Combustion of methane over palladium catalyst in the presence of inorganic compounds: inhibition and deactivation phenomena

Applied Catalysis B: Environmental 47 (2004) 85–93

Combustion of methane over palladium catalyst in the presence of inorganic compounds: inhibition and deactivation phenomena Paloma Hurtado, Salvador Ordóñez∗ , Herminio Sastre, Fernando V. D´ıez Department of Chemical and Environmental Engineering, University of Oviedo, Julián Claver´ıa s/n, 33006 Oviedo, Spain Received 3 April 2003; received in revised form 23 July 2003; accepted 27 July 2003

Abstract The influence of the presence of different inorganic gases, often found in methane-containing industrial off-gases (NH3 , NO2 , H2 , H2 O, SO2 , H2 S, CO and CO2 ) on the catalytic combustion of methane over a Pd/Al2 O3 catalyst is studied in the present work, in a range of concentrations corresponding to typical industrial emissions, such as those of coke oven facilities. Results show that the effect of SO2 and H2 S on the catalyst is similar, both compounds causing partially irreversible poisoning, whereas water (present in the feed or formed by combustion of H2 or CH4 ) causes reversible inhibition in the absence of sulphur compounds. Nitrogen-containing compounds increase methane conversion in the absence of sulphur-containing compounds, but ammonia has the opposite effect when sulphur compounds are present. The other compounds studied do not affect appreciably the catalytic combustion of methane. © 2003 Elsevier B.V. All rights reserved. Keywords: Methane combustion; Coke-oven emissions; Pd catalysts; Poisoning; Inhibition

1. Introduction Catalytic combustion has become a very interesting alternative for the treatment of gaseous emissions containing diluted organic compounds because of its low energy consumption (especially if compared with thermal oxidation) and the insignificant formation of noxious by-products, such as thermal NOx [1,2]. Among the organic compounds present in gaseous emissions, increasing attention is being paid to methane. This interest is justified by two considerations: on the one hand, methane is an important contributor to the greenhouse effect (its global warning potential is 21 times higher than that corresponding to CO2 ); on the other hand, the catalytic oxidation of methane is more difficult than the oxidation of most volatile organic compounds [3]. Thus, methane has been chosen by many authors as a model compound for catalytic oxidation studies, as the catalytic combustion of most organic compounds would be ensured if methane is quantitatively abated.

∗

Corresponding author. E-mail address: [email protected] (S. Ord´oñez).

0926-3373/$ – see front matter © 2003 Elsevier B.V. All rights reserved. doi:10.1016/S0926-3373(03)00328-X

In many gaseous emissions (i.e. from carbochemical and petrochemical processes, treatment of solid wastes and wastewater, etc.), methane is accompanied by other inorganic and organic compounds. So, in the case of coke oven emissions, benzene, toluene, CO, CO2 , NO, H2 , H2 O, NH3 , H2 S and SO2 are present, as can be seen in Table 1, which presents a typical composition of the emissions from the larry car (a system for delivering coal on the top of the ovens) [4]. The complete oxidation of methane can be performed over either noble metals or transition metal oxides. Both families of catalysts have been studied extensively with a view to develop catalytic combustion applications. The main advantage of noble metals over metal oxides is their superior specific activity, which makes them the best candidates for low-temperature combustion of hydrocarbons [5]. This is especially important in the case of methane because, as indicated previously, it is the most difficult hydrocarbon to activate. Among noble metal catalysts, Pd may be the best option. So Schlangen et al. [6] showed that Pd is more active than Pt for the combustion of linear hydrocarbons of molecular weight lower than pentane, whereas Oh et al. [7] and Burch and Loader [8] found that the relative activity of both metals depends on the matrix, alumina-supported Pd being more

86

P. Hurtado et al. / Applied Catalysis B: Environmental 47 (2004) 85–93

Table 1 Typical composition of emissions from coke ovens (coal charging), and values used in the experiments for single compounds in this work (ppm V, except O2 and N2 ) Component

Coke oven emissions

Experiments

CH4 CO2 CO H2 O H2 Nox NH3 SO2 H2 S Other organics (C6 H6 , C2 H2 , C2 H4 ) O2 (% v/v) N2 (% v/v)

3900–6000 3000–276000 2570–6500 1000000–26000 6500–7500 60–80 20–200 50–60 25 0–1000

