Introduction

Polycyclic aromatic hydrocarbons (PAHs) featuring non-hexagonal rings have attracted significant interest in the realm of organic optoelectronics1,2,3,4. Among these, acenaphthylene (AN)-containing PAHs stand out as exceptional structural units for organic functional materials, courtesy of their non-alternant electronic structure that enhances electron affinity (Fig. 1a)5,6,7,8,9,10,11,12. Consequently, the efficient synthesis of these compounds has garnered considerable attention. Traditional synthetic routes to AN-PAHs typically begin with the substitution, oxidation or annulation reactions of acenaphthene13,14,15,16. Nevertheless, these methods usually require multistep manipulation, and suffer from limitation of structural diversity. Recently, the advent of transition metal-catalyzed annulations has propelled significant advancements in the synthesis of AN-PAHs through catalytic strategies17,18,19,20. For instance, such PAHs could be obtained through palladium-catalyzed cyclopentaannulation reactions of halogenated naphthalenes with alkynes21,22,23 or 2-halogenated arylboronic acids24,25,26 (Fig. 1b). Despite significant progress, the development of a streamlined approach to access a variety of AN-PAHs, ideally utilizing abundant feedstocks, remains an appealing task in synthetic chemistry.

Fig. 1: Acenaphthylene (AN)-containing polycyclic aromatic hydrocarbons (PAHs) and their synthetic strategies.
figure 1

a Selected AN-containing PAHs. b Previous work: Pd-catalyzed cyclopentannulation reactions of halogenated naphthalenes. c This work: Rh-catalyzed tandem penta- and hexaannulation reactions of aryl ketones.

Full size image

Aryl ketones are readily accessible chemicals and fundamental building blocks in organic synthesis. In recent years, transition metal-catalyzed ortho-C−H activation and functionalization of aryl ketones have been studied extensively27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44. Among these transformations, the direct C−H annulation reactions of aryl ketones with alkynes have emerged as a versatile strategy for the construction of π-conjugated molecules, as exemplified by the pioneering works of Glorius27, Cheng28, Wang29, Maji45, and our group46,47,48. Despite these advances, most studies have used aryl or alkyl alkynes as substrates for annulation reactions. The aryl or alkyl fragments are often difficult to transform or remove from the formed PAHs, limiting the practical utility of these reactions.

Dimethyl acetylenedicarboxylate (DMAD), an electron-deficient alkyne diester, is widely used in Michael reactions, cyclizations (Diels–Alder and 1,3-dipole) but rarely in transition metal-catalyzed C–H activation-annulations49. Recently, we demonstrated its capability in C–H activation-annulation reaction with naphthalene ketones to generate aromatic polycarboxylic esters, establishing an efficient strategy for the synthesis of graphene-like molecules by utilizing the removability of ester groups as a key point50. Inspired by these studies, we envisaged that its annulation reaction with aryl ketones could rapidly access AN-containing PAHs. In the proposed reaction, a fulvene species, generated from the transition metal-catalyzed C–H activation-annulation of aryl ketone with alkyne, could be a key intermediate. The main challenge lies in effectively coupling the C–H activation-annulation with the Diels–Alder reaction. As proof of concept, we herein disclose a tandem C−H penta- and hexaannulation reaction of aryl alkyl ketones with acetylenedicarboxylates, offering access to a diverse array of AN-PAHs in an atom- and step-economical manner (Fig. 1c). This reaction demonstrates high chemo- and regioselectivity, with the extension of five- and six-membered rings proceeding in an orchestrated pathway.

