Environ. Sci. Technol. 2004, 38, 4664-4670
Activated Carbon and Tungsten Oxide Supported on Activated Carbon Catalysts for Toluene Catalytic Combustion M. A. ALVAREZ-MERINO,† M. F. RIBEIRO,‡ J. M. SILVA,§ F . C A R R A S C O - M A R IÄ N , | A N D F. J. MALDONADO-HO Ä D A R * ,| Departamento Quı´mica Inorga´nica y Orga´nica, Facultad de Ciencias Experimentales, Universidad de Jae´n, 23071, Jae´n, Spain, Departamento Engenharia Quı´mica, Instituto Superior Te´cnico, Avenue Rovisco Pais, 1049-001, Lisboa, Portugal, Departamento Engenharia Quı´mica, Instituto Superior Engenharia de Lisboa, Rua Cons, Emı´dio Navarro, 1949-014 Lisboa, Portugal, and Departamento Quı´mica Inorga´nica, Facultad de Ciencias, Universidad de Granada, 18071, Granada, Spain
We have used activated carbon (AC) prepared from almond shells as a support for tungsten oxide to develop a series of WOx/AC catalysts for the catalytic combustion of toluene. We conducted the reaction between 300 and 350 °C, using a flow of 500 ppm of toluene in air and space velocity (GHSV) in the range 4000-7000 h-1. Results show that AC used as a support is an appropriate material for removing toluene from dilute streams. By decreasing the GHSV and increasing the reaction temperature AC becomes a specific catalyst for the total toluene oxidation (SCO2 ) 100%), but in less favorable conditions CO appears as reaction product and toluene-derivative compounds are retained inside the pores. WOx/AC catalysts are more selective to CO2 than AC due to the strong acidity of this oxide; this behavior improves with increased metal loading and reaction temperature and contact time. The catalytic performance depends on the nonstoichiometric tungsten oxide obtained during the pretreatment. In comparison with other supports the WOx/ AC catalysts present, at low reaction temperatures, higher activity and selectivity than WOx supported on SiO2, TiO2, Al2O3, or Y zeolite. This is due to the hydrophobic character of the AC surface which prevents the adsorption of water produced from toluene combustion thus avoiding the deactivation of the active centers. However, the use of WOx/AC system is always restricted by its gasification temperature (around 400 °C), which limits the ability to increase the conversion values by increasing reaction temperatures.
1. Introduction The catalytic combustion of volatile organic compounds (VOCs) to control gaseous industrial emissions is one of the * Corresponding author phone: 34 958 240444; fax: 34 958 248526; e-mail:
[email protected]. † Universidad de Jae ´ n. ‡ Instituto Superior Te ´ cnico. § Instituto Superior Engenharia de Lisboa. | Universidad de Granada. 4664
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 38, NO. 17, 2004
most promising environmental technologies. VOCs are typically oxidized using a catalyst at temperatures below 600 °C, while thermal incineration processes need temperatures higher than 1000 °C. The catalytic process permits energy saving and minimizes other negative effects such as NOx formation, although the formation of intermediate oxidation products, such as dioxins, should be controlled. Catalysts based on noble metals (Pt, Pd, Rh) supported on different materials present high activity and selectivity to the complete pollutant oxidation (1-4). However they are expensive, susceptible to poisoning (2, 5-8) and show, in some cases, a poor thermal stability. Therefore, there is a clear need to develop cheaper active and selective catalysts by improving either the support or the active phase. It has been pointed out previously that the complete oxidation of VOCs is achieved at lower temperatures using hydrophobic supports (activated carbon, polymers) (9, 10). The water vapor generated from VOCs oxidation, mainly at low temperatures, can deactivate the active sites of inorganic supports, which show a hydrophilic behavior. Moreover, organic compounds are more readily adsorbed on hydrophobic surfaces (3). Acidity is also expected to be a favorable factor for combustion catalysts since it is known that the acid character of supports or catalysts favors the oxidation reactions of organic compounds (1, 11, 12). Transition metal oxides have been used for VOCs oxidation (13-16). The catalytic performance shown by these oxides is in general lower than that for noble metals, although they could tolerate higher levels of poisons. Many catalytic oxidation behaviors can be represented by a simple oxidation-reduction mechanism:
VOC + oxidized catalyst f reduced catalyst + oxidized product (1) reduced catalyst + O2 f oxidized catalyst
(2)
Thus, the ability of oxide catalysts to develop reduction/ oxidation cycles is an important step in the combustion process and controls catalytic behavior (17, 18). Tungsten oxide is a solid with an acidic character which has many applications in heterogeneous catalysis and has an ability to form a large variety of stoichiometric and nonstoichiometric oxides depending on the atmosphere and thermal treatment used (19-24). Taking these factors into account, we expect the WOx/AC system to be a good combustion catalyst. In previous papers, we have described the preparation, characterization, and catalytic behavior in the skeletal isomerization of 1-butene (22), decomposition reactions of alcohols (23), or ethylene hydrogenation (24) of different series of WOx/AC catalysts. The aim of this work is to investigate the catalytic behavior of tungsten oxide deposited on activated carbon for deep oxidation of toluene. The influence of tungsten loading and pretreatment conditions and the role of the support were analyzed. The role of the support was also evaluated by comparison of the WOx/AC system with catalysts based on tungsten impregnated in different materials, namely alumina, silica, anatase, and Y zeolite.
