Catalytically-Active Oxides

stuclied by

Low-Energy Ion Scattering

PROEFSCHRIFT

TER VERKRIJGING VAN DE GRAAD DOCTOR AAN DE TECHNISCHE

UNIVERSITEIT EINDHOVEN, OP GEZAG VAN DE REcTOR MAGNIFICUS,

PROF.DR. J.H. VAN LINT, VOOR EEN COMMISSIE AANGEWEZEN DOOR

HET COLLEGE VAN DEKANEN IN HET OPENBAAR TE VERDEDIGEN OP

DINSDAG 26 SEPTEMBER 1995 OM 16.00 UUR

DOOR

JEAN-PAUL JACOBS

GEBOREN TE HEERLEN

Dit proefschrift is goedgekeurd door de promotoren:

prof.dr. H.H. Brongersma

en

prof.dr. V. Ponec

CIP-DATA KONINKLIJKE BIBLIOTHEEK, DEN HAAG

Jacobs, Jean-Paul

Catalytically-active oxides studied by low-energy ion

scattering I Jean-Paul Jacobs. -Eindhoven: Technische

Universiteit Eindhoven

Thesis Technische Universiteit Eindhoven. - with ref. -

With summary in Dutch.

ISBN 90-386-0246-4

Subject headings: catalysis I oxides I low-energy ion

scattering.

Wel, mijn werk, ik riskeer er mijn leven voor en mijn

verstand is er half aan ten onder gegaan ..... .

Vincent van Gogh Auvers-sur-Oise, 24 juli 1890

Voor mijn ouders Voor Gabriela

The work described in this thesis has been carried out at the Schuit Institute of

Catalysis, Faculty of Physics, Eindhoven University of Technology, The Netherlands.

The investigations were supported in part by the Netherlands' Foundation of Chemical

Research (SON) with fmancial aid from the Netherlands' Organization for the

Advancement of Scientific Research (NWO).

1 General Introduetion

1. OXIDE SURFACES IN SURFACE SCIENCE

2. OXIDE SURFACES IN CATALYSIS

3. ATOMIC LAYER EPITAXY

4. SCOPE OF THIS THESIS

References

2 Applications of Low-Energy Ion Scattering to Oxide Surfaces

1. INTRODUCTION 1.1 An introduetion to LEIS on oxides

1.1a Surface sensitivity

1.1b Quantification; rough surfaces

1. 1 c lsotopes

1.1 d Insuiators

1.1e High temperature

1.2 Principle of LEIS

1.3 Applications of LEIS 2. EXPERIMENTAL DETAILS

2.1 Introduetion 2.2 E1ectrostatic energy analyzers in LEIS

2.3 Experimental factors complicating the analysis

2.3a Charging effects

2.3b Target preparation

2.3c Other experimental problems

2.4 Interpretation of energy spectra 2.5 Choice of experimental conditions

2.6 Fitting of LEIS spectra

2. 7 Change of surface composition by the analysis

2.8 Compositional depth profiling

3. SURFACE STRUCTURE OF NON-SINGLE CRYSTALS

3.1 Introduetion

3.2 Surface structure from information depth

3.3 Signalas function of toading

3.4 The energy method

4. GROWTH AND WETTING

Contents

page 1

2 3 5 9

10

13

14

14

16 16

16

16

16

17

21

22 22

22 24 24 28

28

28 31

34

34

35

37 37 37

41

43

43

Contents

4.1 Oxides on oxides 43 4.2 Oxides on metals and metals on oxides 47

References 49

3 Quantification of Surface Composition using Low-Energy 57 Ion Scattering

1. INTRODUeTION 58 2. EXPERIMENTAL 60 3. RESULTS AND DISCUSSION 63

3.1 Introduetion 63 3.2 Alloys 63 3.3 Oxides 67 3.4 Metals: standards and characteristic veloeities 69 3.5 Surface roughness 71

3.6 Surface composition of powders 73 4. CONCLUSIONS 75 References 76

4 The Surface of Catalytically-Active Spinels 79

1. INTRODUCTION 80 2. EXPERIMENTAL 84

2.1 Catalyst preparation 84 2.2 Catalyst characteri2'Ation 85

3. RESULTS 86 3.1 Introduetion 86 3.2 Simple spinel surfaces 88 3.3 Znt.xMnxA1204 spinels 89 3.4 Cot+xAl2-x04 spinels 91

4. DISCUSSION 92 5. CONCLUSIONS 95

References 96

5 The Growth Mechanism of Nickel Oxide in tbe Preparation 97 of N'J/ All03 Catalysts studied by LEIS, XPS and Catalytic Activity

1. INTRODUCTION 98

Contents

2. EXPERIMENTAL 99 2.1 Catalyst preparati.on 99 2.2 Catalyst characterization 99

3. RESULTS AND DISCUSSION 101 3.1 The nickel growth mechanism in the preparation 101

3.1a Activity durlog the hydrogenation of toluene 101 3.1b Surface characterization 101 3.1c Model for the growth mechanism 103

3.2 Effect of alumina pretTeatment on activity and surface composition 106 4. CONCLUSIONS 107 References 108

6 Surface Analysis and Activity of Chromia/alumina Catalysts 109 prepared by Atomie Layer Epitaxy

1. INTRODUCTION 110 2. EXPERIMENTAL 111

2.1 Catalyst preparation 111 2.2 Catalyst characterization 111 2.3 Activity measurements 112

3. RESULTS 113 3.1 Catalyst preparation 114 3.2 Surface characterization 117 3.3 Activity in dehydrogenation of i-butane 119

4. DISCUSSION 119 4.1 Dispersion of chromia/alumina catalysts prepared by ALE 119 4.2 Oxidation state of Cr in chromia/alumina catalysts prepared by ALE 121 4.3 Catalytic activity of chromialalumina catalysts prepared by ALE 123

5. CONCLUSIONS 124 References 125

Acronyms 127 Samenvatting 128 Summary 130 Publications 132 Epilogoe/Nawoord 135 Curriculum Vitae 136

General Introduetion

1

Chopter 1

1. OXIDE SURFACES IN SURFACE SCIENCE

Since mankind's earllest awareness of bis environment the im}X)rtance of the surface was recognized. However, it was only when ultra-high vacuum conditions could be routinely acbieved, that it became JX)ssible to study the surface layers of solids on an atomie level. At the same time surface analytical techniques were being developed wbich made it JX)SSible to enter a new field of physics and cbemistry: surface science [1 ,2]. Or as one of the students in our group once put it: "the sixties gave us the Beatles, the Rolling Stones, the sexual revolution, ultra high vacuum and with it, surface science". Tbe fact that the number of papers and dedicated scientific joumals is still increasing enormously, shows that surface science bas lost little of its challenging and innovative ebameter in tbe interveDing yea.rs.

Tbe majority of UHV studies was, and is, performed on metal and semiconductor systems, which are relatively easy to clean and to examine by surface science techniques (smooth, flat, conductive). Tbe single crystal surfaces of oxides, on the contrary, still constitute a relatively unexplored area in modern surface science [3], because of tbe unavailability of well-defined systems, the difficulty of surface preparation (e.g. high annealing temperatures, in-situ cleaving) and problems related to the surface analysis techniques (e.g. charging).

Tbe fundamental study of well-cbaracterized metal-oxide single crystals can provide im}X)rtant and very detailed information about the oxide surface, such as the surface electrooie and atomie structure, surface lattice vibrations and surface com}X)sition. Also tbe influence on these properties of the adsorption of molecules from the gas pbase (NO, CO, NH3, C~, etc.) or of the deposition of metallayers or particles are of great interest.

Although our knowledge of the behavior of oxide surfaces on an atomie level remains incomplete, tbis does not mean tbat the application of surface analysis to oxides is very limited. Indeed surface analysis techniques can still provide valuable information despite a lack of detailed knowledge of the system being studied, and have found widespread use since the earliest days of surface science. Of course it is still essential that the surfaces can be prepared reproducibly. Although the information is not as detailed as can be obtained on the model single-crystal systems, the study of JX)lycrystalline oxides, "real world systems", have revealed im}X)rtant information about the chemistry and physics tbat determine parameters like catalytic activity (or JX)Îsoning of the catalytic activity), adhesion, electron emission. These microscopie studies can often be directly related to macroscopie behavior in practical situations, e.g. performance of catalysts,

2

Genera/Introduetion

coatings, electron multipliers, etc. Major effort bas been expended to bridge the gap between the conditions under a

surface science study (UHV, single crystals) and "the real world". Latest trends in the

field of catalysis attempt to close thematenals gap by simulating "real world" catalysts by the use of thin oxide layers on flat (single crystal) surfaces either by evaporation and oxidation [4,5] or by wet-chemical methods [6,7]. Employing such model systems also allows more use to be made of the different surface analytical techniques. Although these model systems provide many details of e.g. the oxidation state of the active species and partiele size effects, one should not lose sight of the fact that for practical

applications it is the "real world" surfaces that we are really interested in. Therefore this thesis concentrates on "real world" systems. It will be shown by surface analysis using

LEIS that, even for highly porous polycrystalline materials used in catalysis, a detailed description of the surfaces is possible.

2. OXIDE SURFACES IN CATALYSIS

Heterogeneons catalysis bas a major impact on the chemical and petrochemical

industry and on environmental controL Since in heterogeneons catalysis the gas or Iiquid

phase reactantslproducts interact with the solid surface of the catalysts, it is clear that a detailed characterization of the surface is extremely important, if not essential, to describe

the catalytic process (6,8,9]. Heterogeneons catalysts can be found in all shapes and forms. Metal grids,

metal/alloy particles or metal oxides supported by other porous oxides, zeolites, sulfides, polymers, etc. An important and interesting part of heterogeneons catalysis deals with oxides. Oxides can be applied to function as catalysts (e.g. Mn30 4, FeSb-oxides, BiMo­oxides), as oxide promotors (NaO, ThO:z), as solid acids (zeolites), or as structural

promotors by stabilization of the catalytic active material on porous supports (e.g. V20sfTi0a, Cr()JAI20 3) (10,11].

In catalytic oxidation by oxides the reaction mechanism can, for convenience, be divided into two regimes. The first, extrafacial reactions, deals with reactions via adsorbed radicals (Langmuir-Hinshelwood) or reactions of free (desorbing) radicals in combination with surface reactions. The second, intrafacial reactions, involves the incorporation of lattice oxygen into the product in a single or two step process, the latter being evident from Fig. 1 (Mars van Krevelen). However, the low temperature and the medium temperature oxidation reactions can also be subdivided into two alternative regimes

3

Chapter 1

Extrafacial reactions (Langmuir-Hinshelwood)

co i ~ - --

Intrafacial reactions (Mars-van Krevelen)

co

t i

- ioi i 1 i !ei

--

Fig. 1 Schemotic representation of the oxidation reacrions on catalytical/y-active oxides via interfacial or extrafacial reactions, respectively.

according to the electrophilicity of the active species: electrophilic oxidation, with ~. 0 2-

or o-, and nucleophilic oxidation which involves lattice oxygen [12, 13]. Electrophilic oxidation proceeds through the activation of oxygen on adsorption on

the surface of an oxide catalyst. These reactions often proceed as total oxidation. In a model initially proposed by Mars and Van Krevelen [14] for the oxidation of naphtahlene catalyzed by V 20 5 and since validated for a large number of different systems, nucleophilic reactions introduce the oxygen into the hydracarbon reactant

4

General Introduetion

through a mechanism which donates an oxygen anion (QZ-) extracted from the lattice of the oxide catalysts. Nucleophilic reactions can provide high selectivity and are applied in the synthesis of many different products in industry. Well-established examples [15] are Bi-Mo oxides, Fe-Sb oxides and U-Sb oxides in the catalytic propene oxidation to acrolein (C3~ + <>z..., H2C=CH-CHO + H20) or in the preparation of acrylonitrile by catalytic ammoxidation of propene (~ + NH3 + 3/2 0 2 ..., CH2CHCN + 3 H20). The vacancies created in the oxidation reaction are replenisbed by molecular oxygen from the gas phase. Ponec and coworkers [16] showed that the vacancies can also be used to perfarm selective catalytic reduction on spinel-structured oxides. This will be discussed in more detail in chapter 4 of this thesis.

Transition metal oxides on oxide carriers (supports) or in mixtures with other oxides are widely used as heterogeneous catalysts in many different industrial processes. It was found that the surface structure of catalytically active oxides such as chromia and vanadia, dispersed on supports such as alumina, titania and silica, may differ considerably from that of the bulk phase. The nature of the support and the preparation metbod used play a major role in the resulting catalytic behavior due to the formation of particular active surface phases [10]. For example, vanadia supported by alumina shows a strong interaction between the active vanadia phase and the alumina support, resulting in a monolayer catalyst [17], while for silica supported vanadia only a weak interaction exists, causing the catalyst to behave like bulk V20 5• Titania and zirconia exhibit a medium strength of interaction [18]. All systems show a different catalytic behavior and the most suitable system is chosen depending on the application. When studying the literature the following trend can be distinguished for the strength of the interaction between the catalytic active oxide phase and its support: 'Y-alumina > titania, zirconia > silica. Unravelling the specitïc surface structure of the active phase is important to understand the catalytic behavior. In this thesis the surface structures of Ni-oxide and Cr­oxide on 'Y-alumina have been studied using the new preparation technique of atomie layer epitaxy.

3. ATOMIC LAYER EPITAXY

In general, supported catalysts are prepared by conventional methods from salt solutions such as wet impregnation, coprecipitation and ion exchange, followed by drying, oxidation and reduction cycles, to create the catalytically active phase at the surface. Higher demands for the catalyst performance create the need for manufacturing methods,

5

Fig. 2

Chapter 1

0._.::" CI-Zn-CI -'J:!!tl- 'VJ,(}

B -~ ~~p!!i\1* 'l&ir ::!J

D

E

Zn Zn Zn Zn Zn Zn Zn Zn 1\1\1\ I\ 1\1\1\1\

Cl CIC! CIC! CIC! CIC! CIC! CIQ CIC! Cl

N2 -'J:!!tl-

rw~··•l Zn SZnS Zn S ZnS Zn SZn SZnSZn S Zn S

Zn Zn Zn 1\1\ 1\

Cl CIC! Cl Cl Cl

CI-Zn-Cl __..:;>-..;:.. '?,C)

j ZnC12+ H:§ ~zns + 2 HCl I The ALE process illustrated by the growth ofZnS layers (see text).

6

General Introduetion

where the active metal oxide species can be bonded to the surface in a more controlled

manner. In this way the structure and electronic properties of the catalytic active species

can be tailored for specific uses. A number of alternatives have been developed to meet

these challenges [19,20]. Besides the use of new materials such as zeolites, one of

the promising new preparation techniques is ALE [21 ,22].

ALE has been developed as a controlled layer-by-layer deposition metbod for the production of thin films and the epitaxial growth of single crystals. ALE has been applied

with great success to the manufacturing of lli-V and ll-VI semiconductors and to thin

oxide and nitride films. In ALE deposition occurs from the vapor phase. However, in

contrast to conventional CVD methods, the deposition is not based on the contact time,

dosing of the reactive gases or on decomposition of the precursor, but on chemisorption

and saturating surface reaelions which determine the deposition process.

For a more extensive introduetion into ALE the reader is referred to the review

papers of Lakomaa [21] and Suntola [23,24]. Briefly, the ALE technique, as applied to the growth of thin ZnS films, is illustrated in Fig.2. In step A, ZnC12 is chemisorbed

on the surface of the support, e.g. silicon, until one monolayer is deposited. The surplus

of ZnC12 is not adsorbed by the monolayer dosed surface and is removed in the next step

B from the reaction chamber with a purge of inert gas. In step C, the surface is exposed

to H2S. At the surface ZnS is formed while HCl is released until the full monolayer is

transformed. The reaction is again self terminating. HCl and surplus HzS are removed by yet another purge of inert gas (step D). The ALE-sequence can be continuedas shown in

step E, and the whole process can be repeated to form a multilayered structure. In

conclusion, the whole process relles on the fact that the chemisorption is self terminating

after each monolayer coverage has been formed. This allows for a stepwise epitaxial

growth of the layers with a digital accuracy.

The ALE process can extend a bomogeneaus deposition over a large area. Since

the pressure is not crucial, once the precursor can be vaporized, less restraints are put on

the necessary equipment compared to e.g. MBE or MOCVD techniques. Furthermore

complex shapes and sharp edges can be covered uniformly since self-terminating, at

saturated monolayers, gas-salid reaelions and not the contact time determine the

deposition. The ALE process has been in industrial use over 10 years (since early 1980's)

for the manufacturing of display panels. Equipments for ALE processes have been

commercialized by Microchemistry Ltd. (Espoo, Finland).

Recently the ALE technique has been applied to catalyst manufacturing [21]. Here

catalytically active metal oxide species are deposited on a highly porous support from the

vapor phase. The chemisorption of the species is mostly determined by the surface reaction

7

Fig. 3

Cr(acac)3

Me Me / \

__.c /c'cH HC :"·.'o 0 ,---., I', :I c:· \ I :.C,

M~ 0-Çr-0 Me OH 0 HO I I \

Oz---

Chapter 1

Me Me I I

__.C /C, H(\(··,'o 0 ,·-·: fH

c·~- ' I ::C.... 1 0-Çr-0 Me

Me 0 OH OH A I I I ' Si

0 OH 0 0 '-/ol'on o O"'"cr =ei 'er""

I I I 0 0 0 I I

1he ALE process illustrated by the preparation of CrO)Si02 polymerization catalysts using Cr(acac)3 [25}.

with the surface OH groups. As discussed above, the high level of surface control achieved by ALE could provide new possibilities for site specitic positioning of these active species, which conventional impregnation techniques cannot provide. Bither blocking, or removal by thermal treatment, of specific adsorption sites can be used to regulate both the nature of the bonding, and the quantity, of the active species on the surface. The separation distance between the deposited species can be controlled by the size of the ligand(s) on the chemisorbed species. For example, large ligands, such as

8

General Imroduction

acetylacetonate groups, can be used to guamntee a high dispersion. Upon removal of the ligands the catalytically active species are formed. When requiered, the metal toading can

be increased by performing subsequent ALE sequences, and therefore the dispersion can

be carefully controlled. The ALE process for the preparation of CrOJSiÜ:l polymerization

catalysts is schematically shown in Fig. 3 [25].

Difficulties related to inhomogeneous deposition of the metal-oxide species, such

as condensation and accessibility of the adsorption sites in the pores, decomposition of the

reactants on the support, have been effectively dealt with by determining the suitable

reactants and processing window [21]. Several catalysts with high selectivity and activity have been prepared as shown by Lakomaa and coworkers [26] and as reported in

chapters 5 and 6.

4. SCOPE OF TIIIS THESIS

It is the aim of this thesis to explore the possibilities of LEIS and further develop this technique for catalytic research, with the emphasis on "real world" systems. The potential of quantitative surface analysis by LEIS bas only recently been tapped for

catalytic research on "real world" systems [27]. In this thesis it will be shown that the

combination of LEIS and catalytic activity measurements constitute a powerfut

combination in catalytic research. Although the title of the thesis might suggest that only

LEIS has been applied to the analysis of the catalysts, this is certainly not the case. A number of other widely used analytical techniques, e.g. XPS, SEMIEDS, XRD, XRF and

AAS, have also been used for characterization of solids. Por an introduetion totheuse of surface analysis in catalytic research the reader is referred to the recent review by

Niemantsverdriet [6].

In chapter 2 the basic principles and characteristics of the LEIS technique and

experiment will be discussed and an overview of the application of LEIS to oxides will be given. In the following chapter the quantification of surface composition by LEIS will be

treated. Chapter 4 deals with catalytically active spinels. The combination of catalytic

activity, in the selective reduction of nitrobenzene to nitrosobenzene, and LEIS measurements provide a model of the oxide surface where the octahedral sites are almost

exclusively exposed at the surface. In chapter 5 the growth mechanism of Ni-oxide on 'Y­

alumina, during preparanon from the vapor phase using ALE, will be discussed. This will

be related to the catalytic activity in the hydrogenation of toluene to methylcyclohexane.

Surface analysis and the catalytic behavior of chromialalumina monolayer catalysts,

9

Chapter 1

prepared by ALE, for the dehydrogenation of i-butane to i-butene, will be the subject of chapter 6.

1he contents of the following chapters have been individually prepared for publication in international scientific joumals and can be reod independently. Accordingly, some overlap between chapters is unavoidable.

References

[1] C.B. Duke (Ed.), Surface Science: the.jirst thirty years, Surf. Sci. 299/300 (1994) 1. [2} A. Zangwill, Physics aJ Surfaces, Cambridge University Press, Cambridge, 1988. [3] V.E. Henrich and P.A. Cox, 1he Surface Science of Metal Oxides, Cambridge University Press,

Cambridge, 1994. [4) S. Ladas, H. Poppa and M. Boudart, Surf. Sci. 102 (1981) 151. [5] D.W. Goodman, Surf. Sci. 299/300 (1994) 837. [6) J.W. Niemantsverdriet, Spectroscopy in Catalysis: An lntroduction, VCH, New Yotk, 1993. [7] E.W. Kuipers, C. Laszlo and W. Wieldtaaijer, Catal. Lett. 17 (1993) 71. [8] H.H. Brongersma and R.A. van Santen (Eds.), Funtkunental Aspects of Heterogeneaus Catalysis

Stutlied by Partiele Beoms, NATO ASI series B 265, Plenum press, New York, 1991. [9] C.M. Friend, Scientific American (April 1993) 42. [10) H. Knözinger, in Fundamental Aspects of Heterogenous Catalysis Studied by Partiele Beoms, p. 7,

Eds. H.H. Brongersmaand R.A. van Santen, NATO ASI series B 265, Plenum press, New York, 1991.

