Hydrogenation catalysts based on nickel and rare earth oxides

Hydrogenation catalysts based on nickel and rare earth oxides

Applied Catalysis A: General, 101 (1993) 73-93 Elsevier Science Publishers B.V., Amsterdam 73 APCAT A2546 Hydrogenation catalysts based on nickel a...

2MB Sizes 0 Downloads 10 Views

Applied Catalysis A: General, 101 (1993) 73-93 Elsevier Science Publishers B.V., Amsterdam

73

APCAT A2546

Hydrogenation catalysts based on nickel and rare earth oxides. Part II: XRD, electron microscopy and XPS studies of the cerium-nickel-oxygen-hydrogen system G. Wrobel, M.P. Sohier, A. D’Huysser and J.P. Bonnelle Laboratoire de Catalyse H&+ogPne et Homo&e, URA CNRS N” 402, Universite des Sciences et Technologies de Lille, 59855 Villeneuve d’Ascq Cedex (France)

and J.P. Marcq Soci& de la raffinerie B.P. et ELF de Dunkerque, BP 4519,59381 Dunkerque Cedex 1 (France) (Received 18 December 1992, revisedmanuscriptreceived26 March 1993)

Abstract

In a previous article we selected a preparation process of cerium and nickel oxides which permits catalyststo be obtained which contain as much hydrogenas the intermetahiccompounds with the same composition, and which are more active in benzene hydrogenation.Moreover the catalysts’ behaviour varies with the Ni/Ce ratio (n) and at least two zones can be distinguished.In the present paper we have found correlations between catalytic activity, hydrogencontent and some physical and chemical characteristicsof the solids, both in the oxidized and reducedstates.The techniquesused wereelectron microscopy (TEM, SEM, EPMA), X-ray diffraction and X-ray photoelectronspectroscopy.Whatever the method, the catalystsare classifiedinto two families.For x Q 0.5 nickel is insertedin the ceria lattice to form a solid solution. Above 0.5, both crystallized nickel oxide and solid solution coexist. In the reduced state anionic vacancies able to receive hydrogen,probably in an hydridic form, are created in the bulk and at the surfaceof the solid solution. The cat&tic resultscan then be explainedby assuming the existence of three kinds of active sites which differ from each other in terms of the environmentof nickel, and an explanation for the higher efficiency of the catalyst with z = 5 is advanced. Finally the situation is shown to be almost identical in ceria-supportednickel catalysts. Key words: electron microscopy; hydrogen storage; nickel-cerium oxides; unsaturated hydrocarbon hydrogenation; XPS; XRD

INTRODUCTION

Intermetallic compounds based on nickel and rare earths lie at the basis of numerous studies in catalysis. In a previous paper [ 1] we have demonstrated Correspondence to: Dr. G. Wrobel, Univ. des Sci. et Tech. de Lille, Lab. de Cat. Hetirogene et Homogene, URA CNRS No. 402, 59655 Villeneuve d’Ascq, France. Tel. ( +33) 20434949, fax. (+33) 20436501.

0926-860X/93/$06.00

0 1993 Elsevier Science Publishers B.V. All rights reserved.

74

G. Wrobel et al./Appl. Catal. A 101(1993) 73-93

the strong beneficial effect due to the nickel-cerium association in a set of partially reduced mixed oxides, called oxhydrides, and prepared according to different routes: from intermetallics alloys (route A); mixtures of cerium and nickel nitrates (route B); coprecipitation of hydroxides (route C); and mechanical mixtures of pure CeOz and NiO obtained by route C. A ceria-supported nickel has also been prepared for comparison with the bulk catalysts. All the compounds, provided that they are oxidized before the reduction treatment, can store hydrogen and are active in the hydrogenation of alkadienes, both under atmospheric pressure and at high pressure. However, among the bulk catalysts the series C give the best results and, according to the Ni/Ce atomic ratio, there is an almost linear relationship between the activity in benzene hydrogenation and the amount of hydrogen stored in the solids. In order to explain such behaviour, physicochemical studies have been undertaken using X-ray diffraction (XRD), electron microscopy and X-ray photoelectron spectroscopy (XPS ). All the results lead us to propose a model of active sites which is valid for bulk and supported catalysts and which could be generalized to other supported catalysts based on ceria or some other supports and metals of Group VIII. EXPERIMENTAL

Catalysts The solids were prepared according to process C, described previously [ 11, by coprecipitation of nickel and cerium hydroxides from mixtures of their nitrates with triethylamine, followed by calcination under an air flow at 723 K and reduction with hydrogen at 573 K. The atomic ratio x = Ni/Co varied between 0 and 7 and the corresponding samples will be denoted by C, below. The ceria-supported nickel catalyst was prepared by wet impregnation: ceria prepared in the same way as the coprecipitates was stirred with a nickel nitrate solution so that the surface in the reduced state was expected to be covered by a monolayer of nickel atoms. That corresponds to an overall Ni/Ce atomic ratio equal to 0.19. XRD X-ray powder diffraction analysis was carried out with a D 5000 Siemens diffractometer using a copper target and a secondary beam monochromator. The Kcx, contribution was eliminated from the spectra by computer post-processing. The crystahite sizes were obtained from the strongest reflections for NiO and CeOz using the Scherrer equation.