2. Experimental Section A sample of activated carbon was prepared from almond shells in two steps: carbonization at 850 °C under N2 flow and activation on steam for 13 h at 750 °C, using in both cases a gas flow of 18 L h-1. The activation degree was 40 wt %. Commercial Al2O3, TiO2, and SiO2 were used, as well as 10.1021/es034964c CCC: $27.50
2004 American Chemical Society Published on Web 07/27/2004
TABLE 1. Main Characteristics of the Tungsten Catalysts Supported on Activated Carbon Treated in Inert Atmosphere at 475 °C for 4 h sample
SN2 a (m2 g-1)
V2 b (cm3 g-1)
V3 c (cm3 g-1)
AC 4.8W/AC 14.9W/AC 23.1W/AC
929 ( 7 834 ( 6 680 ( 3 576 ( 5
0.17 0.20
0.36 0.33
0.10
0.23
rpd (µmol g-1 min-1)
NH3 desorbede (µmol g-1)
(W/C)tf
(W/C)sf
9.8 20.8
41 74 112
0.0033 0.0114 0.0196
0.0032 0.0123 0.0224
a
SN2. BET surface area determined from the N2 adsorption isotherms at 77 K. b V2. “Mesopore” volume; volume of pores with diameter between 3.6 and 50 nm. Determined by mercury porosimetry. c V3. Macropore volume; volume of pores with diameter greater than 50 nm. Determined by mercury porosimetry. d rp. 2-Propanol dehydration rate. Catalyst weight: 0.5 g pretreated in N2 at 475 °C. Reaction conditions: 110 °C; 3.6 L h-1 of N2 saturated on 2-propanol at 0 °C. e NH3 desorbed. Amount of NH3 desorbed during DTP experiments, after saturation in NH3 flow at 100 °C. f (W/C)t and (W/C)s. Total and surface tungsten atomic ratio, as determined by XPS.
a USHY zeolite provided by Grace Davison. The preparation method and characterization of tungsten catalysts supported on activated carbon, Al2O3, TiO2, or SiO2 are given in detail elsewhere (19-22). Tungsten was deposited on the different supports by incipient impregnation, using aqueous solutions of ammonium tungstate, to get a metal loading between 5 and 23 wt %. All catalysts were textural and chemically characterized by different techniques. The BET equation was applied to the N2 adsorption isotherms, obtaining the N2 surface area (SN2). The volumes of pores with a diameter between 3.6 and 50 nm (mesopores, V2) and wider than 50 nm (macropores, V3) were obtained by mercury porosimetry. The surface acidity of the catalysts was determined by TPD of ammonia and by decomposition of 2-propanol, according to the method developed by Ai (25). These authors consider, from a mechanistic point of view, that 2-propanol dehydration to propene is catalized by acid sites, while the dehydrogenation to acetone is catalized by both acid and basic sites. Thus, the determination of the propene formation rate (rp) is indicative of the sample acidity. The chemical state and dispersion of the tungsten phases were studied by powder XRD and XPS. The catalytic combustion of toluene was carried out at atmospheric pressure in a fixed bed flow reactor, using 0.5 g of catalyst, pretreated under N2 flow at 475 °C for 4 h. The reactant gas mixture, air containing toluene (500 or 800 ppm), was prepared by passing a flow of air through a saturator containing toluene which was kept at -3 °C. The reactant mixture was fed with a flow rate of 4 or 7 L h-1. Due to the different stability of the supports, catalytic activity was evaluated in the temperature range between 300 and 350 °C for the tungsten catalysts supported on activated carbon and between 300 and 500 °C for tungsten catalysts supported on other supports. For each temperature, the reaction time was 8 h to reach the steady state. The thermal stability of AC and WOx/AC samples was studied by TG using a C. I. Electronics thermobalance. The samples were heated in air flow (3.6 L h-1) at 10 °C min-1 up to 850 °C, keeping at this temperature