Reaction mechanisms of carbon dioxide reforming of methane with Ru-loaded lanthanum oxide catalyst

Applied Catalysis A: General 179 (1999) 247±256

Reaction mechanisms of carbon dioxide reforming of methane with Ru-loaded lanthanum oxide catalyst Na-oko Matsui, Kengo Anzai, Noriyasu Akamatsu, Kiyoharu Nakagawa, Na-oki Ikenaga, Toshimitsu Suzuki* Department of Chemical Engineering and High Technology Research Center, Kansai University, Suita, Osaka 564-8680, Japan Received 2 September 1998; received in revised form 30 September 1998; accepted 30 September 1998

Abstract A pulsed reaction technique was applied to discuss the effect of support on the activities and mechanisms in the CO2 reforming of methane over Ru catalyst. The reaction was carried out using a ®xed bed reactor equipped with an on-line mass spectrometer. Four supports: La2O3, Y2O3 and ZrO2 which showed high activity and Al2O3, commonly used one in the reforming reaction, were compared when loaded with Ru. After feeding CO2 at 6008C, we introduced a pulse of CH4 over Ru/La2O3 catalyst under Ar steady ¯ow. We observed the response of CO which was generated from the reaction with CHx on the ruthenium and the Ru±Ox formed during CO2 treatment or during the reaction of Ru±CHx with adsorbed CO2 onto the La2O3. Over Ru/Al2O3 catalyst, however, very small response of CO was observed. A pulse of 13 CO2 was introduced under CH4 steady ¯ow over Ru/La2O3, Ru/Y2O3 and Ru/ZrO2 catalysts. Symmetrical 13 CO responses were observed, but a small response of 12 CO from 12 CHx continued to evolve after generation of 13 CO from 13 CO2 ceased. The following reaction cycle is believed to occur in the CO2 reforming of methane on active supports: A part of metallic ruthenium reacted with CH4 to give Ru±CHx; simultaneously ruthenium metal could be oxidized with CO2 to give Ru±Ox and CO; and then, oxygen transfer from Ru±Ox to Ru±CHx took place to give CO and metallic ruthenium. Distinct temperature increases in the catalyst bed for La2O3, Y2O3 and ZrO2 supports were observed with the introduction of CO2 pulses under Ar ¯ow. On the other hand, a very small increase in the temperature of the catalyst bed was observed on Al2O3. These results indicate that CO2 reforming of CH4 with ruthenium loaded catalysts was strongly assisted by the activation of CO2 adsorbed on the basic sites. # 1999 Elsevier Science B.V. All rights reserved. Keywords: Pulse reaction; Temporal analysis of products;

1. Introduction Synthesis gas is industrially produced by steam reforming of methane or light hydrocarbons with a *Corresponding author. Tel.: +81-6-388-8869; fax: +81-6-3888869; e-mail: [email protected]

13

CO2 ;

13

CH4 ; Zirconia; Yittria

supported nickel catalyst at a high temperature. A high loading Ni catalyst shows high activity (reaction (1)). However, due to carbon deposition on the catalyst, an excess amount of water must be supplied for the steady operation [1±4]. CH4 ‡ H2 O@CO ‡ 3H2

0926-860X/99/$ ± see front matter # 1999 Elsevier Science B.V. All rights reserved. PII: S0926-860X(98)00323-8

(1)

248

N.-o. Matsui et al. / Applied Catalysis A: General 179 (1999) 247±256

Recently, CH4 reforming reaction using CO2 instead of steam has attracted much attention (reaction (2)) [5,6]. This pathway can provide a more suitable H2/CO ratio for Fischer±Tropsch or methanol syntheses in combination with steam reforming. Moreover, CO2 and CH4 are believed to be the major global warming gases, consequently, CO2 reforming of CH4 is interested in view of their recovery and recycle use. CH4 ‡ CO2 ! 2CO ‡ 2H2

(2)

