G. Centi and F. Trifiro' (Editors),New Developments in Selective Oxidation 0 1990 Elsevier Science Publishers B.V., Amsterdam - Printed in The Netherlands
SELECTIVE CATALYSTS J.M.
LOPEZ NIETO, R. BIELSA*, G. KREMENIC'l and J.L.G. FIERRO
Instituto de Catalisis y Petroleoquimica, C.S.I.C., Serrano 119, 2 8 0 0 6 Madrid (Spain) *On leave from INTEC-CONICET, 3000 Santa Fe (Argentina) ABSTRACT Catalytic activity for the selective oxidation of propylene over Mo-RE-0 (RE=Pr,Sm,Tb,Yb) catalyst series, with Mo/(Mo+RE) atomic ratios ranging between 0 and 1, has been studied. For all catalyst series, both activity and selectivity to partial oxidation products exhibited a relative maximum in the Mo-rich compositions region. These data are interpreted in terms of surface and bulk characteristic of the catalysts as revealed by X-ray diffraction, temperature-programmed reduction, laser-Raman and X-ray photoelectron spectroscopic methods. INTRODUCTION Molybdenum-based catalysts are commonly used in many industrial processes which involve selective oxidation of olefins [I]. Rare earth (RE) oxides catalyse a great variety of reactions and promote the partial oxidation of light hydrocarbons [ 2 ] . With the only exception of Ce-containing catalysts , the role of rare earth oxide on the selective oxidation process is not well understood [ 4 ] . Recent studies carried out in our laboratory [ S - 8 1 revealed that catalytic behaviour markedly depends on the composition and type of phases present in the catalysts. This work is part of a broad study to investigate the effects of the rare earth promoters on the structure and reactivity of Mo-based catalysts. For this purpose, the information revealed by several bulk and surface sensitive techniques is compared with activity and selectivity of the binary Mo-RE-0 preparations. EXPERIMENTAL The catalysts were prepared by mixing ammonium heptamolybdate and/or RE nitrate solutions of selected
concentration and volume to obtain fixed Mo/ (Mo+RE) ratios. The solutions were evaporated until dryness and then the remaining solids calcined in a forced flow of air at 823 K for 14 h [ 5 ] . Catalyst testing. Details of the experimental technique used for catalytic activity experiments have been given elsewhere [5-81. In short, 1 . 0 g-samples (particle size between 0.42 and 0.59 mm) were mixed with Sic (in a volume ratio, catalyst:SiC= 1:4). The molar ratio of the components in the reactant mixture was C3H6 : O 2 :He:H20 = 20:30:30:20 and the contact time W/F= 30-90 g.h (no1 C3H6). '- Experiments were carried out over the temperature range 623-723 K, at atmospheric pressure. The efluents of the reactor were analyzed by gas chromatography. Catalyst characterization. Specific surface areas of catalysts were calculated by the BET method from the Kr adsorption isotherms at 77 K. X-ray diffraction (XRD) patterns were obtained using a Phillips PW-1100 diffractometer using Ni-filtered CuKa radiation ( A = 0.15406 nm). Raman spectra (LRS) were recorded using a Jarrell-Ash 25-300 spectrometer equipped with halographic gratings. x-ray photoelectron spectra (XPS) were acquired with a Leybold Heraeus LHSlO electron spectrometer equipped with a magnesium anode (MgK, = 1253.6 eV) and a hemispherical electron analyzer. The binding energies were referenced to the Cls line at 284.6 eV. Details of all these techniques are given elsewhere [581. TPR experiments were made in a Cahn microbalance.
1 Mo/(Mo+RE) Figure 1. Reaction rate for C3H6 at 673 K over Mo-RE-0 (RE= Pr, Tb, Sm, Yb) catalyst series. Contact time W/F= 30 g.h.mo1-l.
+ U W
Mo -T b
1.0 Mo/(Mo + RE1
Figure 2. Selectivity to acrolein ( 0 ) and acetaldehyde + acetic acid ( A ) at 673 K for a propylene conversion = 5 mole %.
