Gold‐Catalyzed Annulation of 1,8‐Dialkynylnaphthalenes: Synthesis and Photoelectric Properties of Indenophenalene‐Based Derivatives

Abstract

A simple gold‐catalyzed annulation of 1,8‐dialkynylnaphthalenes utilizing a cationic gold catalyst was developed. Such a peri‐position of two alkynyl substituents has not been studied in gold catalysis before. Dependent on the substrate, the reactions either follow a mechanism involving vinyl cation intermediates or involve a dual gold catalysis mechanism which in an initial 6‐endo‐dig‐cyclization generates gold(I) vinylidene intermediates that are able to insert into C−H bonds. Indenophenalene derivatives were obtained in moderate to high yields. In addition, the bidirectional gold‐catalyzed annulation of tetraynes provided even larger conjugated π‐systems. The optoelectronic properties of the products were also investigated.

The 6‐endo‐cyclization of the first diyne unit (blue) with two alkynes in peri‐position induces a switch to a 5‐exo‐cyclization for the second peri‐diyne unit (red) in the fast modular synthesis of new extended π‐systems.

Introduction

Over the last two decades, homogeneous gold catalysis ^[1] has been utilized for the electrophilic activation of alkynes. Gold most commonly acts as a redox‐neutral and carbophilic π‐acid that activates carbon‐carbon multiple bonds towards nucleophilic attack. ^[2] Alkynes are still the dominating class of substrates, among them diyne systems ^[3] are very important as starting materials for gold‐catalyzed annulations to extended π‐systems. Different from the majority of gold‐catalyzed reactions, diyne substrates lead to high‐energy intermediates, in the case of two internal alkynes to vinyl cations as intermediates, ^[4] when at least one terminal alkyne is present, gold vinylidene intermediates ^[5] are generated. Both types of intermediates are able to insert into non‐activated aliphatic C(sp³)−H or aromatic C(sp²)−H bonds. This provides easy access to a diverse set of interesting organic scaffolds,[ ³ , ⁴ ] including π‐extended structures. Polycyclic aromatic hydrocarbons (PAHs), which might be the largest class of organic molecules, have been intensively investigated, due to their small HOMO–LUMO gaps these highly conjugated systems are promising for applications in the field of organic electronics. ^[6] Therefore, the development of modular synthetic methods to access polycyclic π‐extended conjugated systems is a promising research topic.

1,8‐Dialkynyl naphthalene structures are flexible and have been extensively been used as substrates in intramolecular ^[7] and intermolecular ^[8] annulations. Recently Wang's group reported an iodine‐mediated electrophilic poly‐cyclization in one step, which provides two different products, probably generated from vinyl cation intermediates or radical intermediates. ^[7i] Additionally, the synthesis of indeno[2,1‐a]phenalene derivatives by an iodine‐mediated electrophilic cyclization of 1,8‐dialkynylnaphthalenes was published (Figure 1).[ ^7j , ^7k ] Tobe's group elegantly designed a synthesis of a variety of rigid carbocyclic platforms, which are stabilized by flanking aromatic groups. ^[7o] Such fused five‐ or six‐membered carbocyclic rings are key substructures of π‐extended systems or organic functional materials.[ ^7j , ^7k , ⁹ ] Gold catalysts represent mild carbophilic π‐Lewis acids, ^[10] we here for the first time use 1,8‐dialkynyl naphthalenes with the parallel peri‐arrangement of the alkynes.

Figure 1.

Known iodocyclization of 1,8‐(bisalkynyl)naphthalene by a tandem radical/cationic pathway and the new gold‐catalyzed annulation reaction.

