Advances in Oxidative Desulfurization of Fuel Oils over MOFs-Based Heterogeneous Catalysts

: Catalytic oxidative desulfurization (ODS) of fuel oils is considered one of the most promising non-hydrodesulfurization technologies due to the advantages of mild reaction conditions, low cost and easy removal of aromatic sulfur compounds. Based on this reason, the preparation of highly efﬁcient ODS catalysts has been a hot research topic in this ﬁeld. Recently, metal-organic frameworks (MOFs) have attracted extensive attention due to the advantages involving abundant metal centers, high surface area, rich porosity and varied pore structures. For this, the synthesis and catalytic performance of the ODS catalysts based on MOFs materials have been widely studied. Until now, many research achievements have been obtained along this direction. In this article, we will review the advances in oxidative desulfurization of fuel oils over MOFs-based heterogeneous catalysts. The catalytic ODS performance over various types of catalysts is compared and discussed. The perspectives for future work are proposed in this ﬁeld.


Introduction
Fuel oils are widely used in human life and are the main energy consumption products in the world [1]. With the continuous growth of the world population and the gradual development of industrial levels, the energy crisis and environmental protection have attracted more and more attention [2,3]. It is well known that there are some sulfur-containing molecules in fuel oils. The combustion of such molecules releases SO x compounds, which not only greatly hinders the processing and production but also brings about serious environmental problems [4]. The composition of sulfur-containing molecules is complex and includes thiols, sulfides, disulfides and thiophene compounds [5]. In particular, thiophene compounds such as thiophene (T), benzothiophene (BT), dibenzothiophene (DBT) and their alkyl derivatives account for a relatively high proportion of the sulfides in fuel oils. The removal of these compounds is of great significance to protect the environment and produce clean fuel oils.
Nowadays, countries around the world have imposed strict restrictions on the sulfur content in fuel oils. Many have demanded that the sulfur level be below 10 ppm. This means that the sulfur-containing molecules like 4,6-dimethyldibenzothiophene (4,6-DMDBT) must be desulfurized. However, it is very difficult to realize this process by industrial hydrodesulfurization (HDS) technology. This drives researchers to develop new desulfurization techniques. Indeed, several non-hydrodesulfurization technologies including adsorption desulfurization (ADS) [6], extraction desulfurization (EDS) [7], biological desulfurization (BDS) [8] and oxidative desulfurization (ODS) have been widely investigated. Therefore, ODS technology is considered one of the most promising nonhydrodesulfurization technologies due to advantages such as mild reaction conditions, low cost and easy removal of aromatic sulfur compounds. Thus, the development of highly efficient ODS heterogeneous catalysts has been pursued.
Currently, various types of heterogeneous catalysts involving metal oxides [9], boron nitride [10], reduced graphene oxide (rGO) [11], Ti-containing porous materials [12,13] and metal-organic frameworks (MOFs) [14,15] have been attempted for ODS reactions of fuel oils. Among them, MOFs have received much attention due to advantages such as abundant metal centers, high surface areas and varied pore structures, which are composed of secondary building units (SBUs) connected by organic linkers to form crystalline porous materials. Such materials possess both the rigidity of inorganic materials and the flexibility of organic materials. Moreover, rich metal centers in the structure of MOFs could be catalytic active sites for some chemical reactions [16]. For this, the synthesis and catalytic performance of the ODS catalysts based on MOFs materials have been widely studied [17][18][19]. Among this research work, MOFs have been used mainly as ODS catalysts or composite catalysts with additional active species or sacrificial templates producing ODS active centers. The ODS performance of catalysts has mostly been evaluated by the catalytic oxidation reactions of BT, DBT and 4,6-DMDBT. In the reaction system, the used oxidants involve H 2 O 2 , tert-butyl hydroperoxide (TBHP), cumene hydroperoxide (CHP), O 2 and air. The reaction temperature is generally below 80 • C, although the oxidation temperature is used up to 150 • C in the case of O 2 or air as an oxidant.
In this review, we will summarize the advances in oxidative desulfurization of fuel oils over MOFs-based heterogeneous catalysts. The catalytic ODS performance over various types of catalysts is compared and discussed. The opportunities and challenges for future work are proposed in this field.