Group VIII metals are reported to be effective catalysts in this reaction. The conventional Ni based steam reforming catalysts tend to be coked due to the formation of stable nickel carbide on the surface of Ni catalysts to give whisker carbon [7]. On the other hand, platinum group metals afforded a high activity and long time stability without any or with only slight carbon deposition [8]. The combination of a metal and a support greatly affected catalytic activity and carbon deposition of this reaction. Rostrup-Nielsen and Bak Hansen [9] reported that the catalytic activity of the MgO-supported catalyst decreased in the following order: Ru>Rh>Ir>PtˆPd. On the other hand, Solymosi et al. [10] reported that the order is Ru>Pd>Rh>Pt>Ir when Al2O3 was used as a support. Zhang et al. [11] reported the effect of support on the Rh catalyst and concluded that the activity order is YSZ (yittria stabilized zirconia)> Al2O3>TiO2>SiO2>La2O3>MgO. With Pd catalyst, the activity order is reported as TiO2>Al2O3> SiO2>MgO [12,13]. With supported Ir catalysts, TiO2 is reported to be the best support by Solymosi et al. [14] and by our group [15]. There have been a large number of arguments about the reason why the combination of metal and support affects the catalytic activity. In general, a basic support seems to exhibit better results than an acidic support. Thus the mechanistic aspects in the CO2 reforming of CH4 on the supported transition metal catalyst needs attention. Methane is believed to be adsorbed and dissociated on the transition metal surface of Ni [16], Rh [8], Pt [10], and Pd [12,13], as shown in reaction (3). CH4 ad: ! CHx ad: ‡ …4 ÿ x†=2H2

(3)

Reactivities of CHx species formed on the metal surface varied with support, even if the same metal was employed [8,17].

In the CO2 reforming of CH4 over Rh catalysts, the dissociation of CO2 is reported to be the rate determining step; the role of the support was to promote dissociation of CO2 [18]. Mechanisms of this reaction are discussed with Ni or Rh catalyst [16,19±22]. Nonsteady state reaction technique, using isotope labeled CH4 or CO2, was reported to explain mechanisms of this reaction over Ni/SiO2 [23]. However, little has been discussed on Ru catalyst which showed high activity and stability without carbon deposition. This paper deals with pulsed reactions of ruthenium loaded catalysts, in order to compare differences in the active La2O3 and less active Al2O3 support by using CO2, 13 CO2 , CH4, and 13 CH4 pulsed reactions. 2. Experimental 2.1. Materials As a catalyst, RuCl3
nH2O (Ru: 38 wt% minimum) (Mitsuwa Kagaku) was used. g-Al2O3; ALO-4 (supplied from Catalysis Society of Japan), ZrO2 (Japan Aerosil), Y2O3 (Nacalai tesque), La2O3 (Wako Pure Chemical Industries) were used. All chemicals were used without further puri®cation. Supported Ru catalysts were prepared by impregnating acetone solution of RuCl3
nH2O to yield a 0.5 or 5.0 wt% ruthenium to various supports. All the catalysts were dried under reduced pressure and calcined in air at 5508C for 5 h and were pelletized and sieved to a size between 60 and 100 mesh. 2.2. Apparatus and procedure for pulse response reactions All reactions were performed by using a ®xed bed quartz reactor having an internal diameter of 4 mm and a length of 200 mm; the reactor was set in a horizontal position in an electric furnace. In the center part of the reactor, a quartz wool plug was packed and 50 mg of the catalyst was charged. Before the reaction, the catalysts were reduced under H2 ¯ow for 60 min at 6008C. CH4 or CO2 pulse except 13 CO2 was introduced with a 6-port gas sampling valve equipped with measuring tubes, under a stream of Ar carrier gas. The reaction temperature was controlled by monitoring the outside temperature of the reactor wall with chromel

