Mechanistic Studies of the Palladium-Catalyzed Desulfinative Cross-Coupling of Aryl Bromides and (Hetero)Aryl Sulfinate Salts.

Pyridine and related heterocyclic sulfinates have recently emerged as effective nucleophilic coupling partners in palladium-catalyzed cross-coupling reactions with (hetero)aryl halides. These sulfinate reagents are straightforward to prepare, stable to storage and coupling reaction conditions, and deliver efficient reactions, thus offering many advantages, compared to the corresponding boron-derived reagents. Despite the success of these reactions, there are only scant details of the reaction mechanism. In this study, we use structural and kinetic analysis to investigate the mechanism of these important coupling reactions in detail. We compare a pyridine-2-sulfinate with a carbocyclic sulfinate and establish different catalyst resting states, and turnover limiting steps, for the two classes of reagent. For the carbocyclic sulfinate, the aryl bromide oxidative addition complex is the resting state intermediate, and transmetalation is turnover-limiting. In contrast, for the pyridine sulfinate, a chelated Pd(II) sulfinate complex formed post-transmetalation is the resting-state intermediate, and loss of SO2 from this complex is turnover-limiting. We also investigated the role of the basic additive potassium carbonate, the use of which is crucial for efficient reactions, and deduced a dual function in which carbonate is responsible for the removal of free sulfur dioxide from the reaction medium, and the potassium cation plays a role in accelerating transmetalation. In addition, we show that sulfinate homocoupling is responsible for converting Pd(OAc)2 to a catalytically active Pd(0) complex. Together, these studies shed light on the challenges that must be overcome to deliver improved, lower temperature versions of these synthetically important processes.

Introduction

The Suzuki–Miyaura reaction,[1] in which a boron-based coupling partner is combined with an aryl halide using a palladium catalyst, has revolutionized synthetic chemistry in the pharmaceutical industry.[2] However, applying these reactions to the preparation of heteroaromatic derivatives has long remained a challenge, because the corresponding heterocycle-derived boronic acids (and related boron-based reagents) can be challenging to prepare, unstable, base-sensitive, and prone to protodeboronation and decomposition under cross-coupling conditions. This is particularly true for pyridine and related diazene-derived boronic acids.[3] In this context, the Willis laboratory has recently reported the use of pyridyl[4] and heteroaryl[5] sulfinate salts as nucleophilic reagents in palladium-catalyzed cross-coupling reactions with aryl and heteroaryl halides (Scheme a). These sulfinate salts, several of which are now commercial products, are stable reagents and deliver efficient reactions with challenging coupling partners, thus addressing many of the limitations of their boron counterparts. More recently, the use of related heteroaryl allylsulfones, which function as latent sulfinate coupling partners, has also been reported.[6]

Scheme 1

Heterocyclic Sulfinate Coupling Reactions, a Generalized Mechanism, and Aims of the Study

More generally, transition-metal-catalyzed desulfinative cross-coupling has emerged as an effective method for C–C bond formation.[7] First employed for homocoupling with stoichiometric palladium(II),[8] sulfinate salts can now be employed as the electrophilic coupling partner in Mizoroki–Heck,[9] Suzuki–Miyaura,[10] and Hiyama[11] cross-coupling processes. Their use as nucleophilic coupling partners was first reported in a 1992 patent,[12] before being more widely developed for both aromatic[13] or five-membered O- and S-heteroaromatic[14] systems. Catalytic versions of homocoupling reactions have also been reported.[15] Together with decarboxylation-based methods, desulfination can be considered as a potentially more sustainable approach to cross-coupling chemistry, relative to processes that require the preparation and use of costly and sensitive organometallic reagents. The release of SO2 gas as the main byproduct should also be considered in this context. While decarboxylation has been studied both experimentally[16] and theoretically,[17] the intrinsic difference between an sp2-hybridized carbon center and an sp3-hybridized sulfur center precludes direct transposition of these studies to the desulfinative variants. Similarly, although carbonylation and sulfur dioxide insertion are related processes, they can follow different mechanistic pathways.[18] To date, no mechanistic investigation of a desulfinative cross-coupling process using sulfinate salts as nucleophilic coupling partners has been reported.[19]

A generalized mechanism for desulfinative cross-coupling is shown in Scheme b. Reduction of the Pd(II) precatalyst to the active Pd(0) species is followed by oxidative addition of the carbon-bromide bond to the transition metal. Transmetalation between the alkali metal sulfinate salt and the oxidative addition complex generates a putative palladium sulfinate intermediate. Extrusion of SO2 followed by reductive elimination regenerates the Pd(0) catalyst and delivers product. In the present study, we explore this mechanism using two model sulfinate reagents, allowing the comparison of carbocyclic and heterocyclic substrates. We describe the first synthesis, characterization, and reactivity study of the putative palladium sulfinate intermediates. Using a combination of stoichiometric experiments and kinetic studies, we propose different resting-state intermediates and turnover limiting steps for the two different classes of sulfinate reagent. We also investigate the role of the base, which is a crucial additive for reactivity. The aim of these studies was to improve the understanding of these synthetically important transformations, and also to provide insights for the development of reactions which proceed using milder conditions, particularly with regard to the high temperatures (120–185 °C) generally used in these reactions.[7]

