Effect of 6,6′-Substituents on Bipyridine-Ligated Ni Catalysts for Cross-Electrophile Coupling

Abstract

A family of 4,4’-^tBu₂-2,2’-bipyridine (^tBubpy) ligands with substituents in either the 6-position, 4,4’-^tBu₂-6-Me-bpy (^tBubpy^Me), or 6 and 6’-positions, 4,4’-^tBu₂-6,6’-R₂-bpy (^tBubpy^R2; R = Me, ⁱPr, ^sBu, Ph, or Mes), was synthesized. These ligands were used to prepare Ni complexes in the 0, I, and II oxidation states. We observed that the substituents in the 6 and 6’-positions of the ^tBubpy ligand impact the properties of the Ni complexes. For example, bulkier substituents in the 6,6’-positions of ^tBubpy better stabilized (^tBubpy^R2)Ni^ICl species and resulted in cleaner reduction from (^tBubpy^R2)Ni^IICl₂. However, bulkier substituents hindered or prevented coordination of ^tBubpy^R2 ligands to Ni⁰(cod)₂. In addition, by using complexes of the type (^tBubpy^Me)NiCl₂ and (^tBubpy^R2)NiCl₂ as precatalysts for different XEC reactions, we demonstrated that the 6 or 6,6’ substituents lead to major differences in catalytic performance. Specifically, while (^tBubpy^Me)Ni^IICl₂ is one of the most active catalysts reported to date for XEC and can facilitate XEC reactions at room temperature, lower turnover frequencies were observed for catalysts containing ^tBubpy^R2 ligands. A detailed study on the catalytic intermediates (^tBubpy)Ni(Ar)I and (^tBubpy^Me2)Ni(Ar)I revealed several factors that likely contributed to the differences in catalytic activity. For example, whereas complexes of the type (^tBubpy)Ni(Ar)I are low spin and relatively stable, complexes of the type (^tBubpy^Me2)Ni(Ar)I are high-spin and less stable. Further, (^tBubpy^Me2)Ni(Ar)I captures primary and benzylic alkyl radicals more slowly than (^tBubpy)Ni(Ar)I, consistent with the lower activity of the former in catalysis. Our findings will assist in the design of tailor-made ligands for Ni-catalyzed transformations.

Graphical Abstract

Introduction

Transition metal-catalyzed processes represent a powerful class of synthetic methods and are used extensively in a variety of industrial settings.¹ The majority of these processes utilize precious metal catalysts, such as Pd, and there are relatively few commercial reactions that utilize more sustainable first-row metals, such as Ni.² Apart from providing economic and environmental benefits, Ni can catalyze reactions that are difficult for precious metal systems. For example, Ni catalysts are vastly superior to Pd catalysts for many contemporary reactions such as cross-coupling reactions using sp³-hybridized substrates,³ dual Ni-photoredox catalyzed processes,⁴ reductive coupling reactions, such as cross-electrophile coupling (XEC),⁵ reductive difunctionalization reactions involving unsaturated hydrocarbons,⁶ and addition reactions to carbonyl-type compounds.⁷ These transformations enable the formation of bonds that are difficult to access using other synthetic methods.⁸ However, although modern Ni-catalyzed reactions could be transformative in the synthesis of active pharmaceutical ingredients, there are often limitations relating to scope or practicality that prevent their widespread utilization in process chemistry.^2g

One of the reasons that precious metal catalyzed transformations are commonly used in industry is because the effects of the ancillary ligand on the elementary steps in catalysis are well understood.⁹ This enables the rational design of tailor-made ligands with specific properties. For example, Buchwald type biarylphosphine ligands are widely used in Pd-catalyzed cross-coupling because they promote elementary steps such as oxidative addition and reductive elimination, while preventing catalyst decomposition.¹⁰ In contrast, the design of tailor-made ligands to promote Ni-catalyzed reactions, such as XEC, is in its infancy.¹¹ For example, even though 2,2’-bipyridine (bpy) ligands have been extensively used to support Ni catalysts for XEC and metallaphotoredox based transformations,^11d there are limited studies exploring the impact of modifying the bpy ligand.¹² A notable exception is the extensive use of 4,4’-^tBu₂-2,2’-bipyridine (^tBubpy), which is proposed to generate metal complexes with greater solubility.¹³

In precious metal catalysis, the steric bulk of the ancillary ligand is known to dramatically affect the rates of elementary processes.¹⁴ Therefore, it is surprising that ortho-substituted bpy ligands have not been broadly studied in Ni-catalyzed reactions, especially given a small number of studies that have demonstrated they can lead to improvements in catalysis (Figure 1a). For example, Martin and co-workers reported that a Ni catalyst supported by a 4,4’-^tBu₂-6-Me-bpy (^tBubpy^Me) ligand gives higher yields for the site-selective carboxylation of unsaturated hydrocarbons with CO₂ and water than systems supported by ^tBubpy.¹⁵ Similarly, Martin et al. recently demonstrated that coordination of a bulky 4,4’-^tBu₂-6,6’-ⁱBu-bpy (^tBubpy^iBu2) ligand to a Ni-catalyst enables direct carboxylation of unactivated secondary alkyl bromides without chain walking.¹⁶ In addition to reductive carboxylation reactions, it has also been reported that the ortho-methyl substituents in 6,6’-Me₂-bpy (bpy^Me2) are crucial for achieving high yields in the Ni-catalyzed XEC of 1,3-dienes with aldehydes and aryl bromides¹⁷ and the Ni-catalyzed reductive cross-coupling of alkyl halides with arylthiosilanes.¹⁸ However, it is unclear how the ortho-substituents influence the elementary steps in catalysis, and bpy ligands with bulkier 6,6’-substituents are typically not used in catalysis in part due to their limited availability.

Figure 1:

a) Prior examples of Ni-catalyzed XEC reactions where the optimized conditions include a 6 or 6,6’-substituted bpy ligand. b) Our work systematically exploring the effect of 6,6’-substituents on bpy ligated Ni catalysts for XEC.

In this work, we describe the synthesis of complexes in the Ni⁰, Ni^I, and Ni^II oxidation states containing the 6 or 6,6’-substituted ^tBubpy ligands, ^tBubpy^Me or ^tBubpy^R2 (R = Me, ⁱPr, ^sBu, Ph, or Mes) (Figure 1b). We show that small changes to the ligand structure make a large difference to the properties of the complexes including thermal stability, reduction potential, and spin state. For example, Ni^I complexes of the type (^tBubpy^R2)NiCl, which are proposed as catalytic intermediates in several Ni-catalyzed transformations,^2i,6f decompose more slowly as the size of the ^tBubpy ligand increases (Figure 1b). Further, whereas complexes of the type (^tBubpy)Ni(Ar)I (Ar = aryl) containing an unsubstituted ^tBubpy ligand are low spin and relatively thermal stable, complexes of the type (^tBubpy^Me2)Ni(Ar)I containing a ^tBubpy ligand with methyl groups in the 6 and 6’-positions are high spin and quickly decompose to form biaryl. As part of this study, we compare the rates of radical trapping between complexes of the type (^tBubpy)Ni(Ar)I and (^tBubpy^Me2)Ni(Ar)I and demonstrate that species containing unsubstituted ^tBubpy ligands trap radicals faster (Figure 1b). Additionally, by using complexes of the type (^tBubpy^Me)NiCl₂ and (^tBubpy^R2)NiCl₂ as precatalysts for different XEC reactions, we connect our stoichiometric experiments to catalysis. The 6 or 6,6’-bpy-substituents lead to major differences in catalytic performance, especially in relation to turnover frequencies (TOFs). Specifically, we show that for some XEC reactions (^tBubpy^Me)NiCl₂ is a more active catalyst than systems previously described in the literature and can facilitate room temperature XEC. Overall, our work provides information about the factors that are important in the design of tailor-made ligands for Ni-catalyzed reactions, many of which are different to those that are crucial in ligand design for precious metal-catalyzed reactions.

