Advances in Designing Au Nanoparticles for Catalytic Epoxidation of Propylene with H2 and O2

Au nanoparticles, which can be used in various industrial and environmental applications, have drawn substantial research interest. In this review, a comprehensive background and some insights are provided regarding recent studies concerning the use of Au nanoparticles for catalytic propylene epoxidation with H2 and O2. Over the last two decades, substantial progress has been made toward the efficient production of propylene oxide (PO); this includes the design of highly dispersed Au catalysts on Ti-modified mesoporous silica supports, the optimization of catalytic epoxidation, and the determination of the mechanisms and reaction pathways of epoxidation. Particularly, the critical roles of catalyst synthesis, the types of material support, Au nanoparticle sizes, and the dispersion amounts of Au nanoparticles are emphasized in this review. In future studies, novel, practical, robust, and highly PO-selective Au nanoparticle catalyst systems are expected to be continually designed for the enhanced catalytic epoxidation of propylene.

Although the chlorohydrin processes play the most significant role in epoxide propylene in the industry currently, these processes involve the dehydrochlorination of chlorohydrins, leading to the formation of various harmful byproducts (brine of chlorine salts). Hydroperoxide (indirect oxidation or coproducts) processes-including styrene monomer propylene oxide (SMPO), ter-butyl alcohol (TBA), and cumene hydroperoxide (CH)-are more friendly to the environment. However, these routes produce a significant amount of coproducts (styrene, ter-butyl alcohol, and dimethyl benzyl alcohol), which heavily depend on market demand. Other processes, such as direction oxidation (catalytic epoxidation and photocatalytic epoxidation), electrochemical, and biochemical processes, also receive considerable attention [1,2,[6][7][8][9][10][11][12][13]. It is noted that an excellent turnover frequency (TOF, 12 s −1 ) could be observed via the biological process [6]. However, the significant challenges of this process are the product toxicity toward microorganisms and the stability of the enzymes. For the direction oxidation, their current TOF is still far from the requirement for industrial-scale production. Among alternative processes for the epoxidation of propylene, the hydrogen peroxide propylene oxide (HPPO) route has been gaining attention, which offers several advantages compared to the traditional chlorohydrin and coproduct processes. For example, there are no chlorinated compounds involved that could potentially be harmful to the environment, as observed in the chlorohydrin processes, and there is no production of coproducts with lower demand and market values, as observed in the hydroperoxide process. However, the high production cost of hydrogen peroxide makes the processes involving hydrogen peroxide unsustainable. Fortunately, there is a proposed HPPO two-step process that involves the in situ production of hydrogen peroxide via the presence of O2 and H2. Hayashi et al. firstly reported that using a mixture of O2 and H2 with Au nanoparticles (2-5 nm) dispersed in titanium dioxide (TiO2) could effectively epoxidize propylene to yield PO with favorable results (PO selectivity > 90%, propylene conversion = 1%-2%) [14]. The proposed reaction pathway for the epoxidation of propylene has two steps: (1) hydrogen peroxide is firstly produced on the Au surfaces, and then (2) hydrogen peroxide is transported close to (−OOH) on the tetrahedralcoordinated Ti 4+ centers to react with propylene ( Figure 2) [15][16][17]. Since the proposal of the Although the chlorohydrin processes play the most significant role in epoxide propylene in the industry currently, these processes involve the dehydrochlorination of chlorohydrins, leading to the formation of various harmful byproducts (brine of chlorine salts). Hydroperoxide (indirect oxidation or coproducts) processes-including styrene monomer propylene oxide (SMPO), ter-butyl alcohol (TBA), and cumene hydroperoxide (CH)-are more friendly to the environment. However, these routes produce a significant amount of coproducts (styrene, ter-butyl alcohol, and dimethyl benzyl alcohol), which heavily depend on market demand. Other processes, such as direction oxidation (catalytic epoxidation and photocatalytic epoxidation), electrochemical, and biochemical processes, also receive considerable attention [1,2,[6][7][8][9][10][11][12][13]. It is noted that an excellent turnover frequency (TOF, 12 s −1 ) could be observed via the biological process [6]. However, the significant challenges of this process are the product toxicity toward microorganisms and the stability of the enzymes. For the direction oxidation, their current TOF is still far from the requirement for industrial-scale production. Among alternative processes for the epoxidation of propylene, the hydrogen peroxide propylene oxide (HPPO) route has been gaining attention, which offers several advantages compared to the traditional chlorohydrin and coproduct processes. For example, there are no chlorinated compounds involved that could potentially be harmful to the environment, as observed in the chlorohydrin processes, and there is no production of coproducts with lower demand and market values, as observed in the hydroperoxide process. However, the high production cost of hydrogen peroxide makes the processes involving hydrogen peroxide unsustainable. Fortunately, there is a proposed HPPO two-step process that involves the in situ production of hydrogen peroxide via the presence of O 2 and H 2 . Hayashi et al. firstly reported that using a mixture of O 2 and H 2 with Au nanoparticles (2-5 nm) dispersed in titanium dioxide (TiO 2 ) could effectively epoxidize propylene to yield PO with favorable results (PO selectivity > 90%, propylene conversion = 1%-2%) [14]. The proposed reaction pathway for the epoxidation of propylene has two steps: (1) hydrogen peroxide is firstly produced on the Au surfaces, and then (2) hydrogen peroxide is transported close to (−OOH) on the tetrahedral-coordinated Ti 4+ centers to react with propylene ( Figure 2) [15][16][17]. Since the proposal of the aforementioned pathway, extensive research has been carried out with regard to designing novel supported Au catalysts (e.g., Au/Ti-contained catalysts) for driven epoxidation.
Catalysts 2020, 10, 442 3 of 23 aforementioned pathway, extensive research has been carried out with regard to designing novel supported Au catalysts (e.g., Au/Ti-contained catalysts) for driven epoxidation. Herein, we report a comprehensive background and some insights on the use of Au nanoparticles for catalytic propylene epoxidation with O2 and H2. Special attention is directed toward the fabrication of Au nanoparticles for the catalytic epoxidation of propylene. Furthermore, substantial progress on the matter is reviewed, including the design of highly dispersed Au catalysts on Ti-modified mesoporous silica supports, the optimization of the catalytic epoxidation conditions, and the determination of the mechanisms and reaction pathways of epoxidation.