Results and Discussion

Model Reactions

The coupling of the unsubstituted sodium pyridine-2-sulfinate salt 1 with 1-bromo-4-fluorobenzene 2 proceeds in 95% yield as shown in Scheme . Although these reactions were initially developed for pyridine sulfinate salts,[4] before being extended to a larger class of heteroaromatic substrates,[5] they can also be applied to carbocyclic variants. For example, sodium 4-methylbenzenesulfinate 4 can be combined with 1-bromo-4-fluorobenzene using Pd(OAc)2, PCy3, and K2CO3 at 150 °C to deliver the cross-coupling product 5 in 92% yield. Encouraged by the robustness of these reaction conditions, we chose to compare the reactivity of heterocyclic and carbocyclic sulfinate salts, as the carbocyclic variants of this cross-coupling also require high temperatures.[7] Indeed, such carbocyclic sulfinate salts are inexpensive, bench-stable reagents that can be easily prepared via a broad range of methods.[20] A lower temperature cross-coupling of such reagents would potentially establish sulfinates as attractive alternatives to the more-expensive boron-based reagents. Heterocyclic sulfinate 1 and carbocyclic sulfinate 4, together with aryl bromide 2, were therefore selected as the substrates for our proposed mechanistic study.

Scheme 2

Substrates and Reactions To Be Studied

Synthesis and Characterization of Organometallic Intermediates

Formation of the Active Palladium(0) Species

Palladium(II) acetate is employed as the palladium source for these desulfinative cross-coupling reactions, and therefore the active Pd(0) species must be generated via a reductive process. This step is often overlooked in mechanistic studies, although the Pd(0) species could be generated from several different pathways and, thus, give rise to different kinetic profiles,[21] making the understanding of how Pd(0) is formed crucial for the design of accurate kinetic experiments (section ). Phosphine oxidation has long been known as a pathway for palladium zero generation,[22] but nucleophilic coupling partners such as organolithium reagents,[23] organostannanes,[24] alcohols,[25] and amines[26] can also mediate the reduction of Pd(II) precatalysts. Having observed such reduction pathways, Buchwald and co-workers judiciously used substoichiometric amounts of an aryl boronic acid in order to ensure complete reduction of their Pd(II) precatalysts.[27] Furthermore, a synthetically useful dimerization of two sulfinate salts using stoichiometric amounts of Pd(II) was reported in the early 1970s.[8] Therefore, we examined the generation of the active catalyst in the presence of 4-methylbenzenesulfinate 4 (Figure ). The two coupling partners 1-bromo-4-fluorobenzene 2 and 4-methylbenzenesulfinate 4 were introduced to a Young’s NMR tube, together with K2CO3 and stoichiometric amounts of Pd(OAc)2 and PCy3 in a 2:1 ratio. The NMR tube was then gradually heated at different temperatures and the reaction was monitored by 31P{1H} and 19F{1H} NMR spectroscopy. The known Pd(PCy3)2(OAc)2 complex[28]6 was rapidly formed at room temperature. Upon heating to 60 °C, the 31P{1H} resonance of 6 started to decay to a broad signal at 29.83 ppm, corresponding to a set of broad doublets at 8.43 and 6.63 ppm in the 1H NMR spectrum (see the Supporting Information (SI)).

Figure 1

31P{1H} NMR spectra of the reaction between Ar–Br (2) and Ar′SO2Na (4) catalyzed by stoichiometric Pd(OAc)2 and PCy3 in a Young’s NMR tube at various temperatures. [Legend: 6, Pd(OAc)2(PCy3)2; 7, (Ar′SO2)Pd(PCy3)2(OAc); 8, (Ar′)Pd(PCy3)2(OAc); 8b, (Ar′)Pd(PCy3)2(Br); and 9, (Ar)(Pd)(PCy3)2(Br); Ar = 4-F-C6H4; Ar′ = 4-Me-C6H4.]

The complex corresponding to these signals was isolated from a separate reaction that was performed for 16 h at 60 °C. Single crystals suitable for an X-ray diffraction (XRD) study were grown from CH2Cl2/hexane and revealed a sulfinato-S palladium complex 7 (Figures a and 2b). The structure displays a square planar geometry around the Pd center, with a trans arrangement of the phosphines. The Pd–S bond length of 2.2755(4) Å in 7 is shorter than those in (dppf)Pd(SO2Me)Cl[29] and (BINAP)Pd(SO2C6H4Me)Cl[30] (2.3262(9) and 2.3331(7) Å, respectively). While the Pd–O bond length of 2.0975(13) Å is notably longer than that in (PCy3)2Pd(OAc)2 (2.0288(17) Å), reflecting the strong trans influence of the sulfinate ligand. The arrangement of the acetate and the 4-methylbenzenesulfinate about the palladium affords a close contact (2.578(2) Å) between the nonbound O atom of the acetate and the 2-aryl-hydrogen of the 4-methylbenzenesulfinate.

Figure 2

(a) Synthesis of the palladium sulfinate complexes 7 and 10. X-ray crystal structures (50% displacement ellipsoids) of (b) 7 and (c) 10. All H atoms have been omitted for the sake of clarity.

