In Situ XAS and XRD Studies on the Formation of Mo Suboxides

In Situ XAS and XRD Studies on the Formation of Mo Suboxides during Reduction of MoO3 .... The Journal of Physical Chemistry B 2002 106 (31), 7719-772...
38 downloads 3 Views 375KB Size
Download PDF
6360

J. Phys. Chem. B 2000, 104, 6360-6370

In Situ XAS and XRD Studies on the Formation of Mo Suboxides during Reduction of MoO3† T. Ressler,* R. E. Jentoft, J. Wienold, M. M. Gu1 nter, and O. Timpe Fritz-Haber-Institut der Max-Planck-Gesellschaft, Department of Inorganic Chemistry, Faradayweg 4-6, D-14195 Berlin, Germany ReceiVed: February 22, 2000

Reduction of MoO3 with hydrogen (5-100 vol %) in the temperature range from 573 to 823 K was studied by in situ X-ray diffraction and X-ray absorption spectroscopy. The experiments performed focused on elucidating phase composition and evolution with time under isothermal as well as temperature-programmed reduction conditions. At reaction temperatures below 698 K, the reduction of MoO3 to MoO2 is a one-step process without the formation of crystalline intermediates. At reduction temperatures above 723 K and H2 concentrations higher than 10 vol %, Mo metal is the final product of the reduction of MoO3. In addition, at temperatures higher than 698 K, the formation of Mo4O11 was observed. However, Mo4O11 is not an intermediate in the reduction of MoO3 but is being formed in a parallel reaction from MoO3 and MoO2 at temperatures above 698 K. It is shown that Mo4O11 can be obtained from a reaction of MoO3 and MoO2 at temperatures above 773 K, affording the same phase ratio of monoclinic and orthorhombic Mo4O11 as the reduction of MoO3 with hydrogen. Quantitative XRD analysis reveals a sigmoidal shape of the evolution of the MoO3 and MoO2 phases during reduction of MoO3 and an increase in the crystallite size of the phases present. This Oswald ripening indicates that a nucleation-growth kinetic mechanism governs the reduction of MoO3 under the conditions studied. The results presented in this work clearly demonstrate the potential of a combined application of in situ XRD and XAFS to reveal phase composition and kinetics of solid-state reactions.

Introduction Molybdenum is known to form a number of crystallographically well-defined suboxides (MonO3n-1, e.g., Mo9O26, Mo8O23, Mo5O14, and Mo4O11) whose average valences are between those of hexavalent molydenum trioxide and tetravalent molybdenum dioxide.1,2 With respect to the chemical reduction of MoO3 to MoO2, there is considerable speculation that some of these suboxides constitute intermediate steps of the reaction. However, although the preparation of molybdenum suboxides is described in detail in the literature,3,4 substantial debate prevails regarding the existence of any intermediates in the reduction of MoO3 to MoO2. Schematic representations of the crystal structures of three molybdenum oxides, i.e., MoO3, Mo4O11, and MoO2, are depicted in Figure 1. MoO3, a layered structure, and MoO2, a slightly distorted rutile structure, are composed of corner- and edge-sharing MoO6 octahedra, whereas the suboxide Mo4O11 contains layers of MoO4 tetrahedra that connect areas of merely corner-sharing octahedra (i.e., areas of idealized ReO3 structure). The reducibility of MoO3 or molybdenum mixed oxides has been the subject of numerous studies in the past (see refs 5 and 6 and references therein). The early stage of the MoO3 reduction, prior to the formation of any crystalline products, was investigated by Tho¨ni and Hirsch7 and Ga.8 These authors proposed that oxygen vacancies originating in the course of reduction † This work is dedicated to Prof. Dr. Wolfgang Metz on the occasion of his 65th birthday. * Author to whom correspondence should be addressed. E-mail: [email protected]. Phone: (+49) 30 8413 3192. Fax: (+49) 30 8413 4405.

resulted in the formation of shear planes in molybdenum trioxide. For the reduction of MoO3 and formation of MoO2, Arnoldy et al.5 postulated a one-step mechanism (MoO3 f MoO2, TPR, 67 vol % H2, 400-1200 K). Spevack and McIntyre found MoO2 to be the only well-defined product of the reduction of MoO3 in 100% hydrogen (623-1000 K).9 Regalbuto et al.6 studied the temperature-programmed reduction (TPR, 4 vol % H2, 400-1200 K) of diluted (SiO2) and supported MoO3 and reported a one-step reduction of MoO3 to MoO2. In contrast to the one-step reduction mechanism (MoO3 f MoO2), Burch 10 was the first to suggest that Mo4O11 is an intermediate product of the reaction. Following this line, Sloczynski and Bobinski11 proposed a consecutive autocatalytic reaction (CAR) model to describe the reduction kinetics of MoO3. The authors claimed that, in the temperature range 723823 K, the reduction of MoO3 is a consecutive reaction (i.e., MoO3 f Mo4O11 f MoO2), orthorhombic Mo4O11 is an intermediate product, and MoO2 is the final product.12 The dissociative adsorption of hydrogen was proposed to be the ratedetermining step, whereas the remaining elementary processes, i.e., the reaction of the surface atomic hydrogen with lattice oxygen, the crystallization of the products of the reduction, and the transport of reactants, are fast. In their model, the surface area of reduced grains is not blocked by the reduction product, and it diminishes in the course of reduction according to the shrinking core model.13 In addition to being a long-debated and still unresolved question in solid-state chemistry, the existence of suboxides during the reduction of MoO3 is closely related to the ongoing quest for the active molybdenum oxide phase in heterogeneous catalysis. Molybdenum oxide-based catalysts are, for instance,

10.1021/jp000690t CCC: $19.00 © 2000 American Chemical Society Published on Web 06/20/2000

Formation of Mo Suboxides

J. Phys. Chem. B, Vol. 104, No. 27, 2000 6361 Experimental Section

Figure 1. Schematic representation of the crystal structures of (i) MoO3 (Pbnm [5-508], layer structure), (ii) Mo4O11 (P21/a [13-142], layers of MoO4 tetrahedra in Mo4O11 are highlighted), and (iii) MoO2 (distorted rutile structure, P21/a [32-671]).

used extensively in the selective oxidation of hydrocarbons.14,15,16 It appears to be well established that the catalytically active phase is neither MoO3 nor MoO2 but rather a partially reduced molybdenum oxide. However, the nature of this “suboxide” is the subject of intense research efforts. It can be expected that only suitable in situ techniques can provide further experimental evidence regarding the active phase of molybdenum oxide-based catalysts under reaction conditions. To further elucidate the mechanism of MoO3 reduction and the existence of molybdenum suboxides, we have performed in situ structural studies of the reducibility of MoO3 under different H2 partial pressures and under isothermal as well as temperature-programmed reaction conditions. The potential of time-resolved X-ray absorption spectroscopy for structural studies on systems under rapidly changing reaction conditions has been presented in a recent publication.17 In this work, a combination of two complementary techniques, namely, in situ X-ray diffraction and X-ray absorption spectroscopy, was employed to elucidate phase compositions (long-range to shortrange order) and reaction kinetics under varying reaction conditions.

