Topics in Catalysis Vols. 4243, May 2007 ( 2007) DOI: 10.

1007/s11244-007-0148-4

43

NH3-NO/NO2 SCR for diesel exhausts after treatment: mechanism and modelling of a catalytic converter
Isabella Novaa,*, Cristian Ciardellia, Enrico Tronconia, Daniel Chatterjeeb, and Michel Weibelb
a

Dipartimento di Chimica, Materiali e Ingegneria Chimica G. Natta, Politecnico di Milano, P.zza L. Da Vinci 32, 20133 Milano, Italy b DaimlerChrysler AG Abteilung RBP/C, HPC: 096-E220, D-70546 Stuttgart, Germany

We review herein the key mechanistic and kinetic features of the reactions involved in the NH3-NO/NO2 SCR system investigated by dynamic reactive experiments over a V-based commercial powdered catalyst, eventually leading to the proposal of an original redox scheme which accounts for stoichiometry, selectivity and intrinsic kinetics of the global SCR process. KEY WORDS: dieselurea SCR; SCR; selective catalytic reduction; standard SCR; fast SCR; redox kinetic scheme; modelling; dynamic methods; ammonium nitrate.

1. Introduction The Selective Catalytic Reduction (SCR) using urea as reducing agent is currently becoming the technology of choice for the abatement of NOx emissions from heavy-duty diesel engines [1]. Although several car manufactures [24] are currently considering the SCR technology as a possible solution to meet the emission limits of the forthcoming legislations, few papers are available that discuss modelling of SCR monolith reactors for mobile applications [5, 6]. In recent years a joint project started between Politecnico di Milano and DaimlerChrysler aimed at constructing an unsteady mathematical model of SCR converters applicable to the design and control of integrated aftertreatment systems for mobile applications. The investigation of the NH3NO reacting system over V-based catalysts in the presence of O2 and H2O and the derivation of a related transient kinetic model [7, 8] contributed to the development of the BlueTec technology, commercialized by DaimlerChrysler in February 2005 [1]. The possible presence of DOC devices upstream of the SCR catalyst, however, creates the necessity of including also NO2 among the reactants in the kinetic model. The major benet of this solution is that, in addition to the wellknown standard SCR reaction 1 O2 ! 2 N2 + 3 H2 O 2

occurs over the catalyst, which grants a higher DeNOx eciency at low temperature. The research has then recently focused on the study of the complete NH3-NO/NO2 reacting system, which involves a more complex and less explored chemistry as compared to the NH3-NO/O2 system. In this work we present a unifying summarizing report, including also new results, of our extensive fundamental investigation of the reactivity of NH3-NO/ NO2 over a V-based commercial catalyst [911]: it provides the basis on which a dedicated numerical model of SCR converters has been recently derived and developed.

2. Methods Materials and Experimental All the transient reactive experiments were performed over a single commercial V2O5WO3/TiO2 SCR catalyst with medium-high V content. 160 mg of catalyst powder (140200 mesh), obtained by crushing and sieving the original extruded monolith sample, were loaded into a quartz ow-microreactor (6 mm i.d.) placed into an electric furnace, whose temperature was measured and controlled by means of a thermocouple directly immersed in the catalyst bed. The system was operated at atmospheric pressure with a total ow rate of either 120 or 280 cm3/min (STP), the latter corresponding to a Gas Hourly Space Velocity (GHSV) of about 2.1 105 h)1. NH3, NO, NO2, O2 and He feeds were controlled by mass ow controllers while water vapor was added by means of a saturator. The reactor outlet was directly connected both to a UV analyzer (ABB Limas 11-HW) and to a quadrupole mass spectrometer (Balzers QMS 200), which operated in parallel. Typical feed concentrations were 1000 ppm, for NH3 and NOx (with different NO/NO2 feed ratios), 2% for O2, and 1% v/v for
1022-5528/07/0500-0043/0 2007 Springer Science+Business Media, LLC

