Chemical Engineering - Industrial Organic Chemistry

Journal_Tronconi

Etc

Journal of Catalysis 245 (2007) 1–10 www.elsevier.com/locate/jcat Redox features in the catalytic mechanism of the “standard” and “fast” NH 3-SCR of NO x over a V-based catalyst investigated by dynamic methods Enrico Tronconi a,∗ , Isabella Nova a, Cristian Ciardelli a, Daniel Chatterjee b, Michel Weibel b aDipartimento di Chimica, Materiali e Ingegneria Chimica “G. Natta”, Politecnico di Milano, Piazza Leonardo da Vinci 32, I-20133 Milano, Italy bDaimlerChrysler AG Abteilung RBP/C, HPC: 096-E220, D-70546 Stuttgart, Germany Received 29 June 2006; revised 20 August 2006; accepted 12 September 2006Available online 16 October 2006 Abstract The redox mechanism governing the selective catalytic reduction (SCR) of NO/NO 2by ammonia at low temperature was investigated by transient reactive experiments over a commercial V 2O5/WO 3/TiO 2catalyst for diesel exhaust aftertreatment. NO+NH 3temperature-programmed reaction runs over reduced catalyst samples pretreated with various oxidizing species showed that both NO 2and HNO 3were able to reoxidize the V catalyst at much lower temperature than gaseous O 2: furthermore, they significantly enhanced the NO+NH 3reactivity below 250 ◦C via the buildup of adsorbed nitrates, which act as a surface pool of oxidizing agents but are decomposed above that temperature. Both such features, which were not observed in comparative experiments over a V-free WO 3/TiO 2catalyst, point out a key catalytic role of the vanadium redox properties and can explain the greater deNO xefficiency of the “fast” SCR (NO+NH 3+NO 2) compared with the “standard” SCR (NO+NH 3+O 2) reaction. A unifying redox approach is proposed to interpret the overall NO/NO 2–NH 3SCR chemistry over V-based catalysts, in which vanadium sites are reduced by the reaction between NO and NH 3and are reoxidized either by oxygen (standard SCR) or by nitrates (fast SCR), with the latter formed via NO 2disproportion over other nonreducible oxide catalyst components. ©2006 Elsevier Inc. All rights reserved. Keywords:Diesel-urea SCR; Selective catalytic reduction; Standard SCR; Fast SCR; Redox properties; Redox catalytic kinetics; Dynamic methods 1. Introduction Selective catalytic reduction (SCR) with NH 3/urea is emerg- ing as the most promising technology for the abatement of NO x emissions from diesel vehicles[1–8]. This has stimulated a re- newed interest in investigating the fundamental aspects of SCR catalytic chemistry, also in view of the need for the transporta- tion industry to develop design and simulation tools incorporat- ing suitable SCR kinetic schemes. Indeed, NH 3SCR over vanadia-type catalysts, in which 1 molecule of NO is reduced by 1 molecule of ammonia in the presence of oxygen to give dinitrogen and water, 2NH 3+2NO+(1/2)O 2→2N 2+3H 2O, (1) has represented for the last two decades the most effective com- mercial deNO xing process for stack gases from power plants *Corresponding author. Fax: +39 02 2399 3318. E-mail address:[email protected](E. Tronconi). and other stationary sources[9]. However, the specific demands of mobile applications, associated with, for example, volume limitations and dynamic operation, do not permit straightfor- ward transposition of the technology. Because the working con- ditions for mobile applications may be much colder than in stationary installations, increasing the deNO xactivity at low temperatures represents a major development goal. A possible solution to this issue is represented by the so-called “fast” SCR reaction known since the early 1980s, when Kato et al.[10] found that the reaction involving an equimolar NO and NO 2 feed mixture, 2NH 3+NO+NO 2→2N 2+3H 2O, (2) is significantly faster than the “standard” SCR, reaction(1), at low temperature. In practical terms, a preoxidation catalyst located upstream of the SCR catalyst could convert a fraction of NO in the engine exhausts to NO 2to approach the optimal NO/NO 2equimolar feed ratio of reaction(2) [11]. Neverthe- less, adding NO 2to the SCR reacting system introduces consid- 0021-9517/$ – see front matter©2006 Elsevier Inc. All rights reserved. doi:10.1016/j.jcat.2006.09.012 2E. Tronconi et al. / Journal of Catalysis 245 (2007) 1–10 erable complexity resulting from the multiplication of primary and secondary reaction routes and may result in the formation of such undesired byproducts as NH 4NO 3and N 2O[12,13]. From a more fundamental standpoint, the relationship between the catalytic chemistry of the standard SCR and that of the less well-investigated fast SCR, as well as the reasons why NO 2dra- matically accelerates the reduction of NO by ammonia, have not yet been fully elucidated.Because of its widespread application in the abatement of NO xemissions from stationary sources, the mechanism of the standard SCR reaction has been extensively investigated[9,14]. It is generally agreed that the reaction proceeds according to a redox scheme, in which gaseous oxygen is needed for reox- idation of the V-related catalyst sites reduced by the reaction between NO and NH 3. It is also established that at low temper- ature, the catalyst reoxidation by gas-phase oxygen is the rate- determining step of the overall reaction mechanism[15–20]. Concerning the fast SCR reaction(2), a much more limited number of papers analyzing the NH 3–NO/NO 2reacting system are available in the scientific literature. This field was pioneered by Koebel and co-workers[11,17,21,22], who extensively in- vestigated the reactivity of NH 3–NO/NO 2over powdered and monolithic vanadia-based catalysts. Addressing possible rea- sons for the observed higher rates of the fast SCR reaction at lowT, they proposed that gaseous NO 2would replace oxy- gen as a more effective oxidizing agent, thus allowing faster reoxidation of the vanadium sites. The NO 2-enhanced reoxi- dation of the catalyst was demonstrated by in situ Raman