Ag/Pd Cocatalyzed Direct Arylation of Fluoroarene Derivatives with Aryl Bromides.

Diverse C-H functionalizations catalyzed by Pd employ Ag(I) salts added as halide abstractors or oxidants. Recent reports have shown that Ag can also perform the crucial C-H activation step in several of these functionalizations. However, all of these processes are limited by the wasteful requirement for (super)stoichiometric Ag(I) salts. Herein, we report the development of a Ag/Pd cocatalyzed direct arylation of (fluoroarene) chromium tricarbonyl complexes with bromoarenes. The small organic salt, NMe4OC(CF3)3, added as a halide abstractor, enables the use of a catalytic amount of Ag, reversing the rapid precipitation of AgBr. We have shown through H/D scrambling and kinetic studies that a (PR3)Ag-alkoxide is responsible for C-H activation, a departure from previous studies with Ag carboxylates. Furthermore, the construction of biaryls directly from the simple arene is achieved via a one-pot chromium tricarbonyl complexation/C-H arylation/decomplexation sequence using (pyrene)Cr(CO)3 as a Cr(CO)3 donor.

Introduction

In recent years, direct C–H arylation has become a versatile tool for the construction of biaryls. This transformation typically involves in situ metalation of an aryl C–H bond and subsequent coupling with an aryl donor. Many different transition metals have been found to catalyze this transformation, including Rh,[1] Ru,[2] Ir,[3] Co,[4] Cu,[5] Fe,[6] and Au,[7] but the most frequently used continues to be Pd.[8]

In contrast to the mild conditions possible for C–H arylations using bis(aryliodonium) salts and iodoarenes as coupling partners,[9] the use of bromoarenes in Pd-catalyzed direct arylation reactions typically requires high temperature and electron-donating phosphine ligands to facilitate the more challenging oxidative addition step.[10] Intriguingly, Hartwig demonstrated that the electron-donating phosphine ligand actually inhibits the Pd(II)-mediated C–H activation step of simple arenes.[11] This presents a quandary for how to approach reaction design for direct arylation with bromoarenes, as any ligand used must strike a delicate balance between promoting oxidative addition and inhibiting C–H activation. With this issue in mind, we envisaged that a cooperative bimetallic catalytic system, in which an electron-rich Pd undergoes oxidative addition and mediates C–C bond formation while another metal performs C–H activation, could provide a resourceful solution.

Ag(I) salts are often employed as stoichiometric halide abstractors in Pd-catalyzed direct arylation reactions, particularly those using iodoarenes as coupling partners, to prevent inhibition of the Pd catalyst.[9,12] We recently demonstrated through H/D exchange, stoichiometric and kinetic studies that phosphine-ligated Ag-carboxylate salts are actually responsible for the crucial C–H activation step in the Pd-catalyzed arylation of electron-deficient arenes (Scheme a).[13] Simultaneously, Sanford reported similar observations on the oxidative homocoupling of thiophenes and fluoroarenes (Scheme b).[14] Hartwig further demonstrated that AgOPiv mediated the C–H activation in a regioselective Pd-catalyzed allylation of simple arenes (Scheme c).[15] Recently, we uncovered an unexpected mechanistic switch in the direct arylation of (benzo)thiophenes, in which a Pd-catalyzed β-arylation was outcompeted by a Ag-mediated (but still Pd-catalyzed) α-arylation at low concentrations of Pd (Scheme d).[16] In all of these methodologies, however, super-stoichiometric Ag salts were found to be required for reactivity.

Scheme 1

Ag(I)-Mediated C–H Functionalizations in Pd/Ag Systems

More recently, a Ag/Au cocatalyzed oxidative coupling between fluoroarenes and pyrazoles was demonstrated to proceed via Ag-catalyzed C–H activation of the electron-deficient coupling partner.[17] Although no analogous Ag/Pd cocatalyzed system has yet been developed, we reasoned that such a system would make an ideal direct arylation with bromoarene coupling partners. However, a number of challenges would need to be addressed to develop such a process: (1) Ag(I) salts are oxidizing agents and can oxidize electron-rich Pd(0)-species to Pd(II), which could present a problem when using bromoarenes instead of the more reactive iodoarenes; (2) Ag(I) forms highly insoluble bromide salts, which would quickly remove the catalyst as the reaction progresses. Herein, we report the development of the first Ag/Pd cocatalyzed C–H arylation system and its application to the mild arylation of fluorobenzene derivatives with bromoarenes (Scheme e).

