In Situ Study of the Formation of Crystalline Bismuth Molybdate

146

Chem. Mater. 2003, 15, 146-153

In Situ Study of the Formation of Crystalline Bismuth Molybdate Materials under Hydrothermal Conditions Andrew M. Beale and Gopinathan Sankar* Davy Faraday Research Laboratory, The Royal Institution of Great Britain, 21 Albemarle Street, London, W1S 4BS United Kingdom Received July 29, 2002. Revised Manuscript Received October 8, 2002

The three well-known and catalytically important phases of bismuth molybdate have been prepared, in a phase pure form, employing hydrothermal methods at temperatures below 200 °C. An in situ study of the formation of these phases using the time-resolved energydispersive X-ray diffraction (EDXRD) technique revealed that the R and γ crystalline phases are formed directly from the amorphous gel mixture without forming any intermediate phases. Kinetic data analysis using the Avrami-Erofe’ev nucleation-growth model suggests that the mechanisms for formation are different for each phase. This hydrothermal method of preparation yielded materials with a higher surface area as compared to the solids prepared by conventional techniques.

Introduction Bismuth molybdates, having the general chemical formula Bi2O3‚nMoO3, where n ) 3, 2, 1, and their metal-ion-doped analogues, have been the subject of considerable interest due to their catalytic application in the area of selective oxidation/oxidative dehydrogenation or ammoxidation of lower olefins. Many patents, research papers, and reviews can be found in the literature detailing the preparation, characterization, and catalytic studies of these materials.1-19 Among the various compositions, R-Bi2Mo3O12, β-Bi2Mo2O9, and γ-Bi2MoO6 have been found to be the most catalytically active phases. The conventional route for preparing these catalysts involves the use of a precipitation method conducted at specific pH values and heated at temperatures in the range of ca. 400-700 °C to obtain (1) Idol, J. D. U.S. Patent, 2,904,580, 1959. (2) McClellan, W. R.; Stiles, A. B. U.S. Patent, 3,678,139, 1972. (3) Grasselli; Hardman, H. F. U.S. Patent, 3,642,930, 1972. (4) Grasselli, R. K.; Suresh, D. D.; Friedrich, M. S. U.S. Patent, 4,424,141, 1984. (5) Sasaki, Y.; Mori, K.; Moriya, K. E.P. Patent, 0,383,598, 1990. (6) Sasaki, Y.; Mori, K.; Moriya, K. E.P. Patent 0,389,255, 1990. (7) Kope, J.; Kripylo, P.; Hohlstamm, I.; Hoepfner, R.; Knaack, K. E.; Mai, H. G. DE Patent, 4,124,666, 1993. (8) Izumi, J.; Watanabe, S.; Yoshioka, H. E.P. Patent, 0,799,642, 1997. (9) Schirmann, J. P.; Descat, G.; Etienne, E.; Pham, C.; Simon, M. FR Patent, 278,251,2, 2000. (10) Batist, P. H. A.; Bouwens, J. F. H.; Schuit, G. C. A. J. Catal. 1972, 25, 1-11. (11) Trifiro, F.; Hoser, H.; Scarle, R. D. J. Catal. 1972, 25, 12-24. (12) Burban, P. M.; Schuit, G. C. A.; Koch, T. A.; Bischoff, K. B. J. Catal. 1990, 126, 326-338. (13) Buttrey, D. J.; Jefferson, D. A.; Thomas, J. M. Philos. Mag. A 1986, 53, 897-906. (14) Wachs, I. E.; Hardcastle, F. D.; Buttrey, D.; Jefferson, D. A.; Thomas, J. M. Abstr. Pap. Am. Chem. Soc. 1987, 194, 328-INOR. (15) Hardcastle, F. D.; Wachs, I. E. J. Phys. Chem. 1991, 95, 1076310772. (16) Grasselli, R. K.; Burrington, J. D. Adv. Catal. 1981, 30, 133163. (17) Grasselli, R. K. Catal. Today 1999, 49, 141-153. (18) Snyder, T. P.; Hill, C. G. Catal. Rev.-Sci. Eng. 1989, 31, 4395. (19) Buttrey, D. J. Top. Catal. 2001, 15, 235-239.

