Aerobic Oxidative Coupling of o-Xylene: Discovery of 2-Fluoropyridine as a Ligand to Support Selective Pd-Catalyzed C–H Functionalization

Abstract

An improved method for direct oxidative coupling of o-xylene could provide streamlined access to an important monomer used in polyimide resins. The use of 2-fluoropyridine as a ligand has been found to enable unprecedented levels of chemo- and regioselectivity in this Pd-catalyzed aerobic oxidative coupling reaction. Preliminary insights have been obtained into the origin of the effectiveness of 2-fluoropyridine as a ligand.

The synthesis of biaryls via direct oxidative coupling of arenes [Eq. (1)] is a prominent contemporary challenge in organic chemistry, and significant recent efforts have been directed toward homocoupling[1] and cross-coupling[2]

$$\underset{\text{Ar}-\text{H}+\text{Ar}\text{'}-\text{H}}{\overset{2\text{Ar}-\text{H}}{or}}+[\text{O}]\stackrel{\mathit{cat}.}{\to }\underset{\text{Ar}-\text{A}{\text{r}}^{\prime }}{\overset{\text{Ar}-\text{Ar}}{or}}+[\text{O}]{\text{H}}_2$$ (1)

methods. Reactions of this type first emerged in the mid-1960s, when the oxidative homocoupling of simple arenes was reported by van Helden and Verberg using stoichiometric Pd^II salts, [3] and later by Davidson and Triggs using catalytic Pd^II.[4] Homocoupling reactions of simple substituted arenes (e.g., toluene, anisole, o-xylene) typically suffer from low regioselectivity, and therefore such reactions have limited utility in organic synthesis. An important exception is the oxidative coupling of dimethyl-o-phthalate. This Pd-catalyzed coupling reaction is an important step in the commercial synthesis of 4,4′-biphthalic anhydride (2), [5,6] a monomer used in the high-performance polyimide resin, Upilex^® (Scheme 1, Route A), [7] and is the basis for recent industrial interest in the development of improved arene homocoupling methods.[8] Production of the bisanhydride monomer 2 could be streamlined significantly by carrying out direct homocoupling of o-xylene, followed by aerobic oxidation of the benzylic methyl groups of 4,4′-bixylyl 3 (Scheme 1, Route B).[9] A major barrier to implementing this improved synthetic route is the poor selectivity obtained in the homocoupling of o-xylene. Selectivity challenges include regioselectivity, which result in a mixture of bixylyl isomers, and chemoselectivity, associated with overoxidation of bixylyl to afford oligomeric xylyl byproducts. Here, we describe the development of a new chemo- and regioselective method for aerobic oxidative homocoupling of o-xylene [Eq. (2)] and the discovery of 2-fluoropyridine (^2Fpy) as a highly effective ligand in this Pd-catalyzed reaction.

Scheme 1.

Existing and proposed alternate routes for the commercial production of the polyimide monomer 2.

(2)

The oxidative coupling of dimethyl-o-phthalate and o-xylene has been the subject of numerous investigations, and the best reported results are presented in Eqs. (3) and (4), respectively.[10] Large-scale reactions of this type are typically stopped at low conversion (≤ 10%) to avoid overoxidation and formation of oligomeric byproducts. The starting materials are recycled to ensure efficient feedstock utilization. In such reactions, product selectivity often becomes a more important measure of the success of the reaction than the single-pass yield. The oxidative coupling of dimethyl-o-phthalate proceeds with sufficient chemo- and regioselectivity [88% and 91%, respectively; Eq. (3)] to be used in the commercial production of 4,4′-biphthalic anhydride 2. In contrast, the oxidative coupling of o-xylene exhibits poor selectivity [40% and 77%, respectively; Eq. (4)].