Results and Discussion

Development of a Ag/Pd Cocatalyzed C–H Arylation Process Using Bromoarenes as Coupling Partners

We began our investigation testing conditions we previously developed for the direct arylation of (fluoroarene)Cr(CO)3 with iodoarenes.[18] When these conditions were applied to the arylation of 3-fluorotoluene chromium tricarbonyl 1a with bromobenzene 2a, no reaction was detected (Table , entry 1), in line with our previous observation.[18] Our group recently reported that adding up to 60 mol % of PPh3 had a beneficial effect on the reaction rate for the Pd-catalyzed direct C–H arylation with iodoarenes, due to a faster Ag-mediated C–H activation step.[13] In this case, however, this additive did not enable reactivity (entry 2). The results in entries 1 and 2 suggested that the oxidative addition step with less reactive bromobenzene (vs iodoarenes) may be problematic. Accordingly, the addition of 20 mol % of SPhos, a more electron-donating ligand than PPh3, resulted in a small amount of desired product 3aa (entry 3). We postulated that deactivation of Pd(0) via oxidation to Pd(II) by the Ag(I) salt could be competing with the oxidative addition of bromoarene 2a. Thus, a reduction of the amount of Ag(I) salts employed could be beneficial by reducing the rate of undesired Pd oxidation. Our group has previously reported on the use of NMe4 salts as halide abstractors as an alternative to Ag(I) salts in a range of Pd- and Ru-catalyzed direct arylations.[19] We hypothesized that NMe4 salts may be able to regenerate active Ag(I)-species from unreactive AgBr and thus be suitable stoichiometric bromide abstractors for this process (Scheme e, box). This would allow the use of the Ag(I) salt to mediate C–H activation in a catalytic fashion, while slowing down the deleterious Pd(0)-oxidation pathway. To test this hypothesis, we lowered the Ag(I) salt loading to 0.1 equiv of Ag2CO3, in combination with 1.5 equiv of NMe4OC(CF3)3 (entry 4). Gratifyingly, the 7.5-fold reduction in the Ag loading led to a marked increase in reactivity, providing a 12% yield of 3aa. Further optimization revealed that Buchwald’s precatalyst Pd-G2 led to a significant improvement in the yield of 3aa to 51% (entry 5). Replacing the Ag2CO3 catalyst and AdCO2H additive, with just 20 mol % of AgOCOAd, in combination with PBuAd2 as a ligand further increased the yield of 3aa (entry 6). The use of a more active Pd precatalyst, Pd-G4, in which Cl is avoided, led to an additional small yield increase (entry 7). Increasing the loading of K2CO3 to 4 equiv resulted in 73% of 3aa (entry 8). Remarkably, the use of AgBr instead of AgOCOAd resulted in a significant increase in the reaction efficiency, affording an excellent 90% yield of 3aa (entry 9). This result reveals that the carboxylate base is not needed to carry out the C–H activation in this system, a role that may be played instead by either carbonate or the perfluoro-tert-butylalkoxide.[20] Conversely, the reaction did not proceed at all in the absence of either the Pd or the Ag catalysts (entries 10 and 11), confirming that both are essential for the observed reactivity. Note that the mass balance of the reaction was accounted for with unreacted 1a being recovered.

Table 1

Effect of the Change in Reaction Variables on the Direct Arylation of 3-Fluorotoluene Chromium Tricarbonyl 1e with Bromobenzene 2a

entry	[Ag] (equiv)	[Pd]	additives (equiv)	3aa (%)a
1	Ag2CO3 (0.75)	Pd(PPh3)4		0
2	Ag2CO3 (0.75)	Pd(PPh3)4	PPh3 (0.6)	<1
3	Ag2CO3 (0.75)	Pd(PPh3)4	SPhos (0.2)	6
4	Ag2CO3 (0.10)	Pd(PPh3)4	SPhos (0.2), NMe4OC(CF3)3 (1.5)	12
5	Ag2CO3 (0.10)	Pd-G2	SPhos (0.2), NMe4OC(CF3)3 (1.5)	51
6b	AgOCOAd (0.20)	Pd-G2	PBuAd2 (0.2), NMe4OC(CF3)3 (1.5)	61
7b	AgOCOAd (0.20)	Pd-G4	PBuAd2 (0.2), NMe4OC(CF3)3 (1.5)	65
8b	AgOCOAd (0.20)	Pd-G4	PBuAd2 (0.2), NMe4OC(CF3)3 (1.5), K2CO3 (4.0)d	73
9b,c	AgBr (0.20)	Pd-G4	PBuAd2 (0.2), NMe4OC(CF3)3 (1.5), K2CO3 (4.0)d	90
10b,c		Pd-G4	PBuAd2 (0.2), NMe4OC(CF3)3 (1.5), K2CO3 (4.0)d	0
11b,c	AgBr (0.20)		PBuAd2 (0.2), NMe4OC(CF3)3 (1.5), K2CO3 (4.0)d	0

Reaction carried out on a scale of 0.1 mmol of 1a and yield determined by 1H nuclear magnetic resonance (NMR) using nitrobenzene as internal standard.

