Chapter 5 A there-function model reaction for designing DeNOx catalysts

Chapter 5 A there-function model reaction for designing DeNOx catalysts

Chapter 5 A THREE-FUNCTION MODEL REACTION FOR DESIGNING DeNOx CATALYSTS G. Dj6ga-Mariadassou a'*, M. Berger b, O. Gorce c, J. W. Park d, H. Pernot a, ...

4MB Sizes 0 Downloads 51 Views

Chapter 5 A THREE-FUNCTION MODEL REACTION FOR DESIGNING DeNOx CATALYSTS G. Dj6ga-Mariadassou a'*, M. Berger b, O. Gorce c, J. W. Park d, H. Pernot a, C. Potvin a, C. Thomas a and P. Da Costa a aLaboratoire R~activitg de Surface, CNRS UMR 7609, Universitd Pierre et Marie Curie, 4 Place Jussieu, Case 178, 75252 Paris Cedex 05, France bGaz de France, Direction de la Recherche, 361 av. President Wilson, B.P. 33, 93211 La Plaine Saint Denis, Cedex, France CRenault sas, Centre Technique de Lardy, 1 all~e Cornuel, 91510 Lardy, France dOn leave for Korea *Corresponding author: Laboratoire R~activitg de Surface, CNRS UMR 7609, Universit~ Pierre et Marie Curie, 4 Place Jussieu, Case 178, 75252 Paris Cedex 05, France. Tel.: -t-33 (0)1 44 27 36 24, Fax.: +33 (0)1 44 27 60 33, E-mail: gerald, dje ga_mariadassou @ upmc.fr (dje ga. gerald @wanadoo.fr)

Abstract A three-function catalyst model for hydrocarbon SCR of NO x is described, based on experimental evidence for each function, during temperature-programmed surface reactions (TPSR). The release of N: occurs within function 3. It involves the dissociation of NO (via a dinitrosyl-adsorbed intermediate), followed by subsequent formation of N e and scavenging of the adsorbed oxygen species left from NO dissociation. The removal of adsorbed oxygen is due to the total oxidation of an activated reductant (CxHyOz). This reaction corresponds to 'a supported homogeneous catalytic process' involving a surface transition metal complex. The corresponding catalytic sequence of elementary steps occurs in the coordinative sphere of the metal cation. A function 2 has to turn over simultaneously to function 3. It has to deliver the active reductant species CxHyOz to function 3, at the temperature where function-3 cycle tums over. Function 2 is the mild oxidation of HC (or any initial oxygenate) by NO:, through organic nitrogen-containing intermediates (RNOx). The very important feature is that these RNO x species are quite thermally instable: they decompose with temperature to CxHy0 z + NO (not to Ne), according to the following global equation: HC

(or CxHyOz) + N O 2 = Cx, Hy, O z, + N O

It is therefore obvious that functions 2 and 3 have to turn over simultaneously. Nevertheless, at the molecular level, these two functions are not in the same catalytic cycle.

Past and Present in DeNOx Catalysis

P. Granger and V.I. Pdrvulescu (Editors)

9 2007 Elsevier B.V. All rights reserved.

146

Past and Present in DeNO x Catalysis

A function 1 has also to turn over simultaneously with functions 2 and 3, as it has to provide NO2 to function 2. The oxidation of NO to NO2 is therefore the first function of any efficient catalyst. The concept of 'simultaneous turn over' between catalytic functions in multi-functional catalysis is widely accepted (for instance, in bi-functional metal/acid transformation of alkanes), and this aspect of the proposed mechanism is a 'normal' behaviour in steady state.

Keywords Three-functioncatalyst; oxygenates; dinitrogen formation; supported homogeneous catalysis; metal cation active sites.

1. I N T R O D U C T I O N There can be some apparent contradiction in the general concept of NO reduction to N 2, in the presence of an excess of oxygen, according to the following global equation: 2 NO + Reductant + 02,excess, - - N 2 + C O / C O

2+

H20

(1)

In lean conditions, reductants are found to be either hydrocarbons (HCs) or, in some industrial processes or academic studies, some more active oxygenates CxHyOz. It should be recognized that the global reaction (1) is confusing, as nitrogen is globally reduced in oxidizing atmosphere, its formal oxidation state going from '+II' (in NO) to '0' (in N2). Nevertheless, at a molecular level, there are several assumptions explaining this contradiction. In any case, the key of the problem still remains: H o w is dinitrogen (N2) f o r m e d ? This question has to be answered to completely understand the DeNOx process, and design the final efficient catalyst, according to the nature of reductant and the experimental conditions (more particularly, temperature window). Three main approaches can be found in literature to describe the N - N bond formation: (1) An intermediate organic nitroso compound 'RNOx' is formed, leading to N 2 during its decomposition [1-5]. The mechanistic studies by Sachtler and co-workers [1-4] for the reduction of NOx by light alkanes over Fe/ZSM-5 involved adsorbed RNOx species which further react with gas-phase NO x to produce N 2, through the decomposition of diazo compounds [2,4]. (2) A ' N - N ' bond pre-exists in RNOx compound, which then decomposes, leading to N 2. Over zeolite-supported Ag and Cu catalysts, the formation of nitrosonium and N-nitroso-N-alkylhydroxylamate intermediates have been assumed by Martens et al. [6], Brosius and Martens [8] and Kharas [7], respectively, to explain the N - N bonding before the release of molecular N 2. In fact, the authors [6 and 7] also explain the formation of N 2 on a mechanism that involves organo-nitrogen (RNOx) intermediates, but the pairing of nitrogen atoms can involve different reaction pathways, through the formation of unstable diazonium compounds or by the transformation of the initially formed organo-nitrogen molecules to ammonia, and reduction of NOx via an ammonia-SCR process [8]. So, they did not seem to have evidences for RNO x compounds with pre-existing N - N bonds (as shown

Three-Function Model Reaction

147

in the Figure 5 of [8]). The proposed intermediates can have two atoms of N, but they are generally not directly linked. Finally, Iwamoto and Takeda [9] assumed that the decomposition of RNOx species leads to the formation of oxygenated compounds CxHyOz which could be the intermediates of the global DeNOx process. This last assumption corresponds to a fundamental step of the model described in this chapter. (3) A dinitrosyl (NO)2 species forms, whose dissociation leads to N2; the remaining adsorbed oxygen species have to be scavenged by the activated reductant, CxHyO z to recover the free active site, permitting the catalytic cycle to turn over (Figure 5.1, function 3) [ 10]. At a molecular scale, a three-function model can be defined with no direct interaction between the so-called 'reductant' and NO. In this case, the apparent contradiction to the global equation (1) can be ruled out, and the DeNO x reaction by itself occurs owing to the third function of the model (Figure 5.1).

2. GLOBAL PRESENTATION OF THE THREE-FUNCTION MODEL [10-13] 2.1. General concepts and definitions The present model deals with a supported transition metal cation which is highly dispersed, at the molecular scale, on an oxide, or exchanged in a zeolite. In the case of zeolite-supported cations, the formation of different metal species in metal/zeolite catalysts (metal oxides, metal oxocations, besides cationic species) has been considered by different authors who have suggested these species to play key roles in SCR catalysis [14,15]. This supported cation can also be considered as located at a metal oxide/support interface. The important feature is the formation of a coordinatively unsaturated site (cus), permitting the reaction to occur in the coordinative sphere of the metal cation. The cus is a metal cationic site that is able to present at least three vacancies permitting, in the DeNOx process, to insert ligands such as NO, CO, H20, and any olefin o r CxHyO z species that is able to behave like ligands in its coordinative environment. A cus can be located on kinks, ledges or comers of crystals [16]; in such a location, they are unsaturated. This situation is quite comparable to an exchanged cation in a zeolite, as studied by Iizuka and Lundsford [17] or to a transition metal complex in solution, as studied by Hendriksen et al. [ 18] for NO reduction in the presence of CO. The first pathway proposed by Iizuka and Lunsford [17] considers the reduction of nitric oxide by CO over rhodium/Y-zeolite. It leads only to N20 as follows: Rh I (CO)2 "~-NO ~ Rh I (CO)2 (NO) Rh I (CO)2 (NO) 4- NO ~ Rh I (CO)-(v) 4- N20 4- CO2 where (v) is an oxygen vacancy in the coordinative sphere of the metal.