the crystalline catalytic material; this method is preferred over the heating of a mixture of solid oxides. One of the main drawbacks of this approach is that some of these oxides, for example γ-Bi2MoO6, undergo a further transformation to the γ′-Bi2MoO6 phase.20-22 Below 600°C this transformation is a very slow process, but once this temperature is exceeded, the change becomes more rapid.21,23 To overcome some of the difficulties encountered in the preparation of crystalline bismuth molybdate phases, we employed the use of hydrothermal methods to produce phase-pure, crystalline binary oxide phases, under mild conditions. Hydrothermal methods are traditionally used to obtain meta-stable zeotype materials, which include aluminosilicates, aluminophosphates, mesoporous silicas, and their metal-ion-substituted variants.24,25 More recently, this type of soft chemistry has been explored for the preparation of several advanced oxide materials.26,27 This synthetic approach appears to be a highly attractive option considering that it is easily reproducible, is good for obtaining a homogeneous mix, and allows for easy addition of other dopants. A particular advantage of the use of hydrothermal over precipitation methods is that because the operating temperatures are generally below 200 °C, compounds that may decompose at elevated temperatures can now be utilized. (20) Buttrey, D. J.; Vogt, T.; White, B. D. J. Solid State Chem. 2000, 155, 206-215. (21) Sankar, G.; Roberts, M. A.; Thomas, J. M.; Kulkarni, G. U.; Rangavittal, N.; Rao, C. N. R. J. Solid State Chem. 1995, 119, 210215. (22) Kodama, H.; Watanabe, A. J. Solid State Chem. 1985, 56, 225229. (23) Egashira, M.; Matsuo, K.; Kagawa, S.; Seiyama, T. J. Catal. 1979, 58, 409-418. (24) Feng, S. H.; Xu, R. R. Acc. Chem. Res. 2001, 34, 239-247. (25) Cheetham, A. K.; Fe´rey, G.; Loiseau, T. Angew. Chem. Int. Ed. 1999, 38, 3269-3292. (26) Whittingham, M. S.; Guo, J. D.; Chen, R. J.; Chirayil, T.; Janauer, G.; Zavalij, P. Solid State Ion. 1995, 75, 257-268. (27) Whittingham, M. S. Curr. Opin. Solid State Mater. Sci. 1996, 1, 227-232.

10.1021/cm020463z CCC: $25.00 © 2003 American Chemical Society Published on Web 11/21/2002

Formation of Crystalline Bismuth Molybdate Materials

Chem. Mater., Vol. 15, No. 1, 2003 147

Table 1. Synthesis Conditions for the Hydrothermal Preparation of r-Bi2Mo3O12, β-Bi2Mo2O9, and γ-Bi2MoO6a amount of Bi2O3 (g/mol)

amount of (NH4)6Mo7O24‚4H2O (g/mol)

amount of HNO3 (g/mol)

amount of NH4OH (mL/mol)

1.16 (0.0025) 1.16 (0.0025) 1.16 (0.0025)

1.32 (0.0011) 0.88 (0.00071) 0.44 (0.00036)

4.08 (0.045) 4.08 (0.045) 4.08 (0.045)

5 (0.125) 3.75 (0.094) 2.5 (0.063)

final gel pH

preparation temp (°C)

preparation time (h)

6.0

140

20

γ

phaseb

a The number of moles of each ingredient used in the reaction are given in parentheses. b Note, product yields were determined to be ca. 90% (R phase), 85% (β phase), and 70% (γ phase).