Ti-MOFs as ODS Catalysts
In the earlier studies, the Ti species in titanium-containing zeolites and mesoporous silica have been demonstrated to be active in the ODS reactions of sulfur-containing compounds [20,21]. However, the content of the introduced Ti species into the framework is low (<5 wt.%) due to the limitation of the structure. Compared with these Ti-containing porous materials, Ti-MOFs are promising candidates for ODS reactions due to the presence of rich Ti sites in their structure. As a member of Ti-MOFs, MIL-125(Ti) is attractive. In its structure, the basic unit of MIL-125(Ti) is Ti 8 O 8 (OH) 4 -(O 2 C-C 6 H 5 -CO 2 ) 6 and a cyclic octamer is composed of titanium octahedral units shared by corners or edges. These cyclic octamers are connected to the other 12 cyclic octamers through BDC linkers to form a porous three-dimensional periodic array with two types of cages. One type of cage is an octahedron (12.5 Å) and the other is a tetrahedron (6 Å) [22].
In 2013, Se-Na Kim and coworkers first reported that MIL-125(Ti) was active in the oxidative desulfurization of DBT [23]. In this work, the catalytic ODS performance of MIL-125(Ti) was evaluated and compared with traditional microporous and mesoporous titanium silicates. The results showed that the order of DBT conversion under the same reaction conditions was Meso-TS-1 > MIL-125(Ti) > Micro-TS-1 (Table 1), which is in accordance with the window size in porous materials. This result suggests that pore size also plays an important role in the ODS reactions of sulfur compounds besides Ti species. To overcome the limitation of micropore size, meso-MIL-125(Ti) with mesopores was synthesized in the presence of surfactant by vapor-assisted crystallization method [24]. The catalytic results further demonstrate that the introduction of mesopores could improve the catalytic ODS performance of MIL-125(Ti). Additionally, Li et al. investigated the catalytic performance of MIL-125(Ti) and NH 2 -MIL-125(Ti) with different crystal sizes; MIL-125(Ti)-L with large crystal size showed the best catalytic performance when H 2 O 2 was used as oxidant [25]. It is thought that MIL-125(Ti) with large crystal size hould possess more coordination-unsaturated Ti(IV) sites, which is more favorable for catalytic oxidative desulfurization. Although some progress on MIL-125(Ti) has been made, its structural stability is still an issue during the ODS reactions.  [26]. This material used the composite layer of TiO 6 octahedron as two-dimensional secondary structural units to form a three-dimensional framework through a bpdc 2− linkers connection. The catalytic results showed that 99% DBT over COK-47 S was oxidized in 120 min at 60 • C by TBHP as an oxidant. Such catalytic performance could be attributed to the presence of many methoxy groups (Me-O-Ti) and open-metal Ti sites. Moreover, the catalyst remained structurally stable after three recycles.
More recently, Sun and coworkers reported that new porous titanium terephthalates with hierarchical porosity (Ti-BDC) and amorphous nature have been successfully synthesized for deep oxidative desulfurization of fuel oil [27]. The catalytic results indicate that Ti-BDC exhibited superior catalytic activity in the ODS reactions of BT, DBT and 4,6-DMDBT than MIL-125(Ti) and meso-MIL-125(Ti), which might be attributed to the introduction of partial mesopores and abundant coordination-unsaturated Ti sites. Especially in the above synthetic system, a novel porous titanium terephthalate (Ti-BDC-A) with more defective Ti sites and higher surface area was produced with the assistance of acetic acid ( Figure 1). When the material is used as an ODS catalyst, DBT (500 ppm) in model fuel oil can be completely oxidized in 10 min at room temperature [28].

