Influence of the precursor and the support on the catalytic properties of ruthenium for alkane hydrogenolysis

Influence of the precursor and the support on the catalytic properties of ruthenium for alkane hydrogenolysis

Applied Catalysis, 60 (1990) 33 33-46 Elsevier Science Publishers B.V., Amsterdam - Printed in The Netherlands Influence of the Precursor and th...

1MB Sizes 0 Downloads 3 Views

Applied

Catalysis, 60 (1990)

33

33-46

Elsevier Science Publishers B.V., Amsterdam -

Printed in The Netherlands

Influence of the Precursor and the Support on the Catalytic Properties of Ruthenium for Alkane Hydrogenolysis B. COQ*, A. BITTAR, R. DUTARTRE and F. FIGUERAS Laboratoire de Chimie Organique Physique et Cinktique Chimique Appliqubes - U.A. 418 du C.N.R.S. E.N.S.C.M. 8, Rue de I’Ecole Normale - 34053 Montpellier CCdex 2 (France)

(Received 1 July 1989, revised manuscript received 14 November 1989)

ABSTRACT The influence of the ruthenium compound used as precursor to prepare ruthenium catalysts, and of the carrier, have been investigated using the hydrogenolysis of n-hexane, 2-methylpentane and 2,2,3,3_tetramethylbutane as model reactions. Similar activities and selectivities are reproduced for catalysts obtained from Ru(NH,),Cl, or Ru(acac), and a highly crystalline alumina. When RuCl,(NH,), is used as precursor, the catalytic properties are shifted towards those of large particles. At similar dispersions, the catalysts obtained by deposition of Ru(acac), on AlaOa,MgO or TiOa (reduced at 573 K) present similar properties. By high-temperature reduction of the Ru/ TiO, sample, the hydrogen chemisorption decreases, but the specific activity of surface ruthenium atoms is not changed, activation energies for hydrogenolysis decrease and selectivities are shifted towards those of smaller particles. The SMSI effect on Ru/TiOz cannot be reduced to a simple dilution of the ruthenium surface by titanium species. Key words: alkane hydrogenolysis, ruthenium/magnesia, ruthenium/titania, mina, catalyst preparation (ligand exchange), adsorption.

ruthenium/alu-

INTRODUCTION

The influence of precursor and support on the catalytic properties of transition metals is still not well understood. This subject has been studied mainly using Fischer-Tropsch synthesis as a catalytic test, in order to clarify the effect of strong metal-support interaction (SMSI ) on catalysis [ 11. The so-called SMSI refers to the peculiar interaction occurring between small metallic particles and a reducible carrier, like TiOz. In the SMSI state, specific catalytic and chemisorptive behaviours appear: hydrogen adsorption and alkane hydrogenolysis decrease, whereas carbon monoxide hydrogenation increases. The case of titania-supported platinum, palladium and rhodium has been widely studied. Sadeghi and Heinrich [ 21 reported a net electron transfer from Ti3+ to metallic rhodium, in agreement with Xcu computations by Horsley [ 31. Hermann et al. [ 41 reached the same conclusion for Pt/TiOz, from electrical con-

0166-9834/90/$03.50

0 1990 Elsevier Science Publishers B.V.