1-AdCO2H not added to this reaction.

Reaction carried out at 75 °C.

4 equiv of K2CO3 was used in total.

Scope of the Ag/Pd Cocatalyzed C–H Arylation

Next, the scope of the direct C–H arylation of fluoroarene–chromium tricarbonyl complexes 1 with aryl bromides 2 as coupling partners (Table ) was investigated. The reaction tolerated a variety of functionalities at the para positions of the aryl bromide, affording the corresponding biaryls (4ba–4bp) in moderate to excellent yields. Halo groups such as −Cl (4be) and −F (4bf) showed excellent reactivity, widening the opportunity for further functionalization. Bromoarenes with trifluoromethyl (4bg) and nitro (4bm) electron-withdrawing groups afforded 82 and 65% yields, respectively. Aldehyde- and methylketone-bearing bromoarenes 2h and 2i were not compatible with our reaction conditions, with starting material recovered unreacted in both cases. On the other hand, a protected aldehyde (4bj) and a phenylketone (4bk) exhibited very good reactivity. Low reactivity was observed with cyano (4bl) substituent; nonetheless, CO2Me (4bn) and alkene (4bo) functional groups were compatible with the system. More electron-donating SMe (4bd), Me (4bb), and OMe (4bc) substituents gave good yields. This methodology was also applicable to meta- (4bq–4bu) and ortho-substituted bromoarenes (4bv–4bx). Additionally, bromoheteroarenes containing benzofuran (4by) and indole (4bz) cores reacted to afford biaryl products in 99 and 38% yields, respectively. Aryl triflates and tosylates were found to be unreactive in this system. Finally, the reaction of 1b with 2a was scaled up to a 2 mmol scale, giving the desired arylation product 4ba in an 87% yield.

Table 2

Scope of the Direct C–H Arylation of Fluoroarene–Chromium Tricarbonyl Complexes 1 with Bromoarenes 2a,b,c,d

Reactions carried out on a scale of 0.5 mmol of 1.

Performed on a 2 mmol scale.

Yield determined by 1H NMR analysis using nitrobenzene as an internal standard.

3 equiv of bromoarene used.

We then turned our attention to the exploration of the scope of fluoroarene–chromium complexes with bromobenzene 2a as the coupling partner. The reaction was compatible with both electron-donating and -withdrawing functionalities such as OMe (4ca), CH2OTIPS (4da), and CO2Me (4ga). When a methyl group was present at the meta position, the arylation primarily took place at the less hindered site (4aa–4aa′). Finally, when a substituent was present at para position, bisarylated adducts 4ga–4ha were obtained with 3 equiv of bromobenzene 2a in good to very good yields.

Mechanistic Proposal and Mechanistic Studies

The mechanistic proposal for our Ag/Pd cocatalyzed C–H arylation of fluorobenzene complex 1 has been outlined in Scheme . We hypothesized that phosphine-ligated Ag(I)-alkoxide III, formed in situ by the reaction of PBuAd2, AgBr, and NMe4OC(CF3)3, performs C–H activation on arene complex 1, forming Ag(I)-aryl species IV. These, in turn, enter into a Pd(0)/Pd(II) catalytic cycle through transmetalation with Pd(II)-alkoxide VI to form VII, while regenerating Ag(I)-species III. Bis-aryl-Pd species VII then undergoes reductive elimination to form the product 3.

Scheme 2

Proposed Mechanism for Ag(I)/Pd Cocatalyzed C–H Arylation of Fluorobenzene Derivatives with Bromoarenes

