Probing the Oxidative Addition Transition State in Nickel Metallaphotoredox Catalysis Using Intramolecular ¹³C Kinetic Isotope Effects

Abstract

The merger of nickel and photoredox catalysis has enabled a wide range of C–C and C–X bond formations under mild conditions. Central to the mechanistic understanding of these catalytic reactions is the identification of the nickel species involved in the key oxidative addition step. Intramolecular ¹³C kinetic isotope effects (KIEs) in conjunction with DFT analysis are a reliable, quantitative approach to identify the oxidation state and ligand environment at the metal center during oxidative addition. This methodology is applied to evaluate the mechanism of two distinct nickel metallaphotoredox-catalyzed reactions. Specifically, we probe the oxidative addition step of nickel metallaphotoredox catalyzed C(sp²)–C(sp³) cross-coupling reactions that utilize either neutral bipyridyl ligands (NiCl₂(dtbbpy)(H₂O)₄) or anionic diketonate ligands (Ni(THMD)₂) and utilize this information to elucidate the operational catalytic cycles in these reactions. In the N-methyl selective arylation of trialkyl amines catalyzed by NiCl₂(dtbbpy)(H₂O)₄, a Ni(I)(dtbbpy)(α-aminoalkyl) complex is invoked as the most likely species involved in the oxidative addition step. An alternative mechanism is also identified that is consistent with the experimental KIEs, involving oxidative addition to a Ni(0)(dtbbpy)(methyleneiminium) bromide complex. In the deaminative arylation of sterically hindered primary amines catalyzed by Ni(THMD)₂, our results indicate that an anionic Ni(I)(THMD)(Cl) complex is likely involved in the oxidative addition step, a finding that accounts for the rate acceleration observed in the presence of tetrabutyl ammonium chloride as an additive. These results illustrate that intramolecular ¹³C KIEs are an effective experimental probe to address challenging mechanistic questions in nickel metallaphotoredox catalysis.

Graphical Abstract

INTRODUCTION

Over the past decade, the merger of photoredox and nickel catalysis has enabled a broad range of novel transformations under relatively mild conditions.¹ The ability of nickel to access a variety of oxidation states (Ni⁰, Ni^I, Ni^II, Ni^III, and Ni^IV) and redox manifolds imparts unique reactivity that is not available to the more widely used palladium-based catalytic systems. On the flip side, these attributes make the mechanistic understanding of nickel-catalyzed reactions challenging.² The merger with photoredox catalysis and the associated single-electron transfer (SET) events further complicate the elucidation of key details of these catalytic cycles. Computational studies can provide valuable insight; however, the possibility of low- and high-spin states of putative transition states and catalytic intermediates significantly increases the cost and limits the widespread application of computational approaches for the mechanistic elucidation of these reactions. Moreover, energetic comparisons across spin states and catalytic species that are not isoelectronic are complicated.

An important subset of reactions in dual Ni-photoredox catalysis is the cross-coupling of aryl halides with photoredox-generated alkyl radicals to make new C(sp²)–C(sp³) bonds.^3,4 A key elementary step in these transformations is the oxidative addition (OA) of aryl halides to the nickel center. The different redox manifolds accessible in dual nickel-photoredox catalysis make the mechanistic evaluation of this important step challenging (Scheme 1A). Does the photoredox-generated alkyl radical add to the nickel center before/after OA, or does it form the C(sp²)–C(sp³) bond via an outer-sphere mechanism? Does the mechanism of OA vary with the identity of the ligand system? Knowledge of the identity of the specific nickel species that undergo OA can provide valuable clues regarding the overall sequence of events in these complex catalytic cycles. Accordingly, a variety of physical, organic, and photophysical probes have been utilized to investigate the fine details of this key elementary step in dual Ni-photoredox catalysis.^2,5 While these efforts have yielded valuable insight, there is still a critical need for an experimental probe that provides insight into the transition state (TS) structure of OA in these reactions.

Scheme 1.

