Isotope Effects Reveal the Catalytic Mechanism of the Archetypical Suzuki-Miyaura Reaction

Abstract

Experimental and theoretical ¹³C kinetic isotope effects (KIEs) are utilized to obtain atomistic insight into the catalytic mechanism of the Pd(PPh₃)₄-catalyzed Suzuki-Miyaura reaction of aryl halides and aryl boronic acids. Under catalytic conditions, we establish that oxidative addition of aryl bromides occurs to a 12-electron monoligated palladium complex (Pd-(PPh₃)). This is based on the congruence of the experimental KIE for the carbon attached to bromine (KIE_C–Br = 1.020) and predicted KIE_C–Br for the transition state for oxidative addition to the Pd(PPh₃) complex (1.021). For aryl iodides, the near-unity KIE_C–I of ~1.003 suggests that the first irreversible step in the catalytic cycle precedes oxidative addition and is likely the binding of the iodoarene to Pd(PPh₃). Our results suggest that the commonly proposed oxidative addition to the 14-electron Pd(PPh₃)₂ complex can occur only in the presence of excess added ligand or under stoichiometric conditions; in both cases, experimental KIE_C–Br of 1.031 is measured, which is identical to the predicted KIE_C–Br for the transition state for oxidative addition to the Pd(PPh₃)₂ complex (1.031). The transmetalation step, under catalytic conditions, is shown to proceed via a tetracoordinate boronate (8B4) intermediate with a Pd–O–B linkage based on the agreement between an experimental KIE for the carbon atom involved in transmetalation (KIE_C-Boron = 1.035) and a predicted KIE_C-Boron for the 8B4 transmetalation transition state (1.034).

Graphical Abstract

INTRODUCTION

Palladium-catalyzed cross-coupling reactions, between an organic halide and an organometallic reagent, that forge carbon–carbon bonds constitute one of the central pillars of modern-day organic synthesis.¹ Three sequential mechanistic events define the general catalytic cycle of most palladium-catalyzed cross-coupling reactions: (a) oxidative addition (OA) of the organic halide to palladium(0) to form a palladium(II) complex; (b) transmetalation (TM), i.e., transfer of the organic portion of the organometallic reagent to the palladium(II) center; and (c) reductive elimination (RE) to form the new carbon–carbon bond and regenerate the palladium(0) catalyst.² High-resolution insight into the fine details of these fundamental steps is critical to the optimization of existing methodologies and development of novel catalytic cross-coupling reactions.

The palladium-catalyzed Suzuki–Miyaura (S-M) reaction between an organic halide (1) and a boronic acid (2) has emerged as a versatile and robust cross-coupling process with broad application in the synthesis of high-value pharmaceuticals and fine chemicals (Scheme 1).^3–6 The privileged status of the S-M reaction has inspired several high-quality experimental and theoretical investigations into the mechanism, leading to a constantly evolving view of this catalytic process.^7–16 OA has been extensively studied by a number of research groups,^17–26 and the general consensus is that for simple phosphine ligands such as triphenylphosphine OA of 1 occurs to a palladium(0) center ligated to two phosphine ligands (PdL₂) via PdL₂-OA_TS (Scheme 2) to yield a PdL₂ArX OA (4) complex.¹⁷ Alternatively, experimental^20,27 and computational studies^24,25 have also suggested the possible involvement of a coordinatively unsaturated 12-electron PdL₁ complex as the active catalytic species in OA (Scheme 2, PdL₁-OA_TS).

Scheme 1.

Widely Accepted Mechanism of the Prototypical Suzuki–Miyaura Reaction of Aryl Halide and Aryl Boronic Acid Catalyzed by a Palladium(0) Catalyst

Scheme 2.

