents radical addition to Ni center. Taken together, these results are consistent with Mechanism III.

By comparing the catalytic systems of Weix and Diao, significant differences were found. (1) The types of electrophilic substrates used are different: Weix employed simple linear alkyl halides such as iodobenzene, while Diao used structurally complex alkyl halides substituted with heteroatoms and olefins; (2) The ligands used are different: although both ligands belong to the pyridine-type nitrogen ligands, they have different substitution patterns, one being a bipyridine and the other being a phenanthroline; (3) The reducing agents used are different: Weix used Mn, while Diao used Zn. These differences in catalytic systems are important factors that lead to changes in the reaction mechanism. More importantly, Ju et al. discovered that the use of biOx ligand excludes the reduction step of (biOx)Ni^II[86]. This can be attributed to the lack of ligand redox activity, resulting in more negative reduction potentials of (biOx)Ni^II complexes, rendering them unable to be reduced by Zn and Mn. This further highlights the significant impact of ligands on the mechanism^[87].

Similar to Weix, Diao found that the cross-selectivity also originates from the different activation sequences of the two electrophiles. Unlike the Weix’s system, which utilizes Ni⁰ and Ni^II species, Diao employs Ni^IBr and Ni^IPh to activate different electrophiles, respectively. Specifically, the OA is mainly influenced by steric effects, while the formation of radicals via halogen-atom abstraction is related to electronic effects. Therefore, sterically assessable Ni^IBr preferentially undergoes two-electron OA with PhBr 5a to give C(sp²) Ni^II species, while electron-rich but sterically hindered Ni^IPh predominantly activates alkyl bromides 8b via halogen-atom abstraction to forms C(sp³) radicals [Figure 9], thus resulting in cross-selectivity.

Figure 9. The origin of the cross-selectivity.

Meanwhile, by combining experimental and DFT calculations, Shu et al. investigated the mechanism of dipyridine-ligated nickel-catalyzed reductive dicarbofunctionalization of propene with tert-butyl iodide and iodobenzene with the use of Zn reductants^[88]. Their findings confirmed the feasibility of the pathway involving Ni^I species. More recently, Zhu et al. developed an RCC reaction of α-oxy halides enabled by Mn reductants, photocatalysis, electrocatalysis, or mechanochemistry in the presence of nickel and phenanthroline ligands^[89]. Surprisingly, through detailed experimental and theoretical studies, they found the mechanisms of all four catalytic systems are consistent with Mechanism III. Besides, they noted that the Ni^III intermediate, obtained through OA, may trigger comproportionation with Ni^I species to afford the Ni^II intermediate. Intriguingly, Day et al. disclosed that polypyridine-ligated Ni^II halide complexes can undergo the comproportionation with Ni⁰ to form Ni^I species^[90]. These electron-transfer events were corroborated by electrochemical techniques and detailed quantum mechanical calculations. It is worth noting that Tang et al. conducted comprehensive mechanistic studies on the OA of Ni(I) to aryl iodides using electroanalytical and statistical modeling techniques^[91]. And predicted OA rates can be utilized not only to interpret observed reactivities but also to rationalize the mechanism. This indicates that an expanding array of techniques can assist us in predicting reaction models, understanding reaction mechanisms, and guiding the design of novel catalytic systems.

THE INVESTIGATION OF MECHANISM IV

In contrast to the aforementioned initial three mechanisms, Mechanism IV fundamentally encompasses an S_N2 process and the Ni⁰/Ni^II catalytic cycle. As depicted in Scheme 5, commencing with Ni⁰, the reaction proceeds sequentially through OA, TM, an S_N2 reaction, a second TM, and ultimately RE. Notably, this particular mechanism employs organic Grignard reagents as reductants, as opposed to their conventional use as coupling reagents, thereby deviating significantly from traditional XC reactions. Mechanism IV is chiefly observed in the intramolecular reductive coupling reaction system, which is facilitated by Ni⁰ catalysts in the presence of phosphine ligands and organic Grignard reagents. The primary substrates for this mechanism encompass a range of active pyrrole and amine derivatives. It is important to note that Mechanism IV demonstrates fundamental differences in terms of reaction conditions and substrates when contrasted with the previously discussed three mechanisms.

Scheme 5. The catalytic cycle of Mechanism IV.

Chen et al. conducted a comprehensive mechanistic investigation on Ni⁰/BINAP-catalyzed stereospecific intramolecular RCC reactions of benzylic ethers through a combination of experimental and computational approaches, thereby confirming the feasibility of this particular mechanism^[92]. In this discussion, we concentrate on the key transition states for the product formation, as illustrated in Figure 10A. The reaction proceeds sequentially through the OA transition state TS6, TM transition state TS7 and intramolecular S_N2-like attack transition state TS8 to achieve C−O bond cleavage, alkyl transfer, and C−Cl bond cleavage, ultimately leading to the product 16c. The energy barriers associated with these steps are 21.9, 12.0, and 9.3 kcal/mol, respectively. This process takes place under a closed-shell system without involving radicals and exhibits stereospecificity. Among these steps, the OA of the C−O bond occurs with stereoinversion of the benzylic stereogenic center, which is facilitated by the assistance of Grignard reagents, constituting the rate- and selectivity-determining step for product formation. Concurrently, they performed a ¹³C KIE study of the benzylic ether of 2-naphthyltetrahydropyran 9b, which revealed a significant (> 1.01) KIE numerical distribution for benzylic C11 compared to other atoms [Figure 10B]. This further substantiated that the activation of the C−O bond is the rate-determining step.

