Convert Katalis

8/2/2019 Convert Katalis

1/30

ll. ExperimentalThe experiments were carried out utilizing severalcombined ultrahigh vacuum/microreactor systems,described elsewhere,1 with capabilities for X-rayand UV photoelectron spectroscopy (XPS, UPS), ionscattering spectroscopies (ISS), high-resolution elec-tron energy-loss spectroscopy (HREELS), infraredreflection adsorption spectroscopy (IRAS), Augerelectron spectroscopy (AES), low-energy electron dif-fraction (LEED), and temperature-programmed de-sorption (TPD), and facilities for sample heating andcooling.The details for the cleaning and handling of thesingle-crystal catalysts are given in the appropriaterelated references. For the simulated supportedmetal catalysts, Mo(110) and Ta(100) substrates12*3were chosen for preparing the thin SiO24'15 and A1203films. Sample temperatures were monitored witha pair of W~5%Re/W-26%Re thermocouple wiresspot-welded to the edge of the rear surface. Cu and

Pd were evaporated from sources containing thecorresponding high-purity metal wire tightly wrappedaround a tungsten filament. Scanning tunnelingmicroscopic (STM) and atomic force microscopic(AFM) studies were performed with a Digital Instru-ments Nanoscope II.lll. Results and DiscussionA. Model Single-Crystal Catalystsa. CO Methanation over Ni, Ru Single CrystalsEarly studies in our laboratories35*17'13 of themethanation reaction or the reaction of CO with H2to make methane showed that metal single crystalscould be used to model reactions taking place onsupported metal particles. In these studies similari-ties among the data for the close-packed Ni(111)5 andRu(001)1~13 surfaces, the more open Ni(100) and Ru-(110)18 crystal planes, and data for alumina-sup-ported nickel particles demonstrated that the meth-anation reaction is indeed quite insensitive to thesurface structure of nickel and ruthenium catalysts.b. C0 Oxidation over Flh, Flu, lr, Pt, and PdMore recently this structure insensitivity has beenextended to other reactions including the CO oxida-tion reaction. The relative simplicity of CO oxidationmakes this reaction an ideal model of a heteroge-neous catalytic reaction. Each of the mechanisticsteps (adsorption and desorption of the reactants,

stuface reaction, and desorption of products) has beenprobed extensively under UHV conditions with themodern techniques of surface science.19 Becausemany of the reaction parameters determined in UHVcan be applied directly to the kinetics at higherpressures, CO oxidation to date best illustrates thecontinuity between UHV and elevated pressure stud-ies.Goodman1000. 1 if 9 +,-


2/30

,5 oRh (111) J'L onnhool ;co, FORMAT on RATE (TOF).=>9.101 Rh/M 0 PcO= a Torr-L2 3 Po = a Torri 2.. 1+ 5. \0.1+ , -.1 \\ 1\\"' \ T\ ,. H1 \0.0014" \

Rh/A1203 \~l \ l0.0001.- .4 4 4-4 11.2 1.6 2.0 2.4 2.81/1' 1 103 (K)Figure 1. Comparison of the specific rates of the CO-O2reaction measured over Rh(111) and Rh/A1203 at P90 = P0,= 0.01 atm. Model predictions were obtained using twodifferent oxygen adsorption kinetics.Figure 1 compares specific reaction rates [(productmolecules)(substrate surface atom)'1(second)"1] orturnover frequency (TOF) of CO oxidation measuredover single crystals of Rh with those observed onsupported Rh/A1203 catalysts.2*21 There is remark-able agreement between the model and supportedsystems with respect to the specific reaction rates andapparent activation energies. These results showthat the kinetics of CO oxidation on Rh is notsensitive to changes in catalyst surface morphology,even from bulk Rh single crystals to relatively smallsupported Rh particles. Under the conditions ofFigure 1, the surfaces are predominantly coveredwith CO so that the reaction is limited by theadsorption rate of oxygen.2~21 As the temperatureis increased, the reaction rate increases because morevacant sites become available for oxygen adsorptionas a result of the higher CO desorption rate. In

Figure 1 the CO oxidation rate increases with tem-perature following an apparent activation energyvery similar to that for CO desorption.The behavior of the reaction can be analyzed byusing a kinetic model established from UHV surfacescience studies of the interactions of CO and O2 withRh.2 By using the rate constants for adsorption anddesorption of CO and O2 measured at UHV condi-tions, the kinetics of CO oxidation over Rh can bevery accurately predicted at elevated pressures. The


3/30

dashed line of Figure 1 represents the predictedArrhenius behavior from this model?" This closecomparison between the single-crystal results andthose of the model represents an exceptional exampleof the continuity between UHV surface science andreal world catalysis.Further examples of the correspondence betweenmodel single crystals and supported metal catalystsfor CO oxidation have been found for Ru,22*23 Pt, Ir,and Pd.226 In the case of Ru,22'23 post-reaction

lModel Studies in Catalysis Using Surface Science Probesiiiiiwl; fif*"ilT TF1 lF*W I l%;"TIR i11/ AUGER HAT O48eV)l>El

\+4 -J_...|._..iP-1.UlN CKELv)/N'(aA" mnoo) -;- 1T26P3i/ iCARBON TOC(2o.osNl(111) _~4_. -f o_ sO_________,_,,o-f O JQ90 _ kn: *4 l_._._. L it - $_20 4o so0EFFECTIVE REACTION TIME (seconds)Figure 2. Methane decomposition kinetics on low indexnickel single crystal surfaces at 450 K and methane

pressure Pmgthane of 1.00 Torr.9surface analysis indicates that the optimum rate ofCO oxidation on Ru(001) is observed when thesurface is covered by almost a monolayer of oxygen.This surface condition contrasts with that of Rh,where the optimum activity is obtained for surfacesessentially free of oxygen adatoms.2-21 IIn contrast to methanation and CO oxidation, thereare reactions whose reactivity and selectivity dependmarkedly on the surface geometry or the metallic


4/30

particle size of the catalyst. The reactive sticking andhydrogenolysis or cracking of alkanes are importantexamples of structure sensitive processes.c. Alkane Reactive Sticking on Nickel Single CrystalSurfacesThe reactive sticking of alkanes on nickel single-crystal surfaces is strongly dependent on the surfacestructure. For example, methane reactivity, shownin Figure 2 as the time-dependent carbon buildupfrom 1 Torr of methane in contact with the variouslow index nickel single crystal surfaces at 450 K, isseen to increase in the order Ni(111) < Ni(100) < Ni-(110).27 Initial reaction rates for the N i( 110) and Ni-(100) surfaces are very similar and are ~7-10 timesgreater than the initial rate for Ni(111) surfaces at450 K. However, both the Ni( 100) and Ni(11l)surfaces exhibit a strong coverage dependence in themethane decomposition rate, as evidenced by thedeviation from linearity in the plots for these surfacesshown in Figure 2. This behavior is in contrast with

that of the Ni(110) surface, which does not exhibitthe same downward curvature, possibly indicatingcarbon islanding or a reduced carbon coverage de-pendence for the methane reactivity on this surface.A comparison with the results of molecular beamstudies suggests that the effects of vibrational energyon sticking probabilities must be accounted for andChemical Reviews, 1995, Vol. 95, No. 3 525the sticking probabilities of molecules with very lownormal kinetic energies must be accurately knownwhen attempting to model high-pressure processesusing molecular beam techniques. While the dis-sociation of methane indicates that the directchannel to dissociation might dominate for certaincrystal planes, e. g. Ni(111), the results for the higheralkanes imply that dissociation of these moleculesproceeds primarily via a trapped molecular precur-sor.28'29These studies on the alkane-reactive sticking onnickel single-crystal surfaces were carried out underthe high incident flux conditions. The elevatedpressures, in general, are required to produce mea-surable products, not because of the greater avail-ability of higher velocity molecules, but rather be-cause of the competition that is inevitably presentbetween desorption from the precursor or adsorbedstate and dissociation. Since the activation energies

for desorption of many reactants of interests (par-ticularly hydrocarbons) are frequently smaller thanthe activation energies for reaction, desorption domi-nates and reaction probabilities are quite small, oftentoo small to measure at UHV conditions. For thesereactants, the greater number of collisions at higherpressures simply serves to overcome this limita-tion.23=29By extrapolating the rates of dissociative adsorp-tion of alkanes measured in thermal bath experi-


