Allylic C–H Acetoxylation with a 4,5-Diazafluorenone-Ligated Palladium Catalyst: A Ligand-Based Strategy to Achieve Aerobic Catalytic Turnover

Abstract

Pd-catalyzed C–H oxidation reactions often require the use of oxidants other than O_2. Here, we demonstrate a ligand-based strategy to replace benzoquinone with O₂ as the stoichiometric oxidant in Pd-catalyzed allylic C–H acetoxylation. Use of 4,5-diazafluorenone (1) as an ancillary ligand for Pd(OAc)₂ enables terminal alkenes to be converted to linear allylic acetoxylation products in good yields and selectivity under 1 atm O₂. Mechanistic studies reveal that 1 facilitates C–O reductive elimination from a π-allyl-Pd^II intermediate, thereby eliminating the requirement for benzoquinone in this key catalytic step.

The selective oxidation of C–H bonds in organic molecules is an important and growing field.¹ Many of the emerging methods utilize palladium catalysts in combination with stoichiometric organic or transition-metal oxidants, such as PhI(OAc)₂, benzoquinone, Cu^II or Ag^I.2 Mechanistic studies suggest that these oxidizing agents often react with organopalladium(II) intermediates and promote reductive elimination of carbon-carbon or carbon-heteroatom bonds.³ Replacement of these oxidants with molecular oxygen represents a significant fundamental challenge that has important implications for environmentally-benign, large-scale applications of these methods.^4,5 Here, we present a ligand-based strategy to replace benzoquinone (BQ) with O₂ as an oxidant in Pd-catalyzed allylic C–H acetoxylation reactions. Preliminary mechanistic studies reveal that the use of 4,5-diazafluorenone (1) as an ancillary ligand facilitates C–O bond formation from π-allyl-palladium(II) species and, thereby, permits O₂ to be used as the oxidant in a Pd^II/Pd⁰ catalytic cycle without requiring BQ.⁶

Pd-catalyzed allylic acetoxylation reactions represent a prominent class of C–H oxidations that have been the subject of extensive historical⁷ and contemporary^8,9 interest. The vast majority of these reactions utilize BQ as a stoichiometric or cocatalytic oxidant. Fundamental studies implicate at least three possible roles for BQ in the catalytic mechanism (Scheme 1): (1) to promote nucleophilic attack by acetate on a π-allyl-Pd^II species (step II),^3a,d (2) to displace the allylic acetate product from Pd⁰ following reversible C–O bond formation (step III),^8d and (3) to oxidize Pd⁰ to Pd^II (step IV).¹⁰ Because molecular oxygen is also capable of oxidizing Pd⁰ to Pd^II,⁶ we reasoned that an aerobic allylic C–H oxidation process could be achieved by identifying a ligand for Pd that could eliminate the requirement for BQ in steps II and III.

Scheme 1.

Proposed Mechanism for Palladium-Catalyzed Allylic Acetoxylation.

Pyridine, phenanthroline (phen) and related nitrogenous ligands have been widely used in Pd-catalyzed aerobic oxidation reactions.¹¹ Such ligands were tested recently by Lin, Labinger and Bercaw in allylic acetoxylation reactions,^8d and a bipyrimidine-Pd(OAc)₂ catalyst was shown to be effective with 2 equiv of BQ as the oxidant. We reevaluated ligands of this type under 1 atm of O₂ to test their ability to support aerobic Pd-catalyzed acetoxylation of allyl benzene (Table 1). Pyridine (entry 2), bipyridine (bpy, entry 3), phenanthroline (phen, entry 8), bipyrimidine (bpm, entry 11) and a number of bpy and phen derivatives (entries 4–7, 9, 10) were almost completely ineffective; mostly unreacted allyl benzene was obtained at the end of the reaction. The lack of Pd black in these reactions suggested the problem is a lack of reactivity, not catalyst decomposition. A noteworthy exception to these poor results was obtained with 4,5-diazafluorenone (1), which led to an 81% yield of the linear acetoxylation product (entry 12). This ligand was included in the screen based on the hypothesis that the carbonyl group could promote backbonding from the Pd^II center. We reasoned that contribution from the Pd^IV resonance structure (eq 1) might facilitate C–O reductive elimination from a π-allyl-Pd^II intermediate and enable BQ-free catalytic turnover (cf. Scheme 1, step II). Subsequent control experiments failed to validate this hypothesis, however. For example, use of di-2-pyridyl ketone, 2, another ligand capable of backbonding, afforded the allylic acetoxylation product in only 5% yield, whereas use of another diazafluorene ligand, 3, in which the carbonyl group in 1 was replaced with two methyl substituents, resulted in a moderate yield of the acetoxylation product (50%). The latter result is not as good as the acetoxylation yield obtained with 1, but it is substantially better than the yield with 2 or the other ligands in Table 1. These results suggest that geometric properties of 1 (e.g., the bite angle), rather than backbonding, underlie the effectiveness of this ligand.

