Introduction

Though cross-coupling reactions have become tremendously enabling in the past decades, further expansion of the electrophilic partners beyond conventional aryl halides, sulfonates, or carbonates, remains a considerable challenge1,2,3,4,5,6,7,8,9,10. The use of organosulfur compounds in cross-coupling is a new horizon that merits study due to their availability, chemical robustness, and structural versatility11,12,13,14,15,16,17,18,19,20,21. To date, a number of cross-coupling reactions utilizing aryl sulfides or sulfones have been reported, as exemplified by the well-known Liebeskind-Srogl cross-coupling reaction16,19,20,22. In sharp contrast, employment of sulfoxides as electrophiles in transition-metal catalyzed cross-coupling reactions via C-S bond activation remains rare (Fig. 1).

Fig. 1: Cross-coupling reactions employing sulfoxides as electrophiles.
figure 1

a Nickel-catalyzed Kumada coupling reaction of diaryl sulfoxides. b Palladium-catalyzed Sonogashira coupling reaction of diaryl sulfoxides. c Palladium-catalyzed borylation of diaryl sulfoxides. d Buchwald-Hartwig amination of diaryl sulfoxides and amines. e Nickel-catalyzed Negishi coupling reaction of aryl methyl sulfoxides. f This work: transition-metal-free formal cross-coupling of aryl methyl sulfoxides and alcohols.

Full size image

In 1979, Wenkert and coworkers pioneered the first nickel-catalyzed Kumada coupling reaction employing diaryl sulfoxides as electrophiles (Fig. 1a)23. In 2017, the Yorimitsu group reported a palladium-catalyzed Sonogashira coupling employing diaryl sulfoxides as electrophilic partners (Fig. 1b)24. The same group also described a second generation Buchwald-type palladacycle precatalyst that facilitated the double borylation of diaryl sulfoxides with diborane species (Fig. 1c)25. Both of the protocols shared some drawbacks, primarily low yields and limited functional group tolerance. Very recently, the same group reported a Buchwald-Hartwig amination between diaryl sulfoxides and amines catalyzed by an N-heterocyclic carbene-ligated palladium complex (Pd/SingaCycle-A1) (Fig. 1d)26. They noticed that aryl methyl sulfoxides, unfortunately, suffered from very low yields. The first systematic study employing alkyl aryl sulfoxides as electrophiles emerged in 2017 from the Yorimitsu group27. They demonstrated that synthetically accessible aryl methyl sulfoxides could undergo nickel-catalyzed oxidative addition of C–S bond, which was compatible with a Negishi coupling procedure to afford biaryl products (Fig. 1e). However, considerable effort is needed to separate the homocoupling byproducts of the zinc reagents from the desired heterocoupling products. In reports of aryl sulfide cross-couplings, selected examples are also shown where the corresponding sulfoxides can act as the electrophilic partner28,29,30.

In general, the previous transition-metal catalyzed cross-coupling reactions of sulfoxides with different nucleophiles rely on transition metal oxidative addition to the C–S bond23,24,25,26,27. Transition-metal catalysts can require complex ligands, which may add significantly to the overall costs. In addition, the removal of trace transition metals in pharmaceutical and other applications remains a challenge31,32,33,34. In contrast, transition-metal-free cross-coupling strategies can overcome many of these limitations35,36. Some Friedel-Crafts-type protocols fall into this category, but suffer from limited substrate scope (electron rich arenes) and limited control over the reactive position on a given aromatic substrate.

We envisioned that a neglected S-tetracovalent dialkoxyl alkyl aryl intermediate (Fig. 1f, see key intermediate), which was pioneered by Martin group several decades ago37,38,39, could potentially be utilized to activate the C–S bond, facilitating alkyl aryl sulfoxides as electrophilic coupling partners in the subsequent transition-metal-free coupling reactions with alcohols. This strategy would allow selective conversion of thio or sulfoxide groups to alkoxy groups. Furthermore, the utilization of sulfoxides in coupling reactions represents an alternative and complimentary pathway to conventional partners, such as aryl (pseudo)halides or alcohols, since methyl sulfoxides are prepared from thiophenol derivatives, which are manufactured from benzenesulfonic acid or benzenesulfonyl chloride, independent of aryl (pseudo)halides or alcohols. Herein, we report a transition-metal-free cross-coupling strategy enabled by a base-substrate interaction mode and apply it to the reaction of aryl methyl sulfoxides with alcohols to afford a general and complementary entry to alkyl aryl ethers (Fig. 1f).