(3)

(4)

We speculated that the o-xylene coupling reaction could proceed with higher selectivity improved by carrying out the reaction under milder reaction conditions, and that this goal could be achieved by identifying a more active catalyst. With these targets in mind, catalyst screening efforts focused on evaluating reactions at 80 °C, 60 degrees lower than that of the previously optimized conditions (140 °C). This work took place in two phases. Initial studies focused on evaluating combinations of Pd salts (2.5 mol %) with ligands, acid and base additives, and solvents. These screening studies and a brief mechanistic assessment of the optimized conditions were then followed by reoptimization of the catalyst system at 0.1 mol % Pd loading. These efforts are presented in sequence below.

Our ligand screening efforts focused on the evaluation of oxidatively stable nitrogen ligands, which have found significant utility in other Pd-catalyzed aerobic oxidation reactions (see Supporting Information for details).[11] Chelating ligands such as bipyridine and phenanthroline, which have been used successfully in the oxidative coupling of dimethyl-o-phthalate [cf. Eq. (3)], [6d] almost completely inhibited the oxidative coupling of o-xylene at 80 °C (< 1% yield of bixylyl). In contrast, promising results were obtained with monodentate pyridine derivatives (Figure 1). Although little coupling product was obtained with pyridine itself, significant catalytic activity was observed with pyridine derivatives bearing electron-withdrawing groups in the 2-pyridyl position. The best results, by far, were obtained with 2-fluoropyridine as the ancillary ligand (Figure 1). The identity of the anionic ligand also had a significant influence on the reaction, with Pd(OAc)₂ exhibiting a lower reactivity than Pd(TFA)₂ (TFA = trifluoroacetate). The most effective catalyst systems, however, featured a mixture of acetate and trifluoroacetate anionic ligands (Figure 2). Similar results were obtained by combining Pd(OAc)₂ and Pd(TFA)₂ (Figure 2A) or by adding CF₃CO₂H to Pd(OAc)₂ (Figure 2B).

Figure 1.

Ancillary ligand effects on the yield of Pd-catalyzed homocoupling of o-xylene. Conditions: o-xylene (3.8 mmol), Pd(OAc)₂/CF₃COOH, (0.09/0.12 mmol), ligand (0.19 mmol), propylene carbonate (0.4 mL), 1 atm O₂, 80 °C, 17 h, GC yield.

Figure 2.

Anionic ligand effects on the oxidative coupling of o-xylene, based on mixtures of Pd(OAc)₂/Pd(TFA)₂ (A) and Pd(OAc)₂/CF₃CO₂H (B). Conditions: o-xylene (3.8 mmol), Pd(OAc)₂ + Pd(TFA)₂ (0.09 mmol), 2-fluoropyridine (0.19 mmol), CF₃COOH (0–0.12 mmol), propylene carbonate (0.4 mL), 1 atm O₂, 80 °C, 17 h, GC yield.

NMR spectroscopic studies provide some insight into the ligand effects (see Figure 3 and Supporting Information). The ortho-fluorine substituent of ^2Fpy greatly attenuates its basicity/coordinating ability relative to pyridine, [12] and ¹H and ¹⁹F NMR spectra of Pd(OAc)₂/^2Fpy mixtures reveal that ^2Fpy does not bind to Pd(OAc)₂.[13] The ^2Fpy resonances in this spectrum are only those associated with free ligand (cf. Figures 3A, 3B and S1–S2). In contrast, ^2Fpy coordinates to Pd(TFA)₂ with a 2:1 stoichiometry, and the NMR spectroscopic and X-ray crysallographic data establish the identity of this complex as trans-(^2Fpy)₂Pd(TFA)₂ (Figures 3C and S3-S4).[14] Titration experiments establish that the 2:1^2Fpy:Pd(TFA)₂ stoichiometry is retained even when fewer than two equivalents of ^2Fpy are present in solution. The combination of ^2Fpy with Pd(OAc)₂/Pd(TFA)₂ and with Pd(OAc)₂/CF₃CO₂H are more complex. ¹H and ¹⁹F NMR spectra reveal the presence of (^2Fpy)₂Pd(TFA)₂ in addition to two other species (Figures 3D, 3E and S5-S8). The identities of the latter species have not been fully established, but the data are consistent with ^2Fpy coordination to Pd species with both AcO⁻ and CF₃CO₂⁻ ligands, either as mononuclear or higher-nuclearity Pd complexes.