Samples (0.2-0.3 mg) were first heated to 773 K in helium flow (7.2 dm3 h-l) , and the cooled to ambient temperature. After this, they were contacted with hydrogen (7.2 dm3 h") and heated at a rate of 240 K h-l to a final temperature of 793 K. This temperature was maintained about 0.5 h. RESULTS AND DISCUSSION The selective oxidation of propylene over Mo-RE-0 (RE= Pr, Sm, Tb, Yb) catalysts has been shown to depend strongly on the catalyst composition. As Fig.1 shows, all four catalyst series display a clear maximun for the rate of propylene conversion in the Mo-rich composition region. However, the compositions (expressed by the Mo/(Mo+RE) atomic ratios) at which the maximum appears, varies according to sequence Mo-Yb-0(0.89), Mo-Pr-0(0.89), Mo-Tb-0 (0.70) and Mo-Sm-0 (0.60) With the only exception of the Mo-Pr-0 catalyst series, a further decrease of the Mo/(Mo+RE) ratios, e.g. increasing the fraction of RE oxide added, induces a marked decrease of the specific catalytic activity. Beside that, from the data of Fig.1 the activity sequence for the pure RE oxides ( (Mo/(Mo+RE)= 0.0) follows the order, Pr6011 > Tb4O7 >
Sm203 > Yb203, which agrees with the one reported by Minachev et al. 191 for the same reaction. Selectivity values to acrolein and acetyl (acetic acid + acetaldehyde) (Fig. 2) also show a similar maximun to the one found on the activity profiles in the Mo-rich composition region (Mo/ (Mo+RE) between 0.60 and 0.89), while carbon oxides are almost the unique C-containing molecules. As already shown by the XRD patterns, formation of quite disimilar crystalline phases occurs as catalyst composition is varied (Table 1). In agreement with literature findings [ i O , i i ] , the Mo-rich composition range exhibits the Moog phase as the major crystalline entity, in parallel with small amounts of stoichiometric molybdates, and probably some type of tetra- and hexamolybdates [ i l l , whose abundance decreases for the less Morich preparations. One important point to be considered is that catalysts with maximun in activity profiles are those having the largest proportion of molybdates among the overall crystalline phases. Of course, the Mo-Pr-0 is the exception as no crystalline phases were detected along all compositions range. TABLE 1 Crystalline Phases as Identified from X-Ray Difrattion Patterns. Mo
PrsOll Pr6011 Pr6011 Mo03(e) Mo03(e) Moo3
Tb407 Sm203( a ) TbsMoO12(c) Sm203(b) Sm2M020g(el Tb2 (Moo4) ( f l Tb2Mo4Ol5(g) ns Moo3 (h) Moo3 (hl Moo3 (k) Moo3 (k)
Yb Yb203 (a) Yb203(dl Yb2 (Moo4) (e) ns Yb2Mo4ol5(h) Moog (h)
a= cubic; ns= not studied. Minor phases: b= 9Sm203.4Mo03: c= Tb4O7; d=Yb2(Mo04)3; e= RE20g: f=Tb2Mo209; g= Moog; h’RE203.4Mo03 (likely): k= RE203.6Mo03 (likely). Laser Raman spectra of the Mo/(Mo+RE)= 0 . 8 (RE= Pr, Sm, Tb, Yb) catalyst samples were also recorded to monitor the presence of molybdate structures. A s shown in Fig. 3 , all spectra show the bands at 998 and 820 cm-l characteristic of Mo=O stretch and antisymmetric Mo-0-Mo stretching, respectively in Moo3 isolate
Figure 3 . Laser Raman Spectra of Mo-RE-0 catalysts (atomic ratio Moj(MO+RE)= 0 . 8 ) : a) Mo-Yb-0: b) Mo-Tb-0; C) Mo-SHI-0; a) MoPr-0 catalysts.