Results and Discussion

The 1,8‐diyne 1 a was chosen as a model substrate (Table 1). 5 mol % of various gold catalysts with sterically demanding ligands were examined in CDCl₃. Even at room temperature the conversions were fast (2.5 h or less, Table 1, entries 1–5). With the IPr ligand the product 2 a was obtained in 94 % yield (Table 1, entry 1). ^[11] With the triphenylphosphine ligand no improvement in the yield was achieved, but the yield was slightly higher than in the case of the electron‐poor phosphine ligand (CF₃CH₂O)₃P (94 % versus 90 %) (Table 1, entries 2 and 3). Despite a complete conversion of all starting material, the SPhos‐ligated cationic gold catalyst gave a lower yield (88 %, Table 1, entry 4). In C₆D₆, the reaction was completed within 15 min to afford 2 a in 96 % yield (Table 1, entry 5). (IPr)AuCl gave poor results even after prolonged reaction times (4 h), (Table 1, entry 6), and (IPr*)AuCl/AgSbF₆ delivered 2 a in 90 % yield (Table 1, entry 7). With a silver(I) catalyst only low yields of 2 a were obtained (Table 1, entry 8).

Table 1.

Optimization of the reaction conditions.^[a]

Entry[a]	Catalysts	Time	Yield [%][d]
1	(IPr)Au(NCMe)SbF6	1 h	94
2	Ph3PAuNTf2	1 h	94
3[b]	(CF3CH2O)3PAuCl/NTf2	2.5 h	90[e]
4	(SPhos)AuCl/AgNTf2	1 h	88
5[c]	(IPr)Au(NCMe)SbF6	15 min	96
6	(IPr)AuCl	4 h	10 (89)[f]
7	(IPr*)AuCl/AgSbF6	1 h	90[e]
8	AgNTf2	5 h	48 (36)[f]

[a] Reactions were run on 0.01 mmol scale in CDCl₃, and were afterwards directly analyzed by ¹H NMR spectroscopy. [b] 60 °C. [c] In [D₆]benzene. [d] Yield was determined by ¹H NMR by using hexamethylbenzene as the internal standard. [e] Isolated yield. [f] Recovery of 1 a.

Then we explored the reaction scope under the optimized conditions from Table 1, entry 1. A series of diynes with phenyl groups with various substituents at the aromatic system were synthesized, all diynes provided the corresponding products (2 a–2 k, Table 2). Electron‐donating substituents (Me and OMe) at para‐ and ortho‐positions of the phenyl ring gave yields between 89–95 % (2 a–2 d), while for arene 2 e the yield dropped significantly (73 %). The structure of 2 e at the solid state was confirmed by the single‐crystal X‐ray structural analysis. ^[12] Also substrates with fluorine substituents were smoothly converted to the corresponding phenalenes 2 f and 2 j in 89 and 84 % yield, but for 2 j heating to 50 °C for 48 h was necessary. The mesitylene‐substituted substrate 1 g was also converted to the corresponding phenalene 2 g in 94 % yield via the carbocation rearrangement. Amazingly, product 2 h was directly obtained from the general procedure for the Sonogashira‐coupling to form 1 h, no gold catalyst was needed, the palladium did the conversion; for none of the other substrates this was observed. This probably is induced by the significantly more electron‐rich thiophene ring, which allows the palladium species present during the cross coupling to interact with the alkyne units more strongly—unlike in the other substrates, which are dependent on the gold catalyst. Acenaphthene substrates with different substituents (Me and F) at para‐position on the phenyl groups were successfully converted to the corresponding phenalene 2 i and 2 j in 89 and 84 % yield. The reaction of 5,6‐diethynylacenaphthelene 1 k did not give the corresponding phenalene at all.

Table 2.

Scope of the reaction.^[a]

[a] Reaction conditions: 1 a (0.05 mmol), catalyst (5 mol %), solvent (CHCl₃, 1 mL), r.t. [b] Product 2 h was obtained directly from the palladium‐catalyzed coupling, which should have provided 1 h. [c] 50 °C.

