- Email: [email protected]
Characterization and regeneration of Pt-catalysts deactivated in municipal waste flue gas
Applied Catalysis B: Environmental 69 (2006) 10–16 www.elsevier.com/locate/apcatb
Characterization and regeneration of Pt-catalysts deactivated in municipal waste flue gas Søren Birk Rasmussen a,*, Arkady Kustov a, Johannes Due-Hansen a, Bernard Siret b, Frank Tabaries b, Rasmus Fehrmann a a
Department of Chemistry and Center for Sustainable and Green Chemistry, Department of Chemistry, Technical University of Denmark, Denmark b LAB S.A., 25 rue Bossuet, 69455 Lyon Ce´dex 06, France Received 2 February 2006; received in revised form 8 May 2006; accepted 12 May 2006 Available online 27 June 2006
Abstract Severe deactivation was observed for industrially aged catalysts used in waste incineration plants and tested in lab-scale. Possible compounds that cause deactivation of these Pt-based CO oxidation catalysts have been studied. Kinetic observations of industrial and model catalysts showed that siloxanes were the most severe catalyst poisons, although acidic sulfur compounds also caused deactivation. Furthermore, a method for on-site regeneration without shutdown of the catalytic flue gas cleaning system has been developed, i.e. an addition of H2/N2 gas to the off-gas can completely restore the activity of the deactivated catalysts. # 2006 Elsevier B.V. All rights reserved. Keywords: Platinum; CO oxidation; Poisoning; IR spectroscopy; Regeneration; Hexamethyldisiloxane; Catalytic oxidation; Hydrogen
1. Introduction Burning municipal waste or sludges instead of landfill depositing is an important issue from an environmental point of view, since this ‘‘waste to energy’’ conversion also implies several benefits such as control of toxic emissions and heavy metal waste, etc. In this connection the treatment of gaseous emissions containing carbon and nitrogen oxides as well as volatile organic compounds has been of increasing concern in recent years. Thermal incineration, catalytic oxidation and adsorption are commonly used for removing these pollutants. Thermal incineration requires high operating temperatures and high capital cost facilities. If the gaseous stream also includes halogenated compounds, thermal incineration can produce toxic halogenated compounds under certain operating conditions. In some cases, adsorption by adsorbents such as carbon is an alternative. However, this process does not destroy pollutants, but merely concentrates them. Moreover, adsorption
* Corresponding author. Tel.: +45 4525 2390; fax: +45 4588 3136. E-mail address: [email protected] (S.B. Rasmussen). 0926-3373/$ – see front matter # 2006 Elsevier B.V. All rights reserved. doi:10.1016/j.apcatb.2006.05.009
efficiency can be adversely influenced by fluctuating concentrations of the gaseous components. Catalytic oxidation of gaseous organic emissions operates at significantly lower temperatures and shorter residence time than thermal incineration. It is also a common method for removal of CO emissions from industrial or municipal waste gases. Platinum supported on different oxide carriers is traditionally used as the catalyst for the oxidation of CO and hydrocarbons. At high temperatures it does not dissolve in the washcoat, but sintering into larger particles occurs, especially if the catalyst is exposed to temperatures above 700 8C. This leads to a substantial loss of a platinum surface area and a less efficient catalyst. This work focuses on a problem at a sewage sludge plant utilizing a Pt-based catalyst dispersed on an oxide support of proprietary nature for CO oxidation and operating at 250 8C in the tail end of the flue gas cleaning system. It was observed several times that the catalyst deactivated few days after startup of the gas cleaning system. The usual causes of deactivation of the catalyst during exposure to the flue gases are thermal damage as mentioned above, mechanical damage of the monolith or poisoning by
S.B. Rasmussen et al. / Applied Catalysis B: Environmental 69 (2006) 10–16
contaminants. Since the first parts are not relevant in this case, the poisoning of the catalyst by contaminants by components originating from the flue gas is suspected to be the reason for deactivation. According to the literature, possible deactivating flue gas components can be divided into four major classes [1– 13]:
silicium containing compounds; halogenated compounds; hydrocarbons and carbon deposits (coke); minor deactivating agents like SO2, Pb, As, K, phosphates, etc.
The printing industry is a typical example where organosilicon compounds often are present in the waste flue gas. The deposition of silicon residues on the active sites causes poisoning and deactivation of the catalyst. Cosmetics seem to be another important source of these compounds in waste. In early investigations [1,2], hexamethyldisiloxane (HMDS) was found to almost irreversibly deactivate alumina supported Pt and Pd catalysts for methane oxidation, whereas the oxidation of propene and butane over these catalysts was only reversibly inhibited in the presence of HMDS. The authors [1,2] explained these two types of deactivation processes by proposing a dualsite mechanism in which HMDS adsorbs irreversibly on the most active (high-energy) sites, necessary for methane oxidation, while in the case of less active (low-energy) sites used for propene and butane oxidation, HMDS adsorbs reversibly. More recently, it has been reported that HMDS deactivates the catalytic function of platinum based catalysts by forming a SiO2 overlayer which covers the Pt surface [3,4]. The catalytic activity can be recovered during regeneration by: reorganization of SiO2 films into islands which recovers the Pt surface [3]; release of silicon species from the surface [4]. The researchers performed regeneration procedures by exposure of the deactivated samples to air at 25 8C for 15 h [5]. In the case of long-term exposure to the silicon compounds, the Si content becomes too high, and the deactivation of the catalyst seems to be irreversible. Halogenated hydrocarbons are included in the list of volatile organic compounds. Oxidation of these compounds yields acidic gases as reaction products. These products are known to have a substantial poisoning effect on catalysts [6]. In this case catalysts with high noble-metal content (generally 0.2–1.5% Pt) are more poison resistant. For the catalysts with lower Pt content the activity can be completely regenerated by treatment with hot air at 350 8C. It has been reported that exposure of Pt/Al2O3 catalysts in a chlorine atmosphere may result in either sintering or redispersion of the platinum, depending on the nature and concentration of the chlorine-containing species. Adsorbed chlorine on the catalyst support may also produce platinum chloride complexes. It has also been reported that halogenated hydrocarbons may act as catalytic inhibitors, adsorbing on the