148

Past and Present in DeNO x Catalysis

Complexes such as RhI(CO)(NO) and RhI(NO)2 were also considered. In solution chemistry, at low temperature, Hendriksen et al. [18] assumed that the following sequence can occur, leading again to N20 formation.

RhI (CO)2 -~-2 NO ~ RhIII(co) (NO-)2 -~-CO RhIII(CO) (NO-)2 ~ RhI-~- N20 + CO 2 Cataluna et al. [19] also proposed the formation of N20 over ceria alone as follows: 2[Ce3+-(v)l + 2 NO ~-- Ce4+-(ONNO)2--fe 4+

Ce4+- (ONNO)2--Ce4+ ~-- (Ce4+-O2-) -+-[Ce4+(v)] -+N20 where (v) stands again for an oxygen vacancy of ceria. Let us note that in these three examples, the reaction leads to N20 and not to N 2 due to either low-temperature experiments or low catalyst activity. As can be seen, the catalytic process over a zeolite-supported cation, or an oxidesupported cation, can be considered as a supported homogeneous catalysis, as far as adsorbed reactants and products behave like reactive ligands. The model developed for lean DeNO x catalysts over supported cations (function 3), as well as this supported homogeneous catalysis approach, is also suitable for stoichiometric mixture (TWC) comprising CO and H 2 as reductants over supported transition metal cations [20-22].

2.2. Metal active sites We shall mainly consider, in the present chapter, non-precious transition metals, but the model can be extended to precious metals presenting an oxidation state higher than zero [10,11], such as Rh x+, Pd x+, Pt x+ and Ir x+. The model also applies to some oxides alone, such as ceria (CeO2) [19] or mixed oxides such as ceria-zirconia (CeZrO2) able to present redox properties and oxygen vacancies during catalytic reactions.

2.3. The three-function catalytic system To be active in DeNOx reaction, a catalyst has to present two functions to assist function 3, in which the N - N bonding and N 2 releases occur (Figure 5.1). This figure presents the three-function design of a DeNO x catalyst [12,13]. According to this model, the oxygenated intermediates (CH3OH, HCHO) produced from initial HC (CH4) suffer total oxidation in function 3. One of the important targets of DeNOx reaction is that the catalyst has to produce by itself these oxygenates; they correspond to a mild oxidation of n c tO CxnyO z (function 2). The direct (faster, parallel or successive) total oxidation of HC to [CO/CO2 + H20] has to be avoided in function 3. The existence of these CxHyO z is in agreement with Iwamoto and Tanaka's conclusions [9]. If the total oxidation occurs in function 3, then the system will lose a great part of reductant. Let us note that the NO 'reduction' by oxygenates (CxHyOz) is s t o i c h i o m e t r i c .

149

Three-Function Model Reaction NO2 CH3OH CO 2 NO2 CH:~. HCHO 2 NO N2 H20

NO

1:NO+O 2~NO 2 2:NO 2+HC=>CxHyOz NO Oxidation

HC Partial oxidation

3:2 NO + Mx+ + CxHyOz=> N2 + xCO 2 + y/2H20 + M x§ % DeNO x

Figure 5.1. The three-function model for designing DeNOx catalysts in the presence of methane as reductant [12]. So, we do not need a very high quantity of oxygenates. This can be experimentally demonstrated in the absence of the total HC oxidation reaction, by removing dioxygen in the feed. Furthermore, lower the reaction temperature, lower will be the total oxidation of the reductanc In contrast, lower the reaction temperature, higher will be the production of N20, escaping from the catalytic cycle of third function. As always, there is some compromise, and all the work consists in adjusting these parameters.

2.4. Global approach of function 3 Function 3 will be first globally described through two main catalytic stages [10-12]. (1) Stage 1 of function 3: dissociation of NOads tO Nads and

O.d s

It occurs via two adjacent adsorbed NO molecules, leading to an adsorbed 'dinitrosyl' species. These last two co-adsorbed NO species made the two N - O bonds weaker, and the successive two N - O bond scissions led to N2. According to a general kinetic model [ 12], the NzO intermediate can desorb before dissociating to N 2, if the desorption rate constant, kdes, is higher than the reaction (dissociation) rate constant, kreaction , as presented in the following set of rate constants (Figure 5.2): These rate constants are representative of the 'rake mechanism' of Germain [24]. Let us note that there is no need for any 'organic nitroso intermediate' for N2 formation. Therefore, two adsorbed oxygen species 'Oads' (ex-NO) are remaining, strongly adsorbed kdesorption I J k formation ib

Adsorbed intermediate

Intermediategas phase

k~ 9

Product

Figure 5.2. Set of rate constants for the catalytic activity of a reaction intermediate [13].

150

Past and Present in DeNO x Catalysis

on the catalytic site where function 3 is occurring, inhibiting any further NO adsorptiondissociation. (2) Stage 2 of function 3 If the t w o Oad s species are not scavenged, then the reaction will stop. This is the case, for instance, of NO decomposition on Cu/ZSM-5 [25]. Adsorbed oxygen species have to be scavenged either by an activated form of the initial 'HC' reductant, such as CxnyO z (alcohol, aldehyde, etc.) or by the initial HC if their total oxidation is simultaneous with NO decomposition-reduction to N 2. These 'oxygenates' and/or HC suffer a total oxidation to CO/CO 2 and H20, regenerating the active site: this is the principle of catalysis. Once the active site is recovered, the reaction continues to turn over. This is the 'catalytic cycle'. These two stages have to be included in function 3 of the catalyst (Figure 5.1) during the design of any DeNOx catalyst: they give a key to the global NO reduction and N 2 formation.

2.5. Global approach of function 2 The catalyst has to get two more functions to assist the third one. It has to produce by itself, the active reductants, i.e. CxHyO z (CH3OH/HCHO in the case of Figure 5.1) by activating the initial HC of the feed. This mild oxidation of HC occurs owing to the presence of NO2, actually well recognized as a good oxidant of HC at relatively low temperatures. As dioxygen generally provokes the HC total oxidation to CO/CO 2 at quite higher temperatures, there is a new target for designing a good DeNOx catalyst: the lower the temperature of mild oxidation, when compared to the temperature of HC total oxidation, the most efficient will be the catalyst. In some cases, the total oxidation of HC is occurring simultaneously with both the CxHyO z oxidation and the decomposition of NO to N 2.

2.6. Global approach of function 1 Figure 5.1 shows that function 1 is therefore the oxidation of NO to NO2, this last one being subsequently delivered to function 2 to oxidize HC to CxHyO z which will be delivered to function 3 to scavenge the adsorbed oxygen species left by the (NO)2 adsorbed dinitrosyl species- decomposition. In order to experimentally demonstrate the model, this chapter will give evidence for each of the three functions. Once convinced by the model, the target will be to design a catalyst permitting the three functions to run simultaneously. It is finally easy to understand why this simultaneity of functions 2 and 3 can lead to a complete misunderstanding of the DeNO x reaction. As function 2 is producing RNOx compounds, simultaneously with function 3 which is producing N 2, it is rather difficult to discriminate between the elementary steps leading to either RNO x or N 2. The present chapter will demonstrate that, in lean conditions, RNOx leads to [CxHyO z + NO] and N 2 is formed in another catalytic cycle.

Three-Function Model Reaction 3. M A I N F E A T U R E S

151 OF EXPERIMENTAL

CONDITIONS

3.1. Model catalyst preparations Various catalytic systems obey the present general model and will be used in this chapter. References to their synthesis are given in the text. The three-function model introduced in the preceding section has been established on an H-mordenite (HMOR) supported cobalt-palladium catalyst [12]. For the sake of demonstration, model catalysts with a unique function, i.e. F1, F2 or F3, (Figure 5.1), were prepared to separately give evidence of the major role of each active site (Figure 5.1). Let us note that 'three functions' does not necessarily mean three different active sites, but in the case of CoPd/HMOR material, three different sites were identified. Exchanged mordenites (CoPd/HMOR, Co/HMOR, Pd/HMOR) were purchased from the 'Institut R6gional des Mat6riaux Avanc6s (I.R.M.A.)', located in Ploemeur (France). They were prepared according to Hamon et al.'s patent [23] by exchanging a NH 4mordenite with the appropriate amount of metallic precursors, respectively, cobalt(II) acetate and Pd(NH3)4C12. Two kinds of pre-treatment were subsequently applied: 9 2 h at 773 K (500~ in flowing nitrogen or argon (inert gases), with a flow rate of 100 mL min -~ . 9 4 h at 573 K (300~ followed by 4 h at 773 K (500~ under synthetic air. The flow rate was 100mLmin -~ and the increasing temperature rates 1 Kmin -~.

