Characterization and Catalytic Activity for the Oxidation of Ethane and Propane on Platinum and Copper Supported on CeO2/Al2O3

Characterization and Catalytic Activity for the Oxidation of Ethane and Propane on Platinum and Copper Supported on CeO₂/Al₂O₃

R. Cataluña¹, N.R. Marcilio¹, J. Soria², V. Cortés Corberán² and A. Martínez-Arias²

¹ Escola de Engenharia - Departamento de Engenharia Química, UFRGS - Av. Luiz Englert, s/n° -

CEP 90040-040, Porto Alegre, RS, Brazil

² Instituto de Catálisis y Petroleoquímica, CSIC, Campus UAM, Cantoblanco, 28049 -

Madrid, Spain

(Received: November 5, 1997; Accepted: April 6, 1998)

Abstract - Ethane and propane oxidation on platinum and copper supported on Al₂O₃ and CeO₂/Al₂O₃ catalysts were studied comparatively by examining reaction rates as a function of temperature. Results show that the addition of cerium oxide shifts the catalytic activity to higher temperatures. This negative influence is less pronounced in the case of supported copper samples, which on the basis of EPR and FTIR of adsorbed CO results is attributed to the low relative amount of this metal is in contact with ceria. The decrease in activity the presence of ceria might be due to changes in metal particle size or to the stabilization of the oxidized states of the metals, induced by their interactions with cerium oxide. The higher activity of platinum, in comparison with copper, is attributed to its higher reducibility along with an easier hydrocarbon activation on that metal.

Keywords: Catalytic activity, oxidation, copper.

INTRODUCTION

A significant portion of environmental pollutants originate from automotive sources. Of these, hydrocarbons can be eliminated by means of catalytic oxidation methods. Hydrocarbons emitted by automobiles are mainly saturated hydrocarbons, which are usually more difficult to oxidize than olefines or aromatics (Yu Yao, 1980).

The use of three-way catalysts (TWC) is a universally accepted method of automobile emissions control. These are formed of five major components: the substrate, the support, stabilizers, base metal promoters and platinum group metals (Harrison et al., 1988). Model systems of these catalysts include high surface area g -Al₂O₃ as a support for the precious metals, with CeO₂ being added as a promoter. The promoting effect of cerium oxide is believed to include (Trovarelli, 1996) both structural aspects, such as the enhancement of the metal dispersion and stabilization of the g -Al₂O₃ support for thermal sintering, and chemical aspects such as, enhancement of the oxygen storage capacity of the systems, participation in the water-gas shift reaction and decomposition of nitrogen oxides. More recent efforts are devoted to elucidating the participation of ceria in important metal/support interactions that can substantially affect catalyst properties; it is generally recognized that these interactions are mainly responsible for the important promoting effect of ceria on these systems, these effects being connected to the particular redox properties acquired by the systems upon establishment of metal-ceria contacts (Trovarelli, 1996; Martínez-Arias et al., 1997).

In this work, catalytic activity for oxidation of ethane and propane on copper and platinum supported on Al₂O₃ and CeO₂/Al₂O₃ is studied. Of particular interest is the study of the supported copper catalysts, since this metal is one of the most promising potential substitutes of noble metals in TWC. The characteristics of the supported copper and platinum samples used for the catalytic tests (in their initial preoxidized state) are examined by means of EPR and FTIR of adsorbed CO, respectively, while differences in their reducibilities are analysed by TPR experiments using CO as the reducting gas.

EXPERIMENTAL

The incipient wetness impregnation method was applied to catalyst preparation using g -Al₂O₃ (in the form of 1.8 mm diameter spheres, supplied by Condea; S_BET= 200 m²g^-1) as the starting material. In order to avoid internal diffusion effects on the reaction rate measurement (Cataluña, 1995), the spheres were crushed and sieved to give particles with an average diameter between 0.50 and 0.84 mm.

The mixed support CeO₂/Al₂O₃ (containing 10 wt% CeO₂ hereafter referred to as 10CA) was prepared by impregnation with an aqueous solution of Ce(NO₃)_3× 6H₂O. The resulting solid was first dried in air for 24 h at 100^oC and then calcined at 550^oC during 4 h.

The Pt-containing catalysts were prepared by impregnation with an aqueous solution of [Pt(NH₃)₄](OH)₂ (Johnson Matthey), in order to obtain 0.5 wt% of Pt, followed by drying and calcination under the same conditions as those used for the 10CA sample. The platinum supported catalyst on g -Al₂O₃ will be referred to as 0.5Pt/A and that supported on 10CA as 0.5Pt/10CA.

In the preparation of Cu-containing catalysts, aqueous solutions of Cu(NO₃)_2× 3H₂O (Merck) were used in order to get 1 wt% of Cu; these was followed by drying and calcination, as in the usual method. The catalysts supported on g -Al2O3 and 10CA will be referred to as 1Cu/A and 1Cu/10CA, respectively.