11
sites normally required for the activation and adsorption of CO and oxygen [6]. On the other hand in Ref. [7] it was shown that the presence of HCl in the flue gas has no inhibiting effect on the oxidation of hydrocarbons on Pt/Al2O3 catalyst, and no loss of Pt-dispersion was observed in this case. Moreover, according to the Pt-dispersion measurements, it was found that in some cases the presence of chlorine compounds prevents formation of larger platinum crystallites (sintering). Sulfur, although a potential poison for all metals, interacts relatively weakly with platinum. Moreover, some beneficial effect of SO2 on the activity of Pt/Al2O3 for oxidation of hydrocarbons was observed in Refs. [10,11]. A small fraction of SO2 that is oxidized further to SO3 may react with the alumina to give Al2(SO4)3, causing loss of surface area of the washcoat. Small negative effects of sulfation were observed for Pt/TiO2 in Ref. [12]. In addition lead and phosphorous are considered as traditional poisons for the catalysts [13]. The influence of water on the stability and activity of Pt/g-Al2O3 catalysts was studied in Ref. [14]. Measurements made in this work showed that the effect of water vapor in the feed gas (10%) on the activity and stability of the catalyst is quite negligible. This work reports on experimental studies of poisoning of a model catalyst prepared in the lab, and the results are compared with kinetic data obtained for actual industrial catalysts— either as fresh catalysts or as deactivated catalysts after actual use on industrial scale. Based on the literature summarized above and some preliminary investigations Na2S, H2SO4, HCl and hexamethylsiloxane (HMDS) were selected as model poisons. 2. Experimental 2.1. Catalysts preparation Samples of fresh and industrially deactivated catalysts of proprietary nature were obtained from the incineration plant. Model Pt/TiO2 catalysts were prepared according to the following procedure: a 100 ml solution of 131.6 mg H2PtCl66H2O (38% Pt, Fluka) in water was prepared and 10 g of TiO2 (anatase, Millenium Chemicals, pre-calcined at 500 8C) was added to the solution, similar to the procedure described by Benvenutti et al. [16], Green et al. [17], Rasko´ [15] and Hadjiivanov [18]. The mixture was stirred overnight. Water was removed in a vacuum rotational evaporator at 60 8C, and after drying residual (HCl) was removed by calcination at 400 8C for about 30 h, ending up with a 0.5 wt.% Pt/TiO2 model catalyst. Samples of this model Pt/TiO2 catalyst were poisoned by hexamethyldisiloxane (HMDS, >98.5%, Fluka), HC1, H3PO4, and H2SO4 (all from Merck, analysis grade). The poison in question was transferred to an ampule containing 1 g of catalyst sample without being in direct contact, sealed in dry air and heated to 250 8C for several days to ensure equilibrium through gas phase transport. The amount of each poison transferred to a catalyst sample was adjusted such that the bulk molar ratio between poison and platinum was 2:1.
12
S.B. Rasmussen et al. / Applied Catalysis B: Environmental 69 (2006) 10–16
2.2. CO oxidation activity measurements The catalytic activity of all samples was tested in laboratoryscale fixed-bed reactors according to the following procedure: about 50 (0.6) mg powder sample was placed in a tubular Pyrex reactor, fixed with quarts wool. The gas stream, controlled by Bronkhorst Mass Flow Meters, consisting of 79.8% N2, 20% O2 and 0.2% CO, was passed through the reactor at a total flow rate of 100 ml/min. The reactor was placed and preheated in a furnace at 250 8C for 1 h in order to equilibrate the system. Equilibrating the sample at 250 8C before direct measuring of the catalytic activity was found to be critical for obtaining reproducible results relating realistically to industrial conditions. Then, the temperature was lowered with the rate of 1 8C/min and the CO signal was monitored in the temperature interval 140–250 8C. The CO concentration was determined using a Hewlett Packard Gas Chromatograph, HP6890, with a 60 m micro-capillary column equipped with a TCD detector. Data acquisition was controlled by HP Chemstation software. The outlet gas composition was measured every 10 min, and the areas of N2, O2 and CO in the output-signal was used for the calculation of the CO concentration. Prior to loading, the sample was pressed into pellets then crushed and sieved to obtain catalysts with particle size in the interval 0.295–0.71 mm. 2.3. Catalyst characterization Self-supporting disks of the sample were prepared for high temperature-FTIR spectroscopy by pressing about 30–50 mg of the sample powder into 1.3 cm diameter discs. About 10 tonnes pressure was applied for 5 min. FTIR-measurements at high temperature were carried out on a Paragon 1000 Perkin-Elmer FTIR spectrometer, equipped with a water cooled, electrically heated furnace with argon flow through the sample chamber. All spectra were measured in the range 4000–450 cm 1 with 64 scans and a spectral resolution of 8 cm 1, in the temperature range 25–350 8C.
Fig. 1. Temperature dependence of CO conversion for fresh and deactivated industrial catalyst (before and after 1 month of exposure to flue gas). Gas composition: 10% O2, 0.2% CO, N2; GHSV = 30,000 H 1.
50% conversion was obtained, changed from 200 8C for the fresh catalyst to around 255 8C for the exposed catalyst. This extent of deactivation is quite critical, since the plant is operated at 250 8C, and therefore insufficient CO removal was experienced during normal plant operation. At the same time no significant change in the BET surface area was observed between fresh and deactivated catalysts. In order to make a systematic study of the Pt/TiO2 system with respect to poison and regeneration possibilities, a model catalyst was synthesized as described in Section 2. The resulting catalyst
2.4. Flue gas siloxane analysis Siloxanes were collected on-site by using a standard sampling train. The flue gas was bubbled through a chilled liquid (n-heptane/methanol) in an impinger. The solution was sent to the specialized laboratory CERTECH, Belgium. By GC/ MS, utilizing the relevant fragments obtained by MS, the different siloxanes were identified and quantified. CERTECH was selected because it is one of the few laboratories that have experience with siloxane quantification. 3. Results and discussion 3.1. Deactivation studies The commercial catalyst samples before and after exposure to incineration plant flue gas were tested for CO conversion versus temperature (Fig. 1). The T50%, the temperature where
Fig. 2. Temperature dependence of CO conversion for model poisoned catalysts. Gas composition: 10% O2, 0.2% CO, N2; GHSV = 30,000 H 1.
