Hydrogenation catalysts based on nickel and rare earths oxides

Hydrogenation catalysts based on nickel and rare earths oxides

169 Applied Catalysis A: General, 84 (1992) 169-186 Elsevier Science Publishers B.V., Amsterdam APCAT 2260 Hydrogenation catalysts based on nickel...

1MB Sizes 5 Downloads 26 Views

169

Applied Catalysis A: General, 84 (1992) 169-186 Elsevier Science Publishers B.V., Amsterdam

APCAT

2260

Hydrogenation catalysts based on nickel and rare earths oxides I. Relation between cations nature, preparation route, hydrogen content and catalytic activity M.P. Sohier, G. Wrobel and J.P. Bonnelle Laboratoire de Catalyse Hkte’rogkze et Homo&e, URA CNRS No. 402, Universitk des Sciences et Technologies de Lille Flundre Artois, 59655 Villeneuue d’Ascq Ceden (France)

and J.P. Marcq B.P. France, B.P. 4519,593&U Dunkerque Cedex 1 (France) (Received 10 February 1992)

Abstract Several methods of preparation are reported for compounds based on transition metals (Me = Ni,Cu,Co) and rare earth elements (Re=La,Ce,Pr,Nd . ..). The starting materials were intermetallics or nitrate mixtures. In the latter case the precursors of the oxides were obtained by evaporation or coprecipitation with triethylamine or were mechanical mixtures. A ceria-supported nickel catalyst was also prepared for comparison with the first series. Whatever the preparation method, the calcined solids mainly consisted of NiO and La,O, and/or CeO,; in the reduced state all the catalysts were hydrogen reservoirs. From a comparison taking into account specific areas, hydrogen content and catalytic activities in the hydrogenation of unsaturated hydrocarbons, it appears that the coprecipitation route produced compounds with the highest catalytic performances. When cations were varied the most active combination was Ni-Ce and in a series of Ni-Ce-0 catalysts produced by coprecipitation the behaviour was found to be dependent on the Ni/Ce atomic ratio. These results lead us to propose the existence of two types of active sites.

Keywords: catalyst preparation (coprecipitation, evaporation), hydrogen storage, hydrogenation, nickelrare earths, unsaturated hydrocarbons hydrogenation.

INTRODUCTION

Some intermetallic compounds containing rare earths (RE ) and transition metals (Me) have been widely studied for their ability to absorb hydrogen and Correspondence to: Dr. G. Wrobel, Laboratoire de Catalyw H&.&gene No. 402, Universite des Sciences et Technologies Cedex, France.

0926-860X/92/$05.00

et Homogene, URA CNRS de Lille Flandre Artois, 59655 Villeneuve d’Ascq

0 1992 Elsevier Science Publishers B.V.

All rights reserved.

170

M.P. Sohier et al./Appl. Catal. A 84 (1992) 169-186

form hydrides under high pressure (40 to 200 bars) [ 11. Among those solids, compounds such as RENi5 exhibit a significant storage capacity and these materials have therefore been studied as catalysts in many reactions such as carbon monoxide conversion, ammonia synthesis [ 2-71, hydrocarbon hydrogenolysis [ 7,8], dealkylation of aromatic compounds [ 41 and hydrogenation of unsaturated hydrocarbons and carbonyl compounds [ 9-12 1. The influence of oxygen on intermetallic hydrides has been described by several authors. Exposure of LaNi5 to oxygen before or in the course of the hydrogen absorption-desorption cycling facilitates rapid pulverization of the alloy [ 13-141. Depending on the oxidizing conditions, LaNi5 is turned superficially into La( OH), or La,O, with nickel metal or even nickel monoxide [ 15 181. Concerning the REMe, (Me = Fe,Co,Ni) catalysts studied in carbon monoxide conversion, Shamsi and Wallace reported that the converted material rather than the original alloy appears to be the active catalyst [ 191. In our laboratory compounds obtained from a simple oxidation reduction treatment of RENi, are shown to be hydrogen reservoirs which are as good as corresponding intermetallic hydrides [ 201. Hydrogenation of isoprene and benzene at atmospheric pressure and hydrogenation of naphtalene at high pressure were used to study catalytic behaviour. Based on these features, other preparation procedures are undertaken in this work. The compounds obtained are compared with the aim of improving both their ability to store hydrogen and their catalytic performance. Although this paper is mainly concerned with nickel catalysts, other transition metals such as cobalt and copper, which play an important role in hydrogenation reactions, are also considered. EXPERIMENTAL

Preparation procedures Three main preparation routes were used (Scheme 1) depending on whether the initial step was: (i) hydridation of metallic alloys RENi5 (series A); (ii) evaporation and decomposition of an aqueous solution of rare earth and nickel nitrates (series B); (iii) coprecipitation of rare earth and transition metal hydroxides (series C ) . In order to complete the comparison, nickel supported on a ceria catalyst and a fourth group of materials (series D ) obtained by mixing pure nickel and rare earth hydroxides or oxides produced in series C, were also considered. Except for some series A samples, all the precursors were submitted to a further decomposition in a flow of dried air at 723 K for 4 h; finally they were reduced in hydrogen at 573 K for 16 h.

