The realization of simple and green synthetic procedures is an important goal in organic synthesis [1, 2]. In this context, tandem reactions have been widely adopted to combat the tedious operational procedure associated with multistep synthesis [3-8]. While many acid-catalyzed tandem reactions have been developed [9, 10], the ever-increasing demand for novel molecules with biological and material uses and the laborious process of the conventional stepwise synthesis have resulted in the continuous search for developing simple and efficient tandem reactions.

Incorporation of renewable chemicals into heterocycles constitute one of the important ways to valorize biomass, which also follow the principles of green chemistry [11-14]. Acetol can be readily produced through the staged pyrolysis of biomass at low temperatures of 200-300 ℃ [15]. Although some acetol-derived heterocycles including pyrazine [16], oxazoline [17], furan [18], and quinoxaline [19] have been reported, incorporating acetol into novel heterocycles is still desired.

The synthesis of benzo[a]carbazoles has garnered much attention due to their wide application in medicinal chemistry (compounds A and B in Scheme 1) and photographic materials (compound C in Scheme 1) [20-24]. Numerous synthetic methodologies have been developed to construct this privileged core material in the past few decades [25-38]. Among those, protocols with 2-phenylindole as the starting material have recently emerged as one of the most appealing methods since elaborate design and multistep synthesis of the starting material is avoided. However, until now there have been only four approaches to implement this transformation as shown in Scheme 1: (1) In(ONf)₃-catalyzed [4+2] benzannulation with propargyl ethers (Eq. (1)) [39, 40], (2) palladium-catalyzed aerobic-oxidative cycloaromatization with internal alkynes (Eq. (2)) [41], (3) Rh(Ⅲ)-catalyzed cascade reactions with α-diazo carbonyl compounds (Eq. (3)) [42, 43], and (4) BiCl₃-catalyzed benzannulation with α-bromoacetaldehyde (Eq. (4)) [44]. Considering the versatility of benzo[a]carbazoles, the development of novel methods, such as metal-free methods, to synthesize this structural motif from readily available starting materials are in demand. Herein, we report for the first time, the Brönsted acidic ionic liquid (BAIL)-catalyzed synthesis of benzo[a]carbazole from 2-phenylindole (Eq. (5), Scheme 1) in a biphasic system with the following features: (1) renewable counter-reagent, (2) recyclable catalyst, (3) high atom economy (with water as the sole byproduct), and (4) high chemoselectivity.

We commenced our study with acetol and 2-phenylindole in nitroethane at 95 ℃. As shown in Table 1, no reaction occurred by employing weak acid catalysts such as LiBr and H₃BO₃ (Table 1, entries 1 and 2). ZnCl₂ did not catalyze the reaction either (Table 1, entry 3). However, the product (3a) was formed, albeit in a very low yield, when BiCl₃ catalyst was used (Table 1, entry 4). Strong acids such as Al(OTf)₃, Sc(OTf)₃, Cu(OTf)₂, PTSA and TfOH were also examined for their catalytic activity (Table 1, entries 5–9). However, the maximum yield reached only 37% with TfOH (Table 1, entry 9). We then attempted to examine the change in the yield using organic solvents such as nitromethane, acetonitrile, 1, 4-dioxane, and 1, 2-dichloroethane, but there was no increase in the yield (Table 1, entries 10–13). The non-polar solvent, toluene, and the protic solvent, ethanol, also failed to improve the yield (Table 1, entries 14 and 15). The reaction also gave two byproducts, 3a-a and 3a-b. Thus, in order to improve reaction selectivity, we performed the reaction under biphasic conditions. Unfortunately, neither the aqueous biphasic system (Table S1) nor the non-aqueous biphasic systems (Table S2) proved successful in enhancing the yield of 3a. BAILs have been widely used in organic catalysis [45-48]. Thus, some triflic acid-derived BAILs were examined in this study. Considering the hygroscopic property of BAILs, they were prepared in situ by mixing the corresponding zwitterions with TfOH. The imidazolium- (4a) and pyridinium-type (4b) BAILs showed similar catalytic activity as TfOH (Table 1, entries 16 and 17). However, in the presence of the ammonium-type (4c) BAIL, 3a was obtained in 45% yield (Table 1, entry 18). In addition, the yield of 3a increased with an excess amount of 4c, based on triflic acid (Table S3). When 20 equivalents of 4c were used, 3a was obtained in 92% yield (Table 1, entry 19). Since 3a was formed through a dehydration process, 8 equivalents of water, based on 4c, was added to check whether the excess amount of zwitterion played the role of a reservoir. However, the reaction yield did not change significantly (Table 1, entry 20). With 8 equivalents of water, 4c was completely dissolved in the aqueous phase and a biphasic system was formed in conjunction with nitroethane (Fig. S1). After the reaction, the TfOH remained in the aqueous phase and both the acid and 4c could be reused for five times (Fig. 1). Nitroethane was necessary for this reaction and replacement of nitroethane with other organic solvents resulted in reduced yields (Table S4). The ability of the amphiphilic zwitterion (4c vs 4d), reaction temperature, and time also affected the yield (Table 1, entries 21–23). Thus, the optimized conditions for the reaction was obtained with TfOH (10 mol%), 4c (200 mol%), and H₂O (1600 mol%) in EtNO₂ (1.0 mL) at 95 ℃ for 1 h.

