Influence of lanthanide oxides on the catalytic activity of nickel

Influence of lanthanide oxides on the catalytic activity of nickel

Applied Catalysis A: General 232 (2002) 121–128 Influence of lanthanide oxides on the catalytic activity of nickel Andreea Gluhoi, Petru M˘arginean∗ ...

176KB Sizes 0 Downloads 1 Views

Applied Catalysis A: General 232 (2002) 121–128

Influence of lanthanide oxides on the catalytic activity of nickel Andreea Gluhoi, Petru M˘arginean∗ , Dan Lupu, Emil Indrea, Alexandru Radu Biri¸s National Institute for Research and Development of Isotopic and Molecular Technologies, P.O. Box 700, R-3400 Cluj-Napoca, Romania Received 18 June 2001; received in revised form 28 January 2002; accepted 29 January 2002

Abstract The catalytic activity of supported Ni/Ln2 O3 (Ln = La, Ce, Gd, Ho, Yb) catalysts for the hydrogen/water isotopic exchange reaction has been studied. The metallic and total surface areas were measured by hydrogen and krypton adsorption, respectively. The Ni surface area evaluated by hydrogen chemisorption is higher than the total BET surface area for Ni/La2 O3 . The Ni surface area does not change significantly while the total surface area increases with the decreasing Ln3+ ionic radius. The thermodesorption spectra for the adsorbed hydrogen with and without preadsorbed H2 O are reported. The mobility of the adsorbed hydrogen decreases from La2 O3 to Yb2 O3 as support and in the same direction decreases the water uptake normalised to the total surface area. The catalytic activity normalised to the Ni surface area decreases exponentially with the decreasing Ln3+ ionic radii from La2 O3 to Yb2 O3 as support, in close relationship with the basicity and hydration degree of the support oxide. © 2002 Elsevier Science B.V. All rights reserved. Keywords: Hydrogen–water deuterium exchange; Intrinsic activity; Metal-support interaction; Nickel/lanthanide oxides

1. Introduction Metal catalysts are usually deposited on oxide supports in order to enhance the dispersion of the active phase and to improve its resistance in the heat treatment during reduction and catalytic processes. The nature of the support can exert a significant influence on the catalytic properties of the metal for certain reactions, such as the isotope exchange reaction between hydrogen and water vapour [1–4], H2 + HDO  HD + H2 O

(1)

which is of interest for the separation of the hydrogen isotopes [5]. The reaction (1) can also be used as a powerful method to evaluate both the hydrogen mobility on the catalyst surface (useful to evaluate these ∗ Corresponding author. Fax: +40-64-420042. E-mail address: [email protected] (P. M˘arginean).

catalyst in hydrogenation reactions) and the degree of metal-support interaction. As concerning nickel catalysts for reaction (1), it was shown that by supporting the metal on various metal oxides, the reaction rate normalised to metal surface area is strongly affected by the nature of the support [1–4]. Generally, the presence of the support enhances the intrinsic activity (i.e. the activity of the metal surface area unit) of the catalyst compared to that of nickel black. The most efficient supports are TiO2 and ZrO2 , which increase the intrinsic activity by about three orders of magnitude; the least efficient are ZnO and SiO2 , with an increase of only one order of magnitude [4]. The results were interpreted in terms of a model invoking the creation of new highly active sites at the nickel-support interface. This model was proved in the case of the heat-treated Ni/Cr2 O3 samples where it was shown that the increase of the catalytic activity is proportional to the length of the nickel-chromia frontier [6]. The activity enhancement

0926-860X/02/$ – see front matter © 2002 Elsevier Science B.V. All rights reserved. PII: S 0 9 2 6 - 8 6 0 X ( 0 2 ) 0 0 0 9 1 - 1