2 NH 3 + 2 NO +

also the Fast SCR reaction 2 NH 3 + NO2 + NO ! 2 N2 + 3 H2 O 2

* To whom correspondence should be addressed. E-mail: isabella.nova@polimi.it

44

I. Nova et al./Mechanism of NH3-NO/NO2 SCR

water. Intraparticle gradients and gas-solid mass transfer limitations were ruled out by theoretical criteria. A detailed description of equipment and procedures can be found in [7, 11]. Modeling In order to extract the intrinsic kinetics of the SCR process, a set of transient runs performed over the powdered SCR catalyst was analysed according to a dynamic one-dimensional isothermal heterogeneous PFR model of the micro reactor [7, 8]. The kinetic model was coupled to a global multi-response non linear regression routine [12] and applied to the estimation of the rate parameters in the adopted rate equations. Further details are given in [7, 8, 13].

agreement with results reported by other authors [14, 15], the highest NOx conversion was achieved with a 1/1 NO2/NO feed ratio. Notably, the greatest benets from NO2 addition to the NO/NH3 mixtures were evident in the low T-range (T < 250 C), where the activity of the standard SCR reaction (2) is limited. The gure shows, however, that the presence of NO2 in excess with respect to the optimal 1/1 feed ratio also resulted in a decreased NOx conversion. In fact, NO becomes the limiting reactant of the Fast SCR in this case, and the excess NO2 reacts with ammonia at low temperature to form ammonium nitrate according to 2 NO2 + 2 NH3 ! NH4 NO3 + N2 + H2 O 3

3. Results and discussion The reactivity of the full NH3-NO/NO2 system was systematically investigated by dynamic methods over a wide range of temperatures (160 < T < 425 C) and feed mixtures (0/1 < NO/NO2 < 1/0), with the aim of identifying the main reactions prevailing in the dierent operating regions [11]. Figure 1 shows, for example, the transient data collected at 200 C when feeding 1000 ppm NH3 + 500 ppm of NO + 500 ppm of NO2 (NO/NO2 feed ratio = 1/1) in terms of outlet concentrations of NO, NO2, N2 and NH3 (symbols) versus time. Conversions of roughly 700 ppm of NH3, 350 ppm of NO, 350 ppm of NO2, along with production of about 700 ppm of N2 were measured at steady state. Such values agree with the stoichiometry of the Fast SCR reaction (2). In gure 2 the steady state DeNOx eciencies measured at dierent temperatures upon varying the NO2 feed content (symbols) are displayed. In all cases, in

while for T > 275 C the NO2-SCR reaction 4 NH 3 + 3 NO2 ! 7/2 N2 + 6 H2 O 4

prevails: both these reactions are less eective than the Fast SCR (2). The analysis of the reactivity data provided signicant insight into the chemistry prevailing over V-based SCR catalysts when feeding NH3-NO/NO2 mixtures, which was further extended and validated on the basis of transient response experiments addressing both the overall reactivity in the Fast SCR reaction and also individual reaction steps [9, 10]. One of the most signicant ndings was the identication of the key role of nitrates as intermediates in the Fast SCR: this was clearly demonstrated by a set of specically designed dynamic experiments, during which NH4NO3 was rst formed upon the catalyst surface, and then reduced in the presence of NO. Notably, ammonium nitrate had been regarded so far only as a possible by-product formed by a side reaction in parallel to the Fast SCR

1200

T = 200C
1000

NH3 N2

Concentration (ppm)

100

800 600 400 200 0 0 2000 4000

350C
80

NO2

NOX conversion

60

200C 225C

275C

40

NO
20

175C

N2O
6000 8000 10000 12000
0 0.0 0.2 0.4 0.6 0.8 1.0

Time, s
Figure 1. NOx + NH3 TRM. Feed: 1000 ppm NH3, 500 ppm NO, 500 ppm NO2, 1% H2O, 2% O2, balance He, T = 200 C, GHSV = 210000 h)1. Symbols outlet concentration of NH3 (triangles), NO2 (full circles), N2 (squares) and N2O (stars); solid lines kinetic t.

NO 2/NO x ratio

Figure 2. Measured (symbols) and simulated (lines) steady state SCR NOx conversion at dierent T for varying inlet NO2/NOx ratios. Feed: 1000 ppm NH3, 1000 ppm NOx, 1%H2O, 2%O2, balance He, GHSV = 210000h)1.

I. Nova et al./Mechanism of NH3-NO/NO2 SCR

1800 1600
500 600

45

1400

Concentration (ppm)

1000
300

800 600 400 200 0 3800 3900

T NO NH3 NO2
200

100

N2O
4000 4100 4200 4300 4400

4500

Time (s)

Figure 3. TPR experiment: 1000 ppm NH3, 500 ppm NO2, 1000 ppm NO in He at T = 150 C; T-ramp 150300 C at 20 C/min; GHSV = 92000 h)1.

