Influence of the Support on Propene Oxidation over Gold Catalysts

Abstract

The epoxidation of propene without forming a substantial amount of byproducts is one of the holy grails of catalysis. Supported Cu, Ag and Au catalysts are studied for this reaction and the activity of the supported metals is generally well understood. On the contrary, limited information is available on the influence of the support on the epoxide selectivity. The reaction of propene with equal amounts of hydrogen and oxygen was tested over gold nanoparticles deposited onto CeO₂, TiO₂, WO₃, γ-Al₂O₃, SiO_2, TiO₂-SiO₂ and titanosilicate-1. Several metal oxide supports caused further conversion of the synthesized propene oxide. Strongly acidic supports, such as WO₃ and titanosilicate-1, catalyzed the isomerization of propene oxide towards propanal and acetone. Key factors for achieving high PO selectivity are having inert or neutralized surface sites, a low specific surface and/or a low density of surface -OH groups. This work provides insights and practical guidelines to which metal oxide support properties lead to which products in the reaction of propene in the presence of oxygen and hydrogen over supported gold catalysts.

1. Introduction

Propene oxide (PO) is a chemical intermediate that is widely used in the production of a variety of derivates, including polyether polyols and propene glycol [1,2]. Currently, PO is produced via two industrial processes: the chlorohydrin process and the hydroperoxide process. The chlorohydrin route suffers from the production of chlorinated byproducts that pose environmental issues, while the hydroperoxide route suffers from the mismatch in the market demand of co-products such as styrene monomer and methyl-tert-butylether [1]. Hence the direct epoxidation of propene with O₂ (and H₂) to offer PO without byproducts would be a very attractive option and has received much attention in academia and industry.

In their initial study, Haruta and co-workers reported that Au/TiO₂ and Au/TiO₂-SiO₂ were active and selective catalysts for propene epoxidation with O₂ in the presence of H₂. From the supported metals studied, gold was the only metal active and selective for PO formation [3]. Since then, the Au catalyzed propene epoxidation with H₂ and O₂ has been studied in more detail. Scheme 1 shows the generally accepted mechanism. Au activates H₂ and O₂ to form a peroxide, this is considered the rate determining step [4]. The peroxide is spilt over to a Ti⁴⁺-OH site at the interface between the Au nanoparticle and TiO₂ support. Electrophilic oxygen transfer can occur from the Ti⁴⁺-OOH to the C=C bond to yield PO and Ti⁴⁺-OH and form a stochiometric amount of H₂O as the only by-product [5,6,7,8]. Au is thus essential for the activation of H₂ and O₂, while PO is formed at the interface between Au and the surface of the support.

Au-Ti⁴⁺ catalyzed propene epoxidation is structure sensitive: only gold particles smaller than 5 nm are active and selective for propene epoxidation [9]. The PO turn-over frequency (TOF) of titanosilicate-1 (TS-1) supported gold catalysts with varying gold particle size depends as d^−2.7 on the average Au diameter d, being related to the corner sites of the Au particle. The larger particles of 5.1 nm in the study by Yuan and co-workers show a TOF of ca. 0.02 mole PO produced per mole Au surface-atoms per second (s⁻¹) at 200 °C. The smaller gold particles of 2.6 nm show a TOF of ca. 0.14 s⁻¹ at the same temperature, which clearly indicates the size dependency [9,10]. In addition, Oyama et al. showed that only neutral gold particles catalyze the propene epoxidation while cationic gold(III) oxide particles of 3 nm catalyze propene hydrogenation under the same reaction conditions [11]. Next to gold particle size and charge, also the surface properties of the Ti⁴⁺-based supports such as reducibility and acidity/basicity are important. The effect of metal oxide supports on Au-catalyzed propene epoxidation is not yet well understood.

Metal oxide supports show a wide range of physicochemical properties including different morphology (pore structure and accessibility of active sites), reducibility of the metal cations in the oxide and acid/base properties. Firstly, transition metal oxides are often redox active, which is essential for catalysis of oxidation reactions that follow a Mars-van-Krevelen-mechanism [12,13]. The redox-activities of transition metal oxides enhance the availability of active O-species of the support and explain the high activity of Au/CeO₂ catalysts in CO and volatile organic chemical oxidations [14,15,16]. Besides CeO₂, other examples of reducible supports include many transition metal oxides, including CrO₃, V₂O₅, WO₃, MnO₂ and Fe₃O₄ [12,17,18]. Conversely, α-Al₂O₃, γ-Al₂O₃ and SiO₂ are typical examples of inert and non-reducible oxides [12,17].

Secondly, the acid/base properties of metal oxide supports can be divided by their nature, e.g., protonic/non-protonic, their strength and density. Due to the high electron deficiency of the metal cations that are present in metal oxides, oxides of transition metals are well known to accept electrons and to have non-protonic/Lewis acids sites on the surface, such as is the case on for instance TiO₂, MoO₃ or γ-Al₂O₃ [19,20,21,22]. On the contrary, surface hydroxyl groups tend to act as protonic/Brønsted acids. Adsorption of pyridine or ammonia on Brønsted acid metal oxides, such as Nb₂O₅ or WO₃, results in the formation of pyridinium and ammonium ions, respectively [23,24,25]. The strength of both types of acid sites depends on structural parameters, composition and contamination of the metal oxide and oxidation state and electronegativity of the elements that compose the metal oxide. Adsorption with probe molecules can be used for the determination of the acid strength and is in addition one of the tools to determine the density of acidic sites on the surface of metal oxides [26]. Another viable parameter that describes the acidity is the point-of-zero-charge (PZC). The PZC indicates that a metal oxide is overall acidic with a PZC << pH 7 or basic with a PZC >> pH 7 or amphoteric with a PZC ± pH 7. An amphoteric metal oxide contains both acidic and basic sites [27].

Most studies on Au catalyzed propene epoxidation are based on the use of Ti⁴⁺ containing supports, such as TiO₂, TiO₂-SiO₂ and titanosilicate-1 (TS-1). The formation of Ti⁴⁺-OOH species is crucial for the propene epoxidation reaction as depicted in Scheme 2. Ti⁴⁺-OOH species are both observed in the gold catalyzed propene epoxidation in the gas phase [7,28] and in the liquid phase epoxidation of propene with aqueous H₂O₂ [29,30]. Next to TiO₂, other metal oxides with Lewis acid sites are also reported to activate H₂O₂ in a similar fashion, but are typically studied in liquid phase reactions. Examples of such metal oxides include WO₃ and soluble W⁶⁺ compounds [25,31]. Solid WO₃ has been studied as a support for Ag catalyzed epoxidations, but not for the gold catalyzed propene epoxidation [32]. Moreover, γ-Al₂O₃, an oxide with both acidic and basic surface sites, is reported to activate H₂O₂ and catalyze styrene epoxidation [33] and Au supported on γ-Al₂O₃ is reported to form H₂O₂ from H₂ and O₂ at elevated pressures [34]. A theoretical study by Thomson and co-workers showed isolated trimeric Au clusters supported on γ-Al₂O₃ can be active in the epoxidation of propene with H₂ and O₂ [5], but experimental studies on the use of Au/γ-Al₂O₃ for propene epoxidation are lacking.

