Influence of the preparation method and the nature of the support on the stability of nickel catalysts

Influence of the preparation method and the nature of the support on the stability of nickel catalysts

applied catalysis A ELSEVIER Applied Catalysis A: General 109 ( 1994) 167-179 Influence of the preparation method and the nature of the support on t...

666KB Sizes 0 Downloads 18 Views

applied catalysis A ELSEVIER

Applied Catalysis A: General 109 ( 1994) 167-179

Influence of the preparation method and the nature of the support on the stability of nickel catalysts* * A. Gil, A. Diaz, L.M. Gandia, M. Mantes* Grupo de Ingenierh Quimica, Departarnento de Quimica Aplicada, Facultad de Quhica de San Sebastidn, Universidad del Pais Vasco, Apdo. 1072, 20080 San Sebastih, Spain

(Received 7 January 1993, revised manuscript received 17 November 1993)


The influence of the preparation method and the nature of the support on the stability of the metal dispersion in nickel catalysts was studied. Three different preparation methods, incipient wetness, ion exchange and precipitation-deposition using three different commercial supports, silica, alumina and silica-alumina were used. The metallic dispersion stability was evaluated by hydrogen adsorption measurements after high-temperature calcination of the samples before and after reduction. Results have shown that the behaviour of the samples depends on the balance between positive and negative effects in the metal-support interaction, i.e., the increase of the initial metal dispersion and its resistance against sintering and, the loss of nickel in the form of difficult to reduce interaction compounds. Interaction depends on the preparation method, but can be considerably modified during catalyst life. The reactivity of the support towards nickel plays an important role in these changes. The catalyst prepared by precipitation-deposition over silica showed the best metallic dispersion and stability. Key words: alumina; catalyst preparation (ion exchange/incipient wetness/deposition-precipitation); metal-support interaction; nickel; silica


1. Introduction

The metallic


at the end of the preparation method used is one of the of nickel catalysts. In addition, the initial metallic dispersion


most important characteristics

*Corresponding author. Tel. ( + 34-43) 216600, fax. ( + 34-43) 212236, e-mail [email protected] **Paper presented at the XIII Iberoamerican Symposium, Segovia (Spain), 610 July 1992. 0926-860X/94/$07.00 0 1994 Elsevier Science B.V. All rights reserved SSDIO926-860X (93)E0257-D


A. Gil et al. /Applied Catalysis A: General 109 (1994) 167-179

must also be resistant to the operation conditions during catalyst life and to the possible regeneration treatments that the catalyst may undergo after some deactivation processes [ 141. Both these characteristics, metallic dispersion and its stability depend on the metalsupport interaction [ 5,6]. The loss of metallic dispersion during catalyst life or regeneration treatment is due to sintering or solid state reactions between the nickel and the support. Therefore, the interface between them, usually formed by remaining amounts of interaction compounds, plays an important role. The reduction treatment, that is the last step of the activation process, can modify this interface and in this way, the metallic dispersion resistance. Therefore, the classical stability test, high-temperature calcination, must be carried out with both the oxidized precursor sample and the reduced catalyst. In the present work the influence of the metal-support interaction on the metallic dispersion of nickel catalyst was studied. The metal-support interaction was related to the preparation method and the nature of the support. Three different supports, silica, alumina and silica-alumina were used, each in three different preparation techniques: incipient wetness, ion exchange and precipitation-deposition. In order to evaluate the metal dispersion stability, samples were subjected to a high-temperature calcination treatment, before and after reduction. Metallic dispersion was measured by hydrogen chemisorption but total surface area, reduction degree, X-ray diffraction and temperature-programmed reduction measurements were also carried out.

