10180
J. Phys. Chem. 1994, 98, 10180-10188
Zirconia-Supported Monometallic Ru and Bimetallic Ru-Sn, Ru-Fe Catalysts: Role of Metal Support Interaction in the Hydrogenation of Cinnamaldehyde Bernard Coq; Pramod S. Kumbhar, Claude Moreau, Patrice Moreau, and Fransois Figueras Laboratoire de Chimie Organique Physique et Cinktique Chimique Appliqukes, URA 418 CNRS, ENSCM, 8 rue de 1'Ecole Normale, 34053 Montpellier Cedex 01, France Received: February 8, 1994@ The hydrogenation of cinnamaldehyde was studied over a series of Zroz-supported monometallic and bimetallic Ru catalysts. High surface area zirconia was prepared by precipitation of ZrOClz in an alkaline medium. Depending on the thermal treatments, amorphous, tetragonal or monoclinic ZrOz phases were obtained. The metal catalysts were characterized by HZchemisorption, transmission electron microscopy, X-ray diffraction, and infrared spectroscopy of adsorbed CO. When compared at very similar sizes of the metal particles, Ru/ZrOz catalysts are much more selective to cinnamyl alcohol than Ru/A1203. Moreover, among the different Ru/ZrOz samples, that supported on amorphous Z r o z exhibits the highest selectivity to the unsaturated alcohol. It is proposed that, during deposition of Ru(acac)s, the acetylacetone liberated extracts some zirconium as Zr(acac)4. During the reduction step, this Zr acetylacetonate decomposes and covers the Ru atoms, or forms a bimetallic Ru-Zr"+ type of complex. The presence of this adspecies on the Ru particles is in line with the low HZuptake of Ru/ZrOz(amorphous) with respect to the size of the Ru particles (dp FZ 1.8 nm). This kind of zirconium species is not present on Ru particles supported on crystalline Z r o z ; sizes of Ru particles deduced from TEM pictures and HZchemisorption are in good agreement. The increase of cinnamyl alcohol selectivity could be due to interfacial Ru-ZP+ sites at the periphery of the particles. On all these Ru/ZrOz catalysts the presence of mixed Ru-ZP+ sites decreases the strength of the C-0 bond, making easier the attack by hydrogen. This hypothesis is supported by a broad peak of C O adsorbed on Ru/ZrOz (tetragonal) at low frequency (uc0M 1860 cm-I).
Introduction Zirconia as a support for metal catalysts has received attention due to its unusual properties.' Metal/ZrOZ catalysts exhibit a different type of SMSI phenomenon from that observed for TiOz-supported metal as ZrOz is not an easily reducible support like TiOz. However, it has been shown for Rh/ZrOz that after high temperature reduction (HTR), there is some suppression in hydrogen chemisorption, which is characteristic of SMSI.2 This has been attributed to the partial reduction of ZrOz at the metal-support interface as shown by temperature programmed reduction (TPR). In terms of catalytic properties of rhodium, the use of Z I O z as a support was found to improve the selectivity to ethanol in the CO HZreaction as compared to Rh/AlzO3 catalyst^.^ However, it was shown that a higher reduction temperature did not further improve the activity and selectivity to alcohol. From XANES studies Yoshitake et aL4 showed that for WZrOz catalysts as the reduction temperature increases an electron transfer from ZrO2 to Pt occurs. In contrast, this W ZrOz reduced at high temperature showed only a marginal suppression in hydrogen chemisorption as compared to the Rh/ ZrO2 catalysts mentioned above.2 From STM studies Azakura et aL5 showed that Ru particles grow epitaxially on ZrOz. The same was put in evidence for Pt film vapor deposited onto a Zr02( 100)crystaL6 These observations allow one to guess that the nature of the metal-support interaction in ZrOz-supported catalysts could be different from that of classical SMSI catalysts, as both electronic effect and metal-support interfacial models have been suggested to explain the properties of these metal/ zirconia catalysts. Considering the above studies, we guessed that the ZrOzsupported metal catalysts might show interesting behavior for
+
* Author to whom correspondence should be addressed. @
Abstract published in Advance ACS Abstrucrs, August 1, 1994.
0022-365419412098-10180$04.50/0
hydrogenation of unsaturated aldehydes to the corresponding unsaturated alcohols and would help to shine some light on the type of metal-support interaction existing in these catalysts. This speculation was based on earlier works7-10 which showed that selectivity to an unsaturated alcohol is prone to electron transfer from support to the metal, as well as the environment of metal particles. In particular, Vannice et aL7 for Pt/Ti02, Yoshitake et a1.* for Pt/NbzOs, Gallezot et aL9 for PtFe/C, and Poltarzewski et al. lo for RuSdC concluded that the existence of mixed sites at the boundary between metal atoms and partially reduced Tinf, Nbn+,Fen+, and Sn*+ species is the key point to yield high selectivity to an unsaturated alcohol. In our preliminary study on Ru/ZrO:! we found this to be true." Considering this, we undertook a detailed investigation of Ru/ ZrOz catalysts, for which little information is available in the literature. The present study was aimed at the hydrogenation of cinnamaldehyde over zirconia-supported Ru, RuSn, and RuFe catalysts, a reaction well documented in the literature and of considerable commercial importance due to the use of cinnamyl alcohol in fine and perfumary chemicals.