3.1.1. The complete CAT I, three-function catalyst- C o P d / H M O R (Figure 5.3) It was obtained by a pre-treatment of fresh impregnated HMOR in flowing air, up to 773 K. In these conditions, as detected by TEM, EDS and UV-visible spectroscopy (not shown, [12]), a fraction of Co 2+ species, exchanged in the pores of HMOR, migrates on the outside of the zeolite grain, to form Co304 on the external surface of the HMOR grain. The material being impregnated by the palladium precursor, once all the internal exchanged positions have been already occupied by Co 2+, PdO is formed on the external surface of the zeolite grain, as observed by TEM (not shown, [12]).

3.1.2. 'Cat H' (Figure 5.3): Co/HMOR pre-treated in flowing air It corresponds to the cobalt initially exchanged into the HMOR porosity. Nevertheless, a fraction of cobalt oxide - C 0 3 0 4 is produced after calcination, as previously seen in the case of Cat I, on the surface of zeolite grains.

3.1.3. Cat III (Figure 5.3): Co/HMOR pre-treated in flowing argon It was prepared by a thermal pre-treatment of the fresh exchanged material, in flowing argon. As a consequence, it maintains all the Co 2+ pre-exchanged species in their exchange position, inside the pore of HMOR. As it will be seen hereafter, the comparison

152

Past and Present in DeNOx Catalysis

Cat I" Co-Pd/HMOR Pre-treated in air nlvlUn

I

:::

Cat I1: Co/HMOR Pre-treated in air

t

nlvlurl

]

Cat II1" Co/HMOR Pre-treated in ar

Cat IV: Co304/SiO 2 Cat V: PdO/SiO 2 Si02

sio~

.... :

Figure 5.3. Schemes of the five model catalysts used in the present study [12].

of the catalytic activity of Cat II and Cat III gives a clear evidence of the role of external cobalt oxide Co304 [12].

3.1.4. 'Cat IV':

C0304supported

over silica

(Co304/Si02)

It was prepared by incipient wetness impregnation using Co(NO3)2, 6H20, 99.9%, as cobalt precursor. Silica Aerosil 380 was purchased from Degussa. The sample was then dried at 573 K (300~ and subsequently calcined at 773 K (500~ under synthetic air. The quantity of cobalt introduced was 2 wt.%.

3.1.5. 'Cat V': PdO supported over silica (PdO/Si02) It was prepared by incipient wetness impregnation using Pd(NH3)4C12. The sample was then dried at 573 K (300~ and subsequently calcined at 773 K (500~ under synthetic air. The amount of palladium introduced was 0.5 wt.%.

3.2. Model catalyst major role In order to determine the major catalytic activity of the preceding model catalyst, in the three functions of the model, the three reactions were studied separately on each catalyst (Table 5.1). The comparison of the results permits to identify the most active site, for each function, when the complete 'Cat I' CoPd/HMOR catalyst is working.

Three-Function Model Reaction

153

Table 5.1. Major role and function of model catalysts [10,12] Model catalyst

Co/HMOR (Cat III) PdO/SiO2 (Cat V) Co304/8iO 2 (Cat II, Cat IV)

Feed

Major role

Main function

NO/He 1

NO adsorption HC oxidation NO to NO2

F3 F2 F1

CH4/O22 NO/O23

11000ppm NO in helium, flow rate: 50cm 3 min -1, T = 35~ (308 K) 2CH4 400ppm+ 5 vol.% 02 in helium. VVH: 62 000h -1 3400ppm NO + 5 vol.% 02. VVH: 62000h -1

3.2.1. Function 3 Table 5.1 shows that the major NO adsorption is occurring over Co/HMOR (Cat III) when compared to PdO/SiO2 and C o 3 0 4 / S i O 2. This adsorption on Co cationic sites, located inside the zeolite (Cat III) leads [during temperature-programmed desorption (TPD) studies (Figure 5.5a, [12])] to the desorption of NO, and production of both N20 and N 2. The formation of N 2 (function 3) shows that the reduction of NO to N 2 goes through its decomposition, with N20 being the adsorbed intermediate, able to suffer a desorption at low temperature according to Figure 5.2. 3.2.2. Functions 2 and~or 3

The oxidation of methane (functions 2 and/or 3) is mainly occurring over PdO when compared to the activities of [Co304 -3t- Co2+/HMOR] (Cat II) and Co2+/HMOR (Cat III). 3.2.3. Function 1

The major oxidation of NO to NO2 is occurring over C0304 (Cat IV) when compared to the activities of C02+/HMOR (Cat III), and PdO/SiO 2 (Cat V).

3.3. Catalytic activity measurements This chapter reports the results from transient experiments (mainly, TPD or TPSR) coupled with on-line analysis of reaction mixture at the outlet of a well-stirred reactor. It means that the gas composition detected at the outlet of the reactor is in contact with the catalyst inside the reactor. Catalytic runs in isothermal conditions were also proceeded in order to avoid strong adsorptions of reactants or intermediates. Catalytic runs were performed after a pre-treatment of catalysts in flowing hydrogen, synthetic air or nitrogen, at 773 K. After returning to room temperature (RT) and flowing the following reaction mixtures: 9 NO (400ppm), CH 4 (400ppm), 8vo1.% 02, in the case of DeNOx over CoPd/HMOR [12], 9 200ppm NO/9vol.% O2/1000ppm CHsOH, total flow rate: 250cm 3 min -1, VVH: 50000h -1, in the case of DeNO x in the presence of methanol over C0/A1203) [26],

154

Past and Present in DeNO x Catalysis

Figure 5.4. Catalytic device for DeNO x reaction coupled with non-thermal plasma.

TPSR were carried out from RT to 770 K, with a heating rate of 10 Kmin -1. Before TPD, catalysts were pre-treated as above, then flushed by reactant (generally NO in either N 2 or O2), and flushed by N 2 to remove any physisorbed species at RT. Experiments were carried out in a U-type quartz reactor. The sample (0.025-0.2 g) was held between plugs of quartz wool and the temperature was monitored through a WET 4000 or Eurotherm 2408 temperature controllers. Reactant gases were fed from mass flow controllers (Brooks 5850TR). The reactor outflow was continuously analyzed by a set of detectors (Figure 5.4): 9 A 'NOXMAT CLD 700 AL' or a 'Thermo Environmental Instruments 42CHL NO x Chemiluminescence analyzer' for NO, NO2 and NO X (NO + NO2). 9 An Ultramat 6E IR analyzer to monitor N20. 9 An Ultramat 6E to follow CO/CO2. 9 A Fidamat 5E-I to follow the total concentration in HC (and CxHyOz) with a FID detector. 9 A GC/MS equipped with a GC 6890N, Agilent Technologies, using a 'Fused SilicaCP7351' Varian Column, a Chemstation, Agilent Technologies and a Mass Spectral Libraries NIST Rev. D.03.00, Agilent Technologies for analyzing all organic and organic nitroso compounds. 9 A micro-GPC 'G2890A' from Agilent, for quantitatively measuring N 2. Signals from detectors were monitored on-line by Virtual Bench-Lab View computation programs or labVIEW 7 programs. Treatments of data were done using Origin 6.1.

155

Three-Function Model Reaction

The limits of detection were considered to be 1 ppm for NOx and HC, 3 ppm for N20 and less than 5 ppm for CO/CO2.