[11] J.A. Moulijn, P.W.N.M. van Leeuwen, R.A. van Santen (Eds.), Catalysis: An lntegraJed Approach to Homogeoous, Heterogeneaus and lildustrial Catalysis, Elsevier, Amsterdam, 1993.

[12] J. Haber, lecture notes, NIOK Post Graduate Course "Oxidation Catalysis", June 6-9, Rolduc, The Nether lands.

[13] B.E. Nieuwenhuys, V. Ponec, G. van Koten, P.W.N.M. van Leeuwen and R.A. van Santen, in CaJalysis: An lntegrated Approach ta Homogenous, Heterogeneaus and lildustrial Catalysis, p. 140, Eds. J.A. Moulijn, P.W.N.M.van Leeuwen, R.A. van Santen, Elsevier, Amsterdam, 1993.

[14] P. Mars and D.W. van Krevelen, Cbem. Eng. Sci. Suppl. 3 (1954) 41. [15] B.C. Gates, J.R. Katzee and G.C.A. Schuit, Chemistry of Catalytic Pracesses, McGraw-Hill

Chemica! Engineering Series, McGra.w-Hill Book Company, New-York, 1979. [16] T.L.F. Favre, PhD. Thesis, Leiden University, The Netherlands, 1991.

A. Maltha, PhD. Thesis, Leiden University, The Netherlands, 1994. S. Meijers, PhD. Thesis, Leiden University, The Netherlands, 1995.

[17] A.W. Stobbe-Kreemers, G.C. van Leerdam, J.-P. Jacobs, H.H. Brongersma and J.J.F. Scholten, J. Catal. 152 (1995) 130.

[18) M. Niwa, S. Inagaki and Y. Murabami, J. Phys. Chem. 89 (1985) 3869. [19] F.H. Ribeiro and G.A. Sommjai, Recueil des Travaux Chimiques des Pays-Bas 133 (1994) 419. [20] H.C. Foley, Catal. Today 22 (1994) 235. [21] E.-L. Lakomaa, Appl. Surf. Sci. 75 (1994) 185. [22] S. Haukka, A. Kytökivi, E.-L. Lakomaa, U. Lehtovirta, M. Lindblad, V. Lujala and T. Suntola,

in Preparation of Catalysts VI. Scientific Basis for the Heterogeneaus Catalysts, p. 957, Ed. G. Poncelet, Elsevier Science B.V., Amsterdam, 1995.

[23] T. Suntola, Mater. Sci. Rep. 4 (1989) 216.

10

General Introduetion

[24] T. Suntola, in Handhook ofCrystal Growth 3, p. 601, Ed. D.T.J. Hurle, Elsevier Science B.V., Amsterdam. 1994.

[25] S. Haukka, PhD. Thesis, University of Helsinki, Finland, 1993. [26] E.·L. Lakomaa, M. Undblad and T. Suntola, Proc. 2nd Tokyo Conf. on Advanced Catalytic

Science and Technology, Tokyo, Japan, August 21-26, 1995, in press. H • .Knuuttilaand E.-L. Lakomaa, US Pat. 5,290,748 (1994). M. Undblad, L.P. Lindfurs and T. Suntola, Catal. Lett. 27 (1994) 323.

[27] G.C. van Leerdam, PhD. Thesis, Eindhoven University of Technology, Tbe Netherlands, 1991.

11

12

Abstract

Applications of Low-Energy Ion Scattering to Oxide Surfaces*

The possibilities of low-energy ion scattering spectroscopy for surface analysis are

discussed. Low-energy ion scattering is used for the determination of the atomie

composition and structure of oxide surfaces. lts extreme surface sensitivity enables the selective analysis of the outermost atomie layer. It is this layer that is largely responsible

for many ehemical and physical properties of oxides. The LEIS technique will be introduced and experimental details will be discussed. PUTthermore this ehapter will give

an overview of the applications of the technique to oxides.

"The contentsof this chapter bas previously appeared in H.H. Brongersma, P.A.C. Groenen and J.-P. Jacobs, in Science ofCeramic Interfaces 11, p.l13-182, Ed. J. Nowotny, Material Science Monograplis 81, Elsevier Science B.V., Amsterdam, 1994.

13

Chapter 2

1. INTRODUCTION

In this chapter the reader is introduced to the surface analysis technique of low­energy ion scattering (LEIS). Briefly its characteristics and the experimental aspects will be discussed, where the eropbasis is put on the equipment and procedures used in

Eindhoven. PUTthermore the review is restricted to non- and poly-crystalline materials.

For a more complete description of LEIS on oxidic surfaces and LEIS in general the

reader is referred to Ref. [1] and Ref [2], respectively.

1.1 An introduetion to LEIS on oxides

Elastic binary collisions of inert gas ions with energy of 500 eV to 5 k:eV with

surface atoms provide energy spectra that are characteristic of the distribution of the

masses of the surface atoms. AJready in the early sixties, Panin [3] and Walther and Hintenberger [4] had demonstrated that for inert gas ions a clear correlation exists

between the energy loss of a scattered ion and the identity of the surface atoms, and the work of Smith [5,6] gave a strong impulse to the use of low-energy ion scattering in surface science. While the energy spectra of the scattered ions are mainly used for compositional analysis, shadowing and blocking effects enable one to use angular

dependent studies to determine the location of surface atoms on the surface of single

crystals [2,7]. For general applications the Low-Energy Ion Scattering technique is called LEIS

or ISS (Ion Scattering Spectroscopy). Here, LEIS is preferred since the emphasis is on

the low energies and the acronym ISS does not indicate that distinction. For more specific

applications, especially for various types of LEIS in structural analysis of surfaces, many other abbreviations exist (ICISS [8], NICISS [9], NBISS [10], LENS [11], TOF [12], TOP-SARS [13]).

The most important feature of LEIS is its extreme surface sensitivity when inert

gas ions are used. This feature has its origin in the relatively high neutralization

probability and scattering cross sections for these ions at energies of at most a few k:eV.

The high neutralization probability ensures that almost all ions that penetrate beyond the

fust atomie layer are neutralized. When energy analyzers are used that only detect ionized particles, scattered partiele contributions from deeper layers are effectively eliminated.

In contrast to other commonly used surface analytical methods such as Auger Electron Spectroscopy (ABS), X-ray Photoelectron Spectroscopy (XPS) or Electron

Spectroscopy for Chemical Analysis (ESCA), and Secondary Ion Mass Speetrometry

14

Applications of Low-Energy Ion Scattering to Oxide Surfaces

(SIMS) that probe several atomie layers, it was shown by Brongersma and Mul [14, 15] that LEIS can be used to selectively study the outermost atomie layer of a surface. It is this atomie layer that plays an important role in many applications of oxides. Typical examples of such applications are:

catalysts

- identification of active sites or of the species causing poisoning - spreading of the active phase over the support - quantification of the composition of the active phase

ceramics - surface active species promoting sintering - oxygen transport in fuel cells - oxygen sensors

catbodes - determination of origin of low work:functions

adhesion - promotion or prevention

corrosion - proteetion of metals

Segregation of impurities or dopants may dominate surface properties, especially with single erystals and sintered ceramics, which have a low specific surface area. The driving force for segregation of alkalis is often so strong that after moderate heating a large fraction of the surface is covered by alkalis (even if the bulk concentration is only in the ppm or sub-ppm range) and thus the surface properties are altered significantly. Since thermodynamics often restricts the enrichment to the outermost atomie layer, LEIS is the

ideal technique to detect and quantify segregation. In most applications, the main reason for using LEIS is its monolayer sensitivity.

In general, the ability of LEIS to perform compositional analysis is related to the follo­wing features:

- selectivity for lst atomie layer - straightforward quantification - very rough surfaces can be studied - detection of isotopes. - samples can be conductors as well as insuiators - in-situ studies possible at high temperatures

Some of these features are particularly advantageous for studies of ceramics and catalysts

15

Chopter 2

and will be discussed in more detail below.

l.la Surface sensilivity In the early days of LEIS a full advantage of its surface sensitivity was not always taken,

because the ion beams were often so intense that the first monolayer was destroyed before

the analysis was completed. Nowadays, the fluence required for a composition analysis

can be as low as 3 ·1()1° ions/cm2 [16], thus allowing many measurements before the

damage beoomes detectable (so-called Static LEIS). This also allows one to follow

processes such as surface segregation in real time.

l.lb Quantification; rough surfaces Quantification of the atomie composition by calibration against standards is now well

established. This means that, in genera!, the sensitivity to a given element is not affected

by the identity of the neighboring atoms {"no matrix effects") as long as the element is

present in the outermost atomie layer. Some exceptions to this rule exist, but these can be

avoided by proper choice of the ion and energy. The quantification can even be carried

out for very rough surfaces, such as of supported catalysts. Por a more detailed

discussion the reader is referred to chapter 3 of this thesis.

l.lc lsotopes The scattered ion energy depends on the mass of the target atom. Thus, at least in

principle, it is possible to distinguish the isotopes of elements. This property is used for

studies of oxidation mechanisms [17] of metals, quantification of oxygen diffusion in

ceramics for sensor applications [18] and to distinguish oxygen surface sites via

selective oxygen exchange [19].

l.ld Insuiators The relative ease of producing electroos of a few eV and the fact that, in contrast to most

SIMS experiments, no draw-out fields are being used, makes it simple to compensate for

charging effects of the incoming ion beam. LEIS is also relatively insensitive to surface

charging of a few volts. This enables one to investigate insuiators as well as conductors.

l.le High tempemture Since high temperatures of the sample do not affect the analysis, it is possible to follow

in-situ, for instance, surface segregation in ceramics. A restrietion is, of course, that the

vapor pressure of the material under investigation is low enough to allow for undisturbed

16

Applications of Low-Energy Ion Scattering to Oxide Sulfaces

passage of the ion beam (pressure lower tban 1()4 mbar) and, more importantly, to prevent significant contamination of the equipment.

1.2 Principle of LEIS

When inert gas ions are used for the analysis, neutralization is very high. Thus, the probability that these particles are still in an ionized state after interaction with more tban one target atom is strongly reduced. If ionized particles alone are detected, as is

often the case, singly-scattered ions are strongly favored over doubly- and multiply­scattered ions. For the singly-scattered particles, a simpte two-body ooilision model is sufficient to describe the interaction of an ion with a solid target.

If one considers an ion with incident energy & and mass Mw.. which is scattered through an angle e by a target atom of mass Mat. the final energy & of the ion will be

E. (cose±Jr2-sin

29 )2 = E .• k

' 1+r ' (1.1)

where the positive sign applies to r ~ 1 and the + and - are both solutions if 1 ~ r ~

I sine I· The mass ratio r = Mat I Mro.,, and k is the so-called kinematic factor. The equation is simply obtained by solving the equations for conservalion of energy and momenturn upon elastic ooilision with the surface.

During a given experiment, the energy spectrum of the scattered ions at a particular scattering angle is recorded for a fixed energy of the incident ions. The energies of the scattered ions are then characteristic of the masses of the target atoms. Fig. 1 illustrates this for 136° elastic scattering of He+ ions by an 180-exchanged Sr-Sm­Co-oxide [18]. The peaks for Sr, Sm, Co, 0 18 and 0 16 are clearly observable.

Often ions can also be observed with a most probable energy near 0 eV. These particles are ions sputtered from the target. Since light ions, such as H+, contribute to this peak, an intense peak at low energies is generally a strong indication of a contami­nated (water, hydrocarbons) surface (e.g. Fig. 5). Since the angles of the incoming and

outgoing ion with the target surface (a; and af, respectively) do notenter into Bq. 1.1, the

inclination of the target surface with respect to the ion beam does not influence the energy of a binary ooilision peak. This explains why LEIS can still be used for very rough surfaces. At very low angles (00 - 200) of the incoming or outgoing beam, shadowing and blocking will, however, influence the intensities of the peaks. This dependency is used in

structural analysis. In a number of LEIS studies, scattering of non-inert gas ions such as H+, u+ and

17

Chapter 2

1500 Sm

-.!!2 1000 Sr 0 -ç '(i)

180 c: Co Q) -.E ~

x4 Cf)

500 160

w .....J \

0 0 500 1000 1500

Final energy (eV)

Fig. 1. Energy spectrum of 1500 eV 4He+ scattered from 160PO exchanged Sm0.sSro.2Co03

[18].

Na+, bas been used. Since neutralization no longer limits the information depth, a very wide, featureless spectrum of scattered ion energies is obtained. In such cases it is important to restriet the studies to single crystals and to use shadowing and blocking directions to further restriet the scattered ions. Although such studies can be quite successful in special cases, very few applications to oxides or oxidation [20,21] are known. The present discussion is, therefore, restricted to the use of inert-gas ions.

For compositional analysis of the surface by LEIS, a large scattering angle (14<1'-180~ is favored to facilitate good mass resolution, although the scattering cross sectionis at its lowest here. Both the incoming (often perpendicular to the surface) and outgoing trajectory of the ions make large angles (preferably at least 400) with the surface of the sample (see Fig. 2a) to reduce additional neutralization, shadowing and blocking effects

18

Applications of Low-Energy Ion Scattering to Oxide Surfaces

Fig. 2 Illustration of incoming and outgoing ion beams for a) composition analysis b) structure analysis (IOF-SARS) c) structure analysis (IC/SS).

that complicate a quantitative analysis. In back:scattering, all atoms having masses greater than the lightest inert gas ion eHe+) can be detected (Eq. 1.1). Thus, apart from

hydrogen, it should be possible to detect all elements of the periodic table. Although this

is essentially true, the signals for the elements with a mass close to that of the prohing ion

(Li, B, Be, C and N) are quite low. The diffieulties of detecting the light elements

present a potential danger to the quantification of the surface composition. Although these

elements cannot be detected, even atoms as small as hydrogen on the surface do neutralize or physically shield the incoming or outgoing ions and thus prevent the detection of the underlying atoms.

At low angles with the surface, the local atomie structure dominates the scattering

process (Fig. 2b). To simplify the interpretation, one of the beams (generally the

incoming beam) is often at larger angles with the surface, while the other (outgoing)

beam is glaneing. Another attractive choice is to use 180° scattering (Impact Collision Ion

Scattering Spectroscopy, ICISS) so that the incoming and outgoing beam coincide (see

Fig. 2e for an illustration). The angular distribution (azimuthal and/or polar) gives direct

information on the atomie structure, defects, etc. of a surface. For the light elements the detection as a recoil partiele is generally mueh more

sensitive than in a (back-) scattering geometry. This is especially true for hydrogen

whieh, since it is lighter than any inert gas ion, cannot even be observed in

baekscattering. Rabalais and coworkers [13] have developed a technique to perform surface and absorbate structural analysis using time-oHlight scattering and recoiling

speetrometry (I'OF-SARS).

19

Table l.la

Chapter 2

22, 24, 25, 51, 55, 57, 62, 63, 115, 116, 101, 143, 181, 186, 188, 190, 229, 256, 260, 273

105, 106, 113, 136, 149, 159, 181, 190, 191, 202, 205, 206, 212,218

22, 156, 169, 173, 174, 205, 209, 211, 215, 216, 237, 243, 272, 281, 284

178, 19, 56, 63, 96, 97, 145, 194 99, 100, 144, 146,

145, 190, 278

26, 27, 28, 92, 93, 189, 100, 104, 105, 106, 210 108, 111, 112, 113,

162, 174

178

145

195

147, 155, 39, 148, 263, 180, 65, 247, 149, 135

114, 190, 205, 212, 228, 234, 244, 245, 246,249,257,258, 259,261,265,269, 274

97 217, 238

192, 235, 272, 155, 282, 288

118, 119, 120, 142, 92, 107, 109, 206, 154, 183 220, 228, 246, 245, 238

261

25, 101, 128, 129, 131, 132, 133, 134, 142, 156, 162, 164, 171, 182

99, 158 234, 244 153, 162, 171, 173, 213, 231, 235, 279

23, 199, 201, 248, 91, 95, 270 154, 169, 209, 211, 254, 263, 276 177 237, 250, 281, 288

47, 52, 55, 56, 62, 91, 112, 187, 197 142 117, 142, 185, 247, 255, 275, 283, 290

147, 190 92, 102, 107, 190, 218, 227, 228, 234, 274

141

39, 48,49

120, 229

128, 149, 202, 154, 22, 150, 151, 152, 214, 234, 258, 259 238 153, 156, 169,

171, 173, 174, 182, 183, 192, 200, 207, 208, 211, 213, 231, 233, 235, 267, 268, 271, 277, 280, 285, 286

37, 92, 107, 179, 191, 212, 226, 228, 242

137, 138, 139, 140

A selection of papers on LEIS and oxidic suifaces.

20

Applications of Low-Energy Ion Scattering to Oxide Surfaces

dfffu~i@f .. · ... · ...•...•...• 33, 157' 163, 170, ~~,~~9it i 187, 236, 289

84lli>l'P1ifiii 67, 187, 219, 230 ~~w#?if ) .. , .... d~ptjl p~l)til# ····. 64, 80, 253, 254 .W4iliifmi ? ···

®#~tidîl> 62, 67,247

>i 196, 230 i LiifL •:< }·••\

? l ;~9 101, 120, 124,

:,:.:,, .. , .. ,:.,,,··········i·········'·' i Table l.lb

22,67,160

92, 96, 218, 240, 257

135, 232, 239, 258

162, 166, 167, 168, 173, 174, 233

67

204

62, 67, 287

92

168

25, 37, 74, 92, 106, 107, 108, 110, 120, 122

175,193, 195

9, 175, 184, 195, 193, 203

9, 36, 232,239

175, 184

193, 225

203

179

A selection of papers on LEIS and oxidic surfaces.

i9i i. i

180, 241

180

40, 180

40, 57, 118, 119, 207, 223

118, 207

180, 204, 223

57, 185

180

40, 180, 207

1.3 Applications of LEIS

~'""''

253

34,253 ' 264 110, 127

264

34,253

,.,ii! .. n•·.•i·•·:···:··········

105, 131, 165, 190, 198

19,105, 146, 190

41

208

208

190

131, 208

The specific advantages of LEIS for surface investigations of ceramic materials were recognized from the early beginning of the technique. Since the beginning of LEIS

some 300 papers have been published on its use, in various guises, for studies of oxides, or oxidation. In Table l.la and l.lb a selection of these papers is classified according to the material and aim of the studies.

Most studies concentrale on the composition of the first atomie layer. The main application has been for supported catalysts. There are only a few studies of the surface structure of monocrystals of oxides, although some of them are very detailed. Review

papers on the combination of LEIS and oxides have been given by Honig [23], McCune

21

Chapter 2

[24], Carver et al. [25], Horeli and Cocke [26], Brongersma et al. [1,27] and

by Taglauer [28].

In most studies discussed in the present review, various types of LEIS are

combined with other surface techniques. Such combinations are extremely important since these techniques often provide essential complementary information. Por instance, the

larger information depth of XPS, AES and SIMS provide in combination with LEIS,

information on the first and the deeper atomie layers. XPS and SIMS also provide

chemica! information. Low-Energy Electron Difftaction (LEED) is essential for studies of

single crystal surfaces to establish the occurrence of long range periooicity and the dimen­

sions of the unit cell. On the other hand, LEIS provides, in a relatively easy and straight

forward way, the short range order within the unit cell.

2. EXPERIMENT AL DETAilS

2.1 Introduetion

An analysis of the composition of a surface is based on the application of Eq. 1.1.

For a well-defined ion beam (&, M;mJ the mass resolution would thus be fully determined

by the precision with which the scattering angle e and the energy of the scattered ions & are determined. In reality this is not completely true since ine1astic processes (giving an

extra energy loss) and vibration of the surface atoms (giving a Doppier type of

broadening in the energy [29]) will also influence the energy of the scattered ions.

However, for most experimental conditions the inaccuracy in e and the energy resolution

dominate the results.

At present, most analyses of oxides by LEIS use compositional rather than structural analysis. The emphasis in section 2.2 will, therefore, also be on experimental

arrangements for studies of composition. Also, most materials that are used in applications do nothave a structure with long range order.

2.2 Electrostatic energy analyzers in LEIS

In an electrostatic energy analyser the charged particles are deflected in an

electrostatic field. The deflection depends on the energy of the ions which can thus be

determined. In the early years of LEIS, 1211 analyzers [30,31] 180° Hemi-Spherical

Analyzers (HSA or CHA) [32] and 9QO electrostatic analyzers [33] were used.

22

Applications of Low-Energy Jon Scattering to Oxide Surfaces

In modem times, cylindrical (Cylindrical Mirror Analyzer, CMA) or a hemi­

spherical electrostatle analyzer with improved transmission have been used for studies of

oxides (CMA: e.g. Refs. [34,35,36,37] HSA: e.g. Refs. [38,39.40,41]). The CMAhas

the advantage that when the ion beam is coaxial, the scattering angle is the same for the full 3(/;j of the azimuth. For composition analysis, for which no azimuthal information is

required, one can thus use the full 3(/;j azimuth of the analyzer. This greatly increases

the detection efficiency. In commercial systems the ion souree is incorporated in the inner

cylinder of the analyzer. In the fust CMA for LEIS analysis, developed by Brongersma

et al. [42], a somewhat different design with a ring detector was used. This enables one

to separate the ion souree from the analyzer and thus to differentially pump and rnass­filter the primary beam. The exclusion of ions such as H+ and c+ from the primary ion

beam results in lower background in the spectra (such ions could still leave the surface as an ion after many collisions in deeper layers [27]). An improved version of this set-up

("NODUS"), see Fig. 3, is still used for most of LEIS studies of oxides in Eindhoven.

einzellens

deflection plates

Fig. 3 Diagram of the CMA of the NODUS LEIS-machine.

