Palladium-catalyzed synthesis of N-arylated carbazoles using anilines and cyclic diaryliodonium salts

The direct synthesis of N-arylated carbazoles through a palladium-catalyzed amination of cyclic iodonium salts with anilines is described. In particular, electron-poor aniline derivatives reacted smoothly with only 5 mol % of Pd(OAc)2 as catalyst to give the desired products in up to 71% yield. Furthermore, the reactivity of cyclic iodonium salts is compared with the reactivity of the corresponding cyclic bromonium analogues.


Introduction
Carbazoles play an important role as core structural elements in natural products (e.g., alkaloids) and pharmaceuticals [1]. In addition, the carbazole motif constitutes an immense class of materials in the rapidly growing field of molecular electronics. In particular N-arylcarbazoles have promising electroluminescent properties and have subsequently found diverse applications as hole-transport, or as host or luminescent-materials in electronic devices (OLEDs) (Figure 1) [2][3][4][5][6][7]. Representative examples are the host molecules mCP, CBP and CBZ1-F2, the hole transporter BCz2 [8] or the recently described thermally activated delayed fluorescence (TADF) emitter 4CzIPN [9]. Therefore, the efficient synthesis of N-arylated carbazoles is an attractive goal and numerous synthetic methods are known so far from the literature. The main synthetic routes are shown in Scheme 1. The majority are transition-metal mediated. Starting from functionalized 2,2'-biphenyls (path A) [10][11][12][13] or the direct arylation [14,15] of the free NH-functionality of carbazole (path B).
In the past decade, hypervalent iodine chemistry has undergone a renaissance and has developed to become a powerful area in synthetic organic chemistry. Open-chained iodonium salts are well explored in transition-metal-mediated reactions to construct new C-N bonds [16][17][18][19], whereas examples dealing with cyclic iodonium salts are underrepresented [20]. Our group is interested in the development of new C-X coupling strategies via (hypo)iodite or hypervalent iodine catalysis [21][22][23]. Here, we wish to present an alternative Pd-catalyzed method for the construction of N-substituted carbazoles based on a stable, cyclic iodonium salt and electron-deficient anilines [24,25].
In the initial C-N bond-forming step of this cascade reaction, a ring opening of the cyclic iodonium salt through the amine is proposed to give 2'-iodobiphenyl-2-phenylamine (I). In a second, Pd-mediated intramolecular cross-coupling, 9-phenyl-9H-carbazole (3a) should be observed (Scheme 2).