Experimental Section Chemicals. Ruthenium acetylacetonate (Aldrich), iron acetylacetonate (Aldrich), and tetrabutyltin (Merck) were used as precursors for the catalysts in order to avoid the effect of residual ions. Cinnamaldehyde (Aldrich, 99%) and isopropyl alcohol (SDS, 99.7%) stored at 4 "C were used without further purification. Hydrogen of high purity (AGA, 99.99%) was used for the catalytic experiments and hydrogen of ultrahigh purity (AGA, 99.995%) for adsorption measurements. Carbon monoxide was further purified by molecular sieve traps. 0 1994 American Chemical Society
J. Phys. Chem., Vol. 98, No. 40, 1994 10181
Zirconia-Supported Mono- and Bimetallic Ru Catalysts
TABLE 1: Main Features for the Preparation, and some Characteristics, of the Catalysts % tetragonal catalysts support phase (m2g-l) wt % Ru wt % Sn or Fe WRu 65 35
amorphous 70 70 70 70
120 80 35 190 120 120 120 120 200
Preparation of ZrO2. High surface area zirconia was prepared according to procedures reported in detail elsewhere.l2 In short, ZrO2 was obtained by slow addition at constant pH ( ~ 1 0 . 0 5 of ) ZrOC12.8H20 (0.4 mol L-l) and aqueous N b O H (2 mol L-I) in excess water. The precipitate formed was washed with high-purity water to remove C1 ions. The washed solid was dried at 383 K for 24 h and then crushed and sieved below 125 pn. This hydroxylated zirconia has a specific surface of 290 m2 g-l. The amorphous form of zirconia was obtained by calcination at 573 K, whereas the crystalline zirconia (mainly tetragonal form) was obtained by calcination at 673 K for 3 h. The supports are referred in the text as ZrO2(A) and ZrOdC), respectively. Preparation of Monometallic Ru/ZrOz. The supported monometallic Ru catalysts were prepared by adsorption of ruthenium acetylacetonate precursor on both amorphous and crystalline ZrO2 carriers. The requisite amount of the precursor was dissolved in dry toluene and then contacted with the support for 72 h, followed by filtration. The solids thus obtained were treated under a stream of nitrogen at 423 K followed by reduction in a stream of diluted hydrogen (H2/N2:10/90) at different temperatures. More details of the preparation method are given e1~ewhere.l~ A Ru/A12O3 catalyst used for comparison was prepared by the same method reported e1~ewhere.l~ The main features for the preparation of the catalysts are given in Table 1. Preparation of Bimetallic RulZrO2 Catalysts, All the bimetallic catalysts were prepared from the prereduced Ru573/ ZrOz(C) catalyst. The RuSnlZrO2(C) samples were prepared by using the controlled surface reaction method as described in detail p r e v i o ~ s l y .In ~ ~short, the basic principle of this method involves reacting the parent Ru/ZrOz (in situ reduced before use) catalyst with the desired amount of (C4H9)4Sn in n-heptane solution under hydrogen. The surface reaction was carried out at 353 K. The RuFe/ZrOz was prepared by impregnating the prereduced Ru/ZrO;! with Fe(acac)3 in toluene. All the bimetallic catalysts were dried at room temperature under vacuum and then reduced at 573 K under pure hydrogen. The samples were stored in sealed bottles under air without any extra precaution. The metal contents were determined from the W absorbance of the precursor solutions before and after contact. For a few samples this was counterchecked by atomic absorption spectroscopy. The values determined by both of these methods were comparable. Characterization. The catalysts were characterized by X-ray diffraction (XRD), nitrogen physisorption, hydrogen chemisorption, and transmission electron microscopy (TEM). XRD patterns were recorded on a CGR Theta 60 instrument using monochromatized Cu K a l radiation. The surface areas of the solids were determined by nitrogen adsorption at 77 K (BET method) using a Micromeritics ASAP 200 instrument. The hydrogen chemisorption was carried out in a static volumetric apparatus. The hydrogen adsorption isotherm was determined at 373 K in the pressure range of 0-33 kPa. The sample was first reactivated in situ under H2 at 573 K overnight and then
1.01 1.01 1.01 1.2 1.01 1.01 1.01 1.01 1.1
0.80
0.27 0.38 0.53 0.15
0.61 0.46 0.33 0.5 1 0.45 0.45 0.52 0.66
reduction temp (K) 573 773 973 623 573 573 573 573 623
mean size from "EM (nm) 2.1 2.1 4-6 1.7 1.9 2.0 2.2 2.0