23

Chapter 2

A new development by Brongersma and coworkers in the area of surface analysis of oxides is the LEIS-apparatus ERISS (Energy Resolved Ion Scattering Spectrometer). In Fig. 4 a cross-section of the ERISS is shown. The analyzer is rotationally symmetrie

around the primary ion beam. The scattered ions are energy separated in a double toroidal

electrostatic analyzer. The contiguration is such that a linear image in energy is obtained.

A special 2-dimensional detector allows the simultaneons detection of 100 energy

channels. Because of this simultaneons detection and some other improvements in the

design, the sensitivity of this instrument is about three orders of magnitude higher than that of conventional electrostatic analyzers. This allows one to use ion currents as low as 10- 100 pA, thus reducing damage of the target surface by the ion to negligible amounts

("statie" LEIS). ERISS is analogous to the earlier developed EARISS [43,44,45]

but bas been specially modified for composition analysis of insuiator surfaces. Amongst

other features, a special charge neutralizer compensates for surface charging (see section 2.3a).

2.3 Experimental factors complicating the analysis

A brief discussion is given of the main experimental factors that may influence energy spectra and thus their interpretation. Some are related to the set-ups, others to the

target preparation and the poor electrical conductance of most oxides.

2.3a Charging eflects. The primary ion beam will positively charge the surface of insulators. For good

insulators, when no charge compensation is performed, the charging continnes until no further ion cao reach the surface. For perpendicular incidence of the ions this means that

the surface potenrial becomes equal to the accelerating potenrial applied to the ions.

Subsequent ions are reflected by the electrostatic field. Such reflected ions, which have

not reached the target and thus did not lose any energy during scattering, are sametimes

used for energy calibration of the analyzer or the ion source. When the charging is less severe, the scattered ion peaks will still be shifted to higher energies by the charging.

Spraying the target with low-energy electroos cao rednee the effects of the charging. This

is illustrated in Fig. 5 for a cabalt aluminate surface with and without proper charge neutralization. Surface charging oot only shifts the peaks in a LEIS spectrum, it may also enhance the mobility of atoms [46] in the target and thus change the surface

composition (especially for alkali ions).

Even with a charged surface, pea.k identification is still possible. In a first

24

Applications of Low-Energy Ion Scattering to Oxide Surfaces

ion souree

detector

channelplates ··j

analyzer

variabie slits ······················.:...._ .. , .. -

neutralyzer ·········

Fig. 4 Diagram of the analyzer and detector of the ER/SS.

approximation, charging to a potentlal vch will have decelerated the incoming ion to :& -eVch when it reaches the target. According to Eq. 1.1 the energy after scattering will be k

(Eo - eVc.J. Upon leaving the surface the ions are post-accelerated by the surface charge

to their final energy :

E/ = Er + (1-k) eVch (2.1)

25

Chapter 2

In agreement with Fig. 5 the effect of charging is thus largest for the lowest kinematic factor k (the lightest elements). For large scattering angles, such as encountered in a CMA with a coa.xia1 ion gun (see section 2.2), Bq. 2.1 is a good approximation. A more accurate analysis should be used, however, to take account of the precise experimental conditions in a more general situation [47].

Fig. 5

--:- 600 ::I

e ~

.€' 400 en c !! .5 en 200 LU ....J

0 0 1000 2000 3000

Final energy (eV)

LEIS spectra (3 keV He+) from a Co~/04 catalyst withand without proper charge compensation.

Surface charging will not only accelerate the scattered but also the sputtered ions. In the case of oxides, the sputtered ion signal is generally quite large. The most probable energy of these ions is just above 0 eV. The intensity of this peak thus decreases with increasing energy. Since the minimum kinetic energy of a sputtered ion that leaves the surface is 0 eV, these ions will attain a minimum energy of eV eh after acceleration. This will lead to a sharp onset on the low-energy side of the spectrum. Although the precise energy of the onset will depend on the geometry of the set-up (on the dimensions of the

26

Applications of Low-Energy Ion Scattering to Oxide Surfaces

front end of the analyzer, sample holder and beam spot), the value of eV eh (and thus the energy correction for the scattered ions) can be easily be estimated for a given

experimental set-up.

A quantitative analysis of surface composition becomes difficult if surface charging

is appreciable. This is especially the case for very rough surfaces, such as those

encountered with catalysts, where surface charging is very inhomogeneous. This is

illustrated in Fig. 6. Proper charge neutralization is then imperative. Since the primary

ion current that bas to be compensated for is only of the order of 0.1 - 100 nA, a simple

+ + ++ + ++

ll j j j ll

Fig. 6 lllustration of rough surface with ion beam and partial charge compensation by an electron beam.

filament easily produces enough electroos to do this. Preferably these electroos should

come from the same direction as the ions. For the ERISS set-up (Fig. 4) this is

approximately achieved by using a ring filament that surrounds the primary beam. A

simple electrostatic deflection keeps the filament out-of-sight for the target. This prevents

heating of the target by radiation and the contamination of the target by evaporation from

the hot filament. The problems of charging can be circumvented by the use of a beam of neutrats

[9], but this makes quantification much more difficult. Another possibility is to increase

the electrical conductivity by heating the materiaL Grünert et al. [48,49] applied this

metbod to zeolites, where the surface charging effectively could be reduced to enable the

study of the surface by LEIS, UPS and XPS.

27

Chapter 2

2.3b Target preparation The extreme surface sensitivity of LEIS makes proper target preparation a

necessity for quantitative compositional analysis. Powders are pressed into pellets to

reduce the macroscopie roughness and thus facilitate charge compensation. It also

provides a more reproducible and higher density. The surface can be cleaned by sputtering or by chemical reactions. Sputtering generally affects, however, the atomie

structure of the surface and subsurface layers. The most suitable chemical reaction depends, of course, heavily on the target

composition. This is especially important for oxides where the surface composition

depends strongly on the oxygen partial pressure.

Margraf et al. [37] use a H2 beam having an effective pressure at the target of 10""

mbar during one hour at a target temperature of 600 K. This effectively removes

hydracarbon contaminants from the surface. Remaining hydrogen on the surface will strongly neutralize the scattered ions and physically shield the underlying atoms. This affects the quantification (see section 2.5). It is believed, however, that a small amount of

sputtering will remove most of the hydrogen before sputtering of the target itself beoomes

too significant. The hydrocarbons, water and most of the hydroxyl-groups can generally

also be removed by treating the target for 15 minutes with 20 mbar of oxygen at 2500C.

After such a treatment the surface can be analyzed without any sputter cleaning.

2.3c Other experimental problems When the LEIS signals are calibrated against those of reference targets,

reproducible scattering conditions, beam current and beam profile are important. The

precise location of the target will influence the value of the scattering angle e somewhat.

For small scattering angles in particular, this can affect the interpretation. Helbig et al. [47] have given a more detailed analysis of this problem.

2.4 lnterpretation of energy spectra

The energy loss of the ion is mainly due to momenturn transfer in an elastic

collision between the ion and a surface atom. In addition to this, however, inelastic

processes may further lower and broaden the energy of the "elastic" peak. Sametimes one

can even distinguish a discrete inelastic loss of about 25 eV (see e.g Refs. [50,51]).

Although the absolute value of this inelastic loss is not very high, it may complicate the interpretation of the spectra (especially at low ion energies). Wheeler [52] reported, for instance, that for 500 eV 4He+ scattering through 9ff the inelastic energy loss shifted the

28

Applications of Low-Energy Ion Scattering to Oxide Sulfaces

peak for Ta (mass 181) to an even lower value than that of Ag (mass 108). In general, however, there is no serious problem in assigning the peaks. This is especially true for impactenergiesof 1 keV and higher where the inelastic losses are relatively weak.

When the forward scattering of heavy ions fONe+, 40Ar+) is studied at low kinetic energies, multiple surface scattering is often observed and it may even beoome the dominant feature in a spectrum. Under such conditions, the final energy for a given total

scattering angle can be higher than that expected for single elastic scattering (Eq. 1.1 ). For single scattering conditions, however, inelastic processes shift the peak to values slightly lower than that predicted by Eq. 1.1. In practice, the high-energy onset of a peak

is generally a good measure for the elastic value and should thus be used to calculate the mass of the surface atom.

Fig. 7

600

'ih' ö 400 -ti c

.!:.? Cl)

en w ..J

200

0 0

Al

blnary colllsion

reionized neutrals double collislons

/ 500 1 000 1500 2000 2500 3000

Final energy (eV)

LEIS-spectrum for He+ scattering by aluminum.

In Fig. 7 a characteristic spectrum for 3 keV 4He+ ion scattering by Al is given. In addition to the elastic scattering peak around 1700 eV, a small tail is observed at the high energy side, which is the result of ions that collided several times at the surface. This contribution is very small in the backscattering geometry and can in most cases be

29

Chapter 2

neglected. However, there is an intense background (tail) extending to lower energies

This tail results from ions that penetmte into the target, lose various amounts of energy in

multiple rollisions with target atoms and are finally scattered back into the direction of

the analyzer. In genera!, the inroming ions are already neutralized during their fust interaction with the target surface and will thus escape detection in an electrostatic energy

analyzer (only charged particles are detected). It is known, however, that reiomzation can occur in binary rollisions of certain ions with certain atoms (see Ref. [50,53] for a list of rombinations where reionization is observed). The tail results from ions that suffered

such a rollision upon teaving the target. Since reionization occurs as a result of the

overlap of the inner electrooie levels of the incident partiele and the target atom, there is

an energy threshold for this process. As the onset of the tail in Fig. 7 illustrates, the threshold for He+ scattered from Al is around 0.5 keV.

De Wit et al. [54] observed for Ne+ scattering from Cu that there is an

increased background on the low-energy side of the Cu elastic peak when oxygen is adsorbed. This was interpreted as a decrease of the neutralization probability. The importance and origin of the tails have already been described by Nelson [55]. Since

the ions have lost additional energy while passing through the deeper layers, the tails are always on the low-energy side of a peak. The background on the high-energy side of the

peaks is thus always much lower and gives a good impression of the intrinsic noise level

of the set-up. For LEIS studies of oxides, the reionization of He and Ar by oxygen is

particularly relevant. Nelson [55] pointed out that the 2s level in oxygen (23. 7 eV) is

closely matebed with the 1s level in helium (24.5 eV) and the 3s level in argon (25.3

eV). Although Tongson et al. [56] claim in their experiments of inert-gas ion scattering from various gadolinium rompounds, that the low intensity of the tails in the Ne+

scattering are due to effective removal of the oxygen by sputtering, tails are generally

observed for He and Ar but not for Ne scattering from oxides. The presence of a tail seriously affects the detection of light elements. This is

particularly serious since theelemental sensitivities (see section 3.3) for light elements are relatively low. At 3 keV He+ scattering (8 = 14<fl), surface roncentrations of heavy

elements can, therefore, still be detected for roncentrations as low as 10 ppm, while a value of about 1 at.% holds for elements such as oxygen. One can improve the relative

sensitivity for oxygen, by lowering the primary energy [24,52].

lons penetrating into thick targets may loose their full kinetic energy during

straggling through the target. Thus scattered particles will have energies from zero up to

the elastic binary rollision peak. Since only ionized particles are detected, the onset of the

30

Applications of Low-Energy Ion Scattering to Oxide Surfaces

tail is determined by the threshold for reionization. Martin and Netterfield [57] used 2 keV He+ ions to follow the growth of zirconia on silica. lt was found that the low-energy onset of the tail of the elastic Zr peak: shifts to lower energies until a thickness of 70 Á is reached. Since the tail extends downward from the Zr peak:, it is clear that scattering from Zr in deeper layers is involved.

Van Leerdam et al. [58] used the tails in 3 keV He+ scattering to obtain information on the composition of layers as deep as 60 Á. Computer simulations of ion

trajectories in the solid using the SISS-92 program [59] suggest that not only the atomie composition but also the crystalline structure can influence the intensity of a tail. For crystalline targets the ions are found to be deflected into channelling directions after passing only a few atomie layers. This reduces the chance of backscattering and thus the intensity of the tail in comparison to that of a truly amorphous material. 1t seems feasible to exploit this phenomena to obtain unique information on the local crystallinity of a target. This may be very relevant information if the local crystallinity determines oxygen

ditfusion through oxides. However, recent experiments [60] on silicon, where a thin amorphous layer was crystallized upon annealing, could oot confum this.

2.5 Choice of experimental conditions

In principle, any element that is heavier than the ion used for the analysis can be detected by LEIS. The optimal conditions result from a trade-off between scattering probability and neutralization. For large scattering angles, as encountered in a CMA, deleetion is easy for atoms that are at least 2.5 times heavier than the probing ion.

Heavy ions and large scattering angles are desirabie to improve the mass resolution. For

instance, with backscattering of Ne+ and Ar+ ions it is possible to determine the ratio of

the isotapes in a Cu target, while this is very difficult with He+ ions. A drawback of the use of heavy ions such as Ne+ and Ar+ is that their sputter rate is about ten times higher than that of He+. Of course when one is interesteel in the oxygen signal only He+ can be used in a back-scattering geometry.

To benefit from both the light and the heavy ions, mixtures of ions are sometimes used [23,61]. For samples that are not too complex, it is still possible to assign the

peaks to the proper incident ion-target atom combination. In fact, in set-ups that do not allow for differential pumping of the ion source, ion mixtures are often automatically present to a eertaio degree, because of contamination by inert gas from previous experi­

ments. A wide range of energies (200-5000 eV) is used for compositional analysis by

31

Chapter 2

LEIS. Since the sputter rate at these energies is almost proportional to the energy, many authors prefer the lower energies. Often, the argument used is that increased surface

sensitivity is obtained at lower energies. For oxides it has been shown [62], however, that there is no detectable difference in screening between ions of 500 and 3000 eV.

ligher energies have the advantage of somewhat better control of the ion beam, while the

effects of charging and contamination are less severe. In practice, energies of between

500 and 3000 eV are commonly used and constitute a good oompromise for the

conflicting demands. Bea.m currents of a few pA to a few p.A are used with spot sizes of the order of a

mm2• Although sputter rates vary, of course, from target to target; a characteristic removal rate of 1-2 atomie layers for 500 eV 4He+ [37] fora fluence of 2 ·1016 ions/cm2

can be taken as a guideline.

The extreme surface sensitivity of LEIS implies that surface contamination greatly

influences the signals. This has often confused the interpretation of LEIS. In particular, the presence of hydrogen atoms poses a problem. Since hydrogen bas a lower mass than

the lightest inert gas ion (He), conservation of momenturn forbids backscattering of He by

H atoms (although H can be observed in forward scattering or as recoil ,see Ref. [13]).

This means that although neutralization and scattering by hydrogen atoms may reduce the signals from the underlying atoms in a backscattering experiment, it cannot be detected itself. The complication of the quantification by the presence of hydrogen is well-known

[28,63,64,65,66,67,68,69]. The presence of an intense sputter peak (see section 1) is generally a good

indication of a high concentration of hydrogen on the surface [67]. Hydrogen is a special

problem for surfaces such as metallic Pd that dissociate molecular hydrogen or hydrocarbons which are generally present as residual gas in a vacuum system. This may selectively lower the signals for such elements, although spill-over reactions may also

lead to adsorption of hydrogen on neighboring atoms, with a corresponding decrease of

their signals. Another important souree of hydrogen atoms is the adsorption of water,

which will forrn hydroxyl groups on the surface. Most of the hydrogen (as hydride or

hydroxyl) can generally be effectively removed by a treatment in oxygen at elevated temperatures (e.g. a 15 min treatment at 2500C in oxygen of 20 mbar). Since hydrogen

can be sputtered quite efficiently, a small amount of sputtering will remove most of the

hydrogen before the sputtering of the substrate beoomes appreciable. In Fig. 8 an example is given of a palladium surface on which hydrogen has been

adsorbed [70]. The surface composition has been analyzed in the ERISS instrument (see

section 2.2). The sensitivity of this instrument is so high that a very low intensity ion

32

Fig. 8

Applications of Low-Energy Jon Scattering to Oxide Sulfaces

-:::1

..à ,_ a:s -~ '(i) c: Q) -c:

(IJ

w ....I

10

8

6

4

2

0 0

clean Pd surface H-saturated Pd surface

1000 2000 3000 Final energy (eV)

4000

Energy spectra of 4 keV He+ from a Pd sulface before and qfter J(J L H2 exposure [70].

beam can be used to study the surface (statie LEIS). Without sputtering, the hydrogen completely masks the presence of the underlying Pd. This example underlines the extreme surface sensitivity of LEIS. Due to the high sputter probability of hydrogen, regu1ar LEIS

equipment bas no hinder of this effect, even when very low ion energies are being used.

In such experiments a much higher ion dose than in the ERISS is required to obtain even

a frrst spectrum. By that time, most of the hydrogen will have been removed.

Exept from hydrogen, all elements of the periodic table can be detected by

backscattering of He+ ions. Elements such as lithium, beryllium, boron and carbon have

been studied by LEIS, but their elemental sensitivities are very low. The strong matrix effect observed for carbon [71] complicates the detection of the different types of carbon even more. When Li, Be, B or C are present on the surface in significant quanti­

ties, quantification is difficult, although not impossible. It is quite possible that many of

the matrix effects observed in the early years of LEIS were in fact due to varying

amounts of contarnination on the surface. If none of these very light elements are present

33

Chapter 2

in significant amounts, it should be possible to obtain an accuracy of the order of a few

percent for the surface concentrations of the main elements.

2.6 Fitting of LEIS spectra

As discussed above, the LEIS spectra can be divided in most cases into three

contributions. The surface peak, which contains the surface sensitive information, and a

background, which originates from reionized particles scattered from deeper layers.

Generally, the backgrounds are relatively large in oxides. This greatly hampers the

quantification of the LEIS signals, especially for the lighter elements. To extract the

surface sensitive information from the spectra a number of methods have been put

forward. All are based on (semi) empirical considerations since the neutralization and

reionization processes are not well understood.

For materials where the tails in the spectra do not depend too strongly on the

matrix (which is more the exception than the rule), Baun and Solomon [72] proposed

subtraction techniques to analyze the spectra.

A linear background subtraction bas proven a very effective and easy way to

obtain peak areas, see for example Refs [73,74] for a nice illustration. Also the use

of peak heights can provide in many cases a reliable indication for the changes in surface

composition. It bas been found that the fmal results are identical for a number of cases

[62, 75] when peak heights are used, instead of peak areas obtained after linear

background subtraction.

Empirical fitting procedures of background subtraction and peak identification have

been introduced by Nelson [76] and were extended by Young and coworkers [77, 78] This procedure allows for deconvolution of overlapping peaks but, due to

the number of parameters, convergence is sometimes difficult.

Creemers et al. [61] describe a relatively simpte but fast iterative procedure to

obtain peak areas.

2. 7 Changes in the surface composition induced by the analysis

The analysis itself may change the surface composition of a target. Firstly,

vacuum is necessary to perform the LEIS measurements. The stability of the surface

under such conditions is a necessary requirement to obtain relevant information. For

oxides the oxygen partlal pressure is known in some cases to influence segregation

equilibria [79]. A general requirement for LEIS measurements is that the pressure of

34

Applications of Low-Energy Ion Scattering to Oxide Surfaces

the residual gas is low enough to prevent discharges between lens elements that are at high voltage. Another, generally less important, requirement is that oollisions of the

primary and scattered ions with the residual gas can be negleeteeL Unless special precantions are taken, the maximum background pressure is around w-s mbar.

So, in principle, it is possible to vary the oxygen partia1 pressure between 10·5 and

to-10 mbar. Up to now there are hardly any studies carried out where this has been actually done during the measurements.

For a surface sensitive technique such as LEIS, preferential sputtering and damage

by sputtering can also change the surface oomposition. For large enough fluences, preferential loss of oxygen is observed for most oxides. Pitts and Czanderna [80] found

that even a very stabie oxide such as Si02 is reduced by ion bombardment. A fluence of 7

mC/cm2 (4·1016 He+ ions/cm2) of 1 keV 3He+ reduces the oxygen content by 4 at.%.

Both a oollisional [81] and a thermal model [82] have been described (see also Refs.

[83,84]). Acoording to Saied et al. [85] one can reduce the damage by using beams of neutralized ions rather than the ions themselves. The quantification of the

oomposition from the resulting scattered ions is, however, more complex than for incident

ions.

2.8 Compositional deptb profiling

Besides the surface information, the in-depth distribution of the elements in a solid

are of great interest. Usually depth profiling is performed oombining a sputter ion beam

with surface analysis by ABS or SIMS. The bombardment time can be converled into a depth scale. It is clear that also LEIS oould provide this information. Here the primary

ion beam can also be used to sputter the material. However, it is advantageous to use a separate ion beam for sputtering, as will be discussed below.

In this section it is not intended to give a detailed description of the processes that

rule the sputter assisted compositional depth profiling, but a few pittfalls and the use of

LEIS on oxides will be discussed. For a more thorough approach the reader is referred to

Refs. [86,87,88].