MoO3 Preparation and Characterization. Molybdenum trioxide (MoO3) was prepared by thermal decomposition of ammonium heptamolybdate (AHM), (NH4)6Mo7O24‚4H2O (Aldrich), in He (room temperature to 773 K, 2 K/min), followed by oxidation in flowing synthetic air at 773 K for 2 h. For comparison, commercially available MoO2 (Aldrich) and MoO3 (Aldrich) were measured as-purchased. The phase purity of the compounds used was verified by ex situ XRD. Specific areas were calculated by applying the BET method to the nitrogen adsorption isotherms obtained at liquid nitrogen temperature for samples outgassed at 473 K using a Quantachrom adsorption instrument. A BET surface area of 5.4 m2/g was determined for MoO3 prepared from AHM. Peak profile analysis of transmission XRD patterns yielded a crystalline material with an average crystallite size of ∼75 nm (primary particles, compared to ∼120 nm for the as-purchased MoO3). Assuming cubic crystallite morphology, this primary particle size corresponds to a surface area of about 15 m2/g, excluding substantial formation of larger agglomerates (secondary particles), with the primary particles not accessible by the gas phase. Raman measurements of MoO3 (from AHM) used in this work indicated a “low” defect density compared to highly crystalline MoO3 obtained via gas-phase transport reactions.18 In contrast to the MoO3 obtained from AHM, which did not show deviations from an isotropic crystal morphology, the as-purchased MoO3 exhibited a pronounced preferred orientation in the [010] direction (platelets). This is in agreement with previous publications.12 In Situ X-ray Absorption Spectroscopy. For in situ XAFS experiments, molybdenum oxide was mixed with boron nitride (in a ratio of 1:3), and 30 mg of the mixture was pressed with a force of 1 ton into a 5-mm-diameter self-supporting pellet. The absorption jump, ∆µx, at the Mo K edge was ∼2. In situ XAS experiments were performed in transmission in a flow reactor19 at atmospheric pressure. Reaction temperature and reactant mass flow were controlled with a Eurotherm PID temperature controller and a Bronkhorst mass flow controller, respectively. For isothermal reduction experiments, the reactant (H2, diluted in helium) flow was controlled at 20 mL/min, and flow was switched from oxygen (20 vol % in helium) to hydrogen with an intervening helium purge (20 mL/min). Because of the small volume of the in situ cell (3.8 mL), switching between reactants results in a rapid introduction and depletion of reactants.17 Temperature-programmed reduction or reaction experiments were conducted in flowing reactants (20 mL/min) at a constant heating rate of 5 K/min. The product composition at the in situ cell gas outlet was continuously monitored using a mass spectrometer in a multiple ion detection mode (QMS200 from Pfeiffer) with a time resolution of ∼2 s per spectrum for time-resolved experiments and ∼12 s per spectrum for conventional in situ studies. Absolute temperature measurements were calibrated with the same reference compounds used for the calibration of the in situ XRD setup (see below). A maximum deviation of 5 K between tabulated transition temperatures and measured temperatures was obtained. In situ transmission XAS experiments were performed at the Mo K edge (19.999 keV20) at beamline X1.1 at the Hamburger Synchrotron Radiation Laboratory, HASYLAB, utilizing a Si(311) double-crystal monochromator in the Quick-EXAFS mode21 (measuring time, ∼2 min/scan). The storage ring operated at 3.6 GeV with injection currents of 150 mA. Timeresolved in situ XAS experiments were carried out at the Mo K edge utilizing an energy-dispersive spectrometer (European

6362 J. Phys. Chem. B, Vol. 104, No. 27, 2000

Ressler et al.

Figure 2. Weight fraction calibration curve obtained from a least-squares fit of MoO2 and MoO3 reference XANES spectra to spectra of mechanical mixtures of MoO2 and MoO3. A nearly linear correlation between weight fraction of MoO3 and XANES fit results can be seen. The inset shows a refinement of MoO2 and MoO3 reference spectra to the XANES spectra of a selected mixture (MoO2/MoO3 ratio of 3:1).

Synchrotron Radiation Facility, ESRF, ID2422) equipped with a curved Si(111) polychromator in a transmission mode23 (measuring time, ∼3 s/scan). The storage ring operated at 6.0 GeV with injection currents of 200 mA in a 16 bunch mode. XAFS Data Analysis. Details on data reduction procedures for energy-dispersive X-ray absorption spectra (i.e., energy calibration, etc.) can be found in refs 24 and 25. X-ray absorption fine structure (XAFS) analysis was performed using the software package WinXAS97 v1.326 following recommended procedures from the literature.27 Background subtraction and normalization were performed by fitting linear polynomials to the pre-edge and post-edge regions of an absorption spectrum, respectively. The Mo K edge absorption threshold was determined from the first root in the first derivative of the near-edge region (XANES) (in the following, edge shifts are reported relative to the first inflection point in the Mo metal K edge XANES at 19 999 eV). Reduction or oxidation of Mo causes a shift in the Mo K absorption edge to lower or higher photon energies, respectively, and hence, the edge shift can be used to identify changes in Mo valence during reaction.17 In addition to information on the average Mo valence or coordination geometry, the fingerprint character of the nearedge region can be employed to extract qualitative and quantitative information regarding the occurrence of different phases during in situ XAFS experiments. Principal component analysis (PCA, aka “factor analysis”) can be utilized to identify Mo oxide phases present during the reduction and reoxidation of MoO3. Given a set of molybdenum reference spectra, PCA can identify those references that constitute probable components in the original set of experimental XANES spectra. Subsequently, after determination of the number and types of phases present, a leastsquares fitting procedure can be applied to obtain the fraction of each reference phase under oxidation/reduction conditions. Hence, XANES analysis can afford information on lowconcentration (∼1 wt %) or amorphous phases not readily

