Introduction

The charge-transfer theory was formulated in 1952 by Robert Mulliken to rationalize the appearance of strong color on bringing together two colorless organic compounds1. This theory explains how the association of an electron-rich substrate (a donor D) and an electron-accepting molecule (an acceptor A) can induce the formation of a molecular aggregation in the ground state. This new intermediate, termed electron donor–acceptor (EDA) complex (Fig. 1a)2, has physical properties that differ from those of the separated substrates. For example, its formation is generally accompanied by the appearance of a new absorption band, the charge-transfer band (hνCT), which is associated with a transfer of a single electron (SET) from the donor to the acceptor. Often, the energy of this transition lies in the visible frequency range.

Fig. 1
figure 1

The EDA complex activation strategy for photochemical synthetic applications. a The principles of EDA complex activation and synthetic use enabled by light. b In situ generated catalytic intermediates can form an intermolecular EDA complex with a substrate. c A catalyst, adorned with an electron-rich unit, reacts with a substrate to generate an electron-poor intermediate prone to intramolecular EDA complex formation. KEDA association constant of EDA complex formation, D electron-rich molecule, A electron-poor molecule, SET single-electron transfer

Full size image

The photo-physics of EDA complexes have been extensively studied since the 1950s3,4,5, while their use in chemical synthesis was exploited in a more intensive way only 20 years later6,7,8,9. Recently, the resurgence of visible light-driven processes has motivated chemists to reinvestigate the potential of EDA complex activation for promoting photochemical processes10,11,12,13,14,15,16,17,18,19,20,21,22,23,24. For example, our laboratories applied this strategy for enantioselective catalysis25,26,27,28. Specifically, we exploited the ability of a chiral catalyst to interact with a weakly polarized substrate (e.g. aldehydes and β-ketoesters) and transform it into an electron-rich chiral organocatalytic intermediate (such as enamines25,26,27 or enolates28). This enabled the aggregation with an electron-poor reagent to form a colored EDA complex, whose photoactivity provided access to reactive open-shell intermediates (Fig. 1b). At the same time, the chiral organocatalytic intermediate ensured effective stereochemical control over the ensuing radical bond-forming process. In general, all the synthetic methods reported so far6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28 relied on the excitation of intermolecular EDA complexes formed upon aggregation of two substrates/intermediates.

Here, we further expand the synthetic potential of this activation strategy demonstrating the possibility to form an intramolecular EDA complex and using its photochemical activity to drive a stereoselective radical reaction (Fig. 1c). Specifically, we used a chiral catalyst to generate an electron-poor intermediate upon condensation with a weakly polarized substrate. To induce the formation of an intramolecular EDA aggregation29,30,31,32, the catalyst was adorned with an electron-rich moiety. The resulting intramolecular charge transfer π–π interaction elicited a broad absorption band in the visible region. Irradiation with visible light initiated a stereoselective radical process. To our knowledge, this study offers the first demonstration of the potential of photon-absorbing intramolecular EDA complexes in synthetic applications.

Results

Background

The present study was motivated by our interest in developing enantioselective radical conjugate additions33,34,35,36 via iminium ion activation37. This organocatalytic activation mechanism has found many applications for facilitating conjugate additions in ionic domains. In contrast, it was difficult to develop a stereoselective iminium ion trap of open-shell species. Recently, our laboratories reported a strategy, based on the combined action of a chiral organic catalyst 3, bearing a redox-active carbazole moiety, and a photoredox38 catalyst (PC), that enabled enantioselective radical conjugate additions to β-substituted cyclic enones 1 (Fig. 2a)39,40. During these studies, we isolated stable tetrafluoroborate salts of the chiral iminium ion A-1, generated upon condensation of a catalyst of type 3 (bearing a 3,6-di-tert-butyl-carbazole moiety) and an aliphatic enone (R′ = Me in 1, Fig. 2b), which were characterized by X-ray single-crystal analysis. Surprisingly, these crystals showed an intense bright-yellow coloration (see UV–Vis spectrum in Fig. 2c). Typically, aliphatic iminium ions of type A can only absorb in the UV region41,42 (below 350 nm, see for example, the absorption spectrum of iminium ion A-2 in Fig. 2c).