M.P. Sohier et al./Appl. Catal. A 84 (1992) 169-186

171

Scheme 1. Me = Ni, Co, Cu.

Series A The starting materials were rare earth intermetallic alloys: LaNi5, MmNi5 (Mm: mishmetal), CFMmN& (CFMm: cerium free mishmetal) produced by the HY-STOR Company, obtained by induction melting of pure components under vacuum. The composition of the rare earths based on Mm in weight percent was: Mm: 48% Ce; 33% La; 13% Nd; 4% Pr; 1.5% other rare earths. CFMm: 66% La; 25% Nd; 8% Pr; 1% other rare earths. For the materials in series A, the alloys were submitted to hydriding-dehydriding cycles, the conditions of absorption and desorption which are related to thermodynamic and kinetic factors, being specific for each solid. Oxidation of hydrides was carried out in a flow of dried synthetic air for a period of 20 h. Two types of oxides were produced depending on the calcination temperature. Materials produced at 293 K (denoted A I) are “partially oxidized” and materials produced at 703 K (denoted A II) are “fully oxidized”.

Series B The precursors of series B were produced by evaporation at about 373 K and decomposition at 703 K of an aqueous solution containing rare earth nitrates and nickel nitrate. This solution can be obtained in two ways: either by dissolving the original metallic alloy in nitric acid, or dissolving the rare earth nitrate and nickel nitrate (rectapur grade from Prolabo) in the desired proportions in water.

172

M.P. Sohier et aC/Appl. Catal. A 84 (1992) 169-186

Series C The series C solids were prepared by coprecipitation of rare earth and transition metal hydroxides using triethylamine (TEA) (normapur grade from Prolabo) starting from an aqueous solution of nitrates with constant concentration (285 g 1-l). The precipitation conditions were established from preliminary studies of pH changes during the precipitation of hydroxides for each cation (Fig. 1) . Thereafter the procedure used can be described as follows. The solution of nitrates was added dropwise to an excess of TEA constantly agitated by a magnetic stirrer so that each drop precipitated immediately forming a homogeneous mixture of hydroxides. Filtration was performed using a glass filter, and the precipitate was washed first with ethanol or methanol to eliminate the remaining TEA, and finally with doubly-distilled water. The compound was dried in an oven for 15 h, crushed in an agate mortar and calcined and reduced as described previously. Chemical analysis of the series C oxides was performed by the CNRS Microanalysis Centre using atomic absorption spectroscopy. Series D The catalysts in series D were produced from the series C pure hydroxides (Dn) or oxides (Do). Mixtures of the series C materials were combined in the

4o*d

4

6 8 10 12 Vol TEA (cd)

14

16

18

60’

2

4 Vol TEA (cm”)

6

Fig. 1. pH evolution during the precipitation of hydroxides with triethylamine. The titration is performed on pure nitrates solutions (285 g 1-l). (a) Me hydroxides, (0) Ni(N03)2, (A) Cu(NOA, (u) Co(NO,L. (b) RE hydroxides, (0) Ce(NO&, (ml La(NO&.

M.P. Sohier et al./Appl. Catal. A 84 (1992) 169-186

173

desired proportions in a Retch grinder (for between 1 and 3 h) before being calcined and reduced, or simply reduced. Ceria-supported nickel catalyst Ceria-supported nickel was prepared according to the classical wet impregnation method: ceria arising from the series C was stirred with a nickel nitrate solution so that the surface of the catalyst could be expected to be covered by one nickel atom per 0.1 nm’ to achieve a monolayer loading. The overall atomic ratio Ni/Ce was calculated to be 0.19. Thermogravimetric analysis (TGA) The TGA experiments were carried out with a Sartorius 4102 microbalance, equipped with a flow-gas system, which has been previously described [ 211. Specific areas @A) measurements The measurements were performed by nitrogen or argon adsorption according to the BET method on samples degassed at 393 K under vacuum. For the reduced samples, SA were obtained with an apparatus equipped with a flow system which enabled a reduction treatment before adsorption. X-ray diffraction (XBD) analysis Preliminary analysis was conducted by XRD. Films were obtained from a PW 1008 Philips generator equipped with a Debye and Scherrer cell using a copper anode with a nickel filter. Hydrogen contents Measurement of the stored hydrogen was based on the ability of this class of solids to hydrogenate isoprene (2-methyl-1,3-butadiene) in the absence of gaseous hydrogen, at atmospheric pressure, according to a method already described [ 221. All of the hydrogenation reactions were carried out in an all-glass greasefree flow unit [ 231. From 50 to 100 mg of sample were loaded in an isothermal reactor. After reduction of the oxide precursor, the catalyst was flushed with helium (purity 99.9995% from Air Liquide) for 15 min in order to eliminate gaseous hydrogen remaining in the whole glass flow unit. This step was checked using a thermal conductivity detector. After cooling to 423 K isoprene (Fluka, more than 99% pure) was introduced into the helium flow line, at constant pressure (5 Torr) obtained by distillation in a subambient trap. The product distribution was determined with a gas chromatograph, equipped with a flame ionisation detector and a capillary column (squalanne 0.2 mm x 100 m) operated at 318 K.