ex- periments over V 2O5/TiO 2[17], but no direct kinetic evidence was provided to confirm that this effect could explain the order- of-magnitude increment of the SCR. Moreover, a mechanistic scheme in which NO 2simply replaces oxygen in reoxidiz- ing the V sites cannot account for the significant role played by other species in the NO/NO 2–NH 3system. Recently we showed by transient experiments over a commercial V-based catalyst that the reduction of nitrate species (related to previ- ously deposited ammonium nitrate formed by reaction between NO 2and NH 3) by nitric oxide proceeds at the same rate as the fast SCR at 170◦C[23,24]. This suggests that such a reaction pathway may be involved (and actually be rate-controlling) in the low-temperature catalytic mechanism of the fast SCR reac- tion over V-based catalysts. Such data have been rationalized according to a reaction scheme in which NO 2forms nitrite and nitrate species by disproportion, NO reduces nitrates to nitrites, and NH 3reacts with nitrites to form unstable ammonium nitrite, which readily decomposes to nitrogen and water[24]. Simi- lar indications, obtained in this case mostly by spectroscopic techniques, have been reported by Sachtler and co-workers for NO/NO 2–NH 3SCR over zeolite catalysts[25]. This scheme is able to account for an extensive set of kinetic observations, including the influence of the NO/NO 2feed ratio on the selec- tivities to N 2,NH 4NO 3, and N 2O[13]; however, it does not explicitly reflect the redox nature of the catalytic mechanism of SCR. Herein we address redox mechanistic features of standard SCR and fast SCR over an industrial V 2O5–WO 3/TiO 2cat- alyst for mobile applications, considered a representative V-based SCR catalyst with optimized composition. The reactivity of NO +NH 3in the low-temperature range is investigated by temperature-programmed reduction (TPR) methods under con- ditions typical of real aftertreatment devices over catalyst sam- ples subjected to various pretreatments, to examine the influ- ence of different reducing and oxidizing agents on SCR activity. Specifically, the relation between surface nitrates and the redox mechanism of fast SCR is explored. Comparative experiments over a V-free WO 3/TiO 2catalyst are also performed to identify the catalytic role of vanadium. 2. Materials and methods Unsteady SCR reactive experiments were performed at 50– 250 ◦C over a commercial extruded V 2O5–WO 3/TiO 2catalyst (A s≈70 m 2/g) with intermediate V content, originally sup- plied as a honeycomb monolith. A significant portion of the catalyst was crushed and ground to powder (140–200 mesh); for each experiment, a small sample (160 mg) was collected from the mixed powders, diluted with 80 mg of quartz, and loaded in a flow microreactor consisting of a quartz tube (6 mm i.d.). Using the same procedures, a commercial WO 3/TiO 2catalyst (Thann et Mulhouse S.A.) was also tested in specific runs for comparison purposes.The test reactor was operated at atmospheric pressure with a total flow rate of either 120 or 280 cm 3/min (STP), correspond- ing to a gas hourly space velocity (GHSV) of about 9×10 4 or 2.1×10 5h−1 , respectively. Diluted gas streams of NO, NO 2,NH 3, and O 2in He from bottled calibrated mixtures were mixed in suitable proportions by means of mass flow controllers (Brooks 5850 E) to achieve the desired feed composition for each run. For H 2O-containing feeds, the feed stream was passed through a saturator maintained at a controlled temperature be- fore entering the microreactor. For catalyst pretreatment with HNO 3, the feed stream was saturated with a suitable aqueous solution of nitric acid.The reactor outlet was directly connected to both a UV an- alyzer (ABB Limas 11-HW) and a quadrupole mass spectrom- eter (Balzers QMS 200) operating in parallel. In each experi- ment, the UV analyzer monitored the temporal evolution of the outlet NO, NO 2, and NH 3concentrations. Helium was used as the carrier gas, so that nitrogen (the main SCR product) could be detected by the MS, along with the byproduct N 2O, thus al- lowing evaluation of overall N balances, which always closed within±5% at steady state. As a consistency check, the outlet concentration traces of NO, NO 2, and NH 3also were estimated from the MS signals after proper calibration; the steady-state levels agreed satisfactorily with those measured by the UV an- alyzer, whereas the transient phases were typically associated with a slower MS response, particularly in the case of ammonia. In view of the purposes of the present work, herein we primar- ily report the outlet concentration traces of NO generated by the UV analyzer.Typical experiments were NO+NH 3TPR runs. A feed stream consisting of NH 3(1000 ppm)+NO (1000 ppm) with H 2O (1% v/v), balance He, and no oxygen was initially ad- mitted to the reactor at 50 ◦C. Then the catalyst temperature E. Tronconi et al. / Journal of Catalysis 245 (2007) 1–103 was continuously increased at 20 ◦C/min up to 250–550 ◦C, depending on the purposes of the experiment. Such TPR exper- iments were carried out over reference reduced and oxidized catalyst samples, as well as over other catalyst samples sub- jected to different pretreatments.The reference reduced catalyst samples were obtained by exposure to NH 3(1000 ppm feed stream) at 550 ◦C for about 1.5 h. Catalyst reduction was confirmed by a peak of N 2evolu- tion during the first few minutes, the integral amount of which was compatible with reduction of the vanadium catalyst load. The reference oxidized catalyst samples were obtained by ex- posure to O 2(500 ppm) during a temperature ramp (20 ◦C/min) up to 550 ◦C; oxidation of the catalyst was confirmed by O 2 consumption and H 2O evolution at temperatures above ca. 200 ◦C. Various other pretreatments were also applied to the cata- lyst to investigate their influence on its oxidation state and its SCR reactivity, as discussed in the following