Results

Optimization study

Readily accessible methyl 2-naphthyl sulfoxide (1a) which possess no electronic or steric bias was chosen as the model substrate for the study (Table 1). The investigation was initialized by treating 1a with three methoxide bases (LiOMe, NaOMe and KOMe) in 2-Me-THF at 110 °C for 12 h (Table 1, entries 1–3). Use of KOMe as base as well as the nucleophile afforded product 3a in 12% yield. To improve reactivity, four different solvents [toluene, dimethoxyethane (DME), cyclopentyl methyl ether (CPME), and dioxane]) were examined using 1a and KOMe (Table 1, entries 4–7). DME was the best solvent, generating 3a in 33% yield (Table 1, entry 5). We next probed the roles of KOMe by using MeOH (2a) in the presence of different potassium bases [KOtBu, K2CO3, KOH, KN(SiMe3)2, Table 1, entries 8–11]. Notably, the yield of 3a increased to 67% when KOtBu was employed as the base (Table 1, entry 8). Furthermore, the potential byproduct, 2-naphthyl tert-butyl ether, was observed only in trace amounts; presumably, the steric hindrance of OtBu limits its participation as a nucleophile. In the optimization, concentration was determined to be a key factor. When the concentration of the reaction was elevated to 0.5 M, product 3a could be prepared in an optimal 95% yield (Table 1, entry 12). However, the yield of 3a slightly dropped to 91% when the concentration was raised further to 1.0 M (Table 1, entry 13). The amounts of both base and nucleophile could be successfully decreased to 2 equivalents, and 3a was still obtained in 94% assay yield, 90% isolated yield (Table 1, entry 14). Further decreasing the equivalents of either component or the temperature resulted in lower yields (Table 1, entries 15–17). Therefore, the optimal transition-metal-free cross-coupling reaction was 1 equivalent of 1a as electrophile, 2 equivalents of methanol as nucleophile, and 2 equivalents of KOtBu as base in DME (0.5 M) at 110 °C for 12 h.

Table 1 Optimization for transition-metal-free cross-coupling reaction between 1a and 2aa.
Full size table

Substrate scope

Having identified optimal reaction conditions, we next explored the scope of substrates of aliphatic alcohols (Fig. 2). Methanol (2a) and ethanol (2b) were successfully utilized as coupling partners, generating 3a and 3b in 90% and 93% yield respectively. A secondary alcohol, isopropanol (2c), furnished the corresponding product 3c in good yield (82%). Aliphatic alcohols possessing phenyl substituents either at the α- or γ- position were well tolerated, providing 3d and 3e in 61% and 86% yield, respectively. The reactions proceeded smoothly between 1a and alcohols bearing cyclic substituents (2f, 2h), affording the corresponding products in good to excellent yields. Even sterically hindered 1-adamantylmethanol (2h) could be utilized and produced 3h in 80% yield. Moreover, alkyl alcohols with ether or amine functional groups were well tolerated and led to the formation of the desired products 3i-l in excellent yields. Remarkably, the potentially competitive amination did not occur at all with either primary amine 2k or secondary amine 2l. Our protocol could be utilized to functionalize more complex structures including natural products to afford 3m and 3n in excellent yields.

Fig. 2: Substrate scope alcohols with 1a.
figure 2

All reactions were conducted with 1a (0.2 mmol), 2 (0.4 mmol), KOtBu (0.4 mmol) in DME (0.4 mL) under argon atmosphere at 110 °C for 12 h.