Complex 7 then undergoes extrusion of SO2 to give complex 8 (δP = 18.50 ppm) having a molecular formula of (4-Me-C6H4)Pd(PCy3)2(OAc). The corresponding phenyl complex was reported and characterized by X-ray crystallography by Eastgate and co-workers.[21a] Above 100 °C, complex 8 is then likely to undergo a second acetate/aryl sulfinate exchange to form a palladium sulfinate complex (Ar′)Pd(PCy3)2(O2SAr′), followed by extrusion of SO2 and reductive elimination to generate a palladium zero intermediate and 4,4′-dimethylbiphenyl, the homocoupling product. Neither of these two intermediates were observed by NMR spectroscopy as the rate of extrusion of SO2 from such palladium sulfinate complexes (sections and 2.5), and of oxidative addition (section ), are expected to be fast at such temperatures. Instead, the oxidative addition complex (4–F-C6H4)Pd(PCy3)2(Br) 9 (δP = 20.08 ppm) was observed. Complex 8 can also undergo an acetate/bromide exchange giving complex 8b (δP = 19.79 ppm), which can then undergo similar transformations leading to the formation of palladium zero and homocoupling product. Phosphine oxidation products were detected in trace amounts only, which is consistent with the observation that Pd(PCy3)2(OAc)26 is not easily reduced to palladium(0) via phosphine oxidation in the absence of hydroxide bases.[21a]

When the cross-coupling was performed using 5–18 mol % of Pd(OAc)2 and PCy3 in a 1:2 ratio, respectively, a linear correlation was observed between the amount of homocoupling product (4,4′-dimethylbiphenyl) formed and the amount of Pd(OAc)2 introduced (see the SI), which supports the proposed reduction pathway of the precatalyst.

When sodium pyridine-2-sulfinate 1 was employed as the nucleophilic coupling partner, a different palladium complex 10 was isolated (Figures a and 2c). This complex also displays a square planar geometry around the Pd center, with a pyridine residue lying trans to the phosphine instead of a second phosphine ligand as in 7. Two 2-pyridyl sulfinate groups are present in 10. One chelates the palladium in a κ2N,O-mode, affording a five-membered metallocycle. This mode of ligand binding has precedence in copper 2-pyridyl sulfinate complexes.[31] The second 2-pyridyl sulfinate coordinates the palladium by the sulfur, trans to the oxygen binding. The Pd–S bond length in 10 (2.2307(5) Å) is notably shorter than that in 7, possibly reflecting the decreased steric congestion in the former.

Oxidative Addition Complex

Once the active Pd(0) species have been generated, the next step of the presumed mechanism is the oxidative addition of the carbon–halide bond to the Pd(0) intermediate. Oxidative addition of bromobenzene to Pd(PCy3)2 has been investigated by Baird and co-workers[32] and Hartwig and co-workers.[33] Although these studies were conducted in nonpolar solvents and at room temperature, both concluded that the oxidative addition was an irreversible process that involved the aryl bromide substrate and the two-coordinate palladium zero complex Pd(PCy3)2, leading to the formation of compounds with the generic formula (Ar)Pd(PCy3)2(Br). The oxidative addition complex (4-F-C6H4)Pd(PCy3)2(Br) 9 was synthesized independently, according to literature procedures,[34] and proved to be a competent catalyst for our sulfinate coupling reactions, giving similar reaction rate and yield, compared to the mixture of Pd(OAc)2 and PCy3 (see the SI).

Palladium Sulfinate Complexes

For cross-coupling processes in general, the transmetalation step, which involves the oxidative addition complex and the nucleophilic coupling partner, differs for each nucleophile, leading to a wide variety of mechanisms and palladium intermediates. Such post-transmetalation complexes have been characterized for several nucleophiles, including alcohols,[35] amines,[36] thiols,[37] and carboxylic acids.[38] However, despite having been reported as efficient coupling partners, to the best of our knowledge, there is no report of a post-transmetalation intermediate with a nucleophilic sulfinate. A dimeric palladium sulfinate complex with bridging Cl– anions has recently been reported by Shavnya and co-workers;[39] however, it is not relevant to an oxidative addition/transmetalation sequence. In the studied cross-coupling, putative palladium sulfinate intermediates would arise from the displacement of the Br atom from oxidative addition complex 9 by a metal sulfinate salt, as described in Scheme .

Scheme 3

Putative Formation of the Palladium Sulfinate Complexes

Unlike carboxylate salts, metal sulfinate complexes can have different coordination modes, binding through sulfur, oxygen, or both.[40] In the previous section, we have described two types of palladium sulfinate complexes relevant to the reduction of Pd(II) to Pd(0). In the following section, we describe the first synthesis and characterization of post-transmetalation aryl palladium sulfinate complexes.

Synthesis and Characterization

We first attempted to form the transmetalation complexes by heating the oxidative addition complex 9 at moderate to elevated temperatures (50–120 °C) in the presence of the carbocyclic sulfinate salt 4. However, formation of biaryl 5 and/or decomposition were observed. In the presence of silver nitrate, the oxidative addition complex was entirely consumed at room temperature, and we isolated the dimeric palladium sulfinate complex 11 (Figures a and 3b). When the pyridine-2-sulfinate substrate 1 was subjected to the same reaction conditions, the monomeric palladium sulfinate complex 12 was obtained (Figures a and 3c). The geometry about the palladium center in 12 closely resembles that present in 10. The Pd–O bond length in complex 12 (2.132(2) Å) is elongated, relative to that in complex 10 (2.0975(13) Å), reflecting the greater trans influence of the aryl ligand. A similar square planar geometry about the palladium is observed in complex 11; however, the absence of a pyridyl substituent on the sulfinate results in the vacant site being occupied by the S atom of a neighboring sulfinate, affording a dimeric structure. The structure is similar to that of [(PMe3)Pd(CH2CMe2C6H4SO2)]2, which likewise features a trans-phosphorus–sulfur configuration.[41] The bond lengths and angles observed in 11 also resemble those in [(PMe3)Pd(CH2CMe2C6H4SO2)]2.