2. Experimental

2.1. Catalysts Silica (Ketjen F7, 269 m2 gg ‘), alumina (Rhone-Poulenc SCS 250, 243 m* g-‘) and silica-alumina (Ketjen L.A. 30-50 P, 129 m* g-i) were used as supports. Samples were prepared by three different methods: incipient wetness, ion exchange and precipitationdeposition. For the catalysts prepared by the incipient wetness technique, the required amount of an aqueous solution of Ni( N03) ,*6H,O (Merck, analytical-reagent grade) was slowly added to the support at room temperature. This amount was calculated by using the water pore volume of the support. The concentration of the solution was adjusted to obtain catalysts of ca. 10 wt.-%. The ion-exchange catalysts were prepared by adsorption of the nickel amino complex on the supports, from a solution of nickel nitrate in ammonia (30%) [ 71. The process was carried out at room temperature. The pH reached by the preparation mixture was 10.4. The kinetics of the adsorption were followed by EDTA-murexide titration. When the process finished, the solid was washed and filtered. The precipitation-deposition silica and silica-alumina supported catalysts were prepared by heterogeneous precipitation of nickel from an aqueous solution of nickel nitrate slurried with the support [ 8s. The precipitation was produced by a slow and homogeneous change

A. Gil et al. /Applied Catalysis A: General 109 (1994) 167-179


in the pH induced by the thermal decomposition of urea at 90°C. The alumina supported catalyst was prepared by means of a technique described by Schaper et al. [ 91. In this way, the catalyst was prepared by precipitation-deposition of nickel compounds on the activated surface of the support. To achieve this activation, the support was previously treated with a high pH ( 10.4) nickel ammonia solution for 30 min at 30°C. Nickel deposition-precipitation was then effected by thermal decomposition of amino complexes stirred by a carbon dioxide gas flow at 90°C. All the precursors were dried at 120°C for 16 h and calcined in a 150 cm3 min- ’ g- ’ air stream for 16 h at 450°C in a fixed bed reactor. The catalysts prepared by these techniques will be referred to by using Si, Al and SiAl for the support used (silica, alumina and silicaalumina, respectively) and I, A and PD for the preparation method (incipient wetness, ion exchange and precipitation-deposition), e.g. ( Si-I), ( Si-A) , ( Al-PD) . Reduction of the samples was carried out in a hydrogen stream at 500°C for 12 h with a heating rate of 8°C mm . The conditions of passivation have been previously reported [ 101. The reduced and passivated catalysts are designated with an R, subscript, e.g. (S-I)np, (SiAl-A)np. The stability of the metal dispersion was evaluated by subjecting both the calcined precursors and the reduced passivated samples to a calcination treatment at 780°C in a 125 cm3 mini gg’ air stream for 3 h (heating rate 8°C min-‘). The calcined samples are designated with a C subscript, e.g. (Al-PD)c, ( Si-A)R,c. 2.2. Characterization

In order to determine the nickel content (wt.-% Ni), the catalysts were dissolved in hydrofluoric acid and titrated with EDTA-murexide. The surface areas ( SnET) were measured by nitrogen adsorption (BET method) in a Micromeritics Pulse Chemisorb 2700. The metallic surface area (S,) and the degree of reduction v) were measured by hydrogen adsorption and oxygen uptake using the dynamic pulse method with a Micromeritics Pulse Chemisorb 2700. After standard reduction (49O”C, 16 h, 200 cm3 min- ’ g - i of hydrogen), the hydrogen on the nickel surface was removed with a nitrogen stream (SE0 99.998%, treated with a Chrompack clean-oxygen filter) at 490°C for 20 min and cooled to room temperature in the same nitrogen stream. Hydrogen pulses (0.105 cm3) were injected at 20°C until the eluted area of consecutive pulses was constant. Reproducibility of these measurements was better than 10% for the metallic surface area and better than 3% for the degree of reduction. Metallic surface areas were calculated assuming a stoichiometry of one hydrogen molecule adsorbed per two surface nickel atoms. An average cross sectional area per surface nickel atom of 6.33 A’ proposed by Coenen and Linsen [ 1 l] was used, which is in the range of most of the values reported in the literature (6.14-6.77 A’) [ 12-151. The degree of reduction (f, was determined, after hydrogen chemisorption, by removing the chemisorbed hydrogen with helium (SE0 99.998%, treated with a Chrompack clean-oxygen filter) for 1 h at 430°C and then injecting oxygen (SE0 99.995%) in 1.052 cm3 pulses until the signal was saturated. At 430°C the metallic nickel would have been oxidized with oxygen to form NiO [ 161.