available with XRD. However, to convert the fraction of each phase as determined by XANES fitting into a weight percent, an appropriate calibration curve must be assembled. From the calibration curve in Figure 2, obtained for mechanical mixtures of MoO3 and MoO2, a nearly linear relationship between the two quantities is apparent, facilitating the described analysis. Details on PCA analysis and the numerical procedures employed can be obtained from the literature.28,29 Because the scope of this paper is the identification of different molybdenum oxide phases and the evolution of their concentration as a function of the reaction conditions, no detailed structural analysis (EXAFS) of the phases observed will be presented here. This analysis will be the subject of a future publication. X-ray Diffraction. Ex situ X-ray diffraction measurements for phase analysis were conducted using a STOE transmission diffractometer STADI-P (Ge primary monochromator, Cu KR1 radiation) equipped with a position-sensitive detector. Crystalline phase identification based on XRD patterns was aided by the ICDD-PDF-2 database. In situ XRD experiments were carried out in Bragg Brentano scattering geometry on a STOE STADIP P diffractometer equipped with a secondary monochromator (Cu KR1 radiation) and a scintillation counter operated in a stepping mode. The in situ cell consisted of a Bu¨hler HDK S1 high-temperature diffraction chamber mounted onto the goniometer (volume of ∼400 mL) and connected to a gas-feed system (He, O2, and H2) employing mass flow controllers (Bronkhorst). A resistively heated stainless steel ribbon is used as a sample holder, and the sample temperature is measured with a stainless steel clad thermocouple (type K) in contact with the bottom side of the heating band. The sample temperature is controlled with a programmable Eurotherm 818 PID controller, allowing controlled heating to a maximum of 1000 K. The temperature of the sample was calibrated using reference compounds (i.e.,

Formation of Mo Suboxides

J. Phys. Chem. B, Vol. 104, No. 27, 2000 6363

Figure 4. Evolution of XRD patterns during reduction of MoO3 in 10 vol % hydrogen in a temperature range from 573 to 773 K. The onset of reduction can be observed at 673 K. A holding time of 12 h at 773 K resulted in a mixture of MoO2 and Mo metal. Characteristic diffraction lines of MoO2, MoO3, and Mo metal are indicated.

Figure 3. (a) X-ray diffraction patterns and (b) molybdenum K nearedge spectra of reference molybdenum oxides (MoO3, MoO2, and m-Mo4O11). Characteristic (hkl) diffraction lines are indicated in (a).

RbNO3, Ag2SO4, and R-Quarz) with characteristic phase transition temperatures. Deviations between tabulated and measured temperatures were on the order of 5 K. Typically, 15 mg of sample are uniformly spread on the heating band. Because of this preparation mode, some preferred orientation of platelets on the sample holder may be present. The gas-phase composition at the cell outlet was analyzed on line with a Pfeiffer Prisma 200 quadropole mass spectrometer in a multiple-ion detection mode using a Faraday cup and allowing a quantitative analysis of the gas phase. Further details on the setup are published elsewhere.30 XRD measurements reported here were conducted under atmospheric pressure in flowing reactants (∼100 mL/min total flow). These conditions permit an efficient removal of water, desirable in studies of the reducibility of MoO3.5 Unless stated otherwise in the Results section, temperature-programmed experiments were performed at a heating rate of 2 K/min with intermediate isothermal holding periods of about 1 h every 25 K for XRD pattern acquisition. Thus, the uncertainty in given onset temperatures of reduction or phase formation is 25 K. Results X-ray diffraction (XRD) patterns and X-ray absorption nearedge (XANES) spectra of the three molybdenum oxide reference compounds (MoO3, Mo4O11, and MoO2) in Figure 1 are shown in Figure 3a and b, respectively. Evidently, the apparent differences in the XRD patterns and XANES spectra (fingerprint) of the three compounds permit a reliable detection and identification of the corresponding phases in reaction mixtures. In Situ X-ray Diffraction. A large number of in situ XRD experiments were performed in a temperature range from 573 to 823 K and at hydrogen concentrations between 5 and 100

Figure 5. Evolution of XRD patterns during reduction of MoO3 in 5 vol % hydrogen at 773 K. A diffraction pattern of unreduced MoO3 at 573 K is shown for comparison. Characteristic diffraction lines of the constituent Mo oxides are indicated. The formation of MoO2 can be seen prior to the formation of Mo4O11.

vol %. A selection of experiments, which emphasize those points in reaction parameter space (i.e., temperature and hydrogen concentration) that constitute boundaries in terms of reaction products or intermediate phases observed, will be described in detail in the following. A summary of all experiments, i.e., a two-dimensional “reduction product diagram”, will be presented in the Discussion section. Figure 4 shows the evolution of XRD patterns during temperature-programmed reduction of MoO3 in 10 vol % hydrogen in the range from 573 to 773 K and a subsequent isothermal period of 12 h at 773 K. At temperatures below 623 K, no reduction of MoO3 to MoO2 was observed. The onset of reduction of MoO3 and formation of MoO2 can be seen at 673 K. When the system was heated further to 773 K and held at this temperature for 12 h, reduction of the intermediate MoO2 to Mo metal was observed. No additional phases, in particular no intermediates in the reduction of MoO3 to MoO2 were detected. In Figure 5, the evolution of XRD patterns during reduction of MoO3 in 5 vol % hydrogen at 773 K is presented. Under these conditions, the reduction of MoO3 resulted in the formation of both MoO2 and the molybdenum suboxide Mo4O11. The formation of MoO2 is observed prior to that of the suboxide. Furthermore, in all reduction experiments in which the formation

6364 J. Phys. Chem. B, Vol. 104, No. 27, 2000

Figure 6. Evolution of XRD patterns during reduction of MoO3 in 5 vol % hydrogen at 673 K. The formation of MoO2 can be seen after 2 h. After an isothermal period of 40 h, reduction of MoO3 to MoO2 amounted to more than 70%. No intermediate phase was observed during reduction. Characteristic (hkl) lines for MoO3 and MoO2 are indicated.

