Pre-transmetalation intermediates in the Suzuki-Miyaura reaction revealed: The missing link

See allHide authors and affiliations

Science  15 Apr 2016:
Vol. 352, Issue 6283, pp. 329-332
DOI: 10.1126/science.aad6981

Where Pd and B meet in Suzuki coupling

The Suzuki-Miyaura reaction is widely used to form carbon-carbon bonds. It operates by transferring a carbon center from boron to palladium, although the handoff itself happens too quickly to see. Thomas and Denmark have caught sight of the previously elusive intermediate complexes linking palladium to boron through an intervening oxygen. Using low-temperature nuclear magnetic resonance spectroscopy, they characterized the intermediate structures just before the carbon transfer event.

Science, this issue p. 329


Despite the widespread application of Suzuki-Miyaura cross-coupling to forge carbon-carbon bonds, the structure of the reactive intermediates underlying the key transmetalation step from the boron reagent to the palladium catalyst remains uncertain. Here we report the use of low-temperature rapid injection nuclear magnetic resonance spectroscopy and kinetic studies to generate, observe, and characterize these previously elusive complexes. Specifically, this work establishes the identity of three different species containing palladium-oxygen-boron linkages, a tricoordinate boronic acid complex, and two tetracoordinate boronate complexes with 2:1 and 1:1 stoichiometry with respect to palladium. All of these species transfer their boron-bearing aryl groups to a coordinatively unsaturated palladium center in the critical transmetalation event.

Palladium-catalyzed cross-coupling reactions have fundamentally changed the practice of organic synthesis. These reactions forge carbon-carbon bonds through the migration of a carbon-based substituent from a main group element to palladium, as exemplified by the Kumada-Tamao-Corriu (Mg) (1), Suzuki-Miyaura (B) (2), Stille-Migita-Kosugi (Sn) (3), Negishi (Zn) (4), and Hiyama-Denmark (Si) (5) reactions. The Nobel Prize–sharing Suzuki-Miyaura reaction (6) is currently the premier cross-coupling process and has been widely applied in organic (7), medicinal (8), and materials (9) chemistry. It is also frequently used in the industrial syntheses of fine chemicals (10) and pharmaceuticals (11) because of its demonstrated reliability, its functional group compatibility, and the low cost and ease of handling of a wide variety of commercially available boron-based reagents.

Despite the preeminent status of the Suzuki-Miyaura reaction, a fundamental understanding of the critical migratory transmetalation event from boron to palladium has thus far been lacking (1218). For decades, chemists have considered two pathways (path A and path B in Fig. 1) that differ in the role that the hydroxide ion plays in initiating the transmetalation event. Path A proceeds through the combination of a negatively charged aryltrihydroxyboronate (compound 1) and a palladium halide complex (2), which form a hypothetical intermediate containing a Pd-O-B unit (3). The alternative path B proceeds through the combination of a neutral arylboronic acid (4) and a palladium hydroxide complex (formed through the displacement of the organopalladium halide by hydroxide; 5), ultimately converging to the same intermediate 3, which is then poised to transfer the aryl group to palladium in an intramolecular β-aryl elimination step, followed by reductive elimination (Fig. 1). Species such as 3 represent the missing link between the starting organoboron reagents and the diorganopalladium intermediates that are known to afford cross-coupling products.

Fig. 1 Palladium-catalyzed cross-coupling reactions and proposed transmetalation pathways in the Suzuki-Miyaura process.

R indicates an organic group. Compound numbers are shown in bold in the lower panel.

The role of the base in the Suzuki-Miyaura reaction was investigated initially by the Soderquist laboratory (12) and, more recently, by the laboratories of Hartwig (13), Amatore and Jutand (1416), and Schmidt (17, 18). The kinetic analysis in (13) established that path B is favored over path A by more than four orders of magnitude, a conclusion that is reinforced by the extensive kinetic studies in (1416), which clearly identified multiple antagonistic roles for the hydroxide ion. Although nuclear magnetic resonance (NMR) spectroscopic and kinetic studies have provided independent evidence for these pathways, the actual composition and structure of the transmetalation precursors have not been unambiguously defined, despite being widely assumed. Because of the transient nature of these intermediates, traditional methods (e.g., electrospray mass spectrometry and traditional NMR spectroscopy) have proven incapable of characterizing highly reactive intermediates such as 3 (19, 20). Although the intermediacy of a species containing a Pd-O-B linkage has been proposed, its observation and characterization have eluded chemists for over 30 years (21). A recent review by Lennox and Lloyd-Jones (22) states that “[t]he barrier of this process was predicted computationally to be low (14–22 kcal mol−1), suggesting specialist techniques will need to be applied to detect and confirm the identity of [3] experimentally.” One such technique that has proven valuable for providing structural and kinetic data in similar mechanistic studies is rapid injection NMR (RI-NMR) (23).