Size Dependent Reactivity of Au Nanoparticles
As the pioneers in this field, Haruta et al. firstly proposed that Au nanoparticles (<5 nm) with various supports, such as α-Fe2O3, Co3O4, and NiO, effectively promote the oxidation of carbon monoxide [18]. This study revealed a new concept for several catalytic applications, including catalytic epoxidation. Haruta et al. also found that Au nanoparticles (2-5 nm) deposited on anatase TiO2 [14], Ti-MCM-41 [19], and Ti-MCM-48 [15] were crucial for the successful epoxidation of propylene. Hayashi et al. reported that Au hemispherical particles (<4 nm) enhance epoxidation, with excellent selectivity to PO (>90%) and that Au particles with sizes less than 2 nm promote the production of propane [14]. Feng et al. reported that the influence of average-sized Au particles on catalytic epoxidation (e.g., PO selectivity and PO formation rate) increases as the dispersion amount of Au nanoparticles increases (0.06-0.40 wt.%) (Figure 3a-e) [20]. PO and H2 selectivity and the PO formation rate increase remarkably as the Au particle size decreases (Figure 3f-g). Continuing their work on Au supported titanosilicates, Taylor et al. showed that significant activity is attributed to Au nanoparticles (<2 nm), and 116 gPO•kgcat −1 •h −1 is the highest PO formation rate that could be observed for 0.05 wt.%Au/TS-1 (Si/Ti = 36) at 200 °C [21]. Huang et al. clearly indicated that the catalytic activity of Au clusters (1-2 nm) on exterior TS-1 surfaces was higher than that of smaller Au clusters (<0.55 nm) inside TS-1 micropores [22]. The results are summarized in Table 1. In most of these studies, Au particles with their sizes in the range of 2-5 nm are considered as the active sites for the epoxidation of propylene [21,23]. This suggests that controlling the size of Au nanoparticles strongly influences the catalytic epoxidation performance. Herein, we report a comprehensive background and some insights on the use of Au nanoparticles for catalytic propylene epoxidation with O 2 and H 2 . Special attention is directed toward the fabrication of Au nanoparticles for the catalytic epoxidation of propylene. Furthermore, substantial progress on the matter is reviewed, including the design of highly dispersed Au catalysts on Ti-modified mesoporous silica supports, the optimization of the catalytic epoxidation conditions, and the determination of the mechanisms and reaction pathways of epoxidation.

