Computational, Mechanistic, and Experimental Insights into Regioselective Catalytic C-C Bond Activation in Linear 1-Aza-[3]triphenylene.

C-C bond activation by transition metal complexes in ring-strained compounds followed by annulation with unsaturated compounds is an efficient approach to generate structurally more complex compounds. However, the site of catalytic C-C bond activation is difficult to predict in unsymmetrically substituted polycyclic systems. Here, we report a study on the (regio)selective catalytic cleavage of selected C-C bonds in 1-aza-[3]triphenylene, followed by annulation with alkynes, forming products with extended π-conjugated frameworks. Based on density functional theory (DFT) calculations, we established the stability of possible transition metal intermediates formed by oxidative addition to the C-C bond and thus identified the likely site of C-C bond activation. The computationally predicted selectivity was confirmed by the following experimental tests for the corresponding Ir-catalyzed C-C cleavage reaction followed by an alkyne insertion that yielded mixtures of two mono-insertion products isolated with yields of 34-36%, due to the close reactivity of two bonds during the first C-C bond activation. Similar results were obtained for twofold Ir- or Rh-catalyzed insertion reactions, with higher yields of 72-77%. In a broader context, by combining DFT calculations, which provided insights into the relative reactivity of individual C-C bonds, with experimental results, our approach allows us to synthesize previously unknown pentacyclic azaaromatic compounds.

Introduction

The selective activation of unreactive C–C bonds is a key challenge in organic chemistry by opening up new and shorter synthetic paths to functionalized compounds such as natural products, various polyaromatic substances, and nanorings, among others.[1−8] Yet, this area remains mostly unexplored despite the abundance of potential targets and numerous possibilities for further development and expansion.

C–C bond activation studies should target substrates with several relatively equivalent cleavable bonds. However, cleavage may nevertheless occur at various sites, giving rise to mixtures of regioisomeric products. In such cases, computational modeling can theoretically explain the most likely reaction site by helping us to understand the reactivity of C–C bonds under specific experimental conditions and rationally designed reaction conditions. Therefore, theoretical modeling of the spatial arrangement and interaction energies of materials is crucial for a more detailed comprehension of molecular interactions between reactants and catalysts.[9−13]

In this respect, [n]phenylenes—compounds formed by alternating benzene and cyclobutadiene rings—can be envisioned as convenient candidates for C–C bond cleavage studies. Strain-relieve-driven catalytic activation of their C–C bonds yields species capable of participating in annulation reactions with unsaturated compounds and thus expands their molecular framework. Such ring expansion/annulation (REAP) or cut-and-sew processes in smaller aromatic precursors stand out among the possible strategies for constructing polyaromatic hydrocarbons (PAHs) with extended planar π-conjugated systems considering the potential applications of these compounds in different fields of chemistry, e.g., materials science or organic electronics.[14,15]

These applications, however, require developing a new approach to synthesize specifically substituted products. The most well-known representatives of this class of compounds are biphenylenes and angular and linear [3]phenylenes. Biphenylenes have been used as substrates in numerous reports of a transition metal complex-catalyzed activation of the cyclobutadiene C–C bond followed by alkyne or nitrile insertion giving rise to 9,10-substituted phenanthrenes[16−18] or 9-substituted phenanthridines,[19] respectively. In turn, Ni-[20] or Ir-catalyzed[21] insertion to angular [3]phenylenes can give rise to picenes and other aromatics such as benzo[3,4]cyclobuta[1,2-a]phenanthrenes and benzo[c]chrysene, among others. In contrast, only one report has addressed a catalytic C–C bond activation in linear [3]phenylene (Ni-catalyzed alkyne insertion[3]Phenylenes. Synthesis. 2013 ">22]), yielding a mixture of mono- and di-insertion products—benzo[3,4]cyclobuta[1,2-b]phenanthrene, benzo[k]tetraphene, and benzo[m]tetraphene. Similar insertions of their heteroaromatic analogues, which would lead to larger Ne-PAHs (N-embedded PAHs),[23−27] have surprisingly remained overlooked thus far.

