Recent Advances in Selective Photo-Epoxidation of Propylene: A Review

: The epoxidation of propylene to produce propylene oxide (PO) has a vital role in the industrial production of several commercial compounds and the synthesis of numerous intermediates, fine chemicals, and pharmaceuticals. However, the current PO production processes pose significant problems regarding the environment and economy. The direct photo-epoxidation of propylene using molecular oxygen (an ideal oxidant with active oxygen of 100 wt %) under light irradiation is a promising technology to produce PO. This process offers numerous advantages, including the use of simple technologies, low-cost methods, and environmental friendliness. Many efforts have focused on the design of new photocatalyst systems, optimizing the conditions for a photocatalytic reaction, and elucidating the mechanisms of photo-epoxidation. This review is expected to serve as a comprehensive background, providing researchers with insight into the recent developments regarding the direct photo-epoxidation of propylene.


Introduction
Among epoxides, propylene oxide (PO) is known to be an extremely reactive substance and an important chemical intermediate [1]. It is sometimes called 1,2-epoxypropane or 2-methyloxirane. This compound is widely used as the starting material to synthesize numerous commercial materials; these include polymers (polyurethanes and polyesters), oxygenated solvents (propylene glycol ethers), and industrial fluids (monopropylene glycol and polyglycols) [1]. Based on a market overview from Mordor Intelligence, PO is forecasted to gain at a compounded annual growth rate of approximately 5.9% during the forecast period of 2019-2024 [2].
Currently, the existing PO production processes are technologically divided into chlorohydrin, a coproduct (indirect oxidation or organic hydroperoxide), hydrogen peroxide, direct oxidation, and hydro oxidation processes [3]. A significant amount of PO is traditionally made using the chlorohydrin process, including chlor-alkali PO, lime PO, and PO (Lummus) routes, which suffer from environmental liabilities and have high capital costs. Alternative processes through indirect oxidation, including Tert-butyl Alcohol (PO/TBA), styrene monomer propylene oxide (SMPO), and cumene hydroperoxide (CHP), have been commercialized. Since there is a long-term level of demand for their coproducts, it is expected that the organic hydroperoxide processes will maintain an essential role in propylene epoxidation. New PO technologies, namely hydrogen peroxide propylene oxide (HPPO) [4] and hydro oxidation [5], have recently been gaining in popularity. However, the low efficiency of hydrogen and safety considerations require hydrogen peroxide production facilities to apply such processes as only a temporary solution.
Among all alternative processes available, the direct photocatalytic propylene epoxidation process, using molecular oxygen as an oxidant. is the most feasible, because (1) molecular oxygen, which has active oxygen of 100 wt %, is an ideal oxidant; (2) solar light is a reliable, abundant, and green energy source; and (3) solar-driven photo-epoxidation can be easily performed at room temperature without a heat requirement. Although there are numerous advantages to photoepoxidation, the practical applications for this approach are still challenges with many limitations, such as an inefficiency to exploit visible light, an unsatisfactory PO yield (low PO selectivity and inefficient propylene conversion), or the possible deactivation of a photocatalyst [6]. Therefore, although serious efforts have been made regarding the design of novel photocatalytic materials, the optimization of the reaction conditions required to meet these technical remains a challenge.