G. Wrobel et al./Appl. Catal. A 101 (1993) 73-93

75

Electron microscopy

The aspect and the structure of the materials were examined by means of several electron microscopy techniques appropriate to the information required. The TEM (transmission electron microscopy) observations were performed on a conventional Philips apparatus (EM 300) at 100 kV and on an analytical microscope (Philips CM 30) at 300 kV. SEM (scanning electron microscopy) studies were done with a Jeol 1OOCXmicroscope at 20 or 80 kV. Electron probe X-ray microanalyses (EPMA) were carried out using a CAMECA (Cabemax) apparatus equipped with a TAP monochromator crystal. The electron gun was operated at 15 kV and the transitions chosen for this study were Ka for Ni and Lcr for Ce. The sample preparation was adjusted according to the technique used. For the TEM observations the ultrasonic agitated powder in alcohol suspension was deposited on a copper grid covered with a thin carbon film. A gold film was vaporized onto the samples for the SEM study. EPMA experiments were performed on thin foils of solids obtained by ultramicrotomy using a diamond knife on grains embedded in an epoxy resin.

XPS

XPS spectra of the samples were run on a Leybold Heraeus LHS 10 instrument using a non-monochromated Al Ka! radiation. The anode was operated at 300 W power and the FRR (fixed retardation ratio) mode was applied. The base pressure during analysis was 1.33 x 10e6 Pa. Samples were lightly pressed on a stainless steel holder, and mounted on a sample rod. Calcined samples were analyzed directly. When necessary, reduction treatments (with Hz/ NP=10:90 (V/V)) were carried out in situ up to 573 K, in a preparation chamber attached, via an ultrahigh vacuum chamber, to the analysis chamber of the spectrometer. XPS nickel 2133/Z,cerium 3d, oxygen 1s and carbon 1s were recorded. In no case was X-ray-induced reduction observed on the analyzed species. Binding energies (BE) were calibrated by assuming 285 eV for the 1s level of hydrocarbon contaminant carbon. The reproducibility of the peak position was estimated to be ? 0.2 eV. The data analysis, in general, involved smoothing and background subtraction by means of a linear integral profile. Surface atomic ratios were obtained on the basis of the peak area intensities after correction for instrumental parameters, photoionixation cross sections and electron mean free paths [ 2 1.

76

G. Wrobel et al./Appl. Catal. A 101 (1993) 73-93

RESULTS

Electron microscopy study of the oxide precursors This part is essentially concerned with the solids in the oxidized state because most of the compounds are pyrophoric in the reduced state, and are then modified when placing them in air once again. The first TEM and SEM investigations showed that all the pure and mixed oxides have the same appearance. They are constituted by large opaque particles (up to 0.4 mm) (Fig. la) and aggregates of small grains (4-60 nm). Observations at higher magnification, however, show that, as a matter of fact, one cannot distinguish between the surface of a large particle and the surface of an aggregate (Figs. Id-e ) . In order to visualize the repartition of Ni and Ce, an atomic analysis was performed on the C!, sample. At the magnification scale used, the distribution of the two elements appears homogeneous, even on large particles, as shown in Figs. la-c. To complete the microscopic information, electronic microdiffraction was used to determine the chemical structure of the small grains. The selected area was nearly 1 pm2 in most cases. All the diffraction spots corresponding to aggregates or to the edges of large particles are distributed on circles which seem continuous (C ) or punctuated (P). An example is reported in Fig. If and Table 1 for the C, oxide. By comparison with the experimental diagrams of pure oxides prepared according process C and the JCPDS data, it can be concluded that this sample is constituted by a mixture of the two species NiO and Ce02 within the accuracy of the microdiffraction technique. The punctuated diagram is related to NiO, suggesting that the corresponding particles are larger. Furthermore, this last point is completely confirmed by TEM examination at high magnification (Fig. 2a). In this picture the grains of NiO are recognizable by their almost hexagonal shape, and their mean size is clearly higher than the size of the roundish ceria particles, which is evaluated to about 5 nm. Thus the surface of the C, mixed oxide, which appeared homogeneous from the first SEM and EPMA studies, is in fact constituted by grains of a CeO,like compound intimately mixed with - or deposited on - bigger NiO particles. The morphology of some other C, oxides is shown in Fig. 2. The Co.5 sample looks like CeO, while CZ differs from C5 only in terms of the relative amount, and perhaps the size of the NiO particles. In microdiffraction the oxides appear to be constituted by a mixture of Ce02 and NiO in proportions consistent with their composition, except for the C& oxide where the CeO, diagram alone was detected; the few spots attributed to NiO occasionally visible in this last sample are not significant. The Ni/Ce02 sample has also been examined. Here the distinction between the aggregates of Ce02 and the NiO grains is more visible, but in zones where

