Room-Temperature Metathesis of Ethylene with 2-Butene to Propene Over MoOx-Based Catalysts: Mixed Oxides as Perspective Support Materials

We investigated the effect of supports based on ZrO2, TiO2, Al2O3, and SiO2 on the rate of propene formation in the metathesis of ethylene with 2-butene at 50 °C over Mo-containing catalysts possessing highly dispersed MoOx. Large improvements in this rate were achieved when using supports composed of mixed oxides (ZrO2–SiO2, ZrO2–PO4, TiO2–SiO2; Al2O3–SiO2) rather than of individual oxides (ZrO2, TiO2, Al2O3, SiO2). Although previous literature studies dealing with the metathesis reaction over Al2O3- or SiO2-suppported catalysts at higher temperatures suggest the importance of redox or acidic properties of supported MoOx species for catalyst activity, we were not able to establish any general direct correlation in this regard. Contrarily, the rate of propene formation can be significantly enhanced when promoting supports with an oxide promoter. We suggest that the created support lattice defects may facilitate the transformation of MoOx to Mo carbenes under reaction conditions or improve the intrinsic activity of the latter.


Introduction
Metathesis of ethylene with 2-butene is one of the industrial processes for on-purpose production of propene [1]. Typical heterogeneous catalysts include oxides of Re, W, or Mo supported on SiO 2 , Al 2 O 3 , or SiO 2 -Al 2 O 3 [2]. WO x / SiO 2 applied on the large scale is characterized by low price, long lifetime, and resistance to poisoning [3][4][5]. WO x -based catalysts are, however, active at relatively high temperatures (∼350-450 °C), while MoO x -or ReO x -based catalysts demonstrate high activity at lower temperatures (25-200 °C) [2,6].
The activity of MoO x -containing catalysts depends on polymerization degree of MoO x species, surface acidity, pretreatment conditions or presence of co-catalyst [7][8][9]. Isolated di-oxo or oligomeric mono-oxo supported MoO x show high activity for propene production, while crystalline MoO 3 is not active [10,11]. The influence of surface acidity on the activity of Mo-based catalysts was investigated in several independent works [9,[12][13][14]. Apparently, Brønsted acidic sites in close vicinity to MoO x species are relevant for high propene production at temperatures above 100 °C. The activity can also be enhanced after thermal catalyst pretreatment in inert gas [15], propene, ethylene, methane [16], methanol [13] or after photoreductive treatment in CO or H 2 . [17] The influence of co-catalysts such as CaO, Al 2 O 3 or SiO 2 -Al 2 O 3 on the metathesis of ethylene and 2-butene to propene over MoO x /Al x Si y O z was investigated by Goelden et al. [8]. Although these co-catalysts are inactive in this reaction, the activity of Mo-based catalysts is improved when one of them is used as an upstream-located pre-bed for the main catalyst.
To the best of our knowledge, most of previous studies of the metathesis of ethylene with 2-butene to propene were carried out with catalysts based on SiO 2 , Al 2 O 3 , or SiO 2 -Al 2 O 3 supports. Thus, we decided to check the potential application of catalysts based on TiO 2 -or ZrO 2 -containing supports. From a fundamental viewpoint, it was important to check if the general activity-governing relationships established for the often-tested catalysts are also relevant for the-above-mentioned alternatives and particularly at 50 °C. To this end, we prepared a series of Mo-containing catalysts possessing highly dispersed MoO x species on the surface of TiO 2 -or ZrO 2 -based supports as well as of SiO 2 , Al 2 O 3 and SiO 2 -Al 2 O 3 . They were tested for their activity in the metathesis of ethylene with 2-butene and characterized by complementary state-of-the-art techniques.