122

A. Gluhoi et al. / Applied Catalysis A: General 232 (2002) 121–128

is due to the intimate contacts between the metal and support, arising during the preparation process of the catalyst. That is why the simple mechanical mixtures obtained by crushing together the supports with nickel powder show catalytic activities not higher than that of the unsupported nickel [7]. The aim of this paper is to supply additional data for a better understanding of the influence of the metal-support interaction on the catalytic activity: which is the role of the support and in which way the characteristics of the oxide modifies the rate of reaction (1). For this purpose, some rare-earth oxides have been chosen as support for nickel; they have essentially similar chemical properties but there is a decrease of the trivalent rare-earth ionic radius from La3+ to Lu3+ . Another reason for this study lies in the importance of the LnNi5 type alloys as hydride electrodes in Ni/metal hydride batteries. After the activation, the alloy surface consists mainly of nickel clusters and rare-earth oxides [8–14], the same couple (metal/oxide) playing in this case the role of an electrocatalyst for the hydride electrode reversible process. M + H2 O + e −

charge

 MHad + OH−

(2)

discharge

The reactions (1) and (2) have in common the adsorption and mobility of the reacting species (hydrogen and water) on the composite metal/oxide surface layer. For hydride electrodes in commercial Ni/MH batteries, LnNi5 type alloys are largely used so that a better understanding of Ni/various rare-earth oxide catalysts should be useful to obtain Ni/MH batteries with high-rate dischargeabilities.

2. Experimental Five supported nickel samples on different rareearth oxides (La2 O3 , CeO2 , Gd2 O3 , Ho2 O3 , Yb2 O3 ) were prepared by coprecipitation method. A water solution containing a mixture of nickel and lanthanide nitrates was added to a sodium hydroxide solution, with vigorous stirring. The ratio between the salts has been chosen in such a way to obtain 80 at.% Ni and 20 at.% Ln. The resulted precipitate was filtered and

thoroughly washed with doubly distilled water in order to remove the sodium ions down to a very low content (<50 ppm). The resulted precipitates were dried at 105 ◦ C, calcined in a nitrogen stream at 340 ◦ C and reduced in a hydrogen flow at 350 ◦ C. The catalyst was passivated at room temperature, in flowing nitrogen with low oxygen content. The catalyst samples were reactivated prior to use by in situ reduction, in hydrogen flow, at 350 ◦ C. BET surface area, St , of the catalyst samples was measured in an all-glass apparatus, using krypton adsorption at liquid nitrogen temperature. The hydrogen chemisorption was measured in an all-glass apparatus, at room temperature, in the hydrogen pressure range 30–200 Torr. The back extrapolation of the saturation region of the hydrogen adsorption isotherm was done and the intercept at zero pressure was taken as monolayer uptake. The metallic surface area, SNi-chem , was calculated assuming that one hydrogen atom is chemisorbed on each surface metallic nickel atom and each nickel atom occupies 6.5 Å2 . The TPD measurements were performed in an all-metallic set-up. The catalyst sample (about 30–50 mg) was reactivated in hydrogen flow at 350 ◦ C. After hydrogen (dried gas) adsorption for about 1 h, at room temperature, the sample was cooled to about −80 ◦ C and argon was passed through the sample in order to sweep the hydrogen for about 20 min. The thermodesorption was then started at a heating rate of 5 K/min and with an argon flow rate of 25 ml/min. The TPD of hydrogen was monitored by a thermal conductivity detector. In order to evaluate the influence of water co-adsorption on the hydrogen adsorption/desorption by TPD spectroscopy, the same experiments were also made for each sample, using hydrogen saturated with water vapour at the room temperature. The catalytic activity was measured in a flow microreactor system operated at atmospheric pressure. It consists in a glass reactor (5 mm i.d.). The catalyst sample (about 0.1 g) was first re-reduced in situ under hydrogen flow, at 350 ◦ C; after 2 h, the reactor was cooled to the room temperature. The hydrogen flow, containing about 50 ppm deuterium, was bubbled through a heavy water solution, containing about 4 at.% deuterium, and passed through the catalyst bed; 63.5 and 79 ◦ C were selected as saturation and reaction temperatures, respectively. The deuterium content in

A. Gluhoi et al. / Applied Catalysis A: General 232 (2002) 121–128

the hydrogen flow, at the outlet, was measured on-line using a gas chromatograph equipped with a 1.5 m column (activated charcoal) and thermal conductivity detector. The results were plotted according to the appropriate reversible first-order equation in order to obtain the rate constant.