34

ductivity measurements. However Huizinga and Prins [ 51 concluded that the XPS spectrum of titania-supported platinum was mainly determined by the size of the metallic particles and not by interaction with the support. The precursors used for these works on platinum were different: hexachloroplatinic acid in one case [4], and diammino platinum in the other case [5] ; this could explain these discrepancies. Moreover Van’t Blik et al. [ 61 demonstrated that SMSI indeed modifies the initial activity for carbon monoxide hydrogenation over Rh/TiOs, but this effect disappears at the steady state. From these points of view Ru/TiO, exhibits a particular behaviour, since Taniguchi et al. [ 71 reported that the turnover frequency for carbon monoxide hydrogenation was independent of the conditions of reduction, whereas a tenfold increase occurred upon high-temperature reduction of titania-supported platinum and palladium catalysts. Conesa and Soria [ 81 also observed peculiarities for hydrogen adsorption on ruthenium compared to platinum, palladium or rhodium supported on TiOa. Moreover, several independent reports [g-11] agree that chloride ions affect both catalytic and chemisorptive properties of supported ruthenium particles. Burch et al. [ 91 suggested that residual chlorine on the catalyst, after the first reduction, lowers the activity for alkane hydrogenolysis and prevents creation of the SMSI state. After a cycle including reduction, reoxidation and further reduction steps, Ru/TiOz was converted to the SMSI state and became inactive for n-butane hydrogenolysis [ 121. The effect of precursor and support is also well illustrated in the work of Moyes et al. [ 131, in which it was reported that the activity pattern for ethane hydrogenolysis is Ru/SiOa > Ru/TiOB >> Ru/A1,03 when RuB(CO ) 12was used as precursor, whereas it became Ru/A1203 > Ru/SiO, N Ru/TiOg when H4Ru4 (CO ) 12was the precursor. This large body of work, sometimes conflicting, points out the difficulty of analysing in a simple way the influence of support and precursors on the catalytic properties. Moreover it emphasizes the necessity for a careful control of the catalyst preparation, when comparing different work. The cornerstone seems to be the particle morphology and the surface topology, which would be determined by the nature of precursor and support, respectively, as suggested by Bond et al. [ 121. In a recent study by our group the conversion of a series of hydrocarbons, including n-hexane, methylpentanes, 2,2dimethylbutane, 2,2,3,3_tetramethylbutane (TeMB ) and methylcyclopentane, was investigated over a series of Rh/A1203 of widely varying dispersions [ 141. The splitting of C-C bonds was highly sensitive to the metallic dispersion, but the magnitude of this effect depends on the hydrocarbon used as substrate. The fundamental reason for that behaviour is the formation of different surface intermediates, which depend on both the nature of the alkane, and the coordination of the surface sites. For example, TeMB gives mainly demethylation on small rhodium particles, and splitting to iso-butane on large rhodium particles. The same trends are observed for Ru/A1203 catalysts [ 151. Assuming that the catalytic properties

35

are controlled by the coordination number of surface metal atoms, this set of model reactions was applied for the study of the surface topology of bimetallic ruthenium aggregates [ 161, and of the influence of chloride on rhodium catalysts [ 171. A consistent description of rhodium, ruthenium and some of their alloys, supported on alumina, could then be obtained [ 181. In the present work, we apply this method to the investigation of the influence of precursors and supports on the catalytic properties of ruthenium for the conversion of n-hexane, 2-methylpentane and TeMB. The influence of the dispersion of ruthenium has previously been established on a series of catalysts prepared from ruthenium acetylacetonate (acac), thus chlorine-free [ 151. EXPERIMENTAL

Reactants Hydrogen of high purity grade (99.99% ) was used for the catalytic experiments, and hydrogen of ultra-high purity (99.9995% ) for adsorption measurements. The hydrocarbons used were n-hexane (nH, Fluka, purity > 99.6% ) 2methylpentane (2MP, Aldrich, purity > 99.4% ) and 2,2,3,3_tetramethylbutane (TeMB, Aldrich, purity> 99.99% ). Ruthenium acetylacetonate from Aldrich (purity > 97% ) , RuCls ( NH4 ) 3 from Strem Chemicals and Ru (NH, ),Cl, from Fluka were used as precursors. The supports used included a highly crystalline y-A1203 from Woelm (surface area 200 m’/g), MgO prepared by pyrolysis of Mg (O&H,), in air at 573 K (surface area 25 m”/g); the sample of TiO:, was prepared by hydrolysis of Ti (OC3H7)4 in water, thorough washing, drying at room temperature, then calcination at 923 K for 5 h (surface area 130 m”/g) . Catalysts Table 1 summarizes the preparation and characteristics of the catalysts. Ru (acac), reacts with the acidic protons of the support and a ligand exchange is obtained [ 19,201. In that case a benzene solution of the precursor was contacted with the carrier for several hours. The solution was then filtered to obtain a good dispersion, or evaporated to reach medium dispersion [ 157. Two samples of Ru/y-AlzOB (Ru15, Ru21) were prepared by ionic exchange either in acidic medium (precursor RuCl, (NH,) *, pH ca. 2 ) or in basic medium (precursor Ru (NH,)&&, pH ca. 9 ). The precursor solution (10 ml) was contacted with alumina, with stirring, for three hours at room temperature. The filtered solid was washed several times to remove non-exchanged ionic species and then oven-dried at 353 K overnight. Whatever the ruthenium precursor and the carrier the different samples