N.-o. Matsui et al. / Applied Catalysis A: General 179 (1999) 247±256

249

alumel thermocouples by using a programmable controller. 13 CO2 was generated by acidifying Ba13 CO3 with perchloric acid solution and was stored in a gas burette. Analyses of the gases during the pulsed reactions were done by an on-line quadrupole mass spectrometer (HAL201, Hiden Analytical). The mass spectrometer scanned parent peaks of the following eight compounds: H2, CH4, 13 CH4 , H2O, CO, 13 CO, CO2, 13 CO2 within 1s; repeated scans were collected in a personal computer. Measured intensities were corrected for the relative sensitivities of the respective ions. 3. Results and discussion 3.1. Comparison of support between La2O3 and Al2O3 The combination of metal and support affected the catalytic activity in the CO2 reforming of methane at a low temperature. With Ru (0.5 wt%)/La2O3 at 6008C, CH4 conversion reached 28.7% under stoichiometric CH4 and CO2 feed ratio in a continuous ¯ow reactor. On the other hand, with Ru (0.5 wt%)/Al2O3, it decreased to 5.7%. Among other supports, Y2O3 and ZrO2 exhibited high catalytic activities in the CO2 reforming of methane (25±29%), but SiO2 or TiO2 gave low conversions of CH4 (less than 12%). Pulsed reactions were carried out in order to compare the effect of support between La2O3 and Al2O3. Fig. 1 shows responses to a CH4 pulse under Ar steady ¯ow after sweeping out CO2 for 120 s. Before the pulsed reaction, the catalyst was treated with a CO2 steady ¯ow for 5 min at 6008C. When the CH4 pulse was introduced onto Ru/La2O3, H2 was generated immediately. This indicates that CH4 rapidly decomposed to CHx species, and then a small amount of CO was produced. After the CH4 pulse passed through the catalyst bed, a very small amount of H2 and CO continued to evolve (Fig. 1(a)). These results indicate that CO2 was stored on the catalyst, although the CO2 supply was terminated. By contrast, with Ru/Al2O3, generations of CO and H2 were smaller as compared to the amounts observed in Ru/La2O3 and decreased rapidly after the CH4 pulse passed through the catalyst

Fig. 1. Responses to CH4 pulse after CO2 steady flow treatment of Ru/La2O3 (a) and Ru/Al2O3 catalysts. Reaction temperature: 6008C; catalyst: 50 mg; Ru: 5 wt%; carrier gas Ar: 10 ml/min; CH4 pulse size: 1.0 ml (ca. 41 mmol); interval between CO2 flow and CH4 pulse: 120 s; CO2 flow !300 s Ar !120 s CH4 pulse .

bed (Fig. 1(b)). These results indicate that no or only a very small amount of CO2 is adsorbed on the Ru/ Al2O3 catalyst. Zhang and Verykios [24] reported that a large CO2 pool was formed on the Ni/La2O3 catalyst in the form of La2O2CO3 under CO2 reforming reaction conditions. XRD patterns of the Ru/La2O3 catalyst, after the steady state reforming reaction at 6008C for 2 h, showed diffraction peaks ascribed to La2O2CO3, indicating that the reaction of CO2 onto La2O3 did occur with the Ru-loaded case. Such results suggest that CO2 was adsorbed on the La2O3 in the Ru/La2O3 catalyst and CH4 was activated and dissociated to CHx species over metallic ruthenium. CO2 from the La2O3(CO2) reacted with CHx species to give CO and H2.

250

N.-o. Matsui et al. / Applied Catalysis A: General 179 (1999) 247±256

Fig. 2. Responses to CO2 pulse after CH4 steady flow treatment of Ru/La2O3 (a) and Ru/Al2O3 catalysts. Interval between CH4 flow and CO2 pulse: 120 s, other conditions are the same as shown in the caption to Fig. 1; CO4 flow !300 s Ar !120 s CH2 pulse .