5/30

ments to the pressure and temperature conditionsused in the hydrogenolysis and steam reformingstudies, it has been found that ethane dissociationrates on the clean surface are 1-2 orders of magni-tude larger than the rates of ethane hydrogenolysisand reforming and n-butane dissociation on the cleansurface is 2-~3 orders of magnitude faster thann-butane hydrogenolysis under comparable condi-tions.28 Since the alkane dissociation rates measuredin these thermal bath experiments are initial ratesmeasured in the limit of zero carbon coverage, theyrepresent a theoretical upper limit to the rates ofhydrogenolysis and steam re-forming of these alkaneson unpromoted nickel catalysts.In summary, these studies have shown a directrelationship between the atomic corrugation of asurface and its activity toward the dissociative stick-ing of alkanes. Since dissociative sticking is the keyinitial step in the hydrogenolysis of alkanes, then asimilar correlation between the atomic roughness

and activity for this reaction might be anticipated.This indeed is the case as is shown in the followingsection.d. Hydrogenolvsis of Small Alkanes on Nickel and IridiumSingle Crysla SurfacesThe reactivity for ethane hydrogenolysis to meth-ane on nickel has been shown to depend critically onthe particular geometry of the surface. Figure 3shows the specific reaction rates for methane forma-tion from ethane over Ni(100) and N i(11l) surfacesplotted in Arrhenius form.3 It can be seen that themore open (100) surface is far more active than theclose-packed (111) surface. For the Ni(100) surface,the data give an apparent activation energy of 100kJ/mol, which is remarkably close to the 105 kJ/mol

l526 Chemical Reviews, 1995, Vol. 95, No. 3if*'l':7"" If ' T 7 v w IF fifiii H2 2s.6.ra (TOF)H2/c;H6=1ooP1=1oo TOHR I1o"i- IT m(1oo) .,4 m(111) METHANE FORMAT ON RATE,.-_ SMALL ii

. PQHTICLES T-210 i CARTER,cusummo,ii 81 S|NFELTI uses - i1 0.3? PARTICLESy, Bl/c5H=6:s I * P1-511s1'onR i-4


6/30

10 [H .. mi eh ~ .- J; - --is .---ii1.s 1.a 2.0 2.21000/T (K")Figure 3. Arrhenius plot for ethane hydrogenolysis reac-tion on Ni(100) and Ni(111) surfaces at a total reactantpressure PT of 100 Torr and Hp/C2H6 = 100.12 Also shownis the result on supported nickel catalysts at PT of 175 Torrand H?/CQHG = 6.6.22 Aobtained for the methanation reaction over the samesurface.5 Furthermore, the specific rates observed forboth methanation and ethane hydrogenolysis on thissurface are virtually identical for the same partialpressure of hydrogen. These observations stronglysuggest that these two reactions over the Ni(100)surface are following the same reaction pathway andare limited by the same reaction step. As previously@ n-(110)-(112)TOP V EWI -if l"i'}.i'li- To

gig. ._-.1 .f.T ._-,sf . I ,6o e -.\ c-' ay >2-I >2 sl' 10 nm). By assumingparticle shapes, the general behavior of decliningselectivity with larger particle size can be accuratelymodeled, as illustrated in Figure 4.The stoichiometry of the surface intermediateleading to high ethane selectivity, based on kineticsand surface carbon coverages subsequent to reaction,is suggested to be a metallocyclopentane.3435 The Ir-(110) surface undergoes a reconstruction, describedas the Ir(110)-(1 x2) or missing-row structure,resulting in rows of the highly coordinatively unsat-urated C7 sites, as schematically shown in Figure4. These sterically unhindered C7 sites can form ametallocyclopentane species (e.g., a 1,4-diadsorbedhydrocarbon species) which has been proposed as anintermediate in the central scission of butane toethane. On the basis of analogous chemistry re-ported in the organometallic literat1n'e,3'3-39 the mech-anism responsible for the hydrogenolysis of n-butane

on the Ir(110)-(1x2) surface has been postulated tobe the reversible cleavage of the central CC bondin this metallocyclopentane intermediate. On theother hand, butane hydrogenolysis on the Ir(111)surface appears to operate via a different mechanism.First, dissociative chemisorption of butane and hy-drogen occurs followed by irreversible cleavage of theterminal carbon--carbon bond of the adsorbed hydro-carbon. Further C-C bond cleavage prior to productdesorption leads to the methane and ethane observedas initial products.For both iridium surfaces, the extent to whichhydrogenolysis proceeds increases with increasingreaction temperature. This is in keeping with thegeneral trend for increased cracking at higher tem-peratures for alkane reactions. The term roll-overhas been used to describe the fall in overall activityat the high temperatures which leads to a decreasein the selectivity for the production of ethane (shownin Figure 5) in the hydrogenolysis of propane overthese two iridium surfaces. Decreasing the partialpressure of H2 at the temperature of onset of roll-over induces the same selectivity change as observedfor an increase in reaction temperature. The originof this effect is believed to be as follows. As thereaction temperature is raised beyond a criticaltemperature, the hydrogen surface coverage falls

below a saturation or critical coverage. The lowerhydrogen coverage then reduces the efficiency of thehydrogenation of surface hydrocarbon fragments. Itis shown in Figure 5 that the roll-over onset occursat a higher temperature on the Ir(110)-(1x2) thanthat on the Ir(111) surface. From previous studies,it is known that hydrogen desorbs at a highertemperature (390 K at the saturation of the high-temperature desorption state) from Ir(110)-(1>


10/30

gen).4 Therefore, the higher temperature of onsetof roll-over on the more open Ir(110)-(1 x2) surfacecorrelates with the higher binding energy of hydrogenadatoms on this surface. This suggests that S theactive chemisorbed hydrogen is present on metal sitesrather than carbon-covered sites.The atomic-level details of the structure sensitiv-ity of alkane hydrogenolysis are still incomplete;however, this work shows clearly that a large com-ponent of this important catalytic effect is a struc-tural one and relates to the partitioning of the facetdistribution as a function of A particle size. All of theresults summarized above correlate very well withmeasurements on supported real-world catalystsand emphasize the important role that single crystalstudies can play in defining relationships betweensurface structure and catalytic activity/selectivity.

