Tailored quinones support high-turnover Pd catalysts for oxidative C–H arylation with O₂

Abstract

Palladium(II)-catalyzed C–H oxidation reactions could streamline the synthesis of pharmaceuticals, agrochemicals, and other complex organic molecules. Existing methods, however, commonly exhibit poor catalyst performance with high Pd loading (e.g., 10 mol%) and a need for (super)stoichiometric quantities of undesirable oxidants, such as benzoquinone and silver(I) salts. The present study probes the mechanism of a representative Pd-catalyzed oxidative C–H arylation reaction and elucidates mechanistic features that undermine catalyst performance, including substrate-consuming side reactions and sequestration of the catalyst as inactive species. Systematic tuning of the quinone co-catalyst overcomes these deleterious features. Use of 2,5-di-tert-butyl-p-benzoquinone enables efficient use of molecular oxygen as the oxidant, high reaction yields, and >1900 turnovers by the palladium catalyst.

One Sentence Summary:

Mechanistic studies lead to a high-performance palladium/quinone co-catalyst system for oxidative C–H arylation using O₂ as the oxidant.

Homogeneous palladium-catalyzed oxidation reactions of organic molecules originated in 1959 with the discovery of the Wacker process for oxidative coupling of ethylene and water (1). Pd-catalyzed C–H oxidation reactions for oxidative homocoupling of arenes to biaryls were reported shortly thereafter (2), including examples compatible with O₂ as the terminal oxidant (3). The poor catalytic efficiency and low regioselectivity of these methods contributed to development of Pd-catalyzed cross-coupling reactions with pre-oxidized substrates, such as aryl halides, as coupling partners. While Pd-catalyzed cross-coupling reactions have achieved extraordinary success (4), methods for direct oxidative functionalization of C–H bonds could substantially streamline synthetic access to diverse chemical structures (5–8).

Most Pd^II/0-catalyzed coupling reactions, including both oxidative and non-oxidative examples, fit into two general classes: (i) olefinations, such as Heck, Fujiwara-Moritani, and related reactions that involve coupling with alkenes and generate products via β-hydride elimination from a Pd^II-alkyl intermediate, and (ii) arylations, such as Suzuki-Miyaura, Negishi, and related reactions that involve coupling with arenes or aryl nucleophiles and generate the products via reductive elimination from a Pd^II–aryl intermediate. The oxidative reactions (cf. Fig. 1A) are mechanistically similar to non-oxidative cross-coupling reactions, but they often exhibit worse catalytic performance and require the use of one or more stoichiometric oxidants, such as Cu^II, Ag^I, and benzoquinone. The representative collection of oxidative arylation reactions in Fig. 1B, emphasizing cases in which the C–H substrate is the limiting reagent, highlights typical catalytic turnover numbers and stoichiometric oxidant(s) (9–16). The initial indole arylation example (9) achieves comparatively high turnovers (32 TOs); however, it requires 3 equiv Cu(OAc)₂ as the oxidant and 30 equiv of benzene as a coupling partner. Many other examples feature < 10 TOs (see Fig. S1A in the Supplementary Materials for additional examples). Oxidative olefination reactions with limiting C–H substrate often exhibit similarly poor catalytic performance (see Fig. S1B in the Supplementary Materials for examples). We have noted, however, that Pd-catalyzed oxidation reactions that proceed via β-hydride elimination tend to be more compatible with O₂ as an oxidant than reductive elimination reactions (17). This phenomenon is clearly demonstrated in a pair of reactions reported by Yu and coworkers involving Pd^II/0-catalyzed C–H olefination (18–20) and arylation (21). Both were demonstrated with 1 atm O₂ as the oxidant, but the olefination reaction accessed 455 TOs (Fig. 1C, left), whereas the arylation reaction (Fig. 1C, right) achieved only 13 TOs under comparable conditions. Understanding the factors that limit catalyst turnovers in oxidative arylation reactions of this type could play a major role in expanding their utility, and accessing improved catalytic performance could set the stage for large-scale applications that would greatly benefit from low catalyst loading and use of O₂ as an oxidant (22). Here, we present a mechanistic study of the Pd-catalyzed C–H arylation reaction in Fig. 1C and use the insights to develop a modified catalyst system capable of accessing >1900 TOs and high yields with O₂ as the oxidant.

