New User Special Price Expires in

Let's log you in.

Sign in with Facebook


Don't have a StudySoup account? Create one here!


Create a StudySoup account

Be part of our community, it's free to join!

Sign up with Facebook


Create your account
By creating an account you agree to StudySoup's terms and conditions and privacy policy

Already have a StudySoup account? Login here


by: Hailey Halvorson
Hailey Halvorson
GPA 3.8


Almost Ready


These notes were just uploaded, and will be ready to view shortly.

Purchase these notes here, or revisit this page.

Either way, we'll remind you when they're ready :)

Preview These Notes for FREE

Get a free preview of these Notes, just enter your email below.

Unlock Preview
Unlock Preview

Preview these materials now for free

Why put in your email? Get access to more of this material and other relevant free materials for your school

View Preview

About this Document

Class Notes
25 ?




Popular in Course

Popular in Physics 2

This 68 page Class Notes was uploaded by Hailey Halvorson on Thursday October 22, 2015. The Class Notes belongs to PHYS 22 at University of California Santa Barbara taught by Staff in Fall. Since its upload, it has received 21 views. For similar materials see /class/227143/phys-22-university-of-california-santa-barbara in Physics 2 at University of California Santa Barbara.

Similar to PHYS 22 at UCSB


Reviews for GEN PHYS


Report this Material


What is Karma?


Karma is the currency of StudySoup.

You can buy or earn more Karma at anytime and redeem it for class notes, study guides, flashcards, and more!