Figure 3

(a) Synthesis of the palladium sulfinate complexes 11 and 12. X-ray crystal structures (50% displacement ellipsoids) of (b) 11, (c) 12, and (d) 13. (e) Structure of 13. All H atoms have been omitted for the sake of clarity.

The diffusion coefficients obtained by DOSY NMR spectroscopy predicted masses for complexes 11 and 12 in a 2:1 ratio, indicating that the former maintains its dimeric structure in solution. The degree of aggregation in these complexes is also maintained when either an electron-donating or electron-withdrawing substituent is introduced at the position para to the sulfinate group (see the SI).

A notable exception is when steric bulk, in the form of a tert-butyl group, is placed at the 6-position of the pyridine. This prevents the κ2N,O-binding mode of the pyridine sulfinate, thereby affording a dimeric solid-state structure 13 which resembles that of the 4-methylbenzenesulfinate substituted complex, 11 (Figure d). X-ray-suitable crystals could be obtained from CH2Cl2/hexane recrystallization of crude reaction mixtures; however, attempts to obtain analytically pure samples for this compound were unsuccessful.

Reactivity

When the dimeric complex 11 was subjected to 1.1 equiv of PCy3, its resonance peak at δP = 27.77 ppm disappeared and was replaced by a doublet at δP = 21.95 ppm, JP–F = 3.0 Hz. This new signal is very similar to the resonance peak observed for the oxidative addition complex 9 (δP = 20.20, JP–F = 3.0 Hz). Although the conversion was quantitative according to the 19F{1H} NMR spectrum, the new complex was only isolated in 57% yield. Crystals suitable for XRD were obtained, and the identity of complex 14 was confirmed as a square planar palladium complex bearing two phosphine ligands (Figure a). The solid-state structure of 14 closely resembles that of 7; however, unlike in the latter, the sulfinate binds to palladium via the O atom, rather than a S atom. This is consistent with a nucleophilic displacement of the S atom from dimeric complex 11. While the addition of a phosphine to the dimeric complex 11 occurs at room temperature, the pyridine sulfinate containing palladacycle 12 was unreactive toward PCy3, and compound 15 was not observed, suggesting that the Pd–N bond in 12 is much stronger than the Pd–S bond in dimeric complex 11 (Figure b). Accordingly, formation of complex 12 was observed at room temperature upon combining dimeric complex 11 and sodium pyridine-2-sulfinate 1, whereas the reverse reaction was not observed under the same conditions (Figure c). Exchange between two carbocyclic sulfinates was not observed at room temperature either, as dimeric complex 17 was not observed when the more nucleophilic 4-methoxybenzene-sulfinate salt 16 was combined with dimeric complex 11 (Figure d). However, the Pd–S bond of 11 could be cleaved upon addition of excess pyridine (Figure e). A new complex was observed by NMR spectroscopy and assigned as complex 18, but its isolation was not possible, since 11 was recovered after workup.

Figure 4

Reactivity of the palladium sulfinate complexes 11 and 12 at room temperature in 1,4-dioxane/benzene-d6 5:1. X-ray crystal structure (50% displacement ellipsoids) of 14. All H atoms have been omitted for the sake of clarity. 16 = 4-MeO-C6H4–SO2Na.

When 4-methylbenzenesulfinate palladium complex 11 was heated at 90 °C in toluene-d8, it was fully consumed within 20 min with concomitant formation of biaryl 5 (Figure , gray diamonds). Complex 14 (generated in situ from 11 and PCy3) reacted at a slower rate, but gave 84% conversion within 45 min (Figure , yellow triangles). Increasing the concentration of PCy3 resulted in a significant decrease of the rate of SO2 extrusion at 90 °C, suggesting that ligand dissociation is necessary prior to desulfination (see the SI). The pyridine-2-sulfinate palladacycle 12 in which the sulfinate chelates the palladium in a κ2N,O-mode only gave 11% conversion after 45 min at 110 °C (Figure , green circles). Weakening the Pd–N bond by adding an electron-withdrawing group on the pyridyl ring, as in complex 20, increased the rate of extrusion of SO2 in toluene-d8, as a 70% conversion was obtained after 45 min at 110 °C (Figure , blue squares). The relative rates of consumption of the complexes shown in Figure have been shown to correspond to the rates of formation of the cross-coupling products (see the SI).

Figure 5

Reactivity of the palladium sulfinate complexes upon heating. Reactions were performed in toluene-d8 in a Young’s NMR tube. Complex 14 was generated in situ from complex 11 and tricyclohexylphosphine.

Overall, these experiments show that the S–Pd bond present in dimer 11 can readily be displaced by nitrogen- or phosphorus-based nucleophiles. The formation of five-membered-ring palladacycles with the pyridine-2-sulfinate substrates renders the derived palladium sulfinate complexes more thermodynamically stable, and dramatically slows the extrusion of SO2. In addition, we have shown that complexes 11, 12, and 14 are competent catalysts for their respective coupling reactions (see the SI).