Results

Optimization of the Reaction Conditions

At the onset of the optimization study, 1-(naphthalen-2-yl)ethan-1-one (1a) was chosen as model substrate to react with DMAD (2a). To our satisfaction, systematic evaluation of reaction parameters led to the highest yield of the desired product 3 at 62% along with the recovery of 1a in 24% yield under an optimized reaction conditions: 1a (0.1 mmol, 1.0 equiv.), 2a (0.4 mmol, 4.0 equiv.), [Cp*RhCl2]2 (5 mol%), AgOTf (20 mol%), CuO (0.3 mmol, 3.0 equiv.), 1-methylcyclohexane-1-carboxylic acid (1-MeCHA) (50 mol%), and DCE (2.0 mL) in a sealed 25 mL Schlenk tube under N2 atmosphere at 150 °C (oil bath) for 16 h (Fig. 2a). The structure of 3 was confirmed by X-ray crystallography (Fig. 2a). Notably, after the reaction, the resulting mixture mainly contained 3 and unconverted 1a. Blank experiments revealed the indispensability of [Cp*RhCl2]2, CuO, and 1-MeCHA for this transformation (Fig. 2b, entries 2–3). [(p-cymene)RuCl2]2 showed negligible catalytic activity (Fig. 2b, entry 4). Replacing AgOTf with AgSbF6 or AgBF4 led to diminished yields of 3, and the reaction ceased completely when AgTFA was used as an additive (Fig. 2b, entry 5). Other copper oxidants such as Cu(OAc)2, Cu2O, and CuBr2 were proven less effective than CuO (Fig. 2b, entry 6). Moreover, this reaction exhibited robustness towards various organic acid additives, as replacing 1-MeCHA with PivOH, MesCO2H or 1-AdCO2H (Fig. 2b, entries 7-8). Furthermore, the choice of solvent significantly influenced the transformation. Replacement of DCE with PhCl or toluene resulted in yields of 3 in 26% or 8% (Fig. 2b, entry 9). The reaction did not occur at all when using THF, DMF or MeCN as solvents (Fig. 2b, entry 10). Additionally, a reaction condition-based sensitivity screening indicated that this process was relatively robust in the face of small changes in concentration and temperature, high oxygen level, moisture, and scale-up (Fig. 2c and Table S1)51.

Fig. 2: Optimization of the reaction conditions.
figure 2

a Standard reaction conditions: 1a (0.1 mmol), 2a (0.4 mmol), [Cp*RhCl2]2 (5 mol%), AgOTf (20 mol%), CuO (0.3 mmol), 1-MeCHA (50 mol%), and DCE (2.0 mL) in a sealed 25 mL Schlenk tube under N2 atmosphere at 150 °C (oil bath) for 16 h. b Impact of other reaction parameters. Isolated yields are given. c Reaction condition-based sensitivity screening. Me: methyl; Ph: phenyl; Cp*: 1,2,3,4,5-pentamethylcyclopentadiene; OTf: trifluoromethanesulfonate; PivOH: pivalic acid; 1-MeCHA: 1-methylcyclohexane-1-carboxylic acid; DCE: 1,2-dichloroethane.

Full size image

Substrate Scope

With the optimized reaction conditions in hand, we embarked on a comprehensive exploration of the substrate scope using various aryl ketones 1. As summarized in Fig. 3, 1-(naphthalen-2-yl)ethan-1-ones, equipped with electron donating groups such as methyl, methoxy, phenoxy, benzyloxy, pivaloyloxy, hydroxyl, 4-bromobutoxy, and even methylthio on the naphthyl ring, smoothly underwent this tandem annulation reaction, providing the corresponding AN-PAHs in moderate to good yields (Fig. 3, 3-11). The hydrolysis of compound 3 was achieved by refluxing it in a KOH solution composed of water and THF, resulting in the corresponding tetraacid 3’ in a 96% yield. The versatility of the method was further demonstrated with the successful engagement of substrates featuring halogen substituents (fluoro, chloro, bromo) and electron withdrawing groups (ester and formyl), showing the potential for late-stage modification of the annulation products (Fig. 3, 12–16). The applicability of the reaction extended to naphthalene ketones with aryl or heteroaryl substituents, affording the desired products in moderate yields (Fig. 3, 17–20). Encouragingly, even simple phenyl ketones, such as acetophenone (1s) and 1-(4-methoxyphenyl)ethan-1-one (1t), participated in the reaction to yield products 21 and 22. The structure of 21 was confirmed by X-ray crystallography (Table S7). In addition, the protocol successfully accommodated larger π-conjugated ketones, including 1-(phenanthren-2-yl)ethan-1-one (1u), 1-(pyren-2-yl)ethan-1-one (1v), 1,1’-(pyrene-2,7-diyl)bis(ethan-1-one) (1w), 1-(anthracen-2-yl)ethan-1-one (1x), and 1-(chrysen-2-yl)ethan-1-one (1y), yielding the corresponding AN-PAHs in good yields (Fig. 3, 23-27). Moreover, other electron-deficient alkyne diesters (2b-2c) or diketone (2d) also participated in this reaction, delivering the corresponding products in moderate yields (Fig. 3, 28-30). It is worth to note that the low yields of the above reactions are primarily due to the low conversion of ketone substrates 1. Furthermore, for a comprehensive overview of the substrate scope, details on unsuccessful substrates such as bis(4-(methoxycarbonyl)phenyl)acetylene (2k), methyl 2-butynoate (2n), methyl phenylpropiolate (2o), and 1,2-bis(2-pyridyl)acetylene (2p), are given in Table S2.