Date Created: 10/22/15
ADVANCES IN CATALYSIS VOLUME 44 The Surface Science Approach Toward Understanding Automotive Exhaust Conversion Catalysis at the Atomic Level BERNARD E NIEUWENHUYS Leiden Institute of Chemistry Gorlaeus Laboratories Leiden University 2300 RA Leiden The Netherlands This review focuses on the reactions and catalysts used for control of emissions of exhaust gases for gasolinefueled automobiles with emphasis on fundamental understanding of the surface processes Attention is paid to threeway catalysts which simultaneously enhance the conversion of CO hydrocarbons and nitrogen oxides The mechanisms of the CO oxidation and nitrogen oxide reactions the speci c differences in behavior of Pt Pd and Rh the effect of alloy formation and the role of ceria used as additive in threeway catalysts are discussed Results of surface science studies are compared with results reported for supported catalysts For CO oxidation there is excellent agreement between results obtained for single crystal surfaces and supported catalysts The kinetics of the reactions on pure metals can be understood on the basis of the kinetics parameters obtained from singlecrystal studies For the NO reduction reactions there is qualitative agree ment between results obtained with singlecrystal and supported catalysts The major effects of alloy formation can be understood on the basis of the surface com position 39 39 Abbreviations AES Auger electron spectroscopy Ed activation energy for desorption ESDIAD electron stimulated desorption ion angular dependence he hydrocarbon HRBELS highresolution electron energy loss Spectroscopy LH Langmuir Hinshelwood LEED lowenergy electron diffraction RAIRS re ection absorption infrared spectroscopy SERS surfaceenhanced Raman spectroscopy SIMS secondary ion mass spectrometry TDS thermal desorption spectroscopy spectrum or spectra TPRS temperature programmed reaction spectroscopy TOF turnover frequency TWC threeway catalyst or catalysis uhv ultrahigh vacuum UPS UV photoelectron spectroscopy XPS Xray photoelectron spectroscopy 39 259 Copyright 2000 by Academic Press All rights of reproduction in any form reserved 03600564100 3000 BERNARD E NIEUWENHUYS I Introduction Severe emission limits for motor vehicles were introduced rst in the United States and later in many other countries starting in the mid19605 Meeting the increasingly stringent emission requirements in subsequent years forced the installation in motor vehicles of progressively more ad vanced emission control devices The focal point of emission control is the catalytic converter in which the desired chemical reactions occur The pollutants carbon monoxide and unburned hydrocarbons hc are converted by oxidation into the desired C02 and water i 02 9 aCnHm bOz gt CC02 dHZO 2 and the pollutant nitrogen oxides are reduced to dinitrogen and CO or H20 2N0 2C0 gt N2 2C0 H2 N2 H20 aNO th gt cho dco2 eH20 3 The rst generation of automotive catalysts catalyzed only the two oxida tion reactions Eqs 1 and 2 and hence these39are called twoway catalysts At that time much knowledge was available concerning oxidation catalysis and it was relatively easy to develop the rst automotive catalyst Both noble metal and transition metal oxides were available to oxidize the CO and hc Many transition metal oxides exhibit good catalytic properties for the oxidation reactions However the noble metals have superior properties with signi cantly lower light off temperatures the lightoff temperature is the minimum temperature needed to start the reaction and with better resistance to poisoning by sulfur compounds and other compounds present in the fuel or lubricant The noble metals Pt andor Pd were the active components of these catalysts Small noble metal particles were supported by yalumina with high surface area and good thermal stability A new generation of catalysts was needed when the exhaust emission regulations in the United States changed requiring lower levels for NOx After many years of research it was determined that the most effective technology is a threeway catalyst TWC system which simultaneously accelerates the NO reduction reactions and the oxidation of CO and hydrocarbons The principle of TWC is illustrated in Fig 1 which shows the change of the conversion of the three major pollutants as a function of the airfuel ratio The airfuel ratio is usually expressed as the weight UNDERSTANDING AUTOMOTIVE EXHAUST CONVERSION 261 100 90 80 7O 6O 50 4o 80 EFF AF RC1in 2 Window 39 Conversion 30 20 1 10 0 2 Stoichiometric AF Ratio I r 39 391 I I 143 144 145 146 147 148 149 Air I Fuel Ratio FIG 1 Ef ciency of a threeway catalyst for the conversions of CO NO and hydrocarbons at various AF ratios of air per weight of fuel It is easily convertable into the equivalence ratio A the value of which is unity at the stoichiometric point when all the fuel is completely converted to C02 and water Obviously the oxidation reactions Eqs 1 and 2 are favored 39under conditions of excess air A gt 1 whereas complete reduction of nitrogen oxides requires reducing conditions A lt 1 Fortunately simultaneous conversion of NO CO and he can occur in a narrow window around the stoichiometric composition A 1 as demonstrated in Fig 1 Precise control of the airfuel ratio is required to achieve high conversions of CO he and NO The composition of the airfuel mixture introduced to the engine is electronically controlled by a feedback system with an Oxygen sensor to monitor the oxygen concen tration in the exhaust Most of the current convertersconsist of a ew through ceramic monolith with its channel walls covered with a highsurfacearea yA1203 layer the washcoa39t which contains the active catalyst particles The monolith is composed of cordierite a mineral with the composition 2Mg0 2A120339 SSiOZ The chemical composition of a modern TWC is quite complex In addition to alumina the washcoat contains up to 30 wt base metal oxide additives added for many purposes The most common additives are ceria and lanthana in many formulations BaO and ZrOZ are used and in some converters NiO is present The major active constituents of the washcoat are the noble metals Pt Pd and Rh typically 13 g Most of the TWC systems in use today are still based on Pt and Rh ina ratio of about 10 1 262 BERNARD E NIEUWENHUYS In the past 5 years some of the Pt Rh TWC formulations have been replaced by Pdbased TWC the Pdonly TWC and Pd Rhbased TWC particularly in the United States In the past 30 years automotive catalysis has become the greatest novel application of heterogeneous catalysis in the world Automotive catalysis is a major application for the precious metals as illustrated in Table I which shows that the relative importance of Pt Pd and Rh for automotive catalysis changed considerably in the past decade The current usage of Pd for automotive catalysis exceeds that of Pt However the Pd based convert ers contain more precious metal than the Ptbased converters Many reviews 1 4 give detailed descriptions of the fundamentals of automotive exhaust catalysis Shelef and Graham 3 gave a broad view of the unique properties of Rh in automotive three way catalysis This chapter focuses on fundamental processes taking place at the catalySt surface Atten tion is given to the adsorption of the relevant gases mechanisms of the relevant reactions speci c differences in the surface properties of the vari ous noble metals effects of alloy formation and the chemistry of the additives in particular ceria The chemistry of the reactions shown in Eqs 1 3 is understood in considerable detail as a result of recent studies using models of the TWC and surface science techniques Relevant literature of diesel and lean exhaust gas control is also brie y discussed It is not the aim of this chapter to review all the available data The data mentioned and the papers cited here are presented only to the extent that the ndings illuminate the discussion TABLE I Western World Demand for Pt Pd and Rh 10 6 X automobile Percentage of total 10 6 X total catalyst demand for automobile Metal Year demand g gross g catalysts Pt 1987 96 39 40 1992 119 48 40 1996 152 57 37 Pd 1987 98 84 8 1992 121 152 13 1996 212 72 34 Rh 1987 97 70 72 1992 l 18 95 81 1996 148 129 87 Data from Platinum 1997 Johnson Matthey l UNDERSTANDING AUTOMOTIVE EXHAUST CONVERSION 263 ll Adsorption A ABSORPTION OF CO HYDROGEN OXYGEN AND NO The adsorption of diatomic molecules on a metal surface may be consid ered as competition between molecular and dissociative adsorption ABgasaABadsaAads Bads Dissociative adsorption can occur when the bonds formed between the fragments of the dissociated molecule and the surface are much stronger than the bonds between the molecule and the surface For example molecu lar hydrogen oxygen or nitrogen is only weakly adsorbed on transition metals Oxygen hydrogen and nitrogen adatoms on the other hand are strongly bound on many metal surfaces Therefore dissociative adsorption is often thermodynamically possible as discussed later Molecular adsorp tion of CO and of NO is relatively strong on many metal surfaces These adsorbates may undergo both dissociative and molecular adsorption on the same surface depending on the experimental conditions It is often obServed that molecular adsorption prevails at lower tempera tures and that dissociative adsorption occurs at higher temperature This pattern may be caused by kinetics the activation energy for dissociative adsorption is too high for dissooiation at lower temperatures There may also be a thermodynamic reason If the number of surface sites at which adsorption can take place is equal for molecular and dissociative adsorption the surface can accommodate twice as many molecules in the molecular state as it can in the dissociated state Hence molecular adsorption will prevail if the heat of dissociative adsorption is not much greater than the heat of molecular adsorption The entropy change for adsorption is negative and consequently at suf ciently high temperatures desorption 39will occur In the case considered previously only half the number of molecules can be adsorbed in the dissociative state as can be adsorbed in the molecular state As a result the entropy of the system will be lower for molecular adsorption and dissociation can occur at higher temperatures Dissociative adsorption requires a cluster of several free and adjacent metal atoms on the surface Therefore often dissociation occurs when the surface coverage is low and molecular adsorption occurs above a certain coverage provided that both dissociative and molecular adsorption can occur under the experimental conditions considered It is assumed that a diatomic molecule adsorbed parallel to the surface is in a transition state for dissociation The more favorable adsorption complex for molecules such as CO on a group VIIII metal surface is that in which the molecular axis is bonded perpendicular to the surface or BERNARD E NIEUWENHUYS slightly tilted It has been demonstrated by electron stimulated desorption ion angular distribution ESDIAD that the molecular axis vibrates around the surface normal and that its amplitude increases with increasing tempera tures 5 Fig 2 The temperature at which dissociation occurs is probably reachedwhen the vibrational amplitude becomes suf ciently large so that a bond between the O atom and the metal surface is formed resulting in dissociation Many of the adsorption data presented in this review are based on thermal desorption spectroscopy TDS measurements This technique yields in a simple way information about the number of binding states and their bond strengths with the surface For illustration TD spectra are shown for mole cules adsorbed on a39Rh lament 39 rTDS measured for polycrystalline surfaces are often much more compli cated than those representing singlecrystal surfaces A polycrystalline sur face exhibits all the adsorption sites of the crystal faces of which it is composed Since these sites are simultaneously present a TDS represents an average of the spectra of the different surface sites weighted according to the relative concentrations of these sites 1 Adsorption of Hydrogen Hydrogen is known to be dissociativer adsorbed on transition metal surfaces and the initial sticking probability varies from ab ut 005 tounity at T 100 300 K depending on the metal and its surface structure For example the initial sticking probability for hydrogen on a Ni111 surface is only 005 on the stepped Ni8111 X 100 surface it is 02 and on a Ni110 surface it is essentially unity 6 In general the sticking probability is signi cantly smaller on at surfaces than on surfaces with high densities of steps surface defects etc It appears that hydrogen prefers to be adsorbed on sites on which it is coordinated with many metal atoms On 111 surfaces 39 M O O I l l l C D C D CO C O 777777777u7777 TITrTi r TrTrnr a b c d Fig 2 Molecules such as CQ are bonded on most of the group VIII metal surfaces with the molecular axis perpendicular to the surface a or slightly tilted The amplitude of the vibration ar39Ourid the surface normal increases with increasing temperature b A t39a certain temperature an O metal bond is formed c and dissociation occurs immediately d UNDERSTANDING AUTOMOTIVE EXHAUST CONVERSION 265 of the facecentered cubic FCC lattice for example hydrogen is adsorbed at low coverage on threefold sites Hydrogen desorbs from the group VIII metals at about 350 400 K its initial heat of adsorption lies in the range 80 100 kJ mol391 and is the same within 20 on similar surfaces of these metals More details are given in Nieuwenhuys 7 2 Adsorption of Oxygen The interaction of oxygen with Pt surfaces has been studied in detail with modern surface analytical techniques 7 Since the TDS of oxygen on most of the other metals of group VIII are qualitatively similar to those for oxygen on Pt one can expect that essentially the same species occurs on all these metals At 100 K oxygen is adsorbed on Pt111 with a sticking probability of about unity in a molecularly adsorbed state 7 state with a low heat of adsorption 37 kJmol 8 Electron energy loss spectroscopy EELS combined with UV photoelectron spectroscopy UPS indicates an essentially single O O bond with a signi cant electron transfer from the valence band of Pt into orbitals derived from the 7r antibonding oxygen levels and with the 0 0 bond axis parallel to the surface Heating the surface to temperature above 170 K results in the formation of adsorbed atomic oxygen O and desorption of some oxygen y The heat of adsorption of 8 0 decreases rapidly with increasing coverage from 500 to 160 kJmol at 0 08 0m On Pt111 surfaces oxygen is adsorbed at 300 K with a low sticking probability lt01 A third state is sometimes observed upon heating Pt111 in the presence of oxygen in the 900 K temperature range 8 This subsurface oxygen which desorbs at temperatures higher than 1200 K has a very low reactivity toward CO and hydrogen There is controversy concerning the nature of this state 7 Many authors believe that this state is a socalled surface oxide Fig 3 whereas others attribute its appearance A Molecular 2 Oxygen a P1 111 U 9 Atomic 9 Oxygen a c x 5 quotOxidequot b 0 Km 3 5 200 600 1000 1400 Temperature K FIG 3 TDS of oxygen from Pt111 reproduced with permission from Gland et aL 8 266 BERNARD E NIEUWENHUYS to oxideforming bulk impurities such as Si or Ca and its segregation to the surface during oxygen exposure at high temperature On polycrystal line surfaces and on open surfaces the formation of an oxide layer at higher temperature is quite common as is demonstrated by eld emission microscopy 7 In contrast to adsorbed CO the density of O adatoms at saturation is not determined by the size of the adsorbate more open overlayer structures are usually formed The kinetics of oxygen adsorption at ambient tempera ture varies greatly from metal to metal and from plane to plane For example oxygen is adsorbed with a sticking probability near unity on Ni surfaces whereas on Pt111 the sticking probability is lt01 The initial heat of adsorption of 3 0 is approximately 250 kJmol on the group VIII metals corresponding to a metal O bond strength of about 370 kJmol Because of the high desorption temperature the appreciable decrease of the heat of adsorption Q with increasing coverage and the incorporation of oxygen into the bulk it is not easy to nd a correlation of Qmuial with the position of the metal in the periodic table A careful examination of reliable data suggests that the heat of adsorption increases in the following order Pt 230 kJmol Pd lt Ir lt Rh lt Ru lt Ni 330 kJmol The variation in the metal O bond strength is about 55 kJmol or 15 The order in the heat of adsorption is similar to that in the heat of formation of the oxides The tendency of the metal to form surface or bulk oxides also increases in this sequence It can be concluded that in contrast to hydrogen on group VIII metals there are signi cant differences in the metal O bond strength on these metals 3 Adsorption of CO CO on metals is the most extensively studied adsorption system 9 I highlight some of the features that are important for this discussion TDS of CO on group VIII metal surfaces in general show various peaks which can be classi ed into three main groups as shown schematically for CO on polycrystalline Rh in Fig 4 and corresponding to the following 1 LowT 7 states desorbing at temperatures much lower then room tem perature 2 1 states with a peak maximum in its TDS between 350 and 500 K 3 Sometimes depending on the choice of the metal its surface structure and the experimental conditions such as the temperature additional states 3 are observed with a maximum of approximately 900 K These B states arise from C0 dissociation followed by recombination of C and O at the desorption temperature UNDERSTANDING AUTOMOTIVE EXHAUST CONVERSION 267 PCO nu I If I I I I I I 150 150 850 Temperature K FIG 4 TDS of CO from Rh reproduced with permission from Nieuwenhuys 7 At room temperature adsorption normally takes place in the a state corresponding to molecularly adsorbed CO Usually two or more a peaks are observed depending on the surface structure and the CO coverage These peaks can be rationalized on the basis of distinct species adsorbed on different sites and in terms of lateral interactions A model of the a CO metal bond was rst proposed by Blyholder 10 According to this model the bond is formed by electron transfer from the highest lled molecular orbital of C0 50 which is essentially a nonbonding orbital with respect to the C 0 bond to unoccupied metal orbitals im plying a donation of electrons to the metal accompanied by backdonation of electrons from occupied d orbitals into the lowest un lled CO molecular orbital the strongly antibonding 217 orbital The relative contributions of the 039 and the 71 bonding to the metal CO bond strength and to the net charge transfer have been discussed in many papers Suf cient evidence based on theoretical calculations 1 I 12 and inverse photoemission studies 13 indicates that for Pd and Ni metal 71 donation to CO is much more important for the CO metal bond than is the COtometal a donation However the CO Pt bond has a signi cant a donation component 13 Furthermore work function measurements indicate that the net electron transfer from metal to CO is much larger for Rh Pd and Ni than for Pt 14 Experimental and theoretical results support the fact that the CO mole cules are adsorbed on the densely packed surfaces of Pt Pd Rh and Ir with the C 0 axis normal to the surface and with the carbon atom directed to the surface Experimental evidence is derived from angular resolved UPS ESDIAD ion scattering lowenergy electron diffraction LEED intensity analysis and EELS LEED IR or EELS and TDS have been applied for CO on many single 268 BERNARD E NIEUWENHUYS crystal surfaces The general features of these are very similar First a simple overlayer structure is formed up to medium coverage For example on the FCC111 surfaces V5 X V3R30 structures are observed which are fully developed at 6 13 The molecules are then located on most of the densely packed surfaces in identical sites At higher coverage the overlayer unit cell is compressed new surface structures are observed and a fraction of the CO molecules are forced to move into other sites At the highest coverage a closepacked overlayer is formed which is largely determined by the C0 C0 mutual repulsion and not by the substrate sites The maximum density of adsorbed CO molecules is about 1 X 1015 moleculescm The sites on which CO molecules are bound are re ected in the stretching vibration frequencies vof the CO bond In the pioneering work of Eischens et al 15 bands at frequencies below 2000 cm 1 were assigned to CO acting as a bridging ligand M2CO and bands between 2000 and 2100 cm 1 were attributed to linear CO species such as MC 0 Numerous results have shown that Eischens et al s interpretation of the bands was essentially correct and more comprehensive correlations of v with bonding sites have been suggested 16 A substantial increase in v occurs with increasing surface coverage Both experimental observations based on the combined use of TDS LEED and EELS IR and theoretical predictions suggest that the energy difference of CO bonded in ontop positions or in multifold sites is quite small The only metal on which bridged or multifold sites are the most favorable positions for CO is Pd On Ir Rh and Ru linearly adsorbed CO is preferred Also on Pt111 ontop positions are preferred but the energy differences with the other positions are small resulting in a high degree of disorder at room temperature as observed by LEED 7 The differences in heat of adsorption on the group VIII metals for the various singlecrystal surfaces of a metal are about 20 kJmol or 20 7 Dissociative adsorption of CO has been found on a variety of transition metal surfaces Broden et al 17 and Nieuwenhuys I4 correlated the tendency for CO N2 and NO to dissociate with the position of the transition metal in the periodic table the tendency for dissociation increases the further to the left the metal appears in the table and it decreases from 3d to 5d metals Furthermore the borderline for dissociative or molecular adsorption moves to the right in the sequence CO N2 NO to 02 being the same order as the bond strength in the free molecules There is suf cient evidence for the proposed correlation For example W and Mo surfaces dissociate CO easily at room temperature dissociative adsorption has not been reported for Pt Ir and Pd111 surfaces and CO dissociation has been reported to occur on Ni Co and Ru at elevated temperatures Ben zinger 18 suggested that the state of adsorption molecular or dissociative UNDERSTANDING AUTOMOTIVE EXHAUST CONVERSION 269 is determined by thermodynamic criteria In his paper the heat of dissocia tive adsorption was estimated from the heats of formation of metal carbides nitrides and oxides In this analysis a similar correlation as that in Refs 14 and 17 relating dissociative adsorption with position in the periodic table was found The surface structure may have an additional in uence on the dissociation In general closepacked surfaces are the least reactive and rough surfaces or surfaces with steps or kinks are the most reactive for dissociation The effect of surface structure on dissociation can be attributed to the variation of eg metal carbon bond strength with sur face structure F or our purposes it is relevant to conclude that CO adsorption is predom inantly molecular on Pt Ir Pd and Rh and that desorption occurs at about 500 K Controversy exists in the literature regarding CO dissociation on Rh surfaces 19 On supported Rh catalysts and Rh clusters CO dissocia tion has been observed Some authors also reported CO dissociation on certain singlecrystal surfaces whereas other studies indicate no or insig ni cant dissociation on clean Rh surfaces J 9 4 Adsorption of N0 Figure 5 presents a TD spectrum for NO adsorbed on a Rh lament at 80 K 20 showing desorption peaks of N0 N2 and 02 The heat of TDS NORh 3 0 3 392 8 J 2 mez32 me28 me 30 I I l I I l I if f I 200 400 600 800 1000 1200 Temperature K FIG 5 TBS of N0 N2 and 02 following adsorption of NO on a Rh lament at 80 K reproduced with permission from Hendrickx and Nieuwenhuys 20 gt PN0GU 270 BERNARD E NLEUWENHUYS NONORh533 b N2NORh533 k5 r I 100 600 800 400 600 800 1000 gt gt PN2IGu gtTemperoture K gtTemperoture K N2NORhILIOl O NONORMMO gt PN2uu 019L 