of Mo4O11 was observed, a mixture of monoclinic and orthorhombic Mo4O11 was obtained (doublet at ∼22°/2θ in Figure 5). In the following, “Mo4O11” without the prefix o or m, will denote a mixture of the two phases. Figure 6 shows the evolution of XRD patterns during reduction of MoO3 in 5 vol % hydrogen at 673 K. The formation of MoO2 was observed after an “induction period” of about 2 h at this temperature. The total conversion of MoO3 to MoO2 after 40 h was about 70 wt %. The experimental XRD patterns in Figure 6 can be entirely described by a mixture of MoO2 and MoO3. From the in situ XRD experiments performed at temperatures below 773 K, it was found that the reduction of MoO3 resulted directly in the formation of MoO2. No additional crystalline phases (e.g., molybdenum suboxides) were detected in this temperature range. Under more severe reduction conditions (50 vol % hydrogen), the onset of reduction of MoO3 is shifted to lower temperatures (∼623 K), and MoO2 is the only observable product (at temperatures < 773 K) or is an intermediate in the reduction of MoO3 to Mo metal (at temperatures > 773 K). In addition to the reduction of MoO3 in hydrogen, the reaction of MoO3 and MoO2 in He was studied by in situ XRD. Figure 7 shows the evolution of XRD patterns during the temperatureprogrammed reaction of MoO3 and MoO2 (molar ratio of 3:1) in He (1 atm). For this experiment, each step consisted of an increase in the reaction temperature by 100 K at 2 K/min and a holding period of 10 h at constant temperature. During the 10-h holding time, five diffraction patterns were recorded. The last pattern of each step is displayed in Figure 7. From Figure 7, it can be seen that minor amounts of Mo4O11 can be observed after 10 h at 773 K. After 2 h at 823 K, however, a considerable yield of Mo4O11 from a mixture of MoO2 and MoO3 was obtained. From XRD measurements of the reaction of MoO3 and MoO2 (ratio of 3:1) under isothermal conditions, it was found that, at a reaction temperature of 773 K after 40 h, about 40 wt % of the mixture had reacted to Mo4O11. After 5 h at a reaction temperature of 823 K, more than 90 wt % Mo4O11 was obtained. In all cases, the reaction of MoO2 and MoO3 yielded a mixture of orthorhombic and monoclinic Mo4O11, which, however, transformed almost completely (>95 wt % at 823 K in 20 h) into the monoclinic phase after a prolonged reaction time. This phase transformation with time is to be expected as orthorhombic Mo4O11 is the thermodynamically stable high-temperature

Ressler et al.

Figure 7. Evolution of XRD patterns during temperature-programmed reaction of MoO3 and MoO2 in He. The patterns shown were taken after 10 h at a given reaction temperature. Characteristic diffraction lines for MoO3, MoO2, and Mo4O11 are indicated. After 10 h at 773 K, only minor amounts of Mo4O11 can be observed. After 2 h at 823 K, the reaction afforded a conversion to Mo4O11 of about 50 wt %.

Figure 8. Evolution of Mo K edge XANES spectra during reduction and reoxidation of MoO3 at 823 K in 100 vol % H2 and 100 vol % O2, respectively. The normalized mass signals of H2 and O2 are schematically depicted. The majority phases MoO3 and MoO2 are indicated.

phase3 (T > 1000 K), whereas monoclinic Mo4O11 is the lowtemperature phase. Reoxidation of MoO2 at elevated temperatures (>723 K) proceeded too quickly to be followed by the in situ XRD setup employed. At temperatures below 723 K, no additional crystalline phases were observed during reoxidation (i.e., MoO2 f MoO3). In Situ X-ray Absorption Near-Edge Spectroscopy. In situ XAFS measurements were performed to elucidate the amount of amorphous or minority phases that were not detected by the in situ XRD measurements (less than 5.0 wt %). In addition, time-resolved in situ XAFS experiments (measuring time ∼ 3 s) were performed particularly under those reaction conditions where the reduction of MoO3 proceeded too rapidly to be followed by in situ XRD [T > 773 K, c(H2) > 50 vol %]. A number of TPR and isothermal reduction experiments were carried out employing both conventional and energy-dispersive XAFS setups. A summary of the results will be provided in the Discussion section. In this section are described representative experimental runs that were performed under reaction conditions that are located at certain boundaries in reaction parameter space (temperature and hydrogen concentration) with respect to the onset of reduction or molybdenum oxide phases formed. Figure 8 displays a series of Mo K near-edge spectra measured during reduction and reoxidation of MoO3 at 823 K with hydrogen and oxygen. The XANES spectra were measured over a period

Formation of Mo Suboxides

Figure 9. Mo K edge position (relative to the first inflection point in the Mo metal K edge XANES spectra at 19 999 eV) of MoO3 during reduction and reoxidation at different temperatures [i.e., 773, 798, and 823 K (Figure 8)]. The arrows indicate the edge shift direction that corresponds to a reduction or oxidation of molybdenum. Normalized O2 and H2 mass signals measured during reduction/reoxidation at 823 K are shown for illustration.

of 10 min with a time resolution of 2.8 s per spectrum. The noticeable increase or decrease in the first Mo K edge absorption feature corresponds to a reduction or oxidation of molybdenum, respectively. It can be seen that the reduction of MoO3 at 823 K (100% H2) proceeded rapidly (i.e., several minutes) relative to the above-shown XRD results (5 vol % H2, reaction time of several hours). In Figure 9, Mo K edge-shift data obtained from XANES spectra measured during the reduction of MoO3 at different temperatures are presented. Different rates of reduction can be observed. In addition, at 798 and 823 K (Figure 9), the corresponding Mo K edge-shift curves appear to exhibit two separate regions with different slopes, indicating that more than one reaction mechanism is active at temperatures above 773 K. Furthermore, from Figure 9, it can be seen that the reoxidation proceeded much more rapidly than the reduction17 and that it appears to be only gas-transport limited. Principal component analysis of XANES spectra measured during the reduction of MoO3 in pure hydrogen at 823 K revealed three primary components to be present in the experimental XANES data. By means of target transformation, the three oxides MoO3, MoO2, and monoclinic Mo4O11 were found to be suitable references. However, it must be recalled that a phase mixture of monoclinic and orthorhombic Mo4O11 was observed by in situ XRD measurements during the reduction of MoO3. Thus, it must be assumed that either the two Mo4O11 structures cannot adequately be distinguished by their XANES features or that they exhibit a constant phase ratio during the reaction. Therefore, in the quantitative XANES analysis presented in the following, “Mo4O11” refers to a mixture of monoclinic and orthorhombic Mo4O11. Figure 10 shows the results of a quantitative phase analysis of Mo K near-edge spectra measured during the reduction of MoO3 at 823 K. A short induction period (IP in Figure 10) seems to be present, as previously reported by Gajardo et al.31 who found an induction period of ∼2 h in the isothermal reduction of bulk MoO3 at 673 K. After the induction period, the formation of Mo4O11 can be seen prior to the formation of MoO2. A complete conversion of MoO3 to a mixture of MoO2 and Mo4O11 can be observed after ∼5 min, and a complete reduction of MoO3 to MoO2 after ∼15 min under the conditions employed. During the very rapid reoxidation of MoO2, the formation of Mo4O11 can also be observed. From XAFS experiments performed at temperatures below 773 K (50 vol % to 100 vol % hydrogen), it was found that the