The conversion of bombardement time into depth scale is not trivial. The sputter

rate can be constant, when only one major oomponent is present, but can also vary in

depth. Assuming the rate is constant, measuring the erater depth after analysis can give the erosion rate. Another metbod is to measure the weight loss during profiling using a

thickness monitor. However, in many cases no depth scale is given or the conversion is

based on a rough estimate. When the analyzing and spottered area are the same the signal

35

Chapter 2

from the erater edge will distort the depth proflle. The faet that the ion-eurrent

distribution is not uniform can enhance this effect. Rastering of the ion beam and

redueing the analysed area will overcome this problem. lt is elear that, when in LEIS the

primary beam is also used for the sputtering, the analyzing spot should be smaller than the diameter of the ion beam. The composition measured at the surface does not have to reflect the in-depth information for a number of reasons. Firstly, as discussed in the previous section, preferential removal of one component is possible. While in the analysis this effect can be minimized by decreasing the ion dose, this problem is inevitable when

performing ion-assisted depth proflling. Ion beam mixing is a well-known phenomena

which can damage the structure, see Ref. [89]. Radiation enhanced diffusion in

partieular of alkali metals can deplete the solid of this material [90]. Surface roughness, intrinsic or induced by the ion beam, can hinder accurate depth profiling.

Nevertheless when one takes these effects in consideration, acurate depth profiling

can be performed, although for this SIMS and ABS are most commonly used. For very shallow depth information, the sensitivity of LEIS for the outermost atomie layer is a

distinct advantage. LEIS is, for instance, very well suited for determining the dispersion

of a deposited species. For monolayer dispersion an exponentlal decay of the LEIS-signal

is expected, while for a cluster the signal will stay constant until the top Iayers are

sputtered away, after whieh a transition to an exponential decay is expected. In catalysis, LEIS is frequently used in combination with depth proftling to obtain

the near-surface compositonal proftle. This is possible, even, with the highly porous very rough surfaces the "real world" catalysts usually have. But the depth resolution is only

small since the strueture studied is not a nicely Iayered strueture. Therefore, only near­

surface (few atomie Iayers) depth proflling on supported catalysts is meaningful

Brinen et al. [91] used the combination of quantitative XPS and LEIS to study

depth proflling in HDS catalysts. Van Leerdam et al. [92] (see also section 4.1) studied

the stabilisation of 'Y-Al20 3 by La. The groups of Taglauer and Knözinger have presented a number of studies where the spreading and dispersion of catalytically aetive oxides on

their supports is rationalized (for a review see Ref. [93]). Also the SMSI (strong metal­support interaction) effect, well-known in catalysis, of Rh on a titania model support was identified by comparison of the near-surface depth profiles as a funetion of the calcination

temperature [94].

Besides the porosity and roughness which will hinder the proflling of catalysts,

Den Otter et al. [95] showed that for the study of small clusters extra care sbould be taken. By molecular dynamics calculations on a small Rh cluster (148 atoms) on a

support, they showed that one probing ion can lead to a complete explosion of the cluster.

36

Applications of Low-Energy Ion Scattering to Oxide SUlfaces

They state that only a very small dose can be used to study the surface of these clusters and that depth profiling of these clusters is not possible, since the sputtered particles do not necessarily originate from the surface. This also has implications for the SIMS studies of these kinds of systems. In LEIS the interaction time with the cluster is so small that the prohing ion is already far away from the cluster when the cluster begins to deform. As stated above, in LEIS the problem can be overcome by using low doses ( < 7 • 1012

ionslcm2 for 2 keV Ne+). This requirement cannot be fulfilled by the conventional LEIS machines but the new generation, like the ERISS, can provide analysis already with ion doses well below this limit. However, in the investigation of Den Otter et al. [95] the metal particles had no interaction with its (metal) support. The results should therefore only be extrapolated to small metal clusters on a support material where the interaction on the interface is small.

3. SURFACE STRUCTURE OF NON-SINGLE CRYSTALS

3.1 Introduetion

For amorphous and polycrystalline materials, which is the case in most applications, the information gained is averaged over all azimuthal angles and for powders over many angles of incidence. Nevertheless various authors have explored other possibilities to obtain structural information on such materials.

3.2 Surface structure from information depth

In low-energy ion scattering from powders, shadowing and blocking effects are averaged over many crystal orientations (various surface planes, all azimutlts and many angles of incidence). At first sight one might, therefore, expect that for a homogeneons powder a LEIS spectrum would just give the bulk stoichiometry of this powder. This is certainly not the case. Although LEIS is not able to detect the shielding by oxygen of cations within the toplayer, the neutralization, shadowing and blocking are very effective in shielding the deeper layers. This property can be used to obtain very valuable information on the surface structure of these powders and thus have a direct bearing on the understanding of sintering and chemical reactions at the surface.

The importance of shielding in oxides is shown in Fig. 9 where the spectra for He+ ion scattering from f:X- and 'Y-Al20 3 are compared [96]. The Al signal of the 'Y-

37

Chapter 2

modification is only 75 ± 5 % of that of the a-modification. Similar differences between the a- and 'Y-phase have been previously observed for F~03 [97].

To compensate for differences in grain size and pore structure (a- and 'Y-alumina have specific areas of 5.5 and 280 m2/g resp.) the spectra have been normalized to the oxygen peaks (see also chapter 3 of this thesis). The same oxygen signals are expected since a-alumina bas the corundum structure with hexagonal close packing of the oxygen ions, while 'Y-alumina bas a defected spinel structure with almost perfect cubic close pac­king of the oxygen ions. The number densities of the oxygen ions in the topmost layer wi11 thus be the same. For clarity the spectra in Fig. 9 have been shifted somewhat in

energy.

600 1200 1800 2400

Fig. 9 LEIS-spectra of the surfaces of a-AIP3 and 'Y-Alp3• (3 keV 4He+) [96].

For the 'Y-alumina the Al and 0 intensities initially increase rapidly with bombardment time, after which they develop more gradually. If the powder is first heated at 4000C the initia! Al intensity increases by 35%. With higher pretreatment temperatures the initial Al and 0 signals are higher, but the final values are the same. 1t is generally known that hydroxyl groups are present at the surface of 1-alumina. lt is, therefore, assumed that the initia! increase of the signals is due to the removal of these groups. lncreasing pretreatment temperatures wi11 lead to a higher degree of dehydroxylation. Moreover, as a result of the thermal treatment mainly bridging OH groups will remain

38

Applications of Low-Energy Ion Scattering to Oxide Surfaces

[98], which hardly shield the underlying Al atoms. After prolonged ion bombardment the Al signal slowly increases to the value of the a-alumina, while the oxygen signals of both modifications decrease due to preferential sputtering.

In the mid-seventies, Shelef et al. [63,99] used LEIS to investigate the relation between surface composition and chemical reactivity of normal spinels. These compounds

have the general formula:

(3.1)

in which the forma! valencies (roman numerals) and cation coordinations are indicated. Although the sensitivity of LEIS equipment in those days was not yet very high and although the authors stressed that their experiments were still of preliminary nature, their results suggest that tetrabedrally coordinated cations may not be accessible to LEIS. 1t is found that in the first spectrum of a fresh target (little damage by the ion beam) no Co or Zn are detectable in CoA120 4 and ZnA120 4• The cations in tetrabedral surface sites are

believed to be less stabie and will, therefore, move to sites below the surface and thus be shielded more than the cations in octahedral sites. Another explanation is that only certain crystallographic planes may be preferentially exposed in spinels. Recently, the findings of Shelef et al. [63,99] have been confirmed in LEIS studies of various catalytically active spinels ([100] and chapter 4 of this thesis). As illustrated in Fig. 10 [101], the Zn signa! in ZnA120 4 is less than one percent of that for ZnO; see also Ref. [102]. When

oomparing different Co-containing spinels it was found that for CoA120 4, where almost all

Co-cations are present as co2+ in the tetrabedrally coordinated sites, the Co signa! is only 13 % of that for ZnC~04, where almost all Co-cations are present as cif+ in the octahedrally coordinated sites. The Co signals for CÜJ04, where Co is present in both the tetrabedral and octahedral sites, show the same amount of Co on the surface as the ZnC~04 • The surface composition as determined by LEIS was found to correlate directly with the catalytic activity [100]. The results strongly support the idea that in II-III spinels only the octahedral sites are occupied in the surface. For a series of Mn-containing

spinels it was found that due to an oxidative transfer much more Mn was found on the surface [100].

LEIS showed that the preferential exposure of the octahedrally coordinated cations also holds for substituted ferrites (ZnxMg1.xF~04) [103], which are inverse spinels. Inverse spinels have the same structure as normal spinels except that their cation distribution is different: B!dr(III)A ""1(II)B""'(ffi)04, in accordance with Eq. 3.1. While

F~04 (inverse spinel) showed the same amount of iron on the surface as ZnFe.zÜ4 ( normal

39

Chapter 2

spinel), MgF~04 exhibited only half of the iron in comparison to the normal spinel,

which is in complete agreement with the model where only octahedral sites are exposed.

Furthermore, in ZnF~04 no Zn was detected, while in MgF~04, where the doubly

charged Mg-cation is in the octahedrally coordinated interstices, Mg was clearly visible.

1000 1500 2000 2500 Final energy (eV)

Fig. 10 LEIS-spectra of 3 keV He+ scattering from ZnAIP4 (solid line) and ZnO (dashed line) [101}.

Since the pioneerlog work of Shelef, many other groups have combined LEIS

with catalytic activity measurements. In particular the groups of Hercules and Houalla

[104], Taglauer and Knözinger [28], Hoflund [105], Bertrand and Delmon

[106], Bonnelle [107] and of Brongersma [27] have made important contributions;

see also the reviews by Horreil and Cocke [26], Brongersma and Jacobs [lOl] and of

Taglauer and Knözinger [93]. In general the LEIS signals were found to correlate very

nicely with catalytic activity, since both refer to the top atomie layer. When the necessary

40

Applications of Low-Energy Ion Scattering to Oxide Surfaces

experimental precautions are taken, this correlation is even quantitative [101]. This also suggests that there is no change of the cation concentration in the surface due to chemically induced segregation during the catalytic reaction.

3.3 Signal as a function of toading

Another metbod that bas found widespread application in studies of catalysts is to follow the ion scattering signals as a function of the loading of the catalyst. At low loadings (well below monolayer coverage) of the support it is not unreasonable to assume that adsorbed species ("the loading") do notcluster and are thus visible for LEIS. If this is the case, one would expect a linear increase in the LEIS signal of the adsorbed species as a function of loading and a linear decrease of that of the support.

Because of various experimental reasons (see section 2.3) signal ratios are generally used instead of the absolute signals. This may lead to erroneous interpretations. Zingg et al. [108] observed for instance, for catalysts of a -y-Al20 3 support a clear increase in the Mol Al signal ratio at a loading of about 7 wt. %, when the molybdena loading was further increased. 1t was assumed that at low loadings the Mo atoms occupy mainly tetrabedral sites and are thus less visible. Above 7 wt. % loading the additional Mo atoms were assumed to be in octahedral sites. This coverage-dependent site occupation was not observed by Kasztelan et al. [107]. Van Leerdam et al. [75] observed the same discontinuity in the Mol Al signals as Zingg et al. [108]. The change occurred at somewhat higher loadings since their specific area of the -y-alumina was 270 m21g instead of the 196 m21g used by Zingg et al. [108]. However, they were able to determine both the Mo and Al signals as a function of surface loading (see Fig. 11). The Mo signal is found to be strictly linear with Mo03 coverage up to 20 wt.% (above this loading 3-dimensional structures of Mo03 are formed). This suggests that up to this loading the Mo is always accessible to LEIS and thus not bidden in a tetrabedral site. The discontinuity in the Mol Al signals is due to an abrupt change in the Al signal. At low loadings the Al is apparently not shielded by the Mo. At fust sight this may seem surprising. The structure of a cation-deficient spinel such as -y-Al20 3 is rather open. lt is thus possible to find sites for the Mo where it is on top but does not shield any Al atoms. At a loading of 10 wt.% these sites are full. When the loading is increased further, all Mo atoms shift to new locations. Van Leerdam et al. [75] have indicated possible sites for the Mo atoms at low and high loadings.

Both the constant Al signal at low loadings and the abrupt change around 10 wt. % are difficult to comprehend for a powder that exposes many different crystallographic

41

0

Fig. 11

Chapter 2

'--~

20 Mo03 - content (wt %)

• •

30

The LEIS-intensity (S) of Mo (squares) and Al (circles) as a .function of Mo03

content [75].

planes. For this reason and others it bas been suggested [75] that powders that have been annealed at high temperature for long periods may have reached a (kind of) thermodynamic equilibrium at which only one or two of their most stabie surfaces are exposed. Whether such an equilibrium is actually reached may also depend on the structure of the starting material, surface impurities, etc. If the hypothesis of the existence

of only one or at most a very limited number of crystallographic surface planes is true, it

would enable the use of LEIS for atom location on powders. The research on the spinels [100] supports this hypothesis.

In LEIS studies in the past, the support has been regarded as an inert base for the active compound, notwithstanding the fact that it is well known in catalysis that there are strong metal-support interactions that completely change the chemistry. The LEIS signals were normalized to the Al signal of the support. lt was expected that compounds like Mo03 were adsorbed on top of A~03 and shielded the underlying Al atoms and consequently one corrected for this shielding [104]. The results of van Leerdam et al. [75], however, clearly indicate the importance of measuring and evaluating the signals themselves rather than signal ratios. The use of absolute signals also proved to be

42

Applications of Low-Energy Ion Scattering to Oxide Surfaces

important in establishing the mechanism of growth of Ni on A~03 in Atomie l.ayer Epitaxy (ALE) [101,109]. It is presently believed that abrupt changes in the signa! of the support as a function of toading are the rule rather than the exception (see also [110]). The changes in the support signa! can be caused by rearrangements in the

surface layer of the other atoms but also by phase transitions and other types of restructuring of the atoms in the support. It is quite possible that varloos results of LEIS experiments on bimetallic catalysts, where abrupt changes in the signal ratios of toading to support have been observed [111,112,113,114], are due to changes in the signal of the support.

3.4 The energy metbod

One metbod that has received attention (e.g. Refs. [24,25,55,57,115]) is based on the dependenee of signal ratios on the incident energy. It was found that the energy dependenee of the ratio of the signal for scattering from a cation to that of an anion (C/ A) is very characteristic for a compound. This was attributed to the screening of the metal cations by the oxygen anions. So by varying the incident energy, information of the atomie structure on the surface could be determined. Van Leerdam and Brongersma did show [62,116], however, that the origin of the energy dependenee of C/A is not a measure for the screening but generally an intrinsic property of the elements involved. When taking the ratio of the silicon signal from pure silicon (bere Si is certainly not screened) and the oxygen signal of Ba03, the same energy dependenee of C/ A was obtained as for Si~ (Fig. 12, [62]). Also, for Al20 3 and ZnO it was found that the characteristic energy dependenee of Cl A is the same for the compounds as for the constituting elements. Mashhoff et al. [117]) confmned the absence of shielding effects on the energy dependenee of the C/ A ratio in Ti~.

4. GROWTH AND WETTING

4.1 Oxides on oxides

Growth of oxides on oxidic supports finds widespread application in the fabrication of catalysts and electrooie devices. Whether the growth takes place via evaporation, decomposition of volatile compounds or via some liquid phase impregnation, in all cases the wetting properties of the oxides greatly influence the growth mode.

White and coworkers [118,119] used LEIS to study the growth of Si~

43

Chopter 2

5

• •

-~· )( x

0 -V)

2

0 5

Fig. 12 Energy dependendes of the SilO ratio jor Si02 (x) and the reference compounds (•) [62].

films on Z~, Ti02 and ZnO via decomposition of Si(OC2H5) 4• The rednetion of the

LEIS signal of the substrate provides an easy tooi to monitor the growth. The behavior

was found to depend strongly on the nature of the substrate. While silica forms a

monolayer film on ZnO, it forms a monolayer plus three-dimensional clusters on Z~

and a mixed silica-titanate overlayer on Ti~. Martin and Netterfield [57] grew Z~ by

evaporation on silica. When a monolayer of zirconia was reached, as determined by in­

situ ellipsometry, the Si had completely disappeared. The zirconia clearly wets the silica.

Leyrer et al. [120] have discussed the wetting of oxides by other oxides. Wetting is

expected when 'Ya + "! .. < "~•• where "/. and 'Ys are the surface energies of the adsorbate and substrate, while the interface energy "/ .. is the excess energy in the interface (thus "/ ..

is defined here as zero for the interaction between the same oxides). When the reaction

44

Applications of Low-Energy Ion Scattering to Oxide Sulfaces

between the oxides is exothermic, -r .. is negative. Wbile surface energies of a number of oxides are nowadays reasonably wen known [121], interfacial energies are not. Since interfacial energies are generally smaller than the surface energies, the surface energies often determine the outcome. Moreover, from the occurrence or absence of mixed oxides one can at least estimate the sign of the chemical contribution to the interfacial energy. For adsorbates such as Mo03, W03 and V20 5, which all have very low surface energies, the spreading on a high surface energy material such as -y-alumina [25,37,106,122] is thus understood [120]. From the low surface energy of titania it is also clear why silica did not spread, while it did spread on ZnO. Comparing the results of Martin and Netterfield [57] with those by White et al. [118,119], it is surprising that not only does silica spread on zirconia, the reverse is also observed. It may be that a strong interfacial interaction is responsible, since a stabie ZrSi04 is known [123].

600 26 wt% MoOJSi02 Mo

' 11 I I - I I

(/) ..... I I 0 400 0 I I ->. I I - I I

"ëi) I c: (J) Si ..... . 5 Cl) 200 w ....I

0 1000 1500 2000 2500

Final energy (eV)

- Calcined in air at 723 K for 24 h. - - · After subsequent annealing in vacuum at 860 K for 4 h. ------ ·· Aft er subsequent treatment in 92 mbar Hp at 320 K for 2 h.

Fig. 13 LEIS spectra of 3 keV 4He+ from molybdenum oxide on silica [101].

45

Chopter 2

The spreading of molybdena on -y-alumina is well established. The spreading of molybdena on silica is more controversial. Wbile Stencel et al. [124] used LEIS to

demonstrate that under calcination conditions molybdena spreads on silica, exposure to water at low temperatures leads again to clustering. This can be done reversibly. In more

• 6ooe t:. Sh. 1050 oe

0.4 0 23h. 1050 oe

.s • 145h. 1050 oe -e ~ ....1

0.2

3

La-content (rot%)

Fig. 14 Dependenee of the LEIS La/Al-ratio on the thermal treatments at various bulk La­contents of La-added "(-Alz(J3 [92].

recent LEIS experiments [27,101] spreading of molybdena was also clearly established, see Fig. 13. However, the spreading of M~ is not expected on the basis of solid-solid wetting of M~ on SiOz, as was shown by Leyrer et al. [120]. In the higly dispersed Mo/Si02 catalyst, as studied by Stencel et al. [124] and by Brongersma and coworkers [27,101], the spreading can be assigned to the break up of so-called Keggin units

(silicomolybdic acid, H2SiMo120 40 ·2H20) which can be restored upon water treatment. This microspreading [125] is reversible.

The growth of oxides on oxides may also lead to the formation of surface spinels. In studies of Cu oxide impregnated -y-alumina, Strohmeier et al. [126] stress the importance of a surface spinel that resembles CuAI20 4• This spinel is formed for low

46

Applicati.ons of Low-Energy Ion Scattering to Oxide Surfaces

loadings of copper ( <4% Cu for a 100 m2/g support) and for temperatures that are not too high (below 5()(fC). On the basis of these and other studies they argue that, in con­trast to the regular spinel, the Cu2+ ions occupy predominantly (90%) the (somewhat

distorted) octahedral sites. Por bulk CuA120 4 only 40% of the Cu2+ ions are found in

octahedral sites. At high Cu loadings bulk-like CuO is formed on the surface.

In studies of lanthanum oxide on ')'-alumina, van Leerdam et al. [92] deposited the

lanthanum via a [La(EDTA)J complex. In Fig. 14 [92] the dependenee of the La/Al

signal ratio in LEIS is plotted as a function of the La content for a number of treatments.

The linear increase of the signal ratio for the as-prepared sample indicates complete spreading (all signals for La are somewhat reduced due to neutralization by the complex). During calcination at 5500C, the lanthanum oxide agglomerates to small 3-dimensional

particles (especially at the higher loadings), while the lanthanum spreads again (formation

of a lanthanum aluminate surface layer) during further calcination at 10500C. When

')'-Al20 3 is impregnated by V20 5 [110], the vanadium species is optimally dispersed at

loadings up to 7 wt%. When the vanadium toading is further increased (7-11 wt%), both

the V and the Al signals remain constant. Apparently the extra V shields the previously

deposited V. For still higher V loadings the surface beoomes fully covered by vanadia. The detailed information is again characteristic of the monolayer information depth of

LEIS. The fmdings of Jacobs et al. [110] have recently been confirmed by Eberhardt et al. [127].

4.2 Oxides on metals and metals on oxides

Various groups have used LEIS to obtain detailed insight into the growth of metals

on oxides (see e.g. Refs. [25,128,129,130,131]) and of oxides on metals (e.g. Refs. [132,133,134]).

The growth of iron oxide on Pt(lll) and Pt(lOO) was found to be a layer-by-layer

growth by Vurens et al. [132, 134] using LEIS, LEED and AES. The FeO orders on both

substrates. Although for both substrates an FeO (111) overlayer is proposed, the iron

signalis much more intense for the overlayer on the Pt(lOO) surface. The shielding by the oxygen is apparently more effective for the contracted layer on the Pt(lll) surface. It is

suggested that the open nature of the Fe0(111) allows the Fe, which is located in the second layer in contact with the Pt substrate, to be still accessible to the impinging He+ ions. Iron oxide on Pt is also used as a model system for studying sintering under the

influence of alkalis.