The smaller response of CO against CH4 pulse in the Al2O3-loaded case could be accounted for by a smaller adsorption of CO2 on Al2O3 or by lower reactivity of CHx species formed on Ru/Al2O3. The role of La2O3 in the Ni catalyst was studied by FT-IR, XPS and SIMS. It was concluded that CO2 adsorbed on La2O3 (La2O2CO3) would prevent the nickel phase from being shielded by carbon deposition [24,25]. Similarly, on Ru-loaded La2O3, no carbon deposition was observed. Fig. 2 shows responses to pulsing CO2 under Ar steady ¯ow, after sweeping off initially reacted CH4 for 120 s. Before the pulsed reaction, CH4 was supplied to the catalyst for 5 min at 6008C. When the CO2 pulse was introduced, no detectable amounts of H2 and H2O were generated in either catalyst. The generation

of H2O would be expected between the reaction of CHx and CO2 [23], but no H2O was detected possibly due to the adsorption onto the wall of the reaction apparatus. In the case of Ru/La2O3 catalyst, an asymmetric and broad peak of CO2 and a small amount of CO generation continued after CO2 pulse passed through the catalyst bed (Fig. 2(a)). This result indicates that CO2 was adsorbed on the La2O3 and desorbed gradually. During this stage with Ru/La2O3 catalyst, CO2 reacted with carbon species (CHx) formed on the Ru or with oxidized metallic ruthenium to give RuOx and CO. On the contrary, with Ru/Al2O3 catalyst, CO2 was detected as a sharp and symmetrical peak and a very weak response of CO was observed. These results suggest that adsorbed and activated CO2 on the support would preferentially react with carbon species on ruthenium or oxidize metallic ruthenium to Ru±Ox. Fig. 3 shows responses to repeated pulses of CH4 after CO2 was adsorbed for 5 min at 6008C and 8008C with the ruthenium catalysts at loading levels of 0.5 and 5.0 wt%. The amounts of CO and H2 produced were of the same order, irrespective of the ruthenium loading level onto the La2O3. Such a behavior is consistent with the results obtained in the steady ¯ow reaction, where conversion of CH4 did not increase above Ru loading level of 0.5 wt% to La2O3. At 8008C, a larger amount of CO2 would be adsorbed, as indicated by the larger amounts of CO and H2 against the ®rst CH4 pulse. However, against responses to successive CH4 pulses, the amounts of CO and H2 and CH4 conversion decreased rapidly, due to consumption of adsorbed CO2 or rapid desorption of CO2 from La2O3 surface. On the other hand, at 6008C, conversions of CH4 were much lower as compared to those at 8008C. However, in the successive pulses, CH4 conversions were kept constant, indicating slower desorption of CO2 at 6008C. Fig. 4 shows responses to CH4 pulse into a CO2 steady ¯ow with Ru/La2O3 and Ru/Al2O3 catalysts at 6008C. In the CH4 conversion and responses of H2 and CO, no differences were seen between two supports, under the ¯ow of a high concentration of CO2. In both cases, H2 responses were symmetrical, and after CH4 passed through the catalyst bed, H2 responses were not observed. On the contrary, CO was generated after CH4 pulse passed through the catalyst bed. Osaki et al.

N.-o. Matsui et al. / Applied Catalysis A: General 179 (1999) 247±256

251

Fig. 4. Response to CH4 pulse into CO2 steady flow. Flow rate of CO2: 10 ml/min, CH4 pulse size: 0.4 ml; other conditions are the same as those shown in the caption to Fig. 1.

Fig. 3. Responses to the repeated pulses of CH4 after CO2 steady flow for 5 min. Solid lines indicate Ru loading level of 5 wt%, dotted lines indicate Ru loading level 0.5 wt%, open symbols indicate reaction at 6008C, solid symbols indicate reaction at 8008C. Interval of repeated pulse: 60 s; other conditions are the same as those shown in the caption to Fig. 1.

found similar behavior over Ni or CO loaded on Al2O3 catalyst, for the production of CO and H2 by pulsing CH4 under CO2 steady ¯ow. They concluded that CH4 was dissociatively adsorbed on the catalysts to release gaseous H2 and that the resultant adsorbed hydrocarbon species (CHx) gradually reacted with CO2 to

produce CO and H2 [18,26,27]. These results suggest that CO2 reforming of methane proceeded by rapid hydrogen abstraction from CH4 to give H2 and CHx± Ru species, followed by a slow reaction of CO2 with Ru±CHx. Fig. 5 shows responses against pulsing CO2 into a CH4 steady ¯ow. Over Ru/La2O3 catalysts, formation of CO was detected immediately when CO2 was introduced. A large H2 response started to appear after a few seconds; a small amount of unreacted CO2 was detected with delay. CO and H2 continued to appear after the supply of CO2 was terminated (Fig. 5(a)). These results indicate that the following reactions occur: Ru=La2 O3 ÿCO2 ad: ! CO…gas† ‡ RuÿOx =La2 O3 …fast CO formation† (4)