[10, 11, 16]. One of such experiments is illustrated in gure 3. At 150 C, when 1000 ppm NH3, 1000 ppm NO, 500 ppm NO2 (so with excess NO with respect to Fast SCR stoichiometric feed) were fed to the reactor, the steady state outlet concentrations indicated the simultaneous occurrence of the Fast SCR (2) (25% NO conversion) and of the ammonium nitrate formation (3) (25% NH3 conversion): under these conditions ammonium nitrate was being partially stored onto the catalyst surface, as clearly demonstrated in [911]. Afterwards, at t = 3850 s a T-ramp at 20C/min was started: increasing consumption of NO and production of N2 were observed, meaning that the rate of the Fast SCR reaction (2) was growing. As the T-ramp started, desorption of NH3 was also observed, as well as a small production of NO2. Both such species were then consumed as the temperature was increased further. The analysis of the transient behaviour during the temperature ramp points out that another reaction involving NO consumption must have been occurring in addition to the Fast SCR: with a feed mixture of 1000 ppm NH3, 1000 ppm NO and 500 ppm NO2, the highest possible conversion of NO according to (2) should have been 50%, with a production of 1000 ppm of N2 (the dashed lines in gure 3 indicate such values). Nevertheless, the measured NO conversion exceeded 50%, reaching a peak of about 80% with a simultaneous production of 1600 ppm of N2. Such a behaviour is explained by invoking the reaction between ammonium nitrate (which, as mentioned above, had been previously formed and stored onto the catalyst surface) and NO, according to the following stoichiometry NH 4 NO3 + NO ! NO2 + N2 + 2 H2 O 5

Temperature (C)

1200

N2

400

t = 4150 s, and indeed at the end of the experiment steady state concentration values of reactants and products were eventually in agreement with the occurrence of the Fast SCR reaction (2) only. It appears thus that the Fast SCR reaction was proceeding only before t = 4000 s while, as it became limited by the NO2 feed concentration, reaction (9) occurred afterwards. However, no discontinuity and no change of slope can be observed in the neighbourhood of t = 4000 s, thus indicating that reaction (2) and reaction (9) share the same rate as well as the same apparent activation energy. The same conclusion was also achieved performing a transient experiment at constant temperature (170 C) [9, 10]: the NO conversion levels measured either when the Fast SCR reaction was occurring or when the reaction between NO and surface nitrates was occurring were practically unaltered, indicating that the two reactions were actually progressing at the same rate. Accordingly, it can be ruled out that the reaction between NH4NO3 and NO and the Fast SCR reaction proceed in parallel, and an alternative, sequential scheme is proposed, whereby the Fast SCR reaction (2) results from the formation of surface species related to NH4NO3, reaction (3), and their subsequent reduction by NO, reaction (5), which is rate determining at low temperature. Thus, the roles of the reactants in the overall reaction scheme of the Fast SCR can be summarized as follows: (i) NO2 is responsible for the formation of nitrate and nitrite adspecies, likely via disproportion and heterolytic chemisorption, 2 NO 2 + H2 O , HONO + HNO3 6

(ii) NO acts as a reductant, converting nitrates to nitrites, HNO 3 + NO , HONO + NO2 7

(iii) NH3 eventually reacts with nitrites to give harmless nitrogen and H2O via decomposition of the unstable ammonium nitrite, HONO + NH 3 ! N2 + 2 H2 O 8

(iv) in defect of NO, NH3 can also react with nitrates to form ammonium nitrate, which, under specic lowT conditions, builds up onto the catalysts. NH 3 + HNO3 , NH4 NO3 9

The depletion of NH4NO3 is responsible for the subsequent drop of NO conversion observed after

A similar reaction scheme has been recently proposed on the basis of an IR study of the NO/NO2-NH3 reactivity over a BaNaY zeolite SCR catalyst [17]. It is worth emphasizing that the reaction scheme above is able to explain not only the stoichiometry of the Fast SCR reaction, and specically the optimal

