Visible-light-driven spirocyclization of epoxides via dual titanocene and photoredox catalysis

Abstract

We describe the synergistic utilization of titanocene/photoredox dual catalysis driven by visible light for the radical opening/spirocyclization of easily accessible epoxyalkynes. This environmentally benign process uses the organic donor–acceptor fluorophore 2,4,5,6-tetra(9H-carbazol-9-yl)isophthalonitrile (4CzIPN) as a photocatalyst and Hantzsch ester (HE) as an electron donor instead of stoichiometric metallic reductants. The photocatalytic conditions showed exceptionally high reactivity for the synthesis of privileged and synthetically challenging spirocycles featuring a spiro all-carbon quaternary stereocenter. Cyclic voltammetry (CV) studies suggest that Cp₂Ti^IIICl is the catalytically active species.

We describe the synergistic utilization of titanocene/photoredox dual catalysis driven by visible light for radical opening/spirocyclization of easily accessible epoxyalkynes.

Over the last few decades, radical-based transformations have been increasingly used in organic synthesis due to their salient features, such as ease of generation, mild reaction conditions, and broad functional group compatibility.^1,2 As a mild single-electron-transfer (SET) reagent, titanocene monochloride (Cp₂Ti^IIICl) is considered a formidable tool in contemporary radical chemistry due to its ability to promote various fundamental radical-based transformations.^3–7 Cp₂Ti^IIICl was first introduced by Nugent and RajanBabu as a very mild stoichiometric reagent for the reductive opening of epoxides.^8–11 Later, the catalytic conditions developed by Gansäuer et al. (Scheme 1a)¹² employing stoichiometric amounts of active metals in combination with 2,4,6-collidine·HCl further expanded its applications and led to the discovery of a number of novel transformations.^13–16 The key to success was the formation of a stable complex A in reactions while decreasing the concentration of active Cp₂Ti^IIICl.^17,18 We were interested in the radical opening/cyclization reaction of epoxides which has attracted considerable attention from the synthetic community and has been used numerous times in the synthesis of natural products.^19,20 Nevertheless, this reaction required stoichiometric metallic reductants and proceeded slowly particularly with sterically hindered substrates even with high catalyst loading.²¹ Therefore, the development of an eco-friendly and efficient catalytic system with an expanded substrate scope is highly desirable.

Scheme 1. Cp₂Ti^IIICl mediated radical opening/spirocyclization of epoxides; (a) generation of Ti^IIIvia a metal reduction approach; (b) dual titanocene/photoredox catalysis; (c) examples of drugs and natural products containing heterospirocycles.

In recent years metallaphotoredox catalysis has been a new and rapidly growing research subject.^22–29 Photoredox processes can directly modulate the oxidation state of metals by electron transfer (ET).^30–33 Given that the generation of Ti^III is a SET process, we envisioned that the reduction could be facilitated by a photoredox-controlled process while overcoming the aforementioned limitations. On the other hand, spirocycles bearing a chiral spiro all-carbon quaternary carbon are particularly attractive synthetic targets in pharmaceutical development (Scheme 1c).^34–36 Such privileged rigid 3D structures offer the concomitant ability to project functionalities in all three-dimensional orientations and led to enhanced pharmacological activities of molecules. Thus significant attention has been paid to their synthesis.^37,38 Against this backdrop, here we describe our efforts on the synthesis of various heterospirocycles with the aid of photoredox catalysis.

We chose epoxyalkyne 2a as a model substrate for optimization of reaction conditions. After a systematic variation of different reaction parameters, we were pleased to identify the optimal reaction conditions in which a mixture of Cp₂TiCl₂ (5.0 mol%), [Ir(dtbbpy)(ppy)₂]PF₆ (1a, 1.0 mol%, E^III/II_1/2 = −1.51 V vs. SCE in MeCN), HE (1.2 equiv.) and 2a (1.0 equiv.) in THF at room temperature under the irradiation of a 10 W 450 nm light emitting diode (LED) lamp for 12 hours afforded the desired product 3a in an excellent yield of 96% (13 : 1 d.r.) upon isolation (entry 1). Using a commercial 23 W compact fluorescent lamp (CFL) instead of the 10 W 450 nm LED did not compromise the overall yield of the reaction (entry 2). Notably, when the loading of Cp₂TiCl₂ was decreased to as low as only 2.0 mol%, the reaction still led to full conversion and produced 3a in 95% yield (entry 3). Further screening of other photosensitizers revealed that the cheap and readily obtained organic dye 4CzIPN 1b is a competent alternative, which led to full conversion with 94% isolated yield (entry 4). Importantly, the reaction did not proceed in the absence of Cp₂TiCl₂, HE, the photocatalyst, or visible light (entries 5–8). Various solvents, including DMF, MeOH, DMSO, and MeCN, were screened, and they all resulted in poor conversion. The use of other organic electron donors, such as triethylamine, triethanolamine, and ascorbic acid, afforded the product in poor yield.