Zr-MOFs as ODS Catalysts
It is generally well known that Zr-MOFs possess high structural stability compared with other metal-based MOFs. One example is UiO-66 (Zr), whose structure composed of Zr 6 O 4 (µ 3 -OH) 4 clusters in coordination with 12 terephthalic acids [29]. In 2015, Granadeiro et al. first reported that UiO-66(Zr) as a heterogeneous catalyst was used in the ODS of diesel. In this work, UiO-66(Zr) samples were prepared by different synthesis methods and their catalytic ODS performance were evaluated [30]. It is noted that the catalytic performance had a close relationship with the crystallinity degree and linker defects in UiO-66 (Zr). The samples with low crystallinity displayed superior catalytic performance. Meanwhile, the good catalytic activity can be kept in the dual phase catalytic system with H 2 O 2 as an oxidant and acetonitrile as an extractant for the ODS of model and real diesel. Similarly, some researchers obtained UiO-66(Zr) with lower crystallinity by shortening the synthesis time, but not surprisingly its reusability was poor [31]. A detailed comparison of catalytic performance over various Zr-MOFs is shown in Table 2.
In general, the defects in crystal materials are closely related to their properties. Thus, it is of great importance to prepare the defective crystal materials [32][33][34][35][36][37][38][39][40]. As for UiO-66(Zr), some methods have been developed to fabricate defective UiO-66(Zr). For example, Sun et al. adopted solvent-free synthesis to produce defective UiO-66(Zr) [34]. The catalytic data indicate that UiO-66(Zr) free with rich defects exhibited much better catalytic activity than UiO-66(Zr) solvent with high crystallinity prepared by conventional solvothermal route in the ODS reactions. Such catalytic performance has a good correlation with the number of Lewis acid sites in UiO-66(Zr) because the defects that resulted from the loss of organic linkers may promote the formation of Lewis acid sites [35,36]. Based on this study, this research group synthesized defective amino-and nitro-functionalized UiO-66(Zr) by solvent-free method. The results showed that nitro-functionalized UiO-66(Zr) exhibited excellent catalytic activity in the oxidative desulfurization of DBT and 4,6-DMDBT. The ODS activity could almost be kept after five cycles [17]. The possible reason is that the introduction of electron-withdrawing groups reduces the electron density around Zr and enhances its electron-withdrawing ability [37,38]. On the other hand, most of MOFs are microporous materials, which would limit the mass transfer of reactants and hinder the accessibility of catalytic active sites in the pores [41]. For this, Hao et al. synthesized hierarchical porous UiO-66(Zr) (HP-UiO-66(Zr) and evaluated its ODS performance. The results showed that HP-UiO-66(Zr) could complete the oxidation of DBT and 4,6-DMDBT at room temperature at a low oxidant dosage (O/S = 4) [42].
Based on the structure of UiO-66(Zr), bimetal-centered UiO-66(Zr) materials were also prepared. Sun et al. firstly reported Ti-UiO-66(Zr) obtained by ion-exchange method for ODS reactions [43]. It is well known that Ti ions have stronger oxidation ability than Zr ions [44]. Consequently, Ti-UiO-66(Zr) showed better catalytic ODS performance than the parent UiO-66(Zr). In addition, Ye et al. prepared Hf-incorporated UiO-66(Zr) under solvent-free conditions [45]. In the structure of this material, 0.7 Zr atoms are replaced by Hf atoms to form Zr-Hf-oxo clusters with a considerable number of Zr/Hf-OH active sites ( Figure 2). The obtained catalyst may complete the oxidation of 4,6-DMDBT (1000 ppm) from model oil within 15 min at room temperature. Theoretical studies indicate that the exposed Hf-OH centers can easily react with H 2 O 2 to form Hf-OOH intermediates and thus control the reactivity.

