Comparison of Monophosphine and Bisphosphine Precatalysts for Ni-Catalyzed Suzuki–Miyaura Cross-Coupling: Understanding the Role of the Ligation State in Catalysis

Abstract

Practical advances in Ni-catalyzed Suzuki-Miyaura cross-coupling (SMC) have been limited by a lack of mechanistic understanding of phosphine ligand effects. While bisphosphines are commonly used in these methodologies, we have observed instances where monophosphines can provide comparable or higher levels of reactivity. Seeking to understand the role of ligation state in catalysis, we performed a head-to-head comparison study of C(sp²)–C(sp²) Ni SMCs catalyzed by mono and bisphosphine precatalysts using six distinct substrate pairings. Significant variation in optimal precatalyst was observed, with the monophosphine precatalyst tending to outperform the bisphosphines with electronically deactivated and sterically hindered substrates. Mechanistic experiments revealed a role for monoligated (P₁Ni) species in accelerating the fundamental organometallic steps of the catalytic cycle, while highlighting the need for bisligated (P₂Ni) species to avoid off-cycle reactivity and catalyst poisoning by heterocyclic motifs. These findings provide guidelines for ligand selection against challenging substrates and future ligand design tailored to the mechanistic demands of Ni-catalyzed SMCs.

Graphical Abstract

INTRODUCTION

Mechanistically-driven phosphine ligand design for transition metal-catalyzed cross-coupling reactions has enabled the development of robust methodologies for carbon–carbon and carbon–heteroatom bond formation under mild conditions with low catalyst loading.^1,2,3,4 In Suzuki-Miyaura cross-coupling reactions (SMCs), extensive studies of Pd-catalyzed methodologies have identified that unsaturated L₁Pd intermediates (where L is a monophosphine ligand) are generally necessary to promote the organometallic steps of catalysis.^5,6,7,8 These findings have prompted the design and implementation of electron-rich and sterically-bulky monophosphine ligands to preferentially form L₁Pd intermediates that promote efficient catalysis while protecting against off-cycle reactivity (Figure 1a).^9,10,11 While this design principle has been highly enabling for Pd, recent efforts to develop analogous Ni-catalyzed methodologies have revealed that these ligand design principles do not necessarily translate between the two metals.¹² Namely, the ligands that are most effective for Pd, including bulky trialkyl (e.g., P(t-Bu)₃, P(Ad)₃), triaryl (e.g., P(o-tol)₃), and highly-substituted dialkyl-biaryl phosphines (e.g., t-BuBrettPhos) are not similarly privileged in Ni-catalyzed SMCs.^13,4,14,15

Figure 1.

Phosphine ligands effects in Ni and Pd-catalyzed Suzuki-Miyaura cross-coupling (SMC). b. Monophosphines for Ni-SMCs. c. Role of catalyst ligation state. d. This work.

By contrast, bisphosphine ligands like 1,1’-bis(diphenylphosphino)ferrocene (dppf) are commonly used in Ni-catalyzed SMCs and have been incorporated into air-stable precatalysts that enable efficient C(sp²)–C(sp²) bond formation with (hetero)aryl substrates (Figure 1a).^16,17,18 A simple rationalization of the use of bisphosphines over monophosphines in Ni-catalyzed SMCs is that, by virtue of being able to occupy two coordination sites in an entropically-favorable chelate, these ligands can avoid the formation of vacant coordination sites that can enable unproductive reactivity.^13,19,20 Additionally, bulky bisphosphines, like those developed by Stradiotto and coworkers, can facilitate a more facile reductive elimination step.⁴

While these ligands have been enabling for Ni, previous work by our lab in developing the Ni-catalyzed SMC of benzylic acetals revealed an instance where both state-of-the-art bisphosphines and monophosphines were inactive or low-yielding, respectively, in catalysis (see SI for details).^12,13 Development of a series of monophosphines possessing remote steric bulk, quantified by high values of cone angle and low values of percent buried volume (%V_bur), enabled high yields for this transformation (Figure 1b). These ligands, collectively referred to as the DinoPhos ligands, have subsequently been demonstrated by our group and others to be top performers in other Ni-catalyzed SMCs.²¹ Seeking to understand the significance of this steric profile in Ni-catalyzed SMCs, we undertook a high-throughput experimentation (HTE) and ligand featurization campaign, in collaboration with the Sigman Laboratory and Merck Sharpe and Dohme. Univariate analysis of reaction yields against ligand steric features revealed a threshold in Ni-catalyzed SMC reactivity as a function of monophosphine minimum percent buried volume (%V_bur (min)),²² wherein yields above those attained under ligand-free conditions were only observed for ligands with values of %V_bur (min) below ~32%. We observed experimentally that this reactivity threshold is a readout of ligation state, with ligands below the threshold forming L₂Ni(0) species and those above forming L₁Ni(0).¹³ This analysis demonstrated that successful monophosphines for Ni catalysis must be able to form a bisligated Ni species. Extending our analysis to a series of Pd-catalyzed cross-coupling datasets, we again observed reactivity thresholds with opposite directionality, reflecting higher catalytic performance for monophosphines that form L₁Pd species (Figure 1c).