174

M.P. Sohier et al./Appl. Catal. A 84 (1992) 169-186

Catalytic activity measurements The same unit was used to measure the catalytic activities in hydrogenation reactions at atmospheric pressure: (i)isoprene hydrogenation (isoprene, 20 Torr, in pure hydrogen, 2.6 1h-l at 298 K); (ii) benzene hydrogenation (benzene, Merck spectra quality; 40 Torr, in pure hydrogen, 2.6 1 h-’ at 373 K). In both cases the catalyst sample (about 50 mg) was reduced at 573 K for 16 h, then cooled to 373 K under hydrogen before reaction. The reaction products were analyzed by gas chromatography. Naphtalene hydrogenation was performed using a fixed-bed, trickle-phase pilot plant at high pressure (80 to 100 bars) and various temperatures (373 to 623 K) with catalysts (16 to 24 g) previously activated at 623 K under hydrogen at 150 atm for 20 h. The feed composition was 2% of reactant in n-heptane. RESULTS

Chemical composition and structure of solids The use of intermetallic alloys as starting materials fixes the atomic ratio Ni/Re as 5, but the amount of oxygen contained in the A samples is dependent on the calcination procedure. The stoichiometries of A I and A II samples can be determined by thermogravimetric analysis. For example Fig. 2 shows the weight increase during the oxidation of the intermetallic hydrides. The nickel and rare earths are only partially oxidized in the A I samples (L~N&Nw, MmNiJ&,, CFMmNi,O,,,). These elements are in their highest oxidation states in the A II samples (LaNi,O,,, MmNi506.76,CFMmNi,Os.,,). These stoichiometries have been established by TGA of the samples in their final oxidized states and by XRD. Indeed at low temperature around 300 K, only a superficial oxidation occurs and the original hydride keeps its crystallographic structure [ 241. Each grain is composed of a bulk of metallic alloy covered with a layer of amorphous oxides; this has been proven by X-ray photoelectron spectroscopy (XPS ) analysis which shows the presence in the surface layer of La3+ ions in the case of LaNi, and Ce3+ and Ce4+ in MmNi,. For calcination temperatures above 703 K, the initial structure is destroyed and the compound is made of a mixture of nickel oxide and rare earth oxides with cations in their highest oxidation states. These results agree with features reported by several authors [ 15-181: according to the temperature of treatment and the oxygen partial pressure, Laz03 or La (OH), or CeO, are detected in the surface layers of intermetallic alloys between nickel and lanthanum or cerium. The samples Ce-Ni belonging to the series C are noted C, (x: atomic ratio Ni/Ce). The percentage of each element was checked experimentally in order to assess the quality and the reliability of the preparation process. The maxi-

M.P. Sohier et al./Appl. Catal. A 84 (1992) 169-186 WEIGHT % 1c.l

175

,-,N’

i

:

Fig. 2. Weight evolution with temperature during the oxidizing treatment of LaNi, intermetallic compound.

mum error between the experimentally determined composition and the expected values is less than 6%. According to the X-ray data, the C, samples are made up of a mixture of NiO and CeO, in the oxidized state except for x < 0.5 where NiO is not detected. For x > 0.5 NiO is mainly reduced in nickel metal during the activation process and CeO, appears unchanged; here the presence of small amounts of NiO is probably due to the pyrophoric character of this kind of catalyst. The same results are obtained for series B and D. In order to compare the different preparation procedures, three experimental measurements are taken into account: (i) the specific area of the oxides and of some materials called oxhydrides, that is the solids obtained after the reduction treatment previously described; (ii) the hydrogen content of the oxhydrides; (iii) the activity of the catalysts in the isoprene or benzene hydrogenation. Surface areas The surface areas of the solids with ratio Me/RE = 5 are reported in Table 1.