Identification of the Resting-State Intermediates

To identify the resting-state intermediates of the catalytic reactions, the desulfinative cross coupling was monitored by 31P{1H} NMR and 19F{1H} NMR spectroscopy. The reactions were performed in a Young’s NMR tube with a higher loading of Pd(OAc)2 and PCy3 (15 mol % and 30 mol %, respectively). The observed resonance peaks were compared to the signals of reference samples acquired in a 5:1 mixture of 1,4-dioxane/benzene-d6. Because of poor stirring of the heterogeneous mixture in an NMR tube, the reaction kinetics were found to be very slow. After 14 h at 150 °C, at which point biaryl 5 was formed in 40% yield (∼2.7 turnovers), the resonance peak of the oxidative addition complex 9 (δF = −124.26 ppm) was the only other signal observed on the 19F{1H} spectrum along with starting material, internal standard, and product (Figure a). The corresponding characteristic doublet at δP = 20.08 ppm was also observed in the 31P{1H} spectrum, along with free PCy3 (δP = 9.24 ppm), phosphine oxide (δP = 45.65 ppm) and compound 8 (δP = 19.80 ppm) (Figure c). Therefore, the oxidative addition complex 9 is likely to be the resting-state intermediate when using carbocyclic sulfinate 4. The same experiment was performed with the pyridine-2-sulfinate substrate 1. After 14 h of reaction at 150 °C, the conversion to product 3 was lower than for the carbocyclic substrate (20%, ∼1.3 turnovers). The oxidative addition product 9 was observed in a 1:1 ratio with palladium sulfinate complex 12 (Figure b). The corresponding phosphorus resonance peaks (δP = 38.03 and δP = 20.08), along with free PCy3 oxidation products, were also observed (Figure d).

Figure 6

Cross-coupling reactions performed in a Young’s NMR tube. (a) 19F{1H} NMR spectrum of the reaction mixture involving sulfinate 4. (b) 19F{1H} NMR spectrum of the reaction mixture involving sulfinate 1. (c) 31P{1H} NMR spectrum of the reaction mixture involving sulfinate 4. (d) 31P{1H} NMR spectrum of the reaction mixture involving sulfinate 1.

In section , we have assessed the reactivity of the palladium sulfinate complexes and demonstrated that weakening of the N–Pd bond facilitates the extrusion of SO2 from the palladium pyridine-2-sulfinate complexes. When sodium 5-(trifluoromethyl)pyridine-2-sulfinate 19 was subjected to the cross-coupling conditions in a Young’s NMR tube, the corresponding palladium sulfinate complex 20 could not be detected by 19F{1H} or 31P{1H} NMR, and the oxidative addition complex 9 was the only observable complex by 19F{1H} NMR spectrum (see the SI). Similarly, sodium pyridine-4-sulfinate 21, which does not have the appropriate substitution pattern to form a stabilizing N–Pd interaction, only exhibited the resonance of the oxidative addition complex 9 (see the SI). Unidentified minor resonances were detected at δP = 19.81 ppm and δP = 20.17 ppm, together with free ligand and minor phosphine oxide peaks. Finally, the oxidative addition complex 9 was once again the only resonance observed in the 19F{1H} spectrum when the bulky lithium 6-(tert-butyl)pyridine-2-sulfinate 22 was used. This is consistent with the observation of the dimeric crystal structure of the 6-tert-butylsulfinate containing palladium complex 13, as the formation of a κ2N,O-chelate is hindered by the tert-butyl group, showing that the bulky pyridine sulfinate salt behaves in a manner similar to that of a carbocyclic substrate.

Since the concentration, stirring, and therefore kinetics in an NMR tube are not accurately representative of the actual reaction conditions, the two cross-coupling reactions involving sodium pyridine-2-sulfinate 1 and sodium 4-methylbenzenesulfinate 4 were also performed in microwave tubes at a relevant scale. The tubes were heated for 2 h at 150 °C, then 500 μL of the reaction mixture were transferred to an NMR tube containing 100 μL of benzene-d6. The observed resonance peaks were in agreement with the reactions performed in the Young’s NMR tubes. One difference was observed for the cross-coupling of pyridine-2-sulfinate salt 1 for which the ratio between the oxidative addition complex 9 and the pyridine sulfinate complex 12 was found to be 1:3.8 instead of 1:1 (see the SI). The use of an appropriate reaction vial equipped with a stirrer bar instead of Young’s NMR tube must ease the transmetalation step, which involves the poorly soluble sulfinate salt, while the rate of extrusion of SO2 remains unaffected and difficult, because of the strong κ2N,O-chelation mode. As a consequence, the resting-state intermediate is shifted from the oxidative addition complex 9 to the pyridine-2-sulfinate containing palladium complex 12.

Overall, these experiments show that the resting state of the catalyst is highly dependent on the nature of the sulfinate coupling partner and is more likely to be the oxidative addition complex 9 for carbocyclic sulfinates or for pyridine-sulfinate substrates that cannot form strong N–Pd bonds, due to geometrical, steric, or electronic reasons. A strong κ2N,O-chelation as observed for the unsubstituted pyridine-2-sulfinate group in the palladium sulfinate complex 12 was observed to cause a change in the resting-state intermediate.