S.B. Rasmussen et al. / Applied Catalysis B: Environmental 69 (2006) 10–16 Table 1 T50% values for the fresh and poisoned model catalysts Sample
T50% (8C)
Pt/TiO2 H3PO4 HCl H2SO4 HMDS
213 218 238 242 250
had a surface area of 70–80 m2/g and was subsequently tested for CO oxidation activity. The results of this test are given in Fig. 2, where it is shown that the model catalyst reproduces quite well the behavior of the fresh industrial catalyst (plotted with solid lines for comparison) which has a more sophisticated support oxide mixture. Apparently this does not affect the CO oxidation properties to a significant extent. The model catalyst has T50% = 213 8C, compared to 200 8C for the commercial catalyst. Since the behavior of the model catalyst resembled closely the industrial catalysts, it is further reasonable to perform all poisoning studies only for the model Pt-based catalyst. Activity data for poisoned catalysts are given in Fig. 2, and Table 1 summarizes T50% values obtained from these experiments. The temperature of the samples never exceeded 250 8C, since this might lead to evaporation of certain species, yielding a system irrelevant to this industrial case—a CO oxidation catalyst running at 250 8C. HMDS was found to be the most severe poison among all deactivating compounds tested. According to the literature [5] this volatile compound can deactivate a Pt-site being oxidized by the catalytic site followed by the deposition of resulting oxygenated species onto the Pt. This is the most likely reason for deactivation of the Pt-catalyst at plant operating conditions. Both HMDS poisoned model catalysts and industrially deactivated catalysts seem to have T50% in the area of 260– 270 8C. Analysis of flue gas samples taken on the industrial site showed that certain types of siloxanes, named D5 and D6—see Fig. 3, indeed were present in the off-gas. H2SO4 appeared to be a rather severe poison, as did (though to a lower degree) other types of acidic compounds. In fact there seems to be a direct correlation between acid strength and degree of deactivation. These observations, rule out possible deactivation by reduction of the catalyst. Considering the high O2 partial pressure in the off-
13
gas and that platinum is a good sulfur oxidation catalyst, it is most likely that any sulfur compounds reaching the catalyst in lower oxidation state (as SII , S0, SIV) will be oxidized to SVI on-site [19,20]. Thus, sulfur is most likely to interact with a platinum-site forming sulfate compounds. Furthermore, since the acidity has such a strong deactivating effect, it is most reasonable to suggest a poisoning mechanism with combined sulfatization and hydroxylation of the active platinum-site. This means that H2S and H2SO4 present in the off-gas after the scrubber system represent potential severe poisons. This poisoning effect of sulfur resembles that of a three-way catalyst, where the catalyst is partly poisoned by sulfate in lean conditions, but regenerate in rich conditions due to reduction of sulfate by alkanes to the more volatile H2S, which leaves the catalytic site. Due to the possible hydroxylation reactions, H3PO4 (not shown here) and HCl are also considered to be possible poison sources, though weaker. 3.2. FTIR-measurements In order to backup the kinetic data, a series of high temperature-FTIR-measurements were performed at the operation temperature on the model catalysts impregnated with the different poisons. In Fig. 4, the different poisoned catalysts are compared to the two industrial samples. All spectra present a band in the range 1640–1620 cm 1, originating from surface Ti–OH. The deactivated industrial catalyst shows bands of CO adsorbed on surface TiO2 at 2145 cm 1. Sulfur and chloride adsorption gives characteristic bands at 2120–2125 cm 1. A careful inspection of these bands showed than none of them were present in the spectrum of the industrial deactivated catalyst. Neither the siloxane deactivated catalyst had any distinct similarities with the industrial sample. However, as discussed in the introduction, siloxane will probably deactivate by fixing one or more Pt-sites to a partially oxidized siloxane molecule. More careful inspection of the IR region around 2950 cm 1, see Fig. 5, of the siloxane deactivated model catalyst indeed reveal a very small peak attributed to a sp3 hybridized C–H stretch of the partially oxidized, adsorbed siloxane. Furthermore, contrary to the industrial sample, significant peaks of silanol can be observed for the model catalyst, pointing towards the fact that our model poison has been added in far bigger amount than necessary to deactivate
Fig. 3. Types of siloxane structures: (A) HMDS, probe molecule for poison studies (L2); (B and C) 5- and 6-ring type siloxanes found in the flue gas on-site (D5 and D6).
14
S.B. Rasmussen et al. / Applied Catalysis B: Environmental 69 (2006) 10–16
reduces the Pt–O bond back to metallic platinum, it seems likely that the Pt–siloxane bond will be broken, and siloxane will either leave the catalyst surface as a re-volatized compound—or instead bind to a carrier oxide or hydroxide species and become a spectator species on the surface of the catalyst. Fig. 5 shows such a regeneration procedure performed in situ in the high temperature IR furnace, and it is clearly seen at the background corrected blow-up to the right that the siloxane bond indeed disappears by the treatment with formier gas (10% H2/N2). The difference spectrum to the left also clearly shows that this treatment results in the appearance of various types of silanole species (terminal, geminal as well as hydrogen bonded). These results provide a reasonable explanation of the siloxane deactivation phenomena, but moreover they also provide means for a possible easy regeneration of the catalyst by formier gas treatment. 3.3. Regeneration of siloxane poisoned catalysts
Fig. 4. High temperature-FTIR (250 8C) spectra of poisoned model catalysts pressed into self-supporting wafers.
the catalyst. Additional tests has proved that the double amount of siloxane poisoning yielded the same level of deactivation (i.e. the Pt-catalyst was saturated with siloxane). Thus, the scenario with a modified siloxane molecule bound to a catalytic Pt–O site seems to be the most probable explanation regarding the case of the incineration plant catalyst. However, this gives a possible method of easy regeneration of the catalytic site by H2 treatment. If the hydrogen molecule
Two types of regeneration procedures were applied on deactivated Pt samples in order to restore the catalytic activity. It was found that regeneration in air at 500 8C, only partly regenerates the catalyst. More noticeable is the effect of treatment with the formier gas (10% H2, 90% N2), where the catalyst became more active even compared to the fresh Ptcatalyst. However, there is a problem in a low temperature setup, since, for material reasons, the temperature of the catalytic beds cannot exceed 250 8C. As seen in Fig. 6, it is indeed also possible to restore the catalytic activity with formier gas at 250 8C. The regenerated catalyst also exhibits slightly higher activity than the fresh one. However, a second run of the same sample (cooling and reheating of the sample)
Fig. 5. High temperature-FTIR (250 8C) spectra of HMDS poisoned Pt/TiO2 catalyst. (Left) Poisoned catalyst before and after treatment with 10% H2/N2. The difference spectrum shows the conversion of HMDS-derivates into silanoles. (Right) Blow-up of background corrected Si–CH3 band disappearing over time due to the formier gas treatment.