Fig. 1.

Pd-catalyzed aerobic C–H oxidation reactions, highlighting challenges in catalyst performance. (A) Mechanisms for Pd^II-catalyzed oxidative olefination and arylation reactions using O₂ as the oxidant. (B) Catalytic metrics for representative oxidative arylation reactions (9–16). (C) Comparison of catalyst turnovers in Pd-catalyzed olefination and arylation reactions (19, 21). MPAA = Mono-N-protected amino acid.

The reported arylation reaction in Fig. 1C was adapted to facilitate mechanistic study. Use of the potassium salt of the arene substrate [2-(trifluoromethyl)phenylacetic acid (1)] eliminated the requirement for heterogeneous KHCO₃ as a Brønsted base, and arylboronic acid pinacol ester (2) provided a soluble alternative to the ArBF₃K coupling partner. An 83% NMR assay yield of the arylation product 3 was obtained with these modifications when the reaction was conducted with 1 atm O₂ (Fig. 2A), representing an improvement over the originally reported yield of 65%. The initial experiments confirmed that 1,4-benzoquinone (BQ) is a crucial co-catalyst for the reaction. In the absence of BQ, negligible arylation product was formed, with the majority of the arylboronate undergoing protodeboronation to afford fluorobenzene (Ar^F–H) (Fig. 2A). Analysis of the reaction time course revealed a kinetic burst at the start of the reaction, with a magnitude corresponding to one turnover of the Pd catalyst, followed by a much slower steady-state rate (Fig. 2B). Increasing the oxygen pressure had no effect on the burst, but it led to an increase in rate following the burst roughly proportional to the change in O₂ pressure (0.46 × 10⁻³ mM/s at 1 atm; 2.5 × 10⁻³ at 6.4 atm).

Fig. 2.

Kinetic and spectroscopic analysis of the Pd-catalyzed C–H arylation reaction. (A) Oxidative arylation conditions (R = 2-(trifluoromethyl)C₆H₄CH₂-). (B) Kinetic data showing a burst and the influence of O₂ pressure on the steady state arylation rate. (C) Simplified mechanism rationalizing the kinetic burst and BQ inhibition of steady state turnover. (D) Enhanced steady state rate observed with 2,5-^tBu₂BQ. (E) Reaction progress obtained from ¹⁹F NMR spectroscopic analysis of the reaction mixture to which different reaction components were added sequentially. Interpolative curve fits are included as a guide. (F) Kinetic data [2,5-^tBu₂BQ] for formation of 3 and Ar^F–H, with correlations reflecting a hyperbolic curve fit [y = ax/(1 + bx)]. (G) Spectroscopic observation of palladacycle resting state under 6.9 atm O₂, with linear and interpolative curve fits as a guide. Conditions: (A) [1] = 200 mM, 24 h (B, D, F) [1] = 100 mM, [Pd]_t = 5 mM, [Boc-Val-OH] = 10 mM, [quinone] = 20 mM, 85 °C, 1 mL, 1 atm O₂. (E, G) [1] = 100 mM, [Pd]_t = 10 mM, [Boc-Val-OH] = 20 mM, [2,5-^tBu₂BQ] = 40 mM, 64 °C, 0.55 mL.