outgassed to 1.2 x Pa at the same temperature for 3 h. For supported Ru catalysts prepared from Ru(acac)3, we found that the extrapolated value at zero pressure for the isotherms determined at 373 K gave a good estimation of the monolayer hydrogen coverage of Ru ( m u , = l).15 This conclusion was reached by comparing the chemisorption with EXAFS experiments and TEM examination^.'^ The dispersion of Ru was determined from these values. For all the catalysts the results of chemisorption were checked by TEM using a JEOL lOOCX microscope. The main features of the catalysts prepared are given in Table 1. Programmed-TemperatureStudies. TPR experiments have been carried out on Ru/ZrO2. For TPR measurements a 3% HZ/Ar mixture was used. The catalyst sample was usually heated to 973 K with a heating rate of 5 K min-'. The hydrogen consumption was monitored using a thermal conductivity detector (TCD) placed after a cold trap. FTIR Spectroscopy of Adsorbed CO. The IR spectra were recorded on a Nicolet FTIR spectrometer using a Pyrex hightemperature (673 K) vacuum cell with two NaCl windows. The catalysts (X20 mg) were pressed into a self-supporting wafer and placed into the cell. The catalyst wafer was reactivated in situ at 573 K (or 623 K) in a hydrogen flow overnight and then evacuated to 1.2 x Pa at the same temperature for 2 h and cooled to room temperature. The carbon monoxide (20 Torr) was adsorbed at room temperature, and the spectra were recorded after 3 h of equilibration. The final spectrum was obtained by subtracting the background. At least 200 scans were recorded at a resolution of 2 cm-'. The spectra were also recorded after evacuation for 30 min at room temperature. Catalytic Experiments. Hydrogenation of cinnamaldehyde was studied in batch mode in a 100 mL autoclave (Autoclave Engineers) equipped with a Rushton turbine hollow stirrer, a sample port, gas inlet, and vent. The sample tube was modified by placing a filter at its end to prevent carryover of catalyst particles during sampling. The catalyst (0.26 g, unless otherwise specified) was reactivated in isopropyl alcohol (40 mL) at 383 K and 4.5 MPa hydrogen pressure for 2 h. After the autoclave was cooled to room temperature, 0.07 mol of cinnamaldehyde in 10 mL of isopropyl alcohol was injected through the charging port. The autoclave was purged with hydrogen, and the temperature was raised to 383 K rapidly. At this temperature hydrogen (4.5 MPa) was introduced and the stirring was started. Samples (0.5 mL) were withdrawn periodically and analyzed on a gas chromatograph (Varian 3300 equipped with FID detector) using a DB wax capillary column (30 m x 0.32 mm i.d.). The calibration was done by using synthetic mixtures of pure components.
Results Characterization of the Catalysts. The composition and the specific area of the ZrO2 phases, the Ru dispersion determined from HZchemisorption, and the Ru particle size determined from TEM are reported in Table 1. Figure 1 shows the XRD patterns
10182 J. Phys. Chem., Vol. 98, No. 40,1994
IM k
1I"."
30
40
50
60
70
2 Theta (degree)
Figure 1. X-Ray diffraction patterns of different catalysts: (a) Ru623/ ZrOz(A); (b) R~573/ZrOz(C); (c) Ru773/ZrOz(C);(d) Ru973/ZrOz(C).
and Figure 2 the TEM pictures of some catalysts. Two distinct behaviors appear, depending on whether ruthenium is supported on crystalline or amorphous zirconia. On crystalline zirconia, the Ru573ErO~(C)sample reduced at 573 K shows good agreement between the mean size of the Ru particles determined by TEM and the dispersion calculated from Hz chemisorption. Moreover, no Ru phase was detected in XRD in agreement with the other techniques. On the catalyst reduced at 773 K, only a small suppression in hydrogen chemisorption capacity occurred without any sintering of the Ru particles as observed from TEM and XRD. However, the decrease of H2 uptake is not as strong as for the TiOz-supported Ru catalysts, wherein a large suppression is usually observed.'' Further increase in the reduction temperature to 973 K induced a phase change of ZrO2 from tetragonal to monoclinic as well as some sintering of the Ru particles pointed out in XRD patterns (Figure Id) and TEM pictures (Figure 2b). From these facts it emerges that, for Ru catalysts supported on crystalline Z r O 2 , the occurrence of some SMSI state, as that for RdTiOZ, remains questionable. For the Ru catalysts prepared from amorphous ZrO2, the amorphous