Design of Experiments for Probing the Transition State Structures of the Oxidative Addition Step in Nickel Metallaphotoredox Catalysis Using the “Designed” Reactant, 1,4-Dibromobenzene, To Measure Intramolecular ¹³C KIEs

¹³C Kinetic isotope effects (KIEs) determined at natural abundance (Singleton method) are a powerful technique to interrogate the TSs of key steps in catalytic reactions.⁶ Recently, we illustrated that experimental ¹³C KIEs, in conjunction with predicted KIEs from DFT calculations, can effectively probe the ligand environment at palladium during the TS for the OA step in Suzuki–Miyaura and Buchwald–Hartwig amination reactions.⁷ In the latter reaction, we also introduced the use of symmetric reagents (such as 1,4-dihaloarenes) as ‘designed’ substrates for the rapid evaluation of reaction mechanisms under synthetically relevant conditions; the reaction affords an intramolecular ¹³C KIE from analysis of the monofunctionalized desymmetrized product (Scheme 1B). We envisioned that this combined experimental and theoretical approach would also be effective in evaluating the OA step in nickel catalysis. We report herein the successful application of this novel approach to gain high-resolution atomistic insight into oxidative addition TS in two distinct manifolds of dual Ni photoredox catalysis.

RESULTS AND DISCUSSION

In a catalytic reaction involving a symmetric reactant, intramolecular ¹³C KIEs report on the first step that desymmetrizes the reactant.^7a,8 Since OA of aryl bromides to Ni is usually irreversible, intramolecular KIEs determined for these reactions will likely probe the transition state of this key elementary step. Following the determination of experimental intramolecular ¹³C KIEs for a particular reaction, we utilized DFT calculations to locate OA TSs for various possibilities, including different oxidation states, charges, spin states, and ligand environments at the nickel center. A mismatch of the predicted KIEs for a TS model to the experimentally measured values invalidates the model as a mechanistic possibility, while the model that best fits the experimental KIEs is considered the most likely pathway for OA.

For the ¹³C KIEs study, we chose Ni-metallaphotoredox reactions that utilize two distinct nickel catalyst systems as model reactions: (1) the N-Me selective arylation of tertiary amines to form benzyl dialkylamines (Scheme 2)⁹ – a reaction that uses a Ni bipyridine (neutral ligand) based catalyst system and (2) the deaminative cross-coupling of sterically hindered amines with aryl bromides to form benzylic quaternary centers (Scheme 4),¹⁰ a reaction that utilizes a Ni catalyst with an anionic diketonate ligand (Ni(THMD)₂). In each of these reactions, we used 1,4-dibromobenzene (2′) as the ‘designed’ symmetric reactant for the determination of intramolecular ¹³C KIEs.

Scheme 2.

Experimental Conditions and Proposed Catalytic Cycle for N-Methyl Selective Arylation of Trialkylamines Enabled by Dual Nickel Photoredox Catalysis

Scheme 4.

Experimental Conditions and Proposed Catalytic Cycle for Deaminative Cross-Coupling of α-3° Amines with Aryl Bromides Enabled by Nickel Metallaphotoredox Catalysis

Transition State Analysis for Oxidative Addition to Nickel with Neutral Ligands.

In 2021, we reported an N-methyl selective arylation of trialkylamines enabled by Ni/photoredox dual catalysis. In a prototypical reaction, dicyclohexylmethyl amine (1, 2 equiv) reacts with p-cyanobromobenzene (2, 1 equiv) to deliver the methyl C–H arylated product (3) in 75% yield (Scheme 2A). The reaction is catalyzed by a NiCl₂(dtbbpy)(H₂O)₄ (Ni1), an organic photocatalyst 4CzIPN (PC1), and utilizes an inorganic base (Na₂CO₃) in dioxane under blue light irradiation. In our original report, we proposed a mechanism where oxidative addition of aryl bromide occurs to a Ni(0)bipyridyl complex (Scheme 2B, Ni1-A), resulting in a Ni(II) oxidative addition complex (Scheme 2, Ni1-B). Simultaneously, photoexcited PC1* oxidizes 1 to form the amine radical cation 1•⁺, which is easily deprotonated to form α-amino radical 4. This carbon-centered radical adds to Ni1-B to form Ni(III) intermediate Ni1-C. Reductive elimination from Ni1-C delivers product 3 and Ni(I) intermediate Ni1-D. Single electron transfer from the reduced photocatalyst to Ni1-D, followed by loss of bromide, closes both catalytic cycles, regenerating Ni1-A and PC1. We also presented some evidence for the intermediacy of an iminium ion intermediate in this reaction, Ni1-C_im, that might be complexed to nickel as an off-cycle intermediate, that is in equilibrium with Ni1–C. We proposed that stable complexes between the nickel and methylene iminium ions, such as Ni1-C_im, could play a role in the observed selective arylation of methyl over other alkyl groups.