8L A8 1 t 0 k 5L l00AL 0quot OO3L 035Lx50 39 I I I I 1 I I I I I I I 000 600 00 100 600 800 1000 gt Temperature K gt Temperature K1 UNDERSTANDING AUTOMOTIVE EXHAUST CONVERSION 271 adsorption of molecularly adsorbed NO is on the order of 110 ltJmol 1 for the group VIII metals The energy barrier for dissociation is much lower for NO than for CO due to the presence of one electron in the antibonding 277 orbital The temperature at which dissociation starts depends strongly on the NO coverage and is very sensitive to the surface structure For example when a NOcovered Rh331 surface is heated dissociation starts even at 240 K for a low NO coverage 21 However a temperature of 350 K is required for a surface that starts fully covered These results show that some of the molecularly adsorbed NO must desorb in order to create free sites needed for dissociation Van Hardeveld 22 investigated the adsorption of NO on Rh111 with secondary ion mass spectrometry SIMS a technique that measures the composition of the adsorbate layer and TDS At temperatures higher than 250 K a substantial fraction of the N O dissociates during adsorption At temperature higher than 350 K NONs is absent the dissociation is complete The decrease of the dissociation rate with increasing coverage can be explained by assuming that an ensemble of three or four empty sites is required for NO dissociation 23 The authors argued that a more likely explanation is that lateral interactions between NOads and 03dS or Nads cause the decrease in dissociation rate with increasing coverage EELS spectra 7 for NO on Ru001 at temperatures higher than 180 K are consistent with a model in which NO is adsorbed in bridged or threefold sites state 21 VNO 1500 cm l and in ontop sites state 12 VNO 1800 cm At low coverage NO is adsorbed in the more strongly bound 11 state and at coverages gt1 3 of saturation the ontop sites become populated Exposure at 280 K causes dissociation of a1 NO ESDIAD studies suggest that the N O bond is predominantly normal to the surface in both molecular states Ku et a1 For ref see 7 concluded from LEED observations following NO dissociation on a Ru1010 surface that the N and O adatorns form separate islands on this surface Many investigations indicate that molecularly adsorbed NO is highly inclined at low coverage whereas at high coverage this adsorbate has a perpendicular orientation Examples are NO on Ni111 NilOO Pt100 and Rh100 24 The tilted species is usually considered to be a precursor for dissociation Somorjai 25 Gorte et al 26 and Masel 27 found that NO dissocia tion is neglegible on perfect Pt111 surfaces Surface defects bind NO FIG 6 TDS of NO on Rh533 and 410 surfaces a NO from 533 b N from 533 0 NO from 410 and d N2 from 410 Heating rate 20 Ks39l reproduced with permission from Janssen et al 28 BERNARD E NIEUWENHUYS more tightly on Pt and induce dissociation Gorte et al reported that the Pt100 surface has stronger bonding and much higher decomposition activ ity than the Pt110 and 111 surfaces The major tightly bound state on Pt100 dissociates to yield 50 N2 and 02 whereas the fraction decom posed on the Pt111 surface is lt2 The stepped Pt410 surface with 100 terraces is much more reactive in NO dissociation than the 100 surface 27 On Pd111 the rate of NO dissociation becomes signi cant at temperatures higher than 500 K 7 In addition to desorption of NO NZ and 02 production of N20 has been observed on many Pd and Pt surfaces On Rh surfaces only desorption of N2 and NO is observed in the TDS Figure 6 shows TDS for NO on two stepped Rh surfaces Rh533 structure 4111 X 100 and Rh410 structure 4100 X 100 The adsorption of NO and the effect of preadsorbed O and N on the adsorption of NO have been studied on these surfaces and on many other Rh surfaces by Janssen et a1 28 The N atoms are markedly more strongly bound on 100 terraces than on 111 terraces The presence of steps does not affect the thermal stability of Nad on Rh At higher NO exposures repulsive N N and N O interactions lower the thermal stability of Nads Following saturation NO exposure N2 desorbs in a single state from 100 terraces at 750 K From 111 terraces several desorption states of nitrogen appear at temperaturs between 450 and 700 K An important observation is that recombination of Nad and Dad to give N O is more favorable than the 2 Nads gt N2 reaction when Rh with precovered Oads is exposed to NO The literature suggests that on Pd and Pt surfaces adsorption of N O is predominantly molecular at room temperature Dissociation is observed at elevated temperatures and the surface structure has a signi cant 39m u ence on the extent of dissociation In contrast to the adsorption of CO dissociation of NO can easily be detected on Ir Rb and Ru surfaces at room temperature B ADSORPTION OF N H20 C02 N20 AND NH3 Various techniques have been used to investigate the interaction of N2 H20 C02 and NH3 with clean metal surfaces The adsorption of N20 has not been studied in detail The interaction of C02 with group VIII metals was reviewed by Solymosi 29 At 80 K the adsorption on a Rh eld emitter exhibits an interesting crystal face dependency 30 In addition to molecular adsorption dissocia tion occurs at an appreciable rate on the stepped surfaces around 111 and 100 at temperatures higher than 220 K The eld electron microscopy FEM patterns suggest that the surface structure of Rh has a striking in uence on the ability of the metal to dissociate the C02 molecule From UNDERSTANDING AUTOMOTIVE EXHAUST CONVERSION Pt C02 desorbs at a much lower temperature 90 K corresponding to a heat of adsorption of about 20 kJmol and dissociation was not detected 30 No evidence was found for signi cant dissociation of NH3 on Pt111 and Ru001 under lowpressure conditions However decomposition of NH3 occurs at temperatures higher than 600 K in a NH3 atmosphere 3 The effect of the surface structure on the dissociation is large For Pt the order in reactivity is 210 gt 110 gt 100 gt 111 3 In the temperature range 80400 K nitrogen is only molecularly adsorbed on the group VIII metal surfaces with the exception of Fe and Co on which slow dissociation is observed on some faces at temperatures higher than 300 K 7 The nature of the adsorption bond is similar to that of the CO rnetal bond The heat of adsorption however is signi cantly lower for N2 40 kJmol than for CO 130 kJmol on the same surfaces as demonstrated by the data in Fig 7 It can be concluded from the available data that N2 H20 N20 and C02 molecules leave the surface immediately upon formation C COADSORPTION To understand the catalytic properties of the various group VIII metal surfaces the interactions among neighboring adsorbed species and their ros NzRh A P au gt I I l l 1 100 200 300 100 500 Temperature K FIG 7 TBS of nitrogen from a Rh lament reproduced with permission from Nieuwen huys 7 BERNARD E NIEUWENHUYS effects on the surface processes must be known since the surfaces may be covered with different kinds of adsorbates during the reaction Conrad et al 32 studied in detail the mutual interaction of coadsorbed O and CO on a Pd111 surface Some of their relevant results are summarized here Oxygen adsorption is inhibited by preadsorbed CO At coverages below Oco 13 LEED patterns show that O andCO form separate surface do mains However the behavior is different when 0 is preadsorbed CO can be adsorbed on the Pd111 surface covered with O which is less densely packed than a saturated CO layer The O adatom islands are then suppressed to domains of a V3 X V5R30 structure 6 13 with a much larger local coverage than can be reached with 0 alone which orders in a 2 X 2 structure 0 025 After further exposure the LEED patterns suggest the formation of mixed phases of Dads and COMs with local coverages of 90 Eco 05 which are embedded in CO domains When these mixed phases are present C02 is produced even at temperature lower than room temperature Coadsorption studies of other noble metal surfaces are consis tent with this scenario preadsorbed CO inhibits the dissociative adsorption of oxygen whereas CO is adsorbed on a surface covered with O Lambert and Comrie 33 concluded from TDS on Pt111 and 110 that gaseous CO displaces molecularly adsorbed NO a1 and also causes the conversion of this state to the other more weakly bound state 012 The a1 state was reported to be the only one which is reactive with CO Displacement of molecularly adsorbed NO by gaseous CO has also been observed on Pd surfaces Thiel et a1 34 used EELS to observe directly the competition of CO adsorption with one of the two molecular states of NO on Ru001 The stretching frequencies suggest that i CO is linearly bonded to a single Ru atom whereas NO prefers the bridging or multifold sites and ii at higher coverages linear sites are also populated The EEL spectra indicate that conversion from linear to multifold bonded NO occurs during adsorption of CO Similar observations have been made for CO NO mixtures on Pd110 35 For each single phase system the bridge site is preferred When the two are coadsorbed however a mixed phase is formed in which the stable CO site is linear In the presence of a CO NO atmo sphere Pt surfaces favor CO adsorption as expected on the basis of the heats of adsorption 36 Rh surfaces however adsorb both CO and NO 36 Blocking of hydrogen adsorption by NO is observed on Rh surfaces 21 The occurrence of strong repulsive or attractive interactions between coadsorbed H and NO has not been reported In conclusion the main effects found for coadsorption are blocking island formation displacement by the component with the higher heat of adsorption and site conversion Strong attractive interactions between the two adsorbates do not take place UNDERSTANDING AUTOMOTIVE EXHAUST CONVERSION 275 III 00 02 Reaction Usually temperatures of 300 600 K are required for the CO Oz reaction on the active metal catalysts Pt Pd Ir Rh Ru and Ni CO oxidation catalysts are used mainly for automotive emission control In addition air puri cation devices for respiratory protection and CO gas sensors com monly employ CO oxidation catalysts A novel application is in sealed C02 lasers used for weather monitoring and in other remotesensing applica tions For the development of longlife sealed C02 lasers novel catalysts must be developed which are able to oxidize CO at temperatures near ambient Active catalysts for this purpose are based on gold and a transition metal oxide eg Au manganese oxide 37 The mechanisms of the pro cesses responsible for the high activity are not fully understood In my opinion the reaction may occur at the Au MnOx interface with CO ad sorbed on Au reacting with O on MnOx An alternative explanation is that O atoms spill over from MnOx to gold Gold is not very active in 0 0 bond scission However 0 adatoms are stable on gold at room temperature In the presence of CO the O adatom can easily react with CO due to the low Au O and Au CO bond strengths Two mechanisms have been suggested for reactions such as oxidation of CO 1 An Eley Rideal mechanism in which one of the components reacts in the adsorbed state with the other molecule in the gaseous or in a physically adsorbed state eg Oads 39i COgas gt 2 Langmuir Hinshelwood mechanism in which both components react with each other in the adsorbed state eg Oads COads It has been established 38 that the dominant mechanism of the reaction is the Langmuir Hinshelwood L H mechanism CO 9 COads 02 gt ZOads Oads COads gt The alternative Eley Rideal mechanism fails to explain all the experimental results Direct evidence in favor of the L H mechanism has been obtained BERNARD E NIEUWENHUYS from molecular beam experiments Figure 8 39 shows a typical example of the variation in rate of C02 production with time for a CO beam imping ing on an O adlayer on Pd111 at a constant temperature of 374 K and also the variation of the surface concentrations Of Dads and cows The reaction exhibits an induction period and reaches its maximum rate after a few seconds Obviously the data of Fig 8 con rm that both reactants must be adsorbed in accordance with the L H mechanism The reaction may thus be written as dpcogdl kecoeo vecoeo CXp Eaeco where 0CD and 60 are respectively the surface coverages by cows and Oads Ea is the activation energy for the reaction which may vary with OCO and 00 k is the rate constant and v the frequency factor Both for Pd111 and for Pt111 E is equal to 101 kJmol at lowcoverage of CO and O 40 A much lower value of the activation energy 50 kJ was found when a Pt111 surface saturated with O was exposed to a CO molecular beam This decrease in E2 was attributed to a decrease in heats of adsorption of CO and 0 due to repulsive interactions At higher 0 coverages deviations from Eq 10 can be expected because Temperature K 10 11 dCO2dt Particles cm392 s391x1043 e a P 2 CO 0 articles cm x 1 Time 5 FIG 8 Rates of C02 formation on Pd111 and O and CO coverages as a function of time at 374 K reproduced with permission from Engel and Ertl 39 UNDERSTANDING AUTOMOTIVE EXHAUST CONVERSION 277 the O atoms form islands on the surface due to attractive interactions 41 Suf cient evidence exists that the reaction proceeds at the boundaries of O and CO islands 42 under both low and highpressure conditions and on Pt singlecrystal surfaces as well as supported catalysts 43 Recently very interesting results were reported by Wintterlin et al 44 The reaction was imaged on Pt111 by a variable temperature scanning tunneling microscopy SIM Small 0 islands with 2 X 2 structure and at high CO coverage CO adatoms were resolved The results show clear evidence of separate CO and 0 domains with the reaction taking place at the island boundaries The activation energy for C02 formation found from these microscopic measurements was equal to the value reported by Campbell et al 40 using macroscopic measurements Figure 9 shows the rate of C02 production catalyzed by the 111 100 410 and 210 surfaces of a Ptozs Rhojs single crystal as a function of temperature in the presence of a 2 1 mixture of CO and Oz at a total pressure of 2 X 10 7 mbar 45 It shows that the reaction rate increases rapidly between 400 and 500 K to a temperature Tm at which a maximum is reached and beyond which it gradually decreases The temperature at which the maximum occurs increases with increasing CO pressure Similar behavior was found for many Pt Pd Ir Rh and Ru surfaces Xray photo electron spectroscopy XPS measurements show that the maximum rate gt AP CLLJ C02 Temperature K gt FIG 9 Steadystate rates of C02 production for the C0 0 reaction on the 111 100 410 and 210 surfaces of a Ptogs Rhojs single crystal reproduced with permission from Siera et al 45 278 BERNARD E NIEUWENHUYS is reached at a temperature at which the CO coverage is already small 7 These and other observations indicate that adsorbed CO acts as an inhibitor for 0 adsorption and hence for the catalytic reaction The value of Tm is determined by the sticking probability the enthalpy of adsorption of oxy gen and the enthalpy of adsorption of CO The relatively small CO inhibi tion of the reaction on the 111 surface is consistent with the low heat of adsorption of CO on that surface Oscillations in the rate of C02 production have been observed for many supported metal catalysts and singlecrystal surfaces Similar oscillations have been observed for most of the reactions discussed in this review An example is shown in Fig 10 for the NO H2 reaction on Pt100 46 Several models have been proposed to explain these oscillatory rates Sales at al 47 associated the oscillation with a slow and reversible modi cation of the catalyst surface slow oxidation and reduction of the metal surface induces transitions between the two branches Ertl 38 reported in detail the oscillations on the 100 and 110 surfaces of Pt The clean 100 and 110 surfaces reconstruct ie the atomic ar rangement in the topmost layer is not that of the corresponding bulk plane However the reconstruction is lifted by adsorption of CO and NO when PN2 nu gt Time s gt FIG 10 Oscillatory behavior of the NO H2 reaction on Pt100 reproduced with permission from Cobden el al 46 UNDERSTANDING AUTOMOTIVE EXHAUST CONVERSION 279 the coverage exceeds a certain critical value The driving force is the higher heat of adsorption of CO on the nonreconstructed surface which overcom pensates the energy required for altering the surface structure The sticking probability of oxygen is negligibly small on the reconstructed surface whereas it is much higher on the nonreconstructed surface Ertl and coworkers 38 proposed the following model for oscillations The surface structure changes to the nonreconstructed surface due to CO adsorption Oxygen can then be adsorbed due to its relatively high sticking probability on this surface The adsorbed O reacts with adsorbed CO mole cules to yield C02 that is immediately released into the gas phase As a result more sites are created for oxygen adsorption and the reaction rate increases The CO coverage decreases because of its rapid reaction with 0 At the critical CO coverage the surface structure transforms back into the reconstructed phase Oxygen is not adsorbed anymore and hence the reaction rate decreases The CO coverage then increases and beyond the critical coverage the surface structure transforms again to the nonrecon structed phase This transformation completes the cycle of the oscillation Recent studies suggest that rate oscillations under isothermal conditions can also o39ccur on surfaces for which the reconstruction model does not apply A more general model was proposed involving the key role of vacant sites on the surface required for dissociation of molecules such as NO and Oz 48 For recent reviews concerning the oscillatory behavior of surface reactions see Refs 48 49 New regulations in the United States and Europe mandate that automo tive emissions must decrease substantially from current levels As a result there is a strong incentive to develop improved TWC with better oxidation activity at low temperatures since most of the hydrocarbons and C0 are emitted immediately following cold starts of engines As previously men tioned the addition of transition metal oxides can have a bene cial effect on the performance of Au catalysts in CO oxidation Combinations of Pt or Pd with transition metal oxides are also active in CO oxidation at low temperatures 50 Figure 11 shows examples of the reaction over PtMO SiOz catalysts Preoxidation does not signi cantly affect the temperature of 50 conver sion T5095 on PtSiOz Oxidation increases the T50 on PtCoOxSioz However T50 is still much lower for PtMO catalysts than for PtSiOz Similar effects were found for other COOz ratios These experiments were continued for many hours The T5095 of a PtCoOxSioz catalyst shifts to a value between T50 for the prereduced catalyst and the T5096 for the preoxi dized catalyst depending on the COOz ratio PtCoOxSiOZ is the most active of the catalysts investigated in our laboratory At room temperature CO is oxidized to C02 The value of T5095 are 100 200 K lower than those BERNARD E NIEUWENHUYS PtSiOZ C0202 2 39l D PtCoOxSiOZ RedCO02 2 l A PtCoOxSiOZ 0xCO02 2 1 0 co l ou 0 m 1 Conversion O 5 4L 02 Hquot 1 l I O 100 200 300 A00 Temperature C FIG 11 CO conversion at various temperatures during CO oxidation with CO 02 21 catalyzed by PtSiOz and by PtCoOxSioz after a reductive and an oxidative pretreatment reproduced with permission from Mergler et al 50t for PtSiOz depending on the COOz ratio CoOxSiOz does not exhibit signi cant activity at temperatures below 498 K The PtMO catalysts also exhibit improved performance in NO reduction reactions 51 53 IR 53 and temporal analysis of products TAP 54 have been used to investigate the origin of the improved performance of the PtMO catalysts in CO oxidation The TAP experiments shown in Fig 12 demonstrate that the high activity of PtCoOxSiOz in CO oxidation is related to the absence of CO inhibition effects at low temperatures On the basis of these results it was proposed that C00 is the supplier of O which reacts with CO adsorbed on Pt It is likely that the reaction takes place at the Pt CoOx in terface An interesting observation is that CO can form on Pt111 at the low temperature of 160 K upon heating the surface which is covered with UNDERSTANDING AUTOMOTIVE EXHAUST CONVERSION 022 a Intensity GU 011 0 6 39 10 I 360 39 A o 39 660 Time s 070 b i O gt 2 035 3 C O M ilLAMkrr I I i I 77 I I 1 o 37 75 112 150 Timeisi FIG 12 Multipulse experiments with PtSiOz and PtCoOxSiOZ C02 signal intensity 3 COlprecovered Pt catalyst at 327 K during 0 pulses b a COprecovered Pt CoO catalyst at 313 K during Oz pulses reproduced with permission from Mergler et a1 54 molecularly adsorbed 02 and CO 55 The authors showed that both 0 atoms in Ozvads react to give C02 Ozvads ZCOads gt The CO formation temperature coincides with the temperature at which Ozyads dissociates Therefore the origin of C02 formation at this low temper BERNARD E NIEUWENHUYS ature may be attributed to a reaction of the recently discovered hot 0 atoms with adsorbed CO Recent articles have discussed chemical energy which is released during dissociation and subsequent formation of the strong metal O bonds and is transformed in part into kinetic energy This excess kinetic energy could cause motion of the adsorbed particles or could induce chemical reaction STM is an obvious technique to use to characterize the distance that the adsorbed atoms can travel across the surface This tech nique was applied by Wintterlin et al 56 to investigate the diffusion distance of adsorbed O atoms when formed on Pt111 by 02 dissociation The authors found evidence for the existence of hot 0 atoms However the O atoms created by dissociation appear in pairs with an average distance of only two lattice atoms which is much smaller than that found by the same group 57 for dissociation of 02 molecules on Al111 On Al111 the distance exceeds 80 A 39 The existence of hot 0 adatoms was also believed to be responsible for the desorption of C02 at 140 K from the Pt100 hex surface partly covered with com and Duds Fadeev et al 58 used TPR and highresolution EELS HREELS to investigate CO oxidation on Pt100 Some of the results are summarized in Fig 13 At 90 K 02 is adsorbed both on the reconstructed Pt100 hex phase and on the nonreconstructed Pt100 1 X 1 surface as peroxide 02 with the 0 0 bond axis parallel to the surface This molecularly adsorbed 0 CO and 02 bl cmoz co2 reaction let CO adsorption at C02 Pt100hex 50 Pt100 hex 290K vtoz39 vc0 190K 350K 140K Intensity nu Intensity nu 7 10L CO I l l l I l I 17 O 500 1000 1500 2000 100 300 500 700 1 Energy Loss cm Temperature K FIG 13 Lowtemperature CO and 0 adsorption a and C02 desorption as a result of COWs 02Ms reaction b on Pt100hex surface adopted with permission from Fadeev et IL 58 UNDERSTANDING AUTOMOTIVE EXHAUST CONVERSION 283 state desorbs at temperatures near 140 K from the hex phase without dissociation However on the 1 X 1 surface partial dissociation occurs simultaneously with dissociation at 160 K Interaction of Dad with COMs results in the formation of two C02 peaks at 290 and 350 K In addition to these C02 formation peaks two lowtemperature C02 peaks were ob served at 140 and 190 K 39 It has been reported that the C02 molecules produced on Pt surfaces can eject into the gas phase with a translational vibrational and rotational energy in excess of that expected from the surface temperature 59 60 The group of King 61 applied the technique of singlecrystal adsorption microcalorimetry to the investigation of CO oxidation on Pt110 It was found that when CO is closed onto a saturated O overlayer the product C02 molecules have an additional 9 i 17 kJ mol 1 energy in excess of that expected for thermally accommodated molecules However when 02 is dosed onto a CO overlayer the product C02 molecules have an excess energy of 52 t 21 kJ mol l It was suggested that these highly excited C02 molecules are formed by reaction of cows with hot 0 adatoms produced by the 02 dissociation process Many of the results discussed in this paper were obtained on idealized models of real catalysts usually singlecrystal surfaces investigated at low pressure The obvious advantage of this approach is that these surfaces can bebharacterized on the molecularatomic level by use of surface science techniques Can we extrapolate these results to the behavior of supported catalysts with high surface areas at elevated pressures In this context