4. EVIDENCE FOR ELEMENTARY STEPS OF FUNCTION 3 ON CoZ+/HMOR (CAT III) Function 3 can be summarized as follows:

(1) NO dissociation, via an adsorbed dinitrosyl '(NO)2' leading to N 2 formation and desorption + 20ad s (ex-NO) (2) Recovering of the active site by scavenging of Oads species by HCxOy (total oxidation to CO/CO2)

4.1. 'NO' dissociation to N2, during TPD of NO pre-adsorbed at RT on the active site of the third function catalyst: supported Co 2+ on HMOR (Cat III, function 3 alone) selected for the sake of demonstration [12] Figure 5.5a corresponds to a catalyst presenting only supported C o 2-t- o v e r HMOR, the active site of function 3 [10,12]. It shows that the pre-adsorbed NO is able to suffer a decomposition/reduction to N 2 at two different temperatures: N 2 is clearly detected at about 100~ and N 2 nt- N 2 0 between 280 and 370~ The reduction of NO to N 2 is already done in these conditions:

2 NOads = N 2 + 20ad s

(2)

'There is no need for a reductant for this step'. It can be seen that methane is not consumed: it plays the role of an inert gas. This result corresponds to already published data of Li and Armor [25]. As can be seen in Figure 5.5a, the N20 intermediate can also desorb before dissociating, even at high temperature. The reason is that the temperature is too low to get the scission of the second and remaining N - O bond of N20 for obtaining N2: adsorbed N20 desorbs before reacting to give N 2 (Figure 5.2) [13]. Only a fraction of adsorbed N20 goes to N 2.

Past and Present in DeNO x Catalysis

156

(a) 2.5

I

----4--- NO

I '

2.0

"~ Zone 1 J l !

1.5

I TPD

E o~

Nc "

I - * - N2

I Zone21 i

Zone3

1.0

o~ 0.5 r 0.0 Temperature~C

I

(b) 600 E 500

TPSR

:

................. i

400

NO

',

................~L........

O

T(TPD) = T(DeNOx) =

I-i ...................

!! NO

........:

-E 300 O O

cO 200 O

100 i

0

~ - r

-

100

,

|

200

,

i

300

-t

-

i

i

400

I

500

Temperature/~ Figure 5.5. (a) Temperature programme desorption (TPD) of NO pre-adsorbed alone at RT in the absence of oxygen (function 3 alone: Co/H-MOR 'Cat III') (b) Temperature programme surface DeNOx reaction (TPSR) over a complete three-function catalyst CoPd/HMOR 'Cat I' (NO (400 ppm), C H 4 (400 ppm), 8 vol.% 02, 93 000h -1, Heating rate TPSR:10 Kmin -1) [12].

The elementary steps for this first part of catalytic reaction are:

o"i (1) Dinitrosyl formation on the metal cation 2 NO + * ~ N O * N O (2) 1st N - O bond scission of dinitrosyl, 1st remaining Oads N O * N O ~ N 20* O (3) 2nd N - O bond scission, 2nd remaining Oads N20* O ~ N 2 + O * O Balance : 2 NO = N 2 -+-O * O

where * is the cation cus, and tri the stoichiometric number of i step. Let us note that N2Oao s c a n desorb before dissociation (Figure 5.2) [12,13]. These elementary steps perfectly represent the phenomena observed on Figure 5.5a.

Three-Function Model Reaction

157

It is important to note that the temperature of NO desorption on 'Cat III' (F3 alone), at 340~ (613 K), corresponds to: 9 The temperature of thermal activation of NO on the active site (significance of TPD experiments, Figure 5.5a). 9 The temperature where NO is able to dissociate on the Co 2+ active site (cus). 9 The temperature of DeNOx reaction (function 3): comparing Figure 5.5a (NO TPD, 'Cat III') to Figure 5.5b - TPSR in the presence of a three-function catalyst (CoPd/HMordenite, 'Cat I'), in a complete flowing feed NO/HC(CH4)/O2 (excess)the temperature of DeNOx is that of the NO thermal desorption. According to the model, the catalyst will have to produce CxHyOz (CH3OH, HCHO) (function 2) to proceed to the DeNO~ process, as discussed in Section 4.2. 9 The prediction of the DeNO~ temperature range can therefore be done and it should correspond to the temperature at which the HC is activated by a mild oxidation to CxHyOz [109] (see Section 4.2). It will be also verified, in this chapter, on other catalytic systems, that the temperature at which NO dissociates is the temperature of the DeNOx reaction (see Section 7).

4.2. The catalytic cycle of function 3 is able to turn over, when flowing the activated form of HC (CxHyOz" alcohol, aldehyde, etc.) according to the model of Figure 5.2 4.2.1. Case o f CoPd/HMOR (three-function catalyst 'Cat 19 [121 The CoPd/HMOR three-function catalyst is able to produce by itself mild oxygenates of methane, CH3OH and HCHO, above 100~ (373 K), as seen in Figure 5.6. It can be

NO 400 ppm + CH4400 ppm + 5vo1% 02 During TPSR 10 K.min -1 3

5

Or)

~ 2

4

o

2

9_. 0

o% -I-2

--1

,

I

'

1O0

I

200

'

I

300

'

I

400

t--

'

500

Temperature/~

Figure 5.6. Evidence for CxHyOz species during transient DeNOx reaction in the presence of Methane over CoPd/HMOR ('Cat I').

Past and Present in DeNOx Catalysis

158

(a)

(b)

2OO

2OO - t - NO x - e - NO + NO 2

'~, E

E 150

150

\

CL C)..

183~

i'M

O

116 ppm

z

C)

"~

~

NOx

100

9

Z

6 loo

Z 0 Z

50

pm 9 ..

460~

' 30 ppm

Z

50

*~*--*~*-~'~,

_

1 O0

200

300

Temperature/~

400

500

19 ppm

_/r

50 100 150 200 250 300 350 400 450 500 550

Temperature/~

Figure 5.7. TPD and TPSR over ~/-A1203 alone (a) TPD of NO pre-adsorbed at RT, in the presence of oxygen (function 3 alone). (b) DeNOx activity of ~/-A1203 in the presence of CH3OH (total flow rate: 250cm3 min -1, catalyst weight: 0.2g, VVH: 50000h -1, 200ppm NO/9vol.% O2/1000ppm CH3OH) [26]. observed, above 280~ (553 K), that both consumptions of these oxygenates and DeNO x reactions are occurring.

4.2.2. Case of T-A1203 alone: role o f CxHy 0 z on function 3 [26] Following a 'step-by-step' methodology, and according to the previous prediction of DeNOx temperature by TPD of NO, Figure 5.7a shows that the reaction can be expected either at 200~ or in the 400-500~ temperature range. Nevertheless, according to the model, the reaction will occur only if the catalyst is able to produce, at the favourable temperature, the appropriate C~HyOz necessary for the function 3 to work. Methanol is not active at 200~ (473 K) (there is no total oxidation by 'O species' adsorbed at this temperature), but it appears to be a good 'reductant' at high temperature, i.e. methanol is scavenging oxygen species left by previous NO decomposition (Figure 5.7b). The NO~ conversion over T-A1203 in the presence of CH3OH (function 3) is shown in Figure 5.7b. The reaction proceeds through successive isotherms, to avoid any preadsorption of reactants or intermediates. A consumption of NO~ (NO + NO2) higher than 50% is observed between 350 and 500~ (623 and 773 K), i.e. in the 'high-temperature window' predicted by TPD in Figure 5.5a. At 400~ (673 K), the NO~ consumption is maximum and is equal to 90%. Let us note that there is no supported metal in the material, the active cus being a cation, probably A13§ unsaturated site.

4.2.3. Case o f 0.5 wt. % Co/'y-Al20 3 DeNOx in the presence o f C H 3 0 H [26] When 0.5 wt.% Co is supported over 'y-A1203 (Figure 5 . 7 ) - cobalt being mainly dispersed as CoO [26] - the 'consumption' of NOx is found to be higher than 50% between

Three-Function Model Reaction

159 SUCCESSIVE ISOTHERMS

o~" 100 z

80

-

......

/T

100

0

60

z ,,i.-. 0

N 2

/

40

o

3 ~.

....................................................................................

=

..'~..iii!iiljiiiiigiiiiii iiii

40

9

20

~

iiiiiiiiiiiiiiiii

',~

20

o_., Z 0

o~

e-

0

co t0

0

~-

0 0

80

0

!

100

150

,

i

200

,

i

,

i

,

!

