Starazo triple switches – synthesis of unsymmetrical 1,3,5-tris(arylazo)benzenes

Multistate switches allow to drastically increase the information storage capacity and complexity of smart materials. In this context, unsymmetrical 1,3,5-tris(arylazo)benzenes – ‘starazos’ – which merge three photoswitches on one benzene ring, were successfully prepared. Two different synthetic strategies, one based on Baeyer–Mills reactions and the other based on Pd-catalyzed coupling reactions of arylhydrazides and aryl halides, followed by oxidation, were investigated. The Pd-catalyzed route efficiently led to the target compounds, unsymmetrical tris(arylazo)benzenes. These triple switches were preliminarily characterized in terms of their isomerization behavior using UV–vis and 1H NMR spectroscopy. The efficient synthesis of this new class of unsymmetrical tris(arylazo)benzenes opened new avenues to novel multistate switching materials.


Introduction
The reversible photochemically induced structural change of azobenzenes (ABs) opens various ways towards systems manipulations on the molecular level [1,2]. Upon irradiation with UV light (ca. 350 nm), the thermodynamically more stable (E)-AB isomerizes to the higher-energy Z-isomer [3]. The isomerization can be reversed either by irradiation with visible light (ca. 450 nm) or thermally, upon heating. The thermal back-isomerization rate can be controlled by various factors: Functional group substitutions on the phenyl rings determine the thermal half-lives of (Z)-ABs [1]. For example, o-fluoro substitution prolongs the thermal stability up to years [4], while electron-donating groups attached to the different phenyl rings decrease the thermal half-lives below the time scale of seconds. Even subtle interactions, such as London dispersion in alkyl-substituted ABs, can have a significant influence on their isomerization properties [3,4]. Also, the incorporation of AB units into cyclic [5] or macrocyclic structures can control the switching, depending, i.a., on symmetry and ring strain [6][7][8][9]. By combining these approaches, half-lives can be tuned from milliseconds to years.  [11,12] and present work.
The incorporation of multiple AB units into one molecule allows to access multiple states upon isomerization, which dramatically increases the potential for information storage using this photoswitch. Ideally, one molecule can be treated with multiple different inputs, leading to defined, detectable outputs. An example of such compounds are 1,3,5tris(arylazo)benzenes -'starazos' -introduced by Cho and co-workers in 2004 ( Figure 1) [10]. Despite their successful synthesis, using Pd-catalyzed coupling reactions of aryl halides and arylhydrazides [11] followed by Cu(I)-mediated oxidation, the photochemical properties of such compounds have not been studied yet. Going one step further, this type of compounds could be substituted in an unsymmetrical way with different azo units, which allowed individual switching using light of different wavelengths [12]. These compounds were then investigated theoretically by Dreuw and co-workers, who suggested tris(arylazo)benzene 2 ( Figure 1) to feature spectrally separated absorption bands for each AB branch. The authors could show that the excited states of the AB branches in tris(arylazo)benzenes were electronically decoupled, despite the spatial overlap. Hence, only four phenyl rings were sufficient for the construction of three individually photoisomerizable azo units.
Motivated by the promising theoretical results by Dreuw and co-workers [12], an efficient synthetic strategy for asymmetric tris(arylazo)benzenes 3 was developed. With an efficient preparative access, this highly interesting class of multistate photoswitches would be accessible for detailed spectroscopic investigations, leading to fundamental insights applicable to the design of intelligent photoresponsive materials.

Results and Discussion
Initially, two different synthetic strategies were evaluated for the preparation of tris(arylazo)benzenes 3. The first one relied on the consecutive condensation of anilines with nitroso compounds, Baeyer-Mills reactions [13]. With a suitable protecting group strategy, the selective construction of the individual AB branches should be achievable (Scheme 1). Starting from 3,5disubstituted nitrosobenzene E, consecutive Bayer-Mills and deprotection reactions would lead to the target compound 3 in five steps. Furthermore, by using nitrosoarene E, the selective installment of the amine groups, e.g., via acetamide-and nitro group-carrying intermediates, could be achieved [14][15][16]. Such a strategy had already been used successfully in previous syntheses of o-, p-, and m-bis(arylazo)benzenes in our laboratory [14]. However, multiple protection/deprotection steps lowered the atom economy and increased the step count of this strategy.