528 Chemical Reviews, 1995, Vol. 95, No. 3B. Model Oxide-Supported Metal Catalysts

Although single crystals can be used successfullyto model a variety of heterogeneously catalyzedreactions, there is a clear need to develop morecomplex models to address those important issuesspecifically related to very small metal particles andto interactions between the metal particle and thesupport.The first challenge in the synthesis of an oxide-supported metal model catalyst is the preparationand characterization of a suitable oxide catalystsupport. Unfortunately, oxide surfaces such as thosefrequently encountered as catalyst supports areinsulating materials and present problems of varyingdegrees to many charged-particle surface probes.Recently the difficulty associated with surface charg-ing has been eliminated in our laboratories bypreparing an ultrathin, highly ordered, oxide film onthe surface of a metal substrate. Any charginginduced in the thin film during charged-particlemeasurements is dissipated via the conducting sub-strate. This section begins by describing the experi-mental approach to the preparation of ultrathin filmsof silica and alumina and then moves to the synthesisand characterization of supported metal particles ofcopper and palladium.a. Preparation and Characterization oi the OxideSupports

Thin SiO2 films have been synthesized on a Mo-(110) substrate by evaporating silicon into an oxygenbackground. A Mo(110) single crystal was chosen asthe substrate because of its relative ease of cleaningand the ability to thermally desorb the thin films ofSiO2 from the su1'face.4142Silicon dioxide is formed on the substrate at roomtemperature when silicon is evaporated in a ~4 x10* Torr O2 backgi-ound.4143 Figiu'e 6 shows theAuger electron spectra for the films prepared under


11/30

oxygen background pressures of 0, 1 x 10"6, and 4 x10-6 Torr. Clearly, both silicon and silicon dioxidell fur I If if TIT TTITW" 7* if I 0*unrrsll 76 eVSiO: 3: -_ r:,_,_,n_ _. _. El _- J il___ Mir VJ . - 1- _ I ' 1 {"1 "7 it S llE ARB;_-E/ddI""'h-1I111 .l .dNy Mo C . i 91 12V ll

1 SilLVVl30 130 230 330 430 530KINETIC ENERGY [eV)Figure 6. Auger electron spectra for thin films of siliconand silicon dioxide on Mo(110) prepared by evaporatingsilicon at oxygen partial pressures of (A) 0, (B) 1 x 10',and (C) 4 x 10'6 Torr.GoodmanF --_ _ ..._.---..-.____.__.,__,,_.i,.i,.,__,,____.i,,l sooc pll ll 1? 14 12 10.5 eVMUN TSE ARB.sooc f _H l/ dNE/d: l 30C 7 eVA; | 5 eVR | , M F .l . ._____l _ _.___4-__ 44:; x_i__*.*,n1_m____ _ ____1 1, . __m10 5 025 20 15ELECTRON ENERGY LOSS (eV)Figure 7. First-derivative electron energy loss spectra fora ~50 A silicon dioxide film on Mo(110) annealed to 80,

500, and 900 C. The resolution of the analyzer is ~0.6 eV.are deposited in the intermediate oxygen pressurerange of 1 x 10" Torr. The Auger spectra and therelative intensities of silicon and oxygen do notchange for films prepared at oxygen pressure greaterthan 4 x 10* Torr, consistent with the productionof only a silicon dioxide film at the higher oxygenpressure conditions. The silicon and silicon dioxidespecies are differentiated on the basis of their char-acteristic Auger transition energies and line shapes.


12/30

Silicon dioxide has characteristic LVV Auger transi-tions at 76, 63, and 59 eV, whereas silicon has amajor Si(LVV) peak at 91 eV.43-*4 In addition, anAuger transition energy of ~85 eV has been observedfor the SiO species on platinum and on silicon.The Auger spectra shown in Figure 6 furthersuggest that the silicon oxide films are stoichiometric,i.e., SiO2. The presence of SiO (silicon monoxide) inthe silicon oxide films can be ruled out on the basisof the absence of an Auger transition at ~85 eV.Other siliconoxygen species, such as SigO andSi2O3, are not likely to be formed, since thesespecies do not exist as stable compounds; thesespecies have only been proposed to exist at the Si!SiO2 interface.-43 Fiuthermore, the line shapes, thepeak energies of the Auger electronics, and theelectron energy loss spectra (Figure 7) are consistentwith those of silicon dioxide.Figure 7 shows the first-derivative electron energyloss spectra of the silicon dioxide films annealed to

various temperatures. Vitreous silicon dioxide ex-hibits optical excitations at 10.3, 11.8, 14.0, 17.0, and22-23 eV.49 In addition, electron energy loss fea-tures at 10, 12, 14, 17, 20, and 23 eV have also beenreported for thermally grown silicon dioxide on asilicon substrate.-51 There are no electron transi-tions below 10 eV for vitreous silicon dioxide.The electron energy loss spectrum for the silicondioxide films after annealing to 900 C displaysfeatures identical to those of vitreous silica. Thefeatures at ~10, 12, 14, 17, and 20 eV are evident,although not well resolved. These electron energyfeatures are attributed to interband transitions insilicon dioxide whereas the 23 eV feature is due to abulk plasmon excitation.

Model Studies in Catalysis Using Surface Science Probesll ' ' ' '7 ll Ml .. l10/0 _ _ 1252 cm 1 i. | .I-1160 cm'1 12ooc11ooc if1oooc I aooc A l

l sooc 400 c0 l 1l soc ;* -W 1; __.i ~ _1500 1400 1300 1200 1100 1000FREQUENCY(cm")Figure 8. Infrared reflection absorption spectra for a ~50A silicon dioxide film eh Mo(110) annealed to 50, 400, coo,


13/30


14/30

(LO) vibration modes are at 1076 (TO) and 1256 cm(LO) for AS1, and at 1200 (TO) and 1160 cm1 (LO)for AS2.56'57 The TO modes are optically active,whereas the LO modes are optically inactive for aninfinite crystal. These selection rules arise becausethe electromagnetic waves are transverse, and thusdo not interact with the longitudinal phonons. How-ever, for thin films with thickness much smaller thanthe wavelength of the lattice vibration, the vibra-tional modes normal to the surface have the fi'equen-cies of the longitudinal modes. For the vibrationalmodes parallel to the surface the frequencies cor-respond to the transverse modes. Since only thevibrational modes with a component normal to themetal surface are IR active, the IRAS spectra for thesilicon dioxide films exhibit only these featurescorresponding to the longitudinal modes. Thus thespectra from the films should differ from the corre-sponding spectra taken with transmission IR absorp-tion that mainly probes the transverse features.

The increase of the peak IRAS frequency with theannealing temperature for silicon dioxide films onMo(110) (Figure 8) suggests some ordering of filmsupon heating. The broad peak in the IRAS spectrafor the SiO; films is attributed mainly to a convolu-tion of the longitudinal optical modes of the asym-metric stretches of AS1 (1256 cm1) and AS2 (1160cm) of silicon dioxide. These two modes are coupledvia disorder in the SiO; films.565" The AS2 mode isoptically weak in crystalline on-quartz and its inten-sity increases with a concomitant decrease of the AS1intensity in amorphous silicon dioxide. Therefore,the increase of the intensity at 1250 cm and thedecrease of the intensity at 1160 cm stronglyindicate that the SiO2 film becomes more orderedwith heating. This is also consistent with the EELSresults.Thin A1203 films were grown at room temperatureon a freshly cleaned Mo(110) surface. The growth ofthe films was controlled at a rate of approximatelyone-half aluminum equivalent monolayer per minute.The background pressure of oxygen was 7 x 10" Torrduring film growth. The films were annealed to 1200K in oxygen ambient to improve their crystallinequality. Shown in Figure 9 are Auger spectraacquired following film synthesis at two representa-tive film thicknesses: (a) dA]2Q3 = 4.4 A and (b)dA1,0,