S.B. Rasmussen et al. / Applied Catalysis B: Environmental 69 (2006) 10–16
Fig. 6. Temperature dependence of CO conversion for the industrially deactivated catalyst after regeneration with formier gas (10% H2/N2) and with 1:1 mixture of formier gas and flue gas. Activity curves of fresh and deactivated catalysts are given for comparison. Gas composition: 5% H2, 5% O2, 13% H2O, 0.2% CO balanced with N2; GHSV = 30,000 H 1.
showed that the high activity seems to decrease over time. This is probably because the reduction procedure reforms Pt metal on the surface with a good dispersion. After equilibration in the oxidizing off-gas, the activity of the regenerated catalyst will presumably become the same as that of the fresh sample over time due to some re-structuring of the Pt particles. Thus, the results of these studies actually show that it is possible to regenerate the catalyst on-site in the plant. However, in this case it means that the catalytic unit will have to be out of operation while the regeneration is performed. It is more desirable to perform regeneration while the plant is actually operating. A scenario simulating a normal flow rate of off-gas going through the catalyst together with an equal volumetric amount of formier gas means a doubling of the total space velocity for 2 h, but otherwise no change in operation at the plant. In this case the off-gas would contain 5% H2 and 5% O2, which could lead to formation of water by oxidation of H2 instead of regeneration of the catalyst, but as can be seen in Fig. 6, this is not the case. Apparently the regeneration reactions proceed faster than hydrogen oxidation over a Pt-site. Also, the oxidation of hydrogen with oxygen would be expected to occur over an active Pt-site, but at least until the catalytic site is regenerated, this does not occur. Thus, regeneration can be done even in the presence of oxygen and water. Similar behavior has earlier been observed with Pt-oxide catalysts, where catalysts exposed to 1% H2, 0.25% NO, 1% Ar, 2–5% O2, He selectively reduced NO in the temperature range 100–200 8C [21]. Two samples of the artificially poisoned model catalyst (siloxane and H2SO4) were subjected to the same regeneration treatment. And indeed this treatment regenerated the catalytic activity in both cases, as seen in Fig. 7. So if an incineration plant experience a siloxane or a sulfur poisoning of the Pt-
15
Fig. 7. Temperature dependence of CO conversion of HMDS and H2SO4 poisoned model catalysts after treatment with formier gas for 1 h at 250 8C, compared to the activities of industrial catalysts.
catalyst, this regeneration procedure will successfully restore catalyst activity. A possible explanation regarding the reactivation by hydrogen is a reduction of a surface Pt-oxide layer to metallic Pt, whereby the oxide bridge anchoring of the poison to the surface is broken, liberating the poison molecule either for vaporization or movement from the Pt-site to a support TiO2 site. The regeneration procedure can easily be further optimized with respect to minimizing time of exposure and hydrogen content (the latter introduces some extra safety issues regarding gas handling on the plant). It is obvious that compared to the bulky IR-furnace with large dead space compared to sample size, the regeneration can be much faster. The result for experiments where CO was added to the regeneration gas are plotted in Fig. 8.
Fig. 8. Dependence of CO conversion of deactivated industrial catalysts over time after initiation of regeneration procedure with mixture containing both CO and H2. Gas composion: 3–5% H2, 5% O2, 13% H2O, 0.2% CO balanced with N2; GHSV = 30,000 H 1.
16
S.B. Rasmussen et al. / Applied Catalysis B: Environmental 69 (2006) 10–16
These results indicate that even few minutes and a few percent H2 in the flue gas is needed for sufficient regeneration of the catalyst. At 200 8C the exposure time will have to be some minutes longer than at 220 8C. However, this regeneration time is quite small compared to what is expected for switching back and forth from flue gas to regeneration gas mixture in the industrial plant.
References [1] [2] [3] [4] [5]
4. Conclusion The deactivation performance of Pt-catalysts for CO oxidation has been studied in relation to use in sewage sludge municipal waste burners. HMDS is poisoning model Pt–TiO2 catalysts in a similar way as the industrial catalyst is deactivated by the off-gas in a real waste plant. Other compounds, especially H2SO4, could also poison the catalyst to some degree. Siloxane type compounds, D5 and D6, have indeed been found in the flue gas from an industrial waste plant, in a separate analysis of the flue gas during this work. Furthermore, a promising regeneration procedure has been developed based on hydrogen containing gas mixtures administrated in short pulses, either on-line directly in the flue gas or during a bypass of the catalyst bed by the plant flue gas. Acknowledgment LAB SA, France is thanked for financial support of this investigation.
[6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16] [17] [18] [19] [20] [21]
S.J. Gentry, A. Jones, J. Appl. Chem. Biotechnol. 28 (1978) 727. C.F. Cullis, B.M. Willatt, J. Catal. 86 (1984) 187. J.-J. Ehrhardt, L. Colin, D. Jamois, Sens. Actuators B 40 (1997) 117. M. Matsumiya, W. Shin, F. Qiu, N. Izu, I. Matsubara, N. Murayama, Sens. Actuators B 96 (2003) 516. K. Arnby, M. Rahmani, M. Sanati, N. Cruise, A.A. Carlsson, M. Skoglundh, Appl. Catal. B 54 (2004) 1. B. Mendyka, A. Musialik, K. Syczewska, Catal. Today 11 (1992) 597. M.M. Farris, A.A. Klinghoffer, J.A. Rossin, D.E. Tevault, Catal. Today 11 (1992) 501. M.C.N.A. de Carvalho, E. Morgado, J. Henrique, S. Cerqueira, N.S. de Resende, M. Schmal, Ind. Eng. Chem. Res. 43 (2004) 3133. A. Sirijaruphan, J.G. Goodwin Jr., R.W. Rice, J. Catal. 221 (2004) 288. H.C. Yao, H.K. Stepien, H.S. Gandhi, J. Catal. 67 (1981) 231. A. Hinz, M. Skoglundh, E. Fridell, A. Andersson, J. Catal. 201 (2001) 247. K. Ruth, M. Hayes, R. Burch, S. Tsubota, M. Haruta, Appl. Catal. B 24 (2000) L133. I. Chorkendorff, J.W. Niemantsverdriet, Concepts of Modern Catalysis and Kinetics, John Wiley & Sons, 2003,, p. 384. Y.-F. Han, M.J. Kahlich, M. Kinne, R.J. Behm, Appl. Catal. B 50 (2004) 209. J. Rasko´, Catal. Lett. 56 (1998) 11. E.V. Benvenutti, L. Franken, C.C. Moro, C.U. Davanco, Langmuir 15 (1999) 8140. R. Green, P. Morrall, M. Bowker, Catal. Lett. 87 (2004) 129. K.I. Hadjiivanov, J. Chem. Soc. Faraday Trans. 94 (1998) 1901. Z. Paa´l, M. Muhler, K. Matusek, J. Catal. 175 (1998) 245. M.-D. Appay, J.-M. Manoli, C. Potvin, M. Muhler, U. Wild, O. Pozdnyakova, Z. Paal, J. Catal. 222 (2004) 419. C.N. Costa, A.M. Efstathiou, J. Phys. Chem. B 108 (2004) 2620.