Size Dependent Reactivity of Au Nanoparticles
As the pioneers in this field, Haruta et al. firstly proposed that Au nanoparticles (<5 nm) with various supports, such as α-Fe 2 O 3 , Co 3 O 4 , and NiO, effectively promote the oxidation of carbon monoxide [18]. This study revealed a new concept for several catalytic applications, including catalytic epoxidation. Haruta et al. also found that Au nanoparticles (2-5 nm) deposited on anatase TiO 2 [14], Ti-MCM-41 [19], and Ti-MCM-48 [15] were crucial for the successful epoxidation of propylene. Hayashi et al. reported that Au hemispherical particles (<4 nm) enhance epoxidation, with excellent selectivity to PO (>90%) and that Au particles with sizes less than 2 nm promote the production of propane [14]. Feng et al. reported that the influence of average-sized Au particles on catalytic epoxidation (e.g., PO selectivity and PO formation rate) increases as the dispersion amount of Au nanoparticles increases (0.06-0.40 wt.%) (Figure 3a-e) [20]. PO and H 2 selectivity and the PO formation rate increase remarkably as the Au particle size decreases (Figure 3f, [22]. The results are summarized in Table 1. In most of these studies, Au particles with their sizes in the range of 2-5 nm are considered as the active sites for the epoxidation of propylene [21,23]. This suggests that controlling the size of Au nanoparticles strongly influences the catalytic epoxidation performance.   Since determining the driving factor of the size-dependent catalytic activity could provide guidelines to continuously improve epoxidation performance, various factors such as the corner, edge, perimeter, and surface sites of Au nanoparticles were determined via theoretical calculations that allow the imaging of the active catalyst [20,24,25]. Feng et al. proposed a simple Au truncated cuboctahedron shapes physical model with (1 1 1) and (1 0 0) top facets (Figure 4a) [20]. The principles of the model are as follows: (a) the metal atoms occupy crystallographic positions, (b) the numbers of atoms of particles are fixed and predetermined, and (c) the crystallites are efficiently shaped to minimize free energy. Feng et al. assumed that only a single type of Au site plays an active role in the catalytic activity. Based on the above assumptions, the Au atom fraction over different types of site was determined, as shown in Figure 4b. Clearly, perimeter and corner sites played a significant role in the low Au nanoparticle size (<2 nm). Except for the surface site model, other models showed that Au nanoparticles would be inactive when the Au nanoparticle was larger than 5 nm. Based on this hypothesis, the correlation between the PO formation rate and the Au nanoparticle size for different types of site model was established, as illustrated in Figure 4c. Clearly, the catalytic performance in surface, edge, and perimeter site models did not strongly depend on the Au nanoparticle size. On the other hand, the performance on the corner site not only gave a strong size dependence but also agreed with the experimental results, suggesting that the Au corner site played the leading role in the promotion of catalytic epoxidation. Some possible reasons could be used to explain the enhanced performance by the corner model as follows: oxygen could be more readily adsorbed on the Au corner [26]; effectively using hydrogen peroxide, which was formed on small Au nanoparticles, could provide a shorter pathway to Ti 4+ sites for the catalytic epoxidation of propylene.
Catalysts 2020, 10, 442 6 of 23 with the experimental results, suggesting that the Au corner site played the leading role in the promotion of catalytic epoxidation. Some possible reasons could be used to explain the enhanced performance by the corner model as follows: oxygen could be more readily adsorbed on the Au corner [26]; effectively using hydrogen peroxide, which was formed on small Au nanoparticles, could provide a shorter pathway to Ti 4+ sites for the catalytic epoxidation of propylene.  [20], copyright 2014, Elsevier B.V.

Impact of Support Materials for Au Nanoparticles
Since the early work by Yap et al. [14], various Au nanoparticle catalysts deposited on different supports were proposed for the epoxidation of propylene [15,[19][20][21][22][23]. There are essentially three functional advantages of support materials for Au nanoparticles. In the first, support materials have a great influence on selective epoxidation products. Haruta et al. reported an important interaction between the Au nanoparticles and the Ti-sites on the support [36]. The anatase structure promoted