The diffusion of metals through oxides can depend strongly on the stoichiometry

47

Chapter 2

of the oxide. Ocal et al. [135] studied diffusion of metals through thin aluminum

oxide on its metallic support at a temperature of 80 K. Using LEIS it is shown that

deposited Au atoms diffuse through the oxide to the oxide-metal interface (loss of Au at

the surface), only when the oxide is deficient in oxygen. Deposited K does not diffuse

through the oxide. The results are interpreled in terms of the Cabrera Mott mechanism

for low temperature oxidation of metals. According to this mechanism a potential

difference is created across the thin oxide. This acts as the driving force for anion

diffusion through the oxide. The electron affinity of Au is high, so anions are formed that

are driven into the oxide by the field. Potassium does not form anions and is thus not

affected. The high diffusion barrier in stoichiometrie oxides prevents diffusion at these

low temperatures.

Carver et al. [25] followed the clustering of Pt andlor Pt-oxide on alumina as a

result of heat treatments in oxygen. Skoglundh et al. [136] characterized two Pt/Pd

catalysts on oordierite monoliths with and without prior hydrothermal treatment of the

washcoat. From LEIS a surface enrichment of Pd was found in both catalysts. LEIS

showed that the hydrothermal treatment increased the dispersion of the Pt.

Pitts et al. [129] used variabie angle LEIS to determine the position of silver on

tin-sensitized silica. Hoflund et al. [128] used LEIS in combination with AES and XPS to

establish that annealing in hydrogen of Pt/Ti~ reduces the titania to a lower oxide which

then moves over the Pt and covers it. The encapsulating titanium species migrate again

off the Pt upon oxidation.

Madey and co-workers made a detailed study of the growth of ultrathin metal fJlms on a Ti02 (110) surface using LEIS [137,138,139,140]. Due to the

monolayer sensitivity of LEIS, one can clearly distinguish between growth modes where

3-dimensional clusters are formed (Volmer-Weber) and there where initially a 2-

dimensional monolayer is deposited (Stranski-Krastanov). The Frank-van-der-Merwe

growth, layer by layer, is limited to some metal-on metal systems. The wetting, where

metals were deposited at room temperature by thermal vapor deposition, seemed to follow

the reactivity of the 3d-metals with oxygen. Their results show that the wetting ability

decreases as follows: Cr > Fe > Cu. Copper grows in 3D-clusters with the least

spreading ability. Fe grows in flat 3D-islands while complete wetting is observed for Cr

on Ti02 (110). SMSI effects could be observed: upon annealing Fe on Ti~ (110) a

complete encapsulation by Ti02 was observed, while the Fe clusters remained in the

metallic state [141].

48

Applications of Low-Energy Ion Scattering to Oxide Surfaces

Keferences

[l] H.H. Brongersma, P.A.C. Groenen and J.-P. Jacobs, in Science Q.{ Ceramic Interfaces H, p. 113, Ed. J. Nowotny, Material Science MonograpiJs 81, Elsevier Science B.V., Amsterdam, 1994.

[2] H. Niehus, W. Heiland and E. Taglauer, Surf. Sci. Reports 17 (1993) 213-303. [3] B.V. Panin, Sov. Phys. JETP 15 (1962) 215. [4] V. Walther and H. Hintenberger, Z. Naturlorsch. 18A (1963) 843. [5] D.P. Smith, J. Appl. Phys. 38 (1967) 340. [6] D.P. Smith, Surf. Sci. 25 (1971) 171. [7] J.F. van der Veen, Surf. Sci. Rep. 5 (1985) 199. [8] M. Aono, Nucl. lmltr. Meth. Bl (1984) 374. [9] R. Souda, T. Aizawa and Y. lshizawa, J. Vac. Sci. Technol. A8 (1990) 3218. [10] R. Souda, M. Aono, C. Oshima, S. Otani and Y. 1shizawa, Surf. Sci. 179 (1987) 199. [11] H. Eschenbacher, A. Richard and V. Dose, Appl. Phys. A34 (1984) 19. [12] T.M. Buck, G.H. Wheatly and L. Marchut, Phys. Rev. Lett. 51 (1983) 43. [13] J.W. Rabalais, Surf. Sci. 300 (1994) 219. [14] H.H. Brongersmaand P.M. Mul, Chem. Phys. Lett. 14 (1972) 380. [15] H.H. Brongersmaand P.M. Mul, Surf. Sci. 35 (1973) 393. [16] R.H. Bergmans, W.J. Huppertz, R.G. van Welzenis and H.H. Brongersma, Nucl. lmltr. Meth.

DM (1992) 584. [17] A.W. Czandema, A.C. Miller, H.H.G. Jellinek and H. Kachi, J. Vac. Sci. Technol. 14 (1977)

227. [18) I.C. Fullarton, J.-P. Jacobs, H.E. van Benthem, J.A. Kilner, H.H. Brongersma, P.J. Scanion and

B.C.H. Steele, to be published. [19) D.F. Cox and T.B. Fryberger, Surf. Sci. 227 (1990) L10S. [20) W. Englert, E. Taglauer and W. Heiland, Surf. Sci. 117 (1982) 124. [21] S.H. Overbury and R.J.A. van den Oetelaar, Surf. Sci. 301 (1994) 313. [22] J.-P. Jacobs, S. Reijne, R.J.M. Elfrink, S.N. Mikhailov, M. Wuttig and H.H. Brongersma, J.

Vac. Sci. Techn. AU (1994) 2308. [23] R.E. Honig and W.L. Harrington, Thin Solid Films 19 (1973) 43. [24] R.C. McCune, J. Vac. Sci. Technol. 18 (1981) 700. [25] J.C. Carver, S.M. Davis and D.A. Goetsch, in Catalyst Characterization Science, Chapter 12, Ed.

Publ. Co. American Chemical Society, 1985. [26] B.A. Horreiland D.A. Cocke, Catal. Rev.-Sci.Eng.l9 (1987) 447. [27] H.H. Brongersma and G.C. van Leerdam in Fundomental Aspects of Heterogeneaus Catalysis

Studied by Partiele Beams, p. 283, Eds. H.H. Brongersma and R.A. van Santen, NATO ASI series B 265, Plenum press, New York, 1991.

[28] E. Taglauerin Fundamental Aspects of Heterogeneaus Catalysis Studied by Partiele Beams, p. 301, Eds. H.H. Brongersmaand R. A. van Santen, NATO ASI series B 265, Plenum press, New York, 1991.

[29] J.B. Theeten and H.H. Brongersma, Rev. de Phys. Appl. 11 (1976) 57 (in French). [30] R.F. Goff and D.P. Smith, 1. Vac. Sci. Techn. 7 (1970) 72. [31] W. Heiland, H.G. Scbaeffier and E. Taglauer, Surf. Sci. 35 (1973) 381. [32] H.H. Brongersma and P.M. Mul, Chem. Phys. Lett. 14 (1972) 380. [33] J.F. Kelso, C.G. Pantano and S.H. Garofalini, Surf. Sci. 134 (1983) L543. [34] R. de Gryse, J. Landuyt, L. Vandenbroucke and J. Vennik, Surf. Interf. Anal. 4 (1982) 168. [35] J.F. Kelso and C.G. Pantano, J. Vac. Sci. Techno!. A3 (1985) 1343. [36] R.C. McCune and P. Wynblatt, J. Am. Ceram. Soc. 66 (1983) 111. [37] R. Margraf, J. Leyrer, H. Knözinger and E. Taglauer, Surf. Sci. 189/190 (1987) 842. [38] C. Ocal, B. Basurco and S. Ferrer, Surf. Sci. 157 (1985) 233.

49

Chapter 2

[39] M.T. Schmidt, Z. Wu, C.F. Yu and R.M. Oswood, Jr., Surf. Sci. 226 (1991) 199. [40] P.E. West and P.M. George, J. Vac. Sci. Teelmol A5 (1987) 1124. [41] J.F. Geiger, K.O. Schierbaum and W. Göpel, Vacuum 41 (1990) 1629. [42J H.H. Brongersma, N. Hazewindus, J.M. van Nieuwland, A.M.M. Otten and A.J. Smets, Rev.

Sci. Instr. 49 (1978) 707. [43] O.J.A. Hellings, H. Ottevanger, S.W. Boelens, C.L.C.M. Knibbeler and H.H. Brongersma, Surf.

Sci. 162 (198S) 913. [44] O.J.A. Hellings, H. Ottevanger, C.L.C.M. Knibbeler, J. van Engelshoven and H.H. Brongersma,

J. Electron. Spectrosc. Relat. Phenom. 49 (1989) 3S9. [4S] R.H. Bergmans, A.C. K.ruseman., C.A. Severijos and H.H. Brongersma, Appl. Surf. Sci. 70nl

(1993) 283. [46] A. Miotello and P. Mazzoldi, J. Phys. Cl' (1983) 221. [47] H.F. Helbig, P.J. Adelman, A.C. Milier and A.W. Czanderna, NucL Instr. Meth. 149 (1978) S8l. [48] W. Griinert, R. Schlögl and H.O. Karge, Surf. Interf. Anal. lO (1993) 603. [49] W. Grünert, R. Schlögl and H.G. Karge, J. Phys. Chem. 97 (1993) 8638. [SO] R. Souda and M. Aono, NucL Instr. Meth. B15 (1986) 114. [S1] S.N. Mikhailov, R.J.M. Elfrink, J.-P. Jacobs, L.C.A. van den Oetelaar, P.J. Scanion and H.H.

Brongersma, Nucl. Instr. Meth. B93 (1994) 149. [S2] W.L. Baun, Phys. Rev. Al7 (1978) 849. [S3] T.M. Thomas, H. Neumann, A.W. Czanderna and J.R. Pitts, Surf. Sci. 175 (1986) L737. [54] A.O.J. de Wit, R.P.N. Bronekers and J.M. Fluit, Surf. Sci. 81 (1979) 177. [SS] G.C. Nelson, J. Vac. Sci. TeelmoL 15 (1978) 7CY2. (S6] L.L. Tongson, A.S. Bballa and B.E. Knox, Mat. Res. Bull. 16 (1981) 775. [57] P.J. Martin and R.P. Netterfield, Surf. Interf. Anal. 10 (1987) 13. [S8] G.C. van Leerdam, K.-M.H. Lenssen and H.H. Brongersma, Nucl. Instr. Meth. B45 (1990} 390. [S9] C.A. Severijos, G. Verbistand H.H. Brongersma, Surf. Sci. 279 (1992) 297. (60] J.-P. Jacobs and 0. Oielkens, unpublished. [61] C. Creemers, 0. Royer and P. Schryvers, Surf. Interf. Anal. lO (1993) 233. [62] G.C. van Leerdam and H.H. Brongersma, Surf. Sci. 254 (1991) 1S3. [63] M. Shelef, M.A.Z. Wheeler and H.C. Yao, Surf. Sci. 47 (1975) 697. [64] E. Haeussler and G.R. Sparrow, Silic. lnd. 43 (1978) 253. [65] H.H. Brongersmaand T.M. Buck, Nucl. Instr. Meth. B149 (1978} 569. [66] J.R. Diehl, J.M. Stencel, C.A. Spitterand L.E. Malmvsky, Surf. lnterf. Anal. 6 (1984} 56. [67] J.F. Kelso and C.G. Pantano, Surf. Interf. Anal. 7 (1985} 228. [68] E. Taglauer, U. Beitat and W. Heiland, Nucl. Instr. Meth. 149 (1978) 605. (69] F. Delannay, E.N. Haeussler and B. Delmon, J.Catal. " (1980) 469. [70] R.H. Bergmans, M. Vanhommerig, 0. Nooij, A.W. Oenier van der Gon and H.H. Brongersma, to

be published. [71] L.C.A van den Oetelaar, S.N. Mikhailov and H.H. Brongersma, Nucl. Instr. Meth B8S (1994)

420. S.N. Mikhailov, L.C.A. van den Oetelaar and H.H. Brongersma, NucL Instr. Meth. 893 (1994) 210.

[72] W.L. Baun, J.S. Solomon, Anal. Chem. 48 (1976) 931. [73] Ch. Lindsmeier, E. Taglauer and H. Knözinger, in Fundamental Aspects of Heterogeneous

Catalysis Studied by Partiele Beams, p. 413, Eds. H.H. Brongersma and R.A. van Santen, NATO ASI series B 265, Plenum press, New Y ork, 1991.

[74] K. Josek, Ch. Lindsmeier, H. Knözinger and E. Taglauer, Nucl. Instr. Meth. 864 (192) 596. [75] G.C. van Leerdam, J.-P. Jacobs and H.H. Brongersma, Surf. Sci. 268 (1992) 45. [76] G.C. Nelson. J. Vac. Sci. TeelmoL A4 (1986) 1567. [77] V.Y. Young, G.B. Hoflundand A.C. Miller, Surf. Sci. 235 (1990) 60. [78] V.Y. Young, N. Welcome, G.B. Hoflund, Phys. Rev. B48 (1993) 2981.

50

Applications of Low-Energy Ion Scattering to Oxide Surfaces

[79] J. Nowotny, in Scûmee ofCeramic Interfaces, p. 79, Elsevier, Amsterdam, 1991. [80] J.R. Pitts and A.W. Czanderna, Nucl. Instr. Meth. B13 (1986) 245. [81] J.B. Malheibe, S. Hoffmann and J.M. Sanz, Appl. Surf. Sci. 27 (1986) 355. [82] R. Kelly, Surf. Sci. 100 (1980) 85. [83] T. Choudhury, S.O. Saied, J.L. Suilivan and A.M. Abbot, J. Phys. D: Appl. Phys. 22 (1989)

1185. [84] J.L. Sullivan, S.O. Saied, T. Choudbury, J. Phys.: Condens. Matter 5 (1993) A291. [85] S.O. Saied, J.L. Sullivan, T. Choudhury and C.G. Pearce, Vacuum 38 (1988) 917. [86] A. Benninghoven, F.G. Rüdenauer and H.W. Wemer, in Secondary Ion Mass Spectroscopy,

lnstrumental Aspects, Applications fJIId Trends, Chemical Analysis vol. 86, John Wiley & Soos, New York, 1987.

[87] M. Grasserbauer and H.W. Wemer, in Analysis of Microelectronic Materials and Devices, John Wiley & Soos, New York, 1991.

[88] A.W. Czanderna, in Ion Spectroscopies for Surface Analysis, p. 11, Eds. A.W. Czanderna and D.M. Hercules, Plenum Press, New York, 1991.

[89] Y.-T. Cheng, S.J. Simko, M.C. Militello, A.A. Dow, G.W. Auner, M.H. Alkaisi and K.R. Padmanabhan, Nucl. Instr. Meth. B64 (1992) 38.

[90] W.C.A.N. Ceelen, J.-P. Jacobs, H.H. Brongersma, E.G.F. Sengers and F.J.J.G. Janssen, Surf. Interf. Anal., submitted.

[91] J.S. Brinen, D.A. D'Avignon, E.A. Meyers, P.T. Deng, D.W. Behnken, Surf. Interf. Anal. 6 (1984) 295.

[92] G.C. van Leerdam, H.H. Brongersma, I.I.M. Tijburg and J.W. Geus Appl. Surf. Sci. SS (1992) 11.

[93] E. Taglauer and H. Knózinger, in Surface Science, Principles and Applications, Eds. R.F. Howe, R.N. Lamband K. Wandelt, Springer Proc. in Phys. 73, p. 264, Springer-Verlag, Berlin, 1993.

[94] Ch. Linsmeier, H. Knózinger and E. Taglaner, presented at EUROPACAT-I, September 12-17, 1993, Montpellier, France.

[95] W.K. den Otter, H.H. Brongersma, and H. Feil, Surf. Sci. 306 (1994) 215. [96] G.C. van Leerdam, PhD Thesis, Eindhoven University of Technology, The Netherlands, 1991. [97] H.H. Brongersma, L.C.M. Beirens and G.C.J. van der Ligt in "Materials Characterization using

Ion Beams", p. 65, Eds. J.P. Thomas and A. Cachard, Plenum press, New York, 1978. [98] H. Knózinger and P. Ratnasamy, Catal. Rev.-Sci. Eng. 17 (1978) 31. [99] H.C. Yao and M. Shelef, J. Phys. Chem. 78 (1974) 2490. [100] J.-P. Jacobs, A. Maltha, J.G.H. Reinljes, J. Drimal, V. Ponec and H.H. Brongersma, J. Catal.

147 (1994) 294 and chapter 4 of this thesis. [101] H.H. Brongersma and J.-P. Jacobs, Appl. Surf. Sci. 75 (1994) 133. [102] B.M. Strohmeier and D.M. Hercules, J. Catal. 86 (1984) 266. [103] M.R. Anantharaman, J.-P. Jacobs, H. Rosink, S. Reijne and H.H. Brongersma, unpublished. [104] F.M. Mulcahy, M.Houalla and D.M. Hercules, Anal. Chem. 62 (1990) 2232. [lOS] J.E. Drawdy, G.B. Hoflund, S.D. Gardner, E.A. Yngvadottir and D.R. Schryder, Surf. lnterf.

Anal. 16 (1990) 369. [106] R. Prada Silvy, J.M. Beuken, J.L.G. Fierro, P. Beetrand and B. Delmon, Surf. Interf. Anal. 8

(1986) 167. [107] S. Kasztelan, J. Grimblotand J.P. Bonnelle, J. Phys. Chem. 91 (1987) 1503. [108] D.S. Zingg, L.E. Makovsky, R.E. Tischer, F.R. Brown and D.M. Hercules, J. Phys. Chem. 84

(1980) 2898. [109] J.-P. Jacobs, L.P. Lindfors, J.G.H. Reinljes, 0. Jylhä and H.H. Brongersma, Catal. Lett. 25

(1994) 315. [110] J.-P. Jacobs, G.C. van Leerdam and H.H. Brongersma in Fundamental Mpects of Heterogeneous

Catalysis Studied by Partiele Beams, p. 399, Eds. H.H. Brongersma and R.A. van Saoten, NATO ASI B 265, Plenum press, New York, 1991.

51

Chapter 2

A.W. Stobbe-Kreemers, O.C. van Leetdam, J.-P. Jacobs, H.H. Brongersmaand J.J.F. Schotten, J. Catal. 152 (1995) 130.

[111] B. Horrell, D.L. Cocke, 0. Sparrow and J. Murray, J. Catal. 95 (1985) 309. [112] H. Jeziorowski, H. Knözinger, E. Taglauer and C. Vogdt, J. Catal. 80 (1983) 286. [113] J.-M. Beuken and P. Bertrand, Surf. Sci. IQ (1985) 329. [114] C.P. U and D.M. Hercules, J. Phys. Chem. 88 (1984) 456. [115] R.C. McCune, J. Vac. Sci. Tecbnol. 16 (1979) 1569. [116] O.C. van LeenJam and H.H. Brongersma, Proc. Symp. Surf. Sei., La Plagne (1990) 161. [117] B.L. Maschhoff, 1.-M. Pan, R.A. Baragi.ola and T.E. Madey, in Fundamental Aspects of

HeterogetleQus Calalysis Stutlied by Partiele Beams, p. 419, Eds H.H. Brongersmaand R.A. van Santen, NATO-AS! series 8265, Plenum Press, New York, 1991.

[118] T. Jin and J.M. White, Surf. Interf. Anal. 11 (1988) 517. [119] S.K. Jo, T. Jin and I.M. White, Appl. Surf. Sci. 40 (1989) 155. [120] J. Leyrer, R. Margraf, E. Taglauer and H. Knözinger, Surf. Sci. 201 (1988) 603. [121] S.H. Overbury, P.A. Hertrand and O.A. SomoJjai, Chem. Rev. 75 (1975) 547. [122] L. Salvatti Jr., L.E. Makovsky, J.M. Stencel, F.R. Brown and D.M. Hercules, J. Chem. Phys. 85

(1981) 3700. [123] W.C. Butterman and W.R. Foster, Am. Mineratogist Sl (1967) 884. [124] J.M. Stencel, J.R. Diehl, J.R. D'Este, L.E. Makovsky, L. Rodrigo, K. Marcinkowska, A. Adnot,

P.C. Roberge and S. Kaliaguine, I. Chem. Phys. 90 (1986) 4739. [125] M. de Boer, A.J. van Dillen, D.C. Koningsberger, J.W. Geus, M.A. Vuurman and I.E. Wachs,

Catal. Let. 11 (1991) 227. [126] B.R. Strohmeier, D.E. Leyden, R. Scott Field and D.M. Hercules, 1. Catal. 94 (1985) 514. [127] M.A. Eberbardt, M. Houalla and D.M. Hercules, Surf. Interf. Anal. 20 (1993) 766. [128] O.B. Hoflund, A.L. Orogan, Jr., and D.A. Asbury, J. Catal. 109 (1988) 226. [129] J.R. Pitts, T.M. Thomas and A.W. Czandema, I. Vac. Sci. Tecbnol. A4 (1986) 1653. [130] U. Bardi, 0. Prela.zzi., A. Santucci and 0. Rovida, Catal. Lett. 5 (1990) 315. [131] S.D. Oardner, O.B. Hoflund, M.R. Davidson and D.R. Scbryer, I. Catal. 115 (1989) 132. [132] O.H. Vurens, V. Maurice, M. Salmeron and O.A. SomoJjai, Surf. Sci. 268 (1992) 170. [133] R.J. Oorte, E. Altman, O.R. CoraUo, M.R. Davidson, D.A. Asbury and O.B. Hoflund, Surf. Sci.