Characterization and regeneration of Pt-catalysts deactivated in municipal waste flue gas Søren Birk Rasmussen a,*, Arkady Kustov a, Johannes Due-Hansen a, Bernard Siret b, Frank Tabaries b, Rasmus Fehrmann a a
Department of Chemistry and Center for Sustainable and Green Chemistry, Department of Chemistry, Technical University of Denmark, Denmark b LAB S.A., 25 rue Bossuet, 69455 Lyon Ce´dex 06, France Received 2 February 2006; received in revised form 8 May 2006; accepted 12 May 2006 Available online 27 June 2006
Abstract Severe deactivation was observed for industrially aged catalysts used in waste incineration plants and tested in lab-scale. Possible compounds that cause deactivation of these Pt-based CO oxidation catalysts have been studied. Kinetic observations of industrial and model catalysts showed that siloxanes were the most severe catalyst poisons, although acidic sulfur compounds also caused deactivation. Furthermore, a method for on-site regeneration without shutdown of the catalytic flue gas cleaning system has been developed, i.e. an addition of H2/N2 gas to the off-gas can completely restore the activity of the deactivated catalysts. # 2006 Elsevier B.V. All rights reserved. Keywords: Platinum; CO oxidation; Poisoning; IR spectroscopy; Regeneration; Hexamethyldisiloxane; Catalytic oxidation; Hydrogen
1. Introduction Burning municipal waste or sludges instead of landfill depositing is an important issue from an environmental point of view, since this ‘‘waste to energy’’ conversion also implies several benefits such as control of toxic emissions and heavy metal waste, etc. In this connection the treatment of gaseous emissions containing carbon and nitrogen oxides as well as volatile organic compounds has been of increasing concern in recent years. Thermal incineration, catalytic oxidation and adsorption are commonly used for removing these pollutants. Thermal incineration requires high operating temperatures and high capital cost facilities. If the gaseous stream also includes halogenated compounds, thermal incineration can produce toxic halogenated compounds under certain operating conditions. In some cases, adsorption by adsorbents such as carbon is an alternative. However, this process does not destroy pollutants, but merely concentrates them. Moreover, adsorption
* Corresponding author. Tel.: +45 4525 2390; fax: +45 4588 3136. E-mail address: [email protected] (S.B. Rasmussen). 0926-3373/$ – see front matter # 2006 Elsevier B.V. All rights reserved. doi:10.1016/j.apcatb.2006.05.009
efficiency can be adversely influenced by fluctuating concentrations of the gaseous components. Catalytic oxidation of gaseous organic emissions operates at significantly lower temperatures and shorter residence time than thermal incineration. It is also a common method for removal of CO emissions from industrial or municipal waste gases. Platinum supported on different oxide carriers is traditionally used as the catalyst for the oxidation of CO and hydrocarbons. At high temperatures it does not dissolve in the washcoat, but sintering into larger particles occurs, especially if the catalyst is exposed to temperatures above 700 8C. This leads to a substantial loss of a platinum surface area and a less efficient catalyst. This work focuses on a problem at a sewage sludge plant utilizing a Pt-based catalyst dispersed on an oxide support of proprietary nature for CO oxidation and operating at 250 8C in the tail end of the flue gas cleaning system. It was observed several times that the catalyst deactivated few days after startup of the gas cleaning system. The usual causes of deactivation of the catalyst during exposure to the flue gases are thermal damage as mentioned above, mechanical damage of the monolith or poisoning by
S.B. Rasmussen et al. / Applied Catalysis B: Environmental 69 (2006) 10–16
contaminants. Since the first parts are not relevant in this case, the poisoning of the catalyst by contaminants by components originating from the flue gas is suspected to be the reason for deactivation. According to the literature, possible deactivating flue gas components can be divided into four major classes [1– 13]:
silicium containing compounds; halogenated compounds; hydrocarbons and carbon deposits (coke); minor deactivating agents like SO2, Pb, As, K, phosphates, etc.
The printing industry is a typical example where organosilicon compounds often are present in the waste flue gas. The deposition of silicon residues on the active sites causes poisoning and deactivation of the catalyst. Cosmetics seem to be another important source of these compounds in waste. In early investigations [1,2], hexamethyldisiloxane (HMDS) was found to almost irreversibly deactivate alumina supported Pt and Pd catalysts for methane oxidation, whereas the oxidation of propene and butane over these catalysts was only reversibly inhibited in the presence of HMDS. The authors [1,2] explained these two types of deactivation processes by proposing a dualsite mechanism in which HMDS adsorbs irreversibly on the most active (high-energy) sites, necessary for methane oxidation, while in the case of less active (low-energy) sites used for propene and butane oxidation, HMDS adsorbs reversibly. More recently, it has been reported that HMDS deactivates the catalytic function of platinum based catalysts by forming a SiO2 overlayer which covers the Pt surface [3,4]. The catalytic activity can be recovered during regeneration by: reorganization of SiO2 films into islands which recovers the Pt surface [3]; release of silicon species from the surface [4]. The researchers performed regeneration procedures by exposure of the deactivated samples to air at 25 8C for 15 h [5]. In the case of long-term exposure to the silicon compounds, the Si content becomes too high, and the deactivation of the catalyst seems to be irreversible. Halogenated hydrocarbons are included in the list of volatile organic compounds. Oxidation of these compounds yields acidic gases as reaction products. These products are known to have a substantial poisoning effect on catalysts [6]. In this case catalysts with high noble-metal content (generally 0.2–1.5% Pt) are more poison resistant. For the catalysts with lower Pt content the activity can be completely regenerated by treatment with hot air at 350 8C. It has been reported that exposure of Pt/Al2O3 catalysts in a chlorine atmosphere may result in either sintering or redispersion of the platinum, depending on the nature and concentration of the chlorine-containing species. Adsorbed chlorine on the catalyst support may also produce platinum chloride complexes. It has also been reported that halogenated hydrocarbons may act as catalytic inhibitors, adsorbing on the