Fig. 2
figure 2

Background and initial insights. a Original strategy for the enantioselective iminium ion trapping of radicals based on the combination of a carbazole amine catalyst 3 and an exogenous photoredox catalyst (PC). b X-ray crystal structure of the carbazole-based iminium ion A-1. c UV–Vis spectra of the carbazole catalyst 3b (structure in Table 1) and iminium ions A-1 and A-2 [1.0 mM] in CH3CN

Full size image

The X-ray structure of A-1 provided a rationalization for its optical properties. The broad absorption band in the visible region is induced by a stabilizing intramolecular charge-transfer π–π interaction between the electron-deficient iminium ion double bond and the electron-rich carbazole nucleus. This type of intramolecular EDA29,30,31,32 association generally results in a considerable redshift of the absorption bands. The intramolecular EDA complex also accounted for the interatomic separation between the carbazole nitrogen and the sp2 α-carbon of the iminium ion (3.10 Å), which was significantly shorter than the van der Waals distance.

Given our interest in the photochemical activity and the synthetic utility of intermolecular EDA complexes generated from chiral organocatalytic intermediates25,26,27,28,43, we wondered if the visible-light-absorbing properties of the iminium ion-based EDA complex A could be harnessed to design an enantioselective light-triggered process44 without the need for an external photoredox catalyst.

Design plan

Figure 3 details our strategy for exploiting the photoactivity of the intramolecular iminium-ion-based EDA complex A, which is transiently generated in the ground state upon condensation of the carbazole aminocatalyst 3 with cyclic enone 1.

Fig. 3
figure 3

Mechanistic proposal. Central to this study is the ability of the carbazole radical cation within B, generated upon visible-light excitation of the intramolecular EDA complex A, to drive the generation of radicals acting as an SET oxidant. SET single-electron transfer, BET back-electron transfer, Solv solvent, TMS trimethylsilyl

Full size image

Visible-light excitation of A would trigger an intracomplex SET from the electron-rich carbazole (the donor) to the iminium double bond (the acceptor), furnishing the chiral radical intermediate B. Using the photoexcitation of EDA complexes in chemical synthesis is not simple, mainly because of the unproductive, fast back-electron transfer (BET) which restores the ground-state EDA complex, making further reactivity difficult. Therefore, a central element of our approach was to identify a suitable mechanism to interrupt the BET process45 between B and A. We envisioned that the long-lived carbazole radical cation in B, which is known to be a persistent species46, could act as an effective oxidant. Specifically, we hoped that an SET would occur from an easily oxidazible electron-rich alkyl trimethylsilane 4 to B, furnishing the neutral 5π-electron β-enaminyl radical intermediate C and thus precluding the BET. This SET manifold would also generate the silyl radical cation D. The latter intermediate can readily undergo rapid desilylation in the presence of weak nucleophiles, including water47,48, to form a carbon radical, which can then enter a radical conjugated addition manifold39. Importantly, after the stereoselective iminium ion radical trap, the electron-rich carbazole moiety would be positioned at a strategic position to promote a rapid intramolecular SET reduction of the unstable α-iminyl radical cation E49. This regenerates the carbazole radical cation within F, which would be able to oxidize a new molecule of the organic silane 4. By regenerating the radical, this SET event would be the propagation step of a radical chain pathway. Hydrolysis of the neutral imine intermediate G would turn over catalyst 3 while providing product 2 bearing a quaternary carbon stereocenter.