In this context, we have recently reported a regioselective catalytic cleavage of proximal or distal C–C bonds in bistrimethylsilylated azabiphenylene.[28] The cleavage was followed by subsequent annulation with alkynes furnishing benzo[f]- and benzo[h] quinolines, depending on the catalytic conditions, and the presence of trimethylsilyl groups generally enhanced the C–C bond activation given the lower oxidative addition barrier to the C–C bond.[28] Our computational study helped us understand and rationalize the catalyst-dependent regioselective C–C cleavage. Based on our results, we explored the scope of selective C–C bond activation to bistrimethylsilylated 1-aza-[3]phenylene 1a—a compound with a more intricate molecular architecture—for several fundamental and practical reasons. First, the X-ray diffraction data[29] of the parental 1-aza-[3]phenylene clearly showed that all four bonds connecting the aromatic rings are more or less of the same length (1.508–1.510 Å), thus making it difficult to identify the most reactive bond.[30] Second, assessing the relative bond reactivities in this molecule could provide information on regioselective C–C bond activation in even more complex molecules. Third, C–C bond activation and alkyne insertion could give access to tailored Ne-PAHs with specific architectures. Bistrimethylsilyl 1-aza-[3]phenylene 1a has four nonequivalent C–C bonds, unlike bistrimethylsilylated azabiphenylene, which contains two nonequivalent bonds. As such, four different regioisomers could be formed upon the first C–C bond activation followed by mono-insertion of an alkyne (Scheme ).

Scheme 1

Plausible Insertion Products: Four Regioisomeric Products Can Be Formed after the First C–C Activation and Insertion of an Alkyne Moiety into Aza[3]phenylenes

Considering the above, we hypothesized that one-pot selective insertion(s) of alkynes in azaaromatic compounds bearing two cyclobutadiene rings could open a short pathway to new specifically polysubstituted azaaromatic compounds with complex architectures that would otherwise be difficult to prepare. In addition, these compounds could be candidates for new applications in materials science because they are regioisomeric to structurally related compounds such as dibenzoacridine.[31−33] To test our hypothesis, we aimed (i) to establish the order of the C–C bond cleaved and (ii) to assess whether one-pot multiple insertions can lead to compounds with five aromatic rings. More specifically, using a combined computational and experimental approach, we studied the selective C–C bond activation in bistrimethylsilylated 1-aza-[3]phenylene 1a. Following our theoretical findings, we experimentally verified the reactivity of each C–C bond, thereby devising a rapid and modular strategy for selective twofold C–C bond activation followed by alkyne insertion, forming Ne-PAHs with extended π-systems. The experimental results matched with the computational ones.

Results and Discussion

DFT Calculations

At the outset, theoretical calculations were made to shed light on a possible course of the reaction (for details see the Supporting Information). The first step of the analysis was to establish the most stable isomer of the insertion reaction between substrate 1a and the alkyne catalyzed by the Ir complex. Although we did not identify any substantial energy variation, our calculations revealed that the two most probable activation sites are the C–C bonds near the Si-substituted ring in 4 and 5, with a quasi-negligible stabilization in favor of the 2-distal C–C bond in 5 (Table ).

Table 1

Relative Free Energy (kcal·mol–1) for the Four Different Isomers Calculated at Two Different Temperatures, 298.15 and 443.0 K, Respectively

isomer	compound	relative free energy (kcal·mol–1, 298.15 K)	relative free energy (kcal·mol–1, 443.0 K)
1-proximal	2	+3.3	+3.6
1-distal	3	+0.9	+1.4
2-proximal	4	+0.09	+0.13
2-distal	5	0	0

Since the first screening did not provide any useful information on selectivity, we focused on the most favorable C–C bond cleavage in the cyclobutadiene rings of the substrate. The most probable process of catalyst activation is the dissociation of the dinuclear precursor [IrCl(COD)]2, which, in the presence of dppe, allows the formation of two coordinatively unsaturated trigonal planar IrCl(dppe) units releasing two COD molecules (for details, see the Supporting Information). This IrCl(dppe) unit is thus ready to interact with one of the four C–C bonds in the four-membered rings to promote the formation of the oxidative addition product.

In the first step of our computational analysis, we optimized the adducts of the IrCl(dppe) unit with the pyridine nitrogen atom and with the double bonds of the skeleton of 1a as a potential starting point. The formation of the adduct between the coordinative unsaturated fragment IrCl(dppe) and the substrate, namely, Nadd-1, through the pyridine nitrogen has been estimated to be particularly exergonic by −22.3 kcal·mol–1, shown in Figure S1. On the contrary, the formation of any adduct between the metal fragment and one of the C=C bonds of the substrate has been estimated to be slightly exergonic by 1–3 kcal·mol–1; thus, the temperature should be fundamental for overcoming the energy differences and promoting in some way the development of the reaction. In the second step, we analyzed all possible isomers of the oxidative addition intermediates—oxidative addition products (OAP)—provided by the interaction between the unsaturated IrCl(dppe) fragment and one C–C bond of the cyclobutadiene rings. Figure reported all of the eight optimized compounds (two isomers for each product, depending on the N position above or under the coordination plane) together with their relative free energy (in kcal·mol–1) considering the most stable structure of the 5-dist-up isomer OAPu-5 as the zero-energy point.