Initial Rate Studies

Productive catalytic reactions are performed at 150 °C in 1,4-dioxane (boiling point (bp) = 101 °C) and involve partially soluble reagents such as the sulfinate salts and potassium carbonate, leading to suspensions rather than homogeneous solutions. Therefore, the reactions could not be accurately monitored by NMR spectroscopy and were monitored using HPLC analysis of aliquots taken from the reaction mixture via syringe, and the partial orders were obtained using the initial rates of the reactions.

Order in Aryl Bromide

The reactions of 4-methylbenzenesulfinate 4 and pyridine-2-sulfinate 1 were performed with different concentrations of aryl bromide 2 (0.025–0.15 mM).

The initial rates were not affected by the concentration of the electrophilic coupling partner (Figures a and 7b). This zero-order behavior suggests that oxidative addition of the C–Br bond into the Pd(0) species is not the turnover limiting step under the reaction conditions. This observation is in agreement with the observed resting-state intermediates in section .

Figure 7

Initial rate plots for cross coupling between aryl bromide 2 and (a) sodium 4-methylbenzenesulfinate 4 (orange diamonds) and (b) sodium pyridine-2-sulfinate 1 (orange circles).

Order in Sulfinate Salt and K2CO3

It was observed that 0.010–0.10 M solutions of sodium 4-methylbenzenesulfinate 4 or sodium pyridine-2-sulfinate 1 are heterogeneous at 150 °C. Therefore, the partial orders in sulfinate coupling partner were not measured. Similarly, the order in potassium carbonate was not obtained.

Order in Catalyst

To get the partial order in catalyst, the initial rate of the reaction was measured for different concentrations of Pd(OAc)2 and PCy3, keeping the ligand to palladium ratio fixed at two. The rate of the reaction involving 4-methylbenzenesulfinate sodium salt 4 was observed to be approximately proportional to the square root of the total concentration of palladium introduced, as palladium acetate: rate ∝ [Pd(OAc)2]0.5 (Figure a). Half order in catalyst can generally be rationalized by an equilibrium between a catalytically active complex and an off-cycle species. This can be caused by dimerization of the active catalyst to form an inactive dimer.[42] Half-order dependence of the catalyst can also arise from ligand dissociation.[43] When the concentration of Pd(OAc)2 was varied while the phosphine ligand was in excess, such as [Pd(OAc)2]/[PCy3] < 0.10, the half-order dependence on the concentration of palladium acetate was no longer observed. Instead, the rate was observed to be directly proportional to the concentration of Pd(OAc)2: rate ∝ [Pd(OAc)2] (Figure b).

Figure 8

Initial rate plots for cross coupling between aryl bromide 2 and (a) 4-methylbenzenesulfinate 4 (green circles), (b) 4-methylbenzenesulfinate 4 using an excess of PCy3 (blue squares), (c) 4-methylbenzenesulfinate 4 using an excess of PCy3 (gold triangles), and (d) pyridine-2-sulfinate 1 (red diamonds).

Therefore, the half-order dependence observed can be taken into account by a fast equilibrium between the catalytically inactive oxidative addition intermediate (Ar)Pd(PCy3)2(Br) and the reactive intermediate (Ar)Pd(PCy3)(Br) generated after the loss of a phosphine ligand (Scheme ). However, the rate of oxidative addition is known to decrease dramatically upon the addition of ligand, as the inactive tris-ligated Pd0(PCy3)3 can be formed.[44] In order to verify that the switch between half-order and first-order is not due to a change of turnover-limiting step, the order in aryl bromide was determined for a Pd/L ratio <0.10 (Figure c). A zero-order case was obtained under these conditions, showing that the oxidative addition is not rate-determining in the presence of excess ligand.

Scheme 4

Interpretation of the Initial Rates for the Carbocyclic Sulfinate 4

In order to assess the lability of PCy3 under the reaction conditions, the analogous oxidative addition complex 23 was synthesized with the electronically and sterically similar tri-isopropylphosphine Pi-Pr3.[45] When the two oxidative addition complexes were heated at 75 °C in a 5:1 mixture of 1,4-dioxane/benzene-d6, statistical scrambling was obtained within 2 h and the mixed complex 24 was observed (Scheme ). Therefore, PCy3 is likely to be labile under the reaction conditions, which supports the phosphine dissociation/transmetalation sequence.

Scheme 5

Lability of PCy3 under the Reaction Conditions

The rate of the reaction involving sodium pyridine-2-sulfinate 1 was directly proportional to the concentration of palladium acetate: rate ∝ [Pd(OAc)2] (Figure d). There is no longer a half-order dependence in Pd(OAc)2 as observed for the carbocyclic sulfinate 4. This is in agreement with the observation of different resting-state intermediates for reactions involving substrates 1 and 4. For the former, the first order dependence in Pd(OAc)2 is consistent with the catalyst resting state being palladium sulfinate complex 12 and the rate-determining step being the extrusion of SO2. As a half-order dependency is not observed for pyridine-2-sulfinate 1, this suggests that the transmetalation is irreversible under these reaction conditions.