Then we tested the reactivity of the unsymmetrical diyne 1 l to investigate the effect of the mesitylethynyl and 2‐benzo[b]thiopheneethynyl substituents (Table 3). First different steric hindered ligands were examined. The reaction of 1 l under the standard conditions (Table 3, entry 1) through 5 mol % of (IPr)Au(NCMe)SbF₆ in 1,2‐dichlororethane (DCE)_, resulting in azuleno[1,2‐a]acenaphthylene 3I (36 %) and phenalene 4 I (18 %) as a major products due to a 5‐exo pathway as well as a small amount of [b]fluorantheno[8,9‐d]thiophene 2I (36 %) due to a 6‐endo pathway (Scheme 1). The structures of 2 l and 4 l at the solid state were unambiguously confirmed by X‐ray structural analysis (Figure 2). ^[12] The more sterically hindered gold catalyst (IPr*)AuCl/AgNTf₂ dramatically diminished the formation of the phenalene 2I and improved regioselectivity of products 3 I and 4 I (Table 3, entry 2). Switching the counter ion from NTf₂ ⁻ to SbF₆ ⁻ were able to efficiently produce the compound 4 I with less of 3 I, but the product 2 l was not observed at all (Table 3, entry 3). Changing the solvent to benzene and toluene improved regioselectivity towards 3 l (Table 3, entries 4 and 5). Interestingly, when the reaction was conducted with electron‐poor ligands, P(OCH₂CF₃)₃ and P(C₆F₅)₃, azuleno[1,2‐a]acenaphthylene 3 I was obtained as a major product (Table 3, entries 6 and 9). Further screening showed that in the present of the bulky phosphine ligand tBuXPhos the reaction proceeded faster and three isomers were obtained in 89 % yield in a ratio of 1:2:2 (Table 3, entry 7). In contrast, neither (IPr)AuCl, nor AgSbF₆ alone promoted the desired reaction, showing that a cationic gold species is necessary for the annulation (Table 3, entries 10 and 11).

Table 3.

Optimization of the reaction conditions for diyne 1 k.^[a]

Entry[a]	Cat.	Time [h]	Solvent	Ratio[b] 2 l/3 l/4 l	Yield [%][c]
1	(IPr)Au(NCMe)SbF6	2.5	DCE	2:2:1	90
2	(IPr*)AuCl/AgNTf2	5	DCE	3.4:1:1	92
3	(IPr*)AuCl/AgSbF6	5	DCE	3.5:1:0	94
4	(IPr*)AuCl/AgSbF6	3	PhH	3:1:0	70
5	(IPr*)AuCl/AgSbF6	4	toluene	3:1:1	70
6	(CF3CH2O)3PAuCl/AgSbF6	3	DCE	1:3:0.5	95
7	(tBuXPhos)AuCl/AgSbF6	1.5	DCE	1:2:2	89
8	(IPr)AuCl/AgSbF6	3	DCE	1:2:2	88
9[d]	[(C6F5)3PAuCl]SbF6	4	DCE	1:2.5:0.5	73
10	AgSbF6	5	DCE	ND	ND
11	(IPr)AuCl	5	DCE	ND	ND

[a] General reaction conditions: 0.05 mmol 1 L, 5 mol % of Au catalyst, 5 mol % of Ag catalyst, solvent (1 mL). [b] Ratio of 2 l/3 l/4 l were determined by ¹H NMR analysis of the crude product. [c] Yield of the mixture. [d] 10 mol % [(C₆F₅)₃PAuCl]SbF₆ was used at 50 °C.

Scheme 1.

Synthesis of 5 m and 5 n by dual gold‐catalyzed cascade transformations.

Figure 2.

Solid‐state molecular structures of compounds 2 l and 4 l.