3. Results and discussion 3.1. Surface characteristics The surface area values evaluated by various methods are reported in Table 1, revealing an increase of the total surface area, St , with the increasing atomic number from La to Yb. The evaluation of SNi-XRD must be taken with care because this method allows determination of the average crystallite size and the intercrystallite area is not exposed in fact to the gas chemisorption; moreover, Ni crystallites with the effective average size <20 Å can not be detected by XRD. The examination of the results shows that for Ni/Ln2 O3 , SNi-chem < SNi-TPD and it can be explained by taking into account that in TPD experiments the saturation of the sample with adsorbed hydrogen has been carried out at 1 atm H2 pressure (at −80 ◦ C) while in chemisorption the maximum H2 pressure was not higher than 200 Torr and room temperature, as described in the experimental section. The hydrogen adsorption experiments at H2 pressure up to 3.5 atm reported in Fig. 1 show a typical Langmuir isotherm with a significant adsorption/desorption hysteresis. The adsorption rate is rather slow, requiring at least 3 h to reach the equilibrium values. The Ni

123

surface area evaluated by TPD is higher than that resulting from chemisorption method as shown in Table 1, except for Ni black. This difference results from the fact that the saturation of the catalyst with hydrogen is achieved only at higher pressures than those used here in chemisorption, as shown in Fig. 1. Moreover, our TPD experiment starts from sample saturation at −80 ◦ C, so that the metal surface areas evaluated by this two methods are different. It is known [15] that on Ni(1 1 0) surface, a saturation coverage of 1.5 monolayers at hydrogen adsorption can be achieved. Thus, a calculated surface area higher than the real one can result for Ni by hydrogen chemisorption if on these oxides as support, the Ni(1 1 0) surface preferentially appears. On the other hand, it is worth to note that for Ni/La2 O3 S Ni > S t suggesting that there are adsorption sites available for hydrogen which are not available for krypton adsorption (or do not exist) on Ni black. For the Ni black St is practically the same with SNi-chem and SNi-TPD . Differences between the BET surface area measured with nitrogen and hydrogen could arise when nanostructures are present. For carbon nanofibers the unexpected behaviour—high adsorption capacity for hydrogen but relatively low surface area measured with nitrogen adsorption was explained taking into account that the “nitrogen adsorption does not reach the same surface area like hydrogen does [16].” It might be suspected that, at least for Ni/La2 O3 , where the discrepancy is obvious, nanostructures or whiskers could be present, like those observed for La2 O3 and CeO2 [17,18], as support for metallic nickel. However, electron microscopy did not reveal whiskers in the Ni/La2 O3 sample.

Table 1 Surface area of Ni/Ln2 O3 (Ln = rare earths) (80 at.% Ni, 20 at.% Ln) evaluated by various methodsa Catalyst

Composition (Ni wt.%)

SNi-XRD (m2 /g)

SNi-chem (m2 /g)

SNi-TPD (m2 /g)

St (m2 /g)

Ni black Ni/La2 O3 Ni/CeO2 Ni/Gd2 O3 Ni/Ho2 O3 Ni/Yb2 O3

100 59 57.7 56.4 55.4 54.4

15.6 23.7 15.0 42.2 36.6 16.5

6.8 29.6 38.7 39.2 23.2 41.7

6.4 46.0 44.1 52.6 50.5 50.5

6.7 16.9 42.0 68.3 105.2 124.9

For some of the pure oxides the BET surface area measured were: 36.7 for La2 O3 , 165.4 for CeO2 and 147.2 m2 /g for Yb2 O3 . aS Ni-XRD —nickel surface area from average particle size evaluated from XRD data, SNi-chem —nickel surface area from hydrogen chemisorption, SNi-TPD —nickel surface area from hydrogen thermodesorption spectroscopy, St —total surface area (nickel + oxide) by krypton adsorption (BET).

124

A. Gluhoi et al. / Applied Catalysis A: General 232 (2002) 121–128

Fig. 1. Hydrogen adsorption/desorption isotherm for Ni/La2 O3 .

A higher hydrogen uptake (at 1 atm H2 or above) might result from a possible penetration, at a slower rate, of the adsorbed H atoms between the Ni crystallites and the oxide surface [19]. This assumption is justified by the fact that in the absence of these additional sites, i.e. for Ni black SNi-chem ≈ SNi-TPD ≈ St .