Since the formation of the α-amino radical occurs via deprotonation of 1•⁺, an obvious experiment to obtain mechanistic information is a competitive k_H/k_D measurement for 1. A competitive k_H/k_D experiment probes the first irreversible step in the catalytic cycle of a reactant. Accordingly, we performed the reaction of 2 under standard reaction conditions (Scheme 2A) but with a 50:50 mixture of 1 and 1-d₃ (Figure 1). Aliquots were drawn from the reaction mixture and purified to isolate all isotopologues of product 3. At 1 h, ¹H NMR analysis of the isolated product mixture revealed a 50:50 ratio of 3:3-d₂, indicating a k_H/k_D of 1.0. At later time points, we observe a steadily increasing formation of the monodeuterated isotopologue 3-d₁. At 24 h, our analysis revealed a 42:44:14 ratio of 3:3-d₁:3-d₂-suggesting that H/D exchange into the benzylic position of the product was occurring under reaction conditions (Figure 1). The presence of 3-d₁ clearly complicates the interpretation of the competitive k_H/k_D experiment.

Figure 1.

Intermolecular k_H/k_D measurement to probe involvement of C–H bond cleavage in first-irreversible step with respect to 1 in the N-methyl selective arylation of trialkylamines enabled by dual nickel photoredox catalysis.

Since the interpretation of the competitive k_H/k_D experiment is ambiguous, we sought to obtain clarification by designing an experiment to measure ¹³C KIEs on the methyl carbon of the amine that is arylated. We envisioned performing this reaction with both the tertiary amine and the aryl bromide components being symmetric for intramolecular ¹³C KIE experiments. Accordingly, we performed this reaction with N,N-dimethylisopropyl amine (1′) and 1,4-dibromobenzene (2′). The N-methyl arylated product (3′) isolated from the reaction mixture can be used to determine intramolecular ¹³C KIEs for the bond-forming carbon atoms of both the aryl bromide and the trialkylamine (as shown in Figures 2A and 3A). Under standard reaction conditions, we isolated product 3′ without any double functionalization for the determination of intramolecular ¹³C KIEs. The results from these measurements tell a compelling story; there is a unity KIE on the bond-forming carbon atom of the trialkyl amine (Figure 2A, KIE_C–Me = 0.999(1)), and a significant normal KIE on the bond-forming carbon atom of the aryl bromide (Figure 3A, KIE_C–Br = 1.024(1)).¹¹ The qualitative interpretation of KIE_C–Br is that the first irreversibly desymmetrizing step for 2′ involves C–Br bond cleavage (likely oxidative addition). In contrast, the qualitative interpretation of KIE_C–Me is that the methyl carbon is not undergoing a bonding change at the first irreversibly desymmetrizing step for 1′.

Figure 2.

Experimental intramolecular ¹³C KIEs obtained for N-methyl selective arylation of trialkylamines using symmetric N,N-dimethylisopropyl (6′). Also shown are the transition structures for various pathways for 6’ along with theoretical ¹³C KIEs using the B3LYP-D3(BJ)/6–31+G** CPCM(1,4-dioxane) level of theory. All bond distances are in Angstroms (Å).

Figure 3.

Experimental intramolecular ¹³C KIEs obtained for N-methyl selective arylation of trialkylamines using symmetric 1,4-dibromobenzene (2′). Also shown are the transition structures for various oxidative addition pathways for 2′ along with theoretical ¹³C KIEs using the B3LYP-D3(BJ)/6–31+G** CPCM(1,4-dioxane) level of theory. All bond distances are in Angstroms (Å). The free energy barriers for each TS is relative to the enegy of the separated reactants i.e the preceding nickel species and 2′.

For the quantitative interpretation of the experimental ¹³C KIEs (KIE_C‑Me and KIE_C–Br) for 1′ and 2′, we modeled all relevant steps in plausible catalytic cycles using B3LYP-D3(BJ)/6–31+G** CPCM(DMSO) level of theory as implemented in Gaussian 16.^12–14 The oxidative addition transition structures were labeled using the template Ni1-OA-TSX^y where Ni1 refers to the nickel catalyst used (Scheme 2A), OA refers to the steps being modeled i.e. oxidative addition, X indicates the different mechanistic models considered for oxidative addition, and y refers to the spin state corresponding to the lowest energy TS for that particular model. From the scaled vibrational frequencies of the computed transition structures, ¹³C KIE predictions were obtained from the reduced vibrational frequencies using the equation of Bigeleisen and Mayer,^15,16 and a Wigner tunneling correction was applied to the predicted values. The predicted KIEs (Figures 2, 3, and 4) were normalized against the carbon atom assumed to be the “standard” carbon for the experimental measurements. To ensure an unbiased interpretation of experimental KIEs, the transition structures that gave the best predictions for KIE_C–Br in the N-Me selective arylation were calculated using six distinct functionals (B3LYP-D3(BJ), B3LYP-D3,¹⁷ M06,¹⁸ MN15,¹⁹ B97D,²⁰ and ωB97X-D²¹) each with either Pople (6–31+G**) or Karlsruhe (def 2TZVP)²² basis sets on all atoms as implemented in Gaussian (See pages S24–S26 in Supporting Information). The approach reported herein is routinely utilized and is well-established to evaluate reactivity and selectivity in similar catalytic systems.^2,23–25

