Introduction

Disulfides are widely utilized across various disciplines, including biology1, food chemistry2, and pharmaceutical sciences3. Of particular note is the pharmaceutical industry’s effective employment of S-S bond linkers in formulating a range of therapeutics aimed at treating diabetes, alcoholism, and cancer, through their incorporation into antibody, small molecules and peptides (Fig. 1A)4,5,6,7,8.

Fig. 1: Development of a perthiyl radical generation strategy for the synthesis of unsymmetric disulfides.
figure 1

A Representative biorelevant disulfide compounds. B Strategies for unsymmetric disulfides construction and perthiyl radical generation. C This work: one-pot unsymmetrical disulfides synthesis via perthiyl radicals addition.

Full size image

Traditional methods to prepare disulfides primarily rely on the disulfide exchange or the couplings of two sulfur-containing compounds. Contemporary strategies for disulfide synthesis often employ disulfurating reagent (RSS−LG) to forge C-SS bonds, with persulfide cations (RSS+) or persulfide anions (RSS−) serving as crucial intermediates9,10,11,12,13,14,15,16,17,18,19,20,21 (Fig. 1B). However, there have been no reports on a method for directly synthesizing unsymmetrical disulfides using perthiyl radicals (RSS•)22,23,24,25. To date, several methods have been explored for generating perthiyl radicals (Fig. 1B), and one of the earliest methods involved the thermal decomposition of tetrasulfides26,27,28,29. However, the reverse reaction, i.e., the dimerization of RSS•, occurs with minimal or no activation energy, resulting in low overall conversion30. An alternative approach involves carbon radical substitution on tetrasulfides31,32. Additionally, the hydrogen-atom-transfer (HAT) process emerges as a powerful and effective strategy for radical production, as the relatively weak RSS–H bond dissociation enthalpy (~70 kcal/mol) enables the oxidation of hydropersulfides to RSS• by mild oxidants33,34,35. Another potent and efficient strategy is photochemistry, which activates S-Cl bonds in RSSCl to form perthiyl radicals26. Unfortunately, these methods for generating perthiyl radicals often require stringent experimental conditions and can only be applied to the pre-synthesized polysulfides. Furthermore, the preparation of polysulfides is often complex and time-consuming. Hence, the development of a versatile and convenient approach for generating perthiyl radicals in a single reaction remains a significant priority. This will not only greatly simplify the process of preparing unsymmetric disulfides but also improve synthesis efficiency.

On the other hand, sulfur dioxide (SO2) is widely used in the pharmaceutical and organic chemical synthesis sectors36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51,52,53,54,55,56,57,58,59,60,61,62,63,64,65,66. In recent years, our research group has focused on the versatile chemical properties of SO2 and developed several synthetic methods in which are carbon radicals capture SO2 molecule to form sulfonyl radicals. Building on our previous work involving sulfonyl radicals, the objective of this study is to develop a general approach using SO2 as the sulfur source to obtain perthiyl radicals for the synthesis of unsymmetric disulfides from easily accessible starting materials in one pot. To achieve this desired reaction, our hypothesis is that a sulfonyl radical can be reduced to a sulfur radical, which can then capture the second SO2 molecule and undergo subsequent reduction to form the target perthiyl radical (Fig. 1B). Implementing of this protocol presents several challenges: (1) A suitable reducing agent which can effectively reduce sulfonyl radicals to sulfur radicals is in need. (2) The sulfur radicals have to capture SO2 specifically. (3) Perthiyl radicals are thermodynamically stable and readily form dimers, making the search for a suitable perthiyl radical receptor an intriguing task. (4) The question arises as to whether the perthiyl radicals would continue to capture one more SO2 molecule. (5) Sulfur radical addition products and perthiyl radical homocoupling products may be produced instead of the desired disulfides.

In this study, perthyl radicals are generated from easily accessible alkyl halides and SO2, enabling the one-step synthesis of unsymmetric disulfides. This process entails two instances of C–S bond formation and one instance of S–S bond formation (Fig. 1C).