Figure 1

Optimized structures of the eight different isomers with the relative free energy (in parenthesis in kcal·mol–1). The hydrogen atoms were hidden for clarity.

Although the complex (OAPu-5) is only slightly favored at room temperature, the free energy differences are too small to prefer one to the other. In this view, we optimized all of the possible transition states, one for each of the eight isomers shown in Figure , and we found that the transition states for the isomers 2-prox-up (OAPu-4) and 2-dist-up (OAPu-5) are at lower energy than the other ones involving the two C–C linkages more proximal to the nitrogen center. In particular, the calculations highlight a free energy limiting difference between the TSu-5 or TSu-4 and the TSd-2 of ca. 12 kcal·mol–1. The three previously mentioned transition states are reported in Figure while all of the other five are shown in the Supporting Information, Figure S2.

Figure 2

Optimized structures for the transition state: (a) 1-prox-down (TSd-2), (b) 2-dist-up (TSu-4), and (c) 2-prox-up (TSu-5), respectively.

The free energy pathways for the three different isomers of the oxidative addition process are shown in Figure together with the associated free energy costs and gains. The first adduct is particularly favored for its anchorage to the pyridinic nitrogen over any other interaction between the unsaturated Ir(dppe)Cl and any unsaturated bond of the substrate. After the cleavage of the Ir–N bonding, the system reasonably evolves toward the coordination of a C–C bonding, followed by its cleavage. The free energy barriers associated with the breaking of the 2-dist-up (TSu-5) C–C bond (+9.6 kcal·mol–1) are lower than that of the first adduct, followed by the free energy barrier of 2-dist-down (TSu-4) (+11.5 kcal·mol–1). The highest value is associated with the 1-prox-down (TSd-2) isomer (+21.5 kcal·mol–1) and thus clearly indicating that 2-dist and 2-prox bonds are preferentially cleaved. The difference of 0.9 kcal·mol–1 between TSu-4 and TSu-5 indicates that they should be formed in a ∼3:1 ratio. This result agrees with the calculated Wiberg index bond order, which predicted the lowest bond order for the 2-dist and 2-prox positions in the starting adducts.

Figure 3

Free energy (kcal·mol–1) profile of the evolution from the separated reactants up to the transition state and to the final oxidative addition (OA) products. Black: TSd-2 and OAd-2; red: TSu-4 and OAu-4; and blue: TSu-5 and OAu-5.

In summary, our DFT calculations showed that the first cleavage should occur at distal C–C bonds, with a slight preference for the 2-distal C–C bond, followed by the cleavage of the 2-proximal C–C bond.

Insertion Reactions

The starting 7,8-bis(trimethylsilyl)1-azatriphenylene 1a was prepared according to the procedure described by Vollhardt et al.,[29] with small modifications from 5,6-bis(trimethylsilyl)-1-azabiphenylene, in an overall yield of 40%. Its nonsilylated analogue 1b was prepared by desilylation of 1a with CF3COOH in dichloromethane in 89% yield. For synthesis details, see the Supporting Information.

Our preliminary calculations on the catalytic action of Ir/dppe indicated that the most vulnerable bonds in 1a should be the 2-proximal and 2-distal bonds, where C–C bond activation likely occurs, giving rise, after a reaction with an alkyne, to the respective insertion products 4 (5,6-diR’-2,3-diRphenanthro[3’,2’:3,4]cyclobuta[1,2-b]pyridines) and 5 (5,6-diR’-2,3-diRphenanthro[2’,3’:3,4]cyclobuta[1,2-b]pyridines). With this knowledge, we studied insertion reactions with diphenylethyne (Table ). At the outset, we performed the insertion of diphenylethyne into 1a under the previously used conditions using an Ir/dppe catalytic system[21] at 130 °C (entry 1). However, only traces of products 4aa and 5aa were formed. By contrast, when MW heating (170 °C) was applied, we observed a rapid C–C bond cleavage followed by insertion of diphenylethyne providing products 4aa and 5aa in promising isolated yields of 8 and 24% (32% overall yield), respectively (entry 2). The 1:3 ratio of 4aa (the product of the 2-proximal C–C bond cleavage) and 5aa (the product of the 2-distal C–C bond cleavage) is in agreement with the prediction based on the DFT calculation. Along with these two major products, other minor aromatic side-products were detected as well, judging from our TLC analysis. However, isolation did not provide them in sufficient amounts for full and unequivocal characterization. In turn, we did not detect traces of di-insertion products of the second C–C bond cleavage followed by diphenylethyne insertion. The combination between experimental and computational results suggested that also in the case of Ir in place of Rh the sites of preferential C–C bond activation should be the same.