the two main questions are i What is the effect of the pressure gap and ii What is the effect of the structure gap In many studies both questions have been addressed because of their general importance to the understand ing of catalysis with the aid of idealized models based on the ultrahigh vacuum surface science approach Fortunately suf cient information is available for the CO oxidation on both real and model catalysts to discuss these topics in detail for this reaction In modern surface science equipment the kinetics of the reaction can be followed in a microcatalytic reactor coupled to but isolated by a valve from the ultrahigh vacuum chamber The pressure gap in the kinetics is then eliminated and the structure effect can be examined The chemical and physical state of the model catalyst is analyzed in the ultrahigh vacuum chamber by transfer under vacuum from the reactor to the analysis chamber The CO oxidation is usually considered to be a typical example of a structure insensitive catalytic reaction For example the catalytic activity is almost equal on PtA1203 catalysts with widely varying particle sizes at high CO concentrations 7 However the reaction becomes structure sensitive in excess oxygen The observed temperature dependences of the BERNARD E NIEUWENHUYS steadystate rates are almost similar on Pd111 100 110 and 210 and polycrystalline Pd surfaces 38 Boudart 62 concluded that under similar experimental conditions the rate per Pd atom is equal on small Pd clusters 5 nm and Pd111 Hence data observed with a number of both singlecrystal and supported metal catalysts indicate that the reaction is essentially surface insensitive However Goodman et al 63 64 concluded that the CO oxidation on Pd is affected by the surface structure in a subtle manner These authors combined kinetic and re ection absorption infrared spectroscopy RAIRS studies of CO oxidation on Pd111 and 100 surfaces in the pressure range up to 10 mbar RAIRS was used to follow the CO coverage during reaction In the lowtemperature range studied with rstorder oxygen and negative rst order CO partial pressure dependencies of the rate the apparent acti vation energy equals the heat of CO adsorption The variation in heat of CO adsorption with coverage differs for Pdlll 110 and 100 For example at 0CD 050 the heats of adsorption are 71 139 and 122 kJ mol for the 111 110 and 100 surfaces respectively The authors argued that the reported apparent structure insensitivity of the reaction on Pd catalysts may be due to the varying CO coverages of the different surfaces under similar reaction conditions Clearly Fig 9 illustrates that the reaction may be very sensitive to the surface structure under certain experimental conditions A reactive mixture of CO and 02 could also modify the catalyst particle shape in such a way that one type of surface structure dominates leading to structure insensitivity Another explanation could be that only the activity of one type of site is measured under certain experimental conditions Results of Ramsier and Yates 5 are consistent with this model They found from temperature programmed reaction experiments that the C0 02 reaction takes place at lower temperatures on the terraces than on the steps of the Pt211 surface However at higher temperatures the high diffusion rate of O adatoms over the surface obscured the differences of step and terrace sites Detailed information is also available for the CO oxidation reaction on Rh Peden 65 Peden et al 66 and Oh eta 67 compared the turnover rates of CO oxidation on Rh111 and Rh100 surfaces with those obtained on supported Rh catalysts with high surface areas Figure 14 is a summary of some of the main results obtained for a CO and 02 pressure of about 10 2 bar Obviously there is no intrinsic sensitivity of the reaction to the surface structure under these experimental conditions Furthermore Fig 14 demonstrates that the singlecrystal surfaces are in fact very good models for the practical catalysts Su et al 68 investigated the reaction and the nature of CO adsorbed on Pt111 at pressures up to 1 bar characterizing the surface with sum BERNARD E NIEUWENHUYS o O O O CO Q31quot aim 3952 780K 0amp0 1 ago a 0 0 1100 a o 5 7 mbor angel s 0 9 U 0 lt7 0 w 642K 9 50 W bg 1 5 Mar o 039 o 0 565 6 3 lt12 C as 2 U LL m j 590K an em 17 mbar C O on 540K o 08 0 amp mbar 1800 1900 2000 2100 2200 2300 2400 Frequency cm391 FIG 15 In situ sum frequency generation spectra of CO oxidation on Pt111 at different temperatures The corresponding turnover rate TOR is also shown The initial reaction conditions were 130 mbar Oz50 mbar CO adopted with pemission from Su et al 68 cm 1 peak It was suggested that the metal carbonyl clusters are the reaction intermediates The activation energy of the reaction decreases from 176 kJmol 1 at temperatures below 600 K with a negative order of reaction in FCC to 59 kJ mol 1 at higher temperatures with a positive reaction order in Pm Su et al 68 concluded that the reaction mechanism changed at 600 K I believe that the observed change in apparent activation energy is in fact consistent with those reported earlier for lowpressure measure ments The Eact for the reaction 59 kJ mol l is within experimental accu racy equal to that found by Campbell et a1 40 for the C0 0 reaction on Pt111 at low pressure starting with an O covered surface STM studies show signi cant reconstruction of Pt110 in both CO and 02 whereas the surface structure of Pt111 exhibits only minor changes 68 As discussed previously at relatively low temperatures the surfaces of Pt Rh and Pd are predominantly covered by adsorbed CO At higher temperatures and or under strongly oxidizing conditions the metal surface UNDERSTANDING AUTOMOTIVE EXHAUST CONVERSION 287 can become oxidized CO oxidation rates on the surfaces of oxidized Pd and Rh are much lower than those on reduced surfaces This effect is large for Rh 65 However Pt showed no signi cant deactivation even under severely oxidizing conditions Bowker et a1 69 reported that the reaction is sensitive to the surface structure at high temperatures when Rh surfaces are covered by 0 Recent STM results demonstrated the high mobility of surface metal atoms as manifested in adsorbateinduced surface reconstruc tions in particular by oxygen CO can also induce changes in surface struc ture and even in morphology of the metal particles This effect was demon strated with the eld ion microscope An original hemispherial Rh tip becomes faceted by reaction with 10 4 Pa CO at 420 K The resulting tip shape is that of a polyhedron with large 100 and 111 facets 70 IV Reduction of NC by CO and H In Section ILA it was shown that the decomposition of NO proceeds on Pt Pd and Rh surfaces However it is not an ef cient process and probably contributes only slightly to the NO removal Exhaust gas contains the reducing gases CO H2 and various hydrocarbons with H2 produced by the water gas shift reaction co H20 s H2 co2 12 and by cracking of hydrocarbons It is assumed in many papers that CO is mainly responsible for the NO conversion 71 The role of he in conversion iNQhaslargelybeerenegleetedeeent worlrby van den Brink and McDonald 72 suggests that CO mainly contributes to the conversion of 02 whereas hcs have a signi cant role in NO conversion Hydrogen the concentration of which is about onethird that of CO in the exhaust is a more ef cient reducing gas for NO than CO at low temperatures 73 Unfortunately due to the presence of various gases in the exhaust dini trogen is just one of a variety of reaction products that can be formed in the catalytic converters used in automobiles The main undesirable reaction products are N20 and NH3 which are formed especially under reducing conditions Some of the processes that have been proposed involve ad sorbed isocyanate NCO and adsorbed HNCO as intermediates The for mation of NCOads complexes has been observed on supported metals by IRSpectroscopy It has been established that the presence and the Stability of the NCO groups depend on the support 74 It is most likely that the isocyanate species resides mainly on the support and its role is merely that of a spectator The reduction of NOx by he on noble metal surfaces has not been investi BERNARD E NIEUWENHUYS gated in detail using the surface science approach The recent nding that NO reduction by be signi cantly contributes to NO conversion is likely to stimulate fundamental investigations of NO hc interactions on metal surfaces IR studies by Bamwenda et IL 75 point to the presence of NCO and CN species on RhA1203 during the NO propene reaction Van Hardeveld et al 76 investigated the reaction of CZH4 and NO on Rh111 by temperatureprogrammed reaction spectroscopy TPR and SIMS No indication was found of a direct reaction between the molecular species The rst steps are NO dissociation and ethylene decomposition The dominant reaction products are as eXpected H20 C02 and N2 At low NO QH4 ratios lt3 signi cant amounts of H2 CO and HCN are produced Surface cyanide is formed by reaction of Cads with Nads in the absence of Oads Depending on the availability of hydrogen CNNs Can be hydrogenated to give HCN that desorbs at the reaction temperature In earlier papers on automotive catalysis the possible formation of HCN was proposed but it has not been detected in the exhaust gas Kobylinski and Taylor 73 studied the NO CO and NO H2 reactions on supported noble metals and found that the activity for the rst reaction increases in the order Pt lt Pd lt Rh lt Ru and that for the second reaction Ru lt Rh lt Pt lt Pd The rst reaction is slower than the second only for Ru was the order reversed Ru is an excellent catalyst for the NO reduction with a minimum of NH3 production However Ru forms volatile oxides under operating conditions resulting in an unacceptable catalyst loss The most ef cient catalyst appears to be Rh 1 Many techniques have been applied to examine the reaction pathways of the NO CO and NO Hz reactions and to elucidate the reaction mecha nisms In this section some of the relevant results are discussed with empha sis on the reaction mechanism Both the NO CO and NO Hz reactions are discussed here since it is likely that the mechanisms of N2 and N20 formation are independent of the type of reducing agent Note that CO dissociation is not considered to be involved in the mechanism However dissociative adsorption of NO into N and O adatoms is an important process on the relevant metals as discussed in Section ILA Possible mechanisms of N2 N20 and NH3 formation can then be evaluated on the basis of the following hypothetical mechanisms involving all the possible elementary steps in which NOads Nads Om COads and Hads can participate N2 formation 1 Without direct NO dissociation on the surface the reduction may proceed via a bimolecular reaction between two molecularly adsorbed N0 molecules 2N0 gt N 203ds UNDERSTANDING AUTOMOTIVE EXHAUST CONVERSION or via a bimolecular reaction of NOads with com 2NOads 2COadS gt N2 2COZ 13b For hydrogen as reducing gas a similar mechanism has been proposed via NOads HadS gt HNOads 13c followed by HNOadS Hads gt Nads H20 and combination of 2Nads to N2 II Dissociation of NOadS followed by combination of 2N adatoms 2Nads gt N 14 III Dissociation of NOads followed by reaction of Nads with NOads Nads NOads Oads 15 with N20 s as an intermediate The oxygen adatoms formed by 111 may then react with Hads or COMs as they do in the H2 02 and C0 02 reactions although the presence of NMs and NOads may modify the activation energies for these reactions due to lateral interactions The N2 C02 and H20 desorb as soon as they are formed IV A fourth proposed mechanism proceeds as follows ZNOads 2 Oads 2N02ads S N204ads N204ads gt Nz w 4 Oads39 17 NHS formation V Hydrogenation of molecularly adsorbed NO with HNOads as an in termediate NOadS 3HadS gt NH3 Oads 18 VI Hydrogenation of Nads formed by NOads dissociation Nads Hads quot9 NHads NHads Hads NH2ads NH2 Hads gt NH3ads 21 NHMds gt NHngas 22 290 BERNARD E NIEU WENHUYS N20 formation VII Reaction of Nads with NOads NONs Nads gt N20 23 VIII Via reaction between two molecularly adsorbed N0 molecules 2N0 S gt N20 Oads 24 IX Via dissociation of N Oads and reactions between two Nads and Om 2N s Oads gt N20 25 The steps listed previously are the possible steps at very low NO conver sions At higher NO conversions reactions of N2 NH3 N20 and H20 with each other and the reactants should also be considered In particular decomposition of NH and N20 and reduction of N20 may contribute to N2 formation N20 gt NZOMS gt N Om 26 XI NH3 gt NHMds 27 2NH339adS gt N2 6Had5 28 via NHz ads NHads Nads and recombination of 2N8 In addition to the previous steps 02 and NO formation via 2Oads 02 29 and XIII Nads Oads gt NOads have been documented in the literatuni39e In the presence of both H2 and CO as reducing agents the situation will be even more complicated due to competition between C0 and H2 and possibly formation of speci c products and intermediates such as HCN This topic was addressed brie y at the beginning of this section Ample evidence exists for the major role of process II in N2 formation on the noble metal surfaces with the possible exception of Pt111 as discussed quotlater Figure 5 shows that TDS of NO from a Rh lament following UNDERSTANDING AUTOMOTIVE EXHAUST CONVERSION a saturation exposure to NO exhibits two distinct N2 peaks BNz with a maximum at about 625 K and aNg with a maximum at about 500 K Similar behavior has been reported for Rh singlecrystal surfaces although the TDS of nitrogen from Rh100like surfaces differ signi cantly from those of the Rh111like surfaces 28 as has been illustrated by the data of Fig 6 At low NO exposures no NO desorbs and N2 is evolved only in the B peak which exhibits secondorder desorption Following higher NO exposures aNz desorption is observed at those NO COVerages at which N O desorption is also observed with a peak maximum only slightly lower than the aNz peak On the basis of these results many authors assigned the B peak to N N recombination process II and the a peak to process III N20 formation has been found for some supported Rh catalysts at the temperature regime of the oz Nz and NO desorption peaks 77 This correlation between aNz N20 and NO desorption may point to a common N201ike surface intermediate for aNz and N20 Lambert and Comrie 33 investigated the CO NO reaction on Pt111 and 110 surfaces and concluded that the reaction proceeds by a L H mechanism between Dads and molecular CONS NO dissociation is also the prime step in the NO H2 reaction on a Pt foil at a pressure of 10 7 mbar 7 NH Was found to be the major product at temperatures lower than 600 K and N2 was the major product at temperatures higher than 600 K when the NOHz ratio was 15 Siera et a1 45 investigated the C0 NO reaction on the same surfaces and under the same experimental conditions as described for the C0 02 reaction At low temperatures lt500 K the reaction rate decreases in the order Pt Rh111 gt 100 gt 410 gt 210 This is the same order as was observed for the C0 02 reaction and it corresponds to the order in enthalpy of adsorption of CO which increases from the 111 to the 210 surface NO dissociation decreases in the order 410 210 gt 100 gt 111 For both reactions the rate at low temperatures is controlled by CO desorption and not by NO dissociation activity Similar effects were found for the NO reduction by hydrogen on Rh singlecrystal surfaces Wolf et a1 21 studied the NO H2 reaction on several Rh surfaces by means of a scanning eld emission probehole micro scope The temperature at which the reaction starts is strongly dependent on the NO coverage and is sensitive to the surface structure At low NO precoverages and with hydrogen in the gas phase the reaction starts already at 240 K on Rh331 a surface with 111 terraces and 111like steps On Rh211 a surface with 111 terraces and 100 steps the reaction starts at a signi cantly higher temperature On Rh321 a surface with 111 terraces and both 100 and 111type steps the reaction features charac teristic of both 111 and 100 steps are observed During reaction an BERNARD E NIEUWENHUYS electropositive adsorbate assigned as NHxads is formed that dissociates at a temperature of approximately 300 K At high NO precoverages NO desorption is the initiation process vacancies required for NO dissociation and hydrogen adsorption are then created Under these conditions surface structural effects are related to differences in enthalpy of adsorption of NO However hydrogen has a bene cial effect on the initiation temperature increases of hydrogen pressure result in decreases of the onset temperature of the reaction This effect can be explained in terms of hydrogenassisted NO desorption ie displacement of NOadS by hydrogen It is not a direct hydrogenassisted NO dissociation for example by formation of HNOads At low NO precoverages the in uence of the surface structure is related to the intrinsic d ferences in NO dissociation reactivity of the various surfaces The observed onset temperatures of reaction did not differ signi cantly from the NO dissociatiOn temperatures The authors concluded that hydrogenassisted NO dissociation proposed by Chin and Bell 77 does not seem to be important under their conditions 21 Recently spectroscopic techniques and the use of isotopically labeled N have provided a wealth of novel information about the mechanisms of N2 NH3 and N20 formation It is illustrative to discuss the detailed studies reported by Hirano et al 78 80 concerning the mechanisms of the NO H2 reactions in light of recent studies The surface investigated in most detail by Hirano et al was a Pt Rh100 alloy surface 75 Rh and 25 Pt in the bulk The 1 X 1 structure representative of the clean Pt Rh100 surface changes into the c2 X 2 surface structure after exposure to a reaction mixture of 1 X 10 7 mbar NO and S X 10 8 mbar H2 Analysis of the c2 X 2 surface by means of TDS auger electron spectroscopy ABS and HREELS indicated that atomic nitrogen was the main species present For comparison similar experiments were carried out on the Rh100 Pd100 and Pt100 surfaces 7880 Figure 16 illustrates some of the re sults Accumulated N atoms ordered in the c2 X 2 surface structure were found on both Pt Rh100 and Rh100 but not on Pt100 In the presence of H2 or D2 at a crystal temperature of 400 K a nitrogen hydrogen vibration was found Its intensity depends strongly on the partial pressure of hydro gen Evacuation of H2 and D2 results in a quick loss of the signal On the basis of these results it was concluded that hydrogenation of N can occur via the reaction Nuds xI lads S NHxads39 The hydrogenation of atomic nitrogen is largely reversible By AES it was established that the NRh signal ratio was decreased only slightly UNDERSTANDING AUTOMOTIVE EXHAUST CONVERSION 293 NDX NHx 2420 3240 PiRhmo M 1 Counts Rhl lOO l l l 0 2000 4000 Energy Loss cm 1 FIG 16 In situ EEL spectra of c2 X 2 N on the Rh100 and Pt Rh100 surface in equilibrium with a hydrogen deuterium mixture at 10 7 mbar and a temperature of 400 K The intensity of the N H signal is 16 times larger on Rh100 than on Pt Rh100 reproduced with permission from Nieuwenhuys et a1 80 during the hydrogen treatment This result indicates that complete hydro genation of the atomic nitrogen to NH can occur but with a very low rate and only under these experimental conditions The intensity of the N H EELS signal varies with the square root of the hydrogen pressure INH Furthermore N H scissor vibrations were absent from the EELS spectra On the basis of these observations it was concluded that the dominant NHx adsorption complex is NHMs and not NHZYMIs or NHlads 294 BERNARD E NIEUWENHUYS Attempts to form a N layer on the Pt100 surface by means of the NO Hz reaction were unsuccessful The NO and H2 partial pressures were varied from 10398 to 10 mbar and the temperature from 400 to 600 K Formation of a nitrogen overlayer was not observed under these conditions Uscful information concerning the reactivity of adsorbed species is ob tained by TPR In the presence of a NO hydrogen H2 or D2 ow 2 X 10 7 mbar the greatest ammonia formation was observed at approx imately 500 K heating rate 3 K 54 78 80 However if the surface was covered with 15N prior to exposure to the 14NO hydrogen ow ammonia resulting from 15N was found even at 435 K 78 These results show that N atoms on the surface can react with hydrogen to give ammonia at 435 K On the basis of these observations and the HREELS results it was concluded that hydrogenation of atomic nitrogen via NHMS is most likely the reaction route to ammonia formation process VI Evidence for the presence of NHmds species has been obtained for many surfaces including those of Pt Rh 78 80 81 Pt 82 84 Rh 70 81 85 Pd 8 and Ru 86 Sun et a1 84 investigated the stability of NHx species on Pt111 formed by electroninduced dissociation of NHgmS using HREELS and XPS NHMs formed following electron irradiation at 100 K Upon heating many of the NHZMs species are rehydrogenated and desorb as NH3 at temperatures of about 200 K whereas a small fraction dehydrogenates forming NHads at temperatures near 300 K At tempera tures exceeding 400 K NH S dehydrogenates leaving NMS on the surface which then desorbs as N2 at temperatures between 500 and 700 K Recent SIMS spectra of the Rh111 surfaces obtained during hydrogena tion of atomic Nads indicate that Nads and NH2ads not NHads are the predominant surface intermediates on Rh111 85 According to the au thors the third hydrogenation step in process VI ie the hydrogenation of NHZMs to give NHlads is rate limiting for NH3 formation with an activation energy of 69 kJmol Because of the interest in ruthenium as a potential catalyst for ammonia synthesis from N2 and Hz the hydrogenation of Ned on Ru surfaces has been investigated 86 Both NHads and NHL s were found in addition to NH3ad3 Dietrich et al 86 reported that the thermal stability of NHMs is the highest of the three NHWs species x 1 2 and 3 on Ru0001 and on Ru1121 at temperatures up to 400 450 K On Ru 1010 however the thermal stability of NHL is higher than that of NHads and much higher than those on other Ru surfaces The reason for the enhanced thermal stability of NHL on Ru1010 is not clear It was also found that coad sorbed N has a positive effect on the thermal stability of NHMIs on Ru0001 whereas NHMs is stable at temperatures up to 400 K and NHads in the NNH coadsorbate is stable at temperatures up to 460 K UNDERSTANDING AUTOMOTIVE EXHAUST CONVERSION 29S Yamada and Tanaka 81 found that the N c2 X 2 structure on the 100 surfaces of Pd Rh and Pt Rh does not change signi cantly upon the formation of NHMs in the presence of hydrogen This observation was interpreted on the basis of formation of NHMs at the borders of c2 X 2 islands during exposure to hydrogen An interesting effect of the surface structure was also found for Pt by Zemlyanov et a1 82 83 HREELS and TDS con rmed that NO desorbs completely from Pt111 without dissociation However N2 and NH3 forma tion were found upon heating the NOcovered Ptl 11 surface in a hydrogen atmosphere HREELS results were interpreted in terms of formation of HNOads It was proposed that NH3 and N2 are formed without direct dissociation of NO HNOads Hads gt Nads H20 followed by reaction 11 and hydrogenation of Nads to NH3 On Pt100 NMs formed by NO dissociation at 300 K reacts readily with hydrogen to give NHZMs on the Pt100 1 X 1 surface and to give NHMs on the reconstructed Pt100hex surface It was proposed that the NHadS and NHZMs species are the intermediates for NH3 formation However NH3 formation was not observed under the experimental conditions the only product observed was dinitrogen The presence of HNOad5 was also proposed by Williams et a1 87 on the basis of surfaceenhanced Raman spectra SERS as an intermediate for the NO H2 reaction on polycrystalline Rh lms The authors demon strated that SERS can be used to characterize adsorbed species for C0 NO and NO H2 reaction 87 88 on ultrathin Rh lms deposited on a roughened gold substrate The Rh surface contained carbonaceous con taminants and possibly gold atoms This type of measurement is interesting because it yields vibrational data in situ even at high pressures In the presence of a NO N2 ow ratio NOHz 1 total pressure1 bar a band characteristic of Nads is present and at temperatures from 420 to about 620 K a band attributed to HNOads is present The feature observed was attributed to HNOads because the