Given our demonstration that bisligated L₂Ni is required for reactivity, it is unsurprising that bisphosphines, in addition to monophosphines below the %V_bur (min) threshold (e.g., PCy₃), are commonly employed in Ni-catalyzed SMCs.^23,24,25 Interestingly, in our HTE studies, we found that the highest yielding ligands were those with %V_bur (min) values between 29 and 32%. This region corresponds to an area where ligands spectroscopically observed to form L₂ species were active in both Ni and Pd catalysis (Figure 1c, dotted vertical lines). We hypothesized that this overlap signifies a region where ligands thermodynamically favor L₂M but can kinetically form L₁M intermediates. Moreover, this would suggest that successful ligands for Ni-catalyzed SMCs must balance these two requirements, potentially to both minimize off-cycle reactivity while accelerating productive chemistry, respectively.

Considering this mechanistic possibility, we sought to compare mono- and bisphosphine performance in Ni-catalyzed SMCs to understand how ligand structure impacts catalyst ligation state and reactivity. For clarity, monoligated L₁Ni monophosphine and [κ¹-L]Ni bisphosphine complexes will be referred to as P₁Ni, while bisligated L₂Ni monophosphine and [κ²-L]Ni bisphosphine complexes will be referred to as P₂Ni. If successful catalysis requires balancing the need for both P₂Ni and P₁Ni intermediates, we hypothesized that these two classes of phosphines would have their respective strengths and weak-nesses which operate as a function of the specific demands of a given substrate pairing. To this end, we have compared the catalytic outcomes of three Ni precatalysts bearing different phosphine ligands—one monophosphine and two bisphosphines—in a series of Ni-catalyzed SMCs of aryl chlorides and aryl boronic acids. No one ligand was observed to be optimal across each substrate pairing surveyed. Further mechanistic experiments and computational studies enabled rationalization of the catalytic outcomes as a function of the required ligation state at different steps in the catalytic cycle (Figure 1d). Our study implicates P₁Ni species in catalysis—including for bisphosphine-ligated Ni—while also demonstrating the need for P₂Ni species as a way of minimizing unproductive side reactivity and catalyst poisoning and provides guidance for ligand selection as a function of substrate properties.

RESULTS

To select representative mono- and bisphosphine ligands, we screened a subset of phosphines in catalysis for two electronically-biased substrate pairings (Figure 2). Reaction 1 (1-chloro-4-trifluoromethylbenzene and 4-methoxybenzene boronic acid) represents an electronically “matched” case, where the electrophile has an inductively withdrawing trifluoromethyl group (CF₃; σ_p = 0.54) and the nucleophile has an electron-donating methoxy group (OMe; σ_p = −0.27).²⁶ By contrast, reaction 2 features the same functional groups but swapped between electrophilic and nucleophilic partners (4-chloroanisole and 4-trifluoromethylbenzeneboronic acid), which represents an electronically “mismatched” substrate pairing. For both reactions, we observed that CyTyrannoPhos (cyclohexylbis(3,5-di-tert-butylphenyl)phosphine) (Figure 2, entry 4) was the highest yielding monophosphine. Similarly, we have previously observed that this ligand was consistently high-yielding across a series of substrate pairings in our HTE evaluation of monophosphines in Ni-catalyzed SMCs (see Supporting Information for details). This ligand is commercially available and can be synthesized in a one-pot, two-step procedure from readily available precursors. For these reasons, it was selected as the representative monophosphine for further study.

Figure 2.

Initial mono- and bisphosphine ligand screen. Yields determined by GC analysis against a dodecane internal standard and are the average of two experimental runs. Reactions run at 0.05 mmol scale (1 mL THF). See Supporting Information for reaction procedures.

In selecting bisphosphines for comparison, we first selected dppf (1,1’-bis(diphenylphosphino)ferrocene) as the representative cis-binding ligand (Figure 2, entry 6). While it was not the highest-yielding bisphosphine, we thought it would provide a meaningful comparison to prior studies, given that it is commonly used as a benchmark for the performance of Ni-catalyzed SMCs²⁷ and in mechanistic studies of Ni-catalyzed cross-coupling.^17,19,28 While dppf was chosen to provide a meaningful benchmark relative to prior studies, we observed that a structurally similar trans-binding bisphosphine, dcypf (1,1’-bis(dicyclohexylphosphino)ferrocene), was a top-performing bisphosphine (Figure 2, entry 7).²⁹ Though this ligand has received less attention in SMC methodologies or mechanistic studies, it has been shown to surpass the performance of dppf in certain instances.^30,31 By directly comparing these two bisphosphines with a high-performing monophosphine, we sought to gain additional insights into how the distinct speciation of these two classes of ligand affects catalytic efficiency and broaden the scope of mechanistic knowledge for Ni-catalyzed SMCs.

