In the past few decades, there has been a growing concern about the environmental pollution issue. The sulfur compounds in diesel can be transformed into SO_xupon burning, causing serious environmental pollution [1, 2]. Therefore, the production of fuels with very low levels of sulfur compounds is a grand challenge for diesel producers to meet stringent national requirements. Hydrodesulfurization (HDS) has been regarded as the most efficient diesel desulfurization process to eliminate sulfides and disulfides, but is less effective for removing 4, 6-dimethyldibenzothiophene and its derivatives with large molecular sizes. In addition, this process usually demands high temperature and pressure [3–6].

Oxidative desulfurization (ODS) is an alternative or supplementary system to HDS for deep desulfurization, which includes the oxidation of sulfides and the extraction separation of oxidation products. It is one of the most effective methods to dispose of aromatic sulfur compounds and their derivatives, and has drawn great attention [7, 8]. Extractants include organic polar solvents, ionic liquids and DESs. Organic solvents have extremely strong volatility and toxicity, which are not conducive to the green, pollution-free development of desulfurization. Because of the shortcomings of organic solvents, the focus has shifted to more environmentally friendly solvents (ionic liquids and DESs) [9–14]. Recently, the use of ionic liquid extraction-oxidizing fuel to remove sulfur-containing compounds has been extensively studied because ionic liquids have many advantages as solvents (environmentally friendly, non-toxic) [3, 15–17]. However, the drawbacks of ionic liquids, including valuableness, low biodegradability, and difficulties in purification, limit their further development. Widespread attention has been paid to much cheaper, green, and environment-friendly solvents (DESs) [18–20], which are liquids usually formed by hydrogen bonding interactions of two or three green and inexpensive constituents. Their physicochemical properties are similar to those of traditional ionic liquids, but they have the advantage of being cheap and biodegradable. Recently, the use of DESs as extractants and catalysts has been widely studied because of their special properties, non-toxicity and harmless characteristics [18–27].

In order to thoroughly remove refractory S-compounds, many types of oxidants have been tested, such as tert-butyl hydroperoxide, hydrogen peroxide (H₂O₂), and ozone (O₃) [17, 28, 29]. ODS processes usually use H₂O₂ as a highly efficient oxidant since this has the advantages of low cost and no by-product formation [3, 7, 8, 30–33]. However, H₂O₂ has low stability, being easily decomposed by light, heat, microwave rays, and tiny particles [7]. H₂O₂ is a liquid that is highly corrosive at room temperature and requires high-level equipment for transport, storage, and subsequent application. Moreover, H₂O₂ is not cheap relative to molecular oxygen.

Over the past few years, scientists have been working hard to find novel ways for aerobic oxidation desulfurization (AODS) to remove sulfur compounds from diesel. For instance, emulsion catalysis [C₁₈H₃₇N(CH₃)₃]₅[PV₂Mo₁₀O₄₀] could achieve a 99% desulfurization rate at 60 ℃ in the presence of isobutyl aldehydes (sacrificial agents) under 1 bar O₂ within 4 h. However, the use of sacrificial agents limits its further application [34]. H₅PV₂Mo₁₀O₄₀/SiO₂ afforded a desulfurization efficiency of 99% at 120 ℃ within 5 h under 1 atm O₂ [35]. In addition, Zhu and co-workers prepared a series of boron nitride (BN) material catalysts, which exhibit surprising catalytic activity. The sulfur compounds in diesel can be completely removed without using additives [36–38]. However, the disadvantage is that the desulfurization process needs to be carried out under high temperature conditions. A high-temperature desulfurization process leads to a decrease in oil quality, which is not conducive to further industrial applications. A liquid-liquid extraction and photochemical oxidative desulfurization system (EPODS) composed of air, isobutylaldehyde (IBA), DESs, and model oil was also assembled. The system achieved excellent desulfurization activity. The DBT removal rate could reach 96.3% at 30 ℃ within 3 h with the sacrificial agent IBA [39]. In our previous work, we found that Anderson-type catalysts displayed excellent activities in desulfurization systems for oxidizing sulfur compounds in diesel to their corresponding sulfones using molecular oxygen as an oxidant without adding any sacrificial agents under mild conditions [40–42]. However, these homogeneous catalysts are difficult to be separated and recycled. Therefore, the development of an AODS system with high efficiency and simple recycling, in the absence of any sacrificial agents under mild conditions, remains an important challenge.

