Reforging Mitsunobu reaction via sulfoxides-catalyzed C(sp³)-O bond activation of alkanols

Abstract

The stereospecific conversion of alkanols plays a pivotal role in synthetic chemistry and drug discovery. Among the substantial advances, the Mitsunobu reaction has long stood out as a powerful platform for C(sp³)-O bond activation to achieve the transformation of alkanols. Due to the inherent environmental drawbacks of phosphorus chemistry, an alternative element is pressingly demanded. Herein, we develop well-designed sulfoxide catalysts for an alternative catalytic Mitsunobu-type reaction with broad substrate scope, generating H₂O as the sole by-product. Kinetic studies, mechanistic investigations, and density functional theory (DFT) calculations provide comprehensive mechanistic insights and uncover key structure-reactivity relationships. This approach establishes a sustainable and atom-economical pathway for the Mitsunobu-type reaction, and expands the conceptual scope of sulfur catalysis.

Due to the inherent environmental drawbacks of phosphorus chemistry, an alternative to the Mitsunobu reaction is pressingly demanded. Herein, the authors develop well-designed sulfoxide catalysts for an alternative catalytic Mitsunobu-type reaction with broad substrate scope, generating H₂O as the sole by-product.

Introduction

The conversion of alkanols with precise stereochemical control is critical for the synthesis of bioactive molecules, natural products, and functional materials^1–3. Among the strategies available, the Mitsunobu reaction^4–6 has been witnessed as a uniquely powerful platform for C(sp³)-O bond activation and stereospecific functionalization of alkanols via the key phosphorus(V) intermediates⁷ which is involved in the Mukaiyama condensation⁸ and the Appel reaction^9,10 as well (Fig. 1a). In those cases, stoichiometric phosphine and activating agent were required, and the problems caused by side products remain significant. Although various methods have been developed to address the disadvantages, including catalytic protocols, but stoichiometric oxidant or reductant was still required^11–16. Meanwhile, other organocatalysts were also capable of deoxychlorination of alkanols aided by acyl chloride^17–20. In 2019, a fully catalytic and redox-neutral Mitsunobu reaction was pioneered by Denton through a well-designed phosphine oxide catalyst (Fig. 1b)²¹, followed by a DFT-assisted catalyst modification of phosphine oxide catalysts revealed by Houk²², and a detailed kinetic analysis revealing the key point for water removal process in the above reactions²³. In 2023, an example of phosphine oxide catalyzed deoxyhalogenation of alkanols was reported by Song and co-workers²⁴. Efforts to address these challenges through phosphine oxide-catalyzed Mitsunobu reactions have provided important advances in phosphorus chemistry^25,26, but limitations remain in terms of catalyst efficiency. An alternative element is urgingly demanded.

Fig. 1. Previous work and this work.

a The Mitsunobu reaction. b The redox-neutral catalytic Mitsunobu reaction. c C-O bond activation enabled by sulfuranes. d Mechanistic evaluation of sulfoxide catalyst. e This work.

Sulfur catalysis offers an attractive alternative. Sulfur reagents are abundant, inexpensive, and possess rich redox chemistry^27–29 capable of mimicking phosphine activation modes without their toxicological and waste-related drawbacks^30,31. Harnessing sulfur-based catalysis to achieve a Mitsunobu-type C(sp³)-O bond activation would merge the conceptual power of a classic reaction with a sustainable, atom-economical framework. However, similar chemistry of sulfur remains rarely reported. The utility of Martin sulfurane^32–38 in stoichiometric alcohol dehydration and the diaryldialkoxysulfuranes for the activation of alcohols, aldehydes, and carboxylic acids reported by Ishikawa^39,40 and co-workers illustrate sulfur’s potential (Fig. 1c). The first example of sulfoxides-catalyzed deoxychlorination of alkanols, reported by Huy⁴¹ and co-workers suffered from low stereoselectivity, poor feasibility with aliphatic and electron-poor alcohols, and catalyst decomposition via the Pummerer rearrangement⁴². In 2025, a phenolic sulfoxide catalyzed Fischer esterification reactions was developed by Kancharla⁴³ and co-workers. However, Only the C-O bond cleavage of carboxylic acids was observed under this catalytic protocol, rather than the deoxygenation of alkanols as in Mitsunobu-type reactions.

