Catalytic Evaluation Of Dioxins

Applied Catalysis B: Environmental 74 (2007) 223–232 www.elsevier.com/locate/apcatb

On the impact of the choice of model VOC in the evaluation of V-based catalysts for the total oxidation of dioxins: Furan vs. chlorobenzene D.P. Debecker, F. Bertinchamps, N. Blangenois, P. Eloy, E.M. Gaigneaux * Universite´ catholique de Louvain, Unite´ de catalyse et chimie des mate´riaux divise´s, Croix du Sud 2/17, B-1348 Louvain-la-Neuve, Belgium Received 30 November 2006; received in revised form 21 February 2007; accepted 22 February 2007 Available online 25 February 2007

Abstract V-based catalysts, widely developed for the catalytic abatement of dioxins, are usually studied and optimized by investigating the oxidation of model chlorinated aromatic compounds (e.g. chlorobenzene). Even though the oxygenated function included in the central aromatic ring of the molecular structure of a dioxin could influence major aspects of the catalytic process, it has never been taken into account in the reported works. In this study, furan is chosen as a model for the central oxygenated ring of a polychlorinated dibenzo furan (PCDF) and its oxidation is compared to the case of chlorobenzene. The strategy was to check systematically if the improvements of formulations enlightened from our previous investigation on chlorobenzene also remain beneficial with furan. It turned out that the use of a sulfate containing TiO2 as support for the active VOx phase as well as the doping of the formulation with Mo or W oxides had very different impacts in the two cases. Some improvement strategies prove to be inefficient or deleterious in the case of furan. Competition tests further suggest that the adsorption behavior of dioxin could be better imitated by furan than by chlorobenzene. These observations highlight, in the case for which working with the target pollutant is difficult (as with dioxins), that the choice of the model molecule is critical. # 2007 Elsevier B.V. All rights reserved. Keywords: Dioxin; Chlorobenzene; Furan; Catalytic oxidation; VOC combustion; VOx/TiO2

1. Introduction In the field of atmospheric pollution, it is evident that polychlorinated hydrocarbons, and in particular dioxins, constitute a major concern. The word ‘‘dioxin’’ collectively refers to polychlorinated dibenzo furans (PCDF) and polychlorinated dibenzo dioxins (PCDD). Dioxins are very stable structures consisting in three aromatic cycles. The central ring includes one (PCDF) or two (PCDD) oxygen atoms and the external cycles are substituted with chlorine atoms at various positions (Fig. 1). These compounds are formed during incineration and combustion processes [1]. As a response to their high persistence and their subsequent toxicity, catalysis is the most promising solution since it allows to directly destroy these pollutants at the source of emission [1–3]. Moreover, the catalytic total oxidation can exhibit an excellent selectivity

* Corresponding author. Tel.: +3210473665; fax: +3210473649. E-mail address: [email protected] (E.M. Gaigneaux). 0926-3373/$ – see front matter # 2007 Elsevier B.V. All rights reserved. doi:10.1016/j.apcatb.2007.02.016

towards harmless products and can operate at relatively low destruction temperature [2,4–6]. For this application, two families of catalysts have been developed. Noble metal-based catalysts are highly active [7– 10] but suffer from high cost and low stability toward the produced HCl and Cl2 [2,11–13]. Moreover, they also catalyze the further poly-chlorination of the pollutants [14]. Transition metal-based catalysts are the alternative. In particular, vanadiabased catalysts are recognized as active and stable in a chlorinecontaining environment [2,15,16]. Indeed, Bertinchamps et al. showed that their activity in benzene oxidation was the same as in chlorobenzene oxidation [17]. So, V-based formulations constitute a promising solution for dioxin total oxidation and efforts are made to improve their formulation. A large number of fundamental works have been done in order to understand the oxidation mechanism that is involved during total oxidation of dioxins on V-based catalysts. However, very few authors have reported works realized directly on real dioxins [18–20]. Real dioxins are usually not used in mechanistic studies because they are hard to handle and of course very toxic. Model volatile organic compounds (VOC) are thus often used. The chosen

224

D.P. Debecker et al. / Applied Catalysis B: Environmental 74 (2007) 223–232

Fig. 1. Molecular structures of PCDD, PCDF, chlorobenzene and furan.

models need to be structurally similar to dioxins, less toxic and easier to handle. Even if they always differ in some ways from the target compound, model compounds prove to be essential in order to test and optimize the proposed formulations. Classically the authors want to evaluate the activity of the catalyst in the oxidation of an aromatic ring, as well as its resistance to chlorine poisoning. Therefore, simple chlorinated aromatic compounds (e.g. chlorobenzene) are very often chosen [1,11,12,15,17,21–25]. As interesting lines to optimization of the catalysts for such model VOCs, it has been found that titania supports are able to spread the vanadia very well on their surface, which leads to high activity [22,24]. Sulfate containing TiO2 are also claimed to be even more efficient [15,17,24]. Additionally, several authors have shown that other transition metal oxides (essentially MoOx, WOx) can act as dope for the active phase of VOx [11,17,21,23,26–29]. Molecules like chlorobenzene are useful models for dioxins. However, one should bear in mind that the central aromatic ring of a dioxin molecule is oxygenated. In a PCDF this ring is a furan (Fig. 1). This essential characteristic is not taken into account when a simple chlorinated VOC like chlorobenzene is chosen as a model. Nevertheless this moiety can influence major aspects of the catalytic process (adsorption of the pollutant, coking, etc.) as it is suggested from the DRIFTS study of Larrubia and Busca [30] on the adsorption of various compounds on a VOx–WOx/TiO2 catalyst. Taralunga [31] also intended to take this oxygenated moiety of dioxins into account when investigating the oxidation of benzofuran in addition to the oxidation of dichlorobenzene over zeolite and noble metalbased catalysts. To our knowledge though, no other reported study did investigate the importance of this central ring in the catalytic abatement of dioxins on V-based catalysts. Improvements of the catalyst formulations are proposed but they are never checked in the perspective of an oxygenated model. We thus, propose a comparative study of the oxidation of chlorobenzene – as a chlorinated aromatic model molecule – and the oxidation of furan—as an oxygenated aromatic model molecule. It should be noted that the issue of the total oxidation of oxygenated VOC is already well documented in the literature. Studies on ketones [32–38], alcohols [38–41] or esters [42–44] total oxidation show that O-compounds are quite easily destroyed and that the behavior of the VOC differ with their