Figure 4.

(A) Experimental intramolecular ¹³C KIEs obtained for deaminative arylation using symmetric 1,4-dibromobenzene (2’). (B) Transition structures for various oxidative addition pathways for 2’ along with theoretical ¹³C KIEs using the B3LYP-D3(BJ)/6–31+G** CPCM(DMSO) level of theory. All bond distances are in Angstroms (Å).

It is important to note that oxidation of 1′ to the corresponding radical cation 1′•⁺ does not desymmetrize 1′. Therefore, KIE_C–Me reports on the first irreversible step in the catalytic cycle for 1′ after amine oxidation. All three steps after formation of 1′•⁺ in our originally proposed catalytic cycle (Scheme 2B)–(a) deprotonation to form 4′ (the α-amino radical corresponding to 1′), (b) addition of 4′ to Ni1–B to form Ni1–C, and (c) reductive elimination from Ni1–C to deliver the product 3′–involve bonding changes at the methyl carbon of the tertiary amine. The relevant transition structures and predicted KIE_C–Me for each of these steps are shown in Figure 2B. Deprotonation of 1′ by either carbonate (TS-dep1) or another molecule of 1′ (TS-dep2) was found to have a predicted KIE_C–Me of 1.009 and 1.011, respectively, values that are inconsistent with the experimental KIE_C–Me. The transition state for the addition of 4′ to Ni1–B to form Ni1–C (TS-add-TS²) also gave a predicted KIE_C-Me that is higher (1.013) than the experimental value. Finally, the TS for reductive elimination from Ni1-C (Ni1-RE-TS²) gives a predicted KIE_C–Me of 1.029, which is also in poor agreement with experimental KIE_C–Me.

We also recognize that an alternative mechanism can be envisioned for this reaction, based on prior mechanistic studies of related reactions,² where the α-amino radical 4 (or 4′) adds to Ni1-A (instead of Ni1-B) to form a Ni(I) intermediate Ni1-E (shown in red in Scheme 2B). Oxidative addition of 2 (or 2′) to Ni1-E will result in the same prereductive elimination intermediate Ni1-C that was invoked in our original proposal. All attempts to locate the TS for the addition of 4′ to Ni1-A resulted in the formation of intermediate Ni1-E, suggesting that this is a barrierless step. Therefore, if this is the first irreversibly desymmetrizing step in the catalytic cycle for 1′, then the predicted KIE_C-Me would resemble the intramolecular equilibrium isotope effect (EIE) for the formation of the α-amino radical (since the TS for barrierless addition of 4′ to Ni1-A will be early and resemble the geometry of 4′). The predicted intramolecular EIE for formation of 4′ is 1.021, which is inconsistent with the experimental KIE_C–Me, suggesting that this is not the first irreversibly desymmetrizing step in the catalytic cycle for 1′ (Figure 2C). Finally, we modeled the oxidative addition of 2′ to Ni1-E, the Ni(I) intermediate resulting from the addition of 4′ to Ni1-A (Ni1-OA-TS3²). The predicted KIE_C–Me for this TS is 1.016, which is in poor agreement with the experimental KIE_C–Me of 0.999 (Figure 2C). At first glance, the predicted KIE_C–Me of 1.016 for this TS might seem surprising since this carbon atom is not undergoing a bonding change at this transition state. The most likely origin of this sizable normal KIE is the difference in the bond strength of the C–Ni bond in Ni1-OA-TS3² compared to the methyl C–H bond in 1′.