36

were then subjected to a direct reduction (without calcination) at 673 K under hydrogen for four hours. The dispersion of ruthenium was determined by hydrogen chemisorption following the method proposed by Yang and Goodwin [ 211. The hydrogen adsorption measurements were made at room temperature in a conventional volumetric apparatus. The sample was reactivated in situ overnight at 673 K. It was then outgassed to 5*10e3 Pa for three hours, at the same temperature. Isotherms of total hydrogen adsorption on the fresh catalyst were determined from 6.6 kPa to 33 kPa. The time for the equilibration at each pressure ranged from three to four hours. The catalysts were then evacuated for 10 min at the same temperature and a second adsorption was carried out in the same manner. The difference between the two isotherms held for irreversibly adsorbed hydrogen, for which the stoichiometry Hi,,/Ru, = 1 has been reported [ 221. The results of chemisorption were checked by transmission electron microscopy. The agreement was generally good. Actiuity tests Catalytic activities were determined on an aliquot (50-100 mg) of the sample used for chemisorption measurements. The sample was reactivated in situ under flowing hydrogen at 673 K overnight. The rates were measured in a conventional flow reactor at low conversion in order to avoid heat and mass transfer limitations. Depending on the reactivity of the sample, reaction temperatures ranged from 423 K to 493 K. Partial pressures of the reactants were: nH = 5.8 kPa, 2MP =9.7 kPa, TeMB = 1.8 kPa. The composition of the effluent was determined by sampling on-line by a gas chromatograph equipped with an apolar capillary column from J & W (30 m x 0.53 mm I.D., DB-1 phase). RESULTS

Catalytic activities and selectivities were compared at 458 K for nH and 2MP, and 473 K for TeMB. The following parameters were evaluated: conversion:a!=(ti/nCi) 1

/ (CS:+~i/nC;)XlOO 1

selectivity in Cj compound: Si = Ci/ ( i Ci) x 100 1

where Ci = mole percent in the effluents of products with i carbon atoms, and Cz = mole percent in the effluents of reactants with n carbon atoms. The specific activity, or turnover frequency (TOF), is defined as the number of molecules of reactant transformed per unit time for one surface metal atom.

37

After a small decrease of the initial rate (30-50% ) a steady state activity was reached after 30 minutes under stream. No noticeable change of selectivity occurred during this period. The conversion level was kept low enough (a < 2% ) in order to avoid readsorption of products. The selectivity values thus referred to primary products. The depth of hydrogenolysis, or number of fragments obtained from one molecule of reactant was characterized by the fragmentation factor & defined, following Paal and Tetenyi [ 23 ] as: n-1 5=

T

n-l Ci/

C i

i/n

G

A value of 2 reflects the splitting of one single bond, whereas & 2 shows the occurrence of multiple fragmentations (methanisation). Influence

of the precursor

The effect of the precursor is expected to increase when the size of the metallic particles decreases. This was indeed observed in the case of Rh/A1203 catalysts [ 171. In the present case, special attention was devoted to obtaining well-dispersed Ru catalysts, starting from the three different precursors and the same y-Al,O, carrier. In all cases, a ratio of H,,/Ru of > 0.9 was obtained (Table 1). The higher hydrogen uptake on the Ru15 sample (H,,/Ru ca. 1.4) TABLE 1 Characteristics of the ruthenium catalysts H,,/Ru represents the amount of hydrogen irreversibly held by ruthenium, with the procedure used for dispersion measurements [ 201 Sample

Precursor

support

Ruthenium (wt.-%)

Reduction temperature (K)

H,,/Ru ratio

Ru7 Ru15 Ru2 1 Ru43 Ru45 Ru37 RuMgO Ru44 Ru44-1 Ru44-2 Ru44-3

Ru(acac), Ru(NH,W, RuCls(NH,), Ru(acac), Ru(acac)3 Ru(acac), Ru(acac), Ru(acac), Ru(acac), Ru(acac)s Ru(acac),