Designing Photocatalysts and the Role of Silica Supports
Many developed photocatalysts that function efficiently under UV and visible light have been successfully proposed for photo-epoxidation, and have remained a subject of significant interest [7][8][9][10][11][12][13][14]16,18,[20][21][22][23][24][25][26][27]. Pichat et al. were the first to screen a series of oxides, such as TiO2, ZrO2, V2O5, ZnO, SnO2, Sb2O4, CeO2, WO3, and an Sn-O-Sb mixed oxide, for the photo-epoxidation of propylene at 320 K under UV-light [21]. Their results indicate that total oxidation predominates over CeO2, TiO2, ZrO2, and ZnO. In contrast, SnO2, WO3, and an Sn-O-Sb mixed oxide photo-catalyze only partial oxidation, whereas Sb2O4 yields a small percentage of CO2 in addition to mild oxidation compounds. However, TiO2 and ZrO2 can produce PO with a selectivity of 10.5% and 5%, respectively. In addition, it was found that V2O5 is photo-inactive under experimental conditions. Surprisingly, V2O5 supported on SiO2 efficiently promotes the partial photo-epoxidation of propylene into aldehydes [26,28]. Yoshida et al. systematically screened 50 types of silica-supported metal oxides for the photo-epoxidation of propylene (as shown in Figure 1) [10]. They found that silica modified with some candidate elements, such as Li, Na, Mg, Ca, Sr, Ba, Y, and La, could achieve a high selectivity to PO; however, their C3H6 conversions were quite low, and similar to those of the bulk silica. Silica loaded with some candidates, including Ti, Zn, Pb, and Bi can achieve high PO yields. Among the metal oxide supports, Ti and Zn are the most effective systems for the photo-epoxidation of propylene. Afterward, photocatalysts based on silica support have received considerable attention, owing to their photo-epoxidation. Some metal oxides supported by silica include Nb2O5 [25], MgO [20], ZnOx [10], PbOx [10], CrOx [10,29,30], BiOx [10], ZrOx [10], V2O5 [14,29], and TiO2 [9,10,12,29]. Silica-based photocatalysts have several advantages [31], including the following: 1. Extensive, highly dispersed, tetrahedrally coordinated metal oxides: some candidates, such as TiO2, V2O5, Mo2O3, and Cr2O3 moieties, could be successfully dispersed and isolated when implanting on the silica matrix. It has been noted that these dispersed and isolated Ti, V, Mo, and Cr transition metal oxides can be easily excited under light irradiation to construct corresponding charge-transfer excited states that engage an electron dispatch from O 2− to M n+ , as shown in Equation This particular property concerns a single-site photocatalyst, which differs from a conventional photocatalyst [29,32]. The highly reactive and selective catalytic epoxidation of propylene under these charge-transfer excited states constructed, in which the states of electron-hole pairs are localized nearby on single-site heterogeneous catalysts, leads to such a significant photo-epoxidation. 2. Localization of the photoexcitation at the moiety because of the electric non-conductance of silica. Silica is a well-known electrical insulator, in theory. Thus, its photoexcitation (electronhole pairs) is localized at the moiety of metal-oxygen and has a long lifetime, which favors photoexcitation. 3. Transparency support material: silica allows UV-visible light to pass through the material, and does not hamper significantly when the light reaches the photoactive sites. 4. Depression of side reaction: silica support is mostly inactive for a photocatalytic reaction. Zeolites and mesoporous silica materials, such as titanium silicalite 1 (TS-1), mobile composition of matter (MCM) [33], Santa Barbara amorphous (SBA-15) [34], FSM-16 [35,36], and Technische Universiteit Delft (TUD-1) [37], are perfect hosts for the preparation of single-site photocatalysts, owing to their highly ordered and open structure. Among the above MCM materials, the structures of which are shown in Figure 2a-c, MCM-41 has received considerable attention [38]. Regarding the synthesis process of these composite materials, two approaches have used different structuredirecting agents, including a true liquid-crystal and a cooperative liquid-crystal [38]. For the true liquid-crystal approach, the presence of precursor inorganic framework materials (normally tetraethyl (TEOS) or tetraethylorthosilicate (TMOS) is not required, since a lyotropic liquidcrystalline phase could be formed at high concentration of the surfactant. On the other hand, it is also possible for this phase to form even at lower concentrations of surfactant molecules for the cooperative liquid-crystal approach.  [38].
The most exciting feature properties of MCM-41 mesoporous material is its regular hexagonal array and hexagonally shaped pore (2-10 nm) structure, a high surface area (up to 1,500 m 2 ٠g −1 ), and excellent thermal stability [39]. Nguyen et al. developed TS-1 and MCM-41 as matrixes to achieve single-site structure photocatalysts [7]. Interestingly, a notable synergetic photocatalytic performance over MCM-41 supporting binary V-/Ti-oxides was achieved during the photo-epoxidation of propylene [6]. Figure 3 clearly shows that two physically mixed photocatalysts (PxV0.05Ti0.2 and PxV0.05Ti0.6) achieved the catalytic performance within the range of the predicted activity. Hence, it is reasonable to conclude that a synergetic enhancement contributes to the significant difference between the predicted and observed performances. Among the candidates, V0.05Ti0.3/MCM-41 photocatalyst performed optimal synergetic photocatalytic activity on both PO formation and C3H6 consumption. Reproduced with permission from [6].
A bismuth-based material, Bi2WO6-TiO2, was recently developed, and has shown an excellent potential performance in photo-epoxidation [16]. Figure 4 clearly shows that two supporting materials with Bi2WO6 (BiWO-Ti50i and BiWO-Ti50i/GS) achieve better PO rates than a simple photocatalyst (Bi2WO6, BW/GS, BW/Sil, and TiO2-P25). However, PO selectivity is low, which becomes the greatest disadvantage of this catalyst system.