The results demonstrate: (i) the superiority of the coprecipitation route in

M.P. Sohier et al./Appl. Catal. A 84 (1992) 169-186

176 TABLE 1

Effect of the preparation route of RE Me6 compounds on surface areas, hydrogen content and activity in benzene hydrogenation Sample

LaNi, A I MmNi, A I LaNi, A II MmNi, A II CFMmNi,A I CFMm Nir, A II MmN& B CeNi, B LaNi, C CeNi, C CeNi, Do CeNi, Du CeCu, C CeCo, C

Surface Area (m2 g-l) Oxide

Oxhydride

1.1 0.7 4.2 2.2 0.9 3.5 17.6 18.5 44.0 49.0 35.3 33.1 13.0 46.0

1.1 0.7 8.0 48.0 0.9 7.5

65.0 43.0 48.0

H/RE

0.04 0.04 0.3 8.0 0.4 2.1 4.3 4.2 6.7 5.2 6.1 #O 0.1

Qn-lo3 mol H (g oxhydride)-’

an-lo3 mol (g oxhydride)-’

0.1 0.1 0.65 18.0

1.4

0.8 4.7 9.15 9.1 14.8 11.0 12.9 #O 0.2

5.3 5.4 33.6 9.9 12.0 5.3 9.8

producing the largest surface areas in the oxide state; (ii) an increase of the surface area during the reduction treatment except in the case of the series A I. This effect is particularly important for MmNi, A II. Among the other CeMe, compounds prepared by route C, it is noted that the surface area of CeCu, is particularly low. The surface areas of the C, oxides are reported in Table 2. As x increases, the specific area decreases slightly, it then remains nearly constant and higher than that of pure nickel monoxide. A surprisingly high value is obtained for CZ. On the other hand during the reduction treatment, the specific areas of the pure oxides NiO or Ce02 decrease counter to that of CeNi oxides. Hydrogen content Titration of stored hydrogen as described in the experimental part has been used to study Cu-Cr-0 and Cu-Al-0 catalysts [ 251 and more recently MO&based systems [ 26,271. This method is appropriate to access both surface and bulk hydrogen introduced in the solid during the reduction step. The amounts of extracted hydrogen are obtained by integrating the curve of the activity in isoprene hydrogenation versus time between t = 0 (time when isoprene is introduced in the helium flow) and tf (time when the activity is zero) [ 221.

177

M.P. Sohier et al./Appl. Catal. A 84 (1992) 169-186 TABLE 2

Evolution of the surface area and hydrogen content with the relative proportions of nickel and cerium in series C Sample

co

C 0.05

C0.2 C0.5

Cl G Cd G C5.5

G G Ni Ni/CeOz”

Surface Area ( m2 gg ’ ) Oxide

Oxhydride

66.2 49.6 52.0 31.2 40.0 89.0 49.9 49.0 58.3 53.0 61.8 25.0 45.8

35.2 50.2 51.5 38.4 57.4 93.3 76.6 65.5 72.7 64.2 87.7 16.0 58.4

H/RE

0

0.32 0.84 0.87 1.28 2.5 4.6 6.7 8.3 7.3 9.4

“rhe supported nickel is included here for comparison with the bulk oxides catalysts.

Numerous precautions were taken to determine the hydrogen content. To ascertain the validity of the method and the absence of parasitic reactions two techniques were used: thermogravimetry and deuterium tracing. In the latter case the oxide was reduced with gaseous deuterium and the distribution of deuterium in products after reaction with isoprene was established. The only products detected were isoprene and C,-monohydrogenated hydrocarbons which contain on average two atoms of deuterium per unsaturated bond having reacted. This means that no self-hydrogenation or polymerization or loss of hydrogen in the gas phase occurs. The same results were obtained by thermogravimetry and it was verified that no hydrogen was lost after reduction when the catalyst was submitted to a pure helium flow at 573 K. Moreover, during the consumption of hydrogen by isoprene, no significant increase in sample weight was evidenced, excluding carbon deposition or polymerization reactions. The amounts of hydrogen removed from the partially reduced oxides are compiled in Tables 1 and 2. They are described in terms of H/RE ratio or mols of hydrogen atoms per mol of oxhydrides or hydride (Qn ) . As the same results are obtained after a second reduction treatment in similar conditions, one can conclude that most of the oxhydrides behave like reversible hydrogen reservoirs whose content can be titrated by isoprene hydrogenation. From Table 1,the hydrogen content of the A I samples is low and quite independent of the nature of the rare earths. In contrast, the A II oxhydrides show a higher hydrogen storage capacity. In fact intermetallic hydrides ob-