A. Gil et al. /Applied

Catalysis A: General 109 (1994) 167-179

Following Bartholomew and Panel1 [ 131, it was assumed that the metallic particles of the catalysts with weak metal-support interaction are spherical and wholly exposed to adsorption and that the unreduced nickel phase is separated from the reduced particles. Under these circumstances, the mean particle size (defined as the cube root of the volume) was determined by means of the following equation [ 171: d = 543 f/S,, where d is the mean size (nm), S, the metallic surface area ( m2 g - ’ of Ni) andfis the degree of reduction. According to Coenen and Linsen [ 1 I], nickel catalyst particles which have strong interactions between the metal and the support are hemispherical and are attached both to the support and to the interaction compound by their equatorial plane. This model leads to the following equation for the mean size of the nickel particles (defined as the cube root of the volume) [ 171: d = 43 1f/S, . Metallic dispersion (% 0) was calculated as the percentage of surface nickel atoms (hydrogen adsorbed) with respect to the total nickel atoms. Temperature-programmed reduction (TPR) behaviour was followed gravimetrically with a Setaram TG 85 thermobalance. The sample, either in the starting, in the reduced and passivated or in the high-temperature calcined state (28 mg), was loaded into the thermobalance and heated in a hydrogen stream (SE0 99.998%) at 120°C until constant weight was achieved. Then, TPR was carried out from 120 to 1000°C at 10°C mini in a 1400 cm3 min-‘g-l hydrogen stream. It should be pointed out that only the positions of the maxima, in combination with other characterization techniques, were used to evidence the metalsupport interaction. X-ray diffractograms (XRD) were recorded with a Philips PW 1710 instrument, using Cu K, radiation filtered through nickel.

3. Results

Table 1 presents the total surface area (S,,,), nickel content (wt.-% Ni) , metallic surface area (S,), degree of reduction cf,, mean size of nickel particles (d), and metallic dispersion (0) of the different catalysts prepared. Fig. 1 shows the TPR results obtained for all the starting samples. Fig. 2 shows the Xray diffraction spectra of the starting samples before reduction and that of the supports. Fig. 3 shows the X-ray diffraction spectra of the catalyst prepared by precipitation-deposition over silica, before reduction (Si-PD), and after the two types of stability tests, ( Si-PD), and (Si-PD),,. Fig. 4 shows the X-ray diffraction patterns of the (Al-PD) precursor before and after calcination and that of the support used. Fig. 5 shows the TPR results obtained for (Si-A), that of the same sample after the stability tests (Si-A)c and (SiA)R+Y, and that of the support. Fig. 6 shows the TPR patterns obtained for all the samples prepared over alumina before and after stability tests, and that of the support.

A. Gil et al. /Applied Catalysis A: General 109 (1994) 167-179



A. Gil et al. /Applied Catalysis A: General 109 (1994) 167-179

A. Gil et al, /Applied Catalysis A: General 109 (1994) 167-179

Table 1 Characterization

of the samples



(Si-I) (Si-l)c

203 207 200 273 243 247 288 262 213 200 165 153 260 185 183 225 173 174 108 103 104 165 145 142 181 166 152

(SW,,, ( Si-A) (Si-A)c (Si-A),,, ( Si-PD) ( Si-PD), (Si-PD),& (Al-I) (Al-I), (Al-1)~~ (Al-A) (Al-A), (Al-A)r+c (Al-PD) (Al-PD),

( Al-PD) ~~~ (SiAl-I) ( SiAl-I), ( SiAl-I),, ( SiAI-A) ( SiAl-A), ( SiAl-A) RpC ( SiAl-PD) ( SiAl-PD), ( SiAl-PD) nK





Ni (wt.-%)

SH (m’/gNi)

11.5 12.5 12.9 1.5 8.1 8.1 8.2 9.6 9.6 10.2 12.2 12.3 8.0 8.0 9.6 13.5 13.5 13.5 10.3 10.5 11.1 5.9 6.1 7.0 10.2 10.5 10.5