[email protected], ALO, &0, Al&‘, ALO, Al&s M&

0.3 0.19 0.23 1.0 1.0 0.9 0.13 0.4 0.4 0.4 0.4

673 673 673 673 673 673 673 573 673 773 873

1 1.4 0.90 0.37 0.65 0.76 0.50 0.50 0.13 0.14 0.14

TiOp TiO, TiO, TiOz

38

agrees with previous studies [ 24,251 stating that the stoichiometry of hydrogen adsorption can be greater than 1 on very small particles. The specific activities of these catalysts are reported in Table 2. Only moderate variations of the TOF appear, but the tendency observed for the three reactions is a higher activity for the catalysts prepared from RuCls ( NHI) 3. The selectivities measured at 2% conversion are reported in Tables 3-5. The two samples, Ru7 and RuEi, prepared from Ru(acac), andRu(NH,)&l, show comparable selectivities for nH and 2MP conversions, with a fragmentation factor close to 2. The sample Ru21 prepared from RuCls ( NH4)2 has a different behaviour since it shows a higher fragmentation factor for the three reactions, a lower selectivity for isomerisation of n-hexane and a higher selectivity to iC, in the conversion of TeMB. The low selectivity for 2,2,3_trimethylbutane (TrMB ) indicates that methane is mainly formed by successive splitting in the adsorbed phase of iC4, as suggested by the results of kinetic analysis reported elsewhere [ 151. It can be concluded that the surface structure is similar when using Ru7 and Ru15 precursors, whereas RuCl, ( NH4)3 induces a different behaviour. Indeed, the catalytic properties observed in that case are those of much larger particles, exhibiting dense planes at the surface. Such catalytic behaviour can be reproduced with sample 43 of low dispersion (Hi,/Ru = 0.37) prepared from Ru ( acac)3, thus chlorine free. TABLE 2 Turnover frequencies obtained for the conversion of n-hexane and 2-methylpentane at 458 K, and 2,2,3,3-tetramethylbutane at 473 K, over supported ruthenium catalysts Sample

Ru2 1 Ru15 Ru7 Ru37 Ru45 Ru43 RuMgO Ru44 Ru44- 1 Ru44-2 Ru44-3 RuSn/Al,O,” RuGe/A120sb

HiJRu ratio

0.90

1.40 1.0 0.76 0.65 0.37 0.50 0.50 0.13 0.14 0.14 0.39 0.41

TOF (h-l) 2MP

nH

TeMB

25.4 19.2 8.7 30 40 50 19 53 132 39 21 13 4.6

29.4 9.2 10.5 42 50 63 41 54 101 58 40 21 5.2

1.9 0.5 0.4 1.3 1.5 1.2 0.7 0.7 2.8 2.7 1.4 0.9 0.02

“Bimetallic catalyst prepared from Ru37 sample as parent material [ 161; Sn/Ru (at/at) =0.26 bBimetallic catalyst prepared from Ru37 as parent material [ 161; Ge/Ru (at/at) =0.36.

39 TABLE

3

Selectivity

in different products

for the conversion

of n-hexane

over supported

ruthenium

cata-

lysts at 458 K Fragmentation

Products

(% )

