MoO3 on zeolites MCM-22, MCM-56 and 2D-MFI as catalysts for 1-octene metathesis

Highly active olefin metathesis catalysts were prepared by thermal spreading MoO3 and/or MoO2(acac)2 on MWW zeolites (MCM-22, delaminated MCM-56) and on two-dimensional MFI (all in NH4+ form). The catalysts‘ activities were tested in the metathesis of neat 1-octene (as an example of a longer chain olefin) at 40 °C. Catalysts with 6 wt % or 5 wt % of Mo were used. The acidic character of the supports had an important effect on both the catalyst activity and selectivity. The catalyst activity increases in the order 6MoO3/HZSM-5(25) (Si/Al = 25) << 6MoO2(acac)2/MCM-22(70) < 6MoO3/2D-MFI(26) < 6MoO3/MCM-56(13) < 6MoO3/MCM-22(28) reflecting both the enhancing effect of the supports‘ acidity and accessibility of the catalytic species on the surface. On the other hand the supports‘ acidity decreases the selectivity to the main metathesis product C14 due to an acid-catalyzed double bond isomerization (followed by cross metathesis) and oligomerization. 6MoO3/2D-MFI(26) with a lower concentration of the acidic centres resulting in catalysts of moderate activity but with the highest selectivity.


Introduction
Molybdenum oxide on silica, alumina or silica-alumina belongs to the well-known and long-time used metathesis catalysts [1]. Albeit typical ill-defined catalysts they are still popular as relatively cheap catalysts finding industrial applications especially in the treatment of low olefins [2][3][4][5]. Their catalytic activity depends on many factors, especially on Mo loading, support acidity, and pre-reaction activations. Surface isolated MoO 4 tetrahedra were proved as the main precursors of the catalytic species [6,7], thus the perfect dispersion of MoO 3 on the surface is a crucial precondition for a high catalytic activity. The mechanisms of transformation of these precursors to the surface Mo carbenes as real catalytically active species has been suggested [6,7]. The replacement of ordinary silicas for mesoporous molecular sieves SBA-15 or MCM-41 increased the Scheme 1: 1-Octene metathesis reaction. catalyst activity substantially, which allowed performing the metathesis of long chain olefins under mild reaction conditions [8][9][10]. The positive effect of these supports on the catalyst activity was ascribed to their high surface areas enhancing the spreading of MoO 3 molecules on the surface and large pores increasing the substrates/products transport rate.
Microporous zeolites like HZSM-5 impregnated by ammonium heptamolybdate solutions were used for the metathesis of low olefins (ethylene, propylene, butenes) [11][12][13]. In the case of bulkier substrates they suffer, however, of micropore size limitations. To overcome these limitations a decrease in crystal size and the application of two-dimensional zeolites can be used [14][15][16][17]. Two dimensional 2D-MFI and MWW delaminated zeolite MCM-56, which have been prepared recently [18][19][20][21], represent two types of these materials, which exhibit relatively high surface areas and high accessibility of catalytic sites on the surface as well [22]. Therefore, we supported MoO 3 and/or MoO 2 (acac) 2 on (i) 2D-MFI (and ordinary HZSM-5 for comparison) and similarly on (ii) MCM-56 and its 3D analogue MCM-22 (both in NH 4 + form) and examined their activity in the metathesis of neat 1-octene (Scheme 1) under ambient pressure and 40 °C. According to our best knowledge, none of these materials have been tested as supports for MoO 3 based catalysts for metathesis of higher alkenes up to now. MoO x on MCM-22 combined with γ-Al 2 O 3 was used in cross metathesis of 2-butene and ethylene in a stream (125 °C, 1 MPa) [23]. MCM-22, and MCM-56 were also used as supports for Hoveyda-Grubbs type hybrid catalysts active in metathesis of long-chain unsaturated esters [24].

