The CO methanation on Rh/CeO2 and CeO2/Rh model catalysts: a comparative study B. Jenewein *, M. Fuchs 1, K. Hayek * Institut f€uur Physikalische Chemie, Leopold-Franzens-Universit€aat, Innrain 52a, Innsbruck A-6020, Austria Abstract Like other reducible oxides, ceria promotes the CO methanation reaction on noble metals, but after high-temperature reduction the promotion is usually reduced and limited to transient conditions. We studied the effect of low- and high- temperature reduction on two types of Rh/ceria model catalysts: ‘‘thin film catalysts’’ consisting of well-defined regular Rh nanoparticles partly embedded in the crystalline ceria support, and ‘‘inverse catalysts’’, i.e. UHV grown ceria submonolayers on polycrystalline Rh surfaces. The turnover rates on either catalyst were related to the free Rh surface area and to the dimensions of the metal–oxide boundary. On thin film catalysts the rates decrease strongly with re- duction temperature up to 723 K while no significant structural changes are detectable by ex situ electron microscopy. On ceria-modified surfaces the reaction is initially favoured after reduction below 573 K, but promotion converts to inhibition with increasing reduction temperature. The changing number of CeIII/CeIV ions on the surface, the oxygen transport to and from the interface, and changes in the free Rh surface area resulting from spreading and reordering of the ceria overlayer are discussed as possible reasons for the observed effects. � 2003 Elsevier Science B.V. All rights reserved. Keywords: Catalysis; Cerium; Rhodium; Electron microscopy 1. Introduction The hydrogenation of CO and related reactions on reducible oxide-supported noble metals are sensitive to metal–support interaction [1]. High- temperature reduction may cause a change of the active metal surface area, either by decoration of the metal surface by (sub)oxide species or by sin- tering and/or spreading of the metal particles on the support. On the other hand, electronic inter- action between metal and support, resulting in ‘‘adlineation sites’’ (e.g., oxygen vacancies or low- valent cations) at the metal–support boundary or in direct metal–metal bonds, may alter the cata- lytic properties. Depending on the reduction con- ditions geometric and electronic effects may also overlap. Attempts to discriminate between morphologic and electronic catalyst changes are facilitated by parallel experiments on different model systems. In the past, experiments on ‘‘inverse’’ model cat- alysts, i.e. metal surfaces partially covered with *Corresponding authors. Tel.: +43-412-4075062; fax: +43- 512-5072925 (K. Hayek). E-mail addresses:
[email protected] (B. Jenewein),
[email protected] (K. Hayek). 1 Present address: Infineon Technologies GmbH, Siemensstr. 2, A-9500 Villach, Austria. 0039-6028/03/$ - see front matter � 2003 Elsevier Science B.V. All rights reserved. doi:10.1016/S0039-6028(03)00095-5 Surface Science 532–535 (2003) 364–369 www.elsevier.com/locate/susc reducible metal oxide layers [2,3], have proved useful for exploring metal–oxide interactions by UHV-compatible techniques. A different ap- proach, via small metal particles supported by thin oxide films (‘‘real’’ model systems) [4] is well suited for high-resolution microscopy and has been employed to study structural and morpho- logic changes of noble metal particles on various supports [5]. A parallel investigation of real and inverse model systems may yield complementary information about the interaction between metal and support and about special catalytic sites at the interface. Currently we investigated the methanation rates on regular noble metal Rh particles in epitaxial relation with the ceria support, and on inverse CeO2/Rh model systems, consisting of ceria sub- monolayers on the noble metal surface [3]. Ceria is generally recognized as a promoter for CO hy- drogenation, but after high-temperature reduction the promoting effect is less significant and limited to transient conditions [6]. It must be expected that real and inverse model catalysts exhibit a different promoting effect if oxygen transport between the metal–support interface and the ceria bulk be- comes important. 