800 700 Ag (cm-11
phase. Other bands in the region 800-960 cm'l, very intense for Mo-Yb-0, moderately intense for Mo-Tb-0 and very low for Mo-Sm-0 catalysts have been assigned, in agreement with XRD patterns, to that vibrations in RE molybdates 1121 as its intensity increased with decreasing Mo-loading. The exception is Mo-Pr-0 catalyst in which small bands in the same region seem to be due to polymolybdates in a separate phase 161. To obtain an estimate of the metal-oxygen strength as well as to explain activity and selectivity changes as a function of catalyst composition. TPR profiles were obtained for all preparations. Table 2 summarizes the reduction degree of catalysts obtained at 793 K. One important point to be considered is the strong dependence of TPR profiles upon catalyst composition. For example in the RE-rich preparations, mostly Mo-Pr-0 [S] and Mo-Tb-0 catalysts series, the reduction degree is larger than in Mo-rich preparations, and also the kinetics of reduction decreases continuously with time indicating that this process takes place
Mo Mo+RE 0.00 0.20
0) Mo-Pr-0 1.04
0.70 0.40 0.13
Mo-Tb-0 1.20 0.70 0.75
0.68 0.75 0.31
1.27 1.27 1.28 0.91
(a) Calculated by the ratio between the experimental weight loss and the theoretical one espected for the quantitative reduction of Moog to MOO? ( a = l ) .(b) Reducible oxides such as Pr6O11 and Tb4O7 present in the catalysts were considered to be reduced to Pr203 and Tb2O3, respectively.
according to the contracting sphere model. However, Mo-rich catalysts begin to reduce at higher temperatures and present S-shaped TPR profiles, i.e., they reduce according to a nucleation model. Photoelectron spectroscopy (XPS) has also been used from a quantitative point of view to reveal the surface composition of catalysts. The dependence between the Mo/(Mo+RE) XPS ratios and those corresponding to the chemical analysis are given in Fig.4. As can be observed, for the Mo-RE-0 (RE= Pr, Sm, Yb) catalyst series there is, in general, a good correlation between surface XPS and chemical compositions, while for Mo-Tb-0 series an important RE surface enrichment is clearly observed throughout the explored compositions. In this latter case a Tb molybdate-phase a few layers thick seems to be formed over Moo3 nuclei, as also suggested by the well resolved LRS spectra of Tb-molybdates (Fig.3). When comparing activity and selectivity data for oxidation of propylene with those of catalyst characterization it results that partial oxidation products are more likely to occur on catalysts with lattice oxygen of a lower reactivity, viz., more difficult to be reduced. Moro-oka et al.[lS] found the more active oxides for total oxidation of hydrocarbons to be those with lower heat of formation of the oxide ( A H M - O ) . Pr6011 and Tb407 have low A%-o values and an important part of unstable lattice oxygen of a high mobility, thus explaining their tendency to form deep oxidation products when present as separate phases in RE rich Mo-RE-0 (RE= Pr, Tb) preparations (Figs.1 and 2 ) . A s already
5 0.f 0
1 R E)chem
Figure 4 . Dependence between the surface XPS and chemical Mo/ (Mo+RE) atomic ratios: RE= Pr (V);Sm ( 0 ) ; Tb(0); Yb ( A ) . In this calculation, the integrated Mo3d and RE4d intensities and published sensitivity factors [ 1 4 ] were considered. shown by TPR, AHM-o tends to be larger for catalyst which are more difficult to reduce. The reduction degree ( a ) at 793 K in the region Mo/ (Mo+RE)= 0.7-0.8 is the lowest but simultaneously selectivity to partial oxidation products is the highest (Fig.2). A similar correlation among catalyst reduction and conversion and selectivity were found by Sachtler and de Boer [lS] in the propylene oxidation over metallic molybdate catalysts. These results are closely related to those reported by Trifiro' et al. [l?], who found that the most selective catalysts (within a series of molybdates) for the same reaction are those exhibiting the lowest diffusion rate of lattice oxygen. Oxygen may be removed by diffusion of lattice oxygen to the interface reduced phase in all the ternary catalyst systems employed in this study. Thus, the diffusion rate of oxygen ions will be lower and the selectivity will be higher for catalysts with lower reducibility, as it effectively occurs. The fact that maximun selectivity to partial oxidation products occurs for Mo/(Mo+RE) ratios in the region 0.7-0.8, where XRD patterns and LRS spectra
revealed excess of Moo3 and several kinds of molybdates, indicates that nucleophilic oxygen species, which then would lead to allylic oxidation, are optimized. AKNOWLEDGEMENTS The authors are indebted to CSIC and CAICYT for sponsorship of this work (Project No. 120). REFERENCES a) R.K. Grasselli, J.D. Burrington, A d v . C a t a l . , 111
r21 131 141
b) C.F. Cullis, D.J. Hucknall, in G. Bond & G. Webb (Eds.), ggCatalysisgl, Vol. 5, Specialist Periodical Reports The Chemical Society, London, (1982) ch. 7, p. 273. a) M.P. Rosynek, C a t a l . R e v . - B c i . Eng., 16 (1977) 111. b) P. Pomonis, R e a c t . Kinet. C a t a l . R e v . , 18 (1981) 247. a) J.C.J. Bart, N. Giordano, J. C a t a l . , 75 (1982) 134. b) J.F. Brazdil, R.K. Graselli, J. C a t a l . , 79 (1983) 1 0 4 . a) J.J. Kim, S.W. Weller, A p p l . C a t a l . , 33 (1987) 15. b) V.M. Khiteeva, Sh.M. Rzakulieva, RUBS. J. Phys. C h e m . , 55 (1981) 1202.