Comparison of the Rates of Transmetalation and SO2 Extrusion

Stoichiometric reactions involving the oxidative addition complex 9 or the palladium sulfinate complexes 11 and 12 were performed in order to qualitatively compare the rates of transmetalation and extrusion of SO2. For the reaction involving the oxidative addition complex 9, 2 equiv of the potassium sulfinate salts 25 or 26 were employed (Figure ). The potassium salts were chosen for these stoichiometric studies, since evidence suggests that these are the salts formed in the catalytic reactions when K2CO3 is used as the base (section ). When the dimeric complex 11 was heated at 150 °C with 1–4 equiv of PCy3, with respect to the amount of palladium, full conversion to the corresponding biaryl 5 was obtained within 2 min (Figure a). Although we have demonstrated that added phosphine slows the rate of extrusion of SO2 at 90 °C in toluene-d6 (see the SI), these data show that the concentration of ligand does not affect the rate of extrusion of SO2 from the in-situ-generated complex 14 at 150 °C. However, the rate of the transmetalation/SO2 extrusion sequence was dramatically slowed upon the addition of PCy3 (Figure b). An excess of phosphine is likely shifting the equilibrium between the transmetalation inactive (Ar)Pd(PCy3)2(Br) 9 and the transmetalation active (Ar)Pd(PCy3)(Br) toward 9, slowing the overall rate of product formation. These observations are consistent with a turnover-limiting transmetalation after the loss of ligand from oxidative addition complex 9, as exposed in section .

Figure 9

Comparison of the rates of the SO2 extrusion and of the sequence of transmetalation/SO2 extrusion for both carbocyclic sulfinate 4 and pyridine-2-sulfinate 1. Pdtotal (1.0 equiv), PCy3 (0–4.0 equiv), Ar2SO2K (2.0 equiv), Ar3SO2K (2.0 equiv). Ar1 = 4–F-C6H4; Ar2 = 4-Me-C6H4; Ar3 = 2-pyridyl. [Legend: green circles, 0 equiv PCy3; blue squares, 1.0 equiv PCy3; gold triangles, 2.0 equiv PCy3; red diamonds, 3.0 equiv PCy3; and purple circles, 4.0 equiv. PCy3.]

For pyridine-2-sulfinate palladium complex 12, the rate of extrusion of SO2 was much slower than for carbocyclic complex 11 (Figure c). Indeed, independent of the concentration of ligand, at least 20 min were necessary to reach >90% conversion, compared to <2 min for the carbocyclic system. The concentration of ligand did not affect the rate of SO2 extrusion at 150 °C. Compared to the reaction with the carbocyclic sulfinate in Figure b, ligand concentration had little impact on the observed rate of product formation for the transmetalation/extrusion of SO2 sequence with potassium pyridine-2-sulfinate 26 (Figure d). This is consistent with the rate-determining step of the transmetalation/SO2 extrusion sequence being cleavage of the Pd–N bond of the palladacycle 12, which should be independent of ligand concentration.

This set of stoichiometric experiments is in good agreement with the partial orders in catalyst obtained using substoichiometric amounts of Pd(OAc)2 in section .

Role of K2CO3

Unlike for the coupling of aryl halides with nucleophiles such as amines, alcohols, or thiols, there are no acidic protons to be removed in the coupling of sulfinate salts. Yet, a carbonate base, and, more specifically, potassium or cesium carbonate, is a crucial additive for the reaction. In the following section, we investigated the role of the carbonate base and of its countercation.

Role of the Carbonate

One molecule of sulfur dioxide is extruded every time a molecule of the biaryl product is generated. Therefore, since 5 mol % of Pd(OAc)2 is used, a 20:1 ratio between the gaseous byproduct and the catalyst is reached at the end of the reaction. Since SO2 gas and SO2 surrogates have been utilized in many palladium-catalyzed C–S bond-forming reactions,[20b,49] the SO2 generated in the present reactions is likely to coordinate to the palladium center, potentially disrupting catalysis.

Carbonate bases are reported to be efficient traps for SO2 gas in the waste stream from combustion in several industrial processes.[50] Given this, we hypothesized that a carbonate base was necessary to trap the liberated SO2 in the reaction under study, preserving the catalyst and allowing reasonable turnover to be achieved. In order to test this, 0.05 M solutions of SO2 gas in 1,4-dioxane were prepared and divided into an empty flask (A) or a flask containing K2CO3 (B). These flasks were then heated at 150 °C for 16 h, which are the standard reaction time and temperature employed in productive catalytic cross-coupling reactions (Figure a). The obtained solvents A and B were then used for reactions between aryl bromide 2 and carbocyclic and pyridine sulfinates 1 and 4, catalyzed by the oxidative addition product 9. For both substrates, 3% product (less than one turnover) was obtained when using solvent A. Conversely, solvent B, which contained K2CO3, provided 81% and 45% yields of products 5 and 3, respectively (Figure b). These experiments show the importance of the carbonate base to sequester SO2. Based on literature reports,[50,51] we conclude that sulfite anions are generated together with carbon dioxide, as shown in Figure c.

Figure 10

Influence of SO2 and K2CO3 on the reaction: (a) generation of gaseous SO2, solvents A and B; (b) reaction outcome; and (c) rationalization of the role of K2CO3.

Role of the Cation

We demonstrated in sections . and 2.5 that the reaction involving pyridine-2-sulfinate sodium salt 1 is limited by the rate of SO2 extrusion, which does not involve the original sulfinate cation. Therefore, the role of the cation was studied for the reaction involving the carbocyclic sulfinate 4, in which the cation is likely to be involved between the resting-state intermediate and the turnover-limiting step. We also showed in section that the rate of extrusion of SO2 is extremely fast, compared to the rate of transmetalation for the carbocyclic substrate 4. Therefore, in the following section, we have considered that the rate of product formation in the reaction between the oxidative addition complex and a sulfinate salt is representative of the rate of the transmetalation step. The reactions were performed at 120 °C using 2 equiv of sulfinate to 1 equiv of oxidative addition product 9 (Figure a).