For our comparison study, we opted to use L_nNi^II(o-tolyl)(Cl) precatalysts (1-3) for their ease of synthesis, air stability, and defined speciation. These precatalysts have been demonstrated to be viable in Ni-catalyzed SMCs, undergoing activation through transmetalation with excess boronic acid followed by reductive elimination to generate the active Ni⁰ intermediate.^13c,^32,33 Further, the use of this precatalyst platform obviates the need for air- and temperature-sensitive Ni(COD)₂.

Both (dcypf)Ni^II(o-tolyl)(Cl) (2) and (dppf)Ni^II(o-tolyl)(Cl) (3) were synthesized according to literature procedures disclosed by Jamison and coworkers.¹⁶ Structural characterization of these compounds demonstrates the contrasting cis and trans binding of the dppf and dcypf complexes, respectively. With a larger bite angle (dcypf: 144.0°; dppf: 102.0°) and higher percent buried volume (%V_bur) (dcypf: 59.8%; dppf: 56.7%) relative to dppf, we hypothesized that dcypf would be more labile in catalysis and represent an intermediate point of comparison between cis-binding dppf and the monophosphine, CyTyrannoPhos.^34,35 We successfully synthesized and isolated the corresponding CyTyrannoPhos₂Ni^II(o-tolyl)(Cl) precatalyst (1) for use in the catalytic reactions as the representative monophosphine (Scheme 1).¹⁶

Scheme 1.

Synthesis of monophosphine precatalyst 1.

The full set of substrate pairings were chosen to incorporate functionalities that would provide a readout of ligand performance in different catalytic scenarios (Figure 3c). Reactions 1 and 2 were revisited with the precatalysts to represent electronically-biased substrate pairings. Similarly, reaction 3 (4-chloroanisole and 4-ethoxycarbonylphenylboronic acid) has a similarly “mismatched” electronic profile (σ_p(CO₂Et) = 0.45),²⁶ but also features an ester functionality that can coordinate to Ni⁰ catalyst intermediates. Previous reports of Ni-catalyzed SMCs have demonstrated that substrates containing ester functional groups are lower yielding in catalysis^{21, 27}. Reactions 4 (2-chloro-meta-xylene and 4-fluoro-phenylboronic acid) and 5 (1-chloro-4-fluorobenzene and 2,4,6-trimethylphenylboronic acid) feature di-ortho substitution on the electrophile and nucleophile that we hypothesized would have slow oxidative addition and transmetalation steps, respectively, that affect the overall efficiency of catalysis and flux of catalytic intermediates. Lastly, reaction 6 (3-chloropyridine and 2-thienylboronic acid) was chosen to include synthetically relevant heterocyclic motifs, which can be catalytically challenging with Ni due to catalyst poisoning and relatively higher rates of nucleophile protodeboronation competing with productive chemistry.

Figure 3.

Comparison of precatalyst reactivity in Ni-catalyzed SMCs. a. General scheme for catalytic reactions studied. Reactions run at 0.05 mmol scale (1 mL THF). See Supporting Information for reaction procedures. b. Ni precatalysts used for comparison study. c. Substrate pairings used in Ni-catalyzed SMCs. The time of the first timepoint is shown next to each reaction label. Each plot shows the yields for each of the three precatalysts at the first time point (solid bar) and 24 h (dashed bar). Reactions 1-3 and 6 were analyzed by GC against a dodecane internal standard. Reactions 4 and 5 were analyzed by ¹⁹F NMR against a 1-fluoronaphthalene external standard. ^aRun with 4 mol% Ni and 5 equivalents boronic acid. ^bRun at 80 °C in 1,4-dioxane.

Each substrate pairing was tested against all three precatalysts from a common set of conditions, with minimal alterations made to facilitate sufficient yields for meaningful comparisons (Figure 3a). Yields were determined at two time points: (1) the time at which the highest yielding catalyst either reaches ~90% yield or stalls, as indicated for each reaction in Figure 3c, and (2) 24 hours. The second timepoint enabled us to differentiate between cases where one catalyst system was slower in forming product and those where one catalyst became inactive over the course of the reaction. The substrate pairings and yields, depicted graphically, are shown in Figure 3c.

As expected, given the electronically activated coupling partners, reaction 1 was high yielding for each precatalyst with a relatively short reaction time (6 h), confirming that these catalysts were competent under these reaction conditions. However, when the substituents were flipped (reaction 2), deactivating both substrates, 1 was the highest yielding (88%) after 9 hours, with 2 achieving similar yields with additional time while 3 reached a maximum yield of 17% over 24 hours. A similar trend was observed for reaction 3, though the reaction was overall slower and lower yielding for each catalyst, with 1 reaching 84% yield over 24 hours.