2. Experimental 2.1. Ceria overlayers polycrystalline rhodium This experimental setup consists of an UHV- AES chamber and an attached all-glass reactor cell [7]. The samples were Rh foils, either side covered with submonolayers of ceria, exposing a total surface area of �8 cm2. After ceria deposition the catalyst is moved in the position for Auger elec- tron spectroscopy (AES) and for transfer to the high-pressure cell. This recirculation reactor (vol- ume 60 ml) is attached to the outside of the chamber by a glass-to-metal seal. Due to the clean environment (all glass, no leads connected to the sample) the reaction rates are well reproducible, and the effect of structural and compositional changes on the catalytic activity can be easily as- sessed. Ceria adlayers were prepared by evapora- tion of cerium metal from a tungsten wire in 2.0� 10�7 mbar O2 with the sample at room tem- perature, followed by annealing in oxygen at 673 K. The resulting surface stoichiometry is CeO2, but partial reduction to CeIII occurs upon an- nealing under vacuum and upon treatment in hy- drogen [8–11]. The coverage was determined with a quartz microbalance and correlated to plots of Auger signal intensity vs. deposition time. The surface coverage was related to metallic Ce: ‘‘1 ML Ce’’, implying a Ce/Rh ratio of 1, corresponds to 1.58� 1015 surface cerium atoms/cm2 or about 1.2 nominal monolayers of CeO2. After reduction in hydrogen at given tempera- ture and pressure the samples were exposed to a CO/H2 mixture (generally 40 mbar CO+120 mbar H2, He added to 1 bar) in the reactor. Reaction rates at 573 K as a function of nominal metal coverage (between 0.16 and 0.6 ML Ce) were determined from conversion vs. time plots, and related to the rate on the clean Rh surface. 2.2. Thin film model catalysts [5] Nanosized rhodium particles were grown on in situ deposited NaCl films (100 cm2 area) by physical vapour deposition at 623 K substrate temperature and a base pressure of 1� 10�5 Pa. The nominal Rh film thickness was 2–3 nm and the mean particle size between 10 and 15 nm. There- after, the samples were covered with 20 nm of ceria evaporated from a W boat. One part of the re- sulting metal/oxide film was removed from the vacuum system, floated and rinsed in water and mounted on gold EM grids. For kinetic measure- ments the film was further stabilised by depositing additional silica layers on top (up to 500 nm thick) before removal from the NaCl substrate. The structure and morphology of the metal particles and their changes due to oxidation and reduction were studied in parallel by electron microscopy and selected area electron diffraction (Zeiss EM10 and JEOL 4000). The kinetic measurements on the Rh/ceria thin film model catalysts were performed in a computer controlled recirculation reactor [12] under similar reaction conditions as above. B. Jenewein et al. / Surface Science 532–535 (2003) 364–369 365 3. Results Under the given deposition conditions Rh grows on NaCl(0 0 1) as well-shaped nanocrystals, mainly truncated octahedra, and the successive coating by the support leaves the (1 0 0) base planes uncov- ered [5]. Fig. 1 shows the Rh particles supported by amorphous carbon (a), and by crystalline CeO2 (b). Due to the similar electron density of Rh and CeO2 the electron optical contrast in Fig. 1b arises mainly from Bragg contributions, but the inserted SAED pattern reveals an epitaxial relation be- tween Rh and ceria [13]. Although the Debye– Scherrer rings indicate some random growth of rhodium as well as of ceria it can be stated that ceria grows preferentially with its (1 1 1) planes on the (1 1 1) planes of the octahedral Rh particles, resulting in an almost perfect model surface of arrays of ceria-surrounded (1 0 0) Rh facets. After reduction in hydrogen up to 673 K no significant alterations of the microstructure are detectable by (ex situ!) electron microscopy. As discussed else- where [14,15], only reduction above 773 K will induce irreversible changes like particle decora- tion, agglomeration and alloy formation. On the other hand, the (initial) rate of CO methanation does change significantly already under low-temperature reduction (Fig. 2a and Table 1). If every reduction is preceded by an- nealing in 1 bar oxygen at 673 K (standard pro- cedure) the reaction rate as a function of Tred declines continuously between 473 and 673 K and is immeasurable after reduction at 723 K (Table 1). This effect is partly reversed if each reduction is followed by annealing in vacuum at 773 K. In this case the activity is more decreased after low-tem- perature reduction and less decreased after high- temperature reduction (Table 1). The catalytic behaviour of the ceria-modified Rh metal surface depends on the ceria coverage and on the temperature of annealing prior to re- duction. Reproducible rates were only obtained after annealing the as-grown ceria adlayers at about 873 K for 10 