We first investigated the proposed Ag-catalyzed C–H activation step. By treating complex 1b with various components from our reaction system, in the presence of D2O, we studied the H/D exchange resulting from a reversible C–H activation process (Table ). The use of 20 mol % of AgBr only and in combination with either 20 mol % of PBuAd2 or 1.5 equiv NMe4OC(CF3)3 afforded no deuteration of 1b (Table , entries 1–3). As predicted, when all three components were used in combination, 50% deuteration of 1b was observed after only 2 h (Table , entry 4). This result indicates that perfluoro-tert-butoxide is a suitable base to carry out C–H activation in combination with Ag(I), despite its monodentate nature, but only in the presence of the phosphine ligand. Further addition of K2CO3 as a terminal base further enhanced the deuteration of 1b, yielding 83% of deuterated starting material 1b (Table , entry 5). The beneficial effect of K2CO3 could be explained through the removal of perfluoro-tert-butanol from the mixture, as shown in Scheme , thus favoring the C–H activation equilibrium toward IV. When 1b was instead treated with D2O under the standard coupling conditions but in the absence of AgBr, no deuteration or biaryl coupling was observed (Table , entry 6), further demonstrating key role of the Ag(I) salt in the C–H activation step. These results point toward a similar overall mechanism to those observed in our previous studies of Ag-mediated C–H activation,[13,16] albeit with subtly different kinetically relevant steps. Here, instead of rate-limiting transmetalation or C–H activation, we observe a situation where both of these steps are kinetically relevant.

Table 3

H/D Scrambling of Fluoroarene–Chromium Complex 1b

entry	conditions	1b/1b-da
1	AgBr (20 mol %)	100:0
2	PBuAd2 (20 mol %) + AgBr (20 mol %)	100:0
3	AgBr (20 mol %) + NMe4OC(CF3)3 (1.5 equiv)	100:0
4	PBuAd2 (20 mol %) + AgBr (20 mol %) + NMe4OC(CF3)3 (1.5 equiv)	50:50
5	PBuAd2 (20 mol %) + AgBr (20 mol %) + NMe4OC(CF3)3 (1.5 equiv) + K2CO3 (4.0 equiv)	17:83
6	Pd-G4 (5 mol %) + PhBr (1.5 equiv) + K2CO3 (4.0 equiv) + NMe4OC(CF3)3 (1.5 equiv)	100:0

Ratios were determined by 1H NMR.

Given the well-known affinity of Ag(I) for bromide,[21] we carried out stoichiometric NMR studies to determine whether NMe4OC(CF3)3 is indeed able to abstract the bromide and generate the proposed Ag(I)-alkoxide. When PBuAd2 and AgBr were mixed with NMe4OC(CF3)3, the formation of (PBuAd2)AgOC(CF3)3 could be confirmed by 31P and 19F NMR (see Figures S3 and S4 in the Supporting Information). The identity and structure of (PBuAd2)AgOC(CF3)3 (III) were confirmed by single-crystal X-ray diffraction analysis (Figure ). These experiments are consistent with both the formation of Ag-alkoxide III from AgBr and its proposed role on the C–H activation of 1 (Scheme ).

Figure 1

ORTEP plot of (PBuAd2)AgOC(CF3)3 (III). Selected bonds and angles: P–Ag, 2.332 Å; O–Ag, 2.086 Å; O–Ag–P 173.6°. All hydrogen atoms are omitted for clarity.

To further investigate the reaction mechanism, we applied reaction progress kinetic analysis (RPKA)[22] and variable-time normalization analysis (VTNA).[23] We began by examining the stability of the catalysts during the reaction by carrying out two kinetic runs under “same-excess” conditions (Figure ). A time-adjusted plot[24] (Figure a) from reactions carried out with initial concentrations of either 0.1 or 0.15 M of 1b (and an [excess] of 0.05 M) showed a small but reproducible deviation between the two traces. This suggested that either product inhibition and/or catalyst deactivation were occurring. We thought that perfluoro-tert-butanol, formed in the C–H activation step, may be responsible for the slight inhibition observed. When the same-excess experiment was repeated under standard conditions with 0.05 M of added perfluoro-tert-butanol, a good match was obtained (Figure b) between the two kinetic runs, confirming the small inhibitory effect of this side product and indicating that no significant catalyst decomposition is occurring. Therefore, VTNA could still be applied to determine the orders on the different components in the reaction (Figure ).[23] The kinetic analyses revealed positive orders on both catalysts, Pd and Ag, and arene 1b and an order zero on bromoarene 2b. These data are consistent with both the transmetalation (IV and VI to III and VII) and the C–H activation (III and 1 to IV) steps contributing to the rate of the reaction simultaneously, with the C–H activation step being in quasi-equilibrium. An order of 0.5 in Ag catalyst suggests the existence of an inactive Ag(I)-dimer resting state (such as II) in equilibrium with active monomeric Ag(I)-alkoxide III, which performs kinetically relevant C–H activation.[25] The partial order in Pd is typical of cocatalyzed processes where the order may range from 0 to 1, depending on the rate of the interconnecting step. For example, in our case, the partial order in Pd is consistent with a relatively fast, but still relevant, transmetalation step. An order of 0.7 in 1b indicates that the C–H activation step is an equilibrium displaced toward 1b. An order of 0 in bromoarene 2b indicates that the oxidative addition to Pd(0) is not involved in the rate of the reaction and suggests that this step is fast. The order in NMe4OC(CF3)3 salt could not be determined due to its very low solubility in toluene; however, kinetic runs with different stirring rates provided identical reaction profiles, suggesting that the order may be zero.[26] Taken together with the H/D exchange experiments in Table , these results are strongly consistent with the proposed mechanism. In support of our analysis, we simulated the proposed mechanism using COPASI,[27] which allowed the identification of a set of kinetic constants replicating the observed kinetic orders for all reaction components (see Figure S13 in the Supporting Information).