252

N.-o. Matsui et al. / Applied Catalysis A: General 179 (1999) 247±256

Fig. 5. Response to CO2 pulse into CH4 steady flow. Flow rate of CH4: 10 ml/min, CO2 pulse size: 0.25 ml; other conditions are the same as those shown in the caption to Fig. 1.

Fig. 6. Responses to 13 CO2 pulse into CH4 steady flow. Conditions are the same as those shown in Fig. 5.

3.2. Isotope labeled pulse reaction RuÿCHx ‡RuÿOx ! CO…gas†‡H2 …gas†‡Ru…metal† …late CO formation†

(5)

When the same reaction sequence was conducted on Ru/Al2O3, only about 30% of the supplied CO2 reacted and very small amounts of CO and H2 were observed (Fig. 5(b)). Effective adsorption of CO2 is very weak on Al2O3. Unlike the case on La2O3 support, where CO2 is able to be concentrated. Ruthenium was covered with Ru± CHx species during CH4 steady ¯ow in both supports, but due to weaker activation of CO2 on Al2O3, the reaction of CO2 and Ru±CHx occurred only slightly and the response of CO was very weak as compared to that for La2O3.

The same reaction sequences as above were carried out using isotope labeled 13 CO2 or 13 CH4, in order to distinguish CO formed from CHx species and CO2. Fig. 6 shows responses to 13 CO2 pulse into a 12 CH4 steady ¯ow. Over Ru/La2O3 catalyst, simultaneous responses of 13 CO and 12 CO were observed with different patterns. After generation of 13 CO decreased to a lower level, 12 CO continued to appear (Fig. 6(a)). The symmetrical 13 CO response is ascribed to the reaction of 13 CO2 and a weaker and continued response of 12 CO is ascribed to 12 CHx . CH4 pulsed reaction was carried out by Hu and Ruckenstein [28] over the reduced NiO/MgO catalyst after the reaction of CO2 reforming of methane for 2 h. The CO and CO2 were detected during the successive CH4 pulses over the used catalyst under an He steady

N.-o. Matsui et al. / Applied Catalysis A: General 179 (1999) 247±256

253

¯ow, but their responses decreased with increasing number of pulses. This indicates that a certain amount of the Ni active sites were oxidized with CO2 during the reaction [28]. The results of XRD analyses of Ru/La2O3 catalyst after CO2 reforming exhibited diffraction peaks assignable to RuO2 and to Ru metal. A pulsed reaction with labeled CO2 indicates that ruthenium metal could be oxidized with 13 CO2 to give 13 CO and Ru oxides (reaction 6). In the latest stage of the reaction, ruthenium oxides reacted with Ru±CHx to give 12 CO and metallic ruthenium (reaction 7). Ru …metal† ‡13 CO2 ad: !13 CO …gas† ‡ RuÿOx 12

RuÿOx ‡ Ruÿ CHx !

12

(6)

CO …gas†

‡H2 …gas† ‡ Ru …metal†

(7)

