Introduction

The Ferrier rearrangement (FR) enables the conversion of glycals 1 (Fig. 1a) into synthetically versatile 2,3-unsaturated glycosides 31, important building blocks in numerous bioactive molecules such as spliceostatin2, paclitaxel3, and several antibiotics like capuramycin4,5 (Fig. 1b). Classical FR protocols rely on strong Brønsted or Lewis acids (e.g. CF3CO2H6, TfOH7, BF3·Et2O8, SnCl49, I210) that promote the formation of allylic oxocarbenium ion A from a suitably activated glycal derivative 1, decorated with a good leaving group at the C-3 position. Alternatively, single electron oxidation can generate radical cation B, which subsequently expels a radical fragment to form intermediate A. Such oxidative variants employ (sub)stoichiometric amounts of oxidants, including (NH4)2[Ce(NO3)6]11,12 or K5CoW12O413. While effective, these protocols often rely on corrosive acids and hazardous oxidizing reagents, both notorious for their impact on environmental sustainability, and responsible for a considerable amount of waste generation. Some mild and eco-friendly FR methodologies employing microwave irradiation14 and deep eutectic solvents15 have also been reported. However, their broader applicability is still constrained by scale-up challenges and a relatively limited substrate scope.

Fig. 1: Ferrier rearrangement and its application.
Fig. 1: Ferrier rearrangement and its application.
Full size image

a Classical and oxidative Ferrier rearrangement. b Bioactive compounds obtained via Ferrier rearrangement as a key step.

Organic electrochemistry has recently re-emerged as a powerful platform for redox transformations16, replacing chemical oxidants and reductants with electric current as an inexpensive, renewable, and inherently safe reagent17,18. Recently, we have demonstrated, the feasibility of performing the FR under undivided electrochemical conditions using inexpensive electrodes and mild reaction conditions19. This approach exhibited broad substrate compatibility, accommodating a wide range of glycal derivatives 1 and nucleophiles 2. Nonetheless, this approach suffered from long reaction times, large amounts of supporting electrolyte, and a high excess of charge input, common drawbacks of batch electrolysis.

Continuous-flow electrochemistry, by contrast, offers superior mass and heat transfer due to short diffusion paths20. Moreover, a larger surface-to-volume ratio significantly reduces the reaction time, while the short interelectrode gap in flow reactors lowers Ohmic voltage drop, thereby minimizing the need for supporting electrolytes21. Flow technology has thus become integral to modern organic electrochemistry, offering enhanced efficiency, sustainability and scalability, making it highly attractive for industrial implementation22. The Nagaki group23 reported the development of a novel flow electrochemical reactor20, capable of producing cationic intermediates for FR within very short residence times by adopting the “cation pool” strategy (Fig. 2a)24. In this approach, they afforded a stoichiometric amount of the oxonium ion A by flash electrolysis and subsequently intercepted it by nucleophiles 2 before its decomposition. Another flow electrochemical variant of FR was later reported by Sato and Suga featuring one synthetic example 3a with moderate yield (Fig. 2b)25. However, these protocols are constrained by the need for highly sophisticated divided-cell flow setups, costly platinum electrodes,26 cryogenic conditions, and a limited substrate scope, overall compromising their potential to be employed in an industrial application.

Fig. 2: Electrochemical Ferrier rearrangements in flow.
Fig. 2: Electrochemical Ferrier rearrangements in flow.
Full size image

a Nagaki’s protocol in a divided cell under cryogenic conditions. b Sato and Suga’s protocol in a divided cell with one example. c This work.

Building on our experience in handling sensitive intermediates and products under electrochemical flow conditions27,28,29, we now report an operationally simple, efficient, and scalable FR in an undivided flow setup. The process requires inexpensive electrodes and only minimal amount of supporting electrolyte and electrical charge input (Fig. 2c). Overall, this approach advances FR toward more sustainable and industry-relevant electrochemical synthesis.