All the gases employed are of commercial purity. For adsorption experiments, they had been further purified by vacuum distillation techniques before storage in appropiate containers attached to a conventional high vacuum line (capable of maintaining a dynamic vacuum of ca. 6 × 10^-3 N m^-2) in which these experiments are performed.

Measurements of catalytic activity of hydrocarbon oxidation were conducted in a quartz flow reactor maintained at the desired temperature by a tube furnace. Helium containing the reaction gases was sent through the catalyst at atmospheric pressure. In these experiments the space velocity used was 30,000 h^-1 and the catalyst volume was 0.5 cm³. The composition of the outlet gas was analyzed by gas chromatography, using porapak Q and molecular sieve columns in series/by-pass arrangements. Before catalytic testing, the samples were subjected in situ to a standard pretreatment consisting of heating under 3% mol O₂ at 500^oC during 1 h, cooling in the same gas to room temperature and then purging briefly (5 min) with N₂.

For the experiments temperature programmed reduction in CO (CO-TPR), after a standard precalcination treatment consisting of heating in a 3% O₂:N₂ flow at 500^oC during 1 h, cooling in the same gas flow to room temperature and then purging briefly (5 min) with N₂, the mixture was switched to a 1% CO/N₂ mixture at a gas flow rate of 200 cm³.min^-1 and, after a short equilibration period, a programmed temperature ramp was initiated at a rate of 5 K.min^-1 up to 700^oC. Analysis of the feed and outlet gas streams was performed using a Perkin-Elmer FT-IR spectrometer (model 1725X) coupled to a multiple reflection transmission cell (Infrared Analysis Inc. "long path gas minicell", 2.4 m path length, ca. 130 cm³ internal volume) or a VG 100-D quadrupolar mass spectrometer, using Ar as carrier gas.

EPR spectra were recorded at 77 K with a Bruker ER 200 D spectrometer operating in the X-band and calibrated with a DPPH standard (g = 2.0036). Portions of about 40 mg of sample were placed inside a special quartz probe cell that had greaseless stopcocks. A prolonged outgassing treatment at room temperature was performed before recording the spectra. Relative intensities were evaluated by computer double integration of the corresponding signals.

FTIR spectra were recorded at room temperature with a Nicolet 5ZDX Fourier Transform spectrometer with a 4 cm^-1 resolution and taking 128 scans for each spectrum. Thin self-supporting discs (ca. 10 mg cm^-2) were prepared and placed in standard greaseless cells provided with NaCl windows, where they could be subjected to thermal or adsorption treatments. The procedure for CO adsorption consisted in a prolonged outgassing of the corresponding sample at room temperature, followed by adsorption of CO up to an equilibrium pressure of 10 Torr.

RESULTS AND DISCUSSION

Reducibility of the samples

Figure 1A shows the CO consumption profiles obtained for the 1Cu/A and 1Cu/10CA samples. A small degree of CO consumption is observed immediately after beginning the TPR runs for both samples. More significant CO consumption is produced at temperatures higher than ca. 150^oC or 300^oC for 1Cu/10CA or 1Cu/A, respectively, and is larger for 1Cu/10CA. Larger CO consumption steps are detected at 450 and 500^oC for 1Cu/10CA and 1Cu/A, respectively. In order to perform a thorough analysis of these profiles, several processes leading to CO-consumption phenomena should be considered. Thus, in addition to the simple CO + O_s ® CO₂ reduction process (where Os denotes a surface oxygen atom), the water-gas shift (WGS) reaction involving reaction between CO and surface hydroxyl groups or CO disproportionation (Boudouard reaction: 2CO ® CO₂ + C), should also be taken into consideration. In addition, in the cerium-containing catalysts, both metal and ceria reduction processes can be produced. Recent results (Martínez-Arias et al., submitted), using EPR and FTIR techniques, show that reduction of most of the copper is produced at T_r ³ ca. 300^oC in both copper-containing samples. A smaller amount of copper, in contact with ceria, is reduced at lower temperatures. The CO-consumption step onset at 150^oC for 1Cu/10CA is mainly due to ceria reduction, although reduction of copper in contact with dispersed ceria entities is also involved in this consumption step. On the other hand, the consumption steps observed at higher T_r (450 and 500^oC for 1Cu/10CA and 1Cu/A, respectively) are mainly due to CO disproportionation phenomena, as revealed by TPD experiments (Martínez-Arias et al., submitted), although the contribution of the WGS reaction should also be considered.