The second approach towards tris(arylazo)benzenes 3 relied on Pd-catalyzed coupling reactions and Cu(I) oxidation, as presented by Cho and co-workers (Scheme 2) [10]. This route offers the advantage that the preparation of a wide variety of N-(tert-butoxycarbonyl)phenylhydrazides has already been reported [17][18][19]. Additionally, starting from easily accessible 3,5-dibromoazobenzenes H, only two consecutive coupling reactions, followed by oxidation, would be required to obtain the target starazo 3. However, selectivity might be problematic in the first coupling reaction, which would lead to lower yields of the desired monocoupled intermediates.
the Bayer-Mills route was followed. The azobenzene building block 8, with an unsubstituted phenyl ring and two orthogonal nitrogen substituents in the 3-and 5-position (Scheme 3), was prepared. The synthesis commenced with the acetylation of 3,5dinitroaniline (4) in 95% yield, followed by the selective reduction of one nitro group using an aqueous ammonium sulfide solution to furnish aniline 6 in 65% yield [20]. After oxidation of 6 to its nitroso analogue 7 [21], a Baeyer-Mills reaction with aniline yielded the targeted azobenzene building block 8 in 87% yield (i.e., 53% yield over four steps).
After the successful synthesis of (E)-N-(3-nitro-5-(phenyldiazenyl)phenyl)acetamide (8), the reduction of the nitro group was attempted. However, the literature-known procedure of using sodium hydrosulfide did not yield the desired (E)-N-(3amino-5-(phenyldiazenyl)phenyl)acetamide (9). Alternative reduction attempts, such as using Pd/C-catalyzed reduction by H 2 in different solvents, SnCl 2 , or iron under acidic conditions were not successful either. In all cases, the reactions produced complicated mixtures, and the target compound could neither be identified nor isolated. At this stage, the second strategy via Pd-catalyzed coupling reaction was assayed [10,11].
Nevertheless, both AB 15 as well as the arylhydrazides could be prepared on a multigram scale, which allowed to access the desired monocoupling products 16a/16b on a scale of several hundred milligrams. Their 1 H NMR spectra indicated the presence of mixtures of several isomers, not only E-and Z-configuration of the AB, but also different configurations of the Boc groups. However, the coupling products could be unequivocally identified by high-resolution mass spectrometry. Next, 16a/16b were coupled with N-Boc-N-phenylhydrazine to afford the corresponding tris(arylazo)benzene precursors 17a in 23% and 17b in 52% yield, respectively (Scheme 5). Again, like for Scheme 6: Coupling reactions of 15 with the corresponding arylhydrazides to access monocoupled (16a/16b) and biscoupled compounds (17c/17d) as major products.
the first coupling reaction, it was not possible to isolate the products as uniform isomers using chromatographic methods. Hence, the tris(arylazo)benzene precursors 17a/17b were used without further purification.
Initially, a coupling reaction of N-Boc-N-phenylhydrazine with AB 15 was attempted. However, with this inversed reaction sequence, inseparable mixtures of the desired coupling product and the AB starting material 15 were obtained. Hence, coupling of the electron-poor and electron-rich arylhydrazides was performed first to allow the isolation of 16a/16b by chromatographic methods.
Finally, the target tris(arylazo)benzenes 3a/3b were obtained by oxidation of the precursors 17a/17b using CuI under basic conditions in DMF at elevated temperature (Scheme 5). For the cyano-substituted derivative 3a, a yield of 20% was achieved, whereas the methoxy derivative 3b could be isolated in a good yield of 59%.
Having both tris(arylazo)benzenes 3a/3b in hand, UV-vis spectroscopy was used for preliminary investigations on the photophysical properties of the molecular triple switches 3. Both tris(arylazo)benzenes 3a and 3b showed characteristic UV-vis spectra of ABs with strong π-π* absorption, having their absorption maxima at 337 nm (3a) and 349 nm (3b), respectively (Figure 2a). In contrast to AB, the π-π* and n-π* bands of both compounds overlapped, forming shoulders at ca. 450 nm. The samples were irradiated at 365 nm with a high-power LED (see Supporting Information File 1 for specifications) to induce E-to-Z photoisomerization.