Results and discussion

Our investigation commenced with the introduction of commercially available tri-O-acetyl-D-glucal 1a and benzyl alcohol 2b into an undivided electrochemical microflow reactor (300 µL internal volume) equipped with an impervious-graphite anode and regular graphite cathode (Table 1) (see Supplementary Information, Figure S1)30. Based on our previously reported batch electrochemical FR19, where acetonitrile was identified as the optimal solvent, we rapidly optimized the reaction to the undivided continuous-flow setup, affording the desired product 3b in 94% yield, in only 2 min of residence time (tR) under galvanostatic conditions in acetonitrile (entry 1). The excess of benzyl alcohol was required to obtain high yield of the desired product, while lowering the amount of nucleophile resulted in the formation of side products. Notably, the 250 µm inter-electrode gap enabled a drastic reduction of electrolyte loading (0.1 equiv.) without compromising the yield22. Decreasing the substrate concentration (entry 2) or applied charge (entry 3) led to slightly lower yields, whereas increasing the charge (entry 4) led to undesired oxidation of benzyl alcohol 2b to benzaldehyde. The reaction remained highly efficient even at shorter tR (entries 5 and 6), underscoring the potential for high productivity under continuous-flow conditions, though a 2-minute residence time was selected for practicality. Variations in electrolytes from nBu4NClO4 to nBu4NBF4 (entry 7) or nBu4NPF6 (entry 8) proved to be detrimental for the yield, while the use of inert or anhydrous conditions (entries 9 and 10) had little effect on the overall process. Alternative electrode combinations (entries 11 and 12) led to reduced yields. As expected, the reaction was confirmed to be electrochemical in nature, as no conversion occurred in the absence of applied current (entry 13).

Table 1 Optimization of reaction conditions
Full size table

Having established the optimal reaction conditions, we next turned our attention to evaluating the generality of this process (Fig. 3). We began by exploring the scope of various glycal derivatives in combination with benzyl alcohol 2b. Consistent with the batch reaction19, our electrochemical flow protocol was not limited to glycals bearing an acetyl leaving group at the C-3 position (compound 3b) but was also compatible with the benzyl-protected derivative, affording the desired product 3c in high yield (see Supplementary Information, Figure S2 for starting materials). Furthermore, extending the methodology to D-galactose-derived glycal provided the corresponding 2,3-unsaturated glycoside 3d in good yield, confirming the applicability of this microflow electrochemical approach. We then investigated the scope of nucleophiles. Various alcohols were tested: ethanol, n-hexanol and L-menthol afforded 3e-3g in excellent yields. When tetra-O-acetyl-D-glucose31 was employed as the nucleophile, the corresponding disaccharide 3h was obtained. Next, we explored carbon-based nucleophiles to enable C–C bond formation. Electrochemical glycosylation with trimethylsilyl cyanide yielded 3a in significantly higher yield compared to the previously reported method by Sato and Suga25. When allyltrimethylsilane and silyl enol ether were used, the desired alkylated products 3i and 3j were obtained in high yields. Finally, using 1-phenyl-2-trimethylsilylacetylene as the reaction partner enabled the construction of a C(sp³)–C(sp) bond (compound 3k). Furthermore, the oxidation of glycal 1a in the presence of Et₃SiH furnished the product 3l in excellent yield. Notably, reaction with trimethylsilyl azide as the nucleophile was also successful, providing glycosyl azide 3m exclusively in 94% yield, assessed by 1H NMR spectroscopy of the crude mixture. However, during workup procedure, glycosyl azide 3m partially underwent a [3,3]-sigmatropic rearrangement and was isolated as C-1/C-3 isomeric mixture in 89% yield32,33,34. Additionally, employing p-toluenesulfonamide as the nucleophile afforded the corresponding 2,3-unsaturated glycoside 3n in very good yield. Importantly, aliphatic sulfur nucleophiles were also found to be compatible with the reaction conditions (products 3o and 3p). However, the yields were significantly lower compared to those obtained via the previously reported batch protocol.19 This indicates that, although aliphatic sulfur nucleophiles can participate in the transformation, their diminished reactivity sets the limitation for our methodology. On the contrary, when thiophenol was tested as nucleophile, the corresponding product 3q was isolated in excellent yield, despite the low oxidation potential of thiophenol. Some unprotected glycals and nucleophiles were not compatible with the reaction conditions giving a range of side products or stayed unreactive (see Supplementary Information, section 5). Lastly, the products were obtained in diastereomeric ratios in line with the previous electrochemical contributions19,23,25, while the yields are considerably higher, underscoring the efficiency of this electrochemical flow transformation.

Fig. 3: Substrate scope.
Fig. 3: Substrate scope.
Full size image

Reaction conditions: glycal 1 (1.0 equiv.), nucleophile 2 (2.0 equiv.), nBu4NClO4 (0.1 equiv.), CH3CN (0.2 M), impervious graphite anode/graphite cathode, 1.0-8.0 mA·cm−2, 0.025-0.20 F·mol−1, room temperature, 0.15 mL·min−1 flow rate, 2 min residence time, collected for 33.3 min. Yields correspond to isolated products. Diastereomeric ratios were determined by 1H NMR analysis of isolated products. a Scale up experiment: 1.0 mL·min−1 flow rate, 18 s residence time, collected for 50 min. b Yield was determined by 1H NMR analysis of the crude mixture, using diglyme as an internal standard.