C,

C2

C3

nC,

Ru2 1

36.5

25.0

16.0

13.0

7.5

1.5

2.50

Ru15

24.0

22.0

19.5

18.0

11.5

4.5

2.10

Ru7 Ru37

16.5 25.5

24.0 22.0

23.5 20.0

21.5 17.0

10.0 12.0

3.0 2.5

2.10 2.25

Sample

factor nC,

iC,

Ru45

28.5

21.0

19.0

15.5

12.0

2.5

2.30

Ru43

35.0

20.5

16.5

13.0

12.0

2.0

2.20

RuMgO Ru44

29.0 21.5

16.5 25.5

17.5 21.0

12.5 19.5

17.5 9.0

6.5 1.5

2.20 2.25

Ru44-1 Ru44-2

19.0 14.0

26.0 26.0

23.0 24.5

21.5 23.0

9.5 7.5

3.5

2.15 2.10

Ru44-3 RuSn/A1,03 RuGe/Al,O,

13.0 53.5 21.5

25.5 20.0 21.0

26.0 10.5 23.0

23.5 8.0 18.5

8.0 6.0 9.0

2.5 1.0 6.0

2.05 3.15 2.20

TABLE

4

Selectivity

in different products

for the conversion

of 2-methylpentane

over supported ruthenium

catalysts at 458 K Sample

Fragmentation

Products

factor CI

C,

C3

iC,

iC,

nCS

iC,

Ru21 Ru15

35.5 25.5

16.0 18.5

3.5 3.0

20.5 23.0

18.5 20.0

1.5 2.5

2.5 6.5

2.20 2.10

Ru7 Ru37 Ru45

23.0 26.5 31.5

20.5 15.0 15.5

3.0 4.5 5.0

24.5 18.0 17.0

18.5 20.5 18.5

2.5 4.5 3.0

6.0 11.0 7.5

2.00 2.10 2.20

Ru43

35.0

13.0

5.0

16.0

19.5

2.5

7.0

2.20

RuMgO Ru44

37.5 26.5

12.0 23.0

4.0 3.5

13.5 25.5

23.0 12.0

2.0 2.0

6.0 5.5

2.20 2.20

Ru44-1

22.0

21.0

2.5

24.5

15.0

2.50

11.3

2.05

Ru44-2 Ru44-3 RuSn/A1,OB

19.5 20.5 44.0

23.0 25.5 10.5

3.0 2.5 3.5

27.5 28.0 19.0

13.5 14.5 14.5

3.0 3.0 1.5

9.5 5.0 5.5

2.00 2.00 2.40

RuGe/Al,O,

26.5

20.5

3.5

22.5

14.5

5.0

5.5

2.10

40 TABLE

5

Distribution

of products

for the conversion

of 2,2,3,3-tetramethylbutane

over supported

ruthe-

nium catalysts at 473 K Products

Sample

(%

Fragmentation

)

factor C,

C,+C,

iC,

TrMB”

18.0 32.0 21.0

-

74.5 42.0

4.5 25.5

2.30 2.10

-

60.5 45.0 64.0

2.05 2.00 2.25

-

68.5 56.5 56.0 48.5

18.0 26.5 12.5 7.5 12.5

31.0

-

37.0

24.0 42.0

16.0

RuSn/Al,O, RuGe/Al,O,

39.0

11.0 _

Ru21 Rul5 Ru7 Ru37 Ru45 Ru43

28.0 23.0 22.0 25.0 25.5 24.0

RuMgO Ru44 Ru44-1 Ru44-2 Ru44-3

“TrMB TABLE

17.5 27.0

2.30 2.05 2.15 2.0

36.5

31.0 22.0

2.0 2.20

42.0 28.0

4.0 32.5

2.95 2.05

is 2,2,3_trimethylbutane. 6

Activation Sample

energies for the conversion Activation

of alkanes in the temperature

range 433-493

K

energies

(kJ/mol) nH

2MP

TeMB

131 112

127

116 104

95 96

175 153 172

Ru44 Ru44-1 Ru44-2 Ru44-3

111

99 108

RuSn/Al,O, RuGe/Al,O,

110 130

Ru7 Ru37 Ru45 Ru43

120 54 54

100

58 50 114 117

196 247 221 130 160 156

Influence of the support The catalytic properties of alumina, magnesia and titania-supported ruthenium, prepared from Ru ( acac)3, are also reported in Tables 2-6. In that case, Ru/TiO, was reduced at 573 K in order to prevent SMSI. The dispersions of

41

Ru/MgO and Ru/TiO, were similar (Hi,,/Ru ca. 0.5)) and two Ru/A1203 samples in the same range of dispersion were added for comparison purposes. With regard to selectivities, the extent of deep hydrogenolysis is the same whatever the carrier, as deduced from the fragmentation factor. Ru/MgO shows selectivities patterns typical of ruthenium aggregates of medium size, e.g. Ru45, supported on alumina [ 151; therefore, it is expected that the surface structure of ruthenium will be similar on these two supports, Ru/TiO, favors the splitting of internal C-C bonds for nH and 2MP hydrogenolysis. The selectivity for demethylation of TeMB is also highest on Ru/ Ti02, and characteristic of very small ruthenium aggregates. These results therefore agree with the idea that the dispersion of ruthenium is underestimated when TiO, is used as carrier. An electron micrograph of Ru44 is given in Fig. 1. The presence of the lattice fringes of the support gives an internal calibration of the microscope, and the size of the smallest particles appears to

Fig. 1. Electron

micrograph

of Ru/TjO,

reduced at high temperature

(773 K).

