Introduction

There has been mounting evidence to suggest that the number of saturated carbons and chiral centers in an organic molecule correlate strongly to its clinical prospect1,2,3,4. Because of this, saturated heterocycles have become increasingly crucial scaffolds for the development of new pharmaceutical compounds. However, unlike heteroaromatics, which can be synthesized conveniently via a variety of cross-coupling reactions, functionalized saturated heterocyclic ring systems have remained challenging to produce5,6,7. For example, the generation of 1,4-dioxane derivatives, which are prevalent in natural products and bioactive compounds (Fig. 1a)8,9,10,11,12,13, usually requires a lengthy synthetic procedure and/or complex starting materials that are themselves hard to obtain8,9,10,14,15.

Fig. 1
figure 1

Reaction design. a Selected bioactive molecules containing the 1,4-dioxane moiety. b Oxidation of styrenyl alkenes via direct electrolysis. c Synthesis of O-heterocycles via annulation reactions of alkenes with diols

Annulation reactions, in which two bonds are formed in a single step, are among the most-efficient methods for the synthesis of cyclic compounds16. Particularly, dehydrogenative annulation reactions via X–H (X=C or heteroatom) functionalization provide straightforward access to cyclic scaffolds from easily available substrates17,18,19. Although conventional dehydrogenative annulation reactions often involve the use of stoichiometric chemical oxidants, recent advances in organic electrochemistry20,21,22,23,24,25,26,27,28,29,30 have led to the development of safer and more environmentally sustainable alternatives that operate under oxidant-free conditions31,32,33,34,35,36,37,38. However, to the best of our knowledge, the synthesis of saturated heterocycles with 1,4-diheteroatoms through alkene annulation reactions has not been reported. Although the intramolecular anodic dioxygenation of heteroatom-substituted alkenes proceeds efficiently39,40, the intermolecular dimethoxylation of styrenyl alkenes with MeOH under similar conditions resulted in a mixture of compounds generated from 1,2-addition and alkoxydimerization (Fig. 1b)41,42, probably owing to the relatively high concentration of the alkene radical cation intermediates that were formed on the electrode surface. Moreover, the dimethoxylated product can undergo oxidative decomposition via C–C bond cleavage43,44,45,46.

To minimize the side reactions mentioned above, we argue that the use of redox catalysis47,48,49,50,51,52,53,54,55,56,57,58 can facilitate the formation of the desired radical cation in the bulk solution and reduce the electrode potential. Herein, we report a triarylamine-catalyzed electrochemical dehydrogenative annulation reaction of alkenes with 1,2- and 1,3-diols for the synthesis of 1,4-dioxane and 1,4-dioxepane scaffolds (Fig. 1c).

Results

Reaction optimization

The annulation of 1,1-diphenylethene 1 and ethylene glycol 2 was chosen as the model reaction for optimization (Table 1). The electrolysis was conducted in an undivided cell equipped with a reticulated vitreous carbon (RVC) anode and a platinum plate cathode. The highest yield of the 1,4-dioxane product 4 was 91%, achieved when the reaction system consisted of triarylamine (2,4-Br2C6H3)3N (3)46,59 (Ep/2 = 1.48 V vs SCE) as the redox catalyst, iPrCO2H as acidic additive and an excess of 2 in refluxing MeCN (entry 1). Other redox mediators that had a lower oxidation potential than 3, such as (4-BrC6H4)3N (3a, Ep/2 = 1.06 V vs SCE), (4-MeO2CC6H4)3N (3b, Ep/2 = 1.26 V vs SCE) and the imidazole derivative 3c53 (Ep/2 = 1.19 V vs SCE), were less catalytically effective (entries 2–4). When no catalyst was used, the yield of 4 dropped to 58% and two additional byproducts, 5 and 6, were obtained in 16 and 6% yields, respectively (entry 5). Running the reaction at RT (entry 6), with only two equiv of 2 (entry 7), in the absence of iPrCO2H (entry 8), or with an alternative acidic additive such as AcOH (entry 9), EtCO2H (entry 10), CF3CO2H (entry 11), and TsOH (entry 12), all led to a significant decrease in reaction efficiency. The replacement of RVC with an anode material that had a smaller surface area, such as Pt (entry 13) or graphite (entry 14), also showed a detrimental effect on the formation of 4.

Table 1 Optimization of reaction conditionsa

Substrate scope

We explored the substrate scope under the optimized conditions by first varying the substituents on the alkene (Table 2). The reaction was shown to be broadly compatible with different 1,1-diphenyl alkenes bearing substituents of diverse electronic properties at the para- or meta-position of the benzene ring (714). Meanwhile, alkenes functionalized with a 2-thiophenyl or 2-thiazolyl group were also tolerated (15, 16), albeit with slightly decreased reactivity. When a trisubstituted olefin whose C–C double bond was embedded in a five or six-membered ring was employed, the reaction afforded the cis-fused products (1719) as the only diastereomer. However, the employment of a seven-membered cyclic alkene resulted in the generation of a mixture of diastereomers (20, dr = 1.8:1). The structure of the minor diastereomer of 20 was subsequently confirmed by X-ray crystallographic analysis. Trisubstituted acyclic alkenes (2131) bearing a halogen (23, 24), free alcohol (25), silyl ether (26), tosylate (27), ester (28), sulfonamide (29, 30), or phthalimide (31) were all found to be suitable substrates. Furthermore, dioxanes 32 and 33 could be obtained in moderate yields from their corresponding enyne and diene precursors, respectively. It is worth emphasizing that a previous attempt at the anodic reaction of dienes with ethylene glycol generated a mixture of addition and dimerization products instead of the annulation product60. Current limitation of the annulation reaction included the inefficient reaction of α-methylstyrene (34, 19% yield) and the complete failure of styrene (35) and a tetrasubstituted alkene (36)

Table 2 Substrate scope

On the other hand, ethylene glycol could be replaced with other vicinal diols such as (2R,3 R)-(−)-2,3-butanediol (37), cis-1,2-cyclohexanediol (38), 2-methyl-1,2-propanediol (39, 40), 2,3-dimethyl-2,3-butanediol (41), and 1,3-diols (4245). Notably, the unsymmetrically substituted diol 2-methyl-1,2-propanediol reacted regioselectively with different trisubstituted alkenes to afford 39 and 40 bearing two tetrasubstituted carbon centers.