Aided by the RI-NMR apparatus developed in our laboratories (24), we have undertaken the generation and structural and kinetic characterization of these elusive intermediates. We hypothesized that combining stoichiometric amounts of arylpalladium complex trans-(i-Pr3P)2(4-FC6H4)Pd(OH) (6) with 4-fluorophenylboronic acid (7) (Fig. 2, route 1), or trans-(i-Pr3P)2(4-FC6H4)Pd(I) (8) with thallium 4-fluorophenylboronate (10) (Fig. 2, route 2), should converge on a species whose structure and kinetic competence can be examined. The choice of triisopropylphosphine (i-Pr3P) was critical to allow the preparation of discrete, stable precursors and also to facilitate structural assignments (25, 26).

Fig. 2 Formation of a 6-B-3 complex containing a Pd-O-B linkage.

The crossing lines for compound 12 signify a cyclic trimer. In the middle panel, the red bonds indicate HMBC cross peaks. Throughout, h indicates hours; equiv, equivalents; conv., conversion; quantitative, 100% conv.

The synthesis in route 1 involved the addition of a tetrahydrofuran-d8 (THF-d8) solution of 4-fluorophenylboronic acid (7) to a THF-d8 solution of trans-(i-Pr3P)2(4-FC6H4)Pd(OH) (6), together with 2.0 equivalents of i-Pr3P, at −78°C (Fig. 2). Although no new species were observed at −60°C, warming the solution to −30°C resulted in the quantitative conversion of compounds 6 and 7 to a new species. The combination of one- and two-dimensional NMR spectroscopic techniques executed at −30°C led to the structural elucidation of the newly formed species as complex 11, containing a Pd-O-B linkage (supplementary materials, figs. S1 to S10).

The coordination geometry was assigned to a trans-bisphosphino square planar palladium complex. This assignment was based on the observation of the 13C NMR signal (PCH) at 25.38 parts per million (ppm) as an apparent triplet (JP-C = 10 Hz; J is the coupling constant between the phosphorus and carbon atoms) attributable to virtual coupling (27) and the 31P NMR signal at 29.98 ppm (a solitary singlet), which is shifted slightly upfield compared with the corresponding resonance of 6 at 33.00 ppm (figs. S2 and S4).

The bonding connectivity of complex 11 was established by the observation of strong through-space interactions [nuclear Overhauser effect (NOE) spectroscopy] of both Hb and Hd (hydrogens in the b and d positions) with the methyl hydrogens on the i-Pr3P group (Fig. 2, blue arrows, and fig. S10). In addition, cross peaks between the B-OH group and the ipso-carbon [C(1)]–bearing boron (3JBOH-C(1)) were observed in the heteronuclear multiple-bond correlation (HMBC) spectrum (Fig. 2, red bonds, and fig. S9). The resonances for Ha and Hb (Fig. 2, blue aryl) and Hc and Hd (green aryl) are shifted slightly downfield (+0.07 to +0.16 ppm) in complex 11, compared with substrates 6 and 7 (table S1). The resonances for the fluorine atoms in 11 [Fa (blue aryl) and Fb (green aryl)] are both shifted upfield relative to those associated with 6 and 7, but the change was much more pronounced for Fb (–4.54 to –0.91 ppm), which facilitated their identification.

Most importantly, the boron atom in complex 11 was assigned to a tricoordinate geometry [6-B-3 (28)] on the basis of the 11B NMR signal, which appeared as a broad singlet at 29 ppm (table S1)—well within the chemical shift regime for 6-B-3 boron compounds (29). Related 6-B-3 complexes of arylboronic and diarylborinic acids with Pt and Rh have been synthesized and exhibit 11B NMR resonances similar to that of 11 (30, 31).