sections. Specifi- cally, pretreatment procedures were performed by feeding H 2O (1% v/v) in a temperature ramp up to 550 ◦C, H 2O (1% v/v) at 200 ◦C, NO (1000 ppm) at 200 ◦C, NO (1000 ppm)+H 2O(1% v/v) at 200 ◦C, NH 3(1000 ppm) at 200 ◦C, NO (1000 ppm)+ NH 3(1000 ppm) at 200 ◦C, NO 2(100 or 500 ppm) at 150 ◦C, and HNO 3(about 100 ppm)+H 2O (1% v/v) at 100, 150, and 200 ◦C. In addition to TPR runs, specific tests were carried out ac- cording to the transient response method (TRM experiments). This involved performing a step change of the inlet concen- tration of one reactant (1000 ppm NH 3, NO, or NO 2) while feeding other reactants in the presence of H 2O (1% v/v) and bal- ance He at 170 ◦C. In line with the SCR literature for V-based systems[14], no significant activity in the oxidation of NO to NO 2up to 450 ◦C was observed over the present catalysts, even in dedicated diagnostic runs. 3. Results and discussion 3.1. SCR behavior of reference reduced and reference oxidized catalyst samples The activity in the standard SCR reaction of reference re- duced and oxidized catalyst samples was first analyzed by NO+NH 3TPR. Over the reference reduced catalyst, no con- version of NO was evident up to 200 ◦C, and only minor SCR activity was apparent at 250 ◦C, when about 980 ppm of NO was still present in the outlet stream.A different picture was observed for NO+NH 3TPR over the reference oxidized catalyst. The NO outlet concentration began to decrease at≈130 ◦C, and the NO conversion was close to 8% at 250 ◦C. Occurrence of the SCR reaction was also confirmed by stoichiometric evolution of nitrogen, the integral amount of which was found to be in good agreement with that measured during reference reduction of the catalyst sample.The data were reproducible, as confirmed by several repli- cated runs over both prereduced and preoxidized catalyst sam- ples. The data indicate that the reduced commercial V catalyst is virtually inactive in the standard SCR reaction when no oxy-gen is present in the feed stream; the minor NO conversions measured above 200 ◦C are attributed to the so-called “slow” SCR reaction[12,14], 2NH 3+3NO→(5/2)N 2+3H 2O. (3) In fact, replicated temperature ramps performed up to higher temperatures over the reduced catalyst confirmed conversion of both nitric oxide and ammonia, as well as production of nitro- gen, according to the stoichiometry of reaction(3).Itiswell known that under practical conditions, that is when oxygen is present, slow SCR is negligibly slow compared with standard SCR.In contrast, the catalyst sample pretreated according to the reference oxidizing procedure appeared to be significantly more active. Indeed, analysis of the outlet concentrations of nitric ox- ide, ammonia, and nitrogen confirmed the occurrence of the standard SCR reaction(1)at temperatures as low as 130 ◦C. This is in line with literature reports[15]that standard SCR over V-based catalysts proceeds by consuming the catalyst lat- tice oxygen, in agreement with a redox mechanism. If oxygen is fed to the reactor, then the lattice oxygen consumed by the reaction can be replenished, and the reaction proceeds to steady state. Otherwise (i.e., if gaseous oxygen is not available), the reaction stops once all of the reactive lattice oxygen of the cat- alyst is depleted[15]. In fact, starting from around 250 ◦C, the evolution of NO conversion measured over the reference oxi- dized catalyst increased only slowly, paralleling that observed over the reference reduced catalyst, thus indicating that only the slow SCR reaction(3)was proceeding. 3.2. Standard SCR reaction versus catalyst reoxidation We now apply the tests assessed in the previous section to clarify aspects of the redox mechanism associated with the stan- dard SCR reaction.Fig. 1illustrates results from three different TPR runs: (A) NO concentration profile during NO+NH 3+O 2 TPR over a reference reduced catalyst, (B) O 2concentration profile during a temperature ramp for reoxidation of a refer- ence reduced catalyst, and (C) NO concentration profile during NO+NH 3+O 2TPR over a WO 3/TiO 2catalyst. It is apparent that curve (B), although very noisy due to ex- perimental limitations with the signal acquisition from the MS, essentially overlaps curve (A). Accordingly, the same activa- tion threshold prevailed for both reoxidation of a reference re- duced catalyst (curve (B)) and standard SCR in the presence of gaseous oxygen over a reference reduced catalyst (curve (A)). In contrast, the NO+NH 3+O 2TPR experiment over the V-free catalyst (curve (C)) showed the onset of some NO conversion only above roughly 300 ◦C, a threshold temperature well above that measured for the V-based catalyst, thus confirming that the low-temperature activity in the standard SCR reaction can be ascribed to the well-known redox properties of the vanadium catalyst component[9,12,14]. Accordingly, the data inFig. 1support that V reoxidation is the rate-limiting step in the standard SCR reaction. Of course, such a result has significant implications for SCR kinetic mod- eling and is in agreement with indications reported previously. 4E. Tronconi et al. / Journal of Catalysis 245 (2007) 1–10 Fig. 1. Curve (A) NO concentration profile during NO+NH 3+O 2(1000/ 1000/500 ppm+He) TPR over a reference reduced catalyst. Curve (B) O 2 concentration profile during reoxidation (O 2feed=500 ppm+He)ofaref- erence reduced catalyst. Curve (C) NO concentration profile during NO+ NH 3+O 2(1000/1000/20,000 ppm+He) TPR over a WO 3/TiO 2catalyst; flow rate=280 cm 3/min (STP). Heating rate=20 ◦C/min. Using different approaches, Lietti and Forzatti[15]and Marsh- neva et al.[20]concluded that the slow step in the SCR reaction over V 2O5–WO 3/TiO 2catalysts is associated with V catalyst oxidation by gaseous oxygen.Having established the redox nature of the standard SCR re- action, we next assess the roles of different species in the related reduction and oxidation steps, respectively. 