structure. In this paper, the total oxidation of an aromatic Ocompound is addressed. Even though it can be expected that our catalyst would destroy furan easily, this remained to be checked. Also, the basic oxygen of furan is likely to play an important role in the adsorption of the pollutant on the acid VOx/TiO2 based catalyst. Since the adsorption of this polar group (similar to those present in PCDD and PCDF) can influence major aspects of the catalytic process it is important to have a better insight on this matter. Besides, the main goal of this comparative study is to check the beneficial effect brought by ‘‘improved’’ formulations of Vbased catalysts in the case of the oxidation of furan. In particular, two parameters of the formulation are studied: (i) the use of a sulfate containing TiO2 support instead of the classical TiO2 support and (ii) the use of Mo and W oxides as doping phases of the active phase of vanadium. Therefore, the catalytic performances of V-based catalysts supported on two different supports and possibly upgraded with MoOx or WOx are investigated in the oxidation of chlorobenzene and furan. Also, in addition to the one-pollutant tests, the respective behaviors of the two models are investigated in the course of a competition test, namely a test involving both pollutant models simultaneously. The observed performances are correlated with the results of various characterizations. FTIR analysis with adsorbed pyridine allowed us to study the acidity of the catalysts. As suggested by Larrubia and Busca [30], the acidity of the catalyst should indeed play a crucial role in the adsorption of the VOCs on the surface of the catalyst. The tendency of the various impregnated oxides to spread on the supports is considered thanks to XPS and XRD analysis. The weight loadings (ICP-AES) and the surface area (N2 adsorption) are also measured. 2. Experimental 2.1. Preparation of the catalysts Catalysts are supported on two different TiO2 supports: a 70% anatase-30% rutile TiO2 (Degussa P25: 48 m2/g, denoted hereafter T) and a pure anatase TiO2 containing 1.4% wt of sulfate (Millenium PC 100: 91 m2/g, denoted hereafter Ts). The active phase is VOx (V) and the secondary phases are MoOx (Mo) and WOx (W). The precursors of these oxides are respectively NH4VO3 (Vel, 99.9%); (NH4)6Mo7O244H2O (Aldrich, >99%); and (NH4)2WO4 (Aldrich, 99.99%). To obtain the impregnation solutions, the precursors are dissolved in distilled water and complexed with 2 moles of oxalic acid for 1 mole of transition metal. All impregnation solutions had a pH below 3. A classical wet impregnation method was used and the amounts of precursor were calculated in order to obtain 0.75 theoretical monolayer of transition metal oxide on the surface of each support. The theoretical monolayer coverage was calculated based on the cross sectional area of a unit composed of one transition metal atom and its oxygen coordination ˚ 2 for VOx, 17 A ˚ 2 for MoOx and sphere. The values are 12 A 2 ˚ for WOx [24]. The suspension of support in the 15 A

D.P. Debecker et al. / Applied Catalysis B: Environmental 74 (2007) 223–232 Table 1 Abbreviation and compositions of the catalysts Name

Support

Active phase (0.75 mL)

Doping phase (0.75 mL)

TV TVMo TVW TsV TsVMo TsVW TMo TW

TiO2 TiO2 TiO2 Sulfated containing TiO2 Sulfated containing TiO2 Sulfated containing TiO2 TiO2 TiO2

VOx VOx VOx VOx VOx VOx / /

/ MoOx WOx / MoOx WOx MoOx WOx

225

TCD and two FID) was used with He as carrier gas in order to quantify chlorobenzene, furan, O2, CO, CO2 and to detect other hydrocarbons. The analysis parameters were set as to allow one analysis each 15 min and to obtain measured performances accurate within a range of about 1% (in relative) for the conversion of chlorobenzene and furan. To calculate the conversion at a given temperature, only the concentrations of reactants measured and averaged in the period of time from 100 to 150 min after stabilization were taken into account. The conversion is defined as the ratio reactant transformed/reactant in the inlet. 2.3. Characterization

impregnation solution was stirred for 2 h at room temperature before the solvent was evaporated under reduced pressure in a rotavapor at 45 8C. The obtained solids were dried overnight in an oven at 110 8C and then calcined at 400 8C at atmospheric pressure in air for 20 h in a muffle furnace. Six different V-based catalysts have thus been prepared. Their compositions are described in Table 1. The reference catalyst contains 0.75 monolayer (ML) of the active phase wetimpregnated on TiO2 (TV). Doped formulations were obtained by simultaneous impregnation of vanadium oxide with molybdenum or tungsten oxides on TiO2 in order to obtain 0.75 theoretical ML of each deposited oxides (TVMo and TVW). The same preparation protocol has been applied to the sulfate containing support, leading to the three formulations denoted TsV, TsVMo and TsVW. Also, the two doping oxides have been impregnated alone on T (TMo and TW). 2.2. Catalytic tests All the catalytic tests were performed in a metallic fixed-bed micro-reactor (PID ENG & Tech, Spain, Madrid) operating at atmospheric pressure. The reactor was made of an inconel tube of 1 cm internal diameter. The catalytic bed was composed of 200 mg of catalyst powder selected within the granulometric fraction 200–315 mm and diluted in 800 mg of inactive glass spheres with diameters in the range 315–500 mm. The gas stream contained 100 ppm of chlorobenzene or/and 150 ppm of furan in He (Praxair), 20% of O2 (Praxair; 99.995%) and He (Praxair; 99.996%) as diluting gas to obtain 200 ml/min (VVH = 37000 h1). The reaction was run from 100 to 400 8C in a step mode. At each temperature investigated, the catalyst was stabilized for 150 min. Two start-up procedures have been envisaged. In the ‘‘direct’’ procedure, the test is readily started from 100 8C and proceeds to the next temperature steps. In the ‘‘sat’’ procedure, the catalyst is first contacted with a concentrated flow of the VOC (750 ppm) for 4 h at 100 8C. After this period, the composition of the flow is stabilized at the usual concentration and the usual temperature-programmed test proceeds. Analysis of reactants and products was continuously performed by on-line gas chromatography (GC). The CP3800 gas chromatography apparatus from Varian equipped with four columns (one Hayesep G, one Hayesep T, one Molsieve and one CP-Sil 8CB), a methaniser and three detectors (one