Effect of the Light Source
For the photo-epoxidation of propylene, the features of the light source (wavelength spectrum, types of light, light intensity, etc.) are crucial factors for efficient photocatalysis. Yamashita et al. used a high-pressure Hg lamp equipped with various UV cut-filters (λ > 250 nm, λ > 340 nm, λ > 450 nm) to conduct the photo-epoxidation experiment over mesoporous molecular sieve photocatalysts at 295 K [40]. The authors found that the reaction only proceeds on the V/Ti-HMS (hexagonal mesoporous silica) and Cr-HMS photocatalysts when using UV with λ > 340 nm, whereas no reaction occurs over the Ti-HMS photocatalyst. As the reason for this, V ions effectively shift the absorption band of Ti containing a mesoporous molecular sieve (Ti-HMS) toward longer wavelengths, whereas the chromium oxide moieties (Cr 6+ ) dispersed in mesoporous silica are responsible for absorbing the long wavelength of visible light irradiation. In another approach, Nguyen et al. have successfully used artificial sunlight for the first time (a 300 W Xe lamp equipped with an AM1.5G filter) and UV light (a Hg Arc lamp equipped with various filters (365, 320-500, and 250-400 nm)) to drive the photoepoxidation of propylene [19]. Since there is a limit on the light delivered to the photocatalyst that can be successfully absorbed, the authors proposed normalized light utilization (NLU) to indicate that the possible absorbed light could be consumed to activate the photoreaction. NLU = Intensity of light emitted × Normalized absorption capability of photocatalyst (2) As expected, the rate of PO formation and C3H6 consumption correlate with NLU in the log-log scale (as shown in Figure 5). This might conclude that photo-epoxidation over V-Ti/MCM-41 has a similar reaction mechanism under either UV light or UV-visible light/artificial sunlight.  [19].