46

I. Nova et al./Mechanism of NH3-NO/NO2 SCR

equimolar NO to NO2 feed ratio, but also the selectivity to all of the observed products, namely N2, NH4NO3 and N2O (which derives from thermal decomposition of ammonium nitrate [911]); furthermore it is in agreement with the observed kinetics of the Fast SCR reactions, which at low temperature would be limited by the rate of the reaction between nitrate and NO. In additional dedicated mechanistic experiments we found that, contrary to the NH4NO3 formation, the rate limiting step (i.e. reaction (5) or (7)) as well as the global Fast SCR reaction (2) did not proceed over a V-free WO3/TiO2 catalyst, thus identifying the catalytic role of vanadium redox sites in the Fast SCR mechanism. It was also proved that surface nitrate species, formed from NO2, are much more eective than gaseous oxygen in reoxidizing the reduced V-sites, which explains the greatly enhanced DeNOx activity with respect to the standard SCR at low temperature. Accordingly, an original, unifying mechanistic redox scheme could be proposed for both Standard and Fast SCR reactions, i.e. for the global NH3-NO/NO2 SCR process [18]. On the basis of such a reaction chemistry a detailed kinetic model was derived, whose intrinsic rate parameters were estimated from global nonlinear regression of a set of 42 transient runs collected on varying temperature, reactant concentrations and NO/NO2 feed ratio, according to the procedures described in [7, 8]. The good quality of the resulting t is illustrated in gures 12, where the calculated (lines) and measured (symbols) temporal evolutions of NH3, NO, NO2, N2 and N2O are compared. Figure 2 also conrms that the proposed rate equations are able to represent eectively the observed inuence of the NO2 feed content on the whole product distribution, including N2, N2O and HNO3. To our knowledge, no such a comprehensive mechanistic kinetic description of the NH3 + NO/NO2 reacting

system has been reported previously. On a methodological basis, it is also worth emphasizing the informative power demonstrated herein by the use of dynamic methods in elucidating the mechanism of catalytic reactions. References
[1] http://www.mercedes-benz.com/specials/scr/en/scr_start_e.swf [2] http://www.sae.org/automag/features/futurelook/012005/1113 184.pdf [3] http://www.volvo.com/group/scr/en-gb/ [4] http://www.renault-trucks.it/ [5] C. Winkler, P. Flo rchinger, M.D. Patil, J. Giesho, P. Spurk and M. Pfeifer, SAE 2003-01-0845 (2003). [6] A.P.E. York, T.C. Watling and J. P. Cox et al., SAE 2004-01-0155 (2004). [7] C. Ciardelli, I. Nova, E. Tronconi, B. Konrad, D. Chatterjee, K. Ecke and M.Weibel, Chem. Eng. Sci. 59 (2004) 5301. [8] D. Chatterjee, T. Burkhardt, B. Bandl-Konrad, T. Braun, E. Tronconi, I. Nova and C. Ciardelli, SAE paper 2005-01-965 (2005). [9] C. Ciardelli, I. Nova, E. Tronconi, D. Chatterjee and B. BandlKonrad, Chem. Commun. 23 (2004) 2718. [10] I. Nova, C. Ciardelli, E. Tronconi, D. Chatterjee and B. BandlKonrad, Catal. Today 114(1) (2006) 3. [11] C. Ciardelli, I. Nova, E. Tronconi, B. Bandl-Konrad, D. Chatterjee, M. Weibel and B. Krutzsch, Appl. Catal. B Envir. 70(14) (2007) 80. [12] P.L. Villa, P. Forzatti, G. Buzzi-Ferraris, G. Garone and I. Pasquon, Ind. Eng. Chem. Prod. Res. Dev. 24 (1985) 12. [13] D. Chatterjee, T. Burkhardt, M. Weibel, T. Braun, E. Tronconi, I. Nova and C. Ciardelli, SAE Technical Paper 2006-01-0468 (2006). [14] A. Kato, S. Matsuda, T. Kamo, F. Nakajima, H. Kuroda and T. Narita, J. Phys. Chem. 85 (1981) 4099. [15] M. Koebel, G. Madia and M. Elsener, Catal. Today 73 (2002) 239. [16] G. Madia, M. Koebel, M. Elsener and A. Wokaun, Ind. Eng. Chem. Res. 41 (2002) 351. [17] Y.H. Yeom, J. Henao, M. Li, W.M.H. Sachtler and E. Weitz, J. Catal. 231 (2005) 181. [18] E. Tronconi, I. Nova, C. Ciarolelli, D. Chatterjee and M. Weibel, J. Catal, 245(1) (2007) 1.