surface species was formed only from NOadS and hydrogen and not from Nads and hydrogen The authors proposed an additional pathway for NO dissociation illustrated by Eq 33 This proposed process may explain the hydrogenassisted NO dissocia tion rst proposed by Hecker and Bell 89 Investigations of Rh single crystal surfaces have also indicated that the presence of hydrogen may have a bene cial effect on NO dissociation 21 90 It was also proposed that this alternative process for NO dissociation may be fast under conditions of slow direct decomposition because of the absence of free adjacent sites needed for NO dissociation 87 It was argued that a high coverage of BERNARD E NIEUWENHUYS Had may block the vacant sites and slow the direct dissociation Blocking of sites needed for NO dissociation by hydrogen is an unlikely process in contrast to blocking by NO CO N or 0 Furthermore the hydrogen assisted NO dissociation can also be attributed to the removal of OM and or Nads by hydrogen 21 The NO H2 reaction catalyzed by Rh533 which is composed of four atomwide 111 terraces separated by 100 steps was investigated in situ by Cobden et al 9 in the 10 6 mbar pressure range by using fast XPS and mass spectrometry The emphasis of this study was on understanding the oscillatory behavior of the NO H2 reaction on Rh surfaces Some of the results are also relevant to the current discussion It was shown that NO s is not present on the surface in the temperature range in which N2 and NH3 formation is observed Two different Nads species were observed in addition to Oads One NMS species N1 is probably Ned on the 111 terraces and the other N 11 is either NadS adsorbed on the steps or NHmdS OadS plays an important role in controlling the surface reactivity Oads pres ent in small amounts destabilizes the NI species and at high OmS coverage it favors the formation of N11 The results summarized previously show that the surface structure and the presence of coadsorbed species have a large effect on the relative stability of NH S and NH2d However more detailed studies are required to understand the relative stabilities of NH and NH on noble metal sur faces In conclusion there is ample evidence that process V1 is a major mechanism for NH formation The intermediates N H have been identi ed The only possible exception may be Pt111 As stated in Section ILA NO dissociation is negligible on Pt111 at low pressure However PtIOO is active in NO bond breaking Polycrystalline Pt surfaces have low activity for NO dissociation The group of Schmidt 92 93 investigated the NO CO reaction on polycrystalline and single crystal Pt surfaces At low pressures Pt111 is unreactive for the reaction The kinetics of the reaction was investigated on polycrystalline Pt at temper atures from 300 to 1200 K and pressures from 10 8 to 1 mbar The authors proposed that the reaction is a true bimolecular reaction between NOads and CO Eq 13b rather than NO decomposition with CO scavenging of Oads mechanism 11 For the reactive Pt100 surface the reaction of NO H2 was very similar to that of NO CO and mechanism 11 qualita tively explains the obServations The possible relevance of mechanisms I and IV to the formation of N2 in TWC has not been addressed except for polycrystalline Pt Mechanism IV has only been added for the sake of completeness since it has not been proposed for NO reduction catalyzed by TWC This and related mechanisms have been proposed for the selective catalytic reduction of NOx by hydrocar UNDERSTANDING AUTOMOTIVE EXHAUST CONVERSION 297 bons 94 96 and hydrogen 97 in the presence of 02 and catalyzed by various zeolites containing metal ions in their cavities The rate of N2 formation is enhanced in the presence of excess 02 In several proposed mechanisms NOW 5 with y gt 2 plays a key role These mechanisms although probably not important for TWC may play a role when dissocia tion of NO cannot occur due to the absence of adjacent metal atoms required for NO dissociation Shelef and Graham 3 proposed that the high selectivity for N2 production that distinguishes Rh from Pt and Pd catalysts may be rooted in the promotion of N pairing in 2N0 s molecules before the N O bond is broken These authors noted that IR spectra of NO on welldispersed RhA1203 catalysts point to the presence of dinitrosyl NO2ads a species that is not found on Pt and PdA1203 catalysts These adsorbed dinitrosyl species could represent a locus for the event of pairing of the nitrogen atoms It is probable that NO2adS is formed on ionic Rh sites No experimental evidence supporting this mechanism is available for TWC The formation rates of N2 and N20 on Pt Rh100 as a function of crystal temperature are shown in Fig 17a for a mixture of 12 X 10 7 mbar NO and 36 X 10 7 mbar H2 under steadystate conditions 78 80 The N2 and NH3 formation rates increase rapidly as temperature rises from 400 K until the maximum rates are obtained at approximately 600 K for N2 and at the signi cantly lower temperature of about 525 K for NHa Nitrous oxide N20 is only a minor product under these conditions N2 production remains high and almost constant until a temperature of about 1000 K is reached indicating that the reaction rate in this temperature range may be controlled by the collision frequency of NO on the surface At higher temperatures the formation rate drops For reaction mixtures with excess hydrogen the N2 formation rate is almost independent of the hydrogen pressure zero order at temperatures from 600 to 800 K The ammonia production is approximately rst order in hydrogen over the whole temperature range of 400 to 800 K N20 formation is observed only at relatively high NO pressures and it is only slightly affected by changing the hydrogen pressure Under the conditions of these experiments PNO lt 5 X 10 6 mbar the N2 formation remains at least a factor of 100 larger than the N20 production rate In another series of experiments of H2 in the gas ow was replaced by NH and the 14NO by 15NO to allow examination of the distribution of the two types of N atoms over the Ncontaining products Figure 17b shows the variation of dinitrogen formation rates with increasing temperature in the presence of a ow of 12 X 10 7 mbar of 15NO and 30 X 10 7 mbar of 14NH3 In the lowtemperature range T lt 800 K the rate of 14N15N formation is higher than the rate of 15N15N formation which in turn is much BERNARD E NIEUWENHUYS a 2395 Pt RhIlOOI E o g Tm 012 o 0quot LL I 0 U 515 0 a 3 2 0 E 0 3 NH3 0 o 905 QDDDDUDUDUDDDQ m y u 5 c N20 0 00 5 Fu9rm l i 4 r l Ij f 39 00 800 1200 Temperature K C 10 2 A 1b Pt RhHOOI S E w 7 u T E 29 395 DECIDED S U CI 30 l E 039 30 a o 9 D39AAAAA AA D a i i g v A AAA A A A28 A DU 0 I I I I I I I I I r 400 800 1200 Temperature K FIG 17 a Steadystate formation of N2 0 NH3 Cl and N2 A on the Pt Rh100 surface in the presence of 12 X 10 7 mbar NO and 36 x 10 7 mbar Hz b Steadystate formation rates of N on PtRh100 with amu A 29 CI and 30 O in the presence of 12 X 10 7 mbar 15N0 and 30 X 10 7 mbar NHg reproduced with permission from Hirano et al 78 higher than the rate of 14N14N formation The 14N 1 NH3 concentration in the gas phase was a factor of 25 larger than the 15N 15NO concentration The relatively low production of 14NMN is in qualitative agreement with results reported by Otto et al 98 for the NO NII3 reaction catalyzed by supported Pt Using N isotopes these authors showed that N2 formed UNDERSTANDING AUTOMOTIVE EXHAUST CONVERSION 299 on Pt exclusively from NH3 is only a minor path Under the conditions used in the experiments shown in Fig 17b N2 is formed on Pt Rh100 exclusively from NH3 albeit at a much lower rate than from N0 NH3 and from N0 only At temperatures higher than 700 K the rate of 14leN formation decreases rapidly with increasing temperature and the 15leN production increases The 14NMN production shows the same temperature dependence as the 1 N15N production The temperature dependence of the N2 formation is the same for the NH3 15NO and the H2 NO reactions suggesting that the same processes are involved The Ncombination reactions can occur over the whole tem perature range as demonstrated by the rate of 14N2 formation The produc tion of 14N2 is much smaller than the production of 14N15N and 15N2 whereas the 14NH3 concentration in the gas phase is a factor of 25 greater than that of 15N0 This comparison may indicate that NH3 is a slower producer of N ds than is NO However this argument is in contradiction with the high rate of 14NISN formation It was suggested that the relatively high rate of production of 14NISN may point to a large contribution of the reaction of Nads with NOads process III to the N2 formation in the lower temperature range T lt 700 K At 1100 K the rate of 14N2 formation is a factor of 15 less than at 600 K the rate of 14N15N formation is a factor of 25 less and the rate of 15N2 formation only slightly lower On the basis of this result it was concluded that in the hightemperature range 1100K the impor tan39Ce of the N combination reaction is much greater than in the low temperature range and that ammonia becomes a less effective N producer than NO The same authors investigated the NO H2 reaction catalyzed by Pt Rh100 in the 10mbar pressure range 78 80 Figure 18 shows some results obtained at 550 K At a temperature less than 500 K and a NOHz ratio of 15 most of the NO reacts at low conversions to give N2 and the formation of NH3 is greater than that of N20 However as the temperature increaSes above 500 K the NH3 production exceeds the N2 production and N20 formation essentially ceases The rate of NO conversion is much lower for a NOIIz ratio of unity than for the ratio of 15 N2 production is only slightly affected by lowering of the H2 pressure However NH3 formation decreases drastically as a result of lowering of the hydrogen pressure Consequently the selectivity toward dinitrogen is much improved by lowering of the H2 NO ratio The hydrogen pressure was varied from 12 mbar to 24 mbar at a constant NO pressure of 12 mbar It appeared that the reaction order in hydrogen is essentially zero for formation of N2 and N20 and about unity for formation of NH3 under these conditions It is emphasized that NO NH3 and N20 decomposition and the NO H2 and NH3 NO 300 BERNARD E NIEUWENHUYS U o E w L 3 In In 3 CL E t CE Time min 010 b a 0 Pi RhiiOO D 600K 15NO r e o S l u i0395 0 4 E 005 I O 0 O o OiLNISNO r o 00 O L a 39 39d39 39 5N2o B O D E I H DUI H Ci N20 C 0quotquotrquot quotIquot AIJPLII LAI PI I I I 0 Z 0 D 11 15 0201 15 0000 N N 39 3 NO o 05 3 A D o E d D 15 VN 01cc OBIIII N2 c 3912 j 0 I DAA A1LN2 P A o 2AA DODD 39 IAA 0 r I I m l I I I I 0 0 5 10 Time min FIG 18 a Formation of N2 NHg and N20 on Pt Rh100 at 550 K at a NOHz ratio of 15 with total pressure of 3 mbar b The observed reaction products for the 15N0 lquotNH reaction on Pt Rh100 as a function of the reaction time at 600 K with a lsNO NH ratio of 1 and total pressure of 3 mbar reproduced with permission from Hirano el al 78 reactions can occur simultaneously Hence useful information concerning one speci c reaction is obtained only at the beginning of the reaction when the conversion is still low The most striking differences between the low and higher pressure regimes are the following i The relatively high selectivity to N20 in the UNDERSTANDING AUTOMOTIVE EXHAUST CONVERSION 301 higher pressure regime and ii the relatively high selectivity to ammonia The rst observation is completely in line with the expectations based on the mechanism for N20 formation via process VII requiring a high concentration of molecularly adsorbed NO The dissociation of NO is inhib ited by high concentrations of adsorbed N O as discussed previously It has also been reported that during the NO hydrogen reaction a silicasupported Rh catalyst is largely covered with NO On the other hand in the low pressureregime Rh surfaces are covered with Nads As the temperature increases the concentration of NOMS decreases as a result of desorption and the selectivity to N20 is expected to decrease as observed The same argument can be used to interpret the relatively high selectivity to NH3 A high concentration of NMs favors a high selectivity to N2 via process 11 A low concentration of Nads favors the formation of NH process VI Obviously the Nads concentration at 600 K in the 10mbar pressure regime is lower than that in the 10 7 mbar regime It has been stated that the selectivity for NH formation on supported metals such as Pt and Rh increases with increasing temperature up to a maximum Tm beyond which it decreases for Pd and Pt Trn is about 600 K The maximum temperature used in one investigation 78 was 600 K and hence the increasing selectiv ity to NH3 observed when the temperature increases is in agreement with these data The 15NO 14NH3 reaction was also investigated in the pres sure range of 10 mbar 78 Some results are shown in Fig 18b for a 15N014NH3 ratio of 1 at 600 K 14N20 is not formed whereas 14N15NO and 15N20 are both formed showing that N20 is formed only via process VII At 600 K the formation of 14NISNO is faster than that of ISNZO which suggests that NH3 is a better producer of N than NC at this temperature N2 formation is not detected at 550 K whereas at 600 K the rate of N2 formation is much less than that of 14N15N and 15N2 On the basis of this observation it was suggested that process III is the dominant path for N2 production in the lowtemperature regime The molecular picture that the authors proposed from their results is the following 78 80 At low temperatures T lt 450 K the majority of the adsorbed species are NO and N is only a minor species Adsorbed hydrogen is very mobile and can easily react with either 0 or N provided it can nd a vacant site in the neighborhood of an adsorbed O or N atom The concentration of vacant sites is low in this temperature range As the temperature increases the concentration of NO becomes lower and the concentration of Nads increases When the NO concentration becomes much lower than the Nads concentration the rate of process VII and hence the selectivity to N20 becomes low For N2 formation via process III high concentrations of both Nad and NOadS are required for N2 formation via process 11 a high concentration of Nad5 is required Consequently process 302 BERNARD E NIEUWENHUYS III may contribute to the N2 formation in the lower temperature range and process II dominates in the higher temperature range The selectivity to NH3 increases with increasing temperature in the temperature range 300 600 K again suggesting that the dominant mechanism of NH3 forma tion is via process VI and not via hydrogenation of Now The selectivity of NH will be the greatest at a temperature at which both the NMs and Had5 concentrations are suf ciently high The strong positive reaction order in hydrogen pressure shows that the concentration of HadS is the limiting factor under the experimental conditions At some very high temperature the reaction rates of both NH3 and N2 formation become low because of the low Nads and Hads concentrations Obviously this temperature depends strongly on the absolute NO and H2 pressures and on the metal surface Rh with its much stronger metal N bond strength will exhibit a much better selectivity toward N2 than Pt which has a weaker metal N bond strength The observed decrease in selectivity toward NH3 as temperature becomes very high can be attributed to a combination of limited availability of H at high temperature and the high rate of NH decomposition In contrast to the apparent structure insensitivity of the CO 02 reaction catalyzed by Rh and reported by Peden 65 Peden et a1 66 and Oh et al 67 these researchers found substantially different kinetic behavior for the NO CO reaction on Rh111 Rh100 and supported RhA1203 66 67 as illustrated in the Arrhenius plots of Fig 19 The corresponding activation energies are 124 kJmol for Rh111 and 101 kJmol for Rh100 The authors postulated that the structure sensitivity of the C0 NO reaction catalyzed by Rh is due to different rates of NO dissociation On Rh111 the rate would be limited by the formation of N from process 11 and the surface would be largely covered with Nads On the supported catalyst the ratelimiting step was judged to be NO dissociation STM is an attractive technique for elucidating reaction mechanisms since in addition to the observation of the surface structure it allows imaging of adsorbed atoms and molecules under some conditions Many recent exam ples indeed show that STM can provide useful additional information 44 99 100 Leibsle et al 100 observed a onedimensional C0 0 reaction on Rh110 Xu and Ng 99 imaged NONs molecules coadsorbed with O or N atoms on Rh111 NOMs molecules form ordered islands with 4 x 4 structure Repulsive interactions between NO and O or N lead to segregation of NO and O or N islands The results also suggest the formation of Rh N1 X 2 added row reconstruction through NO dissociation Some confusion exists in the literature concerning N20 formation as a product of the C0 NO reaction In the papers of Oh et a1 67 and Peden et al 66 N20 formation was not reported for the C0 NO reaction on Rh singlecrystal surfaces However Hecker and Bell 101 UNDERSTANDING AUTOMOTIVE EXHAUST CONVERSION 303 103 co NO gtC02 12 N2 A PCO PNO10mbar LL O 02 I Rh 100 I 1 o Rh 111 B Rh Al O 203 m C E 1011 U E 0 LL N 0 8 10 104 12 16 39 1398 20 3 1 1 IT x 10 K 1 FIG 19 Comparison of the speci c rates of the NOCO reaction on Rh111 Rh100 and RhA1203 at Pco PNO 10 2 bar adapted with permission from Oh et a1 67 reported that N 20 is a major Ncontaining product of the C0 NO reaction catalyzed by RhSiOz Belton and Schmieg 102 studied the reaction on a Rh111 catalyst and found that N20 is formed with a selectivity of 70 from a mixture of 106 kPa of NO and 106 kPa of CO at temperatures between 525 and 675 K Both Rh111 and RhSiOz give similar product distributions with more N20 than N2 and an activation energy of 139 k at temperatures higher than 480 K To what extent does reaction III contribute to N2 formation Whereas Hirano etal 78 80 concluded from that reaction III may make a signi cant contribution at low temperatures Belton et al 103 concluded that there is no experimental evidence for the reaction step NOads Nerds 9 NZOads 9 N2 Oads and they recommended that in future NO reduction mechanisms this step should be omitted Belton et al prepared a Rh111 surface covered with 04 monolayers Of 15N by electron beam dissociation of 15N39Oad5 and satu rated this surface with 14NO at 200 K TDS did not reveal 15NO showing that scrambling of 15N and 14NO did not occur during their experiments In addition all three N2 isotopic mixtures masses 28 29 and 30 have 304 BERNARD E NIEUWENHUYS been observed in the TDS indicating that the aNz feature is also due to N atom recombination The apparent rst order behavior of the aN2 peak and the correlation of aNz and NO desorption were attributed to rate limiting formation of N by the dissociation of NONS at high NO coverage Now inhibition of its own dissociation Another view was presented by Borg et al 23 and by Makeev and Slinko 104 Repulsive interactions in the adlayer are required to produce a Nz Bugyi and Solymosi 105 prepared a N overlayer with high coverage on Rh111 from atomic N in the gas phase by means of a discharge tube and observed a Nz and 8 Nz desorption states very similar to those observed after NO exposure Apparently the presence of NOW is not required to produce aN2 in TDS Makeev and Slinko 104 were able to simulate the TDS reported by Root et al 106 and Borg et al 23 of N2 and NO from Rh111 following NO exposure on the basis of repulsive interaction between Nads and NOW together with the strong inhibition of NO dissociation by NOads Nads and Oads The NO CO reaction on Rh111 was investigated by Permana et al 107 by RAIRS The measurements were performed with NO and CO pressures in the 10mbar range and hence in the same range as in the automotive exhaust gas contacting the TWC IR spectra taken under reac tion conditions showed only atop CO and multiply bonded either twofold or threefold NO Changes in the surface correrages of NOMS and C0 s correlated well with the observed changes in N20 selectivity at tempera tures below 635 K NOadS dominates and N20 formation is favored At temperatures above 635 K at which N2 formation is preferred CO is the majority surface species Obviously these RAIRS data support the model according to which N20 and N2 are formed by parallel pathways by reaction of Nads with either NONS or Nads respectively By adding 15N20gas to the 1 NO and 12CO reactant mixture it was shown that 15N20 was not consumed during the reaction Therefore readsorption of N20 is not an important path to produce N2 under the conditions used in Permana et al 107 650 K and low conversion However the same authors also showed that N20 is readily converted to N2 at temperatures higher than 700 K 102 The C0 N20 reaction runs only after complete NO conversion at tempera tures higher than 700 K The mechanism and kinetics of the NO CO reaction on Rh111 have been discussed in detail by Zhdanov and Kasemo 108 They showed that simulations based on surface science data obtained at low pressures reproduce the scale of the reaction rate at the pressure regime of interest for the TWC but fail to predict accurately the apparent activation energy and reaction orders Can we conclude that process III does not at all contribute to N2 forma UNDERSTANDING AUTOMOTIVE EXHAUST CONVERSION 305 tion In this context it is relevant to review recent work of Ikai et a1 109 111 on N desorption from Pd110 They shOWed that the behavior of Pd110 toward NO adsorption is very different from that of Pd100 TDS experiments with an NOcovered surface rapidly heated in uhv characterizing NO adsorbed on Pd110 showed N0 N2 and N20 peaks with a maximum at about 490 K In addition to the 490K peak a second NO desorption peak was observed at 370 K The authors also measured the spatial distribution of the molecules desorbing during TDS and TPR This technique can provide additional information about the nature of the chemical processes involved Nam5 can accumulate on Pd100 during NO exposure at higher temperatures but not on Pd110 NMs accumulation on Pd110 was obtained by exposing the surface to NVN ions The resulting N2 TDS are distinctly different from the N2 TDS observed following NO adsorption In addition its spatial distribution is very different from the offnormal desorption found for N2 formed from N0 adsorption and also from N2 formed by the NO H2 reaction In the same way aPd110 surface precovered with 15N was prepared Exposure to 14NO resulted in 15N15N peaks at 460 and 520 K The spatial distribution is very different from that of 15N2 at 460 K Ikai 11 suggested that reaction III may be responsible fortthe offnormal desorption of N at 490 K On the basis of the results discussed previously it is concluded that the formation of N2 on Rh111 can be understood solely in terms of Nads combination However it is premature to conclude that process III does not play any role in the N2 formation on other metal surfaces The C0 NO reaction on various kinds of Pd catalysts was investigated by Rainer et al 112 Kinetics data for PdAIZO3 and planar model PdAlegTa110 catalysts were compared with those for C0 NO on Pd111 100 and 110 surfaces Figure 20 is a comparison of the Arrhen ius plots measured at partial pressures of about 1 mbar in each reactant for the model catalysts and at a CO pressure of 59 and a NO pressure of 68 mbar in a helium carrier for the PdA1203 powder catalyst The reported apparent activation energies are 67 kJmol for Pd111 and 100 71 kJmol for Pd110 and 155 kJmol for the PdA1203 powder catalyst The authors concluded from the comparison of rates per surface Pd atom TOF at 560 K that the supported PdA1203 powder catalysts exhibit a pronounced particle size effect with an increase in activity with increasing particle size The Pdlll surface is more active than the 100 and 110 surfaces The authors argued that the smaller Pd particles with their higher stepedgedefect densities have more in common with the Open singlecrystal faces whereas the larger particles have more Pdlll character PdIOO is more effective at dissociating NOads than Pd111 As a result PdlOO yields a higher ratio of NZNZO than Pd111 TDS of N2 306 BERNARD E NIEUWENHUYS A Sin le Cr stal g E 5 o 3 gt300 R Pd111 D S 8 6 1 25 2x Pcmoo E 2 Model 2 PdAIZO3To 110 U m j C 2 5 01 t 1200 X E loved o LL N 8 125 1 PdAl203 001 