Synthesis of Triarylmethanes via Palladium-Catalyzed Suzuki-Miyaura Reactions of Diarylmethyl Esters

Abstract

The synthesis of triarylmethanes via Pd-catalyzed Suzuki-Miyaura reactions between diarylmethyl 2,3,4,5,6-pentafluorobenzoates and aryl boronic acids is described. The system operates at mild conditions and has a broad substrate scope, including the coupling of diphenylmethanol derivatives that do not contain extended aromatic substituents. This is significant as these substrates, which result in the types of triarylmethane products that are prevalent in pharmaceuticals, have not previously been compatible with systems for diarylmethyl ester coupling. Further, the reaction can be performed stereospecifically to generate stereo-inverted products. On the basis of DFT calculations, it is proposed that the oxidative addition of the diarylmethyl 2,3,4,5,6-pentafluorobenzoate substrate occurs via an S_N2 pathway, which results in the inverted products. Mechanistic studies indicate that oxidative addition of the diarylmethyl 2,3,4,5,6-pentafluorobenzoate substrates to (IPr)Pd(0) results in the selective cleavage of the O–C(benzyl) bond in part because of a stabilizing η³-interaction between the benzyl ligand and Pd. This is in contrast to previously described Pd-catalyzed Suzuki-Miyaura reactions involving phenyl esters, which involve selective cleavage of the C(acyl)–O bond, because there is no stabilizing η³-interaction. It is anticipated that this fundamental knowledge will aid the development of new catalytic systems, which use esters as electrophiles in cross-coupling reactions.

Graphical Abstract

Introduction

Triarylmethanes are common motifs in natural products,¹ fluorescent probes,² organic dyes,³ metal ion sensors,⁴ and active pharmaceutical ingredients.⁵ For example, triarylmethanes have anti-viral,⁶ anti-tuberculosis,⁷ and anti-breast cancer⁸ properties (Figure 1). The most common method to synthesize triarylmethanes is through Lewis or Brønsted acid catalyzed Friedel-Crafts reactions of diarylmethanols or their derivatives with arenes, however, these reactions often have limited substrate scopes and poor selectivity.^5b In recent years, a number of transition metal-catalyzed C(sp²)–C(sp³) cross-coupling reactions have been developed for the synthesis of triarylmethanes. These methods typically utilize non-classical electrophiles, such as diarylmethyl -ammonium salts,⁹ -sulfones,¹⁰ and -alcohol derivatives.¹¹ In particular, alcohol derivatives, for example esters, are attractive as electrophiles for the synthesis of triarylmethanes because alcohols are abundant, stable, and diverse building blocks, which can easily be derivatized.¹²

Figure 1.

Examples of medicinally relevant triarylmethanes.

In 2005, Kuwano et al. demonstrated that benzyl acetates can be used in Pd-catalyzed Suzuki-Miyaura reactions to generate diarylmethanes (Figure 2a).¹³ Although this work was not extended to diarylmethyl esters, subsequent studies by Watson et al. and Jarvo et al. showed that diarylmethyl esters can be used in Ni-catalyzed Suzuki-Miyaura reactions to generate triarylmethanes stereospecifically (Figure 2b).^11,14 The later reactions are currently limited to electrophiles with extended aromatic systems, such as naphthyls and biaryls, and are not compatible with diphenylmethanol derivatives, which are prevalent in natural products and pharmaceuticals.^7,8b Nevertheless, they are the first examples of a synthetically valuable transformation and are fundamentally interesting because of their selectivity. Specifically, the high yields of triarylmethane products suggest that the diarylmethyl esters undergo oxidative addition exclusively across the O–C(benzyl) bond (Figure 3a).^11,13,15 In contrast, depending on the catalyst and exact nature of the substrate some aryl esters can undergo oxidative addition across either the O–C(aryl) (Figure 3a)¹⁶ or C(acyl)–O bond (Figure 3b),¹⁷ to give either biaryl or ketone products, respectively, after transmetallation and reductive elimination. Further, in some cases following C(acyl)–O bond cleavage a decarbonylative step can occur, which ultimately leads to biaryl products (Figure 3c).¹⁸ At this stage, the exact reasons for the differences in selectivity are not clear, especially in regard to why the O–C(benzyl) bonds of diarylmethyl esters are cleaved and not the C(acyl)–O bonds.

Figure 2.

Summary of previous work on Suzuki-Miyaura reactions of diarylmethyl esters (a, b) and comparison to this work (c).

Figure 3.

Bonds that can be cleaved in oxidative addition reactions involving esters.