Effect of Reaction Temperature
In general, gas-solid heterogeneous reactions are significantly influenced by the reaction temperature; however, there is no obvious impact on a heterogeneous photocatalysis [41]. Very early on, Pichat et al. found that a rise in reaction temperature would harm the activity of photocatalytic epoxidation [21]. The results suggest that the temperature significantly influences the distribution in partial oxidation products. In a previous study, Nguyen et al. also found that the reaction temperature obviously alters the distribution of products in the photo-epoxidation of propylene [15]. Figure 6a shows that epoxidation selectivity firstly increases with the rise in temperature within the range of 298-323 K. Subsequently, a drop of selectivity toward PO is observed when the reaction temperature is maintained over 323 K. It should be noted that a rising reaction temperature favors the forming of PA (propionaldehyde), one of the most reactive aldehydes. This result is consistent with the finding by Yoshida et al., who reported that the reaction temperature influences the selectivity to the product and is favorable toward the formation of an aldehyde [28]. In another study, Li et al. also found that increasing the reaction temperature does not favor epoxidation selectivity [42]. The competition of multiple reactions toward different products might be attributed to this phenomenon. Therefore, the distribution of the products will be rearranged (as shown in Figure 6b). In contrast, the reaction temperature also influences the conversion of propylene, by altering the adsorption of reactants and the desorption of products on the surface of photocatalyst. Generally, high temperatures tend to inhibit adsorption, because it is an exothermic process. Contrarily, increased temperature is beneficial to the desorption (endothermic process) and accelerates the surface reaction. As shown in Figure 7a, the reaction temperature has a dual action in terms of the reaction rate: (1) enhancing the desorption of products, which provides more active sites for the photocatalytic reaction; and (2) reducing the adsorption of reactants, which decreases the photocatalytic efficiency [15]. In another approach, Li et al. also observed the dual relationship between the reaction temperature (323-473 K) and the photocatalytic performance for the C3H6 conversion rate over an Au-Ag/TS-1 photocatalysts, as shown in Figure 7b [42].
In conclusion, the effect of reaction temperature on the photo-epoxidation of propylene is relatively complicated, with competing contributions. On the one hand, increasing the reaction temperature can accelerate the surface reaction and favors the desorption of products (endothermic process) to release more active sites for a reaction, resulting in the promotion of a photocatalytic reaction. On the other hand, raising the reaction temperature might suppress the adsorption of reactants (exothermic process), which might inhibit the efficiency of the reaction.

Effects of Oxygen/Propylene Ratio
Previous studies have also tested the effects of the different oxygen/propylene ratios on the photo-epoxidation system [13,20]. Figure 8 clearly illustrates the dependence on the different ratios of oxygen/propylene [13,20]. Obviously, a high oxygen/propylene ratio is favorable for the performance of photo-epoxidation. This result is highly consistent with an observation of direct epoxidation over silver catalysts [43,44]. Without the presence of oxygen, only a slight oxidation pathway towards acetaldehyde (AA) takes place over Mg-loaded silica. The photo-epoxidation of propylene towards PO in the presence of oxygen has been found. The optimal condition for achieving the best PO selectivity among the experiments studied was oxygen/propylene = 2. However, an excess amount of oxygen would push for the formation of other partial oxidation products.

Effects of Co-Feeds
A further important parameter is the effect of co-feeds. This has been carefully investigated by Nguyen et al. in the photo-epoxidation of propylene over V-Ti/MCM-41 under UV-light irradiation of 0.3 mW٠cm -2 at 323 K [27]. Figure 9 presents the correlation between H2O vapor pressure and catalytic activity. It clearly shows that the photocatalytic activity is extremely influenced by the presence/absence of co-feeds; these include H2O and H2. For H2O, the photo-activity is quite complicated. And depends on the partial pressure of the co-feed. In the range of 0.0-0.6 kPa H2O, the photocatalytic activity is increased with the increase of co-feed. On the other hand, the photocatalytic performance is not further promoted when the H2O pressure is over 0.6 kPa. For H2 (5.6 kPa), the presence of H2 could promote both of photocatalytic activity and stability of photocatalyst. It should be noted that the mechanism of photo-epoxidation in the presence/absence of co-feeds might not be similar; therefore, an increase in both the AA selectivity and AA formation rate has been observed in the presence of co-feeds.