Coq et al. nature of zirconia remained intact during the reduction step. The TEM pictures show that the particles of Ru are quite small (1.7 nm), in contrast to the chemisorption data which show a reduced value for hydrogen uptake. Something like SMSI can be postulated in this case. As reported in previous work," the size of Ru particles has a dramatic influence on cinnamyl alcohol selectivity. It was therefore essential to compare bimetallic catalysts having similar particle sizes to avoid these effects. The "controlled surface reaction" technique (CSR) that we have used to prepare the RuSn/ZrO:! samples allows us to reach this goal, as reported in our earlier paper.14 The particle size was further c o n f i e d from TEM pictures for Ru573/ZrOz(C) and Ru573Sn3/ZrOz(C) (Figure 2). This was the same when iron was added to the Ru573/ZrOz(C) parent sample. Thus the discrepancy between the hydrogen uptake (Table 1) and the TEM observations is due to partial coverage of the Ru surface of the parent catalyst (Ru573/ZrOz(C)) by the second metal and is not due to the sintering ofthe metallic phase. The state of Sn and Fe in the catalysts was not determined before or after the experiments. However, by using Mossbauer spectroscopy, Vertes et al. l6 showed that, for PtSn prepared from the CSR method, pure PtSn, phases are formed and that only a fraction of Sn transformed to SnO, after exposure to air at room temperature. Catalytic Experiments. A stirrer speed of 1550 rpm was choosen after it was confirmed that there was no external mass transfer limitation. As far as internal mass transfer limitation was concerned, Weisz's criterion was determined for the reaction carried out on the most active catalyst. This criterion is given by the expression Cp = ( 1/Co)(R2/DeR)(rate),in which COis the ) , the average reactant concentration (6 x loW4mol ~ m - ~ R particle radius (0.005 cm), and D,ff the effective diffusivity of the reactant (typically cm2 s-l for the Knudsen regime). In these conditions, the value of @ x 0.025 shows that there was no internal mass transfer limitation. Figure 3 shows as a function of reaction time a typical product distribution obtained during the hydrogenation of cinnamaldehyde (CAL) over the Ru573/ZrOz(C) catalyst. Under the experimental conditions used, the main reaction products are hydrocinnamaldehyde (HCAL), cinnamyl alcohol (COL), hydrocinnamyl alcohol (HCOL), hydrocinnamaldehyde diisopropylacetal, P-methylstyrene, and 1-phenylpropane. As reported earlier by Galvagno et uZ.,'~ the diacetal is formed by reaction between HCAL and the solvent, isopropyl alcohol, and was always in equilibrium with HCAL. Therefore, for simplification HCAL and the acetal are reported together. It is well established that the hydrogenation of a,P-unsaturated aldehydes occurs through the classical pathway shown in Scheme 1. Our results are fully in line with this network of parallel and consecutive reactions. The results of activity and selectivity for the catalysts reduced at different temperatures are listed in Table 2. For comparison the results obtained previously" on a RdA1203 catalyst with similar Ru particle sizes are also given. The selectivity to COL increased until 50-70% of CAL was converted, whatever the catalyst used. Several explanations were proposed for that b e h a ~ i o r . ~ Among .'~ them, the HCAL formed at the beginning of the reaction remains adsorbed in large amounts on the metal surface and modifies the properties of the active phase either by a ligand effect which increases the charge density on the metal, and results in decreased probability for C=C activation, or by a steric effect which forces an oncoming CAL molecule to adsorb via the C=O group at the extremity of the molecule. The influence of HCAL formed at the beginning of the CAL transformation was checked by studying the hydrogenation of
Zirconia-Supported Mono- and Bimetallic Ru Catalysts
J. Phys. Chem.. Vol. 98, No. 40. 1994 10183
20 nm -
2 0 nm
(:
>.,/
4 0 nm -
2 0 nm -
.-,
(b)
(d)
Figure 2. E M micrographs: (a) Ru573iZr02(C): (b) Ru973iZKMC); ( c ) Ru623iZrOAA); (d) Ru573Snl/ZrOi(C).
CAL on RdA1203 in the presence of HCAL (HCAUCAL = 1/10, mollmol)." In that case a 3-fold decrease of the rate occurred, and the initial selectivity to COL increased from 22 to 79%. RflrO2 catalysts are more selective than the RdA1203 catalyst. For Ru catalysts supported on crystalline ZrO2, the
reduction temperature had only a marginal effect on the catalytic properties. This type of behavior has been reported earlier for CO hydrogenation on RhErO? catalysts and is different from the usual SMSI catalysts which show an increase in activity at high reduction temperature.' In another respect, when ruthenium is supported on amorphous zirconia, the selectivity to
10184 J. Phys. Chem., Vol. 98, No. 40, 1994
Coq et al.
0 0
+. . . 0
1 1 1 . 5
10
15
A .
20
. A ,
25
30
I
I
35
40
A 45
50
Reaction time fh\ Figure 3. Hydrogenation of cinnamaldehyde over Ru573/ZrO~(C)catalyst as a function of time. TR= 383 K,C u h h y h = 7.9 x mol ~ m - ~ , Phy&,gen = 4.5 MPa, catalyst weight = 0.78 g, (0)cinnamaldehyde, (m) hydrocinnamaldehyde hydrocinnamaldehyde diisopropylacetal, (0) cinnamyl alcohol, (0)hydrocinnamyl alcohol, (A) B-methystyrene.