Table 2

Catalytic Insertion Reaction of Diphenylethyne and 4-Octyne with 1a

entry	catalyst	ligand	R1	(equiv)	yieldsa,bc
					4aa (%)	5aa (%)	C.Y. (%)b	6aa (%)	7aa (%)	C.Y. (%)b
1d	[Ir(cod)Cl]2	dppe	Ph	2	<3	<3		0	0
2		dppe	Ph	1	8	24	32	0	0
3		dppe	Ph	2	0	0		25	45	70
4		dppe	Ph	10	0	0		22	50	72
5		dppp	Ph	2	8	12	20	4	22	26
6		dppp	Ph	10	10	10	20	8	17	25
7	[Rh(cod)Cl]2	dppe	Ph	2	0	0		17	33	50
8		dppe	Ph	10	0	0		26	46	72
9		dppp	Ph	2	<3	<3		<3	<3
10e	[Rh(cod)2]BF4	dppef	Ph	10	15	21	36	<3	<3
					4ab (%)	5ab (%)		6ab (%)	7ab (%)
11	[Ir(cod)Cl]2	dppe	n-Pr	1	14	20	34	0	0
12		dppe	n-Pr	10	0	0		26	51	77

Isolated yields.

Combined yields.

<3% denotes the amount of the product that was detected by TLC but could not be isolated in sufficient purity.

Oil bath.

THF instead of toluene.

10 mol %.

After developing this approach for selective mono-insertion, we then tried to increase the yields of the mono-insertion product and also to induce the second C–C bond cleavage to achieve a twofold insertion. For this purpose, we performed the reactions with 2 and 10 equiv of the alkyne with respect to 1a (entries 3 and 4). Workup of the reaction mixtures in both instances provided only mixtures of the double-insertion products 6aa and 7aa in isolated yields of 25 and 22, and 45 and 50%, respectively (70 and 72% overall yield, respectively). Traces of monoinsertion products were not detected. Gratifyingly, compounds 6aa and 7aa were readily separable by simple column chromatography. In short, the predominant (sole) formation of the double-insertion products indicates that the 1-distal C–C bond in the mono-insertion products 4 and 5 is relatively readily activated in comparison with the cleavage of C–C bonds in 1a, confirming the DFT results. As far as the regioselective aspect is concerned, the preferential activation of the 1-distal C–C bond is in line with previous observations obtained from studies of alkyne insertion to 1-azabiphenylene. Next, we decided to change the Ir/dppe catalytic system for the Ir/dppp one because it gave better results for the distal C–C bond activation in azabiphenylene.[28] However, its use in a reaction of 1a with diphenylethyne provided a mixture of mono- and di-insertion products 4aa–7aa regardless of 2- or a 10-fold excess of diphenylethyne (entries 5 and 6). Workups of the respective reaction mixture enabled us to isolate the mono-insertion products 4aa and 5aa in 8 and 12% and in 10 and 10% yields and the di-insertion products 6aa (7,8,12,13-tetraRphenathro[3,2-h]quinolines) and 7aa (5,6,8,9-tetraRphenanthro[2,3-h]quinolines) in 4 and 22% and in 8 and 17% yields, respectively. The structures of 6aa and 7aa were unequivocally confirmed by single-crystal X-ray analyses (Figures and 5).

Figure 4

ORTEP plot of 6aa. Ellipsoids are shown at 50% probability.

Figure 5

ORTEP plot of 7aa. Ellipsoids are shown at 50% probability.

To assess the catalytic activity of other group IX transition metal-based catalysts, we tested the Rh/dppe catalytic system. The reactions were run with 2 and 10 equiv of diphenylethyne, obtaining mixtures of double-insertion products 6aa and 7aa in both cases (entries 7 and 8). The former yielded 6aa and 7aa in 17 and 33% (50% overall yield) and the latter in 26 and 46% (72% overall yield), respectively. The mono-insertion products were not detected in the respective reaction mixtures, thus confirming that the 1-distal C–C bond in 4 or 5 can be more easily activated than the 1-proximal bond. This result seems to be in line with the obtained computational findings of a very low barrier associated with the second insertion process compared to the first one, providing a reasonable justification for the missing detection of mono-insertion products.