Table 2 The amounts of hydrogen desorbed from Ni/Ln2 O3 catalysts using dry and wet hydrogen gas for sample saturation for adsorption Catalyst

Dry hydrogen (cm3 /g)

Wet hydrogen (cm3 /g)

3.2. Hydrogen thermodesorption spectroscopy and the influence of preadsorbed H2 O

Ni black Ni/La2 O3 Ni/CeO2 Ni/Gd2 O3 Ni/Ho2 O3 Ni/Yb2 O3

1.83 12.86 11.96 15.03 12.28 14.45

2.40 13.39 11.69 18.75 12.50 14.44

The amount of adsorbed hydrogen does not change significantly with the lanthanide oxide used as support, as can be seen from Table 2. However, it is much higher compared to nickel black. Fig. 2 reveals that generally the hydrogen desorption occurs in two steps except for Ni/La2 O3 which shows a single desorption peak at slightly lower temperatures. The second peak (or shoulder) appears more or less pronounced above 100 ◦ C showing a slight tendency to move towards higher temperatures from Ce to Yb.

The preadsorbed water shows almost no influence on the amount of adsorbed hydrogen but it changes the peak profile, narrowing the desorption range and increasing sharply the peak intensity; this suggests a narrowing of the range of M–Had bond energies for the available adsorption sites. The nature of M–Had bond may be affected by the neighbouring sites occupied by water molecules; changes of the M–H bond strength in solution compared to the gas phase has been reported in literature [20].

A. Gluhoi et al. / Applied Catalysis A: General 232 (2002) 121–128

125

Fig. 2. Thermodesorption spectra of hydrogen from Ni/Ln2 O3 catalyst in the absence (solid line) and in the presence of coadsorbed water (broken line).

Table 3 shows that the amount of adsorbed water normalised to the total surface area (in the presence of hydrogen), decreases with the decreasing Ln3+ ionic radius, in agreement with the character of the Ln2 O3 oxides for which the hydration degree decreases from La2 O3 to Lu2 O3 .

3.3. The catalytic activity for the isotopic exchange reaction H2 /HDO The values of the reaction rate measured for the Ni black and supported Ni samples are reported in Table 4. The specific activity value, rg , increases

126

A. Gluhoi et al. / Applied Catalysis A: General 232 (2002) 121–128

Fig. 2. (Continued).

Table 3 Water adsorption by Ni/Ln2 O3 catalysts in weight percent and normalised to the total surface area Catalyst

H2 O (wt.%)

H2 O (g/m2 ) ×103

Ni black Ni/La2 O3 Ni/CeO2 Ni/Gd2 O3 Ni/Ho2 O3 Ni/Yb2 O3

2.89 8.04 9.30 10.79 11.60 12.20

4.3 3.45 2.20 1.58 1.10 0.98

Table 4 The catalytic activity of Ni/Ln2 O3 catalysts for the isotopic exchange H2 /HDO measured per gram of catalyst—rg and normalised to the Ni surface area—rs Catalyst

rg × 105 (mol HD/g s)

rs × 106 (mol HD/m2 s)

Ni black Ni/La2 O3 Ni/CeO2 Ni/Gd2 O3 Ni/Ho2 O3 Ni/Yb2 O3

0.023 33.7 13.1 3.8 2.69 2.1

0.034 11.4 3.4 0.97 1.16 0.50

by about two orders of magnitude in the case of Ni/Yb2 O3 , and more than three orders of magnitude for the Ni/La2 O3 sample compared to Ni black. For the other oxides as support, intermediate rg values are obtained. It should be emphasised that any support used in these studies has now measurable catalytic activity for reaction (1) under the conditions described above. The reaction rate has been, therefore, normalised to the Ni surface area, SNi-chem . The intrinsic activity (i.e. the activity per metal surface area unit), rs , is enhanced by all the supports and is strongly dependent on the rare-earth ion in the support. While Yb2 O3 enhances rs by about one order of magnitude, compared to Ni black, La2 O3 enhances it more than two orders of magnitude. The effect of the support on the catalytic properties of metals is much discussed in literature [21,22]. Thus, the CO, CO2 [23,24] and acetone [25] hydrogenation was extensively studied and enhancement of the intrinsic activity by to two orders of magnitude was reported. The explanation oh this behaviour is based on: (i) a direct involvement of the support via a chemical bonding with adsorbed reactants by oxygen vacancies [24,26] or Lewis acid sites [27–29] and (ii) an indirect effect of the support which induces changes both