20 15 16 92 61 65 119 70 90 28 1.5 1.2 27 1.3 2.8 50 12 12 27 16 16 56 13 12 49 22 18




f(%) 100 100 100 85 69 89 91 83 96 71 6 4 63 8 14 94 32 43 98 84 68 96 55 66 92 65 80


d (nm)

D (%)

28 31 34 4.0 4.5 5.9 3.3 5.1 4.6 11 21 19 8.9 27 19.7 8.1 12 15 20 28 23 7.4 18 22 8.1 13 20

3.0 2.3 2.4 14 10 9.9 18 10 14 4.2 0.3 0.2 3.8 0.2 0.4 7.6 1.8 1.9 4.1 2.5 2.5 8.6 2.1 2.0 7.5 3.4 2.7


Fig, 3. XRD patterns of the sample prepared by precipitation+ieposition over silica; before, (Si-PD), stability tests, ( Si-PD), and ( Si-PD)R,c. ( 0) NiO and (0) nickel antigorite like compound

and after


A. Gil et al. /Applied Catalysis A: General 109 (1994) 167-179





2 8 (j”


Fig. 4. XRD patterns of the alumina support used, that of the sample prepared by precipitation+Ieposition alumina prior to calcination, (Al-PD)p.c., and after calcination, ( AI-PD). (0) NiO and (A) Feitknecht pound.






T/“C Fig 5. TPR patterns of ( Si-A) before and after stability tests.

over com-

A. (Gilet al. /Applied Catalysis A: General 109 (1994) 167-179



A. Gil et al. /Applied Catalysis A: General 109 (1994) 167-179

4. Discussion 4.1. Influence of the preparation dispersion

method and the nature of the support on the metallic

4. I. 1. Silica XRD, TPR and S,,, data confirm that an interaction compound, nickel antigorite-like type [ 5,111 was produced in ( Si-PD) whilst only NiO on silica can be detected in (Si-I). An interaction product was also produced in ( Si-A) like that in (Si-PD) but in a lower amount or less well crystallized as shown by XRD (Fig. 2) but producing a similar SBET increase. The formation of this compound was related to the high pH used during preparation [ 71. The positive effect of the metal-support interaction that increases the metal dispersion prevails over the negative one that decreases the degree of reduction. Therefore, metallic surface areas of ( Si-PD) and (SGA) are significantly higher than that of (Si-I). 4.1.2. Alumina The TPR results and measured reduction degree suggest the formation of a large amount of interaction product, difficult to reduce, during the preparation of (Al-I) and (Al-A). The SBETincrease of this samples with respect to that of the support also confirms the metalsupport interaction [ 18,191. This compound has been described as nickel aluminate [ 18,191. Nevertheless, this product cannot be detected by XRD which may be due to a poor crystallinity. Metal-support interaction, produced even in the sample prepared by incipient wetness (Al-I), must be related to the high reactivity of alumina with nickel. In (Al-I) and (Al-A) the metal-support interaction removes a significant amount of nickel from the reduced form (30-40%) and does not produce good metal dispersions. Therefore, the measured metallic surface areas are low. The results obtained with the precipitation-deposition method on alumina are more interesting. The XRD pattern of the dried precursor before calcination shows (Fig. 4) the presence of a Feitknecht compound [ 18,191. This compound consists of brucite type hydroxide layers, containing nickel and aluminium ions, alternated by layers containing water and carbonate ions. After calcination, XRD lines corresponding to this product disappeared (Fig. 4) due to the carbon dioxide and water loss that produce a corresponding crystallinity loss [ 9,19,20]. Nevertheless, the resulting sample, (AI-PD), is quite different, easier to reduce (see TPR andfdata) and showing higher metal dispersion than either (AlI) or (Al-A). 4.13. Silica-alumina The presence of interaction compounds in (SiAl-A) and (SiAl-PD) cannot be confirmed by XRD, which only indicates the presence of poorly crystalline materials. Nevertheless, the metallic surface area of those samples are double that of ( SiAl-I), but all three are easy to reduce cf> 0.9). In addition to that, the three preparation methods used produced the same effect on the S,,, for all the supports: impregnation reduces the SBETby pore plugging, and precipitation-deposition and ion-exchange increase the SBETdue to the formation of nickel-support interaction compounds. The interaction between nickel and silica-alumina