5000 8000 3000 20000 7000 65 180 40 25 0

17–21 76–81

21 79

active in an oxidizing atmosphere (O2 :CH4 ratio greater than 2); and Dodgson and Webster showed that Pd catalysts are more resistant to deactivation by sintering than Pt catalysts [9]. On the other hand, it is generally accepted that Pt is less sensitive to sulphur poisoning [10]. However, previous results of our research group showed that the activity of Pt for the feed considered in this work is too low for practical use [11]. Concerning the effect of different compounds on the combustion of methane over noble metal catalysts, several studies on the effect of water and carbon oxides can be found in the literature. So it was observed that sulphur compounds (see for example [10,12–15]) and water seriously affect the catalytic activity, whereas carbon dioxide has no effect [16]. The effect of nitrogen compounds has been studied to a lesser extent: Öcal et al. [17] studied the effect of NO on the catalytic oxidation of methane over hexaluminate-supported Pd, reporting that the influence of this compound (positive or negative) depends strongly on the operating temperature and the oxygen concentration. The simultaneous presence of nitrogen- and sulphurcontaining molecules, together with carbon oxides, H2 , H2 O, and other organic compounds is expected to affect the combustion of methane (i.e. the reaction kinetics, and the kinetics and reversibility of catalyst deactivation). In order to evaluate the industrial applicability of Pd for the catalytic combustion of effluents that contain such compounds, either continuously or intermittently (e.g. coke oven emissions), or that can be contaminated occasionally with them (e.g. by impurities in feedstocks or the failure of separation processes), studying the behaviour of such complex mixtures is of the highest interest. But in spite of this, such studies are very rare in the literature. Mowery et al. [18] studied the oxidation of methane in a simulated gas engine exhaust emission, but focused their studies on the effect of SO2 and water on Pd, finding that the presence of water increases the pernicious effect of SO2 on the catalyst activity and stability. Auer et al. studied the behaviour of more complex

mixtures using perovskite-based catalysts [19]. They found that La0.9 Ce0.1 CoO3 , the most sulphur-tolerant among the catalysts tested in this study, was irreversibly deactivated in the presence of sulphur compounds. In the present work, the behaviour of a Pd/␥-alumina catalyst for the combustion of methane in the presence of several inorganic compounds (NH3 , NO2 , H2 , H2 O, SO2 , H2 S, CO and CO2 ) is studied, both for binary, methane–inorganic compound mixtures in air, and for mixtures containing several inorganic compounds simultaneously. The reversibility of the effects is also studied.

2. Experimental A commercial 0.5% (w/w) Pd on ␥-alumina catalyst (BASF RO-25) was used in this work, with textural characteristics (measured by nitrogen adsorption with a Micromeritics ASAP 2000 apparatus) and composition (given by the manufacturer) listed in Table 2. The catalyst, available in extruded forms, was milled and sieved, and the fraction between 250 and 350 ␮m was selected for the reaction. The reactor, a lab-scale fixed bed, consisted of a stainless steel tube of 40 cm length and 0.9 cm internal diameter, placed inside a PID-controlled electric furnace. Five thermocouples measured the external wall temperature of the reactor at different heights. Temperature in the reaction zone (the height of the zone charged with 0.3–1.2 g catalyst was 0.5–1 cm) was measured by a thermocouple inserted on top of the catalytic bed. This thermocouple provided the signal for the PID temperature controller. Discrepancies between temperatures measured in the centre of the bed and at the external reactor wall for the same bed height were always lower than 10 ◦ C. The experimental device used in this work ensured the absence of relevant thermal effects, due to the small reactor diameter and the high thermal conductivity of the stainless steel wall. The isothermicity of the reactor was verified experimentally in a previous paper [21] for the oxidation of 5000 ppm of toluene (a reaction much more exothermic than the one studied here). Mass flow regulators supplied by Brooks (5850 TR/DA1B6 D1, 5850 TR/DB1B3D1 provided with electronic device 0152) and by Bronkhorst HI-TEC (F-201C-FA-22 provided Table 2 Composition and principal properties of the catalyst used in this work Composition (wt.%)

0.5% Pd, balance ␥-Al2 O3

BET specific surface areaa (m2 /g) BJH desorption pore volumeb (cm3 /g) Average pore diameter (nm) Metal dispersion (%, H2 chemisorption)

108.5 0.43 15.5 75.2

a

Determined by the method of Brunauer–Emmet and Teller. Mesopore volume, determined by the method of Barret–Joyner and Halenda (desorption). b