,

250 300 350 400 Temperature/~

450

500 550

0

Figure 5.8. Formation of N2 during DeNOx reaction (feed: 200ppm NO, 1000ppm CH3OH, 9% 02 and Ar complement) over 0.5 wt.% Co2+/~/-A1203. N2 has been measured by micro GPC. The reaction was conducted by successive isotherms. '?' = acetonitrile [26].

225 and 425~ (498 and 698 K) and 100% (total conversion) between 300 and 350~ (573 and 623 K). For a temperature higher than 400~ (673 K), no inhibition of the reaction by water was observed (not shown here). Water molecules do not compete on active sites (A13+ and Co 2+) with NO, methanol or HCHO oxygenates resulting from the interaction between CH3OH and NO2 (function 2).

4.2.4. Is the consumption of NOx corresponding to the DeNOx reaction, i.e. to the formation of N2? [26] In the case of A1203 alone, the higher conversion of NO to N 2 was 50% at 500~ (773 K) (not shown here). In the case of 0.5 wt.% Co2+/A1203 alumina, it can be observed in Figure 5.8 that the plot of formation of N 2 does not overlap that of NOx conversion for T < 400~ (673 K). The highest NOx conversion t o N 2 is 73% at 350~ (623 K). Furthermore, the selected zone between 400 and 500~ (673 and 773 K) shows that the selectivity to N 2 is 100%, the two previous plots overlapping. For temperature lower than 400~ (623 K), the difference between the two plots, based on NOx concentration, was found to correspond to the formation of acetonitrile (GC/MS detection).

4.3. Conclusion on function 3 Similar results have been obtained over alumina alone, in the presence of propene [27]. The initial HC of the feed (propene) has to first transform to oxygenates (alcohol, aldehyde, etc.) - simultaneously to the NO decomposition (function 3) - to scavenge adsorbed oxygen species left by NO decomposition and regenerate the active sites of function 3. The mild oxidation of HC to oxygenates is the role of function 2 of the present model.

160

Past and Present in DeNO x Catalysis

5. FUNCTION 2: A C T I V A T I O N , HC (INITIAL

REDUCTANT)

MILD BY NO 2

OXIDATION-,

OF

Function 2 can be summarized as follows:

I via RNOxl

The net reaction for a HyC x hydrocarbon is: HyC x -F NO2 = CxHyO + NO

(3)

The interaction between HC [the initial and 'global' reductant in reaction (1)] and NO 2 leads to the formation of an organic nitrogen-containing compound (RNOx), which subsequently suffers decomposition, releasing some 'CxHyO' oxygenate and NO. This is the crucial point: we shall see that NO is formed, but not N 2 in this fundamental step.

According to literature, one of the models [6] considers the following global scheme: HC + NOx ~ RNOx ---> N2 where dinitrogen comes from the direct 'decomposition' of RNO x (see Martens et al. [6,8], Section 1). In contrast, the present model is based on an experimental result, showing that the 'decomposition of RNO x in the presence of oxygen' leads to oxygenates CxHyOz, necessary for function 3 to turn over, with the release of NO and not N 2 [5,28]. (Let us note that the generalization of the present model to the oxidation of carbon particulates by NO 2 leads to the same result: carbon partial oxidation and release of NO [31]). At this level of discussion, some confusion can be done on the origin of N 2 formation during what is often considered as the 'oxidation of RNOx releasing N2'. The present explanation has two parts: (1) The RNO x decomposition leads to the formation of oxygenates CxHyO z with release of NO (function 2). (2) Furthermore, any efficient catalyst, in DeNO x reaction, has to more particularly present functions 2 and 3 turning over simultaneously. As soon as CxHyO z species (alcohol, aldehyde, etc.) are formed, by function 2, they initiate the rotation of the catalytic cycle corresponding to function 3. So, it is this last cycle which releases N 2 as shown experimentally, simultaneously with RNO x decomposition in another cycle. These two reactions do not occur in the same catalytic cycle.

Three-Function Model Reaction

161

The present chapter intends to demonstrate experimentally, with even more details, all these phenomena. The basis of the demonstration can be based on already published data on the surface reaction between NO 2 and adsorbed organic compounds. Yokoyama and Misono have shown that the rates of NO2 reduction over zeolite or silica are proportional to the concentration of adsorbed propene [29], whereas Ii'ichev et al. have demonstrated that NO2 reacts with pre-adsorbed ethylene and propylene on H-ZSM-5 and Cu-ZSL-5 to form nitro-compounds [30]. Chen et al. [2-4] have observed the same nitrogencontaining deposits on MFI-supported iron catalysts. The question on the pairing of nitrogen atoms is not considered here. A more evident demonstration has been done on supported precious metals [5,28]. The interaction between NO2 and HC is clearly located in the second cycle (Figure 5.1). Baudin [32] have detected the RNOx and showed the decomposition of RNO x t o CxHyO z over an Ir/CeZrO2 catalyst. The dinitrogen is formed in the third cycle and is not linked to any organic-nitro compound. Following Yokoyama and Misono, II'ichev et al. and Chen et al. (Figure 5.9) show the decomposition of an RNO x compound pre-synthesized over an Ir/CeZrO2 catalyst, where iridium is cationic. A first adsorption step was done at R T by flowing the following gas mixture: NO (150ppm), C3HTOH (550ppmC), 02 (8 vol.%), complement: N 2. The RNO x results in the interaction between adsorbed propan-2-ol and NO2 formed and adsorbed. RNOx are formed and stored on the catalyst surface. C3H7OH has been selected for simulating the

100

9

200

,

"

NO

'E" 700 600

t,;-:

Z 400

400

500 S

1' :/1

I

C 300 . ' " " ...............[N20 Z .... :: N=O I./"

NO 2

.....

i

........... :~I~i !1:

2000

200

300 S T (~ S 200 ~ ---',~'30'0 9

9 100 , 9

~" 1800

J

002

ss

d

FID

~

~"~ "'~OO-1600140~ O 1200 C) 1000 O

,,

14~0S poo

400

.... 100 500

,400 9 , 500

signal

(CxHyOz)

"~ ,,~ "'i

/

~'J

002

8oo

1"7400 00~600 I[ ~ / 200 i

-/- ........ J

100

~

_

200

300 T

400

Decomposition step RNOx --> CxHyOz + ?:,::i(i)

I :1"~oo I ~',:;~:'~s-~' ,

.............................................................. S

100

.... :::

t 1 700

NO"" q'~0e ..

!11

/

........................

500

. . . . . " " " " " "

:=iI

~\

o.

o soo

300

2000 1800 1600 1400 1200 1000 800 600 400 200 0

TPSR (10 KJmin) - NO (150 ppm) - C3HFOH (550 ppm C) - 02 (8%) - N2

500

(~

Figure 5.9. Concentration (ppm) vs. T; pre-adsorbed RNOx decomposition: RNOx = CxHyO z -+NO; TPSR (10~ min-1), NO (150ppm), 02 (8 vol.%), complement: N2. First plot: N-containing species; second plot: FID signal including CxnyO z and CO, CO2 species [32].

162

Past and Present in DeNO x Catalysis

reaction of RNOx decomposition, as it is one of the CxHyO z corresponding to the mild oxidation of propene by NO2, as shown elsewhere [32]. A second step concerns the TPSR in the presence of flowing NO (150ppm), 02 (8 vol.%), complement: N 2, in the absence of propanol, to simulate a TPSR reaction in the presence of NO/C3H7OH]O 2, where NO is oxidized t o NO2,ad s reacting with C3HvOHaos to lead to RNOx,ads. During TPSR, it can be seen that the decomposition of the nitro compound is quite violent and leads to the formation of NO (Figure 5.9, NO~ graph). Let us consider the following reaction: RNO~

= CxHyO z -at- NO

This reaction is one of the steps in cycle 2 (Figure 5.1), when starting with the initial HC of the feed. The two-step sequence is: HC + NO 2 ~ 'RNOx' 'RNO~'

--+ CxHyO z +

NO

Net reaction: HC + NO 2 : CxHyO z + NO where 'RNO~' is the surface intermediate. This reaction can also occur without catalyst, in the gas phase [5,12]. RNO x is an intermediate of cycle 2 (not 3 !), leading to NO and not t o N 2. The simultaneity between cycles 2 and 3 can lead to the confusion that N 2 c o m e s from the RNO~ decomposition, as often assumed in literature for the N - N pairing. For the sake of simplicity, a CeOz-ZrO 2 (70/30) mixed oxide will be now used as material. This mixed oxide has been previously shown to be able to proceed to three-way catalysis, the general concept for N 2 formation over a metal cation being the same: NO decomposition and oxygen species scavenging, in stoichiometric conditions, by CO as reductant [ 10,11 ].

