Thermal Spreading As an Alternative for the Wet ... - ACS Publications

Oct 14, 2010 - Thermal Spreading As an Alternative for the Wet Impregnation Method: ... 18059 Rostock, Germany, and Laboratoire de Chimie des Matéria...
4 downloads 0 Views 2MB Size
18664

J. Phys. Chem. C 2010, 114, 18664–18673

Thermal Spreading As an Alternative for the Wet Impregnation Method: Advantages and Downsides in the Preparation of MoO3/SiO2-Al2O3 Metathesis Catalysts Damien P. Debecker,*,† Mariana Stoyanova,‡ Uwe Rodemerck,‡ Pierre Eloy,† Alexandre Le´onard,§ Bao-Lian Su,§ and Eric M. Gaigneaux*,† UniVersite´ catholique de LouVain. Institute of Condensed Matter and Nanoscience - IMCN, DiVision of Molecules Solids and ReactiVity - MOST, Croix du Sud 2/17, B-1348 LouVain-la-NeuVe, Belgium, Leibniz-Institut fu¨r Katalyse e.V. an der UniVersita¨t Rostock, Albert-Einstein-Str. 29a, 18059 Rostock, Germany, and Laboratoire de Chimie des Mate´riaux Inorganiques (CMI), Faculte´s UniVersitaires Notre-Dame de la Paix (FUNDP), 61 rue de Bruxelles, B-5000 Namur, Belgium ReceiVed: August 9, 2010; ReVised Manuscript ReceiVed: September 23, 2010

Silica-alumina-supported MoO3 catalysts are classically prepared via impregnation of the support with a molybdenum salt solution, usually ammonium heptamolybdate, and subsequent drying and calcination (three steps). The downsides of such a route for the synthesis of heterogeneous metathesis catalysts are linked to the limited control on the nature of the MoOx stabilized at the surface, to the uneven distribution of the deposit in the pores of the support, and to the build up of inactive species that find their origin in the wet step of the preparation. In opposition, the direct thermal spreading of molybdenum oxide onto the support is a straightforward (one step) method involving no wet stage. It allows the conversion of bulk MoO3 crystals to amorphous molybdate species dispersed at the surface of the silica-alumina support. This contribution shows that the catalysts obtained via both methods exhibit similar performances in the self-metathesis of propene to butene and ethene. However, based on XRD, XPS, Raman spectroscopy, ICP-AES, N2 physisorption, TEM, and MAS-NMR spectroscopy, it is shown that the origin of active and inactive species in the two systems is different. Whereas the activity of wet-made catalysts is limited by the formation of bulky MoO3 crystals and of aluminum molybdate, the performances of dry-made catalysts are limited by the incomplete spreading of MoO3 nanocrystallites. 1. Introduction The heterogeneous metathesis of light alkenes is an important reaction for the petrochemical industry, allowing us to regulate the stocks of light olefins (ethene, propene, butene) as a function of the market and at low energy and environmental cost.1 For that purpose, robust catalysts based on MoO3,2,3 WO3,4,5 or Re2O36-8 are intensively studied. The preparation of MoO3-based metathesis catalysts is classically realized by impregnation of a support with an aqueous solution of a Mo precursor, usually ammonium heptamolybdate (AHM), followed by drying and calcination.9,10 In general, the dispersion of the active phase deposited on a support by wet impregnation is not really controlled. Mastering the nature of Mo surface species that form on the support is not easy. That is, there is no guarantee to obtain a homogeneous deposit at the surface of the support. Three main reasons for this can be cited.9 First, strong interactions between the species in solution and the support can prevent the species from reaching the bottom of pores.11 In consequence, the Mo deposit is not dispersed on the whole available surface. Agglomerated Mo species are prone to sinter and to form MoO3 crystals upon calcination. Second, the drying step (e.g., in a rotavapor) implies that part of the support is already dry, sticking on the edge of the flask, while * Corresponding authors. E-mail: [email protected] (D.P.D.), [email protected] (E.M.G.). Tel: 0032 10473665. Fax: 0032 10473649. IMCN and MOST are new research entities involving the group formerly known as “Unite´ de catalyse et chimie des mate´riaux divise´s”. † Institute of Condensed Matter and Nanoscience. ‡ Leibniz-Institut fu¨r Katalyse e.V. an der Universita¨t Rostock. § Faculte´s Universitaires Notre-Dame de la Paix (FUNDP).

