Ethers are indispensable structural motifs present in numerous natural products, synthetic pharmaceuticals, and functional materials [1]. Over the past decade, intensive efforts have been devoted to developing several methods for the construction of such compounds. Among these developed methods, the C–H/O–H cross-dehydrogenation coupling (CDC) of heterocycles with alcohols is an ideal choice for the preparation of various heterocyclic ethers given its inherent step-economy and atom-economy [2]. Moreover, alcohols are perfect coupling reagents because they are naturally abundant, have low cost and high stability, and are easy to handle. However, most of these CDC reactions depend on transition-metal catalysts and/or strong oxidants, which often lead to unwanted metal residues and chemical wastes. In addition, the C–C bond formation competitive reaction adjacent to the oxygen atom of alcohols could compromise the yield and/or selectivity.

Quinoxalin-2(1H)-ones are prevalent nitrogen-heterocyclic units [3-18] in a broad range of pharmaceutically and biologically active compounds [19, 20]. As a consequence, the development of novel synthetic protocols for the facile diversification of quinoxalin-2(1H)-one has been actively explored over the past few years [21]. In particular, a variety of efficient methods for synthesizing various 3-functionalized quinoxalin-2(1H)-ones through the direct C–H functionalization of quinoxalin-2(1H)-ones has been well-established, which include region-selective alkylation [22-30], arylation [31-35], acylation [36, 37], amination [38-44], and phosphonation [45-47] reactions. We were aware that the first example of a copper-catalyzed C(sp²)-H/C(sp³)-H CDC reaction of quinoxalin-2(1H)-ones and alcohols, with 3 equiv. of tert-butyl peroxybenzoate (TBPB) as the oxidant, was reported by Yuan and Mao (Scheme 1a) [24]. However, in this CDC reaction, the alcohols acted as radical partners to afford corresponding alcoholic 3-hydroxyalkylated quinoxalin-2(1H)-ones, which are structurally restricted. Recently, Zhang and Li pioneered a metal-free construction of 3-fluoroalkoxylquinoxalin-2(1H)-ones from quinoxalin-2(1H)-ones and fluoroalkyl alcohols with 2 equiv. of [bis(trifluoroacetoxy)iodo]benzene as an oxidant (Scheme 1b) [48]. In spite of the above achievements, an efficient catalytic protocol for the synthesis of various 3-alkoxylquinoxalin-2(1H)-ones under metal- and strong oxidant-free conditions is in great demand.

Ambient air is a low-cost, abundant, and safe oxidant that ideally fulfils the requirements of green and sustainable chemistry [49-56]. However, the efficient utilization of dioxygen (or ambient air) is challenging, because the existence of a triplet-ground-state of molecular oxygen largely blocks oxygen reduction and selective transformation with a large kinetic barrier, and the four-electron reduction of molecular oxygen involves more electrons than the two-electron substrate oxidation. In the past decade, the photocatalytic reaction using dioxygen (or ambient air) as a terminal oxidant has become a research hotspot of green chemistry [57-69], because it is very consistent with the 12 principles of green chemistry. However, to the best of our knowledge, no example of the synthesis of 3-alkoxylquinoxalin-2(1H)-ones through the visible-light photocatalytic CDC reaction with ambient air as the sole oxidant exists. As part of our ongoing interest in green chemistry [70-78], herein, we report an efficient and practical approach to the preparation of various 3-alkoxylquinoxalin-2(1H)-ones through a photoredox-catalyzed C(sp²)−H/O−H cross-dehydrogenative coupling of quinoxalin-2(1H)-ones and alcohols in the presence of ambient air as an oxidant, under metal- and strong oxidant-free conditions (Scheme 1c).

Our preliminary exploration started from the photocatalytic CDC reaction of N-methylquinoxalin-2(1H)-one (1a) as the template substrate, ethanol (3 equiv., 2a) as the coupling reagent, and 3 mol% of Rhodamine B as the photocatalyst. When the reaction mixture was irradiated by a 3W blue LED lamp (465 nm) in acetonitrile under air exposure for 12 h, the expected 3-ethoxy-N-methylquinoxalin-2(1H)-one (3aa) was generated in 21% NMR yield (Table 1, entry 1). Inspired by this experimental result, various photocatalysts were screened (entries 2−6). Among these examined photocatalysts, 9-mesityl-10-methylacridinium (Acr⁺-MesClO₄^-) was the optimal catalyst, furnishing product 3aa in 95% yield (entry 5). Varying the loading of the catalyst (entries 7–9) revealed that the Acr⁺-MesClO₄^- loading could be reduced to 2 mol% without any impact on the yield of 3aa (entry 8). Next, the screening of the amount of EtOH did not afford better results (entries 10–11). A series of solvents was investigated (entries 12–16), and the results showed that the solvents played a pivotal role in this transformation. Only in acetonitrile and ethyl acetate did the reaction occur smoothly (entries 8 and 12). Other reaction media, such as THF, DMF, DMSO, and DCE, did not support the visible-light photocatalytic CDC reaction (entries 13−16). Changing MeCN to EtOH resulted in the production of the expected product (3aa) in a 75% yield (entry 17). When the model reaction was performed under irradiation by the 3W green LED lamp or sunlight, only trace or 16% yield of the desired product (3aa) was observed, respectively (entries 18 and 19). The reaction with dioxygen as an oxidant did not provide a better result (entry 20). No CDC reaction was observed in the absence of Acr⁺-MesClO₄^- (entry 21).