P. Hurtado et al. / Applied Catalysis B: Environmental 47 (2004) 85–93

with electronic device E-7100) controlled the flow rate of the gas mixtures fed to the reactor. A Cole Palmer syringe pump (74900-00-05) was used to add water to the reaction mixture. The reactor feed line was traced to avoid condensation. A dumping tank provided with a relief valve avoided possible fluctuations in the inlet stream to the reactor. The gas feed flowed downward through the reactor. Further details on the experimental set-up are given in [21]. The catalyst, diluted with 1 g of glass ground and sieved to 355–500 ␮m particle size, was placed in the middle section of the reactor, supported on a metallic mesh. The upper part of the reactor was filled with 355–500 ␮m glass particles, in order to pre-heat the reaction mixture. The lower part of the reactor remained empty, in order to limit the pressure drop through the reactor. In this way, the reactor was operated at nearly atmospheric pressure (Pmax = 0.3 bar). The reactor gas feed consisted of methane in synthetic air, with inorganic compounds added. The feed was made up by mixing synthetic air (Air Liquide), oxygen, nitrogen, a 2.5% (v/v) methane in synthetic air mixture, several gaseous mixtures containing one or more inorganic gases (Air Products), all with purity higher than 99.9995, and liquid water. Prior to the reactions, air at 350 ◦ C was circulated overnight through the reactor charged with the catalyst, in order to provide a clean catalyst surface and to eliminate the influence of thermal processes on the catalyst surface, or of any phenomena caused by the products emitted from the catalyst surface, such as HCl, a strong catalyst poison [20]. Exhaust gas was analysed by gas chromatography (Hewlett Packard HP 5890 Series II). Methane in the inlet and outlet streams was analysed using a 30 m fused-silica capillary column with an apolar stationary phase SE-30, and a FID detector. H2 , CO and CO2 were analysed using HayeSep N 80/100 and molecular sieve 45/60 columns connected in series, with a TCD detector. The remaining inorganic gases (H2 S, SO2 , SO3 , NH3 and NOx ) were analysed using an electrochemical detector, Dräger MiniWarm. Methane conversions were calculated from both reactor inlet and outlet methane concentrations (measured with the FID), and from CO2 concentrations at the reactor outlet (measured with the TCD), data calculated in both ways being in good agreement. When the reaction was stopped, the catalyst was cooled “in situ” by circulating air at room temperature, separated from the glass particles by sieving, and collected for subsequent characterisation. Fresh and used catalysts were characterised by different techniques. Nitrogen Porosimetry measurements were performed in a Micromeritics ASAP 2000 apparatus. Crystallographic changes were studied by X-ray diffraction in a Philips PW1710 diffractometer, working with the K␣ line of Cu. Exposed metal was evaluated by hydrogen chemisorption at 70 ◦ C (in order to avoid the formation of hydrides) using the pulse technique in a Micromeritics 2900 apparatus, equipped with a TCD detector. Prior to chemisorption analysis, catalysts samples were reduced at

87

350 ◦ C in hydrogen for 1 h. Hydrogen flow was then stopped and the sample cooled to 70 ◦ C in Ar.

3. Results and discussion 3.1. Effect of single compounds The scope of this work initially was to study the effect of single inorganic compounds on the performance of the selected catalyst for methane combustion. In a preliminary set of experiments, the effect of benzene at low concentration (150 ppm) was studied, and this effect was found to be negligible, as expected, because of the low concentration and high reactivity of benzene compared to methane. The effect of the inorganic compounds listed in Table 1 on the combustion of 5000 ppm V methane was studied, working at a space time of 5.8 g cat min/mmol of CH4 (WHSV = 50 h−1 ). The experiments consisted, first, of obtaining the light-off curves (methane conversions versus temperature), from room temperature to 550 ◦ C at a ramping rate of 2.5 ◦ C/min. Then, the catalyst was maintained in the reactor at 450 ◦ C for 50 h. This part of the procedure will be called “aging experiments”. Finally, a second light-off curve was obtained for the aged catalyst. Results are shown in Figs. 1–4 (light-off curves for fresh and aged catalysts), Table 3 and Fig. 5 (evolution of conversion versus time for 50 h at 450 ◦ C, constant temperature). Results indicate that, when no inorganic compound is added (only the CO2 and water formed in the reaction are present), the catalyst deactivated slightly (methane conversion decreased less than 8% after 60 h on stream, remaining then almost constant after 100 h). This slight deactivation is produced by water formed during the reaction, as observed by other researchers [22]. In the same way, it was observed that CO2 leads to no noticeable decrease in the activity of the catalyst. The observed effect of CO2 is in a good agreement with the findings of

Fig. 1. Light-off curves for the oxidation of methane alone (), in the presence of 3000 ppm CO (䊐) and in the presence of 8000 ppm CO2 (䊊). Open symbols correspond to fresh catalyst and closed symbols correspond to light-off curves recorded after 50 h on stream at 450 ◦ C.

88

P. Hurtado et al. / Applied Catalysis B: Environmental 47 (2004) 85–93 Table 3 Summary of the results obtained in the aging experiments carried out in the presence of inorganic compounds

Fig. 2. Light-off curves for the oxidation of methane alone (), in the presence of 25 ppm H2 S (䊐) and in the presence of 40 ppm SO2 (䊊). Open symbols correspond to fresh catalyst and closed symbols correspond to light-off curves recorded after 50 h on stream at 450 ◦ C.