Considering the sterically hindered substrates, we observed that 1 was the highest yielding against the di-ortho-substituted electrophile (reaction 4, 69% after 12 h). When using 2,4,6-trimethylbenzeneboronic acid as the nucleophile (reaction 5), 2 was highest yielding at 12h (94%), while 1 required additional reaction time (78% at 12h; 95% at 24 h). Finally, when employing two heterocyclic coupling partners in reaction 6, we observed that 3 was most efficient, reaching nearly quantitative yield after 7 hours. Precatalyst 1 was lowest yielding for this reaction (49%, 24 h).

Overall, we observed that (i) no one precatalyst was uniformly optimal for all substrates and (ii) a precatalyst could be high yielding in one case and low yielding in another (e.g., precatalyst 3 with an 8% yield in reaction 4 and 99% yield in reaction 6). Further, we observed that 2 and 3 exhibited notably different performances in several of the reactions surveyed in spite of their common backbone motif. The lower yields for 3 in reactions 2 through 5 relative to both 1 and 2 prompted us to consider the role of ligation state in promoting these transformations; in particular, we sought to evaluate the intermediacy and role of P₁Ni in catalysis. While mechanistic studies of P₁M species have often focused on monophosphines in Pd SMCs, P₁M bisphosphine species have previously been proposed for both Pd and Ni-catalyzed processes.^36,37,38 Though our prior work demonstrated that formation of P₂Ni is essential for successful catalysis, likely to protect the Ni center against off-cycle reactivity and catalyst deactivation, we sought to determine if the transient formation of unsaturated P₁Ni intermediates plays a role in accelerating the oxidative addition and transmetalation steps of the catalytic cycle. To this end, we utilized reactions 4 through 6 as case studies, as they highlight cases where each of the three ligands is highest performing.

Sterically hindered oxidative addition

In reaction 4, which features a di-ortho substituted aryl chloride, we observed that the bisphosphine catalysts were significantly lower yielding than the monophosphines (vide supra).³⁹ Considering the potential role of ligation state in the oxidative addition step, we sought to monitor the rate of formation of the oxidative adduct of 2-chloro-meta-xylene and a (CyTyrannoPhos)₂Ni⁰ species in the presence of increasing equivalents of added ligand. Due to the complicated speciation of Ni(COD)₂ with CyTyrannoPhos, we elected to use a (CyTyrannoPhos)₂Ni(naphthalene) complex (4) synthesized by the reduction of the corresponding L₂NiCl₂ complex with magnesium in the presence of naphthalene.^40,41 High temperature ¹H NMR enabled us to track the formation of (CyTyrannoPhos)₂Ni(xylyl)(Cl) (5) in situ (Figure 4). As shown by the reaction profiles, the rate of oxidative addition is attenuated in the presence of increasing amounts of free ligand. The observed inverse order in ligand indicates that a ligand dissociation event occurs prior to oxidative addition, implicating a P₁Ni species in catalysis.⁴² This observation is consistent with proposed intermediates in Pd-catalyzed C(sp²)–Cl oxidative addition, where P₁Pd is invoked as the active species.^43,44,45

Figure 4.

Oxidative addition of 2-chloro-meta-xylene to 4 in the presence of added equivalents of ligand. Formation of 5 was monitored by ¹H NMR. See Supporting Information for synthesis of 4 and experimental details.

Having observed this inverse dependence on ligand in oxidative addition, we then computationally probed the relative energetics of mono- and bisligated oxidative addition pathways using density functional theory (DFT) to determine which pathway is more favorable for each of the three ligands surveyed. Oxidative addition transition states for P₁Ni and P₂Ni species of each ligand were identified (Figure 5a). Barriers were determined from the P₂Ni⁰(arene) (where arene = 2-chloro-meta-xylene) ground state modeled with η²-arene coordination. Both transition states were modeled as concerted oxidative addition processes into the C(sp²)–Cl bond. Bisphosphine P₁Ni structures were computed with κ¹ binding of the ligand, where the second arm is rotated away from the metal about the ferrocenyl backbone.

Figure 5.

Computational analysis of P₁Ni and P₂Ni oxidative addition pathways. a. Schematic of two pathways explored. The P₁Ni pathway for the monophosphine was calculated with one equivalent of free ligand. The P₁Ni pathway for the bisphosphines are represented with κ¹ binding of the ligand, where the curved dotted line represents the ferrocene backbone. b. Preferred oxidative addition pathway for each ligand. The free energies of activation were determined by DFT methods at the M06/def2-TZVP//B3LYP-D3/6-31G(d,p) [SDD for Ni and Fe] level of theory with a SMD solvation model (THF).