Powder 60 ii I l I F 15 16 17 18 19 1000T K FIG 20 Arrhenius plots for the rate of the NO CO reaction on various kinds of Pd catalysts The planar model and singlecrystal surface data were taken in a batch reactor 1 mbar of each reactant and the powder catalyst in a ow reactor Poo 59 mbar PNO 68 mbar adopted with permission from Rainer et al 112 following NO exposure and TPR of C0 NO show essentially two peak maxima for N2 production a at 450 K and 8 between 530 and 590 K for Pd111 and between 575 and 625 K for Pd100 The desorption tempera ture for BN is higher than the reaction temperature employed in these experiments The Pd100 surface more active for N0 dissociation than Pd111 exhibits the lower activity for the reaction This correlation implies that the removal of 8N is more important than NO dissociation in determin ing the reaction kinetics Obviously the dissociation of NO is not the rate limiting step on Pd The origin of the huge difference in activity between the model catalysts and the powder catalyst is not clear UNDERSTANDING AUTOMOTIVE EXHAUST CONVERSION 307 V Effects of Alloy Formation In most of the early papers the formation of alloy particles was consid ered to be undesirable For example Kummer 113 mentioned that each noble metal is used for a speci c purpose which can be served by keeping it separate from other metal components However according to Nieuwen huys et al 114 116 alloy formation cannot be avoided at the extreme conditions under which the catalyst must operate Many studies have shown that nearly all the noble metal particles are alloyed in real vehicleaged catalysts 117 118 Many possible detrimental effects of alloy formation have been proposed 113 including reduction of the effectiveness of Pt for alkane oxidation because of Rh surface segregation In addition Pt alloy formation may result in a loss of the availability of Rh or Pd to the exhaust gas since Pt readily sinters to large crystallites at temperatures higher than 870 K whereas Rh and Pd alone can remain dispersed under usual operation conditions 113 Alloy formation in general may also result in a distinctly different behavior relative to those of the pure component catalysts 119 Therefore it is of particular interest to review the available data concerning the effects of alloy formation on the reactions of importance in automotive catalysis The rst step in understanding the results of activity measurements with alloy catalysts is understanding knowledge concerning the surface composi tion Under vacuum Pd segregates to the surface of PtPd 120 and RhPd alloys 121 The surface composition of Pt Rh alloys has been investigated by several groups with techniques such as AES ionscattering spectroscopy work function measurements atomprobe FIM and STM 114116 119 122 131 Most of the results point to a strong Pt surface segregation of clean Pt Rh crystals under vacuum The Pt enrichment in the top layer is accompanied by Pt depletion in the second layer 123 124 126 Calcula tions of LeGrand and Treglia 126 and Schoeb et al 127 show that the experimentally found Pt surface enrichment can be understood on the basis of a difference in surface energy Pt surface enrichment should be expected whatever the bulk composition and equilibrium temperature According to van Delft er al 114 115 the surface composition of Pt Rh alloys may be extremely sensitive to the presence of adsorbate atoms due to the very small differences in factors resulting in Pt surface segregation of the clean surfaces Experiment show that the surface composition may easily change in the presence of other elements Carbon monoxide and hydrogen do not exert a measurable in uence on the surface composition 114 115 For oxygen however a strong oxygeninduced Rh surface segre gation was found that was caused by the much stronger Rh O bond strength relative to that of Pt O 114 115 124 Similar results were obtained for BERNARD E N IEUWENHUYS NO adsorption NO dissociates on the surface resulting in oxygeninduced Rh surface segregation The presence of sulfur also causes Rh surface segregation 123 In view of the relative thermal stabilities of the oxides Rh203 gt PdO gt PtOz it is not surprising that under oxidizing conditions and at tempera tures higher than 870 K metallic Pt particles exist in combination with Rh oxide 132 133 and that PdO crystallites separate from Pt crystallites in Pd Pt alloys in the temperature range 670 870 K 134 A more compli cated behavior was found for Pd Rh 135 Beck et al 124 reported that the PtoJo Rhom 111 single crystal surface which has a surface composition of about 30 Pt in vacuum remains Pt rich even under 50 mbar of hydrogen at temperatures typical of automotive catalytic converter operation 770 870 K A few studies of supported Pt Rh A1203 catalysts have been pursued using NMR spectroscopy Wang et al 136 used 13C NMR of adsorbed CO as well as 195Pt NMR and concluded that the surface is slightly enriched in Rh in the presence of CO Savar gaonkar et al 1 3 7 reported the use of 1H NMR spectroscopy to determine the surface composition of Pt RhA1203 catalysts in the presence of hydro gen The authors concluded that their catalysts are slightly enriched in rhodium relative to the adsorbatefree catalysts which are enriched in Pt Simulations indicated that the heat of adsorption of hydrogen must be 13 kJmol higher on Rh than on Pt to achieve the reported Rh surface segrega tion It should be noted that the interpretation of the observed Knight shifts in terms of surface composition is not straightforward Furthermore literature data characterizing hydrogen adsorption on pure wellde ned singlecrystal surfaces suggest that the initial heat of adsorption of hydrogen on Rh 80 kJmol is similar to that on Pt within 10 138 In my opinion there is no doubt of the correctness of the earlier conclusion Pt Rh surfaces are Pt enriched under reducing conditions and Rh enriched under oxidiz ing conditions The most direct determination of both the surface structure and the composition of Pt Rh alloy surfaces was obtained with STM 128 129 The 100 surfaces of Ptolzs Rhms and PtasoRhaso single crystals were imaged with atomic resolution and with discrimination of the Pt and Rh atoms The STM image shown in Fig 21 demonstrates that there is a limited tendency for RhRh and Pt Pt clustering on the surface and PtRh ordering is absent Interestingly Pt preferentially populates the step edges Oxygen causes a large reconstruction of the surface 130 I31 For example Fig 22 shows the STM image of the 100 surface of Ptms Rho exposed to 02 at 770 K and subsequently cooled to room temperature in oxygen The STM image points to the formation of linear Rh O chains Adsorption and reactivity investigations of Pt Rh alloy surfaces were UNDERSTANDING AUTOMOTIVE EXHAUST CONVERSION 309 aquot 0 n V r quot 1 A a 39 1 no A v quott f i t39 quot quotquot l I wquot vg quot s 4 I V I I 1 i 1 U quot I FIG 21 STM image of Pt05 Rh05100 image size 200 X 200 A Bright spots correspond to Rh atoms reproduced with permission from Wouda et al 129 carried out in a number of laboratories particularly to examine the effect of alloy formation on the reactions of importance in automotive catalysis oxidation of CO oxidation of hydrocarbons and reduction of NO 45 79 80 114 116 139 147 NO bond breaking is usually considered the rst step in the reduction of NO by CO and hydrogen and therefore many papers deal with studies of NO adsorption and dissociation see Section IIA4 NO adsorption is also a sensitive probe for examining the possible effect of alloy formation since there are large differences in the behavior of Pt and Rh toward NO interaction The extent of dissociation of NO on Pt is sensitive to the surface structure Rh has a much greater reactivity for NO bond breaking and the effect of the surface structure is smaller At temperatures of about 210 K complete dissociation of NO occurs on Rh100 and on Pt Rh100 alloy surfaces at low NO coverages and partial 310 BERNARD E NIEUWENHUYS FIG 22 STM image of Ptols Rho 5100 exposed to O at 770 K and subsequently cooled to room temperature in oxygen reproduced with permission from Wouda et al 131 dissociation Occurs during heating following saturation The extent of N2 desorption is a measure of the activity of a surface in NO bond breaking On the 100 surfaces of Pt Rh and Ptms Rhojs NO desorption occurs at about 450 K Figure 23 illustrates the effect of alloy formation on NO dissociation under TDS conditions It shows the relative amount of NO desorbing as NO as a fraction of the total coverage following exposure at a temperature of 210 K or lower On Rh100 and on Pt Rh100 alloy surfaces complete dissociation occurs at low NO coverages and partial dissociation occurs during heating following saturation 14 In contrast to the behavior on Rh100 and Pt Rh100 surfaces the fraction of NO decomposing on Pt100 does not change dramatically with coverage The behavior of the Pt Rh100 surface resembles that of the pure Rh100 surface at low NO coverage and that of pure Pt100 at high coverage This pattern illustrates that Rh atoms on the surface are very effective in NO dissociation and that NO dissociation occurs mainly on Rh sites These results also indicate that mixed Pt Rh sites are not very reactive in NO bond breaking 141 NO dissociation is sensitive to the surface structure of Rh and in particu UNDERSTANDING AUTOMOTIVE EXHAUST CONVERSION 311 10 013a 31 or B quotU 11 0 3 01 m CI 3 g 02 39 Pi Rh100 Rh100 Oa T T l l 0 02 01 06 08 10 NO Adsorbed aul FIG 23 Molecular NO desorption from Pt Rh Pt and Rh100 The abscissa is NO adsorbed as a fraction of saturation NO coverage reproduced with permission from Siera et al 141 Pt and Rh100 data from Root et al 106 and Gorte and Schmidt 148 lar to that of Pt For example it has been reported that the Pt111 surface cannot break the NO bond whereas the Pt410 surface is very reactive in NO bond breaking 27 XPS has been used to investigate the N O dissociation on PtRh singlecrystal surfaces 114 the results are summa rized in Fig 24 This gure shows the following i The dissociation activity is sensitive to the surface structure The 410 and 210 surfaces are more reactive in NO bond breaking than the 321 surface the activity of which is larger than that of the 111 surface The effect of the surface structure is greater for Ptrich than for Rhrich surfaces and ii the dissociation reactivity is higher for Rhrich than for Ptrich surfaces Fisher et al 143 144 investigated NO adsorption and reduction on the 111 surface of a Pt010 Rh090 single crystal with a surface composition of about 30 Pt in vacuum The presence of 10 atom Pt in the bulk signi cantly reduces the ability of the surface to dissociate NO The activation energy of NO dissociation was reported to be intermediate between those of the reaction on Rh111 and on Pt111 An interesting behavior was found for the Pt Rh100 surface after exposure to NO at 500 K 139 Initially a c2 X 2 surface structure is BERNARD E NIEUWENHUYS 50039 Tom 1100K 321 I 3 2 210 I E 400 a 410 E l 39 Tam 800K son 300 400 500 Temperature K FIG 24 The temperature at which a dissociation percentage of 25 is obtained for Pt rich Pt Rh alloy surfaces versus the same parameter for Rhrich Pt Rh alloy surfaces repro duced with permission from Wolf el al 114 formed which via a combination of c2 X 2 and 3 X 1 changes slowly into a 3 X 1 surface structure The time required for formation of the 3 X 1 structure depends strongly on the initial Rh surface concentration On Rhrich surfaces the 3 X 1 structure is formed much faster than on an originally Pt rich surface The 0 formed by NO dissociation slowly replaces the N adatoms also formed by NO dissociation The O atoms extract Rh atoms to the surface Since the second layer is Rh enriched the slow formation of the 3 X 1 structure is related to an exchange of Rh and Pt atoms between the second and the top layer 39 Similar observations were made by Tanaka et al 146 147 for Pt Rh surfaces prepared by electrochemical deposition of Pt on Rh100 or Rh on Pt100 All these studies demonstrate that the rst two layers of Pt Rh samples are very exible both composition and structure change easily with changing experimental conditions caused by exchange of atoms between the rst and second layer A pure Rh100 surface is not very reactive for NO reduction by hydrogen at temperatures lt650 K A Ptenriched Pt Rh100 surface shows reactivity at temperatures higher than 650 K However on a Rhenriched Pt Rh100 surface with 3 X 1 structure NO reacts with hydrogen even at 500 K The authors suggested that a Pt Rh100 hybrid UNDERSTANDING AUTOMOTIVE EXHAUST CONVERSION surface with 3 X 1 structure is the active catalyst which can be formed during the catalytic process 146 147 In contrast to the activity of Pt100 the activity of the Pt110 surface is very low for the NO H2 reaction I49 Deposition of a submonolayer of Rh on this surface results in a drastic enhancement of the activity which is also higher than that of pure Rh110 The resulting pl X 2 Pt Rh110 catalyst exhibits an almost equally high activity as p3 X 1 PtRh100 It was proposed that a speci c Pt RhO Pt arrangement is needed for high activity 149 The previous results show that Rh sites are those with the higher activity for NO dissociation What do we know about CO adsorption Does CO prefer Rh or Pt sites These questions were addressed by Rutten et al J 45 in an investigation of CO adsorption on a Ptozs Rho75 111 surface with RAIRS TDS and work function measurements It was found that CO shows a pronounced preference for Rh sites at both100 and 300 K up to a coverage of 0 027 Adsorption was found to be predominantly on atop sites and only weak bridge bands were detected Unfortunately information concerning coadsorption of CO and NO on Pt Rh single crystal surfaces is missing However IR spectra of NO CO mixtures are available for Pt Rh supported on silica Heezen et al 150 concluded that for CO NO mixtures the N0 molecules are primarily adsorbed on Rh and the CO molecules on Pt atoms The CO Oz and CO NO reactions have been investigated in the pres ence of several surfaces of Ptozs Rhojs and PtOJO Rho39go single crystals 45 119 142 144 The results were compared with data for pure Pt and Rh surfaces to examine the effect of alloy formation This comparison is of interest for understanding the synergistic effects reported by many groups for the CO oxidation catalyzed by supported Pt Rh 151 153 Siera et al 142 argued that bene cial effects of Pt Rh alloy formation might be related to the strong preference of oxygen for the Rb sites on the surface This effect might result in a better mixing of CO and O on the surface with the O bound to Rh and CO adsorbed on Pt and hence in a faster reaction at lower temperature Similar effects might be expected for C0 NO 142 because CO has a slight preference for Pt and NO for Rh However Siera et al 45 80 142 did not nd any indication of a synergistic effect due to alloy formation This result is consistent with those obtained in the same laboratory for supported Pt Rh alloy catalysts 80 114 150 Large differences were found by Siera et al 45 in the steady state reaction rates of both reactions catalyzed by four Ptogs Rhogs alloy surfaces in the low pressure range An example is shown in Fig 25 for the NO CO reaction and in Fig 9 for the C0 02 reaction BERNARD E NIEUWENHUYS gt N O r 111 j 9N iE H00 9 8 x 2 410 PM 1 1 300 500 700 900 Temperature K FIG 25 Steady state reaction rates of C02 production from C0 NO on Pt Rh111 100 410 and 210 under stoichiometric conditions reproduced with permission from Siera et al 45 This large effect of the surface structure is a remarkable observation for the C0 02 reaction since this reaction is usually considered to be a typical example of a reaction that is insensitive to the surface structure The in uence of the surface structure can be explained on the basis of the interaction of these surfaces with CO NO and 02 A large CO inhibitiOn occurs in the low temperature range lt400 K especially for the C0 0 reaction This effect is larger for the 210 surface than for the 111 surface due to the higher heat of adsorption of CO on the open 210 surface The reaction order in CO becomes positive in the higher temperature range 550 K A positive reaction order was found for oxygen in the whole temperature range just as for pure Pt111 However for pure Rh111 a negative reaction order in oxygen was found in the higher temperature range 154 This observation suggests that the Rh atoms in the Pt Rh alloy are more dif cult to oxidize in the presence of Pt atoms Similar observations were made for the C0 NO reaction 45 In the low temperature regime the order in CO is negative However the reaction starts at signi cantly lower temperature especially on the 111 surface which is the least reactive surface for the N O bond breaking Hence N O dissociation is not the ratedetermining step at low temperatures For pure Rh100 a reaction order of zero was reported by Oh et al 155 Oh et a UNDERSTANDING AUTOMOTIVE EXHAUST CONVERSION 315 concluded that the Rh100 surface is predominantly covered by NO Pt Rh alloy surfaces on the other hand are covered with both NO and CO at low temperatures The CO molecules reside predominantly on the Pt atoms and the N0 molecules on the Rh atoms in the alloy surface 150 Hence it is likely that the Pt atoms in the alloy surface are covered with CO at low temperatures causing the negatiVe dependence of reaction rate on C0 pressure Ng et at 144 examined the NO CO CO Oz and CO NZO reactions on a Ptmo Rhogo 111 surface The NO CO activity of this alloy surface is similar to that of Rh111 at temperatures from 573 to 648 K in that the two surfaces are represented by the same activation energy reaction orders and selectivity The turnover frequencies are slightly lower than those for Rh111 when compared on a per surface atOm basis however the rates per surface Rh atom are virtually unchanged The authors suggested that the primary effect of Pt is to dilute the Rh surface concentration Hirano et al 78 80 investigated the reaction of NO with hydrogen on the 111 100 and 410 surfaces of a Ptozs Rhms single crystal and for comparison the pure Pt100 and Rh100 surfaces Both the activity and the selectivity depend strongly on the surface structure and composition as shown in Fig 26 In the lowtemperature range 520 K the activity of the 100 surfaces decreases in the order Pt100 gt Pt Rh100 gt Rh100 The activity of pure Rh100 is drastically enhanced by alloying with Pt For the alloy surfaces the order in activity at 520 K is 100 gt 410 gt 111 The order of intrinsic reactivity for NO bond breaking is Pt Rh410 gt Rh100 gt Pt Rh100 gt Pt100 gt Pt Rh111 Hence these results indicate that NO bond scission is not the ratedetermining step of the reaction The selectivity toward N2 decreases in the order Rh100 Pt Rh410 gt Pt100 gt PtRh111 at 520 K It is well established that many vacant metal sites are required before NO dissociation can occur If the surface is largely covered with molecularly adsorbed NO or with N adatoms the reaction is inhibited by blocking of active sites The conversion of NO at temperatures below 600 K is low for the Rh100 surface due to the presence of strongly bound N atoms The high selectivity of this surface is also caused by the high concentration of N adatoms On Pt100 the Pt N bond strength is much weaker resulting in a much lower concentration of N adatoms at 520 K Free Pt sites remain available and consequently the reaction is fast but with a much lower selectivity toward N2 The Pt Rh100 alloy surface shows a behavior intermediate between those of pure Pt and Rh100 The CO oxidation and NO reduction reactions have also been investi gated in the presence of supported Pt Rh catalysts The results seem confus ing Whereas Oh and Carpenter I51 and Lyman et al 153 156 reported 316 BERNARD E NIEUWENHUYS 80 NO Conversion Vol 07 O PtRh111 PtRh100 P1000 PtRh410 Rh100 100 Selectivity PtRh111 Pt100 PtFih100 Pt Rh410 Rh100 FIG 26 Relative activity NO conversion after 3 min at 520 K and selectivity for N2 at a conversion of 10 for the reduction of NO by hydrogen on Pt Rh and Pt Rh alloy single crystal surfaces reproduced with permission from Hirano et a1 79 a better performance of some bimetallic PtRhA1203 catalysts than those of individual Pt or Rh results of other groups did not support the evidence for a synergistic effect 80 114 157 158 For illustration the results for the NO reduction reactions reported by Nieuwenhuys et a1 80 114 157 are summarized in Fig 27 which shows the temperature required to achieve a constant turnover frequency versus the bulk catalyst composition Both UNDERSTANDING AUTOMOTIVE EXHAUST CONVERSION 317 TK r r rr I I t I o 20 40 so 80 100 Rhtbuikiea TiK b D O 400i 3504 39 1 300 regs 0 20 A 60 80 100 Rhbulk FIG 27 3 Temperature required for a constant turnover frequency of 005 s391 for the NO CO reaction on PtRh on SiOz catalysts as a function of the bulk composition for three NOCO ratios I NOCO 14 0 NOCO 1 O NOCO 4 b Temperature required for a constant turnover frequency of 0005 s 1 for the N0 H2 reaction on Pt Rh on SiOz catalysts as a function of the bulk composition for two NOHz ratios 0 NOHz 1 I NOHz 13 reproduced with permission from Nieuwenhuys e aL 80 the activity and the selectivity of the Pt Rh alloy catalysts are between those of the constituent metal catalysts RhSi02 is more active in the NO CO reaction than PtSiOz For this reaction the effect of changing feed composition is small for Rh and large for Pt In uhv studies it was 318 BERNARD E NIEUWENHUYS also found that the reaction rate is a very weak function of the CONO ratio For PtSiOz the results point to a signi cant CO inhibition as has been observed under lowpressure conditions on Pt singlecrystal surfaces Only in a large excess of NO can a CO precovered Rh100 surface be saturated with NO 159 On Rh the reaction occurs at temperatures much higher than the NO dissociation temperature Apparently the reaction rate is limited by the presence of N on the Rh catalyst The NO H2 reaction starts at a much lower temperature on PtSiOz than on RhSiOz an observa tion that is consistent with the singlecrystal work discussed previously The catalytic performance of Pt Rh alloys has been correlated with the surface composition expected on the basis of the single crystal work 80 114 The surface composition of Pt Rh alloys varies strongly with condi tions such as the gas composition Clean Pt Rh alloy surfaces are enriched with Pt Adsorbates can easily induce segregation of Pt or Rh to the surface Oxygen in the gas phase induces Rh surface segregation For the NO CO and NO H2 reactions the activity of the Ptoj Rho s alloy catalysts resembles that of pure Rh under net oxidizing conditions whereas under net reducing conditions its activity is between those of Pt and Rh This comparison suggests that the surface composition varies with the experi mental conditions from almost pure Rh under net oxidizing conditions to perhaps a bulklike or Pt rich surface composition under net reducing conditions However the selectivity of this catalyst for the NO H2 reac tions is different from those of both Pt and Rh This comparison indicates that this catalyst contains Pt atoms in the surface that in uence the Selectiv ity The activity of the Ptol75 Rho alloy catalyst is almost equal to that of pure Pt under net reducing conditions whereas its activity is between those of Pt and Rh under stoichiometric and net oxidizing conditions Again this comparison suggests that the surface composition changes with the feed composition The properties of the Pam Rho alloy catalysts are notable For both the NO CO and NO H2 reactions the activity is equal to that of pure Rh under net oxidizing stoichiometric and net reducing conditions Under net oxidizing conditions a behavior such as that of pure Rh may indeed be expected However under net reducing conditions the presence of both Pt and Rh atoms should be expected on the surface The relatively high activity of this catalyst for the NO CO reaction might be caused by a bene cial effect of the presence of both Rh and Pt ie the low CO inhibition on Rh sites and a bene cial effect of Pt on for example the amount of N or NO on the surface For the NO H2 reaction however the presence of Pt atoms does not seem to diminish the inhibition effect of NO probably because of the lower reaction temperatures The synergistic effect reported by Lyman et al 153 156 was found