The two-stage catalytic mechanism in Fig. 2C provided a framework for preliminary interpretation of these results. The kinetic burst could arise from fast stoichiometric oxidative coupling of the substrates by Pd^II/BQ, while the post-burst phase could arise from turnover-limiting aerobic oxidation of the reduced Pd catalyst. BQ is commonly used to oxidize Pd⁰ in Pd-catalyzed oxidation reactions (23); however, Brønsted acid is needed to promote this reaction (24). We speculated that the basic conditions associated with the present arylation reaction prevent BQ from serving as an effective oxidant, and it instead coordinates to Pd⁰ and inhibits its reoxidation by O₂. This mechanistic hypothesis prompted us to test sterically modified benzoquinone derivatives that might undergo more facile displacement by O₂ (Fig. 2D) (25), thereby facilitating catalytic turnover. Use of tert-butyl-p-benzoquinone (^tBuBQ) had virtually no impact on the reaction time course; the reaction still exhibited a kinetic burst, followed by slow steady-state turnover. In contrast, use of 2,5-di-tert-butyl-p-benzoquinone (2,5-^tBu₂BQ) led to a sustained high rate, without a slowdown after the initial catalyst turnover (Fig. 2D, blue).

The reaction was then interrogated by ¹⁹F NMR spectroscopy in order to gain insights into the role of the different catalyst components. The C–H substrate 1 and Pd^II(OAc)₂ were dissolved in ^tAmylOH and added to an NMR tube under ambient air. No reaction was observed upon warming this mixture to 64 °C, consistent with previous results showing that C–H activation is very slow in the absence of a mono-N-protected amino acid (MPAA) ligand (8,19,26,27). Addition of the MPAA ligand initiated immediate formation of the previously characterized palladacycle derived from activation of the substrate C–H bond (Fig. 2E, black, see Supplementary Materials Section 5 for details) (27). Subsequent addition of the arylboronic ester coupling partner led to formation of protodeboronation side product, Ar^F–H (Fig. 2E, blue). No C–H arylation product 3 was observed at this stage, but arose only after addition of 2,5-^tBu₂BQ (Fig. 2E, red). These observations align with the catalytic data in Fig. 2A showing that Ar^F–H is the product obtained when the reaction is conducted in the absence of BQ. Systematic variation of the 2,5-^tBu₂BQ concentration provided evidence for direct competition between the formation of Ar^F–H and 3, with the latter promoted by higher quinone concentrations (Fig. 2F).

These experiments were further complemented by spectroscopic and kinetic analysis of the fully constituted catalytic reaction mixture. Use of a sealed NMR tube with higher initial O₂ pressure (6.9 atm) minimized complications arising from depletion of dissolved O₂ in the NMR solution (e.g., which accounts for decay of the palladacycle after 2.5 h in Fig. 2E). Under these conditions, the palladacycle forms rapidly and accounts for all of the Pd in solution (Fig. 2G, see Supplementary Materials Section 6 for details). This species persists as the catalyst resting state, while the arylation product 3 steadily appears in parallel with smaller quantities of the Ar^F–H byproduct. Initial-rate kinetic studies revealed a negligible deuterium kinetic isotope effect (k_H/k_D = 1.2 ± 0.1), based on independent rates measured for arylation of 1 and 1-d₁, the latter deuterated at the site targeted by C–H activation, ortho to the directing group. Under conditions of saturating [2,5-^tBu₂BQ], the catalytic rate law exhibits a first-order dependence on [Pd], a zero-order dependence on O₂ pressure, and saturation dependences on the coupling partners [1] and [2] (Figs. S2, S8-10 in the Supplementary Materials).

The experimental data presented above indicate that the palladacycle is the catalyst resting state and transmetalation of Ar^F from the arylboronate to palladium is the turnover-limiting step when 2,5-^tBu₂BQ is used as the quinone co-catalyst. The dependence of the rate on substrate 1 is rationalized by the ability of basic additives to promote the transmetalation step (28). The proposed mechanism in Fig. 3A incorporates these features, together with other steps in the catalytic reaction and steps leading to byproduct formation. The reaction is initiated by MPAA-promoted C–H activation of the substrate in I to form palladacycle II (26), which is observed by ¹⁹F NMR spectroscopy. A recent study indicates that the MPAA ligand plays a catalytic role in the C–H activation step and does not influence other steps in the mechanism (27). Moreover, MPAA undergoes facile exchange with substrate-derived carboxylates and likely does not remain coordinated to Pd throughout the cycle. Transmetalation from the arylboronate to the palladacycle generates the diaryl-Pd^II species III, which can undergo protodemetalation to form byproduct Ar^F–H and regenerate the palladacycle II or undergo quinone-promoted reductive elimination to generate the arylation product 3 and the Pd⁰-quinone adduct IV. The Pd^II catalyst is regenerated via reaction of IV with O₂.