another portion is still in contact with the remaining solution. In these cases, the amount of Mo that is actually deposited on different particles of support can be different, which results in different surface densities and uneven distribution of dispersed and agglomerated species. Third, in the case of the preparation of MoO3/Al2O311 or MoO3/SiO2-Al2O39 catalysts, the wet step results in the formation of a mixed aluminum-molybdenum phase, which itself is prone to instability upon calcination.12 For these reasons, the formation of MoO3 crystallites is sometimes unavoidable, even when the theoretical monolayer coverage is not reached. In our targeted reaction, such MoO3 crystals are undesired.2,13,14 This was identified as a limitation for wet-made MoO3/SiO2-Al2O3 metathesis catalysts.9 It thus appears attractive to search for alternative methods to deposit a Mo oxide phase at the surface of the support. Several reports from the literature showed that molybdenum trioxide can be transported in the gas phase and spread at the surface of another solid if an appropriate thermal treatment is applied.15,16 Recently, such direct thermal spreading (TS) of MoO3 has been proposed for the preparation of metathesis catalysts.13,17,18 The technique allows us to get rid of the wet state step, which inherently is one of the causes for the lack of homogeneity of catalysts prepared by wet impregnation. An additional justification for investigating this technique is the fact that the nuclearity of the transported species is low. Gu¨nther et al. proposed that (MoO3)3 molecules desorb from bulk MoO3 and are transported during TS.19,20 Indeed, MoO3 trimers are the major species found in MoO3 vapor under the range of temperature (typically 500 °C) used for TS. This is an important difference with the situation encountered in the liquid phase,

10.1021/jp1074994  2010 American Chemical Society Published on Web 10/14/2010

MoO3/SiO2-Al2O3 Metathesis Catalysts where dissolved polyoxomolybdenum species with high nuclearity may lead to aggregated Mo deposit and may constitute nucleation points for the sintering to MoO3. Interestingly, the TS technique is very easy and cheap: only one step, no wastewater, and no energy spent for (vacuum) drying. For the conversion of large amounts of light alkenes in a refinery, such cheap preparation of heterogeneous metathesis catalysts would be highly regarded. In this Article, catalysts obtained by the TS of molybdenum oxide on a silica-alumina support are described in detail and compared systematically with similar catalysts obtained via wet impregnation of AHM on the same support. The performances of the catalysts in the self-metathesis of propene at low temperature are also compared. The purpose is to describe the physicochemical properties of both systems and to understand in each case how the metathesis activity is generated and how it is limited. This study results in an inventory of the respective advantages and downsides of both methods. 2. Experimental Section 2.1. Preparation of the Catalysts. 2.1.1. Support. The support is a commercial mesoporous silica-alumina catalyst support purchased from Aldrich (Grade 135). This support contains 13 wt % of alumina and has a specific surface area of 490 m2 g-1. This composition is close to the optimum determined in a previous recent work.3 Prior to use in the preparations, the support was calcined at 500 °C under static air for 15 h. 2.1.2. Wet Impregnation. The catalysts are prepared by wet impregnation of AHM.9 An initial precursor solution was prepared by dissolving 12.268 g of AHM (Aldrich, 99.98% purity) in distilled water (6.66 g of Mo per liter). An appropriate amount of this precursor solution (depending on the nominal MoO3 loading targeted for each synthesis) was diluted in distilled water to yield a 200 mL of impregnation solution. Calcined support (4 g) was then suspended in the impregnation solution for 2 h under magnetic stirring. Water was then evaporated under reduced pressure in a rotavapor at 40 °C. The recovered solid was dried at 110 °C for one night and calcined at 500 °C for 2 h in a muffle furnace under static air. The samples are denoted WIx, where WI stands for wet impregnation and where x is the nominal MoO3 weight loading (from 4 to 16 wt % of MoO3). 2.1.3. Thermal Spreading. The catalysts were obtained by direct TS of MoO3 (Aldrich, 99.5% purity) on the support. The desired proportions of Mo trioxide and support were handground in an agate mortar for 5-10 min. The mixture (4 g) was then placed in a muffle furnace and heated to 500 °C for 8 h under static air. Samples are denoted TSx, where TS stands for TS and x is the MoO3 nominal weight loading (from 5 to 20 wt % of MoO3). 2.2. Catalyst Characterization. The Mo loading was measured by inductively coupled plasma-atomic emission spectroscopy (ICP-AES) on an Iris Advantage apparatus from Jarrell Ash Corporation. The materials were dried at 105 °C prior to measurements. N2 physisorption experiments were performed at -196 °C on a Micromeritics Tristar apparatus. The samples were outgassed at 150 °C under vacuum (2 Pa). The specific surface area was determined from the BET method in 0.05 to 0.30 P/P0 range. The pore size distribution was derived from the desorption branch using the BJH method. The average pore diameter is calculated as (4 × pore volume/BET specific surface area). X-ray diffraction (XRD) measurements were performed with a Siemens D5000 diffractometer using the KR radiation of Cu