X-ray diffraction (XRD) measurements were performed on the fresh catalysts with a Siemens D5000 difractometer using ˚ ). The 2u range was the Ka radiation of Cu (l = 1.5418 A recorded between 5 and 758 at a rate of 0.028 s1. The ICDDJCPDS database was used to identify the crystalline phases. X-ray photoelectron spectroscopy (XPS) was performed on a SSI X-probe (SSX-100/206) spectrometer from Surface Science Instruments working with a monochromatic Al Ka (1486.6 eV) radiation (10 kV, 22 mA). The analysis chamber was operated under ultrahigh vacuum with a pressure close to 5  109 Torr. Charge compensation was achieved by using an electron flood gun adjusted at 8 eV and placing a nickel grid 3.0 mm above the sample. Pass energy for the analyzer was 150 eV and the spot size was approximately 1.4 mm2. For these measurements, the binding energy (BE) values were referred to the C-(C, H) contribution of the C 1s peak at 284.8 eV. The surface atomic concentrations were calculated by correcting the intensities with theoretical sensitivity factors based on Scofield cross sections [45]. Peak decomposition was performed using curves with 85% Gaussian type and 15% Lorentzian type, and a Shirley non-linear sigmoid-type baseline. The following peaks were used for the quantitative analysis: O 1s, C 1s, V 2p, Ti 2p, Cl 2p, Mo 3d and W 4d. Based on the XPS analysis, we estimated the surface ratio impregnated oxide/support by the formula: XPS ratio ¼

atomic concentration of impregnated metal ð%Þ atomic concentration of Ti ð%Þ

Fourier Transformed Infra Red spectra of pre-adsorbed pyridine (FTIR-pyridine) were recorded using an IFS55 Equinox spectrometer (Bru¨cker) equipped with a MCT detector and working with a resolution of 4 cm1. The amount of Brønsted and Lewis sites is determined by the integration of the area of the peaks at 1537 and 1446 cm1, corresponding respectively to pyridine adsorbed on Brønsted sites and Lewis sites. The details of this experiment are given elsewhere [17]. The estimation of the weight percentages of V, Mo, W and Ti was performed through inductively coupled plasma-atomic emission spectroscopy (ICP-AES) on an Iris Advantage apparatus from Jarrell Ash Corporation. N2 adsorption measurements have been performed at 196 8C, under a N2-partial pressure P/P0 = 0.3 on a

226

D.P. Debecker et al. / Applied Catalysis B: Environmental 74 (2007) 223–232

Micromeritics ASAP 2000 in order to determine the BET surface area (SSA) of our catalysts and supports. Prior to the measurement, the samples were degassed under N2/He (30:70% vol.) at 120 8C for 1 h. 3. Results and discussion 3.1. Data on the catalytic tests 3.1.1. Effect of the start-up procedure In the case of chlorobenzene, the start-up procedure did not affect the performances measured afterwards. Fig. 2 shows the GC results obtained during a test on chlorobenzene with the ‘‘direct’’ procedure. At each step of temperature a short transition period leads to a stable plateau of observed conversion. It is thus easy to evaluate an average conversion for each investigated temperature. For furan on the other hand, the two start-up procedures lead to very different results. When the test is started with the ‘‘direct’’ procedure (Fig. 3), the conversion is almost complete at the very beginning of the 100 8C-step and then it decreases continuously. This decrease of the conversion is observed during the whole temperature step. After 150 min, the temperature is raised to the next step. As a consequence, the conversion reaches a higher conversion and again decreases progressively. Complementary tests have shown that this instability lasts for more than 15 h. A similar long-lasting transition period has also been reported in the oxidation of butanol on noble metal based catalysts [3]. Thanks to the ‘‘sat’’ procedure, we were able to get rid of this instability. The catalyst that has first been contacted with a concentrated flow of furan for 4 h at 100 8C shows stable values of conversion for each temperature (Fig. 4). These observations reveal the occurrence of a strong adsorption of furan on our catalysts. This is in agreement with Spoto et al. [46] who already showed that furan interacts very strongly with the acidic sites of a zeolite. In the ‘‘direct’’ procedure, our hypothesis is thus that furan is removed from the flux both through real oxidation and through simple adsorption. The surface is progressively covered with a carbonaceous deposit. The extent of the elimination through adsorption is thus

Fig. 2. Evolution of the down-stream chlorobenzene concentration during a catalytic test (‘‘direct’’ start-up procedure) on TsVMo.

Fig. 3. Evolution of the down-stream furan concentration during a catalytic test (‘‘direct’’ start-up procedure) on TsVMo.

predominant at the beginning of the temperature plateau and then decreases as the coverage of the surface is progressing. This leads to a long-lasting transition period during which no stable conversion can be measured. The effect of the ‘‘sat’’ start-up is to saturate the surface of the catalyst with furan. The saturated catalysts acquire a pronounced brown-red color. This was also observed on zeolites after adsorption of furan by Spoto et al. [46]. The saturated catalysts exhibit a stable conversion depending only on the applied temperature. It is noteworthy that Paulis et al. [37] used a similar start-up procedure in order to cope with the discrepancies observed in catalytic results and caused by the strong adsorption of acetone on manganese oxide-based catalysts. The nature of the observed deposit has not been investigated extensively. Yet, an enrichment of the atomic surface concentration of carbon has been detected by XPS measurements on saturated catalysts. The proportion of C linked to oxygen atoms is also higher in saturated samples than in fresh catalysts. This can originate from the presence of various Ocontaining species produced from the reaction of furan on the catalyst. The occurrence of polymerization of furan or its derivatives on the catalyst surface would not be surprising as Spoto et al. [46] evidenced the high reactivity of furan in reactions.