J. Phys. Chem. B, Vol. 104, No. 27, 2000 6365

Figure 10. Evolution of phase composition during reduction (100 vol % H2) and reoxidation (100 vol % O2) of MoO3 at 823 K. The number and type of phases present (i.e., MoO3, MoO2, and Mo4O11) were determined by PC analysis. The amount of each phase was obtained from a least-squares fit of reference spectra to the corresponding experimental XANES spectra. Hydrogen (dashed) and oxygen (dotted) mass signals are shown. A short induction period (IP) can be seen in the reduction of MoO3. It can be seen that MoO3 is entirely consumed after ∼5 min. The formation of Mo4O11 can be observed during reduction and reoxidation.

Figure 11. Evolution of phase composition (MoO3, MoO2, and Mo4O11) during temperature-programmed reaction of MoO3 and MoO2. The amount of each phase was determined from experimental XANES spectra as described above. The onset of the reaction can be observed at 773 K.

reduction of MoO3 did not result in the formation of any detectable well-defined molybdenum suboxide phases. A onestep reduction of MoO3 to MoO2 was observed in these cases. To corroborate XRD measurements of the reaction of MoO3 and MoO2, the same reaction was followed by in situ XAFS. PC analysis of the corresponding Mo K near-edge data afforded the reference spectra of three different phases, namely, MoO3, MoO2, and Mo4O11, to comprise the set of experimental XANES spectra. The evolution of the concentration of the three molybdenum oxide phases with increasing temperature, as obtained from a least-squares refinement of a sum of the three reference spectra to the experimental spectra, is displayed in Figure 11. At a temperature of about 773 K, the onset of the reaction can be seen in a strong increase in the Mo4O11 concentration. The final concentration of Mo4O11 amounted to about 20 wt %, which is on the same order of magnitude as the conversion observed in the XRD data. Furthermore, a Mo K edge-shift of ∼0.2 eV was measured during the reaction of MoO2 and MoO3. This indicates that the Mo average valence did not change considerably during the reaction (complete reduction of MoO3 to MoO2 results in a Mo K edge shift of

6366 J. Phys. Chem. B, Vol. 104, No. 27, 2000

Figure 12. Formation of molybdenum oxide phases during reduction of MoO3 as a function of hydrogen concentration and reaction temperature. Light gray region below 623 K (MoO3): no reduction of MoO3 is observed within 10 h. Gray region above 623 K (MoO2): MoO3 is reduced to MoO2 without the formation of other oxides being detected by XRD or XAFS. Dark gray region above 673 K (Mo): MoO3 is reduced to Mo metal in a two-step reaction (i.e., MoO3 f MoO2 f Mo). Vertical dotted lines (Mo4O11): formation of Mo4O11 observed during reduction of MoO3 to MoO2 (gray) or Mo metal (dark gray).

∼4.0 eV17), which, in turn, excludes reduction of MoO3 alone as the source of the Mo4O11 formation observed. Discussion The in situ XRD and XAFS experiments presented in the Results section indicate that, in the range of temperature from 623 to 823 K and of hydrogen concentration from 5 to 100 vol %, molybdenum suboxides (i.e., Mo4O11) form only under certain reaction conditions during the reduction of MoO3. Both X-ray diffraction and absorption results show the formation of Mo4O11 at temperatures above 698 K (Figures 5 and 10). However, at temperatures below 698 K, the formation of suboxides was not detected under any reaction conditions (hydrogen concentration and reaction time) whatsoever (Figure 6). To facilitate the discussion of the results presented, a schematic summary of the different molybdenum oxide phases (final as well as transient) observed during reduction of MoO3 as a function of hydrogen concentration and reaction temperature is depicted in Figure 12. However, the chosen representation in Figure 12 must not be mistaken as a phase diagram, because it does not show only the thermodynamically stable phases or products at a given temperature. In Figure 12, two major regions can be seen at temperatures below 673 K. The first region is denoted “MoO3” and corresponds to reaction conditions under which no reduction of MoO3 was detected. The second region is denoted “MoO2” and encompasses those reaction conditions that resulted in a onestep reduction of MoO3 to MoO2 without the formation of other crystalline phases (Figure 6). It can be seen from Figure 12 that the onset temperature of reduction of MoO3 (623 K) is not significantly affected by an increasing hydrogen concentration. Under reduction conditions ranging from 673 K and 100% hydrogen to 773 K and 10 vol % hydrogen (dark gray area denoted “Mo” in Figure 12), Mo metal is the final product of the reduction of MoO3 (Figure 4) regardless of intermediate steps or other crystalline phases formed. In contrast to the reduction of MoO3 to MoO2, a strong influence of the hydrogen concentration on the onset temperature of MoO2 reduction to Mo metal is evident from Figure 12.