until constant weight. The thermal stability and the metal loading were determined from the corresponding thermograms. The results indicate that gasification of the catalyst support becomes important above 400 °C. Below 350 °C, the gasification degree is smaller than 1%, corresponding to the release of less stable surface groups on carbon surface such as carboxylic acid (26). According to these results, the maximum reaction temperature used with these catalysts was fixed in 350 °C. Blank tests carried out during the toluene combustion experiments (without toluene in the stream and maintaining the other reaction parameters) show that the amount of CO or CO2 from AC gasification are negligible in the chosen temperature range. Reactant and reaction products were analyzed with an on-line gas chromatograph equipped with a thermal conductivity detector (TCD) and a flame ionization detector (FID)
in series. Only CO and CO2 were detected as reaction products for all experimental conditions. The hydrocarbons (toluene and possible derivatives) and carbon dioxide were analyzed with a Paraplot Q capillary column (25 m × 0.32 mm × 10 µm) and carbon monoxide with a Molecular Sieve 5A capillary column (25 m × 0.32 mm × 30 µm), with both columns being connected in parallel by a 10-port valve. Both reactant (toluene) and products (CO and CO2) were previously calibrated using adequate standards. Calibration curve fittings are always greater than r2 ) 0.99 and the uncertainties estimated from different injections are below 5%. The catalytic activity was calculated in terms of toluene conversion as
xtoluene )
[toluene]in - [toluene]out [toluene]in
× 100
The selectivity to produce CO2 was calculated as
SCO2 )
[CO2]out/7 [toluene]in - [toluene]out
× 100
and the toluene partial conversion into CO2 (XCO2) corresponds to
XCO2 )
[CO2]out 7 × [toluene]in
× 100
The partial conversion and selectivity to obtain CO were calculated in a similar way.
3. Results and Discussion 3.1. WOx/AC Catalyst Characteristics. Three tungsten-based catalysts supported on AC containing different amounts of W (4.8, 14.9, and 23.1%) have been used in this work. The AC support was also studied and compared with metal-based catalysts. As commented above, the preparation and physicochemical characteristics of AC and catalysts have been previously described (19, 24). The most relevant data observed for support and tungsten catalysts supported on activated carbon are briefly summarized in Table 1. Both the surface area (SN2) and meso + macro porosity (V2 + V3) of the catalysts decreased when the tungsten content increases. Conversely, the catalyst acidity determined either by ammonia TPD or 2-propanol dehydration (reaction rate rp) increases with tungsten loading. XPS showed that tungsten oxide is well dispersed on the carbon surface of all catalysts. Thus, the tungsten surface concentration detected by XPS (W/C)s and the total tungsten concentration (W/C)t show a linear relation (19). XRD diffraction patterns of catalysts after pretreatment at 475 °C under inert atmosphere are shown in Figure 1. The results show that crystallinity increases with metal loading, with a mixture of two oxides, WO3 and W20O58 (20-1324 and VOL. 38, NO. 17, 2004 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
4665
FIGURE 1. XRD patterns of tungsten catalysts supported on activated carbon pretreated at 475 °C in inert atmosphere: (9) WO3; (b) W20O58.
FIGURE 3. Toluene conversion obtained (a) and selectivity to CO2 (b) on activated carbon with TOS: sample weight 0.5 g pretreated at 475 °C in N2; toluene concentration 500 ppm in air; total flow 7 L h-1.