As a strong oxidant, peroxide sulfonate has been studied in the field of oxidation [32, 43, 44]. Schulz et al. [45] found that p-chlorotoluene peroxysulfonate formed in situ is highly selective in the oxidation reaction, preferentially attacking electron-rich atoms. Unfortunately, the conditions for the formation of peroxosulfonate are too harsh and usually require the presence of H₂O₂ and OH^–. Recently, Piera et al. [46] have found that the activation energy of the direct oxidation of oxygen can be greatly reduced by a biomimetic path, thereby overcoming the kinetic disadvantages. The use of electron transfer mediators (ETMs) generally promotes the reaction by transporting electrons from the catalyst to the oxidant along a low energy pathway, thereby increasing oxidation efficiency. The transition metal Mo in the highest oxidation state is capable of forming metal-oxygen cluster anions, commonly referred to as polyoxometalates (POMs) [47]. The molecular and electronic structural diversities of these species are significant in biomimetic catalysis [46, 48, 49]. Therefore, the use of POMs as ETMs to generate peroxysulfonate in situ with molecular dioxygen is the focus of our research. In our previous work [50], Co polyanion catalysts and p-TsOH-based DESs which can oxidize sulfur compounds in diesel fuel by a biomimetic approach to corresponding sulfones were reported. Despite the great progress that has been made, the DES biomimetic system still has a lot of room for development. For example, the role of POMs in biological simulation systems has not been clearly stated. Establishing the relationship between the physicochemical properties of the eutectic mixtures and the oxidation desulfurization efficiency of the biomimetic system is more difficult. Therefore, solving the above problems is one of the most critical challenges to the highly desired development of a biomimetic process for AODS.

In this study, a biomimetic method was used to design a novel and efficient AODS system that can significantly improve the efficiency of ODS. An Anderson-type catalyst, Na₃Fe(OH)₆Mo₆O₁₈, was prepared by recrystallization and applied in an AODS system of diesel. An Fe-Anderson type POM, Na₃Fe(OH)₆Mo₆O₁₈, and BSA-based DESs were respectively used as the ETM and the electron donors to mimic the enzyme oxidation system. The AODS system exhibits outstanding catalytic activity. With O₂ as the oxidant, the DBT removal can reach 95% at 60 ℃ in 60 min. The effects of different catalysts, DES composition, and reaction conditions (reaction temperature, DES dosage, catalyst dose, S-compounds, commercial diesel, and catalytic system recycling) on desulfurization were studied.

Poly(ethylene glycol) with average M_n= 2000, ammonium molybdate tetrahydrate (AR, 99%) and sodium molybdate dehydrate (AR, 90.0%) were purchased from Macklin Reagent (Shanghai, China). Benzenesulfonic acid (≥90%), dibenzothiophene (≥98%) and tetradecane (≥98%) were purchased from Aladdin Reagent (Shanghai, China). Phosphotungstic acid hydrate (AR) was purchased from Sinopharm Chemical Reagent Co. Ltd. (Shanghai, China). Iron(Ⅲ) nitrate nonahydrate (AR) was purchased from BASF Chemical Reagent (Tianjin, China). All reagents were reagent grade and used as received.

Na₃Fe(OH)₆Mo₆O₁₈ was synthesized according to established procedures described elsewhere with minor modifications [51]. The synthesis of DESs was carried out by a simple mixing method. PEG2000 was selected as the donor and benzenesulfonic acid was selected as the acceptor. The donor and acceptor of the hydrogen bond were mixed according to molar ratios of 1/1, 2/1, 2.5/1 and 3/1. The two original components were vigorously stirred at 80 ℃ for 1 h using a magnetic stirrer in the oil bath.

Fourier transform infrared (FT-IR) spectroscopy of the sample was performed on a Shimadzu IRAffinity-1S FT-IR spectrometer using KBr pellets at room temperature. The ultraviolet (UV) spectra were recorded on a TU-1901 (Beijing General Analytical Instrument Ltd Co, China) at room temperature. The oxidized diesel was tested by gas chromatography-flame photometric detector (GC-FPD) (Agilent 7890B; HP-5).

