Disulfide bonds are prominent building blocks in molecules with multiple biological activities [1-3]. Moreover, the unique and flexible characteristics of disulfides have enabled their utilization in many fields; for example, the ability to modify an intra/intermolecular construction [4, 5] highlighted the importance of disulfide bridges in self-assembled medicine. Disulfide linkages are also popular as dopants in technical rubber and polymeric materials [6, 7] where they enhance the self-healing characteristic of polymers. Therefore, numerous advances in the synthesis of disulfides have been made during the past decades. Of the available strategies for the synthesis of disulfides, the oxidation of thiols still dominates this field. While classical strategies achieved the desired transformations smoothly in the presence of stoichiometric oxidants (such as H₂O₂ [8], halogens and derivatives [9], and sulfoxide [10]), they still suffered from the drawback of waste generation. This issue could be resolved by the use of air or molecular oxygen (O₂) as the terminal oxidant [11-14]. Thermally promoted pathways have been investigated to overcome the drawbacks of low activity and selectivity stemming from activation of O₂. The functional catalyst material displayed outstanding reaction activity in heterogeneous systems, but they held the disadvantage of noble material participation [15-17] or thermal deactivation of catalytic nanoparticles [18]. As for homogeneous catalysis, most of which involved organic compound catalysts [19-21], elevated temperature was necessary. However, many biologically active molecules are very sensitive to high temperature and thus, their modification with disulfide bonds should proceed at room temperature. In contrast to thermal catalysis, the visible-light photocatalytic pathway was superior in fulfilling reactions at room temperature in a green and energy-saving manner [22-24], particularly when light was introduced into heterogeneous photocatalytic schemes [25].

Recently, heterogeneous photocatalysis has developed into a promising route for organic synthesis [26-33]. Typically, TiO₂ has displayed great potential in heterogeneous photocatalysis [34]; specifically, TiO₂ responds to visible light as a versatile platform to realize some conversions via complexation with heteroatoms (including O, N, and S) [35]. In fact, the non-covalent interactions give rise to visible-light absorption and activation; this is a new frontier in visible-light-promoted transformations [36]. Complexation caused a reduction in the oxidation potential of TiO₂ and an upshift in its highest occupied molecular orbital (HOMO). Subsequently, under light irradiation the excited electron of the heteroatom would be efficiently injected into the conduction band of TiO₂ through weak coordination, followed by transfer to the electron acceptor, along with the release of a positive charge (h⁺). Therefore, both of the highly active species are suited for the reductive or oxidative pathways in selective synthesis. In a previous report [37], we successfully realized the aerobic oxidation of amines into imines by a complexation system, in which the complex was afforded by anatase TiO₂ and amines under visible-light irradiation (Scheme 1(a)). Subsequently, we were delighted to find that the complexation strategy, occurring between P25 TiO₂ and triethylamine (TEA) to form a visible-light-harvesting surface complex, could also efficiently facilitate the aerobic oxidation of sulfides into sulfoxides (Scheme 1(b)) [38]. In these two instances, O-atom transfer from O₂ to substrates facilitates the visible-light-induced selective formation of oxidation products on TiO₂. It would be of great interest to identify new types of substrates and enrich the role of O₂ in TiO₂ visible-light photocatalysis.

Herein, we sought to establish an attractive alternative for the visible-light-promoted selective oxidation of thiols to disulfides with O₂ on anatase TiO₂. The facile formation of disulfides could be accomplished with no additives and minimal product contamination. In our two previous reports, only visible light around 400 nm could be utilized. In this work, green LEDs at 520 ± 10 nm were explored as the light source. More importantly, O₂ was found to act as the acceptor for electrons and protons rather than to incorporate into the substrates. The transformation was extremely efficient for the selective oxidation of various thiols, especially with substrates bearing electron withdrawing groups (less than 10 min). This protocol has been proven to be an energy-saving and atom-economic route to green chemistry.

We tested the new issue with 4-methylbenzenethiol 1a as the substrate, TiO₂ (ST-01) as the photocatalyst, and CH₃CN as the solvent, and maintained the reaction mixture at an initial O₂ pressure of 0.1 MPa. The following tests revealed that TiO₂, O₂, and visible light were all indispensable components of the reaction protocol (entries 1–3, Table 1). The results obtained after 20 min of visible-light irradiation are shown in entry 4, Table 1. It was determined that the transformation from thiol to disulfide 2a proceeded with 92% conversion and excellent selectivity under 520 nm green LED irradiation. It was hypothesized that the reaction exhibited an efficient complexation wherein the thiol substrate chemisorbed onto the surface of TiO₂ through the S atom in the initial stage, following which the application of mild visible light induced the subsequent reaction. In effect, UV-vis analysis (Fig. 1) indicated that the formation of the 4-methylbenzenethiol-TiO₂ complex could significantly shift its light absorption to the visible-light region as compared to the uncomplexed TiO₂ and thiol 1a. When CH₃OH was employed as the solvent (entry 5, Table 1), the transformation declined with a moderate conversion of 74%. This might be due to the interaction between CH₃OH and the substrate/TiO₂, resulting in a noticeable effect on the surface complexation between the substrate and TiO₂, whereas the polar aprotic solvent CH₃CN had a negligible effect on complexation.