Fig. 3: Scope of the reaction.
figure 3

Reactions run with 1 (0.1 mmol), 2 (0.4 mmol), [Cp*RhCl2]2 (5 mol%), AgOTf (20 mol%), CuO (0.3 mmol), 1-MeCHA (50 mol%), and DCE (2.0 mL) in a sealed 25 mL Schlenk tube under N2 atmosphere at 150 °C (oil bath) for 16 h. Hydrolysis conditions: 3 (0.1 mmol), KOH (2.0 mmol), and THF/H2O (2.0 mL, 1:1, v/v) in a sealed 25 mL Schlenk tube under N2 atmosphere at 100 °C (oil bath) for 24 h. Me: methyl; Et: ethyl; iBu: iso-butyl; Ph: phenyl; Bn: benzyl; Piv: pivaloyl; Cp*: 1,2,3,4,5-pentamethylcyclopentadiene; 1-MeCHA: 1-methylcyclohexane-1-carboxylic acid; OTf: trifluoromethanesulfonate; DCE: 1,2-dichloroethane.

Full size image

To further demonstrate the versatility of this protocol, a bidirectional annulation reaction of 1,1’-(naphthalene-2,6-diyl)bis(ethan-1-one) (1z) with 2a was conducted, affording a two acenaphthylene-containing PAH 32 in a synthetically useful yield (Fig. 4a). Subsequently, we conducted density functional theory (DFT) calculations to explore the photophysical and aromatic properties of compound 32. Figure 4c illustrates that the lowest unoccupied molecular orbital (LUMO) of 32 is distributed across the entire molecule skeleton, while the highest occupied molecular orbital (HOMO) predominantly resides along the long axis of 32. The LUMO and HOMO energy levels of 32 are -3.47 and -6.42 eV, respectively, demonstrating promise for semiconductor applications. The UV-vis and fluorescence spectra revealed maximum absorption and emission peaks of 32 at 430 nm and 543 nm (green fluorescence), respectively (Fig. 4d). To further investigate the stereo-electronic structure and aromaticity of 32, we conducted two-dimensional nucleus-independent chemical shift (2D-NICS)52 and anisotropy of the induced current density (ACID)53 analysis (Fig. 4e and f). The results of these calculations indicate that the six-membered rings (rings 1, 2, 3 and 4) exhibit typical aromatic properties, as evidenced by negative 2D-NICS values (Fig. 4b and e) and a clockwise ring current along the perimeter of the chrysene core (Fig. 4f). In contrast, positive 2D-NICS values (Fig. 4e) and an anticlockwise direction of the ring current suggest the anti-aromaticity of the five-membered rings (rings 5 and 6) (Fig. 4b and f). In addition, the UV-vis and fluorescence spectra of other representative products 3, 23, 24, 26, and 27 were measured. As shown in Fig. S9 and Fig. S10, the absorption and emission maxima of these products vary from 369 nm to 499 nm, and 519 nm to 597 nm, respectively, depending on the degree of π-extension and shape of the products. Due to a fusion of a pyrene unit, a remarkable red-shift is observed for both the absorption and the emission maxima of 24 (λabs = 499 nm, λem = 597 nm). To our delight, these compounds exhibit large Stokes shifts, ranging from 70 nm to 150 nm (Table S4). These excellent photophysical properties demonstrate that these compounds may have potential applications in molecular probes.

Fig. 4: Bidirectional annulation reaction.
figure 4

a Bidirectional annulation reaction of 1z with 2a. Reaction conditions: 1z or 31 (0.1 mmol), 2a (0.4 mmol), [Cp*RhCl2]2 (5 mol%), AgOTf (20 mol%), CuO (0.3 mmol), 1-MeCHA (50 mol%), and DCE (2.0 mL) in a sealed 25 mL Schlenk tube under N2 atmosphere at 150 °C (oil bath) for 16 h. b The structure of 32. c The optimized molecular geometry and molecular orbitals (HOMO and LUMO) of 32, calculated at the B3LYP/6-31 G(d) level of theory. d The absorption and emission spectra of 32 in dichloromethane (1 × 10−5 mol/L). e The two-dimensional nucleus-independent chemical shift (2D-NICS) calculations of 32, calculated at the B3LYP/6-31 G(d) level of theory. f The anisotropy of the induced current density (ACID) plots of 32, calculated at the B3LYP/6-31 G(d) level of theory. The red and blue arrows indicate the clockwise and anticlockwise ring currents, respectively. Me: methyl; HOMO: highest occupied molecular orbital; LUMO: lowest unoccupied molecular orbital; Eg: energy gap.