Recently, our group and others have reported Pd-catalyzed Suzuki-Miyaura reactions of phenyl esters in which the C(acyl)–O bond is selectively cleaved to generate ketones in high yields.^18c,19 Based in part on Kuwano et al.’s work with benzyl acetates,¹³ we hypothesized that if the substrate was changed to a diarylmethyl ester we could instead selectively cleave the O–C(benzyl) bond, which would ultimately lead to the formation of triarylmethanes. Herein, we report the first examples of Pd-catalyzed Suzuki-Miyaura reactions of diarylmethyl 2,3,4,5,6-pentafluorobenzoates to selectively generate triarylmethanes (Figure 2c). Importantly, the reaction is compatible with diphenylmethanol derivatives that do not contain extended aromatic substituents and enantioenriched diarylmethyl 2,3,4,5,6-pentafluorobenzoates can be coupled with high stereospecificity to give stereo-inverted products. The latter observation, supported by DFT calculations, suggests that oxidative addition of the substrate to the (IPr)Pd(0) active catalyst occurs via an S_N2-type mechanism and we were able to isolate the oxidative addition product from the reaction of a diarylmethyl 2,3,4,5,6-pentafluorobenzoate to an (IPr)Pd(0) complex. Additional DFT calculations demonstrate it is kinetically and thermodynamically more favorable to cleave the O–C(benzyl) bond of benzyl benzoate during oxidative addition to (IPr)Pd(0) in part due to an η³-interaction between the benzyl group and the Pd. In contrast, it is kinetically and thermodynamically more favorable to cleave the C(acyl)–O bond of phenyl benzoate during oxidative addition to (IPr)Pd(0), which proceeds via a concerted mechanism. Overall, this work provides a notable synthetic advance due to the improved substrate scope for the formation of triarylmethanes and valuable fundamental information about selectivity in the cleavage of benzoates.

Results and Discussion

Reaction Discovery and Development

In preliminary work, we demonstrated that we could couple benzyl benzoate with phenyl boronic acid to selectively generate a diarylmethane product using (η³-1-^tBu-indenyl)Pd(IPr)(Cl) (IPr = 1,3-bis(2,6-diisopropyl-phenyl)-1,3-dihydro-2H-imidazol-2-ylidene) as a precatalyst (Figure 4). This is in agreement with Kuwano et al.’s results¹³ and is indicative of selective cleavage of the O–C(R) (R = benzyl) bond. It is notable that when benzyl benzoate is replaced with phenyl benzoate under the same reaction conditions we selectively form benzophenone, which is consistent with selective cleavage of the C(acyl)–O bond (Figure 4). Our results with benzyl benzoate suggested that we could leverage the observed cleavage of the O–C(R) (R = benzyl) bond to develop a selective method for the synthesis of synthetically valuable triarylmethanes from diphenylmethyl esters through the careful selection of the leaving group on the ester and the optimization of the reaction conditions.

Figure 4.

Comparison of Pd-catalyzed Suzuki-Miyaura reactions of phenyl and benzyl benzoate using (η³-1-^tBu-indenyl)Pd(IPr)(Cl) as the precatalyst under identical conditions.

We initially synthesized diphenylmethyl-acetate, -benzoate, and −2,3,4,5,6-pentafluorobenzoate via a straightforward reaction between diarylmethanol and the appropriate carboxylic acid (see SI).^17a Subsequently, using 1 mol% of (η³-1-^tBu-indenyl)Pd(IPr)(Cl) as the precatalyst, a 4:1 mixture of toluene/ethanol as the solvent, and 2 equivalents of K₂CO₃ as the base, we assessed these esters as substrates in Suzuki-Miyaura reactions with 4-methoxyphenylboronic acid at room temperature (Figure 5a). After 6 hours, Suzuki-Miyaura reactions with diphenylmethyl-acetate and -benzoate give 28% and 35% of (diphenyl)(4-methoxyphenyl) methane, respectively, while the reaction with diphenylmethyl 2,3,4,5,6-pentafluorobenzoate yields 85% of the desired product. The observed trend in reactivity of the diphenylmethyl esters correlates with the pK_as of the corresponding carboxylic acids, with acetic acid and benzoic acid having similar pK_a values and 2,3,4,5,6-pentafluorobenzoic acid having a much lower pK_a value.²⁰ We suggest that the pK_a values are reflective of the substrates relative ability to be activated by the Pd center via an S_N2-type mechanism, as well as the strength of the interaction between Pd and the anion (vide infra).¹⁵ⁱ These reactions are some of the first examples of Suzuki-Miyaura reactions of diphenylmethanol derivatives that do not contain extended aromatic substituents.^15k,21 We also compared the reactions using esters as electrophiles to those using diphenylmethyl chloride and diphenylmethyl bromide as substrates (Figure 5a). Although diphenylmethyl chloride gives a high yield (77%), the reaction results in the formation of (ethyl)(diphenylmethyl)ether as a side product (~5%) (Figure 5b), presumably as a result of base mediated nucleophilic substitution of the alcohol solvent on the halide. In agreement with this proposal, the more reactive diphenylmethyl bromide gives an even lower yield (21%), and an increased quantity of the ether side-product (~33%).

Figure 5.

(a) Pd-catalyzed Suzuki-Miyaura reactions of diphenylmethyl esters and halides and (b) side reaction observed with aryl halides. ^aConditions: diphenylmethyl-X (0.05 mmol), 4-methoxyphenylboronic acid (0.075 mmol), K₂CO₃ (0.1 mmol), (η³-1-^tBu-indenyl)Pd(IPr)(Cl) (0.0005 mmol), toluene (0.4 mL), and EtOH (0.1 mL). ^aYields are an average of two runs determined by GC-FID using the conversion of starting material to product.