Recently, Zhang's group ^[3a] and our group[ ^3b , ^3c , ^3d , ^3e , ^3f , ^4a ] showed that gold vinylidenes, accessible from an internal and a terminal alkyne, are highly reactive intermediates and can efficiently insert into C(sp²)−H bonds or C(sp³)−H bonds. Thus, we next tested our naphthalene‐linked diyne systems (1 m, 1 n) with a dual activation gold catalysts (TDAC[PF₆]) in DCE at 90 °C. These were efficiently converted into the targets 5 m and 5 n in 89 and 97 % yields, respectively (Scheme 1). On the other hand, for yet unknown reasons no conversions were observed for the electron‐rich substituents (R²=OMe, tBu) 1 o and 1 p (Scheme 1).

Next, the bidirectional gold‐catalyzed conversion of the 1,4,5,8‐tetra‐(phenylethynyl)naphthalenes 6 a and 6 b was explored (Scheme 2). First we tested the standard reaction conditions, when the solution of the tetrayne 6 a in DCE was treated with (IPr)Au(NCM)SbF₆ (10 mol %) at 50 °C, the two dark purple isomers 7 a and 8 a were isolated in 32 % and 17 % yield (Scheme 2). The lack of symmetry of the compound visible from the ¹H NMR spectra of these products indicated that neither of them are the expected symmetrical products. Interestingly, an unexpected mode of intramolecular cyclization was observed. The expected 6‐endo‐dig‐cyclization proceeded in one side to construct a phenalene moiety, but the unexpected 5‐exo‐dig‐cyclization in the other side occurred to form a fluoranthane moiety. Regioisomers 7 and 8 were potentially produced via vinyl cation intermediates Va and Vb (Scheme 2). The connectivity of 8 a was unambiguously confirmed by single‐crystal X‐ray diffraction analysis. ^[12]

Scheme 2.

Gold‐catalyzed conversions of the substrates 6 a and 6 b.

In the case of 1,4,5,8‐tetrakis‐(phenylethynyl)naphthalene 6 a and 6 b, additional mode of distortion to reduce steric repulsion is effect to obtain the unsymmetrical products. According to the X‐ray structure of phenylethynylnaphthalene derivatives, ^[8a] while the structure of diene 1 is involved the expanding mode of distortion in the naphthalene core, in the case of tetrayne 6, both the twisting and expanding modes are observed together with bending of phenylethynyl units, in particular the carbons attached directly to the naphthalene core move away from each other compared to 1. ^[13]

We used density functional theory (DFT) calculations at the SMD/M06/def2‐TZVP//SMD/B3LYP‐D3//6‐31G(d), SDD(Au) level of theory in DCE to understand the mechanistic details of the bidirectional gold‐catalyzed reaction by considering 6 c as a model for 6 b (Figure 3). This combination of methods and basis sets has been recently confirmed by us to provide results having a better consistency with the experimental data. ^[14] Accordingly, the reaction commences with coordination of the gold complex to one of the alkyne moieties ^[15] of 6 c in an endergonic fashion with ΔG=3.5 kcal mol⁻¹. The resultant intermediate VI is a branching point for two routes, A 6‐endo‐dig‐cyclization via TS^VI‐a and a 5‐exo‐dig‐cyclization via TS^VI‐b. The calculations indicate that TS^VI‐a is lower in energy than TS^VI‐b, implying that in agreement with the experimental finding, the cyclization prefers the 6‐endo‐cycliztion mode. Once the first cyclization has occurred, intermediate VII is formed by another cyclization via transition structure TS^VII, furnishing intermediate VIII. Since TS^VII lies much lower in energy than TS^VI‐a and intermediate VIII is highly stable with a relative free energy of −17.8, the transformation VI→VII→VIII is not reversible and thus the regioselectivity of the cyclization is determined by the energy difference between TS^VI‐a and TS^VI‐b. Subsequently, the organic molecule IX is produced with ΔG=−66.1 kcal mol⁻¹ followed by a series of chemical steps from intermediate VII, as shown in Figure 3.

Figure 3.

Calculated mechanism for cyclization of 6 c catalyzed by [(IPr)AuNCMe]⁺. The relative free energies (in red) are given in kcal mol⁻¹.