Results and Discussion

Synthesis of 6,6’-Subsituted Bipyridine Ligands

To probe the impact of varying the steric properties of the ancillary ligand on the speciation, stability, and catalytic activity of Ni complexes, we synthesized a family of 6,6’-substituted ^tBubpy ligands. We included tert-butyl groups in the 4,4’-positions because this improves the solubility of the Ni complexes in organic solvents (see SI Section E). Two different strategies were used to synthesize the 6,6’-substituted ^tBubpy ligands. For ligands with aryl substituents in the 6,6’-positions, commercially available ^tBubpy was converted to 4,4’-^tBu₂-6,6’-Cl₂-bpy (^tBubpy^Cl2) through initial oxidation with H₂O₂ followed by chlorination with POCl₃ (Figure 2a).¹⁹ Subsequently, a Pd-catalyzed Suzuki-Miyaura reaction between ^tBubpy^Cl2 and either phenyl or mesityl boronic acid was used to generate ^tBubpy^Ph2 or ^tBubpy^Mes2 in yields of 72 and 61%, respectively.^{20 tBu}bpy^Mes2 was prepared on 0.54 g scale, demonstrating that the procedure is scalable.

Figure 2:

a) Synthesis of 4,4’-di-tert-butyl-2,2’-bipyridines with aryl substituents in the 6,6’-positions. b) Synthesis of 4,4’-di-tert-butyl-2,2’-bipyridine with alkyl substituents in the 6,6’-positions. XPhos = Dicyclohexyl[2’,4’,6’-tris(propan-2-yl)[1,1’-biphenyl]-2-yl]phosphane.

The cross-coupling approach did not work well for generating ^tBubpy ligands with alkyl groups in the 6,6’-positions. Instead, ^tBubpy was directly functionalized with different alkyllithium reagents as was previously described for the synthesis of an asymmetric ligand ^tBubpy^Me containing a single methyl group in the 6-position and the synthesis of 2,9’-substituted phenanthroline (phen) ligands (Figure 2b).^15,21 In addition to reproducing the synthesis of ^tBubpy^Me, we used alklylation with alkyllithium reagents to synthesize the symmetric ligands ^tBubpy^Me2, ^tBubpy^iPr2, and ^tBubpy^sBu2, which contain methyl, iso-propyl and sec-butyl groups, respectively in the 6,6’-positions. Although the conditions used for the alkylation reactions were similar, each reaction needed to be individually optimized. Despite the relatively low yields, 39–49%, we were able to prepare each ligand on scales above half a gram. Notably, although ^tBubpy^sBu2 contains chiral alkyl substituents, a racemic mixture was generated and used in this work.

Synthesis and Characterization of LNi^IICl₂ Complexes

Ligated Ni^II dihalide complexes are often used as precatalysts in Ni-catalyzed reactions.^2d Initially, we prepared (^tBubpy^Mes2)NiCl₂ through the reaction of ^tBubpy^Mes2 with anhydrous NiCl₂ at reflux in EtOH (Figure 3a). Although (^tBubpy^Mes2)NiCl₂ was generated in 72% yield, the reaction required almost two days to reach completion, and the improved solubility of (^tBubpy^Mes2)NiCl₂ in organic solvents results in a significant loss in yield when a solvent mixture of THF and diethyl ether was used to wash the crude product. A better method to prepare (^tBubpy^Mes2)NiCl₂ involves sonication with (dme)NiCl₂ (dme = dimethoxyethane) in dme (Figure 3b).²² In this case, the reaction is complete in less than one hour at room temperature, and (^tBubpy^Mes2)NiCl₂ was obtained in 94% yield by precipitating the complex using pentane and then isolating via benchtop filtration. We used the sonication method to prepare (^tBubpy^Me)NiCl₂, (^tBubpy^Me2)NiCl₂, (^tBubpy^iPr2)NiCl₂, (^tBubpy^sBu2)NiCl₂, and (^tBubpy^Ph2)NiCl₂ in high yields. All the ^tBubpy ligated Ni^II dichloride complexes are paramagnetic in solution (S = 1 from the Evans method²³), consistent with a high spin d⁸ system. Despite the paramagnetism of the complexes, it was possible to obtain high quality ¹H NMR spectra, which contained the correct number of resonances and integrated accurately. The new complexes were also characterized using UV-Vis spectroscopy and elemental analysis (see SI Section D).

Figure 3:

Synthesis of Ni^IICl₂ complexes ligated with 6,6’-substituted ^tBubpy ligands using a) heat or b) sonication.

One of the characteristics of Ni-catalyzed reactions is that they are more likely to proceed via one electron pathways compared to their Pd and Pt congeners which typically operate through elementary steps involving two electron changes.² We used cyclic voltammetry to elucidate the impact of the 6,6’-substituted ^tBubpy ligands on the redox properties of our family of Ni complexes. The cyclic voltammograms and the reduction potentials are shown in Figure 4.

Figure 4:

Cyclic voltammograms of a) (^tBubpy)NiCl₂, b) (^tBubpy^Me)NiCl₂, c) (^tBubpy^Me2)NiCl₂, d) (^tBubpy^iPr2)NiCl₂, e) (^tBubpy^sBu2)NiCl₂, and f) (^tBubpy^Mes2)NiCl₂ collected at 1 mM concentration in MeCN at 100 mV/s under an Ar atmosphere. 0.1 M TBAPF₆ was used as supporting electrolyte. Voltammograms were collected in the presence of 1 mM Ferrocene (Fc) as an internal reference. Reduction potentials recorded vs. Fc⁺/Fc in MeCN. Arrows indicate the direction of the cyclic voltammetry scans.

The cyclic voltammogram of the unsubstituted complex (^tBubpy)NiCl₂ in acetonitrile (MeCN) shows many features, as described previously (Figure 4a).²⁴ We assign the irreversible peaks at −1.84 and −1.99 V vs Ferrocenium/Ferrocene (Fc⁺/Fc), respectively, which are relatively ill-defined and overlap with each other, as the Ni^II/Ni^I and Ni^I/Ni⁰ couples. Additionally, we assign the quasi-reversible one electron reduction at −2.23 to reduction of the ^tBubpy ligand, based on a previous report investigating the related complex, (bpy)₂NiBr_2.²⁵ We suggest that other features, for example the peaks at −1.25 and −1.5 V, arise from decomposition of electrogenerated species or the presence of complexes with solvent coordinated, as has been proposed previously.²⁴ Consistent with this hypothesis, peaks other than those at −1.84, −1.99, and −2.23 V show significantly lower current, suggesting that they do not arise directly from (^tBubpy)NiCl₂ or its direct reduction products (see SI Section F).

The cyclic voltammograms of all the Ni complexes containing 6,6’-substituted ^tBubpy ligands, except (^tBubpy^Ph2)NiCl₂, are more clearly resolved and feature lower baselines than the cyclic voltammogram of (^tBubpy)NiCl₂. Two consecutive irreversible or quasi-reversible one-electron reductions assigned as Ni^II/Ni^I and Ni^I/Ni⁰ couples, respectively, and a third reversible couple corresponding to the reduction of the ^tBubpy ligand are observed (Figure 4b–4f). This assignment is consistent with previous reports for structurally related complexes.^24b,25 We hypothesize that the improved electrochemical response of these complexes is due to the stabilization of the Ni^I species formed after the first reduction of the corresponding Ni^II complex. Specifically, the steric bulk of the 6,6’-substitutents prevents disproportionation of Ni^I at the electrode. Interestingly, there is also an anodic shift in the Ni^II/Ni^I couple, which becomes approximately 0.5 V less negative for the 6,6’-substituted complexes relative to the unsubstituted complex (^tBubpy)NiCl₂. The changes in the Ni^I/Ni⁰ couple are smaller, shifting anodically by 0.21 and 0.26 V for the mono- and dimethyl substituted complexes, respectively, and changing by less than 0.1 V for the other derivatives. As expected, the changes in the more negative bpy-based reduction are the smallest when comparing the 6,6’-substituted complexes to the unsubstituted species, with shifts between 0.05–0.10 V in the cathodic direction observed. These shifts in the negative direction are expected for electron donating alkyl groups but are surprising for aryl groups, which are typically electron withdrawing. Finally, it is notable that the voltammograms of all (^tBubpy^R2)NiCl₂ complexes exhibit a relatively small pre-wave near −1.2 V. These types of features have been observed for other transition metal complexes and are often interpreted as adsorption of the complex to the electrode or some other type of electrochemically induced modification of the complex.²⁶

The differences between the Ni^II/Ni^I and Ni^I/Ni⁰ potentials for the 6,6’-substituted Ni complexes range from 0.42 to 0.68 V, setting them apart from the unsubstituted (^tBubpy)NiCl₂ complex for which these two reductions differ by only 0.15 V. This difference is potentially consequential for any reaction that relies on the chemical reduction of these Ni^II complexes. For unsubstituted (^tBubpy)NiCl₂, most chemical reductants capable of reducing the complex to Ni^I will also reduce it to Ni⁰. In contrast, for the 6,6’-substituted ^tBubpy complexes, the larger separation of the two redox events may enable access to the Ni^I oxidation state via chemical reduction without overreduction to Ni⁰. Furthermore, the less negative potentials of the Ni^II/Ni^I couple of the 6,6’-substituted complexes may allow access to the respective Ni^I species using relatively milder reductants.