At this point, we have discounted every step involving 1′ shown in both pathways in Scheme 2 as the first irreversibly desymmetrizing step that accounts for experimental KIE_C–Me. The near-unity KIE_C–Me raises the possibility that the formation of a desymmetrized H-bonded complex, between 1′•⁺ and the base, prior to the deprotonation event (that forms 4′) could be the first irreversibly desymmetrizing step for 1′. Accordingly, we evaluated the free energy surface for the approach of 1′•⁺ and 1′ to form a H-bonded complex by performing a series of fixed-distance optimizations that vary the distance between the methyl α-CH of 1′•⁺ and the nitrogen of 1 (that acts as the base) from 1.7 to 4.1 Å (Figure 2D). There is a free energy maximum at an α-CH…N distance of ~3.9 Å (TS_H-bond), which is 4.0 kcal/mol above the separated reactants and leads to the H-bonded complex that lies 2.7 kcal/mol below the separated starting materials. The deprotonation TS (TS-dep2) is 7.2 kcal/mol lower in energy than that of TS_H-bond, which becomes the first irreversible step that desymmetrizes 1′. The predicted EIE_C–Me for TS_H-bond is 1.002, which is in good agreement with the experimental KIE_C–Me value of 0.999 (1).

For the quantitative interpretation of the KIE_C–Br of 1.024(1) in the reaction of 1′ and 2′, we modeled various possibilities for oxidative addition (Figure 3B). Initially, we located Ni1-OA-TS1³–the lowest energy transition structure for oxidative addition of 2′ to Ni(0)(dtbbpy) complex Ni1-A, our originally proposed mechanism for oxidative addition (Scheme 2). This TS gave a predicted KIE_C–Br of 1.028, which is slightly higher than the experimental KIE_C–Br. Next, we modeled the oxidative addition of 2′ to Ni(I)(dtbbpy)(Br) complex Ni1-D (Ni1-OA-TS2²); the predicted KIE_C–Br for this TS is 1.020, which is slightly lower than the experimental KIE_C–Br. We also modeled Ni1-OA-TS3² involving oxidative addition of 2′ to the Ni(I)(dtbbpy)(α-aminoalkyl) complex (Scheme 2, Ni1-E). The predicted KIE_C–Br for this model is 1.025, which is in good agreement with the experiment.

As previously described, we suggested that nickel complexes of iminium ions (such as Ni1-C_im, Scheme 2) are off-cycle intermediates in this reaction. We questioned whether an analogous complex could be the active catalyst that engages the aryl bromide in the key oxidative addition step. Specifically, we envisioned that the addition of a α-amino radical (4/4′) to Ni(I)(dtbbpy)(Br) (Ni1-D) could result in a Ni(II) intermediate (Scheme 3, Ni1-F) that could be in equilibrium with a Ni(0) methylene iminium bromide (Scheme 3, Ni1–F_im) species (similar to the equilibrium between Ni1–C and Ni1–C_im described in Scheme 2). Based on this hypothesis and a literature precedent of stable isolable Ni(0)-iminium ion complexes,^9,26 we investigated the oxidative addition of 2′ to a Ni(0)-methylene iminium bromide complex as a possible mechanism for oxidative addition. The transition structure corresponding to this proposed mechanism for oxidative addition was located (Figure 3B, Ni1-OA-TS4¹) and the predicted KIE_C–Br for this TS was 1.025, which again is in excellent agreement with the experimental KIE_C–Br of 1.024(1).²⁷

Scheme 3.

Plausible Catalytic Cycles for N-Methyl Selective Arylation of Trialkylamines Based on Identification of Ni1-OA-TS3² and Ni1-OA-TS4¹ as the Two Most Likely Transition States for Oxidative Addition of Aryl Bromide

We recognize that all four models for oxidative addition presented in Figure 3 have predicted KIE_C–Br values that are relatively close to the experimental KIE_C–Br.²⁸ Therefore, we evaluated the free energy barrier (ΔG^‡) for each of the four transition structures relative to separated reactants (1′ + nickel intermediate that reacts with 1′, see Figure 3). One of the transition structures that shows the best agreement between experimental and predicted KIE_C–Br–Ni1-OA-TS3² had the lowest calculated ΔG^‡ values of 2.8 kcal/mol, suggesting that this is the most likely mechanism for the oxidative addition step. Thus, from the experimental and predicted ¹³C KIEs for both reactants and the detailed DFT analysis of competing mechanisms, the most likely catalytic cycle for the N-methyl selective arylation reaction of 1 and 2 is shown in Scheme 3A. This mechanism involves the initial addition of the α-amino radical 4 to Ni1-A to form Ni1-E. This is followed by oxidative addition of 2 to Ni1-E (via Ni1-OA-TS3²), resulting in Ni1–C, which undergoes reductive elimination to deliver 3 and Ni(I)(dtbbpy)(Br) complex Ni1-D. Reduction of Ni1-D by the reduced photocatalyst (PC1•⁻) regenerates Ni1-A and PC1.