Ressler et al. In contrast to the results at reaction temperatures below 673 K where a one-step reduction of MoO3 to MoO2 was found, the formation of Mo4O11 is observed at temperatures above 673 K (Figure 5, vertical dotted lines in Figure 12). However, as can be seen from the slope of the dashed line in Figure 12, the reaction temperature at which the formation of Mo4O11 can be observed in isothermal experiments is shifted toward higher temperatures with an increasing hydrogen concentration. From the in situ XRD and XAFS studies performed on the reaction of MoO2 and MoO3 (Figures 7 and 11, respectively), it was learned that the reaction to Mo4O11 proceeded at a considerable rate at temperatures above 773 K. Hence, the formation temperature of Mo4O11 from MoO3 and MoO2 coincides with the temperature regime in which the considerable formation of Mo4O11 (>50 wt %) was observed during the reduction of MoO3 with hydrogen. In addition, Mo4O11 was the only product formed during the reaction of MoO2 and MoO3 regardless of the phase composition of the starting material. None of the other crystallographically well-defined suboxides (e.g., Mo8O23, Mo9O26, and Mo5O14) was obtained in any detectable amount. This again coincides with the phases observed during the reduction of MoO3 for which no suboxides other than Mo4O11 were formed in considerable amounts at temperatures above 773 K. Furthermore, the same mixture of monoclinic and orthorhombic Mo4O11 was obtained during both the initial stage of the reaction of MoO3 and MoO2 and the reduction of MoO3. In the former, the prolonged reaction time eventually resulted in the nearly complete transformation of the orthorhombic into the monoclinic phase. In addition, it was found that the reaction proceeded best for an undiluted sample, whereas a diluted sample (MoOx:BN ) 1:3) yielded a much lower conversion to Mo4O11 (∼6 wt %) after the same reaction time. Thus, an intimate mixing and close contact of the individual molybdenum oxide grains is required for the solidstate reaction to proceed at a substantial rate. Taking these results into account, it seems that the formation of the well-defined suboxide Mo4O11 during the course of the reduction of MoO3 with hydrogen is not due to a two-step reduction mechanism but rather to a parallel reaction of the product MoO2 with unreacted MoO3. Apparently, the high reaction rate of the reduction of MoO3 under severe conditions [c(H2) > 20 vol %] and the formation of crystalline Mo4O11 from MoO3 and MoO2 only under moderate hydrogen concentrations (773 K) explains the seemingly contradicting reports in the literature regarding the formation of suboxides during the reduction of MoO3. From the results presented here, a competition between the reduction of MoO3 and the formation of Mo4O11 can be assumed. At low hydrogen concentrations ( 698 K)

3MoO3 + MoO2 f Mo4O11 Mo4O11 + 3H2 f 4MoO2 + 3H2O

(I) (II)

To access the reaction kinetics that govern the reduction of MoO3, a quantitative phase analysis based on in situ XRD patterns was performed. For such an analysis, the patterns are fit32 with a mixture of MoO3 and MoO2 (e.g., Figure 6) or MoO3, MoO2, and Mo4O11 (e.g., Figure 5). The resulting phase composition (MoO3, MoO2, and Mo4O11) during isothermal reduction of MoO3 as a function of reduction time is depicted in Figure 13a and b for different reaction conditions. Figure 13a shows the evolution of the MoO2 and Mo4O11 phase fractions during reduction at 723 K in 5, 10, and 20 vol % H2.

Reduction of MoO3 in 5 vol % H2 at 723 K yielded about 10 wt % of Mo4O11, whereas in 10 vol % H2, the fraction of Mo4O11 formed amounted to more than 60 wt %. However, with an increase in the hydrogen concentration to 20 vol %, a considerable decrease in the weight fraction of Mo4O11 can be seen (∼20 wt %). Evidently, in 20 vol % H2, the reduction of MoO3 to MoO2 proceeds too rapidly to permit the formation of large amounts of Mo4O11 from MoO2 and unreacted MoO3. Phase evolution during reduction of MoO3 in 5 vol % H2 at 748 and 773 K is shown in the top and bottom parts, respectively, of Figure 13b. A strong increase in the fraction of Mo4O11 can be seen at 773 K compared to that at 748 K, together with an earlier onset of Mo4O11 formation at higher temperatures. Eventually, a further increase in the reaction temperature can result in the detection of Mo4O11 prior to that of MoO2 (Figure 10), indicating that, under these conditions, MoO2 “immediately” reacts with MoO3 to form Mo4O11. An intermediate stage can be observed at 773 K, where MoO2 is first formed (below 1.0 in Figure 13b, bottom) and subsequently consumed by the reaction with MoO3 to yield Mo4O11 (between 1.0 and 2.0 in Figure 13b, bottom). In addition, it can be seen from Figure 13 that the reduction of Mo4O11 proceeds much more slowly than that of MoO3. The apparent sigmoidal shape of the MoO3 reduction curve (Figures 13 and 14) can be described by Avrami-Erofeyev kinetics indicating that nucleation of the solid product of reduction is the rate-determining step.33 In Figure 14, the normalized evolution of the MoO3 phase fraction during reduction in 5 vol % H2, measured by XRD at different reaction temperatures, as well as in 100% H2 at 823 K, measured by XAS, is depicted. A very similar phase evolution with respect to MoO3 was obtained in all cases studied. Under no reaction conditions a zero-order reaction (dR/dt ) 0, with R representing the degree of reduction) observed, as has been previously reported and attributed to an oxygen-diffusion-controlled reduction mechanism.5 Least-squares refinements of the Avrami-Erofeyev equation [R ) 1 - exp(-kt)n, with R representing the “reduction parameter” (ranging from 1 to 0), k the rate constant, t the reduction time, and n the exponential factor (ranging from 2 to 3)] to the evolution of the MoO3 phase measured in isothermal reduction experiments were employed to determined the MoO3 reduction rate constants kMoO3. In the following, results obtained

6368 J. Phys. Chem. B, Vol. 104, No. 27, 2000

Ressler et al.

Figure 15. Evolution of half-life of MoO3 [t1/2(MoO3)] during reduction in 40 vol % hydrogen with increasing reaction temperature. Detection of Mo4O11 during in situ XRD measurements is indicated (open circles).