In a typical experiment, 4 g of DES was added to a three-necked flask, and 20 mL of model diesel with a sulfur content of 500 μg g^–1 (DBT, BT and 4, 6-DMDBT dissolved in the organic solvent decalin) was added. The extraction sample was taken at a constant temperature for 30 min. Then 20 mg of Fe-Anderson type POM was added and 60 ml min^‒1 of oxygen was bubbled from the bottom of the reactor. The upper oil phase of the reaction was sampled and analyzed by GC with flame ionization detector (GC-FID, Fuli 9750, HP-5). The sulfur content was measured using an internal standard method with tetradecane as the internal standard.

POMs catalysts with different structures have different catalytic effects on DBT removal. We investigated the desulfurization activity of Keggin-type POM H₃PW₁₂O₄₀, Anderson-type POM (NH₄)₆Mo₇O₂₄ and Na₃Fe(OH)₆Mo₆O₁₈ (Table 1). It was found that the Keggin-type POM H₃PW₁₂O₄₀ catalyst exhibits very low removal of DBT (only 28%) (Table 1, entry 3). In comparison to the low efficiency of Keggin-type POM catalysts, the Anderson-type POM (NH₄)₆Mo₇O₂₄ catalysts show better catalytic activity. The removal of DBT can reach 63% within 1 h (Table 1, entry 1), which may be due to the fact that the Anderson-type POM has more terminal oxygen than the Keggin-type POM and is more susceptible to oxygen activation [41, 42, 52]. It is worth noting that the DBT removal of Anderson-type POM catalysts Na₃Fe(OH)₆Mo₆O₁₈ reached 75%. A possible reason for the different desulfurization activities may be the difference in the atomic center of the Anderson structure [53]. The composition of a DES plays an important role in its physicochemical properties [21, 27]. The effect of the compositions of DESs on DBT extraction and oxidation was also studied (Table 1). BSA was chosen as HBD and PEG2000 was chosen as HBA. The AODS system exhibited very low desulfurization activity (29% and 1%, respectively) when either of the DES or POM was absent (Table 1, entries 4 and 5). To our surprise the removal of DBT was extremely high with DES (n(HBDs)/n(HBAs) = (1/2.5) (Table 1, entry 8), where DBT removal could reach 88% after 60 min of reaction, compared to 75% and 82% for n(BSA)/n(PEG2000) = 1/2 and 1/3 (Table 1, entries 1 and 7), respectively. Moreover, when the DES composition n(BSA)/n(PEG2000) = 1/1 (Table 1, entry 6), the desulfurization system shows the lowest extraction capacity, indicating that PEG2000 is a key component of DESs for extraction ability. Therefore, considering the desulfurization efficiency, n(PEG2000)/n(BSA) = 2.5/1 was selected as the optimal molar ratio.

The effect of the amount of catalyst on desulfurization is shown in Fig. 1. The DBT removal was studied with different amounts of catalysts of 0, 5, 10, and 20 mg. Before the reaction, about 26% DBT was transferred into the DES phase. The reactivity of the desulfurization system shows a trend of increasing first and then decreasing with an increasing amount of catalyst. When the amount of the catalyst is 5 mg, the desulfurization rate is only 69% in one hour. It is worth noting when 10 mg of the catalyst dosage was used, the AODS system showed excellent activity and the removal of the sulfur-containing compound DBT could reach 95% after 60 min of reaction. This phenomenon can be attributed to the increase in the active site of the catalyst as the amount of the catalyst increases, resulting in a stronger reactivity. However, when the catalyst amount increased to 20 mg, DBT removal decreased from 95% to 87% in 60 min. This behavior can be explained by the so-called "catalyst inhibitor conversion" phenomenon, which occurs in some oxidation reactions when the catalyst concentration is too high [53–56]. Thus, 10 mg of the catalyst is considered to be the optimum amount for the catalytic oxidation system.