Profiles of the temperature programmed reduction in CO of the 0.5Pt/A and 0.5Pt/10CA samples are shown in Fig. 1B. In the case of 0.5Pt/A, an initial weak CO consumption peak is centered at ca. 150^oC, while significant consumption is observed for T > 275^oC. For 0.5Pt/10CA, a low temperature CO uptake is first observed at around 130^oC, while the most significant consumption occurs from ca. 175^oC, producing a maximum centered at ca. 275^oC. These profiles, with the support of separate characterization results (Martínez-Arias et al., in preparation), indicate that platinum oxide reduction is involved in the CO consumption peaks observed at ca. 130-150^oC, although for 0.5Pt/10CA, certain ceria reduction is produced as well in that range; on the other hand, both platinum and ceria reduction and the WGS reaction are involved in the CO consumption at T_r ³ ca. 300 ^oC and 200 ^oC for 0.5Pt/A and 0.5Pt/10CA, respectively.

Figure 1:
CO-TPR profiles of different samples. (A) supported copper-samples, (B) supported platinum samples

On the basis of these results, the lower temperature needed to reduce platinum, as compared to copper, is noteworthy; while CO begins to reduce the platinum at room temperature, it does not reduce most of the copper until T_r ³ ca. 300^oC.

Characteristics of the Initial Preoxidized Samples

Figure 2 shows EPR spectra of 1Cu/A and 1Cu/10CA. Both spectra are very similar (only differing in the slightly lower intensity shown by the 1Cu/10CA spectrum) and are mainly formed by the overlapping of two Cu²⁺: a broad anisotropic signal showing extremes at g = 2.24 and g = 2.05, signal A, and a narrower axial signal with gº = 2.321 and gÁ = 2.057, signal B, in which a hyperfine pattern of four lines can be resolved in its two components, yielding coupling parameters Aº = 17.1 × 10^-3 cm^-1 and AÁ = 1.9 × 10^-3 cm^-1. The relative contribution of signal A is significantly higher for both cases (roughly 80% of total spectrum intensity, as ascertained by simulating the full spectrum with a combination of both signals and evaluating their integrated intensities). Signal A, showing a relatively broad linewidth and, as a consequence, unresolved hyperfine splitting, must be due to Cu²⁺ species interacting with one an other (i.e., generically clustered Cu²⁺), where the signal is broadened by magnetic dipolar interactions. Separate XPS experiments show the presence of peaks due to Cu²⁺ at 935.4 to 935.7 eV, along with satellite peaks at ca. 944 eV, which are typical of local environments for these ions of the copper aluminate spinel type. On the other hand, signal B, in which the hyperfine structure is well resolved, can be attributed to isolated Cu²⁺ ions in a tetragonally distorted (axial lengthening and planar shortening) symmetry (Berger and Roth, 1967). These results suggest that good copper dispersion is produced for both samples, which might explain the difficulty in reducing the samples, as noted above. On the other hand, the similarity between both spectra shows that most of the oxidized copper in these samples interacts with the alumina part of the catalysts, with only a small amount remaining in contact with ceria on the 1Cu/10CA catalyst.

FTIR spectra using CO adsorption at room temperature on the preoxidized platinum-supported samples are shown in Figure 3. No band becomes apparent after CO adsorption at room temperature for 0.5Pt/A. This result contrasts with that obtained for 0.5Pt/10CA, which presents a band at 2090 cm^-1. Separate experiments show no shift in frequency with changes in CO partial pressure. The absence of carbonyl bands in the 0.5Pt/A catalyst suggests that relatively large platinum oxide crystals showing a small concentration of defects are present in that sample. In the case of 0.5Pt/10CA, the characteristics of the band at 2090 cm^-1 suggest that it is due to carbonyls adsorbed on relatively dispersed oxidized platinum entities (Martínez-Arias et al., in preparation), most likely Pt-O surface complexes, stabilized by interactions with underlying ceria. This result indicates the presence of interactions between platinum and cerium oxide in this sample which, as seen below, produce important differences in the catalytic behaviour of the supported platinum samples.

Figure 2:
EPR spectra of 1Cu/A and 1Cu/10CA calcined samples.

Figure 3:
FTIR spectra following adsortion of 10 Torr of CO on the calcined samples.

Table 1:
Initial reaction temperatures (T

	Ethane		Propane
	Eap	Tir	Eap	Tir
1Cu/A	81	445	90	480
1Cu/10CA	95	460	83	490
0.5Pt/A	84	300	79	230
0.5Pt/10CA	83	345	76	280

_ir) and apparent activation energies (E_a) of the ethane and propane oxidation reactions

Catalytic behaviour of the samples

The chromatograms obtained for each alkane showed that, in all cases, complete conversion of the hydrocarbons to CO₂ was the only reaction. The profiles of reaction rate do not present any discontinuity, suggesting that there are no changes in the reaction mechanism in the course of the experiment. Table 1 shows characteristic points on the activity profiles. The results are reported as mol of HC converted to CO₂ per gram of catalyst versus temperature. Apparent activation energies (E_a) were calculated at conversion levels lower than 15%. Initial reaction temperature (T_ir) is arbitrarily attributed to the temperature at which a noticeable reaction rate is observed (1 × 10^-6 mol.s^-1.g^-1).