One outlier in our cyclic voltammetry investigation of 6,6’-substituted ^tBubpy-ligated Ni^II complexes is (^tBubpy^Ph2)NiCl₂ (see SI Section F). The cyclic voltammogram of this complex shows no clear features in the Ni^II/Ni^I region, with an irreversible wave peaking near −1.76 V and no reversible bpy-based reduction, instead a large irreversible wave is seen at −2.61 V. At this stage, the reasons why (^tBubpy^Ph2)NiCl₂ exhibits such distinct electrochemical behavior from the other complexes in the series are unclear, but it is likely undergoing distinct chemical changes, such as cyclometallation, after the initial reduction. Overall, the large differences in redox properties observed across the family of 6,6’-substituted Ni complexes, especially in comparison to the unsubstituted species, is surprising, as the 6,6’-subsititions were expected to have relatively small effect on the electronic properties of the complexes.

The solid-state structures of (^tBubpy^Mes2)NiCl₂, (^tBubpy^Ph2)NiCl₂, (^tBubpy^sBu2)NiCl₂, and (^tBubpy^Me)NiCl₂ were determined using single crystal X-ray crystallography (Figure 5). (^tBubpy^Mes2)NiCl₂ (τ₄ = 0.83, τ₄’ = 0.76), (^tBubpy^Ph2)NiCl₂ (τ₄ = 0.80, τ₄’ = 0.75), and (^tBubpy^sBu2)NiCl₂ (τ₄ = 0.80, τ₄’ = 0.74) all possess a distorted tetrahedral geometry around the Ni center, which is consistent with their paramagnetic nature in solution. Tetrahedral geometries around Ni have also been observed for other Ni^II complexes with different 6,6’-substituted bpy ligands and halide groups.²⁷ The bond distances and angles around Ni are similar (within three standard deviations) for (^tBubpy^Mes2)NiCl₂, (^tBubpy^Ph2)NiCl₂, and (^tBubpy^sBu2)NiCl₂, with the only major difference that the N(1)-Ni bond in (^tBubpy^sBu2)NiCl₂ is slightly shorter (~0.03 Å) than those in (^tBubpy^Mes2)NiCl₂ or (^tBubpy^Ph2)NiCl₂.

Figure 5:

Solid-state structures with thermal ellipsoids at 50% probability of a) (^tBubpy^Mes2)NiCl₂, b) (^tBubpy^Ph2)NiCl₂, c) (^tBubpy^sBu2)NiCl₂, and d) {(^tBubpy^Me)NiCl₂}₂ with selected bond lengths and angles. Hydrogen atoms omitted for clarity. Additional bond angles for {(^tBubpy^Me)NiCl₂}₂ are shown below: N1-Ni1-Cl1: 96.61 (6), N1-Ni1-Cl2: 98.93 (4), N1-Ni1-Cl3: 114.89 (4), N2-Ni1-Cl1: 88.80 (4), N2-Ni1-Cl2: 172.16 (4), N2-Ni1-Cl3: 93.14 (4), Cl1-Ni1-Cl3: 148.332 (18), Cl2-Ni1-Cl3: 94.215 (16), N3-Ni2-N4: 80.57(6), N3-Ni2-Cl1: 99.91(4), N3-Ni2-Cl2: 105.28(4), N3-Ni2-Cl4: 100.91(4), N4-Ni2-Cl1: 173.69(4), N4-Ni2-Cl2: 89.66(4), N4-Ni2-Cl4: 94.65(4), Cl1-Ni2-Cl2: 84.143(15), Cl1-Ni2-Cl4: 91.437(17), Cl2-Ni2-Cl4: 153.809(19).

In contrast to the monomeric structures observed for (^tBubpy^Mes2)NiCl₂, (^tBubpy^Ph2)NiCl₂, and (^tBubpy^sBu2)NiCl₂, (^tBubpy^Me)NiCl₂ adopts a dimeric structure in the solid state with each Ni bound to a ^tBubpy^Me ligand, a terminal chloride ligand, and two bridging chloride ligands. The geometry around each Ni center is distorted square-based pyramidal (τ₅ = 0.39 for Ni1, 0.33 for Ni2) with one nitrogen atom of the ^tBubpy^Me ligands occupying each of the axial positions. Presumably, the lower steric bulk of ^tBubpy^Me allows (^tBubpy^Me)NiCl₂ to dimerize, and similar dimeric structures have been observed for Ni dihalide complexes with bpy ligands that do not have a large steric profile.²⁸ Despite the fact that (^tBubpy^Me)NiCl₂ is a dimer in the solid state, based on its solution state magnetic moment (~ 3.25 μ_B) which is consistent with an S = 1 system, we propose that it is a monomer in solution. If the dimeric structure was still intact in solution, there would likely be electronic communication between the Ni centers, and either an S = 2 system from ferromagnetic coupling between the Ni centers or an S = 0 system from antiferromagnetic coupling between the Ni centers would be expected.

In Situ Formation of 6,6’-Ligated Ni^I Chloride Complexes

The cyclic voltammograms of the majority of our 6,6’-substituted ^tBubpy Ni^II dichloride complexes indicate that reduction to well-defined Ni^I complexes, without overreduction to Ni⁰, should be feasible using relatively mild chemical reductants. Studying the reactivity of Ni^I complexes is important because Ni^I species are proposed to be intermediates in a wide-range of Ni-catalyzed transformations.^3–7 Although a number of groups have investigated the reactivity of bpy^20,24b,29 and phen³⁰-ligated Ni^I species with alkyl and aryl electrophiles that are relevant to catalysis, systematic studies exploring the steric effect of the ancillary ligand on the stability and reactivity of Ni^I complexes are lacking. Our family of 6,6’-substituted ^tBubpy ligands provides an opportunity to examine ligand effects on Ni^I complexes.

To generate Ni^I complexes, we treated our compounds of the type (^tBubpy^R2)NiCl₂ (R = H, Me, ⁱPr, ^sBu, Ph, and Mes) and (^tBubpy^Me)NiCl₂ with an excess of Zn in 2-MeTHF,³¹ which mimics a proposed elementary step in several Ni-catalyzed XEC reactions.^5c We stirred the Ni^II complexes with Zn³² at room temperature until near full conversion of the Ni^II complex was reached as determined by ¹H NMR spectroscopy (see SI Section G). Interestingly, the rates of consumption of the Ni^II dichloride complexes varied as a function of the 6,6’-substituted ^tBubpy ligand. However, no trends were observed based on the properties of the ^tBubpy^R2 ligand, and factors such as stirring speed and particle size of the reductant are likely important.^30a After filtering off the unreacted Zn and other salt byproducts, we recorded the EPR spectra of the filtrate at 25 K and quantified the amount of EPR active species generated using a CuSO₄•5H₂O EPR standard (Figure 6). The reaction between (^tBubpy)NiCl₂ and Zn afforded an EPR silent species, which is consistent with either reduction to Ni⁰ or the initial formation of unobserved (^tBubpy)NiCl, followed by rapid dimerization to form the EPR inactive dimer [(^tBubpy)Ni^I(μ-Cl)]₂. Several groups have previously formed this dimer via different synthetic routes, which are all proposed to involve the monomeric complex (^tBubpy)NiCl as an intermediate.^29a,29d,29e Reduction of (^tBubpy^Me)NiCl₂ afforded only a small amount of an EPR active species that we were unable to quantify accurately (Figure 6a). We hypothesize that the EPR active species is (^tBubpy^Me)NiCl, which rapidly decomposes presumably because of the lack of steric bulk around the Ni center.