We performed a series of Suzuki-Miyaura reactions between diphenylmethyl 2,3,4,5,6-pentafluorobenzoate and 4-methoxyphenylboronic acid to evaluate the factors that are important in obtaining high yields. In the absence of precatalyst no product is observed, and the majority of the starting material is converted to ethyl 2,3,4,5,6-pentafluorobenzoate, indicating that the diphenylmethyl 2,3,4,5,6-pentafluorobenzoate is not stable under the reaction conditions (Table 1, Entry 2). In the presence of (η³-1-^tBu-indenyl)Pd(IPr)(Cl), the desired coupling outcompetes the side-reaction and quantitative conversion occurs in 24 hours (Entry 3). Increasing the temperature to 40 °C, results in a quantitative yield after 4 hours (Entry 4). Other common precatalysts, such as (η³-cinnamyl)Pd(IPr)Cl and PEPPSI-IPr, give moderate but reduced yields relative to (η³-1-^tBu-indenyl)Pd(IPr)(Cl) likely due to their slower rates of activation under our reaction conditions (Entries 5 & 6).²² A similar yield was obtained when (η³-1-^tBu-indenyl)Pd(IPr)Cl was replaced with [Pd(IPr)(μ-Cl)Cl]₂ but this precatalyst was not pursued further, as it lacks a precursor that can be readily used for ligand screening (Entry 7).²³ Changing the ancillary ligand to other frequently utilized NHC ligands, such as SIPr, IMes, and IPr*^OMe (Entries 8–10), or phosphine ligands, such as PCy₃, XPhos, or SPhos (Entries 11–13), gives reduced yields compared to IPr. We have observed similar trends in related Suzuki-Miyaura reactions involving phenyl benzoates.^19f The choice of base is also crucial and although high yields are obtained with K₂CO₃, lower yields are observed with other common inorganic bases, such as K₃PO₄ or Na₂CO₃ (see SI). In the absence of a protic co-solvent no product is observed (Entry 14). Although the reaction worked comparably in the presence of methanol or ethanol (Entries 16 & 1), faster transesterification occurs in methanol to generate the methyl 2,3,4,5,6-pentafluorobenzoate by-product, which unnecessarily consumes starting material. Further, the yield suffered when water or isopropanol are used instead of ethanol (Entries 15 & 17). The role of the alcohol co-solvent may be in solubilizing the reagents, assisting in precatalyst activation,^22b or facilitating transmetallation by displacing the carboxylate with an alkoxy ligand.¹³

Table 1.

Optimization table for the Pd-catalyzed Suzuki-Miyaura reaction of diarylmethyl 2,3,4,5,6-pentafluorobenzoate with 4-methoxyphenylboronic acid.

Entry	Deviation from optimized conditions	GC conversiona
1	No change	85%
2	No precatalyst	0%b
3	24 h instead of 6 h	>99%
4	40 °C instead of r.t. and 4 h instead of 6 h	>99%
5	(η3-cinnamyl)Pd(IPr)Cl instead of (η3-1-tBu-indenyl)Pd(IPr)Cl	73%
6	PEPPSI-IPr instead of (η3-1-tBu-indenyl)Pd(IPr)Cl	64%
7	0.5 mol% [Pd(IPr)(μ-Cl)Cl]2 instead of (η3-1-tBu-indenyl)Pd(IPr)Cl	86%
8	SIPr instead of IPr	62%
9	IMes instead of IPr	16%
10	IPr*OMe instead of IPr	40%
11	PCy3 instead of IPr	2%c
12	XPhos instead of IPr	3%
13	SPhos instead of IPr	46%d
14	Toluene only	<1%
15	H2O instead of EtOH	69%
16	MeOH instead of EtOH	89%e
17	iPrOH instead of EtOH	56%

Conditions: diphenylmethyl ester (0.05 mmol), 4-methoxyphenylboronic acid (0.075 mmol), base (0.1 mmol), precatalyst (0.0005 mmol), solvent (0.5 mL).

^aYields are an average of two runs determined by GC-FID using the conversion of starting material to product.

^b64% conversion to F₅C₆COOC₂H₅ and 36% starting material remaining.

^c67% conversion to F₅C₆COOC₂H₅.

^d32% conversion to F₅C₆COOC₂H₅.

^e9% conversion to F₅C₆COOCH₃.