Interestingly, metal oxides that are readily reduced and re-oxidized catalyze the H₂O₂ decomposition reaction. MnO₂ reacts violently with aqueous H₂O₂ involving reduction of Mn⁴⁺ to Mn³⁺ and subsequent reoxidation to Mn⁴⁺ in the catalytic mechanism to yield H₂O and O₂ [35]. Sheldon reported that the readily reducible CrO₃, that also has strongly Lewis acidic surface sites, catalyzed the decomposition of peroxide reagents [36]. TiO₂-based supports are of particular interest, since they show both reducibility of the metal oxide and Lewis acidic surface sites. TiO₂ is proposed to exist of tetrahedral Ti⁴⁺ species in the case of TiO₂-SiO₂ and TS-1. Reduction of these tetrahedral Ti⁴⁺ species yields instable Ti³⁺ species that readily oxidize to Ti⁴⁺ in air or Ti⁴⁺-OOH species in the presence of peroxides. However, during propene oxidation over Au/TiO₂-SiO₂ catalysts, Ti³⁺ is not detected [7,28,37]. In short, metal oxides with Lewis acid sites are proposed to activate H₂O₂, which is a crucial intermediate in the epoxidation reaction. H₂O₂ is, however, decomposed by readily reducible metal oxides.

Several individual studies have noticed that some oxides affect the products formed from propene oxidation. This is more extensively studied for supported Ag and Cu catalysts than for supported gold catalysts. Supported Ag and Cu catalysts have been studied using a variety of supports, including α-Al₂O₃, SiO₂, WO₃ and CaCO₃ [32,38,39,40,41]. These catalysts often operate at low conversion and low selectivity. Apparently, the acidic supports catalyze the isomerization and total oxidation of PO [42,43,44]. For supported gold catalysts, the support influence is only reported for Au/TiO₂ and Au/TiO₂-SiO₂ catalysts. Au supported on TiO₂ is reported to deactivate within hours of catalytic testing. The deactivation is proposed to be due to the strong adsorption of PO on Ti-O-Ti sites [45,46]. Dispersed TiO₂ on SiO₂ and a TS-1 with low titanium content show no Ti-O-Ti sites. These catalysts possess a high degree of stability [45,46,47,48], but PO is reported to adsorb on the epoxidation centers of a Au/TS-1 catalyst, thus inhibiting the epoxidation reaction [49]. Acidic Ti⁴⁺ centers of a Au/TiO₂-SiO₂ was shown to catalyze subsequential reactivity of PO [50]. These reports suggest a contribution of the support acidity on the propene epoxidation reaction, but a systematic study is still missing.

In this study the reactivity of PO over a series of frequently used catalyst supports was examined in the gold-catalyzed propene epoxidation in the presence of H₂ and O₂. We focus on how physicochemical properties of metal oxides affect the selectivity during propene epoxidation.

2. Results and Discussion

2.1. Metal Oxide Supported Gold Catalysts

Gold nanoparticles were deposited on metal oxide supports by two different methods: deposition precipitation-urea was performed for the synthesis of Au/TiO₂ and Au/CeO₂, while the other catalysts were prepared using cation adsorption of Au(en)₂Cl₃ [15,51,52,53].

Table 1. Characteristics of the as prepared Au catalysts supported on different metal oxides.

Catalyst	Au Loading (wt%) a	Gold Particle Size (nm) b	PZC (pH)
Au/CeO2	1.89	2.0 ± 0.7	6.2
Au/WO3	0.82	5.2 ± 1.5	3.5
Au/TiO2	0.10	2.7 ± 0.6	5.5
Au/SiO2	0.16	2.3 ± 0.7	6.7
Au/γ-Al2O3	0.08	n.d.	7.8

n.d. not determined, ^a The gold loading was determined using elemental analysis, ^b number averaged gold particle diameter as derived from transmission electron microscopy.

lists the structural properties of the supported Au catalysts. Both methods led to Au nanoparticles of 2–5 nm on the metal oxides and quantitative deposition of the gold with a nominal loading of 0.1 wt% for γ-Al₂O₃ and TiO₂ and 0.2 wt% for SiO₂ and TS-1. The Au/WO₃ and Au/CeO₂ were prepared at a higher gold loading of 1.0 and 2.0 wt%, respectively. No gold particles were observed in the Au/WO₃ catalyst at a lower loading of 0.1%.

The acid–base properties of the gold catalysts were studied by means of their point-of-zero-charge (PZC) [27]. The Au/CeO₂ catalyst is amphoteric with a PZC value of 6.2. The Au/WO₃ is acidic with a PZC value of 3.5. The Au/TiO₂ catalyst is mildly acidic with a PZC value of 5.5. The Au/SiO₂ catalyst is amphoteric with a PZC of 6.7. Finally, the Au/γ-Al₂O₃ is mildly basic with a PZC value of 7.8. The titration curves of the catalysts are included in Figure S1. The observed PZC of the catalysts do not differ much from the pristine supports (Table 2), indicating preservation of the original surface acidity/basicity, except for Au/SiO₂. The SiO₂ support itself is acidic with a PZC of 4.9, much lower than the final Au/SiO₂ catalyst. This can be explained by the presence of Na⁺ from the synthesis, which causes a decrease in the surface acidity [50].

Figure 1 shows representative High-Angle Annular Dark Field Scanning Transmission Electron Microscopy (HAADF-STEM) images of catalysts supported on different metal oxides. The gold particles are visible as bright features on the metal oxide supports. Additional HAADF-STEM images at lower magnification are included as Figures S2 and S3 and show that the gold nanoparticles are homogeneously distributed over the supports. Scanning Electron Microscopy (SEM) images (Figure S4) show the three-dimensional structure of the catalysts and do not show any large gold agglomerates.

Figure 2 shows the HAADF-STEM images of the Au/γ-Al₂O₃ catalyst at low and high magnification. Elemental analysis confirmed the presence of Au, but almost no nanoparticles with sizes above 2 nm were observed, while according to literature this is the minimum size required for propene epoxidation [9].