only for a Ptrich catalyst with a very homogeneous composition of the particles UNDERSTANDING AUTOMOTIVE EXHAUST CONVERSION 319 Bimetallic catalysts with a bimodal particle composition distribution Rh rich and Ptrich particles did not show synergistic performance Cai et al 153 suggested that the absence of synergism in the observations of Nieuwenhuys et al 80 114 157 may be attributed to the presence of bimodal particle composition distribution or to the low operating tempera ture regime used by Nieuwenhuys et al due to the high activity of their catalysts It is emphasized that the synergism observed by Cai et al 153 is very small smaller than 9 K expressed in terms of the temperature needed to achieve a conversion of 50 In addition the effect was not observed for the Pt Rh catalysts with other compositions than 96 Pt including a very Ptrich Ptogg Rhogl catalyst The synergistic effect found for the Pt096 Rh0m catalyst was explained by a bene cial effect of Pt on Rh reduc tion and by a higher probability of having both reactant species on neigh boring sites In conclusion the catalytic properties of supported Pt Rh alloys are strongly dependent on the gasphase composition the bulk composition of the alloy particles and the reaction temperature A combination of factors determines the activities of the different catalysts Under oxidizing condi tions at suf ciently high temperatures Rh segregation to the surface occurs This is re ected in the catalytic activity of the Pt Rh alloys which can vary between that of pure Pt and that of pure Rh Large synergistic effects caused by alloying of Pt with Rh have not been observed Differences in activity and selectivity for catalysts with different Pt Rh compositions can be explained in terms of speci c properties of the pure metals toward the adsorbates Although most results show that the behavior of the alloy catalysts varies between those of Pt and Rh care must be taken when interpretingthe resultsinrtermsmfabsoiutfsui eicoic tm i ti and Rh The activity of an alloy catalyst depends on the surface composition of the metal particles However the relationship between activity selectiv ity and surface composition may be complicated for many reasons The catalytic behavior varies with the concentration and the size of the various ensembles of atoms on the surface All reaction steps molecular adsorption dissociation and reaction between the species on the surface may depend in different ways on the distribution of atoms over the surfaces Further more the small effect of alloying on the binding energies of adsorbed atoms and molecules may play a role in the catalytic performance of the alloys VI Effects of the Additives Cerium and Lanthanum Oxides It was noted in the introduction that the washcoat of the automobile catalysts contains several other oxides mainly cerium and lanthanum oxide 320 BERNARD E NIEUWENHUYS These rare earth oxides make an important contribution to catalyst perfor mance and durability and have multiple functions 160 The ceria content of current threeway catalyst formulations is about 25 wt as cerium of the washcoat Ceria was originally added to increase the stability and dispersion of the noble metal 161 Ceria and in particular lathana are thought to stabilize the 39yA1203 support by inhibiting its structure change from the y to the 0 modi cation 162 164 An important difference be tween the two rare earth oxides is the facile change of the oxidation state of Ce whereas La is valence invariant La203 under operation conditions Cerium oxide can be partly reduced from CeOZ to give oxygende cient Ce02x x is near onehalf and reoxidized to CeOz under oxidizing condi tions Ceria is a chemically active component as an oxygen buffer compo nent 165 as a promoter for the noble metal catalyst and as a catalyst for the water gas shift reaction 166 C0 H20 1 CO2 H2 35 This reaction leads to an additional increase in CO conversion and to the formation of hydrogen which has a bene cial effect on NO reduction The oxygen storage capacity of ceria results in a widening of the effective airfuel ratio at which the reduction and oxidation reactions can operate during the oscillatory cycle Under fuelrich reducing conditions the stored oxygen is released and is available for the oxidation of CO and he oxygen is stored during fuellean oxidizing conditions thereby enhancing NO reduction to N2 Recently direct promoter effects of ceria on the catalytic properties of noble metals have been reported 167 168 Addition of ceria to RhA1203 was found to improve NO reduction at low temperatures with a decrease in the apparent activation energy for the C0 NO reaction 168 A large effect of the particle size of ceria was found for NO reduction catalyzed by a PtceriaA1203 169 The authors proposed that the Pt ceria interaction increases by reduction of the ceria particle size Although these promoter effects of ceria on the noble metals are well documented the origin of the effects is not fully understood The few surface science studies directed at understanding these promoter effects provide some novel information Zafiris and Gorte 170 examined the structure and adsorption properties of Pt and Rh deposited on amorphous ceria hns They reported that ceria supported Pt exhibits adsorption properties very similar to those of A1203 supported Pt for CO NO and H2 In contrast large effects were found for Rh on ceria adsorption of CO and of NO is affected by ceria and CO adsorption results in a C02 formation peak during TDS It was suggested that O can migrate from ceria to the Rh surface at temperatures near room temperature UNDERSTANDING AUTOMOTIVE EXHAUST CONVERSION 321 Schmidt and Krause 171 and Schmidt and Schwartz 172 observed dramatic changes in the microstructures of Rh supported on thin lms of SiOz and A1203 during heating in an atmosphere of NO CO or a mixture of NO CO Rh on silica is volatilized even by treatment in NO CO at 570 K but the addition of ceria retards this effect Heating Rh on alumina in NO CO had much less effect However heating in NO alone resulted in redispersion of Rh into small 20A particles on the A1203 It was speculated that a mobile nitrosyl complex is formed during heating in NO resulting in redispersion and that volatilization occurs via formation of metal carbonyls Hardacre et al 173 J 74 investigated the properties structure and composition of cerium oxide lms prepared by cerium deposition on Pt111 nding that the activity for CO oxidation is enhanced on Pt111 that is partially covered by ceria It was suggested that new sites at the Pt oxide interface become available for reaction A remarkable observa tion is the high activity for CO oxidation when the Pt111 sample is fully encapsulated by ceria Pt was undetectable by XPS and AES It was proposed that an ultrathin disordered ceria lm becomes the active catalyst It was also demonstrated by XPS and AES that Pt dramatically increases the reducibility of cerium oxide that is in intimate contact with Pt This result suggests that intimate contact between the noble metal and oxide phases is indeed crucial to facile oxygen release from ceria Highresolution electron microscopy demonstrated the presence of direct contact between ceria and noble metal for supported Pt Rh catalysts 175 Hardacre et al 173 1 74 related the catalytic activity of the ceria phase to partially reduced cerium oxide In conclusion there is overwhelming evidence for the bene cial effect of cerium oxide on the activity of the noble metal catalyst However the nature of the promoter effect of ceria is not fully understood Most likely the noble metal cerium oxide interface is of crucial importance for some of the effects observed More studies with model systems are needed for a better understanding of the promoter effects of ceria VII Summary Assessment and Forecast The automotive TWC is one of the major achievements of modern re search in heterogeneous catalysis There have been major efforts to eluci date the fundamental reaction pathways and the catalyst characteristics that account for the success of the TWC The chemistry involved is understood in considerable detail as a result of work with idealized models of the TWC The mechanisms of the CO oxidation and NO reduction reactions with 322 BERNARD E NIEUWENHUYS CO hydrogen and he have been discussed in detail in this review In particular the application of various modern surface science techniques has advanced our understanding of the principles of the reactions on the atomic scale and the relationship of surface composition and structure to catalytic properties Surface science offers many opportunities in catalysis research because a variety of techniques are available to characterize in detail the composition and structure of the catalyst surface and to identify the adsorbed species A frequent criticism of the surface science approach is that it is far removed from real catalysis since most of the surface science techniques can only be applied at low pressures and with model catalysts often single crystal surfaces The so ealled pressure gap has been bridged by combining in the same apparatus the techniques needed for surface analysis and character ization with the ability to measure reaction rates at elevated pressures In addition many techniques can also be applied in situ at elevated pressures In this review literature data concerning CO oxidation and NO reduction on model catalysts have been reviewed and compared with those reported for supported catalysts The major differences in behavior of the three noble metals Pt Pd and Rhused in TWC have been assessed It is concluded that the major mechanisms are reactions of the L H type be tween Oads Cows and the dissociation products of NO viz N ads and Dads with N2 formed by combination of 2 Nads NH3 by hydrogenation of Nads and N20 by reaction between Nad and NOads Although other mechanisms have been proposed and their possible existence cannot be ruled out the effects of the surface composition and structure the speci c differences in behavior of Pt Pd and Rh the effect of changes in temperature and variations in partial pressures can be fully understood on the basis of these reaction pathways The effects of alloy formation and the chemistry of the additive cerium oxide although less well understood have also been evaluated in detail It was shown that the catalytic properties of Pt Rh alloy catalysts are strongly dependent on the gasphase composition and the bulk composition of the alloy particles Many factors determine the activities of these catalysts Under oxidizing conditions Rh segregation to the surface occurs This is re ected in the catalytic activity of the PtRh alloys which can vary between that of pure Pt and that of pure Rh No large synergistic effects resulting from the alloying of Pt with Rh have been observed The differences in activity observed for the different catalysts39can be explained in terms of the speci c properties of the pure metals toward the adsorbates The activity and the selectivity of the Pt Rh and Pt Rh surfaces depend on the surface structure and composition The selectivity is determined by the relative concentrations of CO NO N and H adsorbed on the various surfaces UNDERSTANDING AUTOMOTIVE EXHAUST CONVERSION 323 In fresh automotive exhaust catalysts the precious metals are well distrib uted in the form of small crystallites with sizes between 05 and 5 nm During aging a severe loss of Pt surface area occurs due to sintering Rh however remains in a highly dispersed state because it is partly in an oxidized state whereas the Pt particles remain metallic The morphology andcomposition of the catalyst particles change continuously during opera tion Pt Rh alloy particles are formed under reducing conditions Under oxidizing conditions dealloying occurs resulting in rhodium oxide particles being separated from metallic Pt and Pt Rh alloy particles During reaction cerium switches between oxidation states leading to changes in the noble metalA1203 especially Rh interactions Cenoble metal and Cesupport interactions can occur Hence a continuous restructuring and modi cation of catalytic behavior can occur All these effects make the automotive TWC one of the most dynamic catalysts in use Interest in the control of emissions from automobiles will continue to grow because of increasingly stringent legislation worldwide for emission of he CO NOx and particulates Major new developments include i the introduction of a small catalytic converter combined with the main con verter containing the TWC close to the engine manifold enabling quicker lightoff and therefore overall better CO and hc conversion and ii the development of catalytic converters for leanburn gasolinefueled engines and for diesel engines Leanburn versions of the fourstroke engine have been in development for many years From the perspective of fuel economy and hence COZ39 emission the use of leanburn engines is desirable The dif culty is in achieving NOx conversion under such oxidizing conditions The most thoroughly investigated catalysts are zeolites containing metal ions in particular CuZSMS Col ZSMS and other metalzeolite systems Unfortunately these zeolites are deactivated at high temperatures in the presence of water because of destruction of the zeolite structure which is an inevitable component of automotive emissions Although some reports suggest that many metalcontaining zeolites such as FeZSMS 176 177 and ceriumexchanged mordenite 178 are stable in water vapor it is not certain that a zeolitebased catalyst that meets the legislation standards for emission and catalyst lifetime will be developed New technologies that are developing include NOx catalysts that can reduce NOx under oxidizing conditions adsorbers absorbers for NOx and he particulate traps that are regenerated by catalytic oxidation using Pt and novel catalysts that can convert CO immediately after the cold start REFERENCES I Taylor K C Catal Sci Technol 5 119 1984 2 Taylor K C Catal Rev Sci Eng 35 457 1993 324 NQMiS 0 10 l 12 I I4 15 16 94 I7 18 19 20 21 2 2 540K 2 25 26 27 28 R 29 30 31 3 3 34 3 36 37 sum PI BERNARD E NIEUWENHUYS Shelef M and Graham G W Catal Rev Sci Eng 36 433 1994 Bowker M and Joyner R W Insights Spec Inorg Chem 145 1995 Ramsier R D and Yates J T Surf Sci Rep 12 243 1991 Winkler A and Rendulic K D Surf Sci 118 19 1982 Nieuwenhuys B B Surf Sci 126 307 1983 NA TOASI Ser C 398 155 1993 and references therein Gland J L Sexton B A and Fisher G B Surf Sci 95 587 1980 Campuzano J P Chem PhysSolid State Surf Sci Catal 3A 389 1991 Blyholder G J Phys Chem 68 2772 1964 Bagus P S Nelin C J and Bauschlicher C W J Vac Sci Technol A2 905 1984 Muller W and Bagus P S J Vac Sci Technol A3 1623 1985 Rangelov G Memmel N Bertel B and Dose V Surf Sci 251 965 1991 Nieuwenhuys B E Surf Sci 105 505 1981 Eischens R P Pliskin W A and Francis S A J Chem Phys 22 1786 1954 Sheppard N and Nguyen T T in Advances in IR and Raman Spectroscopy R E Hester and R J H Clark Eds Vol 5 Heyden London 1978 Broden G Rhodin T N Brucker C Bendow R and Hurych Z Surf Sci 59 593 1976 Benzinger J B Appl Surf Sci 6 105 1980 Gorodetski V V and Nieuwenhuys B E Surf Sci 105 299 1981 Hendrickx H A C M and Nieuwenhuys B B Surf Sci 175 185 1986 Wolf R M Bakker J W and Nieuwenhuys B B Surf Sci 246 135 1991 van Hardeveld R M Thesis Technology University of Eindhoven 1997 Borg H J Reijerse J F C J M van Santen R A and Niemantsverdriet J W J Chem Phys 101 10051 1994 Sandell A Nilsson A and Martensson N Surf Sci 241 L1 1991 Somorjai G A Surf Sci 107 573 1981 Gorte R J Schmidt L D and Gland J L Surf Sci 109 367 1981 Masel R 1 Catal RevSci Eng 28 335 1986 Janssen N M H Cholach A R Ikai M Tanaka K and Nieuwenhuys B E Surf Sci 382 201 1997 Solymosi F J Mol Catal 65 337 1991 van T01 M F H Gielbert A Wolf R M Lie A B K and Nieuwenhuys B E Surf Sci 287288 201 1993 Lof er D G and Schmidt L D Surf Sci 59 195 1976 Conrad 11 Ertl G and Kuppers J Surf Sci 76 323 1978 Lambert R M and Comrie C M Surf Sci 46 61 1974 Thiel P A Weinberg W H and Yates J T J Chem Phys 71 1643 1979 King D A Faraday Disc Chem Soc 87 303 1989 van Slooten R F and Nieuwenliuys B E J Catal 122 429 1990 Gardner S D Ho and G B Upchurch B T Schryer D R Kielin E J and Schryer J J Catal 129 114 1991 Ertl G Adv Catal 37 213 1990 Engel T and Ertl G J Chem Phys 69 1267 1978 Campbell C 1 Ertl G Kuipers H and Segner J J Chem Phys 73 5862 1980 Gland J L Sexton B A and Fisher G B Surf Sci 95 587 1980 Su C X Cremer P 8 Shen Y R and Somorjai G A J Am Chem Soc 119 3994 1997 Li Y Boecker D and Gonzalez R D J Catal 110 319 1988 4 k 45 4 4 48 49 5 Q 5 5 5 5 55 5 57 kWNN O 5 00 5 6 61 ex Q SUN 6 6 6 66 5quot 67 68 6 7 7 72 N9 7 7 7 7 ownuw 77 7 00 79 8 81 9 UNDERSTANDING AUTOMOTIVE EXHAUST CONVERSION 325 Wintterlin J V lkening S Janssens T V W Zambelli T and Ertl 0 Science 278 1931 1997 Siera J Rutten F and Nieuwenhuys B E Catal Today 10 353 1991 Cobden P D Siera J and Nieuwenhuys B E J Vac Sci Technol A10 2487 1992 Sales B C Turner J B and Maple M B Surf Sci 114 381 1982 Janssen N M H Cobden P D and Nieuwenhuys B E J Phys Condens Matters 9 1889 1997 lmbihl R and Ertl 6 Chem Rev 95 697 1995 Mergler Y J van Aalst A van Delft J and Nieuwenhuys B E Appl Catal BIO 245 1996 Mergler Y J and van Aalst A ACS Symp Ser 587 196 1995 Mergler Y J and Nieuwenhuys B E Appl Catal 1312 95 1997 Mergler Y J van Aalst A van Delft J and Nieuwenhuys B EJ Cazal 161 310 1996 Mergler Y J Hoebink J and Nieuwenhuys B E J Catal 167 305 1997 Matsushima T Surf Sci 127 403 1983 Wintterlin J Schuster R and Ertl G Phys Rev Lett 77 123 1996 Brune H Wintterlin J Trost J Ertl G Weichers J and Behm R J J Chem Phys 99 2128 1993 Fadeev E Gorodetslcii V and Smirnov M in Proceedings of the Boreskov Memorial Symposiumquot 1997 N0vosibirsk Russia Watanabe K Uetsuka H Ohnuma H and Kunimori K Catal Lett 47 17 1997 Mullins C B Rettner C T and Auerbach D J J Chem Phys 95 8649 1991 Wartnaby C E Stuck A Yeo Y Y and King D A J Chem Phys 102 1855 1995 Boudart M J M01 Catal A 120 271 1997 and references therein Szanyi J and Goodman D W J Phys Chem 98 2972 1994 Szanyi J Kuhn W K and Goodman D W J Phys Chem 98 2978 1994 Peden C H F ACS Symp Ser 482 143 1992 Peden CH F Goodman D W Blair D S Berlowitz P J Fisher G B and Oh S H J Phys Chem 92 1563 1988 Oh S H Fisher G B Carpenter J B and Goodman D W J Catal 100 360 1986 Su X Jensen J Yang M X Sameron M B Shen Y R and Somorjai G A Faraday Disc 105 263 1996 Bowker M Guo Q and Joyner R W CataL Lett 18 119 1993 Kruse N and Gaussmann A Surf Sci 266 51 1992 Schwartz S B Fisher G B and Schmidt L D J Phys Chem 92 389 1988 van den Brink P J and McDonald C R SAE Tech Paper No 952399 1995 Appl Catal B 6 L97 1995 Kobylinski T P and Taylor B W J Catal 33 376 1974 Solymosi F Vblgyesi L and S rk ny J J Catal 53 336 1978 Bamwenda G R Obuchi A Ogata A and Mizuno K Chem Lea 2109 1994 van Hardeveld R M Schmidt A J G W van Santen R A and Niemantsverdriet J W J Vac Sci Technol A15 1558 1997 Chin A and Bell A T J Phys Chem 87 3700 1983 Hirano H Yamada T Tanaka K Siera J Cobden P D and Nieuwenhuys B E Surf Sci 262 97 1992 Hirano H Yamada T Tanaka K Siera J and Nieuwenhuys B E Stud Surf Sci Catal 75 345 1993 NieuwenhuysB B Siera J Tanaka K and Hirano HACS Symp Ser 552 114 1994 Yamada T and Tanaka K J Am Chem Soc 113 1173 1991 8 N 8 84 8 U k 8 Q 8 l 88 8 9 9 0 9 N 9 9 amp Lugu I 08 109 110 111 112 113 114 11 P 116 117 BERNARD E NIEUWENHUYS Zemlyanov D Y Smimov M Y Gorodetskii V V Block J H Surf Sci 329 61 1995 Zemlyanov D Y Smimov M Y and Gorodetskii V V Surf Sci in press Sun Y M Sloan S Ihm H and White J M J Vac Sci Technol A14 1516 1996 van Hardcveld R M van Santen R A and Niemantsverdriet J W J Phys Chem 101 998 1997 Dietrich H Jacobi K and Ertl G Surf Sci 377 379 308 1997 J Chem Phys 106 9313 1997 Williams C T Tolia A A Weaver M J and Takoudis C 0 Chem Eng Sci 51 1673 1996 Tolia A A Williams C T Takoudis C G and Weaver M J J Phys Chem 99 4599 1995 Hecker W C and Bell A T J Catal 92 247 1985 Kiskinova M Lizzit S Comelli G Paolucci G and Rosei R Appl Surf Sci 64 185 1993 Cobden P D Nieuwenhuys B E Esch F Baraldi A Comelli G Lizzit S and Kiskinova M J Vac Technol A 16 1014 1999 Klein R L Schwartz S and Schmidt L D J Phys Chem 89 4908 1985 Lesley M W and Schmidt L D Surf Sci 155 215 1985 Li Y Slager T L and Armor J N J Catal 150 388 1994 Witzel F Sill G A and Hall W K J Catal 149 229 1994 Adelman B 1 Beutel T Lei G D and Sachtler W M H Appl Catal 811 L1 1996 Salama T M Ohnishi R and Ichikawa M Chem Commun 105 1997 Otto K Shelef M and Kummcr J T J Phys Chem 74 2690 1970 Xu H and Ng K Y S Appl Phys Lett 68 496 1996 Surf Sci 365 779 1996 Leibsle F M Murray P W Francis S M Thornton G T and Bowker M Nature 363 706 1993 Hecker W C and Bell A T J Catal 84 200 1983 85 389 1984 Belton D N and Schmieg S J J Catal 144 9 1993 138 70 1992 Belton D N DiMaggio C L Schmieg S J and Ng K Y S J Catal 157 559 1995 Makeev A G and Slinko M M Surf Sci 359 L467 1996 Bugyi L and Solymosi F Surf Sci 188 475 1987 Root T W Schmidt L D and Fisher G B Surf Sci 134 30 1983 Permana H Ng K Y S Peden C H F Schmieg S J Lambert D K and Belton D N J Catal 164 194 1996 Zhdanov V P and Kasemo B Surf Sci Rep 29 31 1997 Ikai M He H BorroniBird C E Hirano H and Tanaka K Surf Sci 315 L973 1994 Ikai M and Tanaka K Surf Sci 357358 781 1996 Ikai M Thesis University of Tokyo 1996 Rainer D R Vesccky S M Koranne M Oh W S and Goodman D W J Catal 167 234 1997 Kummer J T J Phys Chem 90 4747 1986 Wolf R M Sicra J van Delft F C M J M and Nieuwenhuys B E Faraday Disc Chem Soc 87 275 1989 van Delft F C M J M Nieuwenhuys B E Siera J and Wolf R M Iron Steel Inst Jpn Int 29 550 1989 Nieuwenhuys B E Surf Rev Lett 3 1869 1996 Kim S and d Aniello M J Jr AppL CamL 56 23 45 1989 11 9 11 o 120 121 122 123 124 125 126 127 128 129 130 131 I32 133 134 135 I36 137 138 13 0 140 141 142 143 144 145 I46 1 4 7 J 48 149 150 151 152 153 UNDERSTANDING AUTOMOTIVE EXHAUST CONVERSION 327 Maire R Capella M Meunier G Beziau J F Bazin D Dexpert H Garin F Schmitt J L and Maire G Stud Surf Sci Catal 96 749 1995 Nieuwenhuys B E in The Chemical Physics of Solid Surfaces D A King and D P Woodmff Eds Vol 6 p 185 Elsevier Amsterdam 1993 Kuijers F J Tieman B and Ponec V Surf Sci 75 657 1978 Batirev I G and Leiro J A J Electr Spectr Rel Phen 7 79 1995 Siera J van Delft F C M J M van Langeveld A D and Nieuwenhuys B B Surf Sci 264 435 1992 Tsong T T Surf Sci Rep 8 127 1988 Beck D D DiMaggio C L and Fisher G B Surf Sci 297 293 303 1993 Altenstaedt W and Leisch M Appl Surf Sci 9495 403 1996 39 LeGrand B and Treglia G Surf Sci 236 398 1990 Schoeb A M Raeker T J Yang L Wu X King T S and De Pristo A B Surf Sci Lett 278 L125 1992 Matsumoto Y Okawa Y Fujita T and Tanaka K Surf Sci 355 109 1996 Wouda P T Nieuwenhuys B E Schmid M and Varga P Surf Sci 359 310 1996 Matsumoto Y Aibara Y Mukai K Moriwaki K Okawa Y Nieuwenhuys B B and Tanaka K Surf Sci 377 379 32 1997 Wouda P T Schmid M Hebenstreit W and Varga P Surf Sci 388 63 1997 Zhu Y and Schmidt L Surf Sci 129 107 1983 Wang T and Schmidt L J Catal 70 187 1981 Chen M and Schmidt LJ39 Catal 56 198 1979 Graham G W Potter T Baird R J Ghandi H S and Shelef M J Vac Sci Technol A4 1613 1986 Wang Z Ansermet J P Slichter C P and Sinfelt J H J Chem Soc Faraday Trans I 84 3785 1988 Savargaonkar N Khanra B C Pruski M and King T S J Catal 162 277 1996 Gorodetski V V Nieuwenhuys B E Sachtler W M H and Boreskov G K Surf Sci 108 225 1981 Yamada T Hirano H Tanaka K Siera J and Nieuwenhuys B E Surf Sci 226 1 1990 Tanaka K I Yamada T and Nieuwenhuys B B Surf Sci 242 503 1991 Siera J Koster van Groos M J and Nieuwenhuys B B Reel Trav Chim Pays Bus 109 127 1990 Siera J van Silfhout R Rutten F J M and Nieuwenhuys B E Stud Surf Sci Catal 71 395 1991 Fisher G B Dimaggio C L and Beck D D Stud Surf Sci Catal 75 383 1993 Ng K Y S Belton D N Schmieg S J and Fisher G B J Catal 146 394 1994 Rutten F J M Nieuwenhuys B B McCoustra M R S HolJins P and Chesters M A J Vac Sci Technol A15 11619 1997 Taniguchi M Kuzembaev E K and Tanaka K Surf Sci 290 L711 1993 Tamura H Sasahara A and Tanaka K Surf Sci 303 L379 1994 Gorte R J and Schmidt L D 5qu Sci 111 260 1981 Sasahara A Tamura H and Tanaka K CataL Lett 28 161 1994 J Phys Chem 100 15229 1996 J Phys Chem B 101 1186 1997 Heezen 1 Kilian V N van Slooten R E Wolf R M and Nieuwenhuys B B Stud Surf Sci Catal 71 381 1991 Oh S H and Carpenter J B J Catal 98 178 1986 Tzou M S Asakura K Yamazaki Y and Kuroda H CataL Lett 11 33 1991 Cai Y Stenger H G Jr and Lyman C E J Catal 161 123 1996