min. This treatment, resulting also in a significant increase of the ratio of Ce vs. Rh Auger signals, was taken as the starting point of the following reduction series (Fig. 2b and c). As shown in Fig. 2b, the reduction of a low- coverage ceria overlayer (60.25 ML Ce) between 373 and 573 K immediately before reaction induces a slight (up to twofold) initial rate enhancement, but carbon deposition (revealed by AES) leads to increasing deactivation in the course of the reac- tion. In contrast, (further) reduction between 573 and 673 K is followed by a strong decline of initial rates which can be reversed upon oxidation and (repeated) low-temperature reduction. At higher Ce coverage the reaction rate is significantly Fig. 1. Regular Rh-particles (mean size 15 nm) supported by amorphous carbon (a), and epitaxially grown ceria (b). Inset: Electron diffraction pattern with magnification of ‘‘star-like’’ Rh (2 0 0) spot. 366 B. Jenewein et al. / Surface Science 532–535 (2003) 364–369 reduced already after low-temperature reduction, and it is very low after reduction at 673 K and above, as shown in Fig. 2c for h ¼ 0:6 ML Ce (0.72 ML nominal ceria coverage). Also in this case some additional deactivation occurs during the reaction due to carbon deposition. It was attempted to include measurements under transient reaction conditions on the thin film and the inverse catalyst, since in this case a promotional effect could possibly be observed after reduction at higher temperature [16,17], but the amounts of product obtained are far too small for a quantitative evaluation. In addition, parallel kinetic measurements were carried out under steady state reaction conditions using an impregnated Rh/ceria catalyst with a mean Rh particle size of Fig. 2. CO hydrogenation at 573 K on the thin film model catalyst of Fig. 1(a), and of the inverted CeO2/Rh catalysts with nominal Ce coverage of 0.25 (b) and 0.6 (c). Pre-treatment: O2, (1 bar, 673 K, 1 h); H2, (1 bar, Tred, 1 h); Reaction: (a) 20 mbar CO+200 mbar H2 + 780 mbar He; (b) and (c) 40 mbar CO+120 mbar H2 + 840 mbar He. Table 1 Initial turnover rates (molecules per site and second), related to free Rh surface area, as a function of reduction temperature Tred (K) Inverse (ceria/Rh) Thin film (Rh/ceria) Thin film+post-annealing at 773 K 373 4.70� 10�2 2.9� 10�2 1.2� 10�2 473 2.5� 10�2 573 2.8� 10�2 8.5� 10�3 1.3� 10�2 673 9.1� 10�3 4 nm [18]. Again, a strong suppression of CO hy- drogenation was observed, when the reduction temperature was increased up to 723 K. 4. Discussion and conclusions In agreement with the well-documented pro- motional effect of ceria on Rh in supported cata- lysts [17] we observed a rate enhancement upon depositing ceria on a Rh surface, probably due to the creation of reactive interface sites. Compared with titania and vanadia overlayers under similar conditions [2,3] this enhancement is smaller and, in contrast to the rhodium-vanadia system [3], pro- gressive reduction up to 873 K does not lead to (sub)surface alloy formation. For comparing thin film and inverse model systems upon reduction a reference state is needed. The ‘‘ideal array’’ of ceria-surrounded (1 0 0) Rh facets will represent this reference state for the thin film catalyst if we can assume that oxidation at 673 K and subsequent reduction at 6 473 K do not change the structural properties. As for the ceria adlayers, annealing near 873 K was necessary to obtain a reference state with reproducible surface composition. As recently shown by Eck et al. [11], ceria grows on Rh(1 1 1) at room temperature in a Volmer–Weber mode resulting in multilayers at higher coverage, but reordering accompanied by spreading occurs upon annealing, until at 873 K a reproducible two-dimensional (double layer) island system is observed, with Ce3þ ions mainly in the Rh-near positions. Similar rearrangements are likely to occur on the polycrystalline Rh surface. Therefore, after annealing at 873 K and subse- quent oxidation at 673 K the ceria overlayer is fi- nally converted to large islands in double layer configuration [11] which are more ordered than the as-grown state. The relation of measured conversion–time plots (initial rates) to reduction temperature is very similar for inverse and thin film catalysts, although at first sight the turnover rates are different. However, when discussing absolute turnover rates one must compare real and inverse catalysts of equal surface coverage, for example the thin film catalyst depicted in Fig. 1 (mean Rh particle size 15 nm)––initially exposing a surface area