G. Wrobel et al./Appl. Catal. A 101 (1993) 73-93

b

e

C

f

Fig. 1. Morphology, composition and structure of the C6 oxide. (a, b, c) EPMA pictures, bar = 10 p. (a) Global, (b) Ni Ka, (c) Ce La. (d, e) SEM pictures, bar=200 nm. (d) Aggregate, (e) large particle, (f) microdiffraction diagram obtained by TEM.

G. Wrobel et al./Appl. Catal. A 101 (1993) 73-93

78 TABLE 1

Comparison of C, oxide with pure oxides in microdiffraction VS: very strong, S: strong, m: medium, w: weak, vw: very weak. (C) Continuous circles; (P) Punctuated circles

CeOz" (C)

Go d 3.12 2.72 2.43 2.08 1.91 1.63

(Cl (0 (PI P) (Cl (Cl

1.47 1.35 1.25 1.20 1.10 1.05 0.95 0.94

(PI (Cl (PI (PI (Cl (P) (P) (PI

I

d

Z

m

3.13 2.69

S m

W

S S m W

S VW W W

m vw W

NiO” (P) d

2.42 2.09 1.91 1.63 1.56

W

W

m m

m W W

m 0.91

1.26 1.20 1.04 0.95 0.93

d

Uh

d

UL

3.124 2.706

100 29 2.410 2.088

91 100

1.476

57

1.259 1.206

16 13

1.044 0.9582 0.9338

8 7 21

0.8527

17

0.8040

7

1.913 1.632 1.562

51 44 5

1.353 1.241 1.210 1.1044 1.0412 0.9505

5 15 6 12 9 5

0.9146 0.9018 0.8556 0.8251 0.8158

13 7 7 6 5

S

W W

NiO JCPDS

S S

S S 1.47

1.35 1.24 1.21 1.10 1.04 0.95

Z

Ce02 JCPDS

W W

m

W

0.85

(P)

m

0.85

m

0.81

(PI

W

0.80

W

“Experimental diagram of the pure oxides prepared according to the C process. The d values are in A.

the two types of particles are covering each other, the aspect of the preparation looks like the C5 oxide one (Fig. 2e).

XRD The XRD spectra of the series C in the oxidized state, including pure oxides prepared in the same way, as well as the supported catalyst, are presented in Fig. 3. The experimental lines correspond to the data of NiO and CeO, reference compounds, but the lines widths are different according to the nature of the oxide, NiO or CeO,. Moreover two ranges can be distinguished, depending on x (Fig. 3a): (i) for x < 0.5 no nickel oxide is detected while the width of the lines attributed to CeO, increases with n; (ii) for x> 0.5 the two species NiO

G. Wrobel et al./Appl. Catal. A 101 (1993) 73-93

79

Fig. 2. TEM examination of some samples in the oxidized state. bar=20 nm. (a) C6, (b) CZ, (c) G2, (4 CeOz, (e) Ni/CeOz.

and CeOz are present. The CeO, lines remain broad while the NiO ones are relatively narrow. As far as the supported catalyst is concerned the spectrum is mainly consti-

80

G, Wrobel et al./Appl. Catal. A lOl(l993)

73-93

Fig. 2. (Continued. )

tuted by CeOz without significant modification of the line widths referrilng to the pure oxide (Fig. 3b). Moreover, fromcomparison of XRD spectra, recabrded under the same experimental conditions, of the supported catalyst and the bulk

G. Wrobel et al./Appl. Catal. A 101(1993) 73-93

81

Fig. 2. (Continued.)

C& oxide, the overall atomic ratio of which is nearly the same, it turns out that C,, is more poorly crystallized while the strongest NiO lines appear weakly in the supported catalyst. In order to compare these results with those of the microscopic investigation, an evaluation of the particle size (d) has been made. The mean d values have been estimated from the XRD line widths of the (111)) (200) and (220) peaks for both NiO and CeO, (Table 2). For all the oxides where NiO is crystallized, the NiO particles are always larger than the CeOz ones and their size develops slightly when x increases. One can also notice that CZ is unique since the CeO, lines in this sample are very similar to those of the pure oxide. Although no systematic study has been undertaken on the compounds in the reduced state, some analyses have been done on samples highly charged with nickel and passivated after reduction. They allow us to establish the presence of metallic nickel without any visible modification of the CeO, lines. XPS analysis Chemical states of nickel and cerium in the oxide precursors Cerium. XPS spectra of cerium compounds are known to exhibit rather complex features due to numerous initial and final 4f configurations [ 31. However, differentiation of the valence states of cerium does not present any difficulty

G. Wrobel et al./Appl. Catal. A 101(1993) 73-93 .7-..-.

‘,__7__-____‘I~--

,~~~~_~~~~~~

-_ ~, _ .