Full size image

Next, the aryl group of the methyl sulfoxides was surveyed when cyclohexylethan-1-ol (2g) was employed as cross-coupling partner (Fig. 3). Phenyl methyl sulfoxide (1b) could successfully couple with 2g to afford the corresponding 4a in 76% yield under slightly modified conditions. Methyl sulfoxides bearing electron-withdrawing substituents such as cyano or trifluoromethyl groups underwent the cross-coupling reaction to afford compounds 4b and 4c in excellent yields. Substrates with electron-donating groups reacted more sluggishly; even so, 4d and 4e were successfully prepared using our method in modest to good yields. It is noteworthy that the para-thiomethyl group barely reacted, in contrast to prior reports involving transition metal catalysts23,28,29,30, allowing the formation of 4e and highlighting the unique chemoselectivity of our protocol. Steric hindrance on the aryl sulfoxide portion exerted negligible influence on our protocol as evidenced by the 1-naphthyl derivative 4f forming in 94% yield. Methyl sulfoxides possessing a polymerizable vinyl group (1h) or and electrophilic carbonyl group (1i) were also compatible, albeit giving 4g and 4h in modest yields. Alkyl heteroaryl ethers are ubiquitous in biologically relevant compounds40. Remarkably, a range of heteroaryl groups exhibited good to excellent compatibility under the optimal conditions; pyridyl (4i, 4j, 4k), isoquinolyl (4l), imidazolyl (4m), pyrimidinyl (4n), benzoimidazolyl (4o) as well as quinolyl (4p-t) ethers could be obtained in 66–99% yields. A small number of pyridyl sulfoxides have been found to undergo substitution with alkoxides, but reduction was an accompanying product41,42. Of note, bromo- or iodo-substituents on 4r-t were compatible with this method, providing a means for later modification via orthogonal transition-metal catalyzed coupling strategies. Significantly, this protocol could even be utilized to directly derivatize 2-(methylsulfinyl)-4-(2-thiophenyl)-6- (trifluoromethyl)pyrimidine (1v), a patented compound with anti-apoptosis bioactivity43, to deliver 4u in 72% yield. To establish the scalability of our transition-metal-free coupling procedure, a gram-scale reaction with 1q (5.5 mmol, 1.05 g) and 2g (11.0 mmol, 1.41 g) was performed, affording 4p in excellent yield (99%, 1.48 g).

Fig. 3: Substrate scope of aryl methyl sulfoxides with 2ga.
figure 3

aUnless noted, all reactions were conducted with 1 (0.2 mmol), 2 g (0.4 mmol), KOtBu (0.4 mmol) in DME (0.4 mL) under argon atmosphere at 110 °C for 12 h. b2g (3.0 equiv), KOtBu (3.0 equiv) at 110 °C for 24 h. c2g (3.0 equiv), KOtBu (3.0 equiv). d40 °C; e2g (3.0 equiv), KOtBu (4.0 equiv) at 100 °C for 30 h. fLarge scale reaction: 1q (5.5 mmol), 2g (11.0 mmol), KOtBu (11.0 mmol) in DME (11.0 mL) under argon atmosphere at 110 °C for 12 h. g1u (4.0 equiv), 2g (1.0 equiv), KOtBu (2.0 equiv) at 40 °C for 48 h.

Full size image

Synthetic applications

To illustrate the potential utility of a transition-metal free coupling reaction of aryl methyl sulfoxides with alcohols in medicinal chemistry and drug development, two drug candidates have been prepared employing this method as the key step (Fig. 4). A sphingosine 1-phosphate receptor modulator 6e, which is currently in phase II clinical trials for the treatment of multiple sclerosis44, could be synthesized in 5 steps and 20% overall yield. Notably, no transition metals are utilized in the entire route, simplifying the drug purification process. Another patented compound, opioid delta receptor agonist 7c45, could be generated in only 3 steps and 61% overall yield using our highly chemoselective transition-metal free coupling strategy as the key step.

Fig. 4: Synthetic applications.
figure 4

Application of transition-metal-free cross-coupling of aryl methyl sulfoxides with alcohols to pharmaceuticals.