of about 30% Rh and 70% CeO2––and the ‘‘0.6 ML Ce’’ catalyst of Fig. 2b, with a nominal 72% CeO2 coverage. After low-temperature reduction we may assume that for either case the total conversion is the sum of the reaction occurring at the ceria– rhodium interface and on the bare Rh patches. From STM images of ceria-covered Rh(1 1 1) surface [11] one can deduce that the Rh patches are not too different in size from the (1 0 0) faces of Rh particles in Fig. 1. Indeed, in the reference states, i.e. after low-temperature reduction, the TOF per free Rh surface site agree within a factor of two on both catalysts. In Table 1 the changes of initial TOF with Tred are given with respect to the free Rh surface area in the reference state. The TOF on the thin film catalyst were corrected to equal reactant partial pressures according to the experimentally determined reaction orders ()0.3 in CO and +1 in hydrogen). These results leave two ways to explain the rate decrease upon reduction, particularly at high temperature: From previous work it is known that the number of Ce3þ ions on the surface in- creases significantly between 473 and 673 K [6,8– 11,19]. It is tempting to relate the activity decrease directly to the surface reduction and hence to the formation of oxygen vacancies on the surface. Oxygen transport between bulk and surface is only possible at higher T and would explain the ‘‘reverse’’ activity changes of the thin film catalyst due to post-reduction annealing at 673 K and above (Table 1). On the other hand, the results obtained on the inverse catalysts indicate that the changing Ce3þ/Ce4þ ratio is interconnected with reordering of the ceria overlayer, and that spreading may occur upon annealing or hydrogen treatment at elevated temperature [11]. More- over, it cannot be completely excluded that under comparable conditions at least parts of the Rh surface of the thin film catalyst will be covered by a ‘‘spilt-over’’ Ce suboxide monolayer, which is not easily detectable in the EM and readily re- versed upon reoxidation at room temperature. This latter explanation, to be confirmed by more detailed in situ studies, would imply a combined effect of electronic interaction and surface deco- ration. 368 B. Jenewein et al. / Surface Science 532–535 (2003) 364–369 Acknowledgements This work was supported by the Austrian Sci- ence Fund (Project S 8105). We also acknowledge fruitful discussions with S. Bernal, J.J. Calvino and J.M. Pintado (C�aadiz) in the frame of ‘‘Acciones Integradas 4/2002’’. References [1] G.L. Haller, D.E. Resasco, Adv. Catal. 36 (1989) 173. [2] A.B. Boffa, C. Lin, A.T. Bell, G.A. Somorjai, Catal. Lett. 27 (1994) 243. [3] K. Hayek, B. Jenewein, B. Kl€ootzer, W. Reichl, Top. Catal. 14 (2001) 25. [4] K. Hayek, M. Fuchs, B. Kl€ootzer, W. Reichl, G. Rupp- rechter, Top. Catal. 13 (2000) 55. [5] G. Rupprechter, K. Hayek, H. Hofmeister, J. Catal. 173 (1998) 409. [6] A. Trovarelli, G. Dolcetti, C. de Leitenburg, J. Kaspar, P. Finetti, A. Santoni, J. Chem. Soc. Faraday Trans. 88 (1992) 1311. [7] W. Reichl, G. Rosina, G. Rupprechter, C. Zimmermann, K. Hayek, Rev. Sci. Instrum. 71 (2000) 1495. [8] C. Hardacre, G.M. Roe, R.M. Lambert, Surf. Sci. 326 (1995) 1. [9] K.D. Schierbaum, Surf. Sci. 399 (1998) 29. [10] D.R. Mullins, P.V. Radulovic, S.H. Overbury, Surf. Sci. 429 (1999) 186. [11] S. Eck, C. Castellarin-Cudia, S. Surnev, M.G. Ramsey, F.P. Netzer, Surf. Sci. 520 (2002) 173. [12] C. Zimmermann, K. Hayek, Chem. Ing. Tech. 63 (1991) 68. [13] S. Bernal, F.J. Botana, J.J. Calvino, G.A. Cifredo, J.A. P�eerez-Omil, J.M. Pintado, Catal. Today 23 (1995) 219. [14] S. Penner, D. Wang, D.S. Su, G. Rupprechter, R. Pod- loucky, R. Schl€oogl, K. Hayek, Surf. Sci. 532–535 (2003) 276. [15] S. Bernal, J.J. Calvino, M.A. Cauqui, J.M. Gatica, C. Larese, J.A. P�eerez Omil, J.M. Pintado, Catal. Today 50 (1999) 175. [16] A. Trovarelli, G. Dolcetti, C. De Leitenburg, J. Kaspar, P. Finetti, A. Santoni, J. Chem. Soc. Faraday Trans. 88 (1992) 1311. [17] A. Trovarelli, Catal. Rev. – Sci. Eng. 38 (1996) 439. [18] M. Fuchs, B. Jenewein, K. Hayek, J.J. Calvino, J.M. Pintado, S. Bernal, in preparation. [19] J. Cunningham, D. Cullinane, J. Sanz, J.M. Rojo, X.A. Soria, J.L.G. Fierro, J. Chem. Soc. Faraday Trans. 88 (1992) 3233. B. Jenewein et al. / Surface Science 532–535 (2003) 364–369 369 The CO methanation on Rh/CeO2 and CeO2/Rh model catalysts: a comparative study Introduction Experimental Ceria overlayers polycrystalline rhodium Thin film model catalysts [5] Results Discussion and conclusions Acknowledgements References