188 (1987) 327. [134] O.H. Vurens, D.R. Strongin, M. Salmeron and O.A. SomoJjai, Surf. Sci. 199 (1988) L387. [135] C. Ocal, B. Basurco and S. Ferrer, Surf. Sci. 163 (1985) 335. [136] M. Skoglundh, L.O. Löwendahl, P.O. Menon, B. Stenbom, J.-P. Jacobs, 0. van Kessel and H.H.

Brongersma, Catal. Let. 13 (1992) 27. [137] Th.E. Madey, U. Diebold and J.-M. Pan, in Adsorption on Ordered Suifaces on lonic Solids and

Thin Films, p. 147, Eds. E. Umbach and H.-J. Freund, Springer Proc. in Surf. Sci. 33, Springer­Verlag, Berlin, 1993.

[138] U. Diebo1d, J.-M. Pan and T.E. Madey, Phys. Rev. B47 (1993) 3868. [139] J.-M. Pan, U. Diebold, L. Zhang and T.E. Madey, Surf. Sci. 295 (1993) 411. [140] J.-M. Pan and T.E. Madey, J. Vac. Sci. Tecbnol. All (1993) 1667. [141] J.-M. Pan and T.E. Madey, Catal. Lett 20 (1993) 269. [142] U. Bardi, Appl. Surf. Sci. 51 (1991) 89. [143] P.R. Berning and A. Niiler, Nucl. Instr. Meth. B63 (1992) 434. [144] T.N. Taylor and W.P. Ellis, Surf. Sei. 107 (1981) 249. [145] S. Tanaka, T. Nakamura, M. Iiyama, N. Yoshida, S. Takona, F. Shoji and K. Oura, Appl. Phys.

Lett. 59 (1991) 3637. [146] D.F. Cox, T.B. Fryberger and S. Semancik, Phys. Rev. B38 (1988) 2072. [147] S. Priggemeyer, H. Koschmieder, O.N. van Wijk and W. Heiland, Fresenius J. Anal. Chem. 341

(1991) 343.

52

Applications of Low-Energy Ion Scattering to Oxide Surfaces

[148] C. Creemers, H. van Hove and A. Neyens in Recent Dev. in Condens. Mater. Phys. 2, p. 363, Ed. J.T. Devreese, Plenum press, New York:, 1981.

[149] U. Lindner and H. Papp, Appl. Surf. Sci. 32 (1988) 7S. [lSO] L.C.A. van den Oetelaar, J.-P. Jacobs, M.J. Mietus, H.H. Brongersma, V.N. Semenov and V.G.

Glebovsky, Appl. Surf. Sci. 70/71 (1993) 79. [1S1] E. Taglauer and W. Heiland, Appl. Phys. 9 (1976) 261. [1S2] M. Canepa, P. Cantini, F. Fossa, L. Maltera and S. Terreni, Phys. Rev. B47 (1993) 1S 823. [1S3] H. Niehus and G. Comsa, Surf. Sci. 93 (1980) Ll47. [1S4] J.R. Engstrom, D.J. Bonser and T. Engel, Surf. Sci. 268 (1992) 238. [1SS] M.T. Schmidt, Z. Wu and R.M. Oswood, Jr., Surf. Interf. Anal. 17 (1991) 43. [1S6] U. Bardi, A. Atrei and G. Rovida, Surf. Sci. 239 (1990) LS11. [1S7] W.C.A.N. Ceelen, J.-P. Jacobs, H.H. Brongersma, E.G.F. Sengers and F.J.J.G. Janssen, Surf.

lnterf. Anal. , accepted for publication. [1S8] J.-M. Pan, B.L. Maschhoff, U. Diebold and T.E. Madey, J. Vac. Sci. Technol. AlO (1992) 2470. [1S9] D. Dissanayake, J.H. Lundsford, M.P. Rosynek, J. Catal. 143 (1993) 286. [160] B.G. Baker and M. Jasieniak, in Surface Science, Principles and Applications, p. 279, Eds. R.F.

Howe, R.N. Lamband K. Wandelt, Springer Proc. in Phys. 73, Springer-Verlag, Berlin, 1993. [161] M. Salmeron, Stud. Phys. Theor. Chem. 77 (Comput. Chem., Pt. B) SS. [162] Y.G. Shen, D.J. O'Connor and R.J. MacDonald, Surf. lnterf. Anal. 18 (1992) 729. [163] 0. van Kessel, H.H. Brongersma, J.G.A. Hölscher, R.G. van Welzenis, E.G.F. Sengers and

F.J.J.G. Janssen, Nucl. lnstr. Meth. B64 (1992) S93. [164] M.C. Asensio, M. Kerkar, D.P. Woodruff, A.V. De Carvalho, A. Femandez, A.R. Gonzalez­

Elipe, M. Femanda-Garcia and J.C. Conesa, Surf. Sci. 273 (1992) 31. [16S] J.E. Drawdy, G.B. Hoflund, M.R. Davidson, B.T. Upchurch and D.R. Schryer, Surf. Interf.

Anal. 19 (1992) SS9. [166] R.P. Nettermeier, K.-H. Müller, D.R. McKenzie, M.J. Goonan and P.J. Martin, J. Appl. Phys.

63 (1988) 760. [167] J.R. Pitts, T.P. Massopust, A.W. Czanderna and L.L. Kazmerski, J. Vac. Sci. Techn. A4 (1986)

241. [168] J.R. Pitts, S.D. Bischke, J.L. Falconer and A.W. Czanderna, J. Vac. Sci. Technol. A2 (1984)

1000. [169] A. Rossi, C. Calinski, H.W. Hoppe and H.H. Strehblow, Surf. lnterf. Anal. 18 (1992) 269. [170] W.C.A.N. Ceelen, J.-P. Jacobs, H.H. Brongersma, E.G.F. Sengers and F.J.J.G. Janssen,

NEV AC-blad 4 (1992) 91 (in Dutch). [171] Y. Shen, D.J. O'Connor and R.J. MacDonald, Nucl. lnstr. Meth. B67 (1992) 3SO. [172] S.V. Hattangady, M.J. Mantini, G.G. Fountain, R.A. Rudder and R.J. Markunas, J. Appl. Phys.

71 (1992) 3842. [173] Y.G.Shen, D.J. O'Connor and R.J. MacDonald, Surf. lnterf. Anal. 17 (1991) 903. [174] U. Bardi, A. Atrei and G. Rovida, Surf. Sci. 268 (1992) 87. [17S] X.D. Peng and M.A. Barteau, Catal. Lett. 12 (1992) 24S. [176] W. Mahdi, J. Schütze, G. Weinberg, R. Schoonmaker, R. Schlögl and G. Ertl, Catal. Lett. 11

(1991) 19. [177] A.H. Al-Bayatti, K.G. Orrman-Rossiter, D.G. Armour , J.A. Van den Berg and S.E. Donnelly,

Nucl. lnstr. Meth. B63 (1992) 109. [178] A.A. Dzhurakhalov, E.S. Parilis, N.Yu. Turaev, F.F. Umarov and LD. Yadgarov, Izv. Akad.

Nauk SSSR, Ser. Fiz. SS (1991) 240S. [179] F. Arena, B.A. Horrell, D.L. Cocke, A. Parmaliana and N. Giordano, J. Catal. 132 (1991) S8. [180] R.A. van Hassel and A.J. Burggraaf, Appl. Phys. AS3 (1991) 1SS. [181] S.D. Gardner, G.B. Hoflund, M.R. Davidson, H.A. I..aitinen, D.R. Schryer and B.T. Upchurch,

Langmuir 7 (1991) 2140. [182] U. Bardi, A. Atrei, G. Rovida and P.N. Ross, Surf. Sci. 2Sl/2S2 (1991) 727.

53

Chapter 2

[183] V.D. Borman, E.P. Gusev, Yu.N. Devyatko, Yu. Yu. Lebedinskil, S.V. Rogo:zhkin, V.N. Tronin and V.I. Troyan, Vopr. At. Nauki Tekh. , Ser. Fiz. Radiats 2 (1990) 84.

[184] X.D. Peng and M.A. Barteau, Langmuir 7 (1991) 1426. [185] R. Souda, W. Hayami, T. Aizawa and Y. Isbizawa, Phys. Rev. B43 (1991) 10062. [186] S.J. Hoekje and G.B. Hoflund, Thin Solid Films 197 (1991) 367. [187] A. Miotello, G. Cinque, P. Mazzoldi and C.G. Pantano, Phys. Rev. B43 (1991) 3831. [188] S.J. Hoekje and G.B. Hoflund, Appl. Surf. Sci. 47 (1991) 43. [189] A.H. Al·Bayati, K.O. Orrman-Rossiter, J.A. Van den Berg and D.G. Armour, Surf. Sci. 241

(1991) 91. [190] S.D. Gardner, G.B. Hotlund, D.R. Scbryer and B.T. UpchlllCh, J. Phys. Chem. 95 (1991) 835. [191] Y. Tong, M.P. Rosynek and J.H. Lunsford, J. Catal. 126 (1990) 291. [192] S.P. Jeng, P.H. Holloway, D.A. Asbury and O.B. Hotlund, Surf. Sci. 235 (1990) 175. [193] X.D. Peng and M.A. Barteau, Appl. Surf. Sci. 44 (1990) 87. [194] I.A. Zotov, I.Yu. Panichltin, S.P. Chenakin and V.T. Cherepin, Sverkhprovodimost: Fiz., Khim.,

Tekh. 3 (1990) 233. [195] X.D. Peng and M.A. Barteau, Surf. Sci. 233 (1990) 283. [196] A.M. Then and C.G. Pantano, J. Non-Cryst. Sol. 120 (1990) 178. [197] Q. Guo, L. Gui, H. Huang, B. Zbao and Y. Tang, J. Catal. 122 (1990) 457. [198] R. Cavicchi, M. Tarlov and S. Semancik, J. V~~~.:. Sci. Technol. AS (1990) 2347. [199] B. Baretzky, NATO-ASI ser. E155 (1989) 329. [200] S.D. Gardner, O.B. Hotlund and D.R. Schryer, J. Catal. 1tg (1989) 179. [201] F. Rummens, P. Bertrand and Y. de Puydt, NATO-ASI, ser. E155 (1989) lOl. [202] U. Lindner and H. Papp, Fresenius' Z. Anal. Chem. 333 (1989) 540. [203] H. Nakamatsu, A. Sudo and S. Kawai, Surf. Sci. 223 (1989) 193. [204] H. Bubert, H. Pudeibach, H. Puimand W.A. Roland, Fresenius' Z. Anal. Chem. 333 (1989) 304. [205] S. Houssenbay, S. Kasztelan, H. Toulhoat, J.P. Bonnelle and J. Grimblot, J. Phys. Chem. 93

(1989) 7176. [206] A. Bellare, D.B. Dadybwjor and M.J. Kelly, J. Catal. 117 (1989) 78. [207] G.B. Hotlund, G.R. Corallo, D.A. Ashbury and R.E. Gilbert, J. V~~~.:. Sci.Technol. A5 (1987)

1120. [208] D.A. Ashbury and G.B. Hoflund, J. V~~~.:. Sci. Technol. A5 (1987) 1132. [209] G.B. Hoflund, M.R. Davidson, E. Yngvadottir, H.A. Laitinen and S. Hoshino, Chem. Mater., 1

(1989)625. [210] S.V. Hattangady, G.G. Fountain, D.I. Vitkavage, R.A. Rudder and R.J. Markunas, J.

Electrochem. Soc. 136 (1989) 2070. [211] S. Haupt and H.H. Strehblow, Corros. Sci. 29 (1989) 163. [212] D. Ouafi, F. Mauge, J.-C. Lavalley, E. Payen, S. Kasztelan, M. Houari, J. Grimblotand J.P.

Bonnelle, Catal. Today 4 (1988) 23. [213] V.D. Boeman, E.P. Gusev, Yu.Yu. Lebedinskii and V.I. Troyan, Poverkhnost 11 (1988) 138. [214] E.S. Shpiro, B.B. Dysenbina, O.P. Tk.achenko, G.V. Antoshin and Kh. M. Minlll.:hev, J. Catal.

110 (1988) 262. [215] P.F. Carcia, F.D. Kalk, P.E. Bierstedt, A. Ferretti, O.A. Jones and D.G. Swartz.fager, J. Appl.

Phys. 64 (1988) 1671. [216] S.M. Davis, Catal. Lett. 1 (1988) 85. [217] S.H. Com, J.L. Falooner and A.W. Czandema, J. V~~~.:. Sci. Technol. A6 (1988) 1012. [218] P.J.C. Cbappell, M.H. Kibel and B.G. Baker, J. Catal. 110 (1988) 139. [219] Y. Zhon, M. Nalcasb.ima and J.M. White, J. Phys. Chem. 92 (1988) 812. [220] C.T. Campbell, K.A. Daube and J.M. White, Surf. Sci. 182 (1987) 458. [221] J.M. Van Zoest, I.M. Fluit, T.J. Vink and B.A. Van Hassel, Surf. Sci. 182 (1987) 179. [222] D.A. Ashbury and G.B. Hoflund, J. Vac. Sci. Technol. A5 (1987) 1132. [223] K.H. Muller, R.P. Netterfield and P.J. Martin, Phys. Rev. B35 (1987) 2934.

54

Applications of Low-Energy Ion Scattering to Oxide Sulfaces

[224] T.J. Vink, J.M. Der Kinderen, O.L.J. Gijmman, J.W. Geus and J.M. Van Zoest, Appl. Surf. Sci. u; (1986) 357.

[225] J.W. R.abalais and J.N. Chen, J. Chem. Phys. 85 (1986) 3615. [226] G. Kremenic, V. Cortes Coberan and J.M.O. Tascon, Surf. Interf. Anal. 9 (1986) 207. [227] J.C. Carver, I.E. Wachs and L.L. Murrell, J. Catal. 100 (1986) 500. [228] S. Kasztelan, E. Payen, H. Toulhoat, J. Grimblotand J.P. Bonnelle, Polybedron 5 (1986) 157. [229] J.M.O. Tasoon, P. Bertrand, M. Genetand B. Delmon, J. Catal. 97 (1986) 300. [230] R. Tromp, G.W. Rubloff, P. Balk, F.K. LeGoues and E.J. van Loenen, Phys. Rev. Lett. 55

(1985) 2332. [231] G.B. Hoflund, O.A. Asbury, P. ~n and H.A. Laitinen, Surf. Sci. 161 (1985) L583. [232] R.C. McCune, Mat. Res. Soc. Symp. Proc. 48 (1985) 27. [233] C. Ocal, B. Basurco and S. Ferrer, Surf. Sci. 157 (1985) 233. [234] J.M. Stencel, L.E. Makovsky, J.R. Oiehl and T.A. Sarkus, J. Catal. 95 (1985) 414. [235] M. Erbudak and F. Stucki, Phys. Rev. B32 (1985) 2667. [236] J.F. Kelso and C.G. Pantano, J. Vac. Sci. Technol. A3 (1985) 1343. [237] S. Kannar and P. Mohazzabi, J. Mater. Sci. Lett. 4 (1985) 720. [238] A.E.T. Kuiper, G.C.J. Van der Ligt, W.M. Wijgert, M.F.C. Willemsen and F.H.P.M. Habraken,

J. Vac. Sci. Technol. B3 (1985) 830. [239] R.C. McCune and R.C. Ku, Adv. Ceram. 10 (1984) 217. [240] H. Puderbach and H.J. Goehausen, Spectrochim. Acta 39B (1984) 1547. [241] B.H. Oavis, Appl. Surf. Sci. 19 (1984) 200. [242] R.P. Silvy, J.M. Beuken, P. Bertrand, B.K. Hodnett, F. Delannay and B. Delmon, Bull. Soc.

Chim. Belg. 93 (1984) 775. [243] G.C. Allen, Met. Sci. 18 (1984) 295. [244] J.M. Stencel, L.E. Makovsky, J.R. Diehl and T.A. Sarkus, J. Raman Spectrosc. 15 (1984) 282. (245] L.E. Makovsky, J.M. Stencel, F.R. Brown, R.E. Tischer and S.S. Pollack, J. Catal. 89 (1984)

334. [246] B.R. Strohmeier and O.M. Hercules, J. Phys. Chem. 88 (1984) 4922. [247] H.W. Wemer and N. Warmoltz, J. Vac. Sci. Technol. Al (1984) 726. [248] P. Vargaand E. Taglauer, Nucl. lnstr. Meth. 8230 (1984) 800. [249] S. Kasztelan, J. Grimblotand J.P. Bonnelle, J. Chim. Phys. Phys.-Chim. Biol. 80 (1983) 793. [250] B.P. Loechel and H.H. Strehblow, J. Electrochem. Soc. 131 (1984) 713. [251] P. Bertrand, J.M. Beuken and M. Oelvaux, Nucl. lnstr. Meth. 218 (1983) 249. [252] C.G. Pantano, J.F. Kelso and M.J. Suscavage, Mater. Sci. Res. 15 (1983) 1. [253) J.P. Landuyt, L. Vandenbroucke, R. de Gryse and J. Vennik, Surf. Sci. 12(; (1983) 598. [254) I.S.T. Tsoog, Mater. Sci. Res. 15 (1983) 39. [255) R. Kumar, J.A. Schultz and J.W. Rabalais, Chem. Phys. Lett. 97 (1983) 256. [256) H.J. I..ewerenz, O.E. Aspnes, B. Miller, O.L. Malm and A. Heller, J. Am. Chem. Soc. 104

(1982) 3325. [257] R.L. Chin and O.M. Hercules, J. Catal. 74 (1982) 121. [258] B.C. Rodrigo, H. Jeziorowski, H. Knözinger, X.Z. Wang and E. Taglauer, Bull. Soc. Chim.

Belg. 90 (1981) 1339. [259] J. Abart, E. Delgado, G. Ertl, H. Jeziorowski, H. Knözinger, N. Thiele, X.Zh. Wang and E.

Taglauer, Appl. Catal. 2 (1982) 155. [260] A. Shih and G.A. Haas, Appl. Surf. Sci. 8 (1981) 125. [261] H. Knözinger, H. Jeziorowski and E. Taglauer, Stud. Surf. Sci. Catal. 7 (1981) 604. [262] P.G. Fox, GlassTeelmoL 22 (1981) 67. [263] S.V. Krishnaswamy, L.L. Tongson, N. Said and R. Messier, J. Vac. Sci. TeelmoL 18 (1981) 401. [264) R. De Gryse, J.P. Landuyt, A. Vermeire and J. Vennik, Appl. Surf. Sci. 6 (1980) 430. [265] F. Delannay, E.B. Haeussler and B. Delmon, J. Catal. 66 (1980) 469. [266] W.L. Baun, Surf. Teclmol. 11 (1980) 385.

55

Chapter 2

[267] W.L. Baun, Surf. Technol. 11 (1980) 421. [268) H.H. Strehblow and B. Titze, Electrochim. Acta 25 (1980) 839. [269] F. Delannay, E.N. Haeussler and B. Delmon, Bull. Soc. Chim. Belg. U (1980) 255. [270] F. Delannay and E.N. Haeussler, Ned. Tijdscbr. Vacuumtechnol. 18 (1980) 38. [271] A.C. Milier and A.W. Czanderna. Appl. Surf. Sci. 4 (1980) 481. [272] W.L. Baun, Appl. Surf. Sci. 4 (1980) 374. [273] R.C. McCune, J.E. Cbel.gren and M.A.Z. Wheeler, Surf. Sci. 84 (1979) LS15. [274] M. Wu and D.M. Hercules, J. Phys. Cbem. 83 (1979) 2003. [275] R.C. McCune, Anal. Cbem. 51 (1979) 1249. [276) P.H. Holloway and G.C. Nelson, J. Vac. Sci. Tecbnol. 16 (1979) 793. [277] M. Nagasaka, R. Vasofsky and H.F. Helbig, J. Vac. Sci. Technol. 16 (1979) 151. [278] W.P. EUis and T.N. Taylor, Surf. Sci. 75 (1978) 279. [279] K. Akaisbi, A. Miyahara and A. Sagara, J. Nucl. mater. 76/77 (1978) 378. [280] R. Schubert, J. Electrochem. Soc. 125 (1978) 1215. [28\] A.C. Milier and A.W. Czandema, Vacuum 28 (1978) 9. [282] A.W. Czandema, A.C. Miller, H.H.G. Jellinek and H. Kachi, Ind. Res. 20 (1978) 62. [283] D.L. Cbristensen, V.G. Mosotti, T.W. Rusch and R.L. Erickson, Nucl. Instr. Method. 149 (1978)

587. ' [284] G.C. Nelson, J. Electrochem. Soc. 125 (1978) 403. [285] N.T. McDevitt, W.L. Baun and J.S. Solomon, J. Electrochem. Soc. 123 (1976) 1058. [286] R.P. Frankenthal and D.L. Malm, J. Electrocbem. Soc. 123 (1976) 186. {287] G.C. Nelson, J. Appl. Pbys. 47 (1976) 1253. [288) A. Emmanuel, G.B. Donaldson, W.T. Band and D. Dew-Hughes, IEEE Trans. Magn., MAG 11

(1975) 763. [289] R.A. Kushner, D.V. McCaughan, V.T. Murphy and J.A. Heilig, Pbys. Rev. BlO (1974) 2632. [290] R.F. Goff, J. Vac. Sci. Techno!. 10 (1973) 355.