Here, we report the preparation of R, β, and γ phases of binary bismuth molybdate phases employing mild hydrothermal conditions and we followed the formation of the R and γ phases from an amorphous gel to the crystalline solid using time-resolved energy-dispersive X-ray diffraction (EDXRD). This in situ technique allowed us to monitor and understand the crystallization process associated with these materials, to optimize the conditions for the production of these solids, and to determine the nature of any intermediate phases formed during the synthesis process. The unique advantage of using this technique over conventional diffraction methods is that it utilizes high-intensity polychromatic radiation from a synchrotron, which can penetrate a stainless steel hydrothermal cell without loosing too much intensity. Thus, by careful selection of both the size and thickness of the cell,28 it was possible to follow the crystallization of solids, under reaction conditions, with a time resolution on the order of 1-2 min, therefore enabling a much more detailed and accurate kinetic analysis than that which is obtainable by ex situ quenching methods. We and others have successfully used this EDXRD technique in the study of the formation of microporous materials and other systems.29-33 Here, the in situ EDXRD measurement carried out during the hydrothermal treatment of the synthesis gels containing bismuth nitrate and ammonium heptamolybdate showed that phase-pure crystalline materials are formed within a few hours in the temperature range of 100-150 °C. For the β phase it was necessary to apply further heat treatment to the solid extracted from the hydrothermally treated gel, at ca. 560 °C, to form the final phase. Formation of the pure β phase was followed by in situ combined X-ray diffraction (XRD) and extended X-ray absorption spectroscopy (EXAFS, also referred to as QuEXAFS when a rapid scanning method is employed), an ideal technique to follow the phase formation and transformation in solids.34,35 However, (28) Evans, J. S. O.; Francis, R. J.; Ohare, D.; Price, S. J.; Clark, S. M.; Flaherty, J.; Gordon, J.; Nield, A.; Tang, C. C. Rev. Sci. Instrum. 1995, 66, 2442-2445. (29) Davies, A. T.; Sankar, G.; Catlow, C. R. A.; Clark, S. M. J. Phys. Chem. B 1997, 101, 10115-10120. (30) Rey, F.; Sankar, G.; Thomas, J. M.; Barrett, P. A.; Lewis, D. W.; Catlow, C. R. A.; Clark, S. M.; Greaves, G. N. Chem. Mater. 1995, 7, 1435-1436. (31) Walton, R. I.; Millange, F.; O’Hare, D.; Davies, A. T.; Sankar, G.; Catlow, C. R. A. J. Phys. Chem. B 2001, 105, 83-90. (32) Loh, J. S. C.; Fogg, A. M.; Watling, H. R.; Parkinson, G. M.; O’Hare, D. PCCP Phys. Chem. Chem. Phys. 2000, 2, 3597-3604. (33) Millange, F.; Walton, R. I.; O’Hare, D. J. Mater. Chem. 2000, 10, 1713-1720. (34) Reilly, L. M.; Sankar, G.; Catlow, C. R. A. J. Solid State Chem. 1999, 148, 178-185. (35) Sankar, G.; Wright, P. A.; Natarajan, S.; Thomas, J. M.; Greaves, G. N.; Dent, A. J.; Dobson, B. R.; Ramsdale, C. A.; Jones, R. H. J. Phys. Chem. 1993, 97, 9550-9554.

perhaps the most important observation is that the surface area measurements suggest that the solids prepared by hydrothermal methods possess a much higher surface area as compared to the ones prepared by either sol-gel or solid-state methods. Experimental Section For the hydrothermal preparation of the bismuth molybdate phases we adopted the following procedure. Stoichiometric amounts of acidified bismuth nitrate solution were mixed with ammonium heptamolybdate dissolved in ammonium hydroxide. The pH of this mixture was adjusted to a specific range before introducing this mother liquor into a Teflon-lined autoclave. The autoclave was placed in a preheated oven for several hours. The details of the chemical composition, pH range, temperature, and time of the reaction are given in Table 1. After the reaction for a specific period the contents were washed, dried, and characterized by XRD. XRD patterns of all the as-synthesised materials (dried at 100 °C) were recorded using a Siemens D500 diffractometer equipped with a copper target. Scanning electron micrograph (SEM) pictures were taken using a JEOL 733 Superprobe with an Oxford Instruments ISIS/INCA system for EDX analysis. The BET surface area measurements were carried out on a Micromeritics Gemini III. All samples were subjected to degassing at 100 °C in a flow of nitrogen overnight prior to the surface area measurements taken using nitrogen as the adsorbate at -196 °C. Measurements were carried out on both the as-synthesized materials and calcined at 400 °C (a typical temperature used for catalytic reaction).1 In the case of the β phase, it was necessary to calcine at ca. 560 °C since the phasepure material was formed only after further heat treatment of the hydrothermally treated sample. In situ EDXRD measurements were carried out on station 16.4 of Daresbury Synchrotron radiation source, which operates at 2 GeV with a typical current of 150-250 mA. The details of the experimental setup are given elsewhere.36,37 In a typical experiment, an autoclave containing the appropriate gel mixture was introduced into an oven, preheated to a specific temperature. EDXRD measurements were carried out at temperatures between 110 and 150 °C. Data collection was started 2 min after the introduction of the autoclave and a data collection time of 1 min/scan was employed for this work. The detector was placed at a fixed 2θ angle of 4.619°. Integrated areas of all of the reflections detected in the EDXRD spectra were determined using a Gaussian curve fitting routine using the XFIT program. In situ XRD/QuEXAFS measurements were carried out at station 9.3 of Daresbury Synchrotron radiation source. The station was equipped with a Si(220) double-crystal monochromator and ion chambers for measuring incident and transmitted beam intensities for recording X-ray absorption spectra. For diffraction measurements, a position-sensitive INEL (36) Barnes, P.; Jupe, A. C.; Colston, S. L.; Jacques, S. D.; Grant, A.; Rathbone, T.; Miller, M.; Clark, S. M.; Cernik, R. J. Nucl. Instrum. Methods Phys. Res., Sect. B 1998, 134, 310-313. (37) Muncaster, G.; Davies, A. T.; Sankar, G.; Catlow, C. R. A.; Thomas, J. M.; Colston, S. L.; Barnes, P.; Walton, R. I.; O’Hare, D. PCCP Phys. Chem. Chem. Phys. 2000, 2, 3523-3527.