_...._.r...

b

Fig. 3. XRD spectra of the oxides. (a) Influence of the Ni/Ce atomic ratio. (b) Comparison of the Ni/CeOz with pure ceria and bulk C& oxide. Lines of CeOz ( n ) and NiO ( 0 ) from JCPDS data.

G. Wrobel et al./Appl. Catal. A 101 (1993) 73-93

83

TABLE 2 Mean size of CeOz and NiO grains in the series C Sample

Ce02 C0.2 C 0.5 CI c2 C4 G

NiO

d (A) XRD” CeOz

NiO

63 65 38 40 61 45 47

80 100 110 140 210

d (A) EM* CeOz

60-80 30-60 50-75 40-60 -

“From XRD. *From electron microscopy.

and this has been investigated according to the work of Le Normand et al. [ 41. Similar Ce 3d spectra were obtained for all the oxide precursors. By comparing the observed 3d envelope with those reported for Ce3+ and Ce4+ by Le Normand and for Ce metal [ 51, one can unambiguously ascribe the spin-orbit doublet peaks to Ce4+ in CeOJike species. Nickel. Photoelectron spectroscopic measurements of core electrons levels in nickel compounds are complicated by a shake up satellite structure located at a higher binding energy than the main or ‘principal’ photoemission peak. Recent attempts have been made to explain these satellites in term of ligand-tometal charge transfer [ 61, final state effects after photoemission [ 71, or magnetic interactions [ 81. In the main, the amplitude and exact location of satellite and principal peaks depend critically on valence state and coordination environment, and may allow differentiation of closely related species [ 8,9]. This is illustrated in Fig. 4, which shows the Ni 2p,,, band shapes of the C, oxides whose main spectral features are listed in Table 3 and compared to those of well-known nickel-containing reference compounds. The major conclusion which can be drawn from the peak profiles is that the oxidation state of nickel can be unequivocally assigned to divalent nickel. Moreover, as is shown in Table 3, the characteristic XPS parameters have values varying with the nickel content and which can be roughly ordered in two ranges: (i ) the first range corresponds to low nickel content catalysts with spectral features different from those of tetrahedrally (Td) or octahedrally (Oh) coordinated divalent nickel [ 121; (ii) the second range corresponds to high nickel content catalysts with strong spectral similarities between NiO and the last oxides of the C, series.

84

Wrobel et al./Appl. Catal. A lOl(l993)

co9

Cl

C5

c7

e

B.E.feV) 885 860 855 Fig. 4. Ni 2p,,, XPS lines for C, oxides.

850

TABLE 3 XPS parameters of Ni 2p,,, lines for catalysts and reference compounds Samples

EBo

AE”

I,lI,”

Ref.

CO.Wd C 0.2 C0.5 C, C, C, Cs C7 NiO’ NiC&Of

855.2 855.3 855.4 855.0 854.9 854.4 854.4 854.5 854.5 856.4

nd 6.1 6.6 6.6 6.9 7.3 7.2 7.1 7.1 5.1

nd 0.42 0.46 0.60 0.54 0.57 0.57 0.57 0.59 0.74

SlO 111

“Binding energy (eV) of the main line. ‘Energy separation (eV) between the main line and the satellite line. ‘Ratio of the satellite and principal peak heights. dPoorly resolved spectrum. eNi2+ 100% in Oh. ‘Ni’+ 100% in Td.

73-93

G. Wrobel et al./Appl. Catal. A 101(1993) 73-93

85

Reduced samples The behaviour of the nickel and cerium species with respect to hydrogenation can be a way to obtain useful information on their mutual interactions in the oxhydride state. Typical results are shown in Fig. 5 for the Ni 2pa12and Ce 3d spectra of two representative catalysts - one in the range x < 0.5 and the other in the range x > 0.5 - compared to pure nickel oxide and cerium oxide exposed to hydrogen under the same experimental conditions. The striking feature of this study is a simultaneous change in the shapes of the nickel and cerium lines with increasing nickel content. On one hand the Ce 3d spectra of C&, and C5 are substantially different from the spectrum of pure CeO,, which remains inaltered by reduction in H,; attention must be focused on two new Ce3d

NI 2h2

q.E.