Fig. 3. Light-off curves for the oxidation of methane alone (), in the presence of 180 ppm NH3 (䊐) and in the presence of 60 ppm NO2 (䊊). Open symbols correspond to fresh catalyst and closed symbols correspond to light-off curves recorded after 50 h on stream at 450 ◦ C.

Fig. 4. Light-off curves for the oxidation of methane alone (), in the presence of 7000 ppm H2 (䊐) and in the presence of 20,000 ppm H2 O (䊊). Open symbols correspond to fresh catalyst and closed symbols correspond to light-off curves recorded after 50 h on stream at 450 ◦ C.

Inorganic compound

Aging period (h)

Conversion decrease (%)

None None CO CO2 H2 S SO2 NH3 NO2 H2 H2 O

50 100 50 50 50 50 50 50 50 50

7.9 8.2 8.93 8.76 73.7 77.7 7.1 6.3 15.7 14.0

Cullis et al. [23] and Burch et al. [24], who stated that the presence of CO2 does not affect the catalyst performance in the oxidation of methane. In contrast, Ribeiro et al. reported catalyst deactivation when CO2 was added to the reactor feed, working at lower temperature (275 ◦ C) and conversions [25]. The negligible effect of CO2 is explained by the very weak interaction that can be expected between this fully oxidised compound and the catalytically active phase. CO was found to have an effect similar to CO2 on methane combustion. In the literature it is reported that CO is a strong poison for the oxidation of toluene, benzene, and 1-hexene over Pt catalysts [26], and for the oxidation of polyaromatic compounds over Pd catalysts [27]. In the case of methane combustion, the low effect observed can be explained by considering the much higher reactivity of CO when compared with methane. In fact, in separate experiments it was observed that CO was fully oxidised to CO2 below 130 ◦ C, temperatures at which methane conversion is negligible when it is present in the feed. So at the conditions needed to combust methane, CO is fully oxidised. Concerning the sulphur-containing compounds studied (SO2 and H2 S), both light-off curves and aging experiments

Fig. 5. Aging curves for the oxidation of 5000 ppm methane at 450 ◦ C in the presence of 25 ppm H2 S (䊏), 40 ppm H2 S (䊐) and 40 ppm SO2 (䊊).

P. Hurtado et al. / Applied Catalysis B: Environmental 47 (2004) 85–93

show that they reduce strongly the catalyst performance (Figs. 2 and 5), the decrease being greater as time on stream increases, and as the concentration of sulphur increases (except for the light-off curves for fresh catalyst; this behaviour will be discussed latter). The comparison of the effects of SO2 and H2 S from the experimental data given in Figs. 2 and 5 is difficult, since the concentration of both compounds is different (40 ppm SO2 , 25 ppm H2 S). In order to clarify this point, a subsequent experiment was carried out at 40 ppm H2 S, maintaining the catalyst at 450 ◦ C, constant temperature. Results (Fig. 5) indicate that the effect of both sulphur-containing compounds is very similar, as the deactivation curves for 40 ppm SO2 and 40 ppm H2 S basically coincide. This result is in a good agreement with the observed total oxidation of H2 S to SO2 at reaction conditions, whereas the formation of SO3 was not detected in any case. Similar behaviour is reported in the literature, and it is assumed to be caused by the formation of inactive Pd sulphates from irreversibly adsorbed SO2 , which is then oxidised to SO3 on the catalyst surface and finally forms the sulphate [12–15,28]. It can also be seen in Fig. 5 that, whereas there is a difference between the deactivation curves for sulphur concentrations of 25 and 40 ppm for low time on stream, conversions for high time on stream are much closer. This result suggests that the state attained by the catalyst after 50 h of operation is very similar for all sulphur-containing feeds. The lower deactivation effect of H2 S observed in the light-off curves for fresh catalyst can be explained, taking into account the unsteady nature of the experiment, by the kinetics of the different steps of the deactivation process. Concerning the nitrogen compounds (NO2 and NH3 ), it was observed both in the light-off curves and during the period at constant temperature (Table 3 and Fig. 3) that they increase the methane conversion slightly. In addition, results reveal that the effect of each compound is very similar, even though the concentration of ammonia is almost three times higher than that of NO2 . Concerning the reaction products, ammonia oxidation can produce NO, N2 O, NO2 and N2 . Although in this work it was not possible to quantify the concentrations of the reaction products, Escandón et al. [29] studied the oxidation of ammonia over several catalysts, including Pd supported on alumina, finding that, at similar reaction conditions, ammonia is almost fully converted to a mixture of NO, NO2 and N2 O. Analogous results were presented by Öcal et al. [17], who reported an increase in methane conversion to CO2 on a hexaaluminate-supported Pd catalyst in the presence of much higher concentrations of N-containing compounds than in our case (2.3% NO, that reacted partially to NO2 ). This effect was observed above 410 ◦ C, a temperature at which the possible contribution of homogeneous reactions can be neglected. As indicated by Öcal et al., the favourable effect of NO can be explained either by an increase in the availability of surface oxygen, formed by the decomposition of NO2 , or by the formation of the O• species on the surface, which may initiate the methane combustion.