Key Questions Regarding the Catalytic Mechanism of the S-M Reaction Addressed in This Work Using ¹³C KIEs

Elegant kinetic studies by Carrow and Hartwig, involving stoichiometric reactions of isolated OA complexes, have revealed that 4 is likely converted to a hydroxopalladium species 5 prior to the TM step by reaction with aqueous hydroxide generated in the THF-water reaction mixture containing an inorganic base such as K₂CO₃.²⁸ Groundbreaking low-temperature NMR studies by Thomas and Denmark have shown that association of 2 and 5 leads to the formation of two key pre-TM intermediates containing Pd–O–B linkages—a tetracoordinate boronate (8B4) complex and a tricoordinate boronic acid (6B3) complex.¹³ However, since both intermediates readily convert to product 3 through TM (via 8B4-TM_TS or 6B3-TM_TS, Scheme 2) and subsequent RE upon warming to room temperature, the exact identity of the pre-TM intermediate in the catalytic reaction remains ambiguous. Additional kinetic and computational studies by Denmark and co-workers suggest that (a) both 8B4-TM_TS or 6B3-TM_TS are viable TM pathways depending on the concentration of ligand in solution and (b) the TM event occurs only after dissociation of a ligand from these pre-TM intermediates.¹⁴ The final step in the catalytic cycle, C–C bond-forming RE, is extremely facile and thought to be accelerated by hydroxide ions in solution.²⁹

The currently accepted catalytic cycle for the S-M reaction (vide supra) is based predominantly on mechanistic observations from stoichiometric reactions of putative intermediate complexes.^{13,17–21,28} This approach is necessitated due to the low concentration of these intermediates under typical catalytic conditions (<2 mol % palladium catalyst). Utilization of mechanistic observations from stoichiometric studies to interpret the catalytic mechanism is valid only if the intermediate complexes form under catalytic conditions. Kinetic studies can often provide valuable mechanistic information under catalytic conditions,^8,9 but this does not include atomistic information about the transition state (TS) of the elementary steps. Computational studies provide detailed insight into the TS of each step of the catalytic S-M reaction.^{10,11,23,24,29–32} However, experimental validation of the calculated TSs in the catalytic S-M reaction has remained elusive.

A powerful experimental probe of transition structures in carbon–carbon bond-forming reactions is the determination of ¹³C kinetic isotope effects (KIEs) at natural abundance developed by Singleton and co-workers.^33–35 Despite this, there are limited examples of the use of ¹³C KIEs among the vast literature of mechanistic studies of cross-coupling reactions.^36–44 Given that each step in the proposed catalytic cycle of the S-M reaction involves bonding changes at a carbon atom, determination of ¹³C KIEs for the two bond-forming carbon atoms (e.g., KIE_C-X and KIE_C-Boron, Scheme 2) can provide vital insight into the transition structures of key steps under standard (catalytic) conditions. The typical range of ¹³C KIE values for carbon atoms involving bond-formation/bond-breaking at the KIE-determining step is 1.010–1.060. The Singleton method allows for the determination of ¹³C KIE value with high precision (typically ±0.003). When used in conjunction with predicted ¹³C KIEs obtained from DFT calculations, experimental ¹³C KIEs can be used to quantitatively distinguish between subtly different TS geometries.⁴⁵ We describe herein a combined experimental and theoretical ¹³C KIE study of the S-M reaction that provides unprecedented insight into the catalytic mechanism of this important reaction. Importantly, we find that ¹³C KIEs are an exceptionally sensitive probe of TSs involving bond-making or bond-breaking between carbon and palladium—enabling the first experimental characterization of the transition structure of the OA and TM steps under catalytic conditions.

RESULTS AND DISCUSSION

Mechanism of OA of Aryl Bromides in the Catalytic S-M Reaction.

Oxidative addition of aryl bromides to palladium(0) is highly exergonic and, barring some examples,⁴⁶ is generally considered an irreversible step in the catalytic cycle. Therefore, ¹³C KIEs determined for the aryl bromide in the S-M reaction should reflect the nature of the OA TS. We chose the prototypical S-M reaction of aryl bromide 1a with boronic acid 2a catalyzed by Pd(PPh₃)₄ (Figure 1) for determination of ¹³C KIEs for 1a via quantitative NMR analysis. Under these conditions, we observe a significant normal ¹³C KIE on the carbon atom attached to the bromine (KIE_C–Br = 1.020(1), Figure 1). This result is qualitatively consistent with OA being the first irreversible step in the catalytic cycle and provides an experimental benchmark to probe the exact nature of the OA transition state using density functional theory (DFT) calculations.