Results

Reaction optimization

The disulfide reaction was initially investigated by studying the reaction between the model reactants 1-bromocyclopentane 1, ethyl acrylate 2 and ex-situ generated SO2 gas (from SOgen)36 (Table 1). The optimized conditions were: PhSiH3 (3.2 equiv) as reductant, HCOOLi (3.0 equiv) as the base, a 1:2.5:2.5 ratio of 1-bromocyclopentane to ethyl acrylate to SO2, 2.0 mL mixture of NMP-DMSO (1:1) as solvent and 40 °C as the reaction temperature. Under these conditions, product 3 was obtained in 72% isolated yield (entry 1, Table 1). The reaction did not occur in the absence of PhSiH3 or HCOOLi (entries 2–3). Substituting PhSiH3 with Et3SiH as the reductant resulted in a substantial reduction in the product yield (entry 4), while the use of other reductant sources such as PMHS and NaBH4 directly suppressed the reaction (entries 5–6). It was observed that the choice of base had a significant impact on the reaction. Replacing HCOOLi with K2CO3, DABCO, and HCOOK led to a dramatic decrease in yields (entries 7–9). In addition, two alternative SO2 surrogates and S8 were explored. Using S8 instead of SO2 resulted in product yields of 49% (entry 10). The use of DABSO resulted in a significantly reduced yield (entry 11), while the utilization of another inorganic SO2 surrogate, namely K2S2O5 (entry 12), completely hindered the conversion. Furthermore, the choice of solvent also influenced the reaction outcome. The use of DMSO or NMP resulted in product yields of only 60% and 43%, respectively (entries 13–14). Instead, a mixture of 1:1 DMSO and NMP was found to be the optimal solvent for the reaction.

Table 1 Optimization of the reaction conditionsa
Full size table

Substrate scopes

With the optimal reaction conditions identified, we proceeded to explore the generality of the reaction. First, the scope with respect to the alkyl bromides was investigated (Fig. 2). We observed that both linear and α-branched primary alkyl bromides could be successfully converted into the corresponding disulfides (4-8). The reaction exhibited a high chemoselectivity for activating alkyl bromides over alkyl fluorides, as evidenced by the formation of product 9. In addition, the alkyl bromides bearing remote functional groups, such as benzyl ether and trimethylsilane, were found to be compatible with the reaction conditions (10,11). Moreover, alkyl bromides bearing aromatic rings, such as non-substituted and substituted phenyl and naphthyl rings, proved to be suitable substrates in the transformation (12-16). Notably, we found that heterocycle substituents, such as N-protected pyrrolidines, N-protected piperazine, thiophene, were well-tolerated on the alkyl bromides (17-19). Additionally, substrates containing unsaturated propargyl and allyl derivatives underwent the conversion smoothly, providing the desired products (20, 21). Next, secondary alkyl bromides were tested in the reaction. Apart from the model product 3, some secondary alkyl bromides with various carbon chain length could lead to the desired products (22-25). Interestingly, various secondary benzyl bromides were also suitable substrates, successfully providing products 26-28. In addition, two examples of α-carbonyl alkyl bromides went through the transformation and gave the target disulfides (29, 30). To our delight, tertiary alkyl bromides with large steric hindrance were also proved amenable for the reaction (31-33), indicating that this strategy had potential for the construction of hindered unsymmetrical disulfides.

Fig. 2: Substrate scope of alkyl bromides.
figure 2

Reaction conditions: alkyl bromides (0.2 mmol, 1.0 equiv), α, β-unsaturated esters (0.5 mmol, 2.5 equiv), SO2 (0.5 mmol, 2.5 equiv), PhSiH3 (0.64 mmol, 3.2 equiv), HCOOLi (0.6 mmol, 3.0 equiv), DMSO (1.0 mL) and NMP (1.0 mL), at 40 °C for 9 h under argon atmosphere. Isolated yields were reported. std. cond. = standard conditions.