+
SCHEME 1
w---a+o Hydrocinnamaklehyde (HCAL)
Cinnamaklehyde (CAL)
-
i-propand
\
di-propylacetal of hydrocinnamaldehyde)
\\ Cinnemyl alcohol (COL)
b-methylstyrene
cinnamyl alcohol is the highest. No report is available in the literature on this type of catalyst for any reaction. The effect of bimetallic formulations, RuSn (Sn% = 0.27, 0.38, 0.53) and RuFe (Fe% = 0.15), on activity and selectivity
Hydrocinnamyi alcohd (HCOL)
1-Phenylpropane
to cinnamyl alcohol at different conversions of cinnamaldehyde are reported in Table 3. Adding iron had only a marginal effect on the selectivity, while Sn addition improves it clearly. The specific activity per surface Ru atom goes through a broad
J. Phys. Chem., Vol. 98, No. 40, 1994 10185
Zirconia-Supported Mono- and Bimetallic Ru Catalysts
TABLE 2: Catalytic Properties of Zirconia-SupportedRuthenium Catalysts for the Hydrogenation of Cinnamaldehyde at 383 K" selectivity to cinnamyl alcohol at different conversions of cinnamaldehyde (mol %) reaction rate catalyst WRu (mol s-l gRU-l) x 104 TOF (h-l) ob 25 50 R~573/zrOz(C) R~773/zrOz(C) Ru973/zrOz(C) Ru623/ZrOz(A) RLdA1203 a
0.80 0.61 0.46 0.33 0.66
84 135 119 239 133
1.84 2.26 1.50 2.17 2.44
61.0 61 .O 53.0 69.0 55.0
52.0 54.0 54.0 60.0 22.0
61.0 65.0 52.0 72.0 52.0
Ccmmaldehyde = 7.9 x 1 0 - ~mol ~ m - Phydrogen ~, = 4.5 MPa. Initial selectivity.
TABLE 3: Catalytic Properties of Zirconia-Supported Bimetallic RuSn and RuFe Catalysts for the Hydrogenation of Chamaldehyde at 383 Ka selectivity to cinnamyl alcohol at different conversions of cinnamaldehyde (mol %) Sn, Fe reaction rate ob 25 50 catalyst amt (wt %) WRu (mol s-l gRU-l) x 104 TOF (h-') R~573/zroz(C) Ru573Snl/ZrOz(C) Ru573SnUZr0~(C) Ru573Sn3/ZrO~(C) Ru573Fe/ZrOz(C) a
0.80 0.27 0.38
0.53 0.15
0.51 0.45 0.45 0.52
Cc",,,mal&hyde = 7.9 x 1 0 - ~mol cm+,
Phydrogen
1.84 1.90 1.64 0.94 3.46
84 136 133 76 135
52.0 65.0 61.0 66.0 52.0
61.0 72.5 62.0 72.5
55.0
61.0 74.0 65.0 75.5 55.0
= 4.5 MPa. Initial selectivity.
maximum in the same range of tin content as reported for the gas phase hydrogenation of acrolein over RSn/SiOz catalysts.l8
Discussion The classical SMSI behavior observed in the case of TiO2supported metal catalysts is correlated with the decreased hydrogen chemisorption capacity after high-temperature reduction without any artefacts of sintering of metal particle^.'^ For these types of catalysts it has now been well established that the major reason for SMSI is the blocking of the metallic surface by TiO, adspecies (x 5 2) formed after local reduction of TiOz. In the present study, as the reduction temperature increases, we do observe a slight decrease in hydrogen uptake, but it is not as high as those reported for TiO2-supported catalysts. However, the results we found are in agreement with those reported by Yoshitake et al? for Pt/ZrOz reduced at different temperatures (HPt = 0.75 at 373 K and 0.60 at 773 K). Contrary to this, Dall Agnol et aL2 have reported a large decrease in hydrogen chemisorption for W r O 2 catalysts (89 mL g-' after reduction at 573 K and 5.9 mL g-l at 873 K), similar to the case for Rh/TiOz catalysts. This means that the metal has a strong influence on the type of SMSI observed. We suggest that the decreased chemisorption capacities for ZrOz-supported metal catalysts come from surface reduction of Z r O 2 at the metalsupport interface due to spillover of hydrogen from the metal to the support. This reduction causes spreading of the reduced ZrO, on the edge atoms of the metal. The extent of migration of Zro, is dependent on the metal used. It seems that rhodium is much more active for this reduction than platinum or ruthenium. Most interesting is the case of Ru/ZrOz(A) prepared from amorphous ZrOz. This catalyst shows reduced hydrogen chemisorption capacity even at the low reduction temperature of 623 K; however, the Ru particle size is very small, as determined by TEM. Due to the difficulty for Ru metal to initiate the SMSI state with zirconia, it seems improbable that the low temperature of reduction (623 K) allows the catalysts to enter this state, even in the case of amorphous zirconia. We suggest that the abnormal behavior of this catalyst originates from the preparation steps. The amorphous ZrOz, due to the lower calcination temperatures used, has many free hydroxyl groups on the surface and, during impregnation, these groups react with acetylacetone formed by decomposition of Ru acetylacetonate. During reduction, the Zr acetylacetonate