Figure 6:

Experimental and simulated EPR spectra of proposed Ni^I species obtained through reduction of a) (^tBubpy^Me)NiCl₂, b) (^tBubpy^Me2)NiCl₂ (g_‖ = 2.430 & g_⊥= 2.105), c) (^tBubpy^iPr2)NiCl₂ (g_‖ = 2.443 & g_⊥ = 2.103), d) (^tBubpy^sBu2)NiCl₂ (g_‖ = 2.448 & g_⊥ = 2.103), e) (^tBubpy^Ph2)NiCl₂ (g_x = 2.067, g_y = 2.123, g_z = 2.357) and f) (^tBubpy^Mes2)NiCl₂ (g_x = 2.080, g_y = 2.112, g_z = 2.424) with Zn in 2-MeTHF. The EPR spectra were recorded at 25 K in 2-MeTHF. All yields determined by double integration of the EPR spectra against a CuSO₄•5H₂O external standard.

In contrast to the trace amount of EPR active species formed from the reduction of (^tBubpy^Me)NiCl₂, relatively high yields of EPR active species were generated from the reduction of (^tBubpy^R2)NiCl₂ (R = Me, ⁱPr, ^sBu, and Mes) with Zn. In fact, the most sterically bulky systems, (^tBubpy^sBu2)NiCl₂ and (^tBubpy^Mes2)NiCl₂ gave almost quantitative yields of the EPR active species. The reduction of (^tBubpy^Mes2)NiCl₂ with Zn generated a rhombic EPR spectrum (Figure 6f). This spectrum is similar to that previously observed for a monomeric 2,9-dimesityl-substituted (phen)Ni^ICl complex^30b and on that basis, we propose that the product of reduction is (^tBubpy^Mes2)NiCl. The EPR spectrum from reduction of (^tBubpy^sBu2)NiCl₂, which we propose generates (^tBubpy^sBu2)NiCl, is axial (Figure 6d). Likely the greater flexibility of the sec-butyl groups relative to the mesityl groups results in an axial as opposed to a rhombic spectrum. The shift from axial to rhombic EPR spectra with more steric bulk has previously been observed with phosphine-ligated Ni^I species.³³ In a similar fashion to (^tBubpy^sBu2)NiCl₂, reduction of(^tBubpy^Me2)NiCl₂ and (^tBubpy^iPr2)NiCl₂ gives rise to axial EPR spectra, which we assign to (^tBubpy^Me2)NiCl and (^tBubpy^iPr2)NiCl, respectively (Figures 6b & 6c). The yield of these species is slightly lower (~60%), suggesting that the less sterically bulky systems either give less clean reduction or are more unstable. Reduction of (^tBubpy^Ph2)NiCl₂ resulted in a much smaller amount of EPR active material (~17%) than (^tBubpy^Mes2)NiCl₂ and a significant amount of unidentified product. The observed EPR spectrum is rhombic, consistent with the spectrum obtained from the reduction of (^tBubpy^Mes2)NiCl₂, suggesting that (^tBubpy^Ph2)NiCl is formed (Figure 6e). The different behavior of (^tBubpy^Ph2)NiCl₂ compared with (^tBubpy^Mes2)NiCl₂ (and also 6,6’-substituted ^tBubpy ligands with alkyl substituents) on reaction with a reductant is consistent with the different cyclic voltammograms observed for (^tBubpy^Ph2)NiCl₂ (vide supra).

Despite the near quantitative formation of (^tBubpy^sBu2)NiCl and (^tBubpy^Mes2)NiCl, we were unable to isolate these species or grow single crystals. This is in part because the stability of the Ni^I species is highly solvent-dependent. For example, although a sample of (^tBubpy^Mes2)NiCl in 2-MeTHF is stable at −33 °C for up to 2 months (see SI Section G), the introduction of hydrocarbon or aromatic solvents, such as pentane or toluene, to precipitate the complex results in rapid decomposition. This decomposition likely involves disproportionation of the Ni^I complex, as we were able to identify (^tBubpy^R2)NiCl₂ as a side product in our crystallization attempts using ¹H NMR spectroscopy, along with Ni black precipitate. We propose that changing the 4,4’-tert-butyl substituents to lower the solubility of the Ni^I species might aid with crystallization and slow down bimolecular decomposition as examples of Ni^I halide species with 4,4’-{OC(O)Me}₂-bpy or bpy ligands lacking any 4,4’-substituents have been crystallized.^20,29e,29f Our results indicate that Ni^I species of the form (^tBubpy^R2)NiCl can unequivocally be accessed via the reduction of Ni^II complexes and that the 6,6’-substituents on the ^tBubpy ligands play a key role in the reaction outcome, as larger steric groups lead to the formation of higher amounts of Ni^I species.

In Situ Formation of LNi⁰(cod) and L₂Ni⁰ Species from Ni(cod)₂

In the previous sections, we prepared Ni^I and Ni^II complexes containing our family of 6,6’-substituted bpy ligands. We were also interested in preparing Ni⁰ species, as these are often proposed as intermediates in XEC reactions.^5c Further, a common strategy in Ni-catalyzed reactions is to generate the active catalyst LNi⁰ in situ from the ancillary ligand and a Ni⁰ precursor, such as Ni(cod)₂ (cod = 1,5-cycloctadiene).^2d This allows for rapid screening of ligands in catalysis and avoids the synthesis of precatalysts that might be difficult to prepare and isolate. We treated Ni(cod)₂ with 1.1 equivalents of the different 6,6’-substituted ^tBubpy^R2 ligands with Ni(cod)₂ in THF and tracked the formation of ^tBubpy^R2 ligated Ni⁰ species using NMR spectroscopy after 18 hours at 50 °C (Figure 7). The unsubstituted ligand, ^tBubpy, reacts with Ni(cod)₂ to form a near quantitative amount of (^tBubpy)Ni(cod) and a trace amount of (^tBubpy)₂Ni. In contrast, the reaction between ^tBubpy^Me and Ni(cod)₂ gives only approximately 75% of (^tBubpy^Me)Ni(cod) and a small amount (~7%) of (^tBubpy^Me)₂Ni, indicating that a single methyl group makes a significant difference to the reaction outcome. Treatment of ^tBubpy^Me2 with Ni(cod)₂ results in an even larger perturbation in the product distribution and leads exclusively to (^tBubpy^Me2)₂Ni in approximately 33% yield, with no (^tBubpy^Me2)Ni(cod) observed. Control experiments indicate that the product distributions observed in the reactions of ^tBubpy^Me and ^tBubpy^Me2 with Ni(cod)₂ represent an equilibrium mixture and changing the amount of ligand added shifts the distribution (see SI Section H). In the cases of the more sterically bulky ligands ^tBubpy^iPr2, ^tBubpy^sBu2, ^tBubpy^Ph2, and ^tBubpy^Mes2 no formation of LNi(cod) or L₂Ni is observed upon reaction with Ni(cod)₂.

Figure 7:

Reaction between different bpy ligands and Ni⁰(cod)₂ in THF. ^aConversion was determined using ¹H NMR spectroscopy via the ratio of the target species to the total amount of LNi⁰(cod) + L₂Ni⁰ + Ni⁰(cod)₂.

Our results indicate that for nitrogen donor ligands, the displacement of a cod ligand in Ni(cod)₂ is exquisitely sensitive to steric bulk. As the steric bulk becomes greater, displacement of cod becomes less favorable, and the use of Ni(cod)₂ as a precursor for screening bulky nitrogen donor ligands in Ni-catalyzed reactions will likely give incorrect results about the relative activity of different ligands due to the incomplete formation of the ligated catalyst. We hypothesize that ^tBubpy^Me2 results in the formation of only the homoleptic complex (^tBubpy^Me2)₂Ni because ^tBubpy^Me2 is a worse donor to Ni than ^tBubpy, likely due to steric reasons. Consequently, with ^tBubpy^Me2, there is less back donation from Ni to cod to stabilize a complex of the type LNi(cod)₂ and the formation of a homoleptic complex becomes more favorable. The donor power of ^tBubpy^Me is likely in between ^tBubpy and ^tBubpy^Me2 and therefore in that case, we observe both (^tBubpy^Me)Ni(cod) and (^tBubpy^Me)₂Ni. For larger ^tBubpy^R2 ligands, displacement of cod is presumably thermodynamically unfavorable.