5.1. Detection of CxHyO z and RNOx compounds by GC/MS during DeNO x reaction over CeZrO2 alone [32] It is now quite well recognized, at international level, that N O 2 produces a mild oxidation of HC to CxHyOz, at a temperature lower than that of its total oxidation by O2. Figure 5.10 shows that CeOz-ZrO 2 (70/30) mixed oxide alone is able to proceed to DeNOx reaction. Its activity is quite low [8-10% at 275~ (548 K)], but it shows that a cation is able to proceed to the reaction. Figure 5.11 more particularly shows that oxygenates are detected by GC/MS, as well as RNOx species. Decomposition of RNOx intermediate leads to oxygenates: Baudin et al. [28,32] have shown the simultaneous formation of 'NO' for a reaction over Ir/CeZrO 2. The corresponding reaction: Propene + NO 2 -- CxHyO z + has been also found to occur in the gas phase [28].

NO

(4)

163

Three-Function Model Reaction 100

80 E

.o_

GC/MS analysis

~> 6o 8 o

o 4o 20 0 1O0

200

300

400

500

Temperature/~ Figure 5.10. DeNOx reaction over CeZrO2 alone (500ppm NO; 2000ppmCl of C3H6; 8 vol.% 02). Successive isotherms [32].

O-.N~O

4

6

8

10

12

14

Time/min

Figure 5.11. TPSR DeNO x reaction over CeZrO2; (200ppm NO, 1000ppm CH3OH, 9% 02). GC/MS analysis at 225~ as indicated on Figure 5.7 [32].

Berger has shown that N O 2 produced over a three-function CoPd/HMOR (Cat I) also produces methanol and formaldehyde at a temperature as low as 120~ (393 K) during the DeNOx reaction in the presence of methane [11,12] (Figure 5.7).

5.2. C o n c l u s i o n o n f u n c t i o n 2 Catalyst needs NO 2 to proceed to the mild oxidation of HC to oxygenates such as alcohol or aldehyde, avoiding the total oxidation of HC to CO, C02/I--I20. AS we shall discuss later, alumina is not a good catalyst for oxidation, but can be a good candidate for such a purpose. So, catalyst also needs to oxidize NO to NO2: this is function 1, which clearly have to 'tum over', according to the model, simultaneously with the two other functions 2 and 3, to get an optimized catalyst.

164

Past a n d Present in D e N O x Catalysis

6. FUNCTION 1" CATALYTIC OXIDATION OF NO TO NO2 Function 1 can be summarized as follows:

The global reaction is given as follows" 2 NO + 0 2 = 2 NO2

(5)

In the present case, the system needs an active site able to oxidize NO without total oxidation of HC. Platinum has to be avoided and an oxide can be considered. Let us note that some supports are also able to proceed to NO oxidation (CeZrO 2, even alumina, etc.). This reaction has to occur to a sufficient degree of conversion to get sufficient amount of NO2, and it has to occur at the same temperature as functions 2 and 3. Park [26] has studied the NO oxidation over Co304]A1203. These results will be presented for demonstration. Figure 5.12 shows that the NO oxidation without catalyst is very slow, whereas thermodynamics shows that 100% conversion can be obtained up to about 200~ (473 K). Alumina alone has a very low activity. Higher the amount of cobalt, higher is the density of sites and higher the NO e production. Let us note that the reaction can be done without platinum.

100 o

o

--~

~-o...~

Thermodynamics o

o~('kl 80

O

\ o

z

.9,~ 60

2w t ~ Co _- ~

.f

tO

~ > tO O

o z

"\

40 9 20 0

_

n

0

I

1 O0

n

n

~

I

o

_

a

200

lwto/:co ~

/

I

300

0.5wt oYoCo

i

I

Alumina J

400

I

500

Temperature/~ Figure 5.12. Catalytic oxidation of NO to NO2 over 0.5, 1, 2wt.% Co/AleO 3. Successive isotherms. 200ppm NO/9vol.% O2/Ar; total flow rate: 250cm 3 min -l" catalyst weight: 0.2g; VVH: 50 000h -1 [26].

Three-Function Model Reaction

165

6.1. Partial conclusion on the three functions A partial conclusion can be drawn at this stage of presentation. A larger number of works have been published in literature and the present model can explain the majority of data. An extensive literature review has been already done by Gorce et al. and will not be considered again in this chapter [5]. On this basis, the three functions can be studied separately. A concept of 'composite catalyst' can be considered; using three catalysts, each one bringing one of the three functions of the model. Furthermore, let us remember that one material can also bring one or two functions. The main difficulty is to get the three catalytic cycles - associated to the three functions - turning over simultaneously. The KOCAT Society (no patent reference known by authors), in South Korea, solved the DeNO x problem of big burners, by directly injecting oxygenates on the catalyst at the outlet of the burner. This process involves the third function of our model. This example shows that only one model (the present one) for DeNO x reaction can be used for either mobile or stationary sources. Pathways are the same; what is changing is the nature of the reductant, which has to be activated, through its partial oxidation, at the temperature when N - O bonds (dinitrosyl species) are broken. What does it mean? As it is very difficult to find the best design of material, for the catalyst to simultaneously initiate the three functions by itself, an external device can be developed to substitute functions 1 and 2, providing the catalyst the 'good' oxygenated species, for the full range of temperature. There are different ways to develop such an idea. One of them is the non-thermal plasma-assisted DeNO x reaction. It will be time consuming- and there is no room in the present chapter- to recall all results already published on this topics by the Society of Automotive Engineers (SAE) Congresses. One of the earlier works, by Penetrante and co-workers [33], has shown that the coupling of a non-thermal plasma reactor, in front of alumina catalyst, was able to produce a significant DeNOx of the feed. They did not give, in their paper, any interpretation for the observed results, but published the composition of the stabilized gas mixture at the outlet of the plasma reactor, just before the catalytic reactor containing alumina. The gas mixture contained more oxygenated compounds. Hoard and Balmer [34] and Dora'/and Kushner [35] have also found CxHyO z (CH20, CH202), NO2 and RNOx (CH3ONO2), which are 'intermediates needed for functions 1 and 2 and, furthermore, for the third function itself. Alumina alone, similar to Co/HMOR [Figure 5.5a, reaction (2)], presents two peaks of NO desorption (see hereafter Figure 5.15a): one is at low temperature, 175-325~ (448-598 K) and the second one is at 400-550~ (673-823 K). According to Section 4.1 and reference [10], it can be predicted, for alumina, that: (1) NO is able to dissociate and produce N 2 at low temperature (function 3). (2) DeNO x should occur as soon as active reductants are delivered to alumina at the temperature when NO dissociates, to scavenge Oads left by NO (function 3). By coupling a non-thermal plasma reactor to a catalytic reactor containing alumina alone, Baudin [32] have observed the DeNOx function of alumina at low temperature,

Past and Present in DeNOx Catalysis

166

confirming the present model. In this case, the non-thermal plasma plays the role of both the functions 1 and 2, as will be discussed in next section.

7. NON-THERMAL PLASMA-ASSISTED CATALYTIC NOx REMEDIATION, FOR SUBSTITUTING FUNCTIONS 1 AND 2 OF THE MODEL. ACTIVATION OF THE LOW-TEMPERATURE DeNOx-FUNCTION 3 OF ALUMINA [32-38] 7.1. Effect of a dielectric barrier discharge (DBD) type non-thermal plasma on a synthetic gaseous reaction mixture The synthetic mixture contained: C3H 6 (2000ppm C1) - NO (500ppm) - O2 (8 vol.%) - N 2, energy density: 3 6 J L -1, voltage: 14kV, VVH: 54000h -1 [32]. The resulting gas mixture at the outlet of the plasma reactor is then flowing through the catalytic reactor. Figures 5.13 and 5.14 show the nature of gas species at the outlet of the plasma reactor, in the absence of catalyst.