J.M. Lopez Nieto, J.L.G. Fierro, L. Gonzalez Tejuca, G. Kremenic', J. C a t a l . , l 0 7 (1987) 325. J.M. Lopez Nieto, G-Kremenic', A. Martinez Alonso, J.M.D. Tascbn, J. Mater. S c i . , (in press). G. Kremenic',J.M. Lopez Nieto, J. Soria, J. Marti, Proc. Inter. C o n f . R a r e E a r t h D e V . L A p p l . , Beijing, China, September 1985, Vol. 1, p. 614. G. Kremenic', J.M. Lopez Nieto, J.L.G. Fierro, L.G. Tejuca, J. L e s s - C o m m o n Met., 136 (1987) 95. K.M. Minachev, D.A. Kontratev, G.N. Antoshin, K i n e t .
a) K. Nassau, J.W. Shiever, E.T. Keve, J. S o l i d State
8 (1967) 131.
3 (1971) 411.
b) L.H. Brixner, P.E. Biersted, A.W. Sleight, M.S. Lisic,
Mat. Res. B u l l . ,
6 (1971) 545.
a) F.P. Alekseev, E.I. Get'man, G.G. Koshchoev, M.V. Mokhosoev, R u s s . J. Inorg. C h e m . , 14 (1969) 1558. b) E. Ya Rode, G.V. Lysanova, L.Z. Gokhman, Inorg. Mater., 7 (1971) 1875.
H. Jeziorowski, H. Knozinger, J. Phyo. Chem.,
J.M. Lopez Nieto, A.G. Valdenebro, J.L.G. Fierro, in preparation. C.D. Wagner, L.E. Davis, M.V. Zeller, J.A. Taylor, R.H. Raymond, L.H. Gale, Surf. Interface A n a l . , 3 (1981) 211. Y. Moro-oka, Y. Morikawa, A. Ozaki, J. c a t a l . , 7 (1967)
t 151 I161
W.M.H. Sachtler and N.H. de Boer, Proc. 3rd. I n t . C o n g r . C a t a l . , Amsterdam, 1964 (W.M.H. Sachtler, G.C.A. Schuit and P. Zwietering, Eds), Wiley, New York, 1965, vol.1, p.252. F. Trifiro', P. Centola, I. Pasquon and P. Jiru, P r o c . 4 t h . I n t . C O n g r . C a t a l . , MOSCOW, 1968 (B.A. Kazansky, Ed.), Adler, New York, 1968. Vol.1, p.252.