XEC Reactions using 6,6’-Substituted ^tBubpy Ni^II Dichloride Complexes

Ni/Co Dual-Catalyzed XEC Reactions

The results of our synthetic studies of Ni^I and Ni⁰ species highlight that there are differences in the properties of Ni complexes supported by different 6,6’-substituted ^tBubpy ligands. Given the prevalence of bpy supported Ni complexes in XEC reactions,^11d we evaluated our Ni^II complexes as precatalysts for several XEC reactions. Initially, we assessed the complexes in a Ni/Co dual catalytic coupling of methyl 4-bromobenzoate and 1-bromo-3-phenylpropane in the presence of a homogeneous reductant TDAE under the optimized conditions developed using (^tBubpy)NiCl₂ as the precatalyst (Figure 8).³⁴ We selected this reaction because the mechanism is well established, which allows for easier understanding of how changes to the ancillary ligand around Ni are affecting the elementary steps. Specifically, the Ni catalyst selectively activates the aryl halide, and an intermediate of the type L_nNi(Ar)X traps an alkyl radical, while the Co catalyst activates the alkyl halide to from an alkyl radical.

Figure 8:

Evaluation of the performance of our Ni^II precatalysts in a Ni/Co dual catalytic XEC reaction between an aryl and alkyl bromide. All yields are based on NMR conversion and are the average of two trials. Pc = phthalocyanine. Ar-Ar = biaryl side product from homocoupling of the aryl bromide. DMAc = N,N-Dimethylacetamide.

Despite the large difference in the steric properties of the 6,6’-substituted ^tBubpy ligands, nearly all the Ni^II precatalysts gave similar yields of product after 24 hours (>80%), which are only slightly lower than the quantitative yield obtained using (^tBubpy)NiCl₂ as the precatalyst. Surprisingly, the only precatalyst that gave a substantially different yield is (^tBubpy^Me)NiCl₂, which has the smallest deviation in ligand structure from (^tBubpy)NiCl₂. In this case, the XEC product was formed in only 46% yield, but a significant amount of biaryl product was also formed (~50%), suggesting that a catalytic intermediate of the form (^tBubpy^Me)Ni(Ar)Cl is decomposing before it is able to trap an alkyl radical generated by Co^II(Pc).³⁴

To gain further understanding about the differences in catalytic performance between our precatalysts, we monitored reactions using (^tBubpy)NiCl₂, (^tBubpy^Me)NiCl₂, (^tBubpy^Me2)NiCl₂, (^tBubpy^sBu2)NiCl₂, and (^tBubpy^Mes2)NiCl₂ at shorter reaction times (Figure 9). The unsubstituted complex, (^tBubpy)NiCl₂, gave near complete conversion after 4 hours (~95%), with no significant increase in product yield after this time. In contrast, precatalysts containing substituents on both the 6 and 6’-positions of the ^tBubpy ligand, (^tBubpy^Me2)NiCl₂, (^tBubpy^sBu2)NiCl₂, and (^tBubpy^Mes2)NiCl₂, required more than 4 hours to reach their maximum yields, although a substantial amount of product (~70%) was formed within 4 hours, indicating that the 6,6’-substituents on the ^tBubpy ligand only inhibit catalytic turnover to a small extent. In addition, the rate of product formation was comparable between all three 6,6’-substituted precatalysts, which suggests that increasing the steric bulk from methyl to mesityl has a negligible impact. However, when (^tBubpy^Me)NiCl₂ was used as the precatalyst, all the starting aryl bromide was consumed within 2 hours to give a mixture of product and biaryl in approximately a 1:1 ratio, demonstrating faster but less selective turnover. This is an important observation as a major problem for XEC reactions is the low TOF of most catalysts.

Figure 9:

Time course studies of (^tBubpy)NiCl₂, (^tBubpy^Me)NiCl₂, (^tBubpy^Me2)NiCl₂, (^tBubpy^sBu2)NiCl₂, and (^tBubpy^Mes2)NiCl₂ precatalysts for the coupling of methyl 4-bromobenzoate and 1-bromo-3-phenylpropane. a) Yield of product and b) amount of aryl bromide starting material present. The error on the data points is ±5%.

To leverage the higher TOF of (^tBubpy^Me)NiCl₂ as a precatalyst, we aimed to improve its selectivity for the cross-coupled product. Previously, we demonstrated that the Ni/Co dual catalytic system can be rationally optimized by varying the relative loadings of the Ni or Co catalyst.³⁴ If biaryl is observed, this suggests that radical capture by an intermediate of the form (^tBubpy^Me)Ni(Ar)Br is too slow (relative to decomposition of (^tBubpy^Me)Ni(Ar)Br), and the loading of the Co catalyst needs to be increased to accelerate the generation of alkyl radicals (alternatively the loading of the Ni catalyst can be reduced). At 80 °C, the yield of cross-coupled product could be significantly improved (>80%) either by decreasing loading of (^tBubpy^Me)NiCl₂ to 1 mol% or increasing the Co^II(Pc) loading to 2.5 mol% (Figure 10). More significantly, using 2.5 mol% of (^tBubpy^Me)NiCl₂ and 2.5 mol% of Co^II(Pc), (^tBubpy^Me)NiCl₂ gave a higher TOF than the unsubstituted precatalyst (^tBubpy)NiCl₂ at 40 °C (see SI Section J), and a 91% yield of product was obtained. In fact, (^tBubpy^Me)NiCl₂ could even be used as a precatalyst at room temperature with elongated reaction times (48 hours). To demonstrate the practicality of (^tBubpy^Me)NiCl₂ as a more active precatalyst, we performed a Ni/Co dual-catalyzed XEC reaction between 1-bromo-3-phenylpropane and four complex aryl halides that were once intermediates in drug discovery programs from the MSD Aryl Halide Informer Library³⁵ at room temperature (Figure 11). High yields were obtained despite the presence of diverse functional groups, such as esters, amines, and heteroarenes, and these are some of the first room temperature XEC reactions between an alkyl and aryl halide.

Figure 10:

Optimization of Ni/Co dual catalytic XEC reaction between aryl bromide and alkyl bromide using (^tBubpy^Me)NiCl₂ as the precatalyst. ^a48 h. All yields are based on NMR conversion and are the average of two trials. Ar-Ar = biaryl side product from homocoupling of the aryl bromide.

Figure 11:

Dual-catalyzed XEC reactions of medicinally relevant aryl halides with 1-bromo-3-phenylpropane using (^tBubpy^Me)NiCl₂ as the precatalyst at room temperature. All yields are based on NMR conversion and are the average of two trials.

We propose that the increased activity of (^tBubpy^Me)NiCl₂ compared to (^tBubpy)NiCl₂ is due either to faster generation of the (^tBubpy^Me)Ni(Ar)Br intermediate or its presence in greater concentration. This could be because of faster oxidative addition to (^tBubpy^Me)Ni⁰ than (^tBubpy)Ni⁰ or that there is less decomposition to inactive species in the case of the ^tBubpy^Me ligand. Given that sterically bulkier systems often undergo slower oxidative addition,^14g,24b,36 the latter explanation is likely more plausible. For the more sterically bulky systems, (^tBubpy^Me2)NiCl₂, (^tBubpy^sBu2)NiCl₂, and (^tBubpy^Mes2)NiCl₂, we propose the slower rate of catalytic turnover is related to slower oxidative addition of the aryl bromide, which controls the rate of the formation of the L_nNi^II(Ar)X intermediate. In principle, this could lead to lower yields in catalysis because the alkyl radical can decompose if the formation of L_nNi^II(Ar)X is too slow and there is no appropriate Ni complex to trap the radical. However, we hypothesize that in the Ni/Co dual catalytic system in DMAc, the alkyl radical can reversibly interact with Co^IIPc, forming a cage-rebound radical, which is significantly more stable than an out-of-cage free radical. As a result, despite the slower formation of the L_nNi^II(Ar)X intermediate with precatalysts containing 6,6’-substituted ^tBubpy ligands, the alkyl radical can still be trapped by L_nNi^II(Ar)X, which leads to similar catalytic outcomes across all systems with different ligands.