We further evaluated the scalability of this methodology by performing a multi-gram reaction of 1a (10 mmol) and 2b (see Supplementary Information, section 4). In flow synthesis, productivity is strongly influenced by the flow rates; thus, a faster flow rate without compromising yield is crucial. For this scale-up experiment, a flow rate of 1 mL·min−1 (residence time 18 s) was employed. The reaction proceeded efficiently under these conditions, delivering 3b with a productivity of 10.2 mmol per hour (3.27 g·h−1) and achieving a space-time yield of 10.9 Kg·L−1·h−1. Conversion of starting material 1a and yield of product 3b remained consistent throughout the entire 50-minute collection period. Furthermore, during reaction optimization no residence time limit was reached at which the yield began to decrease. These findings suggest that productivity could be further enhanced at even higher flow rates.

To demonstrate the sustainability of the developed protocol, its green chemistry metrics were compared with other electrochemical FR methods using the standardized CHEM21 toolkit35 (Table 2). In particular, the divided-cell approach reported by Nagaki23 and our previously reported batch method19 were selected for comparison, both applied to the synthesis of compound 3i from glucal 1a and allyltrimethylsilane. Our highly efficient chain process not only avoids the use of highly hazardous compounds and cryogenic conditions but also affords product 3i in higher yield. Moreover, the process mass intensity (PMI) was significantly improved due to the reduced electrolyte loading and the higher reaction mixture concentration.

Table 2 Comparison of the green chemistry metrics
Full size table

After establishing the sustainability and synthetic utility of this undivided microfluidic electrochemical FR, we investigated its reaction mechanism. The exceptionally high faradaic efficiency observed under flow conditions, together with supporting evidence from the previous literature reports25,36 and cyclic voltametric data19, suggests a radical chain mechanism (Fig. 4). The process is initiated by the anodic oxidation of tri-O-acetyl-D-glucal 1a, generating radical cation B. Loss of the acetoxy radical from B affords oxonium ion intermediate A, which is subsequently intercepted by a nucleophile, here benzyl alcohol 2b, to yield 2,3-unsaturated glycosyl derivative 3b upon deprotonation. The acetoxy radical is then reduced to the acetate by oxidizing another molecule of glucal 1a, regenerating radical cation B and thereby propagating the chain process. The chain is terminated through reduction of the acetoxy radical at the cathode. The choice of anode material is critical: only impervious graphite efficiently promoted the formation of the desired product, whereas other electrodes gave no conversion. Although further studies are needed to confirm this, the high performance of impervious graphite is likely due to its resistance to adsorption or surface grafting of chemical species, which can inhibit electron transfer from the substrate to the electrode37. Reversible surface interactions further facilitate efficient charge transfer, while enhanced mass transfer under flow conditions accelerates the overall reaction. We attribute the optimal efficiency of our electrochemical protocol to this combination of reduced surface fouling and improved electron and mass transfer.

Fig. 4: Proposed mechanism.
Fig. 4: Proposed mechanism.
Full size image

SET – single electron transfer oxidation on anode and reduction on cathode.

Conclusion

In summary, we have developed a continuous-flow electrochemical Ferrier rearrangement in an undivided microreactor using inexpensive electrodes. The process proceeds under mild conditions with minimal amount of supporting electrolyte and charge, delivering a broad range of 2,3-unsaturated glycosyl derivatives in high yields and with excellent faradaic efficiency. This advance transforms the Ferrier rearrangement into a sustainable, industrially relevant electrochemical process and establishes a foundation for scalable electrochemical glycosylations.

Methods

General procedure for electrochemical Ferrier rearrangement in flow

A solution of glycal 1 (2.0 mmol, 1.0 equiv.), nucleophile 2 (4.0 mmol, 2.0 equiv.) and nBu4NClO4 (0.2 mmol, 0.1 equiv.) in CH3CN (0.2 M) was loaded into a 10 mL syringe and pumped through an undivided microreactor (300 µL of internal volume) at 0.15 mL·min−1 (2 min residence time) under constant current at room temperature. After reaching steady state after 10 minutes, the product solution was collected for 33 min and 20 seconds, corresponding to 1.0 mmol of product. The resulting reaction mixture was then concentrated under reduced pressure, and the crude was purified by flash column chromatography on silica gel to afford corresponding product 3. The structures of compounds 3a-q were unambiguously confirmed by NMR spectroscopy (see Supplementary Information, Figures S4-S37).