Here, we address this challenge by designing a redox-neutral sulfoxide-catalyzed C(sp³)-O bond activation for Mitsunobu-type reaction enabled by a rigid biaryl diarylsulfoxide skeleton specifically designed to suppress the Pummerer rearrangement. By strategically selecting a monocyclic sulfurane as the key reactive intermediate, this catalyst design balances reactivity and stability, avoids the pitfalls of both highly reactive acyclic sulfuranes and inert spiro-sulfuranes, and optimizes the rate-determining step among multiple reversible steps to minimize side-product formation. We envisaged that catalyst 1 could be pre-activated via reversible dehydration and cyclization with acidic nucleophile to form 2. Subsequently, alkanol would be activated via a substitution and ring-opening process, generating 3 as steady state, which would undergo nucleophilic substitution with counter anionic nucleophile to proceed reaction (Fig. 1e). This approach not only establishes a sustainable, atom-economical pathway for Mitsunobu-type reaction but also expands the conceptual scope of sulfur catalysis in synthetic chemistry.

Results and discussion

To verify the feasibility of this proposal, a stoichiometric reaction with C1 was initially conducted and the desired trichloroacetate product was detected (see Supplementary Information page S54 for more details), demonstrating the possibility of this C(sp³)-O bond activation strategy. Encouraged by these results, the catalytic reactions were then conducted with 4a as nucleophile, toluene as solvent, with water removed by a Dean-Stark trap under reflux condition (Fig. 2). However, only 8% yield was obtained under this catalytic condition. Thus, various modified sulfoxide catalysts were tested. It was proven efficient that electron-withdrawing groups were introduced on the A ring to further increase the acidity of the phenol group (C2), and the electron-donating groups were employed on the C ring to increase the stability of the sulfurane intermediate (C3). To our delight, with the utility of catalyst C5, bearing an -NO₂ group on the A ring and -OMe group on the C ring, the reaction proceeded smoothly to afford the desired product in 50% yield. Notably, only C4 give an improved result of 71% yield, implying the yield was affected by both electronic effect and steric effect (Fig. 2, C4–C9). Compared with C5, C10 that bears an additional methoxy group on the B ring demonstrated no significant difference in yield, suggesting that the electronic effect of the B ring has less impact on the catalyst activity. When a more rigid catalyst (C11) was employed, no desired product was observed, presumably because of the sterically hindered methyl group. Only trace products were observed when C12 and C13, both bearing with an ortho-nitro group, were engaged, which might be attributed to the steric hindrance and the intramolecular hydrogen bond of the ortho-nitro group. Worthwhile, the reaction with the more acidic nucleophile 4b afforded the desired product 6a in higher yield (95%). Subsequently, other reaction parameters were optimized with respect to the ratio of substrates, the loading of catalyst, and the concentration (see Supplementary Information page S22 for more details). Meanwhile, physical removal of water is necessary as Denton²¹, Andrews and Borsley reported²³, the utility of molecular sieve in the reaction mixture gave trace result. After evaluation, the reaction performed with 5 mol% of catalyst, 1.2 equivalent of nucleophile, and a concentration of 0.05 M proved optimal, delivering the desired product in 96% yield.

Fig. 2. Optimization of sulfoxide catalysts.

Isolated yield, trace ( < 5%). See Supplementary Information for experimental procedure.

Mechanistic research

With the optimized reaction conditions in hand, to further understand the reaction process, the kinetic studies were subsequently performed. The kinetic studies under standard condition revealed fractional orders for all components owing to the impact of water removal^23,44. To further reveal the detailed reaction kinetics, the pseudo-first-order kinetics were employed. The kinetic curves demonstrated that the reaction follows pseudo-first-order kinetic with respect to the nucleophile, alkanol, and catalyst under the pseudo-first-order conditions (Fig. 3a) (see Supplementary Information page S68-S85 for details). A comparison of sulfoxide catalyst C4 with phosphine oxide catalyst demonstrated that the sulfoxide catalyst exhibited an approximately 6-fold rate acceleration (see Supplementary Information page S88 for details). Subsequently, the substituent effects on A ring and C ring of the catalyst were examined by means of Hammett plots^45,46 (Fig. 3a, see Supplementary Information page S86 for details). A distinct break was observed both Hammett plots of A ring and C ring. For the A ring, the correlation line for electron-withdrawing substituents (R = H, OCF₃, CF₃) displayed a significantly positive ρ value of ~1.59, whereas the line corresponding to strongly withdrawing substituents (R = CF₃, CN, NO₂) exhibited a much smaller positive ρ value of ~0.14. For the C ring, the correlation line for electron-withdrawing substituents (R = H, Cl, CF₃) displayed a negative ρ value of ~−0.35, while the line for electron-donating substituents (R = OCH₃, CH₃, H) exhibited an even smaller ρ value, close to zero. Such phenomena are typically attributed to three factors: (a) a change in reaction mechanism, (b) variations in a single mechanism, such as differing extents of bond formation/cleavage or altered charge distribution in transition state, and (c) changes in contribution between polar and resonance effects by different substituents in transition states^47,48.