11
sites normally required for the activation and adsorption of CO and oxygen [6]. On the other hand in Ref. [7] it was shown that the presence of HCl in the flue gas has no inhibiting effect on the oxidation of hydrocarbons on Pt/Al2O3 catalyst, and no loss of Pt-dispersion was observed in this case. Moreover, according to the Pt-dispersion measurements, it was found that in some cases the presence of chlorine compounds prevents formation of larger platinum crystallites (sintering). Sulfur, although a potential poison for all metals, interacts relatively weakly with platinum. Moreover, some beneficial effect of SO2 on the activity of Pt/Al2O3 for oxidation of hydrocarbons was observed in Refs. [10,11]. A small fraction of SO2 that is oxidized further to SO3 may react with the alumina to give Al2(SO4)3, causing loss of surface area of the washcoat. Small negative effects of sulfation were observed for Pt/TiO2 in Ref. [12]. In addition lead and phosphorous are considered as traditional poisons for the catalysts [13]. The influence of water on the stability and activity of Pt/g-Al2O3 catalysts was studied in Ref. [14]. Measurements made in this work showed that the effect of water vapor in the feed gas (10%) on the activity and stability of the catalyst is quite negligible. This work reports on experimental studies of poisoning of a model catalyst prepared in the lab, and the results are compared with kinetic data obtained for actual industrial catalysts— either as fresh catalysts or as deactivated catalysts after actual use on industrial scale. Based on the literature summarized above and some preliminary investigations Na2S, H2SO4, HCl and hexamethylsiloxane (HMDS) were selected as model poisons. 2. Experimental 2.1. Catalysts preparation Samples of fresh and industrially deactivated catalysts of proprietary nature were obtained from the incineration plant. Model Pt/TiO2 catalysts were prepared according to the following procedure: a 100 ml solution of 131.6 mg H2PtCl66H2O (38% Pt, Fluka) in water was prepared and 10 g of TiO2 (anatase, Millenium Chemicals, pre-calcined at 500 8C) was added to the solution, similar to the procedure described by Benvenutti et al. [16], Green et al. [17], Rasko´ [15] and Hadjiivanov [18]. The mixture was stirred overnight. Water was removed in a vacuum rotational evaporator at 60 8C, and after drying residual (HCl) was removed by calcination at 400 8C for about 30 h, ending up with a 0.5 wt.% Pt/TiO2 model catalyst. Samples of this model Pt/TiO2 catalyst were poisoned by hexamethyldisiloxane (HMDS, >98.5%, Fluka), HC1, H3PO4, and H2SO4 (all from Merck, analysis grade). The poison in question was transferred to an ampule containing 1 g of catalyst sample without being in direct contact, sealed in dry air and heated to 250 8C for several days to ensure equilibrium through gas phase transport. The amount of each poison transferred to a catalyst sample was adjusted such that the bulk molar ratio between poison and platinum was 2:1.
12
S.B. Rasmussen et al. / Applied Catalysis B: Environmental 69 (2006) 10–16
2.2. CO oxidation activity measurements The catalytic activity of all samples was tested in laboratoryscale fixed-bed reactors according to the following procedure: about 50 (0.6) mg powder sample was placed in a tubular Pyrex reactor, fixed with quarts wool. The gas stream, controlled by Bronkhorst Mass Flow Meters, consisting of 79.8% N2, 20% O2 and 0.2% CO, was passed through the reactor at a total flow rate of 100 ml/min. The reactor was placed and preheated in a furnace at 250 8C for 1 h in order to equilibrate the system. Equilibrating the sample at 250 8C before direct measuring of the catalytic activity was found to be critical for obtaining reproducible results relating realistically to industrial conditions. Then, the temperature was lowered with the rate of 1 8C/min and the CO signal was monitored in the temperature interval 140–250 8C. The CO concentration was determined using a Hewlett Packard Gas Chromatograph, HP6890, with a 60 m micro-capillary column equipped with a TCD detector. Data acquisition was controlled by HP Chemstation software. The outlet gas composition was measured every 10 min, and the areas of N2, O2 and CO in the output-signal was used for the calculation of the CO concentration. Prior to loading, the sample was pressed into pellets then crushed and sieved to obtain catalysts with particle size in the interval 0.295–0.71 mm. 2.3. Catalyst characterization Self-supporting disks of the sample were prepared for high temperature-FTIR spectroscopy by pressing about 30–50 mg of the sample powder into 1.3 cm diameter discs. About 10 tonnes pressure was applied for 5 min. FTIR-measurements at high temperature were carried out on a Paragon 1000 Perkin-Elmer FTIR spectrometer, equipped with a water cooled, electrically heated furnace with argon flow through the sample chamber. All spectra were measured in the range 4000–450 cm 1 with 64 scans and a spectral resolution of 8 cm 1, in the temperature range 25–350 8C.
Fig. 1. Temperature dependence of CO conversion for fresh and deactivated industrial catalyst (before and after 1 month of exposure to flue gas). Gas composition: 10% O2, 0.2% CO, N2; GHSV = 30,000 H 1.
50% conversion was obtained, changed from 200 8C for the fresh catalyst to around 255 8C for the exposed catalyst. This extent of deactivation is quite critical, since the plant is operated at 250 8C, and therefore insufficient CO removal was experienced during normal plant operation. At the same time no significant change in the BET surface area was observed between fresh and deactivated catalysts. In order to make a systematic study of the Pt/TiO2 system with respect to poison and regeneration possibilities, a model catalyst was synthesized as described in Section 2. The resulting catalyst
2.4. Flue gas siloxane analysis Siloxanes were collected on-site by using a standard sampling train. The flue gas was bubbled through a chilled liquid (n-heptane/methanol) in an impinger. The solution was sent to the specialized laboratory CERTECH, Belgium. By GC/ MS, utilizing the relevant fragments obtained by MS, the different siloxanes were identified and quantified. CERTECH was selected because it is one of the few laboratories that have experience with siloxane quantification. 3. Results and discussion 3.1. Deactivation studies The commercial catalyst samples before and after exposure to incineration plant flue gas were tested for CO conversion versus temperature (Fig. 1). The T50%, the temperature where
Fig. 2. Temperature dependence of CO conversion for model poisoned catalysts. Gas composition: 10% O2, 0.2% CO, N2; GHSV = 30,000 H 1.