148

Chem. Mater., Vol. 15, No. 1, 2003

Beale and Sankar

Figure 2. Scanning electron micrograph images (magnified 20 000×) of the bismuth molybdate solids (a) Bi2MoO6, γ phase, and (b) Bi2Mo3O12, R phase, prepared by hydrothermal methods at 140 °C. Table 2. Surface Area Measurements for the As-Prepared Phase Pure Bismuth Molybdate Phases and after Calcinations at 400 °C

Figure 1. X-ray diffraction patterns of (a) Bi2Mo3O12, R phase, (b) Bi2MoO6, γ phase, and (c) Bi2Mo2O9, β phase, recorded using a conventional Cu KR radiation source, are shown along with the lines representing the patterns reported in the JCPDS database, card nos. 21-103, 21-102, and 33-209, respectively. All three materials were prepared by employing hydrothermal methods at 140 °C for 20 h. For the β-phase formation it was necessary to heat the material further to ca. 560 °C. detector was used. About 40 mg of the β-phase bismuth molybdate precursor was pressed into a 13-mm pellet and mounted into the in situ cell that permits the measurement of combined XAS and XRD data. The sample was heated at 5 °C minute from room temperature to 560 °C and held at this temperature for 60 min. Mo K-edge XAS and XRD data were collected sequentially during this activation process. The time taken for the Mo K-edge XAS pattern was 380 and 180 s for XRD data, resulting in a total cycle time of 10 min, which includes 40 s of dead time to move the monochromator back to the starting point. XRD data were collected at a wavelength of 0.928 Å, well below the Mo K-edge absorption to avoid fluorescence effects. The INEL detector was calibrated using a NBS silicon standard and a 10-µm Mo foil was used to calibrate the monochromator position. XAS data were processed using the suite of programs available at Daresbury Laboratory, namely, EXCALIB (for converting the raw data to energy vs absorption coefficient) and EXBROOK (to obtain normalized XANES part of the spectra and for background subtraction to extract EXAFS).

Results and Discussion First, we discuss the results obtained based on the laboratory-based ex situ methods and subsequently we show the results obtained from in situ time-resolved studies. Ex Situ Studies. The diffraction patterns of the asprepared material using hydrothermal methods are shown in Figure 1. It is clear from the XRD patterns that the solids prepared using gel compositions that

phase

calcination temp (°C)

R R γ γ β

as-prepared 400 as-prepared 400 560

calcination time (h) 4 4 8

surface area (m2/g) 8.9 8.2 10.1 9.8 2.9

correspond to R and γ phases can be readily identified as the crystalline R-Bi2Mo3O12 and γ-Bi2MoO6 (JCPDS Card Nos. 21-103 (a) and 21-102 (b)) phases, respectively. The gel prepared with β-phase (Bi2Mo2O9) stoichiometry did not yield a phase-pure material. The XRD pattern of this sample appears to have reflections that correspond to the γ-type phase in addition to an amorphous phase. However, the pure form of the β phase can be prepared from this hydrothermally treated material by calcining in air at temperatures between 500 and 560 °C. The XRD pattern of this calcined material is also shown in Figure 1c along with the index lines for β phase (JCPDS Card No. 33-209). All of the as-prepared materials show a much larger surface area than that which has been previously reported for materials prepared by the conventional sol-gel methods, even after calcination at a typical reaction temperature of 400 °C, employed for propylene ammoxidation. The surface areas for the three materials prepared by hydrothermal methods are given in Table 2. In Figure 2 we show the SEM pictures for the R and γ phases prepared by hydrothermal methods at 140 °C. It is clear that the particle size and shape are different for both phases with the R phase consisting of irregularly shaped particles which appear slightly larger in size than the platelet-type particles found for the γ phase. In Situ EDXRD Studies. To obtain details on the kinetics of the crystallization process and to determine whether the R and γ phases are produced via the formation of any other intermediate phases or by-phase transformation processes during the hydrothermal synthesis, we carried out a detailed in situ study, at various reaction temperatures. The stacked EDXRD patterns (typical data recorded at 140 °C is shown in Figure 3) recorded during the hydrothermal treatment clearly showed that, during the initial stages, two very broad