Fig. 3. Mechanistic researches.

a Kinetic studies and Hammett plots. Pseudo-first-order condition: derivation of standard condition, maintain a 30-fold excess of 4b or 5c. Standard condition: 0.06 M 4b, 0.05 M 5c, 0.005 M C4. Condition of Hammett plots: 0.12 M 4b, 0.1 M 5c, 0.01 M catalyst. b Isotope experiment. c Research of active intermediates.

Then, the isotope labeling experiment with ¹⁸O-enriched alkanol 5an as substrate was conducted (Fig. 3b), resulting in the formation of an ¹⁸O enriched catalyst (75% ¹⁸O containing) and a product 6at with high ¹⁶O in corporation, which suggested that the C(sp³)-¹⁸O bond cleavage was accompanied by S-¹⁸O bond formation. After that, a stoichiometric stepwise experiment was employed to track the formation of plausible intermediates (Fig. 3c). Under this condition, the key monocyclic intermediate 7 was captured with trifluoroacetate bonding to the sulfur center and characterized by HRMS and in situ NMR experiments. Subsequently, 7 was efficiently quenched by the addition of alkanol 5c, affording a ring-opening intermediate 8 and other derivates (see Supplementary Information page S57 for details). When the reaction was completed, C4 was regenerated and precipitated from the reaction system as well as the formation of desired nucleophilic substitution product.

Based on the aforementioned results, density functional theory (DFT) calculations were then conducted to support the proposed mechanism (Fig. 4), which was based on the template reaction in Fig. 3. The reaction was initiated via proton transfer from the acidic nucleophile (4b) to sulfoxide C4, with a low energy barrier (TS1, 5.9 kcal/mol), followed by another proton transfer, forming a four-coordinated sulfurane intermediate (Int2) with a trigonal bipyramid geometry, in which two oxygen atoms occupied the axial position. Specifically, two diastereoisomers (Int2-dr1 and Int2-dr2) were generated due to the rotation of the biaryl skeleton, and the latter one was stabilized by the π-π stacking interaction between A ring and C ring. Subsequently, the dehydration of Int2-dr1 and Int2-dr2 proceeded via a proton transfer from the nucleophile to the hydroxy-sulfurane, affording a more active sulfurane intermediate (Int3). Then, Int3-dr1 and Int3-dr2 were readily quenched by alkanol 5c via a proton transfer process forming Int4-dr1 and Int4-dr2, which could rearrange into the ring-opening intermediates Int5-dr1 and Int5-dr2 with a quite low energy barrier (~1.0 kcal/mol). This process was further supported by the ¹⁹F NMR studies (see Supplementary Information page S64 for details). Eventually, the Int5-dr2 underwent nucleophilic substitution with a reaction barrier of 21.0 kcal/mol, implying that this step was the rate-determining step of this reaction. Based on a simplified model shown in Fig. 4b, the substituent effects were also computationally evaluated to further understand the Hammett plot results (Fig. 3a). Surprisingly, each substituted sulfoxide exhibited the same mechanistic profile (Fig. 4b) according to calculations, implying that the rate-determining step (r.d.s.) remained unchanged across different substituents. In the case of the A ring, the electron-withdrawing groups were positively correlated with the stability of Int5 by enhancing hydrogen bond interaction, but negatively correlated with the nucleophilicity of the anion, leading to higher ∆∆G^‡_r.d.s.. In the case of the C ring, the electron-donating groups were positively correlated with the stability of Int5, but negatively affected the leaving ability of the sulfoxide, leading to higher ∆∆G^‡_r.d.s. as well. Meanwhile, the change of ∆∆G^‡_obs showed an opposite trend compared to ∆∆G^‡_r.d.s in both cases. In addition, the catalytic Fischer esterification process was evaluated as well (see Supplementary Information page S100-S101 for details), which turned out to possess much higher energy barrier for either aryl carboxylic acid (∆∆G^‡_obs = 39.8 kcal/mol) or the alkyl one (∆∆G^‡_obs = 40.6 kcal/mol) to overcome.