42

be below 1 nm. However this sample also contains large particles, in the 5-10 nm range; thereby the dispersion determined by chemisorption is an average between small and large particles. Nevertheless a large extent of the ruthenium surface corresponds to very small particles, which will exhibit their specific properties. Accordingly, the catalytic properties of Ru44, compared to those of a well-dispersed Ru/A1203 catalysts, like Ru7, show a good similarity. However, a second possible interpretation can be suggested, taking into account that, even at this low reduction temperature (573 K) a partial reduction of the support could be initiated, lowering the hydrogen uptake and splitting the selectivity patterns towards those of very small particles, as discussed further. Effect of the temperature of reduction of Ru/TiO, By increasing the reduction temperature, Ru/TiO, can be pushed into the SMSI state, which is achieved around 773 K [ 5,9]. The occurrence of SMSI is evidenced by the large decrease of hydrogen uptake, without sintering of the metallic phase (Table 1). However the decline of TOF, expected from other works [ 1,9,12,24] does not happen (Table 2). Apparently, the Ru/TiO, catalysts prepared from Ru( acac), retain a normal activity, even after high temperature reduction. Nevertheless, the activation energy for all three reactions is lowered, and this could indicate some changes in the mechanism of the reaction (Table 6). The selectivity patterns for alkane conversion are also affected by the reduction at high temperature (Tables 3-5 ) . Whatever the alkane processed, deep hydrogenolysis decreases and only single C-C bond splitting remains in the conversion of TeMB and 2MP. Isomerisation of 2MP reaches a maximum. More generally, the selectivities for hydrogenolysis are shifted clearly towards those of smaller ruthenium particles: internal cleavage of nH and demethylation of TeMB are favored. DISCUSSION

All three hydrocarbon conversions give the same picture of the catalyst surface. This rules out the possibility of contamination of one reactant by trace impurities and also minimizes possible artefacts due to carbon deposits which should vary with the reactant. The discussion of the catalytic properties can be based on two aspects: the changes of TOF and of selectivity. The study of the influence of the dispersion of ruthenium supported on alumina has demonstrated that the changes in selectivity can be accounted for by assuming the occurrence of different surface complexes, depending on both the alkane processes and the coordination of the metallic site [ 151. a8 Complexes can be formed for the splitting of bonds involving primary and secondary carbon atoms. CUS Complexes, or metallocyclopentanes, are readily formed with TeMB, but require a high coordination

43

of the ruthenium surface site. Finally, ay complexes, or metallocyclobutanes, occur whatever the coordination of the site, but with a lower probability than cubcomplexes. Thus, increasing the dispersion usually decreases the TOF and modifies the selectivity patterns: large particles favor deep hydrogenolysis (high values of the fragmentation factor r), splitting of terminal C-C bonds of nH and 2MP and central cleavage of TeMB, whereas small particles favor single hydrogenolysis (<- 2)) splitting of internal C-C bonds of nH and 2MP and demethylation of TeMB. The comparison of Ru/A1203 obtained from various precursors shows that the samples prepared from Ru (acac )3 or Ru ( NH3 ) ,Cl, are similar, but a clear effect of the precursor is observed when using RuCls ( NH4) 3. In that case, the catalytic properties of this latter sample look like those of Ru/A1203 of lower dispersion, but obtained from Ru (acac ) 3, like Ru43. Such a behaviour indicates a higher coordination of the surface ruthenium atoms. Usually high coordination is associated with flat surfaces. The occurrence of both flat surfaces and high dispersion can be reconciled by assuming an epitaxial growth of ruthenium on the support when RuCI, (NH 4 ) 3 is used as precursor. Electron diffraction has given direct evidence of epitaxial growth of Pd particles prepared by deposition of PdCl, on highly crystalline alumina [ 271. The precursor showed a strong influence, since epitaxy was not observed when the impregnation was done with Pd (NO, ) 2. The hypothesis that a similar behaviour occurs with ruthenium appears reasonable. The comparison of ruthenium supported on alumina, magnesia and titania (reduced at 573 K), shows that provided the distribution of particle sizes is taken into account, the support has only a small effect on the catalytic properties of ruthenium catalysts prepared from Ru (acac ) 3. By increasing the reduction temperature of Ru/TiOz, the catalytic properties move in the direction of those of small particles: a decrease of deep hydrogenolysis, and an increase of internal cleavage for nH and of demethylation for TeMB. However the TOF for all three reactions shows a small increase, when the hydrogen adsorption decreases. These changes in selectivity pattern cannot then be attributed to an increase of the ruthenium dispersion. The simple interpretation is to assume a migration of TiO, species onto the surface of the metallic particles. This phenomenon was suggested previously by Tauster and Fung [28] to interpret the decrease in hydrogen and carbon monoxide uptake for Pt/TiO, in SMSI state, and confirmed by other authors [ 26,29-331. The catalytic properties described above could then be understood assuming a dilution of the active ruthenium surface by TiO, species (ensemble size effect). This dilution occurs either by an homogeneous distribution of titanium species on the surface, or by a preferential localization on the dense planes of the particles, thereby keeping free the sites of low coordination, edges and corners, responsible of the specific properties of small aggregates. Biloen et al. [ 341 have proposed an elegant method to check the importance