With the optimized conditions (Table 1, entry 8), the reaction scope of quinoxalin-2(1H)-ones and the alcohols was explored. The representative results are illustrated in Tables 2 and 3, it is evident that our photocatalytic CDC reaction turned out to be rather general. Firstly, a wide range of functionalized quinoxalin-2(1H)-ones was well achieved, and the isolated yields were mostly ≥ 80%. The nitrogen atom of quinoxalin-2(1H)-one with various chain-length substituents (alkyl and functionalized phenyl substituents) on nitrogen atom underwent the transformation efficiently (3aa–3ka). Importantly, a series of valuable functional groups, such as alkenyl (3da), alkynyl (3ea), and ester (3ha–3ia) groups were well compatible in the standard conditions, which made this eco-friendly reaction more useful for further transformations. Quinoxalin-2(1H)-ones containing electron-withdrawing, -neutral, and -donating substituents at the phenyl ring could stereo-selectively and highly efficiently generate the desired products (3la–3ta). A series of functional groups including nitro (3la), fluoro (3ma), chloro (3na), bromo (3oa), and trifluoromethyl (3pa and 3qa) at the different positions of the substrates were well tolerated. Di-substituted quinoxalin-2(1H)-ones also delivered the expected products (3ra–3ta) in good to excellent yields. However, no alkoxylation product was formed when N-free quinoxalin-2(1H)-one was used as the substrate.

Subsequently, a wide range of alcohols was investigated under the standard conditions to react with quinoxalin-2(1H)-ones (Table 3). Much to our satisfaction, the alcohols with various carbon-chain lengths, isomeric structures, and functional groups underwent the present CDC reaction efficiently generating the desired 3-alkoxylquinoxalin-2(1H)-ones in high yields (3ab–3aq). The straight-chain alkyl alcohols (1a–1d and 1k–1o) as well as the bulky cycloalkyl alcohols (1i–1j) afforded the target products in good to excellent yields. Moreover, the alcohol substrates with important functional groups such as alkoxy (3ak), cyano (3al), trifluoromethyl (3am), and alkynyl (3an) groups all afforded the desired products in good yields, which provided an opportunity for subsequent manipulation. The eco-friendly and mild approach was also applied to complex natural alcohols, such as L-menthol and borneol, producing the expected products (3ap and 3aq) in 72% and 77% isolated yields, respectively. When phenol was employed as a substrate, no reaction occurred.

To demonstrate the utility of the present photocatalytic CDC reaction, both a large-scale synthetic reaction and a one-pot transformation were carried out. As shown in Scheme 2a, the photocatalytic alkoxylation reaction of N-methylquinoxalin-2(1H)-one was conducted at 8 mmol scale under the standard conditions, resulting in an 88% yield of 3aa. However, decreasing the amount of EtOH to 2.5 equiv. led to a decrease in the 3aa yield (76% yield). In combination with the clean reaction conditions and the outstanding leaving performance of methoxyl group, a one-pot sequential photocatalytic methoxylation/SN_Ar nucleophilic substitution reaction can be successfully achieved (Scheme 2b).

To gain insight into the reaction mechanism, some control experiments were conducted, as shown in Scheme 3. The addition of a radical scavenger (TEMPO or BHT) into the reaction mixture completely inhibited the photocatalytic reaction, indicating that the present transformation might proceed through a free-radical process (Scheme 3a). When the model reaction was conducted under a nitrogen atmosphere, no product 3aa was generated and the starting material (1a) was quantitatively recovered (Scheme 3b). As shown in Fig. 1, this time-profile of photocatalytic CDC reaction indicated that the developed reaction was entirely suppressed in the absence of visible light irradiation. These experimental results suggested that continuous visible-light irradiation is necessary for the photocatalytic CDC reaction.

A plausible mechanistic cycle for the photocatalytic CDC reaction of quinoxalin-2(1H)-ones and alcohols based on the control experiments and literature reports [22, 48, 57] is shown in Scheme 4. The reaction initiated with the activation of acridinium catalyst A by the irradiation of visible light, generating the excited-state species (B), which subsequently undergoes a single-electron transfer (SET) process along with quinoxalin-2(1H)-one 1 to afford the corresponding radical cation intermediate (C) and the acridine radical (D). Subsequently, alcohol 2 attacks the C3 position of intermediate C as a nucleophile to form radical E, which can be oxidized by dioxygen to generate the expected product (3) with the release of a hydroperoxy radical. Finally, the hydroperoxy radical oxidizes the acridine radical (D) to produce the ground-state catalyst (A) and the hydroperoxy anion through a SET process. The hydroperoxy anion coupled with hydrogen ion to form the side-product, hydrogen peroxide.

In conclusion, we have reported for the first time, a visible-light photocatalytic C(sp²)-H/O-H cross-dehydrogenation coupling of quinoxalin-2(1H)-ones and alcohols employing ambient air as an oxidant at room temperature, providing an eco-friendly and efficient route to access a wide range of 3-alkoxylquinoxalin-2(1H)-ones. The easily available raw materials, low catalyst loading (2 mol%), excellent functional group tolerance, sustainable dioxygen oxidant, metal-free reaction conditions, good to excellent yields, excellent atom-economy, and scalability of the present photocatalytic reaction make it very attractive in the pharmaceutical industry and for organic synthesis.