Optimization of the model reaction

To validate our photochemical plan, we selected β-methyl cyclohexenone 1a as the model substrate, while using the carbazole catalysts 3 to promote the formation of the chiral iminium ion A (Table 1). We performed the experiments in CH3CN at 35 °C under irradiation by a single high-power (HP) LED (λmax = 420 nm) and using an irradiance of 15 mW/cm2, which was controlled by an external power supply (full details of the illumination set-up are reported in Supplementary Figure 1). Water (2 equiv) was added to facilitate the generation of radicals upon desilylation of intermediate of type D. As radical precursors, we selected a pool of organic silanes 4a–c with different oxidation potentials (Eox (4·+/4), as measured by cyclic voltammetry versus Ag/Ag+ in CH3CN). In consonance with the mechanistic requirement that radicals originate upon SET oxidation from the carbazole radical cation in B, only substrates 4 with comparable or lower oxidation potential than the catalyst carbazole unit should be suitable for reaction. This was the exact reactivity trend observed in the photochemical radical conjugate addition promoted by catalyst 3a (entries 1–3). Given the oxidizing power of the carbazole radical cation in 3a (Eox (3a·+/3a) = + 1.15 V), it is no surprise that only the trimethylsilyl carbazole 4a (Eox (4a·+/4a) = + 0.95 V) provided the desired conjugate addition product 2 bearing a quaternary stereocenter (entry 1). Substrates 4b and 4c remained completely unreacted. Substrate 4d, having a dimethyl(phenyl)silyl group, provided better results than the trimethylsilyl (TMS) analog (speculatively because of the enhanced β-silyl stabilization of the intermediate radical cation of type D50,51, compare entries 1 and 4), and was selected for further studies.

Table 1 Optimization studies
Full size table

We then evaluated a family of chiral catalysts 3b–e bearing different substitution patterns, which provided the redox-active carbazole unit with a range of electronic properties. As previously reported by our group39,40, both enantiomers of catalysts 3 are accessible in good yields from readily available carbazoles through a five-step sequence (see Supplementary Note 1 in the Supporting Information for details on the synthesis of catalyst 3e). In congruence with the mechanistic proposal, we found a correlation between the catalytic activity and the oxidation ability of the carbazole unit. The reaction was effectively promoted by catalysts bearing more electron-withdrawing groups (3d and 3e), which magnified the oxidizing power of the carbazole radical cation (entries 7 and 8). In contrast, the presence of an electron-donating group in 3c completely inhibited the process (entry 6). The best results in terms of reactivity and stereocontrol were provided by catalyst 3e, which was adorned with substituents that can enhance both its steric shielding and oxidizing ability (entry 8).

Mechanistic investigation

Catalyst 3e was selected for further investigations to better elucidate the mechanism. No product formation was detected in the absence of light or when replacing catalyst 3e with cyclohexylamine, which lacked the redox-active carbazole core. Optical absorption studies (detailed in Supplementary Figure 4) indicated that only the in situ generated iminium ion A could absorb at 420 nm, the operative λ of the system. When performing the model reaction in the presence of ethene-1,1-diyldibenzene 5, we observed the formation of adduct 6, arising from the trap of the radical photogenerated from substrate 4d (Fig. 4a). We also detected 3-methyl cyclohexanone 7, a compound which likely arose from the β-enaminyl radical intermediate C (Fig. 3) upon hydrogen abstraction. The last observation suggests a possible pathway to turn over the catalyst 3 involved in the light-triggered radical generation manifold.

Fig. 4
figure 4

Mechanistic investigations. a Trapping the radical intermediate generated from 4d. b The presence of an hydrogen donor 8 favors the formation of 3-methyl cyclohexanone 7. c Entrainment experiment supporting a radical chain propagation mechanism

Full size image

We then performed an additional experiment, where the model reaction was carried out in the presence of an excess of 1,4-cyclohexdiene 8, a hydrogen donor that afforded 7 in a better yield and with moderate enantioselectivity (Fig. 4b). Control experiments established that catalyst 3e, substrate 4d, and light were all necessary for the formation of product 7. Overall, these observations are congruent with the proposed mechanism for radical formation, where the photogenerated intermediate B oxidizes substrate 4 to furnish the radical along with intermediate C.