Metallic ruthenium would be reproduced as an active site and this reaction cycle could be repeated in the CO2 reforming of methane. Ross et al. [29] investigated the mechanism of CO2 reforming of CH4 on Pt/ZrO2 catalyst, by using a temporal analysis of products (TAP) reactor system, which can introduce different types of pulses. They concluded that CO2 and CH4 were independently dissociated on Pt/ZrO2. CO2 acted as oxygen supplier, while CH4 abstracts the oxygen. They proposed formation of an oxygen pool on the catalyst. Our results are quite similar to these ®ndings, except that no CO2 formation in the reaction with CH4 and no oxygen pool were observed. As shown in Figs. 5 and 6, weak H2 response (1.210ÿ7±1.510ÿ7 Torr) as compared to the background level of CO and CO2 was observed during CH4 steady ¯ow, and rapid increases in the H2 response were seen after pulsing 12 CO2 or 13 CO2 on Ru/La2O3 catalyst. Methane was decomposed on the surface of Ru metal, and ruthenium was transformed into Ru± CHx species. When 13 CO2 was introduced, after a small amount of 13 CO formation, Ru±Ox species could be formed (reaction (4)), followed by reaction (5) to give metallic ruthenium. This caused a large amount of H2 formation, which was observed during the initial few seconds when CH4 was introduced to the fresh Ru/ La2O3 catalyst. In the Ru/Al2O3 catalyst, the amounts of 13 CO and 12 CO against response to 13 CO2 were almost the same.

Fig. 7. Response to CO2 pulse into 13 CH4 steady flow. Conditions are the same as those shown in Fig. 5.

An XRD analysis of Ru/Al2O3 catalyst after CO2 reforming exhibited diffraction peaks assignable to Ru metal. The result indicates that oxidation of ruthenium with 13 CO2 would be dif®cult to occur, since adsorption of 13 CO2 on the Al2O3 surface would be very weak as evidenced by the large response of unreacted 13 CO2 . Fig. 7 shows responses to pulsing 12 CO2 into the 13 CH4 steady ¯ow. In contrast to pulsing 13 CO2 into 12 CH4 , 12 CO was observed against the 12 CO2 pulse. A rapid and distinct appearance of 12 CO was observed on the Ru/La2O3 catalyst. After 12 CO response decreased to a lower level, 13 CO continued to appear. This again indicates that metallic ruthenium or ruthenium partially covered with CHx would be oxidized with 12 CO2 to give 12 CO and Ru±Ox species, followed by oxygen transfer to CHx species, affording 13 CO. Delayed response of H2 is ascribed to the formation of

254

N.-o. Matsui et al. / Applied Catalysis A: General 179 (1999) 247±256

Fig. 9. Responses of CO2 onto various oxide supports. Reaction temperature: 6008C; Oxide: 50 mg; CO2 pulse size: 1.0 ml; Ar carrier gas flow rate 10 ml/min.

occur. These reactions are characteristic of the catalysts which showed high activities. 3.3. Adsorption of CO2 onto various supports

Fig. 8. Responses to 13 CO2 pulse into CH4 steady flow on Ru/ZrO2 and Ru/Y2O3 catalysts. Conditions are the same as those shown in Fig. 5.

metallic ruthenium sites after removal of surface oxides and CHx species. On Ru /Al2O3 very weak responses of 12 CO and 13 CO were observed with a large amount of unreacted 12 CO2. 13 CO2 pulsed reactions into a CH4 steady ¯ow were done on Ru/Y2O3 and Ru/ZrO2 catalysts which showed high activities in the CO2 reforming of methane. Similar to the Ru/La2O3 catalyst, against 13 CO2 pulse in the CH4 steady ¯ow 12 CO responses were observed with delay, and 12 CO formation continued after generation of 13 CO decreased to a very low level (Fig. 8). Delayed formation of 12 CO after a rapid 13 CO response indicates that over the Ru/Y2O3 and Ru/ZrO2 catalysts, reactions (6) and (7) seem to

Since adsorption of CO2 onto the support at an elevated temperature is an important factor for the higher catalytic activities, adsorption of CO2 on the support was measured by the CO2 pulsed reaction at 6008C. The results are shown in Fig. 9. Decrease in the CO2 peak area as compared to the Al2O3 seems to indicate the adsorption of CO2 onto the support. On ZrO2 and La2O3 supports, very slight adsorption of CO2 was observed, as evidenced by the smaller responses of CO2 compared to pulsing the same amount CO2 without catalyst. In particular, in the case of La2O3, an unsymmetrical response was observed with tailing. This result indicates that adsorption of CO2 and delayed desorption occurred on the La2O3 support. On Y2O3, however, CO2 was not adsorbed under the conditions employed. In the XRD analyses of Ru/Y2O3 or ZrO2, different from Ru/La2O3 catalyst which formed La2O2CO3 after steady state reaction, no changes in the support were observed after CO2 reforming at 6008C. Rapid adsorption and activation of CO2 might have occurred on the support, as seen in the symmetrical responses to 13 CO2 pulses in the CH4 ¯ow. In order to detect weak adsorption of CO2, temperature changes in the catalyst bed was measured, by