Formation of Crystalline Bismuth Molybdate Materials

Figure 3. Typical stacked EDXRD patterns recorded in situ during hydrothermal synthesis, at 140 °C, of (a) γ phase and (b) R phase. The time zero corresponds to the start of the data collection, which was 2 min after introduction of the autoclave into a preheated oven. A data collection time of 1 min was employed for this work. For clarity, only data obtained from the middle detector (the detector was placed at a fixed 2θ angle of 4.619°) is shown over a short range of d spacing.

but distinct reflections appear in the gel with γ stoichiometry (Figure 3a) and four broad reflections appear in the gel with R stoichiometry (Figure 3b), which over a period of time become narrow. These reflections have been identified as the -221, 023, 040, and -204 peaks for the gel with R stoichiometry and the 131 and 200/ 002 reflections for the γ gel. The other reflections that appeared at lower d spacings are not shown here for both phases due to the limited d spacing range of the detector and their lack of intensity. For the subsequent kinetic analysis of this gel-to-solid reaction, the growth of the integrated peak intensity of a particular Bragg reflection Ihkl with time (t) (eq 1) is used to represent the extent of reaction (Rhkl). Similar

Rhkl(t) )

Ihkl(t) Ihkl(t∞)

(1)

analysis procedures have been used for several systems (see, for example, refs 29, 32, and 42-45). In Figure 4 (38) (39) (40) (41)

Avrami, M. J. Chem. Phys. 1939, 7, 1103-1112. Avrami, M. J. Chem. Phys. 1940, 8, 212-224. Avrami, M. J. Chem. Phys. 1941, 9, 177-184. Erofe’ev, B. V. C. R. Dokl. Acad. Sci. URSS 1946, 52, 511-517.

Chem. Mater., Vol. 15, No. 1, 2003 149

Figure 4. Normalized (with respect to the last data set collected during measurement at a specific temperature) extent of reaction (R) curves for the most intense reflections for the bismuth molybdates (top) γ phase and (bottom) R phase versus time. We have plotted the 131 and 200/002 reflections for the γ phase and the -221, 023, 040, and -204 peaks for the R phase.

the extent of reaction (R), for the previously identified reflections (-221, 023, 040, and -204 peaks for the gel with R stoichiometry and the 131 and 200/002 reflections for the γ gel) for both phases seen in the EDXRD data, are plotted as a function of time. It appears that the growth curves obtained from various reflections are superimposable (within experimental error) and it is concluded that uniform crystal growth had taken place with no obvious favored crystallographic direction. The sigmoidal curves seen here are typical for a crystallization process consisting of an initial period of induction/ nucleation followed by rapid crystallization of the phase and gradual growth of the crystallites until a constant value is reached, reflecting a tendency toward completion of crystal growth. Due to the uniformity of the growth process, for the subsequent kinetic analysis the most intense peaks (-221 for R phase and 131 for γ phase) were used. The extent of reaction curves for both R and γ phases recorded at various temperatures are compared in Figure 5. It is clear from Figure 5 that an increase in synthesis temperature results in a reduction in both the rate and onset of crystallization for both phases. It is interesting to note that although the (42) O’Hare, D.; Evans, J. S. O.; Fogg, A.; O’Brien, S. Polyhedron 2000, 19, 297-305. (43) Evans, J. S. O.; Price, S. J.; Wong, H. V.; O’Hare, D. J. Am. Chem. Soc. 1998, 120, 10837-10846. (44) Wilkinson, A. P.; Speck, J. S.; Cheetham, A. K.; Natarajan, S.; Thomas, J. M. Chem. Mater. 1994, 6, 750-754. (45) Fogg, A. M.; Price, S. J.; Francis, R. J.; O’Brien, S.; O’Hare, D. J. Mater. Chem. 2000, 10, 2355-2357.