0.E. e

.

920

900

880

860

860

Fig. 5. Ce 3d and Ni 2p,,, XPS lines of C, samples in the reduced state compared to Ce 3d and Ni 2p,,, lines of pure reduced CeO, (a) and NiO (b). TABLE 4 Correlation of percentage u”’ peak in Ce 3d region with respect to percentages of Ce4+ and Ce3+ in reduced catalysts Samples

u”’ (o/o)

Ce4+ (%)

Ce3+ (%o)

CeOz C 0.05 CWzLI Co.5a Go

13 13 10

100 100 75 50 60

0 0 25 50 40

7

8

“rhe percentages of Ce4+ and Ce3+ are given with an accuracy within 5 10%.

86

G. Wrobel et al./Appl. Catal. A lOl(1993)

73-93

lines (denoted u’ and v’ in Fig. 5) which are characteristic of Ce3+ species [ 41. On the other hand, while NiO is almost completely reduced to nickel metal, the Ni2+ species initially present in the oxides have been only partially reduced to different extents. In a more quantitative way the relative amounts of Ce3+ and Ce4+ on the surface catalyst samples can be roughly estimated from the intensity of the line u”’ centered at 917 eV [ 131. Table 4 shows a progressive increase of the amount of Ce3+ species up to x=0.5, above which the values levelled off. DISCUSSION

As far as we know, no study has been made on mixed nickel-rare earth oxides and more particularly on nickel-cerium oxides from a catalytic viewpoint. Indeed, the intermetallics based on nickel and rare earths, such as LaN&,, CeNi, or MmNi, (Mm = mischmetal alloy) for example, have been extensively reported for their abilities to store hydrogen [ 141, When they are used in catalysis, particularly in methane synthesis from CO and HP,they behave like nickel supported on the corresponding rare earth oxide [ 15,161. On the other hand conventional Ni/Ce02 catalysts have been studied in the hydrogenation of benzene or toluene and CO conversion [ 17-191. Essentially by means of thermodesorption experiments, the authors established their peculiar abilities to absorb hydrogen; this point has been discussed in Part I [ 11. These Ni/Ce02 catalysts lead to selectivities different from those of Ni/A1203, Ni/SiO, or even Ni/La,O,. Hence the mixed oxides presented in this work are interesting in the sense that the information obtained from their physical and chemical characterisation will lead to a better understanding of the supported catalyst.

Bulk oxide characterization XRD and electron microscopy technique agree in distinguishing two ranges, depending on the Ni/Ce atomic ratio: (i) 0
G. Wrobel et al./Appl. Catal. A 101 (1993) 73-93

87

Ce4+ in the CeO, lattice should be also evidenced by a shift of the CeO, peaks positions towards the greater 28 values, since the ionic radius of Ni2+ and Ce4+ are 0.72 and 1.02 A respectively. Such a tendency exists, but the line broadening is too large to estimate the shift values with any precision. (ii) x > 0.5 (zone II ) where two phases are evidenced. In this zone the oxide precursors are formed by well crystallized NiO particles intimately mixed with smaller CeO,-like grains. Additional information is provided by some quantitative XPS features drawn from the comparison of the surface composition with the bulk content, which is a valuable method for detecting structural changes in solids as composition is varied [ 201. Thus the nickel superficial compositions of the C, oxides determined from XPS atomic ratios, are compared in Fig. 6 to the bulk nickel contents. The 45” diagonal line corresponds to the case of homogeneous solids. It can be seen that, except for C2, all the oxides precursors are nickel-poor in the surface. Moreover the difference between surface and bulk nickel contents increases with x until the C,, compound, and then progressively decreases. The observed discontinuity near x= 0.5 then confirms the existence of different structural properties on both sides of this nickel concentration. In zone I the linear relationship between nickel concentration in the bulk and on the surface implies the formation of a well-dispersed nickel species in the solids. Combined with the previous XPS qualitative analysis, which shows NIO

80

-

H a oo-

Y. g

2 8

40

-

20

-

._ i 3

0

20

40

60

80

100

atomic % Ni in bulk Fig. 6. Variation of the nickel surface concentration as a function of the bulk nickel content in the C, series.

88

G. Wrobel et al./Appl. Catal. A 101 (1993) 73-93

that nickel ions are not located in Oh or Td lattice sites, and with the XRD results, it can be admitted that, for low nickel contents, Ni2+ species diffuse into the fluorite lattice and therefore form a solid solution with ceria. In zone II the solid solution becomes saturated with Ni2+ ions. Any further addition of metal ions can be accommodated only by segregation of a separate metal oxide phase, namely NiO. The progressive decrease of the surface nickel depletion versus the bulk for high nickel contents is consistent with this picture.