FIGURE 2. Toluene conversion obtained on activated carbon pretreated at 475 °C in N2 flow: sample weight 0.5 g; toluene concentration 500 ppm in N2; total flow 7 L h-1. 05-0386 JCPDS file, respectively), being detected. Higher pretreatment temperatures can lead to a complete reduction of WO3 which causes the formation of metallic tungsten or tungsten carbide, and a consequent decreasing of the acidity of the catalysts (19). 3.2. Role of the Activated Carbon Support. It is wellknown that activated carbons are typically used to remove organic vapors by adsorption (27), but they can also act as a catalyst in many reactions (28). Therefore, it is necessary to clarify the role of carbon supports in our reaction media: whether toluene is trapped in the porous structure (adsorbent role) or the support can itself act as a catalyst. Different experiments were carried out using samples of activated carbon. The first were performed at temperatures between 300 and 350 °C, using as feed of the reactor a flow of 7 L h-1, 500 ppm of toluene in N2. The results observed are shown in Figure 2 as the percentage of toluene removed (adsorbed or reacted) by the AC sample versus time on stream (TOS). The percentage of toluene removed decreases with the TOS and it increases progressively with reaction temperature, although total conversion was not achieved. Under these experimental conditions (absence of O2 in the feed), the oxidation of toluene is not possible, and in fact, the amount 4666
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 38, NO. 17, 2004
of carbon oxides detected at the reactor exit was negligible. Other organic compounds formed from toluene transformation were not observed either. The phenomenon that occurs under these conditions is not an adsorption process. If adsorption were the predominant effect, the contrary evolution of conversion with the temperature would be found, i.e., toluene adsorption would decrease with increasing reaction temperature. Moreover, at any reaction temperature, conversion decreases until a steady state, which is reached more rapidly for low temperatures. If toluene were removed by adsorption, the saturation of the surface would occur at a smaller TOS with increasing temperature. This means that under these experimental conditions toluene is transformed by activated carbon, probably by polymerization reactions, to heavy products that remain trapped in the porous texture, blocking the catalyst surface area, as will be discussed later. This kind of reaction that occurs on AC, where toluene is not directly adsorbed, but transformed into heavy products (coke precursors), has been observed using other porous supports. On NaHY zeolite (2, 5, 29) these deposits, identified as coke precursors, consist of heavy products composed of polyaromatic hydrocarbons and aromatic oxygenated compounds. Figure 3 shows the catalytic behavior of AC when air was used in the feed mixture, and the evolution with temperature is similar to the one observed when the catalyst is used in a non-oxidant atmosphere (Figure 2). The first consequence is an increase in the toluene conversion (Figure 3a), in such a manner that at 350 °C the toluene is initially completely removed (toluene conversion of 100%), but the conversion decreases with TOS until a steady state of 72% of conversion. At lower temperatures, the initial conversion is close to 50% and it decreases very slowly with TOS. The only desorbed products were CO2, CO, and nonreacted toluene, but the sum of selectivities to CO2 and CO is in general lower than 100%, confirming that part of toluene
TABLE 2. Influence of Space Velocity (GHSV) on the Oxidation Reaction of Toluene Using Activated Carbon (AC) Pretreated at 475 °C in N2 Flow for 4 ha GHSV (h-1)
Tr (°C)
Xtoluene %
XCO2 %
4000
325 350
62 83
40 83
7000
325 350
46 72
20 41
XCO %
SCO2 %
SCO %
65 100 44 58
13
16
TABLE 4. Toluene Conversion (X) and Selectivity to CO2 (SCO2) and CO (SCO) Obtained at Different Reaction Temperatures (Tr) Using WOx/AC Catalystsa Tr (°C)
Xtoluene (%)
XCO2 (%)
XCO (%)
SCO2 (%)
SCO (%)
AC
350 325 300
72 46 33
41 20 8
13
58 44 26
16
4.8W/AC
350 325 300
79 58 28
53 26 11
15 6
67 45 39
17 10
14.9W/AC
350 325 300
82 50 22
63 30 11
17 6
77 60 51
18 10
23.1W/AC
350 325 300
87 44 17
69 35 13
18 7
80 78 75
20 16
sample
a
Reaction conditions: sample weight 0.5 g; toluene concentration 500 ppm in air. TOS 8 h.
TABLE 3. Catalyst Characteristics and Catalytic Performance of 14.9W/AC Catalyst in the Oxidation of Toluene at 350 °C, after Treatment at Different Temperaturesa pretreatment temperature (°C)
tungsten phaseb
atomic (W/O)sc
Xtoluene (%)
SCO2 (%)
SCO (%)
950 475 350
WC + W6C2.54+W WO3 + W20O58 WO3 + WO2.98
1.09 2.83 3.10
84 82 81
42 77 56
13 18 16
a Sample weight 0.5 g; toluene concentration 500 ppm in air. GHSV 7000 h-1. TOS 8 h. b From XRD. c From XPS.