Results and Discussion
Catalyst preparation and characterization XRD patterns and texture properties ( Table 1, Figure 1 A,B,C,D) of prepared MCM-22, MCM-56 and 2D-MFI zeolites proved a high quality of these supports. For catalyst labelling following the mode has been adopted: x MoO 3 /MCM-22(y), where x = Mo concentration in wt % Mo, y = Si/Al molar ratio. After spreading Mo compounds over the support surface areas (S BET , S ext ) as well as void volumes (V) decreased. Similar reduction of these quantities has been already observed earlier [9,10,24]. For x MoO 3 /MCM-22 (28), XRD patterns of catalysts are similar to those of their parents approximately up to x = 6 wt % of Mo (0.9 Mo atoms per nm 2 ). At higher Mo concentrations signals of crystalline MoO 3 appeared (marked with * in Figure 1 A [9]. XRD patterns of 6MoO 3 /MCM-56(13) and 6MoO 3 /2D-MFI (26) indicated also a good MoO 3 spreading, contrary to 6MoO 3 /HZSM-5 (25) where MoO 3 signals were clearly visible, probably as a result of lower external surface area.   MCM-22 (28), however, its Lewis acid site concentration was significantly lower. After supporting Mo compounds the concentrations of Brønsted acid sites decreased significantly which may indicate that MoO x species reacted predominantly with Brønsted acid sites of the supports. It is manifested by intensity decrease of the band in the region 3609-3625 cm −1 , ascribed to OH vibration in the Si-O(H)-Al acid site (see Supporting Information File 1, Figures S1-S5) [25]. On the other hand, the concentrations of Lewis acid sites in the catalysts was slightly higher compared to the parent supports. It may be explained by the formation of some amount of Mo in a lower oxidation state which has been already described for siliceous supports (MCM-41, SBA-15) [9,26].

Catalytic activity MCM-22-based catalysts
Na + forms of zeolites turned out to be unsuitable supports for metathesis catalysts. For example, by supporting MoO 3 on MCM-22 (28) in Na + form (6 wt % of Mo) we obtained material providing only 0.5% 1-octene conversion in 19 h (1-octene/Mo = 320, t = 40 °C). Therefore, we converted Na + forms to NH 4 + forms, which were used for supporting Mo compounds by thermal spreading method.
The time development of 1-octene conversion over 6MoO 3 / MCM-22 (28) is shown in Figure 2. The GC chromatogram of the final product is shown in Figure S6 (in Supporting Information File 1). It is seen that in addition to the main metathesis product (7-tetradecene), alkenes from C13 to C9 are present in considerable amounts. It is a consequence of the 1-octene double bond isomerization followed by cross metathesis. Moreover, a certain amount of oligomers (mainly dimers) were also observed in the reaction mixtures. Both isomerization and oligomerization are due to the acidic character of the support (vide infra). In addition to the total conversion of 1-octene (K tot ), the conversion to all metathesis products (K met ), and the conversion to tetradecene (K C14 ) calculated according to the following equations are plotted in Figure 2.
where m i and M i (i = 9-14) are weight amounts and molecular weights of alkenes from C9 to C14; m d , m t and M d , M t are weight amounts and molecular weights of octene dimers and trimers, respectively; m C8 is weight amount of octene (all isomers) and M C8 is the molecular weight of octene. For comparison, the conversion curve over 6MoO 3 /SBA-15 is added in Figure 2. 6MoO 3 /SBA-15 was prepared from allsiliceous SBA-15 (S BET = 877m 2 /g, V = 1.07cm 3 /g, pore diameter D = 6.4 nm) and it is known as a very active and selective catalyst [9,10]. Under reaction conditions applied the selectivity to tetradecene was about 98% during the whole experiment, and therefore only K tot is plotted in Figure 2 in this case.
Both K tot and K met for 6MoO 3 /MCM-22 (28) were significantly higher than the total conversion for 6MoO 3 /SBA-15 ( Figure 2). Conversions to oligomers (K ol = K tot − K met ) were about 12% (at 2 h) and practically did not change in the further course of the reaction. However, the conversions to tetradecene were rather low (maximum conversion about 20% was achieved).
Higher catalytic activity of molybdenum oxide on zeolitic support in metathesis may be ascribed to the higher acidity of supports. The enhancing effect of Brønsted acidity on the catalytic activity has been already described [6] and it assumed that most of Mo active species in zeolite-based catalysts are formed by reacting molybdenum oxide with Si-O(H)-Al groups [12,27]. Similarly, Lim et al. showed recently [28], that Brønsted acid sites improve dispersion of molybdenum oxide on the surface. Moreover, for related system based on tungsten oxide in zeolite, it was suggested using high resolution STEM that Brønsted acid sites in proximity to metathesis active sites facilitate olefin adsorption and metallocycle formation [29]. Such mechanism may be effective also for Mo catalysts. The decrease in the selectivity due to isomerization and/or oligomerization seems to be an unavoidable cost for this activity enhancement.  It is known for molybdenum oxide catalysts, that with increasing Mo loading the catalytic activities increase up to maximum value [6,10]. At higher loadings the molybdenum oxide spreading on the surface became imperfect and catalytically inactive bulk MoO 3 appears. The effect of increasing Mo loading on catalyst activity for MCM-22(28)-based catalyst is shown in Table 3.
For 8MoO 3 /MCM-22(28) XRD pattern shows a small amount of bulk MoO 3 (marked with asterisk in Figure 1A). In accord with this, the conversions fell down in comparison with 6MoO 3 /MCM-22 (28), the selectivity, however, slightly increased: the amount of oligomers was reduced and the selectivity to the tetradecene approximately doubled. It suggests that more acid sites were covered by MoO x species and oligomerization and isomerization ability of catalysts decreased. However, further increase in the Mo loading to 10 wt % in 10MoO 3 / MCM-22 (28) Table 4.  Figure 1B) and showing a mild metathesis activity. The highest conversion K tot = 35% (after 20 h) was achieved over 5MoO 2 (acac) 2 /MCM-22(70). Oligomerization activity of all these catalysts was considerably lower in comparison with that of 6MoO 3 /MCM-22(28) (K ol = 1% only). However, the isomerization was not suppressed and conversion to tetradecene K C14 was low.