species decompose and cover the Ru atoms or form a bimetallic Ru-ZP+ type of complex. This probably causes the decrease observed in hydrogen chemisorption. A similar kind of deposition of support species during the impregnatioddeposition step of the precursor was suggested by Haller and Resasco.20 With respect to the catalytic properties, the Ru/ZrOz catalyst shows a very interesting behavior. In terms of initial selectivity to COL, the catalyst behaves similar to that of RwTiOz with the same Ru particle size." Several explanations were suggested to explain this behaviour: (1) A bimetallic RuZr phase could be formed during the reduction of Ru/ZrOz, as shown by Szymanski et aLzl in the case of WZrOz. Electronic or geometric modifications of ruthenium would result from the appearance of such a phase. This could produce an effect similar to that in the bimetallic catalysts.' However, the electronic modification of platinum by the ZrO2 carrier is controversial. On the one hand, from XANES studies, Yoshitake et d4concluded that the density of the unoccupied d-state of platinum decreased when the reduction temperature of WZro2 increased to 993 K; i.e. zirconia donates electrons to platinum. On the other hand, from kinetic studies of competitive hydrogenation of toluene and benzene on W W z , Szymanski et ~ 1 speculated . ~ ~ that an electron transfer occurred from platinum to zirconium upon formation of the Rl-,Zr, alloy phase at elevated reduction temperature. (2) The interaction between Ru and ZrOz can induce an epitaxial growth of the metal particles, modifying their morphology from a round to a flat shape, with the flat shape being characteristic of the low index planes of large particles, which are selective to COL f ~ r m a t i o n .Such ~ an epitaxial growth was shown for Ru particles supported on ZQ5 and for Pt films vapor-deposited onto a Zroz(100) crystal.6 (3) Specific sites occur at the periphery of the Ru particles. From TPR studies on Rh/ZrO2, it has been suggested that ZrO2 will be partially reduced, probably at the boundary with Rh particles.2*22These Ru-Zro, mixed sites will reproduce, in part, the behavior of RwTi02. To throw more light on the behavior of Ru/ZrOz, TPR experiments and FTIR spectroscopy of adsorbed CO were done. Figure 4 shows the TPR profiles for ZrOz(C)and Ru573fZr02(C) catalysts. It was found that ZrOz(C) alone is almost not reduced below 973 K. By contrast, two small hydrogen consumption peaks occur around 750 and 900 K on Ru/Zroz-
10186 J. Phys. Chem., Vol. 98, No. 40, 1994
273
373
Coq et al.
473
573
6 73
temperature
773
973
673
(K)
Figure 4. TPR profiles of (a) ZrOz(C) and (b) Ru573/ZrO~(C).
2200
2100
2000
1900
1600
1700
Wavenumber (cm-' )
Wavenumber (cm" )
Figure 5. IR spectra at room temperature of CO adsorbed on (a) RdAl~03and (b) Ru573/ZrO2(C): (-) after 3 h of equilibration; (---) after being outgassed for 20 min at room temperature.
(C). These peaks could be ascribed to a partial reduction of ZrO2 (Z#+ Zr3+), assisted by the presence of Ru, or to the formation of a bimetallic phase Ru(l-,)Zr, (Z#+ Zfl). The formation of this bimetallic phase could be responsible, in part, for the decrease in H2 chemisorption for Ru973/ZrOz(C) (W Ru = 0.33). Szymanski et dZ1 reported the formation of a pt(l-,)Zr, phase upon reduction of WZrOz at 873 K (x = 0.05) and 1023 K (x = 0.14). In another respect, under strongly reducing conditions at 773 K, the formation of Zr3+ ions at the periphery of Rh particles in Rh.lZrO2 was also proposed by Guglielmin~tti.~~ The extent of this surface reduction is low and certainly limited to the vicinity of the Ru particles. In other respects, the temperature of the Ru reduction peak ( ~ 3 9 K) 2 is similar to that found for Ru/A1203 prepared according to the same procedure^.'^ Very likely, the interaction is not much stronger between ruthenium and zirconia than with alumina.
-
-
From the FTIR spectroscopy of adsorbed CO on these catalysts, one can see that the spectra for RdZrO2 are different from those for RdAlzO3 (Figure 5 ) . The IR spectra of CO adsorbed on reduced RdAl2O3 exhibit the main bands usually reported in the l i t e r a t ~ r e . ~According ~-~~ to the recent studies of Yokomizo et aZ.F6 the high-frequency (HF) band at 2132 cm-' corresponds to (C0)3Rus+ isolated species on the support. These species come from some corrosive chemisorption of Ruo atoms, probably located at low coordination sites such as edges and comers. In the present case the amount of these species is very low due to the low acidity of the surface hydroxyl groups and the precursor used to prepare the catalyst. Another vibration band associated with these species does appear around 2080 cm-'; it was seen as a shoulder at about 2100 cm-l and was difficult to extract from the spectra. However, the main component of the spectra is the broad band centered around
J. Phys. Chem., Vol. 98, No. 40, 1994 10687
Zirconia-Supported Mono- and Bimetallic Ru Catalysts
CHART 1
2040 cm-' which corresponds to CO in on-top positions. There is another small component around 2000 cm-l which appears as a shoulder. This feature is difficult to assign but could correspond to linear species different from those vibrating at 2040 cm-', maybe on partially oxidized Ru atoms.27 Finally, a very small contribution appears at 1780 cm-'. This feature is not well documented but could be assigned to bridged species, such a species taking place generally on low-index planes of Pd, Pt, and Rh catalysts. By contrast, the IR spectra of CO adsorbed on Ru/ZrO2 are very different. The narrow band at 2189 cm-' which vanishes upon outgassing is assigned to weakly bonded CO on Z P , in agreement with literature data.23,28The contribution of (C0)3Rud+ species appears only as a small unresolved peak at about 2130 cm-'. The main difference with the spectra on RdA1203 occurs in the 21001700 cm-' region, where a very broad band appears. In the 2100- 1900 cm-' range, the absorption band corresponds certainly to linear CO species adsorbed on sites of different topologies. But the intriguing point is the feature between 1900 and 1700 cm-', which could be assigned to two different species: (i) first, to multibridged species on low-index planes. However, on large Ru particles supported on alumina, mainly populated by low-index planes, there was no evidence for an enhancement of the "bridged species".29 (ii) second, to bridged CO species bonded at the borderline of Ruo particles through the C atom but also bridge-bonded to the coordinatively unsaturated Zf'+ or Zr3+ ions through the 0 atom, as shown in Chart 1. Such a species was proposed to give a broad CO absorption band occurring at 1660 cm-' on Rh/Z1a2.