Impact of Support Materials for Au Nanoparticles
Since the early work by Yap et al. [14], various Au nanoparticle catalysts deposited on different supports were proposed for the epoxidation of propylene [15,[19][20][21][22][23]. There are essentially three functional advantages of support materials for Au nanoparticles. In the first, support materials have a great influence on selective epoxidation products. Haruta et al. reported an important interaction between the Au nanoparticles and the Ti-sites on the support [36]. The anatase structure promoted selectivity to PO, and the TiO 2 caused complete combustion to CO 2 . Additionally, microporous titanium silicates (e.g., TS-1, TS-2, and Ti-β zeolite) prefer the formation of propionaldehyde. In the second, support materials play an effective role in enhancing the reactive adsorption ability of reactants and the hydrogen efficiency. Some studies focused on designing isolated and highly dispersed tetrahedral-coordinated Ti-oxide centers in the ligand-to-metal charge-transferred state ((Ti 4+ −O 2− )*), which is associated with the adsorption of propylene [5,48] and the nucleation sites for Au nanoparticles [49]. In previous studies, Au nanoparticles were deposited on Ti-modified mesoporous silica supports, such as SiO 2 [27,28], TUD [29], SBA-15 [30,31], MCM-41 [19,32,33], MCM-48 [15], MCM-36 [34], YNU-1 [34], MWW [34,35], TS-1 [20][21][22][23][37][38][39][40][41][42][43][44][45][46]50], and TS-2 [47], to enhance the epoxidation of propylene. In the last, support materials play a key role in slowing the deactivation. Currently, one of the economic and most technical challenges that requires further research is reducing the deactivation of Au-modified catalysts over time. Uphade et al. reported that Au/Ti-MCM-48 and Au/Ti-MCM-41 exhibit very high initial conversions (5.6% and 5.1%, respectively); however, these values decrease sharply with time (2.6%-2.7% after 2 h of time-on-stream) (Figure 5a) [15]. A similar phenomenon was observed regarding PO selectivity and H 2 conversion. The loss of activity is speculated to arise from the aggregation of Au nanoparticles [23]. Other possible reasons include the production of the bidentate species and other organic fragments that block the active sites [43] or micropore blocking in Au-supported microporous catalysts [51]. Recently, Xu et al. developed novel hybrid core-shell materials that comprise TS-1 zeolite in the core and mesoporous silica in the shell (Figure 5b). Au nanoparticles were successfully incorporated in the inner mesochannels with the range of 2-4 nm (Figure 5c) [45]. Interestingly, Au/TS-1@meso-SiO 2 exhibited a stable propylene conversion (2.6%-2.9%) and high PO selectivity (93%-95%) after 54 h of time-on-stream ( Figure 5d). As observed via the TEM (Figure 5c), the aggregation of Au nanoparticles could be restrained by the confinement of the mesopores during the catalytic reaction, thereby prolonging the stable performance. The detailed two-step reaction is described in Figure 5e: (a) hydrogen peroxide was produced in situ from H 2 and O 2 via the Au nanoparticles in the shell and (b) hydrogen peroxide was diffused into the Ti-tetrahedral sites located in the micropores of the TS-1 core to create the Ti-peroxo active species (Ti-OOH) that could epoxidize the C=C bond of propylene to generate PO. Feng et al. also proposed that Au dispersed in core-shell TS-1/silicalite-1 catalysts exhibited high catalytic stability and that the PO formation rate could reach up to 126 g·kgcat −1 ·h −1 , with high PO selectivity (>80%), for over 100 h of time-on-stream (Figure 5f,g) [52]. In another approach, Feng et al. designed a novel hydrophobic hierarchical TS-1 support (HTS-1) that maintained a high PO formation rate (150 g·kgcat −1 ·h −1 ) for over 100 h of time-on-stream [53]. The boosting properties of mass transfer and the hydrophobicity of the Au/HTS-1 catalyst diminish coking, extending the high catalytic activity.

Preparation of Dispersed Au Nanoparticle Catalysts
Taking all of the relationships between the structural catalysts and catalytic performance together, it is scientifically interesting to discuss the preparation methods of dispersed Au nanoparticles. Since the efficiency in the propylene epoxidation strongly depends both on the size of gold nanoparticles and the type of supports, it is clear that the preparation of these catalysts is equally crucial. In most cases, the deposition-precipitation method is used to prepare Au-Ti 4+ catalysts (e.g., Au/TS-1), as Au nanoparticles can be controlled such that they selectively disperse close to the Ti sites rather than the Si sites [42,54]. On the other hand, the impregnation method has also been proposed; however, this method only produces large spherical Au particles (several tens of nanometers in

Preparation of Dispersed Au Nanoparticle Catalysts
Taking all of the relationships between the structural catalysts and catalytic performance together, it is scientifically interesting to discuss the preparation methods of dispersed Au nanoparticles. Since the efficiency in the propylene epoxidation strongly depends both on the size of gold nanoparticles and the type of supports, it is clear that the preparation of these catalysts is equally crucial. In most cases, the deposition-precipitation method is used to prepare Au-Ti 4+ catalysts (e.g., Au/TS-1), as Au nanoparticles can be controlled such that they selectively disperse close to the Ti sites rather than the Si sites [42,54]. On the other hand, the impregnation method has also been proposed; however, this method only produces large spherical Au particles (several tens of nanometers in diameter) that are neither selective nor active for propylene epoxidation [55,56]. Furthermore, solid grinding [22] and ionic liquid immobilization [57] methods were also proposed to synthesize Au-Ti 4+ catalysts. The advantages and disadvantages of the proposed methods are as follows.