Substrate Scope

Given the success of Suzuki-Miyaura reactions involving diphenylmethyl 2,3,4,5,6-pentafluorobenzoate, we explored the substrate scope for the coupling of a range of diarylmethyl 2,3,4,5,6-pentafluorobenzoates with phenylboronic acid under our optimized conditions (Figure 6). Electron neutral diarylmethyl esters (6a & 6b), including fluorene-9-ester (6b), result in high yields, 96 and 94% respectively, of the desired product. The latter result is notable because fluorenes are important motifs in biologically active molecules,²⁴ as monomers in polymer chemistry,²⁵ and as dyes for solar cell applications, and our system provides a facile method for functionalization.²⁶ Additionally we were able to generate 6a on a 1 mmol scale in 95% isolated yield (see SI), which demonstrates that our method is scalable. The reaction is also compatible with esters containing electron-donating (6c) and -withdrawing (6d & 6e) substituents. This includes the doubly cyano substituted ester (6e) which is coupled in 86% yield. This is significant because cyano groups often coordinate to Pd and inhibit catalysis. It is more difficult to couple substrates with ortho-substitution on an aryl group of the diarylmethyl ester, such as 6f. Even with higher precatalyst loading (4 mol%) and longer reactions times (16 h), when ethanol is used as the co-solvent, transesterification to generate the ethyl 2,3,4,5,6-pentafluorobenzoate competes with the desired cross-coupling reaction and only a 40% NMR yield of cross-coupled product is observed. However, an 87% yield is obtained by increasing the precatalyst loading to 4 mol%, the temperature to 80 °C, the time to 16 h, and switching to water as the co-solvent to prevent transesterification. Consistent with the challenges in coupling ortho-substituted substrates, (1-napthyl)(phenyl)methyl 2,3,4,5,6-pentafluorobenzoate (6g) gives minimal yield under the optimized conditions (6% NMR yield). When more forcing conditions are utilized, with the use of water in place of ethanol as a co-solvent, a yield of 77% is obtained. Under the same conditions, the even more sterically bulky bis(1-naphthyl)methyl 2,3,4,5,6-pentafluorobenzoate (6h) gives an 87% yield. In contrast, we are able to couple the less sterically bulky substrate (2-napthyl)(phenyl)methyl 2,3,4,5,6-pentafluorobenzoate (6i) without major modification to our optimized conditions (16 h instead of 4 h). Pyridyl groups are common in pharmaceuticals but are often difficult substrates for cross-coupling reactions due to coordination of the heterocycle to Pd.²⁷ Using our more forcing conditions, including the replacement of ethanol with water as a co-solvent, we can couple a 2-pyridyl containing electrophile (6j) in 49%.

Figure 6.

Isolated and NMR yields for Pd-catalyzed Suzuki-Miyaura reactions of diarylmethyl 2,3,4,5,6-pentafluorobenzoates with (4-methoxyphenyl)boronic acid. Conditions for isolated yields: Ester (0.2 mmol), phenylboronic acid (0.3 mmol), K₂CO₃ (0.4 mmol), (η³-1-^tBu-indenyl)Pd(IPr)Cl (0.0002 mmol), toluene (1.6 mL), ethanol (0.4 mL). Conditions for NMR yields: Ester (0.05 mmol), phenylboronic acid (0.075 mmol), K₂CO₃ (0.1 mmol), (η³-1-^tBu-indenyl)Pd(IPr)Cl (0.0005 mmol), toluene (0.4 mL), ethanol (0.1 mL). NMR yields were determined using 1,2,4,5-tetramethyl benzene as an internal standard. ^a4 mol% (η³-1-^tBu-indenyl)Pd(IPr)Cl instead of 1 mol%. ^b16 h instead of 4 h. ^c80 °C instead of 40 °C. ^dWater instead of ethanol.

Enantiomers of chiral molecules often have different biological activity due to variations in their binding affinity with chiral targets.²⁸ Therefore, there are advantages to accessing single enantiomers of triarylmethanes.¹ We hypothesized that our method could be compatible with producing single enantiomers of triarylmethanes if enantioenriched diarylmethyl 2,3,4,5,6-pentafluorobenzoates esters were utilized as substrates. The enantioenriched substrates, (S)-(2-naphthyl)(phenyl)methyl 2,3,4,5,6-pentafluorobenzoate and (S)-(4-methylphenyl)(phenyl)methyl 2,3,4,5,6-pentafluorobenzoate, were readily synthesized from the appropriate (S)-diarylmethanol by following a literature procedure involving the reaction of an in situ generated phenylzinc species and an aldehyde catalyzed by a chiral pyrrolidine ligand (see SI).²⁹ The subsequent coupling of (S)-(2-naphthyl)(phenyl)methyl 2,3,4,5,6-pentafluorobenzoate with 3-furylboronic acid under our optimized conditions results in the formation of triarylmethane 7a in 97% yield and 99% es, indicating the reaction proceeds with high enantiospecificity (Figure 7). Comparison of the polarimetry data of isolated 7a ([α]²⁰_D −23.25 (c 0.400, CDCl₃)) with literature data ([α]²⁹_D −22.0 (c 1.00, CDCl ₃))^11a suggests the formation of the (R)-isomer, which is consistent with a pathway involving inversion. Similarly, the reaction of (S)-(4-methylphenyl)(phenyl)methyl 2,3,4,5,6-pentafluorobenzoate with (4-methoxy)phenylboronic acid generates the enantioenriched triarylmethanes 7b in 88% yield and 90% es. Given that we propose the mechanism of the reaction involves oxidation addition, transmetallation and reductive elimination (vide infra), the stereochemical inversion of the (S)-diarylmethyl 2,3,4,5,6-pentafluorobenzoates to the (R)-triarylmethanes suggests that oxidative addition proceeds via an S_N2 pathway. This is because transmetallation and reductive elimination typically occur with retention of configuration in Suzuki-Miyaura reactions,³⁰ and therefore the stereochemistry of the product is likely based on oxidative addition. Our work is consistent with previous studies utilizing NHC-Ni catalysts for the stereoselective coupling of naphthyl and biaryl diarylmethyl esters, which also proceed with inversion and propose that oxidative addition is responsible for the observed stereochemistry.^11a,15i

Figure 7.

Isolated yields and enantioselectivity for Pd-catalyzed stereospecific Suzuki-Miyaura reactions of diarylmethyl 2,3,4,5,6-pentafluorobenzoates. Conditions for isolated yields: Enantioenriched ester (0.05 mmol), arylboronic acid (0.075 mmol), K₂CO₃ (0.1 mmol), (η³-1-^tBu-indenyl)Pd(IPr)Cl (0.0005 mmol), toluene (0.4 mL), ethanol (0.1 mL).