Surprisingly, the Rh/dppp system did not lead to products (entry 9). We also tested a cationic Rh-based system ([Rh(cod)2]BF4) (entry 10), and it showed high selectivity for the mono-insertion products but low catalytic activity. Products 4aa and 5aa were isolated in only 15 and 21% yields (36% overall yield), respectively, despite the large excess of diphenylethyne (10 equiv). Di-insertion products 6aa and 7aa were detected in trace amounts only. The reaction of 1a was also performed with 1 or 10 equiv of 4-octyne under standard conditions (entries 11 and 12). The former case gave rise to a mixture of mono-insertion products 4ab and 5ab in 14 and 20% yields (34% combined yield). Di-insertion products were not detected. The latter one exclusively yielded a mixture of 6ab and 7ab in 26 and 51% yields (77% combined yield). The mono-insertion products 4ab and 5ab were detected only in trace amounts.

Desilylation under acidic conditions of the maternal silylated compounds 6aa, 6ab, 7aa, and 6ab provided 10a (65%), 11a (59%), 10b (76%), and 11b (88%), respectively (Scheme ).

Scheme 2

Desilylation of 6 and 7 to 10 and 11, Respectively

Finally, the reaction of 1-aza-[3]phenylene 1b (1a without the SiMe3 groups) with diphenylethyne, under the conditions described in entry 4 (Table ), provided a mixture of di-insertion products 10a and 11a in 19 and 40% yields (59% overall yield), respectively (Scheme ). The mono-insertion products 8a and 9a were detected only in trace amounts.

Scheme 3

Insertion of Diphenylethyne into 1b

Since compounds 11a and 11b were obtained in larger amounts by desilylation of 7aa and 7ab, we attempted annulations of 11a and 11b with diphenylethyne and 4-octyne to obtain the respective quinolizinium salts using various catalytic systems (Table , for a full account, see Table S1 in the Supporting Information). At the outset, the standard conditions ([Cp*RhCl2]2, Cu(BF4)2, O2, 50 °C)[34] were applied, but the reaction did not proceed. Replacing the Rh catalyst with Co or Ir catalysts did not yield the expected products either. Ultimately, we tested systems combining Cu(OAc)2 with AgBF4 as an oxidant at 100 °C for the annulations of 11a and 11b with diphenylethyne and 4-octyne, thereby obtaining the expected products 12a and 12b in 53 and 59% yields, respectively (entries 1 and 2). Co- or Ir-based catalysts fared poorly; the former did not provide any product, and the latter gave 12b in a meager 9% yield (entries 3 and 4). Single-crystal X-ray diffraction analysis of 12a unequivocally confirmed its structure (Figure ). Thanks to its higher reactivity, the Rh catalyst was more successful than the Ir catalyst.[35,36]

Table 3

C–H Activation-Based Annulations of 11 with Diphenylethyne and 4-Octyne to 12

entry	compound	catalyst	alkyne	12	isolated yields (%)
1	11a	[Cp*RhCl2]2	Ph	12a	53
2	11b	[Cp*RhCl2]2	n-Pr	12b	59
3	11b	Cp*CoI2	n-Pr	12b	0
4	11b	[Cp*IrCl2]2	n-Pr	12b	9

Figure 6

ORTEP plot of 12a. Ellipsoids are shown at 50% probability.

Subsequent attempts to conduct the reactions of 10a and 10b with diphenylethyne or 4-octyne under the previously mentioned conditions and switching to various catalysts and different reaction temperatures were unsuccessful. Only after changing to the cationic Rh-complex ([Rh(cod)2]BF4), the reactions of 10a and 10b with diphenylethyne and 4-octyne yielded the desired products 13a and 13b, albeit in low isolated yields of ∼5% (Scheme ). We presume that the lower ability of 10a and 10b to undergo annulation with alkynes is caused by steric hindrance exerted by the phenyl and n-Pr groups in position 13 that shield the C–H bond in position 14. This steric hindrance is evident in the X-ray structure of 10b (Figure ).