3.3. Activities of different species in catalyst reduction To identify the species responsible for catalyst reduction dur- ing the standard and fast SCR reactions, we studied the activity of different potential reducing agents participating in the SCR reacting system (NO, NH 3,H 2O, and combinations thereof), according to the following procedure. A fresh catalyst sample was first oxidized, then pretreated with one of the candidate re- ducing agents. After the pretreatment, NO+NH 3TPR was run, and the resulting NO reduction activity was compared with the corresponding data obtained over both the reference oxidized and reduced catalysts.The results are summarized inFig. 2, which compares the NO reduction activity in the TPR runs performed over the catalyst samples after various pretreatments. Specifically, such pretreatments involved feeding H 2O (1% v/v) at 200 ◦C (curve (A)), H 2O (1% v/v) at a temperature ramp up to 550 ◦C (curve (B)), NO (1000 ppm) at 200 ◦C (curve (C)), NH 3 (1000 ppm) at 200 ◦C (curve (D)), NO (1000 ppm)+H 2O(1% v/v) at 200 ◦C (curve (E)), and NO+NH 3(1000+1000 ppm) at 200 ◦C (curve (F)). In addition, curves obtained for the reference oxidized (curve (G)) and of the reference reduced (curve (H)) catalyst samples are reported for comparison.Fig. 2shows that in curves (A)–(E), the observed deNO xac- tivity was essentially identical to that typical of the reference oxidized catalyst (curve (G)) and remained evidently higher than the activity measured over the reference reduced cata- Fig. 2. NO concentration profiles during NO+NH 3TPR (1000/1000 ppm+ He) over catalysts after different pretreatments: (A) H 2O (1% v/v) at 200 ◦C; (B) H 2O (1% v/v) inT-ramp up to 550 ◦C; (C) NO (1000 ppm) at 200 ◦C; (D) NH 3(1000 ppm) at 200 ◦C; (E) NO (1000 ppm)+H 2O (1% v/v) at 200 ◦C; (F) NO (1000 ppm)+NH 3(1000 ppm) at 200 ◦C; (G) reference oxi- dized catalyst; (H) reference reduced catalyst. Other conditions as inFig. 1. lyst (curve (H)). Thus, it appears that exposure of the oxidized catalyst to atmospheres containing either H 2OorNOorNH 3 alone or H 2O+NO together at 200 ◦C did not reduce the cat- alyst. H 2O did not reduce the oxidized catalyst even at 550 ◦C. Further supporting this conclusion is the fact that no nitrogen evolution was detected during all such pretreatments.In contrast, curve (F) inFig. 2shows the results of a simi- lar experiment over an oxidized catalyst sample that had been exposed to a mixture of NO and NH 3(1000+1000 ppm) at 200 ◦C before the TPR run. In this case, deNO xactivity was quite similar to that observed over the reference reduced cat- alyst (curve (H)); moreover, N 2evolution was observed dur- ing the pretreatment. Although additional spectroscopic evi- dence would be needed in this respect, these data suggest that NO+NH 3jointly were able to reduce the present V catalyst al- ready at 200◦C. This indeed seems consistent with the results inFig. 1. As discussed in the previous section, at 200 ◦C, SCR activity is limited by catalyst reoxidation, whereas the reduc- tion step of its redox mechanism was found to proceed already rapidly at the same temperature. 3.4. Mechanistic implications: catalyst reduction In this section we briefly discuss some mechanistic impli- cations of the present results concerning the catalyst reduction phase in the redox mechanism of the standard and fast SCR re- actions. Our data do not support a direct reducing action of NO alone. This is in agreement with the view, supported by many authors, that NH 3rather than NO is activated (i.e., oxidized) by V-based catalysts in the first step of the standard SCR catalytic mechanism[14]. Our data also do not support that at 200 ◦C, ammonia alone is directly activated by deeply reducing the V catalyst. In fact, as suggested byFig. 2and by the data on N 2 evolution during catalyst pretreatment, under the present exper- imental conditions, the catalyst reduction step in the redox SCR E. Tronconi et al. / Journal of Catalysis 245 (2007) 1–105 mechanism seems to require a cooperative action (co-presence) of both NO and NH 3. In this respect it should be also considered that the reducing action of NH 3is enabled by the presence of V atoms in adjacent sites[26,27]; however, exploring the affect of V surface density is beyond the scope of the present study.The foregoing results may be consistent with the following interpretations: 1. Ammonia, already adsorbed onto nonreducible acidic sites, is initially activated (oxidized) by adjacent vanadium re- dox sites, but the reaction proceeds to a significant extent only after interaction with gaseous NO according to an Eley–Rideal scheme, NH * 3+V 5+ =O↔V=O[NH 3], (4a) V=O[NH 3]+NO→N 2+H 2O+V 4+ –OH, (4b) where NH * 3represents adsorbed ammonia. A similar proposal was formulated by Topsøe et al. on the basis of in situ FTIR evidence[19]. However, the chemical na- ture of the initial NH 3activation, which does not reduce the cat- alyst deeply (with nitrogen evolution), as documented herein, remains to be clarified. Moreover, this interpretation is some- what at variance with recent kinetic data showing considerable inhibition of ammonia on the low-temperature SCR of NO and pointing to the existence of an optimal NH 3surface coverage [18,28–31]. It was also found that such an inhibiting action of ammonia, already reported by other authors[32–34], cannot be accommodated by a simple Eley–Rideal kinetic approach assuming a reaction between adsorbed ammonia and gaseous nitric oxide[18,30]. 