A series of DBT removal experiments involving different DES dosages were carried out as shown in Fig. 2. With an increasing dosage of DES, a minimum of 13% and a maximum of 40% extraction desulfurization efficiency were obtained with the 1 g DES and the 10 g DES system, respectively. This result shows that the amount of DES is critical for the extraction capacity of this system. It was found that when the amount of DES is 4 g, the desulfurization efficiency is the highest and reached 95% in 60 min. But DBT removal was only 69% in 60 min compared to the 10 g DES system with the strongest extraction ability. This phenomenon may be owing to the excess DES having a dilution effect on the catalyst phase [57]. Therefore, 4 g of DES was considered to be the optimal dosage.

The influence of temperature on DBT removal over the Fe-Anderson type catalyst Na₃Fe(OH)₆Mo₆O₁₈ was investigated in Fig. 3. When reaction temperatures were 40 and 50 ℃, DBT removal was only 14% and 34%, respectively. With the increase in reaction temperature, the removal of DBT increased significantly, and a 95% removal of DBT was obtained at 60 ℃. It is worth noting that the removal of DBT can reach 100% at 70 ℃ in 60 min. However, given economic and desulfurization efficiency factors, 60 ℃ was chosen in this work as the optimal temperature for the reaction.

The reaction activity of different sulfur-compounds, for example, BT, DBT and 4, 6-DMDBT in model diesel were studied in the AODS system (Fig. 4). It is worth noting that the removal of the two types of sulfur compounds DBT and 4, 6-DMDBT can reach 99% in 180 min whereas the removal of BT reached 95% in 240 min. The order of desulfurization activity is: DBT > 4, 6-DMDBT > BT. In order to explore the reasons for their different desulfurization activities, the sulfur atoms and steric hindrance of three sulfur-containing compounds were analyzed. BT has the lowest electron density (5.696), so its AODS process has the lowest activity [58]. However, the difference in electron density between DBT and 4, 6-DMDBT (5.758 and 5.760, respectively) is negligible, and the difference in AOSD process activity can be attributed to the effect of steric hindrance [59]. This hypothesis is consistent with the results of our previous work [56].

The application of biomimetic AODS systems in real diesel was also investigated. Figs. 5 and 6 show GC analysis profiles of sulfur-containing compounds in two kinds of commercial diesel (abbreviated as Commercial Diesel Ⅰ and Ⅱ) before and after desulfurization, respectively. After desulfurization of Commercial Diesel Ⅰ, the content of sulfur compounds in the diesel oil was reduced from 24 to 3 μg g^–1 (desulfurization rate 88%, Table 2, entry 1). Furthermore, the sulfur content of Commercial Diesel Ⅱ after desulfurization was reduced to only 59 μg g^–1 (Table 2, entry 2). It can be seen from Fig. 5 that most of the remaining sulfur content is due to unoxidized BT and its derivatives. So the obtained Commercial Diesel Ⅱ was subjected to a second AODS treatment. After the secondary desulfurization of Commercial Diesel Ⅱ, the remaining sulfur content was 14 μg g^–1 and the total desulfurization reached 94% (Table 2, entry 2). The result is consistent with the reaction activity results of different sulfur compounds (Fig. 4, BT was more difficult to remove than DBT in this system). The results also indicate that the biomimetic AODS system can be used for real diesel.

In order to detect the stability of DES containing the Fe-Anderson type catalyst, a recovery experiment was carried out as shown in Fig. 7. The system after the reaction was allowed to stand overnight, and then the upper oil was separated from the reaction system by decantation. Therefore, the biomimetic system (the low DES phase containing POM) was recovered and then the fresh model diesel added into it for the next AODS process under the same conditions. The biomimetic AODS system showed no significant decrease in activity after 5 cycles, confirming its good stability.

To investigate the effect of DES physical properties on the desulfurization efficiency of diesel, the conductivity and viscosity of different DESs were tested. It can be seen in Fig. 8 that the conductivity of DES decreases with an increase in the amount of PEG2000 amount at a certain temperature. According to the experimental results (Table 1, entries 1, 6, 7 and 8), the solvent conductivity of DES was not the main factor affecting the desulfurization reactivity of the system. This means that it could not be simply concluded that the correlation between the conductivity of DES and desulfurization capability was positive or negative. Fig. 9 shows the viscosity data of DESs of different compositions at different temperatures. PEG2000/2.5BSA has a maximum viscosity of 60 ℃ and was speculated to have a stronger hydrogen bond, because viscosity data is mainly affected by intermolecular hydrogen bonding at low temperatures [60, 61]. What is particularly noteworthy is that PEG2000/2.5BSA was also the best system for AODS activity (Table 1). Therefore, we speculate that the activity of the biomimetic AODS system is related to the hydrogen bond strength between the DES components. This conclusion is consistent with the results of our previous work [27]. When the temperature was raised to 80 ℃, the viscosity increased when the composition of PEG 2000 increased. Because of the high temperature, the stacking effect plays a dominant role. The longer the carbon chain of the compound, the stronger the filling effect and the more difficult the free movement, so the viscosity of the liquid increases as the length of the carbon chain increases [60, 61].