N.-o. Matsui et al. / Applied Catalysis A: General 179 (1999) 247±256

255

and Y2O3, chemisorption of CO2 did occur, but weak adsorption (physisorption) occurred on Al2O3. In the 13 CO2 pulsed reaction, Ru/Y2O3 and Ru/ ZrO2 catalyst exhibited the same behavior as Ru/ La2O3. Although the oxidized form of ruthenium (RuO2) on Ru/Y2O3 and Ru/ZrO2 catalysts was not detected in the XRD analyses, oxidation of ruthenium on these catalysts would occur to give surface oxides with CO2. 4. Proposed mechanism

Fig. 10. Variation in the catalyst bed temperature against pulsing CO2 at 6008C. Thin-walled sheathed thermocouples (K) were directly inserted into the center of the catalyst bed. Conditions are the same as shown in Fig. 9.

inserting very thin walled sheathed ®ne thermocouples into the catalyst bed. Fig. 10 shows variations of catalyst bed temperatures at 6008C, by pulsing a 1 ml of CO2 in steady ¯ow of Ar. With La2O3, ZrO2, and Y2O3, the catalyst bed temperature increased by about 1.0±1.58C, but the temperature increase was only about 0.38C with Al2O3. The increases in the temperatures seem to be associated with the heat of adsorption of CO2. These results again revealed that, with La2O3, ZrO2

Above results lead to the following reaction scheme in the CO2 reforming over ruthenium loaded catalyst: 1. When Ru loaded on La2O3 or Al2O3 was reduced and reacted with CH4 at 6008C, H2 was generated to give Ru±CHx species on both La2O3 and Al2O3. 2. When CH4 pulse was introduced onto Ru/La2O3 under Ar steady flow after CO2 was supplied for 5 min, responses of H2 and CO were observed. On the other hand, with Ru/Al2O3, generation of CO and H2 was very small. 3. On the contrary to case 2, by pulsing CO2 under Ar after a steady feed of CH4, the response of CO was only observed with Ru/La2O3. 4. Into the steady flow of CH4, responses of CO2 pulses were analyzed by using labeled 13 CH4 and 13 CO2 , in order to distinguish CO from CH4 and CO2. CO2 was believed to oxidize metallic Ru to Ru±Ox to give CO. The reaction of Ru±Ox with Ru± CHx afforded another CO. 5. Differences in the reactivity between La2O3 and Al2O3 loaded cases would be ascribed to the capability of adsorption of CO2 onto the oxides. Heat of adsorption of oxides was directly estimated by the temperature jump measurement at an elevated temperature. La2O3, Y2O3, and ZrO2 exhibited significant temperature rises by pulsing CO2. However, only a slight increase in the temperature of the catalyst bed was observed for Al2O3, indicating only physisorption of CO2 would occur on Al2O3. These ®ndings lead to the conclusion that CO2 reforming of CH4 on Ru-loaded catalyst proceeds as follows: (i) Ru±CHx formation with metallic ruthenium and CH4, (ii) oxidation of Ru with CO2 activated on the La2O3 support to give Ru±Ox and CO, (iii) oxygen transfer from Ru±Ox to Ru±CHx to give