J. Phys. Chem. C, Vol. 114, No. 43, 2010 18665 (λ ) 1.5418 Å). The 2θ range was recorded between 5 and 75° at a rate of 0.02° s-1. The ICDD-JCPDS database was used to identify the crystalline phases. Confocal Raman spectroscopy was done on the InVia Raman microscope (Renishaw) equipped with a diode light (785 nm). The resolution was set to 4 cm-1. Acquisition time was 10, and 10 scans were recorded and averaged for each catalyst. The laser power was set to 10 mW, and the 50× objective was used. X-ray photoelectron spectroscopy (XPS) was performed on a Kratos Axis Ultra spectrometer (Kratos Analytical, Manchester, U.K.) equipped with a monochromatized aluminum X-ray source (powered at 10 mA and 15 kV). The pressure in the analysis chamber was ∼10-6 Pa. The analyzed area was 700 µm × 300 µm. The pass energy of the hemispherical analyzer was set at 160 eV for the wide scan and 40 eV for narrow scans. We achieved charge stabilization by using the Kratos Axis device. The electron source was operated at 1.8 A filament current and a bias of -1.1 eV. The charge balance plate was set at -2.8 V. The sample powders were pressed into small stainless steel troughs mounted on a multi specimen holder. The following sequence of spectra was recorded: survey spectrum, C1s, O1s, Si2p, Al2p, Mo3d, and C1s again to check for charge stability as a function of time and for the absence of degradation of the sample during the analyses. The binding energy (BE) values were referred to the C-(C, H) contribution of the C1s peak fixed at 284.8 eV. Molar fractions (%) were calculated using peak areas normalized on the basis of acquisition parameters after a linear background subtraction, with experimental sensitivity factors and transmission factors provided by the manufacturer. The Bruker Avance 500 apparatus was used to record the Magic angle spinning nuclear magnetic resonance (MAS-NMR) of 27Al, using a flip of 10° (with respect to the vertical) and recording 5000 scans for each sample at intervals of 0.1 s. The sample rotation speed was 15 000 rpm. Most spectra were recorded on samples that were previously stored for several months in a desiccator and often manipulated under ambient conditions. No particular care was thus made about the degree of hydration of the sample. In some dedicated experiments, the sample (support or TS10 catalyst) was analyzed 1 day after the last thermal treatment (calcination or TS). In that case, the samples were stored for 1 night in an oven at 110 °C before NMR analysis. TEM analyses were performed using a LEO922 electron microscope operating at 200 keV. The powdered samples, dispersed in 2-butanol, were deposited on copper grids coated with a carbon film, and the solvent was then evaporated. 2.3. Metathesis Reaction. The evaluation of the metathesis activity of the catalysts was carried out in a multichannel apparatus with a capacity of treating of up to 15 samples under identical conditions.21 The whole design allows fully automated control of gas flows and of three temperature zones (gas preheating, reactor, and post reactor lines with 16-port valve) along with reactor switching and product sampling. All catalysts were sieved and selected in the 200-315 µm granulometric size range. The catalysts (200 mg) were introduced in quartz straight reactors (5 mm i.d.). In each experiment, several samples were pretreated in parallel by heating up to 550 °C (temperature ramp of 5 °C min-1) in N2 (14 mL min-1 flow in each reactor) and keeping this temperature for 2 h. Afterward, the system was cooled to the reaction temperature (40 °C) under the same N2 flow. A propene (99.95% purity) flow of 8 mL min-1 was admitted for 1 h sequentially in each reactor to measure the initial metathesis activity of each sample. During activity