CorFehztionbetween the physicochemical properties and catalytic behaviour In Part I [ 1 ] we established that the activity in benzene hydrogenation ( aH) varies linearly with the amount of hydrogen (Qn) stored in the oxhydrides when x is smaller than 5 (Fig. 7). For x> 5 the linear relationship does not hold any longer and the C5 catalyst, which stores the highest amount of hydrogen, is the most active. Moreover, within the domain x < 5 one can distinguish the solids with x < 0.5 (zone I) from those with 1c> 0.5 (zone II). At this stage it is noting that the value of x near 0.5 seems to be critical whatever the technique used to characterize the solids. In zone I, as soon as nickel is introduced in ceria, hydrogen can be extracted from the solid in the reduced state and the oxhydride becomes active in hydrogenation. In zone II the appearance of nickel metal coming from the reduction of free NiO leads to an increase of the amount 35 ;h g 30 ‘= u 2 25 X ;

20

0

2

4

6 Qx

8

10

12

1000 molH ( g. oxhydride

14

16

18

I-’

Fig. 7. Variation of the catalytic activity with hydrogen content in the C, samples. (m) C,, ( A ) Ni/CeO,.

G. Wrobel et al./Appl. Catal. A 101(1993) 73-93

89

of extractable hydrogen and mainly of the hydrogenation activity. So the linear variation between an and Qu and the break observed for x = 0.5 point out the important role of both nickel metal and hydrogen occluded in the oxhydrides. This behaviour of the oxhydrides according to the nickel content has led us to distinguish several types of hydrogenation sites, depending on whether the active nickel comes from the solid solution or from free NiO. To describe these active sites in a more precise way, we have first looked into the reduction of the solid solution (zone I), keeping in mind that this domain corresponds to the coexistence of Ni’, Ni’+, Ce3+ and Ce4+ species. Taking into account all the experimental results we propose that the reduction proceeds according to the following steps: (i) reduction of Ni2+ ions belonging to the solid solution into Ni” species; (ii) departure of water leading to the creation of anionic vacancies in the Ce02 type lattice; (iii) simultaneous reoxidation of a part of the Ni” species by reduction of Ce4+ ions in their vicinity into Ce3+; (iv) filling of the anionic vacancies with hydrogen, probably in the hydridic form, which corresponds to the so-called ‘occluded hydrogen’, as it was already evidenced in other reduced mixed oxides [ 211. Hence the whole reduction process can be described according to the following equations: Ni2++H2 +02-+Ni”+H20+

Cl

(1)

2Ce4++Ni0=2Ce3++Ni2+

(2)

H2+02-+O+OH-+H-

(3)

This allows us to understand how the incorporation of nickel in the Ce02 lattice can lead to the reduction of some Ce4+ ions and to the creation of anionic vacancies filled with hydrogen. In this picture, some sites belonging to the solid solution, called sites A, would be constituted of superficial nickel associated with hydrogen. Now attention must be paid to the zone II, where the activity is strongly enhanced and reaches a maximum for x = 5. The observed synergetic effect can be due to an interaction between nickel metal and the reduced solid solution. This hypothesis is supported by the microscopic pictures (Fig. 2a, b) which show a very good contact between NiO and CeO, particles before reduction and by the fact that pure nickel metal is active and also contains hydrogen (Fig. 7 ) . The role of the solid solution is assumed to furnish and to enhance the quantity of the appropriate hydrogen species. However, there must be an x value for which the interaction between nickel particles and CeO,-like particles is at a maximum. And referring to the relative crystallites size of the two compounds in the oxidized state (Table 2), the value x= 5 is quite plausible (the volume of a NiO particle is much larger than that of a CeO, one). For x > 5 the contact number between all the nickel particles and the solid solution is less important, which contributes to the decrease of the number of active sites, and the activity falls. Then the atoms of nickel metal in strong interac-

G. Wrobel et al./Appl. Catal. A lOl(1993)

90

73-93

S sdution soiid c sites A

l

sites B

0 sites C

Scheme 1. Model for nickel-ceriuxn catalysts.