The scope of the boronic acid coupling partner was evaluated using diphenylmethyl 2,3,4,5,6-pentafluorobenzoate as the electrophile (Figure 8). The reaction is compatible with electron-neutral, -donating, and -withdrawing aryl boronic acids (8a-h). In a noteworthy result, 4-methylester- (8f) and 4-acetyl- (8g) phenylboronic acids give yields of 87% and 70%, respectively, which are not only electron-withdrawing but can be further functionalized, for example via nucleophilic addition reactions.³¹ The reaction is tolerant of ortho-substituted aryl boronic acids (8i-j). However, while 2,6-dimethoxyphenyl boronic acid (8l) could be coupled in high yield (although difficulty in isolation resulted in a lower isolated yield), minimal product is observed with 2,4,6-trimethylphenyl boronic acid (8m) even with higher precatalyst loadings. 1- and 2-napthyl boronic acids (8j & 8k) are coupled in yields of 83% and 80%, respectively. Heteroaryl-substituted triarylmethanes derivatives are common in pharmaceutical compounds,^7b,32 but cross-coupling reactions of diarylmethanol derivatives with heteroaryl boronic acids are limited,^11a,33 possibly due to the instability of heteroaryl nucleophiles.³⁴ Our system is able to couple 2-furyl (8n) and 3-thiophene (8o) boronic acids in yields of 98% and 97%, respectively, but 2-substituted thienyl boronic acid gives low conversion (see SI). Benzo[b]thien-2-yl- (8p) and boc-protected pyrrole- (8q) boronic acids give yields of 80% and 70%, respectively, but require longer reaction times. Unfortunately, pyridyl boronic acids could not be coupled (see SI), likely due to protodeborylation or coordination of the nitrogen atom of the pyridine ring to the catalyst.

Figure 8.

Isolated and NMR yields for Pd-catalyzed Suzuki-Miyaura reactions of diphenylmethyl 2,3,4,5,6-pentafluorobenzoate with boronic acids. Conditions for isolated yields: Ester (0.2 mmol), boronic acid (0.3 mmol), K₂CO₃ (0.4 mmol), (η³-1-^tBu-indenyl)Pd(IPr)Cl (0.0002 mmol), toluene (1.6 mL), ethanol (0.4 mL. Conditions for NMR yields: Diphenylmethyl 2,3,4,5,6-pentafluorobenzoate ester (0.05 mmol), (4-methoxyphenyl)boronic acid (0.075 mmol), K₂CO₃ (0.1 mmol), (η³-1-^tBu-indenyl)Pd(IPr)Cl (0.0005 mmol), toluene (0.4 mL), ethanol (0.1 mL). NMR yields were determined using 1,2,4,5-tetramethyl benzene as an internal standard. ^a8 h instead of 4 h. ^b16 h instead of 4 h. ^c4 mol% (η³-1-^tBu-indenyl)Pd(IPr)Cl instead of 1 mol%.

Overall, our substrate scope demonstrates that we have developed a mild and general method for the synthesis of triarylmethanes, containing both extended and non-extended aromatic groups, from readily available diarylmethyl esters. In particular, the ability of our system to couple diphenylmethyl esters that do not contain extended aromatic substrates is significant, as these substrates have not previously been utilized. Further, our system is tolerant of a variety of different functional groups, is compatible with heterocyclic substrates, and is stereospecific. As a result, we expect that it will be valuable for the synthesis of medicinally relevant triarylmethanes.

Mechanistic Studies

The high yields of triarylmethanes that are observed in Suzuki-Miyaura reactions involving diarylmethyl 2,3,4,5,6-pentafluorobenzoates indicate that the reaction must be proceeding with high selectivity, consistent with the selective cleavage of the O–C(benzyl) bond. Further, there is no evidence for cleavage of the C(acyl)–O bond, which presumably occurs readily when a phenyl ester is utilized instead of a benzyl ester (Figure 4). Previous studies have hypothesized that oxidative addition to Pd(0) is responsible for the differences in selectivity between phenyl- and benzyl-esters but there are few well-defined examples of the oxidative addition of these substrates and the reaction pathways (e.g. concerted versus S_N2) have not be probed.³⁵ In seminal work, Yamamoto et al. demonstrated differences in oxidative addition for aryl- and benzyl-trifluoroacetates to phosphine ligated Pd(0) complexes.^15b,35b,35c Specifically, aryl trifluoroacetates were shown to oxidatively add across the C(acyl)–O bond, while benzyl trifluoroacetates underwent oxidative addition across the O–C(benzyl) bond. However, the mechanistic origins for this difference were not elucidated and their applicability to our system, which features a different ester and ancillary ligand, was unclear. Therefore, we studied the oxidative addition of the type of esters used in this work, such as phenyl- and benzyl-benzoates, to the proposed active catalytic species (IPr)Pd(0).³⁶