Fig. 4. Evolution of the down-stream furan concentration during a catalytic test (‘‘sat’’ start-up procedure) on TsVMo.

D.P. Debecker et al. / Applied Catalysis B: Environmental 74 (2007) 223–232

3.1.2. Selectivity and carbon mass balance In the case of chlorobenzene oxidation, CO2 is the main oxidation product. The CO selectivity is comprised between 5 and 10%. Traces of maleic anhydride are also detected, which is in agreement with the observations of Larrubia and Busca [30]. As the tests on furan are run in the ‘‘sat’’ procedure, the catalyst surface is saturated with a carbonaceous deposit during the test start-up. When the temperature is raised during the test, this deposit is progressively burned. The outlet product concentrations are thus the sum of (i) the products directly formed from the oxidation of the entering furan and (ii) the products formed or desorbed from the carbonaceous deposit. Moreover, the outlet product concentrations are often unstable, even at the end of the 150 min temperature plateau. It is thus impossible to state a rigorous carbon mass balance in the tests run on furan. Nevertheless, CO2 is the main oxidation product. The estimation of the CO selectivity can vary between 5 and 30%. In addition, the outlet maleic anhydride concentration can reach a few ppm. This partial oxidation activity is not surprising since (i) it is known that VOx/TiO2 catalysts can produce maleic anhydride during the oxidation of various VOC (benzene, chlorobenzene, benzofuran) [30] and (ii) furan is a known intermediate in the production of maleic anhydride from butene on V2O5 containing catalysts [47–49]. 3.2. Evaluation of the catalyst formulations 3.2.1. Catalytic conversion of chlorobenzene Fig. 5 presents the light-off curves of the six catalysts investigated in the course of total combustion of chlorobenzene (‘‘direct’’ start-up procedure). The abatement of chlorobenzene starts significantly at 150 8C. At 350 8C all the formulations convert 100% of the chlorobenzene concentration. The interpolated temperature needed to convert 50% of the chlorobenzene concentration (T50) is comprised between 185 and 230 8C. The activity of the entire set of catalysts can be classified based on the T50 as: TV (230 C8) < TVMo (220 8C) < TVW (215 8C) < TsV (195 8C) < TsVW = TsVMo (185 8C). The three Tssupported formulations are more active than the corresponding T-supported ones. The T50 decreases by about 35 8C for each formulation when Ts is chosen instead of the classical TiO2. For both supports, the formulations that contain MoOx or WOx are

Fig. 5. Chlorobenzene light-off curves (‘‘direct’’ procedure).

227

Fig. 6. Furan light-off curves (‘‘sat’’ procedure).

more active than the formulation containing only VOx. The coimpregnation of the doping phases brings a beneficial effect by decreasing the T50 by about 10–15 8C. 3.2.2. Catalytic conversion of furan Fig. 6 presents the light-off curves of the catalysts in the course of the total combustion of furan (‘‘sat’’ start-up procedure). The abatement of furan starts already significantly from 100 8C. At 250 8C, the six catalysts investigated convert 100% of the furan concentration. The 50% conversion is reached at temperatures comprised between 135 and 180 8C. The easy oxidation of furan, as compared to chlorobenzene, was indeed expected as an oxygen-containing molecule is intuitively easier to oxidize. But the most interesting data we wish to enlighten is the fact that the T50 ranking of the catalysts is obviously different from what is observed in the oxidation of chlorobenzene: TVMo (180 C8) < TVW (170 8C) < TsVMo (165 8C) < TsVW (160 8C) < TV (150 8C) < TsV (135 8C). First, the three Ts-supported formulations are not collectively better than the three T-supported ones. Nevertheless, as regarding a given composition of impregnated oxides, the Ts-supported formulations are still better. But the positive effect is small (decrease of the T50 by about 10 8C for each formulations) as compared to the effect observed in chlorobenzene oxidation. Secondly, doped formulations are systematically less active than the corresponding formulations that were only impregnated with VOx. The negative effect is obvious as the T50 increases by about 25 8C for MoOx or WOxcontaining formulations (for a given support). These two oxides have also been tested when impregnated alone on T. Fig. 7 shows that WOx only has an effect above 250 8C. At lower temperature, TW has the same low activity as the bare support. MoOx on the other hand shows some activity, but is far from competing the performances of the six V-based catalysts as TMo has a T50 of about 210 8C only. 3.2.3. Surface acidity Bertinchamps et al. already investigated the surface acidity of identical catalysts [17]. Their main findings are crucial to the understanding of our results and can be summarized as follows. The study of the Lewis acidity revealed that (i) the sulfate

228

D.P. Debecker et al. / Applied Catalysis B: Environmental 74 (2007) 223–232 Table 2 XPS ratios

TV TVMo TVW TsV TsVMo TsVW

Fig. 7. Furan light-off curves on T, TMo and TW (‘‘sat’’ procedure).