Deposition-Precipitation Method
Among the several techniques used for dispersing Au nanoparticles on supports, the depositionprecipitation (DP) method has yielded considerably successful results. To examine the effect of DP parameters on Au deposition in TS-1, Yap et al. studied the influence of gold concentrations within the range of 0.98-6.37 wt.% and a pH of 4-10 on the morphology of Au nanoparticles and their catalytic activity [23]. They found that a high pH promotes gold loading on the support (Figure 6a); however, it does not substantially alter the average size of the Au nanoparticles (Figure 6c,d). The Au capture efficiency is quite low (<2.5%) and requires further studies for optimization (Figure 6b).
Catalysts 2020, 10, 442 9 of 23 diameter) that are neither selective nor active for propylene epoxidation [55,56]. Furthermore, solid grinding [22] and ionic liquid immobilization [57] methods were also proposed to synthesize Au-Ti 4+ catalysts. The advantages and disadvantages of the proposed methods are as follows.

Deposition-Precipitation Method
Among the several techniques used for dispersing Au nanoparticles on supports, the depositionprecipitation (DP) method has yielded considerably successful results. To examine the effect of DP parameters on Au deposition in TS-1, Yap et al. studied the influence of gold concentrations within the range of 0.98-6.37 wt.% and a pH of 4-10 on the morphology of Au nanoparticles and their catalytic activity [23]. They found that a high pH promotes gold loading on the support (Figure 6a); however, it does not substantially alter the average size of the Au nanoparticles (Figure 6c-d). The Au capture efficiency is quite low (<2.5%) and requires further studies for optimization (Figure 6b). To increase Au capture efficiency, the hydrophobic property of supports should be decreased, the size of the confined Au nanoparticles must be within 2-5 nm, and there could be a post-treatment with promoters (e.g., K + , Cs + , Mg 2+ , Ca 2+ , Sr 2+ , and Ba 2+ ) [43], Cs + [33,37], NH 4+ [58], trimethylamine [59], and Ba 2+ [29]. Lu et al. reported that treatment with K + and Cs + did not affect gold loading, whereas treatment with Mg 2+ , Ca 2+ , Sr 2+ , and Ba 2+ increased gold loading. The probable mechanisms for this phenomenon are shown in Figure 7 [43]. As TS-1 has a negatively charged surface, it cannot easily attract Au anions, as shown in Figure 7a. Furthermore, M + might not have a sufficiently strong positive charge; consequently, the Au anions are not attracted to the support surface (Figure 7b). By contrast, M 2+ possesses a sufficient positive charge to attract Au anions (Figure 7c). To increase Au capture efficiency, the hydrophobic property of supports should be decreased, the size of the confined Au nanoparticles must be within 2-5 nm, and there could be a post-treatment with promoters (e.g., K + , Cs + , Mg 2+ , Ca 2+ , Sr 2+ , and Ba 2+ ) [43], Cs + [33,37], NH 4+ [58], trimethylamine [59], and Ba 2+ [29]. Lu et al. reported that treatment with K + and Cs + did not affect gold loading, whereas treatment with Mg 2+ , Ca 2+ , Sr 2+ , and Ba 2+ increased gold loading. The probable mechanisms for this phenomenon are shown in Figure 7 [43]. As TS-1 has a negatively charged surface, it cannot easily attract Au anions, as shown in Figure 7a  Reproduced with permission from reference [43], copyright 2009, Elsevier B.V.
Cumaranatunge and Delgass used NH4NO3 to post-treat TS-1 supports to enhance the Au capture efficiency [58]. The Au capture efficiency reached up to 69.7%, corresponding to a favorable PO formation rate of 60 g٠kgcat −1 ٠h −1 . In another approach, Feng et al. clearly observed the influence of the slurry aging time in the DP method on the size of the Au nanoparticles [42]. Typically, there is a substantial amount of small-sized Au complexes with a low mass transfer resistance that could migrate into the micropores of the support when the aging time is prolonged. Therefore, the prepared catalyst had a substantial number of small Au nanoparticles inside the micropores, leading to an enhanced PO formation rate (from 4-10 h of aging, Figure 8a). However, further increasing the slurry aging time leads to the dispersion of Au (3.2-3.8 nm in particle size) outside TS-1 (Figure 8b-e), resulting in a decrease in the PO formation rate (18-24 h of aging, Figure 8a). Cumaranatunge and Delgass used NH 4 NO 3 to post-treat TS-1 supports to enhance the Au capture efficiency [58]. The Au capture efficiency reached up to 69.7%, corresponding to a favorable PO formation rate of 60 g·kgcat −1 ·h −1 . In another approach, Feng et al. clearly observed the influence of the slurry aging time in the DP method on the size of the Au nanoparticles [42]. Typically, there is a substantial amount of small-sized Au complexes with a low mass transfer resistance that could migrate into the micropores of the support when the aging time is prolonged. Therefore, the prepared catalyst had a substantial number of small Au nanoparticles inside the micropores, leading to an enhanced PO formation rate (from 4-10 h of aging, Figure 8a). However, further increasing the slurry aging time leads to the dispersion of Au (3.2-3.8 nm in particle size) outside TS-1 (Figure 8b-e), resulting in a decrease in the PO formation rate (18-24 h of aging, Figure 8a).
Notably, Feng et al. found that different charging sequences could influence the transferability of Au complexes into the micropores of supports. Typically, the post-treatment of prefilling the micropores with H 2 O had the effect of directing Au onto a specific area of the supports, thereby altering the catalytic performance [51].