= 20.0 A. The thickness of the dA1,0, = 4.4 A filmwas calculated from the Auger intensity ratio of the

530 Chemical Reviews, 1995, Vol. 95, No. 3i AES._.. ii.1 Al2O3lMo(1 10)UN TSH'-. \


15/30

{-_' -=*-* : _~;.4~4--':1*_-a-jrr;-.*1-_-_.=-:=r-:.*-=-*;;T I?--'"; 1rt 1 b4. 1 lAl a \iENS TY ARBi Q iA i 0*LAUGEFI NT+-iL"T Al Mo /ig; E H . . -4 1__..-.-. - l~ --- 1;-,,,--- -l_, :1,-----10 100 200 300 400 500 600KINETIC ENERGY (eV)Figure 9. Auger spectra of A1202 films grown on Mo-(110); (2) d220, = 4.4 A; (b) am, = 20.0 A. The films were

annealed to 1200 K in oxygen ambient to improve theircrystalline quality. The spectra were acquired at Ep = 2.0keV and at a sample current of 5 ;.A.Al (LVV) and Mo(MNN) peaks using the followingequation:dAl 011 --iiii, _ [1,.,,..] ;_ nllmlIMo Illlloo [ dAl 0 11 012---1-"'MThe value of I A1+3/IM,, was obtained from thestandard Auger spectra of A1202 and Mo, respec-tively. The attenuation lengths, lA1+= and 1.122,, weretaken from the literature to be 3.753 and 6.7 A,59respectively. The thickness of the thick film (dA12Q3= 20.0 A) was estimated by extrapolating the valueof the thin film based on evaporation time andconstant aluminum flux.It is seen in Figure 9 that there is no indication ofthe presence of a metallic Al feature at 68 eV. Thepredominant spectral features are the Al*3 (LVV)transitions at ~54 eV and the 0 (KLL) transition at~500 eV. The spectral features in the 100--250 eVkinetic energy range of spectrum a arise from the Mosubstrate.

Structural studies of very thin A1202 films (dA1,0,5 8 A) using LEED showed a complex pattern inwhich most diffraction spots can be interpreted asarising from electron multiple scattering at theinterface. With an increase in film thickness, themultiple diffraction effects were attenuated. A simplehexagonal structure was seen at dA1,0, 2 15 Aalthough the diffraction spots were not as sharp asthose from multiple diffraction, indicating somedegree of disorder in the thick films. The lattice


16/30

periodicity derived from the LEED photographsrelates closely to the 0-0 distance in bulk A1202. Theobserved hexagonal pattern can therefore be inter-preted as due to the presence of an ordered, close-packed oxygen anion layer associated with either the(001) orientation of (1-A1203 or the (111) face ofy-A1202. A detailed structural analysis of the initialstage of epitaxy of A1202 films on a similar refractorymetal substrate, Ta(110), has been given elsewhere.Goodmanii HREELS.1 1 N203/Mo(1 10)~ ii i B751+-I-\inUN TSI .ii , i

.; I ---0 J__ .-175. 6402 j\ i 1255 1510 1735. j 370 b 1NS TY ARB1 pi asoNTE5 ;l:2::,-,-' * sen* 4 I25 14551685 *. -14 5 .. _ _4%__. WW _ - _ ,7_ __-500 0 ' 500 1000 1500 2000waveuumaea ism")Figure 10. HREELS spectra of A1202 films grown on Mo-(110); (2) aw, = 4.4 A; and (b) am, = 20.0 A. The filmswere annealed to 1200 K in oxygen ambient to improvetheir crystalline quality. The spectra were acquired at E,= 4.0 eV and at the specularly reflected beam direction.The growth of A1202 films has been fiirther exam-ined using HREELS, as shown in Figure 10. Thefundamental modes of the surface optical phonons(.below 1000 cm in frequency) and their multiplesand combinations are evident in the spectra. The

surface optical phonon losses of the thin film (spec-trum a) are characterized by a two-mode pattern,whereas three modes of the phonon losses are typicalfor the thick film (spectrum b). These results are ina complete agreement with previous data in theliterature.6"63 Frederick et al.64'65 have suggestedthat the appearance of the two phonon modes arecharacteristic of very thin A1202 films.Because silica and alumina supports are widelyutilized in practical catalytic applications, these oxide


17/30

films are convenient materials for constructing amore realistic supported-metalpartic1e catalyst,intermediate in complexity between metal singlecrystals and the supported real-world catalysts.b. Copper on SilicaModel silica-supported copper56'63 catalysts havebeen prepared by evaporating copper onto a silicathin film. The preparation conditions define thecorresponding metal particle dispersions or averagesize.7*8 As in the studies described in the previoussections, the silica films are supported on a Mo(110)substrate. The structure of the model silica-sup-ported copper catalysts has been investigated withIRAS and scanning tunneling microscopy (STM)."7The IRAS studies of adsorbed C0 indicate that thereare several types of copper clusters with surfacestructures similar to (111), (110), and other high-index planes of single-crystal copper.66"3 The STMstudies show several types of copper clusters on silicaand reveal images of metal clusters on the amorphous

supsport with atomic resolution as shown in Figure1 1.c. Palladium on SilicaThe adsorption and reaction of C0 on model silica-supported palladium catalysts over a wide range of

IiiiiiModel Studies in Catalysis Using Surface Science ProbesChemical Reviews, 1995, Vol. 95, No. 3 531Figure 11. A STM image (110 x 110 A) of Cu cluster formed by the depositionof 1.2 ML copper on sili_ca(100 A)/Mo(110).The Cluster exhibits a CuCu spacing of (4.4 >< 2.4) A. This image wasacquired after the sample was exposed to waterand dried in air..___.___L_ ._ Ton 5 j T=100K 1i .iPd(111) llPd(100)AbsorbancePdlSilicalMo(1 1 o)16(Pd) A yll

15 ML IF7ML1ML \'/X_1..._.__l_.____1_ .. .1-.. ..1 ..-. . ___2200 2100 2000 1900 1800Frequency (cm '1)Figure 12. A Comparison of infrared reflectionadsorp-tion spectra of C0 on Pd(111), Pd(100), and silica-supported


18/30

palladium surfaces. The spectra were collected at 100 Kwith a C0 pressure of 1 x 10"'5 Torr.temperatures and pressures have demonstrated acontinuity between catalysis on Pd single crystals25=26and Pd small particles69'72 and, as well, between thekinetics of C0 oxidation at low and high pressures.Pd overlayers of varying thickness on the thin silicafilms were annealed to 900 K prior to the C0adsorption experiments. This thermal treatment hasbeen shown to roduce small metal particles in therange 30-500 K, with larger particles being formedfor the higher Pd coverages. AThe IR spectra of adsorbed C0 on model silicasupported Pd catalysts are displayed in Figure 12 forPd coverages of 1.0, 7.0, and 15 monolayer (ML). ThePd particle size was determined by chemisorptionmethods69'7 and verified using scanning probe tech-niques (see, for example, Figure 13). Figure 12 showsFigure 13. An atomic force micrograph of an ~10 nm Si02film on Mo(100) subsequent to the deposition of 10 mono-

layer equivalents of palladium and an anneal to 700 K.IR spectra of adsorbed C0 on model Pd/Si02 catalystswith different Pd coverages show three distinctabsorption features corresponding to C0 adsorbedonto 3-fold hollow (1880 cm), bridging (1990 cm),and a-top (2110 cm) configurations. At 622 = 1.0ML the dominant absorption feature corresponds toC0 adsorbed onto an a-top position, while the peaksoriginating from 3-fold hollow and bridging CO arebroad, suggesting a nonhomogeneous distribution ofthese adsorption sites. Increasing the Pd coverageto 7.0 ML and then to 15 ML results in the sharpen-ing of the absorption features and the dominance ofthe peak originating from C0 adsorbed in the bridg-ing positions. These results demonstrate that thelarger particles have well-defined crystal orienta-tions, whereas the smaller particles have a widedistribution of adsorption sites. A comparison of theIR spectra of equilibrated CO overlayers obtained