Scheme 4

C–H Activation-Based Annulations of 10

X-ray Crystal Structures

Single crystals of 6aa, 7aa, and 11aa were grown by hexane diffusion into their CH2Cl2 solutions, solving the crystal structures by single-crystal X-ray diffraction analysis (Figures –6). The crystal structure of 6aa revealed that the phenanthro[3,2-h]quinoline scaffold is almost planar, as expected. The dihedral angle values for ∠N1-C22-C21-C20 and ∠C9-C10-C11-C12 of the bay regions[37] are only 1.7 and 2.0°, respectively. The latter lies within the interval of dihedral angle values typical of structurally related compounds with the benzo[k]tetraphene scaffold, which range from 0.0 to 5.9°, depending on the respective substitution pattern.[38−41]

The crystal structure of 7aa revealed that the phenanthro[2,3-h]quinoline scaffold adopts a slightly helical arrangement, as shown by two dihedral angles of two neighboring bay regions: ∠N1-C22-C21-C20 is 1.7° and ∠C20-C19-C18-C17 is 12.9°. The distance between H20 and H17 is 2.12 Å. The sum of the angles is two times lower than that of the structurally related compound with the benzo[m]tetraphene scaffold (20.4 and 25.8°).[42]

The molecule 12a partly assumes a helical shape due to its [4]helicene arrangement. The naphthoquinolizinium fragment in 12a is almost planar, with dihedral angles ∠N1-C22-C21-C8 179.1° and ∠N1-C22-C21-C20 of 1.3°, resulting from ring fusion. Four aromatic rings on the right-hand side form the [4]helicene arrangement with dihedral angles ∠C24-C20-C18-C18 and ∠C20-C19-C18-C17 of 27.1 and 23.4°, totaling 50.5°. The sum of the dihedral angles is similar to that of 1-Ph[4]helicene, which is ∼48°.[43] Conjugation between the naphthoquinolizinium moiety and the [4]helicene backbone can nevertheless occur and affect its optoelectronic properties. As for the phenyl substituents, their orientation is unsurprisingly almost perpendicular to the basic framework.

Photochemical Properties

Figure shows the UV/Vis absorption and fluorescence spectra of 12a and 12b in dichloromethane, and the spectral data are summarized in Table S3. The spectra differ only slightly, showing characteristic absorption bands around 380, 458, and 488 nm (for 12a) and 373, 459, and 489 nm (for 12b). The photoluminescence (PL) properties of 12a and 12b were also examined in dichloromethane solutions at room temperature. Both compounds behave as fluorescent emitters, and the fluorescence spectra are unresolved with maxima at λem = 516 (12a) and 506 (12b) nm (Figure ). These values are red-shifted by ∼40 and 50 nm in comparison with their maternal 5,6,10,11-tetraphenyl (λem = 477 nm) and 5,6-dipropyl-10,11-diphenyl (λem = 457 nm) naphtho[2,1,8-ija]quinolizinium salts, which also had unresolved emission spectra.[44] The emission quantum yields (Φ) were Φabs = 0.46 (12a) and 0.45 (12b).

Figure 7

Absorption and emission spectra. Absorption (left) and normalized emission (right) spectra of 12a (blue) and 12b (red) were recorded as CH2Cl2 solutions.

In summary, the combined computational/experimental approaches allow a (regio)selective activation of C–C bonds in substrates with several similar reactive sites. Our results clearly show that a one-pot C–C activation/alkyne insertion reaction sequence is a feasible strategy for transforming strained aromatic molecules (linear aza[3]phenylene) into larger and specifically substituted π-conjugated N-embedded aromatic systems.

First, our DFT calculations, based on thermodynamic analysis, highlighted the preferential sites of C–C activation in the framework of linear 1-aza[3]phenylene. Second, our experiments confirmed all DFT results, according to which two regioisomeric mono-insertion products are formed. Third, the major reaction pathway is a twofold insertion in the presence of excess of an alkyne, giving rise to products with the phenathro[3,2-h]quinolines (6) and phenathro[2,3-h]quinoline (7) frameworks in high yields and selectivity. Finally, our preliminary experiments have shown that 9 can undergo C–H bond activation-triggered annulation, yielding quinolizinium salts. These findings demonstrate that DFT calculations can be used to provide useful hints on a synthetic route based on the ring-opening of azapolyphenylenes as a straightforward access to previously unknown, region-defined, fully aromatic azaPAHs with potential applications in organic electronics. Thus, such a ring expansion/annulation process (REAP) offers a complementary strategy to recently reported procedures, such as aza-APEX based on the reaction of an in situ generated nitrilium cation with aromatic compounds,[45] rhoda-electrocatalyzed domino alkyne annulations[46] or the π-extension of heteroaryl halides[47] because these methods cannot be used in our starting materials.