With satisfactory reaction conditions established, we then explored the scope of the cyclization reaction using 4CzIPN as the photosensitizer. Positively, the cyclization reaction worked well and afforded the desired variably heterospirocyclic products in good to excellent yield (Tables 2 and 3). The reaction allows the rapid construction of various 5/5, 5/6, 5/7 and 5/8 spiro-ring fused systems (3a–3k) bearing tetrahydrofuran or pyrrolidine motifs via the 5-exo cyclization pathway. Interesting, the diastereoselectivity of the cyclization reaction is highly correlated with the ring size in the substrates. Heterospirocycles containing a 5/5 spiro-ring fused system (3a–3f) were obtained with surprisingly high diastereoselectivity. In some cases (3b, 3c, and 3e) only a single isomer was obtained. The product 3d with a sterically hindered t-butyloxy carbonyl (Boc) protecting group on the N atom was obtained with reduced diastereoselectivity (5 : 1 d.r.). The diastereoselectivities dropped in 5/6, 5/7 and 5/8 spiro-ring fused systems. Given that enantioenriched epoxides could be easily obtained (e.g. via sharpless asymmetric epoxidation), this strategy provides access to optically active spirocycles featuring an all-carbon quaternary stereocenter with the transfer of stereochemical information from epoxides (3c, 3e and 3f). Bis-heterospirocyclic scaffolds were frequently employed in pharmaceutical chemistry. For example, bis-heterospirocyclic 3d is the core structure of DLK inhibitors³⁹ and XEN402 (ref. 40) (scheme 1c), which are used for treating neurodegeneration and congenital erythromelalgia respectively. Furthermore, 6-exo cyclization was also investigated under the standard conditions and smoothly produced a serious of drug-like 6-(trifluoromethyl)-3-pyridinesulfonyl piperidine derivatives including 6/5, 6/6 and 6/7 spiro-ring fused systems (5a–5k) in generally excellent yields. Moreover, cyclization reactions with epoxy-alkynes afforded products containing exocyclic-alkenes and free alcohols which were suitable for further functionalization. This approach provides access to a broad range of novel spirocyclic piperidine and pyrrolidine spirocycles which could be of interest to synthetic and medicinal chemists.

Scope of 5-exo and 6-exo cyclization^a,b,c,d.

^aReaction conditions: 2 and 4 (100 mg, 0.1 M in THF).

^bIsolated yield.

^c 3c, 3e and 3f were synthesized from enantiomer pure epoxides.

^dYields within parentheses are based on catalytic conditions using metal as a reductant: CpTi₂Cl₂ (5 mol%), Zn (2.0 eq.), coll·HCl (2.5 eq.), THF, 20 hours.

Additive effect on Ti-catalyzed cyclization^a,b.

^aIn all cases, 2l was used as the substrate and the yield of 3l was determined with ¹H NMR.

^bValues within parentheses are recovery yields of the additives determined with ¹H NMR.

To examine the scalability of the reaction, gram-scale synthesis of 3a and 3c was performed under the standard conditions with 23 W CFL irradiation. Pleasingly, 92% (3a) and 89% (3c) isolated yields were obtained respectively without any deterioration. Furthermore, an additive-based investigation⁴¹ was performed and the results are summarized (Table 3). From this screening, we found that 11 out of 12 additives have no adverse impact on the yield of the reaction. The additives were recovered after the reaction, including benzoxazole 6a, quinazolinone 6b, collidine 6c, tetrahydroquinoline 6d, benzothiazole 6e, indole 6f, and benzofuran 6h. However, quinoxaline 6l strongly inhibited Ti catalysis and 3l was produced in only 23% yield. Notably, the reaction is compatible with various functional groups including phenols 6g, free alcohols and alkene 6j, iodobenzene 6i, ester 3m, ether 3n, dioxolane 5i, lactone 5k, and alkyne 3e. The generality of the dual Ti/4CzIPN catalysis system was further demonstrated by a highly related hydrogen transfer reaction of epoxides which could exclusively provide anti-Markovnikov alcohols (see the ESI S8†). The low-cost of 4CzIPN, broad compatibility with sensitive functional groups, and simple operation conditions are appealing for laboratory and industrial applications.