The UV-visible (UV-Vis) spectrum of Fe-Anderson catalyst dissolved in acetonitrile varies in the presence of nitrogen or molecular oxygen, as shown in Fig. 10a. The disappearance of the absorption peaks at 232 and 258 nm indicates that Fe-Anderson catalyst interacts with oxygen [42]. The UV-vis spectrum of BSA-based DES dissolved in acetonitrile showed little change in the presence of nitrogen or molecular oxygen. However, when the Fe-Anderson catalyst was added to the DES-containing acetonitrile solution as shown in Fig. 10b, the peaks at 232 and 258 nm disappeared markedly after oxygen bubbling for 1 h. This implied that oxygen cannot directly form peroxosulfonate with BSA due to the high energy barrier, but oxygen molecules can be activated when POM and DES coexist [50]. These results are consistent with the experimental observation that the desulfurization system is inactive when POM and DES were used alone, but a high AODS activity could be obtained when POM and DES were used together. (Table 1, entries 4 and 5). In order to further verify the reaction mechanism, the Fourier-transform infrared (FT-IR) spectra of DES and POM in an oxygen or nitrogen atmosphere were investigated. Infrared curves of DES and POM under nitrogen atmosphere are shown in Fig. 11a. The three absorption peaks at 1118, 1145 and 1199 cm^‒1 are attributed to the SO₂ stretching vibration [32], and the 870‒1000 absorption peaks are assigned to POM [42]. It is worth noting that when DES and POM reacted in the oxygen environment for 60 min (Fig. 11b), the absorption peaks of 1118 and 1199 cm^–1 moved to 1110 and 1176 cm^–1 respectively, and the changes in these two absorption peak positions indicate that sulfonic acid groups form new species under the action of oxygen.

Moreover, a new absorption peak is generated at 963 cm^–1, possibly because the heteropolyanion of POM interacts with oxygen to form a peroxopolyanion [62]. The IR results were consistent with the UV characterization and confirmed the formation of peroxosulfonate. Based on the results of UV-vis, FT-IR and GC-mass spectrometry (GC-MS) (see SI), we speculated about the mechanism of AODS by multistep electron transfer through a biomimetic approach (Scheme 1). First, POM and oxygen form peroxopolyacid anions, which then undergo nucleophilic substitution with benzenesulfonic acid in the protic solvent PEG2000 to form phenylperoxysulfonic acid. Because peroxysulfonate is highly active for electron-rich S atoms, it preferentially attacks the S atom. Thereby, DBT is oxidized to DBTO₂. The reduced phenylperoxysulfonic acid and POMs are oxidized by O₂ to initiate a new catalytic cycle. These results indicate that the coupled redox system and ETMs facilitate the reaction process by transporting electrons from the BSA-based DES to the oxidant along a low energy pathway. Eventually, the DBT was susceptible to oxidation to DBTO₂.

A new Fe-Anderson catalyst coupled with a PEG2000/BSA-type DES biomimetic system has been successfully prepared. The viscosity of DES is maximum in the n(PEG2000)/n(BSA) = 2.5 system at 60 ℃, and it is speculated that this results from stronger hydrogen bonding. In addition, the PEG2000/2.5BSA system also possesses the best desulfurization activity. This result suggests that the activity of the desulfurization system is related to the strength of the hydrogen bonding. This biomimetic strategy was applied to AODS of model diesel, showing unprecedented performance with 95% removal of DBT in 60 min. Furthermore, this biomimetic system can also be successfully used for oxidative desulfurization of real diesel. Finally, the catalyst can be recycled 5 times without a significant decrease in reactivity, indicating the potential for commercial use of this catalytic system.