Solid Grinding Method
The solid grinding (SG) method is an effective technique to prepare Au nanoparticles on Timodified mesoporous silica supports. Huang et al. proposed a method of depositing small Au nanoparticles (< 2 nm) on TS-1 supports using dimethyl Au (III) acetylacetonate [39]. In this method (Figure 10a), the alkaline pretreatment of the TS-1 surface creates numerous defects and small

Solid Grinding Method
The solid grinding (SG) method is an effective technique to prepare Au nanoparticles on Ti-modified mesoporous silica supports. Huang et al. proposed a method of depositing small Au nanoparticles (<2 nm) on TS-1 supports using dimethyl Au (III) acetylacetonate [39]. In this method (Figure 10a), the alkaline pretreatment of the TS-1 surface creates numerous defects and small mesopores where organic Au complexes could be actively fixed, leading to the formation of confined Au clusters (<2.0 nm) during the synthesis of the catalyst. Approximately 90% of the Au nanoparticles had diameters in the range 1.0-2.0 nm (Figure 10c,e). Without pretreatment (Figure 10a), the Au clusters (<2.0 nm) could easily aggregate to create several large Au nanoparticles (>2.0 nm, Figure 10b,d). Huang et al. suggested that the small Au clusters (<0.55 nm) deposited on alkaline-treated TS-1 via DP are not active for PO synthesis and that they exhibit a lower PO formation rate than the clusters produced via SG (Au particle sizes in the range 3.2-3.8 nm) [22].

Ionic liquid Immobilization Method
The ionic liquid immobilization (ILI) technique has been recently proposed to prepare Au nanoparticles catalysts [57]. Here, 1-butyl-3-methylimidazolium tetrafluoroborate was used to alter the surface of TS-1 supports, effectively immobilizing the biosynthesized gold nanoparticles onto the

Ionic Liquid Immobilization Method
The ionic liquid immobilization (ILI) technique has been recently proposed to prepare Au nanoparticles catalysts [57]. Here, 1-butyl-3-methylimidazolium tetrafluoroborate was used to alter the surface of TS-1 supports, effectively immobilizing the biosynthesized gold nanoparticles onto the support. Interestingly, ILI could achieve an Au capture efficiency of 100 wt.%, which was higher than that of any other technique. In other studies, Shin et al. successfully used tetrakis(hydroxymethyl)phosphonium chloride/NaOH for stabilizing the system [60].

Effect of Reaction Conditions on Catalytic Performance
The accumulation of knowledge regarding enhanced catalytic epoxidation has led to the addressing of the reaction conditions. Numerous researchers had extensively studied reaction temperature [21,41,43,57], space velocity [41,61], and reactant concentration [61,62].

Effect of Reaction Temperature
Among the reaction parameters for the epoxidation of propylene, reaction temperature plays the most important role, as it correlates to the activation of the reactants and the efficiency of the hydrogen peroxide species in catalytic epoxidation [63]. Du et al. observed that the propylene conversion (2.7%-8.8%) and PO formation rates (36-115 g·kgcat −1 ·h −1 ) increase with a rise in the reaction temperature (in the range 220-300 • C) (Figure 11a,b) [57]. This result is consistent with the results obtained by Taylor et al. (Figure 11c) [21] and Zhan et al. (Figure 11e) [41]. However, high reaction temperatures do not improve PO selectivity (81.7%-73.8%) or H 2 efficiency (54.3%-21.0%) (Figure 11a,b,f). As the reaction temperature increases, the selectivity to propanal (Figure 11d) and ethanol (Figure 11f) increases, while the selectivity to PO decreases [21,43].