A. Gil et al. /Applied Catalysis A: General 109 (1994) 167-179


in (SiAl-A) and (S&l-PD) should be less than that produced over silica or alumina, but intense enough to explain the metallic dispersion increase compared to that in ( SiAl-I). 4.2. Stability tests 4.2.1. Silica Both types of stability tests, namely calcination with or without previous reduction (R,C and C subscript, respectively) produced similar effects on (Si-I) : a slight decrease of the metallic surface area. For samples with considerable metal-support interaction ( Si-A) and (Si-PD), different results can be observed for every treatment. When the sample was reduced before calcination the resulting catalyst was easier to reduce than the one not previously reduced. (Si-A)c and ( Si-A) RpCshowed similar metallic surface area but with a significant difference in the degree of reduction. These results are in agreement with TPR patterns (Fig. 5) that show that in (Si-A) RpCpart of the nickel is easy to reduce (without interaction) but in (Si-A), the reduction peak has a higher-temperature shoulder. From all this information it can be deduced that the difficulty of reducing (Si-A), compensates for a lower degree of sintering, giving the same metallic surface area as that of ( Si-A) RpC. ( SiPD) showed an interesting behaviour in the two types of stability tests. R,C treatment gives a sample easier to reduce than C treatment and even easier to reduce than the starting sample (Si-PD) . Nevertheless, ( Si-PD),, showed a great resistance to sintering, by comparison of nickel particle size, that is smaller for ( Si-PD) ,+c than for ( Si-PD) c as can be seen in Table 1. XRD patterns (Fig. 2) show that the interaction compound in (Si-PD) has disappeared in (Si-PD)c and ( Si-PD)R,c, showing only lines corresponding to NiO. It can be concluded that the nickel-silica interaction on (Si-PD) has undergone a positive change during reduction, giving a catalyst at the same time well dispersed and better anchored to the support. 4.2.2. Alumina (Al-I) and (Al-A:), where an aluminate like interaction compound was formed, suffered a dramatic loss in the metallic surface area related with a very low reduction degree. TPR experiments (Fig. 6)~ showed that a great amount of a nickel compound difficult to reduce was produced, usually identified as nickel aluminate [ 18,20-221. In ( Al-PD) where a Feitknecht type interaction compound was produced, two types of nickel can be seen by TPR in (Al-PD), and (Al-PD)R,c. One high-temperature peak, probably due to nickel aluminate, and an other at a lower temperature explain the higher metallic surface area and reduction degree in these samples in comparison to those of (Al-I) and (Al-A). This easily reducible nickel can be related to the different coordinations of nickel in the Feitknecht compound and the nickel aluminate. It should been pointed out that the Feitknecht compound disappeared after the: first calcination but, nevertheless, it influences the metal stability. As a general conclusion of all the stability tests carried out with alumina supported samples, the high reactivity of alumina against nickel can certainly be chosen as the main factor controlling the metallic dispersion obtained in this type of samples. 4.2.3. Silica4umina Stability tests showed small differences between samples prepared over silica-alumina due to the preparation method. The metallic surface area after the stability test was about 25 to 50% of the starting sample value. Reducibility differences were compensated for by


sintering differences, test.

A. Gil et al. /Applied Catalysis A: General 109 (1994) 167-179

showing finally similar surface area values after both types of stability

5. Conclusions Two different effects can be produced by the metal-support interaction in nickel supported catalysts. One of them negative, the loss of nickel in the active form due to the difficulty of reduction of some interaction compounds. The second one positive, that is the increase of the metal dispersion reached during the preparation and the stabilization of the nickel particles against sintering. The metal surface area of the catalyst depends on the balance between these two opposite effects. Metal-support interaction is basically controlled by the preparation method, but as was shown in the stability test, it can vary significantly during the life of a catalyst. Results obtained using samples prepared on alumina have made evident the main role played by the support reactivity against nickel. The particular interest of samples prepared by precipitation-deposition on silica and alumina should also be pointed out. The different interaction products (antigorite-like and Feitknecht compounds respectively) produced well dispersed catalysts, very resistant to sintering. In particular (Si-PD) can be identified as the best catalyst from this study. It presented a higher metallic surface area and the best sintering resistance, improved by the reduction treatment.