532 Chemical Reviews, 1995, Vol. 95, No. 3Temperature (K)600 550 500 450r r r'** 0 *1 '5'"~ Pd,lSiO2 (100A)lMo(110) ,A 0 Pd ICO2 Format on Rate (TOF

4 a 1L 522222, T0,--- 0.3ML" - osurroot 'I ~ 2ML__ A A Pd(110)' iPd(111) \ML Pd(100)


19/30

1.50 1.70 1.90 1 2.10 2.301000IT (K' )Figure 14. The Arrhenius plots of C0 oxidation rate onPd(111), Pd(110), and Pd(100) and silica-supported pal-ladium. P20 = 1.00 Torr; P0, = 0.50 Torr.from the Pd15M1/Si02 sample and Pd(100) and Pd(111)single crystals reveals striking similarities. AsFigure 12 shows, the stretching frequencies of ad-sorbed C0 on the model Pd/Si02 catalyst are identicalto the combined features observed for Pd(100) andPd(111). The agreement among the IR spectra of themodel supported catalysts and the metal singlecrystals suggests that the metal particles formed onthe thin SiO2 film have facets consisting primarilyof (100) and (111) orientation. This is anticipated inthat the formation of metal particles with theseMiller indices is thermodynamically favored.The oxidation of C0 on this model system has beenstudied as well. Figure 14 shows the specific C02formation rates as a function of the reaction temper-

ature. The specific activities of the single crystal andthe supported catalysts for CO oxidation are es-sentially identical. The apparent activation energiesfor the relatively large particles (6p2 > 2 ML) aresimilar (~27 kcal/mol), but somewhat lower (~25kcal/mol) for the smaller particles (0.6 and 0.3 ML).The apparent activation energies for three Pd singlecrystals with different orientations are similar2526-7but distinctive [29.4 :l: 0.3 kcal/mol for Pd(100), 28.1:l: 0.4 kcal/mol for Pd(111), and 30.7 :l: 0.5 kcal/molfor Pd(110)]. An excellent correspondence is foundamong the results obtained for the model silica-supported Pd catalysts and the Pd single crystals forthe oxidation of C0.The effect of palladium particle size on the catalyticproperties was investigated'"~72 utilizing the decom-position of chemisorbed N0 (Figure 15) and reactionsof N0 with C0 in flowing conditions (Figure 16).N0 was used in order to differentiate N2 from C0and N20 from C02 with mass spectrometry. Duringtemperature-programmed reaction on the large par-ticles following a saturation N0 exposure, N0desorbs in three peaks centered at 315, 515, and 595K; N2 is produced in peaks at 545, 595, and 690 K;and N20 is evolved in a peak at 545 K (Figure 15A).The 545 K N2 peak is attributed to the fragmentationof N20. On small palladium particles (~50 A), N20

is not produced; N0 desorbs in two peaks at 310and 515 K, and N2 is evolved at 530 and 670 K(Figure 15B). The relative yield of N2 with respectGoodmanF l$5ii=dlsi0,r1ooA)lMo(11o) .___ A B8 6(Pd)-10 ML 6(Pd)-1.4 ML ll-l il no ,1 A1 nus-31 ,


20/30

, X1 l *1 """= ""3:9___./-/\L.i X3 N 0 me-46 x3-1-L :;* J: ,-I-_-f=* *__:'::~_;~.-: tr: :;;_%;L r. 4 l 4. 4 M. rugs l ll300 500 700 900 300 500 700 900Mass Spectrome er S'gna (a.u.)Temperature (K)Figure 15. Temperature-programmed reaction for N0on model silica-supported palladium catalysts. N0 wasadsorbed at 290 K to saturation. The palladium coverageswere (A) 10 monolayer, corresponding to a particle size f~250 A, and (B) 1.4 monolayer, corresponding to ~55-77 _"'"~W '*"*" "'T'*;:___ 7 r _ _,, *5 N0+CO(1:1), 10'5Torr i.-.. , PdlSiO2 *T A . B" it 9(Pd)-10 ML 6(Pd)-1 ML.A, co 2 mle-44 *5_ "- l xi A

at on Ra e a.u.lll X5 I *5 N2 mle-3 M"' = x1 iFormProduc200 400 600 soo 400 soo soo 1000Temperature (K)Figure 16. C0 + N0 reaction rates as a function oftemperature for small and large particle Pd catalysts. Thegas phase pressure was 1 x 10* Torr with equimolariC0and N0. The product formation rate was monitored viamass spectroscopy. The palladium particle sizes correspondto ~250 (A) and ~45 A (B).to N0 desorption (above 400 K) decreases as theparticle size grows (from ~50% for 30 A to ~20% for250 A particles), suggesting less N0 decompositionon the larger particles. In addition, the formationof N20 on the larger particles correlates with theappearance of the N0 desorption peak at 595 K.There is no detectable 02 evolution below 1000 K forall sizes of palladium particles. However, 02 de-sorption is observed in a peak at ~1250 K, concurrentwith the desorption of palladium. Therefore, theoxygen from nitric oxide decomposition is apparentlydissolved into the bulk of the palladium particles.

The reaction of C0rwith N0 under flowing condi-tions''72 further demonstrates a particle size effect(Figure 16). The relative rate of product evolutionwas monitored with a quadrupole mass spectrometerduring reaction with flowing N0 and C0 (1:1) at 1x 10" Torr. The reaction rate increases withthecatalyst temperature, reaching a maximum at ~580K and then declines. 0n the large palladium par-ticles (Figure 16A), the maximum rate of N20evolution occurs at 570 K, 10 K lower than the


21/30

maxima for both C02 and N2 production. On thesmall particles (


22/30

The growth of particulate Cu deposits on thesewell-defined A1202 films has been studied in the 80-800 K substrate temperature range." The filmsutilized for supporting Cu particles were typically 20A thick and exhibited excellent thermal stability andchemical inertness toward adsorption. It has beenreported that very thin A1202 films (dA12Q3 < 8 A)grown on A1 substrates react with Ni deposits atelevated temperatures, causing diffusion of Ni intothe substrate through defects in the A1202 film.It has long been known'*3 that particulate metaldeposits prepared at relative low substrate temper-Chemical Reviews, 1995, Vol. 95, No. 3 533TPD lCulAl2O3lMo(1 10; """\ llti li iUN TSillr