tion with the solid solution lead to a second type of very active sites (sites B ) which are different from the other superficial nickel atoms belonging to the metal particles (sites C) (Scheme 1). The sites A and B, containing hydrogen as hydride in an anionic vacancy, can be very similar; in that case the different behaviour of the catalysts in zones I and II would be attributed to the number of active sites solely or to the different nature of nickel in the two sites. Further experiments are necessary to resolve this question. Among all the oxides based on nickel and rare earth (RE = La, Ce) prepared according to the route C, it is now possible to understand why the Ce-Ni-OH system has the highest hydrogen reservoir and leads to the most active catalysts [ 11. In contrast to other rare earth elements, the possible simultaneous presence of cerium in the two oxidation states, Ce3+ and Ce4+, is undoubtedly a factor which is essential to the genesis of the active sites through the creation of anionic vacancies. Nickel supported on ceria catalyst The investigation of this solid by means of the same techniques as for the bulk compounds reveals several interesting features. The XRD analysis shows that crystallized NiO is present in small amounts in the oxidic state and the structure of the support does not appear to be modified in comparison with that of pure ceria. However, the XPS results do not allow us to conclude that this sample is solely the addition of the two oxides. Indeed, the INi/lce intensity ratio expected on the basis of a monolayer coverage by nickel is about 0.090.11. The experimentally measured value is 0.06 (8)) whereas a uniform distribution of nickel into the solid should give a theoretical ratio of 0.05 (4 ). So the experimental value is probably due to an homogeneous distribution of nickel

G. Wrobel et al./Appl. Catal. A 101 (1993) 73-93

91

ions in the CeOP surface layers and to the presence of small cry&all&es of nickel oxide as evidenced by XRD. The supported catalyst could then be depicted according to Scheme 2. From a catalytic point of view, all types of active sites exist, as described in the case of the bulk catalysts, and that can explain why the C& bulk oxhydride stores more hydrogen (more A sites) but has a lower activity (almost no B sites). The lower efficiency of the supported catalyst compared to the C5 catalyst can be due to the smaller amount of nickel and to a more important segregation between the NiO particles and the CeO, aggregates (Figs. 2a-e), consequently leading to a less intimate interaction between the two components. It is now interesting to compare the above model (Scheme 1) with the results and ideas related to nickel or other Group VIII metal catalysts supported on ceria. Bearing in mind the description of the solid solution in the reduced state, one may note that the particular property of the model we propose is to gather in one picture three characteristics of the system which are sometimes displayed separately: (i) the simultaneous presence of Ce3+ and Ce4+ species already pointed out for Ni/CeO, [ 221; (ii) the existence of anionic vacancies; (iii) the abilities of this class of catalysts to absorb more hydrogen than necessary for the reduction of the transition metal oxide [ 23,241. As the B sites are the most active sites for hydrogenation of benzene it supports the idea that ‘the active sites are formed by a surface nickel atom at the metal-support interface in close vicinity to an anionic vacancy on the support’ proposed by Herrmann et al. [ 251 for Ni/CeO, catalysts, and developed by Le Normand et al. [ 261 in the case of Pd/CeO, catalysts. This model also allows us to explain why the catalysts based on nickel and ceria can contain more hydrogen than nickel catalysts supported on more classical supports such as SiO, or A1203, or even Laz03 [ 181 for which the reaction (2) of the reduction process does not occur. It could also be taken into account to explain the metal-support interaction found by several authors [ 27,281. At last, and in a more general way, this model may be extrapolated to numerous

ceo2

-

S solution solid Scheme 2. Model for nickel oxide supported on ceria.

G. Wrobel et al.jAppl. Catal. A lOl(l993)

92

73-93

systems already studied and based on CeOz and metals of Group VIII, such as Pt [29-311, Pd [26], Rh [32,34], Ir [24], and to the same transition metals supported on oxides such as TiOz or ZrO, in which cations are able to have several oxidation states. CONCLUSIONS

XRD, electron microscopy and in situ XPS measurements carried out on bulk hydrogenation Ce-Ni-0 catalysts prepared by a coprecipitation route, provide information which allows us to classify the oxides into two families according to the atomic ratio x=Ni/Ce: (i) the first family, with x< 0.5, corresponds to the existence of a solid solution with incorporation of Ni2+ ions in the CeO, lattice; (ii) in the second family, with x> 0.5, crystallized NiO and solid solution coexist. During reduction, while NiO is completely transformed into nickel metal, only a part of nickel belonging to the solid solution is reduced and the solid solution is the seat of a redox type reaction; this phenomenon leads to the creation in the bulk and at the surface of numerous anionic vacancies able to receive hydrogen, probably in an hydridic form. These results, associated with the linear relationship between the activity in benzene hydrogenation and the hydrogen content reported in Part I [ 11, lead us to propose three types of active sites: (i) sites A, where nickel arising from the solid solution is associated with hydrogen contained in the solid solution; (ii) sites B, corresponding to nickel metal issuing from free NiO in interaction with hydrogen of the solid solution; (iii) sites C, where the nickel species do not interact with the solid solution. When the two crystalline phases are present the synergetic effect observed depends on the quality of the interaction between the two types of particles and is attributed to the B sites. Finally, the best efficiency obtained for the catalyst with x= 5, and the Ni/ CeO, catalyst behaviour, are explained by means of the same concepts.