S.B. Rasmussen et al. / Applied Catalysis B: Environmental 69 (2006) 10–16 Table 1 T50% values for the fresh and poisoned model catalysts Sample
T50% (8C)
Pt/TiO2 H3PO4 HCl H2SO4 HMDS
213 218 238 242 250
had a surface area of 70–80 m2/g and was subsequently tested for CO oxidation activity. The results of this test are given in Fig. 2, where it is shown that the model catalyst reproduces quite well the behavior of the fresh industrial catalyst (plotted with solid lines for comparison) which has a more sophisticated support oxide mixture. Apparently this does not affect the CO oxidation properties to a significant extent. The model catalyst has T50% = 213 8C, compared to 200 8C for the commercial catalyst. Since the behavior of the model catalyst resembled closely the industrial catalysts, it is further reasonable to perform all poisoning studies only for the model Pt-based catalyst. Activity data for poisoned catalysts are given in Fig. 2, and Table 1 summarizes T50% values obtained from these experiments. The temperature of the samples never exceeded 250 8C, since this might lead to evaporation of certain species, yielding a system irrelevant to this industrial case—a CO oxidation catalyst running at 250 8C. HMDS was found to be the most severe poison among all deactivating compounds tested. According to the literature [5] this volatile compound can deactivate a Pt-site being oxidized by the catalytic site followed by the deposition of resulting oxygenated species onto the Pt. This is the most likely reason for deactivation of the Pt-catalyst at plant operating conditions. Both HMDS poisoned model catalysts and industrially deactivated catalysts seem to have T50% in the area of 260– 270 8C. Analysis of flue gas samples taken on the industrial site showed that certain types of siloxanes, named D5 and D6—see Fig. 3, indeed were present in the off-gas. H2SO4 appeared to be a rather severe poison, as did (though to a lower degree) other types of acidic compounds. In fact there seems to be a direct correlation between acid strength and degree of deactivation. These observations, rule out possible deactivation by reduction of the catalyst. Considering the high O2 partial pressure in the off-
13
gas and that platinum is a good sulfur oxidation catalyst, it is most likely that any sulfur compounds reaching the catalyst in lower oxidation state (as SII , S0, SIV) will be oxidized to SVI on-site [19,20]. Thus, sulfur is most likely to interact with a platinum-site forming sulfate compounds. Furthermore, since the acidity has such a strong deactivating effect, it is most reasonable to suggest a poisoning mechanism with combined sulfatization and hydroxylation of the active platinum-site. This means that H2S and H2SO4 present in the off-gas after the scrubber system represent potential severe poisons. This poisoning effect of sulfur resembles that of a three-way catalyst, where the catalyst is partly poisoned by sulfate in lean conditions, but regenerate in rich conditions due to reduction of sulfate by alkanes to the more volatile H2S, which leaves the catalytic site. Due to the possible hydroxylation reactions, H3PO4 (not shown here) and HCl are also considered to be possible poison sources, though weaker. 3.2. FTIR-measurements In order to backup the kinetic data, a series of high temperature-FTIR-measurements were performed at the operation temperature on the model catalysts impregnated with the different poisons. In Fig. 4, the different poisoned catalysts are compared to the two industrial samples. All spectra present a band in the range 1640–1620 cm 1, originating from surface Ti–OH. The deactivated industrial catalyst shows bands of CO adsorbed on surface TiO2 at 2145 cm 1. Sulfur and chloride adsorption gives characteristic bands at 2120–2125 cm 1. A careful inspection of these bands showed than none of them were present in the spectrum of the industrial deactivated catalyst. Neither the siloxane deactivated catalyst had any distinct similarities with the industrial sample. However, as discussed in the introduction, siloxane will probably deactivate by fixing one or more Pt-sites to a partially oxidized siloxane molecule. More careful inspection of the IR region around 2950 cm 1, see Fig. 5, of the siloxane deactivated model catalyst indeed reveal a very small peak attributed to a sp3 hybridized C–H stretch of the partially oxidized, adsorbed siloxane. Furthermore, contrary to the industrial sample, significant peaks of silanol can be observed for the model catalyst, pointing towards the fact that our model poison has been added in far bigger amount than necessary to deactivate
Fig. 3. Types of siloxane structures: (A) HMDS, probe molecule for poison studies (L2); (B and C) 5- and 6-ring type siloxanes found in the flue gas on-site (D5 and D6).
14
S.B. Rasmussen et al. / Applied Catalysis B: Environmental 69 (2006) 10–16
reduces the Pt–O bond back to metallic platinum, it seems likely that the Pt–siloxane bond will be broken, and siloxane will either leave the catalyst surface as a re-volatized compound—or instead bind to a carrier oxide or hydroxide species and become a spectator species on the surface of the catalyst. Fig. 5 shows such a regeneration procedure performed in situ in the high temperature IR furnace, and it is clearly seen at the background corrected blow-up to the right that the siloxane bond indeed disappears by the treatment with formier gas (10% H2/N2). The difference spectrum to the left also clearly shows that this treatment results in the appearance of various types of silanole species (terminal, geminal as well as hydrogen bonded). These results provide a reasonable explanation of the siloxane deactivation phenomena, but moreover they also provide means for a possible easy regeneration of the catalyst by formier gas treatment. 3.3. Regeneration of siloxane poisoned catalysts
Fig. 4. High temperature-FTIR (250 8C) spectra of poisoned model catalysts pressed into self-supporting wafers.
the catalyst. Additional tests has proved that the double amount of siloxane poisoning yielded the same level of deactivation (i.e. the Pt-catalyst was saturated with siloxane). Thus, the scenario with a modified siloxane molecule bound to a catalytic Pt–O site seems to be the most probable explanation regarding the case of the incineration plant catalyst. However, this gives a possible method of easy regeneration of the catalytic site by H2 treatment. If the hydrogen molecule
Two types of regeneration procedures were applied on deactivated Pt samples in order to restore the catalytic activity. It was found that regeneration in air at 500 8C, only partly regenerates the catalyst. More noticeable is the effect of treatment with the formier gas (10% H2, 90% N2), where the catalyst became more active even compared to the fresh Ptcatalyst. However, there is a problem in a low temperature setup, since, for material reasons, the temperature of the catalytic beds cannot exceed 250 8C. As seen in Fig. 6, it is indeed also possible to restore the catalytic activity with formier gas at 250 8C. The regenerated catalyst also exhibits slightly higher activity than the fresh one. However, a second run of the same sample (cooling and reheating of the sample)
Fig. 5. High temperature-FTIR (250 8C) spectra of HMDS poisoned Pt/TiO2 catalyst. (Left) Poisoned catalyst before and after treatment with 10% H2/N2. The difference spectrum shows the conversion of HMDS-derivates into silanoles. (Right) Blow-up of background corrected Si–CH3 band disappearing over time due to the formier gas treatment.