To provide additional evidence for the structure of 11, an independent synthesis was carried out (route 2). This synthesis involved combining trans-(i-Pr3P)2(4-FC6H4)Pd(I) (8) with 3.0 equivalents of thallium 4-fluorophenylboronate (10) in the presence of dibenzo-22-crown-6 (32), together with 1.0 equivalent of i-Pr3P in THF, at −78°C; this was followed by warming to −30°C in the NMR spectrometer (Fig. 2). A small amount of conversion (~10%) to complex 11 was observed, together with cross-coupling product 13 (~30%), by 31P and 19F NMR spectroscopy, demonstrating that intermediate 11 can be formed without the intermediacy of arylpalladium hydroxide complexes (33).

To support the assertion that the boron atom in 11 is tricoordinate (6-B-3), a second independent synthesis of 11 was undertaken (route 3). A solution of arylpalladium hydroxide complex 6 and 4-fluorophenylboroxine 12 (0.33 equivalents) in THF-d8 was combined with 2.0 equivalents of i-Pr3P at −78°C in an NMR tube, which was quickly inserted into the NMR spectrometer that had been pre-cooled to −60°C, whereupon complex 11 was observed (Fig. 2, route 3). A ~50% conversion to 11 was observed at −60°C over 36 hours, along with cross-coupling product 13. The similarity of the spectroscopic data (including the NOE spectroscopy cross peaks and the 11B NMR chemical shifts; figs. S13 to S21) for the species generated from the three independent syntheses provides compelling support for the structural assignment of 11 as a 6-B-3 palladium(II) complex containing a Pd-O-B linkage.

The formation of 6-B-3 complex 11 must proceed via an 8-B-4 complex (such as 3; Fig. 1) that is formed initially which then suffers by the rapid loss of a molecule of water. We attempted to shift the equilibrium toward such a complex by generating 11 in mixtures of THF and H2O (99:1), but we observed no change in the 31P, 19F, or 11B NMR spectra. We considered exploring more strongly coordinating hydroxide sources that are often used in Suzuki-Miyaura reactions, but the addition of inorganic bases to THF-H2O blends is known to form biphasic mixtures (13). Fortunately, CsOH•H2O dissolves readily at −30°C in mixtures of THF and CH3OH (12:1). A freshly generated sample of 6-B-3 complex 11 (from 6 and 7; vide supra) at 0.034 M in THF was cooled to −78°C; this was followed by the addition of 50 μl of a 2 M solution (5.0 equivalents) of CsOH•H2O in methanol. The sample was monitored by NMR spectroscopy at −30°C, but again, no change in the 19F, 31P, or 11B NMR spectra was observed.

The resistance of the boron atom in complex 11 to adopt a tetracoordinate geometry is probably caused by the steric hindrance that results from the presence of two i-Pr3P ligands on the palladium atom (F- and B-strain) (34). Thus, to enable saturation of the boron valences would require a decrease in steric congestion, achieved by removing a i-Pr3P ligand from arylpalladium hydroxide complex 6. This hypothesis led to the investigation of monoligated arylpalladium hydroxy complex, [(i-Pr3P)(4-FC6H4)Pd(OH)]2 (17), which exists in dimeric form in both solution and solid states (35).

The addition of a THF-d8 solution of 7 (2.0 equivalents) to a THF-d8 solution of 17 (1.0 equivalent) at −78°C, followed by warming to −60°C, produced no new complexes. However, upon cooling the solution to −100°C, a new species emerged, with complete consumption of 17 and with 50% of 7 remaining (Fig. 3, route 4). The structure of this species could be assigned as the bridged bis-arylpalladium arylboronate complex (18), which is reminiscent of other palladium acetate and carbonate complexes (36, 37). The stoichiometry of complex 18 was determined by adding a THF-d8 solution of 7 (1.0 equivalent) to a THF-d8 solution of 17 (1.0 equivalent) at −60°C, followed by cooling to −100°C, where a quantitative conversion was observed. This dinuclear complex did not incorporate another molecule of 7, even in the presence of 3 additional equivalents of 7 at −100°C. However, exchange spectroscopy showed cross peaks between 18 and unbound 7 at −100°C, demonstrating that the system was in equilibrium even at this temperature. This stoichiometry (1B:2Pd) is attributed to the thermochemical preference for Pd-(μ-OH)-Pd moieties, which is observed in other bridged mixed-hydroxide complexes (38).