~~ The low C-0 stretching frequency of this adsorbed CO species is an evidence of a weaker CO bond. The position of this band is shifted to higher frequency on RdZrO2, and the assignment of this type of CO complex is not unequivocal. Wagray et aL30 reported a similar broad band around 1900-2000 cm-' in a study of CO adsorption on a RuK+/Si02 catalyst. They concluded that a strong interaction occurred between CO and Ki. Finally, due to the broad component at around 2040 cm-I, it is difficult to put in evidence some shift to lower or higher frequency of linearly adsorbed CO on RdZr02 compared to RdAl203. As a matter of fact, it is impossible to determine from the IR study if an electronic modification of the Ru occurs when supported on zirconia. We have seen from chemisorption data that Ru/ZrOz(C) catalysts enter with difficulty in the SMSI state. Therefore, a Ru/Zr02(C) catalyst efficient and selective for cinnamyl alcohol needs to exhibit small Ru particles in order to develop a larger amount of these interfacial sites. In terms of catalytic behavior, the type of surface complex shown in Chart 2 is expected to give high selectivity to carbonyl group hydrogenations as described by Vannice and Sen.' For clarification, one can consider adsorption of cinnamaldehyde via the carbonyl group attached to the Zf'+ sites at the periphery of the metal particles. This affects the polarization of the carbonyl group and improves the selectivity. This is schematically depicted in Chart 2. At reduction temperatures higher than 773 K, the presence of Ru-Zr3+ interfacial sites probably exists, since a small fraction of the ZrO2(C) surface is reduced, as shown by TPR profiles (Figure 4). In the present case we found that the selectivity to cinnamyl alcohol obtained on Ru573/ZrO2(C) can also be improved by addition of Sn. In contrast, the Ru-Fe
CHART 2
/ catalyst did not show such an improvement in selectivity. The question is, what is so special about Sn? In the literature, the following explanations have been put forward to explain this b e h a ~ i o r : ' ~ ~ " . (1) ' ~ . ~formation ~ of metal-tin sites which selectively activate the carbonyl group by bonding through the oxygen end of the carbonyl group; (2) transfer of electrons from tin to the active metal which increases the electron density at the metal atoms, resulting in decreased electron transfer to the C=C double bond. This results in a decreased probability of C=C hydrogenation; (3) tin occupies the sites of lower coordination, which are thought to be less selective to COL formation. In the case of PtSdSiOz catalysts, Marinelli et aZ.18 argued against the electron transfer model whereas Verbeek and S a ~ h t l econcluded r~~ that a change occured in the d-band density upon tin addition to platinum. In another respect, Wagray et al.30have speculated that potassium occupies the sites of lower coordination in RuIUSi02 catalysts. In quantum chemical studies of and R u S model ~ ~ ~ bimetallic clusters, it was shown that tin is preferentially located at the sites of low coordination. Moreover, from the calculated atomic net charges for RugSn clusters,34it appears that tin remains almost neutral; no "electron transfer" occurs between Ru and Sn. Therefore, an electronic modification of the Ru d-band by tin remains questionable on model catalysts. We suggest that the increase in COL selectivity upon the first addition of tin comes either from formation of Ru-Sns+ sites (but why does iron not have the same influence?) or, better, by the selective poisoning of Ru sites of low coordination, less selective for COL format i ~ n . " , ~A~similar explanation was proposed by Wagray et aL30to interpret the higher selectivity to an unsaturated alcohol during the hydrogenation of 3-methyl-2-butenal over RuK/SiO2 catalysts. In other respects, an additional effect of dilution of the Ru surface cannot be excluded at high tin loading. The isolation by tin of the Ru sites would hinder the adsorption of CAL as a flat species, thus making the C=C double bond less able to be activated."
Acknowledgment. This work was supported by a financial grant from the Indo-French Centre for Promotion of Advanced Research (IFC/306-1). The help of R. Dutartre in TEM experiments and L. C. de Menorval in FTIR experiments is gratefully acknowledged. The authors thank V. Ponec for helpful discussions. References and Notes (1) Mercera, P.D. L.; van Ommen, J. G.; Doesburg, E. B. M.; Burgraaf, A. J.; Ross, J.R.H. Appl. Catal. 1990, 57, 127. (2) Dall Agnol, C: D.; Gervasini, A.; Morazzoni, F.; Pinna, F.; Strukul, G. J. Catal. 1985, 96, 106.