7.1.1. Plasma plays the role of function 1 The NO oxidation to NO2 is already occurring at RT. In the experimental conditions of Figure 5.13, total NO oxidation is observed between 180 and 270~ (453 and 543 K). Above 280~ (553 K), the system follows thermodynamics.

7.1.2. Plasma plays the role o f function 2: formation o f CxHyO z and RNO x compounds Figure 5.14 reports the GC/MS analysis of organic species at the outlet of plasma at 150~ (423 K).

50O

E 400 Q. C~. t-

.o 300 t'-

o 200 co

o

100

"~ )q:__ _ J

--

0 0

50

1O0

150 200 250 Temperature/~

300

350

Figure 5.13. NO oxidation to NO2 in a DBD non-thermal plasma [32].

400

Three-Function Model Reaction I-0

167

t

N2

150~ - Plasma ON

Io H3C-i-N~ 0

,,..,~o

2

0

H3C

!

N=O

/ ON

H30--N ~

NH2

O-N ~o

'

i

'

2.00

'

'

'

i

.

4.00

.

.

.

.

.

.

.

6.00

.

.

.

8.00

.

.

.

.

10.00

.

.

.

.

.

.

12.00

I

.

14.00

.

.

.

i

.

.

.

.

16.00

i

.

.

.

.

18.00

Time/min

Figure 5.14. Detection of organic products oxygenates and organic nitrogen-containing compounds at the outlet of non-thermal plasma reactor without catalyst. Feed: NO (500 ppm) - C 3 H 6 (2000ppm C1) - O2 (8 vol.%) - N 2 [32]. It is clearly shown, as already found by Penetrante and co-workers [33], Hoard and Balmer [34] and DoraY and Kushner [35], that the plasma reactor is able to produce oxygenates and RNOx from RT to 400~ (673 K). Those species correspond to function 2 and they are necessary for the DeNO~ reaction according to the present model.

7.2. Plasma-assisted DeNOx catalysis: case of alumina alone (Figure 5.15) [32]. 7.2.1. Plasma 'OFF': high-temperature DeNO x The feed composition has been given just before, for a VVH = 54 000 h -l. Figure 5.15b, in 'plasma OFF' condition, shows that DeNO x is occurring only at high temperature. The reaction is occurring at the temperature of the 2nd peak of NO activation over alumina [Figure 5.15a, 10% conversion for T > 425~ (998 K)].

7.2.2. Plasma 'ON': activation o f the low-temperature function 3 o f alumina When plasma is 'ON', the NO~ conversion is rising from 2 to 40%, at 280~ (553 K). Figure 5.16 shows the presence, at the outlet of the catalytic reactor, of unreacted CxnyO z and RNO~ species, demonstrating that the plasma substitutes the catalytic function 2 of the DeNO~ process. Theses compounds are separately delivered in the full range of temperature, at the very beginning of the temperature-programmed reaction. This is quite different from

168

Past and Present in DeNO x Catalysis

(a) E140 t

g 120 1

' ' ' f"~~' / s

~ 80J

d

I

/

:'T': NO 'aciiv'ati0n over AI20 2

\

- \

!

20-1 ~ 100'150'

' 0 '250'300/350'4~0'450 . 500/ ...

(b)

Tempe"rature/~ 80

J "

i

t'-

.o 60

OFF

>

t- 40 O O 0 ~ 20 z 0

t .

'

100 150 200 250 300 350 400 450 S00

Temperature/~

Figure 5.15. Non-thermal plasma-assisted DeNOx reaction over alumina alone, according to Baudin [32]. (a) NO, NO2, NOx and N20 TPD plots. (b) NOx conversion vs. T without plasma.

the catalytic process, where functions 1 and 2 have to be activated at the temperature where function 3 begins to work. Consequently, the plasma is producing CxHyO z at a sufficiently low temperature, including a selection of C x n y O z (Figure 5.16) able to activate the low-temperature function 3 of alumina (Figure 5.15b). This activation is linked to the 1st peak of NO desorption, as predicted by our methodology (Figure 5.15a).

7.2.3. Sequence 'plasma ON' (low temperature)- 'plasma OFF' (high temperature) in the presence of a mixture of HCs representative of diesel engine exhaust Figure 5.17 shows that a mixture of HCs and a VVH of 18 000 h -1 corresponding to a volume of alumina three times higher than the precedent case (Figure 5.15b) leads to a NO x conversion between 50 and 60%, in the temperature range 180-425~ (453-698 K), plasma being 'OFF' for temperature higher than 320~ (593 K). The reason for such a process is that at higher temperature, HCs play their role. Let us note that DeNO x was followed by N 2 quantitative measurements for every 3 minutes. The preceding results suggest an advantageous plasma--catalyst coupling effect on the NO x remediation, in full accordance with the proposed mechanism [38]. The C x n y O z and RNO~ compounds, produced by the non-thermal plasma before the catalytic reactor

169

Three-Function Model Reaction

Main gas phase species (GC/MS Analysis) 100

~i~i~i~

z

8 60 g

i.

eo

0

o

100 150 200 250 300 350 400 450 500

Temperature/~

,,

O.lO4_

I. ol

45.10 4 ---

Ic

I

~

H3c/O\N

o

t

I

-i--

2

4

6

8

;

10

12

14

16

18

Time/min

Figure 5.16. Non-thermal plasma-assisted DeNOx reaction over alumina alone, according to Baudin [32]. GC/MS analysis of outlet gases at 215~ (488K) and 25% NO~ conversion (Figure 5.14b). Feed: n-Clo, 07H8, 03H6, 03H8, Ar 100 DeNOx=N 2(p-GC) For T > 200~

#

AI20~ '"~

80

o~ tO

~9

> tO

O

60

40 Plasma ON

z

20

0 100

150

200

250

300

350

400

450

500

Temperature/~

Figure 5.17. Non-thermal plasma-assisted DeNOx reaction over alumina alone, according to Baudin [32]. Reaction in the presence of a mixture of HCs (see feed composition on the top of the figure). VVH = 18 000 h -1 . with alumina, demonstrate the strategy of the model at both low (plasma 'ON') and high (plasma 'OFF') temperatures, as well as the equivalence of the role of plasma and that of functions 1 and 2 of the catalyst, when non-plasma assisted.

170

Past and Present in DeNO x Catalysis

8. CONCLUSION AND M E T H O D O L O G Y It has been chosen, for presenting the three-function model, to start from the true 'DeNO~ catalytic cycle' corresponding to function 3, which leads to the N - N bonding and N2 release from the catalyst. Subsequently, it appears that two other functions are necessary to assist function 3. Function 1, oxidation of NO to NO 2, can be studied separately [3926] and its temperature of activation is adjusted to the temperature at which function 3 is working. (It does not mean that the amount of NO2, at the outlet of the reactor, will be the same in the presence of a reductant, as there is a consumption of NO2 during the interaction HC]NO2). Function 3 can be studied separately by direct injection of the CxHyO z oxygenates (alcohol, aldehyde, etc.) corresponding to the mild oxidation process of HC by NO 2. It is rather difficult to study function 2 separately, as the catalyst generally presents, simultaneously, functions 1 and/or function 3. Nevertheless, the mild interaction of HC with NO 2 can be approached through the direct NOz/HC reaction, even in the absence of dioxygen. It has to be compared to the total oxidation of HC in the p r e s e n c e of oxygen, as it is a competitive reaction for the HC consumption. In contrast, it will be very interesting to compare the two catalytic pathways of elementary steps, and reaction intermediates, in both oxidation reactions (mild and total oxidation of reductant).

8.1. Some important features of the model can be summarized as follows 9 Simultaneity between the three catalytic cycles, as presented hereafter (Figure 5.18):

The three cycles have to turn over in the same range of temperature. This catalytic approach of the DeNOx reaction is not new. There is the same process for isomerization of alkanes, where there are also 3 catalytic cycles which have to turn over simultaneously (bifunctional catalysis). The kinetics of isomerization is given by only one cycle, the other two turning over very rapidly and are near equilibrium [13].

~+ i i o~ i

[i~b-~7

" ..........

.......

"

-'

i+

i

9

9

i HxCyOl

i" co, co~ i

"-'- . . . . . . . . . . . . . "

!"

:: 9 RNO

H20

i

'it ...................

x

Figure 5.18. Catalytic assisted DeNOx reaction: each cycle corresponds to one function (F1, F2 and F3).

