Atroposelective Synthesis of Azobenzenes by Palladium-Catalyzed Cross-Coupling of Racemic Biaryl Triflates and Diazenyl Pronucleophiles

Abstract

A kinetic resolution of racemic biaryl monotriflates by an atroposelective palladium-catalyzed diazenylation enables the synthesis of azobenzene derivatives decorated with an axially chiral substituent. The C­(sp²)–N­(sp²) cross-coupling reaction makes use of silylated diazenes as diazenyl anion equivalents, and chiral ferrocene-based bisphosphine ligands act as effective supporting ligands. The resulting chiral azobenzenes bearing binaphthyl and naphthyl/phenyl backbones undergo reversible trans–cis isomerization under irradiation with LEDs of different wavelengths, highlighting their potential use in photoresponsive chiroptical materials.

Compared to axially chiral C- and O-functionalized biaryls, the synthesis of N-functionalized biaryls is considerably less explored. To date, five main strategies have been established for the construction of axially chiral biaryl amines and their derivatives. Aside from asymmetric cross-coupling to forge aryl–aryl bonds and annulation approaches, modification of existing biaryl amine structures as part of a kinetic resolution (KR) provides an alternative access. Of note, Fernández, Lassaletta, and co-workers introduced an enantioselective Buchwald–Hartwig cross-coupling in the form of a kinetic asymmetric transformation (DYKAT) (Scheme A, left). The fifth approach is based on desymmetrization. , Notably, Gu and co-workers reported a highly atroposelective ring-opening amination of achiral cyclic diaryliodonium salts (Scheme A, middle). Another typical example of enantioselective C­(sp²)–N­(sp³) bond formation is a desymmetrization of 1,1′-biaryl-2,6-dibromides by Cong and co-workers (Scheme A, right).

1. Catalytic Asymmetric C­(sp²)–N­(sp³) Cross-Coupling for the Atroposelective Formation of Biaryl Amines and Planned C­(sp²)–N­(sp²) Cross-Coupling for the Synthesis of Chiral Azobenzenes .

^a dppf = 1,1′-bis­(diphenylphosphino)­ferrocene; Si = triorganosilyl.

A few years ago, our laboratory had begun to investigate diverse reactivity of silicon-masked aryl-substituted diazenes. In 2022, we introduced an unprecedented palladium catalysis where silylated diazenes serve as precursors of diazenyl anions (Scheme B, top), otherwise fleeting intermediates prone to rapid loss of dinitrogen. This rare palladium-catalyzed C­(sp²)–N­(sp²) cross-coupling allowed for the direct installation of an azo unit at a prefunctionalized arene, thereby enabling the synthesis of nonsymmetric azobenzenes. An asymmetric variant of this reaction is not known, but it would be appealing because it could grant access to axially chiral azobenzenes. This class of photoresponsive molecules with their unique physicochemical properties such as reversible trans–cis isomerization have been shown to be particularly attractive for the design of chiroptical materials. Their synthesis typically starts from enantioenriched BINAM and BINOL (for axially chiral systems) or is done by chromatographic separation on chiral stationary phases (for planarly chiral systems). To the best of our knowledge, there have been no reports on the asymmetric synthesis of these axially chiral azobenzenes starting from racemic or prochiral substrates. We hypothesized that racemic biaryl monotriflates could engage in a KR process in the presence of a chiral palladium catalyst (Scheme B, bottom).

Our study commenced with 2-(trifluoromethanesulfonyloxy)-1,1′-binaphthyl (rac-1a) and N-phenyl-N′-trimethylsilyldiazene (2a) as model substrates (Table ; for variation of the substitution pattern at the silicon atom, see Table S2). Guided by our previous work, where dppf had been found to be crucial for suppressing dinitrogen extrusion, we decided to begin with chiral ferrocene-type bisphosphine ligands. We optimized the reaction conditions employing Pd­(OAc)₂ as the precatalyst and CsOAc as the base in toluene as the solvent. Initially, a series of ferrocene-based ligands L1–L5 was examined. Using L1, we achieved the formation of the desired axially chiral azobenzene product (S)-3aa in a good yield of 53% with a promising selectivity factor (s) of 20 (entry 1). In contrast, the 2,5-diisopropyl-substituted ligand L2 resulted in both a low yield and low selectivity (entry 2). Next, we investigated structurally similar 2,4-disubstituted FerroTANE ligands L3–L5. For ligands bearing small R groups (Me and Et), the yield and selectivity remained low (entries 3 and 4). The 2,4-diisopropyl-substituted ligand L5 provided a substantially higher selectivity but lower yield (entry 5). The reaction also proceeded in the absence of a base, albeit with a decreased yield (entries 6 and 7). Additional ferrocene-based Josiphos-type ligands did not lead to any improvement (see the Supporting Information for details). Conducting the reaction at room temperature led to a lower yield of (S)-3aa but an improved s factor (entry 8). K₂CO₃ instead of CsOAc did improve the yield of 3aa and the selectivity, making it the base of choice (entry 9). Given that L5 exhibited a high s factor, we also probed various bases using L5 at room temperature over prolonged reaction times (entries 10–13). To our delight, the s factors of this reaction were close to 40 when CsOAc, KOAc, or K₂CO₃ was used as the base. Considering the high hygroscopicity of CsOAc and KOAc, which may lead to reproducibility issues, we opted for K₂CO₃ as the base. Increasing the amount of K₂CO₃ from 0.3 to 1.0 equiv resulted in a 42% NMR yield of (S)-3aa with s = 40 (entry 14). The absolute configuration was assigned as S for 3aa by comparison of the HPLC traces of the recovered binaphthyl triflate 1a with independently prepared samples of (R)-1a and (S)-1a (see the Supporting Information for details).