Comparing the calculated free energy of activation (ΔG^‡) for both the P₂Ni and P₁Ni transition states (Figure 5b), we identified that CyTyrannoPhos had a lower barrier for the monoligated pathway, while dcypf and dppf exhibited lower barriers for their bisligated pathways. Interestingly, we also identified that the P₁Ni⁰ ground state species is lower in energy than the corresponding P₂Ni⁰ species for CyTyrannoPhos, while dcypf and dppf were determined to have lower energy P₂Ni⁰ species (Table S25).

Taken together, the experimental and computational results provide some insight into the observed discrepancies in reactivity between the two ligand classes for reaction 4. We hypothesize that the increased product yields with the monophosphine precatalyst can be attributed in part to oxidative addition through an energetically accessible P₁Ni pathway. The lower catalytic yields seen for the bisphosphines imply that oxidative addition along the P₂Ni pathway is less efficient, allowing for decomposition of the Ni⁰ intermediates and comproportionation of Ni⁰ with Ni^II species to generate catalyticallyinactive Ni^I.¹⁷ Previous studies of Ni-catalyzed oxidative addition that invoke P₁Ni pathways involve cleavage of C(sp²)–O bonds of phenol derivatives, like carbonates, esters, and sulfamates, that stabilize the unsaturated Ni center through secondary interactions with the substrate leaving group.^46,47,48,49 For example, the Houk group proposed that selective C(sp²)–O activation of an aryl ester by a (PCy₃)Ni catalyst proceeds via a monoligated transition state in which the carbonyl oxygen of the ester is coordinated to Ni.⁵⁰ The results presented here, however, also suggest that P₁Ni species are relevant in oxidative addition processes for which secondary interactions with the electrophile leaving group are not present and contribute to overall catalytic performance of a ligand system. By this analysis, we also anticipated that monophosphines with less-favorable ligand dissociation would be similarly low-yielding with this substrate (vide infra). Moreover, for sterically-challenging oxidative additions, labile monophosphine systems would be expected to outperform more rigid bisphosphines.

Sterically hindered transmetalation

We next shifted our attention to reaction 5, in which the nucleophilic coupling partner has di-ortho substitution. Here, we observed that 2 was the highest yielding precatalyst at 12 hours; precatalyst 1 was also high yielding, though required additional reaction time to achieve comparable yields. The low yields associated with 3 prompted us to probe the stoichiometry of the catalyst during transmetalation.⁵¹ Using CyTyrannoPhos as our model system, we applied the method of continuous variation (MCV) to reaction 5 (Figure 6a).⁵² Holding the overall mole percent of Ni and ligand constant, the mole fraction (χ), or ratio of ligand to metal, was incrementally varied and the initial rates of product formation were collected at each mole fraction. As shown in Figure 6a, we found that the fastest rate of product formation occurred at χ = 0.5, corresponding to 1:1 monophosphine to Ni stoichiometry at the turnover limiting step. Given that the oxidative addition of this electrophile occurs rapidly at room temperature (see Supporting Information), transmetalation is proposed to be turnover limiting for this substrate pairing and thus the step for which 1:1 Ni:phosphine stoichiometry (P₁Ni) is required.^48,53

Figure 6.

a. Job plot analysis of reaction 5. Rates were determined by initial rates, monitoring product formation by ¹⁹F NMR as a function of time against a 2-fluorobiphenyl internal standard. Runs performed in duplicate. See Supporting Information for full procedures. b. Pre-transmetalation intermediates. Ligand dissociation is required to enable the two-point binding of the hydroxyboronate species, which is supported by the Job plot analysis.

These results suggest that ligand dissociation to form an P₁Ni species is necessary for transmetalation, and thus ligand systems with greater lability will accelerate this step. This finding agrees with stoichiometric transmetalation studies performed by Grimaud and coworkers showing kinetic inhibition for biaryl formation from a (PPh₃)₂Ni^II oxidative adduct and phenylboronic acid in the presence of added PPh₃.²⁰ The proposed pre-transmetalation intermediate for both Ni and Pd catalysis exhibits two-point binding of a hydroxyboronate moiety to form a four-coordinate metal center bearing a single phosphine (Figure 6b).^6,7 For bisphosphine ligands, formation of this monoligated intermediate requires dissociation of one ligand arm to form the corresponding P₁Ni^II species. Thus, a bisphosphine with increased lability is expected to be more efficient in the transmetalation step. While 3 evidently can undergo transmetalation, as it is a competent ligand in Ni SMCs, we hypothesize that the inefficiency of transmetalation to this sterically-hindered nucleophile makes the catalyst vulnerable to off-cycle processes, as evidenced by the formation of protodehalogenated substrate (2% fluorobenzene after 12 h).¹⁷