Figure 16. “Arrhenius plot” [ln(k) versus 1/T] for the reduction of MoO3 in 40 vol % hydrogen in the temperature range from 623 to 773 K. Rate constants k were obtained by fitting the Avrami-Erofeyev equation [R ) 1 - exp(-kt)n] to the evolution of the MoO3 phase measured in isothermal reduction experiments. Two regions with different slopes can be distinguished, corresponding to reduction of MoO3 (I) with the parallel formation of Mo4O11 and (II) without the detection of Mo4O11. The apparent activation energies EA,I and EA,II as determined from the two slopes are indicated.

for the reduction of MoO3 in 40 vol % hydrogen are presented. Under such conditions, the temperature ranges with (723-823 K) and without (623-698 K) the detection of Mo4O11 in isothermal XRD experiments are equivalent (Figure 12), permitting a reliable comparison of the half-life of MoO3 and the rate constants for the two regions. The half-life of MoO3 in 40 vol % H2 as a function of reaction temperature is depicted in Figure 15 with two separate regions in temperature clearly visible. An “Arrhenius plot” for the reduction of MoO3 in 40 vol % hydrogen in the temperature range from 623 to 773 K is shown in Figure 16. Evidently, two regions with different slopes can be distinguished, corresponding to the reduction of MoO3 (I) with and (II) without the formation of Mo4O11 (Figure 12). From two linear curves fit to the data points in regions I and II apparent activation energies are determined. For the reduction of MoO3 without parallel formation of Mo4O11, an apparent activation energy of EA,II ) 34 kJ/mol is obtained, whereas for that region in temperature where the formation of Mo4O11 is detected, an apparent activation energy of EA,I ) 103 kJ/mol is calculated. From an “Arrhenius analysis” of the reduction of MoO3 in 5 vol % hydrogen, apparent activation energies of EA,I ) 433 kJ/mol and EA,II ) 133 kJ/mol were obtained, indicating a dependence of the activation energies of reduction of MoO3 on the hydrogen partial pressure. In total, the apparent activation

Figure 17. Evolution of crystallite sizes [Scherrer formula applied to the main diffraction (hkl) line, as indicated in Figure 3a] during reduction of MoO3 (a) at 673 K in 5 vol % hydrogen [MoO3 and MoO2 observed (Figure 6)] and (b) at 773 K in 5 vol % hydrogen [MoO3, MoO2, and Mo4O11 observed (Figure 5)].

energies obtained corroborate the assumption of a parallel reaction of MoO3 and MoO2 to yield detectable amounts of Mo4O11 only at higher reaction temperatures. In addition to phase composition, crystallite sizes of the observed molybdenum oxide phases were determined from XRD patterns employing the Scherrer equation.34 The evolution of the crystallite size of the constituent phases during reduction of MoO3 at 673 K (XRD patterns in Figure 6) and 773 K (XRD patterns in Figure 5) is presented in Figure 17a and b, respectively. The increase in the crystallite sizes of all phases present (Figure 17a and b) indicates that an Oswald ripening35 occurs during the reduction of MoO3 in hydrogen. This is in agreement with the assumption that a nucleation-growth kinetic mechanism governs the reduction of MoO3 under the reaction conditions studied in this work. The different rates of MoO3 crystallite growth at 773 and 673 K (Figure 17a and b) compared to that of MoO2 may indicate an accelerated annealing of MoO3 at higher temperatures. An alternative interpretation for the sigmoidal reduction kinetics was published by Sloczynski who postulated that the dissociative absorption of hydrogen is the rate-determining step in the reduction of MoO3.36 However, a kinetic model that requires a shrinking core characteristic as proposed in the literature11,12,13 cannot properly explain the diffraction domain growth observed. Other than Mo4O11, no additional well-defined suboxides were observed during the reduction of MoO3 within the parameter space investigated. This excludes the formation of long-rangeordered Magnelli structures (sheer structures,2 e.g., Mo8O23) during MoO3 reduction in hydrogen. The lower detection limits of the techniques applied in this work, i.e., XRD and XAFS, are about 5 wt % for a crystalline suboxide phase and about 1

Formation of Mo Suboxides wt % for an amorphous, disordered, or crystalline suboxide phase, respectively. An analysis of the concentration of point defects or disordered sheer structures that may be present in the starting material37,38 as well as during reduction is beyond the scope of this paper. A detailed structural investigation utilizing a combination of EXAFS analysis of absorption spectra measured in situ during reduction and reoxidation and postmortem transition electron microscopy (TEM) will be presented in a forthcoming paper. The results obtained from the in situ XRD and XAFS experiments presented here underline the paramount importance of the effect of reactant concentration, reaction temperature, and particularly time on the observed reaction products. Therefore, to ascertain the transferability of the results of temperatureprogrammed experiments, heating rates and reactant concentrations must be properly reported and kept constant if different experiments are to be compared. In studies on the reducibility of MoO3, high heating rates and reactant concentrations can considerably distort the onset temperature of different reduction steps as well as masking possible parallel reactions. In general, isothermal experiments should be favored over temperatureprogrammed studies to properly elucidate product composition and reaction kinetics. With respect to the application of partially reduced molybdenum oxides in heterogeneous catalysis it can be stated that, under reaction conditions of selective propene oxidation (i.e., ∼673 K), Mo4O11 should not be formed from MoO3 and, hence, should not contribute to the active state of the catalyst. In addition, from the work presented here, it appears that other structurally well-defined suboxides with extensive long-range order are not formed in substantial amounts (>1.0 wt %) under the conditions studied and, hence, should not play a major role as the active state of molybdenum oxide-based catalysts. It appears more likely that oxygen vacancies and possibly the resulting short-to-medium-range disorder (e.g., shear planes) may constitute active sites for selective propene oxidation. However, additional in situ studies under propene atmosphere must be conducted to corroborate this conclusion. Conclusion In the work presented here, the reduction of MoO3 with hydrogen (5-100 vol %) in the temperature range from 573 to 823 K was studied by in situ X-ray diffraction and absorption spectroscopy. The experiments performed focused on elucidating phase composition and evolution with time under isothermal as well as temperature-programmed reduction conditions. The results obtained clearly demonstrate the potential of a combined application of complementary in situ XRD (long-range order) and XAFS (short-range order) studies to reveal the phase composition and kinetics of solid-state reactions. It was found that, at reaction temperatures below 698 K, the reduction of MoO3 is a one-step process (MoO3 f MoO2) with MoO2 being the only detectable product (long-range and short-range ordered). No intermediates, in particular no molybdenum suboxides, or other crystalline reduction products are formed in concentrations larger than ∼1.0 wt %. At reduction temperatures above 773 K and H2 concentrations higher than 10 vol %, Mo metal is the final product of the reduction of MoO3 according to a two-step reduction process (MoO3 f MoO2 f Mo). In addition, at temperatures above 698 K, the formation of Mo4O11 was observed. Furthermore, it was found that crystalline Mo4O11 can readily be obtained from a reaction of MoO3 and MoO2 at temperatures