Table 1.

Identification of a Ligand for Palladium-Catalyzed Aerobic Allylic Acetoxylation.^a

Entry	Ligand	Yieldb	Entry	Ligand	Yieldb
1	None	4%	7		0%
2	c	4%	8		9%
3		4%	9		0%
4		0%	10		0%
5		0%	11		4%
6		3%	12		81%

^a5% Pd(OAc)₂ (3 mg, 0.0134 mmol), 5% ligand (0.0134 mmol), allyl benzene (35 μL, 0.268 mmol), AcOH (1 mL), 1 atm O₂, 80 °C, 18h.

^bGC (internal std = nitrobenzene).

^c10%

(1)

Successful aerobic acetoxylation of allyl benzene provided the basis for further optimization of the reaction conditions and evaluation of the substrate scope. It was possible to decrease the temperature to 60 °C and replace AcOH with dioxane as the reaction solvent; the optimized conditions featured 5 mol % Pd(OAc)₂, 16 equiv of AcOH and a catalytic amount of NaOAc (20 mol%) under 1 atm of O₂ (see Supp. Info. for additional optimization data). These Pd-catalyzed allylic acetoxylation conditions were then tested with a number of other alkenes (Table 2). Successful reactivity was observed with naturally occurring allyl benzene derivatives estragole and methyl eugenol (entries 2 and 3) as well as simple hydrocarbon-based α-olefins (entries 4 and 5). Silyl ethers were well tolerated, as were esters, acetals, amides, and carbamates (entries 6–11). Nearly all of the substrates examined produced the linear acetoxylation product exclusively in good yields and exhibited a strong preference for formation of the E isomer of the alkene. The reactivity appears selective for terminal olefins, as β-methyl styrene, cyclohexene, methyl crotonate and methylenecyclohexane gave little to no desired product.

Table 2.

Aerobic Allylic Acetoxylation of Terminal Olefins.^a

Entry	Substrate	Time (h)	Product	E:Zb	Yieldc
1		24		17:1	81% (89%)
2		24		14:1	79% (87%)
3		24		6:1	68% (68%)
4		48		10:1	76% (82%)d
5		24		19:1	70% (76%)
6		24		21:1	84% (92%)
7		48		6:1	52% (72%)
8		48		29:1	76% (87%)
9		24		7:1	76% (84%)e
10		48		36:1	71% (84%)
11		24		16:1	74% (86%)

^aReaction conditions: 5 mol % Pd(OAc)₂ (0.05 mmol), 5 mol % 1 (0.05 mmol), 20 mol % NaOAc (0.20 mmol), substrate (1.0 mmol), dioxane (2.8 mL), AcOH (0.9 mL, 16 mmol), 1 atm O₂, 60 °C.

^bBased on GC analysis of the crude mixture.

^cIsolated yields. GC yields are given in parenthesis.

^d3:1 mixture of allylic and homoallylic acetates.

^e4 atm O₂. 5:1 mixture of linear and branched isomers.

These allylic oxidation reactions enable anti-Markovnikov hydration of α-olefins to be achieved via tandem O₂/H₂- coupled redox steps. This net transformation represents a long-standing challenge in the field of catalysis,¹² and it can be achieved in a straightforward one-pot process consisting of Pd-catalyzed allylic acetoxylation, removal of the acetate under basic conditions and hydrogenation of the alkene (Scheme 2A). No additional catalyst is required for the hydrogenation step; addition of activated carbon to the crude reaction mixture, which still contains Pd from the acetoxylation reaction, and stirring under 1 atm of H₂ results in efficient hydrogenation of the alkene. Good yields of terminal alcohols were obtained for three representative substrates utilizing this sequence (Scheme 2B). The isolated yield of the alcohol essentially matches that of the initial acetoxylation step.

Scheme 2.

Net Anti-Markovnikov Hydration of Terminal Alkenes via a One-Pot, Three-Step Sequence.