56

Abstract

Quantification of Surface Composition using Low-Energy Ion Scattering*

Low-energy ion scattering is known for its extreme surface sensitivity and its facilitation of relative straightforward quantification of the surface composition using calibrated standards. The validity of quantification via calibration against pure metals relles on the fact that the ion fraction after scattering does not depend on the chemical environment of the atoms in the surface (no matrix effects). It is shown that calibration is possible for different alloys (A&oA-120, NisoP11o, NiAl(llO)) and oxides ('y-Al20 3, a-Al20 3, a-Al2~ sapphire(l102), NiO, 14 wt% Ni/-y-Al2~) containing Ni and Al. One should never, however, discount the possibility of matrix effects fully as is shown in the case of NiAl(llO). For their use as calibration standards the elemental sensitivity factors are presented for different metals (Cu, Rh, Ta, Ir, Ni Al, Si, Cr, Mo, Pd, Pt, W, and Re) using different projectiles eHe+, 4He+). For the same elements characteristic veloeities have also been measured for He+ in the 1000-3500 eV range. It is found that the characteristic velocity decreases with the filling of the d-band in the metal when studying the periodic system. Finally, the influence of surface roughness on the LEIS-signal intensity is quantified.

'This chapter has been hased on the following: J.-P. Jacobs, S. Reijne, R.J.M. Elfrink, S.N. Mikhailov, and H.H. Brongersma, J. Vac. Sci. Technol. Al2 (1994) 2308-2313; S.N. Mikhailov, R.J.M. Elfrink, J.-P. Jacobs, L.C.A. van den Oeteloar, P.J. Scanion and H.H. Brongersma, Nucl. /nstr. Meth. 893 (1994) 145-155.

57

Chapter 3

1. INTRODUCTION

LEIS is a surface sensitive technique that selectively probes the outermost atomie

layer [1.2]. A beam of mono-energetic inert-gas ions fHe+ or 4He+, 1-3.5 keV) is

directed onto the target. Since the energy toss of the scattered ion after a binary scattering

event at a given scattering angle depends on the mass of the target atom, the atomie

composition of the outermost atomie layer of the target can be derived from the energy

distribution of the scattered ions. The high neutralization probability and large scattering

cross section of the low-energy inert-gas ions ensure the monolayer sensitivity. The intensity (S;(E)) of the ions scattered from element i on the surface can be

described by

S;(E) = I • <T;(E) • Pt(E) • n; • eEr • R = 11;• (E) • 0; , (1.1)

where I is the primary ion current, <T;(E) is the differential cross section for scattering

from element i, eEr is an instromental factor taking into account experimental factors

including the transmission of the analyzer (where Er is the final energy of the ion), P;+(E)

is the ion fraction after scattering from element i, n; is the number of surface atoms of element i per unit surface area, R is a constant which takes into account the roughness of

the surface. 0 < R < 1 [3]. 71;"(E) is the absolute sensitivity factor of element i and 8; surface coverage of element i (in at. %) .

Direct quantification of the surface composition, i.e. n;, is only possible when all

the parameters in Eq. 1.1 are known. The differential scattering cross section can be

calculated with a screened Coulomb potential. for example, in the Molière approximation

of the Thomas-Fermi screening function [4]. The ion current and instromental factor can be determined experimentally. which leave the ion fraction P; + and the roughness. These

parameters are poorly known and hamper the direct quantification.

Different models for the ion fraction have been reported in literature. Models

[5.6.7] based on a parametrization of the neutralization rate with a functional form

rely on the fitting of a number of adjustable parameters by the experiment and appeared to be succesfut in descrihing the relative changes in ion fraction as a function of primary

energy. They are all similar to the functional form of the model introduced by Bagstrom

[5], and show dependenee of the interaction time of the ion with the surface in the

following form [7]: 1 1 • [-v •• ( - + -)]

P = e V; Vr , (1.2)

where v. is the characteristic velocity fora given ion-target combination, and V;, vr are

the respective ion veloeities before and after the collision. (Quasi-) ab-initio models

58

Quantijication of Surface Composition using Low-Energy Ion Scattering

[8,9,10], in which a quanturn mechanical description of the charge exchange is given have not, however, been developed sufficiently well to provide absolute numbers for neutralization probabilities. It is not within the scope of this work to fully comprehend the fundamental aspects of the charge exchange processes which determine the ion fraction.

Since only charged particles are monitored after scattering from a surface, the charge transfer processes, neutralization and (re-)ionization, play a key role in the p+ behavior. Fundamental aspects of the charge exchange processes between the He+ projectile and 29 elements have been studied by Souda and coworkers [11,12,13,14]. It was shown that neutralization can occur on both the incoming and outgoing trajectory of the ion, whereas (re-)ionization may take place during the close encounter. The threshold energy for reionization was measured for 29 elements using a He" incident beam. A measure for the ionization probability was estimated from the ratio of the normalized He+ intensity after scattering for He" incidence to that for He+ incidence. It was found that this ratio rapidly increases as the energy exceeds the tbreshold energy. Furthermore, the ionization probability seems to be suppressed as the number of d electroos from the target increases, but it increases steeply for the IIIb group elements.

O'Connor et al. [15] evaluated the role of various electrooie parameters such as workfunction and valenee band width on the neutralization rate during the outgoing trajectory. No clear correlation with workfunction, and some correlation with band width, was observed for the elements studied. It was also found that the characteristic velocity v. is proportional to the final velocity of the ions in a certain primary energy <Ft,) interval. In Refs. [16,17,18] it bas been reported that, when the total trajectory is considered, v. is constantintheEp-range 1-5 keV fora variety of elements.

The quantification through calibration relies on the fact that the ion fraction P; + does notdepend on the chemical environment of element i (no matrix effects). The matter about possible matrix effects is controversial. In alloys, Novaeek and Varga [19] and Ackermans et al. [23], amongst others, reported the absence of matrix effects in NiPt and CuPd alloys. The energy dependenee of LEIS-signals on the primary energy for oxides, which was first attributed to strong matrix effects [20], was proven to be an intrinsic property of the target atoms and not influenced by the oxidation [21]. The adsorption of oxygen on TaNb-alloys and Ni(lOO) also did not show any indication of matrix effects [22]. However, recent results on different carbon species show that for graphitic carbon the sensitivity for 2 keV incident 4He+ and a scattering angle of 136" is about 240 times lower than for carbidic carbon [16]. At present, the status of the theory and the restricted number of experimental data do not allow the general prediction of the presence or

59

Chapter 3

absence of matrix effects in low-energy ion scattering.

Since neutralization and reionization together determine the charge state of

projectiles, and since their relative contributions to the ion fraction are still poorly

understood, more experimental data is needed. Consequently, other purposes of this work

were to determine the elemental specific characteristic veloeities for a number of metals

and to study their dependenee on the electronic structure of the target. Generally quantitative analysis of alloys, oxides or different absorbate/substrate

systems is based on ciilibration against standards (pure elements). For this purpose, the second part of Eq. 1.1 can be used [1,2,23]. The availability of elemental sensitivity

factors is evidently necessary for practical applications, therefore, a number of relative

sensitivities for different metals (Cu, Rh, Ta, Ir, Ni, Al, Si, Cr, Mo, Pd, Pt, W, and

Re), using different projectiles eHe+, 4He+) at different primary energies within the

1000-3500 eV range, have been determined and compared to values in literature [24,25].

A simple test can be performed to check the validity of the calibration by measuring the calibrated surface composition as a function of primary energy of the prohing ions. In this way the influence of the interaction time (i.e. the charge-exchange processes) can be probed. The present work is concentraled on Al because of its

numerous applications as a metal, in alloys, ceramics and catalysts. Samples containing

Ni are also studied for comparison. The aim of this work is to determine the presence or

absence of matrix effects in LEIS for Al and Ni. Therefore, Al and Ni have been studied

in various chemical environments (metal, alloy, oxide).

The surface analysis of these microscopically-rough surfaces encounters all kinds

of problems. It was reported by Margraf et al. [26] that for different aluminum oxides the LEIS signal decreased by more than a factor of 5 due to surface roughness, when studying different aluminas. Calculations of Nelson [27] of the influence of surface

roughness on the intensity of LEIS-signals predict a decrease by a factor of about 1. 7

when introducing roughness. In the present study the influence of the microscopie

roughness on the LEIS signal is studied by oomparing optically flat surfaces with very

rough powders used in catalysis.

2. EXPERIMENTAL

The LEIS experiments were performed using the low-energy ion scattering set-ups,

NODUS and MINl-MOBIS. Their basic design is the same and bas been described in

60

Quantification of Sulface Composition using Low-Energy Ion Scattering

more detail elsewhere [28]. The primary ions CHe+, 4He+) are generateel in a Leybold

ion souree (type IQE 12/38) and are directed perpendicular onto the target. The scattered

ions are energy selected by a modified [28] CMA. The scattering angles are 142"

(NODUS) and 136° (MINI-MOBIS). In the NODUS apparatus charging of insu1ating

samples can be effectively compensated by flooding the surface from all sides with low­

energy electrons. The MINI-MOBIS is equipped with a sputter ion souree (Leybold, type

IQE 12/38) at grazing incidence (15"). Oxides are measured in the NODUS, while most

alloys are studied using the MINI-MOBIS. In both machines there is a facility to heat the

samples by indirect heating up to 1000 K. For the high-melting metals resistive heating was used, which enabled annealing close to the melting point. The nominal base pressure

in the MINI-MOBIS is in the low 10'10 mbar range, while in the NODUS a base pressure

in the low 10·9 mbar can be maintained. During the operation the pressure wiJl increase in

both setups to the high 10·9 mbar range. This increase is due to the inert gas from the ion

source, which will not affect the measurements.

The following samples were used in the analysis: Cu, Al, Ag, Cr, Rh, Ta, Ir, Pd,

Mo, W, Pt, Re (polycrystalline), Si(lOO), Ni(lOO) and NiAl(llO) single crystàis [29],

NisoPtw (polycrystalline), AgsoA-120 (polycrystalline2 [30], -y-Alz03 (AKZO, powder), a.­Al203 (Fluka A.G., powder), a.-Al20 3 sapphire(l102) (Crystal Systems), NiO (Johnson Matthey GmbH, powder), 14 wt%-Ni/-y-Al:P3 (catalysts prepared by ALE [31],

powder).

The metals and alloys were cleaned by sputter-anneal cycles until judged clean by

LEIS, i.e. typically less than the detection limit of 1% and a few percents of a monolayer

of oxygen and carbon respectively. However, in the case of Ta it was rather difficult to

keep a clean surface in vacuum fora long time due to the high reactivity to oxygen. To

obtain a well-structured surface [32,33] the sapphire single crystal was heated in 1 bar oxygen at 1673 K for 24 hrs and then transported through air to the experiment.

However, the surface was then contaminated due to alkali diffusion from the bulk.

Therefore, these oxide surfaces were Ar-sputter cleaned prior to analysis. All powders

were pressed into wafers and prior to measurement chemically cleaned by a calcination in

20 mbar of oxygen at 550 K in a connecting chamber.

The metal polycrystalline samples were measured using 1.0, 1.5, 2.0, 2.5, 3.0 and 3.5 keV incident 3He+ and 4He+ probe ions under identical scattering conditions. A

current density of approximately 3 ·10-6 A/cm2 was used.

In this study Al20/NiAI(110) serves as a suitable model system for a smooth

aluminum oxide surface in comparison to the smooth sapphire and rough oxide powders.It

has been reported in literature [29] that a very thin alumina film ( ""'5 Á) can be grown on

61

(i) 10 ö

T'"" -~ ëi.i r.:: ~ 5 r.:: en w ...J

Chapter 3

Al y-AIPa sapphire

- Ni/y-Aip3

-- AI20JNiAI(110)

Ni

1000 1500 2000 2500

(a)

30

~ ('b 20

>--"(i) r.:: (l) -r.:: 10 en w ...J

0

Fig. 1 (b)

Final energy (eV)

- NiAI(110) -- Ni80P~ ······ AgaoA120

1000

Al

2000

Ni I Jl jl

11

Final energy (eV)

Typical LEIS spectra using 3 keV 4He+: a) oxides,

62

Ag

3000

b) alloys.

Quantijication of Surface Composition using Low-Energy Ion Scattering

a NiAI(IIO) alloy after a saturation with oxygen ( > 1200 L). This system is used in

catalysis as a model system for the alumina support.

3. RESULTS AND DISCUSSION

3.1 Introduetion

Figs. la,b show typical LEIS spectra of the different systems. The spectra shown were all measured using a 3keV 4He+ ion beam. Less than 10 % of a monolayer was

removed to collect one spectrum. Spectra presented in one tigure can be compared on an

absolute scale since the experimental conditloos were kept constant. For the alloys and metals where the backgrounds were relatlvely small, a linear background subtractlon was

used to quantlfy the surface peak areas. For the oxides, where the backgrounds are more

significant, a careful fit and deconvolution of peaks was performed, based on the metbod

introduced by Nelson [34] and adapted for our purpose by Drimal and Koksa [35].

3.2 Alloys

For the alloys and oxides the LEIS signal intensities are normalized to the signal of the pure metal samples. In this way that the signals do not bave to be corrected for

experimental factors such as the transmission of the CMA or the efficiency of the

ehanneltrons used in the detectlon. The results for the Al signal ratlos can be found in

Fig. 2a. The Ni signal ratlos are shown in Fig. 2b.

The surface coverages of the alloys as a funetion of primary energy are presented

in Fig. 3. The signals in Fig. 3 have been corrected for the atomie densities in the

different samples, where for the polyerystalline materials the molair volume (V..,) was used to determine the atomie surface density ((11V..,)213). If (for flat surfaces) no matrix

effects are present, the ion fraetion p+ does not depend on the ehemical environment,

and thus the signal is only dependent on the coverage and a calibrated sensitivity factor.

The LEIS signal of the pure element at that energy (corrected for the atomie density) is

the absolute sensitlvity factor defined in Eq. l.I. From the LEIS signals of the alloy and

the sensitlvity factor, the absolute coverage can be caleulated. If no matrix effects are presentand Vegards law holds, the coverages of the different components must add up to

100%.

For NisoPtzo and AggoAI2,~> it is shown in Figs. 3a,b that the calculated

63

Fig. 2

60

0

(a)

80

60 ~ 0 -ê

~40 z Cl) -.,._

z Cl)

20

0

(b)

Chapter 3

--AgsoA120

0----- __ o

<l x 1.5 ,;i- - --<1- - .:51.. - - - - - -<> oxidized NiAI <1 <J

~ .. ~ .. :-:.:-:.r:~ .. :-:.:-: .. ~-~.:-: .. ~ .. :-:g~.-~ .. :-:.:-: <>- ...• :-:. ~z sapphire 0

v~ ~ ~Y ----"'--J -~ 1

1

.alumin: suppo~s .L

1000 2000 3000

0

Primary energy (eV)

NiAI

-----ä- -1!--~-

x 20 14 wt% NVy-A1p3

NiO

Cl

(7- -o---o- .o-- _o

1000 2000 3000 Primary energy (eV)

LEIS signa/ intensity ratios as a function of primary energy: a) Aluminum LEIS signa/ intensity ratios for the alloys (NiA.l(l/0), Ag8()"1lzo), smooth oxides (sapphire, oxidized NiA/(110)) and the powdered catalysts (a-Alp3, "'(-Alp3, Nii"'(-Alp~. b) Nickel LEIS

signal intensity ratios for the alloys (NiAl(I IO), NisJ>tzJ, powdered oxides (NiO, Nii"'(­

Alp~.

64

Quantijication of Surface Composition using Low-Energy Ion Scattering

compositions do not depend on the ion energy. The results for the Ni8J>tw alloy confirm

the results obtained by Novaeek and Varga [19] that no matrix effect is present. At every

primary energy the results are reproduced quantitatively and the coverages add up to

100%, which means that in this system the ion fraction is not influenced by the alloying.

This is also true for AgsoAJ20 and a1so holds for the experiments with 3He+. The

neutralization behavior is expected to be the same (equal v.) for the two He isotopes, the

only difference being the velocity of the incident and scattered ions [36]. In the NiaoPt20

alloy a surface enrichment of Pt is fonnd, which is due to preferendal sputtering during

the Ar+ bombardement cleaning procedure [19] or surface segregation [37]. The

surface concentradons of Ag8oAJ20 reflect the bulk values in accordance with Dirks and Brongersma [30].

So far the alloys seem to behave in a simpte way: there are no matrix effects. The

results on the NiAl alloy (see Fig. 2a for the Al signal and Fig. 2b for the Ni signal)

show that the calibrated Ni signal is not dependent on the primary energy while the calibrated Al signal decreases with increasing primary energy, thus the surface coverage

could not be determined from the Al signals. When only the Ni-signal ratio is taken,

which does not change with primary energy, a surface concentradon of Ni of about 43

at.% could be determined. This is close to the value of 50 at.% reported in literature e.g.

Ref. [38], where it is stated that the NiAl(llO) surface is bulk truncated.

One could imagine that Al in the alloy would be shielded preferentially by a

contamination (if present) during the measurement. However, no contamination was

detected which can account for this effect, so the decrease cannot be assigned to an

artefact. Another possibility would be that, because NiAl(llO) is an open structure, one

could expect contributions from deeper layers. However, this also would lead to an

increase of the calibrated Al signal when going to higher energies. Furthermore, NiAl

truncates the bulk structure at the surface [38], which means a 1: 1 composition at both

the fust and second atomie layer. Therefore, both metal signals would have to be affected

equally.

A possible explanation is a change in the neutralization behavior of the He ions on

the Al. The explanation is not straightforward. The change of the neutralization behavior

might be related to the change in the electrooie structure from Al-pure to NiAl-alloy. In

references [39,40] it is shown that, for the surface, the total density of states (DOS)

of NiAl(llO) exhibits a tilled d-band. The calculations imply that, effectively, electroos

are transferred from nickel to aluminum. In the alloy the Ni d-band is filled. However

there is a larger transfer of the sp-electrons of Ni to the Al. The strong hybridization of

the sp-electrons in the alloy, which makes the electrooie structure of the aluminum more

65

Chopter 3

r···ö-~H-ë!

total 3 ' • He:

<> - 100 <> • .. ~ -···-

~ e • • 0 .....; l1l - 80 (J) Ag 0) l1l .. Q ........ .. 0 . ···------~·-········-~--------···e. ........... ö .... ... ~ 60 • 0 (.)

Al (J) 4Ö (.) • l1l -o--0

--11-- - - -11- - - -ii- - - .LJ_

't: 0

::I 20 (J)

0 1000 2ooo 3000

(a) Primary energy {eV)

;? 100 0

total 0

[ ___ ·_~--~~·;_·j _<> ____ ~"------------~--~<>~ . ~ .

rJ - 0

80 .. 2 ........... ~------------.----------------········i ........... o .. .

Ni •

60

:f_o ___ + __ : __ ~- __ -.- ___ o_ 0 _ I , I , I

1000 2000 3000

(b} Primary energy (eV)

Fig. 3 The determined sulface coverage presented as a j'unction of the primary energy show the absence of a matrix effect on (a) Nisof't1f) and (b) Ag8{j'il:zo. Also the surface coverages add up to 100% without any normalization.

66

QuantiflcaJion of Surface Composition using Low-Energy Ion Scattering

Nî-like. This is not expected for the other alloys. However, why the obtained value for Al at lower primary energies is closer to the expected 50 at.% then at higher energies is still unclear. More thorough and detailed theoretica! and experimental investigations, currently

underway [41], are needed to give a more conclusive answer on this subject.

3.3 Oxides

Some typical LEIS spectra of the measured oxides are shown in Fig. la. From the spectrum of the oxidized NiAl, one can distinguish the Al and 0 peaks originating from the scattering by the alumina layer, which coveres the NiAl alloy surface. The background which extends to about 2250 eV is related to ions which have penetraled the

bulk, where they are neutralized, and after back scattering have been reionized prior to

leaving the surface. Because the background starts just below the energy position of the Ni single scattering peak, it can be concluded that Ni atoms can be found just below the surface layer. This is in agreement with the models of the oxidized NiAl reported in lirerature [29].

For the oxides Fig. 2 shows that there is no change in the LEIS signal intensity ratios as a function of the incident energy. From this it can be concluded that the neutralization behavior as a function of the incident energy for the metal (Al polycrystalline), the Al in the alloy A~20, and in the oxides (powders, and smooth surfaces) is the same. So although going from metal to oxide, where a large difference in work function and electronic configuration will occur, the signal ratios do not change as a function of incident energy. This is an essential demand in discarding possible matrix effects. Futhermore, it is considered unlikely that, when strong matrix effects would exist in the oxides compared to the scattering from metals, the functional behavior with increasing primary energy of the He ions would be unaffected.

For the flat aluminas 20 % of the Al LEIS intensity with respect to pure Al is

measured. Assuming bulk terminating surface the aluminum densities in the sapphire as well as the oxidized NiAl(llO) differ from that in the polycrystalline Al sample by a factor of between 1.5 and 3.8 depending on the model of the surface. For the alumina two possible explanations can be considered: coverage of the aluminum atoms by an oxygen overlayer or a change in neutralization of the ions (matrix effects). The last explanation seems not to be valid since the signal of the Al as a function of the primary energy is not affected by the change from metal to oxide, as is shown in Fig. 2 and Ref.