Effect of Space Velocity
As shown in Figure 12a,b, with the increase in space velocity, the conversion of propylene abruptly decreased, whereas the PO selectivity and H 2 efficiency increased [41]. The loss of catalytic activity might arise from the inadequate contact time of the reactants when the reaction is conducted at high space velocity. To the contrary, low space velocity favors the total oxidation of reactants and products or the further transformation of PO to secondary products, resulting in a decrease in PO selectivity. In another approach, Lu et al. optimized the reaction process under safe operating conditions. Based on the estimated explosion limits, the simulation was conducted to study the superficial velocity (Figure 12c,d) [61]. To meet the safety requirements, ∆UO 2 must be ≤0. Almost unchanged ∆UO 2 values and hotspot temperatures (462.8-463.2 K) were obtained, owing to the fact that the feed gas mixture was kept constant, and only a slight change in the wall heat transfer coefficient (hw) was investigated. With the rise in superficial velocity, propylene conversion decreased, while PO selectivity increased. Furthermore, the increase in superficial velocity had a negative effect on the H 2 efficiency, PO outlet fraction, and PO yield (Figure 12d).

Effect of Reactant Concentration
From the perspective of chemical safety, attention to the explosion limits of H 2 and O 2 mixtures is required. Schröder et al. reported that these mixtures could be explosive with H 2 concentrations between 3.8 and 95.5 mol.% [64]. Based on the computational model, Lu et al. studied the influence of the inlet propylene-H 2 and O 2 partial pressures on the catalytic performance (Figure 13a,b) [61]. To meet the safety requirements, ∆UO 2 must be ≤0; increases in the inlet propylene-H 2 and O 2 partial pressures increase ∆UO 2 , thereby increasing the probability of a severe explosion. Clearly, the increased inlet H 2 partial pressure would promote propylene conversion but decrease PO selectivity and H 2 efficiency. By contrast, the inlet propylene partial pressure increases PO selectivity and H 2 efficiency; however, it inhibits propylene conversion (Figure 13a). The higher the inlet H 2 and/or propylene partial pressure is, the higher the PO outlet fraction and PO yield are. The increase in the O 2 partial pressure would increase the hotspot temperatures, propylene conversion, H 2 efficiency, PO outlet fraction, and PO yield (Figure 13b). Based on the experiment, Oyama et al. observed an increase in propylene with an increase in the concentrations of H 2 and O 2 , whereas PO selectivity and H 2 efficiency decreased slightly, resulting in a significant increase in the PO rate (Figure 13c,d) [62]. This system could achieve a propylene conversion of up to 5.9%, PO rate of 90 g·kgcat −1 ·h −1 , and H 2 efficiency of 9.4% at the feed composition of H 2 /O 2 /C 3 H 6 /Ar = 40:40:10:10. Compared to O 2 , H 2 had a more significant influence on the PO rate and selectivity.

Effect of Reaction Conditions on Catalytic Performance
The accumulation of knowledge regarding enhanced catalytic epoxidation has led to the addressing of the reaction conditions. Numerous researchers had extensively studied reaction temperature [21,41,43,57], space velocity [41,61], and reactant concentration [61,62].

Effect of Reaction Temperature
Among the reaction parameters for the epoxidation of propylene, reaction temperature plays the most important role, as it correlates to the activation of the reactants and the efficiency of the hydrogen peroxide species in catalytic epoxidation [63]. Du et al. observed that the propylene conversion (2.7%-8.8%) and PO formation rates (36-115 g٠kgcat −1 ٠h −1 ) increase with a rise in the reaction temperature (in the range 220-300 °C) (Figure 11a-b) [57]. This result is consistent with the results obtained by Taylor et al. (Figure 11c) [21] and Zhan et al. (Figure 11e) [41]. However, high reaction temperatures do not improve PO selectivity (81.7%-73.8%) or H2 efficiency (54.3%-21.0%) (Figure 11a,b,f). As the reaction temperature increases, the selectivity to propanal (Figure 11d) and ethanol (Figure 11f) increases, while the selectivity to PO decreases [21,43].