The supported gold catalysts were tested in propene epoxidation at 200 °C with 10% propene, 10% oxygen and 10% hydrogen in helium at a typical flow of 10,000 mL/g_cat/h. The pristine support materials did not show any activity under the studied conditions. Figure 3 shows the propene and hydrogen conversions of the different supported gold catalysts. In addition, Table S1 contains a full overview of the conversions and selectivities to the various oxidation products and Table S2 shows a comparison of the catalysts tested in this work with similar catalysts in literature. The highest conversion of up to 17.4% as was observed for the Au/CeO₂ catalyst, the high conversion of propene under oxidative conditions using similar Au/CeO₂ catalysts is well known in literature, for example Louis et al. observed 50% propene conversion at 294 °C. [15] The Au/γ-Al₂O₃ catalyst showed a propene conversion below 0.1%. While the other catalysts showed propene conversions of up to 0.5%. The Au/γ-Al₂O₃ catalyst was the least active in the combustion of hydrogen and showed a hydrogen conversion of only 8.9%. The Au/WO₃ and Au/TiO₂ catalysts showed a hydrogen conversion of 27.2% and 34.8%, respectively. The Au/SiO₂ catalyst is more active with a hydrogen conversion of 52.0%, while the Au/CeO₂ catalyst showed a 100% hydrogen conversion. The propene and hydrogen conversions of all catalysts were stable during the catalytic test (Figure S5), with the sole exception of the Au/TiO₂ catalyst which deactivated on-stream (Figures S6 and S7).

Figure 4a presents the activity of the gold catalysts in propene oxidation as function of the gold particle size. The Au/CeO₂ catalyst demonstrated the highest turn-over-frequency (TOF) in propene oxidation of 0.28 s⁻¹, defined as molecules of propene converted per surface Au atoms per second [54]. The Au/WO₃ and Au/TiO₂ were almost inactive in propene oxidation; the TOF of the Au/TiO₂ catalyst of 0.01 s⁻¹ is in agreement with the results of Delgass et al. who observed a TOF of 0.003–0.034 s⁻¹ for various Au/TiO₂ catalysts at 200 °C [55]. The Au/SiO₂ catalyst shows a TOF of 0.04 s⁻¹. The TOF of the Au/γ-Al₂O₃ catalyst is based on the total amount of Au atoms and is <0.01 s⁻¹. The gold catalysts supported on SiO₂ and CeO₂ are the two catalysts with the smallest gold particle size and highest activity in propene and oxidation; however, as discussed in a later section no clear overall relation between Au particle size and activity could be established.

The activity of the supported catalysts in H₂ oxidation to H₂O is given in Figure 4b. The Au/γ-Al₂O₃ catalyst was almost inactive with a TOF of 0.03 s⁻¹. The Au/CeO₂ catalyst showed a TOF of 1.28 s⁻¹, defined as molecules of hydrogen converted per surface Au atoms per second. The hydrogen conversion activity increased in the order of Au/WO₃ (1.40 s⁻¹), Au/TiO₂ (3.10 s⁻¹) and Au/SiO₂ (5.17 s⁻¹). For many catalysts, the hydrogen oxidation activity was much higher than the propene oxidation activity and independent of the gold particle size for the studied catalysts. The oxidation of H₂ is generally ascribed to the activation of H₂ and O₂ on the surface of the Au particles primarily forming *OOH and H₂O₂ intermediates [56,57]. In the case of Au/TiO₂ catalysts this has been studied in detail. There, a spill-over of the peroxide intermediates to the epoxidation sites at the interface between the Au particle TiO₂ support is proposed to be crucial for PO formation [56,58]. If no-spillover occurs, these peroxide intermediates decompose to form H₂O, as is predominantly the case for many of the studied catalysts.

The carbon selectivity towards different oxygenates is shown in Figure 5. The Au/CeO₂ catalyst showed 99.8% CO₂ selectivity, with only trace amounts of propanal as side product. This suggests both a strong interaction between gold and the CeO₂ support and readily reducibility of the CeO₂ support [59]. Trace amounts of acetaldehyde from over-oxidation of propene and propanal were formed over the Au/WO₃ catalyst. The Au/TiO₂ produced a mixture of PO (22% selectivity), acetaldehyde and propanal at low activity. The observed mixture of products is most likely due to the relative high reaction temperature of 200 °C compared to literature [37,55] and has been observed by others as well. For example, Nijhuis and co-workers obtained a Au/TiO₂ catalyst that operated at 0.5% conversion and 20% PO selectivity at 90 °C, [37] whereas a higher selectivity was reported at 50 °C of >99% PO. [60] For comparison reasons, we also tested our Au/TiO₂ catalyst at a lower reaction temperature of 100 °C. There our Au/TiO₂ showed a higher selectivity towards PO (Figures S6 and S7), but quickly deactivated as is in accordance with literature [37]. Our Au/SiO₂ is the only catalyst with a high selectivity towards acrolein, this catalyst also shows the formation of trace amounts of PO. Allylic oxidation of propene to acrolein is reported for Au/SiO₂ based catalysts, [60,61,62] but until now only a single report mentions that Au/SiO₂ can also be active in the epoxidation of propene [63]. The Au/γ-Al₂O₃ catalyst produced trace amounts of acetone, propanal and acetaldehyde but no PO, which is consistent with previous reports. [62] The Au/SiO₂ and Au/CeO₂ catalysts were also studied in propene oxidation without H₂ at 200 °C. Under those condition, the propene conversion over the Au/SiO₂ catalyst decreased below 0.1% to offer a mixture of acrolein (76%) and acetaldehyde (24%). Interestingly, the Au/CeO₂ showed an increased propene conversion of 22.9%, while the selectivity did not change (99% towards CO₂), demonstrating the intrinsic differences in catalysis over the Au/CeO₂ and Au/SiO₂ catalysts.

If the rate determining step in C₃H₆ oxidation took place on the gold, a distinct TOF independence of the gold particle size would be expected. However, the observed differences in propene and hydrogen oxidation activity cannot solely be attributed to the varying gold particle sizes. The Au/γ-Al₂O₃ was the only catalyst that was almost inactive in both propene and hydrogen oxidation. This is most likely due to the absence of 2–5 nm gold nanoparticles. In addition, Au catalysts supported on Al₂O₃ are often reported to be less active for other reactions, such as butadiene hydrogenation and CO oxidation compared to other supports [64,65]. The high activity of other, such as the Au/WO₃ and Au/TiO₂ catalysts, in H₂ oxidation illustrate that they can generate peroxide intermediates and thus could potentially catalyze propene oxidation. The readily reducible Au/CeO₂ catalyst showed the highest activity in complete oxidation of propene, while the other catalysts showed a mixture of oxygenates. Only the Au/TiO₂ and Au/SiO₂ catalyst show the formation of PO. This clearly stresses the importance of the support acidity/basicity and reducibility for the formation of PO and other oxygenates.

2.2. Reactivity of Propene Oxide with Metal Oxide Supports

The influence of the support induced reactivity was studied in more detail by exposing the metal oxide supports to PO and relating the observations to the support reducibility and presence of acidic or basic surface sites.