18666

J. Phys. Chem. C, Vol. 114, No. 43, 2010

Debecker et al. TABLE 1: MoO3 Weight Loading (ICP-AES) and Textural Properties (N2 Physisorption)

Figure 1. Evolution of the specific metathesis activity of (b) WI4, ([) WI8, (2) WI11, (9) WI16, (O) TS5, (]) TS10, (4) TS14, and (0) TS20 with time on stream.

measurement, the other reactors are kept under the same N2 flow. Propene and nitrogen were purified over Molsieve 3A (Roth) filters. N2 was also purified by an oxygen filter (Oxysorbglass, Linde). The composition of the reaction gas was analyzed by an Agilent 6890 GC. Product analysis took about 6.5 min for each injection. The separation of hydrocarbons was performed on an HP-AL/M column (30 m length, 0.53 mm i.d., 0.15 µm film thickness) applying a temperature ramp between 90 and 140 °C and FID detection. The experiments were carried out at atmospheric pressure. The selectivity to metathesis products was always very close to 100% (typically 99%). Only traces of secondary metathesis products (1-butene, pentenes, hexenes) and isomerization product (isobutene) were detected. The activity is calculated on the basis of metathesis products (ethene and trans- and cis-butene) formation. The specific activity is defined as the number of moles of propene converted to metathesis products per gram of catalyst and per hour. The standard deviation for activity measurements is 100 nm. In contrast, the sample obtained from this mixture after the TS procedure is totally amorphous. In fact, MoO3 crystals were never detected by XRD on either the TS10 sample (Figure 3) or on the TS catalysts with higher loading. Note that the progressive disappearance of the diffraction patterns of MoO3 crystals under the effect of the thermal treatment of such mixture can be observed by in situ XRD experiments.13,18 3.4. Transmission Electron Microscopy (TEM). Transmission electron microscopy confirms the presence of bulky MoO3 crystals in WI samples. The case of WI16 is presented in (Figure 4) and compared with the bare support. The latter is made of irregular porous particles. In the catalyst, the same type of particles is detected along with aggregates of crystalline particles in some places. The size of these particles is on the order of 20-60 nm, in agreement with the estimation from XRD. In the 10:90 (wt %) mechanical mixture of MoO3 and silica-alumina, the irregular particles of support are detected (Figure 5a). In addition, some large and regular crystals of MoO3 are observed (Figure 5b). Their size is comprised between 100 and 1000 nm. In the TS10 sample (namely, the catalyst obtained after TS of the 10:90 (wt %) mechanical mixture of MoO3 and silica-alumina), only irregular particles similar to those observed in the mechanical mixture are distinguished. The particles in Figure 5c appear, however, slightly different from the ones observed in the fresh mechanical mixture (Figure 5a). It appears that a kind of coating exists around the particles, like cement surrounding the irregularities of the particles. In parallel, the bulky regular structures attributed to MoO3 crystals are not detected anymore in the sample after TS, confirming again the observations made by XRD. 3.5. X-ray Photoelectron Spectroscopy. In XPS (Figure 6), the proportion of Mo detected at the surface of the 10:90 wt % MoO3/SiO2-Al2O3 mechanical mixture is very low. It increases drastically after TS. This evidences the spreading of Mo oxide at the surface of the silica-alumina support. When the MoO3 loading is varied, the Mo/(Si+Al) ratio evolves linearly. The proportion of Mo at the surface is very close to the proportion of Mo in the bulk composition of the samples. It is noteworthy that this ratio remains lower than that in the case of the reference WIx catalysts. 3.6. Raman Spectroscopy. On WI catalysts, Raman spectroscopy revealed the presence of both amorphous polymolybdate species and MoO3 crystals.9 Figure 7 shows the case of WI8 as a point of comparison for TS catalysts. WI samples were found to be heterogeneous in terms of Mo species detected by Raman spectroscopy. The nature and proportion of species detected change when several spectra were recorded from different particles of the same sample. This must be correlated to the fact that the MoO3 crystals form aggregates and are not homogeneously dispersed in the solid, whereas the area analyzed by the confocal tool is small. In TS catalysts, the band at 950 cm-1, which is typically attributed to 2D polymolybdates,23,24 was always observed. This band is growing with increasing MoO3 loading. This indicates

Debecker et al.