Figure 1.

Experimental ¹³C KIEs determined for aryl bromide under standard catalytic conditions. Numbers in parentheses represent the 95% confidence range in the last digit of each experimental KIE as determined from 12 measurements from two independent experiments.

Key bond-distances and predicted KIEs for all transition structures presented in this study are displayed as an average from calculations implemented using 12 DFT methods routinely employed to study these systems (see Supporting Information for computational details). This ensures that the theoretical evaluation of our experimental results is not an artifact of one particular DFT method. For the quantitative interpretation of KIE_C–Br, we calculated the TS for OA of 1a to both monoligated and diligated palladium(0) complexes. The two transition structures PdL₁-OA_TS-Br and PdL₂-OA_TS-Br are similar in terms of bond-making and bond-breaking distances (Figure 2). The Pd–C bond-forming distance in the mono-ligated PdL₁-OA_TS-Br is 2.01 Å compared to a slightly longer Pd–C bond-forming distance of 2.12 Å in the more electron-rich diligated PdL₂-OA_TS-Br. Intriguingly, KIE_C–Br is exceptionally sensitive to the slightly different Pd–C distance in the two TSs; the predicted KIE_C–Br for PdL₁-OA_TS-Br is 1.021 compared to a predicted KIE_C–Br of 1.031 for PdL₂-OA_TS-Br (Figure 2). The shorter C–Pd distance in PdL₁-OA_TS-Br dampens the imaginary mode associated with the reaction coordinate (~133i) relative to PdL₂-OA_TS-Br (~181i), resulting in the lower predicted KIE_C–Br. Finally, the greater extent of C–Pd bond-formation in PdL₁-OA_TS-Br also corresponds to a more advanced C–Br bond-breaking distance of 2.27 Å (C–Br bond-breaking distance in PdL₂-OA_TS-Br is 2.20 Å). Based on the excellent agreement of experimental and predicted KIEs for PdL₁-OA_TS-Br, our results suggest that OA in the catalytic S-M reaction occurs to a PdL₁ species. Importantly, this result quantitatively rules out PdL₂ as the active species that undergoes OA in the catalytic reaction.

Figure 2.

Calculated TSs and predicted ¹³C KIEs for OA of 1a to mono- and diligated palladium(0) with key bond distances shown in angstroms (Å).

To further support our finding that OA occurs to PdL₁ in the catalytic reaction, we conducted three additional experiments. The first experiment involves determination of KIE_C-Br for the stoichiometric OA of 1a to Pd(PPh₃)₄ to form the well-characterized PdL₂(Ar₁)(Br) complex 4a (Figure 3A).⁴⁷ This stoichiometric experiment delivers an experimental benchmark KIE_C-Br value to unambiguously characterize PdL₂-OA_TS-Br.⁴⁸ Gratifyingly, the KIE_C-Br for this reaction is 1.031(2), a value consistent with the predicted KIE_C-Br for PdL₂-OA_TS-Br of 1.031. Importantly, this result provides unequivocal evidence that OA in the Pd(PPh₃)₄ catalyzed S-M reaction occurs via different mechanisms under catalytic (Figure 1) versus stoichiometric (Figure 3A) conditions. In the second experiment, the catalytic S-M reaction of 1a and 2a was performed with added PPh₃ ligand (Figure 3B). Excess ligand is expected to decrease the probability of OA occurring to a PdL₁ complex by driving the PdL₁/PdL₂ equilibrium toward the PdL₂ species. Consistent with our hypothesis, the KIE_C-Br for this reaction is 1.031(1), a value consistent with the predicted KIE_C-Br for PdL₂-OA_TS-Br. The reaction with excess ligand is significantly more sluggish, suggesting that OA to PdL₁ is a faster process than OA to PdL₂, an observation that is consistent with prior computational studies.^24,25,27 Finally, we performed the catalytic S-M reaction of 1a and 2a using the isolated complex 4a as the catalyst (Figure 3B). Observation of an experimental KIE_C-Br of 1.021(1) illustrates that OA occurs via PdL₁-OA_TS-Br even when a PdL₂ complex is explicitly used as the catalyst for the S-M reaction.