2. Nitric oxide is oxidized by the V catalyst to nitrite species, but the equilibrium is highly unfavorable and shifts to the right only in the presence of NH 3(adsorbed onto nearby acidic sites), which reacts with nitrites to give N 2and H 2O via decomposi- tion of unstable ammonium nitrite intermediates, for example, according to V 5+ =O+NO↔V 4+ –ONO, (5a) V 4+ –ONO+NH * 3→N 2+H 2O+V 4+ –OH. (5b) In this case, the inhibiting action of ammonia could be more easily explained by either a competitive adsorption of NH 3onto the V sites involved in NO activation or an adverse electronic interaction of adsorbed NH 3with the vanadium oxidizing cen- ters[18]. In any case, both(4a)+(4b)and(5a)+(5b)yield the overall catalyst reduction step, NO+NH * 3+V 5+ =O→N 2+H 2O+V 4+ –OH. (6) For the standard SCR reaction, reoxidation of the reduced V sites closing the redox cycle is then carried out by gaseous oxy- gen according to V 4+ –OH+(1/4)O 2→V 5+ =O+(1/2)H 2O. (7) For fast SCR, however, other strong oxidizing agents—namely, NO 2and HNO 3—are present in the reacting environment in addition to O 2. Thus, it is of interest to compare their oxidizing activity, as we discuss in the following sections. Fig. 3. NO concentration profiles during NO+NH 3TPR over catalysts after different pretreatments: (A) O 2(500 ppm) atT=150 ◦C; (B) NO 2(100 ppm) at 150 ◦C; (C) HNO 3(∼100 ppm) at 150 ◦C; (D) reference oxidized catalyst; (E) reference reduced catalyst. Other conditions as inFig. 1. 3.5. Activity of different species in catalyst reoxidation Similar to the investigation of reducing agents in Section3, in this section we test the reactivity of different oxidizing species, namely O 2,NO 2, and HNO 3, in the catalyst reoxida- tion step of the redox SCR mechanism.Fig. 3compares the deNO xactivity observed in NO+NH 3TPR runs over three catalyst samples. Each sample was first reduced according to the reference procedure, then pretreated at 150 ◦C with a dif- ferent oxidizing agent—O 2(curve (A)), NO 2(curve (B)), or HNO 3(curve (C))—and finally tested in a NO+NH 3TPR run. Data collected in the same TPR experiment over the reference oxidized catalyst sample (curve (D)) and the reference reduced catalyst sample (curve (E)) also are displayed for comparison.It is apparent fromFig. 3that the NO evolution recorded when the catalyst sample was pretreated in oxygen flow at 150 ◦C (curve (A)) essentially overlaps with the curve obtained over the reference reduced catalyst (curve (E)): hence, oxygen was apparently unable to reoxidize the prereduced V catalyst at 150 ◦C. This is consistent with the data in curve (B) ofFig. 1, which suggest that the onset of catalyst reoxidation by oxygen occurred above 200 ◦C under our experimental conditions. In contrast, NO evolution traces recorded after exposing the reduced catalyst both to NO 2(curve (B)) and to HNO 3 (curve (C)) show that NO conversion was initiated already at about 130 ◦C, a threshold temperature very similar to that typ- ical of the reference oxidized catalyst (curve (D)). These data then suggest that both NO 2and HNO 3were able to effectively reoxidize the V catalyst at 150 ◦C. Notably, two similar TPR experiments were repeated af- ter exposing prereduced catalyst samples to HNO 3at 200 and 100 ◦C. The results (not reported here) showed behavior quite similar to that of curve (C) for the sample pretreated at 200 ◦C, whereas with pretreatment at 100 ◦C, the catalyst behaved like the reference reduced sample and thus likely was not oxidized. Accordingly, the temperature threshold for catalyst reoxidation 6E. Tronconi et al. / Journal of Catalysis 245 (2007) 1–10 Fig. 4. NO 2and O 2concentration profiles duringT-ramp at 20 ◦C/min af- ter catalyst pretreatment with HNO 3(∼100 ppm) at 150 ◦C. Feed flow rate= 280 cm 3/min (STP). by HNO 3lies between 100 and 150 ◦C under our experimental conditions.Another important feature becomes apparent on inspection ofFig. 3. For the catalyst samples pretreated with NO 2and HNO 3(curves (B) and (C), respectively), the behavior dur- ing the NO+NH 3temperature ramp was identical to that of the TPR run over the reference oxidized catalyst (curve D) up to about 170 ◦C; above this temperature, however, curves (B) and (C) exhibited a marked increment of deNO xactivity, fol- lowed by a sudden drop in NO conversion at around 250 ◦C. Interestingly, such a temperature threshold corresponds to the onset of nitrate decomposition, as indicated by the main de- sorption peaks of O 2and NO 2detected during TPD runs af- ter buildup of nitrates onto the catalyst by exposure to either HNO 3(seeFig. 4)orNO 2(with similar results). The forma- tion/storage of nitrate species onto pure TiO 2,WO 3/TiO 2, and V 2O5–WO 3/TiO 2catalysts from NO 2has been reported by sev- eral authors[35–38]. Thus,Figs. 3 and 4provide evidence in favor of a strong promoting action of adsorbed nitrates on the NO+NH 3SCR reactivity at low temperature. The surface ni- trates would then be decomposed and depleted above≈250 ◦C, explaining the subsequent drop in NO conversion. 3.6. Mechanistic implications: catalyst reoxidation and role of nitrates As for the catalyst reduction step in Section4, herein we discuss the results ofFigs. 3 and 4in relation to their rele- vant mechanistic implications. So far, only Raman evidence has been reported in the literature to support that NO 2affects the reoxidation of V-based SCR catalysts more efficiently than gaseous oxygen[17]. In this respect, the data inFig. 