89

Regarding the effect of water and hydrogen, both light-off curves (Fig. 4) and aging curves (Table 3) show a decrease in methane conversion, the decrease being more pronounced as time on stream increases. This effect is much less marked than for sulphur-containing compounds. Over the Pd catalyst used, hydrogen is fully oxidised to water at low temperature (180 ◦ C), at which methane conversion is still negligible. So the effect of hydrogen takes place through the water formed, and either hydrogen or water added to the reaction feed will produce similar effects on the catalyst. In the literature, water has been found to produce two different effects: inhibition of the oxidation reaction [16,25] and modification of the properties of the support and/or active phase [30], altering the active phase-support interaction, or forming surface Pd hydroxides, less active than the oxide. The experimental observation that the effect of the water produced by the combustion of hydrogen (7000 ppm) is stronger than that of the water added to the feed (20,000 ppm) can be explained taking into account that the combustion of hydrogen produces water directly on the catalytic surface, hence it is very likely that hydrogen combustion produces high local water concentrations near the catalyst surface. 3.2. Characterisation of fresh and aged catalyst Samples of catalyst, fresh and collected after the experiments (which comprised recording of the light-off curves plus the aging run), were characterised by XRD, nitrogen physisorption (BET), and hydrogen chemisorption (pulse technique). No heterogeneity in the catalyst samples coming from different reactor positions was observed. Neither crystallographic changes (both fresh and aged samples showed the same XRD peaks, corresponding to ␥-alumina) nor substantial textural differences between fresh and aged catalysts (surface area and pore size distribution) were found. Concerning the metal dispersion, pulse chemisorption of hydrogen revealed that the metal dispersion of the aged samples was very similar in all cases (43–45%), except for the experiments with sulphur compounds. These values were markedly lower than those corresponding to fresh catalyst (75.2%). In order to investigate the cause of this decrease, a blank experiment, in which the same procedure was followed as previously (light-off heating, aging at 450 ◦ C during 50 h, and second light-off heating), was carried out, but in which the reactor was fed only with air. Metal dispersion for the catalyst aged through this procedure was 43%, which indicates that thermal sintering is responsible for the observed loss of metal dispersion. Despite this sharp decrease in Pd dispersion, the extent of self-deactivation detected was small (Table 3). A more pronounced decrease of metal dispersion was found for the catalysts exposed to sulphur compounds (final dispersions 7% for H2 S and 3.5% for SO2 ). This decrease could be caused by an irreversible poisoning of the catalyst

90

P. Hurtado et al. / Applied Catalysis B: Environmental 47 (2004) 85–93

(formation of PdSO4 ), or by sintering during the process of reduction of the catalyst (carried out in the pulse chemisorption apparatus at 350 ◦ C). An active role of sulphur compounds in the phenomena leading to the migration of Pd atoms is unlikely, as these effects are reported to take place when the inorganic compounds (anionic or nucleophilic species) can form volatile species with Pd (Cl− or CO) [31]. However, the formation of PdSO4 is likely to occur, since surface sulphates could be transformed during reduction into sulphur-reduced species chemisorbed strongly on the Pd surface. Therefore, the amount of Pd atoms available for hydrogen chemisorption would decrease. 3.3. Catalyst performance in the presence of mixtures of inorganic compounds The behaviour of more complex reaction feeds was studied using the mixtures with compositions listed in Table 4. The experimental procedure consisted of maintaining the reactor overnight at 350 ◦ C while feeding air, and then increasing the temperature to 450 ◦ C. When this temperature was reached, the reactor feed was started, and the reaction was then maintained at constant temperature for up to 50 h. The results obtained are shown in Figs. 6 and 7 (feed with no water added) (2.0–2.7% H2 O). Results confirm the pernicious effect of SO2 and H2 O (both produced by combustion of H2 or added to the reaction feed) found in Section 3.1, and in the results of Mowery et al. [18] and Mowery and McCormick [32], who reported an important synergism between both compounds in the deactivation of a PdO catalyst used for methane oxidation. Concerning the effect of moisture, it can be observed that, although in the presence of added water the catalyst deactivates faster, the final conversion is very similar for both cases. So, after 8 h on stream, in both cases methane conversion decreases about 92% (mixtures 6 and 10). This result suggests a saturation of the catalyst, the addition of water above a certain value causing no additional effect on the catalyst performance. The synergistic effect of water and SO2 could be explained in different ways. Water could modify the surface reactivity of the support (alumina), decreasing its affinity for interac-