Fig. 3 Formation of 8-B-4 complexes containing Pd-O-B linkages.

The connectivity of the Pd-O-B linkage in 18 was confirmed by the observation of NOEs between Hb, Hd, and the bridging OH group with the methyl hydrogens on the i-Pr3P group. The observation of NOE cross peaks, along with HMBC (3JBOH-C(1)) cross peaks between the BOH and the ipso-carbon–bearing boron (red bonds), indicates that the arylboronic acid and arylpalladium hydroxide are connected. The 11B NMR chemical shift of 18 was too broad to determine accurately. The broadening of the 11B NMR signal in complex 18 is probably attributable to the chemical exchange between 7 and 18.

To further aid in the structure determination of 18, we combined 3.0 equivalents of thallium arylboronate 10 with [(i-Pr3P)(4-FC6H4)Pd(I)]2 (19) in THF-d8 at −78°C and then warmed the sample to −50°C (Fig. 3, route 5). Cooling the mixture to −100°C resulted in the observation of complex 18 (~50%) by 1H NMR spectroscopy. The ability to forge the Pd-O-B linkage in 18 by two routes provides compelling support for the structural assignment.

In an attempt to arrive at a different stoichiometry (1B:1Pd), 60 μl of CH3OH was injected into a THF-d8 solution of 18 with 1.0 equivalent of 7 (from 17 and 7; vide supra), which resulted in the quantitative formation of a new species (20) (Fig. 3). The presence of a Pd-O-B linkage in 20 was established by the observation of NOE cross peaks between the methyl hydrogens on the i-Pr3P and both Hb and Hd (Fig. 3, blue arrows, and fig. S57). The 11B NMR chemical shift of 20 at 9 ppm is well within the characteristic chemical shift regime of tetracoordinate (8-B-4) complexes (12, 13, 39). The proposed Pd-O-B-O core can be found in an analogous bridging arylpalladium acetate complex (40, 41).

The ability to generate intermediate species containing Pd-O-B linkages provided a singular opportunity to examine the kinetic aspects of the transmetalation event in the Suzuki-Miyaura cross-coupling reaction. The transfer of the aryl group from boron to palladium was investigated by using NMR spectroscopy to follow the decay of complex 20 and the concomitant formation of cross-coupling product 13.

The addition of CH3OH into a THF solution of 18 and 7 at −55°C led to the generation of 20. The subsequent formation of cross-coupling product 13 was monitored by 19F NMR spectroscopy at −30°C (to expedite data collection). First-order plots of [20] and [13] versus time (figs. S62 to S64) were fitted by using the functions [A] = [A]0ekt and [P] = [A]0(1 − ekt), respectively, where [A] is the concentration of 20, [A]0 is the initial concentration of 20, [P] is the concentration of 13, k is the rate constant, and t is time. These functions provided accurate values for kobs (the observed kinetic constant) for the decay of 20 [(1.41 ± 0.02) × 10−3 s−1] and the formation of 13 [(1.55 ± 0.09) × 10−3 s−1; Fig. 4A].

Fig. 4 Kinetic data.

(A) Formation of 13 from 8-B-4 complexes 18 and 20 (negative signs indicate consumption of starting material). (B) Inverse order dependence on [i-Pr3P] for the formation of 13 from 6-B-3 complex 11.

A similar kinetic analysis was performed in pure THF by combining a THF solution of 7 (2.0 equivalents) with a THF solution of 17 (1.0 equivalent) at −78°C, followed by warming the sample to −30°C. First-order decay of arylpalladium complex 18 and the formation of 13 were observed with kobs values of (7.59 ± 0.58) × 10−4 s−1 and (5.78 ± 0.13) × 10−4 s−1, respectively (figs. S65 to S67), indicating that the decay of 18 and the formation of 13 coincide. The correspondence of these rate constants and the clean first-order behavior suggests that at –30°C, 18 is largely converted to 20 (sufficient 7 is present to allow this) (42). Moreover, the similarity of the rate constants in THF and the THF-CH3OH mixture further supports the conclusion that 18 is converted to 20 before transmetalation (Fig. 4A) (43).