Ni-Catalyzed XEC Reactions with Alkyl Pyridinium Salts (Katritzky Salts)

Our catalytic results with the Ni/Co system showed that while alterations to the 6 and 6’ substituents on the ^tBubpy ligand of the precatalyst affect the TOF, the differences in product yield do not vary greatly between the systems. We hypothesize that this is in part due to the ability of Co^IIPc to reversibly stabilize an alkyl radical, which means that the concentration of free alkyl radicals is low and variations in the rate of elementary processes at Ni are less important. To examine the effect of the 6,6’-substituents on ^tBubpy ligated Ni precatalysts in a reaction where a free alkyl radical is proposed to be generated, we compared the performance of our precatalysts in a reductive coupling reaction between alkyl Katritzky salts and an aryl iodide.³⁷ Katritzky salts, which can be generated from alkyl amines and pyrylium cations,³⁸ have been used in a variety of XEC reactions as an alkyl radical precursor.³⁹ Upon single electron reduction, they are proposed to generate alkyl radicals through C–N bond homolysis,^39a,40 which is likely an irreversible process.

Initially, we evaluated all the precatalysts in the coupling of methyl 4-iodobenzoate and a benzylic alkyl Katritzky salt in the presence of the weak homogeneous reductant TME (E° = −0.85 V vs Fc⁺/Fc) under the optimized conditions previously described for (^tBubpy)NiCl₂ (Figure 12).³⁷ In this reaction, the use of TME as the reductant is crucial because reduction of the benzylic Katritzky salt is fast due to the relative stability of the benzyl radical. Only a weak reductant can generate radicals at a slow enough rate to match the rate of formation of the L_nNi^II(Ar)X intermediate, which is responsible for trapping the radical.³⁷ In our test reaction, apart from the unsubstituted complex, (^tBubpy)NiCl₂, only (^tBubpy^Me)NiCl₂ gave a relatively high yield of the product (~74%). All other precatalysts gave essentially negligible product yields, and there was a significant amount of aryl iodide still present at the end of the reaction. The observation of unreacted aryl iodide was unexpected because in the absence of a Katritzky salt, the unsubstituted complex (^tBubpy)NiCl₂ can be reduced by TME and perform a reductive homocoupling reaction of the aryl iodide to form biaryl (see SI Section L). Thus, even if a productive coupling reaction did not occur with the sterically bulky precatalysts, it was expected that after precatalyst activation, homocoupling of the aryl iodide would occur at a slower rate than the consumption of the Katritzky salt. Based on the lack of homocoupling product, we hypothesized that precatalyst activation is a problem for the 6,6’-substituted precatalysts with TME, and we were not forming the active Ni species. Therefore, we performed a reaction using (^tBubpy^Mes2)NiCl₂ as the precatalyst in the presence of the stronger reductant TDAE (E° = −1.11 V vs Fc⁺/Fc). In this case, negligible amount of product was formed, but we did observe an 83% yield of biaryl from homocoupling of the aryl iodide. This suggests that the precatalyst is activated, but the formation of the L_nNi^II(Ar)X intermediate or radical trapping by L_nNi^II(Ar)X is too slow relative to the formation and decomposition of an alkyl radical from the Katritzky salt. Consequently, all our alkyl radical undergoes deleterious side reactions (hydrogen atom abstraction from solvent or alkyl-alkyl coupling), and the Ni catalyst only homocouples the aryl iodide.

Figure 12:

Catalytic performance of Ni^II precatalysts in a Ni-catalyzed XEC reaction between methyl 4-iodobenzoate and a benzylic Katritzky salt. All yields are based on NMR conversion and are the average of two trials. ^aTDAE used as the reductant. ArI is unconverted starting material. Ar-Ar = biaryl side product from homocoupling of the aryl iodide.

To further explore the role of reductant strength and the different 6,6’-substituted precatalysts in XEC reactions with Katritzky salts, we changed the alkyl electrophile from a benzylic Katritzky salt to a primary alkyl Katritzky salt. In this case, a stronger reductant TPyE (E° = −1.32 V vs Fc⁺/Fc) is required to reduce the Katritzky salt (Figure 13).³⁷ Using the conditions previously optimized for (^tBubpy)NiCl₂, our 6,6’-substituted precatalysts gave much lower yields of the product. Instead, biaryl was generated in significant quantities for all precatalysts except (^tBubpy)NiCl₂. This suggests that for the sterically bulky systems, precatalyst activation is occurring but the rate of alkyl radical production is not matched with the rate of formation of the L_nNi^II(Ar)X intermediate or the rate of radical trapping by L_nNi^II(Ar)X. To confirm that our more sterically bulky systems are ineffective in systems where a free alkyl radical is generated, we screened our complexes in a standard Ni-catalyzed C(sp²)–C(sp³) XEC reaction using Zn⁰ as a reductant.⁴¹ In agreement with our hypothesis, the 6,6’-substituted Ni^II complexes gave significantly lower yields than the unsubstituted complex (^tBubpy)NiCl₂ (see SI Section M).

Figure 13:

Catalytic performance of the Ni^II precatalysts in a Ni-catalyzed XEC reaction between methyl 4-iodobenzoate and a primary Katritzky salt. All yields are based on NMR conversion and are the average of two trials. ArI is unconverted starting material. Ar-Ar = biaryl side product from homocoupling of the aryl iodide.

Two main conclusions can be drawn from the catalytic reactions between alkyl iodides and Katritzky salts with our family of Ni^II precatalysts. (i) The trends in precatalyst activation do not follow the reduction potentials that were determined by electrochemistry. For example, cyclic voltammetry indicates that it is easier to reduce both (^tBubpy^Me2)NiCl₂ and (^tBubpy^Mes2)NiCl₂ than (^tBubpy)NiCl₂, and yet with TME only (^tBubpy)NiCl₂ is activated. In fact, the activation of (^tBubpy)NiCl₂ is surprising given that TME has a reducing potential of −0.85 V, whereas the first reduction of (^tBubpy)NiCl₂ occurs at −1.84 V. This suggests that precipitation of a salt, such as [TME]⁺[Cl]⁻ could be crucial to precatalyst reduction. Alternatively, it may be that a very small amount of the Ni^II precatalyst is reduced to Ni^I, and then disproportionation occurs to generate the active Ni⁰ species (along with Ni^II). In this case, the rate of disproportionation may be more important than the initial reduction. Although the exact mechanism of precatalyst activation is not clear, our results demonstrate that small changes to the ligand structure affect activation, and more detailed studies on the pathway for activation are required to design ligands that will promote this key process in catalysis. The differences in precatalyst activation may be even more pronounced with heterogeneous reductants in which mass transfer is a greater issue than with our homogeneous reductants. (ii) Small changes to the 6 and 6’-substituents of the ^tBubpy ligand change the rates of elementary processes at Ni and cause large differences in product yield. This is exemplified by the reaction between a primary alkyl Katritzky salt and an aryl iodide in the presence of TPyE, where changing the precatalyst from (^tBubpy)NiCl₂ to (^tBubpy^Me)NiCl₂ results in a significant decrease in product yield, even though the reductant can activate both precatalysts. Understanding why small changes to the ancillary ligand can have such a dramatic impact on catalytic performance is crucial for the rational design of new ligands and is explored in the next section.