Figure 6 presents the mass titration curves of the supports, which were used for the determination of their point-of-zero-charge. The physicochemical properties of the different supports are shown in Table 3. Several supports were added to the experiment to clearly show the influence of basicity (CaCO₃), low surface area and limited surface -OH groups (α-Al₂O₃) and Lewis acidity (Titanosilicate-1, TS-1) [66,67]. N₂ physisorption isotherms of the studied supports are included in Figure S8.

The acidity of the metal oxide supports was characterized using infrared spectroscopy with adsorption of pyridine (Py-IR, Figure S9) and temperature programmed desorption of ammonia (NH₃-TPD, Figure S10) and the results are summarized in Table 2. For the CeO₂ support, Py-IR shows both adsorption of pyridine on Lewis acid (electron accepting) sites as well as oxidation of pyridine to α-pyridone caused by Ce⁴⁺ to Ce³⁺ reduction. [68] Due to the low specific surface area of WO₃ an acid concentration of only 22.3 μmol/g_cat was obtained from NH₃-TPD with a desorption maximum between 250–400 °C. NH₃-TPD on TiO₂ shows a 10-fold increase in acid concentration compared to WO₃, while the acid sites are much weaker with a desorption maximum at 255 °C. Py-IR showed that the TiO₂ consists primarily of Lewis acid sites. TS-1 contains an acid concentration of 82.3 μmol/g_cat based on NH₃ desorption, with a desorption maximum at 201 °C, showing that the acidic sites are weaker than for TiO₂. Only a minor amount of acid concentration was observed for SiO₂ using NH₃-TPD, while no acid sites were observed using Py-IR. Although the γ-Al₂O₃ is basic with a PZC of 8.2 NH₃-TPD confirms the presence of acid sites with a maximum desorption at 204 °C.

Table 2. Overview of the characteristics of the support materials used.

Support	BET S.A. (m2/g)	PZC (pH)	-OH Coverage (OH/nm2)	Acid Concentration (μmol/gcat)
CeO2	66	7.2	3.8 [69]	17.7 a LA; 0 a BA
WO3	8	3.1	n.d.	22.3 b
TiO2	52	5.7	3.3 [70]	239.3 b 111.3 a LA; 0.4 a BA
TS-1	549	3.8	n.d.	82.3 b
SiO2	499	4.9	5 [71]	38.2 b 0 a LA; 0 a BA
γ-Al2O3	249	8.2	8 [72,73]	385.3 b
α -Al2O3	8	7.1	3.5 [74]	-
CaCO3	1	9.8	n.d.	-

n.d. not determined, ^a Corresponding to the adsorbed μmol of pyridine per g_cat determined using Py-IR, ^LA Lewis acid, ^BA Brønsted acid, ^b Corresponding to the adsorbed μmol of NH₃ per g_cat determined using NH₃-TPD.

Table 2. Overview of the characteristics of the support materials used.

Support	BET S.A. (m2/g)	PZC (pH)	-OH Coverage (OH/nm2)	Acid Concentration (μmol/gcat)
CeO2	66	7.2	3.8 [69]	17.7 a LA; 0 a BA
WO3	8	3.1	n.d.	22.3 b
TiO2	52	5.7	3.3 [70]	239.3 b 111.3 a LA; 0.4 a BA
TS-1	549	3.8	n.d.	82.3 b
SiO2	499	4.9	5 [71]	38.2 b 0 a LA; 0 a BA
γ-Al2O3	249	8.2	8 [72,73]	385.3 b
α -Al2O3	8	7.1	3.5 [74]	-
CaCO3	1	9.8	n.d.	-

n.d. not determined, ^a Corresponding to the adsorbed μmol of pyridine per g_cat determined using Py-IR, ^LA Lewis acid, ^BA Brønsted acid, ^b Corresponding to the adsorbed μmol of NH₃ per g_cat determined using NH₃-TPD.

The pristine supports were tested under reaction conditions (200 °C, 1/1/1/7 ratio of C₃H₆/H₂/O₂/He at a flow of 10,000 mL/g_support/h). Without any Au none of the supports showed the formation of propene oxide or other oxygenates. Subsequently, stacked bed experiments with top to bottom flow were performed with a PO producing Au/TS-1 catalyst on top and the different supports at the bottom. The Au/TS-1 catalyst will be discussed in more detail in the next section. In this way we tested if the observed selectivity was the results of sequential reactions of the PO catalyzed by the supports.

Figure 7 shows the results of the stacked bed experiments. The different support materials are ordered from left to right according to their PZC from acidic to basic. Shown on the far right as a reference is the observed selectivity to the carbon containing products when no extra support was used downstream.

The carbon mass balance was between 25–50%, hence a large fraction of PO was adsorbed on the support surface during the 2 h of testing (Figure 7a). At prolongated reaction times, the mass balance increased. For example, the α-Al₂O₃ shows a carbon mass balance of 50% after two hours of the test, but after 4 h a mass balance of 73% was achieved. This indicates saturation of the specific sites on the support’s surface with PO and was also observed for the other supports. For these relative short testing times the selectivity to gaseous products was examined in more detail (Figure 7b).

WO₃, TS-1, TiO₂ and γ-Al₂O₃ effectively catalyze the isomerization of PO towards mainly propanal. In addition, WO₃ and TS-1 catalyze the further oxidation of PO towards acetaldehyde. The more inert supports SiO₂, α-Al₂O₃ and CaCO₃ hardly influenced the original product distribution compared to the Au/TS-1 catalyst, except for some adsorption on the surface. The CeO₂ support is unique in catalyzing the complete combustion of the original products towards CO₂.

The adsorption of PO over TiO₂ and TiO₂-SiO₂ supported Au catalysts has been reported in literature. Mul and co-workers showed that the adsorption of PO on Au/TiO₂ and Au/TiO₂-SiO₂ catalysts leads to ring opening of the epoxide and irreversible adsorption of the formed propoxy-species on the TiO₂ sites of the support [45,46]. This is generally accepted to cause deactivation of Au/TiO₂ catalysts, but a comprehensive overview of the reactivity of PO with various metal oxides has not yet been reported.

Scheme 3 illustrates the different types of supports and their reactivity with PO. Neither the PZC of the support nor its specific surface area are very influential factors. The redox active supports (CeO₂, WO₃ and TiO₂) [18,75], cause further oxidation of PO to acetaldehyde and at the end complete combustion of PO to CO₂. This is especially pronounced for CeO₂, the Au/CeO₂ catalyst was active in the complete oxidation of propene to CO₂ at 19.7% conversion.

The metal oxide supports that possess Brønsted acidic (WO₃) or Lewis acidic (TS-1, TiO₂) surface sites [25,76] show isomerization of PO towards propanal and are an example of the heterogeneously catalyzed Meinwald rearrangement [51,77]. The isomerization over basic γ-Al₂O₃ towards acetone and propanal is facilitated by either basic surface sites or Lewis acidic surface sites that are present at surface defects. The presence of these acidic sites was confirmed by NH₃-TPD.