Figure 3.

(A) Stoichiometric experiment to establish a benchmark ¹³C KIE value for OA to PdL2 (B) ¹³C KIE experiments to probe the effect of ligand concentration on the OA transition state. Numbers in parentheses represent the 95% confidence range in the last digit of each experimental KIE as determined from 12 measurements from two independent experiments.

Mechanism of OA of Aryl Iodides in the Catalytic S-M Reaction.

Next, we turned our attention to studying the S-M reaction of aryl iodide 1b and boronic acid 2a. We determined KIE_C-I under standard catalytic conditions similar to the conditions used for the bromo derivative 1a (Figure 1). The first key result is the near-unity KIE_C–I of 1.003(1) (Figure 4A, 0 mol % added PPh₃). This suggests that there is no bonding change at this carbon in the first irreversible step of the catalytic cycle, a result qualitatively inconsistent with OA as the first irreversible step for 1b. Oxidative addition of aryl iodides to palladium(0) is generally considered an irreversible process. Therefore, the absence of a significant normal KIE_C–I suggests that the first irreversible step for 1b likely precedes OA, i.e., binding of 1b to palladium(0) is irreversible and has a higher barrier than the ensuing OA step.

Figure 4.

(A) ¹³C KIE experiment to probe the OA TS as a function of ligand concentration for aryl iodides. Each experimental KIE_C-I is an average of 12 measurements from two independent experiments, and numbers in parentheses represent the 95% confidence range in the last digit of each experimental KIE. (B) Predicted ¹³C KIEs for all relevant TSs involved in the OA of 1b under these conditions with key bond distances shown in angstroms (Å).

For the quantitative interpretation of the KIE_C-I, we modeled the OA step of 1b to both PdL₁ (PdL₁-OA_TS-I) and PdL₂ (PdL₂-OA_TS-I) complexes (Figure 4B) using the theoretical methods described previously. Predicted KIE_C-I for these two TSs are 1.013 and 1.023, both of which are inconsistent with the experimental KIE_C-I value. Next, we located PdL₁-bind_TS-I, the transition structure for formation of the η² binding complex between PdL₁ and 1b. The predicted KIE_C–I for PdL₁-bind_TS-I is 1.004, which is in excellent agreement with the experimental KIE_C-I. No such TS could be located for PdL₂ (IRC calculations from PdL₂-OA_TS-I result in separate starting materials), consistent with prior investigations suggesting that OA of aryl halides to PdL₂ occurs without a preceding binding event for monodentate phosphine ligands.^20,24,25 These results lend strong support to OA of 1b to palladium(0) occurring via irreversible binding of 1b to PdL₁ followed by a facile OA event.

To further confirm that the PdL₁ pathway is operative for 1b, we determined KIE_C-I in the presence of 5 mol % added PPh₃ (Figure 4A). Unlike in the case of aryl bromide 1a, under conditions of excess ligand, we did not observe a KIE_C-I corresponding to PdL₂-OA_TS-I; instead, we observed an experimental KIE_C-I of 1.008(1) – a measurement greater than the predicted KIE_C-I of 1.004 for PdL₁-bind_TS-I (the KIE-determining step in the PdL₁ pathway) but significantly smaller than the predicted KIE_C-I of 1.023 for PdL₂-OA_TS-I (likely the KIE-determining step in the PdL₂ pathway). This result can be interpreted as being consistent with ~85% of the reaction proceeding via the PdL₁ pathway and ~15% via the PdL₂ pathway. Increasing ligand concentrations to 20 and 40 mol % resulted in KIE_C-I values of 1.016(1) and 1.022(4), respectively, reflecting the increasing contribution of the PdL₂ pathway in the catalytic reaction (Figure 4A).