Stability and Reactivity Studies of (bpy)Ni^II(Ar)X Intermediates

In Ni/Co dual catalyzed XEC reactions, we propose that there are three main processes at Ni that are crucial to determining the overall catalytic performance of different 6,6’-^tBubpy ligated precatalysts after they are activated: (i) the rate of oxidative addition of the aryl halide to the (^tBubpy^R2)Ni⁰ species generated from the reduction of the Ni^II precatalyst, (ii) the rate of decomposition of the (^tBubpy^R2)Ni(Ar)X intermediate formed from oxidative addition to (^tBubpy^R2)Ni⁰, and (iii) the rate of radical capture by (^tBubpy^R2)Ni(Ar)X. Oxidative addition has been studied by many different groups on both Ni and other metals, and the results demonstrate that increasing the steric bulk of the ancillary ligand leads to slower oxidative addition.^14g,24b,36 Therefore, we decided to focus on understanding the stability of different (^tBubpy^R2)Ni(Ar)X complexes and the ability of these complexes to trap radicals.

Initially, we synthesized a series of different complexes of the type (^tBubpy^R2)Ni(Ar)I (Ar = o-tolyl, 4-F-o-tolyl, and 2,5-xylyl) with ^tBubpy and ^tBubpy^Me2 as ancillary ligands (Figure 14). These complexes were prepared by oxidative addition of an aryl iodide to an in situ generated (^tBubpy)Ni⁰(cod) or (^tBubpy^Me2)₂Ni⁰ species formed from the reaction between either ^tBubpy or ^tBubpy^Me2 and Ni(cod)₂. Due to the difficulty of displacing cod with bulkier ^tBubpy^R2 ligands (vide supra), reactions with other 6,6’-substituted ^tBubpy ligands did not afford the target complexes. In contrast, although 6-subsituted ligand ^tBubpy^Me reacts with Ni(cod)₂ to form (^tBubpy^Me)Ni⁰(cod), the subsequent oxidative addition reaction afforded a complex mixture of products. We propose that this is due to rapid decomposition of intermediates of the form (^tBubpy^Me)Ni(Ar)I (vide infra).

Figure 14:

Synthesis of (^tBubpy^R2)Ni(Ar)I complexes bearing different aryl groups and ^tBubpy and ^tBubpy^Me2 ancillary ligands.

With the unsubstituted ligand, ^tBubpy, all three of the (^tBubpy)Ni(Ar)I complexes we synthesized give diamagnetic ¹H NMR spectra, indicating that they are low-spin square planar d⁸ complexes. This is consistent with previous examples of Ni^II aryl halide complexes bearing bpy ligands but is in contrast with the observation that (^tBubpy)NiCl₂ is paramagnetic and hence a high spin tetrahedral d⁸ complex.⁴² Presumably, the change from a weak-field chloride ligand to a strong-field aryl ligand is sufficient to cause a change in the spin-state at Ni and indicates that different Ni^II intermediates in XEC reactions can exist in different spin states. The need for ancillary ligands that support Ni complexes in two different spin states on the same catalytic cycle, creates additional complexity for ligand design. A stability study using ¹H NMR spectroscopy showed that (^tBubpy)Ni(o-tol)I is relatively stable in solution under N_2, and there is no evidence of decomposition when it is left for 24 hours in C₆D₆ at room temperature.

The introduction of methyl substituents in the 6,6’-positions of the ^tBubpy ligand resulted in the formation of (^tBubpy^Me2)Ni(Ar)I complexes that gave paramagnetic NMR spectra. This is consistent with a high-spin d⁸ tetrahedral geometry at Ni. The change from low-spin to high-spin electron configuration is likely driven by steric factors as the introduction of methyl substituents in the 6,6’-positions disfavors the square planar geometry due to the increased steric congestion in the metal coordination plane. A similar tetrahedral geometry was observed previously for a (4,4’-Me₂-bpy)Ni(Ar)I complex with a very large aryl substituent,⁴³ and a recent computational study showed that a tetrahedral geometry is favored for Ni^II(Ar)X complexes bearing bpy ligands with 6-substituents.^12b Nonetheless, to our knowledge, our complexes are the first examples of isolated Ni^II aryl halide complexes supported by 6,6’-substituted bpy ligands. Given that methyl groups are the smallest substituents we introduced in the 6,6’-positions of our family of ^tBubpy ligands, it also suggests that our other 6,6’-^tBubpy ligands will result in high-spin tetrahedral intermediates of the type (^tBubpy^R2)Ni(Ar)X. It is well known in coordination chemistry that high-spin complexes are more reactive and less stable than low-spin complexes due to weaker metal-ligand bonding,^1e and this is also the case for complexes of the type (^tBubpy^Me2)Ni(Ar)I. A stability study showed a solution of (^tBubpy^Me2)Ni(o-tol)I in C₆D₆ under N₂ fully decomposed over 24 hours at room temperature to form (^tBubpy^Me2)NiI₂, free ^tBubpy^Me2, biaryl, and Ni black. We propose that the decomposition pathway involves ligand disproportionation of two equivalents of (^tBubpy^Me2)Ni(o-tol)I into (^tBubpy^Me2)NiI₂ and (^tBubpy^Me2)Ni(o-tol)₂, which has been observed for other bpy-ligated Ni(Ar)X complexes on a longer time scale.^29a,29e,44 (^tBubpy^Me2)Ni(o-tol)₂ then decomposes to generate biaryl, Ni black and free ligand. The decomposition of (^tBubpy^Me2)Ni(Ar)I was so significant that despite repeated attempts with different aryl species (see SI Section E for the synthesis of additional (^tBubpy^Me2)Ni(Ar)I complexes), we were unable to grow single crystals of (^tBubpy^Me2)Ni(Ar)I for X-ray analysis, but the NMR evidence is unequivocal that these are high-spin tetrahedral complexes. Further, we hypothesize that complexes of the type (^tBubpy^Me)Ni(Ar)I, containing a single methyl group in the 6-position of the ^tBubpy ligand, are also high-spin and as a consequence so unstable that we were not able to isolate them.

The fact that decomposition is faster when moving from (^tBubpy)Ni(Ar)I to either (^tBubpy^Me)Ni(Ar)I or (^tBubpy^Me2)Ni(Ar)I is counter to conventional wisdom for metal complexes (and what was observed for Ni^I species), which is that more sterically bulky ancillary ligands lead to increased stability (vide supra). Further, it was proposed computationally that a change in the geometry of the LNi^II(Ar)X intermediate from square planar to tetrahedral can result in lower selectivity of in XEC dure to formation of biaryl homocoupling products.^12b Indeed, our catalytic results using complexes of the type (^tBubpy^R2)NiCl₂ are consistent with faster decomposition of (^tBubpy^R2)Ni(Ar)I intermediates, as a large amount of biaryl was observed in all systems except the unsubstituted ^tBubpy complex (^tBubpy)NiCl₂ in catalytic reactions between aryl iodides and Katritzky salts (Figure 13). It also highlights a major difference in designing ligands for first-row metals, such as Ni, compared to second- and third-row metals, such as Pd, which is that potential changes in spin-state need to be factored into ligand design and that even small changes can lead to differences in spin state.⁴⁵

Even though the trapping of alkyl radicals by (^tBubpy^R2)Ni(Ar)I is commonly proposed in XEC, there are very few studies on this important elementary step.⁴⁶ We were interested in determining the relative rates of radical capture between complexes of the type (^tBubpy)Ni(Ar)I and (^tBubpy^Me2)Ni(Ar)I to probe if the presence of steric bulk (and the change in geometry) affects the rate of the process. Initially, we performed independent stochiometric radical capture experiments using (^tBubpy)Ni(o-tol)I and (^tBubpy^Me2)Ni(o-tol)I (Figure 15a). In these experiments a benzyl radical was generated by reducing Co^II(Pc) to an anionic Co^I(Pc) complex, which undergoes an S_N2 reaction with benzyl chloride to generate a Co^III benzyl species (Figure 15b).³⁴ The Co^III complex undergoes homolytic Co–C bond dissociation to generate a benzyl radical. The benzyl radical is then trapped by the Ni^II aryl halide complex to generate a Ni^III aryl alkyl species, which undergoes rapid reductive elimination to form 1-benzyl-2-methylbenzene and an unobserved Ni^I complex, which is proposed to decompose to generate free ligand and Ni black. (^tBubpy)Ni(o-tol)I and (^tBubpy^Me2)Ni(o-tol)I generated 1-benzyl-2-methylbenzene with comparable yields. This indicates that both complexes can capture radicals effectively, and that the relative stability of the two complexes does not impact the efficiency of the process on a timescale of 1 hour at room temperature.