An alternative mechanism that is also consistent with the experimental KIEs can be envisioned if, instead of being reduced by the photocatalyst, Ni1-D is intercepted by α-amino radical 4 to form Ni(II) complex Ni1–F. This intermediate is in equilibrium with the corresponding Ni(0)-methylene iminium ion intermediate Ni1–F_im. Oxidative addition of 2 to Ni1–F_im via Ni1-OA-TS4¹ results in a Ni(II) intermediate Ni1-G, which upon single electron reduction from the reduced photocatalyst (PC1•⁻) results in loss of bromide ion, a neutral Ni(III) intermediate Ni1–C, and regeneration of the photocatalyst. Subsequent reductive elimination from Ni1–C delivers product 3 and regenerates Ni1-D for the next catalytic cycle. Both catalytic cycles in Scheme 3 are consistent with experimental KIE_C-Me and KIE_C–Br and might both be operational to some degree, depending on the relative rates of reduction of Ni1-D by SET from PC1•⁻ versus the addition of 4 to Ni1-D. However, we favor the mechanism in Scheme 3A due to the much lower ΔG^‡ for the key oxidative addition step.

Transition State Analysis for Oxidative Addition to Nickel with Anionic Ligands.

In 2021, we reported the deaminative cross-coupling of α-3° amines with aryl bromides enabled by Ni/photoredox dual catalysis. In a prototypical reaction, N-tert-butyl-1-(2,4,6-trimethoxyphenyl)methanimine (5, 2 equiv) reacts with p-cyanobromobenzene (2, 1 equiv) to deliver the sp²-sp³ cross-coupled product (6) in 94% yield. A combination of nickel catalyst with an anionic diketonate ligand Ni(TMHD)₂ (Ni2), photocatalyst (Ir(dF-CF₃-ppy)₂(dtbbpy)PF₆) (PC2), an inorganic base (K₂HPO₄) and tetrabutyl ammonium chloride salt in DMSO was utilized to catalyze the reaction under blue light irradiation (Scheme 4A). The reaction is proposed to proceed via the OA of 2 to the Ni(I) species (Ni2-A) to form a Ni(III) oxidative addition complex (Ni2–B). Simultaneously, the photoexcited Ir(III) photocatalyst oxidizes the redox-active amine, which undergoes deprotonation (by either the inorganic base or another molecule of 5) to form an imidoyl radical, and subsequent β-scission, resulting in a tert-butyl radical 8 and cyanoarene. The oxidative addition complex Ni2–B undergoes outer-sphere reductive elimination with 8 to deliver the cross-coupling product and Ni(II) intermediate Ni2–C. Finally, single electron transfer from the reduced Ir(II) species to Ni2–C, followed by loss of halide ion, regenerates Ni2-A and the Ir(III) photocatalyst to close both catalytic cycles (Scheme 4B).

The originally proposed mechanism is consistent with all experimental data but failed to account for the 2-fold rate acceleration found in the presence of the tetraalkylammonium chloride additive,²⁹ which was an additional stimulus for pursuing more detailed mechanistic studies. In order to probe the key oxidative addition step using intramolecular ¹³C KIEs, we performed the reaction of 1,4-dibromobenzene (2′) with 5 under the optimized reaction conditions. The reaction was stopped at ~50% consumption of 1,4-dibromobenzene to prevent any formation of the double addition product. The monofunctionalized product 6′ was isolated from the reaction, and the intramolecular ¹³C KIEs were determined. The resulting ¹³C KIEs from 12 measurements from two independent experiments are shown in Figure 4A. There is a significant normal ¹³C KIE of 1.020(2) on the carbon atom that undergoes OA (Figure 4A, KIE_C–Br).³⁰ This result is qualitatively consistent with oxidative addition being the first irreversible step in the catalytic cycle that desymmetrizes 2′.⁷

For the quantitative interpretation of the experimental KIE_C–Br for 2′, we modeled various transition structures, and the ¹³C KIE predictions were obtained using the same methods described in the previous section (vide supra). The predicted KIEs (Figure 4B) were standardized against the carbon atom assumed to be the “standard” carbon for the experimental measurements. We evaluated the competing models for oxidative addition using a variety of DFT methods, as discussed in the previous section (see pages S11–S13 in the Supporting Information).