Figure 5. TEM micrographs of (a) a particle of the silica-alumina support found in the 10:90 wt % MoO3/SiO2-Al2O3 mechanical mixture, (b) a bulk MoO3 crystal found in the same sample, and (c) a particle of the TS10 catalyst (i.e., the sample obtained after thermal spreading of the mixture depicted in images a and b). Each scale bar represents 200 nm.

that MoO3 crystals are partially converted to dispersed Mo species during the thermal treatment. However, in contradiction with XRD and from ∼8 wt % MoO3 loading (TS10), Raman spectroscopy enlightens the presence of a growing amount of crystalline Mo oxide (main sharp bands at 666, 819, and 995

MoO3/SiO2-Al2O3 Metathesis Catalysts

Figure 6. Surface Mo/(Si+Al) atomic ratio measured by XPS as a function of the bulk Mo/(Si+Al) ratio for (2) WIx catalysts, (9) TSx catalysts, and (0) the 10:90 (wt %) mechanical mixture of MoO3 with the support (before thermal treatment). The arrow indicates the evolution of the surface Mo/(Si+Al) ratio after thermal spreading. The dotted line represents the value expected for totally homogeneous samples (same amount of Mo at the surface and in the bulk). Note: the bulk ratios are calculated from the experimental composition determined by ICP-AES.

Figure 7. Raman spectra obtained on the MoO3/SiO2-Al2O3 catalysts prepared by wet impregnation or thermal spreading: (a) support, (b,c) WI8 at different locations in the sample, (d) TS5, (e) TS10, (f) TS14, and (g) TS20. The * symbols indicate the signals attributed to crystalline MoO3, and the # symbol is used to indicate the band attributed to surface polymolybdates. Note: the spectra of TS20 has been attenuated (signal divided by 3) to fit in the artwork. As a result, the noise and the band at 950 cm-1 are also graphically attenuated.

cm-1).25 Only the TS5 sample shows absolutely no peak related to MoO3 crystallites. 3.7. 27Al MAS-NMR Spectroscopy. The environment of Al atoms was studied via 27Al MAS-NMR spectroscopy. In the silica-alumina support, the spectrum is characterized by three broad bands attributable to Al atoms in three different coordination states (Figure 8). The signal around 54 ppm is attributed to tetrahedrally coordinated framework Al atoms (AlTET) typically found in silico-aluminic materials.26 The signal at ∼2 ppm corresponds to Al octahedrally coordinated (AlOCT) like that in γ-Al2O327 or in amorphous polymeric aluminum oxide phases.28 The intermediate band at 30 ppm is attributable to

J. Phys. Chem. C, Vol. 114, No. 43, 2010 18669

Figure 8. 27Al MAS-NMR spectra of (a) the support, (b) the WI8 catalyst dried but not calcined, and (c) WI8. The AlTET, AlPEN, and AlOCT broad signals are indicated on the first spectrum. [AlMo6] and Al2(MoO4)3 are marked with * and #, respectively. A more complete discussion on the attribution of these five signals is proposed in ref 9.

five-coordinated Al atoms (AlPEN).26,29 In silica-aluminas, this coordination was assigned to the interface between an aluminatype phase and a truly mixed silica-alumina phase.30 The impregnation of AHM on the silica-alumina support changed the chemical environment of Al atoms. Figure 8 displays the spectra of the catalysts with 8 wt % MoO3 before and after calcination at 500 °C. In the dried (but not calcined) sample, a large peak at ca. 15 ppm is detected. This peak is attributed to the Anderson-type heteropolymolybdate that forms upon complexation of dissolved Al3+ cations by polyoxomolybdic anions in solution, as demonstrated by Carrier et al.31 This [Al(OH)6Mo6O18]3- species, abbreviated [AlMo6], was frequently observed during the preparation of supported MoO3 catalysts with Al-containing supports.9,32-36 The calcination of the catalyst led to the appearance of a new peak at ca. -14 ppm, attributed to aluminum molybdate, Al2(MoO4)337,38 The direct formation of aluminum molybdate from the calcination of [AlMo6] was already demonstrated unambiguously.32 Also, when the Mo loading was increased in WI catalysts, the peak at 15 ppm increased markedly, showing that Al2(MoO4)3 builds up.9 In TS samples, the peak at -14 ppm is never detected (even at high Mo loading), thereby evidencing the absence of the Al2(MoO4)3 phase in dry-made catalysts. The [AlMo6] species is observed with variable intensity (Figure 9). To verify the possible effect of aging on the formation of [AlMo6], one spectrum was recorded directly after the preparation of TS10 (Figure 10). No [AlMo6] is detected in this “fresh” TS sample. Leaving the newly prepared TS10 sample on the bench under ambient condition for 2 weeks led to the reappearance of the peak at 15 ppm. This demonstrates that the [AlMo6] species also forms not only during the wet step of an impregnation but also upon aging. It was also observed that the proportions of AlTET, AlPEN, and AlOCT in the “fresh” TS10 catalysts (Figure 9) are different from those found classically with aged samples (Figure 8). For a comparison purpose, the same procedure was applied on the freshly calcined support and on a 2 week aged support sample. The same changes in the AlTET, AlPEN, and AlOCT proportions were observed, and aging provoked the restoration of the initially obtained spectra. This shows that such modifications are simply due to the extent of the hydration of the support. 4. Discussion 4.1. Applicability of the Thermal Spreading. The TS method appears to be an attractive route because it is very