Initial evaluation of the oxidative addition of phenyl benzoate with the Pd(0) source (IPr)Pd(0)(styrene)₂ resulted in slow decomposition to Pd black and (IPr)₂Pd. The use of more electron withdrawing electrophiles is known to result in more facile oxidative addition and also stabilize the resulting transition metal complexes.³⁷ In this case, treatment of phenyl 2,3,4,5,6-pentafluorobenzoate with (IPr)Pd(styrene)₂ at room temperature resulted in the slow formation of new product(s) as determined by NMR spectroscopy, but this was followed by decomposition, which prevented full characterization. We hypothesized that an ortho-coordinating ligand on the benzoate group, such as diphenylphosphine, would stabilize the putative three-coordinate oxidative addition complex by binding to the open coordination site on the metal center. The reaction of phenyl 2-(diphenylphosphino)benzoate with (IPr)Pd(styrene)₂ at room temperature resulted in the clean formation of a product, 1, with a single resonance in the ³¹P NMR spectrum at 50.4 ppm, which was isolated in 75% yield (Figure 9a). X-ray crystallography confirmed that 1 is the result of oxidative addition of the C(acyl)–O bond to (IPr)Pd(0) with coordination of the phosphine (Figure 9b). Interestingly, the crystal structure shows that the C(acyl) ligand is trans to the OPh ligand and the phosphine ligand is trans to the IPr ligand. It would be expected, however, that after concerted oxidative addition the C(acyl) ligand and OPh ligand would be cis to one another, which suggests that the ligands can rearrange on the metal center either by decoordination of the phosphine or OPh ligands. Nevertheless, the formation of 1 indicates that (IPr)Pd(0) is capable of facile cleavage of the C(acyl)–O bond of a phenyl ester. Notably, this is the first example of a well-defined oxidative addition product from the addition of a phenyl ester to Pd(0) that does not undergo decarbonylation.

Figure 9.

(a) Oxidative addition of phenyl 2-(diphenylphosphino)benzoate to (IPr)Pd(0) to form 1. (b) ORTEP (30% probability) of 1. Hydrogen atoms and methyl groups associated with the iso-propyl substituents of IPr omitted for clarity. Selected bond distances and angles Pd1-C1 1.971(9) Å, Pd1-C46 1.975(7) Å, Pd1-P1 2.287(2) Å, Pd1-O1 2.107(6) Å; C46-Pd1-O1 90.2(2)°, C1-Pd1-C46 89.5(3)°, P1-Pd1-O1 96.0(1)°, P1-Pd1-C1 84.0(2)°.

Next, we investigated the reaction of benzyl benzoate derivatives with (IPr)Pd(styrene)₂. In an analogous fashion to the reaction with the unsubstituted phenyl benzoate, the reaction of benzyl benzoate with (IPr)Pd(styrene)₂ resulted in slow decomposition to Pd black and (IPr)₂Pd. However, when we performed the reaction of (IPr)Pd(styrene)₂ with 2,3,4,5,6-pentafluorobenzoate, a substrate used in catalysis, one new product, 2, is formed and isolated in a 65% yield (Scheme 1).³⁸ In the ¹H NMR spectrum of 2, the resonance associated with the methine proton of the benzyl group is shifted up-field relative to that of the free ester (4.78 vs 7.19 ppm), consistent with the presence of a metal center in close proximity to the methine proton.³⁹ The ¹³C NMR spectrum includes a resonance at 177.0 ppm and two new peaks are observed in the IR spectrum at 1633 cm⁻¹ and 1343 cm⁻¹. This is indicative of the presence of a carbonyl group. Overall, our NMR data suggests that 2 is the oxidative addition product from cleavage of the O–C(benzyl) bond of 2,3,4,5,6-pentafluorobenzoate. However, we were unable to identify the exact structure of 2 as despite repeated attempts single crystals for X-ray diffraction could not be obtained. Possible structures of 2 include those with the diarylmethyl ligand binding in either an η¹ or η³-fashion, and the 2,3,4,5,6-pentafluorobenzoate ligand binding in a κ¹ or κ²-manner (Scheme 1). Further, it is possible that the 2,3,4,5,6-pentafluorobenzoate anion is not coordinated to Pd and an ion pair is generated. Nevertheless, our results indicate that oxidative addition of 2,3,4,5,6-pentafluorobenzoate with cleavage of the O–C(benzyl) bond is facile and we suggest that this is the first step in catalysis. Consistent with this proposal, when 2 is used as a catalyst in the coupling of diphenylmethyl 2,3,4,5,6-pentafluorobenzoate with (4-methoxy)phenylboronic acid under our optimized conditions (Equation 1), complete conversion to (2-methoxyphenyl)(diphenyl)methane is observed, indicating that 2 is a kinetically competent catalyst.

Scheme 1.

Formation of oxidative addition complex (IPr)Pd(CHPh₂)(OOCC₆F₅) (2) and possible structures of 2.

(eq 1)

DFT calculations were performed to help understand the observed selectivity in the oxidative addition of phenyl- and benzyl-benzoates to (IPr)Pd(0).⁴⁰ Phenyl and benzyl-benzoate were used as model substrates given our results in Figure 4, showing that they undergo selective cleavage of the C(acyl)–O bond and O–C(benzyl) bond, respectively. For both substrates, the concerted oxidative addition pathway was computed for the heterolytic cleavage of the C(acyl)–O and the O–C(aryl) bonds. In the case of benzyl benzoate, an S_N2 pathway was also calculated for the cleavage of the O–C(benzyl) bond.