(3) Cerioni, G.; Martinengo, S.; Zanderegh, L.; Tonelli, G.; Iannibello, A.; Girelli, A. J. Chem SOC. Faraday Trans. I 1984, 80, 1605. (4) Yoshitake, H.; Iwazawa, Y. J. Phys. Chem. 1992, 96, 1329. ( 5 ) Asakura, K.;Iwazawa, Y.; Pumell, S.; Watson, B.; Barteau, M.; Gates, B. C. Catal. Lett. 1992, 15, 317. (6) Roberts, S.; Gorte, R. J. J. Phys. Chem. 1991, 95, 5600.
10188 J. Phys. Chem., Vol. 98, No. 40, 1994 (7) Vannice, M.; Sen, B. J. Catal. 1989,115,65. Vannice, M. A. Catal. Today 1992, 15, 255. (8) Yoshitake, H.; Iwazawa, Y. J. Cafal. 1990, 125, 227. (9) Gallezot, P.; Giroir-Fendler, A.; Richard, D. In Catalysis of Organic Reactions; Pascoe, W., Ed.; M. Dekker Inc.: New York, 1991; p 1. (10) Poltarzewski, Z.; Galvagno, S.;Staiti, P.; Pietropaolo, R. J. Catal. 1986, 102, 190. (11) Coq, B.; Kumbhar, P. S.; Moreau, P.; Moreau, C.; Warawdekar, M. J. Mol. Catal. 1993, 85, 215. (12) El Alami, D. Ph.D. Thesis, Montpellier, France, 1993. (13) Coq, B.; Figubras, F.; Geneste, P.; Moreau, P.; Moreau, C.; Warawdekar, M. J. Mol. Catal. 1993, 78, 211. (14) Coq, B.; Bittar, A,; Dutartre, R.; Figubras, F. J. Cafal. 1991, 128, 275. (151 Coa. B.: Crabb, E.: Warawdekar, M.; Bond, G. C.; Slaa, J. C.; Galvagno, i.;Mercadante, L.; Garcia Ruiz, J.; Sanchez Sierra, C. J. Mol. Catal. 1994, 92, 107. (16) Vertes, C.; Talas, E.; Czako-Nagy, I.; Ryczkowski, J.; Gobolos, S.; Vertes, A., Margitfalvi, J. Appl. Catal. 1991, 68, 149. (17) Galvagno, S.; Capannelli, G.; Neri, G.; Donato, A.; Pietropaolo, R. J. Mol. Catal. 1991, 64, 237. (18) Marinelli, T. B. L. W.; Vleeming, J. H.; Ponec, V. In Proceedings of the 10th International Congress on Catalysis; Guczi, L., Solymosi,F., Tbtbnyi, P., Eds.; Akademiai Kiado: Budapest, 1993; Vol. C, p 1211. (19) Tauster, S. Acc. Chem. Res. 1987, 20, 389.
Coq et al. (20) Haller, G. L.; Resasco, D. E. Advances in Catalysis; Academic Press: London and New York, 1989; Vol. 36, p 173. (21) Szymanski, R.; Charcosset, H.; Gallezot, P.; Massardier, P.; Tournayan, L. J. Catal. 1986, 97, 366. (22) Joswiak, W. K. React. Kinet. Catal. Lett. 1986, 30, 345. (23) Guglielminotti, E. J. Catal. 1989, 120, 287. (24) Dalla Betta, R. A. J. Phys. Chem. 1975, 79, 2519. Brown, M. F.; Gonzalez, R. D. J. Phys. Chem. 1976, 80, 1731. Robbins, J. L. J. Catal. 1989, 115, 120. Solymosi, F.; Rasko, J. J. Cafal. 1989, 115, 107. (25) Kellner, C. S.; Bell, A. T. J. Catal. 1981, 71, 296. (26) Yokomizo, G. H.; Louis, C.; Bell, A. T. J. Catal. 1989, 120, 1. (27) Gupta, N. M.; Kamble, V. S.; Iyer, R. M.; Ravindranathan Thampi, K.: Gratzel. M. J. Catal. 1992, 137, 473. (28) Bensitel, M.; Saur, 0.;Lavalley, J. C.; Mabilon, G. Mater. Chem. Phys. 1987, 17, 249. (29) de Menorval, L. C.; Coq, B.; Crabb, E.; Figukras, F. In preparation. (30) Wagray, A.; Wang, J.; Oukaci, R.; Blackmond, D. J. Phys. Chem. 1992, 96, 5964. (31) Galvagno, S.; Donato, A.; Neri, G.; Pietropaolo, R. Catal. Lett. 1991, 8, 9. (321 Verbeek. H.; Sachtler, W. M. H. J. Catal. 1977, 42, 257. (33) Coq, B.; Goursot, A,; Tazi, T.; Figubras, F.; Salahub, D. R. J. Am. Chem. SOC.1991, 113, 1485. (34) Goursot, A.; Pedocchi, L.; Coq, B. J. Phys. Chem., accepted for publication.