For the cyano-substituted tris(arylazo)benzene species 3a, the π→π * absorption band decreased only to a relatively small extent, while for starazo 3b, the expected switching behavior was more pronounced. Furthermore, the irradiation time for reaching the photostationary state (PSS) of 3a was significantly longer compared to 3b or other AB derivatives. Afterwards, the solutions were irradiated with light of 448 nm to photochemically induce Z-to-E isomerization. In both cases, spectra of PSSs with slightly higher E/Z ratio could be reached (Figure 2b and Figure 2c). This is consistent with the fact that the tris(arylazo)benzenes 3a/3b were isolated as mixtures of the all-E-isomers including small amounts of other photoisomers after the synthesis. Furthermore, the all-E states could be reached after only 20 and 7 s of irradiation, respectively, which was significantly faster than the E→Z photoisomerization. All in all, the irradiation experiments revealed that the methoxy-substituted derivative 3b shows reversible photoisomerization, while cyano-substituted tris(arylazo)benzene 3a could only be marginally isomerized.
To get deeper insight into the photoisomerization, 1 H NMR spectroscopy was applied to monitor the isomerization process. For both 3a and 3b, complex spectra were obtained after irradiation at 365 nm (see Figure S2 and Figure S3, Supporting Information File 1). Hence, a high number of photoisomers was generated, which indicated that selective photoisomerization was not possible in both cases under the applied conditions. After irradiating the samples with light of 448 nm wavelength and keeping them in the dark at room temperature, the initial spectra could be restored.

Conclusion
In summary, 1,3,5-tris(arylazo)benzenes 3a/3b were successfully synthesized using Pd-catalyzed coupling reactions of arylhydrazides and aryl bromides followed by CuI-mediated oxidation as key steps. Although the low yield of the first coupling step, due to the preference for double coupling, was a drawback of this strategy, enough material could be prepared via this method to characterize and investigate the isomerization properties of the triple photoswitches 3a/3b. Changing the N-Boc-Nphenylhydrazides used herein to different derivatives opened the possibility to synthesize a wide variety of unsymmetrical starazo species, starting from suitable dibromoazobenzenes.
A preliminary investigation of the presented three-state switches by UV-vis and 1 H NMR spectroscopy revealed that both derivatives 3a and 3b were capable of E→Z photoisomerization. However, no selective photoswitching could be achieved due to overlapping absorption bands of all arylazobenzene moieties (see Figure 2). Furthermore, analyses showed that starazo compound 3b had a higher E/Z ratio in the PSS compared to derivative 3a. In addition, complex spectra were obtained, in which the individual isomers could not be assigned unambiguously. Although the starazo species that were prepared did not show selective switching, the presented synthesis opened an easy access to other analogous to further study the fundamental properties of 1,3,5-tris(arylazo)benzenes in the near future. These new insights will foster the design of novel, multi-state photoresponsive systems for smart materials. [24]: 3,5-Dibromoaniline (9.99 g, 39.8 mmol, 1.00 equiv) was suspended in water (20 mL) and aq HBF 4 (50%, 15 mL, 119 mmol, 3.0 equiv) was added. The suspension was cooled to 0 °C and a solution of NaNO 2 (2.76 g, 40.0 mmol, 1.00 equiv) in water (8 mL) was added dropwise. The grey suspension was vigorously stirred at 0 °C for 45 min. The precipitate was filtered off, washed with Et 2 O (5 × 50 mL), and dried in vacuum to yield the diazonium tetrafluoroborate as grey solid (11.2 g). To a suspension of this diazonium tetrafluoroborate (11.2 g, 32.0 mmol, 1.00 equiv) in MeOH (80 mL), a solution of 1,3,5trimethoxybenzene (5.98 g, 35.6 mmol, 1.11 equiv) in MeOH (80 mL) was added dropwise at rt over 15 min. After standing overnight at −20 °C, the orange precipitate was filtered off, washed with cold MeOH (ca. 25 mL), and dried in vacuum to yield 15 as orange powder (13.8 g, 80%). mp 213-215 °C (dec); 1 H NMR (400 MHz, DMSO) δ 7.91 (t, J = 1.8 Hz, 1H), 7.80 (d, J = 1.8 Hz, 2H), 6.38 (s, 2H), 3.89 (s, 3H), 3.83 (s, 6H); 13