Fig. 6. Aging curves for the oxidation of 5000 ppm V methane in the presence of different complex mixtures (all of them in the absence of external water): 5000 ppm CH4 , (䊏); 5000 ppm CH4 and 65 ppm SO2 (); 5000 ppm CH4 , 65 ppm SO2 and 268 ppm NH3 (䉫); 5000 ppm CH4 , 65 ppm SO2 , 180 ppm NH3 and 7000 ppm H2 (䊐); 5000 ppm CH4 , 65 ppm SO2 , 180 ppm NH3 , 7000 ppm H2 and 107 ppm NO2 (+); 5000 ppm CH4 , 40 ppm SO2 , 25 ppm H2 S, 180 ppm NH3 , 8000 ppm CO2 , 3000 ppm CO, 7000 ppm H2 and 65 ppm NO2 (䊊).

Fig. 7. Aging curves for the oxidation of 5000 ppm V methane in the presence of different complex mixtures (all of them in the presence of external water): 5000 ppm CH4 and 27,000 ppm H2 O (䊏); 5000 ppm CH4 , 65 ppm SO2 and 27000 ppm H2 O (䉱); 5000 ppm CH4 , 65 ppm SO2 , 180 ppm NH3 and 27,000 ppm H2 O (䊉); 5000 ppm CH4 , 40 ppm SO2 , 25 ppm H2 S, 180 ppm NH3 , 8000 ppm CO2 , 3000 ppm CO, 7000 ppm H2 , 65 ppm NO2 and 20,000 ppm H2 O (䉬).

Table 4 Composition of the reaction feeds used in the study of complex mixtures (ppm V) Mixture

CH4

1 2 3 4 5 6 7 8 9 10

5000 5000 5000 5000 5000 5000 5000 5000 5000 5000

SO2 65 65 65 65 40 65 65 40

H2 S

25

25

CO2

8000

8000

CO

H2

NH3

NO2

3000

7000 7000 7000

268 180 180 180

107 65

3000

7000

180 180

65

H2 O

27000 27000 27000 20000

P. Hurtado et al. / Applied Catalysis B: Environmental 47 (2004) 85–93

tion with sulphur oxides, and hence increasing the exposure of Pd to sulphur. In agreement with this explanation, Lampert et al. found that Pd supported on materials with low affinity for sulphur oxides are quickly deactivated [33]. Another possible explanation could be the formation of Pd(OH)2 in presence of water (as proposed by Burch et al. [24]), this species being more reactive with sulphur oxides than PdO. In contrast with the experiments for single compounds (Section 3.1), for which the presence of ammonia enhanced methane conversion, when ammonia is added to a SO2 –methane mixture, methane conversion decreases. This behaviour could be explained considering that, due to its basic character, ammonia can react with acid species (such as SO2 or SO3 ) present on the proximity of the active phase, leading to the formation of ammonia sulphites or sulphates, which would cover the active phase or block pores. This hypothesis is very difficult to demonstrate, since (NH4 )2 SO4 is unstable under vacuum conditions, which makes difficult its detection by nitrogen physisorption or by using the conventional surface analysis techniques. However, leaching of spent catalyst in boiling water, followed by analysis of the leachates using Ionic Chromatography, proved the presence of NH4 + and SO4 2− ions, which supports this hypothesis. The behaviour of the methane–sulphur compounds–hydrogen–ammonia mixtures (mixtures 4 and 9) is very similar to the behaviour of the more complex mixtures. This confirms that the presence of CO, CO2 , and NO2 has no important effect on the catalyst performance at the conditions studied. 3.4. Studies of reversibility The study of the reversibility of the effects of inorganic compounds on the catalytic combustion of methane is very interesting, in evaluating the industrial feasibility of this process. If the deactivating effects are reversible, the catalyst affected by the continuous or occasional presence of deactivating compounds can be easily regenerated. In other cases, such as the emissions from coke ovens, the presence of deactivating compounds is intermittent and periodical, so that the intermediate periods could be used to restore the catalyst activity [4]. The experiments carried out with this purpose consisted of maintaining the reactor at 350 ◦ C overnight with a feed of air, increasing the temperature to 450 ◦ C, then starting to feed the reaction mixture (1 l/min, stp), and maintaining the reaction at constant temperature for 30 min. The reactor feed was then changed for 70 min to (a) 5000 ppm V methane in air, or (b) air only, and maintaining constant temperature and total flow rate in both cases. After this 70 min period the feed was changed again to the initial concentration. Five feed mixtures were studied, with compositions given in Table 5. These experiments were carried out at lower space time (672 g min/mol of CH4 , WHSV = 431 h−1 ), in order to follow easily the catalyst deactivation and regeneration. The