Full size image

Continuously, a series of alkyl iodides were employed in the reaction in place of the model reactant alkyl bromides (Fig. 3). Pleasingly, disulfide products 4, 7, 11, 12, 23, and 27 which could be prepared from primary alkyl bromides were also successfully prepared from alkyl iodides. Primary alkyl iodide with longer carbon chain was tolerated in the reaction, providing the target product 34. Two secondary alkyl iodides also went the reaction smoothly and produced the desired products 23 and 27. In the same way, alkyl chlorides were used in place of bromide analogs (Fig. 3). Alkyl chlorides containing terminal alkene, terminal alkyne or allyl group could afford the corresponding products 21, 35 and 36 respectively. Some secondary alkyl chlorides were also suitable substrates for this protocol (23, 26, 27). It was gratifying that pseudohalides were compatible with the reaction conditions. Various alkyl tosylates could be converted into the corresponding disulfides in 45%-69% yields (Fig. 3). For example, products 4, 12 were obtained again by using primary alkyl tosylates instead of the primary alkyl bromides. This protocol also had good tolerance for alkyl tosylates containing aromatic rings decorated with various substituents (37-41). Tosylate substrates containing heterocycles and terminal alkene were also compatible with this transformation (42-44). Both linear (45, 25, 22) and cyclic (3, 46, 47) secondary alkyl tosylates successfully underwent the desired transformation.

Fig. 3: Substrate scope of alkyl iodides, alkyl chlorides and alkyl tosylates.
figure 3

Reaction conditions: alkyl electrophile (0.2 mmol, 1.0 equiv), α, β-unsaturated esters (0.5 mmol, 2.5 equiv), SO2 (0.5 mmol, 2.5 equiv), PhSiH3 (0.64 mmol, 3.2 equiv), HCOOLi (0.6 mmol, 3.0 equiv), DMSO (1.0 mL) and NMP (1.0 mL), at 40 °C for 9 h under argon atmosphere. Isolated yields were reported. std. cond. = standard conditions.

Full size image

We next began to investigate the scope of alkene derivatives for the reaction (Fig. 4). α, β-unsaturated esters (48-56) were all tolerated under the standard experiment conditions. Interestingly, excellent regioselectivity of the reaction was demonstrated by using vinyl acrylate as a reactant (55). To our delight, vinyl pyridine could be used as an effective perthiyl radical acceptor (57). α-aryl substituted α, β-unsaturated ester was also proved to be a suitable substrate, providing product 58 in moderate yield. Additionally, α, β-unsaturated amide could readily participate in the reaction (59). Furthermore, we aimed to showcase the practical applications of this strategy by utilizing pharmaceuticals and natural products as reaction substrates for modification purposes. As a results, disulfides derived from Diacetonefructose (60), Estrone (61), Paroxetine (62), L-Menthol (63) and DL-α-Tocopherol (64) were successfully prepared respectively.

Fig. 4: Substrate scope of alkenes and coupling with natural products or pharmacologically active molecules.
figure 4

Reaction conditions: alkyl bromides (0.2 mmol, 1.0 equiv), α, β-unsaturated esters (0.5 mmol, 2.5 equiv), SO2 (0.5 mmol, 2.5 equiv), PhSiH3 (0.64 mmol, 3.2 equiv), HCOOLi (0.6 mmol, 3 equiv), DMSO (1.0 mL) and NMP (1.0 mL), at 40 °C for 9 h under argon atmosphere. Isolated yields were reported. std. cond. = standard conditions.