3provide a direct confirmation that NO 2can reoxidize the reference re- duced catalyst at a much lower temperature than O 2under SCR reactive conditions. They further prove that HNO 3is a much better oxidizing agent as well. Fig. 5. NO concentration profiles duringT-ramps at 20 ◦C/min. Curve (A) feed=NO+NH 3+NO 2(1000/1000/500 ppm). Curve (B) feed=NO+NH 3 (1000/1000 ppm) over reduced catalyst pretreated with NO 2(500 ppm) at 150 ◦C. Curve (C) feed=NO+NH 3+O 2(1000/1000/500 ppm). Because catalyst reoxidation is the rate-limiting step of the SCR reaction, as documented in Section2, an important impli- cation is that the overall SCR deNO xkinetics will be strongly enhanced in the presence of NO 2and/or HNO 3. In fact, the pro- moting action of NO 2when added to the NO+NH 3reacting system, leading to the so-called fast SCR, is well known in the SCR literature.Notably, the behavior observed after oxidizing the catalyst with either NO 2or HNO 3was similar. In fact, as presented and discussed in other works from our group[13,23,24], over V- based SCR systems at low temperatures NO 2readily undergoes the disproportion reaction 2NO 2+H 2O↔HONO+HNO 3,(8) so that the reactivities of NO 2and HNO 3can hardly be distin- guished.Fig. 3further suggests that the buildup of nitrates onto the V-based catalyst (possibly onto sites associated with the W and Ti components) may play an important role in enhancing the deNO xefficiency compared with the NO+NH 3+O 2stan- dard SCR below 250 ◦C. In this respect we have previously reported evidence pointing to a significant participation of ni- trates in the mechanism of the fast SCR reaction; specifically, data from dedicated transient reactivity experiments at 170 ◦C showed that the rate of fast SCR(2)equals the rate of reduction of nitrate species (related to NH 4NO 3)byNO[23,24]. Additional transient experiments illustrated inFig. 5provide more insight into the role of nitrates in the catalyst reoxidation step, and also into its relationship with the rate of fast SCR. InFig. 5, curve (A) represents the evolution of NO conversion observed during a TPR run at 150–300 ◦C with a feed includ- ing NO (1000 ppm)+NO 2(500 ppm)+NH 3(1000 ppm). At 150 ◦C, the NO conversion was>20% due to the fast SCR re- action. At this low temperature, however, NO 2and NH 3also reacted to form NH 4NO 3, which partially built up onto the E. Tronconi et al. / Journal of Catalysis 245 (2007) 1–107 catalyst[13,24], as pointed out by the following analysis of am- monia, NO 2, and nitrogen traces (not shown in the figure): 2NH 3+2NO 2→NH 4NO 3+N 2+H 2O. (9) With increasing temperature up to about 250 ◦C, NO conversion was significantly enhanced, eventually exceeding 80%. No- tably, this is higher than the stoichiometric limit of 50% associ- ated with the feed concentration of NO 2, a clear indication that the reaction proceeded at the expense of nitrates initially stored on the catalyst. In fact, NO conversion again dropped suddenly when the temperature exceeded 250 ◦C (i.e., the temperature threshold corresponding to the onset of nitrate decomposition [seeFig. 4]), and eventually approached the steady-state stoi- chiometric 50% limit.Curve (B) inFig. 5displays the evolution of NO conver- sion during a temperature ramp in which only NO and NH 3 (1000 ppm each) were fed over a reduced catalyst sample that had been previously pretreated with NO 2(500 ppm). The simi- larity between curves (A) and (B) is quite evident. In particular, in the range 150–225 ◦C the temperature dependence of the fast SCR reaction in the presence of gaseous NO 2(as inferred from curve (A)) is virtually the same of the NO+NH 3reaction in the presence of adsorbed nitrates (curve (B)). This strongly sug- gests that, as in standard SCR, the rate-determining step in the fast SCR reaction is still associated with reoxidation of the V sites, which is, however, accomplished more effectively by ad- sorbed nitrates generated from NO 2than by gaseous oxygen, as in the case of standard SCR.For comparison purposes,Fig. 5displays also curve (C) (same as curve (A) inFig. 1) associated with NO conversion during a NO+NH 3+O 2temperature ramp. The drastically lower reactivity of NO in standard SCR, in the absence of ei- ther gaseous NO 2(as in curve (A)) or adsorbed nitrates (as in curve (B)), is clearly apparent.We also note that replacing O 2with gaseous NO 2in the role of V-oxidizing agent, as proposed by Koebel et al.[17]to account for the higher fast SCR rates, cannot explain the tran- sient data of the temperature ramps inFigs. 3 and 5. Instead, V reoxidation must be effected by adspecies stored on the cata- lyst, either nitrates or possibly molecularly adsorbed NO 2. Such adsorbed species act as reservoirs of oxidizing agents, thus per- mitting the significant increment of NO conversion observed as the temperature is raised. 3.7. Role of vanadium redox properties in the fast SCR reaction Data reported in the previous sections showed that buildup of nitrates onto the V-based catalyst plays a crucial role in en- hancing SCR deNO xefficiency. It was suggested that this re- sults from the fact that nitrates reoxidize the vanadium sites more efficiently than gaseous oxygen. However, nitrates also are formed onto sites related to W and Ti[35], the most abun- dant components of the SCR catalyst. Accordingly, to assess the role of vanadium (in particular, its redox properties) in the mechanism of the fast SCR reaction, more dedicated transient runs were carried out on a comparative basis over the V-basedcatalyst discussed earlier, as well as over a commercial V-free WO 3/TiO 2catalyst. The experiments were designed to analyze separately the two consecutive reaction steps—nitrate buildup and their reduction by NO—that previous works identified as the two major stages in the mechanism of the overall fast SCR reaction at low temperatures[23,24]. Fig. 6A shows the experiment performed over the V-based catalyst at 170 ◦C[23]. In the first stage, NH 3and NO 2alone were fed to the reactor, thereby forming and depositing ammo- nium nitrate onto the catalyst as confirmed by the consumption of both reactants and the production of nitrogen, in agree- ment with reaction(9). Formation of ammonium nitrate, as estimated from the lack in nitrogen balance, was also in line with(9). DTA-TG measurements confirmed that the endother- mic decomposition of ammonium nitrate did not start until 173–174 ◦C, slightly above the temperature of the present ex- periment. Thus, the formed NH 4NO 3remained on the catalyst sample.The second part of the run involved removing NO 2from the feed stream. Thus NH 3readily recovered its feed concentration level, and the reaction was no longer observed, even though residual ammonium nitrate was still present on