Whereas the metal oxides that contain acidic surface sites show isomerization of the desired PO, the oxide supports that lack acidic surface sites, such as SiO₂ and CaCO₃, show higher PO selectivity. Factors that are beneficial to prevent sequential reactions of the PO once it has been formed are an inert surface as observed for the SiO₂ support, a low specific surface area as for the CaCO₃ support and a low density of surface -OH groups as for the α-Al₂O₃ support

2.3. Au Supported on Ti⁴⁺ Modified Oxides

Until now it has become clear that three important conditions have to be fulfilled simultaneously by a good supported Au catalyst for propene epoxidation: peroxide species should be formed (as is true for almost all supports tested), these species should be able to spill over to the site at which propene is adsorbed (which strongly depends on the Au-support interface) and the support itself should not cause sequential reaction of the propene oxide. Most active and selective gold catalysts for propene epoxidation are often based on TS-1 as support [78,79]. However, as evidenced above, the TS-1 support can catalyze isomerization of PO towards propanal, which is often reported as a side product in propene epoxidation using Au/TS-1 catalysts. Here we compare the activity and selectivity of Au supported on various TiO₂ containing supports in propene epoxidation.

Figure 8 shows a schematic representation of the various TiO₂-supports: TiO₂ grafted on γ-Al₂O₃ (TiO₂-γ-Al₂O₃), TiO₂ grafted on SiO₂ (TiO₂-SiO₂) and two types of TS-1, which differ in having accessible pores (TS-1) or blocked pores (TS-1U). Gold nanoparticles were deposited on TiO₂-γ-Al₂O₃, TiO₂-SiO₂ and the two types of TS-1 by cation adsorption of Au(en)₂Cl₃ with Na₂CO₃ for pH control at a nominal loading of 0.20 wt% for Au/TS-1 and normalized to the specific surface area of the other supports. For the Au/TS-1 catalyst, the template directing reagents were removed prior to gold deposition, gold deposition thus took place in the pores as well as on the surface of the TS-1 crystals. On the contrary, for the Au/TS-1U catalyst, the template directing reagents were removed only after gold deposition, hence gold was present on the surface of the TS-1U crystals rather than in the pores.

Shown in Figure 9 are representative Bright-Field TEM (BF TEM) and HAADF-STEM images of the gold catalysts supported on different TiO₂-based supports. The gold particles are typically between 1.5–4.2 nm in size. Table 3 summarizes the structural properties of the Au catalysts. The PZC of the Ti-modified supports was lower than for pristine γ-Al₂O₃ and SiO₂ (Table 2), which is indicative for the formation of acidic tetrahedral Ti⁴⁺ sites on the supports [45,80]. The absence of oligomeric TiO₂, which is detrimental for propene epoxidation, was confirmed by diffuse reflectance UV-Vis (Figure S11). Upon gold deposition, the PZC typically increased to 6.7–7.7, indicating that Na⁺ was present and neutralized the acidic Ti⁴⁺ sites. The presence of Na⁺ was confirmed by Elemental Dispersive X-ray spectroscopy (Figure S3), but NH₃-TPD shows no change in acid concentration between TS-1 (82.3 μmol_NH3/g) and Au/TS-1 (87.6 μmol_NH3/g) (Figure S10), although their strength is slightly lower with a desorption maximum at 196 °C vs. 201 °C. Only in the case of Au/TS-1U the PZC of the final catalyst does not differ from the TS-1 support. There, the template directing reactants blocked the micropores and Na⁺ did thus not neutralize the acidic Ti⁴⁺ sites in the micropores.

Table 3. Characteristics of the as prepared Au catalysts.

Catalyst.	Au Loading (wt%) a	Gold Particle Size (nm) b	PZC TiO2-Support (pH)	PZC Catalyst (pH)
Au/TiO2-γ-Al2O3	0.08	3.5 ± 0.5	6.0	7.7
Au/TiO2-SiO2	0.16	2.1 ± 0.7	2.9	6.8
Au/TS-1	0.19	3.3 ± 1.0	3.8	6.7
Au/TS-1U	0.22	4.2 ± 1.1	3.8 c	3.8

^a The gold loading was determined using elemental analysis, ^b number averaged gold particle size, ^c the PZC of TS-1.

Table 3. Characteristics of the as prepared Au catalysts.

Catalyst.	Au Loading (wt%) a	Gold Particle Size (nm) b	PZC TiO2-Support (pH)	PZC Catalyst (pH)
Au/TiO2-γ-Al2O3	0.08	3.5 ± 0.5	6.0	7.7
Au/TiO2-SiO2	0.16	2.1 ± 0.7	2.9	6.8
Au/TS-1	0.19	3.3 ± 1.0	3.8	6.7
Au/TS-1U	0.22	4.2 ± 1.1	3.8 c	3.8

^a The gold loading was determined using elemental analysis, ^b number averaged gold particle size, ^c the PZC of TS-1.

The supported gold catalysts were tested in propene epoxidation at 200 °C with 10% propene, 10% oxygen and 10% hydrogen in helium at a flow of 10,000 mL/g_cat/h. Figure 10 shows the propene and hydrogen conversions, while the selectivity to the various oxygenates is shown in Figure 11 and Table S1 contains a full overview of the conversions and selectivities and Table S2 contains a comparison with catalysts reported in literature. The gold catalyst supported on TiO₂-γ-Al₂O₃ showed the lowest propene and hydrogen conversion, of, respectively, 0.5% and 21.9%. Like the Au/γ-Al₂O₃ catalyst, the Au/TiO₂-γ-Al₂O₃ catalyst showed the formation of trace amounts of oxygenates. The gold catalyst supported on TiO₂-SiO₂ showed the highest propene and hydrogen conversion of, respectively, 3.6% and 78.8%, although this catalyst slightly deactivated during the catalytic test. The Au-TiO₂-SiO₂ catalyst showed an initial propene conversion of 4.7% which decreased to 3.6 after 90 min (a loss of 24% of activity). The deactivation of PO producing Au/TiO₂-SiO₂ catalysts is well reported in literature, Haruta et al. showed a deactivation of the propene conversion of 20–30% within 1.5 h over Au/Ti-MCM-41 catalysts. [47] A similar catalyst is reported in literature and studied under the same conditions, there a propene conversion of 1.9% was obtained. [81] The catalysts supported on TS-1 and TS-1U showed a stable propene conversion of 1.7% and 3.0%, respectively, with a hydrogen conversion at 60.5% and 59.2%. The Au/TS-1 catalyst did not show any deactivation during the catalytic test, while the Au/TS-1U showed a deactivation of the propene conversion from 3.2% to 3.0% during the catalytic test. Hence the performance was clearly somewhat better for the TS-1U than for TS-1. The H₂ conversion of all the catalyst was stable during the 1.5 h catalytic test. Compared to literature, Lu and co-workers attempted to deposit Au on uncalcined and calcined TS-1, but due to the low PZC of calcined TS-1 were not able to deposit < 5 nm Au particles on this support. Hence their catalyst was nearly inactive under the studied conditions. [82] Delgass et al. report Au/TS-1 catalysts operating at 2.0–3.5% propene conversion and 3.5–18.9% H₂ conversion (14,000 mL/g_cat/h flow of 10% of each reactant in N₂ at 200–205 °C). The much higher H₂ conversion in our results is likely due to the low Au loading used by Delgass of 0.01 up to 0.09 wt% compared to the 0.22 wt% in our work [83].