The electrochemical annulation reaction could be conducted on a gram scale as demonstrated by the preparation of 2.5 g of 27 in 80% yield from the annulation of the alkene 46 and 2 (Fig. 2). To allow the application of a higher current, a large anode 50 times the size of that used for the abovementioned model reaction was employed. Gratifyingly, the reaction was completed in < 1 hour, which provided rapid access to 27.

Fig. 2
figure 2

Electrochemical gram scale reaction. Gram scale synthesis of 27

Discussion

A reaction mechanism was proposed in Fig. 3a. The triarylamine catalyst 3 (Ep/2 = 1.48 V vs SCE) is first anodically oxidized into the radical cation I, which in turn oxidizes the alkene substrate 1 (Ep/2 = 1.68 V vs SCE) through single-electron transfer to furnish the corresponding radical cation II and regenerate 3. The nucleophilic trapping of II with ethylene glycol (2) and the subsequent deprotonation produce the carbon-centered radical III39,40,61,62,63, which is then oxidized by I to afford the carbon cation IV. The cyclization of IV eventually generates the 1,4-dioxane product 4 (Ep/2 = 1.95 V vs SCE). On the cathode, protons are reduced to produce H2. The addition of iPrCO2H facilitates H2 evolution and probably also plays an important role in reducing unwanted reduction of I, the CH2Cl2 solvent and the alkene substrate. The catalytic role of 3 was confirmed by the detection of a catalytic current47,64 using cyclic voltammetry (Fig. 3b). The inclusion of 3 was also found to inhibit the oxidative decomposition of 4, probably because of the inefficient oxidation of 4 by 3-derived radical cation I (Supplementary Fig. 1). This was supported by the observation that 4 remained largely stable when subjected to electrolysis in the presence of a catalytic amount of 3 (Fig. 4a). In contrast, under similar conditions but in the absence of 3, 32% of 4 was found to have decomposed, resulting in the formation of 1,3-dioxane 5 and benzophenone 6 in 16 and 7% yields, respectively (Fig. 4b).

Fig. 3
figure 3

Mechanistic rationale and cyclic voltammograms. a Mechanistic proposal. b Cyclic voltammograms recorded in MeCN/CH2Cl2 (6:1) with 0.1 M Et4NPF6 as the supporting electrolyte. 3 (2.6 mM). 1 (1.3 mM). SET, single-electron transfer

Fig. 4
figure 4

Electrolysis of compound 4. a Electrolysis of 4 in the presence of triarylamine 3. b electrolysis of 4 in the absence of 3. Yields were determined by 1H NMR using 1,3,5-trimethoxygenzene as the internal standard

In summary, we have developed a triarylamine-catalyzed electrochemical annulation reaction for the synthesis of 1,4-dioxane and 1,4-dioxepane scaffolds from alkenes and diols. The reaction is compatible with a wide variety of functional groups and showed excellent tolerance for di- and trisubstituted alkenes, allowing facile access to functionalized O-heterocycles with tetrasubstituted carbon centers. We are currently investigating whether our alkene annulation reaction can be applied to the synthesis of other 1,4-heterocyclic compounds.

Methods

Representative procedure for the synthesis of 4

A 10-mL three-necked round-bottomed flask was charged with 3 (0.02 mmol, 0.1 equiv), the alkene 1 (0.2 mmol, 1 equiv), ethylene glycol 2 (9 mmol, 45 equiv), iPrCO2H (0.4 mmol, 2 equiv), and Et4NPF6 (0.2 mmol, 1 equiv). The flask was equipped with a reflux condenser, a RVC anode (100 PPI (pores per inch), ~ 65 cm2 cm−3, 1 cm × 1 cm × 1.2 cm) and a platinum plate (1 cm × 1 cm) cathode before being flushed with argon. Then, anhydrous MeCN was added. Constant current (12.5 mA, janode = 0.16 mA cm−2) electrolysis was performed at reflux (internal temperature, 80 °C) until the alkene substrate was completely consumed (monitored by thin layer chromatography or 1H NMR). The reaction mixture was cooled to room temperature (RT) and saturated Na2CO3 solution was added. The resulting mixture was extracted with EtOAc (3 × 20 mL). The combined organic solution was dried with anhydrous MgSO4 and concentrated under reduced pressure. The residue was chromatographed through silica gel and the product 4 was obtained in 91% yield as a white solid by eluting with ethyl acetate/hexanes. All new compounds were fully characterized (See the Supplementary Information).

Procedure for the gram scale synthesis of 27

The gram scale synthesis of compound 27 (80%, 2.5 g) was conducted in a 200-mL beaker-type cell with two RVC plates (100 PPI, 5 cm × 5 cm × 1.2 cm) as anode and a Pt plate cathode (3 cm × 3 cm) at constant current of 625 mA (janode = 0.16 mA cm−2) for 0.9 h. The reaction mixture consisted of compound 2 (15 mL, 262 mmol), 46 (2.2 g, 5.8 mmol), 3 (0.42 g, 0.58 mmol), iPrCO2H (1.1 mL, 11.6 mmol), MeCN (180 mL), and Et4NPF6 (1.6 g, 5.8 mmol).