The generation of 6-B-3 complex 11, with its Pd-O-B linkage, raised the question of whether this complex is a competent intermediate in the Suzuki-Miyaura reaction. Complex 11 was thermally stable at −30°C in the presence of an excess of i-Pr3P for over 24 hours, indicating that added phosphine attenuates the rate of the transmetalation process. A THF solution of this complex was warmed from −30°C in 10°C intervals, whereupon a considerable amount of cross-coupling product 13 was observed by means of 19F NMR spectroscopy at 20°C over the course of 3 to 12 hours. A plot of [11] versus time displayed s-shaped curves (figs. S68 to S91), signifying that the kobs increases during the course of the reaction (which is indicative of autocatalysis) (44).

To confirm the kinetic requirement for phosphine dissociation in the cross-coupling of 11, the kinetic order in phosphine was determined by adding a THF solution of 7 to a solution of 6 with increasing amounts of i-Pr3P, ranging from 97 to 294 mM, at 20°C (Fig. 4B). The s-shaped kinetic profiles were fitted, and a vmax (maximum rate) was extracted from the data (45). A plot of log[vmax] versus log[i-Pr3P] gave a slope of −1.05 ± 0.05 (fig. S92), which is consistent with an inverse dependence on phosphine, indicating that dissociation of a phosphine is a pre-equilibrium process that leads to the hypothetical 14-electron palladium complex 15. Because 15 is formed in such a low-equilibrium concentration, it is not possible to determine whether transmetalation occurs directly from this 6-B-3 species or whether it requires the coordination of another group on boron to form species related to 20. Although very low levels of halide and water are present, the coordination state of boron in the transmetalation event cannot be unambiguously established.

Through the combination of three methods of investigation (spectroscopic analyses, independent syntheses, and kinetic measurements), we have unambiguously identified and characterized three pre-transmetalation species containing Pd-O-B linkages that undergo the Suzuki-Miyaura cross-coupling reaction. Despite the long-held assumption that these types of intermediates are involved in the transmetalation event, our study provides the first definitive evidence for their involvement. We have demonstrated that both tetracoordinate (18 and 20) and tricoordinate (11) boron complexes containing the critical Pd-O-B moieties are able to transfer their B-aryl groups to palladium. Moreover, our investigations establish that an empty coordination site on the palladium atom is needed for the transmetalation event to take place from all three Pd-O-B–containing species. We foresee these results serving as a platform for further investigations of the venerable Suzuki-Miyaura cross-coupling process.


Materials and Methods

Figs. S1 to S118

Tables S1 to S26

References (4651)


  1. Although i-Pr3P is not commonly used in Suzuki-Miyaura reactions, it is structurally similar to c-Hex3P (c-Hex, cyclohexyl), which has widespread application (2, 26).
  2. This nomenclature designates an N-X-L species, where N is the number of formally valence-shell electrons around atom X that are involved in bonding L ligands to X (29).
  3. We cannot unambiguously exclude the possibility that thallium hydroxide dissociates to a minor extent, forming the palladium hydroxide complex 6. These results do not contradict the conclusions of the studies led by Hartwig (13) and Amatore and Jutand (1416). The kinetic incompetence of sodium or potassium boronates in aqueous-organic solutions need not carry over to organic-only solutions of thallium boronates.
  4. We cannot unambiguously exclude the possibility that methanol has incorporated into complex 20.
  5. Although 18 and 7 are both observed at –100°C, at –30°C, these species converge, thus preventing a definitive assignment of whether 18 or 20 is the actual species undergoing decay in the kinetic analysis.
  6. The identity of the species that is inducing autocatalysis has not been determined at this time.
Acknowledgments: We are grateful for generous financial support from the NSF (CHE-1012663 and CHE-1151566). A.A.T. is grateful to the University of Illinois for graduate fellowships. We thank L. Zhu for helpful suggestions regarding NMR spectroscopy. Some of the data presented here were collected in the Core Facilities of the Carl R. Woese Institute for Genomic Biology on a 600-MHz NMR instrument (funded by NIH grant S10-RR028833) or at the Integrated Molecular Structure Education and Research Center at Northwestern University. Full experimental procedures, characterization and kinetic data, and copies of the 1H, 13C, 31P, 19F, 11B, NOE, and exchange spectra can be found in the supplementary materials.
View Abstract

Navigate This Article