The organic molecule IX has two different sites for coordination of the gold complex. If the gold complex coordinates to IX to give X, the ensuing intermediate then is a bifurcation point for two processes, 6‐endo‐dig‐cyclization via TS^X‐a and 5‐exo‐dig‐cyclization via TS^X‐b. In line with the experimental observations, the calculations support favorability of the 5‐exo‐dig‐cyclization, as evidenced by the finding that TS^X‐b is 3.9 kcal mol⁻¹ lower in energy than TS^X‐a. The same is true if we assume that the cyclization occurs from the other π‐complex that is, intermediate XI; similarly, for this case, transition structure of the 5‐exo‐dig‐cyclization (TS^XI‐b) lies 2.1 kcal mol⁻¹ below that of the 6‐endo‐dig‐cyclization (TS^XI‐a), a result which agrees with the regioselectivity observed experimentally. Finally, we want to note that due to the closeness of the relative free energies of TS^X‐b and TS^XI‐b, both the syn‐ and anti‐products are predicted computationally to be produced with a nearly equal amount. This prediction is fully consistent with our experimental observations, the results of which are given in Scheme 2.

The optical properties of phenalenes 2 a–2 k were examined by UV/Vis absorption spectroscopy in dichloromethane (Figure 4). In view of maximum absorption wavelengths of 2 a, 2 f and 2 l, they show strong absorption at long wavelength (about 460 nm). Comparison with phenalenes structures, azulenoacenaphthylene 3 l exhibits the distinct broad peak between 500≈750 nm. The π‐extension has a significant effect on the absorption of 7 a and 8 a (λ=510 and 550 nm). The HOMO and LUMO levels of 2 a, 2 f, 7 a, and 8 a were estimated by cyclic voltammetry (Table 4). The HOMO energy level of 2 a containing an electron donor group (−5.33 eV) was higher than 2 f containing an electron acceptor (−5.47 eV). Based on the compounds 7 a and 8 a, π‐extension with fluoranthane provided lower LUMO levels and smaller HOMO–LUMO energy gaps.

Figure 4.

UV absorption of 2 a, 2 f, 4 l, 7 a, and 8 a.

Table 4.

Cyclic voltammetry data and estimated HOMO and LUMO energies.^[a]

	E ox [V]	E red [V]	E HOMO [b] [eV]	E LUMO [b] [eV]	E gap [c][eV]
2 a	0.53	−1.93	−5.33	−2.87	2.46
2 f	0.67	−1.74	−5.47	−3.06	2.41
7 a	0.41	−1.20	−5.21	−3.60	1.61
8 a	0.54	−1.16	−5.34	−3.64	1.70

[a] Cyclic voltammetry in CH₂Cl₂ containing 0.1 m ⁿBu₄NPF₆. [b] HOMO and LUMO energy levels in eV were approximated using the equation HOMO=−(4.80+E _ox), LUMO=−(4.80+E _red). [c] E _gap=LUMO−HOMO.

In conclusion, an efficient gold‐catalyzed annulation protocol through vinyl cation or ‐vinylidene intermediates via a selective shift towards a 6‐endo‐dig‐cyclization mode for the indeno[2,1‐a]phenalenes synthesis was developed. In addition, the bidirectional gold‐catalyzed annulation of tetraynes provided even larger conjugated π‐systems, the interesting different mode of the first and the second cyclization, leading to unsymmetrical products, was explored by a computational study.

Conflict of interest

The authors declare no conflict of interest.

Supporting information

As a service to our authors and readers, this journal provides supporting information supplied by the authors. Such materials are peer reviewed and may be re‐organized for online delivery, but are not copy‐edited or typeset. Technical support issues arising from supporting information (other than missing files) should be addressed to the authors.

Supplementary

S. T. Fard, K. Sekine, K. Farshadfar, F. Rominger, M. Rudolph, A. Ariafard, A. S. K. Hashmi, Chem. Eur. J. 2021, 27, 3552.