containing support exhibits two times more Lewis sites than the classical TiO2; (ii) the Lewis sites present at the surface of the sulfate containing support are stronger than those of conventional TiO2; (iii) the impregnated oxides either do not exhibit any Lewis sites on their surface or exhibit less Lewis sites than the bare support; and (iv) the decrease of the number of Lewis sites that is observed when VOx, MoOx and WOx are impregnated is due to a coverage effect. The study of the Brønsted acidity revealed that (i) the classical support does not exhibit any Brønsted sites, but Ts exhibits some of them (ii) Brønsted sites are mainly brought by the impregnation of VOx, MoOx and WOx and (iii) the amount of Brønsted sites brought by the impregnated oxides is four times bigger in the case of Ts than of T support. This latter was discussed as due to a better spreading of the oxide phases on Ts and suggests that the presence of stronger Lewis sites on the sulfate containing TiO2 promotes this spreading [17]. 3.2.4. Other characterizations For the entire set of catalysts, vanadium and molybdenum oxides were not detected at all on XRD diffractograms. This suggests the obtention of well-dispersed oxides, and likely the formation of a homogeneous coverage of these phases as submonolayers. WO3 crystallites on the other hand were detected in the TVW and TsVW samples (JCPDS 32-1395). This is in agreement with Leyrer et al. who explained the increasing spreading of the oxides on TiO2 within the series: WO3 < MoO3 < V2O5 [50]. Also, Bertinchamps et al. detected the formation of WO3 crystallites on TiO2 at a theoretical loading as low as 0.75 monolayer [15]. XPS data were analyzed in term of atomic concentration ratios (Table 2). In this way, an increase of the ratio V/Ti can be correlated to a better spreading of the active phase [17]. The slightly higher V/Ti ratio for TsV thus confirms the better tendency of VOx to spread on sulfate containing TiO2. Table 2 also shows that the presence of a secondary phase leads to a decrease of the ratio V/Ti. This could be related to a covering effect of the secondary phase (Mo or W oxides) on the active phase (VOx). But, Bertinchamps et al. rather demonstrated that the spreading of Mo and W oxides on V oxide is almost absent [17]. Thus the lower V/Ti ratios observed for the doped

V/Ti

Mo/Ti

W/Ti

0.17 0.11 0.14 0.18 0.10 0.13

– 0.17 – – 0.17 –

– – 0.15 – – 0.13

formulations is better accounted by the fact that, the secondary phase prevents the active phase from spreading well on the support. It can be noted that the effect is less important in the case of WOx than in the case of MoOx. This is consistent with the fact that WO3 crystallites are detected by XRD on our catalysts, namely the fact the WOx is itself poorly spread on the support and thus does not impede the spreading of VOx as significantly as MoOx does. ICP-AES loadings and specific surface area (SSA) are given in Table 3. 3.2.5. Discussion on the effect of the sulfate containing support The use of Ts has a great positive impact on the performances of the catalysts in the oxidation of chlorobenzene. This effect can be attributed to the higher amount of Brønsted sites present on Ts-supported catalysts [17]. Brønsted sites seem to be implicated in the adsorption of chlorobenzene on the surface of the catalyst, which is the first step of the catalytic abatement process. This statement is in agreement with Sinquin et al. [51] and Ramachandran et al. [52] who both suggested that Brønsted sites are the adsorption sites of chlorinated VOCs. The amount of Brønsted sites is thus a crucial parameter. Their surface concentration should be high in order to provide enough adsorbed chlorobenzene for the sites that are involved in the combustion of the aromatic cycle. In this view, Ts contributes in a direct, as well as in an indirect manner. Ts bears some Brønsted sites on its surface (direct contribution). Nevertheless, the activation effect is mainly indirect, as Ts – exhibiting a higher amount and strength of Lewis sites – favors the spreading of the impregnated oxides. These well spread oxides can in turn expose more Brønsted sites at the surface of the catalyst, which leads to an enhancement of the performances. In the oxidation of furan, the use of the sulfate containing support brings only a small beneficial effect. One should bear in mind that Ts possesses a larger specific surface area than T (Table 3). Also, we always impregnated 0.75 theoretical ML on our catalysts and the catalytic tests are always run with the same amount of catalyst (200 mg). As a result, when the reactor is filled with a Ts-supported catalyst, it contains more active (and doping) phase than when it is filled with a T-supported catalyst. On one hand, this fact is not sufficient to explain the extent to which Ts improves the performances in the oxidation of chlorobenzene. On the other hand, this simple parameter could account for the limited improvement observed in the oxidation of furan. As a consequence, we should safely limit our

D.P. Debecker et al. / Applied Catalysis B: Environmental 74 (2007) 223–232

229

Table 3 ICP-AES weight concentrations and ICP-AES ratios

T Ts TV TVMo TVW TsV TsVMo TsVW TMo TW

SSA (m2 g1)

V (%)

Mo (%)

W (%)

V/X

V2O5 (%)

MoO3 (%)

WO3 (%)

49.3 82.4 45.7 44.9 43.1 68.8 66.6 56.7

– – 2.55 2.11 1.96 3.93 3.20 3.48 – –

– – – 2.82 – – 4.75 – 3.05 –

– – – – 5.95 – – 10.91 – 6.55

– – – 0.75 0.32 – 0.67 0.32 – –

– – 4.55 3.76 3.49 7.01 5.70 6.20 – –

– – – 4.25 – – 7.16 – 4.60 –

– – – – 7.50 – – 13.75 –

interpretation to the fact that Ts has no chemical effect on the oxidation of furan. This is fully consistent with the observed strong tendency of furan to adsorb on the surface of all of our catalysts. Moreover we had to saturate the surface of our catalyst in the start-up of the catalytic test in order to measure the real oxidation activity. Consequently, the first step of the catalytic abatement process, namely the adsorption, is certainly not a limiting step in the case of furan. By increasing the Brønsted acidity of surface, Ts does not promote the adsorption and does thus not have the possibility to improve the entire catalytic process.

Though Fig. 7 shows that WOx is inactive, while MoOx is moderately active, it is noteworthy that MoOx-doped catalysts are even worse than WOx-doped ones. Nevertheless, XRD and XPS results suggest that MoOx spreads more easily on TiO2 supports that WO3. As a consequence, the spreading of the VOx phase is more affected in the case of co-impregnation of Mo oxide. The higher activity of TVW and TsVW as compared to TVMo and TsVMo, respectively suggests that the spreading of the VOx phase is a crucial parameter. Moreover, it confirms the fact that the sites that are active in the oxidation of furan are located on the VOx phase.