Full size image

Mechanism Study

To gain some insight into the reaction mechanism, several control experiments were conducted. Initially, treatment of 1-(1-methylnaphthalen-2-yl)ethan-1-one (1aa) with 2a under the standard reaction conditions afforded product 33 in 7% yield. In contrast, the reaction of 1ab and 2a produced only a byproduct 34 in 38% yield, arising from the trimerization of 2a, with substrates 1ab and 2a being recovered in 92% and 28% yields, respectively (Fig. 5a). These outcomes suggested that the reaction might commence with the C3−H activation and cyclopentaannulation of 1-(naphthalen-2-yl)ethan-1-one to form an intermediate with a five-membered ring. Indeed, a fulvene-containing intermediate 35 was obtained in 44% yield from the reaction between 1-(naphthalen-2-yl)propan-1-one (1ac) and 2a. Simultaneously, the acenaphthylene-containing product 36 was afforded in 17% yield, confirming the above conjecture (Fig. 5b). To clarify the subsequent cyclohexaannulation, compound 35 was subjected to further reaction with 2a in the presence of AgOTf and CuO, excluding [Cp*RhCl2]2, resulting in the desired product 36 in 42% yield (Fig. 5b). The structures of 33 and 36 were confirmed by X-ray crystallography (Table S8 and S9). These findings suggested that the hexaannulation might proceed through a Diels–Alder reaction, and the rhodium catalyst is likely not involved in this step. Subsequently, a series of radical capture experiments were conducted by adding radical scavengers such as 2,2,6,6-tetramethylpiperidine (TEMPO) and 2,6-di-tert-butyl-4-methylphenol (BHT) into the reaction of 1a and 2a. As shown in Table S3, the yield of 3 showed a nonlinear decrease along with the increase of TEMPO addition. However, when BHT was used as a free radical scavenger, the yield of 3 did not decrease significantly, and it could still reach 37% when one equivalent of BHT was added. These experimental results suggest that this reaction may not involve a radical mechanism. Finally, the chemical kinetics of the reaction of 1c and 2a were studied using in situ infrared (IR) measurements (Fig. 5c). Encouragingly, the absorption intensities of 1c, 2a, and 5 could be monitored by in situ IR measurements, tracked by change of the peaks at 1683, 1731, 1540 cm−1, respectively (Fig. 5c, (i)). The kinetic profile showed that this annulation reaction proceeded relatively fast without a discernible induction period, with the majority of 5 formed within 3 h (Fig. 5c, (ii-iii)).

Fig. 5: Mechanism study.
figure 5

a Reactions of 1aa or 1ab with 2a were performed under standard reaction conditions. b Reaction of 1ac with 2a was performed under standard reaction conditions. Reaction of 35 with 2a was performed without the addition of [Cp*RhCl2]2. c in situ IR experiments of 1c and 2a. Me: methyl; Cp*: 1,2,3,4,5-pentamethylcyclopentadiene; 1-MeCHA: 1-methylcyclohexane-1-carboxylic acid; OTf: trifluoromethanesulfonate; DCE: 1,2-dichloroethane.

Full size image

Based on the above observations, we propose a mechanism of the reaction between 1a and 2a, involving C–H activation-annulation and Diels–Alder reaction (Fig. 6a). The reaction initiates with a ketone-directed C3–H activation of 1a with Cp*RhX2 (X = OTf or RCO2) to form a five-membered cyclorhodium intermediate I. Subsequently, intermediate I chelates with 2a, leading to the formation of intermediate II. Alkyne insertion and subsequent intramolecular electrophilic attack of the carbonyl unit generate intermediate IV. Transmetalation of intermediate IV with copper carboxylate salt yields intermediate VI and Cp*RhX2 through a transitional intermediate V. beta-Hydrogen elimination of intermediate VI produces intermediate VII. A Diels–Alder reaction of intermediate VII with 2a delivers intermediate VIII, corroborated by a high-resolution mass spectrometry (HRMS) determination (Fig. S7, m/z calcd: 459.1050, found: 459.1046). Finally, intermediate VIII undergoes further oxidative aromatization in the presence of copper or silver salts to produce the desired product 3.