Fig. 4. DFT calculations at the B3LYP-D3(BJ)/6-311 G(d,p) (CPCM, toluene) level of theory.

a Calculated reaction mechanism. b Analysis of the Hammett plots and the proposed structure-reactivity relationships.

Substrate scope

Having accessed the structure-reactivity relationships of the sulfoxide catalysts, the investigation into substrate scope was conducted to further demonstrate the capability of this chemistry (Fig. 5). For the conversion of primary alkanols, sulfoxide C4 was employed as a more suitable catalysts with only 5 mol% loading in most cases (6a-6q). 5a was smoothly converted yielding 6a in 96% yield. To our delight, the catalyst loading could be further lowered, although a longer reaction time was required (see Supplementary Information page S21 for details). With the utility of long-chain alkanol, product 6b was obtained in 98% yield, implying that the hydroxy group was well discerned in this reaction conditions. Chloride- and bromide-substituted alkanols (5c & 5d) were tolerated well in this reaction, without intramolecular substitution products being observed. In addition, a gram-scale synthesis of 6c was successfully conducted with 95% yield (see Supplementary Information page S27 for details). Typically, amides (5e & 5f) and arylborate (5g) were also compatible, affording 6e, 6f, and 6g in good to excellent yields. Worthwhile, acid-sensitive alkenes (5h-5j) reacted smoothly with minimal C = C double bond rearrangement, and the relatively low yield of 6j was attributed to the high volatility of 5j. Benzyl substrates (6k-6n) were obtained in good to excellent yields as expected. Acid-sensitive alkynyl alkanol (5o) was also tolerated under this condition, delivering 6o in moderate yield. Owing to steric hindrance, 6p was afforded in only 27% yield. Excellent regioselectivity (r.r. = 20:1) was showcased when 5q was employed, affording the functionalization product of primary alcohol. By the meantime, these catalysts were also capable of bisfunctionalization, delivering 6r and 6s in 53% and 44% yield, respectively.

Fig. 5. Substrate scope of alkanols.

Reactions were carried out with 4, 5 (0.5 mmol, 1.0 eq.), sulfoxide catalyst, and toluene (10.0 mL) under reflux condition with a Dean-Stark trap. ^aC4 (5 mol%) as catalyst. ^bC10 (20 mol%) as catalyst. ^cC4 (10 mol%) as catalyst. ^dC2 (10 mol%) as catalyst. ^eC2 (10 mol%) as catalyst and TfOH (10 mol%) as co-catalyst.

With respect to secondary alkanols, sulfoxide C10 was identified as a more suitable catalyst in terms of both high yields and high enantiomeric excess (e.g., see Supplementary Information page S22 for details), and a higher catalyst loading of 20 mol% was required. 6t and 6u were obtained in good to excellent yields with inverted stereochemistries as a key feature of Mitsunobu-type reaction. Owing to the competitive S_N1 and Fischer esterification pathways, the benzyl product 6v was obtained with much lower ee value (59%). Interestingly, several cases showed retention of stereochemistry in the substitution product (5w and 5ac), which were attributed to the anchimeric participation of sulfur atom and alkenyl group, respectively⁴⁹, and supported by further DFT calculations (see Supplementary Information page S84 for details). Cyclic alkanol 5x was converted to configuration-inverted product 6x smoothly. However, cyclooctyl product 6y was obtained in lower yield (26%) owing to the strained medium-size ring. This reaction was found to be compatible with a more rigid substrate 5z, affording the desired product 6z with excellent stereoselectivity. However, due to the high volatility of 5z and 5aa, 6z and 6aa were obtained in poor yield (31% & 43%). Natural-product derivative 6ab was acquired with good yield (86%) and excellent diastereoselectivity (d.r. = 100:1). It is worth mentioning that this catalyst performed well with more challenging glycosylation where the stereoselectivities were affected by the anomeric effect in most cases, affording retention products in good to excellent yields (6ad & 6ae). Furthermore, other nucleophiles such as carboxylic acids, sulfonic acids can also be engaged in this catalytic system, affording 6af to 6ai in good to excellent yields. Nitrogen-centered nucleophile (4f) was effective as well. The O-alkylation product (6ak) of isoxazole substrate was obtained with triflic acid (TfOH) as a co-catalyst, rather than the N-alkylation one. Unfortunately, thiophenol was found ineffective due to either poor acidity or catalyst poisoning probably.