Fig. 3.

Catalytic cycle consistent with the experimental data and computational analysis. (A) The proposed catalytic mechanism for C–H arylation of 1 (R = 2-(trifluoromethyl)C₆H₄CH₂-), and (B) calculated free energy diagram for protodemetalation and reductive elimination pathways in the absence and presence of quinones. See text and Supplementary Materials for details of the computational methods. TOF = Turnover Frequency, TON = Turnover Numbers

The proposed mechanism in Fig. 3A highlights the delicate balance between the beneficial and deleterious effects of the quinone co-catalyst. It is needed to promote reductive elimination (29–31) and avoid catalytic degradation of the arylboronate; however, it can also inhibit catalytic turnover by coordinating too strongly to Pd⁰. Further insights into these roles of quinone were obtained from density functional theory calculations (Fig. 3b). The relative energies presented in Fig. 3B were calculated at the M06-(IEF-PCM)/BS2 [BS2 = 6–311+G(d,p) (for all atoms except Pd) and SDD (for Pd)] levels of theory by using geometries and enthalpy and entropy corrections calculated at the B3LYP-D3BJ/BS1 level of theory [BS1 = 6–31G(d,p) (for all atoms except Pd) and LanL2dz (for Pd)] (see Section 8 in the Supplementary Materials for details).

Specifically, the Pd^II-diaryl species, Pd(Ar)(Ar’) (cf. III in Fig. 3A) was interrogated to compare the energetics of the protodemetalation step with C–C reductive elimination in the absence and presence of quinone (BQ and 2,5-^tBu₂BQ) (Fig. 3b). Neither BQ nor 2,5-^tBu₂BQ form favorable adducts with the Pd(Ar)(Ar’) species (ΔG° = +2.9 and =3.9 kcal/mol, respectively); however, both lower the kinetic barrier for reductive elimination relative to the quinone-free barrier (ΔΔG^‡ = –4.4 and –3.2 kcal/mol for BQ and 2,5-^tBu₂BQ). The transition-state energies for quinone-promoted reductive elimination is very similar to the transition-state energy for the protodemetalation pathway, consistent with the competitive formation of Ar^F–H and product 3 (cf. Fig. 2f).

The mechanistic insights provided above set the stage for efforts to optimize the catalytic performance. Key screening data are summarized in Fig. 4A, with full details provided in Section 9 of the Supplementary Materials. Use of 2,5-^tBu₂BQ with 5% mol Pd(OAc)₂ and 3 atm O₂ led to >99% yield of 3, corresponding 20 TOs (Fig. 4a, entry 1). Reducing the catalyst loading to 0.15 mol% Pd with the same ratio of Pd:MPAA:2,5-^tBu₂BQ (1:2:4) resulted in 147 TOs. The reaction instead generated large quantities of Ar^F– H and 4,4’-difluorobiphenyl (Ar^F–Ar^F) (entry 2). The latter byproduct arises from oxidative homocoupling of arylboronate (32) and is rationalized by inefficient C–H activation of 1 at low Pd/MPAA loading, due to relatively weak binding of MPAA to Pd^II (27). This complication was addressed by increasing the MPAA loading to 10 mol%, which enhances the rate of C–H activation without increasing the Pd loading. These conditions supported effective C–H activation and increased the Pd TOs to 373 (Fig. 4a, entry 3). The quinone loading and identity of the MPAA ligand were then varied in order to access higher reaction yields. The most dramatic effect was observed upon replacing Boc-Val-OH with Ac-Ile-OH as the MPAA ligand, which led to 667 TOs and a quantitative yield of 3 (entry 5).

Fig. 4.