The kinetic data, in combination with the observed similarities in the reactivity profiles of 1 and 2 (vide supra), suggests that dcypf exhibits greater lability in catalysis to enable efficient access to P₁Ni^II intermediates. To probe the role of bisphosphine flexibility in transmetalation, we tested a series of Ni precatalysts bearing 1,2-bis(diphenylphosphino)alkyl ligands with increasing alkyl linker chain lengths in catalysis (Figure 7). Increasing the alkyl linker length increases the conformational flexibility of the ligand and destabilizes the resulting chelate to make dissociation of one arm more favorable, which we hypothesized would enable higher yields. The corresponding Ni^II(2,4,6-trimethylphenyl)(Br) precatalysts (dppe (6); dppp (7); dppb (8), Fig. 7), as well as the corresponding Ni(COD)₂ and ligand catalyst mixtures, were subjected to reaction 5.⁵⁴ Though these precatalysts were not high yielding for reaction 5, we observed that increased linker length resulted in higher yields. In particular, the yield increased significantly between the cis-bound 7 to trans-bound 8 (3% yield versus 24% yield). Taken together, these experiments suggest that increased backbone flexibility is beneficial for catalysis as it enables the formation of an unsaturated P₁Ni center that can more efficiently undergo transmetalation.

Figure 7.

Reaction 5 yields as a function of bisphosphine linker length. Yields determined by ¹⁹F NMR analysis against a 1-fluoronaphthalene external standard. Unless otherwise indicated, conditions were identical to those in Figure 3. See Supporting Information for synthesis and characterization of the precatalyst complexes.

While our kinetic analysis with (CyTyrannoPhos)_nNi invokes a P₁Ni species, precatalyst 2 was found to be the most efficient precatalyst in reaction 5. To rationalize this observation, we considered off-cycle reactivity that impacts the overall efficiency of the monophosphine system by looking at side product formation. We observed greater amounts of mesitylene, the product of protodeboronation of the nucleophile, being formed by the monophosphine precatalyst 1 as compared to the reaction using 2 (Table 1). By contrast, protodehalogenation of the electrophile was not observed with 1, while only trace protodehalogenated electrophile was observed with 2.

Table 1.

Protodeboronation side products formed in reaction 5 for reactions catalyzed by 1 and 2. Mesitylene side product quantified by GC analysis against dodecane internal standard. Remaining boronic acid is not visible by GC after workup. See Supporting Information for full details.

		Yield (%)
Precatalyst	Time (h)	Product	Mesitylene
	12	78	109
CyTyrannoPhos (1)	18	86	139
	24	95	143
	12	90	25
dcypf (2)	18	96	29
	24	96	37

Under common SMC conditions using stoichiometric base and water, base-mediated protodeboronation can occur, particularly for a less-efficient catalyst that is consuming substrate less rapidly and for less stable boronic acids.⁵⁵ If base-mediated protodeboronation was the sole mechanism for formation of mesitylene in the reaction, one would expect that, after forming nearly quantitative yields of product after 12 hours, the reaction employing 2 would still incur significant protodeboronation of the remaining ~4 equiv of substrate over the next 24 hours. However, as shown in Table 1, the overall amounts of mesitylene formed in the presence of 2 remain low even after 90% yield is attained at 12 hours. These results suggest that catalyst-mediated protodeboronation occurs, and that its occurrence is a function of ligand. We hypothesize that protodeboronation occurs to a greater extent with 1 due to the formation of a greater amount of coordinatively unsaturated Ni in solution.⁵⁶ By contrast, while dcypf can access a P₁Ni species to generate a highly reactive catalyst, the proximity of the second arm of the phosphine can shield the Ni center from unproductive chemistry.

Reactivity of monophosphines as a function of steric features

The role of P₁Ni species identified in the previous two case studies helps to rationalize why precatalyst 1 is high-yielding in reactions 4 and 5. However, within monophosphine chemical space, we suspected that ligand steric properties also contribute to the ability to access P₁Ni species. Indeed, in our HTE study of monophosphines in Ni SMCs, we observed low reactivity for monophosphines with %V_bur (min) values below ~28%. Moreover, computational analysis of the ΔG of phosphine dissociation for monophosphine P₂Ni(benzaldehyde) complexes indicated that ligand dissociation becomes increasingly favorable with increasing values of %V_bur (min).¹³

This prompted us to consider how monophosphine steric features impact reactivity in reactions 4 and 5, given the need to form unsaturated P₁Ni species in catalysis. For this comparison, we utilized a series of four monophosphine precatalysts, including 1, that share a common mixed alkyl-aryl scaffold but vary in the magnitude of both their calculated %V_bur (min) and cone angle features. Both MePPh₂ (precatalyst 9) and CyPPh₂ (precatalyst 10) have unsubstituted aryl groups, while MeTyrannoPhos (methyl(3,5-di-tert-butylphenyl)phosphine) (precatalyst 11) and CyTyrannoPhos (precatalyst 1) include bulky t-Bu groups that increase their measured cone angles (Figure 8). Within each pairing in this series, we observed lower yields for the precatalyst with a smaller %V_bur (min). Overall, however, both ligands possessing greater steric bulk, as quantified by their larger cone angles, were higher yielding across the three reactions surveyed, with precatalyst 1 the highest yielding in the series.