Figure 11 shows the selectivity to the various oxygenates. The Au/TiO₂-γ-Al₂O₃ did not produce a significant amount of PO and formed only trace amounts of acetone and propanal. The Au/TiO₂-SiO₂ and Au/TS-1 catalyst both have a high activity and selectivity towards PO, which is in accordance with reports in literature where similar catalysts show >83% selectivity towards PO [81,83]. While Ba(NO₃)₂ promoted and silylated Au/TiO₂-SiO₂ catalysts an increased activity and selectivity towards PO formation. [84] Figure 12 shows the TOF for PO formation, which was 0.06 s⁻¹ for the Au/TS-1 catalyst (0.07 s⁻¹ for propene oxidation) with an average gold Au particles size of 3.3 nm, in line with literature in which Delgass et al. reported a TOF of ca. 0.06 s⁻¹ for 3.4 nm gold particles supported on TS-1 [54]. Remarkedly the Au/TS-1U catalysts showed a much higher propene oxidation activity (TOF 0.15 s⁻¹), but only 9% selectivity towards PO. Instead, this catalyst showed a high selectivity towards propanal, which is probably the result of subsequent PO isomerization, as evidenced from the stacked bed experiments discussed earlier. This is in accordance with previous results by Haruta et al. who showed that an extensively washed Au/TS-1 catalyst show 3% selectivity towards PO and 74% selectivity towards propanal [60]. A preliminary attempt to impregnate the Au/TS-1U catalyst with NaNO₃ or Ba(NO₃)₂ to inhibit the isomerization reaction did not significantly change the activity or selectivity of the catalyst. These results clearly show that Ti⁴⁺-based supports such as TiO₂-SiO₂ and TS-1 are crucial for the epoxidation of propene, it highlights the promoting effect of Na⁺ for the propene epoxidation, and it also shows that other catalysts do show higher peroxide formation rates, and are limited by subsequent reactions of the propene oxide that is formed, hence showing the way to further improvement of these propene epoxidation catalysts.

3. Materials and Methods

3.1. Materials

SiO₂ (GRACE^® Davicat SI 1404) was kindly provided by GRACE, Columbia, MD, USA. CeO₂ (, nanopowder, 99.5% min, Alfa Aesar, Haverhill, MA, USA), TiO₂ (P25, Evonik, Essen, Germany), WO₃ (nanopowder, Sigma-Aldrich, Amsterdam, The Netherlands), γ-Al₂O₃ ( gamma-phase, Alfa Aesar, Haverhill, MA, USA), α-Al₂O₃ (BASF), CaCO₃ (99.95%, Sigma-Aldrich, Amsterdam, The Netherlands), HAuCl₄·3H₂O (ACS reagent, ≥49.0% Au, ACROS Organics, Geel, Belgium), Na₂CO₃ (≥99.5%, Sigma-Aldrich, Amsterdam, The Netherlands), NaNO₃ (≥99.0%, Sigma-Aldrich, Amsterdam, The Netherlands), Ba(NO₃)₂ (ACS Reagent, >99.0%, Sigma-Aldrich, Amsterdam, The Netherlands), NaBH₄ (ReagentPlus, 99%, Sigma-Aldrich, Amsterdam, The Netherlands), KCl (≥99%, Fisher Chemical, Pittsburgh, PA, USA), AgNO₃ (≥99.0%, Sigma-Aldrich, Amsterdam, The Netherlands), TiCp₂Cl₂ (99+%, STREM, Newburyport, MA, USA), Ti(OBu)₄ (≥97%, Fluka Chemie, Buchs, Switzerland), Tween^® 20 (Polyethylene glycol sorbitan monolaurate, for synthesis, Sigma-Aldrich, Amsterdam, The Netherlands), tetrapropyl ammonium hydroxide (TPAOH, 25% in H₂O, Sigma-Aldrich, Amsterdam, The Netherlands), tetraethyl orthosilicate (98%, ACROS Organics, Geel, Belgium), isopropanol (99+%, ACROS Organics, Geel, Belgium), ethanol (99.5%, extra dry, ACROS Organics, Geel, Belgium), urea (ACS Reagent, Sigma-Aldrich, Amsterdam, The Netherlands), chloroform (anhydrous, 99.9%, ACROS Organics, Geel, Belgium), 1,2-dichlroethane (anhydrous, 99.8%, ACROS Organics, Geel, Belgium), triethylamine (anhydrous, ≥99.5%, Sigma-Aldrich, Amsterdam, The Netherlands) and ethylenediamine (99.5%, Sigma-Aldrich, Amsterdam, The Netherlands) were purchased from the indicated providers and used without further purification.

3.2. General Considerations

All manipulations involving gold were performed in the absence of light in glassware that was rinsed with aqua regia prior to use. After preparation the supported gold catalysts were stored in the absence of light in an exicator under Ar. For inert manipulations, common Schlenk techniques and equipment were used under N₂ using glassware with Teflon joints.

3.2.1. Support Preparation: Titanosilicate-1

Following a literature procedure [85]. 2.00 g Tween^® 20 was dissolved in 26.40 g Milli-Q water and added to 24.60 g of a 25% tetrapropyl ammonium hydroxide solution in water. An amount of 36.06 g of tetraethyl orthosilicate was added under vigorous stirring. After addition stirring was continued for 1 h. Subsequently 1.81 g of tetrabutyl orthotitanate in 10 mL isopropanol was added dropwise. After 1 h of stirring, the mixture was transferred to a 250 mL Teflon lined autoclave (Berghof) and heated to 160 °C for 18 h under autogenous pressure. The TS-1 was retrieved via centrifugation (6000 rpm, 10 min) and dried under ambient conditions. The crude and uncalcined TS-1 is denoted as TS-1U. An amount of 10.90 g of the crude TS-1 was calcined at 550 °C for 6 h under a flow of compressed air at a ramp of 5 °C/min to offer calcined TS-1 as a white powder, denoted as TS-1.