18670

J. Phys. Chem. C, Vol. 114, No. 43, 2010

Figure 9. 27Al MAS-NMR spectra of (a) the support, (b) TS5, (c) TS10, (d) TS14, and (e) AT20. The AlTET, AlPEN, and AlOCT broad signals are indicated on the first spectrum. [AlMo6] is marked with *.

Figure 10. 27Al MAS-NMR spectra of (a) the freshly calcined support, (b) the same sample after 2 weeks aging under ambient conditions, (c) the “fresh” TS10 catalysts, and (d) the same “fresh” TS10 catalyst after 2 weeks of aging under ambient conditions. The AlTET, AlPEN, and AlOCT broad signals are indicated on the first spectrum. [AlMo6] is marked with *.

straightforward. However, as observed from chemical analysis (Table 1), part of the MoO3 introduced in the synthesis (typically ca. 15% of the introduced MoO3) was lost during the preparation, presumably as a result of sublimation. This issue would, of course, need to be addressed in the perspective of scaled-up preparation. No such problem was encountered in the case of the wet impregnation route. It can appear to be surprising that the simple physical mixing of silica-alumina and of bulk MoO3 (both being inactive in the reaction), followed by a simple thermal treatment, yields active metathesis catalysts. Physico-chemical characterization data discussed below allow a better understanding of how the TS leads to the formation of supported Mo species that can yield active metathesis centers. TS was only scarcely exploited in the preparation of MoO3-based heterogeneous catalysts, generally on an empirical basis. Here it is shown that the MoO3/ SiO2-Al2O3 metathesis catalysts prepared in this way remain slightly less active than those obtained by the classical WI

Debecker et al. method. This will probably not disqualify the TS method for practical applications because the simplicity of the method may be considered to be a decisive benefit. 4.2. Advantages of the Thermal Spreading - Conversion of Bulky MoO3 Crystals into Dispersed MoOx Species. The intrinsic limitations of the wet impregnation method appear clearly from characterization data. Bulky MoO3 crystals form under the effect of the calcination, especially when the Mo loading is high. These crystals are clearly detected in XRD and Raman spectroscopy. The high proportion of Mo detected by XPS at the surface of WI catalysts was correlated to the formation of MoO3 crystals at the outer surface of the silica-alumina particles.9 The porosity of the support is virtually not affected, indicating that the Mo oxide phase does not enter the pores of the support. Consistently, MoO3 crystals are clearly observed in TEM, agglomerated at the outer surface of the support particles. In the same way, Al2(MoO4)3 builds up in WI catalyst after calcination. The formation of aluminum molybdate can be related to the presence of a subsequent amount of [AlMo6] in the dried catalyst (just after the wet step of the method).11 Both MoO3 crystals14 and Al2(MoO4)337 species have often been identified as inactive in the olefin metathesis. Their formation is correlated with the activity plateau reached when the Mo loading is increased above 8 wt %. Whereas the purpose of the present Article is to explore the potentialities of the TS method as an alternative to the wet impregnation method, it has to be mentioned that several strategies can be developed to improve the wet route to supported MoO3 catalysts. In particular, the use of additives is abundantly documented in the literature as a way to control the pH of the impregnation medium and to minimize the complexing potential of Mo species in solution and prevent the formation of unwanted species (typically, Al-Mo species). Using citrate,11 phosphate,39,40 or nitrilo triacetic acid41,42 additives proved to be a relevant way to better control the properties of the formed Mo oxide phase. A similar strategy for the improvement of the wet impregnation method is presently being explored by us in the case of MoO3/ SiO2-Al2O3 metathesis catalysts and will be the object of future publications. As far as the TS method is concerned, all characterization tools converge to demonstrate that the thermal treatment (partially) converts pure MoO3 crystals to amorphous molybdenum oxide spread over the support surface. The intense diffraction lines detected in the mechanical mixture disappear after the thermal treatment (Figure 3). The effect of the thermal treatment is also demonstrated in TEM. The absence of large MoO3 crystals in the thermally treated samples (while these structures were clearly observed in the fresh mechanical mixture) demonstrates the conversion of MoO3 crystals upon thermal treatment. It also seemed that the support particles after TS are embedded in a Mo oxide film presumably made of Mo oxide. Although this can not be taken as an irrefutable proof of the dispersion of MoO3 over the support, the latter hypothesis is further supported by Raman spectroscopy. A Raman band attributed to amorphous surface polymolybdates was observed, demonstrating the formation of dispersed species (Figure 7). Also, MAS-NMR investigations showed that a somewhat intimate interaction was created between Mo and Al atoms (at least part of them) because an aluminum polymolybdate phase is detected in aged samples (Figure 9). The very low Mo/(Si + Al) atomic ratio determined via XPS at the surface of the raw mechanical mixture reveals the initial heterogeneity of the sample. If the mixture was really homogeneous on the molecular scale, then the surface ratio would