Other Metal-Centered MOFs as ODS Catalysts
The other metal-centered MOFs mainly include the studies of V-MOFs, Co-MOFs and MIL-101(Fe/Cr) for ODS reactions. A detailed comparison of catalytic performance over other metal-centered MOFs in the ODS reactions of model fuel oil is shown in Table 3.
MIL-47(V) is the first example of non-Zr-and Ti-based MOFs for ODS reactions. In this work, the catalytic performance of MIL-47(V) was compared with that of MIL-125(Ti) by the oxidation reactions of T, BT and DBT [53]. The results show that MIL-47(V) exhibited better catalytic ODS performance than MIL-125(Ti) in the oxidation of DBT. Meanwhile, MIL-47(V) is more suitable for the catalytic oxidation of DBT at relatively low reaction temperatures, which should be related to the pore size of the material. Under the same reaction conditions, the removal efficiency of BT over MIL-47(V) is not as good as that of DBT. A reasonable explanation is that the electron density of sulfur atoms in DBT is higher, while the sulfur species with higher electron density have higher catalytic activity [54,55]. Notably, MIL-47(V) exhibited unusual catalytic activity in the oxidation of thiophene, suggesting that metal centers in MOFs have a great influence on the ODS reactions. Although the catalytic ODS performance of MIL-47(V) is better than that of MIL-125(Ti), its structural stability is poor.
Another interesting study about V-MOFs is the study on MFM-300(V). Li et al. found that MFM-300(V) can carry out the ODS reaction by ambient air as the oxidant, and the removal of DBT and 4,6-DMDBT reached 99.6% and 98.1%, respectively [56]. The bridging oxygen atoms in the structure exist on the corner of pores and make MFM-300(V) surface of pores locally electron-rich, which is conducive to capturing protons from hydrocarbons and activating oxygen in the air [57].
Masoomi et al. reported that Co-MOF materials (TMU-10 and TMU-12) were successfully synthesized and their catalytic performance in ODS reactions was evaluated [58]. The removal of DBT (500 ppm) over TMU-12(Co) was 75.2% at 60 • C. Additionally, Abazari et al. studied the catalytic ODS performance over another type of Co-MOF (NH 2 -TMU-53). The results showed that the sulfur content in a model oil for DBT removal can be reduced from 500 ppm to 103 ppm at 60 • C after 2 h [59]. Meanwhile, it is noted that the introduction of amino functional groups may significantly improve the adsorption ability for DBT, and the oxidation rate of DBT increased with adsorption amount.
MIL-101 is an important class of MOFs, which generally possess high surface area and large pore openings. Adrián Gómez-Paricio et al. first investigated the catalytic performance of MIL-101(Fe) and MIL-101(Cr) in ODS of DBT with oxygen as the oxidant [60]. During the reaction, an obvious induction period of 6 h was observed. Further studies disclosed that the induction period of the reaction was related to solvent diffusion and the formation of the first reactive oxygen species. The catalytic results showed that MIL-101(Cr) had better catalytic activity and cycle stability than MIL-101(Fe). Subsequently, the catalytic performance of functionalized MIL-101(Cr) was also studied. It was found that nitro-functionalized MIL-101(Cr) had better ODS activity for DBT in fuel oil [61], which is in agreement with the case of functionalized UiO-66(Zr).

Composites of MOFs and Additional Active Phases as ODS Catalysts
Although MOFs generally have the advantages of high surface area, rich porosity and varied pore structures for catalytic reactions, it is a fact that the number of catalytic active sites in MOFs is limited because most metal centers are in a state of coordination saturation. To remedy this shortcoming, some composite catalysts of MOFs and additional active phases have been designed and synthesized by various strategies. The studied MOFs mostly referred to the MIL family, Zr-MOFs and MOF-199 with good structural stability.
The active phases were mainly focused on various types of polyoxometalates and some metal oxides.