178

M.P. Sohier et al./Appl. Catal. A 84 (1992) 169-186

tained industrially after several hydriding-dehydriding runs at pressures from 40 to 200 bars, have a similar ability to store hydrogen. For example it is possible to introduce similar amounts of hydrogen in the MmN& by the A II method and by the industrial hydriding-dehydriding route (6.7 instead of 8 atoms of hydrogen by MmNi5 unit). Otherwise the results of Table 1 point out: (i) The influence of the preparation method on oxhydrides containing nickel and cerium with an atomic ratio Ni/RE=5 as shown by classifying these solids according to decreasing hydrogen content: MmNi, A II > C > Du > Do > B. (ii) The influence of the nature of the cations for the Me/RE = 5 samples. One can conclude that, among the rare earths cerium plays the principal role, and nickel is reasonably essential for the production of high capacity hydrogen reservoirs. The highest hydrogen contents are obtained in series C for the samples C5_7 (Fig. 3a). It is worth noting that no hydrogen could be extracted from CeO, (Co) and that pure nickel is a relatively good hydrogen reservoir (H/Ni = 0.8). The hydrogen content of Ni/CeO, is also reported in Fig. 3a; its position relative to the curve shows that it behaves somewhat like the oxide C,,, in storing hydrogen. 20

7 15 a ‘C n Z :: s I 10 = E

t

3

t

.

5’-

1 co2

I

,’

,----%___/

J 0

0.2 0.4 0.5 0.0 Atomic ratio Ni/NI+Ce

1

co

,I

1 ?

Cl

.

.U)5 I

1

/

0.4 0.6 0.5 Atomic ratio NI/NI+Ce

i 1

Fig. 3. (a) Variation of hydrogen capacity with nickel content in the C, catalysts. (b) Variation of the catalytic activity in benzene hydrogenation with the nickel content in the C, oxhydrides. (B) CL (A ) Ni/Ce02.

179

M.P. Sohier et al./Appl. Catal. A 84 (1992) 169-186

Catalytic activities Isoprene hydrogenation This reaction is able to differentiate the catalytic behaviour of the A I and A II samples, the A II series based on mishmetal being the most active, achieving total conversion into isopentane at 293 K. But as the B, C, and D samples also approach this level of transformation, another reaction was chosen to classify the catalytic abilities of the solids. Benzene hydrogenation Benzene was chosen since hydrogenation of the aromatic ring is more difficult than hydrogenation of conjugated dienes. A stable stationary state is reached after 30-50 min and the only product obtained is cyclohexane. The catalytic activities ( aH) reported in Tables 1 and 2 are related to this stationary state and are given with an accuracy of within 5%. As mentioned previously, the effects of the preparation method, the nature of the cations and the respective proportions of nickel and cerium are considered. In addition the influence of the crushing time for the series D materials is reported in Table 3. A survey of the an values presented in Table 1 points out the high activity of the Ce-Ni compounds belonging to series C for the benzene hydrogenation. The classification is: C > Du > Do > B > A II. Now comparing catalysts with the same relative content in transition metal and rare earth elements prepared according to method C, one can say that nickel remains the best among the transition metals and cerium is still superior to lanthanum. Finally the specific activity curve referred to the oxhydrides versus the nickel content for the C catalysts (Fig. 3b), shows the efficiency of the Ni-Ce assoTABLE 3 Influence of the crushing time and the starting materials on hydrogen content and catalytic activity in benzene hydrogenation for series D Sample

CeNi,,, D,

Crushing time (h)

QH*103mol H (g oxhydride)-’

aH*103mol (g oxhydride)-’

Experimental

Calculated

Experimental

Calculated

1 2 3

12.3 12.3 12.7

5.2

6.0 7.3 7.7

6.0

1

9.0

5.1 5.2

4.5

CeNi, Do 2

7.1

6.1

0

t

_

-\

.\..

I

+__~_ .--

200400eoosa,

\

-~.-~

__

1

_

--

--

,

a--,

1.ooo

_

._~__

Ttme in min

I

-. ----.

._

_ .._...

..-..

1200

I

1.400

-7

_ _ ----

_~~_..

__-___

1.m

._

a

0

5

10

0

.._. _

---.

-

9.ooo

.._._ .__._._ _..

Ttme in min

2ow

_. __._.._.._..._.__

. \ . .

l.WO

I

15

20

25

b

4.m

5.m

__._._ _... _.____ .._. .-...___.

Fig. 4. Effect of the oxidizing treatment on the naphtalene hydrogenation for the MmN& oxhydride. T=373 K; pH2= 150 atm. (a) MmNi, A I. (b) MmNi, A II. (0 ) Decalin, (0 ) tetralin, (A ) total.