Full size image

From the standpoint of atom economy, aryl methyl sulfoxides are the most appealing of sulfoxide coupling partners. Nevertheless, other alkyl aryl sulfoxides were also explored (Table 2). Ethyl 2-naphtyl sulfoxide (5a) proved to be a suitable coupling partner, despite affording 3a in lower yield (Table 2, 5a). As expected, the other three alkyl sulfoxides under investigation (5b–d) only generated trace amounts of 3a, presumably owing to rapid formation of sulfenate anions after attack of methoxide on the sulfur46,47,48 (Table 2, 5b–5d).

Table 2 Transition-metal-free cross-coupling of alkyl aryl sulfoxides with methanola.
Full size table

To understand the mechanism, a series of control experiments were performed (Table 3). When 2.0 equiv of TEMPO was introduced under the standard reaction conditions, the yield of 3a was not significantly affected (see Supplementary Information for details), which points away from a radical pathway. Next, KOtBu was replaced by two other tert-butoxide bases (LiOtBu, NaOtBu) in the transition-metal-free cross-coupling protocol, with the rest of the conditions kept the same. Despite the similar basicity of these reagents, the yields of 3a were dramatically decreased, which illustrates the crucial role of the potassium cation (Table 3, entries 1–2). To further confirm the effect, 2.0 equivalents 18-crown-6 were introduced into the otherwise standard conditions, and the yield of 3a dropped precipitously to 5% (Table 3, entry 3). In a subsequent study, 2.0 equivalents of potassium fluoride added to the reactions with LiOtBu or NaOtBu restored the reactivity providing 63% and 27% yield, respectively (Table 3, entries 4–5). These control experiments support a pivotal role of potassium cation in the transition-metal-free cross-coupling49,50. Hence, a computational study was conducted to fully understand this unusual C-S bond activation mode for methyl sulfoxides that does not require a transition metal catalyst.

Table 3 Influences of bases and additives on the transition-metal-free cross-couplinga.
Full size table

Mechanistic studies

Computational studies focused on the reaction of 1a with methoxide using different cations (K, Na and Li). DFT calculations were performed using Gaussian 09. All geometry optimizations and vibrational frequency calculations of stationary points and transition states (TSs) were carried out at the B3LYP level of theory with 6-31+G(d) basis set for all atoms. Single point energies and solvent effects in 1,4-dioxane were computed at the M06-2X level of theory with 6-31+G(d,p) basis set for all atoms, using the gas-phase optimized structures. Solvation energies were calculated by a self-consistent reaction field (SCRF) using the SMD solvation model.

Several possible mechanisms were taken into consideration (Fig. 5). The first proposal (Fig. 5a), involves a SNAr reaction between the aryl methyl sulfoxide and the alkoxide, initiated by a π-cation interaction between the potassium and the aromatic ring51,52,53. A second proposal (Fig. 5b) proceeds via concerted addition of the alkoxide and elimination of the methyl sulfoxide group. The third mechanistic possibility involves nucleophilic addition of the alkoxide to the methyl sulfoxide facilitated by coordination of the potassium cation to the aromatic ring and the oxygen from the sulfoxide group. The resultant intermediate can subsequently undergo cyclization via attack of the oxygen atom on the ipso carbon. The final step entails the elimination of the methyl sulfenate to afford the desired product (Fig. 5c).

Fig. 5: Possible mechanistic pathways.
figure 5

The potassium assisted C–S bond activation of aryl methyl sulfoxides (a-c).

Full size image

The first mechanistic proposal was quickly found to be untenable due to far more favorable coordination of the cation to the sulfoxide oxygen rather than via a π-cation complex. Intermediates and transition states for the second mechanistic possibility were located. However, the computational energies indicate that the potassium counterion should decrease the reactivity relative to sodium and lithium (see Supplementary Fig. 1), which is the opposite of what was observed experimentally.