150

Chem. Mater., Vol. 15, No. 1, 2003

Figure 5. Plots of the area under the reflection (131) for the γ phase (top) and (-221) for the R phase (bottom), as a function of reaction time. The various curves in this figure represent the data recorded at different temperatures. Note that the time zero corresponds to the start of the data collection, 2 min after introducing the autoclave into the preheated oven held at a specific temperature.

induction period for R-phase growth is longer, the rate at which the crystallization took place was faster compared to that of the γ phase. The different nature of the growth curves seen for the two phases suggests that the crystallization processes are likely to be different. Furthermore, the growth profiles for each phase showed little variation with temperature, which we conclude is evidence that the growth mechanism is not temperature-dependent for the range of temperatures studied. In Figure 6 we plot the full-width at half-maximum (fwhm) of the same reflections shown in Figure 5 to determine the variation in particle size with temperature. The fwhm values for both systems decreases very sharply with the rate at which the decrease takes place dependent upon the reaction temperature. The interesting feature of these plots is that the fwhm decreases up to a point, which coincides with the change in rate seen in Figure 5. Beyond this point the fwhm remains more or less constant (within the experimental error), while a further increase in the extent of reaction (R) is observed. Due to such variation in the R vs time plot and fwhm with time, and to extract meaningful kinetic information of the initial stages of crystallization, we have analyzed only the first part of the crystallization curve with an upper limit of R close to ca. 0.6 (see Figure 5). Using the well-known Avrami-Erofe’ev equation,38-41 which is widely applied to model solid crystallization

Beale and Sankar

Figure 6. Plots of the fwhm of the reflection (131) for the γ phase (top) and (-221) for the R phase (bottom), as a function of reaction time. The various curves in this figure represent the data recorded at different temperatures. Note that the time zero corresponds to the start of the data collection, 2 min after introducing the autoclave into the preheated oven held at a specific temperature.

and phase transformations (see eq 2), we obtained the rate of growth of R and γ phases:

R ) 1 - exp(-kt)

n

(2)

where k is a rate constant and n related to the mechanism for both nucleation and growth. Common values used for n range from n ) 0.5 to n ) 4 and is known to contain information on both the dimensionality and the process of crystallization. Previous work documents the successful application of this equation to model a variety of solid-state reactions,42-44 but it was generally found to be most appropriate over a data range of 0.15 < R < 0.6.45 To extract both n and k from our data, it is necessary to perform a Sharp-Hancock analysis using eq 3:46

ln[-ln(1 - R)] ) n ln(t) + n ln k

(3)

The resultant Sharp-Hancock plots for both phases for the three different temperatures are shown in Figure 7 and the kinetic data and Avrami exponent (n) derived from this plot are given in Table 3. The n values (close to 1.0 for the γ phase and between 3 and 4 for the R phase) for both phases are also in agreement with the observations made in Figure 5, which suggests that the growth mechanism for the two phases is different and (46) Hancock, J. H.; Sharp, J. D. J. Am. Ceram. Soc. 1972, 55, 74-77.

Formation of Crystalline Bismuth Molybdate Materials

Chem. Mater., Vol. 15, No. 1, 2003 151

Figure 8. Mo K-edge XANES of a starting gel mixture of Bi2MoO6, crystalline Bi2MoO6, and crystalline Bi2Mo3O12 are compared with model compounds containing tetrahedrally coordinated Mo(VI) (Na2MoO4) and octahedrally coordinated Mo(VI) (MoO3) compounds. The pre-edge feature “A” dominates the spectra in the materials possessing four-coordinated molybdenum whereas feature “B” is more prominent for sixcoordinated species.

Figure 7. Sharp-Hancock plots for the data recorded at various temperatures for the γ phase (top) and the R phase (bottom) over a data range of 0.15 < R < 0.6. Table 3. Avrami Exponent (n) and Rate of Crystallization (k) As Derived from Sharp-Hancock Analysis in Figure 7 phase

temp (K)

n

k (m-1)

R

393 403 413 383 413 423

3.13 3.80 3.83 1.06 1.19 0.97

0.0339 0.0603 0.0693 0.0270 0.0471 0.0634

γ

that is not temperature-dependent over the range of temperatures studied. For the R phase the exponent values are higher than those determined for the γ phase. According to the analysis of Hulbert, the difference in the exponent values corresponds to very different growth mechanisms.47 For the γ phase the derived values being close to 1.0 are thought to be indicative of a system experiencing growth in two dimensions which is diffusion-controlled and has a decreasing rate of nucleation. This proposed mechanism appears reasonable given that the γ phase is often described as consisting of Bi2O22+ layers in an Aurivillius48-type structure and, furthermore, supported by the presence of platelets in the SEM picture for the sample synthesized at 140 °C (see Figure 2a). For the R phase the values between 3 and 4 are thought to be representative of three-dimensional growth, which is proposed to be via a phase-boundary-controlled mechanism with a decreasing nucleation rate.47 Although this is not apparent from the SEM picture of (47) Hulbert, S. F. J. Br. Ceram. Soc. 1969, 6, 11-19. (48) Aurivillius, B. Ark. Kemi. 1951, 2, 519-527.