Three-Function Model Reaction

171

9 Regulation of the total oxidation of 'reductants' (HC, CxHyOz) by 02, taking into account the difference of temperature between their mild oxidation by NO2 and their total oxidation by dioxygen. This point requires the choice of a total oxidation function, but not too much active. 9 If the preceding requirements are fulfilled, then the DeNOx process (function 3) does not need a large amount of reductant, as it is very often claimed: the stoichiometry of '2NO + CxHyO z -- N2 + xCO/CO2 + y/2 H20' should be considered. Clearly, it is generally impossible to avoid the competition between the *OadS left by NO and the *Oads due to 02 dissociation, for the total CxHyO z oxidation on function 3 (this competition corresponds to a kinetic coupling of at least two catalytic cycles, through Oads [13]). Both of them contribute to the total oxidation of reductants.

8.2. A methodology can be suggested from these conclusions, for designing the best efficient catalyst in DeNO x reaction For a given catalyst, in the framework of the present model, the proposal is as follows: [I] Prediction of the temperature where DeNO x can take place, by TPD of NO preadsorbed with or without oxygen. In the absence of oxygen, check the formation of N 2, N20, NO2 and NO during TPD. If N20 and/or N 2 are formed, it means that the reaction already took place. If not, the system needs the reductant to take place. This experiment also means that function 3 can work [10,25]. [II] Study of function 3, by studying the reaction NO/O2/CxHyO z (corresponding to the HC initially in the feed) [5,10,26,28]. [III] Checking of function 1:NO/O2 to NO2. Check the domain of temperature compared to that of NO desorption (thermal activation). [IV] Function 2 can be approached in the absence of 02, by the reaction HC/NO2. If function 3 exists on the catalyst, then the reaction can partially proceed, when the temperature is rising, to the total oxidation of HC.

8.3. Finally, some targets and suggestions for the future can be done (1) If cycle 1 is easy to study (NO to NOe) [39], then pathways of functions 2 and 3 still remain to be established in detail. (2) Kinetics of cycles 2 and 3 have to be done. That of cycle 2 will probably have to take into account the presence of cycle 3. (3) What is the catalytic cycle able to drive the kinetics of the whole process (rate determining cycle)? (4) How to control the kinetics of the process, regulating the kinetics of two of the three cycles, to get only one rate determining cycle? (5) The design of a three-function catalyst, for a given application with specific reductants, will be easier in the framework of the model. (6) Considering the model, several possibilities to solve the problem of DeNO x on stationary or mobile sources can be defined, at different levels.

172

Past and Present in DeNO x Catalysis

The model can also be extended to three-way catalysis as far the active site is a cation of transition metal, the reductant being CO and the feed CO/NO/HC being very near stoichiometry [ 10,11 ].

ACKNOWLEDGEMENTS One of the authors (GDM) greatly acknowledges all engineers from industries (PSAPeugeot-Citroen, Renault, Gaz de France, ADEME) who worked with the 'Laboratoire R6activit6 de Surface', as well as Michel Boudart for many fruitful discussions on DeNO x. The Group of Researchers who worked in the framework of the French-Polish 'Jumelage' (granted by CNRS, Polish Academy of Science, Polish Ministry of Scientific Research and Information Technology, French Foreign Office) on 'Catalytic Materials for Environment' are also greatly acknowledged for so many works and PhD thesis which contributed to applying and confirming our model.

REFERENCES [ 1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11]

Wang, X., Chen, H.-Y. and Sachtler, W.M.H. (2001) J. Catal., 197, 281. Chen, H.-Y., Voskoboinikov, T. and Sachtler, W.M.H. (1999) Catal. Today, 54, 483. Chen, H.-Y., Voskoboinikov, T. and Sachtler, W.M.H. (1999) J. Catal., 186, 91. Chen, H.-Y., Voskoboinikov, T. and Sachtler, W.M.H. (1998) J. CataL, 180, 171. Gorce, O., Baudin, F., Thomas, C. et al. (2004) Appl. Catal., B54, 69. Martens, J.A., Cauvel, A., Francis, A., et aL (1998) Angew. Chem. Int. Ed., 37, 1901. Kharas, K.C.C. (1993) AppL Catal., B2, 207. Brosius, R. and Martens, J.A. (2004) Topics CataL, 28, 119. Iwamoto, M. and Takeda, H. (1996) Catal. Today, 27, 71. Dj6ga-Mariadassou, G. (2004) Catal. Today, 90, 27. Dj6ga-Mariadassou, G., Fajardie, F., Temp~re, J.-F. et aL (2000) J. Mol. Cat., A: Chemical,

[12] [13] [14] [15] [16]

Berger, M. (2000) ThOse de l'Universit~ Pierre et Marie Curie. Dj6ga-Mariadassou G. and Boudart M. (2003) J. Catal., 216, 89. Ferreira, A.P., Capela, S., Da Costa, P. et al. (2007) Catal. Today, 119, 156. Ferreira, A.P., Henriques, C., Ribeiro, M.F. et al. (2005) Catal. Today, 107-108, 181. (a) Boudart, M. and Dj6ga-Mariadassou, G. (1982) Cin~tique des R~actions en Catalyse Hgt~rogOne, Masson, Paris. (b) Boudart, M. and Dj6ga-Mariadassou, G. (1984) Kinetics of Heterogeneous Catalytic Reactions, Princeton University Press: Princeton, NJ, USA. Iiuzuka, T. and Lunsford, J.H. (1980) J. Mol. Catal., 8, 391. Hendriksen, D.E., Meyer, C.D. and Eisenberg, R. (1977) lnorg. Chem., 16, 970. Cataluna, R., Arcoya, A., Scoane, X.L. et al. (1995) Catalysis and automotive pollution control III. In Studies in Surface Science and Catalysis (A. Frennet and J.M. Bastin, eds) 96, p. 215. Fajardie, F., Temp~re, J.-F., Manoli, J.-M. et al. (1998) J. Catal., 179, 469. Manuel, I., Thomas, C., Bourgeois, C. et al. (2001) Catal. Lett., 77, 193. Manuel, I., Chaubet, J., Thomas, C. et al. (2004) J. CataL, 224, 269. Hamon, C., Le Lamer, O., Morio, N. et al. (1998) PCT WO 98/15339. Germain, J.E. (1969) Catalytic Conversion of Hydrocarbons, Academic Press: New York, p. 259.

161, 179.

[17] [18] [19]

[20] [21 ] [22] [23] [24]

Three-Function Model Reaction

173

[25] Li, Y. and Armor, J.N. (1991) Appl. Catal., 76, L1. [26] Park, J.-W. (2005) ThOse de l'Universitd Pierre et Marie Curie. [27] L8 Van, T., Thiet, N.-Q., Thomas, C. et al. (2004) Proc. of Global Symposium on Recycling, Waste Treatment and Clean Technology, "REWAS", 1, 785. [28] Baudin, F., Da Costa, P., Thomas, C. et al. (2004) Topics Catal., 30-31, 97. [29] Yokoyama, C. and Misono, M. (1996) J. Catal., 160, 95. [30] II'ichev, A.N., Matyshak, V.A., Korchak, V.N. et al. (2000) Kinet. Catal., 41, 706. [31] Darcy, P. (2005) ThOse de l'Universitd Pierre et Marie Curie. [32] Baudin, F. (2004) ThOse de l'Universitd Pierre et Marie Curie. [33] Penetrante, B.M., Brusasco, R.M., Meritt, B.T. et al. (1998) SAE Technical Paper 982508. [34] Hoard, J. and Balmer, M.L. (1998) SAE Technical Paper 982429. [35] Dora'f, R. and Kushner, M.J. (1999) SAE Technical Paper 01-3683. [36] Gorce, O., Jurado, H., Thomas, C. et al. (2001) SAE Technical Paper 2001-01-3508. [37] Khacef, A., Cormier, J-M., Pouvesle, J-M. et al. (2002) Hakone 8, P5.4. [38] Baudin, F., Schneider, S., Lendresse, Y. et al. (2004) Patent 2.877.693-04 11882-A1. [39] Marques, R., Darcy, P., Da Costa, P. et al. (2004) J. Mol. Cat., 221, 127.