Given the excellent stereoselectivity of this method, we moved on to the synthesis of oxazolines, which are valuable building block in asymmetric chemistry⁵⁰. Under this catalytic reaction condition, a green pattern via intramolecular substitution was suggested without utilizing additional activating agents such as MsCl, TsCl, etc⁵¹. (Fig. 6a). When C2 and TfOH were employed as catalyst and co-catalyst, substrate 5af was readily converted to oxazoline 6al in excellent yield and high diastereoselectivity with inverted configuration. The catalytic intramolecular substitution process was proposed based on a comparsion with the non-catalytic result. Surprisingly, seven-member ring (6am) was facilely constructed with 82% yield. Product 6an and 6ao were afforded in lower yields, revealing that the electronic effect of the benzoyl group was crucial. A more active alkanol 5aj-rac was less effective than the inactive one (5ak-rac), implying that the steric effect at the carbon center outweighed the electronic effect. Notably, N-O bond activation was also feasible under this catalytic system (Fig. 6b). In the intermolecular reaction involving 5al and 5a, a rearrangement product (6ar) was obtained instead of the C(sp³)-O bond functionalization product, indicating the in situ generation of isocyanate intermediate.

Fig. 6. Other applications of sulfoxide catalysis.

Reactions were carried out with 5 (0.5 mmol, 1.0 eq.), C2 (10 mol%), TfOH (10 mol%) and toluene (10.0 mL) under reflux condition with a Dean-Stark trap. a Application in oxazoline synthesis. b Application in N-O bond activation.

In conclusion, we have disclosed a sulfoxide-catalyzed C(sp³)-O bond activation of alkanols for a fully catalytic and redox-neutral Mitsunobu-type reaction with broad substrate scopes and high stereospecificity achieved. This catalysis that proceeds without external oxidants, reductants, or stoichiometric activating agents, is capable of activating primary and secondary alkanols, and tolerates diverse nucleophiles while demonstrating potentials in N-O bond activation. Comprehensive mechanistic insights were obtained and structure-reactivity relationships were investigated via a series of mechanistic studies. This work establishes a sustainable and atom-economical alternative to phosphorus-based Mitsunobu reactions, and reveals new opportunities for sulfoxide catalysis in stereospecific transformations.

Methods

Computational details

Quantum chemical calculations were conducted with the Gaussian 16 package⁵². Geometry optimization and frequency calculations were performed at the B3LYP level of theory with Grimme’s dispersion correction with Becke-Johnson Damping (D3-BJ)⁵³ and the 6-311 G(d,p) basis set. Solvation effects were taken into account applying the CPCM⁵⁴ implicit solvation model with toluene as solvent in all calculations. IRC calculations^55–57 were used to confirm that the transition states found from the optimization calculations connect the related reactants and products. Single point energy corrections were performed at the ωB97-xD⁵⁸ level of theory and the def2-TZVP base set⁵⁹. Structures were generated using CYLview⁶⁰.

General procedure for the synthesis of 6

To an oven-dried 25.0 mL flask equipped with a Teflon-coated stirring bar was added 4, 5 (0.5 mmol, 1.0 eq.), sulfoxide catalyst and toluene (10.0 mL), the mixture was refluxed for 24–48 h under Dean-Stark condition and air atmosphere. Then, the mixture was cooled to room temperature and evaporated in vacuo. The residue was purified by flash column chromatography on silica gel (eluting with ethyl acetate and petroleum ether) to afford the desired product 6.

Author contributions

J.Y. & J.L. directed the project and revised the manuscript. Q.L. wrote the draft of manuscript, prepared the Supplementary Information, carried out the reactions, and performed the DFT calculations. W.Z. assisted the synthesis of substrates.

Peer review

Peer review information

Nature Communications thanks Yohei Shimizu, and the other, anonymous, reviewers for their contribution to the peer review of this work. A peer review file is available.

Data availability

The data supporting the findings of this study are available within the Article, Supplementary Information and Supplementary Data 1. Data supporting the findings of this manuscript are also available from the corresponding author upon request. The X-ray crystallographic coordinates for structures reported in this study have been deposited at the Cambridge Crystallographic Data Centre (CCDC) under deposition number CCDC 2456218 (C4). These data can be obtained free of charge from the CCDC via (https://www.ccdc.cam.ac.uk/structures/).

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s41467-026-75633-1.