REFERENCES M.P. Sohier, G. Wrobel, J.P. Bonnelle and J.P. Marcq, Appl. Catal., 84 (1992) 169. W.J. Carter, G.K. Schweitzer and T.A. Carlson, J. Electr. Spectrosc. Relat. Phenom., 5 (1974) 827. A. Kotani, T. Jo and J.C. Parlebas, Adv. Phys., 37 (1988) 37. F. Le Normand, J. El Fallah, L. Hilaire, P. Legare, A. Kotani and J.C. Parlebas, Solid State Commun., 71 (1989) 885. G. Praline, B.E. Koel, R.L. Hance, H.I. Lee and J.M. White, J. Electr. Spectrosc. Relat. Phenom., 21 (1980) 17. S. Hufner, Solid State Commun., 49 (1984) 1177. B.W. Veal and A. Paul&as, Phys. Rev. Lett., 51 (1983) 1995; Phys. Rev. B, 31 (1985) 5399.

G. Wrobel et al./Appl. Catal. A lOl(l993)

73-93

93

8 M. Lenglet, A. D’Huysser and C.K. Jorgensen, Inorganica Chimica Acta, 133 (1987) 61. 9

10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34

M. Lenglet and A. D’Huysser, C.R. Acad. Sci. Paris, 310 (1990) 483. C.P. Li, A. Proctor and D.M. Hercules, Appl. Spectr., 38 (1984) 880. M. Lenglet, A. D’Huysser, J.P. Bonnelle, J. Durr and C.K. Jorgensen, Chem. Phys. Lett., 136 (1987) 478. M. Lenglet, A. D’Huysser, J. Arsene, J.P. Bonnelle and C.K. Jorgensen, J. Phys. C Solid State Phys., 19 (1986) L363. J.Z. Shyu, K. Otto, W.L.H. Watkins, G.W. Graham, R.K. Belitz and H.S. Gandhi, J. Catal., 114 (1988) 23. J. Less Common Met., Vol73 and 74 (1990). W.E. Wallace, in A.F. Andresen and A.J. Moeland (Editors), Hydrides energy storage, Proc. Int. Symp., Pergamon, Oxford, 1977, p. 501. J. Barrault, D. Duprez, A. Guilleminot, A. Percheron-Guegan and J.C. Achard, Appl. Catal., 5 (1983) 99. J. Barrault, A. Alouche, A. Chafik, V. Paul-Boncour and S. Probst, Proc. 9th. Int. Congres Catal., 4 (1988) 642. J. Barrault, A. Alouche, V. Paul-Boncour, L. Hilaire and A. Percheron-Guegan, Appl. Catal., 46 (1989) 269. J. Barrault and A. Alouche, Appl. Catal., 58 (1990) 255. J.C. Klein and D.M. Hercules, Anal. Chem., 53 (1981) 754. L. Jalowiecki, G. Wrobel, M. Daage and J.P. Bonnelle, J. Catal., 107 (1987) 375. I. Akalay, M.F. Guilleux, J.F. Tempere and D. Delafosse, J. Chem. Sot., Faraday Trans., 83 (1987) 1137. M.D. Mitchell and M.A. Vannice, Ind. Eng. Chem. Fund., 23 (1984) 88. L. Tournayan, N.R. Marcilio and R. Frety, Appl. Catal., 78 (1991) 31. J. Herrmann, E. Ramaroson, J.F. Tempere and M.F. Guilleux, Appl. Catal., 53 (1989) 117. F. Le Normand, J. Barrault, R. Breault, L. Hilaire and A. Kiennemann, J. Phys. Chem., 95 (1991) 257. A. Maubert, G.A. Martin, H. Praliaud and P. Turlier, React. Kinet. Catal. Lett., 24 (1984) 183. P. Turlier, H. Praliaud, P. Moral, G.A. Martin and A. Dalmon, Appl. Catal., 19 (1985) 287. P. Meriaudeau, J.F. Dutel, M. Dufaux and C. Naccache, in B. Imelik et al. (Editors), Metalsupport and metal-additive effects in catalysis, Elsevier, Amsterdam, 1982, p. 95. T. Jin, Y. Zhou, G.J. Mains and J.M. White, J. Phys. Chem., 91 (1987) 5931. F. Le Normand, L. Hilaire, K. Kili, G. Krill and G. Maire, J. Phys. Chem., 92 (1988) 2561. S. Bernal, F.J. Botana, R. Garcia, Z. Kang, M.L. Lopez, M. Pan, F. Ramirez and J.M. Rodriguez-Izquierdo, Catal. Today, 2 (1988) 653. J.G. Nunan, H.J. Robota, M.J. Cohn and S.A. Bradley, J. Catal., 133 (1992) 309. P. Loof, B. Kasemo, S. Anderson and A. Frestad, J. Catal., 130 (1991) 181.