256

N.-o. Matsui et al. / Applied Catalysis A: General 179 (1999) 247±256

another CO and probably H2. Consequently, metallic Ru could be regenerated. Acknowledgements This work was partially supported by the Grand-in Aid on Priority Area no. 09 218 255 from the Ministry of Education, Science, Culture, and Sports of Japan. K. Nakagawa is grateful to his research assistantship from the High Technology Center, Kansai University. References [1] E. Kikuchi, K. Ito, Y. Morita, Bull. Jpn. Petrol. Inst. 17 (1975) 206. [2] E. Kikuchi, A. Machino, Y. Ishikawa, N. Ishii, Y. Morita, Sekiyu Gakkaishi 29 (1986) 469. [3] E. Kikuchi, S. Uemiya, A. Koyama, A. Machino, T. Matsuda, Sekiyu Gakkaishi 33 (1990) 152. [4] M. Chai, M. Machida, K. Eguchi, H. Arai, Chem. Lett. (1993) 41. [5] J.T. Richardson, S.A. Paripatyadar, Appl. Catal. 61 (1990) 293. [6] A.T. Ashcroft, A.K. Cheetham, M.L.H. Green, D.D.F. Vernon, Nature 352 (1991) 18. [7] T. Sodesawa, A. Dobashi, F. Nozaki, React. Kinet. Catal. Lett. 12 (1979) 107. [8] A. ErdoÈhelyi, J. CsereÂnyi, F. Solymosi, J. Catal. 141 (1993) 287. [9] J.R. Rostrup-Nielsen, J.-H. Bak Hansen, J. Catal. 144 (1993) 38.

[10] F. Solymosi, G. Kutsan, A. ErdoÈhelyi, Catal. Lett. 11 (1991) 149. [11] Z.L. Zhang, V.A. Tsipouriari, A.M. Efstathiou, X.E. Verykios, J. Catal. 158 (1996) 51. [12] A. ErdoÈhelyi, J. CsereÂnyi, E. Papp, F. Solymosi, Appl. Catal. A 108 (1994) 205. [13] F. Solymosi, A. ErdoÈhelyi, J. CsereÂnyi, A. FalveÂgi, J. Catal. 147 (1994) 272. [14] A. ErdoÈhelyi, J. CsereÂnyi, F. Solymosi, Stud. Surf. Sci. Catal. 107 (1997) 525. [15] K. Nakagawa, K. Anzai, N. Matsui, N. Ikenaga, T. Suzuki, Y. Teng, T. Kobayashi, M. Haruta, Catal. Lett. 51 (1998) 163. [16] P. Turlier, E. Brum Pereira, G.A. Martin, ICCDU, Bari, Italy, 1993, p. 119. [17] T. Osaki, H. Masuda, T. Mori, Catal. Lett. 29 (1994) 33. [18] J. Nakamura, K. Aikawa, K. Sato, T. Uchijima, Catal. Lett. 25 (1994) 265. [19] T. Osaki, T. Horiuchi, K. Suzuki, T. Mori, Catal. Lett. 44 (1997) 19. [20] Z.L. Zhang, X.E. Verykios, Catal. Lett. 38 (1996) 175. [21] J. Rasko, F. Solymosi, Catal. Lett. 46 (1997) 153. [22] A.M. Efstathiou, A. Kladi, V.A. Tsipouriari, X.E. Verykios, J. Catal. 158 (1996) 64. [23] V.C.H. Kroll, H.M. Swaan, S. Lacombe, C. Mirodatos, J. Catal. 164 (1997) 387. [24] Z.L. Zhang, X.E. Verykios, Appl. Cata. A 138 (1996) 109. [25] Z.L. Zhang, X.E. Verykios, S.M. MacDonald, S. Affrosman, J. Phys. Chem. 100 (1996) 744. [26] T. Osaki, H. Masuda, T. Horiuchi, T. Mori, Catal. Lett. 34 (1995) 59. [27] T. Osaki, T. Horiuchi, K. Suzuki, T. Mori, J. Chem. Soc., Faraday Trans. 92 (1996) 1627. [28] Y.H. Hu, E. Ruckenstein, Catal. Lett. 43 (1997) 71. [29] A.N.J. van Keulen, K. Seshan, J.H.B.J. Hoebink, J.R.H. Ross, J. Catal. 166 (1997) 306.