Finally, we conducted an experiment using two carbazole catalysts with an opposite absolute configuration (Fig. 4c). We mixed 15 mol% of (R,R)-3d, a catalyst that can efficiently promote the model reaction to give product (S)-2a in 80% ee, with 5 mol% of catalyst (S,S)-3b, which is completely inactive in these conditions. This catalyst mixture afforded the product (S)-2a with a greatly reduced enantioselectivity (42% ee) with respect to using 3d alone. This result implies that, while only (R,R)-3d can be involved in the radical generation pattern, the subsequent iminium ion trap is governed by both catalysts, thus causing an erosion of stereocontrol. This entrainment experiment (the non-active catalyst 3b at initiation is instead apt at propagation) supports the mechanism proposed in Fig. 3, where the EDA-complex-based photochemical radical generation serves to initiate a radical conjugate addition sustained by a chain propagation mechanism. Our attempts to determine the quantum yield of the process, which would be useful to support the feasibility of a chain manifold, were frustrated by the impossibility of carefully establishing the photon flux of the high-power violet LED needed for the photochemical processes, a crucial requirement for reliably determining the quantum yield.

Substrate scope and synthetic applications

Adopting the optimized conditions described in Table 1, entry 8, we demonstrated the generality of the radical conjugate addition by evaluating a variety of cyclic enones 1 and α-carbazole silyl reagents 4 (Table 2). The resulting chiral products 2 contain a quaternary carbon center bearing a carbazole moiety, a structure found in optoelectronic materials, synthetic dyes, conducting polymers52, and naturally occurring alkaloids53.

Table 2 Substrate scope for the enantioselective trap of carbazole-derived radicals
Full size table

A wide range of substituents at the carbazole core were well-tolerated, regardless of their electronic properties and position (products 2a–k). Heterocycle rings, including furyl and thienyl moieties, could be included in the products (entries 7 and 8). As for the enones, a wide range of carbocycles and β-olefin substituents could be used. Ring systems that incorporate β-alkyl groups other than methyl reacted with a reduced enantioselectivity (entries 12–14), while the radical conjugate addition performed well for a diverse range of ring sizes, including cyclopentenyl and cycloheptenyl architecture (entries 15 and 16). Interestingly, linear enones reacted smoothly to provide the corresponding compounds 2q-r, albeit with moderated or no stereoselectivity (entries 17–18). Crystals from compound 2a were suitable for X-ray crystallographic analysis, which established the absolute configuration of the quaternary stereocenter.

The general applicability of a chemical strategy is crucial for evaluating its usefulness. We wondered if the photochemical activity of the iminium-ion-based EDA complex could be expanded to successfully generate radicals from α-silyl amines 935,54,55,56, organic silanes with a low oxidation potential (Eox < 1.0 V, see Supplementary Figure 12 for details). As detailed in Table 3, the resulting photochemical radical conjugate addition to cyclic enones efficiently afforded an array of adducts 10 with a quaternary stereocenter. In contrast, both α-indole and α-pyrrole silyl reagents remained unreacted under the optimal conditions (see Supplementary Figure 2).

Table 3 Substrate scope for the enantioselective trap of α-silyl amine-derived radicals
Full size table

Preliminary investigations indicated that our protocol is compatible with different radical precursors other than silanes. For example, N-phenylglycine 1157 (Eox = + 0.42 V vs. SCE in CH3CN) and the (benzyloxy)methyl-substituted dihydropyridine 13 (Eox = + 1.08 V vs. Ag/Ag+ in CH3CN) provided the corresponding adducts 12 and 14, respectively, with promising enantioselectivity (Fig. 5a). We also applied our method for the streamlined preparation of biologically relevant compounds. Figure 5b details a straightforward enantioselective two-step synthesis of compound 15, which is potentially useful to prevent ophthalmic disease58. Last, we could stereoselectively prepare trans-alcohol 16, the epimer of a compound useful for the treatment of influenza (Fig. 5c)59.