1. Optimization of the Reaction Conditions .

entry	ligand	T (°C)	t (h)	base	yield of 3aa (%)	ee of 3aa (%)	ee of 1a (%)	conv. (%)	s
1​	L1	60	4	CsOAc	53	72	92	56	20
2​	L2	60	24	CsOAc	8	47	19	29	3
3​	L3	60	24	CsOAc	15	54	14	21	4
4​	L4	60	24	CsOAc	26	78	34	30	11
5​	L5	60	24	CsOAc	34	82	53	39	17
6​	L1	60	4	none	41	80	63	44	17
7​	L5	60	4	none	19	90	32	26	26
8​	L1	25	60	CsOAc	28	78	88	53	23
9​	L1	25	48	K2CO3	39	85	74	47	27
10​	L5	25	60	CsOAc	33	89	63	41	33
11​	L5	25	72	KOAc	31	92	50	35	39
12​	L5	25	72	Cs2CO3	trace
13​	L5	25	72	K2CO3	21	93	34	26	38
14​	L5	25	72	K2CO3	42	89	77	46	40

^aReactions were performed on a 0.10 mmol scale at a concentration of 0.50 M under an argon atmosphere.

^bYields were determined by ¹H NMR spectroscopy using CH₂Br₂ as an internal standard.

^cThe ee values were determined by HPLC analysis on a chiral stationary phase.

^dCalculated conversion: C = ee_SM/(ee_SM + ee_PR), where ee_SM is the ee of recovered 1a and ee_PR is the ee of 3aa.

^e s = ln­[(1 – C)­(1 – ee_SM)]/ln­[(1 – C)­(1 + ee_SM)].

^f1.0 equiv of K₂CO₃.

^gThe isolated yield was 37% after flash chromatography on silica gel.

With the optimized reaction conditions in hand, we explored the substrate scope of this atroposelective cross-coupling (Scheme ). The coupling reactions of silyldiazenes substituted with halogen atoms provided the corresponding axially chiral azobenzenes (S)-3ab, (S)-3ac, and (S)-3ad in good to excellent yields with excellent enantioselectivity. Electron-donating substituents at the para position were also compatible, and the desired products (S)-3ae with a methyl and (S)-3af with a methoxy substituent were obtained in good yields with good s factors. A lower reaction rate was observed for the silyldiazene bearing an electron-withdrawing ester group, and even after extending the reaction time to 7 days, no increase in yield of (S)-3ag was observed. Next, we gauged the substitution pattern of the biaryl triflate electrophile. Both 6-methoxy- and 6-trimethylsilyl-substituted binaphthyl triflates were well-tolerated using L5 as the ligand, which led to higher reactivity than L1, affording products with s factors of 63 for (S)-3bb and 38 for (S)-3ca, respectively. Substitution at the ortho position with a methyl group resulted in good to excellent enantioselectivity and s factors for products (S)-3da and (S)-3db at an elevated reaction temperature of 60 °C. We attribute this loss of reactivity to steric hindrance. Although the yield was lower, substitution at the ortho position proved to be beneficial for enantioselectivity, and hence, an additional o-fluoro substituent was tested. The yield remained moderate when the reaction time was extended to 7 days, presumably due to the electron deficiency, but the s factor was still good for (S)-3ea. We continued to expand the substrate scope by modifying the other aryl moiety of the electrophile. Replacing the naphthyl group with a phenanthryl substituent as in (S)-3fe resulted in a good yield and a good s factor. The naphthyl group was also successfully replaced by ortho-substituted phenyl groups, which coupled effectively with silyldiazenes. Substitution at the ortho position with methoxy as in (S)-3ga and (S)-3gb, ethyl as in (S)-3ha, or methyl as in (S)-3ia and (S)-3ja gave good yields and selectivities. A halogen substituent was also well-tolerated, affording (S)-3ka in good yield with a moderate s factor. An o-phenyl substituent led to a moderate yield and selectivity for (S)-3la. Finally, a significant decrease in enantioselectivity was observed when the triflyloxy-substituted naphthyl moiety was exchanged for a phenyl group as in 3ma and 3na (gray box). Typically, the reaction predominantly yields the trans isomer, with most trans:cis ratios exceeding 92:8 (deviating trans:cis ratios: 89:11 for 3ae, 88:12 for 3ca, and 85:15 for 3ea). A minor amount of the cis isomer can be observed in the NMR spectrum.