The oxidative addition transition state for C(acyl)–O cleavage in phenyl benzoate, TS_Ph(C_Ac–O), is shown in Figure 10. In addition to the C(acyl)–O bond cleavage (1.71 Å), this transition state also involves the concerted formation of the Pd–C(acyl)Ph (2.08 Å) and Pd–OPh (2.16 Å) bonds. The three atoms involved in the bond rearrangement form a 3-membered Pd–O–C metallacycle, consistent with literature precedent.^17a In contrast, the transition state for O–C(aryl) bond cleavage, in phenyl benzoate TS_Ph(O–C_Ar), has a 5-membered metallacycle geometry, in which one of the two carboxylate O atoms breaks the bond with the Ph ring (2.04 Å), whereas the other forms a new bond with Pd (2.20 Å). The full relaxation of these two transition states to the oxidative addition products yields the complexes P_Ph(C_Ac–O) and P_Ph(O–C_Ar) with the expected Ph(O)C–Pd–OPh and Ph(O)CO–Pd–Ph moieties, respectively (Figure 10).⁴¹ Importantly, the cleavage of the C(acyl)–O bond in phenyl benzoate is both kinetically (8.8 vs 19.0 kcal/mol) and thermodynamically (−4.0 vs 4.6 kcal/mol) preferred compared with cleavage of the O–C(aryl) bond, consistent with our experimental results (Figure 4).

Figure 10.

DFT calculations comparing the oxidative addition of phenyl benzoate via C(acyl)–O and O–C(aryl) bond cleavage. Relative energies in kcal/mol.

The concerted transition states associated with cleavage of the C(acyl)–O and O–C(benzyl) bond of benzyl benzoate are shown in Figure 11, along with the transition state associated with cleavage of the O–C(benzyl) bond via an S_N2 mechanism. The geometry of TS_Bz(C_Ac–O) is very similar to TS_Ph(C_Ac–O), with the cleavage and formation of the C(acyl)–O (1.96 Å), Pd–C(acyl)Ph (2.02 Å) and Pd–OBz (2.11 Å) bonds in a 3-membered palladacycle. The transition state for O–C(aryl) bond cleavage, TS_Bz(O–C_Bz), is also similar to TS_Ph(O-C_Ar), involving both O atoms of the carboxylate group, one cleaving the bond to the benzyl (1.99 Å), and the other forming a new bond with Pd (2.41 Å). However, TS_Bz(O–C_Bz) has a distinct feature. Specifically, there is an η³-interaction between the metal center and the benzyl moiety, with Pd–C interatomic distances of 2.71, 2.34 and 2.17 Å for the C_Bz, C_ipso and C_ortho atoms, respectively.⁴² This interaction becomes a covalent η³-bond in the P_Bz(O–C_Bz) product, with distances shortened to 2.09, 2.27 and 2.39 Å, respectively. This is likely the main reason why the P_Bz(O–C_Bz) product is lower in energy than the P_Bz(C_Ac–O) product (−9.7 vs 6.0 kcal/mol), although the transition state energy associated with TS_Bz(C_Ac–O) is still lower than that observed for TS_Bz(O–C_Bz) (12.5 vs 20.7 kcal/mol).

Figure 11.

DFT calculations comparing the oxidative addition of benzyl benzoate via concerted C(acyl)–O and O–C(aryl) bond cleavage and also an S_N2 pathway. Relative energies in kcal/mol.

The cleavage of the O–C(benzyl) bond can also occur via an S_N2 pathway. The S_N2 mechanism follows a lower energy pathway than either the concerted O–C(benzyl) or C(acyl)–O bond cleavage, as the barrier associated with the S_N2-TS_Bz(O–C_Bz) transition state, 6.4 kcal/mol, is the lowest found for this substrate (Figure 11). The two main structural features of S_N2-TS_Bz(O–C_Bz) are: (1) the η³-interaction between the metal center and the C_Bz (2.32 Å), C_ipso (2.29 Å), and C_ortho (2.25 Å) atoms, similar to that of the concerted TS_Bz(O–C_Bz) transition state, and (2) the linear arrangement between the forming Pd–C_Bz (2.32 Å) and the breaking C_Bz–O (2.08 Å) bonds, with a Pd–C_Bz–O angle of 173.6°, and C_Bz adopting a distorted trigonal bipyramid geometry. The preference for the S_N2 pathway is in agreement with the inversion of configuration observed in the Pd-catalyzed Suzuki-Miyaura reactions of enantioenriched (diaryl)methyl 2,3,4,5,6-pentafluorobenzoate (Figure 7), as well as that seen in the literature with (NHC)Ni catalysts.^11b,15i The product of the S_N2 pathway, S_N2-P_Bz(O–C_Bz), is an ion pair between the (IPr)Pd(η³-Bz)⁺ cation and the PhCO₂⁻ anion (with a natural charge q = ±0.78 value). This ion pair is more stable than the P_Bz(C_Ac–O) complex (1.6 vs 6.0 kcal/mol) and can be further stabilized by rearranging into the P_Bz(O–C_Bz) complex (1.6 vs −9.7 kcal/mol).⁴³ Thus, O–C(benzyl) cleavage is both kinetically and thermodynamically preferred over C(acyl)–O cleavage. Overall, the computational results explain the experimental observation that for phenyl benzoate the C(acyl)–O bond is cleaved whereas for benzyl benzoate the O–C(benzyl) bond is cleaved.