Figure 11

Influence of the cation on the rate of the transmetalation step.

Lithium sulfinate 27 showed almost no reactivity (2% product formation in 2 h). Similarly, sodium sulfinate 4 provided 12% product within 2 h, whereas potassium sulfinate 25 reached 74% conversion to product (Figure b). Moving further down in the alkaline metals group, cesium sulfinate 28 increased the rate significantly, with 84% product formation in the same time. Tetrabutylammonium sulfinate 29 had a similar initial rate, compared to the potassium sulfinate 20. However, the conversion did not rise above 60%, even after 5 h. This could be caused by Hofman-type elimination of the ammonium salt, leading to substrate decomposition. The use of 18-crown-6 only slightly increased the rate of anion exchange with 79% yield after 2 h. However, a dramatic increase in rate was observed with tetramethylammonium sulfinate 30, which led to 86% yield within the first 10 min. The low performance of the tetrabutylammonium sulfinate salt, compared to the potassium, or cesium sulfinates suggests that solubility of the sulfinate salt is not correlated to the rate of the transmetalation step. Indeed, the tetrabutylammonium sulfinate is entirely soluble in 1,4-dioxane at 120 °C, while the alkali sulfinates are not. Tetramethylammonium salts have been used as silver surrogates to abstract halide atoms,[52] which might explain the excellent reactivity of sulfinate 30 in the transmetalation step. When sodium sulfinate 4 was employed in a 1:1 ratio with K2CO3 at 150 °C, the transmetalation occurred almost as quickly as the transmetalation involving potassium sulfinate 25 (Figure c). This suggests that the role of K2CO3, in addition to trapping liberated SO2, is to take part in a cation metathesis with the sodium sulfinate salt, thus facilitating transmetalation.

Toward Lower Temperatures

In section , the use of the silver salt AgNO3 allowed the synthesis of a variety of palladium sulfinate complexes at room temperature, starting from the oxidative addition complex 9. In other words, this additive allowed the transmetalation step to occur 125 °C below the temperatures needed in the catalytic system, presumably because of the high affinity between Ag and Br atoms.[53] To confirm the viability of this additive based strategy for lowering the reaction temperature, stoichiometric reactions were performed between the oxidative addition complex 9 and carbocyclic and pyridyl sodium sulfinates 4 and 1 in the presence of AgNO3. Cross-coupling product 5 was obtained in 70% yield for the carbocyclic sulfinate, showing the potential of silver additives to promote reactivity for these types of substrates (Scheme a). However, for the heterocyclic sulfinate 1, only the corresponding palladium sulfinate complex 12 was obtained (Scheme b).

Scheme 6

Toward a Lower Temperature Cross-Coupling

HPLC yield against 1,3,5-trimethoxybenzene.

19F NMR yield based on starting material consumption.

19F NMR yield against 1-fluoronaphtalene.

Ar = 4-F-C6H4.

This suggests that an additive-based strategy that accelerates transmetalation will not be sufficient to achieve a lower temperature cross-coupling for substrates able to strongly chelate the palladium in a κ2N,O-fashion. For these systems, in addition to the transmetalation, there will be an additional challenge in disrupting the stabilizing chelation to the Pd center. This is consistent with the earlier results and was confirmed by the reaction shown in Scheme c, in which the sterically hindered 6-tert-butylpyridine-2-sulfinate salt 22 underwent a lower temperature cross coupling and delivered biaryl 31 in 72% yield. However, the use of silver salts was noncompatible with the catalytic reaction, presumably due to their ability to also act as oxidants.[54] The tetramethylammonium alternative proved to be promising, providing the cross-coupling product 5 in 85% yield in the stoichiometric reaction between the oxidative addition complex 9 and the tetramethylammonium sulfinate 30 at 100 °C (Scheme d). We are currently investigating the use of such additives in the catalytic reaction, as well as alternative strategies to achieve lower temperature cross-couplings with carbocyclic and heterocyclic substrates.

Conclusion

The mechanism of the palladium-catalyzed desulfinative cross-coupling of aryl halides and sulfinate salts was investigated. The in situ reduction of the Pd(II) source to form the active Pd(0) species was shown to be mediated by the homocoupling of two sulfinate substrates. Novel palladium sulfinate complexes were independently synthesized, characterized, and proven to be competent intermediates in the catalytic reaction. The resting-state catalyst and the rate-determining step were identified. Their identity was heavily dependent on the nature of the sulfinate coupling partner, more specifically on the ability of the sulfinate to strongly chelate the palladium in a κ2N,O-mode. The kinetic data also suggest that both transmetalation and extrusion of SO2 most likely occurs after dissociation of a phosphine ligand in order to free a coordination site on palladium. Finally, we showed that the base has a dual role: the carbonate traps the generated sulfur dioxide and permits catalyst turnover, while the potassium countercation undergoes a cation metathesis with the sulfinate salt to allow a faster transmetalation step (Scheme ).

Scheme 7

Overall Mechanism

In summary, isolation of novel palladium sulfinate complexes and kinetic analysis has allowed us to identify the key challenges to reduce the temperature of these reactions. Efforts toward reactions that proceed using milder conditions are currently underway.