Catalyst Characterization
Specific surface area (S BET ) of the calcined supports was determined by nitrogen physisorption experiments at − 196 °C using a Belsorp mini II setup (Bel Japan). Adsorption isotherms were evaluated according to the BET method using the adsorption data in the relative pressure (p/p 0 ) range of 0.05-0.3.
UV-vis measurements were carried out using an Avantes spectrometer (AvaSpec-2048-USB2-RM) equipped with a high-temperature reflection UV-vis probe, an Ava-Light-DH-S-BAL deuterium-halogen light source and a CCD array detector. Each sample was in situ calcined in flowing air at 500 °C for 1 h and cooled down to 50 °C in the same flow. The UV-vis spectra were recorded at 50 °C in the range of 200-800 nm. BaSO 4 was used as a white standard for calculating the Kubelka-Munk function.
TEM analysis was conducted on a FEI Tecnai G 2 20 S-TWIN instrument operating at 200 kV. Each sample was ground into a fine powder and dispersed in ethanol using an ultrasonic bath. Hereafter, it was transferred onto copper grids coated with a 2 nm carbon layer and dried at 60 °C for 5 min. Additionally, a second copper grid coated with 2 nm carbon layer was placed on top.
Temperature-programmed desorption of ammonia (NH 3 -TPD) was carried out in an in-house developed setup containing eight individually heated continuous-flow fixedbed quartz reactors. Each sample (50 mg) was calcined in flowing air at 500 °C for 1 h, cooled down to 120 °C, and purged with Ar for 15 min. Hereafter, the treated materials were exposed to a flow of 1 vol% NH 3 in Ar (10 mL·min −1 ) at 120 °C for 1 h, flushed with Ar for 5 h to remove weakly bound NH 3 , and cooled down to 80 °C in the same flow. Then, they were heated in Ar flow up to 900 °C with a heating rate of 10 K·min −1 . Desorbed ammonia was registered by an on-line mass spectrometer (Pfeiffer Vacuum OmniStar GSD 320). The signals at m/z of 15 (NH) and 40 (Ar) were recorded, with the latter being a reference standard.
For distinguishing between Brønsted and Lewis acidic sites, we performed IR measurements of adsorbed pyridine on a Tensor 27 spectrometer (Bruker). Prior to the measurement, each catalyst was pressed into a self-supporting wafer with a diameter of 20 mm. Prior to pyridine adsorption, the wafer was calcined in air flow at 400 °C in the IR cell for 10 min and cooled down in N 2 flow to room temperature. Pyridine was adsorbed at 25 °C until saturation. Then the reaction cell was evacuated for removing physisorbed pyridine. After heating the sample in vacuum up to 150 °C, the IR spectrum of adsorbed pyridine was recorded.
Temperature-programmed reduction with H 2 (H 2 -TPR) was carried out in the same set-up used for NH 3 -TPD. 100 mg of each sample were heated in flowing air to 500 °C for 1 h, cooled down to room temperature and purged with Ar for 15 min. Hereafter, the catalysts were heated in a flow of 5 vol% H 2 in Ar (10 mL·min −1 ) up to 900 °C with a heating rate of 10 K·min −1 . MS signals at m/z of 2 (H 2 ) and 40 (Ar) were recorded.
Temperature-programmed surface reaction with trans-2-C 4 H 8 (C 4 H 8 -TPSR) was carried out in the same set-up used for NH 3  In situ Diffuse Reflectance Infrared Fourier Transform Spectroscopy (DRIFTS) measurements were collected using a Thermo Scientific Nicolet iS10 spectrometer equipped with a Harrick Praying Mantis and high-temperature reaction chamber. Prior to collecting the spectra, each sample was heated in N 2 flow (10 mL·min −1 , heating rate of 10 K·min −1 ) from room temperature to 450 °C, then treated in air flow (12 mL min −1 ) at 450 °C for 1 h and cooled down in N 2 flow (10 mL min −1 ) to 150 °C. The spectra were collected with a resolution of 4 cm −1 and an accumulation of 64 scans in the range of 400-4000 cm −1 under flowing N 2 (10 mL·min −1 ).