the R phase (see Figure 2b) (as it is proposed that threedimensional growth should result in spherical particles), the particles appear to be irregularly shaped. It is interesting to note that the similar fwhm values at completion for both γ and R phases appear to indicate that, within the temperature range of the investigation, changing the temperature has very little apparent affect on the final crystallites’ size. On the basis of these findings for both γ and R phases, it is possible to arrest the reaction at any given time to produce smaller particle sizes with a larger surface area. Mo K-Edge X-ray Absorption Spectroscopy. Although it appears from the EDXRD that a poorly crystalline phase of the final product is formed at the initial stages of the reaction, Mo K-edge X-ray absorption spectroscopy reveals that the coordination environment of the molybdenum is completely different for the starting gel as compared to the final phase. Typical Mo K-edge X-ray absorption near-edge structure (XANES) of the starting gel and the final product for Bi2MoO6 are compared with Mo(VI)-containing model compounds in Figure 8. It is well-known in the compounds containing transition-metal ions that a very distinctive pre-edge feature appears in the XANES spectrum, which is due to the forbidden 1s-nd (n ) 3 or 4) transition (by dipole selection rule) but becomes allowed due to the mixing with p orbitals and systems without inversion symmetry.49,50 Thus, the pre-edge intensity for a system in the same oxidation state will be higher when the absorbing atom (here Mo(VI) ions) is present in a tetrahedral environment compared to an octahedral one.51 Although the quantitative analysis of the pre-edge peak is not completely accomplished, this feature has (49) Waychunas, G. A. Am. Miner. 1987, 72, 89-101. (50) Wong, J.; Lytle, F. W.; Messmer, R. P.; Maylotte, D. H. Phys. Rev. B 1984, 30, 5596-5610.

152

Chem. Mater., Vol. 15, No. 1, 2003

Beale and Sankar

Figure 9. (a) Stacked XRD plot obtained using the position-sensitive INEL detector, during the course of heating the solid obtained from hydrothermal synthesis (appropriate gel mixture containing chemical composition related to the β phase was first hydrothermally treated at 140 °C for 20 h), from room temperature to 560 °C at a rate of 5 °C/min. Results are shown from 200 °C onward as below this temperature the material condenses on losing water, resulting in a distortion in the data. XRD patterns were collected at 0.928 Å. In (b) we show the stacked Mo K-edge XANES data (for clarity we show only the small region, although the entire Mo K-edge XAS pattern was recorded within 380 s), recorded sequentially with the XRD pattern shown in (a). In (c) we show the variation in the intensity of the features A and B with temperature. The main change was seen to occur at temperatures above 500 °C, which is clearly due to the loss of the distorted six-coordinate geometry resulting from the presence of γ phase and the formation of β phase with molybdenum in four coordination, consistent with the XRD data.

been found to be sensitive to coordination geometry and metal-oxygen distances.49,50,52 Similarly, another feature on the rising absorption edge is also found to be different for different coordination (first neighbor) geometries.53 The pre-edge peak and the feature on the rising absorption edge are marked as A and B, respectively, in Figure 8. In general, the pre-edge intensity (A) is known to have the lowest intensity and feature B the highest for octahedral coordination geometry and the opposite for tetrahedral coordination. In the absence of a complete theoretical description, models have been developed based on crystallographically well-characterized systems and they have been successfully used for many solid-state systems.52,54 Here, we employed a simple comparison of the features A and B in the Mo K-edge XANES spectra of various bismuth molybdate (51) Corma, A.; Rey, F.; Thomas, J. M.; Sankar, G.; Greaves, G. N.; Cervilla, A.; Llopis, E.; Ribeira, A. Chem. Commun. 1996, 16131614. (52) Farges, F.; Brown, G. E.; Rehr, J. J. Geochim. Cosmochim. Acta 1996, 60, 3023-3038. (53) Antonio, M. R.; Teller, R. G.; Sandstrom, D. R.; Mehicic, M.; Brazdil, J. F. J. Phys. Chem. 1988, 92, 2939-2944. (54) Brown, G. E.; Calas, G.; Waychunas, G. A.; Petiau, J. Rev. Miner. 1988, 18, 432-512.