MCM-56-based catalysts
Conversion curves for the 1-octene metathesis over 6MoO 3 / MCM-56(13) under standard conditions are displayed in Figure 3. In spite of the 2D character of support the conversions over 6MoO 3 /MCM-56(13) were significantly lower in comparison with 6MoO 3 /MCM-22 (28): the initial reaction rate (calculated at reaction time = 2 h) being about a half of the initial reaction rate over 6MoO 3 /MCM-22 (28). On the other hand the extent of oligomerization was practically the same (for final product the oligomerization selectivity was 14%) and the extent of cross metathesis was even higher (the selectivity to tetradecene was only 15%). The crystals of MCM-22 (see SEM image in Supporting Information File 1, Figure S7) consist of very thin platelets and therefore a great amounts of 12-membered ring cups of MWW structure are on crystal exterior [18]. These cups as we assume host MoO x species. Although MCM-56(13) as 2D zeolite consists of very thin layers, these layers may be curled and packed, which prevents the access of substrate molecules to the most of 12MR cups (for MCM-56(13) morphology see Supporting Information File 1, Figure S8). This may explain the lower activity of 6MoO 3 /MCM-56(13) compared with 6MoO 3 /MCM-22 (28). Similarly, a higher activity of MCM-22 in comparison with MCM-56 has been observed in toluene disproportionation [18] and also for RCM of citronellene over immobilized Ru catalysts the activity of catalyst based on MCM-56 was not higher than that based on MCM-22 [24].

The accompanying oligomerization activity
The experiments with Mo-free zeolites (Figure 5a,b) confirmed that the oligomerization activity was connected with the support itself. In these "blank" experiments the reaction conditions, as well as pretreatment mode were the same as for Mo oxide cata- lysts. No metathesis products were observed, only 1-octene oligomerization and double bond isomerization occurred. Figure 5a shows 1-octene oligomerization over MCM-22 (28) and MCM-22(70). Families of dimers and trimers (in weight ratio dimers/trimers approximately 20:1 for the final conversions) were detected, isolation and characterization of individual dimers/trimer was not possible. It was visible from GC, that isomerization of starting 1-octene also occurred, however, the exact quantification was not possible. The oligomerization rate was higher for MCM-22 (28) in accord with its higher acidity as compared with MCM-22(70). The extent of oligomerization in these blank experiments is several times higher than that achieved over metathesis catalysts: it may be due to the partial capping of support acid sites with Mo species catalysts and also due the parallel consumption of 1-octene in metathesis. Figure 5b shows oligomerization of 1-octene, 2-octene (cis + trans), 3-octene (trans), and 4-octene (trans) over MCM-22 (28). It is seen that the initial reaction rate decreases in the order 1-octene ≈ 2-octene > 3-octene > 4-octene. The low-temperature oligomerization of alkenes over zeolite was studied as concerns industrially important low alkenes oligomerization and lower reactivity of internal alkenes in comparison with 1-alkenes was also recognized [30,31]. The reduced activity of 3-and 4-octenes in oligomerization might explain the fact, that in our metathesis experiments the accompanying oligomerization occurred practically only in the beginning of the reaction. In later stages when most of 1-octene was isomerized to 3-and 4-octenes only little increase in oligomer amounts was observed. Due to the catalyst acidity accompanying reactions occurred: (i) 1-octene double bond isomerization followed by cross metathesis and (ii) 1-octene oligomerization (mainly dimerization). The extent of these reactions depends strongly on the support acidity. Highly acidic supports MCM-22 (28) and MCM-56(13) delivered a catalyst of rather low selectivity (up to 14% conversion to oligomers, 15-20% conversion to tetradecene at about K tot = 90%). Less acidic supports -MCM-22(70) and 2D-MFI(26) gave rise catalysts of significantly higher selec-tivity: conversion to oligomers was reduced to 1%, double bond isomerization and cross metathesis proceeded in less extent, so selectivity to tetradecene increased (e.g., for 2D-MFI (26) to 35% at K tot = 90%).