Collectively, these results provide support for OA of both aryl bromides and iodides occurring to a 12-electron PdL₁ species under standard catalytic conditions. This OA step is preceded by the formation of an η² binding complex between PdL₁ and the aryl halide. The key difference between aryl bromides and aryl iodides is the relative energies of the binding versus the OA step. Experimental KIE_C-Br values (Figure 1) suggest that formation of the η² binding complex has a lower barrier than the OA step for aryl bromides. In contrast, the weaker C–I bond results in the lowering of the OA barrier relative to that of the binding step.²³ The PdL₂ pathway is operational only in the presence of excess PPh₃ ligand or under stoichiometric conditions.

Mechanism of TM in the Catalytic S-M Reaction.

Having established the catalytic mechanism of OA, we turned our attention to the investigation of the TM step under catalytic conditions. In particular, we questioned whether we could distinguish between aryl transfer from the 8B4 (8B4-TM_TS) versus the 6B3 (6B3-TM_TS) pre-TM intermediates by determining the ¹³C KIE on the carbon atom of 2a (Scheme 2, KIE_C–Boron) that migrates from boron to palladium during the TM event.⁴⁹ We speculated that KIE_C–Boron will be different for these two TSs due to the difference in nucleophilicity of the aryl group in the respective intermediates. We chose the prototypical S-M reaction of aryl iodide 1b and boronic acid 2a catalyzed by Pd(PPh₃)₄ for determination of KIE_C–Boron. Observation of a significant KIE_C–Boron of 1.035(1) (Figure 5A) is qualitatively consistent with TM being the first irreversible step in the catalytic cycle for 2a.

Figure 5.

(A) ¹³C KIE experiments to probe the TM TS in the catalytic S-M reaction of 1b and 2a. Experimental ¹³C KIEs are shown as an average of 12 measurements from two independent experiments, and numbers in parentheses represent the 95% confidence range for the last digit of the experimental KIE values. (B) Calculated TSs and predicted ¹³C KIEs for possible TM TSs with key bond distances shown in angstroms (Å).

For the quantitative interpretation of this KIE, we modeled the intramolecular TM event (Figure 5B) from both the tetracoordinate boronate complex (8B4-TM_TS) and the tricoordinate boronic acid complex (6B3-TM_TS). In 8B4-TM_TS, the forming Pd–C bond distance (r_Pd-C) is 2.18 Å and the breaking B–C bond (r_B-C) is 2.05 Å. The corresponding distances in 6B3-TM_TS are 2.07 and 2.35 Å. Therefore, TM from the 8B4 intermediate has an earlier transition state than the 6B3 intermediate, consistent with the more nucleophilic migrating aryl group in 8B4. We also observed a clear correlation between r_Pd-C in the transition structures and KIE_C–Boron—increased proximity of the migrating carbon atom to the palladium center in 6B3-TM_TS (compared to 8B4-TM_TS) results in a lower predicted KIE_C–Boron (Figure 5B). The average KIE_C–Boron from 12 DFT predictions for 6B3-TM_TS is 1.024, a value that is in variance with the experimental KIE of 1.035(1). On the other hand, the predicted KIE_C–Boron for 8B4-TM_TS is 1.034, which is in excellent agreement with experimental KIE_C–Boron. On a side note, we also explored the intermolecular TM TS for the transfer of the aryl group from an aryltrihydroxyborate to the PdL₁(Ar₁)(Br) OA complex (inter-TM_TS). While the predicted KIE_C–Boron for this TS is in reasonable agreement with experiment, it is 7.0 kcal/mol higher in energy than 8B4-TM_TS, suggesting that inter-TM_TS is likely not the operative TM pathway.²⁸ The quantitative match of experimental (~1.035) and predicted (~1.034) KIE_C-Boron for 8B4-TM_TS provides the first quantitative evidence for the transition structure of the TM step in the S-M reaction under catalytic conditions. We obtained identical KIE_C-Boron (1.035(2)) when the reaction was performed with 1a and in the presence of 5 mol % added triphenylphosphine, suggesting that the TM step is unaffected by these modifications to reaction conditions (see Table S10 in the Supporting Information for details of these experiments).