In a similar reaction, Gutierrez and Molander^2b have proposed that OA occurs to complex Ni2-A (previously described in Scheme 4B). We modeled this TS and found the predicted KIE_C–Br to be 1.012 (Figure 4B, Ni2-OA-TS1²). This prediction is not in agreement with the experimental KIE_C–Br, suggesting that Ni2-A is likely not the active species undergoing oxidative addition in this system. An alternative pathway considered by Gutierrez and Molander involves an initial reaction of the t-butyl radical with Ni2-A to form a Ni(II)(THMD)(t-Bu) complex, which then undergoes oxidative addition to 2′ to form the corresponding Ni(IV) complex. Subsequent inner-sphere reductive elimination would then deliver product 6′ and Ni(II) intermediate Ni2–C. We modeled the oxidative addition TS in this mechanism (Ni2-OA-TS2³) and obtained a predicted KIE_C–Br of 1.033, which is in poor agreement with the experimental KIE_C–Br of 1.020.¹² Since both pathways discussed above result in a Ni(II)-THMD(Br) complex in the product-forming step, we also modeled oxidative addition to this complex (Ni2-OA-TS3¹) to give a Ni(IV)THMD(Ar)(Br)₂ complex as an alternative mechanism for oxidative addition. Once again, a predicted KIE_C–Br of 1.029 for Ni2-OA-TS3¹ is inconsistent with this pathway as the operational mechanism for oxidative addition. Next, we considered the possibility of oxidative addition of the aryl bromide to an anionic Ni(I)(THMD)(Cl) complex to form the corresponding anionic Ni(III) complex as an operative pathway.³¹ This anionic Ni(I) complex could form by reaction of Ni2-A with chloride ions in solution (from the nBu₄NCl additive). The OA TS for this mechanism is Ni2-OA-TS4_Cl², and the corresponding predicted KIE_C–Br is 1.021, a value that is in excellent agreement with the experimental KIE_C–Br of 1.020. We also modeled the OA TS to a similar anionic complex with a bromide ion instead of a chloride ion (Figure 4, Ni2-OA-TS4_Br²)–the predicted KIE_C–Br for Ni2-OA-TS4_Br² is also 1.021, again, in excellent agreement with the experimental KIE_C–Br.

Even though the predicted KIE_C–Br for the models shown in Figure 4 support an anionic nickel species as the active species involved in the oxidative addition step, one insightful reviewer pointed out that the low imaginary mode for Ni2-OA-TS1² corresponds to an energy surface that is almost flat around the saddle point. Such energy surfaces are uniquely prone to large variations in KIEs while using different functionals, since slight changes in geometries can lead to large changes in isotope effects. Therefore, to ensure an unbiased interpretation of our experimental KIEs, we evaluated Ni2-OA-TS1² and Ni2-OA-TS4_Cl² using a total of 21 different DFT methods and found that the average predicted KIE_C–Br for Ni2-OA-TS1² was 1.021 ± 0.008 while that for Ni2-OA-TS4Cl² was 1.024 ± 0.004 (See Supporting Information, page S14). The good agreement of the average predicted KIE_C–Br for both models with the experimental KIE_C–Br suggests that both of these pathways might simultaneously be operational. However, the larger spread of the predicted KIEs for Ni2-OA-TS1² makes this interpretation relatively less reliable solely on the basis of the predicted KIEs.

To further evaluate these competing mechanisms, we decided to compare the free energy surfaces for the neutral and anionic pathways for oxidative addition. Comparison of pathways with different charges is typically fraught with error. We sought to mitigate these errors by including a tetramethylammonium (TMA) counterion in the calculations of the free energy surface for the anionic pathway leading up to Ni2-OA-TS4_Cl² to enable a comparison to the neutral pathway leading up to Ni2-OA-TS1². Using separated starting materials as the reference, we find that the prereactive complex of the neutral oxidative addition TS (Figure 5, N₀) is downhill by 10.5 kcal/mol. The oxidative addition TS (Ni2-OA-TS1²) has a 3.8 kcal/mol barrier from N₀. For the anionic pathway, the prereactive complex of anionic oxidative addition TS (with TMA counterion, Figure 5, A₀) is downhill from separated starting materials by 11.7 kcal/mol. The anionic oxidative addition TS with the TMA counterion included (Ni2-OA-TS4²_Cl-TMA) has a barrier of 3.7 kcal/mol from that of A₀. Assuming there is no significant barrier for the formation of N₀/A₀ and considering the potential errors associated with this comparison, these calculations suggest that both pathways are equally viable, with a marginal preference for the anionic pathway.