91

Table 5 Composition of the reaction feeds used in the reversibility studies (ppm V) Mixture

CH4

SO2

1 2 3 4 5

5000 5000 5000 5000 5000

65 65 65

NH3

H2 O 27000

180

27000 27000

two regeneration feeds used (air or air plus methane) were selected in order to determine if they produce different regeneration. On the other hand, the presence of methane in the regenerating stream allows us to follow the exit methane conversion, and hence the extension of the catalyst regeneration. Results are shown in Figs. 8 and 9. It can be observed that the effect of water is essentially reversible, since when water is removed from the feed, methane conversion is the same as in operation with no water added. This result is especially interesting from a practical point of view, since it indicates that although methane conversion is affected by variations in the moisture content of an off-gas to be treated by catalytic combustion, the stability of the catalyst remains basically unaffected. In contrast with our results, Mowery et al. [18] reported permanent activity loss for a 1% Pd/alumina catalyst used for the catalytic combustion of 790 ppm methane in the presence of steam. This discrepancy could be caused by either the higher temperature (520 ◦ C), the higher water concentration (up to 9%), or the lower oxygen concentration (6%) used by this researcher, so that the equilibrium between Pd oxides and hydroxides would be more displaced to the hydroxide form. On the other hand, Kikuchi et al. [34], studying the combustion of methane (1%) over Pd/SnO2 and Pd/Al2 O3 at 280 ◦ C in the

Fig. 8. Oxidation of 5000 ppm V methane in air with cyclic feed of inorganic gases: with no inorganic gases (䊐), in the presence of SO2 (), H2 O (䉬), SO2 and H2 O, (䊊) and SO2 , NH3 and H2 O (䉱). See concentrations in Table 5. During the period 30–100 min, the feed was commuted to only 5000 ppm V methane in air. Zone marked with (1) means presence of inorganic gases in the feed, whereas zone marked with (2) means absence of these compounds.

92

P. Hurtado et al. / Applied Catalysis B: Environmental 47 (2004) 85–93

range of concentrations studied, except sulphur-containing compounds, which cause severe and irreversible deactivation of the catalyst.

Acknowledgements

Fig. 9. Oxidation of 5000 ppm V methane in air with cyclic feed of inorganic gases: effect of feed composition during the commutation period (air or air + methane). See concentrations in Table 5. With no inorganic gases (䊐). In presence of SO2 and regeneration with air (䉱) and methane/air (). In presence of H2 O and regeneration with air (䉬) and methane/air (䉫). Zone marked with (1) means presence of inorganic gases in the feed, whereas zone marked with (2) means absence of these compounds.

presence of high amounts of water (up to 20%), found that the effect of steam was essentially reversible. Concerning the effect of SO2 , it is partially reversible, so that methane conversion increases from 10 to 30% when SO2 is removed from the feed. This result would indicate that the surface sulphates, which are assumed to be mainly responsible for the deactivation, are partially decomposed in the presence of a sulphur-free feed. The presence of ammonia in the feed does not affect appreciably the extent of the regeneration. This aspect leads one to think that the ammonium sulphate, assumed to be formed under reaction conditions, is rapidly decomposed in the absence of nitrogen-and sulphur-containing compounds.

4. Conclusions Methane catalytic abatement in gaseous emissions by catalytic combustion on Pd-alumina is severely affected by the presence of sulphur compounds (SO2 and H2 S). This effect is mostly irreversible. Water, both added to the feed or produced by combustion of hydrogen, presents a much milder, entirely reversible inhibitory effect. The presence of water increases the negative effect of sulphur compounds. Carbon monoxide and dioxide do not have a noticeable effect on the catalyst performance, whereas nitrogen compounds (NH3 and NO2 ) improve the catalyst performance slightly in the absence of sulphur compounds, while the effect of ammonia is negative in the presence of sulphur compounds. This behaviour is tentatively explained considering the formation of surface ammonia sulphates. The catalyst studied is then adequate for the combustion of methane in the presence of the compounds considered in the

This work has been supported by a grant from the European Commission (Contract ENV4-CT97-0599). The catalyst has been kindly supplied by BASF-Spain. Profs. B. Delmon and F. Thyrion (Catholic University of Louvain, Louvain-la-Neuve, Belgium) are acknowledged for their fruitful discussions. Ms. L. Escandón is also acknowledged for her help in the catalyst characterisation.