0

1

2

3

4

6

61

_~____~

% Conversion

7

% conversion 30

M.P. Sohier et al./Appl. Catal. A 84 (1992) 169-186

181

ciation and the superiority of the C, catalyst. Pure CeO, is completely inactive and Ni/CeO, slightly superior to C,,,. High-pressure hydrogenation The catalytic abilities of series A have been investigated through naphtalene hydrogenation. The A II oxhydrides are four times better than the corresponding A I ones (Fig. 4). From a kinetic viewpoint, an important fact is the reaction order versus the hydrogen pressure. Indeed this pressure had no effect with oxhydrides while with classical nickel-supported catalysts it led to an order of 2. Moreover the A II catalysts give selective conversion of naphtalene into decaline (Fig. 4b). These results confirm the enhancement of numerous hydrogenation reactions when the precursors are completely oxidized. DISCUSSION

Hydrogen storage analysis In this part we propose to comment and discuss the Qn values as deduced from our method and to compare them with other values drawn from the literature. It is clear, even for the oxhydrides which contain the lowest amounts of hydrogen, that Qn is superior to the superficial adsorption capacity of the solids. For example an estimation assuming that hydrogen is only located at the surface and that one atom of hydrogen occupies 0.1 nm’, leads respectively to a H/RE ratio equal to 0.15 and 0.6 compared with 0.32 and 6.7 for C,,, and C, samples. Then all the studied oxhydrides behave as hydrogen reservoirs and a simple relationship between SA values and Qn is not expected. This point may seem surprising but it is supported by some features reported in the literature concerning the behaviour towards hydrogen of CeO, or ceriasupported metals. For example after temperature-programmed reduction (TPR) experiments Barrault et al. [28] observed that the hydrogen desorption capacity was higher than the quantity adsorbed at room temperature. Two types of hydrogen were evidenced by Cunningham et al. [ 291 by H NMR on a Rh/Ce02 catalyst; the authors concluded the existence of a hydrogen species bound to the support. More recently Tournayan et al. [30] and Frety et al. [ 311pointed out the particular role of CeO, in comparison with the more classical supports such as A1203, MgO, SiO,-Al,O, in supported iridium catalysts. They discussed the hydrogen storage abilities of Ir/CeO, in terms of strong metal-support interaction (SMSI) or hydrogen spill-over phenomenon. However the quantities of desorbed hydrogen are always inferior (by at least one magnitude order) to ours if one compares them with the value found for the Ni/Ce02 catalyst. Pure Ce02 is also able to incorporate hydrogen in the bulk as is demonstrated by thermogravimetric, H NMR, IR and electron spin resonance (ESR) tech-

182

M.P. Sohier et al./Appl. Catal. A 84 (1992) 169-186

niques [32]. The highest value is observed for CeOz reduced at 673 K; it corresponds to 3.4*102’ H g-’ or H/Ce=O.l. Now we will discuss why it is impossible to extract hydrogen by isoprene from CeO,, and why the values obtained for similar compounds are higher using our method than by using temperature-programmed methods (TPM). The mechanism of extraction of the occluded hydrogen established previously for the Cu-Cr-0 systems [22] supposes that: (i) an hydrogenating site is present at the surface of the catalyst; (ii) the hydrogen species can diffuse through the reduced compound (hydride or oxhydride) lattice; (iii) the regeneration of the active site can occur. For pure Ce02 the first condition is not satisfied since it is completely inactive in hydrocarbon hydrogenation. Therefore although it may contain hydrogen this method cannot be applied to this oxide. In contrast, as soon as some nickel is associated to CeO,, the amounts of hydrogen extracted by isoprene become more important than those deduced from TPM. Similar trends have been evidenced on Cu-Zn-Al-O catalysts [ 331. Several explanations can be advanced: (i) Extraction by isoprene is carried out at low temperatures. At higher temperatures a part of the hydrogen species can be modified by reaction with oxygen inside the solid. (ii) The conditions of reduction are different: generally 1% hydrogen in an inert gas in TPR compared to a treatment under pure hydrogen in our case. Then the two methods do not give the same information. The main point is that the method we used permits us to compare the different catalysts provided that the reduction step follows an oxidizing treatment at sufficiently high temperature. By this method the structure is almost similar for all the series and is constituted by a mixture of rare earth oxide and nickel monoxide. If this is not the case the role of the structure can be important. This is particularly clear when comparing series A I and A II. The intermetallic hydrides are known to be good hydrogen stores; however, very small amounts of hydrogen can be extracted by isoprene from these compounds when they are oxidized at room temperature. Under these conditions the superficial layer, alone, is slightly oxidized and the structure based on the CaCu, hexagonal model is maintained. A consequence of this is the extremely difficult diffusion of hydrogen in the lattice. Importance of the Ce-Ni association