A. Gluhoi et al. / Applied Catalysis A: General 232 (2002) 121–128

Fig. 3. The variation of the intrinsic catalytic activity of Ni/Ln2 O3 catalysts with the Ln3+ radius.

in the electronic properties and on the architecture of the neighbouring metal surface [30]. As regards the reaction (1), in Fig. 3, the exponential decrease of the rs value with the diminution of the Ln3+ ionic radius is reported. Concerning the possible question about the valence state of the cerium it must be pointed out that in the experimental conditions used here for the catalyst preparation NiO/CeO2 results after calcination. However, in the case of cerium oxide as support it was shown that it can be reduced by the hydrogen at moderate temperature (623 K, i.e. our working conditions [31]) to Ce3+ state in the neighbourhood of the metal. Therefore, the radius of Ce3+ has been taken into account in Fig. 3. It can be seen also that the decrease of the intrinsic activity follows the same direction with the decreasing of both the basicity and the hydration degree from La2 O3 to Yb2 O3 . Nickel deposited on the most basic oxide, La2 O3 , shows the highest intrinsic activity. This points out to the idea that the basicity of

127

the support oxide and its degree of hydration play an important role on the rate of the reaction (1). These results can be explained in terms of a mechanism in which the activation of the both reactants is required. The surface of metallic nickel activates the hydrogen molecule by dissociative chemisorption, and the water molecule is activated by the hydroxyl groups of the oxide support, which can interact easily and reversibly with water. As a proof for the proposed mechanism it might be invoked that when oxides are present as support, with –OH groups on their hydrated surface, the exchange between –OH and HDO is a very fast stage [32]. The exchange step between adsorbed hydrogen (moving on the metal surface) and the –OH/–OD groups from the oxide support obviously occurs at the metal-support interface. The overall reaction rate depends both on the concentration of the activated adsorbed co-reacting agents (Had and H2 Oad ) and on their mobility. Although Ni black which adsorbs similar amounts of water with Ni/La2 O3 , normalised to the surface area (see Table 3), the intrinsic activity of the former is two orders of magnitude lower than for Ni/La2 O3 . The absence of the –OH groups on the Ni black surface results in rather inactive H2 O molecules on its surface and consequently a low reaction rate. From these it is obviously revealed the role of the –OH groups existing on the support oxide in the enhancement of the intrinsic activity for reaction (1). Finally, it should be noted that rs increases with the amount of adsorbed water on the unit surface area (Table 3) and the basicity of the support, and also with the mobility of the adsorbed hydrogen, qualitatively revealed by TPD, in agreement with the proposed reaction mechanism.

4. Conclusions The rate of the isotopic exchange reaction H2 /HDO heterogeneously catalysed by Ni/Ln2 O3 exhibits a strongly support-dependent catalytic activity. The reaction rate normalised to the active surface area decreases exponentially with the decreasing of Ln3+ ionic radius from La3+ to Yb3+ . This decrease follows the same direction with the decreasing basicity and the hydration degree of the Ln2 O3 support oxide. The results can be explained by the changing mobility of the adsorbed co-reacting species (hydrogen