Figure 8.

a. Monophosphine reactivity in reactions 4 and 5. Ligand steric features were determined by computational methods as detailed in reference 22.²² The cone angle value shown is the Boltzmann average value. Yields determined by ¹⁹F NMR analysis against a 1-fluoronaphthalene external standard. Unless otherwise indicated, conditions were identical to those in Figure 3. b. Ligand exchange with CyTyrannoPhos and MeTyrannoPhos. Progress of each species was monitored by ³¹P NMR. Performed in C₆D₆ (0.007M) at room temperature. ^aRun using Ni(COD)₂ and CyPPh₂

Given the proposed role of P₁Ni species in catalysis, we hypothesize that the improved performance of 1 over similar monophosphine-ligated precatalysts can, in part, be attributed to its steric profile that enables more facile formation of unsaturated intermediates in catalysis. The higher %V_bur (min) of CyTyrannoPhos (27.7%) as compared to MeTyrannoPhos (25.3%), for example, captures its ability to dissociate from P₂Ni species more readily in catalysis. Indeed, when we combine a mixture of 1 with 2 equivalents of MeTyrannoPhos, we see full conversion of 1 in 12 hours to form both the mixed ligand species and 11 (Figure 8b). By contrast, the reverse process occurs very slowly, only partially converting complex 11 to the corresponding mixed complex, suggesting that dissociation of the lower %V_bur (min) phosphine is less thermodynamically and kinetically favorable.

To gain insight into the mechanism of ligand dissociation from 1, we performed ³¹P 2D EXSY experiments with increasing equivalents of added CyTyrannoPhos. Analysis of degenerate ligand exchange between 1 and free CyTyrannoPhos (1.0, 5.0, and 10.0 equivalents) revealed that the rate of ligand exchange does not change with increasing concentration of free ligand (Figure 9, see Figures S20–S22). This supports a dissociative mechanism of ligand exchange, consistent with the hypothesis that 1 is substitutionally labile and enables formation of P₁Ni intermediates during catalysis.⁵⁷

Figure 9.

³¹P 2D EXSY of (CyTyrannoPhos)₂Ni(o-tolyl)(Cl) (1) in toluene-d8 at 60 °C (t_m = 2.5 s). The peak at 23.7 ppm corresponds with 1, with the free ligand at −0.95.

Additionally, the higher yields seen for 11 and 1 as compared to their undecorated analogues (9 and 10, respectively) indicate that the remote steric bulk described by cone angle is beneficial for catalysis, potentially through mediating ligand dissociation dynamics or stabilizing catalytic intermediates. Overall, these observations reinforce that %V_bur (min) is the dominant feature controlling ligation state and the favorability of ligand dissociation.

Reactivity of heteroaryl substrates

Though dppf-ligated precatalyst 3 was lower yielding for reactions 1 through 5, we identified it to be the top performer in reaction 6 (3-chloropyridine with 2-thienylboronic acid), reaching 95% yield in 7 hours. By contrast, this was the lowest yielding reaction for 1 (40% at 7h; 49% over 24 h). The use of heteroaromatic coupling partners in Ni catalysis can be challenging owing to the ability of these substrates to bind to and poison the catalyst. Further, while Ni-catalyzed SMCs commonly require elevated temperatures and stoichiometric water and base, heteroaromatic boronic acids are susceptible to protodeboronation under these conditions.⁵⁸ Hartwig and coworkers demonstrated that a (dppf)Ni(cinnamyl)(Cl) precatalyst enabled the synthesis of several hetero-biaryl compounds under relatively mild conditions (0.5 mol% catalyst loading at 50 °C).⁵⁹ Given this precedent, the observed high yields with 3 in reaction 6 were unsurprising and prompted further investigation into the origin of the increased activity of the bisphosphine precatalysts relative to the monophosphine precatalyst.

We first identified whether one or both coupling partners were low-yielding with CyTyrannoPhos-ligated precatalyst 1 by cross-coupling each with a non-heteroaromatic substrate. Using 3-chloropyridine as the electrophile (reaction 7, Scheme 2), all three precatalysts achieved near quantitative yields. By contrast, we found that the 1 was low-yielding in the reaction of 2-thiophenylboronic acid with 1-chloro-4-fluorobenzene (reaction 8, Figure 10, entry 1), while both 2 and 3 again achieved high yields (Table S14). Considering the possibility of protodeboronation liberating free thiophene in the reaction mixture, we tested catalysis in the presence of 1 and 2 equivalents of added thiophene (Figure 10, entries 2 and 3), which resulted in decreased yields. Increasing the equivalents of boronic acid to 5 equivalents (Figure 10, entry 4) further attenuated the reaction yield. In fact, we identified that the highest yields for precatalyst 1 were observed when the boronic acid loading was lowered to 1.0 equivalent (Figure 10, entry 5). Lastly, we considered the possibility of the product acting as a catalyst poison by adding one equivalent of 2-phenylthiophene (entry 6), which showed no impact on product yield. By contrast, the bisphosphine precatalysts were unaffected by these additives in reaction 8 (Table S14).