MoO3/SiO2-Al2O3 Metathesis Catalysts

J. Phys. Chem. C, Vol. 114, No. 43, 2010 18671

Figure 11. Schematic representation of the mechanical mixture and of the TS catalysts. In the mechanical mixture, gray balls represent support particles and dark squares represent bulky MoO3 crystals. In the catalyst, after thermal spreading, gray balls surrounded by a dark line represent support particle covered by an amorphous MoOx phase and small dark squares represent persistent MoO3 crystallites. Dotted horizontal lines schematize the depth explored in XPS (ca. 10 nm). Note: this illustration is drawn as a guide to the eye only; relative proportions in size are not respected.

equal the bulk ratio (equal to 0.045 in the case of the 10:90 wt % mixture of MoO3/SiO2-Al2O3 and schematically represented by the dotted line in Figure 6 for the whole range of composition). In fact, in the mechanical mixture, MoO3 is present as bulky 3-D crystals surrounded by smaller silica-alumina particles. When pressed together in the sample holder for XPS analysis, only the surfaces of MoO3 crystals that reach the 10 nm external surface of the sample are probed. Considering the large size of these MoO3 crystals (as estimated via the Debye-Scherrer equation in XRD and as observed in TEM), a significant proportion of Mo atoms remains undetected. This situation is depicted schematically in the upper part of Figure 11. The steep rise in the Mo/(Si + Al) atomic ratio after thermal treatment is unambiguous evidence that Mo dispersion increases in the sample. As the MoOx phase gets dispersed over the surface of the silica-alumina material, the divergence between surface composition and bulk composition is reduced. The fact that the Mo/(Si + Al) ratio reaches ca. 0.043 in TS10 shows that a homogeneous distribution of Mo in the sample is reached (lower part of Figure 11). Textural analysis showed a noticeable decrease in specific surface area and in pore volume in TS catalysts as compared with the bare support. In parallel, the mean pore diameter remains relatively stable or tends to slightly increase. From the pore size distribution curves, it appears clearly that the porosity of the support is affected by the TS of MoO3. Consequently, it can be concluded that Mo oxide gets partially dispersed inside the pores of the support under the effect of the TS. The transport of Mo thus seems to operate upon relatively large distances and down to confined spaces. 4.3. Limitation of the Thermal Spreading - Incomplete Conversion of MoO3. Raman spectroscopy, in contradiction with XRD and TEM, revealed the presence of crystalline MoO3 species. The fact that these species are detected in Raman spectroscopy but not via XRD indicates that crystallites smaller than the XRD detection limit (∼5 nm43) actually remained in the mixture after TS. In other words, the spreading was not complete. It should be reminded that the procedure consisted of 8 h of TS, following the method proposed by Topka et al.,13 but in situ XRD experiments18 suggested that longer TS time would be more relevant (10 h). The thermal conversion of crystalline MoO3 to amorphous well-spread MoOx species was complete only in the sample with the lowest MoO3 loading. At higher loading, the Raman signals related to crystalline MoO3