Next, we investigated the effect of different types of TiO₂ on the conversion of thiol to disulfide (Table 2). To this end, 5 types of TiO₂ were tested; it was clear that (ST-01) TiO₂ displayed greater activity than the other types. Noël and co-workers [39] disclosed that P25 TiO₂ was applied as a photocatalyst, aided by base, for this reaction in a previous report. However, anatase TiO₂ with a high specific surface area can deliver nearly 5 times greater conversion of 4-methylbenzenethiol without the need for base. The ideal amount of TiO₂ was determined to be 30 mg (See details in Supplementary Data, the TEM characterization for P25 and ST-01 TiO₂, Fig. S1). Note that the reaction could be carried out with air as the oxidant; however, due to the odor of the substrates, O₂ with initial pressure of 0.1 MPa was selected as the oxidant so that the reaction proceeded in a closed system.

With the optimal reaction conditions established, the scope of applicable thiols was examined (Table 3). When 4-methylbenzenethiol served as the substrate, the oxidative coupling process was completed in 30 min, as monitored by thin-layer chromatography (TLC), and gave a 96% isolated yield. Other derivatives possessing electron-donating groups on the benzene ring (entries 2–7, Table 3) and benzenethiol (entry 8, Table 3) were also implemented in this protocol with excellent yields (> 90%). When considering the influence of the position of the substituent groups (entries 3–5, Table 3) on reaction efficiency, steric hindrance was the main factor necessitating prolonged irradiation time for thiol 1f. Analogous result was seen for the disubstituted thiol 1g (entry 7, Table 3). As for the thiols with electron-withdrawing groups (entries 9–12, Table 3), to our surprise, the transformations were completed in several minutes (less than 10 min) with high isolated yields and functional group tolerance. The protocol was also suitable for the hetero- and poly-aromatic thiols (entries 13 and 14, Table 3). From phenylmethanethiol, an alkyl mercaptan, (entry 15, Table 3) the disulfide 2o was obtained with an inferior isolated yield of 44% after 6 h irradiation. However, this sluggish reaction was greatly improved when triethylamine (TEA) was introduced at the beginning of the reaction as a base additive, affording the target product at a yield of 91% in 40 min.

Unlike symmetric disulfides, asymmetric species were inherently challenging to synthesize. Additionally, the synthesis of some of the asymmetric disulfides was accomplished through this heterogeneous system with the aid of organic base (TEA). We did some investigations of this issue, employing 1a and t-butylthiol 3 as the reagents (Table 4). When proceeding through the above optimized conditions, symmetric disulfide 2a was the sole product. However, in the presence of TEA, the photocatalytic reaction exhibited the unique ability to selectively yield the corresponding asymmetric disulfide 4a, with a 97% isolated yield, in just 5 min. The products obtained from other starting materials, 4b, 4g, 4i, and 4k, indicated that the electronic or steric effects exhibited minimal loss to the present system, as seen by their corresponding yields. In the case of 4k, we also examined 2k as a substrate under dark conditions lacking TiO₂ wherein 4k was found at 98% yield. This finding suggested that the symmetric disulfide was not thermodynamically stable and therefore could transform into the asymmetric disulfide in the presence of mercaptan 3 and base. As for product 4m, the electrical properties of the N atom had a negative effect on the nucleophilic substitution between 2m and 3. Finally, this asymmetric synthesis was also compatible with thiol 1n.

A plausible mechanism for the visible-light photocatalytic selective oxidation of thiols to disulfides on TiO₂ is depicted in Scheme 2. It was assumed that prior to photocatalysis, the thiol, a heteroatom substrate, was easily adsorbed onto the surface of TiO₂ via weak coordination at the initial stage. Then, visible-light irradiation motivated an electron from the thiol to be transferred to the conduction band (cb) of TiO₂ while circumventing the valence band (vb). At the same time, sulfur radical A and H⁺ were released. The unpaired electron was easily trapped by O₂ to form the superoxide radical anion O₂^·-. Subsequently, a three-component transient state, involving O₂^·-, radical A, and a thiol molecule, was formed. This state then transformed into product B and anion C. It should be noted that the thiol bearing an electron-withdrawing group was beneficial for the electrophilic attack on the electron rich cloud density of radical A. Thus, it was rational that thiols bearing electron-withdrawing groups displayed greater activity than those possessing electron-donating groups. Finally, anion C combined with H⁺ to form H₂O₂. The synthesis of asymmetric disulfides proceeded smoothly, wherein mercaptan 3 rapidly performed a nucleophilic attack on the newly formed symmetric disulfide with the assistance of TEA.

In summary, we have developed a visible-light-induced strategy for the aerobic oxidation of thiols to disulfides. In this context, anatase TiO₂ served as a multifunctional platform, not only for complexation with thiols, but also for electron transfer to form O₂^·-. When various thiols were examined, the method exhibited excellent reaction selectivity, functional group tolerance, and substrate scope with high isolated yields (> 90%). Notably, the substrates with electron-withdrawing groups could fulfil the transformation in less than 10 min. The concise design contributed to the atom-economic and energy-saving concepts of sustainable chemistry. Simultaneously, it was also an efficient strategy for the synthesis of some asymmetric disulfides with TEA as an additive.