the catalyst. Af- terwards, NO was admitted to the reactor while ammonia was still continuously fed (t=8500 s); NO and NH 3immediately reacted in equimolar amounts, with production of N 2.Butthe reaction also involved consumption of NH 4NO 3and in fact pro- ceeded only until depletion of the ammonium nitrate deposited onto the catalyst.Data from this and other similar experiments[24]have been interpreted assuming that the conversion of ammonium nitrate occurred via the reaction between NO and nitric acid (in equi- librium with ammonium nitrate), which was thus reduced to nitrous acid with formation of NO 2. Then in the presence of ad- sorbed NH 3, nitrous acid would produce N 2via decomposition of NH 4NO 2, whereas NO 2would react readily with ammonia to form more NH 4NO 3and N 2[24]. Such a scheme is in fact in agreement with the observed overall stoichiometry, NO+(1/2)NH 4NO 3→(3/2)N 2+(5/2)H 2O. (10) To analyze the role of vanadium in this global reactivity, an identical experiment was run over the V-free WO 3/TiO 2cat- alyst (Fig. 6B). The first part of this experiment shows that feeding NH 3and NO 2to the reactor resulted in nitrogen pro- duction, similar to what was seen over the V-based catalysts and in agreement with reaction(9), indicating that formation and deposition of ammonium nitrate occurred over the WO 3/TiO 2 catalyst, just as for the V-based catalyst. Accordingly, vanadium seems unnecessary for the formation of ammonium nitrate onto the catalyst surface. A very different picture appears in the second and final part of the experiment, when NO was admitted to the reac- tor (roughly att=8250 s). In contrast to the V-based catalyst (Fig. 6A), on feeding NO to the WO 3/TiO 2catalyst, where ni- trates had been stored previously, here we found no conversion of either nitric oxide or ammonia, or any nitrogen production. This indicates that the V-free sample did not allow reduction of nitrate species by NO, at least under the experimental condi- 8E. Tronconi et al. / Journal of Catalysis 245 (2007) 1–10 Fig. 6. Transient experiments atT=170 ◦C: formation of NH 4NO 3(feed: 1000 ppm NH 3, 1000 ppm NO 2,1%H 2O); reduction of NH 4NO 3by NO (feed: 1000 ppm NH 3, 1000 ppm NO, 1% H 2OinHe).(A)V 2O5–WO 3/TiO 2catalyst; (B) WO 3/TiO 2catalyst. Inset: TPD run performed at the end of experiment (B). tions used here. Notably, the presence of unreacted NH 4NO 3on the V-free catalyst sample was confirmed by an additional TPD experiment performed at the end of this run (seeFig. 6B, inset). Heating the WO 3/TiO 2catalyst in inert atmosphere provoked the decomposition of the residual ammonium nitrate, leading primarily to the evolution of N 2O[11,22,24]. 3.8. A unifying redox scheme for standard SCR and fast SCRHerein we propose a set of reactions that formally summa- rizes the catalytic chemistry underlying the experiments dis-cussed thus far and, by identifying the related role of the vana- dium redox properties, provides a unifying approach to the mechanisms of the standard SCR and fast SCR reactions. Standard SCR: NO+NH * 3+V 5+ =O→N 2+H 2O+V 4+ –OH, (6) V 4+ –OH+(1/4)O 2→V 5+ =O+(1/2)H 2O. (7) Fast SCR: NO+NH * 3+V 5+ =O→N 2+H 2O+V 4+ –OH, (6) V 4+ –OH+NO − 3+H +→V 5+ =O+NO 2+H 2O, (11) E. Tronconi et al. / Journal of Catalysis 245 (2007) 1–109 Scheme 1. Redox catalytic cycles of the standard and fast SCR reactions over V 2O5–WO 3/TiO 2catalysts. V=O and V–OH are oxidized and reduced vana- dium sites, respectively. S=O is a non-reducible oxidic site. 2NO 2+O 2− ↔NO − 2+NO − 3,(12) NO − 2+NH * 3→N 2+H 2O+O 2− +H +. (13) The steps associated with the NO+NH 3+O 2standard SCR, reactions(6) and (7), were discussed in Section4. Reaction(6) accounts for reduction of the V 5+ -sites; reaction(7)is the rate- limiting reoxidation step involving gaseous oxygen.In the case of the NO+NH 3+NO 2fast SCR, the reduc- tion of the V sites still occurs according to the same global reaction(6), but the rate-determining step in the redox process (i.e., the reoxidation of V sites) is radically changed, in this case being carried out by surface nitrates according to reac- tion(11), which replaces the much slower reaction(7).The introduction of nitrates brings about additional complexity to the SCR chemistry, however, so reaction(12)represents forma- tion of nitrite and nitrate ad-species via disproportion of NO 2, whereas reaction(13)accounts for the decomposition of nitrites to N 2via reaction with NH 3. In agreement with the experimen- tal observations shown inFig. 6, no redox catalyst function is involved in steps(12) and (13), which are assumed to occur over nonreducible oxidic sites, represented here schematically as O 2− , possibly associated with the W- or Ti-catalyst compo- nents. A similar sequence has been invoked in the literature to explain the formation of ammonium nitrate from NO 2+NH 3 observed over TiO 2[35]and over zeolites[25],aswellasthe formation of nitrates from NO 2over Al 2O3[39]. The redox cy- cles of the vanadium sites for the standard SCR and fast SCR reactions are schematically depicted inScheme 1. In summary, the proposed redox mechanism:(i) Explains the higher rate of the fast SCR as due to the faster reoxidation of V sites by nitrates (seeFig. 3). (ii) Agrees with the behavior observed during transient runs,when the fast SCR occurred also in the absence of gaseous NO 2due to the involvement of surface nitrates (seeFig. 5). (iii) Is consistent with the reducing action of NO on adsorbed nitrates and ammonium nitrate, observed and reported herein (Fig. 6A) and in previous publications[13,23,24], and explains why this reaction does not proceed effectively on a V-free catalyst (Fig. 6B). (iv) Can account not only for the deNO xreactions, but also for the formation of observed byproducts, such as ammonium nitrate and N 2O, both originating from reactions of sur- face nitrates with ammonia at low and high temperatures, respectively[12,13,23,24]. 