Fig. 6: Proposed mechanism and stepwise annulation reactions.
figure 6

a Proposed mechanism of the tandem annulation reaction. b Stepwise annulation reactions of 1 s with two electronically differentiated alkynes (see part VIII of Supplementary Information for the details). c Derivatization of 44 (see part IX of Supplementary Information for the details). Me: methyl; Et: ethyl; tBu: tert-butyl; Ph: phenyl; Piv: pivaloyl; Ac: acetyl; Cp*: 1,2,3,4,5-pentamethylcyclopentadiene; OTf: trifluoromethanesulfonate; 1-MeCHA: 1-methylcyclohexane-1-carboxylic acid; DPPP: 1,3-bis(diphenylphosphino) propane; DDQ: 2,3-dichloro-5,6-dicyano-1,4-benzoquinone; DCE: 1,2-dichloroethane.

Full size image

After investigating the reaction mechanism, we speculated that the tandem penta- and hexaannulation reaction could be achieved stepwise by adjusting the reaction conditions. To test this hypothesis, we conducted stepwise annulation reactions of acetophenone (1 s) with two electronically differentiated alkynes (Fig. 6b). Under slightly modified Rh-catalyzed conditions, acetophenone (1 s) reacted with various diphenyl alkynes, affording a series of 1-methylene-2,3-diaryl-1H-indenes (37-43) in good yields. These intermediates then underwent a Diels–Alder reaction with DMAD (2a), producing the desired AN-PAHs (44-50). The structure of 44 was confirmed by X-ray crystallography (Table S10). On the contrary, the Diels–Alder reactions of 37 with thiophene or indole were not achieved likely due to their aromaticity. Additionally, compound 44 was obtained in 60% and 17% yields in the absence of AgOTf or CuO, respectively, and the yield of 44 could still reach 9% in the absence of AgOTf and CuO simultaneously (Fig. S6), which suggests that the hexaannulation does not require an oxidant, but the addition of metal salts, especially copper salts, can significantly promote this process. These results further verified the proposed reaction mechanism and expanded the structural diversity of the AN-PAHs. To demonstrate the practicality of this reaction, several derivatizations of product 44 were conducted (Fig. 6c). Hydrolysis of 44 yielded diacid 51, which further underwent a condensation reaction to produce imide product 52 or a decarboxylation reaction to generate product 53. Finally, treatment of 53 with FeCl3 and DDQ in dichloromethane for 2 h delivered product 54 in a 68% yield. The photophysical properties of 54 were investigated, exhibiting yellow fluorescence and phosphorescence emissions at 544 nm and 524 nm, respectively (Fig. S11). These properties indicate that it may have potential applications in yellow OLED materials.

Discussion

In summary, we have introduced a tandem C–H annulation reaction of aryl alkyl ketones with acetylenedicarboxylates, offering a straightforward and rapid method to access a series of AN-PAHs from abundant feedstocks in a single step. This reaction demonstrates high chemo- and regioselectivity, a broad substrate scope, and excellent functional group tolerance. Mechanism studies has elucidated a comprehensive pathway involving Rh-catalyzed C–H cyclopentaannulation, followed by a Diels–Alder reaction. Additionally, stepwise annulation reactions of acetophenone with two electronically differentiated alkynes verified the proposed mechanism and expanded the structural diversity of AN-PAHs. We anticipate that this work will captivate the synthetic community, offering a valuable toolkit for materials chemists to explore novel acenaphthylene-based organic functional materials.

Methods

General procedure for tandem C−H penta- and hexaannulation reactions

A 25 mL Schlenk tube with a magnetic stir bar was charged with [Cp*RhCl2]2 (3.1 mg, 5 mol%), AgOTf (5.1 mg, 20 mmol%), CuO (24.0 mg, 0.3 mmol, 3.0 equiv.), 1-MeCHA (7.1 mg, 0.05 mmol, 0.5 equiv.), naphthalene ketone 1 (0.1 mmol, 1.0 equiv.) and alkyne 2 (0.4 mmol, 4.0 equiv.) in DCE (2.0 mL) under N2 atmosphere. The vessel was sealed tightly with a screw cap, then stirred at 150 °C in an oil bath for 16 h. The resulting solution was cooled to ambient temperature, diluted with 10 mL of CH2Cl2, filtered through a celite pad, and washed three times with 10 mL of CH2Cl2. The obtained organic extracts were evaporated under reduced pressure and the residue was absorbed into small amounts of silica gel. Purification was performed by column chromatography on silica gel to provide the desired product.