Composites of MOFs and Polyoxometalates
Polyoxometalates (POMs) have been widely studied as the active centers of ODS catalysts because their acidic and redox properties can be effectively designed and controlled through changing their compositions and structures at atomic or molecular scale [62,63]. However, it is well known that POMs are easily soluble in organic medium, which is not beneficial for the recycling and reuse of heterogeneous catalysts. Thus, the composites of encapsulating POMs into MOFs can be an efficient approach to solve this problem. Table 4 shows a detailed comparison of catalytic performance over the composites of MOFs and POMs in the ODS reactions of model fuel oil.
One of the research focuses is the synthesis and catalytic ODS performance of the composite of POMs and MOFs from the MIL family. Hu et al. first reported the encapsulation of phosphotungstic acid (PTA) into the cages of MIL-101(Cr) by using the strategy of "bottle around ship" under static conditions [64]. The obtained material exhibited the order of oxidation activity (DBT > 4,6-DMDBT > BT) for different sulfur compounds. After four consecutive reaction cycles, the catalytic performance decreased by about 4% and PTA loading in the composite reduced by about 5%, indicating that such encapsulation could greatly inhibit the loss of PTA into organic medium. Based on this methodology, some similar POMs [65][66][67][68] were also used as active centers encapsulated in the cages of MIL-101(Cr). For instance, Ribeiro [65,66]. The catalytic results showed that DBT in model oil was almost completely oxidized over both catalysts and could be reused well. The difference is that the former uses acetonitrile as the extractant and the latter is ionic liquids BMIPF 6. It was found that ionic liquids can promote the generation of reactive oxygen species and avoid the decomposition of PW 12 , which is an effective co-catalyst. In addition to the abovementioned Keggin-type POMs, the Weakley-type Na 7 H 2 LaW 10 O 36 (abbreviated as LaW 10 ) was also investigated because its molecular size matches well with the window size of MIL-101(Cr) [69]. Lu   Another research hot spot is the combination of POMs and MOF-199(Cu). MOF-199(Cu) is welcomed mainly because its pore size closely matches with the size of POMs [80,81]. Rafiee et al. first synthesized three composites of POMs and MOF-199(Cu) and demonstrated their potential in ODS reactions [82]. Further, a series of such composites (including, but not limited to, POM@MOF-199@MCM-41, C-POM@MOF-199@MCM-41, HPA@MOF-199@CA, CNTs@MOF-199-Mo 16 V 2 ) were prepared for ODS reactions with O 2 as the oxidant [83][84][85][86][87][88][89][90][91]. The catalytic results indicated that the combination of POM and MOF-199(Cu)@MCM-41 might be helpful for the enhancement of ODS performance under the reaction condition of O 2 as the oxidant. Based on the combination of the same active species with different MOFs, Wang et al. made a comparison on the catalytic ODS activity of PTA@MIL-100(Fe), PTA@UiO-66(Zr), and PTA@ZIF-8(Zn). The results suggested that MOFs with large window size could be more suitable for the encapsulation of active species [92]. In addition, to facilitate the recovery of catalyst, Si-Wen Li et al., synthesized Fe 3 O 4 @MOF-PMoW (FeMP, M = Cr, Cu, Zr) with magnetism by the one-pot method [19]. It has been proposed that this kind of catalyst has a unique spinous structure similar to the virus, which is beneficial for enhancing the catalytic performance by increasing the contact area with sulfur substrate ( Figure 5). The other cases mainly involved the composites of POMs and Zr-MOFs, such as UiO-66(Zr) [92][93][94][95][96][97], UiO-67(Zr) [98] and MOF-808(Zr) [99]. Zhang and coworkers encapsulated H 3 PMo 12 O 40 (PMA) into the pores of UiO-66(Zr) and evaluated its catalytic performance by the oxidation of DBT (500 ppm) in model oil by using TBHP as an oxidant [93]. The excellent catalytic performance was attributed to the synergistic effect between UiO-66(Zr) and PMA. The lack of lattice oxygen and electron transfer led to the easier generation of oxygen vacancies and Mo 5+ on PMA@UiO-66(Zr), which can promote the formation of active free radicals and thus improve its ODS performance. In addition, it has been demonstrated that unsaturated metal sites or defect sites in MOFs are closely related to the catalytic performance of materials [26,31]. For example, Chang et al. introduced thiourea and 1-butyl-3-methylimidazolium chloride ([Bmim]Cl) into the synthetic system of PMA@UiO-66(Zr) to construct more unsaturated metal sites in UiO-66(Zr). As a result, the enhancement of catalytic ODS performance was observed [94]. Sun et al. reported that UiO-66(Zr) with abundant defect sites can be synthesized by the solvent-free method [34]. Based on this work, a composite material of HPW and UiO-66(Zr) (PW/UiO-66(Zr)-green) was produced ( Figure 6) [96]. It should be emphasized that the catalyst synthesized by this method possesses hierarchical pores and double active sites and displays excellent catalytic activity to oxidize DBT and 4,6-DMDBT at room temperature when H 2 O 2 is used as the oxidant.