3.2.6. Discussion on the effect of the doping phase The co-impregnation of MoOx or WOx brings an obvious doping effect in the oxidation of chlorobenzene. We earlier showed that the doped formations present more acid Brønsted sites than the catalysts containing only VOx [17]. These sites are supposed to be involved in the adsorption step of chlorobenzene [17,51,52]. The improved performances of doped catalysts can therefore be correlated with the higher amount of Brønsted sites present on their surface. In the oxidation of furan, however, the effect of the coimpregnation of MoOx and WOx is opposite. Three parameters have to be envisaged when trying to understand the effect of the co-impregnation of secondary phases along with the active phase. First, the presence of a secondary oxide prevents VOx from spreading well on the support. As a consequence, the active phase could be less efficient. Secondly, MoOx and WOx are not listed as good total oxidation catalysts in the literature [1,9,11,12,22,43,44,53]. In a previous paper we showed that these oxides are inactive in the oxidation of benzene [24]. Fig. 7 shows that TW is inefficient in furan oxidation. TMo, though exhibiting some activity, is far from achieving the performances of the V-based catalysts. Thirdly, as mentioned above, doped catalysts exhibit a larger amount of Brønsted acidic sites on their surface. This last parameter is crucial for the promotion of the adsorption of chlorobenzene and seems to overcompensate the two other parameters in the precise case of chlorobenzene. However, the increase in the surface Brønsted acidity has no influence in the oxidation of furan, since this VOC easily adsorbs on all of our catalysts. As a consequence, the negative effects brought by the two other parameters are not compensated anymore. So, the co-impregnation of MoOx and WOx is deleterious for the catalytic oxidation of furan.

3.3. Chlorobenzene versus furan in simultaneous reaction 3.3.1. Competition test The adsorption behavior of furan is very different from what is observed in the case of chlorobenzene. This has been showed to have a dramatic effect on the entire catalytic process. An important remaining question is to know what will remain from these described behaviors if furan and chlorobenzene are reacted together. It is known that the behaviors of the pollutants as well as the performances of the catalyst in their abatement are usually far from additive when mixtures are used. It is thus useful to perform competition tests between the two VOC in addition to the classical one-pollutant test. Fig. 8 presents the light-off curves obtained in the course of catalytic tests run on chlorobenzene and furan, alone and in competition. The tests were run in the ‘‘sat’’ procedure on

Fig. 8. Furan and chlorobenzene light-off curves alone and in competition on TsVW.

230

D.P. Debecker et al. / Applied Catalysis B: Environmental 74 (2007) 223–232

TsVW that was described as the most interesting formulation, based on the study on chlorobenzene oxidation. Furan oxidation is not significantly disturbed by the presence of chlorobenzene. Chlorobenzene conversion, at the opposite, is dramatically inhibited by the presence of furan in the flow. When chlorobenzene is alone in the flow the conversion reaches 19%, 61% and 97% at 150 8C, 200 8C and 250 8C, respectively. The total conversion is obtained at 300 8C. In the case of the competition test, no chlorobenzene is removed from the flow, even at 250 8C. The conversion only starts above 250 8C and reaches 96% at 300 8C. This means that furan and chlorobenzene compete for the same active phase of the catalyst, namely the VOx phase. As furan adsorption is fast, the active sites are blocked and chlorobenzene cannot reach them. Chlorobenzene removal is still inhibited after furan conversion has reached 100%. This is due to the presence of the carbonaceous deposit formed by furan and/or its products of partial oxidation. The sites are free to accept chlorobenzene only at higher temperature, when the ignition of the deposit has occurred. 3.3.2. Discussion on the possible role of the central ring of dioxin Whether the optimization of catalyst formulation is envisaged on the basis of a chlorinated VOC or of an oxygenated VOC will affect the decision of dioxins abatement catalyst designers. The pertinent question is thus to know which model is the most relevant. Knowing that a dioxin bears both an oxygenated ring and two chlorinated rings, what can be the role of these different moieties? The above-presented intermolecular competition results help us having an insight on the possible role of the oxygenated ring of a dioxin in the course of its abatement on a V-based catalyst. Our results confirm that the oxygenated function of a furan has a drastically higher affinity for V-based catalysts than the chlorinated function of chlorobenzene. When these two functions are present together in the flow, the oxygenated one obviously wins the intermolecular competition. This suggests that the central oxygenated ring of a dioxin could also win the intramolecular competition against the chlorinated moieties of the same molecule. The adsorption of the dioxin is thus possibly directed by the oxygen atom(s) of the central ring and likely not by the chlorinated ends. This hypothesis could explain some observations that are reported in the literature. Weber et al. [27] noteworthy showed that, in industrial plants, dioxin removal proceeds under two distinct regimes: in the lower temperature regime, removal mainly proceeds through adsorption and in the higher temperature regime catalytic oxidation is dominant. This behavior is similar to the behavior we describe here in the case of furan and is not surprising in the perspective of our hypothesis. Furthermore, Liljelind et al. [20] note that the adsorption of dioxins is reversible. This means that the molecular structure of the pollutant is not altered during the very first step of the catalytic process. The dissociative adsorption through the chlorine atom should thus not be considered as the initial step during dioxin abatement and certainly not as a spontaneous fast step as when chlorobenzene