S.B. Rasmussen et al. / Applied Catalysis B: Environmental 69 (2006) 10–16
Fig. 6. Temperature dependence of CO conversion for the industrially deactivated catalyst after regeneration with formier gas (10% H2/N2) and with 1:1 mixture of formier gas and flue gas. Activity curves of fresh and deactivated catalysts are given for comparison. Gas composition: 5% H2, 5% O2, 13% H2O, 0.2% CO balanced with N2; GHSV = 30,000 H 1.
showed that the high activity seems to decrease over time. This is probably because the reduction procedure reforms Pt metal on the surface with a good dispersion. After equilibration in the oxidizing off-gas, the activity of the regenerated catalyst will presumably become the same as that of the fresh sample over time due to some re-structuring of the Pt particles. Thus, the results of these studies actually show that it is possible to regenerate the catalyst on-site in the plant. However, in this case it means that the catalytic unit will have to be out of operation while the regeneration is performed. It is more desirable to perform regeneration while the plant is actually operating. A scenario simulating a normal flow rate of off-gas going through the catalyst together with an equal volumetric amount of formier gas means a doubling of the total space velocity for 2 h, but otherwise no change in operation at the plant. In this case the off-gas would contain 5% H2 and 5% O2, which could lead to formation of water by oxidation of H2 instead of regeneration of the catalyst, but as can be seen in Fig. 6, this is not the case. Apparently the regeneration reactions proceed faster than hydrogen oxidation over a Pt-site. Also, the oxidation of hydrogen with oxygen would be expected to occur over an active Pt-site, but at least until the catalytic site is regenerated, this does not occur. Thus, regeneration can be done even in the presence of oxygen and water. Similar behavior has earlier been observed with Pt-oxide catalysts, where catalysts exposed to 1% H2, 0.25% NO, 1% Ar, 2–5% O2, He selectively reduced NO in the temperature range 100–200 8C [21]. Two samples of the artificially poisoned model catalyst (siloxane and H2SO4) were subjected to the same regeneration treatment. And indeed this treatment regenerated the catalytic activity in both cases, as seen in Fig. 7. So if an incineration plant experience a siloxane or a sulfur poisoning of the Pt-
15
Fig. 7. Temperature dependence of CO conversion of HMDS and H2SO4 poisoned model catalysts after treatment with formier gas for 1 h at 250 8C, compared to the activities of industrial catalysts.
catalyst, this regeneration procedure will successfully restore catalyst activity. A possible explanation regarding the reactivation by hydrogen is a reduction of a surface Pt-oxide layer to metallic Pt, whereby the oxide bridge anchoring of the poison to the surface is broken, liberating the poison molecule either for vaporization or movement from the Pt-site to a support TiO2 site. The regeneration procedure can easily be further optimized with respect to minimizing time of exposure and hydrogen content (the latter introduces some extra safety issues regarding gas handling on the plant). It is obvious that compared to the bulky IR-furnace with large dead space compared to sample size, the regeneration can be much faster. The result for experiments where CO was added to the regeneration gas are plotted in Fig. 8.
Fig. 8. Dependence of CO conversion of deactivated industrial catalysts over time after initiation of regeneration procedure with mixture containing both CO and H2. Gas composion: 3–5% H2, 5% O2, 13% H2O, 0.2% CO balanced with N2; GHSV = 30,000 H 1.
16
S.B. Rasmussen et al. / Applied Catalysis B: Environmental 69 (2006) 10–16
These results indicate that even few minutes and a few percent H2 in the flue gas is needed for sufficient regeneration of the catalyst. At 200 8C the exposure time will have to be some minutes longer than at 220 8C. However, this regeneration time is quite small compared to what is expected for switching back and forth from flue gas to regeneration gas mixture in the industrial plant.
References [1] [2] [3] [4] [5]
4. Conclusion The deactivation performance of Pt-catalysts for CO oxidation has been studied in relation to use in sewage sludge municipal waste burners. HMDS is poisoning model Pt–TiO2 catalysts in a similar way as the industrial catalyst is deactivated by the off-gas in a real waste plant. Other compounds, especially H2SO4, could also poison the catalyst to some degree. Siloxane type compounds, D5 and D6, have indeed been found in the flue gas from an industrial waste plant, in a separate analysis of the flue gas during this work. Furthermore, a promising regeneration procedure has been developed based on hydrogen containing gas mixtures administrated in short pulses, either on-line directly in the flue gas or during a bypass of the catalyst bed by the plant flue gas. Acknowledgment LAB SA, France is thanked for financial support of this investigation.
[6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16] [17] [18] [19] [20] [21]
S.J. Gentry, A. Jones, J. Appl. Chem. Biotechnol. 28 (1978) 727. C.F. Cullis, B.M. Willatt, J. Catal. 86 (1984) 187. J.-J. Ehrhardt, L. Colin, D. Jamois, Sens. Actuators B 40 (1997) 117. M. Matsumiya, W. Shin, F. Qiu, N. Izu, I. Matsubara, N. Murayama, Sens. Actuators B 96 (2003) 516. K. Arnby, M. Rahmani, M. Sanati, N. Cruise, A.A. Carlsson, M. Skoglundh, Appl. Catal. B 54 (2004) 1. B. Mendyka, A. Musialik, K. Syczewska, Catal. Today 11 (1992) 597. M.M. Farris, A.A. Klinghoffer, J.A. Rossin, D.E. Tevault, Catal. Today 11 (1992) 501. M.C.N.A. de Carvalho, E. Morgado, J. Henrique, S. Cerqueira, N.S. de Resende, M. Schmal, Ind. Eng. Chem. Res. 43 (2004) 3133. A. Sirijaruphan, J.G. Goodwin Jr., R.W. Rice, J. Catal. 221 (2004) 288. H.C. Yao, H.K. Stepien, H.S. Gandhi, J. Catal. 67 (1981) 231. A. Hinz, M. Skoglundh, E. Fridell, A. Andersson, J. Catal. 201 (2001) 247. K. Ruth, M. Hayes, R. Burch, S. Tsubota, M. Haruta, Appl. Catal. B 24 (2000) L133. I. Chorkendorff, J.W. Niemantsverdriet, Concepts of Modern Catalysis and Kinetics, John Wiley & Sons, 2003,, p. 384. Y.-F. Han, M.J. Kahlich, M. Kinne, R.J. Behm, Appl. Catal. B 50 (2004) 209. J. Rasko´, Catal. Lett. 56 (1998) 11. E.V. Benvenutti, L. Franken, C.C. Moro, C.U. Davanco, Langmuir 15 (1999) 8140. R. Green, P. Morrall, M. Bowker, Catal. Lett. 87 (2004) 129. K.I. Hadjiivanov, J. Chem. Soc. Faraday Trans. 94 (1998) 1901. Z. Paa´l, M. Muhler, K. Matusek, J. Catal. 175 (1998) 245. M.-D. Appay, J.-M. Manoli, C. Potvin, M. Muhler, U. Wild, O. Pozdnyakova, Z. Paal, J. Catal. 222 (2004) 419. C.N. Costa, A.M. Efstathiou, J. Phys. Chem. B 108 (2004) 2620.