Conclusion
It is seen that for the metathesis of longer chain hydrocarbons like 1-octene, supports ensuring a good access of bulkier substrate to the active centers are necessary. The acidity of the support increases the catalyst activity, however, simultaneously with decrease of the catalyst selectivity. 2D-MFI(26) due to its moderate acidity and 2D character results in catalysts of moderate activity but of the highest selectivity.
With the described catalysts 1-octene was converted into a mixture of higher olefins: in addition to tetradecene as a homometathesis product, olefins of 9-13 C atoms from cross metathesis and C16 dimers were formed in various extent. Therefore, the described catalysts may find application especially if a mixture of higher olefins is desired, for example in the preparation of detergents, lubricants etc.

Experimental Catalyst preparation and characterization
The zeolite supports MCM-22 and MCM-56 were prepared according to [32,33], 2D-MFI was synthesized according to [21]. HZSM-5 (CBV 5524) was purchased from Zeolyst. Na + forms of zeolites were converted to NH 4 + form by three-fold treatment with 1.0 M NH 4 NO 3 solution at room temperature for 3 h. The supports were characterized by XRD (Bruker AXS D8 Advance diffractometer with a graphite monochromator and a Vantec-1 position sensitive detector using Cu Kα radiation in Bragg−Brentano geometry) and by N 2 adsorption (77 K, Micromeritics GEMINI II 2370 volumetric Surface Area Analyzer). Molybdenum(VI) oxide (Sigma-Aldrich) and bis(acetylacetonato)dioxomolybdenum(VI) (Aldrich) as sources of Mo oxide species were used for catalyst preparation using the thermal spreading method (500 °C, 8 h). SEM images were recorded using a JEOL JSM-5500LV microscope.
The concentrations of Lewis (cL) and Brønsted (cB) acid sites were determined by FTIR spectroscopy of adsorbed pyridine (Py) using a Nicolet 6700 with a transmission MCT/B detector. The zeolites were pressed into self-supporting wafers with a density of 8.0-12 mg·cm -2 and activated in situ at T = 450 °C and p = 5·10 -5 torr for 4 h. Pyridine adsorption was carried out at 150 °C and a partial pressure of 3.5 torr for 20 min followed by desorption for 20 min at 150, 250, 350 or 450 °C. Before adsorption, pyridine was degassed by freeze-pump-thaw cycles. All spectra were recorded with a resolution of 4 cm -1 by collecting 128 scans for a single spectrum at room temperature.
For elemental analysis ICP OES (iCAP 7000, Thermo Scientific) was used. About 50 mg of the catalyst was digested in a mixture of HF, HCl, and HNO 3 (1:2:2). The samples were placed in a Berghof microwave in a closed vessel at T = 140 °C for 35 min. Saturated solution of H 3 BO 3 was then added for complexation of the excess of HF. After digestion solutions under analysis were collected in 250 mL flasks and diluted with ultra pure water.

Catalytic experiments
Catalytic experiments were carried out in an argon atmosphere using a vacuum argon line. 1-Octene (Aldrich, 98%) was passed through alumina and stored with Na. The content of water in 1-octene was about 5 ppm. 2-Octene (Alfa-Aesar, 98%), trans-3-octene (Alfa-Aesar, 97%) and trans-4-octene (Aldrich) were purified in a similar way. In a typical experiment 50 mg of catalyst (6 wt % of Mo) was used. Before reaction catalyst was pretreated in vacuo at 500 °C for 30 min. After cooling to 40 °C, the reactor was filled with Ar and neat 1-octene (1-octene/Mo ratio = 320) was added under stirring. The reaction progress was followed by GC analysis of reaction mixture samples taken at given intervals. Individual compounds were identified by GC/MS. A high-resolution gas chromatograph Agilent 6890 with a DB-5 column (length: 50 m, inner diameter: 320 μm, stationary phase thickness: 1 μm), equipped with a 7683 Automatic Liquid Sampler and a FID detector and GC/MS (ThermoFinnigan, FOCUS DSQ II single Quadrupole) were used. Conversions were calculated from the mass balance.

Supporting Information
Supporting Information File 1 IR spectra of catalysts, GC of reaction products, and SEM images of catalysts.