In contrast, the third mechanism (Fig. 5c) accounts for the experimentally observed cation effect (Fig. 6). Calculations suggest that the reaction proceeds via a nucleophilic attack of the alkoxide to the sulfoxide, which will lead to coordination of the counterion with the C1 carbon and formation of a three membered ring via attack of the alkoxy oxygen to the ipso carbon of 1a to afford INT1. This intermediate can then undergo an elimination step via TS1 to afford the desired product and methyl sulfenate. The energy barrier to obtain the desired product is significantly lower for potassium than for sodium or lithium. Calculations suggest that the nature of the cation will affect both the stability of INT1 and the subsequent elimination of the methyl sulfenate. Experimental observations showed that the use of LiOtBu completely inhibits reactivity, and that NaOtBu significantly decreases the yield (Table 3, entries 1–2), further supporting the computational findings. Since this mechanism relies on initial nucleophilic attack of the sulfoxide, it also consistent with more electrophilic substrates reacting more quickly (Table 2, 5b and 5c) than those with electron donating groups (Table 2, 5d). For heterocycles with a nitrogen adjacent to the sulfoxide (4i, 4j, 4m, 4n, 4o, 4u), reactivity may be further assisted by bidentate coordination of the potassium to the heterocyclic nitrogen and the sulfoxide oxygen. However, the lack of any distinct trends in the outcomes for these substrates relative to heterocycles with more distal nitrogens (4k, 4l, 4p, 4q, 4r, 4s, 4t) indicates that this effect does not predominate.

Fig. 6: Free energy profile for the reaction of 1a with methoxide in presence of different counterions (M = Li, Na and K).
figure 6

Relative free energy values were calculated with SMD-1,4-dioxane-M06-2×/6-31+G(d,p)// B3LYP/6-31+G(d).

Full size image

Discussion

In summary, a unique transition-metal-free cross-coupling protocol utilizing aryl methyl sulfoxides as electrophilic partners with alcohols to afford alkyl aryl ethers is described. An array of functional groups is well tolerated, and the method can be used for late-stage functionalization of natural products and pharmaceuticals. The methyl sulfoxides utilized as coupling partners in our protocol were generally prepared by methylation of corresponding thiophenol derivatives, followed by oxidation by NaIO4 or m-CPBA. Considering that thiophenol derivatives are typically manufactured by reduction of benzenesulfonic acid or benzenesulfonyl chloride, the method developed herein provides orthogonal entries relative to halides or alcohols, which permits different functional group compatibility and different substitution patterns. The successful application of the transition-metal free coupling reaction reported herein in drug syntheses and derivatization highlights its potential utility in medicinal chemistry as well as drug development. Both experimental results and a DFT-based computational study suggest that the reaction proceeds via a nucleophilic addition of the alkoxide to the sulfoxide accompanied by a cyclization. Subsequently, the alkoxide migrates to the ipso carbon while eliminating the methyl sulfenate. The dramatic impact of the cation on the reactivity arises both from greater stabilization of INT1 via association with the C1 carbon and enhanced ability to fragment via TS1. These results wherein an electrophilic partner is activated by addition of a nucleophile that both enhances its ability as leaving group and activates the aromatic substrate via coordination of a counterion open up possibilities for electrophile activation and coupling in general. The concepts described herein also provide a basis for the construction of systems to allow transformations of organosulfur compounds in other contexts.

Methods

General procedure for catalysis

To an oven-dried microwave vial equipped with a stir bar was added KOtBu (44.9 mg, 0.4 mmol, 2 equiv), and 2-methanesulfinyl-naphthalene (38.0 mg, 0.2 mmol, 1 equiv) under argon atmosphere in a glove box. DME (0.4 mL) was added to the vial by syringe. The microwave vial was sealed and removed from the glove box. Then, methanol (16.2 μL, 0.4 mmol, 2 equiv) was added by syringe under argon atmosphere. Note that solid and viscous oil alcohols were added to the reaction vial prior to KOtBu. The reaction mixture was heated to 110 °C in an oil bath and stirred for 12 h. Upon completion of the reaction, the sealed vial was cooled to room temperature, and opened to air. The reaction mixture was passed through a short pad of silica gel. The pad was then rinsed with 10:1 dichloromethane:methanol. The resulting solution was subjected to reduced pressure to remove the volatile materials and yielded a viscous oil. The residue was purified by flash chromatography as outlined below.