Mechanistic studies have begun to provide insights into the ability of the diazafluorenone ligand (1) to support aerobic BQ-free catalytic turnover. Two similar [(L)Pd^II(η³-allyl)]OAc complexes, 4 (L = diazafluorenone, 1) and 5 (L = tBu₂bpy), were prepared in order to compare the effects of O₂ and BQ on the reactivity of the π-allyl-Pd^II complexes under catalytically relevant conditions (Figures 1 and 2). In the first set of experiments, the ability of 4 and 5 to undergo stoichiometric acetoxylation of the allyl ligand was probed by ¹H NMR spectroscopy under three separate conditions: in the absence of an oxidant (1 atm N₂), under 3 atm O₂¹³ and in the presence of 2 equiv BQ (Figure 1). The diazafluorenone complex reacts to form allyl acetate in good yield under all conditions (70–90%); however, the reaction time varied from 24 h (N₂) to 3 h (O₂) to 1 h (BQ).¹⁴ Complex 5, with the catalytically incompetent tBu₂bpy ligand, is completely unreactive in the presence of N₂ and O₂; however, it undergoes acetoxylation in the presence of 2 equiv BQ (88%).

Figure 1.

Ligand effects on stoichiometric acetoxylation of well-defined π-allyl-Pd complexes in the absence of an oxidant (1 atm N₂) and in the presence of O₂ (3 atm) and benzoquinone (2 equiv). The indicated reaction time indicates the time needed for >95% conversion of 4 or 5.

Figure 2.

Experiments designed to probe the reversibility of C–O bond formation from (L)Pd(η³-allyl) complexex. (A) Acetate exchange into cinnamyl acetate in the presence of a (L)Pd(η³-allyl) complex; (B) Mechanistic basis for (L)Pd(η³-allyl)-catalyzed acetate exchange in cinnamyl acetate and the influence of an oxidant on the extent of exchange; (C) Extent of acetate exchange with different (L)Pd(η³-allyl) complexes in the absence and presence of an oxidant; (D) O₂ pressure effects on the extent of acetate exchange for the reaction in Figure 2A [(L)Pd(η³-allyl) = 4].

A second set of experiments probed the possibility of reversible C–O bond formation by monitoring the ability of π-allyl-Pd^II complexes 4 and 5 to mediate acetate-d₃ incorporation into cinnamyl acetate (Figure 2A). The proposed Pd-mediated mechanism for acetate exchange is shown in Figure 2B, and the data are summarized in Figure 2C. The diazafluorenone complex 4 mediated extensive acetate exchange (76% in 3 h) in the absence of an oxidant; however, this reactivity was eliminated by the presence of O₂ or BQ.¹⁵ Acetate exchange decreased systematically as the O₂ pressure increased (Figure 2D), presumably reflecting the ability of O₂ to trap the Pd⁰-alkene complex (k_ox, Figure 2B) and inhibit exchange. Negligible acetate exchange was observed with the tBu₂bpy complex 5, even in the absence of an oxidant.

These results provide insights into the catalytic steps that have been proposed to involve BQ (Scheme 1, steps II and III). The reactivity of 5 in the stoichiometric acetoxylation study (Figure 1) reveals that BQ can induce acetoxylation of an otherwise unreactive π-allyl-Pd^II complex. O₂ is not an effective BQ surrogate in this reaction. The stoichiometric reactivity of 4 reveals that BQ is not required for the C–O bond-forming step when diazafluorenone is the ancillary ligand. We speculate that the distorted bite angle of the diazafluorenone ligand¹⁶ destabilizes Pd^II and allows the π-allyl ligand to undergo nucleophilic attack in the absence of BQ. Computational studies to probe this hypothesis will be the focus of future studies. The qualitative rate differences for the acetoxylation of 4 in the presence of N₂, O₂ and BQ (Figure 1) can be rationalized in two ways. O₂ and BQ can enhance the rate by trapping the putative Pd⁰-alkene species and preventing reversion to the π-allyl-Pd^II complex.¹⁷ Alternatively, the faster rate with BQ relative to O₂ could reflect the ability of BQ to promote acetoxylation of the π-allyl-Pd^II complex, in addition to trapping the Pd⁰ intermediate.

Further work will be needed to probe the effect of ancillary ligands on other steps of the catalytic mechanism, such as C–H activation, and their influence on the identity of the turnover-limiting step and the catalyst resting state. Nevertheless, the results presented here provide clues into the ability of an ancillary ligand to enable catalytic turnover with O₂, rather than BQ, as the stoichiometric oxidant, and a plausible catalytic cycle is shown in Scheme 3.

Scheme 3.

Proposed Mechanism for Palladium-Catalyzed Aerobic Allylic Acetoxylation.

This study points to a potentially versatile ligand-based strategy to achieve aerobic Pd-catalyzed C–H oxidation. For example, BQ and other oxidants have been proposed to promote C–C reductive elimination in oxidative cross-coupling reactions.3g,j,k On the basis of the present results, we anticipate that it might be possible to identify ancillary ligands that facilitate reductive elimination from Pd^II and thereby eliminate the requirement for undesirable stoichiometric oxidants. Efforts to test this hypothesis have been initiated.