Importantly, the reactions were re-subjected to metal reduction catalytic conditions for comparison with the photocatalytic conditions and the yields are shown within parentheses (Table 2^d). It clearly showed that the yields were generally lower. Particularly, we found that the yields dropped dramatically as the steric encumbrance of the substrates increased. For example, substrates containing dioxolane (2g, 4i) or a 7/8-membered-ring (2j, 2k, 4j) afforded the products (3g, 5i, 3j, 3k, 5j) in less than 10% yield with the recovery of the starting materials.

Catalytic species and mechanistic studies

To further understand the exceptional reactivity of the photocatalytic reaction system and the mechanism, a series of experiments were performed. First, the mixture of Cp₂TiCl₂, the photocatalyst (1a or 1b) and HE in THF was irradiated under a 23 W CFL or a 450 nm LED at room temperature. The color change from red (Cp₂TiCl₂) to green indicates that a Ti^III species was formed (Fig. 1b). The results of analysis by ultraviolet-visible (UV-vis) spectroscopy further revealed that the reduction is fast. It was nearly complete within 20 minutes (Fig. 1a). On the other hand, no change of color was observed in the dark even at elevated temperatures, indicating that ground-state SET was not operative. The initial electron transfer during the SET process was investigated by fluorescence quenching experiments (Fig. 1c). The luminescence of 1a at its λ_max (591 nm) in degassed THF was readily quenched by Cp₂TiCl₂, exhibiting Stern–Volmer kinetics with a rate of 3.8 × 10⁹ L M⁻¹ S⁻¹. HE also quenched the luminescence of 1a with a rate of 4.5 × 10⁷ L M⁻¹ S⁻¹. Therefore, both oxidative and reductive quenching of the excited state of 1a can occur.

Fig. 1. (a) UV-vis spectra showing the reduction of Cp₂TiCl₂ upon photoirradiation of a solution of Cp₂TiCl₂, [Ir(dtbbpy)(ppy)₂]PF₆ and HE in THF; the peak at 515 nm corresponds to Cp₂TiCl₂. (b) Photoirradiation (23 W CFL) of the mixture for 30 min. (c) [Ir(dtbbpy)(ppy)₂]PF₆ emission quenching by Cp₂TiCl₂. Experimental details are found in the ESI, S2.†.

Cyclic voltammetry (CV) studies were carried out and the results are shown in Fig. 2. In line with the previous CV studies of Zn/Cp₂TiCl₂ solutions in THF (green line), the broad oxidation wave appearing at −0.26 V (vs. SCE) is assigned to the (Cp₂Ti^IIICl)₂ dimer and Cp₂Ti^IIICl monomer couple (designated 1/2).^17,42 Interestingly, CV studies of Ti^III species generated under photoreductive conditions (orange line) revealed the main oxidation wave appearing at −0.22 V (vs. SCE) which corresponds to the Cp₂Ti^IIICl monomer. One concern was the intrinsic instability of Cp₂TiCl₂ under visible light which would decompose by losing a cyclopentadienyl ligand.⁴³ Nevertheless, in the preliminary studies we found that the Cp₂Ti^IIICl species exhibits special stability under photoreductive conditions (for detailed Experiments and discussion, please see the ESI S9†). Furthermore, the CV analysis of Zn/CpTiCl₃ showed the main oxidation wave appearing at −0.63 V (vs. SCE) (see the ESI S4†). Therefore, the possibility of generation of the CpTi^IIICl₂ species via losing a cyclopentadienyl ligand under light could be ruled out.

Fig. 2. Cyclic voltammetry of 2 mM Cp₂TiCl₂/Zn (green line) and 2 mM Cp₂TiCl₂/[Ir(dtbbpy)(ppy)₂]PF₆/HE (orange line) recorded on a glassy carbon disk electrode with ν = 0.1 V s⁻¹ in 0.2 M Bu₄NPF₆/THF. Experimental details are found in the ESI, S4.† The values are given in V vs. SCE and can be converted to V vs. Fc⁺/Fc by subtracting 0.52 V.