Composites of MOFs with Other Active Species
It is well known that some metal oxides such as MoO 3 , WO x , MnO 2 , VO x can be active phases in the ODS reactions of sulfur compounds. Minoo Bagheri et al. prepared the composite of MOFs and MoO 3 by the combination of Zn-MOF (TMU-5) and MoO 3 [100,101]. The obtained material was used for aerobic photooxidation of DBT under UV light, resulting in 95.6% oxidation rate of DBT (Table 5). However, the leaching of molybdenum ions would occur after reuse. Comparatively, a new magnetic catalyst Fe 3 O 4 @MoO 3 @MOF-199(Cu) obtained by the combination of MOF-199(Cu) with the magnetic core MoO 3 @Fe 3 O 4 displayed good recycle performance [102]. The catalysts can be collected by magnetic field and recycled 15 times. In the same way, Fe 3 O 4 @W-MoO 3 @MOF-199(Cu) with core-shell structure was also prepared by replacing MoO 3 with W-MoO 3 [103].
On the basis of UiO-66(Zr), various types of active species including MnO 2 [104], MoO 3 [105], WO x [106] and 1-methylimidazolium-3-propylsulfonate hydrosulfate (abbreviated as PSMIMHSO 4 ) [107] were introduced into the pore channels or grafted on the framework of UiO-66(Zr). Therefore, W/UiO-66(Zr) prepared by in-situ injection of a single W site on the node of the defective UiO-66(Zr) structure by forming a Zr-O-W bond under solvent-free conditions exhibited excellent catalytic performance in the ODS reactions of DBT and 4,6-DMDBT. Such performance was attributed to the presence of single-atom W sites [106].
1-Sulfopropyl-3-methylimidazolium hydro-sulphate (expressed as ILs) was loaded in the mesoporous structure of MIL-100(Fe), and the DBT (50 ppm) removal rate of 99.31% was obtained under a high O/S ratio of 25 [108]. However, the performance of the material decreased by about 21% after five reuses. Recently, Qi et al. designed IL@MOF composites by post-synthetic ligand exchange method using carboxyl functionalized ionic liquids as raw materials, and the obtained materials could remove more than 99% DBT under optimized conditions [109]. The improvement of catalytic performance was attributed to the introduction of ionic liquids to enhance the adsorption capacity of materials, which is more conducive to the contact between reactants and active sites.

MOFs-Derived Materials as ODS Catalysts
Based on the advantages of high surface area, uniform and rich metal centers, MOFs have been explored as the precursors of active ODS catalysts. For this, MOF-derived materials with mixed nanostructures have been generated by designed synthesis and pyrolysis. The adjustable metal content in MOF-derived materials and the possibility of combining with highly active metal oxides or carbides make them attractive candidates for efficient ODS catalysts. Table 6 shows a detailed comparison of catalytic performance over MOFs-derived materials in the ODS reactions of model fuel oil.
Kim et al. proposed a strategy of high-temperature carbonization and carbon reduction to prepare titanium oxide nanoparticles supported on amorphous carbon with nanoporosity (abbreviated as Ti/NC) by using Ti(O-IPR) 4 -modified IRMOF-3(Zn) as a template and evaluated its catalytic performance in the oxidation reaction of DBT [112]. The catalytic results indicated that Ti/NC had high ODS reactivity (k app = 3.4 × 10 −3 min −1 ) and good cycling stability compared with the materials synthesized by impregnation method. However, the introduced Ti content was low due to the limitation of pore volume by this strategy. In order to overcome the shortage, this research group attempted a pyrolysis strategy using hierarchical micro/meso-MIL-125(Ti) as a template and studied the properties of pyrolysis products at different calcination temperatures [113]. The authors claimed that the titanium content and phase in the final material could be controlled by changing the pyrolysis temperature. The obtained material has higher porosity and catalytic activity for DBT oxidation than that prepared by the pyrolysis of micropore-MIL-125(Ti). In particular, it has been pointed out that more mesopores and smaller titanium nanoparticles in the material are the main reasons for the improvement on catalytic activity. In the same way, a series of V-based carbon catalysts were obtained by pyrolysis of MIL-47(V) at different temperatures [114]. This series of catalysts have the advantages of high V loading (36-68 wt.%) and high mesoporous ratio (0.84~0.91). The formation of active vanadium oxides or carbides can be controlled by adjusting the pyrolysis temperature. In addition, although MIL-47(V) pyrolysis to form carbon catalysts could enhance their chemical resistance to TBHP, the leaching of active species is still an issue.
Sarker et al., developed a new strategy to synthesize MOFs loaded with Ti precursors inside/outside the pores by regulating the hydrophilic/hydrophobic relationship between the MOFs and Ti precursor. The obtained materials were pyrolyzed at a high temperature to prepare TiO 2 -integrated carbon materials [115]. The results showed that the oxidation of DBT (1000 ppm) over such catalysts can reach 97.2% within 1 h at 80 • C. The authors propose that the particle size of TiO 2 is related to the position of Ti precursor. The pyrolysis of MOF loaded with Ti precursor inside the pore can obtain smaller TiO 2 particles and higher porosity than those loaded outside the pore, which will lead to superior catalytic activity. In addition, a chromium-titanium oxide hybrid nanocomposite was also reported, which was obtained by the pyrolysis of Ti-modified MIL-101(Cr) [116]. When CHP was used as oxidant, the removal of DBT (1000 ppm) in model oil could reach 90% within 30 min.
Besides the abovementioned strategies, there are other methods for preparing carbon materials containing metal oxide nanoparticles involving the pyrolysis of MOF@MOF or bimetallic MOF. Bhadra et al. reported porous carbon materials containing TiO 2 nanoparticles (MDC-P and MDC-C) synthesized by MIL-125(Ti)-NH 2 and ZIF-8(Zn)@MIL-125(Ti)-NH 2 as pyrolysis templates [117]. It can be seen from catalytic data that MDC-C exhibited good ODS activity with 99.5% removal of DBT (1000 ppm) within 2 h and can almost maintain initial activity after five cycles. As an example of the pyrolysis from bimetallic MOFs, Bhadra et al. prepared a series of MnO-loaded porous carbon materials (abbreviated as MDNM) by the pyrolysis of Zn/Mn bimetallic MOF-74 [118]. It was found that the porosity of MDNM increased with the Zn content while the size and content of MnO were reduced, which might be related to the evaporation of Zn at high temperature. The catalytic results show that the performance of MDNM(75Zn25Mn) is the best among materials with different metal ratios. In addition, bimetallic MAF-6(Zn, Co) was also synthesized and pyrolyzed to obtain Co-loaded N-doped porous carbon [119]. The material MDC-6(75Zn25Co)-900 could give a conversion of 93.6% within 2 h in ODS of DBT. Further studies suggest that such performance is attributed to high porosity, the synergistic effect of Co-N-carbon, and the uniform dispersion of small size active Co nanoparticles, which is beneficial for improving mass transfer and charge transfer process [120,121].
More recently, the composite of heteropoly acid and MOFs has also been used as pyrolysis template that can produce nitride-loaded porous carbon materials. Mondol et al. first reported the synthesis of Mo 2 N@C by high-temperature pyrolysis of phosphomolybdic acid supported MAF-6(Zn) [122]. When Mo 2 N@C was used as an ODS catalyst, DBT (1000 ppm) in model oil could be completely oxidized within 20 min at 60 • C. The oxidation of DBT follows a non-radical reaction mechanism. The high catalytic activity is attributed to the contribution of N in Mo 2 N, which leads to the easy formation of Mo-peroxides. By similar synthesis strategy, W 2 N@C was successfully synthesized and exhibited good catalytic ODS performance (Figure 8) [123].