The results of catalytic activity are shown in Figure 4. In the case of the supported platinum samples, the results clearly show lower activity for the ceria-containing catalysts, in agreement with earlier results (Yu Yao, 1980). This might be due either to the stabilization by ceria of oxidized states of platinum or to the formation of smaller metallic platinum particles upon interaction with the reactant mixture of 0.5Pt/10CA. These are two hypotheses based on previous results from the literature which show lower HC oxidation activity for oxidized than for reduced catalysts (Yu Yao, 1980; Oh et al., 1991) and higher activity by larger metallic particles (Yu Yao, 1980; Hicks et al., 1990). In fact, characterization results show the stabilization of Pt-O entities in the 0.5Pt/10CA catalyst. A more difficult reduction of these entities or the formation of smaller platinum particles as a result of its ceria-induced larger dispersion in the cerium-containing catalyst might then explain the results observed.

Figure 4:
Reactivity profiles for the reaction of oxidation of ethane (A) and propane (B).

Smaller differences are observed for supported copper, with the activity of 1Cu/A being slightly higher than that of 1Cu/10CA for the oxidation of propane and somewhat higher for the oxidation of ethane. This correlates well with the mentioned fact that most of the copper interacts with alumina, rather than with ceria, in the 1Cu/10CA sample. The slightly higher activity for copper supported on Al₂O₃ might be due to effects similar to those mentioned for supported platinum. In this case, copper-ceria interactions might induce stabilization of a small amount of copper as partially oxidized Cu⁺ entities, as reported on in previous works (Martínez-Arias et al., submitted; Liu et al., 1995), which would display lower activity than completely reduced metallic copper entities. On the other hand, relatively smaller oxidized copper aggregates interacting with alumina are likely to be present on the initial 1Cu/10CA sample (Martínez-Arias et al., submitted), which could, upon reduction by interaction with the reactant mixture, lead to smaller sizes of metallic copper for the ceria-containing sample.

The slow step in the alkane oxidation reaction has been postulated to be the dissociative chemisorption of the HC on the bare metal surface with the breaking of the weakest C-H bond. This leads to the formation of adsorbed species, which are rapidly oxidized by interaction with co-adsorbed oxygen species. The differences in activation energy between both catalysts reflect the differences in the dissociative energy of the C-H bond broken during the adsorption process. The results obtained for the supported platinum catalysts suggest that in the case of propane, the breaking of the C-H bond of the -CH₂- group is easier than in the case of the C-H bond of the -CH₃ group of ethane, as expected (Martínez-Arias et al., submitted). In the case of the copper-supported catalyst, this result is not observed, suggesting that the limiting step of the reaction is different than thatfor the supported platinum catalysts. This might be related to the more difficult reduction of copper, as evidenced by the TPR experiments (Figure 1). A possible hypothesis to explain this behaviour might be that the temperature needed to generate a sufficient quantity of active reduced metallic copper particles upon interaction of the initial preoxidized catalysts with the reaction mixture, is high enough for the HC molecules to become activated for the reaction, with the formation of metallic copper being thus the limiting step in this case.

REFERENCES

Yu Yao, Y-F., Ind. Eng. Chem. Process Res. Dev. 19, 295 (1980).

Harrison, B.; Diwell, A.F. and Hallett, C., Platinum Metal Rev. 32 (2), 73 (1988).

Trovarelli., A., Catal. Rev. Sci. Eng. 38, 439 (1996), and references therein.

Martínez-Arias, A.; Soria, J.; Cataluña, R.; Conesa, J.C. and Cortés Corberán, V., Preprints of the CAPoC4 Congress, Brussels, vol. 2, p. 271 (1997).

Cataluña, R., Ph.D. diss., Universidad Politécnica de Madrid (1995).

Martínez-Arias, A.; Cataluña, R.; Conesa, J.C. and Soria, J., J. Phys. Chem. B (submitted).

Martínez-Arias, A.; Coronado, J.M.; Cataluña, R.; Soria, J. and Conesa, J.C., (in preparation).

Berger, P.A. and Roth, J.H., J. Phys. Chem. 71, 4307 (1967).

Oh, S.H.; Mitchell, P.J. and Siewert, R.M., J. Catal. 132, 287 (1991).

Hicks, R.F.; Rigano, C. and Pang, B., Catal. Lett. 6, 271 (1990).

Liu, W.; Sarofim, A.F. and Flytzani-Stephanopoulos, M., Chem. Eng. Sci. 49, 4871 (1995).

Publication Dates

• Publication in this collection
  09 Oct 1998
• Date of issue
  June 1998

History

• Accepted
  06 Apr 1998
• Received
  05 Nov 1997