J.C. VEDRINE (I. de Recherche sur la Catalyse, Villeurbanne, France): I was surprised that you concluded that selective molybdate catal st exhibit lower diffusion rate of lattice oxygen. Using l20 labelled C02 as a probe we have observed that lattice 0 of bismuth molybdates ( a or B phases, kown to be very selective in propene oxidation to acrolein) are exceptionally labile involving both surface and bulk lattice oxygen. How did you determine the lattice oxygen lability of your samples? J.M. M P E Z NIETO (I. Catdlisis y Petroleoquimica, Madrid, Spain): The term diffusion rate of oxygen in the rare earth molybdates refers here to the relative ease with which oxygen can be released from the catalyst. We found that the catalyst whose Mo/(Mo+RE) ratio is 0 and 1 are poorly selective to partial oxidation products, viz. carbon oxides and water were the major oxidation products.To explain this behavior, it was assumed that catalysts with these extreme compositions have highly reactive oxygen species, such as oxygen adsorbed. On the contrary, in the region of intermediate Mo/ (Ho+RE) ratios , where molybdates were found to ocour, the bulk lattice oxygen seems to be involved in the selective oxidation of adsorbed hydrocarbon. The mobility of the latter oxygen species must be high as confirmed by the observation that the surface prereduction of the different molybdates at temperatures close to 600 K is faster than the subsequent oxygen adsorption on the partially reduced surface. This particular behaviour has been explained as due to partial restoration of the original surface, upon surface reduction, by diffusion of bulk lattice oxygen to the surface which then adsorbs oxygen slowly until initial state recovery. J.C. VEDRINE (Ins. de Recherche sur la Catalyse, Villeurbanne, France): You also found high selectivity in acetic acid and acetaldehyde which was interpreted as electrophilic attack of propene rather than allylic. The last is giving acrolein. In a recent paper by us on MoO3/SiO2 (ref.1) much allylic attack was detected at low Mo coverage but yielded propanal. Did not you observed any propanal in your products? Acetic acid results from a more complex reaction mechanism with C-C cleavage as for acetaldehyde. J.M. M P E Z NIETO (I. Cathlisis y Petroleoquimica, Madrid, Spain): For the Moo3 and MoOj/Si02 systems, Vedrine et al. (ref.1) found high selectivity toward propanal at conversions levels below 1%. For the MoOj/Si02 catalysts studied early in our laboratory, we did not detect propanal at conversion levels as high as 15-20% (ref.2). In this study working at conversion levels around 10% on Mo-RE-0 systems, no propanal was detected in any case. Only acrolein, acetic acid and acetaldehyde were observed. Acetaldehyde is mainly a primary product (from propene degradation), but it also forms by decomposition of acrolein (ref.3). However, acetic acid is formed by oxidation of C2- and c3-oxigenated products. R. LARSSON (I. Inorganic Chemistry, Lund, Sweden): determined the activation energies of these -action?
J.M. M P E Z NIETO (I. Catalisis y Petroleoquimida, Madrid, Spain): Temperature coefficients for propene oxidation on the various rare earth molybdates have been calculated. They have been not summarized for practical reasons. In general, the values obtained
summarized for practical reasons. In general, the values obtained do not vary significantly along the explored compositions with the exception of the Mo/(Mo+RE) ratios with maxima in activity and selectivity which led to values substantially higher. To illustrate this, the temperature coefficients for the Mo-Pr-0 catalyst series were 106-119 kJ/mole for compositions Mo/ (Mo+RE) < 0.88, while a value of 143 kJ/mole was obtained for the most active no/ (Mo+RE) = 0.91 catalyst.
R. LARSSON (I. Inorganic Chemistry, Lund, Sweden): Have you IR spectra of the catalysts? J.M. LOPEZ NIETO (I. Catdlisis y Petroleoquimica, Madrid, Spain): Exploratory experiments using IR technique revealed the appearance of several M-0-M lattice vibrations, however the unambiguous assignment of that bands to specific compounds was not straighforward. Very recently, an in-depth analysis of these molybdate series was carried out by Laser Raman Spectroscopy in our laboratory. This study will constitute the next step of the research of the bulk and surface properties of the rare earth molybdates.
Liu, M. Forissier, G. Coudurier, J. C. Vedrine, J. Chem. Faraday Trans. 1, 85 (1989) 1607. 2) J. M. M p e z Nieto, G. Kremenic', A. Martinez-Alonso, J. M. D. Tascbn, J. Mater. S c i . , 24 (1989) (in press). 3) J. M. M p e z Nieto, J. M. D. Tascbn, G. Kremenic', Bull. Chem. SOC. Jpn., 61 (1988) 1383. 1) T.