Figure 3.

¹H NMR spectra demonstrating the ability of ^2Fpy to coordinate to Pd^II–trifluoroacetate (TFA) species (spectra C–E), but not Pd(OAc)₂ (spectrum B) in C₆D₆. Conditions: (A) ^2Fpy in C₆D₆; (B) [Pd(OAc)₂] = 14.6 mM, [^2Fpy] = 30.7 mM; (C) [Pd(TFA)₂] = 24.3 mM; [^2Fpy] = 51.9 mM; (D) [Pd(OAc)₂] = 20.8 mM, [CF₃COOH] = 18.7 mM, [^2Fpy] = 41.6mM; (E) [Pd(OAc)₂] = 6.9 mM, [Pd(TFA)₂] = 7.3 mM, [^2Fpy] = 31.2 mM.

A large deuterium kinetic isotope effect (k_H/k_D = 10.7(2.0)) was obtained for the catalytic oxidative coupling reaction by independently measuring the initial rate of bixylyl formation with o-xylene and with o-xylene-d₁₀.[15] This result suggests that C–H activation of o-xylene is the turnover-limiting step of the reaction, and it provides a basis for understanding the empirical ligand screening data in Figures 1 and 2. These data highlight the beneficial effect of electron-deficient ancillary ligands, ^2Fpy and TFA, but they also reveal that acetate (~1 equiv with respect to Pd) benefits the reaction. Such observations can be rationalized within the framework of a “concerted metalation-deprotonation” (CMD) pathway for Pd^II-mediated C–H activation.[16] In such mechanisms, C–H activation is facilitated by two factors: (1) increased electrophilicity of the Pd^II center and (2) the presence of a basic ligand (such as acetate) that deprotonates the arene C–H bond as the Pd–C_aryl bond is forming.

The catalyst system identified from these initial studies [Pd(OAc)₂/CF₃CO₂H/^2Fpy (1:1:2)] has a relatively high catalyst loading (2.5%) and low turnover number (12); however, it exhibits improved product selectivity (60%) and regioselectivity (3:3′ = 88:12) relative to the previous-best results [Eq. (4)]. In an effort to develop a more practical catalyst system, the oxidative coupling of o-xylene was reevaluated at 0.1 mol % Pd loading. The initial result was quite poor (Table 1, entry 1). Further optimization, however, revealed that improved results could be obtained by including 0.1 mol % Cu(OTf)₂ as a cocatalyst and by replacing propylene carbonate with acetic acid as a solvent (Table 1; see Supp. Info. for additional data).

Table 1.

Optimization data for o-xylene homocoupling at 0.1 mol % Pd loading.^[a]

Entry	Additive (0.1 %)	Solvent	TON [b]	Product Selectivity [%]	Regioselectivity [%][c]
1	–	p.c.[d]	7	11	68
2	Cu(OAc)2	p.c.	8	41	65
3	AgOAc	p.c.	4	65	38
4	Cu(OTf)2	p.c.	45	63	67
5	AgOTf	p.c.	5	5	69
6	Zn(OTf)2	p.c.	16	29	65
7	Cu(OTf)2	AcOH	80	94	88
8	Cu(OTf)2	EtCO2H	74	56	88
9	Cu(OTf)2	PivOH	16	19	67

^[a]Conditions: o-xylene (3.8 mmol), Pd(OAc)₂ (0.004 mmol), additive (0.004 mmol), 2-fluoropyridine (0.008 mmol), CF₃COOH (0.005 mmol), solvent (0.4 mL), 1 atm O₂, 80 °C, 17 h.

^[b]Determined by GC yield.

^[c][3/(3 + 3′)] × 100.

^[d]Propylene carbonate.