Full size image

Mechanistic investigations

To shed light on the nature of this reaction system, radical trapping experiments were carried out (Fig. 5A). When the reaction was conducted without ethyl acrylate, the TEMPO-carbon radical adduct compound 65 was detected. When TEMPO was added to the standard reaction in which SO2 was added, the target disulfide 3 was not observed and the S-centered radical was efficiently trapped by TEMPO (66). Moreover, when TEMPO was added after 20 min of the reaction, perthiyl radicals and sulfur radicals were generated and trapped, proved by the detection of 66 and 67. Next, the parallel kinetic isotope effect (KIE) experiments (from initial rate) showed that the kH/kD value was 4.43, and the competitive KIE was determined to be kH/kD = 2.57, indicating that the protonation step was involved in the turnover-limiting step (Fig. 5B). In the template reaction, we have identified two side products: disulfide compound 68 with a yield of 9% and trisulfide compound 69 with a yield of 5%. Subsequently, we performed a control experiment under standard reaction conditions, excluding the addition of ethyl acrylate. Dimers of disulfides 68, trisulfides 69, and tetrasulfides 70 were identified, with yields of 32%, 25%, and 18%, respectively. To gain more insight into the reaction mechanism, symmetrical disulfide 68, trisulfide 69, or tetrasulfide 70 was utilized in place of reactant 1 in reactions with reactant 2, all eventually generating product 3 in yields of 50%, 53%, and 45%, respectively. Additionally, trace amount of thioether 3” was detected in the reaction between compound 68 and ethyl acrylate. In the template reaction, 1.0 equivalent of SO2 was used instead of 2.5 equivalents, and as a result, trace amount of thioether product 3” was detected by GCMS. These results indicate the generation of sulfur radicals and demonstrate that there is an interconversion between the sulfur radicals and compound 68 throughout the reaction process (Fig. 5C).

Fig. 5: Key mechanistic findings.
figure 5

A Radical-trapping experiments with TEMPO; B KIE experiments; C Intermediate experiments; D Proposed mechanism. std. cond. = standard conditions.

Full size image

Proposed mechanism

On the basis of the initial mechanistic studies, a plausible reaction mechanism is proposed herein (Fig. 5D). Initially, the key species I may be generated from the interaction between alkyl bromide and a radical initiator which generated from the silane. Secondly, the alkyl radical I is trapped by SO2 to form the alkylsulfonyl radical species II, which would undergo fast deoxygenation, giving thiyl radical intermediate III, which has an equilibrium with compound 68. Intermediate III could capture an additional molecule of SO2, and then undergoes the same deoxygenation process to yield perthiyl radical IV. IV may either couple with III or undergo homocoupling to form compounds 69 and 70, respectively. This process is reversible. Then, perthiyl radical adds to 2 and generates V which then undergoes HAT with PhSiH3 to produce the disulfuration product 3.

DFT studies

To gain more in-depth insights into the reaction mechanism, density functional theory (DFT) calculations were conducted (Fig. 6). Silicon radical formation through PhSiH3 is the initiation step67,68,69. First of all, we found that the base additive (HCOO−) enables a much more facile reaction pathway. Its binding with PhSiH2 radical is exergonic by -2.8 kcal/mol, resulting in a new radical intermediate PhSiH2-HCOO•−, which is a quite strong reductant (Eox = −2.51 V versus NHE). In addition, the silane-formate complex, PhSiH3-HCOO−, exhibits a significantly elongated Si–H bond (1.57 angstrom, versus 1.48 angstrom in PhSiH3), indicating that it may act as a superstrong hydrogen donor. Then, the halogen-atom transfer (XAT) process between 1 and PhSiH2-HCOO•− resulting in the formation of alkyl radical (INT1) is determined to have a free energy barrier of 14.5 kcal/mol (TS1). Following that, SO2 is rapidly trapped by INT1 without encountering any energy barrier (Supplementary Fig. 8). We have calculated the barrier for the addition of INT2 to alkene, which is 16.6 kcal/mol. As the reduction of INT2 is barrierless, the addition will not form a competitive side product pathway. The hydrogen transfer from PhSiH3-HCOO− to INT2 is barrierless, resulting in the formation of INT3 and PhSiH2-HCOO•−. Then they immediately undergo a single electron transfer event, and the so-generated radical anion INT4 and PhSiH2OOCH rapidly re-associate to afford INT4’. Then, the INT4’ undergoes deoxygenation through the cleavage of the S–OH bond, hydrogen abstraction, and sequential radical elimination to produce sulfur radical (INT6) and poly(phenylsiloxanes), which may exist in linear or cyclic (INT5, detected by HRMS) forms. Its free energy profile has been explored in Supplementary Fig. 9. Next, we have calculated the barrier for the addition of INT6 to alkene, which is 13.8 kcal/mol, as well as determined that the homocoupling of INT6 occurs barrierlessly. However, as shown in the energy profile, once INT6 is transformed into INT7, it is barrierlessly reduced, furthermore, the energy barrier required for the irreversible transformation from INT6 to INT7 is 11.2 kcal/mol. As a result, the thioether product will not form. Considering that compound 68 may represent an important resting state during the reaction, we further investigated the relevant mechanisms of the transformation of compound 68, as shown in Supplementary Fig. 10. Subsequently, INT7 is reduced barrierlessly by PhSiH3-HCOO−, after which INT8 undergoes a single-electron transfer with a free energy barrier of 7.9 kcal/mol. And then, INT9 and PhSiH2OOCH rapidly re-associate to afford INT9’. Next, INT9’ undergoing deoxygenation is similar to that of transforming INT4’ into INT6, resulting in the formation of perthiyl radical (INT10), as shown in Supplementary Fig. 9. We have found that the association of INT10 with the third equivalent of SO2 unable to give any minimum on the potential energy surface, and thus further combination with SO2 is unrealistic (Supplementary Fig. 11). The perthiyl radical has the ability to undergo addition to ethyl acrylate (2), resulting in the formation of a carbon-centered radical (INT11) with a free energy barrier of 20.0 kcal/mol (TS2). Subsequently, this newly formed radical abstracts a hydrogen atom from PhSiH3-HCOO− through transition state TS3 with an activation free energy of 26.1 kcal/mol to yield the desired product 3 and PhSiH2-HCOO•− which participates in the propagating chain process. The overall barrier is 26.1 kcal/mol, as determined by the hydrogen abstraction of the radical INT11 via the transition state TS3. Overall, the pathway calculated by DFT is consistent with the assumption described in Fig. 5D.