44

of dilution of the surface on catalysis: if the size of the active site is the main parameter, dilution of the surface by different modifiers should result in the same behaviour as reported for platinum covered by tin, carbon or sulphur. In the present case, we can compare the effect of TiO,, with that of tin or germanium (Tables 2-6)) added by reacting an organometallic [e.g. Sn (C,H,),] with a Ru/Al,O, catalyst in presence of hydrogen [ 161. Though the extent of the ruthenium surface coverage is unknown, two main characteristics do appear: (1) The addition of tin or germanium decreases the TOF but induces only small changes in the activation energy of hydrogenolysis; by contrast, hightemperature reduction of Ru/TiO, slightly increases the specific activity, but decreases the activation energy. A compensation effect, as proposed by Burch et al. [lo] accounts for this observation: the number of ruthenium sites is lowered, but the sites which are left free are more active. Due to the complexity of the kinetics it is difficult to precisely establish which step was modified. Badyal et al. [ 351 suggested that hydride species of the RuTiH, type would be formed, in which hydrogen will be loosely held. A lower heat of hydrogen adsorption could readily account for a decrease of the activation energy of hydrogenolysis, since the reaction order is negative versus hydrogen [ 151. (2) The selectivities for alkane hydrogenolysis are shifted towards those characteristic of small ruthenium particles for Ru/TiOz, reduced at 773 K, and RuGe/AlzOa, but in the opposite direction for RuSn/Al,O,. The behaviour of RuPb, RuSn, RuSb and RuGe aggregates for alkane hydrogenolysis was interpreted previously assuming a topological segregation of tin, antimony or lead towards the sites of lower coordination, edges and corners, whereas a random distribution of germanium sets up in these bimetallic aggregates [ 161. The similarities in the shifts of selectivity patterns, between the addition of germanium to ruthenium and the increase of reduction temperature for Ru/TiO,, suggest a migration of TiO, species onto the ruthenium surface, as proposed for Rh/TiOa [ 331 and Ru/TiO, [ 9,121. Moreover, we think that these species are randomly distributed over the surface, like germanium. However, the constancy of TOF and the changes in activation energies suggest that SMSI for Ru/TiOz cannot be reduced to a simple dilution of the ruthenium phase. Electronic factors probably interfere. The results reported by Burch et al. [9] concerning the inactivity of chlorine-free Ru/TiOz in the SMSI state could not be reproduced here. Moreover the selectivity for isomerisation of n-hexane (5-10%) cannot be compared to the 30% value reported by these authors. This illustrates the difficulty of reproducing similar Ru/TiO, catalysts, when the support and the salt are different. This may be indicative of their influence on the properties of the final solid.