1 lARB.l lII-P ll-:-Teml Y_ d.6O0 Ac.400 Im/6=63.5 NTENS TY4 b.300a.80i l - I - _- _l -_%0 950 1&0 1000 F1100 1150Fi 773%TEMPERATURE noFigure 17. TPD spectra of Cu following Cu deposition onA1202 films at various substrate temperatures. The Cucoverages (62,, =-= 1.0) for each spectrum were adjusted tobe approximately the same. A linear heating rate of 10 K/swas used. Aatures exhibit smaller average particle size andhigher particle density than those prepared at highertemperatures; however, the low-temperature prepa-rations are unstable and undergo a major change inmorphology and size upon annealing or chemisorp-tion. To prepare thermally and chemically stableparticles of catalytic interest, it is imperative to carry

out the metal deposition at elevated substrate tem-peratures. Figure 17 shows a series of TPD spectraof Cu deposited onto an A1202 film at various sub-strate temperatures, Ts. The Cu coverages for eachspectrum were adjusted to be approximately thesame. Noteworthy features in Figure 17 are theshifts of the peak maxima to higher temperaturesand the narrowing of the TPD peak as T, is increased.The full width at half maximum (FWHM) of the TPDspectra, for example, decreases from ~70 K at T, =


23/30

80 K to ~50 K at T, = 600 K. Since the peak positionand the shape of the desorption spectra determinethe heat of sublimation, it is conceivable that at lowsubstrate temperatures, the broader TPD peak con-tains more components and thus reflects a broadersize distribution of the Cu particles. Our Augermeasurements further showed that Cu particlesprepared at T, = 600 K were thermally stable up toan annealing temperature of 850 K, whereas adecrease in Cu Auger intensity was observed uponannealing for those prepared at T2 = 80 K.Shown in Figure 18 is a family of TPD spectra ofCu deposited at Ts = 600 K as a function of the Cucoverage in equivalent monolayers, 62. 62,, == 1 isdefined as one monolayer (ML) of Cu on Mo(110), thecoverage that can be derived from the integrated TPDarea of the well-defined desorption feature of the Cumonolayer on Mo(110). Since the first monolayer ofCu grown on Mo(110) assumes the lattice structureof the substrate, the surface density of one equiva-

lent monolayer corresponds to 1.43 x 10 cm"? TheTPD spectra of Figure 18 exhibit an interesting trendin that the leading edge of the peak shifts continu-ously toward higher temperatures as 69,, is increased.It is also noteworthy that the TPD peak width,

534 Chemical Reviews, 1995, Vol. 95, No. 3CulAl2O3lMo(1 10) TSW* i.f i,_ .__i_.._.._ -...__.._-5 _ so 0 O 0 1-- . ._ .-. 0 O I._ . O 1.. 5, c I .._ 40 -1. - ___...L._ .4 I - 1+ I. 0 I 2 3 ii Cu EQUIVALENT MONOLAYER1 1*b d __ ., ...;".'/' ~ 3 I _ _ _ IJ-_---__....._-._..__.....L..__._. _ ,_ ~ _ e s e J_:_ _ __ 7 ' E ' ------ -- 1mo soo 900 woo 1100 1200 1300TEMPERATURE (K)Figure 18. A family of TPD spectra of Cu deposited at T,= 600 K as a function of equivalent monolayers, 63,: (a)

0.16; (b) 0.33; (c) 0.50; (d) 0.67; (e) 0.98; (D 1.25; (8) 1.55;(h) 2.09. The insert shows the heat of sublimation, derivedfrom the leading edge analysis of the spectra, as a functionof Cu coverage in equivalent monolayers. A linear heatingrate of 10 K/s was used.4'-in-5 NTENS TY ARB.UNHEAT OF SUBL MAT ONKcavmo e


24/30

8 300nve=630.20 ~ ~~~ ~ ,M Cu/M203/Mo(110)= T =600K, S0.16 .1 50A 100A *0.14-urr I I 1 | I ;B.UN TS4-Inn-AR0.12 -\0.10 f ~..... l \1_. \

TENS TY RAT O0.08 T"- 0.05 i" OAUGER N0.04 L ,@% 0 "0.02 A F.. 1Q00 Lean-_| - l - _L,,-,,,__i..__,_4 -0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5Cu EQUIVALENT MONOLAYEFIFigure 19. Auger intensity ratio of the Cu (LMM) peakto the O (KLL) peak vs the Cu coverage in equivalentmonolayers. The solid line is a theoretical simulation basedon the isotropic growth model, as discussed in the text. Thenumber density of Cu particles, derived from a curve fittingprocedure, is 1.4 >< 10 cm2. The average sizes of the Cuparticles are inserted as vertical bars.comparable to the width of the TPD peaks of Cumultilayers, remains approximately unchanged as69,, is varied. The insert in Figure 18 shows the heatof sublimation, derived from the leading edge analy-sis of the thermal desorption spectra, as a functionof Cu equivalent monolayer. The heat of sublimationof Cu-particles is seen to decrease rapidly from itsbulk value of 80 :l: 3 kcal/mol at 6C, = 1.2 to 49 kcal/mol at 60, = 0.2. Similar behavior has been observed

previously for the Cu/SiOg system. The decreasein the heat of sublimation may be attributed to adecrease in the number of neighboring Cu atoms asthe Cu particles become smaller.\1 TPD y TPD \Goodmanl 15NO+CO/Cu/A|2O3/Mo(1 10)b \g g mle-,=46.'5N2O i ITIIQI44, CO2 I IW


25/30

_ m/e-=31 NO 1n--~ _ 3 UN TSNTENS TY ARB.l;'4" 1"M--P~ mle-30, N2 i. .fi_~*__,~*;_,,_._e__,_~;_, . ,,_lxo.2 1

mle=-28, co . - 1 -;- A ;;&:1:5i~Ii:;:;ii ff'4l:l100 200 300 400 500 600 700 800TEMPERATURE (K)Figure 20. Thermal desorption signals for ml e = 28, 30,31, 44, and 46 from NO + CO adsorbed on Cu particlesat 80 K vs exposure time, t: (a) t = 60 s; and (b) t = 120 s.The exposure at t === 320 s corresponds to saturation. 63,,was 2.7, corresponding to a cluster size of ~105 A. The

15NO and CO reactants were fully mixed in the gasmanifold with a 15NO/CO ratio of 1:1 prior to admissioninto the UHV chamber. A linear heating rate of 5 K/s wasused.The average size of the Cu particles can be esti-mated from Auger measurements." Figure 19 showsAuger intensity ratios of the Cu (LMM) peak to theO (KLL) peak versus Cu-equivalent monolayers.Each data point was collected after a fresh depositionof Cu onto the clean A1203 surface at T, = 600 K. Thesolid curve through the Auger data in Figure 19represents a theoretical simulation based on a simplemodel described as follows:(1) Nucleation takes place homogeneously.(2) After an initial nucleation period, the numberof particles remains constant until the stage wherecoalescence sets in.(3) The particle shape remains constant duringgrowth.Using the algorithm previously discussed" thedensity of Cu particles, N, can be computed from acurve fitting procedure. The calculated number ofN = 1.4 x 1011 cm'2 compares favorably with thoseobtained by direct measurements with transmissionelectron microscopy (TEM): for the gold/mica system,N = 1.0-1.5 >< 1011 cm2 at T, = 623 K;33 forthesilver/amorphous carbon system, N = 1.0 x 1011 cm

at T, = 742 K. The average sizes of the Cuparticles, derived from a curve fitting procedure, areinserted as vertical bars in Figure 19.The reaction of nitric oxide and carbon monoxidewith particulate Cu deposits supported on A1203 filmshas been studied using TPD and HREELS." NOand CO reactants were fully mixed in the gasmanifold with a 15NO:CO ratio of 1:1 prior to admis-sion into the UHV chamber. The adsorption of the