3.2.2. Support Preparation: TiO₂-SiO₂ and TiO₂-γ-Al₂O₃

Following an adapted literature procedure [86], 1.5 g of dried support (SiO₂ or γ-Al₂O₃) was dispersed in 30 mL anhydrous dichloroethane under an N₂ atmosphere followed by the slow addition of an appropriate amount of TiCp₂Cl₂ in 50 mL anhydrous dichloroethane. The red mixture was heated to 60 °C for 1 h under stirring. Subsequently 0.50 mL anhydrous NEt₃ was added to induce the grafting. Heating at 60 °C was continued for 3.5 h. After cooling to room temperature, the solids were collected via filtration and washed three times with 10 mL anhydrous chloroform. The powder was dried under vacuum and calcined under static air at 500 °C for 4 h with a ramp of 10 °C/min to offer the TiO₂-functionalized supports as a white powder.

3.2.3. Bis(ethylenediamine) Gold (III) Chloride

Following an adapted literature procedure [87], 0.15 mL of ethylene diamine was added dropwise under vigorous stirring, to a yellow solution of tetrachloroauric acid hydrate (0.31 g in 1 mL Milli-Q water). Twenty milliliters of ethanol was added immediately to precipitate Au(en)₂Cl₃. The crude product was washed with plenty of ethanol and diethyl ether and dried in the dark at ambient conditions to offer 0.19 g of Au(en)₂Cl₃ as a beige powder.

3.2.4. Catalyst Preparation: Deposition Precipitation Urea

The Au/CeO₂ and Au/TiO₂ catalysts were prepared following the literature procedure by Louis et al. [15,65].

3.2.5. Catalyst Preparation: Cation Adsorption of Au(en)₂Cl₃

One gram of support (SiO₂, WO₃, γ-Al₂O₃ or TS-1U) was dispersed in 50 mL Milli-Q water and an appropriate amount of Au(en)₂Cl₃ in 10 mL Milli-Q water was added under vigorous stirring. (In the case of TS-1, 5.0 g support was dispersed in 250 mL Milli-Q water). Subsequently the pH was raised to 9.0–9.5 by dropwise addition of a 5 wt% Na₂CO₃ solution in Milli-Q water. The mixture was allowed to stir for three hours. The solids were collected via centrifugation (7000 rpm, 10 min) and washed three times with plenty of Milli-Q water, with centrifugation in between (7000 rpm, 10 min). The solids were allowed to dry at room temperature in the dark and subsequently reduced at 300 °C for 2 h (30% H₂ in N₂, 5 °C/min, 80 mL/min) and calcined at 450 °C for 3 h (20% O₂ in N₂, 5 °C/min, 80 mL/min) to offer the Au catalysts as pink to purple colored powders. Adsorption of Au(en)₂Cl₃ was confirmed by addition of NaBH₄ to the supernatant liquid of the first washings, no Au precipitate was formed. Removal of Cl^- was confirmed by adding AgNO₃ to the supernatant liquid of the last washing step: no AgCl precipitate was formed.

3.2.6. Post Synthesis Treatment of Au/TS-1U

An amount of 0.25 g of previously synthesized Au/TS-1U was dispersed in 10 mL Milli-Q water. One milliliter of a 3.5 mM NaNO₃ or Ba(NO₃)₂ solution was added dropwise and the resulting dispersion was stirred in the dark for 2 h. Subsequently the water was removed in vacuo. The resulting catalysts were dried in the dark under vacuum at room temperature.

3.3. Catalyst Characterization

Elemental analysis was performed using inductively coupled plasmon mass spectrometer at the Mikroanalytisches Labor Kolbe, Germany. Bright Field Transmission Electron Microscopy (TEM) images were collected on Talos L120. High-Angle Annular Dark Field Scanning Transmission Electron Microscopy (HAADF-STEM) images were collected on a Talos F200X. Typically 100–200 individual particles were counted for the determination of the average particle size, for Au/WO₃ only 50–100 particles were observed in 10+ images, for Au/γ-Al₂O₃ and Au/TiO₂-γ-Al₂O₃ < 10 particles were observed in 20+ images. Double aberration corrected HAADF-STEM images were collected on a Spectra 300, operating at 300 kV. For the EM images the samples were prepared by drop casting from ethanolic dispersion on lacey carbon Cu grids.

The weight loss of the catalysts was determined using a PerkinElmer TGA 8000 equipped with a mass spectrometer (MS) under 20% O₂ in Ar at 14 mL/min total flow. Diffuse reflectance UV-Vis was recorded under ambient conditions on a Perkin Elmer Lambda 950S coupled with a 150 mm integrating sphere and an InGaAs detector, using BaSO₄ as reference (200–800 nm, 4 nm interval, slit size 4 nm. The point-of-zero-charges of the samples were determined using mass titrations. The supports were suspended in 3.000 mL of a 0.1 M KCl solution in Milli-Q water and equilibrated. The pH of the solution was measured using a Radiometer analytical MeterLab PHM210 standard pH meter. Nitrogen physisorption measurements were performed at 77 K using a Micromeritics TriStar 3000.

Fourier-Transform Infrared Spectroscopy with temperature programmed pyridine adsorption and desorption (FTIR) was performed on a Thermoscientific Nicolet iS5 using a custom-built cell in transmission mode. Sixty-four scans were used for the background, 32 scans for samples at a resolution of 4 cm⁻¹. In brief, self-supporting pellets of approximately 20 mg (12 mm diameter) were evacuated in vacuo at 200 °C for 1 h, ramp 10 °C/min, followed by cooling to 30 °C. At 30 °C 15 mbar of pyridine (Sigma-Aldrich, 99.8%) vapor was introduced and allowed to adsorb on the sample for 30 min, with spectra collected every 5 min. After 30 min the sample was evacuated for 15 min under a dynamic high vacuum. Followed by temperature programmed desorption of the pyridine under dynamic high vacuum (30–500 °C, ramp 5 °C/min) with spectra recorded every 25 °C. Pyridine adsorbed on Lewis acid sites were characterized by vibrations of coordinated pyridine at 1600–1605, 1575, 1490 and 1445 cm⁻¹, respectively, due to the v8a, v8b, v19a and v19b vibrations of the pyridine molecule. Brønsted acid sites were observed for TiO₂ at 1640 and 1545 cm⁻¹ due to the formation of pyridinium. While OH-bound pyridine (weakly Brønsted acid sites) was observed at 1595, 1575, 1490 and 1445 cm⁻¹ due to the v8a, v8b, v19a and v19b vibrations. The acid concentration was determined after evacuation at 150 °C with the integrated molar extinction coefficients as reported by Emeis ε₁₄₅₅ = 2.22; ε₁₅₄₅ = 1.67 [26].