increased steeply. In other words, the proportion of Mo atoms that remained in the form of small MoO3 crystallites was rising. It must be stressed that the relative band intensities for MoO3 and for polymolybdates do not directly reflect the relative amount of each species because crystalline MoO3 is a stronger Raman scatterer than amorphous molybdate species.32 4.4. Intricate Role of Al-Mo Mixed Phases. 27Al MASNMR results evidenced the presence of [AlMo6] in TS catalysts. This could appear to be surprising because the formation of such compound was discovered and is very often discussed for the case of preparation steps involving a solid-liquid interface. It must be remembered, however, that [AlMo6] can actually be formed independently from any wet step. Unambiguous evidence of the conversion of silica-alumina supported MoO3 crystallites into [AlMo6] under a humid atmosphere was provided by in situ Raman spectroscopy.32 These results explained the appearance of [AlMo6] in aged samples stored under ambient conditions.33,34,44 Also, in the present study, the analyses realized on the freshly prepared and on the 2 week aged samples confirm that [AlMo6] was formed during storage and manipulation of the sample under an ambient atmosphere. A clear link can be established between the presence of [AlMo6] in the impregnated catalyst before calcination and the subsequent formation of both Al2(MoO4)3 and MoO3.9 The formation of important amounts of [AlMo6] during impregnation led to the formation of Al2(MoO4)3 and bulk MoO3 crystals upon calcination. Conversely in the TS route, only the support and bulk MoO3 crystals are present when the thermal treatment is carried out. Consistently, the latter preparation method yields catalysts with no trace of Al2(MoO4)3. As far as crystalline MoO3 is concerned, TS and wet-made catalysts also differ significantly. The small crystallites present in TS samples clearly originate from the incomplete conversion of MoO3 crystals, as supported by in situ XRD.18 Bigger crystals (detectable in XRD) were formed in WI catalysts by the sintering of an amorphous deposit of polymolybdates, with a probable role of [AlMo6] decomposition. Therefore, the origin and size of the crystalline MoO3 species in both systems are different. It should be noted that the formation of bulk MoO3 crystals upon calcination of WI catalysts occurred during the 2 h of thermal treatment at 500 °C. Such crystals could, in principle, also be subjected to TS (also carried out in static air at 500 °C). The in situ XRD experiment previously reported,18 however, showed that the TS of MoO3 onto the support is slow. It is thus not excluded that

18672

J. Phys. Chem. C, Vol. 114, No. 43, 2010

better dispersion of MoO3 in the WI catalyst could be obtained if that longer calcination time is applied. 4.5. Correlation with Metathesis Activity. A distinction can be made between TS5 and the other TS samples with higher loading in terms of characterization of the Mo oxide deposit. All characterization techniques show that TS5 is very similar to WI4. In the case of TS5, the complete conversion of MoO3 was achieved, and this translates into good catalytic performances, close to those of the sample prepared by classical wet impregnation. TS catalysts with higher loading are systematically less active than the corresponding WI samples. The increase in loading only leads to a very limited increase in activity. Finally, the performances of TS catalysts level off around 12 mol g-1 h-1. The following paragraph is an attempt to correlate the catalytic behavior of these catalysts with characterization data, in conjunction with the observations made about wet-made catalysts. The absence of bulk MoO3 crystals and of Al2(MoO4)3, which have previously been pointed to as inactive, appears to be encouraging but is not translated into superior performances or steep activity increase upon increasing Mo-loading. The band detected in Raman spectroscopy at ca. 950 cm-1 represents dispersed polymolybdate species that are sometimes proposed to yield the active centers.13 Even if its intensity tends to increase with increasing loading, the increase in activity does not develop in parallel. At the same time, very small crystallites are detected in Raman spectroscopy. These Raman lines increase steeply when the MoO3 loading increases. The occurrence of such small MoO3 crystallites is the only observed characteristic of TS catalysts that can account for their modest performances. In consequence, not only the bulky MoO3 crystals detected via XRD and the significant amount of Al2(MoO4)3 found on the reference WI samples are responsible for imposing a limit in the activity of MoO3/SiO2-Al2O3 catalysts. Also, the very small (