is not oxidized to CO2 or CO and remains in the carbon pores as coke precursors. The selectivity to CO2 (Figure 3b), increases with both reaction temperature and TOS. This increase of selectivity with TOS could be due to (i) a smaller tendency of toluene to polymerize inside the pores as the pore filling progresses or more active sites are deactivated, or (ii) the combustion of previously formed coke precursors. Table 2 shows the results obtained for AC in the steady state (TOS ) 8 h), by varying the space velocity (GHSV) from 4000 to 7000 h-1. Thus, increasing reaction temperature and contact time (or decreasing space velocity), result in both Xtoluene and SCO2 increase. Activated carbon becomes a specific oxidation catalyst at 350 °C and 4000 h-1, where 100% SCO2 is achieved. At 350 °C but 7000 h-1, the toluene polymerization reaction takes place and toluene combustion is incomplete, leading also to the formation of an important amount of CO. This means that the toluene adsorption and polymerization (or coke formation) processes are faster than oxidation. Toluene polymerization reactions, evaluated as Xtoluene - XCO2 - XCO, are favored with respect to the oxidation reaction at low reaction temperatures and contact time. In conclusion, activated carbon is itself a good catalyst for removing toluene from diluted streams. In non-oxidant conditions, toluene is trapped by adsorption and posterior polymerization inside the pores. In the presence of oxygen, activated carbon behaves as a specific oxidation catalyst if contact time and temperature are high enough. Otherwise, incomplete toluene combustion and/or polymerization also take place. 3.3. Catalytic Behavior of WOx/AC Systems. The tungsten active phase strongly depends on the thermal treatment carried out on the catalyst, as we described and discussed in our previous publications (19, 22, 24). To show the influence of the pretreatment and consequent tungsten phases on the catalytic behavior of samples, 14.9W/AC was pretreated at different temperatures. The results obtained are summarized in Table 3. The thermal treatment of the catalyst under inert atmosphere for 4 h, as shown by XRD, leads to a mixture of WO3 and WO2.92 if the catalyst is pretreated at 350 °C; WO3 and W20O58 if it is pretreated at 475 °C; and W, WC, and W6C2.54 if pretreated at 950 °C. However, microcrystals smaller than
a Sample weight 0.5 g; toluene concentration 500 ppm in air; total flow 7 L h-1; TOS 8 h.
4 nm (not detected by XRD) and/or no crystalline phases might be formed also. Thus, when the pretreatment temperature increases, the tungsten oxide formed from decomposition of the precursor salt is progressively reduced by carbon support, and due to the mixture of phases the catalyst system is referred to as WOx/AC. Simultaneously, the oxygenated carbon groups of the carbon support are also progressively removed (19, 22, 24, 26), as shown by XPS analyses, leading to tungsten oxide microcrystals located in a more reducing environment. We observed that toluene conversion and CO selectivity are practically independent of the active tungsten phase, but CO2 selectivity strongly depends on the oxygen content of these phases (Table 3). The formation of nonstoichiometric tungsten oxides during pretreatments at 475 °C seems to favor the toluene oxidation to CO2, comparatively to those obtained at 350 °C. This could be due to a greater ability to exchange oxygen in an oxidation-reduction mechanism or a larger oxygen mobility favored by the larger vacancies concentration. However, if WO3 is completely reduced at 950 °C to tungsten metallic and tungsten carbides, the selectivity to CO2 decreases. These facts confirm the importance of the nature and mobility of the oxygen on the catalyst surface, as was previously observed for many different oxide catalysts (18, 30-31). Therefore, for the rest of the experiments tungsten supported catalysts were always pretreated at 475 °C. The influence of tungsten content of WOx/AC catalysts on the catalytic behavior for toluene combustion is shown in Table 4. All the results were observed after attaining the steady state, i.e., after about 8 h TOS. At 300 °C toluene conversion clearly decreases with increasing tungsten loading, but SCO2 progressively increases. The decrease of toluene conversion with the metal loading can be explained by the blockage of pores provoked by tungsten oxide particles as is reported in Table 1. The increase in SCO2 with metal loading can be favored by both the progressive reduction of porosity, which complicates coke formation inside the pores and by the high activity of WOx for combustion. Thus, the fact that XCO2 increases from 8 to 13%, despite Xtoluene decreasing from 33 to 17%, indicates that WOx are active catalysts for toluene combustion. At 325 °C toluene conversion reaches a maximum for the catalyst with lower metal loading (4.8W/AC) due to the balance between adsorption-polymerization reaction and combustion, and SCO2 increases again with metal loading. At 350 °C both the toluene conversion and SCO2 clearly increase with VOL. 38, NO. 17, 2004 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
4667
FIGURE 5. Apparent toluene conversion to coke precursors against the apparent surface area of tungsten oxide catalysts supported on activated carbon: sample weight 0.5 g; toluene concentration 500 ppm in air; total flow 7 L h-1.