References [1] N. Mukhopadhyay, E.C. Moretti, Reducing and Controlling Volatile Organic Compounds, Centre for Waste Reduction Technologies, AIChE, New York, 1993. [2] R.E. Heyes, S.T. Kolazkowski, Introduction to Catalytic Combustion, Gordon & Breach, Amsterdam, 1997. [3] S.H. Schneider, Global Warning, Vintage Books, New York, 1989. [4] L.S. Escandón, M.A.G. Hevia, J.R. Paredes, P. Hurtado, S. Ordóñez, F.V. D´ıez, NATO/CCMS Pilot Study, EPA Report 625CR-02/003, 2002, p. 22. [5] P. Gélin, M. Primet, Appl. Catal. B 39 (2002) 1. [6] J.W.A. Schlangen, G.W. Neuhaus, M. Madani, W.F. Maier, J. Prakt. Chem. 334 (1992) 465. [7] S.E. Oh, P.J. Mitchel, R.M. Siewert, J. Catal. 132 (1991) 287. [8] R. Burch, P.K. Loader, Appl. Catal. B. 5 (1994) 149. [9] I.L. Dodgson, D.E. Webster, Stud. Surf. Sci. Catal. 1 (1976) 279. [10] V. Maeyoo, D.L. Trimm, N.W. Cant, Appl. Catal. B 16 (1998) 105. [11] P. Hurtado, S. Ordóñez, H. Sastre, F.V. D´ıez, in: H. Geilig (Ed.), Book Extend. Abstracts of the 4th World Congress Oxidation Catalysis, Sept. 2001, Dechema e. V, Frankfurt am Main, 2001, p. 451. [12] L.J. Hoyos, H. Praliaud, M. Primet, Appl. Catal. A 98 (1993) 125. [13] T.C. Yu, H. Shaw, Appl. Catal. B 18 (1998) 105. [14] Y. Deng, T.G. Nevell, R.J. Ewen, C.L. Honeybourne, M.G. Jones, Appl. Catal. 51 (1993) 101. [15] E.M. Johansson, M. Berg, J. Kjellström, S.G. Järas, Appl. Catal. 20 (1999) 319. [16] J.C. Van Giezen, F.R. Vandenberg, J.L. Kleinen, A.J. Vandillen, J.W. Geuss, Catal. Today 47 (1999) 287. [17] M. Öcal, R. Oukaci, G. Marcelin, B.W.L. Jang, J.J. Spivey, Catal. Today 59 (2000) 205. [18] D.L. Mowery, M.S. Graboski, T.R. Ohno, R.L. McCormick, Appl. Catal. B 21 (1999) 157. [19] R. Auer, M. Alifanti, B. Delmon, F.C. Thyrion, Appl. Catal. B 39 (2002) 311. [20] C.F. Cullis, B.M. Willat, J. Catal. 132 (1991) 267. [21] L. Bello, S. Ordóñez, R. Rosal, H. Sastre, F.V. D´ıez, Appl. Catal. B 38 (2002) 139. [22] M. Lyobovsky, P. Pfeferle, Catal. Today 47 (1999) 29. [23] C.F. Cullis, T.G. Nevell, D.L. Trimm, J. Chem. Soc. Far. Trans. I 68 (1971) 1406. [24] R. Burch, F.J. Urbano, P.K. Loader, Appl. Catal. A 123 (1995) 173. [25] F.H. Ribeiro, M. Chow, R.A. Della Betta, J. Catal. 146 (1994) 537. [26] M.J. Patterson, D.E. Angove, N.W. Cant, Appl. Catal. B 26 (2000) 47. [27] F. Klingstedt, A. Kalantar, L.E. Lindfors, T. Salmi, T. Hekkila, E. Laine, Appl. Catal. A 239 (2003) 229. [28] T.C. Yu, H. Shaw, Appl. Catal. B 18 (1998) 105.

P. Hurtado et al. / Applied Catalysis B: Environmental 47 (2004) 85–93 [29] L.S. Escandón, S. Ordóñez, F.V. D´ıez, H. Sastre, React. Kinet. Catal. Lett. 76 (1) (2002) 61. [30] P. Euzen, J.H. LeGal, B. Rebours, G. Martin, Catal. Today 47 (1999) 19. [31] S. Ordónez, F.V. D´ıez, H. Sastre, Appl. Catal. B 31 (2001) 113.

93

[32] D.L. Mowery, R.L. McCormick, Appl. Catal. B 34 (2001) 287. [33] J.K. Lampert, M.S. Kazi, R.J. Ferrauto, Appl. Catal. B 14 (1997) 211. [34] R. Kikuchi, S. Maeda, K. Sasaki, S. Wennerström, K. Eguchi, Appl. Catal. A 232 (2002) 23.