Buy Material

Are you sure you want to buy this material for

25 Karma

Buy Material

BOOM! Enjoy Your Free Notes!

We've added these Notes to your profile, click here to view them now.


You're already Subscribed!

Looks like you've already subscribed to StudySoup, you won't need to purchase another subscription to get this material. To access this material simply click 'View Full Document'

Why people love StudySoup

Jim McGreen Ohio University

"Knowing I can count on the Elite Notetaker in my class allows me to focus on what the professor is saying instead of just scribbling notes the whole time and falling behind."

Allison Fischer University of Alabama

"I signed up to be an Elite Notetaker with 2 of my sorority sisters this semester. We just posted our notes weekly and were each making over $600 per month. I LOVE StudySoup!"

Bentley McCaw University of Florida

"I was shooting for a perfect 4.0 GPA this semester. Having StudySoup as a study aid was critical to helping me achieve my goal...and I nailed it!"


"Their 'Elite Notetakers' are making over $1,200/month in sales by creating high quality content that helps their classmates in a time of need."

Become an Elite Notetaker and start selling your notes online!

Refund Policy


All subscriptions to StudySoup are paid in full at the time of subscribing. To change your credit card information or to cancel your subscription, go to "Edit Settings". All credit card information will be available there. If you should decide to cancel your subscription, it will continue to be valid until the next payment period, as all payments for the current period were made in advance. For special circumstances, please email


StudySoup has more than 1 million course-specific study resources to help students study smarter. If you’re having trouble finding what you’re looking for, our customer support team can help you find what you need! Feel free to contact them here:

Recurring Subscriptions: If you have canceled your recurring subscription on the day of renewal and have not downloaded any documents, you may request a refund by submitting an email to

Satisfaction Guarantee: If you’re not satisfied with your subscription, you can contact us for further help. Contact must be made within 3 business days of your subscription purchase and your refund request will be subject for review.

Please Note: Refunds can never be provided more than 30 days after the initial purchase date regardless of your activity on the site.