Fig. 6: The free energy profile for the alkene hydro-disulfuration reactions.
figure 6

All molecular geometries were optimized at the M06-2X/6-31 + G(d) level of theory using SMD model to account for solvation energies in DMSO.

Full size image

In summary, we have developed a unified protocol for hydro-disulfuration of alkenes with alkyl electrophiles under mild conditions. This reaction uses readily abundant alkyl electrophiles and alkenes as starting materials, providing straightforward access to unsymmetrical disulfides with excellent functional group tolerance. It is demonstrated as an effective example of the synthesis of unsymmetrical disulfides by using SO2 as a divalent sulfur source. Phenylsilane serves both as the radical initiator and reductant. This research provides a convenient method for synthesis of unsymmetrical disulfides which are challenging to access. Further investigations will focus on expanding the perthiyl radical acceptor scope and finding widespread application for this disulfuration strategy.

Methods

General procedure for the synthesis of compounds 3-64: In an argon fulfilled glovebox, perbromothiophene 1,1-dioxide (219.8 mg, 0.51 mmol, 2.5 equiv), tetradecane (1.0 mL), were added into chamber B with a magnetic stirring bar, followed by addition of 1-methyl-4-vinylbenzene (59.3 mg, 0.5 mmol, 2.5 equiv). HCOOLi (0.6 mmol, 31.2 mg), alkyl electrophiles (-Cl/-Br/-I/-OTs) (0.2 mmol) and alkenes (0.5 mmol) were added into chamber A, then anhydrous NMP (1.0 mL) and DMSO (1.0 mL) was added into chamber A. Then, PhSiH3 (0.64 mmol, 79 μL) was added into chamber A by microsyringe. The two-chamber was sealed and removed out of the glovebox. The chamber B was allowed to stir at 100 °C using heating mantle with 600 rpm stirring speed for 10 min (chamber A was not allowed to stir). After then, the chamber A was allowed to stir at 40 °C using heating mantle with 300 rpm stirring speed for 9 h. Upon completion, the reaction was quenched with water (30 mL) and extracted with EA (30 mL). The extract was washed by brine (15 mL) and dried over anhydrous Na2SO4. After filtering, the filtrate was concentrated, and the residue was purified by flash silica gel column chromatography using petroleum ether/ethyl acetate as eluent to afford pure products 3-64.