Photocatalytic oxidative dearomatization of orcinaldehyde derivatives

Abstract

For decades oxidative dearomatization has been employed as a key step in the synthesis of complex molecules. Challenges in controlling the chemo- and site-selectivity of this transformation have sparked the development of a variety of specialized oxidants; however, these result in stoichiometric amounts of organic byproducts. Herein, we describe a photocatalytic method for oxidative dearomatization using molecular oxygen as the stoichiometric oxidant. This provides environmentally benign entry to highly substituted o-quinols, reactive intermediates which can be elaborated to a number of natural product families.

Graphical Abstract

Oxidative dearomatization (OD) is a powerful transformation which allows for the conversion of phenolic starting materials to quinol products (see 2, Figure 1).¹ These reactive intermediates can be readily transformed through a breadth of carbon-carbon and carbon-heteroatom bond forming events and have been leveraged by Nature and chemists alike in the synthesis of complex molecules.² The use of lead(IV) tetraacetate (LTA) for oxidative dearomatization was first reported over half a century ago and continues to be widely used today, despite the generation of stoichiometric quantities of lead waste and lack of site-selectivity.^3,4 More recently, hypervalent iodide reagents have been employed to perform this transformation, such as PIDA,⁵ PIFA,⁵ and IBX⁶. which result in stoichiometric iodide byproducts that can be challenging to separate from the desired quinol. Efforts to attain high levels of chemo and site-selectivity have led to the development of bespoke oxidants (Fig. 1A). Recent advances in catalytic oxidative dearomatization have demonstrated that iodide reagents can be employed in catalytic amounts in conjunction with stoichiometric quantities of inexpensive oxidants, such as m-chloroperbenzoic acid (mCPBA), to afford dearomatized products.⁷ This renders the method catalytic in iodide, but still affords a stoichiometric organic byproduct (m-chlorobenzoic acid).⁸

Figure 1.

A) General scheme for oxidative dearomatization (OD). B) Mechanism for I(IH)-mediated OD. C) Abbreviated mechanism for FMDO-mediated OD. D) Proposed single-electron OD.

Nature employs flavin-dependent monooxygenases (FDMOs) to perform this challenging oxidation with perfect site- and stereoselectivity.⁹ These enzymes use the noncovalent cofactor FADH2 (see 6) which reacts with molecular oxygen to form hydroperoxyflavin (7), an electrophilic source of oxygen that can react with electron-rich arenes posed in the enzyme active site (Fig. 1B).^9,10 This environmentally benign system employs molecular oxygen as the stoichiometric oxidant and water as the solvent, ultimately affording a single product isomer. Recently, we disclosed our computational findings which support a triplet transition state in the reaction between hydroperoxyl-flavin (see 7) and resorcinol 11.¹¹ Homolysis of the peroxy O–O bond followed by a single electron transfer from the electron-rich arene substrate to the radical flavin cofactor was found to be a more energetically favorable pathway than the two-electron mechanism that is commonly proposed for this enzymatic reaction.⁹ These findings inspired us to develop a small molecule-catalyzed method for oxidative dearomatization that mimics Nature’s approach, but could alleviate the requirement for expensive stoichiometric reductants required by Nature’s catalysts. We envisioned developing a catalytic method which employs molecular oxygen as the stoichiometric oxidant, to produce water as the sole byproduct; thus, improving the overall atom economy and sustainability of the transformation.

Upon photoexcitation, riboflavin and derivatives thereof (see Table 1) have been employed as oxidants in the aerobic oxidation of benzylic C–H bonds,¹² cinnamic acids,¹³ benzylic alcohols¹⁴, amines¹⁵, and thiols.¹⁶ With computational support for oxidation of 11 by flavin through a low-barrier single electron transfer (SET) from the substrate,¹¹ we hypothesized photoexcited flavin could be a competent oxidant in the single-electron oxidation of 11. We envisioned trapping the resulting radical with molecular oxygen to form the desired C–O bond.

Table 1.