systems to extract qualitatively the information about the coordination geometry around molybdenum ions. It is clear from the features A and B (see Figure 8), related to the pre-edge and edge,34 that tetrahedrally coordinated Mo(VI) ions are present in the initial stages of the reaction. The presence of a smaller pre-edge intensity for the final, hydrothermally prepared crystalline γ-Bi2MoO6 is consistent with the crystallographically determined structure of Mo(VI) ions in a distorted octahedral coordination. Similarly, the pre-edge and edge features of the crystalline R-Bi2Mo3O12 catalyst are also consistent with the presence of five-coordinated molybdenum species in this material. Thus, it has been clearly shown (in both cases) that there is no evidence for the formation of any other crystalline phase during the course of this hydrothermal synthesis. In Situ Combined QuEXAFS/XRD Study. To follow the formation of β phase from the solid obtained after hydrothermal treatment, a combined QuEXAFS/ XRD study was undertaken. The facile formation of β-Bi2Mo2O9 during heating of the solid obtained from the hydrothermally treated beta gel is shown in Figure 9. Both the XRD patterns and the X-ray absorption spectra can be seen to change during the calcination

Formation of Crystalline Bismuth Molybdate Materials

process. In Figure 9a the XRD pattern changes from the initial mixed phase material (amorphous and γ phase) to the β-Bi2Mo2O9 phase above 500 °C. In Figure 9b,c the stacked Mo K-edge XANES data recorded during the formation of β phase is shown along with the variation in the two commonly used signatures in the XANES data (marked A and B) employed in the interpretation of the Mo K-edge XANES data. The intensities of both the A and B features are seen to increase and decrease, respectively, suggesting the coordination change from predominantly octahedral geometry of the starting gel (due to the presence of γ phase) to the tetrahedrally coordinated Mo(VI). In summary, we have been able to produce, employing hydrothermal methods, over a short reaction period, crystalline Bi2MonO3n+1 (where n ) 1, 2, 3) phases with higher surface areas as compared to that synthesized by conventional preparative methods. Our in situ timeresolved EDXRD studies clearly suggest that there are no intermediate phases involved in the formation of highly crystalline R- and γ-phase bismuth molybdate solids. Furthermore, we have shown that it is conceivable to tailor the material’s properties by controlling the synthesis conditions, namely, temperature, time, and stoichiometry. The crystallization data were modeled using Avrami-Erofe’ev kinetics and we conclude that the growth mechanisms for the two phases are different. For all three of the temperatures studied for the R phase, it is thought that the phase undergoes threedimensional growth. For the γ phase, initially, twodimensional growth was observed. In both cases, the overall rate of reaction was seen to increase with temperature. Although it was not possible to prepare β phase directly using hydrothermal methods, employing the composition and conditions described here, we have

Chem. Mater., Vol. 15, No. 1, 2003 153

been able to produce a phase-pure material from a hydrothermally produced precursor material by heat treatment at ca. 500 °C. Although similar hydrothermal methods for the formation of the γ phase have previously been reported in the literature,55,56 these have normally been undertaken over a much longer time period, at higher temperatures or in metal halide solutions. Our present study offers a more convenient and reproducible route, without the use of any metal halide solutions, for the formation of this industrially important material. This hydrothermal approach has also successfully been applied to the preparation of another catalytically important material, iron phosphate,57 and thus it has been demonstrated that this synthesis procedure appears to be superior to conventional solid-state routes for producing phase-pure catalytic systems at moderate temperatures. Acknowledgment. The authors thank EPSRC for general financial support, Synchrotron radiation beam time, and a quota award for A.M.B. We also thank Daresbury Laboratory for facilities and the Leverhulme Trust for a senior Research fellowship for G.S. We also thank Dr. Eva Valsami-Jones for use of the BET facility, Drs. J. F. W. Mosselmans and I. Harvey for their help with XAS measurements. Professors C. Richard A. Catlow, G. David Price, Sir John Meurig Thomas, and Dr. Lee M. Reilly are acknowledged for useful discussions. CM020463Z (55) Kodama, H.; Izumi, F. J. Cryst. Growth 1980, 50, 515-520. (56) Shi, Y. H.; Feng, S. H.; Cao, C. S. Mater. Lett. 2000, 44, 215218. (57) Beale, A. M.; Sankar, G. J. Mater. Chem. 2002, 12, 3064-3072.