Scheme 2.

Precatalyst reactivity with 3-chloropyridine^a

^a Yields determined by ¹⁹F NMR against a 1-fluoronaphthalene external standard and presented as the average of two runs. Unless otherwise indicated, conditions were identical to those in Figure 3.

Figure 10.

Reactivity of 1 with 2-thiophenylboronic acid. Yields determined by ¹⁹F NMR analysis against a 1-fluoronaphthalene external standard. Unless otherwise indicated, conditions were identical to those in Figure 3. See Supporting Information for full experimental details.

These results suggest that the formation of thiophene by substrate protodeboronation is detrimental to catalysis for the monophosphine system. Attempts to directly observe and quantify the formation of thiophene were unsuccessful due to the instability of the boronic acid. While the protodeboronation data from reaction 5 suggests that a catalyst-dependent process contributes to thiophene formation, we are unable to distinguish this pathway from the known base-catalyzed protodeboronation of this substrate.^{27, 58} Possible mechanisms for catalyst deactivation with thiophene include catalyst trapping by association of the thiophene π-system to a Ni(0) species and oxidative insertion of the Ni catalyst into the C–S bond.^60,61 We hypothesize that the increased lability of the monophosphine-ligated complex results in an increased concentration of P₁Ni species that are susceptible to deactivation pathways, while the bisphosphine systems remain at P₂Ni sufficiently to protect the catalyst. Nonetheless, we have observed cases (e.g. reaction 2) where all three precatalysts are lower yielding in the presence of a thiophene additive (Table S16), suggesting that overall catalytic efficiency influences how susceptible a given catalyst is to deactivation.

Given the synthetic importance of heterocyclic motifs, we sought to determine if 1 is more broadly incompatible with thiophene-containing substrates by surveying a series of related examples (Figure 11). We found that 2-chlorothiophene (reaction 9) was successfully cross-coupled in 90% yield, achieving the identical product to reaction 8 with the opposite coupling partners. Interestingly, both 2 and 3 were low yielding in this instance (Table S17). Similarly, the 3-thiophenyl boronic acid substrate was cross-coupled successfully (reaction 10, 80% yield), in part owing to the increased stability of this substrate as compared to the 2-substituted boronic acid. Cross-coupling with the corresponding 2-furanylboronic acid substrate was also high-yielding with 1 (reaction 11), reaching 98% yield when coupled with 1-chloro-4-fluorobenzene and exceeding the yields attained by both 2 and 3 (Table S18). Notably, each of these substrates is less susceptible to byproduct formation relative to 2-thiophenylboronic acid, enabling successful catalysis with the more labile monophosphine system.⁵⁸ These studies highlight the additional considerations necessary for catalysis with substrates that can generate a catalyst poison, wherein the overall lability of the complex must be accounted for relative to potential catalyst deactivation pathways.

Figure 11.

Reactivity of 1 with heteroaromatic substrates. Yields determined by ¹⁹F NMR against 1-fluoronaphthalene as an external standard and presented as the average of two runs. Unless otherwise indicated, conditions were identical to those in Figure 3.

CONCLUSION

We have compared the catalytic performance of mono- and bisphosphine Ni precatalysts in SMCs of diverse substrate pairings. Catalytic, kinetic, and computational data demonstrate that P₁Ni species are responsible for enabling challenging oxidative addition and transmetalation steps, while P₂Ni species help to prevent off-cycle reactivity and catalyst poisoning. For reactions involving electronically deactivated or sterically hindered substrates, the use of a monophosphine like CyTyrannoPhos is advantageous for generating highly reactive P₁Ni intermediates. To enable ligand dissociation, monophosphines should have a %V_bur (min) that approaches but does not exceed the ligation state threshold (~32%). In reactions employing substrates that are subject to protodefunctionalization or can serve as a catalyst poison, bisphosphines provide a synthetic advantage by preferentially forming chelated P₂Ni species.⁶² Use of a wider bite angle bisphosphine, like dcypf, can furnish a more active catalyst against electronically or sterically challenging substrates.

These findings are summarized in Figure 12 to establish guidelines for selecting a ligand class in a range of synthetic applications. Additional considerations regarding the electronic and steric properties of a given ligand contribute to its overall effectiveness and should be included in ligand design and selection. As defined by our study, the ideal phosphine ligand for Ni-catalyzed SMCs and related reactions proceeding through similar organometallic intermediates must balance the ability to form both mono- and bisligated intermediates during catalysis. Ultimately, this study provides guidance for ligand selection as a function of substrate properties and informs the development of new ligands for Ni-catalyzed SMCs.

Figure 12.

Guidelines for ligand selection across different substrate classes.