Optimization table for PCOD.

entry	base, 0.9 equiv	buffer	light source	% conversiona
1b,x	NaH	--	blue LEDs	0.5
2b,c	KOtBu	--	blue LEDs	1
3	--	water pH 8.0	blue LEDs	0.5
4	--	HEPES pH 8.0	blue LEDs	60
5	--	HEPES pH 6.0	blue LEDs	3
6	--	KPi pH 8.0	blue LEDs	53
7	--	Tris pH 8.0	blue LEDs	95
8	--	Tris pH 6.0	blue LEDs	2
9c	--	Tris pH 8.0	ambient light	10
10	--	Tris pH 8.0	dark	0
11d	--	Tris pH 8.0	blue LEDs	21
12e	--	Tris pH 8.0	blue LEDs	0

^a)determined by UPLC-PDA

^b)in MeCN

^c)with riboflavin

^d)with FAD in place of FMN

^e)no photocatalyst

Irradiation of 11 in MeCN with 0.4 mol % riboflavin, conditions commonly employed in benzylic oxidation of arenes,¹² returned only starting material. We hypothesized that 11 was too electron deficient to undergo single electron oxidation or that the resulting radical was not sufficiently stable to allow for radical-cage escape.¹⁷ Porco and coworkers have shown that less electron-rich arenes undergo oxidative dearomatization by a copper-oxo-sparteine oxidant when the substrate is deprotonated prior to introduction of the oxidant.¹⁸ Inspired by this solution to low reactivity, we next explored the addition of a number of organic and inorganic bases; however, this strategy afforded only trace amounts of product 12 (Table 1, entries 1 and 2).

Taking advantage of the solvent effect offered by moving to aqueous reaction conditions, wherein the pKa of resorcinol 11 was measured to be 7.2,¹⁰ the phenolate could be generated cleanly under mild conditions. We were gratified to find that in a variety of buffers at a pH of 8.0, with 0.4 mol % FMN, 11 was converted to a single, dearomatized product (Table 1, entries 4, 6 and 7). Tris buffer afforded superior results, providing 95% conversion of starting material to a single product, compared to 60% conversion with HEPES and 53% conversion with potassium phosphate (KPi) buffered solutions. We also noted a precipitous drop in reactivity with decreasing pH (Table 1, entries 5 and 8), supporting our hypothesis that the phenolate is the competent species for SET. However, increasing the pH did not afford improved reaction outcomes. Interestingly, we found FMN to be the superior photocatalyst to riboflavin, perhaps due to the increased solubility in our aqueous reaction media. Exclusion of light resulted in no conversion of the starting material and reactions conducted in ambient light afforded diminished yields relative to blue LED irradiation (entries 9–11).

The clean reaction profiles afforded by these mild, photocatalytic conditions were in sharp contrast to the results we obtained with model substrate 11 using traditional methods for oxidative dearomatization. Pb(OAc)₄, IBX and PIFA afforded a number of byproducts including structural isomers, overoxidation to afford quinone products and challenges in separating the stoichiometric oxidant byproduct or low reactivity (Figure S1).

Next, we probed the mechanism of this photocatalytic oxidative dearomatization (PCOD). Phenoxyl radicals such as 10 have been generated under a variety of photocatalytic conditions;¹⁹ to determine if excited FMN was directly interacting with the substrate, quenching rates for each of the reaction components were measured using Stern–Volmer analyses.²⁰ Under both aerobic and anaerobic conditions, Tris was found to quench the FMN excited state, with quenching rates (kq) of 2 × 10⁸ M⁻¹ s⁻¹ (see Figures S7 and S8 and Table S12). In contrast, when the phenolate was employed as the photoquencher in pH 7.0 water the quenching rate was found to be two orders of magnitude faster than the rate of Tris quenching (see Figures S4 and S6 and Table S12). Under aerobic conditions, a slower rate was observed with phenolate 9 due to the formation of product as evidenced in the UV-Vis spectrum (see Figure S2).

To gain experimental support for the formation of radical 10 under the reaction conditions, we attempted to trap the phenoxyl radical with several radical trapping reagents.²¹ In the presence of TEMPO the appropriate adduct mass was observed (Figure 2B and Figure S13). With preliminary support for the intermediacy of a radical such as 10, we sought support for this phenoxy radical being on a productive pathway to dearomatized product. To generate radical 10 by alternative means, 11 was subjected to AIBN under aerobic conditions. These conditions produced 10% of the o-quniol product in addition to a number of other oxidation products, including the Baeyer-Villiger product. With evidence for the intermediacy of the substrate-based radical, our attention turned to the source of the hydroxyl group incorporated into the final dearomatized product.