128

A. Gluhoi et al. / Applied Catalysis A: General 232 (2002) 121–128

and water) and their concentrations on the catalyst surface, the reaction actually occurring at the nickel-oxide boundaries. The results could explain also the importance of the relatively high La and Ce contents for the high-rate dischargeabilities of LnNi5 type alloys used as anodes in Ni/MH batteries [14] or the improved electrochemical activation [33] of the metal hydride electrodes from Laves phase alloys with La or Ce added. In this case, the same catalyst (electrocatalyst) Ni/Ln2 O3 , is present on the surface of the metal hydride electrode. References [1] P. M˘arginean, A. Olariu, J. Catal. 8 (1967) 359. [2] A. Olariu, P. M˘arginean, Rev. Roum. Phys. 13 (1968) 823. [3] N.H. Sagert, P.E. Show-Wood, R.M.L. Pouteau, Can. J. Chem. 53 (1975) 3257. [4] P. M˘arginean, A. Olariu, J. Catal. 95 (1985) 1. [5] M. Shimizu, S. Kioto, N. Ninomiya, in: Y. Fujii, T. Ishida, K. Takeuchi (Eds.), Proceedings of the International Symposium on Isotope Separation and Chemical Exchange Uranium Enrichment, Tokyo, 1992, p. 560. [6] P. M˘arginean, A. Olariu, Appl. Catal. A: Gen. 140 (1996) 59. [7] P. M˘arginean, Isotopenpraxis 301 (1994) 359. [8] M.E. Fiorino, R.L. Opila, K. Konstadimidas, W.C. Fang, J. Electrochem. Soc. 143 (1996) 2422. [9] J. Nan, Y. Yang, Z. Lin, J. Alloys Comp. 316 (2001) 131. [10] E. Higuchi, E. Toyoda, Z.P. Li, S. Suda, H. Inone, S. Nohara, C. Iwakura, Electrochem. Acta 46 (2001) 1191. [11] H. Hisa, K. Moriai, K. Aoyama, H. Kondo, H. Uchida, J. Alloys Comp. 253–254 (1997) 525. [12] F. Meli, T. Sakai, A. Züttel, L. Schlapbach, J. Alloys Comp. 221 (1995) 284. [13] L. Jiang, F. Zhan, D. Bao, G. Qing, Y. Li, X. Wei, J. Alloys Comp. 231 (1995) 635.

[14] J.J. Reilly, in: J.O. Besenhard (Ed.), Handbook of Battery Materials, Wiley, New York, 1999, Chapter 7, p. 209. [15] E. Boschung, Th. Pillo, J. Hayoz, L. Patthey, P. Aebi, L. Schlapbach, Phys. Rev. B 58 (1998) 210. [16] R. Ströbel, L. Jörissen, T. Schliermann, V. Trapp, W. Schütz, K. Bohmhammel, G. Wolf, J. Garche, J. Power Sources 84 (1999) 221. [17] M.A.P. Dekkers, M.J. Lippits, B.E. Nienwenhuys, Catal. Today 54 (1999) 381. [18] F. Maurel, M.J. Hÿtch, B. Knops, M. Backhaus-Ricoult, Eur. Phys. J. AP 9 (2000) 205. [19] B.L. Mojet, M.J. Kappers, J.T. Miller, D.C. Koningsberger, Stud. Surf. Sci. Catal. 101 (1996) 1165. [20] S. Trasatti, Surf. Sci. 335 (1995) 1. [21] K. Hayek, R. Kramer, Z. Paál, Appl. Catal. A: Gen. 162 (1997) 1. [22] A.Yu. Stakheev, L.M. Kustov, Appl. Catal. A: Gen. 188 (1999) 3. [23] G.M. Pajonk, S.J. Teichner, in: I. Cerveny (Ed.), Catalytic Hydrogenation, Elsevier, Amsterdam, 1986, p. 277. [24] S.A. Stevenson, G.B. Raupp, J.A. Dumesic, S.J. Tauster, R.T.K. Baker, in: S.A. Stevenson, J.A. Dumesic, R.T.K. Baker, E. Ruckenstein (Eds.), Metal-Support Interactions in Catalysis, Sintering and Redispersion, Van Nostrand Reinhold, New York, 1987, p. 55. [25] B. Sen, M.A. Vannice, J. Catal. 113 (1988) 52. [26] R. Burch, A.R. Flambard, J. Catal. 78 (1982) 389. [27] G.A. Somorjai, Surf. Sci. 299/300 (1994) 849. [28] A.T. Bell, J. Mol. Catal. 100 (1995) 1. [29] Y. Borodko, G.A. Somorjai, Appl. Catal. A: Gen. 186 (1999) 355. [30] D.C. Koningsberger, B.C. Gates, Catal. Lett. 14 (1992) 271. [31] S. Bernal, J.J. Calvino, M.A. Cauqui, J.M. Gatica, C. Larese, J.A. Pérez Omil, J.M. Pintado, Catal. Today 50 (1999) 175. [32] A. Ozaki, Isotopic Studies of Heterogeneous Catalysis, Kodansha, Tokyo, 1977, p. 54. [33] D. Sun, M. Latroche, A. Pércheron-Guégan, J. Alloys Comp. 248 (1997) 215.