In our experimental work, we demonstrated that a Suzuki-Miyaura reaction with diphenylmethyl 2,3,4,5,6-pentafluorobenzoate is more efficient than the corresponding reaction with diphenylmethyl benzoate (Figure 5). To investigate this observation, the transition state for cleavage of O–C(benzyl) bond of benzyl 2,3,4,5,6-pentafluorobenzoate via an S_N2-type oxidative addition pathway was calculated using DFT. The barrier is 2.1 kcal/mol lower in energy than that of the benzyl benzoate consistent with our experimental results (Figure 12). However, in both cases the barrier for oxidative addition is very low (4.3 and 6.4 kcal/mol), suggesting that oxidative addition is facile. The S_N2 product of the oxidative addition of benzyl 2,3,4,5,6-pentafluorobenzoate to (IPr)Pd(0), P_FBz(O–C_Bz), is also an ion pair (i.e. (IPr)Pd(η³-FBz)⁺PhCO₂⁻, with q = ± 0.82e), which is significantly more stable than that formed by the non-fluorinated substrate (−6.1 vs 1.6 kcal/mol), likely due to the fluorides promoting charge separation.⁴⁴ We propose that the outersphere 2,3,4,5,6-pentafluorobenzoate anion in P_FBz(O–C_Bz) may lead to faster transmetallation, the likely next elementary step in the mechanism after oxidative addition (Figure 13), in part because the formation of P_FBz(O–C_Bz) is not as thermodynamically favorable as the formation of P_Bz(O–C_Bz) (see SI). Thus, utilizing the 2,3,4,5,6-pentafluorobenzoate as the leaving group may result in improvements to multiple elementary steps on the catalytic cycle. By analogy to other cross-coupling reactions,⁴⁵ we expect that the final step reductive elimination is facile, suggesting that transmetallation is the turnover-limiting step.

Figure 12.

DFT calculations comparing the oxidative addition of the O–C(benzyl) bond in benzyl benzoate and benzyl 2,3,4,5,6-pentafluorobenzoate to (IPr)Pd(0). Relative energies are given in kcal/mol.

Figure 13.

Proposed catalytic cycle for Pd-catalyzed Suzuki-Miyaura reactions of diphenylmethyl 2,3,4,5,6-pentafluorobenzoate.

To further probe the nature of the turnover-limiting step in catalysis we investigated the oxidative addition of (diphenyl)methyl-2,3,4,5,6-pentafluorobenzoate (14a) and (2-methylphenyl)(phenyl)methyl-2,3,4,5,6-pentafluorobenzoate (14b) to (IPr)Pd(styrene)₂. Our substrate scope showed that the use of diarylmethylesters with ortho-substitution, such as 14b, on the aryl groups results in reduced yields under optimized conditions and necessitates the use of harsher reaction conditions to obtain satisfactory yields (Figure 6). However, the rates of oxidative addition of 14a and 14b are comparable (Figure 14), with both occurring at room temperature. This is not consistent with oxidative addition being the reason that harsher conditions are required for the coupling of 14b and suggests that a subsequent step in the catalytic cycle is turnover-limiting. It also provides further evidence that oxidative addition is facile.

Figure 14.

NMR yields of oxidative addition product over time for the reaction of diarylmethyl pentafluorobenzoate with (IPr)Pd(styrene)₂.

Conclusions

We have developed a broad method for synthesizing triarylmethanes under mild conditions via Pd-catalyzed Suzuki-Miyaura coupling reactions involving 2,3,4,5,6-pentafluorobenzoate electrophiles. Importantly, our method is not limited to electrophiles containing extended aromatic systems, such as naphthyl or biaryl groups, and as a result represents a significant advance over previous methods. Further, the reaction is stereospecific and is able to generate chiral triarylmethanes with inversion of configuration. Intriguingly, while the Suzuki-Miyaura reaction of diarylmethyl esters involves the cleavage of the O–C(benzyl) bond, the reaction featuring closely related phenyl ester electrophiles involve selective cleavage of the C(acyl)–O bond. DFT calculations show that cleavage of the O–C(benzyl) bond in benzyl electrophiles is both kinetically and thermodynamically preferred. This is because oxidative addition of benzyl electrophiles to (IPr)Pd(0) via an S_N2 mechanism provides a low barrier pathway for cleavage of the O–C(benzyl) bond, while the formation of products with an η³-benzyl interaction thermodynamically stabilizes the Pd(II) products. In fact, in our reactions the oxidative addition of the 2,3,4,5,6-pentafluorobenzoate electrophiles is so facile that transmellation is likely the turnover-limiting step in catalysis. Phenyl ester electrophiles cannot readily undergo oxidative addition via an S_N2 pathway or form products which are stabilized via chelation and as a result cleavage of the C(acyl)–O bond is kinetically and thermodynamically preferred. Overall, apart from the development of a more general method for the synthesis of triarylmethanes, our work provides fundamental information on the selectivity of oxidative addition of ester electrophiles. Given the high level of current interest in using esters as electrophiles in cross-coupling, this is likely to be valuable for the design of new and improved synthetic methods.