Analysis of ancillary ligand effects with this improved catalyst system revealed that 2-fluoropyridine remains the most effective ligand. A Brønsted plot, based on initial rates of reactions with numerous substituted pyridines, exhibits a “volcano plot” appearance with 2-fluoropyridine at the peak (Figure 4). The presence of paramagnetic Cu species and the low catalyst loading has complicated efforts to probe this catalyst system directly by NMR spectroscopy; however, our previous data (cf. Figures 1–3) provide some clues into the origin of the ^2Fpy ligand effect. We hypothesize that pyridines more basic than ^2Fpy attenuate the electrophilicity of the Pd catalyst and reduce the catalytic activity. In contrast, pyridines less basic than ^2Fpy are poor ligands and, therefore, are unable to modulate the identity and reactivity of the Pd catalyst. The latter hypothesis is supported by ¹H and ¹⁹F NMR spectroscopic studies, which reveal that 2,4,6-trifluoropyridine and other more-electron-deficient pyridine derivatives do not coordinate to Pd(TFA)₂ (see Supporting Information).

Figure 4.

Dependence of the rate of o-xylene coupling on the identity of the ancillary pyridine ligands. The pKa values correspond to those of the pyridinium ions in DMSO. [17] Conditions for initial-rate measurements: o-xylene (3.8 mmol), Pd(OAc)₂ (0.02 mmol), Cu(OTf)₂ (0.02 mmol), ligand (0.04 mmol), CF₃COOH (0.024 mmol), AcOH (0.4 mL), 1 atm O₂, 80 °C, initial rate (0–6 h).

Collectively, these results highlight 2-fluoropyridine as a novel, highly effective ligand to support Pd-catalyzed aerobic oxidative coupling of arenes. The high activity of the ^2Fpy/Pd^II catalyst system permits the oxidative coupling of o-xylene to be achieved at temperatures substantially lower than those previously reported, and the improved catalytic results, which include a dramatic increase in the product selectivity (from 40% to 94%) as well as an increase in the regioselectivity (from 77% to 88%) [Eqs. (4) and (5)], provide the potential basis for more-efficient

(5)

feedstock utilization in a large-scale chemical process (cf. Scheme 1). These results have important implications for other reactions as well. The scope and utility of Pd-catalyzed oxidative coupling reactions continues to expand, [1,2] and the identification of oxidatively stable ligands that can support efficient catalytic turnover with O₂ as the terminal oxidant is an increasingly important goal for this field. Preliminary results in our lab suggest that 2-fluoropyridine will find broad application to these types of reactions.

Experimental Section

Representative Procedure for Aerobic Oxidative Coupling of o-Xylene

Method A

In a disposable culture tube, palladium complexes (0.094 mmol), 2-fluoropyridine (0.19 mmol), trifluoroacetic acid (0.113 mmol), o-xylene (0.4 g) and propylene carbonate (0.4 g) were combined. The reaction tubes were placed in a 48-well aluminum block mounted on a Large Capacity Mixer (Glas-Col) that enabled several reactions to be performed simultaneously under a constant pressure of (approx 1 atm) with controlled temperature and orbital agitation. The headspace above the tubes was purged with oxygen gas for ca. 5 min. The reactions were vortexed for 17 hr under 1 atm of O₂. After the reactions were stopped, n-hexadecane was added to the reaction mixture as an internal standard. Samples were evaluated by GC for the products and remaining starting materials.

Method B

In a 6 ml vial, palladium complexes (0.02 mmol), 2-fluoropyridine (0.04 mmol), trifluoroacetic acid (0.024 mmol) and acetic acid (2 g) were combined and stirred at room temperature for 30 min. The mixture was used as a stock solution. In a disposable culture tube, copper triflate (0.04 mmol) and o-xylene (0.4 g) were combined. Then, the stock solution (0.4 g) was added. Reaction tubes were placed in a 48-well parallel reaction mounted on a Large Capacity Mixer (Glas-Col) that enabled several reactions to be performed simultaneously under a constant pressure of (approx 1 atm) with controlled temperature and orbital agitation. The headspace above the tubes was purged with oxygen gas for ca. 5 min. The reactions were vortexed for 17 hr under 1 atm of O₂. After the reactions were stopped, n-hexadecane was added to the reaction mixture as an internal standard. Samples were evaluated by GC for the products and remaining starting materials.