Elucidating Mechanisms of Photo-Epoxidation
Since the intrinsic photocatalytic epoxidation mechanism is still not entirely clear, there is a strong effort to elucidate the reaction pathways of photo-epoxidation. In a previous study, Carter and Goddard suggested the reaction pathway to produce PO via an oxypropenyl intermediate through H-atom abstraction [45]. The critical role of the forming products is controlled by two factors: the reaction conditions and the angle of H-atom abstraction between attacking of the propylene and O * (ads). Based on H-atom abstraction, Nguyen et al. proposed possible reaction pathways to create PO and other intermediate species over an MCM-41-based photocatalyst (as seen in Scheme 1) [7]. They came up with the conclusion that the distribution of the products is affected by H-atom abstraction in a limited space, e.g., inside the pore. Under a restricted space, it would favor the selectivity to PO via the formation of the oxypropenyl intermediate. Advances made in the mechanism were presented by Yoshida and co-workers [9], who successfully proposed that molecular oxygen is activated on the charge-transferred (Ti 3+ -OL − )* radical pair, which is constructed under light irradiation (as illustrated in Scheme 2). First, the Ti 3+ moiety will react with O2 to form O2 − . However, O2 − cannot activate propylene by itself. Then, the O L − moiety, which is known to be the center of the hole in lattice oxygen, will react with O2 to form an O3 − oxygen radical species, resulting in a reaction with propylene to yield PO. The authors proposed that O3 − is an electrophilic oxygen species capable of the PO forming. When the OL − moiety reacts with propylene, there is a possible reaction pathway to produce acrolein (AL) and ethanal through H abstraction and C−C bond fission, respectively. Scheme 2. The proposed reaction mechanism in the photo-epoxidation of propylene over an isolated tetrahedral Ti species. Reproduced with permission from [9].
By contrast, Amano et al. suggested that both (O2 − ) and (O3 − ) oxygen radical species, which are created at the photo-formed hole center, can react with C3H6 over a Ti active site to generate PO and its co-products (as shown in Scheme 3a) [11]. For TiO2/SiO2, the active oxygen species is derived directly from molecular oxygen to promote the photo-epoxidation. For V2O5/SiO2, the active oxygen species is formed by incorporating the OL − lattice oxygen of the surface metal oxide species. As expected, the active oxygen species (OL − ) in the excited state of a (V 4+ -OL − )* radical pair attacks the double bonds of C3H6 to generate a (V 4+ -O-C3H6 + ) species. Subsequently, the molecular oxygen takes the photo-induced electron on the V cation to promote the selection to propylene epoxidation (as shown in Scheme 3b) [11]. In another approach, Amano et al. proposed the impact of dispersed V 5+ species on the distribution of products [46]. They found that V2O5/SiO2, which contains less dispersed V 5+ species, would promote the forming of PA. In detail, Scheme 4 depicts the mechanism of the photoepoxidation of propylene with molecular oxygen over V2O5/SiO2 [46]. The photo-excited (V=O)* species of monomeric VO4 tetrahedral dispersed in SiO2 are photocatalytic active sites in the formation of PO, whereas the lowly dispersed vanadium oxide species in SiO2 promotes the formation of PA via the isomerization of PO. Scheme 4. The proposed reaction mechanism in propylene photo-epoxidation with molecular oxygen over V2O5/SiO2 [46].
As mentioned above, both V and Ti metal oxide play an essential role in the photo-epoxidation of propylene. However, there has been a notable lack of answers regarding how V and Ti metal oxide promote the forming of PO. In their previous study, Nguyen et al. proposed a possible reaction mechanism in the photo-epoxidation of propylene on V-and Ti-oxide-modified MCM-41 photocatalysts (as shown in Scheme 5) [6]. Clearly, either V-or Ti-oxide could individually photoepoxidize propylene under light irradiation. They found that the primary products of Ti-oxides are AC, PO, and PA, whereas those of V-oxides are AC, EtOH, and PA. It should be noted that Ti-oxide could generate PO directly from propylene, resulting in a higher achieved PO selectivity compared with V-oxides. Based on the possible traveling species, two possible reaction pathways of the synergetic photo-activity could be proposed in the binary V-/Ti-oxides. First, the direct photoepoxidation to produce PO, which is promoted by the Ti-oxides, is still preserved in the binary V-/Tioxides. Second, the indirect photo-epoxidation to produce PO is further enhanced by the possible traveling of PA. The PA traveling, which is produced on the Ti-sites, might travel to the V-sites and be photo-transformed on the V-sites. The V-sites, as compared with the Ti-sites, have a higher possibility to produce the PO through an indirect pathway. Hence, when PA traveling is phototransformed in the V-sites, the photo-epoxidation is dramatically promoted. Scheme 5. The proposed synergetic reaction mechanism in photo-epoxidation of propylene over V-Ti/MCM-41. Reproduced with permission from [6].