Figure 2.

A) Proposed catalytic cycle for PCOD. B) Mechanistic studies using model substrate 11, probing the role of a radical intermediate (left) and source of incorporated O-atom. Inset: Stern-Volmer plots of reaction components. 11 was found to be a strong luminescence quencher.

Observation of the peroxy product by MS and UV-Vis supported the role of O2 as the oxygen atom donor. To probe the operation of a second mechanism, which could proceed through a carbocation arising from a second SET, we performed the reaction in isotopically labeled water (Figure 2B). No incorporation of the labeled water in the hydroxylated product was observed, providing support against the intermediacy of a carbocation species en route to dearomatized product 12. It has been reported that in the presence of amine reductants and light, flavin containing species are capable of forming hydrogen peroxide.²² In order to determine if this oxidant might play a role in photocatalytic oxidative dearomatization, model substrate 11 was subjected to the optimized reaction conditions with five equivalents of ¹⁸O-labeled hydrogen peroxide (Figure 2B). Consistent with the proposed mechanism, no incorporation of ¹⁸O was observed in the dearomatized product.

It has been shown that photoexcited flavin can generate singlet oxygen, which could then undergo nucleophilic attack by phenolate 9, in one possible mechansim.²³ The addition of a singlet oxygen quencher, sodium azide, did result in a diminished yield of the product; however, the reaction was not sufficiently suppressed to implicate singlet oxygen as the critical oxidant. Therefore, we propose that triplet oxygen is the catalytically relevant species.

Based on these data and previous reports of flavin mediated oxidation of arenes^12–16 we propose the following mechanism for the photocatalytic oxidative dearomatization (Figure 2A). First, photoexcitation of the flavin photocatalyst could afford 15, which subsequently performs a single electron oxidation of phenolate 9. The resulting neutral substrate radical 10 could then undergo a recombination event with triplet oxygen to form peroxyl radical 17. SET from flavin semiquinone 16 to 17 would afford peroxy intermediate 18, which has been observed by MS. Finally, reduction of 18, possibly through oxidation of tris buffer,²⁴ affords product 2.

Having obtained experimental support for a proposed mechanism, we next investigated the substrate scope of this photocatalytic oxidative dearomatization. Initial preparative-scale reactions were plagued by low conversion of starting material; however, under an O2 atmosphere improved conversions were obtained, although isolated yield of 12 was low due to instability of the product upon concentration of the crude reaction mixture. We were gratified to find several structural perturbations from the model substrate 11 were tolerated. For example, substrates bearing an ester or ketone at C1 were dearomatized cleanly (Figure 3). A dicarbonyl substrate underwent dearomatization to afford the azaphilone bicycle 21. Interestingly, we found that the presence of an electron-withdrawing group at C1 was critical to obtain a dearomatized product rather than the known C–H oxidation product.¹² We hypothesize that the added resonance stabilization of the extended n-system in these compounds is critical to allow for sufficient radical character at the tertiary carbon, rather than on the primary benzylic carbon, to afford the benzylic alcohol product.

Figure 3.

Substrate scope of PCOD. IBX yields in grey. PCOD conditions: 0.4 mol% FMN, 50 mM Tris pH 8.0, white light, rt, sparging O₂. IBX conditions: IBX (1.1 equiv), TBAI (0.1 equiv), TFA (5 equiv), DCE, rt.

In conclusion, we have developed a photocatalytic method for oxidative dearomatization of densely functionalized arenes using molecular oxygen as the stoichiometric oxidant. The clean reaction profile of this photocatalytic transformation makes it an attractive alternative to traditional small molecule-mediated methods. The mechanism of this transformation has been investigated using a number of techniques, including luminescence quenching, radical trapping reagents, and isotopic labeling studies. Current limitations of this method include the requirement for a carbonyl group ortho to the hydroxyl group present in phenolic substrates. This method allows for the generation of highly substituted o-quinol products including azaphilones under mild, photocatalytic conditions. Work to expand this methodology to include less-electron rich arenes by tuning the redox properties of the photocatalyst is currently underway.