TABLE 5. BET Surface Area, SN2 (m2 g-1) of Fresh and Used Activated Carbon and Tungsten Oxide Supported on Activated Carbon Catalysts FIGURE 4. Toluene conversion (a) and selectivity for CO2 (b) on catalysts WOx/AC and support AC with TOS at 350 °C: sample weight 0.5 g; toluene concentration 500 ppm in air; total flow 7 L h-1. metal loading, in such a way that polymerization reactions are avoided for high metal loaded catalysts. The catalytic behavior observed at 350 °C for catalysts WOx/AC and support AC are compared in Figure 4 as a function of the time on stream. At initial time all the catalysts show similar conversions with values close to 100% (Figure 4a), but after some time of reaction, the conversions diverge and at 400-500 min TOS they are practically stable. Under these steady-state conditions, tungsten catalysts are more active than the carbon support, and the conversion increases with the tungsten content of the catalyst. Figure 4b shows that the evolution of CO2 selectivity with TOS is different from the conversion one, since for all catalysts it slightly increases with TOS and the differences between the catalysts are relatively constant throughout the reaction time. At this temperature the most selective catalyst for oxidation to CO2 is also the richest in tungsten. The formation of CO also takes place with increasing temperature and slightly increases with metal content (see Table 4). This is an unfavorable aspect because CO is also a pollutant. As was previously shown, incomplete combustion is also favored on the support for low contact time. Unfortunately, it is difficult to determine accurately the amount of coke deposited in each case. The amount of coke was estimated by directly weighing the reactor before and after reaction, and it was always lower than 6% (wt. %), becoming negligible for some experimental conditions. Due to the small amount of coke and the combustible character of the support, no significant differences were observed between the TG profiles obtained for used and fresh catalysts, because the coke burning is masked by the simultaneous carbon support combustion. Attempting to correlate the tendency of the catalyst to produce coke precursors with catalyst characteristics, Figure 5 depicts the apparent toluene conversion into coke, estimated as Xtoluene - XCO2 - XCO, versus the specific surface area (SBET) of each catalyst. The correlation observed between these parameters is linear for both reaction temperatures, 4668
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 38, NO. 17, 2004
sample
fresh
useda
% reduction
AC 4.8W/AC 14.9W/AC 23.1W/AC
929 ( 7 834 ( 6 680 ( 3 576 ( 5
600 ( 3 504 ( 5 250 ( 1 214 ( 2
35 50 63 63
a After the toluene catalytic combustion at 300 °C. Total flow 7 L h-1. Toluene concentration 500 ppm in air.
confirming that the formation of coke precursor inside the pores of these carbon based catalysts is dependent on the available surface area and decreases with reaction temperature. These effects of coke deposition are also evident from the analysis of the changes on the textural properties reported in Table 5 for fresh and used catalysts. All the catalysts used show reduced surface areas comparatively to the fresh ones. Assuming that support gasification is negligible during reaction, we might conclude that coke deposition provokes blockage of pores and decrease of surface area of catalysts. However, the richest tungsten catalysts showed the strongest reductions (loss of about 60%). In this case, the metal particles in fresh catalysts are blocking pores and therefore modifying the pore-size distribution. Thus, it is possible that a small amount of coke located at the pore entrance could produce strong surface area reduction but does not wrap the metal particles because they are more selective to oxidation and only slightly deactivated. The other factor that can influence the catalyst performance is acidity. The toluene conversion to CO2 is also linearly related with the catalyst acidity (Figure 6). Thus, changes in porous texture, acidity, and oxidation activity are linearly related with metal content, showing the uniform distribution of metal particles and a similar nature independently of the metal loading, which leads to the combustion selectivity increase. Therefore, tungsten oxides are demonstrated to be active and selective for toluene combustion. The catalyst activity depends on the intermediate tungsten oxide generated during pretreatment, where the oxygen vacancies concentration must be optimized at intermediate values. Toluene is eliminated from the stream across two kinds of reactions: polymerization, leading to coke precursor, or combustion, leading to carbon oxides. No intermediate organic products were detected. Total combustion is favored at increasing reaction temperature and metal loading. A linear relationship
FIGURE 6. Variation of toluene conversion to CO2 at different temperatures with the total surface acidity of tungsten catalysts supported on activated carbon: sample weight 0.5 g; toluene concentration 500 ppm in air; total flow 7 L h-1.