Comparison of the Photo-Epoxidation Process and Others
Although the experimental conditions required for photo-epoxidation of propylene differ, it is worth comparing their performance in terms of the selectivity and product yield (as shown in Figure  10). Amano et al. proposed V2O5/SiO2 photocatalyst as an optimal candidate in their group [11]. By modifying the photo-excited lattice oxygen of V2O5/SiO2 through an alkali, Amano et al. observed a high PO yield of 10.1 g·kgcat −1 ·h −1 [14]. In another approach, BiWO-Ti50i/GS performed an optimal performance, with a PO yield of 16.9 g·kgcat −1 ·h −1 [16]. In a recent study, Nguyen et al. proposed a V-Ti/MCM-41 in which its PO yield was boosted more than two-fold compared with BiWO-Ti50i/GS [6].
To further understand the current status of performance over different propylene epoxidation processes, a broad comparison based on the selectivity and turnover frequency (TOF) was successfully collected and evaluated ( Figure 11). As mentioned earlier, HPPO is a primary green technology currently used in the production of industrial compounds. For the two HPPOs of the commercial Halcon styrene monomer and Degussa-Huls-Headwaters hydrogen peroxide process, their TOFs could reach up to 0.11 s -1 and 2.2 × 10 -2 s -1 , respectively [47]. It should be noted that the required TOF for industrial-scale production is approximately 0.1 s -1 . Gold deposited catalysts have recently received considerable attention for propylene epoxidation via hydro oxidation with the presence of O2 and H2 mixtures. Among potential candidates [48][49][50], a 0.05 Au/TS-1(36) catalyst provided a high TOF of 0.33 s -1 . For homogeneous systems, TOF was found to be within the range of 5.7 × 10 -3 to 2.0 × 10 -2 s -1 [47,51]. The aforementioned catalytic processes afford an attractive TOF and selectivity; however, the requirement of either H2O2 or H2 as the oxidant raises economic and safety concerns. The biological process performs an excellent TOF (12 s -1 ) [52]. However, this process also raises some challenges, including the need for life-support systems, the product toxicity for microorganisms, the supply of cofactors, and the stability of the enzymes. To solve the above issues, direct photo-epoxidation with molecular oxygen was successfully proposed. However, the current result still remains far from the required TOF for large-scale production (TOF = 0.1 s -1 ). For more details, Table 1 summarizes all of the above reaction conditions and their performance with regard to propylene epoxidation.

Summary and Future Perspectives
This review is expected to serve as comprehensive background knowledge and to provide researchers with insight into the recent developments of the direct photo-epoxidation of propylene. This process using molecular oxygen (ideal oxidant with 100 wt % of active oxygen), which utilizes light energy in an active reaction, as a promising technology used to produce PO, owing to its potentially clean, low-cost, and straightforward advantages. Numerous efforts have focused on the design of new photocatalyst systems for optimizing the photocatalytic reaction conditions and elucidating the mechanisms of photo-epoxidation. Although significant improvements in photoefficiency have been obtained for photocatalytic epoxidation, the results are still far from the required TOF for industrial-scale production (TOF = 0.1 s -1 ). Therefore, a goal in the coming years will be the continual development of new and effective photocatalyst systems. To promote the feasibility of and open up new opportunities for the photo-epoxidation of propylene, the following criteria should be directly addressed: • The mechanisms and kinetics of the photo-epoxidation of propylene should be elucidated and determined at a fundamental level. Understanding their roles can support the design of effective photocatalysts in the future.

•
Novel photocatalysts should be tailored that can widen visible light absorption, strengthen the forward reaction, depress the reverse reaction, achieve a recombination of photogenerated electron-hole pairs, prolong the lifetime of photogenerated electron-hole pairs, and favor the adsorption of reactants, with a particular focus on the metal oxides supported on zeolites and mesoporous silica photocatalytic materials.

•
Novel photocatalytic systems, including photoreactors and reaction conditions that can upgrade the mass transfer, photon transfer, distribution, and usage of the light source should be designed and prepared for a scale-up through a reconstruction of the current systems.