Catalytic Tests
Catalytic tests were performed at 1.25 bar (abs.) in an inhouse developed set-up equipped with 14 continuous-flow fixed-bed quartz reactors. The catalysts (20 mg for 1.5Mo/ Siral 10, 1.5Mo/Siral 40, and 1.5Mo/Siral 70 or 100 mg for the remaining catalysts, sieve faction of 315-710 µm) were heated in flowing N 2 up to 500 °C and calcined in air flow at 500 °C for 3 h. Hereafter, they were cooled down in N 2 flow to 50 °C and exposed to a flow (22 mL·min −1 per reactor) of C 2 H 4 /trans-2-C 4 H 8 /N 2 = 5/5/1. All gases were purified over molecular sieve 3 Å filters (Roth). Nitrogen was further purified with an additional AlO-750-2 filter (Pure Gas Products).
The feed components and the reaction products (propene, cis-2-butene and traces of C 5 -olefins) were analysed by an on-line gas chromatograph (Agilent 6890) equipped with AL/S capillary column (for hydrocarbons), connected to a flame ionization detector and a PLOT/Q (for CO 2 )/Molsieve 5 (for H 2 , O 2 , N 2 , and CO) capillary column combination connected to a thermal conductivity detector. To calculate the initial rate of propene formation related to one Mo atom, i.e. turnover frequency of propene formation (TOF Mo ) Eq. 1 was used. The rate was determined after 420 s on stream at a degree of ethylene conversion below 15%. No catalyst activation but continuous deactivation was observed during 35 min on stream.
Here, F feed is the volumetric flow rate of the feed gas (mL·min −1 ) under reference conditions (0 °C, 1 atm), x N 2 with superscripts "in" or "out" stand for the molar flow (mol·mL −1 ) of N 2 at the reactor inlet or outlet, x out (C 3 H 6 ) is the molar fraction of propene at the reactor outlet, N A is the Avogadro's number (6.02 × 10 23 mol −1 ), V m is the molar volume (22,414 mL·mol −1 ), and m cat is catalyst weight (g), 1.5 × 10 18 stands for the apparent surface density of Mo (1.5 × 10 18 m −2 ), S BET is the specific surface area (m 2 g −1 ).

Supported MoO x Species and Their Formation
The catalysts were treated at 500°C in air before recording the UV-vis spectra at 50°C in N 2 atmosphere. The UV-vis spectra of the catalysts (Fig. 1) are characterized by strong absorption bands at 250 nm and 295 nm, related to highly dispersed and small linear-chained tetrahedral MoO x species. A band at ~ 330 nm can be assigned to octahedral polymerized MoO x species [9]. As no absorption band at 400 nm ascribed to small MoO 3 crystallites [19] could be identified, all the catalysts should possess highly dispersed MoO x as a result of low surface density of Mo (1.5 nm −2 ). Strong absorption background in the range 200-400 nm for the TiO 2 -containing samples is due to the intrinsic bandgap of TiO 2 (~ 3.16 eV) [20].
To prove additionally high dispersion of MoO x species, selected catalysts (1.5Mo/ZrO 2 , 1.5Mo/ZrO 2 -PO 4 , 1.5Mo/Siral 40) were characterized by transmission electron microscopy (TEM). Some representative TEM images of the catalysts are shown in Figure S1. Since no evidence of the existence of MoO x nanoparticles can be observed, MoO x species seem to be highly dispersed in agreement with the results of our UV-vis analysis (Fig. 1).
Deposition of MoO x species on support material resulted in consumption (partial or complete) of some hydroxyls, which are, therefore, can be considered as anchoring sites. For example, terminal Zr-OH present on the surface of ZrO 2 (band at 3763 cm −1 ) and ZrO 2 -SiO 2 (band at 3764 cm −1 ) should actively participate in the formation of MoO x species. Similarly, terminal Ti-OH groups in TiO 2 are also involved in this process. For other catalysts, no preferential consumption of certain OH groups for binding MoO x could be established.