2. Substrate Scope of the Atroposelective C­(sp²)–N­(sp²) Cross-Coupling .

^a All reactions were performed on a 0.10 mmol scale at a concentration of 0.50 M under an argon atmosphere. Isolated yields are reported; ee values were determined by HPLC analysis on a chiral stationary phase. Calculated conversion: C = ee_SM/(ee_SM + ee_PR), where ee_SM is the ee of recovered 1 and ee_PR is the ee of 3. s = ln­[(1 – C)­(1 – ee_SM)]/ln­[(1 – C)­(1 + ee_SM)].

^b L1 as the ligand.

^c L5 as the ligand.

^d A 10 mol % loading of Pd­(OAc)₂ and a 12 mol % loading of the indicated ligand were used.

^e Not isolated.

^f The ee value after recrystallization from n-hexane.

To investigate the kinetic resolution process, enantiopure (S)-1a and (R)-1a were individually subjected to the standard conditions using L5 as a ligand (Scheme ). When (S)-1a was used, the desired product (S)-3aa was obtained in 47% yield (eq 1). In contrast, only a very low yield of (R)-3aa was observed with (R)-1a (eq 2). In both cases, no racemization of the recovered triflate 1a was detected. These results suggest that ligand L5 matches with the R configuration of the substrate, thereby facilitating efficient enantiomer discrimination in the KR.

3. Match/Mismatch Control Experiments .

^a Reactions were performed on a 0.10 mmol scale under the standard conditions. Isolated yields after flash chromatography on silica gel are reported.

The products trans-3 obtained from our atroposelective cross-coupling reaction are both chemically and thermally stable and exhibit the expected reversible photoswitchable behavior in organic solvents (Scheme A, left). To evaluate the photoresponsiveness, we irradiated the trans-configured model compound trans-(S)-3aa (trans:cis = 91:9) with 390 nm light for 30 min in C₆D₆, successfully obtaining the corresponding cis isomer cis-(S)-3aa (trans:cis = 18:82); the reversible, photoinduced trans-to-cis interconversion was clearly seen by NMR spectroscopy (see the Supporting Information for details). Unexpectedly, when the same experiment with trans-(S)-3aa (trans:cis = 95:5) was repeated in CDCl₃ for 40 min, N-phenyl-7H-dibenzo­[c,g]­carbazol-7-amine (4aa) was obtained quantitatively (Scheme A, right). NMR monitoring suggests that the reaction proceeds via an initial trans-to-cis isomerization followed by a proton-initiated intramolecular electrophilic aromatic substitution. We favor this ionic mechanism over a radical pathway also because the addition of K₂CO₃ suppresses the cyclization event. Moreover, we recorded the UV–vis absorption spectra of several chiral azobenzenes: trans-(S)-3ab, (S)-3ae, (S)-3fe, and (S)-3ia (Scheme B). These compounds exhibit similar spectral profiles, and the absorption maxima are at around 337, 343, 341, and 334 nm, respectively. Irradiation at 390 nm induced a rapid photoisomerization of trans-(S)-3aa to cis-(S)-3aa in toluene, reaching a photostationary state within approximately 5 s (Scheme C). Compared with trans-(S)-3aa, cis-(S)-3aa shows decreased absorbance at 333 nm and increased absorbance at 453 nm. The presence of an isosbestic point at 422 nm, where all absorption curves intersect, indicates a clean interconversion in toluene between two species without the accumulation of other intermediates. These results show that these axially chiral benzenes possess good photoresponsiveness. Additionally, most of them exhibit very high specific rotation values (maximum around 600; see the Supporting Information for details), suggesting their potential applications in chiral sensing and optoelectronic materials.

4. Photophysical Properties.

^a The reactions were performed in an NMR tube at a concentration of 0.056 M.

^b UV–vis absorption spectra of various products (S)-3 (in toluene, 100 μmol).

^c UV–vis absorption spectra of 3aa (in toluene, 100 μmol) after UV irradiation at 390 nm for 0 s, 0.5 s, 1 s, 2 s, 5 s, 10 s, 30 s, and 1 min.

In summary, we developed an atroposelective cross-coupling between silylated diazenes and biaryl monotriflates for the direct formation of C­(sp²)–N­(sp²) bonds. This kinetic resolution exhibits a synthetically useful selectivity factors. The approach enables the asymmetric synthesis of axially chiral azobenzenes from racemic substrates using palladium with a chiral bisphosphine ligand as a catalytic system. The reaction displays broad scope with good functional group tolerance, and reversible trans-to-cis isomerization of these chiral azobenzene derivatives indicates potential applications in materials science.

The data underlying this study are available in the published article and its Supporting Information.

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/jacs.5c09097.

• Experimental procedures, including the synthesis of starting materials and characterization data, and NMR spectral data for all new compounds (PDF)

The authors declare no competing financial interest.