Results and Discussion
First, we decided to prepare cyclic iodonium salt 1 as the triflate salt, to avoid unwanted side-reactions in solution with a concurrent nucleophilic counterion [26]. However, 1 was synthesized from 2-iodobiphenyl according to the established one-pot procedure for the synthesis of diaryliodonium triflates [27,28] by Olofsson and co-workers (Supporting Information File 1).
toluene resulted only in trace amounts of 3a (Table 1, entry 1 and 2). After increasing the catalyst/ligand ratio (palladium to phosphine 5 mol % and 10 mol %, respectively) and using Cs 2 CO 3 as the base, 3a could be isolated in 35% yield (  ( Table 1, entry 5 and 7). When using bidentate ligands with bite angles higher than 100° (DPEphos 104°, Xantphos 108°) the reaction is more efficient and the yield increases significantly. Next, we asked ourselves, whether other palladium salts could be equal or better in efficiency and yield. Changing Pd 2 (dba) 3 to Pd(OAc) 2 had no significant impact on product yields ( Table 1, entry 12). Further increase of the catalyst ratio from 5 to 10 mol % had little effect (Table 1, entry 15).
Next, we decided to analyze the byproducts of this reaction by GC-MS analysis. For this experiment the reaction was performed according to conditions given in entry 12, Table 1.
After a deeper literature research we came to the conclusion, that those byproducts should probably not only arise from side reactions within the catalytic cycle (for instance, 2-iodobiphenyl from β-H elimination) but also from homolytic or heterolytic decomposition pathways of the diaryliodonium salt (2-iodobiphenyl, 2,2'-diiodobiphenyl, and 2-(2,5-dimethylphenyl)-2'-iodobiphenyl) [29][30][31]. To further verify these observations, we reacted 1 in the presence of aniline (2a) for three days at elevated temperature without adding a Pd catalyst. Again, after GC-MS analysis, we could detect 2-iodobiphenyl, 2,2'-diiodobiphenyl, 2-(2,5-dimethylphenyl)-2'-iodobiphenyl, and 2'-iodo-N-phenylbiphenyl-2-amine in the absence of any produced 3a. Contrary to that observation, when we conducted an analogous experiment without aniline, we observed no decomposition or byproduct formation. These results led us to conclude that byproduct formation is, at least partially, induced through the nucleophilic and/or basic nature of aniline [26,[32][33][34][35]. We therefore had to accept that a significant amount of byproducts are formed during the formation of 3a, reducing our isolated yield.
Furthermore, meta-substituted anilines 2l and 2m were tested, giving the N-arylated carbazoles 3l and 3m in good yields of 64% and 61%, respectively. In general, the use of fluorinesubstituted anilines showed the best results so far in this study. However, with the perfluorinated derivative 2p, the isolated yield of 3p was diminished to 37%. Furthermore, we used our protocol to synthesize the N-arylcarbazole based electronic materials 3n and 3o in 40% (3n) and 23% (3o) yield.
After an extensive exploration of the reaction conditions and the substrate scope with iodonium salt 1, we wanted to compare our results with the corresponding bromonium analogue. Cyclic diarylbromonium salts are considerably less explored than their iodonium congeners as can be seen by only a handful of synthetic methods described in the literature [32,[36][37][38][39][40][41]. In general, bromonium salts are more reactive but have similar reaction behaviour [32,36]. Thus they could be helpful substrates for the synthesis of N-arylcarbazoles from anilines. With this in mind, we initially focussed on the synthesis of dibenzo[b,d]bromolium chloride (5) using a procedure published by Sandin and Hay in 1952 [41] (Scheme 3).
The biphenyl derivative 4*HCl was prepared by Suzuki coupling of 2-iodoaniline and 2-bromophenylboronic acid. Diazotation of 4*HCl and cyclization gave the cyclic diarylbromonium chloride 5 as an off-white powder in good isolated  (6) is indeed a more reactive surrogate for the construction of N-arylated carbazoles, we reacted 6 with p-fluoroaniline (2f) according to our previously described optimized conditions (Scheme 4).
However, 3f was obtained in only 25% yield (Scheme 4), compared to 71% when using the corresponding diaryliodonium salt 1 (Figure 3). Apart from the desired product, we were able to isolate two byproducts from the crude reaction mixture. After a systematic structure determination by one-and twodimensional NMR techniques as well as mass spectrometry, we elucidated the two byproducts as the two regioisomers 2'-bromo-N-(4-fluorophenyl)biphenyl-2-amine (7a, 12%) and 2'-bromo-N-(4-fluorophenyl)biphenyl-3-amine (7b, 18%) (Scheme 4). Compound 7a was either formed during the catalytic cycle as a reaction intermediate, which had not reacted further to the final product 3f, or is the result of a nucleophilic attack, caused by the nucleophilic nature of aniline 2f at the electrophilic ipsoposition in 6. The formation of the other regioisomer 7b, is not evident at first glance. One plausible explanation could be the emergence of a benzyne intermediate during synthesis, generated by β-elimination using aniline as a base. Subsequent nucleophilic trapping of the benzyne with aniline, this time reacting as a nucleophile, results in the formation of 7b. A very similar reactivity was described recently for a nitro-substituted diarylbromonium salt [42]. However, the results of these experiments demonstrate, that the higher reactivity of diarylbromonium salts towards nucleophilic ring opening is accompanied, to a significant degree, by an undesired β-elimination pathway, leading to more complex reaction mixtures and subsequently lower yields of the desired N-arylcarbazole.

Conclusion
In summary, we have developed a novel synthesis of synthetically highly useful N-arylcarbazoles starting from cyclic diaryliodonium salts by a ring opening/Buchwald-amination cascade using anilines and aliphatic amines as nitrogencontaining substrates. With 5 mol % of Pd(OAc) 2 the desired N-arylcarbazoles could be isolated in up to 71% yield. Finally, the corresponding cyclic diarylbromonium derivatives were tested in the same reaction. Significantly lower yields were observed due to undesired side reactions involving benzyne intermediates by β-elimination.

Supporting Information
Supporting Information File 1 Experimental procedures and data of characterization of the described compounds.