Summary and Outlook
In summary, the advances in oxidative desulfurization of fuel oils over MOFs-based heterogeneous catalysts have been reviewed. As shown, many research works have been done in the field. The studied ODS catalysts have mainly involved pure MOFs, composites of MOFs and additional active phases and MOFs-derived materials. The reaction system refers to the use of different oxidants including H 2 O 2 , TBHP, CHP, O 2 and air. The reaction temperature is used up to 150 • C, dependent on the reaction system. Generally, the reaction temperature under the conditions of O 2 or air as oxidant is higher than that under other oxidants. In some cases, various extractants also had an influence on the catalytic ODS performance. The opportunities and challenges in this field are summarized as follows.

1.
As for MOFs as ODS catalysts, most of pure MOFs exhibited poor catalytic ODS performance due to the lack of active sites. However, it has been demonstrated that MOFs with defect sites may improve their performance, especially Zr-based and Ti-based MOFs. Therefore, the design and synthesis of novel Zr-MOFs and Ti-MOFs with abundant defect sites is still challenging.

2.
As for composites of MOFs and additional active phases as ODS catalysts, most research works have been focused on the combination of POMs and MOFs. Because the type of POMs is varied, it is undoubted that such combination could supply more chances to develop new catalysts for catalytic ODS of fuel oils, particularly for aerobic ODS.

3.
As for MOFs-derived materials as ODS catalysts, more active sites could be created and well dispersed into final materials by this strategy. However, the stability of active sites is still an issue. Meanwhile, high cost resulting from the collapse of MOFs is not beneficial for industrial application. Certainly, such a study is meaningful for the understanding of reaction mechanisms for basic research.
Researchers are still faced with many unsolved problems in the ODS field, although some achievements have been made. Particularly the ODS processes of real fuel oils are very complicated and challenging with air as an oxidant. However, the development of novel MOFs catalysts may provide a good chance to meet the challenges.