Acknowledgements The Scholarship support for L.M. Gandfa and A. Gil by the Ministerio de Education y Ciencia (Programme FPI) and the financial support by the Basque Government (Grant PGV 87113) and the Excma. Diputacion Foral de Gipuzkoa are gratefully appreciated.

References [ 1] J.R. Donovan, R.D. Stolk and M.L. Unland, in B.E. Leach (Editor), Applied Industrial Catalysis, Vol. 2, Academic Press, 1984, p. 27 1. [2] C.H. Bartholomew and W.L. Sorensen, J. Catal., 81 (1983) 131. [3] S.E. Wanke and P.C. Flynn, Cat. Rev. Sci. Eng., 12 (1975) 93. [4] J.T. Richardson and J.G. Grump, J. Catal., 57 (1979) 417. [S] M. Montes, C. Penneman de Bosscheyde, B.K. Hodnett, F. Delannay, P. Grange and B. Delmon, Appl. Catal., 12 (1984) 304. [6] M. Montes, J.B. Soupart, M. de Saedeleer, B.K. Hodnett and B. Delmon, J. Phys. Chem., Faraday Trans. 1, 80 (1984) 3209. [7] M. Hot&la, F. Delannay, I. Matsura and B. Delmon, J. Chem. Sot., Faraday Trans. 1,76 (1980) 2128. [ 81 L.A.M. Hermans and J.W. Geus, in B. Delmon, P. Grange, P. Jacobs and G. Poncelet (Editors), Preparation of Catalysts II (Studies in Surface Science and Catalysis, Vol. 3), Elsevier, Amsterdam, 1979, p. 113. [9] H. Schaper, D.J. Amesz, E.B.M. Doesburg, P.H.M. de Korte, J. M.C. Quartel and L.L. van Reijen, Appl. Catal., 16 (1985) 417.

A. Gil et al. /Applied


Catalysis A: General 109 (1994) 167-179


A. Gil, A. Diaz and M. Montes, J. Chem. Sot. Faraday Trans., 87 (1991) 791.

[ 111 J.W.E. Coenen and B.G. Linsen, in B.G. Linsen (Editor), Physical and Chemical Aspects of Adsorbents and Catalysts,


Press, London,

1970, p. 495.

[ 121 J.R. Anderson and KC. Pratt, Introduction to Characterization [ 131 [ 141 [ 151 [ 161 [ 171 [ 181 [ 191

[20] [21] [22]

and Testing of Catalysts, Academic Press, Australia, 1985, p. 7. C.H. Bartholomew and R.B. Pannell, J. Catal., 65 (1980) 390. J.W.E. Coenen, Appl. Catal., 75 (1991) 193. C.H. Bartholomew, in Z. P&l and P.G. Menon (Editors), Hydrogen Effects in Catalysis, Fundamental and Practical Applications, Mercel Dekker, New York, 1988, p. 98. C.H. Bartholomew and R.J. Farrauto, J. Catal., 45 (1976) 41. A. Aguinaga, J.C. de la Cal, J.M. Asua and M. Montes, Appl. Catal., 16 (1989) 417. B. Scheffer, P. Molhock and J.A. Moulijn, Appl. Catal., 46 ( 1989) 11. D.C. Puxley, I.J. Kitchener, C. Komodromos and N.A. Perkyns, in G. Poncelet, P. Grange and P. Jacobs (Editors), Preparation of Catalysts III (Studies in Surface Science and Catalysis, Vol. 16), Elsevier, Amsterdam, 1982, p. :237. P.K. Bokx, W.B.A. Wassenbergand J.W. Geus, J. Catal., 104 (1987) 86. M. Wu and D.M. Hercules, J. Phys. Chem., 83 (1979) 2003. Y.J. Huang and J.A. Schwarz, Appl. Catal., 36 (1988) 163.