Fig. 5
figure 5

Expansion of the scope and synthetic utility. a Using different radical precursors than silanes. b, c Enantioselective synthesis of biologically relevant compounds

Full size image

In summary, we have demonstrated that an intramolecular EDA complex can be used to trigger a photochemical asymmetric radical process. Crucial for implementing the process was a chiral carbazole catalyst that do not contain any photosensitive unit, but rather it guides the photochemical formation of radicals by inducing the transient formation of visible-light-absorbing intramolecular iminium-ion-based EDA complexes. The method, which avoids the need for external photoredox catalysts, offers the first example of synthetic applications triggered by the photochemical activity of an intramolecular EDA complex.

Methods

General

For the exemplary reaction setup, see Supplementary Fig. 1. For more information on unreactive substrates, see Supplementary Fig. 2. For UV–Vis absorption studies, see Supplementary Figs. 3, 4. For cyclic voltammetry studies, see Supplementary Figs. 5–13. For the NMR spectra of compounds in this article, see Supplementary Figs. 14–47. For the product analysis with HPLC on chiral stationary phase, see Supplementary Figs. 48–80. For crystallographic information, see Supplementary Fig. 81 and the Supplementary Tables 1, 2, 3. For the synthesis of catalyst 3e and substrates 2, 4, and 9, see Supplementary Note 1. For the preparation of the iminium ion A-2, see Supplementary Note 2. For details of mechanistic studies, see Supplementary Note 3. For details of synthetic applications, see Supplementary Note 4. For general information, general experimental procedure, and analytic data of compounds synthesized, see Supplementary Methods. For additional references pertinent to the Supplementary Information, see Supplementary References.

Catalytic enantioselective radical addition reaction to enones

Exemplary, a 15 mL Schlenk tube was charged with the chiral carbazole-derived primary amine catalyst (R,R)-3e (0.04 mmol, 20 mol%), acid (0.08 mmol, 40 mol%, salicylic acid for substrate 4 and benzoic acid for substrates 9), the organic silane 4 or 9 (0.15 mmol, 150 mol%), enone 1 (0.1 mmol, 100 mol%), H2O (0.2 mmol, 200 mol%), and 200 μL of CH3CN. The mixture was placed under an atmosphere of argon, cooled with liquid nitrogen, and degassed via vacuum evacuation (5 min), backfilled with argon, and warmed to room temperature. The freeze-pump–thaw cycle was repeated three times, and then the Schlenk tube was placed into a support fitted with a 420 nm high-power single LED (λ = 420 nm). The irradiance was regulated at 15 ± 2 mW/cm2, as controlled by an external power supply and measured using a photodiode light detector at the start and the end of each reaction; the temperature was kept at 35 °C with a chiller connected to the irradiation plate (the setup is detailed in Supplementary Fig. 1). This setup secured a reliable irradiation and temperature while keeping a distance of 1 cm between the reaction vessel and the light source. Stirring was maintained for the indicated time (generally 48 h), and then the irradiation was stopped. The reaction volatiles were removed in vacuum and the residue was purified by column chromatography to give the products 2 or 10 in the stated yield and enantiomeric purity. The reported yield and ee are average of two runs per substrate.

Data availability

The X-ray crystallographic coordinate for structures reported in this study have been deposited at the Cambridge Crystallographic Data Centre (CCDC) under deposition numbers CCDC-1819014 (2a). These data can be obtained free of charge from The Cambridge Crystallographic Data Centre via www.ccdc.cam.ac.uk/data_request/cif. The authors declare that all other data supporting the findings of this study are available within the article and Supplementary Information files, and also are available from the corresponding author upon reasonable request.