is concerned. Also, the removal efficiency is always much higher in dioxin abatement than in mono or poly chlorobenzenes abatement [20,27]. For these latter models, the higher the degree of chlorination, the lower the activation energy. Dichlorobenzene, for example, adsorbs more easily on the catalyst than chlorobenzene because it bears two chlorine atoms, which are involved in the adsorption step [21]. For dioxins however, the destruction kinetics decreases with the chlorination degree, which suggests again that the chlorine atoms are not promoting the removal of the pollutant. Moreover, for a given chlorination degree, the destruction rates for PCDDs is always higher as compared to PCDFs [19]. This could be explained by the fact that PCDDs bear two oxygen atoms in their central ring instead of one in PCDFs if we consider that these oxine functions play a key role in the initiation of the process. It was shown that the improvements that are proposed after optimization studies based on chlorobenzene oxidation are useless or deleterious in the case of furan oxidation. Now we suggest that this could also be the case in the course of dioxin oxidation. 4. Conclusions We compared the catalytic performances of V-based catalysts in the oxidation of chlorobenzene and furan. The behaviors of furan and chlorobenzene on V-based catalysts are very different. This difference is linked to marked differences in the adsorption behavior of the two VOCs. Since the adsorption of the molecule on the catalyst is the first step of the entire process of catalytic total oxidation it can dictate the performances offered by the catalyst. As a consequence, given preparation parameters will have different effects on the conversion of two VOC that have a different adsorption behavior. Adsorption of chlorobenzene needs to be promoted and, in this view the amount of Brønsted sites is critical. We confirm here that it is possible to improve catalyst formulation by increasing the amount of Brønsted surface acid sites. It can be achieved by two ways: - Using a sulfate containing TiO2 support with more Lewis sites promotes the spreading of the impregnated phase. These phases exhibit in turn a higher amount of Brønsted sites required for the adsorption of chlorobenzene. - Co-impregnating Mo or W oxides along with the active phase of vanadium. These two oxides do not cover the active phase of VOx and exhibit a huge number of Brønsted sites assisting the adsorption of chlorobenzene. The abatement of furan occurs at lower temperature than the abatement of chlorobenzene. Furan adsorbs strongly on our catalysts. At low temperature, a stable true conversion is only achieved after the catalyst surface has been saturated with adsorbed furan. This behavior is similar on all the formulations tested. In other words, the adsorption of furan occurs easily whatever the formulation. As a consequence, an attempt to

D.P. Debecker et al. / Applied Catalysis B: Environmental 74 (2007) 223–232

improve the catalyst formulation through a promotion of its adsorbent properties will not be efficient. In some cases, these attempts can rather be deleterious. Hence; - The sulfate containing support does not bring any significant activation. The limited beneficial effect brought by Ts is mainly due to its textural properties (higher specific surface area) while the change in the surface chemistry (higher Lewis and Bro¨nsted acidity) does not improve the performance of our catalysts in the oxidation of furan. - Mo and W oxides, as secondary phases have a deleterious effect on the performances of our catalysts. The fact that they bring a huge quantity of Bro¨nsted sites does not favor the catalytic activity. Instead, as these oxides both are inefficient in total oxidation of furan and prevent the active phase from spreading well on the support, their presence inevitably leads to a negative impact on the catalyst performances. These observations can be of great importance in terms of optimization of the catalysts formulations. If an upgraded formulation is proposed, it will not necessarily be beneficial for the abatement of different model molecules. And in turn, the nature of the model molecule used in works that aim at the optimization of catalyst formulation, will affect the decisions. Similarly, changing a given parameter of the catalyst preparation could have unpredicted effects when applied to dioxin abatement. Our results on the competition between furan and chlorobenzene suggest that dioxins could behave in the same way as furan. Indeed, the central oxygenated ring of the molecule could win the intramolecular competiton against the chlorinated moieties, just as furan does against chlorobenzene in the case of intermolecular competition. The oxine function could thus decisively dictate the nature of the pollutant-catalyst interaction and hence the corresponding performances. Acknowledgments This work was supported by the ‘‘Direction Ge´ne´rale des Technologies, de la Recherche et de l’Energie (DGTRE)’’ of the ‘‘Re´gion Wallonne’’ (Belgium) (convention no. 0114825), the ‘‘Fonds National de la Recherche Scientifique (FNRS)’’ of Belgium and the Universite´ catholique de Louvain. The involvement of the laboratory in the Coordinated Action ‘‘CONCORDE’’ as work package leader, in the Network of Excellence ‘‘FAME’’ of the EU 6th FP, in the IUAP network: ‘‘Supramolecularity’’ sustained by the ‘‘Service public fe´de´ral de programmation politique scientifique’’ (Belgium) are also acknowledged. D.P. Debecker also acknowledges the FNRS for his position of Research Fellow. References [1] K. Everaert, J. Baeyens, J. Hazard. Mater. 109 (2004) 113–139. [2] K. Everaert, J. Baeyens, Waste Manage. 24 (2004) 37–42. [3] P. Papaefthimiou, T. Ioannides, X.E. Verykios, Appl. Catal. B: Environ. 13 (1997) 175–184. [4] A. Buekens, H. Huang, J. Hazard. Mater. 62 (1998) 1–33.