Figure 5.

Comparative free energy profiles leading up to the key oxidative addition step of originally proposed (black) and anionic (red) mechanisms for oxidative addition calculated at B3LYP-D3(BJ)/6–31+G** CPCM (dimethyl sulfoxide).

In our original report,¹⁰ we noted a 2-fold increase in the rate in the presence of tetrabutylammonium chloride additive for the reaction of p-bromoacetophenone. Such electron-deficient bromoarenes are privileged substrates for this reaction. For electron-neutral or electron-rich aryl bromides, a chloride additive is crucial for any reactivity–suggesting that oxidative addition into these substrates is quite challenging. We performed the deaminative cross-coupling of 5 and 2′ in the absence of tetrabutylammonium chloride additive and, in line with our previous results with similar substrates, obtained only trace amounts of 6′ (see the Supporting Information, S20–21). This experiment illustrates that the rate acceleration with the chloride additive for electron-neutral bromoarenes, such as 2′, is quite drastic compared to activated bromoarenes such as p-bromoacetophenone. We interpret this result as evidence of the anionic oxidative addition being the more likely pathway for 2′.

Based on the experimental and computational results presented here, the most likely pathway for the deaminative cross-coupling of α-3° amines with aryl bromides proceeds via an anionic Ni(I)THMD(X) complex (Ni2-OA-TS4_Cl²/Ni2-OA-TS4_Br²). Based on this finding, we propose the following revised catalytic cycle (Scheme 5). The first step in the catalytic cycle is oxidative addition of aryl bromide to the anionic Ni(I)(THMD)(Cl) complex (Ni2-D), resulting in the Ni(III)THMD(Ar)(Br)(Cl) anionic complex (Ni2-E). This anionic Ni(III) complex undergoes outer-sphere C(sp²)–C(sp³) bond-forming reductive elimination with the tertiary radical to deliver the product and an anionic Ni(II)THMD-(Br)(Cl) complex (Ni2–F). Single electron transfer from the Ir(II) photocatalyst and concomitant loss of bromide ion closes the catalytic cycle and regenerates the anionic Ni(I)THMD(Cl) complex (Ni2-D). Alternatively, SET from the iridium photocatalyst can result in the loss of a chloride ion to generate the anionic Ni(I)THMD(Br) complex. It must be noted that the originally proposed catalytic cycle (Scheme 4) involving oxidative addition to Ni2-A cannot be discounted, especially for electron-deficient aryl bromides.

Scheme 5.

Catalytic Cycle for Deaminative Cross-Coupling Based on the Identification of Ni2-OA-TS4²_Cl as the Most Likely Transition State for the Oxidative Addition of Aryl Bromide

CONCLUSIONS

In conclusion, we have introduced the application of intramolecular ¹³C KIEs using symmetric reactants as a powerful tool to obtain a high-resolution experimental snapshot of the oxidative addition transition state structures in dual nickel-photoredox catalysis. The results presented in this article have led to a more in-depth understanding of the mechanisms of two C(sp²)–C(sp³) cross-coupling reactions catalyzed by two distinct ligand systems that are commonly used in nickel metallaphotoredox reactions. In the N-methyl selective arylation of trialkyl amines catalyzed by nickel with a neutral bipyridyl ligand, we show that our experimental KIEs are most consistent with a Ni(I)(dtbbpy)(α-aminoalkyl) as the species that engages the aryl bromide in the key oxidative addition step (Scheme 3). In the deaminative arylation of sterically hindered primary amines catalyzed by nickel with an anionic diketonate ligand, our studies suggest the likely involvement of an anionic Ni(I) intermediate in the key oxidative addition step. We anticipate that this simple approach, which yields experimentally validated computational models for the key oxidative addition step, will gain traction as the preferred mechanistic tool among the practitioners of nickel metallaphotoredox catalysis. Current work in our groups is focused on similar descriptions of analogous carbon-heteroatom bond-forming reactions.

Supplementary Material

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/jacs.4c10182.

Experimental procedures, coordinates of all computed structures, and product characterization data (PDF)

Tables containing values (PDF)