J. Phys. Chem. B, Vol. 104, No. 27, 2000 6369 above 773 K (40 wt % Mo4O11 after 40 h at 773 K, ∼95 wt % Mo4O11 after 5 h at 823 K). Hence, the formation temperature of Mo4O11 from MoO3 and MoO2 coincides with the formation temperature of Mo4O11 during the reduction of MoO3 in hydrogen. Thus, Mo4O11 is not an intermediate in the reduction of MoO3 but is being formed in a parallel reaction from MoO3 and MoO2 at temperatures above 773 K. No other crystallographically well-defined suboxides (e.g., Mo8O23, Mo9O26, and Mo5O14) were observed during the reduction of MoO3 under the reaction conditions investigated. Quantitative XRD analysis revealed a sigmoidal shape (Avrami-Erofeyev model) of the evolution of MoO3 and MoO2 phases during reduction of MoO3 and an increase in the crystallite size of the constituent phases. This Oswald ripening indicates that a nucleation-growth kinetic mechanism governs the reduction of MoO3 under the conditions studied. Apparent activation energies calculated for the reduction of MoO3 in 40 vol % hydrogen confirm the existence of two competing reactions, i.e., reduction of MoO3 and formation of Mo4O11, with the latter exhibiting the larger activation energy. Acknowledgment. We acknowledge the Hamburger Synchrotron Radiation Laboratory, HASYLAB, and the European Synchrotron Radiation Facility, ESRF, for providing beamtime for this work. We are grateful to L. Tro¨ger (HASYLAB) and T. Neisius (ESRF) for assistance during the experiments. T.R. thanks the Deutsche Forschungsgemeinschaft, DFG, for financial support (Habilitationsstipendium). We thank K. Hoffman for performing the specific area determination and M. Dieterle for the Raman measurements. E. Kitzelmann is acknowledged for assistance in the XRD measurements. The authors are very grateful to Prof. Dr. R. Schlo¨gl for fruitful discussions and financial support. References and Notes (1) Magneli, A. Acta Chem. Scand. 1948, 2, 501. (2) Magneli, A. Acta Chem. Scand. 1948, 2, 861. (3) Kihlborg, L. Acta Chem. Scand. 1959, 13, 954. (4) Kihlborg, L. Ark. Kemi 1963, 21, 357. (5) Arnoldy, P.; de Jonge, J. C. M.; Moulijn, J. A. J. Phys. Chem. 1985, 89, 4517. (6) Regalbuto, J. R.; Ha, J. W. Catal. Lett. 1994, 29, 189. (7) Tho¨ni, W.; Hirsch, P. B. Philos. Mag. 1976, 33, 639. (8) Gai, P. L. Philos. Mag. 1981, 43, 841. (9) Spevack, P. A.; McIntyre, N. S. J. Phys. Chem. 1992, 96, 9029. (10) Burch, R. J. Chem. Soc., Faraday Trans. 1 1978, 74, 2982. (11) Sloczynski, J.; Bobinski, W. J. Solid State Chem. 1991, 92, 420. (12) Sloczynski, J. J. Solid State Chem. 1995, 118, 84. (13) Levenspiel, O. Chemical Reaction Engineering; John Wiley & Sons: New York, 1972. (14) Grasselli, R. K. Catal. Today 1999, 49, 141. (15) Bettahar, M. M.; Costentin, G.; Savary, L.; Lavalley, J. C. Appl. Catal. A 145, 1 1996. (16) Haber, J.; Lalik, E. Catal. Today 1997, 33, 119. (17) Ressler, T.; Timpe, O.; Neisius, T.; Find, J.; Mestl, G.; Dieterle, M.; Schloegl, R. J. Catal. 2000, 191, 75. (18) Dieterle, M.; Mestl, G.; Schloegl, R., manuscript in preparation. (19) Jentoft, R. E.; Ressler, T.; Wienold, J. Catal. Lett., manuscript in preparation. (20) Bearden, J. A.; Burr, A. F. ReV. Mod. Phys. 1967, 39, 125. (21) Frahm, R. Nucl. Instrum. Methods Phys. Res. 1988, A270, 578. (22) Hagelstein, M.; San Miguel, A.; Fontaine, A.; Goulon, J. J. Phys. IV 1997, 7, C2-303. (23) Hagelstein, M.; Ferrero, C.; Hatje, U.; Ressler, T.; Metz, W. J. Synchrotron Radiat. 1995, 2, 174. (24) Ressler T. J. Phys. IV 1997, 7, C2-269. (25) Ressler, T.; Hagelstein, M.; Hatje, U.; Metz, W. J. Phys. Chem. B 1997, 101, 6680. (26) Ressler, T. J. Synchrotron Radiat. 1998, 5, 118. (27) Koningsberger, D. C., Prins, R., Eds. X-ray Absorption: Principles, Applications, Techniques of EXAFS, SEXAFS and XANES; John Wiley & Sons: New York, 1988. (28) Malinowski, E. R.; Howery, D. G. Factor Analysis in Chemistry; John Wiley & Sons: New York, 1980.

6370 J. Phys. Chem. B, Vol. 104, No. 27, 2000 (29) Ressler, T.; Wong, J.; Roos, J. EnViron. Sci. Technol. 2000, 34, 950. (30) Guenter, M. M.; Ressler, T.; Bems, B.; Schloegl, R., manuscript in preparation. (31) Gajardo, P.; Grange, P.; Delmon, B. J. Chem. Soc., Faraday Trans. 1 1980, 76, 929. (32) Kraus, W.; Nolze, G. PowderCell V2.2; Bundesanstalt fu¨r Materialforschung: Berlin, 1999. (33) Avrami, M. J. Chem. Phys. 1941, 9, 177. Erofeev, B. V. C. R. Acad. Sci. URSS 1946, 52, 511.

Ressler et al. (34) Young, R. A., Ed. The RietVeld Method; IUCr Monographs on Crystallography 5; Oxford University Press: Oxford, U.K., 1993. (35) Schmalzried, H. Solid State Kinetics; VCH-Verlag: Weinheim, Germany, 1994. (36) Sloczynski, J. React. Solids 1989, 7, 83. (37) Mestl, G.; Herzog, B.; Schloegl, R.; Knoezinger, H. Langmuir 1995, 11 (8), 3027. (38) Mestl, G.; Verbruggen, N. F. D.; Knoezinger, H. Langmuir 1995, 11 (8), 3035.