231

[5] E. Finocchio, M. Baldi, G. Busca, C. Pistarino, G. Romezzano, F. Bregani, G.P. Toledo, Catal. Today 59 (2000) 261–268. [6] M. Goemans, P. Clarysse, J. Joannes, P. De Clercq, S. Lenaerts, K. Matthys, K. Boels, Chemosphere 54 (2004) 1357–1365. [7] J. Corella, J.M. Toledo, A.M. Padilla, Appl. Catal. B: Environ. 27 (2000) 243–256. [8] M. Berndt, P. Landri, Catal. Today 75 (2002) 17–22. [9] M. Ferrandon, J. Carno, S. Jaras, E. Bjornbom, Appl. Catal. A: Gen. 180 (1999) 141–151. [10] A. Musialik-Piotrowska, K. Syczewska, Catal. Today 73 (2002) 333– 342. [11] S. Krishnamoorthy, J.P. Baker, M.D. Amiridis, Catal. Today 40 (1998) 39– 46. [12] S.H. Taylor, C.S. Heneghan, G.J. Hutchings, I.D. Hudson, Catal. Today 59 (2000) 249–259. [13] P. Forzatti, L. Lietti, Catal. Today 52 (1999) 165–181. [14] V. de Jong, M.K. Cieplik, W.A. Reints, F. Fernandez-Reino, R. Louw, J. Catal. 211 (2002) 355–365. [15] F. Bertinchamps, C. Poleunis, C. Gre´goire, P. Bertrand, E.M. Gaigneaux, 4th International Conference on Environmental catalysis Heidelberg, Book of Abstracts (5–8 June), (2005), pp. 155–156. [16] L.J. Alemany, F. Berti, G. Busca, G. Ramis, D. Robba, G.P. Toledo, M. Trombetta, Appl. Catal. B: Environ. 10 (1996) 299–311. [17] F. Bertinchamps, C. Gregoire, E.M. Gaigneaux, Appl. Catal. B: Environ. 66 (2006) 10–22. [18] Y. Ide, K. Kashiwabara, S. Okada, T. Mori, M. Hara, Chemosphere 32 (1996) 189–198. [19] R. Weber, T. Sakurai, H. Hagenmaier, Appl. Catal. B: Environ. 20 (1999) 249–256. [20] P. Liljelind, J. Unsworth, O. Maaskant, S. Marklund, Chemosphere 42 (2001) 615–623. [21] J. Lichtenberger, M.D. Amiridis, J. Catal. 223 (2004) 296–308. [22] S. Krishnamoorthy, J.A. Rivas, M.D. Amiridis, J. Catal. 193 (2000) 264– 272. [23] J. Jones, J.R.H. Ross, in: Proceedings of the International Forum on Environmental Catalysis ’95 (IFEC ’95), Catal. Today 35 (1997) 97–105. [24] F. Bertinchamps, C. Gregoire, E.M. Gaigneaux, Appl. Catal. B: Environ. 66 (2006) 1–9. [25] F. Bertinchamps, M. Treinen, N. Blangenois, E. Mariage, E.M. Gaigneaux, J. Catal. 230 (2005) 493–498. [26] S. Lomnicki, J. Lichtenberger, Z. Xu, M. Waters, J. Kosman, M.D. Amiridis, Appl. Catal. B: Environ. 46 (2003) 105–119. [27] R. Weber, M. Plinke, Z. Xu, M. Wilken, Appl. Catal. B: Environ. 31 (2001) 195–207. [28] R. Weber, T. Sakurai, Appl. Catal. B: Environ. 34 (2001) 113–127. [29] M.D. Amiridis, R.V. Duevel, I.E. Wachs, Appl. Catal. B: Environ. 20 (1999) 111–122. [30] M.A. Larrubia, G. Busca, Appl. Catal. B: Environ. 39 (2002) 343–352. [31] M. Taralunga, Oxydation catalytique de compose´s aromatiques chlore´s representatifs de polluants organiques persistants (POP), Ph.D. thesis, Universite´ de Poitiers, France (2005). [32] M. Baldi, V.S. Escribano, J.M.G. Amores, F. Milella, G. Busca, Appl. Catal. B: Environ. 17 (1998) L175–L182. [33] A. Gil, N. Burgos, M. Paulis, M. Montes, L.M. Gandia, Stud. Surf. Sci. Catal. 130 (2000) 2153–2158. [34] A. Gil, S.A. Korili, M.A. Vicente, G. L.M., Stud. Surf. Sci. Catal. 130 (2002) 517–525. [35] L.M. Gandia, A. Gil, S.A. Korili, Appl. Catal. B: Environ. 33 (2001) 1–8. [36] L.M. Gandia, M.A. Vicente, A. Gil, Appl. Catal. B: Environ. 38 (2002) 295–307. [37] M. Paulis, L.M. Gandia, A. Gil, J. Sambeth, J.A. Odriozola, M. Montes, Appl. Catal. B: Environ. 26 (2000) 37–46. [38] K.M. Parida, A. Samal, Appl. Catal. A: Gen. 182 (1999) 249–256. [39] M. Baldi, E. Finocchio, F. Milella, G. Busca, Appl. Catal. B: Environ. 16 (1998) 43–51. [40] J.M. Gallardo-Amores, T. Armaroli, G. Ramis, E. Finocchio, G. Busca, Appl. Catal. B: Environ. 22 (1999) 249–259.

232

D.P. Debecker et al. / Applied Catalysis B: Environmental 74 (2007) 223–232

[41] E. Fernandez Lopez, V. Sanchez Escribano, C. Resini, J.M. GallardoAmores, G. Busca, Appl. Catal. B: Environ. 29 (2001) 251–261. [42] C. Lahousse, C. Cellier, B. Delmon, P. Grange, Stud. Surf. Sci. Catal. 130 (2000). [43] C. Lahousse, A. Bernier, P. Grange, B. Delmon, P. Papaefthimiou, T. Ioannides, X. Verykios, J. Catal. 178 (1998) 214–225. [44] C. Lahousse, A. Bernier, P. Ruiz, C. Gre´goire, B. Delmon, P. Papaefthimiou, T. Ioannides, X. Verykios, Procceedings 5th World Congress of Chemical Engineering, San Diego, vol. 3, 1996. [45] J.H. Scofield, J. Electron Spectrosc. Relat. Phenom. 8 (1976) 129– 137. [46] G. Spoto, F. Geobaldo, S. Bordiga, C. Lamberti, D. Scarano, A. Zecchina, Top. Catal. 8 (1999) 279–292. [47] D. Honicke, J. Catal. 105 (1987) 10–18.

[48] B. Kubias, U. Rodemerck, H.-W. Zanthoff, M. Meisel, Proceeding of the 5th European Workshop Meeting on Selective Oxidation by Heterogeneous Catalysis, Catal. Today 32 (1996) 243–253. [49] W.D. Schroeder, C.J. Fontenot, G.L. Schrader, J. Catal. 203 (2001) 382– 392. [50] J. Leyrer, R. Margraf, E. Taglauer, H. Knozinger, Surf. Sci. 201 (1988) 603–623. [51] G. Sinquin, C. Petit, S. Libs, J.P. Hindermann, A. Kiennemann, Appl. Catal. B: Environ. 27 (2000) 105–115. [52] B. Ramachandran, H.L. Greene, S. Chatterjee, Appl. Catal. B: Environ. 8 (1996) 157–182. [53] S.-W. Baek, J.-R. Kim, S.-K. Ihm, Selections from the presentations of the 3rd Asia-Pacific Congress on Catalysis, Catal. Today 93–95 (2004) 575–581.