A survey of all the results obtained from the A II and C oxhydrides points out the peculiarity of the Ni-Ce combination whatever the parameter studied. Indeed in the series A II, CFMmNi5 which does not contain any cerium, behaves somewhat as LaNi,: both materials have the same specific area and the same hydrogen capacity which are somewhat lower than the corresponding values for MmNi,. In series C preparations the phenomenon is also quite clear,

M.P. Sohier et al./Appl. Cad

A 84 (1992) 169-186

183

comparing lanthanum and cerium on the one hand and nickel, cobalt and copper on the other hand. In this case the peculiarity of the Ni-Ce combination could be partly explained by reference to the curves showing variations of pH during precipitation of hydroxides. From Fig. 1 one can see that the only associations able to provide a good coprecipitation are the Ce-Ni and Ce-Co ones, since the pH corresponding to precipitation of the hydroxides is nearly the same (7.7,8.1 and 7.9). Despite this, CeCo, is a less efficient hydrogenation catalyst and a much poorer hydrogen reservoir than CeNi,, probably because of the chemical difference between the two elements, cobalt and nickel. Now a comparison of the results obtained with the different Ce-Ni catalysts shows the importance of the preparation process on the degree of interaction occurring between the species NiO or nickel and CeO, detected by XRD analysis, and the influence of the Ni/Ce atomic ratio. This feature is supported by the SA, Qn and on values (Table 2, Fig. 3) within series C, and by the results obtained throughout the different series with Ni/Ce = 5 (Table 1). In series D it is obvious that the interaction between the two hydroxides is more efficient than that between the two oxides as demonstrated from the effect of the crushing time on UH(Table 3). This phenomenon is less marked on Qn; however a calculation made by assuming a simple mechanical mixture leads to lower values than experimental ones. Likewise comparison between the intrinsic activities of C5 and CeNi, Do (0.43*10m3 and 0.19*10-3 mol h-l mm2respectively) points out clearly that surface effects alone are unable to explain the results. The better performance of the series C can be explained by the formation of a mixed compound which occurs from the precipitation step under accurate experimental conditions. This compound can be a hydrated hydroxy-salt, or a salt constituted by one cation and a complex anion of the other element leading to solid solutions during further treatments. For the mechanical mixtures the situation is more complex. Is the observed increase of Qn and on due to the migration of part of the Ni2+ ions on the CeO, matrix during crushing, or to contact between the pure hydroxides or oxides which permits a hydrogen spillover phenomenon between the compounds after reduction as is proposed by Delmon [ 341 in terms of the “remote control” model? Additional information on the surface state are necessary to answer this question. In series B the poorer results can be attributed either to the weaker interaction between nickel and cerium or to the smaller values of the surface areas due to the crystals growth during evaporation. Relation between aH and QH From Fig. 5 where an as a function of Qn is reported for series C, one can classify the solids into three zones according to x: (i)O < x < 0.5: the on values are nearly proportional to the Qn values and the increase in activity is weak. (ii) 0.5 < x:< 5: a roughly linear relationship holds between on and QH, but in

M.P. Sohier et al./Appl. Catal. A 84 (1992) 169-186

184

0

2

4

YOOO nk C&x

10 12 14 ( g. oxhydride IV’

16

18

Fig. 5. Variation of the catalytic activity with hydrogen content in the C, samples. (W) C,, (A ) Ni/CeO,.

this case hydrogen concurs to a strong increase of on, the best catalyst being C,. (iii) x> 5: no simple relation is evidenced in this zone. However, these solids remain better catalysts than pure nickel. Fig. 5 and information about the structure of the catalysts suggest the existence of two types of active sites. In the first zone the catalyst is able to retain hydrogen but the active sites are not numerous enough. This corresponds to the absence of nickel monoxide in the oxide state. As will be shown in our next paper [ 351, Ni2+ ions remain partly incorporated in the CeO,-like lattice in a reduced state and Ce4+ is slightly reduced in Ce3+. The presence of Ce3+ has been also reported by Barrault et al. [28] on Ni/Ce02 catalysts and Le Normand et al. [ 81 on the oxidized intermetallic compounds CePd,. When x > 0.5 the increase of activity is probably due to nickel metal coming from free nickel monoxide in the oxide precursor. But the simultaneous presence of two types of sites must be an important factor and may help to explain the difference in behaviour between cerium based catalysts, and catalysts based on other rare earths. CONCLUSIONS

All the solids based on rare earths and nickel behave as hydrogen reservoirs provided that they are oxidized at sufficiently high temperatures. Yet among the different preparation procedures - from intermetallics, coprecipitation of hydroxides, mechanical mixture of pure oxides or hydroxides or evaporation of nitrates - the REN& oxhydride obtained from coprecipitated hydroxide is the most efficient. Among the rare earths, cerium plays a particular role, and all of the work points to a beneficial association between the species based on cerium and nickel both in the oxidized and reduced states. Where the nickel content of catalysts was varied in the coprecipitation route C, changes seen in

M.P. Sohier et al./Appl. Catal. A 84 (1992) 169-186

185

hydrogenation with hydrogen content suggest the existence of two types of catalytic sites. In order to achieve a deeper understanding of the catalytic behaviour of these solids, a detailed study of the physical chemistry of the Ce-Ni-O-H system will be reported in a second paper (which will also include a model for the active sites).