Figure 10. Experimental and computational mechanistic studies. (A) DFT calculations for key transition states; (B) ¹³C KIE experiments and KIE numerical distribution of 9b. Computations at the B3LYP-D3(BJ)-SMD/def2-TZVPP//B3LYP-D3(BJ)/def2-SVP level. Selected bond distances are given in Å. The hydrogen atom was omitted for simplification.

Moreover, the activation of both the C−O and C−Cl bonds collectively control the stereospecificity of overall reactions.

They also discovered that the intramolecular reductive coupling reaction of halogenated sulfonylamine derivatives, catalyzed by air-stable ((R)-BINAP)NiCl₂ and MeMgI organic Grignard reagents, still adheres to the Ni⁰/Ni^II catalytic cycle^[93,94]. Furthermore, Xu et al. conducted a detailed theoretical study on the Ni⁰/XantPhos-catalyzed intramolecular reductive coupling of tetrahydropyrans, indicating that Mechanism IV is applicable to this system^[95]. Similarly, nickel-catalyzed intramolecular reductive coupling of difluoromethyl moiety and benzylic ether, reported by Lucas et al., also follows a comparable reaction mechanism^[96]. It is noteworthy that Sanford et al. found that when utilizing non-cyclic 1,3-diol derivatives featuring two C(sp³)−I bonds for intramolecular reductive coupling, the activation of C(sp³)−I bonds proceeds via a radical pathway rather than the S_N2 mechanism^[97]. This finding highlights the notion that the intrinsic characteristics of the substrates can indeed exert a substantial impact on the reaction mechanism.

CONCLUSION AND OUTLOOK

In summary, nickel-catalyzed reductive coupling reactions exhibit considerable diversity in their mechanisms, distinct from those catalyzed by palladium and platinum. This can be ascribed to the unique characteristics of nickel, including high paring energy, low electronegativity and redox potential, and multiple oxidation states (0, I, II, III, IV)^[35]. These enable nickel catalysts to preferentially undergo both two- and one-electron redox processes, leading to comparatively diverse mechanistic scenarios. Additionally, these mechanisms are closely related to substrates, ligands, and reducing conditions, making it challenging to discern a unifying pattern. From the four potential reaction mechanisms summarized, we can identify some basic trends: (1) C(sp²) electrophiles tend to undergo two-electron OA, while C(sp³) electrophiles prefer a single-electron pathway initiated by halogen atom transfer; (2) When nitrogen ligands are employed and C(sp³) electrophiles are involved, the reaction is inclined to proceed via a radical pathway, whereas a closed-shell reaction generally occurs with phosphine ligands; (3) Reaction systems involving photoredox catalysts or electrocatalysis typically undergo a single-electron transfer process. However, determining the conceivable reaction mechanism requires considering various reaction conditions, such as the choice of ligands and substrates, and whether the reducing system involves conventional metal and organic reducing agents or emerging photoredox and electrocatalysis. It is particularly important to ascertain whether a Ni^II catalyst precursor is ultimately reduced to Ni⁰ or Ni^I, as this determines the active catalyst in the reaction.

As for C(sp²)−C(sp³) XECs, the origin of the cross-selectivity over homo-selectivity can be ascribed to the different activation sequences of two electrophiles. Notably, different systems use different Ni species to activate electrophiles. For instance, in the Ni/dtbpy-catalyzed system for the XEC of iodobenzene with iodooctadecane, with Mn as the reducing agent, Ni⁰ and Ni^II are always used to activate the electrophiles. In contrast, Ni^IBr and Ni^IPh species can activate electrophiles when using more complex alkyl bromide containing olefins and heteroatoms substrate, Zn reductants, and 1,10-phenanthroline ligands. Generally speaking, due to the favorable π-metal interactions and the instability of aryl radicals, C(sp²) electrophiles tend to proceed through OA, while C(sp³) electrophiles are more likely to generate radicals. Additionally, OA and radical formation processes are also influenced by steric hindrance and electronic effects, respectively.

Despite significant progress made by both experimental and theoretical studies in characterizing the structures of key intermediates, providing reaction rates, identifying the activation modes of electrophiles, determining rate- or selectivity-determining steps, and identifying the origin of cross-selectivity, several limitations and challenges in mechanistic studies remain to be addressed: (1) There are limited studies on the mechanism of RCC reactions, especially those involving photoredox and electrochemistry catalysis; (2) The electron transfer process involved in the reduction process is still unclear; (3) So far, there is no data-driven and artificial intelligence linkages to aid further mechanistic exploration and reaction prediction. Therefore, the mechanistic investigation of nickel-catalyzed reductive couplings is far from complete, and concerted endeavors of experimental and computational studies are highly demanded to help chemists design more powerful and novel catalytic systems.