Figure 2

Time-adjusted (a) same-excess experiment and (b) same-excess experiment in the presence of F9-BuOH.

Figure 3

Determination of orders of reaction of the different components using VTNA.

One-Pot Protocol with in Situ Formation of Cr Complex

Finally, we considered whether we could further improve the practical utility of this C–H arylation methodology by developing a one-pot protocol that would start from the fluoroarene, rather than its Cr complex. Achieving such a process would require developing a mild set of complexation conditions that would be compatible with the subsequent arylation step. Typical standard conditions for the synthesis of (arene)Cr(CO)3 complexes involve reflux in Bu2O for prolonged periods and would therefore be unsuitable for a one-pot protocol. Previous studies have shown that polyarene–Cr complexes are able to undergo arene exchange under relatively mild conditions.[28] A survey of such polyarene complexes revealed (pyrene)Cr(CO)3 (1l) as a suitable Cr(CO)3 donor for the efficient formation of fluorobenzene complexes (Scheme S10 and Table S9 in the Supporting Information). Accordingly, a mixture of 1l and fluorotoluene 5b in dioxane heated at 90 °C for 16 h afforded 1b in an excellent 86% yield (Scheme a). Gratifyingly, this protocol for the preparation of 1b was found to be compatible with the subsequent arylation reaction conditions. Thus, simple addition to the same flask of bromoarene 2a and the other arylation components, followed by in situ decomplexation, afforded 3ba in a 54% yield (Scheme b). This proof-of-principle reaction sets the stage for the extension of this methodology with free fluoroarene substrates in one pot without isolating the chromium tricarbonyl complex, thus increasing its practicality and synthetic utility.

Scheme 3

(a) Arene Exchange and (b) One-Pot Sequential Complexation/C–H Arylation of Fluoroarene

Conclusions

In conclusion, we have developed the first Ag/Pd cocatalyzed C–H arylation process. This catalytic system is capable of regioselective direct C–H arylation of fluoroarenes with bromoarenes using π-complexation to Cr(CO)3 to enhance the fluoroarene reactivity. Key to developing this protocol was the identification of NMe4OC(CF3)3 as a halide abstractor capable of removing Br from AgBr, to regenerate the catalytic Ag-species. This system is able to operate under relatively mild conditions using Ag(I) to carry out the C–H activation and Pd to carry out the bromoarene oxidative addition and reductive elimination steps, in a synergistic system. Our proposed mechanism is fully consistent with H/D exchange experiments, in situ NMR identification of catalytic intermediates, and observed partial kinetic orders on both catalysts and the arene substrate. We have also developed a new one-pot sequential arene complexation/arylation technique for the synthesis of biaryls. This investigation opens the door for the future development of Ag/Pd cocatalyzed systems for C–H functionalization.

Experimental Section

General Procedure for Biaryl Synthesis via Pd/Ag Catalytic System

A mixture of Pd-G4 (18.6 mg, 0.025 mmol, 5 mol %), AgBr (18.8 mg, 0.1 mmol, 20 mol %), PBuAd2 (37.7 mg, 0.1 mmol, 20 mol %), K2CO3 (276.4 mg, 4.0 equiv, 2.0 mmol), NMe4OC(CF3)3 (232.6 mg, 1.5 equiv, 0.75 mmol), arene–chromium complex 1 (1.0 equiv, 0.5 mmol), and bromoarene 2 (1.5 equiv, 0.75 mmol) in PhCH3 (0.5 mL) was stirred at 75 °C for 16 h in a flame-dried reaction vial. The reaction setup was covered with aluminum foil to prevent photodecomposition. After this time, MnO2 (130.4 mg, 1.5 mmol, 3 equiv) in AcOH (3.0 mL) was added and stirred at room temperature for 30 min. The resulting mixture was filtered through celite, concentrated under reduced pressure, and purified through column chromatography to afford the desired biaryl products 4.