Acidic and Redox Properties
In order to distinguish between Brønsted and Lewis acidic sites on the surface of catalysts, we performed IR measurements of adsorbed pyridine on all catalysts with the exception of those based on SiO 2 -Al 2 O 3 supports (acidic properties of these catalysts were fully investigated in Ref. [9]). The obtained IR spectra are shown in Figure S3 in ESI. Noticeably, the spectra of all catalysts are characterized by the presence of bands typical for pyridine adsorbed on Lewis acidic sites (1600 and 1450 cm −1 ). The only catalyst also possessing Brønsted acidic sites (band at 1540 cm −1 in IR spectrum) is 1.5Mo/SiO 2 . Comparing the integral intensity of bands characteristic for different kind of acidic sites, it can be assumed that the number of Brønsted acidic sites in 1.5Mo/SiO 2 is much lower than that of Lewis acidic sites. As the catalysts mainly contain Lewis acidic sites, it is possible to use NH 3 -TPD for deeper analysis of their acidic properties. Figure 2a shows the NH 3 -TPD profiles of the catalysts and the corresponding support materials, while the temperature of the maximal rate of NH 3 desorption and the number of acidic sites related to 1 nm 2 of support are listed in Table S2. Among the support materials, TiO 2 is characterized by the highest density of acidic sites (N sup (a.s.)), while SiO 2 does not possess acidic sites. Noticeably, the bare supports composed of a single oxide (except SiO 2 ) have stronger acidic sites (high T max -NH 3 _sup) than their promoted/mixed counterparts. The maximum of NH 3 desorption peak shifts to higher temperatures for most of the  (Fig. 2b). They are due to H 2 consumption through reaction with differently reducible MoO x species and/or bare support (at temperatures higher than 600 °C). The values of the temperature (T max -H 2 ) of the first hydrogen consumption peak are shown in Table S2. With respect to T max -H 2 , MoO x supported on individual oxides can be ordered as follows: 1.5Mo/TiO 2 < 1.5Mo/ZrO 2 < 1.5Mo/ Al 2 O 3 < 1.5Mo/SiO 2 . In most cases, when the support is a promoted/mixed oxide, T max -H 2 shifts to higher temperatures, i.e. reducibility of MoO x is worsened.
To study reducibility of MoO x species on different supports by trans-2-butene, temperature-programmed surface reaction (TPSR) tests with this olefin were also carried out. Oxygenates CH 3 COR, C 3 H 6 , CO 2 and H 2 O were observed as reaction products. Among all catalysts, only 1.5Mo/ ZrO 2 -PO 4 , 1.5Mo/TiO 2 -SiO 2 , 1.5Mo/Siral 10, 1.5Mo/Siral 40, and 1.5Mo/Siral 70 consumed C 4 H 8 at relatively low temperature ( Figure S4a). Such consumption was accompanied by formation of oxygenates and propene ( Figure  S4 (b) and (c) respectively). It is generally assumed that oxygenates are formed upon transformation of MoO x to carbenes through reaction with fed olefins [7,10]. Accordingly, the formation of oxygenates in the present C 4 H 8 -TPSR tests can be considered as a proof of reductive activation of MoO x species. Propene formation can be explained through 1-and 2-butene metathesis. Thus, it can be concluded that activation of MoO x species supported on binary oxides (ZrO 2 -PO 4 , TiO 2 -SiO 2 , Al 2 O 3 -SiO 2 ) proceeds easier than that of those supported on individual oxides (ZrO 2 , TiO 2 , Al 2 O 3 , SiO 2 ) since no oxygenates and propene were formed at low temperature in the C 4 H 8 -TPSR tests with the latter materials. High-temperature C 4 H 8 consumption (T > 400 °C) was observed for all catalysts and is related to oxidation of this olefin by MoO x to CO 2 ( Figure S4 (d)) and H 2 O ( Figure  S4 (e)).

Catalyst Activity in Metathesis of Ethylene with 2-Butene
The rate of propene formation in the metathesis of ethylene with 2-butene was determined at 50 °C and related to one Mo atom to obtain an apparent turnover frequency (TOF Mo ). The calculated values are shown in Fig. 3 (Table S3) [13,18,26,27].
According to the results shown in Fig. 3, there is an obvious impact of the kind of support on catalyst activity. Considering the UV-vis spectra (Fig. 1), TEM ( Figure S1), H 2 -TPR (Fig. 2b) results, and the Mo content well below a monolayer coverage, we are quite sure that our catalysts do not possess inactive MoO 3 , which would negatively affect the TOF values. However, we cannot exclude that Mo-carbenes formed from differently structured highly dispersed and small linear-chained tetrahedral MoO x species may differ in their intrinsic metathesis activity.