Figure 15:

a) Stochiometric reactions of (^tBubpy)Ni(o-tol)I and (^tBubpy^Me2)Ni(o-tol)I with benzyl chloride in the presence of Co^II(Pc) and TDAE. b) Proposed pathway for radical trapping. Yields are based on NMR conversion and are the average of two trials.

To probe the relative rates of alkyl radical capture between (^tBubpy)Ni(Ar)I and (^tBubpy^Me2)Ni(Ar)I, we performed a competition experiment in which (^tBubpy)Ni(o-tol)I and (^tBubpy^Me2)Ni(2,5-xyl)I were exposed simultaneously to a sub-stoichiometric amount of benzyl chloride, Co^II(Pc), and TDAE (Figure 16a). Given that radical capture by Ni^II and subsequent reductive elimination from Ni^III is likely irreversible, we determined the relative rate of radical trapping by the Ni^II complexes through comparison of the relative amounts of 1-benzyl-2-methylbenzene (formed from trapping by (^tBubpy)Ni(o-tol)I) and 2-benzyl-1,4-dimethylbenzene (formed from trapping by (^tBubpy^Me2)Ni(2,5-xyl)Ar)I). In our initial experiment the ratio of 1-benzyl-2-methylbenzene to 2-benzyl-1,4-dimethylbenzene was ~1.5:1 indicating that radical trapping is faster by the less sterically bulky square planar complex (^tBubpy)Ni(o-tol)I. To ensure that the aryl substituents were not responsible for the differences in the relative rates of radical trapping, we completed the same experiment using (^tBubpy)Ni(2,5-xyl)I and (^tBubpy^Me2)Ni(o-tol)I in which the aryl substituents were reversed (Figure 16b).⁴⁷ In this case, the ratio of 2-benzyl-1,4-dimethylbenzene to 1-benzyl-2-methylbenzene was ~1.5:1 again indicating that trapping by the less sterically bulky system is faster. Similar results were also obtained using o-tol and 4-F-o-tol as the aryl substituents for (^tBubpy)Ni(Ar)I and (^tBubpy^Me2)Ni(Ar)I complexes (see SI Section P). Finally, given the ability of Co^II(Pc) to potentially reversibly trap a radical, we performed competitive radical trapping experiments between (^tBubpy)Ni(2,5-xyl)I and (^tBubpy^Me2)Ni(o-tol)I in which the alkyl radical was generated via reduction of a benzylic Katritzky salt (Figure 16c and 16d). In a similar fashion to our results with Co^II(Pc), the less sterically bulky square planar complex (^tBubpy)Ni(Ar)I still exhibits faster radical capture than (^tBubpy^Me2)Ni(Ar)I with the same product ratio. This suggests that the observed relative rates are intrinsic to the complexes and not related to the substrate activation pathway.

Figure 16:

Competition experiments to determine the relative rates of radical capture between (^tBubpy)Ni^II(Ar)I and (^tBubpy^Me2)Ni^II(Ar)I with radicals generated using a) and b) benzyl chloride, Co^II(Pc) and TDAE, and c) and d) benzylic Katritzky salts and TME. All yields are based on NMR conversions.

Our observation that (^tBubpy^Me2)Ni(Ar)I captures alkyl radicals more slowly than (^tBubpy)Ni(Ar)I is consistent with our catalytic results showing that (^tBubpy^Me2)NiCl₂ generates more of the unwanted biaryl product in reductive coupling reactions between primary alkyl Katritzky salts and methyl 4-iodobenzoate (Figure 13). However, the difference in catalytic performance between (^tBubpy^Me2)NiCl₂ and (^tBubpy)NiCl₂ is enormous, whereas the difference in the rates of radical trapping is relatively small. Therefore, we suggest that under the catalytic conditions (DMAc, 80 °C), the instability of (^tBubpy^Me2)Ni(Ar)I compared to (^tBubpy)Ni(Ar)I and the expected slower rate of oxidative addition to catalysts with more sterically bulky ligands are more significant. This is supported by the fact that in the Ni/Co dual catalyzed XEC reactions in which the alkyl radical is stabilized by Co^IIPc, the slower rate of radical capture by (^tBubpy^Me)Ni(Ar)X and (^tBubpy^R2)Ni(Ar)X does not appear to have a major impact on product formation (Figure 8). Finally, given that radical trapping occurs readily at room temperature, whereas most XEC reactions require elevated temperatures, there is a need to focus on improving the rates of other elementary steps to design systems with higher TOFs for XEC.

Conclusions

In this work, we have prepared a family of 6,6’-substituted ^tBubpy^R2 ligands and used them to synthesize complexes in the Ni⁰, Ni^I, and Ni^II oxidation states. The degree of steric bulk on the 6,6’-substituted ^tBubpy^R2 ligand has a profound impact on whether Ni complexes can be prepared and their stoichiometric and catalytic activity. For example, reaction of Ni(cod)₂ with free ^tBubpy^R2 to prepare complexes of the type (^tBubpy^R2)Ni(cod) only works for ligands with small 6,6’-substituents, which suggests that using Ni(cod)₂ and free ligand to evaluate ligands in Ni-catalyzed reactions will only work for a limited number of ligands and likely should be avoided. For complexes of the type (^tBubpy^R2)NiCl, we observe that as the steric bulk of the ligand increases, the Ni^I complexes become more stable. Conversely, for species of the type (^tBubpy^R2)Ni(Ar)I, we show that more sterically bulky ligands lead to less stable complexes. This is because as the steric bulk of the ligand increases, there is a change in the electronic configuration of (^tBubpy^R2)Ni(Ar)I from low spin to high spin. The observation that small changes to the ancillary ligand control whether proposed intermediates in the catalytic cycle are low spin or high spin is a unique feature in ligand design for first-row catalysts, such as Ni, that does not need to be considered for precious metals and may explain the lack of success to date in designing new ligands for Ni-catalyzed XEC reactions. Further, it is notable that when using state-of-the-art ^tBubpy as a ligand for XEC, the starting precatalyst (and in some cases catalytic intermediate) (^tBubpy)NiCl₂ is high spin, whereas intermediates of the type (^tBubpy)Ni(Ar)X (and likely (^tBubpy)Ni(Alkyl)X) are low spin. Thus, an effective ligand for XEC may need to be able to stabilize both low and high spin complexes.

In catalytic studies, we demonstrated that the 6-methyl substituted ^tBubpy ligand, ^tBubpy^Me, leads to a more active precatalyst for Ni/Co dual-catalyzed XEC reactions than a precatalyst stabilized by state-of-the-art ^tBubpy. We leveraged the increased activity of (^tBubpy^Me)NiCl₂ to perform Ni/Co dual catalytic XEC reactions at room temperature. In contrast, systems with 6,6’-substituted ^tBubpy ligands lead to less active precatalysts but are still able to facilitate the reaction. We propose that systems containing 6,6’-substituted ^tBubpy ligands are only active because the Co co-catalyst binds reversibly with the alkyl radical, and in systems where a less stable radical is generated, such as Ni-catalyzed XEC reactions between aryl halides and Katritzky salts, precatalysts containing 6,6’-substituted ^tBubpy ligands have either poor activity or are not active. Using radical trapping competition experiments, we established that the rate of alkyl radical trapping is slower by intermediates of the type (^tBubpy^R2)Ni(Ar)X compared to (^tBubpy)Ni(Ar)X. However, the slower rates of oxidative addition to systems containing ^tBubpy^R2 ligands and the lower stability of (^tBubpy^R2)Ni(Ar)X intermediates are more likely the major reasons for their inferior catalytic activity. These are the elementary processes which should be the focus of further research. Overall, we expect that the fundamental information we provide about how the 6,6’-substituents of ^tBubpy^R2 ligands impact the properties of Ni complexes and their activity in catalysis will be beneficial for the design of new ligands for Ni-catalyzed reactions. Additionally, our results suggest that different ligand design strategies compared to those that have been successful for precious metal systems will need to be applied because of the tendency of Ni complexes to adopt multiple spin states.