TABLE 6. Surface Characteristics of Tungsten Catalysts Supported on Different Materials after Heat Treatments in Inert Atmosphere at 475 °C for 4 h catalyst
SN2 (m2 g-1)
NH3 desorbed (µmol g-1)
4.8W/AC 5W/SiO2 5W/TiO2 5W/Al2O3 5W/USHY
834 250 9 120 720
41 99 38 294 650
between the catalyst acidity and total toluene combustion, as well as between catalyst surface area and coke precursors production, was found. 3.4. Influence of the Metal Supports on the Activity of Tungsten Oxides. To test the influence of tungsten oxide supports, 5 wt % W was also deposited on typical inorganic supports such as silica, alumina, titania, and USHY zeolite. Some textural and chemical characteristics of these catalysts after thermal treatment in N2 at 475 °C for 4 h are summarized in Table 6. 4.8W/AC catalyst presents the highest value of specific surface area, and the catalyst acidity changes according to the sequence: 5W/USHY . 5W/Al2O3 . 5W/ SiO2 > 4.8W/AC ≈ 5W/TiO2. With respect to the metal phase, only XRD peaks corresponding to WO3 were observed when silica was used as support, indicating that it is the less dispersed catalyst (22). The catalytic behavior of the different samples is shown in Figure 7 by the corresponding temperature-light-off curves. At low reaction temperatures, tungsten oxide is more active when it is supported on AC than in other inorganic supports, despite the greater acidity of the latter supports and high porosity of the zeolite. However, under these operating conditions (500 ppm toluene, GHSV 7000 h-1) this carbon-based catalyst with 5 wt % of W is unable to reach the total toluene conversion because temperature cannot be increased due the support gasification. The total toluene oxidation, XCO2 + XCO ) 100, was observed only with WO3/USHY catalyst, but about 450 °C, where WOx/ AC would be burned. In these experimental conditions, WO3/ USHY is highly selective to the complete toluene oxidation to CO2 (Figure 7a). The other catalysts, WO3/Al2O3, WO3/ TiO2, and especially WO3/SiO2, present very low conversion to CO2. The analysis of the product distribution also shows that at 500 °C, WO3/Al2O3 reaches total toluene conversion, but with very high CO production (Figure 7b). It could seem that the higher activity of WO3/AC at low temperature should be due to the promotion of adsorption/
FIGURE 7. Variation of toluene conversion to CO2 and CO against the reaction temperature of tungsten catalysts supported on different supports: sample weight 0.5 g; toluene concentration 500 ppm in air; total flow 7 L h-1.
FIGURE 8. Apparent toluene conversion to coke precursors against TOS for tungsten catalysts supported on different supports at 350 °C: sample weight 0.5 g; toluene concentration 500 ppm in air; total flow 7 L h-1. polymerization processes, i.e., deposition of coke precursors favored by its higher surface area. However, the comparison of the apparent toluene conversion to coke at 350 °C, estimated from the parameter Xtoluene - XCO2 - XCO as a function of TOS for the different supported catalysts (Figure 8), shows that the stronger coke deposition takes place on the WO3/USHY catalyst. This deposition occurs mainly at the reaction beginning, which is due to the high porosity and strong acidity of the zeolite. In all cases, the deposition of coke precursor decreases with TOS until reaching a steady state. Therefore, at low reaction temperature, WOx/AC catalyst is the most active, probably due to the hydrophobic character of AC in comparison with other supports. At high reaction temperatures water chemisorption does not occur, but at lower temperatures the water chemisorption can provoke VOL. 38, NO. 17, 2004 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
4669
blockage of the active sites. The influence of the hydrophobicity on the activity of catalysts at low reaction temperatures has been previously described for both carbonaceous (3, 10, 11) and inorganic supports (32). P. De´ge´ et al. (32), using Pd-Pt/HFAU catalysts for o-xylene oxidation, showed that the increase of the hydrophobicity of dealuminated zeolites plays a positive role in the oxidation activity in the presence of steam: the o-xylene oxidation is faster for zeolites with higher Si/Al ratios. Therefore, at low reaction temperature WOx is more active when supported on AC. In these experimental conditions catalysts are stable, and the AC gasification is avoided, as well the formation of NOx. Other organic byproducts (dioxins) were also not observed. However, to attain total conversion large metal loading or contact times are needed. The low price of AC and tungsten oxides, compared to other supports and noble metals, lead to cheap catalysts. Moreover, the ability of this system to work at lower temperatures (saving energy) and the possibility of easily recovering the metal phase after deactivation by burning the support, may also economically favor the use of these catalysts.