The scope of the substrates was then probed under the optimized conditions (Fig. 2). An apparent electronic effect could be observed by varying the substituents on the indole ring of 2-phenylindole. 2-phenylindoles with an electron-donating substituent at the C5-position produced the corresponding benzo[a]carbazoles in excellent yields (3b and 3c). The congeners with weak electron-withdrawing substituents such as 5-fluoro-2-phenylindole and 5-chloro-2-phenylindole, only gave a moderate yield. However, by increasing the reaction time, the corresponding benzo[a]carbazoles, 3d and 3e, were obtained in 72% and 75% yields, respectively. A similar tendency was also observed by varying the substituents at the benzene ring of 2-phenylindole (3g, 3h, and 3i vs 3j). The N-substituted 2-phenylindole reacted readily with acetol to produce 3k in 86% yield, which reportedly exhibits pronounced antitumor activity against leukemia, renal cancer, colon cancer, and malignant melanoma cell lines [23]. Two 2-heteroarylindoles, 2-(furan-2-yl)-1H-indole, and 2-(thiophen-2-yl)-1H-indole, were also used in this reaction, with which the target products, 3l and 3m, were isolated in 76% and 75% yields, respectively.

To understand the limitation of this protocol we also examined the a-hydroxyketones (Scheme 2). α-Hydroxyacetophenone reacted smoothly with 2-phenylindole to generate the desired product (5a) in moderate yield. 2, 2-Diethoxyethanol as a masked glycolaldehyde [49], which is a biomass-derived chemical compound synthesized from cellulose or glucose, formed 5b in 80% yield. However, substitutions at the alpha position of the hydroxyl group seemed to be unfavorable since no expected products were obtained with 3-hydroxy-2-butanone and benzoin (5c and 5d). However, α-methoxyacetone gave the same product as acetol in 89% yield.

A proposed mechanism for this reaction is depicted in Fig. 3. The reaction was triggered by the Brönsted acid mediated nucleophilic addition of indole to ketone [50, 51], followed by dehydrogenation to generate intermediate Ⅰ. The carbocation intermediate Ⅰ resonates with the iminium intermediate Ⅱ and the oxonium intermediate Ⅲ. Michael addition of intermediate Ⅱ to 1a gave the byproduct 3a-b while the isomerization of intermediate Ⅲ via 2-phenylindole migration [52-54] followed by deprotonation furnished the byproduct 3a-a. We believed that the main product 3a was formed through the intermediate Ⅴ, which was generated from intermediate Ⅰ either by dehydration followed by protonation or hydride migration. Mechanistically, however, the formation of 3a directly from intermediate Ⅳ via 6π-electrocyclization followed by dehydration could not be ruled out. The existence of many electrophiles (intermediates Ⅰ–Ⅳ, 3a-a, and 1a) and nucleophiles (1a and 2a) simultaneously was a challenge to controlling the reaction selectivity. The sulfone group of 4c was assumed to play an important role in stabilizing the carbocation intermediate Ⅰ and thus, favor the formation of 3a.

Inspired by intermediate V, we envisioned that by introducing a nucleophilic site in the structure of -hydroxyketone and using a simple nucleophile, a similar tandem reaction could be possible. Thus, 2b and 2c were synthesized and subject to the reaction with N-methylindole (Scheme 3). Interestingly, the desired 1, 2-dihydronaphthalene derivative 6a was obtained with 2b but a carbazole 6b was formed with 2c, indicating the facile aerobic auto-oxidation of 4, 9-dihydrocarbazole.

In conclusion, an expeditious synthesis of benzo[a]carbazole from readily available 2-phenylindoles and bio-renewable acetol catalyzed by BAIL in a biphasic system, with the aqueous solution of ammonium zwitterions and nitroethane, was accomplished. This reaction proceeded smoothly without the need for costly or toxic metal-based catalysts. The catalytic system could be used for five times without a significant loss in its catalytic activity. We postulate that this novel route has a great potential for the synthesis of pharmaceutically relevant benzo[a]carbazole derivatives.

The authors declare no competing financial interest.

The Cooperative Innovation Center of Hubei province and the testing center of HUST are acknowledged.