Additionally, the reaction profiles were examined and are shown in Fig. 3. The photocatalytic reaction did not show an obvious incubation period. Moreover, the reactions proceeded pretty fast and completed (full conversion of substrates) within around 2 hours with 2a (red solid line) and even more sterically hindered 4i (blue solid line). In contrast, under the metal reduction catalytic conditions, the reactions proceeded slowly for the same substrates 2a (red dashed line) and 4i (blue dashed line). These results clearly showed the advantage of the current photocatalytic system. Kinetically, complex A (Scheme 1a) formed under metal reduction conditions is less reactive presumably due to the fact that it has no free coordination sites. In contrast, Cp₂Ti^IIICl generated under photocatalytic conditions has a vacant site allowing the oxygen atom of the epoxide to perform a SET through an inner-sphere mechanism efficiently. Furthermore, the in situ-generated HP cation 7, which has a low pKa value compared with the collidine cation, can enhance the rate of protonation of the Ti^IV-alkoxy bond in the catalytic cycle. This is supported by the fact that the reaction is significantly inhibited by the addition of Na₂CO₃ or using triethylamine as a reductant (Table 1, entry 10 and 11).

Fig. 3. Reaction profiles for 2a and 4i (200 mg, 0.1 M in THF). [a] Photocatalytic conditions: 450 nm 10 W LED, 1a (1 mol%), Cp₂TiCl₂ (5 mol%), HE (1.2 eq). [b] Metal reduction conditions: CpTi₂Cl₂ (5 mol%), Zn (2.0 eq.), coll·HCl (2.5 eq.). [c] The yields of 3a or 5i were determined with ¹H NMR.

Optimization of the reaction conditions.

Entry	Conditionsa	Yield (%)b
1	[Ir(dtbbpy)(ppy)2]PF61a	96
2	23 W CFL instead of 450 nm LED, 1a, 16 h	95
3	2 mol% Ti, 1a, 24 h	94
4	4CzIPN 1b, 16 h	94
5	No titanocene	0
6	No HE	0
7	No photocatalyst	0
8	No hν	0
9	DCE instead of THF, 1a	93
10	Et3N instead of HE, 1a	9
11	Na2CO3 (1.0 equiv.) as additive, 1a	0

^aReaction conditions: 2a (100 mg, 0.1 M in THF).

^bIsolated yield.

Based on these results, we tentatively propose the following catalytic cycle in Fig. 5. First, Cp₂TiCl₂ is reduced to Cp₂Ti^IIICl by a photocatalyst (such as excited Ir^III). The strong oxidant Ir^IV species is then reduced by HE to Ir^III and produces HE⁺˙ concurrently. Cp₂Ti^IIICl promoted the reductive opening of the epoxides. The resulting carbon radical 8 intramolecularly adds to the pendant alkyne to construct the spirocyclic structure. The newly formed vinyl radical 9 can abstract a C4–H hydrogen atom of the HE radical cation (HE⁺˙). This is because HE⁺˙ has a low homolytic bond-dissociation free energy for the C4–H bond (this energy requirement is 31.4 kcal mol⁻¹ for acetonitrile).⁴⁴ Deuterium labeling experiments also shows that THF may offer a H source for hydrogen atom transfer (Fig. 4). The in situ-produced Ti–oxygen bond is protonated by the HP cation and Ti^IV is released as well as the cyclization product. Finally, Ir^III can regenerate Cp₂Ti^IIICl for the next catalytic cycle.

Fig. 5. Proposed catalytic mechanism.

Fig. 4. Cyclization of 2l in the presence of THF-d_8.

Conclusions

In summary, we report a novel and greener dual catalyst system consisting of titanocene and the organic dye 4CzIPN for the radical spirocyclization of epoxides under photocatalytic conditions. This operationally simple, scalable, and efficient method provides a strategically distinct manner for the assembly of heterospirocycles featuring a spiro all-carbon quaternary stereocenter. Unlike the metal reduction catalytic conditions, mechanistic studies suggest that under photocatalytic conditions, Ti^III is present as Cp₂TiCl which is readily accessible to epoxides. As expected, the photocatalytic conditions showed exceptional high reactivity for the assembly of heterospirocycles. In this respect, the photocatalytic approach complements the metal reduction approach. Development of new transformations based on photocatalytic reactions is currently under way in our laboratory.

Conflicts of interest

There are no conflicts to declare.