Literature Cited (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12)
Lee, C. H.; Chen, Y. W. Ind. Eng. Chem. Res. 1997, 36, 5160. Guisnet, M.; De´ge´, P.; Magnoux, P. Appl. Catal. B 1999, 20, 1. Wu, J. C. S.; Chang, T. Y. Catal. Today 1998, 44, 111. Wu, J. C. S.; Lin, Z. A.; Tsai, F. M.; Pan, J. W. Catal. Today 2000, 63, 419. Tsou, J.; Magnoux, P.; Guisnet, M.; Orfa˜o, J. M.; Figueiredo, J. L. Catal. Commun. 2003, 4, 69. Chu, H.; Lee, W. T.; Horng, K. H.; Tseng, T. K. J. Hazard. Mater. 2001, B82, 43. Spivey, J. J.; Butt, J. B. Catal. Today 1992, 11, 465. Taylor, S. H.; Heneghan, C. S.; Hutchings, G. J.; Hudson, I. D. Catal. Today 2000, 59, 249 Chuang, K. T.; Cheng, S.; Tong, S. Ind. Eng. Chem. Res. 1994, 33, 1680. Zhang, M.; Zhou, B.; Chuang, K. T. Appl. Catal. B 1997, 13, 123. Ishikawa, A.; Komai, S.; Satsuma, A.; Hattori, T.; Murakami, Y. Appl. Catal. A 1994, 110, 61. Maldonado-Ho´dar, F. J.; Madeira, L. M.; Portela, M. F.; Martı´nAranda, R. M.; Freire, F. J. Mol. Catal. A 1996, 111, 313.
4670
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 38, NO. 17, 2004
(13) Cordi, E. M.; O’Neill, P. J.; Falconer, J. L. Appl. Catal. B 1997, 14, 236. (14) Parida, K. M.; Amarendra, S. Appl. Catal. A 1999, 182, 249. (15) Papaefthimiou, P.; Ioannides, T.; Verykios, X. Appl. Catal. B 1997, 13, 175. (16) Prasad, R.; Kennedy, L. A.; Ruckenstein, E. Catal. Rev. Sci. Eng. 1984, 26 (1), 1. (17) Tseng, T. K.; Chu, H. Sci. Total Environ. 2001, 275, 83. (18) Labaki, M.; Sifftert, S.; Lamonier, J. F.; Zhilinskaya, E. A.; Aboukais, A. Appl. Catal. B 2003, 43 (3), 261. (19) Alvarez-Merino, M. A.; Carrasco-Marı´n, F.; Fierro, J. L. G.; Moreno-Castilla, C. J. Catal. 2000, 192, 363. (20) Pe´rez-Cadenas, A. F.; Moreno-Castilla, C.; Maldonado-Ho´dar, F. J.; Fierro, J. L. G. J. Catal. 2003, 217, 30. (21) Moreno-Castilla, C.; Pe´rez-Cadenas, A. F.; Maldonado-Ho´dar, F. J.; Carrasco-Marı´n, F. Carbon 2003, 41, 863. (22) Alvarez-Merino, M. A.; Carrasco-Marı´n, F.; Moreno-Castilla, C. J. Catal. 2000, 192, 374. (23) Moreno-Castilla, C.; Alvarez-Merino, M. A.; Carrasco-Marı´n, F. React. Kinet. Catal. Lett. 2000, 71, 136. (24) Moreno-Castilla, C.; Alvarez-Merino, M. A.; Carrasco-Marı´n, F.; Fierro, J. L. G. Langmuir 2001, 17, 1752. (25) Ai, M. Bull. Chem. Soc. Jpn. 1976, 49, 1328. (26) Moreno-Castilla, C.; Carrasco-Marı´n, F.; Maldonado-Ho´dar, F. J.; Rivera-Utrilla, J. Carbon 1998, 36 (1-2), 145. (27) Atkinson, R. Atmos. Environ. 2000, 34, 2063. (28) Radovic, L. R.; Rodrı´guez-Reinoso, F. In Chemistry and Physics of Carbon, Vol 25; Thrower, P. A., Walker, P. L., Eds.; Dekker: New York, 1997; p 243. (29) Antunes, A. P.; Ribeiro, M. F.; Silva, J. M.; Ribeiro, F. R.; Magnoux, P.; Guisnet, M. Appl. Catal. B 2001, 33, 149. (30) Sciere, S.; Minico, S.; Crisafulli, C.; Satriano, C.; Pistone, A. Appl. Catal. B 2003, 40, 43. (31) Rougier, A.; Soiron, S.; Haihal, I.; Aymard, L.; Taouk, B.; Tarascon, J. M. Powder Technol. 2002, 128, 139. (32) Dege´, P.; Pinard, L.; Magnoux, P.; Guisnet, M. Chemistry 2001, 4, 41.
Received for review September 3, 2003. Revised manuscript received June 16, 2004. Accepted June 16, 2004. ES034964C