26/30

Model Studies in Catalysis Using Surface Science Probes- HREELS ,. I, '5NO+COlCuIAl2O3/Mo(110) 1--. \._ L ~ 1:100. p \ Q0I 15_ bi N20 |sN0 2119W 1"" 255 1465 3H -P P i \ c_ W ii j ~ 2115 i;N _ 1475l 1 bL 1 .P ass AP J 5 Qlei - - I - -__-;;_7*___ __*__:J* _ __ ________-500 0 500 1000 1500 2011' 2500wnveuuusea ism")

Figure 21. HREELS spectra of (a) a clean A1203 film,am, = s A; (b) Cu/A1203, d,,,,,, = s A, 0,, = 0.67, followinga saturation exposure of N0 + C0 at 80 K; and (c) Cu!A1202, d,i,|,Q, = 8 A, 69,, = 2.7, following a saturationexposure of 15N0 + C0 at 80 K. The N0 and C0 reactantswere fully mixed in the gas manifold with a 15N0/C0 ratioof 1:1 prior to admission into the UHV chamber. Thespectra were acquired using an electron beam with aprimary energy of El, =-= 42 eV and at the specularlyreflected beam direction.NTENS TY AFlB.UN TSgas mixture was then carried out via a gas doser atT, = 80 K. Thermal desorption products from N0+ C0 adsorbed 2 on the Cu particles have beenfollowed by monitoring several masses (ml e = 28,30, 31, 32, 44, and 46). Shown in Figure 20 aredesorption signals for m-/e = 28, 30, 31, 44, and 46following two different exposures. 69,, in Figure 20was 2.7, corresponding to a cluster size of ~105 A.No desorption signal for ml e = 32 was observed inthe 80-850 K temperature range, indicating theabsence of gaseous 02 product. The desorption of theparent molecules of N0 and C0 and the gaseousproducts N2 and N20, due to 15N 0 decomposition,are evident in Figure 20. The observation of a smallC02 desorption peak (mf e = 44) at temperaturesbetween approximately 150 and 250 K in the TPD

spectra of Figure 20 is a clear indication for thereaction between C0 and N0 on the supported Cuparticles. At 62, 5 0.67, however, the C02 desorptionpeak (not shown here) was not observed. Since theC02 yield is very small, the effects of particle size onC02 production is still unclear.The adsorption of N0 + C0 on particulate Cudeposits has also been studied using HREELS."Shown in Figure 21 are HREELS spectra acquiredfollowing a saturation exposure of N0 + C0 (15N0/


27/30

C0 = 1:1). For reference, the spectrum of the cleanA1202 surface is also included. Exposure of the cleansurface to the N0/C0 gas mixture gives rise toseveral adsorbate loss features in the 10002500cm frequency range. Very similar to 15N0 adsorp-tion, the 1255 and 1465 cm'1 losses are attributedto the v(15N0) mode of adsorbed N0 and N20,respectively. The loss peak at 2110 cm is due toexcitation of the carbon-oxygen stretch, 1/(C0), ofadsorbed C0.Chemical Reviews, 1995, Vol. 95, No. 3 535IV. ConclusionsAn approach that combines ultrahigh vacuumsurface analytical methods with an elevated pressurereactor can provide new information about the mo-lecular details that define and control the mechanismof reactions at the gas/solid interface. By using thesetechniques, basic concepts of heterogeneous catalysissuch as structure insensitivity and structiue sensi-tivity can be directly addressed. For structure-

insensitive reactions excellent agreement can beobtained between studies on single-crystal surfacesand studies on the corresponding high-surface-areasupported catalysts, demonstrating the relevance ofkinetics measured on well-ordered single-crystalsurfaces for modeling the behavior of practical cata-lysts. For structure-sensitive reactions, the activityof a particular site or set of sites can be examinedand the effects of surface structure explored in atomicdetail.Recent studies have shown that model oxide sur-faces can be prepared in thin-film form, a preparationthat readily enables their exploration with a widearray of charged-particle surface techniques. Theaddition of metals to the above oxides as supportsprovides a convenient method to model importantaspects of supported metal, catalysts such as sup-port-particle interactions and particle size effects.Using such models a host of surface science tech-niques can be utilized to study catalysis by metalsin systems with well-defined particle sizes and mor-phologies. AV. AcknowledgmentsWe acknowledge with pleasure the support of thiswork by the Department of Energy, the Office ofBasic Sciences, the Division of Chemical Sciences,and the Robert A. Welch Foundation.

VI. References(1) Kahn, D. R.; Petersen, E. E.; Somoljjai, G. A. J. Catal. 1974, 34,294.(2) Sexton, B. A.; Somorjai, G. A. J. Catal. 1977, 46', 167.(3) Goodman, D. W.; Kelley, R. D.; Madey T. E.; Yates, J. T., Jr.Proc. of Symposium on Advances in F ischer-Tropsch Chemistry;American Chemical Society: Anaheim, CA, 1978.(4) Bonzel, H. P.; Krebs, H. J. Surf. Sci. 1980, 91, 499.(5) Goodman, D. W.; Kelley, R. D.; Madey T. E.; Yates, J. T., Jr. J.Catal. 1980, 63, 226.


28/30

(6) Campbell, C. T.; Paffett, M. T. Surf. Sci. 1984, I39, 396.(7) Rodriguez, J. A.; Goodman, D. W. Surf. Sci. Rep. 1991, 14, .1.(8) Goodman, D. W. Annu. Rev. Phys. Chem. 1986, 37, 425.(9) Vannice, M. A.; Garten, R. L. J. Catal. 1979, 56, 236.(10) Campbell, R. A.; Goodman, D. W. Rev. Sci. Instrum. 1992, 63-(1), 172.(11) Szanyi, J.; Goodman, D. W. Rev. Sci. Instrum. 1993, 64(8), 2350.(12) Wu, M.-C.; Corneille, J. S.; Estrada, C. A.; He J.-W.; Goodman,D. W. Chem. Phys. Lett. 1991, 182, 472.(13) Wu, M.-C.; Estrada, C. A.; Corneille, J. S.; Goodman, D. W. J.Chem. Phys. 1992, 96, 3892. .(14) Xu, X.; Goodman, D. W. Appl. Phys. Lett. 1992, 61(7), 774.(15) Xu, X.; Goodman, D. W. Surf. Sci. 1993, 282, 323.(16) Chen, P. J.; Goodman, D. W. Surf Sci. 1994, 312, L767.(17) Goodman, D. W.; White, J. M. Surfl Sci. 1979, 90, 201.(18) Kelley, R. D.; Goodman, D. W. Surface Sci. 1982, 123, L743.(19) Engel, T.; Ertl, G. The Chemical Physics of Solid Surfaces andHeterogeneous Catalysis; King, D. A., Woodruff, D. P., Eds.Elsevier: Amsterdam, 1982; Vol. 4.(20) 0h, S. I-l.; Fischer, G. B.; Carpenter, J. E.; Goodman, D. W. J.

Carol. 1986, I00, 360.(21) Berlowitz, P. J.; Goodman, D. W.; Peden, C. H. F.; Blair, D. S.J. Phys. Chem. 1988, 92, 1563.(22) Goodman, D. W.; Peden, C. H. F. J. Phys. Chem. 1986, 90, 4839.