4. Conclusion NO+NH 3TPR runs in the absence of oxygen were used to examine mechanistic features of the SCR reaction over an in- dustrial V-based catalyst for diesel exhaust aftertreatment under representative low-temperature conditions. Through such meth- ods, we have confirmed that reoxidation of the vanadium sites by gaseous oxygen is the rate-limiting step in the redox catalytic mechanism of the standard SCR (NO+NH 3+O 2) reaction. However, the main contribution of the present work con- cerns the mechanism of the less well-investigated fast SCR (NO+NH 3+NO 2) reaction. TPR data offer direct experimen- tal evidence that reoxidation of V sites is significantly acceler- ated in the presence of NO 2and/or HNO 3; this explains why fast SCR is so much faster than standard SCR at low temper- atures. Our data further suggest that this is not a direct effect, as was previously proposed in the literature, but rather results from the crucial participation of nitrates adspecies, formed from NO 2, in reoxidizing the reduced V sites and thus in promot- ing the rate of the SCR reaction over V-based catalysts. This aspect is critical to rationalize the transient behavior of the fast SCR reacting system and for SCR dynamic kinetic mod- eling.The present results open the way to a new interpretation of the complete NO/NO 2–NH 3low-temperature catalytic chem- istry involved in both the standard and fast SCR reactions over V-based catalysts. It relies on a unifying redox approach in which the rate-controlling reoxidation of the vanadium catalyst sites involves gaseous oxygen in the NH 3SCR of NO only, but is carried out more effectively by surface nitrates when NO 2is added to the reacting system. Nitrates are formed via dispropor- tion of NO 2in a secondary catalytic cycle that does not require a redox function.To the best of our knowledge, this redox scheme is consis- tent with most experimental observations concerning NH 3-SCR chemistry over V-based catalysts available in the literature. It may prove useful as well to interpret mechanistic data for NH 3- SCR over zeolite-based catalysts. A transient Mars–Van Krev- elen kinetic model in close agreement with the SCR chemistry discussed herein is currently under development. 10E. Tronconi et al. / Journal of Catalysis 245 (2007) 1–10 References [1] ACEA final report on Selective Catalytic Reduction, June 2003,http:// europa.eu.int/comm/enterprise/automotive/mveg_meetings/meeting94/ scr_paper_final.pdf. [2] M. Koebel, M. Elsener, M. Kleemann, Catal. Today 59 (2000) 335. [3] R.M. Heck, R.J. Farrauto, S.T. Gulati, in: Catalytic Air Pollution Control,second ed., Wiley, New York, 2002. [4]http://www.cleers.org. [5]http://www4.mercedes-benz.com/specials/scr/en/scr_start_e.swf. [6]http://www.sae.org/automag/features/futurelook/01-2005/1-113-1-84.pdf. [7]http://www.volvo.com/group/scr/en-gb/. [8]http://www.renault-trucks.it/. [9] P. Forzatti, L. Lietti, E. Tronconi, in: I.T. Horvath (Ed.), Nitrogen Ox-ides Removal—Industrial. Encyclopedia of Catalysis, first ed., Wiley, New York, 2002, and references therein. [10] A. Kato, S. Matsuda, T. Kamo, F. Nakajima, H. Kuroda, T. Narita, J. Phys.Chem. 85 (1981) 4099. [11] M. Koebel, G. Madia, M. Elsener, Catal. Today 73 (2002) 239. [12] G. Madia, M. Koebel, M. Elsener, A. Wokaun, Ind. Eng. Chem. Res. 41(2002) 351. [13] C. Ciardelli, I. Nova, E. Tronconi, B. Bandl-Konrad, D. Chatterjee,M. Weibel, B. Krutzsch, Appl. Catal. B Environ. (2006), in press, doi:10.1016/j.apcatb.2005.10.041. [14] G. Busca, L. Lietti, G. Ramis, F. Berti, Appl. Catal. B 18 (1998) 1, andreferences therein. [15] L. Lietti, P. Forzatti, J. Catal. 147 (1994) 241. [16] L. Lietti, P. Forzatti, F. Bregani, Ind. Eng. Chem. Res. 35 (1996) 3884. [17] M. Koebel, G. Madia, F. Raimondi, A. Wokaum, J. Catal. 209 (2002) 159. [18] I. Nova, C. Ciardelli, E. Tronconi, D. Chatterjee, B. Bandl-Konrad, AIChEJ. 52 (2006) 3222. [19] N.-Y. Topsøe, J.A. Dumesic, H. Topsøe, J. Catal. 151 (1995) 241. [20] V.I. Marshneva, E.M. Slavinskaya, O.V. Kalinkina, G.V. Odegova, E.M.Moroz, G.V. Lavrova, A.N. Salanov, J. Catal. 155 (1995) 171. [21] M. Koebel, M. Elsener, G. Madia, Ind. Eng. Chem. Res. 40 (2001) 52.[22] G. Madia, M. Koebel, M. Elsener, A. Wokaun, Ind. Eng. Chem. Res. 41 (2002) 4008. [23] C. Ciardelli, I. Nova, E. Tronconi, D. Chatterjee, B. Bandl-Konrad, Chem.Commun. 23 (2004) 2718. [24] I. Nova, C. Ciardelli, E. Tronconi, D. Chatterjee, B. Bandl-Konrad, Catal.Today 114 (2006) 3. [25] G.T. Went, L.-J. Leu, S.J. Lombardo, A.T. Bell, J. Phys. Chem. 96 (1992)2235. [26] Y. Yeom, J. Henao, M. Li, W.M.H. Sachtler, E. Weitz, J. Catal. 231 (2005)181. [27] I. Giakoumelou, C. Fountzoula, C. Kordulis, S. Boghosian, J. Catal. 239(2006) 1. [28] C. Ciardelli, I. Nova, E. Tronconi, B. Konrad, D. Chatterjee, K. Ecke,M. Weibel, Chem. Eng. Sci. 59 (2004) 5301. [29] D. Chatterjee, T. Burkhardt, B. Bandl-Konrad, T. Braun, E. Tronconi,I. Nova, C. Ciardelli, SAE technical paper 2005-01-965 (2005). [30] E. Tronconi, I. Nova, C. Ciardelli, D. Chatterjee, T. Burkhardt, B. Bandl-Konrad, Catal. Today 105 (2005) 529. [31] D. Chatterjee, T. Burkhardt, M. Weibel, E. Tronconi, I. Nova, C. Ciardelli,SAE technical paper 2006-01-0468 (2006). [32] R.J. Willey, J.W. Elridge, J.R. Kittrell, Ind. Eng. Chem. Prod. Res. Dev. 24(1985) 226. [33] F. Kapteijn, L. Singoredjo, N.J.J. Dekker, J.A. Moulijn, Ind. Eng. Chem.Res. 32 (1993) 445. [34] I. Nova, L. Lietti, E. Tronconi, P. Forzatti, Catal. Today 60 (2000) 73. [35] J. Despres, M. Koebel, O. Krocker, M. Elsener, A. Wokaun, Appl. Catal.B Environ. 43 (2003) 389. [36] S. Djerad, M. Crocoll, S. Kureti, L. Tifouti, W. Weisweiler, Catal. To-day 113 (2006) 208. [37] G. Piazzesi, O. Krocher, M. Elsener, A. Wokaun, Appl. Catal. B Envi-ron. 65 (2006) 55. [38] G. Piazzesi, O. Krocher, M. Elsener, A. Wokaun, Appl. Catal. B Envi-ron. 65 (2006) 169. [39] N. Apostolescu, T. Schroder, S. Kureti, Appl. Catal. B Environ. 51 (2004)43–50.