REFERENCES

8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31

J. Less Common Met., 73,74 (1980). V.T. Coon, T. Takeshita, W.E. Wallace and R.S. Craig, J. Phya. Chem., 80 (1976) 1878. T. Takeshita, W.E. Wallace and R.S. Craig, J. Cat&, 44 (1976) 236. A. Allatar, T. Takeshita, W.E. Wallace and R.S. Craig, Science, 80 (1977) 1093. G.B. Atkinson and L.J. Nicks, J. CataI., 46 (1977) 417. W.E. Wallace, Chem. Tech., (1982) 752. J. Barrauh, D. Duprez, A. Guilleminot, A. Percheron-Guegan and J.C. Achard, Appl. Catal., 5 (1983) 99. F. Le Normand, P. Girard, L. H&ire, M.F. Ravet, G. KriU and G. Maire, J. Catal., 89 (1984) 1. K. Soga, M. Imamura and S. Ikeda, Chem. L&t., (1976) 1387. K. Soga, H. Imamura and S. Ikeda, J. Phys. Chem., 81 (1977) 1762. K. Soga, H. Imamura and S. Ikeda, J. Catal., 56 (1979) 119. T. Imamoto and M. Yokoyama, J. Chem. Sot., Chem. Commun., (1984) 163. H. Uchida, H. Uchida and Y.C. Huang, J. Less Common Met., 101 (1984) 459. R.A. Andrievskii, O.N. Goroshchenko and D.N. Kazakov, Izv. Akad. Nauk SSSR, Neorg. Mat., 22 (1986) 237. L. Schlapbach, Sol. St. Commun., 38 (1981) 117. K. Terao, H. Uchida and Y.C. Huang, Proc. Res. Inst. for Industrial Science, Tokai University, 18 (1987) 23. T. Shirotsuka, K. Onoe and A. Yokoyama, J. Chem. Eng. Jap., 19 (1986) 347. H. Uchida and M. Ozawa, Z. Phys. Chem. N. Folge, 88 (1986) 715. A. Shamsi and W.E. Wallace, Ind. Eng. Chem., Proc. Res. Dev. 22 (1983) 582. H. Diaz, J.P. Marcq, M. Pinabiau and Y. Barbaux, European Patent 8 508 210 to B.P. France (1986). G. Wrobel, A. d’Huysser and J.P. Bonnehe, Nouv. J. Chim., 8 (1984) 291. L. Jalowiecki, M. Daage and J.P. Bonnelle, Appl. Catal., 16 (1985) 1. F. Garin and F.G. Gauit, J. Am. Chem. Sot., 97 (1975) 4466. M. Pinabiau, Thesis, LiIle (1987). L. JaIowiecki, G. Wrobel, M. Daage and J.P. Bonnelle, J. CataI., 107 (1987) 375. L. Jalowiecki, A. Aboulaz, S. Kasztelan, J. Grimblot and J.P. BonneIIe, J. Catal., 120 (1989) 108. L. Jalowiecki, J. Grimblot and J.P. Bonnelle, J. Catal., 126 (1990) 101. J. Barrault, A. Alouche, V. Paul-Boncour, L. Hilaire and A. Percheron-Guegan, Appl. Catal., 46 (1989) 279. J. Cunningham, S. O’Brien, J. Sanz, J.M. Rojo, J.A. Soria and J.L.G. Fierro, J. Mol. Catal., 57 (1990) 379. L. Tournayan, N.R. Marcilio and R. Frety, Appl. Catal., 78 (1991) 31. R. Frety, P.N. Da Silva and M. Guenin, Appl. Catal., 58 (1990) 175.

186 32 33 34 35

M.P. Sohier et al./Appl. Catal. A 84 (1992) 169-186 J.L.G. Fierro, J. Soria, J. Sanz and J.M. Rojo, J. Solid State Chem., 66 (1987) 154. A. Sene, G. Wrobel, J.P. Bonnelle, F. Le Peltier and P. Chaumette, in preparation. B. Delmon, Bull. Sot. Chim. Belg., 88 (1979) 979. M.P. Sohier, A. d’Huysser, G. Wrobel and J.P. Bonnelle, in preparation.