536 Chemical Reviews, 1995, Vol. 95, No. 3(23) Peden, C. H. F.; Goodman, D. W. J. Phys. Chem. 1986, 90, 1360.(24) Berlowitz, P. J.; Peden, C. H. F.; Goodman, D. W. J. Phys. Chem.1988, 92, 5213. -(25) Szanyi, J.; Goodman, D. W. J. Phys. Chem. 1994, 98, 2972.(26) Szanyi, J .; Kuhn, W. K.; Goodman, D. W. J. Phys. Chem. 1994,98, 2978. -(27) Beebe, T. P., Jr.; Goodman, D. W.; Kay, B. D.; Yates, J. T., Jr.J. Chem. Phys. 1987, 87, 2305-2315.(28) Sault, A. G.; Goodman, D. W. J. Chem. Phys. 1988, 88, 7232-7239. Sault, A. G.; Goodman, D. W. In Molecule-Surface Interac-tions, Lawley, K., Ed.; Advances in Chemical Physics; John Wiley& Sons: England, 1989; Vol. LXXVI.(29) Goodman, D. W. Catal. Today 1992, 12, 189. Jiang, X.; Goodman,D. W. Applied Physics A 1990, 51, 99. Jiang, X.; Goodman, D.W. Catal. Lett. 1990, 4, 173-180.(30) Goodman, D. W. Surf Sci. 1982, 123, L679-L685.(31) Clarke, J. K A.; Rooney, J. J. Adv. Catal. 1976, 25, 125-183.(32) Carter, J . L.; Cusumano J . A.; Sinfelt, J. H. J. Phys. Chem. 1966,70, 2257-2263.(33) Martin, G. A. J. Catal. 1979, 60, 452-459.(34) Engstrom, J. R.; Goodman, D. W.; Weinberg, W. H. J. Am. Chem.

Soc. 1986, 108, 4653-4655.(35) Engstrom, J . R.; Goodman, D. W.; Weinberg, W- H. J. Am. Chem.Soc. 1988, 110, 8305-8319.(36) Chan, C.-M.; Van Hove, M. A.; Weinberg W. H.; Williams, E. D.Surf Sci. 1980, 91, 440"-448.(37) Foger, K.; Anderson, J. R. J. Catal. 1979, 59, 325-339.(38) Grubbs, R. H.; Miyashita, A. J. Am. Chem. Soc. 1978, 100, 1300-1302.(39) Grubbs, R. H.; Miyashita, A.; Lin, M.; Burk, P. J. Am. Chem.Soc. 1978, 100, 2418-2425.


29/30

(40) Szuromi, P. D.; Engstrom, J. R.; Weinberg, W. H. J. Chem. Phys.1984, 80, 508-517. '(41) Xu, X.; Goodman, D. W. Appl. Phys. Lett. 1992, 61(7), 774.(42) X11, X.; Goodman, D. W. Surfl Sci. 1993, 282, 323.(43) Davis, L. E.; MacDonald, N. C.; Palmberg, P. W.; Riach G. E.;Weber, R. E. Handbook of Auger Electron Spectroscopy; PhysicalElectronics Division, Perkin Elmer: Eden Prairie, MN, 1976.(44) Joyce, B. A.; Neave, J. H. Surf Sci. 1971, 27, 499.(45) Ko, C. S.; Gorte, R. J. Surf Sci. 1985, 155, 296. -(46) Derrien, J.; Commandr, M. Surf: Sci. 1982, I18, 32.(47) Braun, W.; Kuhlenbeck, H. Surfl Sci. 1987, 180, 279.(48) Bianconi, A.; Bauer, R. S. Surf Sci. 1980, 99, 76.(49) Phillip, H. R. J. Phys. Chem. Solids 1971, 32, 1935.(50) lbach, H.; Rowe, J. E. Phys. Rev. B 1974, 9, 1951.(51) Adachi, T.; Helms, C. R. Appl. Phys. Lett. 1979, 35, 199.(52) Olivier, J.; Faulconnier, P.; Poirier, R. in The Physics of SiO2and Its Interfaces; Pantelides, S. T., Ed.; Pergamon Press: NewYork, 1978; pp 8993. -Goodman(53) Liseke, N.; Hezel, R. Thin Solid Films 1979, 61, 217.

(54) Berreman, D. W. Phys. Rev. 1963, 130, 2193.(55) Almeida, R. M.; Pantano, C. G. J. Appl. Phys. 1990, 68, 4225.(56) Kirk, C. T. Phys. Rev. B 1988, 38, 1255.(57) Lange, P. J. Appl. Phys. 1989, 66, 201.(58) Chen, J. G.; Crowell, J. E.; Yates, J. T., Jr. Phys. Rev. B 1987,35, 5299.(59) Battye, F. L.; Jenkin, J. G.; Liesegang, J .; Leckey, R. C. G. Phys.Rev. B 1974, 9, 2887.(60) Erskine, J. L.; Strong, R. L. Phys. Rev. B 1982, 25, 5547.(61) Strong, R. L.; Firey, B.; "de Wette, F. W.; Erskine, J. L. Phys.Rev. B 1982, 26, 3483.(62) Crowell, J. E.; Chen, J. G.; Yates, J. T.,Jr. Surf Sci. 1986,165,37.(63) Chen, P. J .; Colaianni, M. L.; Yates, J . T., Jr. Phys. Rev. B 1990,41, 8025.(64) Frederick, B. G.; Apai G.; Rhodin, T. N. J. Electron Spectrosc.Relat. Phenom. 1990, 54/ 55, 415.(65) Frederick, B. G.; Apai G.; Rhodin, T. N. Phys. Rev. B 1991, 44,1880.(66) Xu, X.; Vesecky, S. M.; Goodman, D. W. Science 1992, 258, 788.(67) Xu, X.; He, J.-W.; Goodman, D. W. Surf. Sci. 1993, 284, 103.(68) Xu, X.; Goodman, D. W. J. Phys. Chem. 1993, 97, 683.(69) Xu, X.; Goodman, D. W. J. Phys. Chem. 1993, 97, 7711.(70) Xu, X.; Szanyi, J.; Xu, Q.; Goodman,'D. W. Catal. Today 1994,21, 57.(71) Xu, X.; Chen, P.; Goodman, D. W. J. Phys. Chem. 1994, 98, 9242.(72) Xu, X.; Goodman, D. W. Catal. Lett. 1994, 24, 31.

(73) Davies, P. W.; Lambert, R. M. Surf. Sci. 1981, 110, 227.(74) Schmick, H. D.; Wassmuth, H. W. Surf. Sci. 1982, 123, 471.(75) Jorgensen, S. W.; Canning, N. D. S.; Madix, R. J . Surfi Sci. 1987,1 79, 322.(76) Wickham, D. T.; Banse, B. A.; Koel, B. E. Surfl Sci. 1991, 243,83.(77) Wu, M.-C.; Goodman, D. W. J. Phys. Chem. 1994, 98, 9874.(78) Poppa, H. Catal. Rev.-Sci. Eng. 1993, 35, 359.(79) Poppa, H. Vacuum 1984, 34, 1081.(80) Poppa, H. Ultramicroscopy 1983, 11, 105.


30/30

(81) Tikhov, M.; Stolzenberg, M.; Bauer, W. Phys. Rev. B 1987, 36,(82) gzlilbhn, A. M.; Niemantsverdriet, J. W. Surfl Sci. 1990, 233,(83) E. H.; Poppa, H.; Pound, G. M. Thin Solid Films 1976, 32,(s4) P7322, H. J. Appl. Phys. 1961, es, ssss.CR940027Z

http://www.ocrconvert.com/download.php
http://www.ocrconvert.com/download.phphttp://www.ocrconvert.com/download.phphttp://www.ocrconvert.com/download.php

Convert Katalis

Documents

Transcript of Convert Katalis