Factors Affecting Metathesis Activity of MoO x Species
Due to technical limitations for identification/quantification of Mo carbenes, we decided to follow a common approach used for metathesis catalysts tested at high temperatures to reveal possible origin(s) of different activity [9]. To this end we tried to correlate redox and acidic catalyst properties with catalyst activity. Redox properties of MoO x might play a crucial role due to the following reason. It is generally accepted that Mo = CHR carbenes are active sites for olefin metathesis [7,13,28,29]. Such sites are assumed to be created through in situ activation of Mo 6+ O x by olefin molecule into Mo 4+ O x−1 followed by oxidative addition of another olefin molecule (Fig. 4).  Table S2 as well as with the temperature of C 3 H 6 release peak observed in C 4 H 8 -TPSR profiles in Figure S4 (c). As no obvious relationships could be established ( Fig. 5a and Figure S5), this catalyst property does not seem to be decisive.
The role of catalyst acidity in olefin metathesis is documented in several works [9,12,30]. Density functional theory calculations of the metathesis of ethylene with 2-butene over Mo/HBeta suggest that the presence of acidic sites could increase propene productivity due to decreasing the activation barrier for MoO x conversion into Mo-carbenes [12]. Based on the results of IR analysis of adsorbed pyridine, Hahn et al. [9] established a correlation between the concentration of Brønsted acidic sites in MoO x /Sirals and TOF Mo determined at 150 °C, i.e. 100 °C higher than in the present study. Moreover, those authors considered the total amount of Brønsted acidic sites and did not pay attention to their strength. It can be however expected that the stronger the acidic sites, the higher the contribution of side reactions is as suggested in Ref. [12] To check if the strength of acidic sites and/or their concentration in the catalysts or support materials are important for propene formation at 50 °C over the present catalysts, we plotted the activity and the acidity data in Fig. 5b-d. Noticeably, most of the catalysts desorbing NH 3 at lower temperatures are more active than those desorbing NH 3 at higher temperatures (Fig. 5b). Accordingly, it can be assumed that weak acidic sites are preferable for metathesis reaction. Nevertheless, no general direct correlation between the activity and the strength of acidic sites could be established. Regarding the influence of concentration of acidic sites in the catalysts (Fig. 5c) or in the support materials (Fig. 5d) on the TOF Mo values, no obvious dependence could also be found.
Accordingly, the direct influence of redox and acidic properties of the catalysts on the intrinsic activity of supported MoO x species in propene formation at 50 °C was not confirmed. It can be assumed that besides redox and acidic properties, the kind of support also influences the molecular structure of highly dispersed and small linear-chained tetrahedral MoO x species and therefore their ability to form carbenes and/or the intrinsic activity of the latter. Therefore, considering the molecular level complexity of the mechanism of carbene formation and the metathesis reaction itself, it is unlikely to establish general factors determining activity of MoO x species supported on different materials. Thus, further studies are required to check if there are universal fundamentals relevant for this reaction. In particularly, the role of support defects created upon promoting of supports with an oxide promoter seems to deserve a special attention.

Conclusions
For the first time, we have demonstrated that Mo-containing catalysts based on doped ZrO 2 or TiO 2 supports are attractive candidates for the metathesis of ethylene with 2-butenes to propene at only 50 °C and outperform their often-investigated counterparts based on individual SiO 2 or Al 2 O 3 supports. Their higher activity might be related to the presence of lattice defects created in support materials after promotion with an oxidic promoter. This knowledge opens a possibility for purposeful preparation of supports and, accordingly, catalysts on their basis for low-temperature propene formation from ethylene and 2-butenes.