Catalytic performance with low catalyst loading. (A) Catalyst optimization data showing product ratios and catalyst turnover numbers (TON) for the oxidative coupling of 1 and 2. (B) Observed turnovers of 3 with different quinones/alkenes. (C) Application of mechanistic insights to other substrates and coupling partners. Conditions: (A,B) see conditions shown in above and below the arrow in A and B, 0.5 mL scale, Data based on NMR assay yield. (C) Identical to the conditions used in B except 0.5 mol% Pd was used with 30 mol% 2,5-^tBu₂BQ. [**] Reported catalyst turnovers with 5 mol % Pd(OAc)₂ and 20 mol % BQ (21). [^] A 93% isolated yield was obtained. [†] with 0.05 mol% Pd(O₂CR)₂ and 72 hours reaction time. [‡] A 96% isolated yield was obtained. [∫] Yields shown are isolated product yields. [#] 30 mol% 2,6-^tBu₂BQ was used. [§] 5 equivalents of methylboronic acid was used. [¶] K₂CO₃ was used as the base and no Ac-Ile-OH added.

With the identification of conditions compatible with low Pd loading, we revisited the influence of the quinone identity. The results reinforced previous observations (Fig. 4b). Little product was observed with the parent BQ and with ^tBuBQ (A and B), while higher catalytic turnover was evident with 2,5-Me₂BQ (C, 230 TOs). Further increase in the number or size of the substituents conferred additional benefits, as evident with Me₄BQ (F, 574 TOs) and 2,6- ^tBu₂BQ, and 2,5- ^tBu₂BQ (G and H, 667 TOs). The latter two examples led to quantitative product yield at 0.15 mol% Pd loading, prompting reassessment with 0.05 mol% Pd. The beneficial effect of these sterically encumbered quinones is evident from the 1520 turnovers with the 2,6-^tBu₂BQ and 1960 turnovers with the 2,5-^tBu₂BQ, the latter corresponding to a near-quantitative product yield.

This survey also provided an opportunity to test an unexpected conclusion of the mechanistic studies above. The proposed catalytic mechanism in Fig. 3A features a non-redox role for the quinone, suggesting that redox-inactive electron-deficient alkenes should also support catalytic turnover. This hypothesis was validated by the use of dimethyl fumarate (D) and diisopropyl fumarate (E), which supported 390 and 440 turnovers respectively. Although this performance is not better than that observed with the sterically encumbered quinones (F–H), it is much better than BQ, the most commonly used co-catalyst in Pd-catalyzed aerobic oxidation reactions.

The optimized catalyst conditions were tested with a representative collection of other substrate partners with 0.5 mol% Pd (Fig. 4C). Several arylboronic esters, including methyl benzoate and protected aniline derivatives, and methylboronic acid react with the parent C–H substrate in good-to-excellent isolated yields (4–7). The 1-naphthylacetate and benzoate C–H substrates (8–10) also proceed in good yield. Substrates with neutral nitrogen direction groups, including the rather heterocyclic sildenafil precursor, were also effective (11, 12), albeit with more modest yields.

Overall, the results above highlight the delicate balancing act of quinone co-catalysts in these reactions. The quinone is needed to promote the kinetically difficult reductive elimination step and prevent deleterious consumption of the arylboronate via protodemetalation, but it also must avoid catalytic inhibition, which can arise if it coordinates too strongly to Pd⁰. Tuning of the quinone structure allows an effective balance to be found among these roles. These considerations provide relevant context for comparison with the closely related olefination reactions (Fig. 1A and 1C). The β-hydride elimination and H–X reductive elimination steps in olefination reactions are kinetically more facile than the C–C reductive elimination step in the arylation reactions, and olefination reactions lack the deleterious protodemetalation side-reaction that hinders the arylation reaction. Thus, olefinations often do not employ a quinone co-catalyst and tend to be more amenable to aerobic catalytic turnover. The insights from this study show how strategic integration of a redox inactive quinone or related co-catalyst in Pd-catalyzed arylation reactions could provide the basis for highly effective catalytic performance, enabling low Pd loading and effective use of O₂ as the oxidant. The promising extension of these results to other substrates (Fig. 4C) set the stage for further application of the concepts described herein.