CONCLUSIONS

In conclusion, we have utilized a combination of experimental and theoretical ¹³C KIEs to delineate the fine details of the catalytic mechanism of the Pd(PPh₃)₄-catalyzed Suzuki–Miyaura reaction of aryl halides and aryl boronic acids. Our studies provide the first experimentally validated transition structures for both the OA (PdL₁-OA_TS-Br/PdL₁-bind_TS-I) and TM (8B4-TM_TS) steps in the catalytic reaction, leading to a more detailed description of the catalytic cycle (Figure 6). The reaction is initiated by dissociation of two triphenylphosphine ligands (L) from PdL₄ to form the 14-electron PdL₂ complex. During the first turnover, our experimental KIE_C-Br from the stoichiometric OA reaction (Figure 3A) suggests that PdL₂ is likely the palladium(0) species that undergoes OA to the aryl halide to deliver the palladium(II) complex 4_L2.⁵⁰ Displacement of the halide ligand by hydroxide likely results in the hydroxo-palladium complex 5_L2, which presumably undergoes ligand dissociation and coordination to 2a to form the pre-TM 8B4 intermediate. Our experimental KIE_C-Boron supports transmetalation occurring from 8B4 resulting in the pre-RE intermediate 6_L1. Subsequent facile RE delivers the cross-coupled biaryl 3 and generates the highly reactive 12-electron PdL₁ complex.

Figure 6.

Catalytic mechanism of the Suzuki–Miyaura reaction as determined herein using a combination of ¹³C KIEs and DFT calculations. Highlighted in red are the key OA and TM steps in the catalytic cycle that are validated by this study.

After the first turnover, the mechanism of subsequent catalytic cycles depends on the nature of the aryl halide and the concentration of the phosphine ligand in solution. Experimental KIE_C-Br and KIE_C-I determined under standard catalytic conditions (1 mol % Pd(PPh₃)₄ and no added PPh₃) support OA occurring to the PdL₁ complex via a two-step process: (i) formation of an η² binding complex between PdL₁ and the aryl halide and (ii) OA of PdL₁ into the C–X bond to form the palladium(II) complex 4_L1. For aryl iodides, we have shown that the binding event is the first irreversible step, whereas aryl bromides proceed via reversible binding followed by OA as the first irreversible step. Displacement of the halide from 4_L1 by hydroxide results in the mono-ligated hydroxopalladium complex 5_L1, which coordinates to the aryl boronic acid to directly yield the pre-TM intermediate 8B4. Subsequent TM and RE completes the catalytic cycle and regenerates the 12-electron PdL₁ complex.

Using the archetypical S-M reaction of aryl halides and aryl boronic acids catalyzed by Pd(PPh₃)₄, we have illustrated the combined use of experimental and theoretical ¹³C KIEs as a highly sensitive probe to evaluate the TS geometry of key steps in palladium-catalyzed cross-coupling reactions. We show that this approach can distinguish between subtly different mechanistic pathways without the need to isolate highly reactive putative reaction intermediates. The highlight of this work is the ability to identify the ligation state of palladium in the “active catalytic species” based on the magnitude of ¹³C KIEs at the center undergoing oxidative addition. We anticipate that this unprecedented insight into the S-M reaction under catalytic conditions will inspire similar investigations of long-standing mechanistic questions in other areas of transition-metal catalysis and provide a valuable complement to existing approaches that interrogate these important mechanistic questions in cross-coupling reactions.