45 REFERENCES

5

6 7 8 9 10 11 12

13 14 15 16

17 18 19

20 21 22 23 24 25 26 27

28 29 30

S.J. Tauster, Accounts Chem. Res., 20 (1988) 389. H.R. Sadeghi and V.E. Heinrich, J. Catal., 87 (1984) 279. J.A. Horsley, J. Am. Chem. Sot., 101 (1979) 2870. J.M. Herrmann, J. Disdier and P. Pichat in B. Imelik, C. Naccache, G. Coudurier, H. Praliaud, P. Meriaudeau, P. Gallezot, G.A. Martin and J.C. Vedrine (Eds.), (Studies in Surface Science and Catalysis, Vol. II), Metal-Support and Metal-Additive Effects in Catalysis, Elsevier, Amsterdam, 1982, p. 27. T. Huizinga and R. Prins, in B. Imelik, C. Naccache, G. Coudurier, H. Prahaud, P. Meriaudeau, P. Gallezot, G.A. Martin and J.C. Vedrine (Eds.), Studies in Surface Science and Catalysis, Vol. 11, Metal-Support and Metal-Additive Effects in Catalysis, Elsevier, Amsterdam, 1982, p. 11. H.F.J. Van’t Blik, J.C. Vis, T. Huizinga and R. Prins, Appl. Catal., 19 (1985) 405. S. Taniguchi, T. Mori, Y. Mori, T. Hattori and Y. Murakami, J. Catal., 116 (1989) 108. J.C. Conesa and J. Soria, J. Phys. Chem., 86 (1982) 1392. R. Burch, G.C. Bond and R.R. Rajaram, J. Chem Sot., Faraday Trans I, 82 (1986) 1985. T. Narita, H. Miura, K. Sagiyama, T. Matsuda and R.D. Gonzalez, J. Catal., 103 (1987) 492. K. Lu andB.J. Tatarchuk, J. Catal., 106 (1987) 166. G.C. Bond, R.R. Rajaram and R. Burch in M.J. Phillips and M. Ternan (Eds.), Proc. 9th Int. Congress Catalysis, Calgary, Vol. 3, The Chemical Institute of Canada, Ottawa, 1988, p. 1130. R.B. Moyes, P.B. Wells, S.D. Jackson and R. Whyman, J. Chem. Sot., Faraday Trams I, 82 (1986) 2720. B. Coq and F. Figueras, J. Mol. Catal., 40 (1987) 93. B. Coq, A. Bittar and F. Figueras, Appl. Catal., in press. B. Coq, A. Bittar and F.Figueras, Studies in Surface Science and Catalysis, Vol. 48,European Conference on Structure and Reactivity of Surface, Trieste, Elsevier, Amsterdam, 1989, p. 327. B. Coq, F. Figueras and T. Tazi, Zeit. Physik. D., 12 (1989) 579. B. Coq, A. Bittar, T. Tazi and Figueras, F., J. Mol. Catal., 55 (1989) 34. J.P. Boitiaux, J. Cosyns and S. Vasudevan, in G. Poncelet, P. Grange and P.A. Jacobs (Eds.), Studies in Surface Science and Catalysis, Vol. 16, Preparation of Catalysts III, Elsevier, Amsterdam, 1982, p. 123. G. Vlaic, J.C.J. Bart, W. Cavigliolo, A. Furesi, V. Ragaini, M.G. Cattania Sabbatini and E. Burattini, J. Catal., 107 (1987) 263. C.H. Yang and J.G. Goodwin, J. Catal., 78 (1982) 182. A. Sayari, H.T. Wang, and J.G. Goodwin, J. Catal., 93 (1985) 368. Z. Paal and P. Tetektyi, Nature (London), 267 (1977) 234. J.C. Vis, H.F.J. Van’t Blik, T. Huizinga, J. Van Grondelle and R. Prins, J. Mol. Catal., 25 (1984) 367. B.J. Kip, F.B.M. Duivenvoorden, DC. Konigsberger and R. Prins, J. Catal., 105 (1987) 26. D.E. Resasco and G.L. Haller, J. Catal., 82 (1983) 279. H. Dexpert, E. Freund, E. Lesage and J.P. Lynch in B. Imelik, C. Naccache, G. Coudurier, H. Praliaud, P. Meriandeau, P. Gallezot, G.A. Martin and J.C. Vedrine (Eds.), Studies in Surface Science and Catalysis. Vol. 11, Metal-Support and Metal-Additive Effects in Catalysis, Elsevier, Amsterdam, 1982, p. 53. S.J. Tauster and SC. Fung, J. Catal., 55 (1978) 29. G.L. Haller, V.E. Heinrich, M. McMillan, D.E. Resasco, H.R. Sadeghi and S. Sakellson, Proc. 8th Int. Congr. Catalysis, Berlin, Vol. 5, Verlag Chemie, Weinheim, 1984, p. 135. D.N. Belton, Y.M. Sun, J.M. White, J. Am. Chem. Sot., 106 (1984) 3059.

46 31 32 33 34 35

A.J. Simoens, R.T.K. Baker, D.J. Dwyer, C.R.F. Lundand R.J. Madon, J. Catal., 86 (1984) 359. S. Takatani and Y.W. Chung, J. Catal., 90 (1984) 75. F.J. Schepers, J.G. Van Senden, E.H. Van Broekhoven and V. Ponec, J. Catal., 94 (1985) 400. P. Biloen, J.N. Helle, H. Verbeek, F.M. Dautzenberg and W.M.H. Sachtler, J. Catal., 63 (1980) 112. J.P.S. Badyal, A.J. Gellman and R.M. Lambert, J. Catal., 11 (1988) 383.