Effect of Reactant Concentration
From the perspective of chemical safety, attention to the explosion limits of H2 and O2 mixtures is required. Schröder et al. reported that these mixtures could be explosive with H2 concentrations between 3.8 and 95.5 mol.% [64]. Based on the computational model, Lu et al. studied the influence of the inlet propylene-H2 and O2 partial pressures on the catalytic performance (Figure 13a,b) [61]. To meet the safety requirements, ΔUO2 must be ≤ 0; increases in the inlet propylene-H2 and O2 partial outlet fraction, and PO yield (Figure 13b). Based on the experiment, Oyama et al. observed an increase in propylene with an increase in the concentrations of H2 and O2, whereas PO selectivity and H2 efficiency decreased slightly, resulting in a significant increase in the PO rate (Figure 13c-d) [62]. This system could achieve a propylene conversion of up to 5.9%, PO rate of 90 g٠kgcat −1 ٠h −1 , and H2 efficiency of 9.4% at the feed composition of H2/O2/C3H6/Ar = 40:40:10:10. Compared to O2, H2 had a more significant influence on the PO rate and selectivity.

Reaction Pathways
In this last section, a brief review focusing on the mechanistic studies of the catalytic epoxidation of propylene with H 2 and O 2 by supported Au nanoparticles is given. Chowdhury et al. conducted in situ UV-vis and electron paramagnetic resonance (EPR) analyses to observe the adsorbed species of oxygen (O 2− ) on the Au surfaces and the structure of the hydroperoxo species on the tetrahedral Ti 4+ in Au/Ti-SiO 2 , which promote the epoxidation of propylene [17]. In this concept, the hydroperoxo species react with propylene to produce the PO, as shown in Figure 14 [65][66][67]. This result is consistent with those of previous studies [15,68,69]. of propylene with H2 and O2 by supported Au nanoparticles is given. Chowdhury et al. conducted in situ UV-vis and electron paramagnetic resonance (EPR) analyses to observe the adsorbed species of oxygen (O 2− ) on the Au surfaces and the structure of the hydroperoxo species on the tetrahedral Ti 4+ in Au/Ti-SiO2, which promote the epoxidation of propylene [17]. In this concept, the hydroperoxo species react with propylene to produce the PO, as shown in Figure 14 [65][66][67]. This result is consistent with those of previous studies [15,68,69]. Based on the products generated during the reaction, Zhan et al. proposed reaction pathways for the epoxidation of propylene using the Au/TS-1 catalyst, as illustrated in Figure 15 [41]. Clearly, acrolein (propenal), along with PO and CO2, were the primary products at the initial stage of the reaction. Acetone and acetaldehyde (ethanal) were formed as secondary products.  Based on the products generated during the reaction, Zhan et al. proposed reaction pathways for the epoxidation of propylene using the Au/TS-1 catalyst, as illustrated in Figure 15 [41]. Clearly, acrolein (propenal), along with PO and CO 2 , were the primary products at the initial stage of the reaction. Acetone and acetaldehyde (ethanal) were formed as secondary products.
Ti 4+ in Au/Ti-SiO2, which promote the epoxidation of propylene [17]. In this concept, the hydroperoxo species react with propylene to produce the PO, as shown in Figure 14 [65][66][67]. This result is consistent with those of previous studies [15,68,69]. Based on the products generated during the reaction, Zhan et al. proposed reaction pathways for the epoxidation of propylene using the Au/TS-1 catalyst, as illustrated in Figure 15 [41]. Clearly, acrolein (propenal), along with PO and CO2, were the primary products at the initial stage of the reaction. Acetone and acetaldehyde (ethanal) were formed as secondary products.

Conclusions
We have comprehensively reviewed and provided insights into the recent developments on Au nanoparticles for the catalytic epoxidation of propylene with H 2 and O 2 . Over the last two decades, substantial progress has been made to improve the production of PO, including the design of highly dispersed Au catalysts on various types of support material, optimization of the catalytic epoxidation conditions, and determination of the mechanisms and reaction pathways of epoxidation. In particular, the critical roles of catalyst synthesis and post-treatment methods, nature of material supports, and size distributions and loading amounts of Au nanoparticles have been emphasized.
Future studies are expected to continually focus on the design of new, practical, robust, and highly PO-selective supported Au nanoparticle catalyst systems for the epoxidation of propylene. To further enhance the catalytic performance, the following criteria should be directly addressed. Firstly, the effects of low and high dispersions of Au nanoparticles, hydrophobic supports, and the mechanisms of loss of stability and deactivation of catalytic activity should be determined at a fundamental level.