Temperature programmed desorption of ammonia (NH₃-TPD) was performed on a Micromeritics AutoChem II supplied with a thermal conductivity detector (TCD) calibrated for NH₃. In a typical experiment, 100 mg of catalyst was dried in a He flow for 1 h at 600 °C (ramp 10 °C/min). Subsequently, the temperature was decreased to 100 °C, after which NH₃ (10 vol% in He) was adsorbed pulsewise until oversaturation occurred. Physisorbed NH₃ was removed by flowing He for 2 h at 100 °C. Subsequently, NH₃ desorption was measured until 550 °C (ramp 10 °C/min).

3.3.1. Catalytic Testing

Propene oxidation was performed in a quartz fixed bed reactor (i.d. 4 mm) loaded with 150 mg of the catalyst (sieve fraction of 90–212 μm) and diluted with 300 mg SiC (sieve fraction 212–425 μm). Prior to catalysis the catalysts were reduced at 250 °C under 10% H₂ in He for 1 h. The reaction was carried out at 200 °C with a typical gas feed of 25 mL/min in total, with a volumetric concentration of 10% of H₂, 10% O₂ and 10% C₃H₆ in He as balance at a flow of 10,000 mL/g_cat/h. Reaction products were quantified with a standard gas mixture with known concentrations using an online Interscience Compact GC 4.0 (analysis time 8 min) equipped with a Porabond Q column and a Molsieve 5A column in two separate channels, both with a thermal conductivity detector. The materials that did not contain gold did not show any activity at the studied conditions.

The conversion of propylene at reaction time i was calculated from the sum of moles of carbon in the effluent gas, denoted C_oxygenates in all oxidized products (propylene oxide, propanal, acetone, acrolein, acetaldehyde and CO₂/CO), with n as the number of moles of carbon and χ as conversion in % as:

$${\chi }_{{\mathrm{C}}_3{\mathrm{H}}_6}(\%)=\frac{\sum {{\mathrm{nC}}_{\mathrm{oxygenates}}}_{\mathrm{i}}}{{\mathrm{nC}}_{{\mathrm{feed}}_0}}·100\%$$

The selectivities in % towards the produced propylene oxide and other oxygenates at reaction time i were calculated as:

$${\mathrm{S}}_{\mathrm{PO}}(\%)=\frac{{\mathrm{nC}}_{{\mathrm{PO}}_{\mathrm{i}}}}{\sum {{\mathrm{nC}}_{\mathrm{oxygenates}}}_{\mathrm{i}}}·100\%$$

$${\mathrm{S}}_{\mathrm{Product}}(\%)=\frac{{\mathrm{nC}}_{{\mathrm{Product}}_{\mathrm{i}}}}{\sum {{\mathrm{nC}}_{\mathrm{oxygenates}}}_{\mathrm{i}}}·100\%$$

The conversion of hydrogen in % was calculated from the moles of H₂ in the effluent gas divided by the moles of H₂ in the feed as:

$${\chi }_{{\mathrm{H}}_2}(\%)=\frac{{\mathrm{nH}}_{2_{\mathrm{i}}}}{{\mathrm{nH}}_{2_0}}·100\%$$

3.3.2. Stacked Bed Experiments

Stacked bed experiments were performed in a lab scale quartz fixed bed reactor (i.d. 4 mm) loaded with 75 mg of the Au/TS-1 catalyst (sieve fraction of 90–212 μm) that was diluted with 300 mg SiC (sieve fraction 212–425 μm) placed on top of a layer of 150 mg of the support materials (sieve fraction 90–212 μm). Prior to the test the materials were reduced at 250 °C under 10% H₂ in He for 1 h. The reaction was carried out at 200 °C with a typical gas feed of 25 mL/min in total, with a volumetric concentration of 10% of H₂, 10% O₂ and 10% C₃H₆ in He as balance at a flow of 10,000 mL/g_support/h. Reaction products were quantified with a standard gas mixture with known concentrations using an online Interscience Compact GC 4.0 (analysis time 8 min) equipped with a Porabond Q column and a Molsieve 5A column in two separate channels, both with a thermal conductivity detector.

3.3.3. Turnover Frequency Determination

The turnover frequency is defined as the moles of propene that are converted per mole surface Au atoms per second [54].

$$\mathrm{TOF} \left({\mathrm{s}}^{-1}\right)=\frac{{\mathrm{rC}}_3{\mathrm{H}}_6}{\mathrm{nAu}}$$

The number of surface Au atoms is based on TEM and ICP analysis.

$$\mathrm{nAu} \left({\mathrm{Au}}_{\mathrm{surface} \mathrm{atoms}}·{\mathrm{g}}_{\mathrm{cat}}{}^{-1}\right)=\frac{{\mathrm{m}}_{\mathrm{cat}}·\mathrm{Au}-\mathrm{loading}·\mathrm{D}}{{\mathrm{MW}}_{\mathrm{Au}}}$$

The volume–area mean diameter (dVA) of the gold nanoparticles is based on counting typically 100–200 individual particles in TEM images.

$$\mathrm{dVA} (\AA )=\frac{\sum {\mathrm{n}}_{\mathrm{i}}·{\mathrm{d}}_{\mathrm{i}}^3}{\sum {\mathrm{n}}_{\mathrm{i}}·{\mathrm{d}}_{\mathrm{i}}^2}$$

The dispersion D is determined as fraction of Au atoms at the surface of hemispherical nanoparticles. V_Au is the volume of a Au atom in the bulk (16.94 ų) and A_Au is the area of a Au surface atom (8.75 Ų)

$$\mathrm{D}=\frac{6·\left(\raisebox{1ex}{${\mathrm{V}}_{\mathrm{Au}}$}\!\left/ \!\raisebox{-1ex}{${\mathrm{A}}_{\mathrm{Au}}$}\right.\right)}{\mathrm{dVA}}$$

4. Conclusions

To conclude, this study for the first time gives an overview of the role of metal oxide supports in the Au catalyzed propene epoxidation reaction in the presence of H2. A first prerequisite is the formation of peroxides, which is much more pronounced on several of the investigated catalysts than on the established ones. However, a next requirement is the cooperativity between the gold nanoparticles and the metal oxide support to allow epoxidation of propene. The investigated metal oxide supported gold catalysts show moderate to high propene oxidation activity, but their selectivity towards PO is low. We observed that supports catalyze subsequent reactions of PO. In particular, metal oxides with a strong acidic character catalyze the isomerization of PO to propanal and acetone, while readily reducible supports show combustion of PO. A requirement for the metal oxide supports to achieve high epoxide selectivity is hence the presence of inert surface sites, a low specific surface area and/or low density of surface -OH groups. Interestingly, Ti4+-based supports react with PO, but their use is necessary to achieve high epoxidation activity, selectivity and stability. With this research we show the impact of the support on the selectivity of the final catalysts and provide practical guidelines for the preparation of active and selective PO producing catalysts.