[21]. This leaves the effective shielding of the Al atoms by oxygen or, to put it a different way, the Al concentration on the topmost surface layer does not reflect the bulk structure.

67

Chopter 3

101 10"1

•Pd Cu Rh• Re - 10° .I -- -Ta.,. 1 o-2 -.

.si,~~/- Î -.-Mo ::) ..... Wlr (/)

..Q Al•,' N"" ..... a o<C as -- 0 r::: 1 o-1 •Si* 10"3

10"2 104

0 10 20 30 40 50 60 70 80

(a) atomie number Z

101 1 o-1 .Pd Pt

•Rh • Re

N1._ • Mo Ta_." • - 10° Cu --- W Ir 1 o-2 -::) SJ '_- -~

..... (/)

..Q . / N"" ..... Air' a as o<( - -r::: 1 o-1 10"3 0 • Si*

10·2 1 o-4

0 1 0 20 30 40 50 60 70 80

(b) atomie number Z

Fig. 4 Relative elemental sensitivity factors at 1 keV for 3He+ (a) and 4He+ (b) at notmal incidence. 1he two values indicated for the single crystal Si are for two different assumed surface densities. 1he scattering angle is 136". 1he dotted line indicates the differential cross section.

68

Quantijication of Surface Composition using Low-Energy Ion Scattering

However, the fact that shielding of the aluminum atoms is still observed for a sputtered

alumina surface is not evident. Since aluminum oxides have not been shown to be reduced under ion bombardment [42], and the oxygen atoms can be assumed to be mobile under

ion bombardment, the surface composition found is plausible.

3.4 Metals: stamlards and characteristic veloeities

A series of metals bas been studied to determine the elemental sensitivities and the

characteristic veloeities which provides new data for the evaluation of theoretical models.

Figs. 4a,b presents the relative sensitivities for the measured elements at 1 keV primary

energy for 3He+ (a) and 4He+ (b). The relative sensitivity factor (in contrast to the

absolute values used insection 3.2 for the alloys, was defmed as:

S; neu • [Ef]Cu 17; = -=----=-=-

SCu n; • [Ef]; (3.1)

where n; is the number of atoms per unit surface area; for single crystals the density was

determined according to the known surface structure, for polycrystalline targets the

density was determined using the volume of the atoms. The instromental factor c was

taken out by taking all sensitivities relative to Cu. Sensitivity factors have been measured

previous1y by other workers [24,25], and similar trends are observed.

The use of the sensitivities presented in Fig. 4 for the quantification of the surface

composition bas been nicely illustrated by Ackermans et al. [23] for CuPd alloys. In a nut

shell, when one considers an alloy with only two elements, A and B, no contaminations

and keeping Vegard's law in mind, one can derive the following expressions for direct

quantification of the surface coverage in atomie percent from the LEIS intensities of the

aHoy using the relative sensitivities from Fig. 4:

(3.2)

(3.3)

In this way the determined sensitivity factors (see Fig. 4) can provide a reference for

future experiments without doing again the careful calibration as performed in section

3.2. In the same figure are shown the differentlal cross-sections u;(E) calculated using

the Molière approximation to the Thomas-Fermi potential. The general behavior of 17;(Z)

appears to be mainly determined by the behavior of u;(E). Deviations from u;(E) are

69

Chapter 3

attributed to differences in the ion fractions. To obtain the characteristic veloeities the logarithm of the LEIS signal divided by

cross section, target current and fina1 energy is plotted for a number of elements as a tunetion of the sum of the reciprocals of the initia! and fina1 velocity of the ions. A linear relationship was found for practically all targets in agreement with Eqs. 1.1 and 1.2 as is shown in Fig. 5. For Cu, some deviation from straight line behavior was observed. The nonlineacity may be associated with changes in the relative contributions of different mechanisms to the neutralization. The determination of v. for elements which exhibit straight line behavior within this narrow energy interval is straightforward.

Fig. s

-""':;::: w t)

:::::. tiJ -c

11

10

9

8 0.40 0.50 0.60 0.70 0.80 0.90 1.00

1/V1+1/Vt (* 10·5 S/m)

Determination of vc for the metals from the logarithm of the scattered ion signa! divided

by beam current, differential cross section and jinal energy versus the sum of the

redprocal veloeities before and qfter scattering; for clarity the lines have been shifted

vertically. Results using 3He+ and 4He+ ions have been indicated by open and closed

symbols, respectively.

70

Quantification of Suiface Composition using Low-Energy Ion Scattering

Fig. 6 shows the differences of the characteristic veloeities with respect to the value for Pd (the lowest measured value for vJ. Differences in the characteristic velocity

are more easily determined than absolute values. For Si and C [16] the absolute values of

v. were measured to be 4.8±0.4 ·lOS m/s and 10± l·lOS m/s respectively, compared with values of 6.4± l·lOS m/s and 11 ± l•lOS m/s calcuJated from the ion ftaction data of Bhattacharya et al. [43]. There is an evident correlation with the filling of the d-band. As is clearly shown in Fig. 6. This is similar to the results of Souda and coworkers [12,13,14] whohave found that the ionization probability is suppressed as the number of target d electroos increases. No suppression of ionization was found for s- and p-orbitals of the target. The correlation between our v. data (which are associated with p+ at a

given energy interval) and the reionization probability measured by Souda et al. [14] can be seen in Fig. 6. Tsuneyuki et al. [44,45] have explained the reïonisation process on the basis of level crossing. At close distances, (avoided) crossing of the levels provides the possibility of charge transfer: He0 +MeHe++M·. If this channel is open, the reverse process (He++M-eHe0+M) should also be possible. Level crossing, thus, provides the possibility for both neutralization and (re-)ionization at the close encounter. The possibi­

lity for level crossing is found todependon the filling of the d-band [44]. The threshold

energies calculated by Tsuneyuki et al. [44] are in good agreement with the experimental results of Aono and Souda [12,13] and our work [46]. Other electrooie properties of the target, such as workfunction and bandwidth did not show a clear correlation with v. values [46], although Van den Oetelaar and Flipse [10] claim, using our v. data and the workfunctions by Michaelson [47], that a correlation does exist. The fact that the v. of oxides such as Al20 3, ZnO and Si02 (see Ref. [21] and section 3.3) are the same within experimental error, makes it difficult to onderstand that the workfunction would be

responsible for the observed behavior.

3.5 Surface roughuess

Both flat and rough alumina surfaces have been measured. The results of the study of the roughness can be extracted from Fig. 2a for the aluminas. The measured AllO signal ratlos are summarized in Table 1. From this it can be concluded that all spottered

alominas have essentially the same ratio, which is not too surprising for solids whose long

range order has been destroyed by the sputtering and which have the same bulk stoichiometry. Also, since it is not expected that the low sputtering dose will influence the microscopie roughness, one can use the Al signals of the spottered surfaces to quantify the influence of the surface roughness, see Table 1.

71

Chapter 3

The influence of roughness on the LEIS signals, calculated by Nelson [27] (Fig. 7,

geometry A), is adapted to our geometry of normal incidence and backscattering (135°). The physical screening by a two dimensional surface of tilted cubes is calculated since no

change of neutralization is expected. A normal distribution for the possible orientations of

the cubes is used following Nelson. The geometry (B) is shown in Fig. 7. The shielding

on a tilted surface covered by spheres (C) is also calculated. Finally geometry B and C are combined in D, tilted cubes with a surface of small spheres. When the spheres are

considered to be small compared to the cubes, only the geometry determines the shielding

1. 11. m.tv. v. vt.vu. VIII lb 11" lllb IV., Vb VlbVIIbO 10° • 4

• • • a ~ • :!::::: 3 -1 o-1 • • • • en ·- • -.c e Cl E CU • .c Cl T • • 10 0 a 2 0 ... 1Q·2 1 • '!"'" c. • .. -c 0 -0 • • "CC ;:: T a.

0 u CU 1 o-3 0 > N 6 é • I

u c 6 • 0 > 0 • -

10-4 I filling of d-states I

·1 0 • K CaSe Ti V Cr MnFeCo NI CuZn Ga GeAs SeBr Kr /:;. .. RbSr Y Zr NbMoTcRuRh PdAgCd In Sn Sb Te I Xe

0 • CsBa Lu Hf Ta W Re Ir Pt Al SI

Target elements

Fig. 6 Comparison between ionization probabilities measured by Souda et al. [14] (closed symbols) and characteristic velocity differences (increasing vc corresponds to increasing neutralization) measured in this work (open circles).

72

Quantification of Surface Composition using Low-Energy Ion Scattering

not the dimensions. The geometry of the surface roughness is considered to be the same

in all azimuthal directions. The calculations show a decrease of about a factor of two at a

rms slope of 40" (see Fig. 7).

The measurements on the very rough powdered wafers and the smooth single

crystals (Fig. 2) are in agreement with the calculations, using a simple physical shielding

model. Of course this does not imply that our simple model is a good representation of

the rough surface. But the combination of the physical shielding and the increasing

density when the surface is tilted shows a decrease of a factor of two in the worst case. This is found for a number of different systems including alumina, titania and silica.

From these results, it follows that the roughness influences the LEIS intensities but

the effect is not dramatic. There is no significant influence of the surface roughness

between the a-Al20 3 (5.5 m2/g) and -y-Al20 3 (269 m2/g), thus the large difference in

specific surface area is not reflected in the LEIS-intensities at the used large scattering

angles (136° and 14n.

This investigation did not confirm the results of Margraf et al. [26] who reported a decrease by a factor of at least five due to roughness using a scatttering angle of 131>. A

possible explanation can be that other factors which can have an influence on the signal

were not effectively dealt with, such as charging of the insulating material when an ion

beam is directed on it.

3.6 Surface composition of powders

The surface compositions of the different aluminum oxides exhibit different Al/0 ratios, as is shown in Table 1. lt is therefore possible to distinguish the -y-A~03 from a­Al203. Van Leerdam [48] points out that the difference in the AllO ratio between the

two alominas can be described quantitatively in terms of the exposure of specific

crystallographic planes of the alumina crystallites. The surface of the -y-A120 3, a defect

spinel, could be assigned to the 0(110) spinel surface plane. This plane contains only

octahedrally coordinated alnminurn cations in addition to the oxygen anions. The results

are in agreement with recent studies on other catalytically active spinels [35]. The fact

that LEIS can provide quantitative information offers new possibilities to check the

validity of the surface stoichiometry of a model system. The oxidized NiAl(llO) for example shows an alumina overlayer which contains more Al than is found in -y-Al20 3,

and even more than for a-Al20 3• An explanation would be that the thin oxide layer is still

-y- Al20 3 but a different plane is exposed in the powder from that in the single crystal

surface, due to pinning of the crystal structure by the underlying NiAl alloy. Efforts to

73

Fig. 7

1.0

0.9

.c 0.8

~ -•o.1 "'"J:

C) :::J 0

_ ... 0.6

0.5

Chapter 3

0 10 20 30 40 <P (rms slope in degrees)

Reduction in the intensity due to shadowing ~ a function of the rms slope of the surface

for different geometries: A: tilted cubes (45" incidence, 9(}' scattering [27]), B: tilted cubes (normal incidence, 135" scattering), C: tilted surface of spheres (nonna/ incidence,

135° scattering), D: tilled cubes covered with spheres (normal incidence, 135" scattering).

74

Quantification of Sulface Composition using Low-Energy Ion Scattering

use different faces of a-A120, sapphire single crystals to study the preferential exposure of one of the low-index planes in the powder a-alumina were unsuccessful. No alkali-tree sapphire surface could be obtained, after annealing at the required annealing temperature [32,33] of 1673 K in oxygen.

Before sputtering After sputtering Roughness (R)

S,./S0 SAJ!So sAlrooghtsAI--

a-A}a03 sapphire(1102) 2.5 1.0

2.3 2.5 0.6

1.7 2.5 0.6

2.8

Table 1 AllO LEIS signal intensity ratiosjor the different aluminum oxides (3 keV 4He+).

4. CONCLUSIONS

The LEIS experiments on the alloys Ni8oPt2o, AgsoAJ20 and NiAl(llO) showed that the neutralization behavior of the scattered He ion was not affected by the alloying in the

NisoPtzo and AggoA120 alloys. With NiAl(llO) the significant deviation for the Al signal ratios was interpreled as a change in the neutralization due to the matrix. The euergy dependenee of the observed compositions is a strong support for quantitative LEIS of oxides. The surface composition of the aluminum oxides determined by LEIS show less alnminurn than expected from bulk densities or surface models. When oomparing a metal

(Al or Ni) to an alloy (AggoA120 or PtsoNi20) and an oxide (A120 3 or NiO) no matrix effects

could be determined. One should, however, never discount the possibility for such effect fully, as is shown in the case of the NiAl(llO).

A series of metals (Cu, Cr, Si, Al, Rh, Re, Pd, W, Ta, lt, Pt, Ni) has been characterized and elemental sensitivity factors were measured for for 1.0, 1.5, 2.0, 2.5, 3.0, 3.5 keV incident energies for 3He+ and 4He+. This can be used as a reference for the quantifïcation by LEIS of the surface composition. For the same elements the elemental specific characteristic veloeities (v.) for the overall trajectory have been determined. For the given scattering conditions v. decreases with the filling of the d-band in the elements.

75

Chopter 3

The (re-)ionization and neutralization at a close encounter seem to be associated with one

another. The characteristic veloeities provide new data to be able to evaluate theoretical

models in future. The results on the influence of the roughness show that when the scattering

conditions are kept constant, different powders can be compared on an absolute scale

without correcting for the exact surface roughness. This greatly facilitates the absolute comparison of LEIS results from different powders, e.g. when studying catalysts of different pretreatment or loading. One should, however, prepare the wafers reproducibly.

LEIS can provide quantitative comparison of the surface composition between oxides,

which offers new possibilities to check the validity of the surface composition of a model

system in comparison to the powder systems used in for example catalysis.

Concluding, on the one hand the whole picture is not as straightforward as was

implied in the beginning of ion scattering but on the other hand the influence on the

quantification by the matrix is in most cases negligible. This explains why the technique, despite its anomalies and the poor understanding of the charge exchange in the scattering process, proved a reliable and quantitative surface analysis technique. In the rest of the

thesis this will be explored further.

Relerences

[1] H.H. Brongersma and G.C. van Leerdam, in FWIII.lmtental Aspects of Heterogeneaus Cmalysis studied by Partiele Beams, Eds. H.H. Brongersmaand R.A. van Santen, NATO ASI Series 8 265 p. 283, Plenum Press, New York, 1991.

[2] H. Niehus, W. Heiland, E. Taglauer, Surf. Sci. Rep. 17 (1993) 213. [3] W.L. Baun, in Quantitmive Surface Analysis of Mmerials, Ed. N.S. Mclntyre, p. 150, American

Society for Testing Matmials, 1978. [4] G. Molière, Z. Naturforsch. 2a (1947) 133. (S] H.D. Hagstrum, in lnelastic lon Surface Collisions, Eds. N.H. Tolk, J.C. Tully, W. Heiland and

C.W. White, p. 1, Academie Press, New York, 1977. [6] D.J. Godfrey and D.P. Woodruff, Surf. Sci. lOS (1981) 438. [7] G. Verbist, J.T. Devreese and H.H. Brongersma, Surf. Sci 233 (1990) 323. [8] E.C. Goldberg and F. Flores, Phys. Rev. 845 (1994) 8657. [9] H. Shao, D.C. Langreth and P. Nordlander, Phys. Rev. B49 (1994) 13 948. [10] R.J.A. van den Oetelaar and C.F.J. Flipse, Phys. Rev. Lett., submitted. [11] R. Souda, T. Aizawa, C. Oshima, S. Otani, Y. Ishizawa, Surf. Sci. 194 (1988) Lll9. [12] M. Aono, R. Souda, Nucl. lnstr. Meth. 827 (1987) SS. [13] R. Souda and M. Aono, Nucl. Instr. Meth. 815 (1986) 114. [14] R. Souda, T. Aizawa, C. Oshima and Y. lshiwa, Nucl. lnstr. Meth. 845 (1990) 364. [lS] D.J. O'Connor, Y.G. Shen, J.M. Wilson and R.J. MacDonald, Surf. Sci. 197 (1988) 277. [16] L.C.A van den Oetelaar, S.N. Mikhailov and H.H. Brongersma, Nucl. lnstr. Meth 885 (1994)

420.

76

S.N. Mikhaitov, L.C.A. van den Oetelaar and H.H. Brongersma, Nucl. lnstr. Meth. 893 (1994) 210.

Quantification of Surface Composition using Low-Energy Ion Scattering

[17] T.M. Buck, W.E. Wallace, R.A. Batagiola, J.B. Rothman, R.J. Gorte and J.G. Tittensor, Phys. Rev. B48 (1993) 774.

[18] H.H. Brongersmaand T.M. Buck, Nucl. Instr. Meth. 132 (1976) 559. {19] P. Novaeek and P. Varga, Surf. Sci. 248 (1991) 183. (20] R.C. McCune, J. Vac. Sci. Technol. 18 (1981) 700. {21] G.C. van Leerdam and H.H. Brongersma, Surf. Sci. 254 (1991) 153. [22] LC.A van den Oetelaar, J.-P. Jaoobs, M.J. Mietus, H.H. Brongersma, V.N. Semenov and V.G.

Glebovsky, Appl. Surf. Sci. 7Gnt (1993) 79. (23] P.A.J. Ackermans, G.C.R. Kndzen and H.H. Brongersma, Nucl. Instr. Meth. B45 (1990) 384. [24] J.S. Brinen, D.A. D'Avignon, E.A. Meyers, P.T. Deng and D.W. Behnken, Surf. Interf. Anal. 6

(1984) 295. [25] E. Taglauer, Appl. Phys. A38 (1985) 161. [26] R. Margraf, H. Knözinger and E. Taglauer, Surf. Sci. 211/212 (1989) 1083. [27) G.C. Nelson, J. Appl. Phys. 47 (1976) 1253. [28) H.H. Brongersma, N. Hazewindus, J.M. van Nieuwland, A.M.M. Otten and A.J. Smets, Rev.Sci.

Instrum. 49 (1978) 700 and Cbapter 2 of this thesis. [29) M. Wuttig, W. Hoffmann, R. Jaeger, H. Kuhlenbeck and H. Freund, Mat. Res. Soc. Symp. Proc.

221 (1991) 143. R.M. Jaeger, H. Kuhlenbeck, H.-J. Freund, M. Wuttig, W. Hoffmann, R. Franchy and H. Ibach, Surf. Sci. 259 (1991) 235.

[30] A.G. Dim and H.H. Brongersma, J. Electrochem. Soc. 127 (1980) 2043. [31] J.-P. Jacobs, L.P. Lindfors, J.G.H. Reintjes, 0. Jylhi and H.H. Brongersma, Catal. Lett. 2S

(1994) 315. [32] L-M. Peng and J.T. Czernus:dc.a, Surf. Sci. 243 (1991) 210. [33] T. Hsu and Y. Kim, Surf. Sci 258 (1991) 119.

Y. Kim and T. Hsu, Surf. Sci 258 (1991) 131. [34] G.C. Nelson, J. Vac. Sci. Techn. A4 (1986) 1567. [35] 1. Drimal and V. Kok:sa, to be published. [36] P.A.J. Ackermans, M.A.P. Creuwels, H.H. Brongersmaand P.J. Scanlon, Surf. Sci. 227 (1990)

361. [37] Y. Ganthier, R. Baudoing-Savois,J.J.W.M. Rosink, and M. Sotto, Surf. Sci. 297 (1993) 193. [38] D.R. Muilins and S.H. Overbury, Surf. Sci. 1!)!) (1988) 141. [39] s.-c. Lui, M.H. Kang, E.I. Mele, E.W. Plummer and D.M. Zelmer, Phys. Rev. 839 (1989)

13149. [40] S.-C. Lui, J.W. Davenport, E.W. Plummer, D.M. Zelmer, G.W. Fernando, Phys. Rev. B42

(1990) 1582. [41] E.C. Goldberg, private commuuication. [42] S. Hofmann and J.M. Sanz, J. Trace and Mieroprobe Techniques 1 (1982-83) 213. [43] R.S. Bhattacharya, W. Eckstein and H. Verbeek, Surf. Sci. 93 (1980) 563. [44] S. Tsuneyuki and M. Tsukada, Phys. Rev. 834 (1986) 5159. [45] S. Tsuneyuki, N. Sbima and M. Tsukads, Surf. Sci. 186 (1987) 26. [46] S.N. Mikhailov, R.J.M. Elfrink, 1.-P. Jaoobs, L.C.A. van den Oetelaar, P.J. Scanion and H.H.

Brongersma, Nucl. Instr. Meth. B!)J (1994) 149. [47] H.B. Michaelson, J. Appl. Phys. 48 (1977) 4927. [48] G.C. van Leerdam, PhD. Thesis, Eindhoven Uuiversity ofTechnology, The Netherlands, 1991.

77

78

The Surface of Catalytically-Active Spinels *

Abstract

By substitution of Mn and Co cations in different sites in the spinel strueture of Mn30 4

and C~04 by cations whieh are not active in the selective rednetion of nitrobenzene to nitrosobenzene, the role of these sites in the catalytie reaction can be studied. By combining aetivity measurements with quantitative surface analysis by LEIS, whieh only probes the outermost atomie layer, the catalytic performance can be related to the surface composition. The results suggest that octahedral sites are exposed almost exclusively at the surface of the spinel oxide powders and that only these sites participate in the reaction. The only two low-index planes of the spinel structure which can satisfy this condition are identified as B(lll) and 0(110). Allthough the present results pertain to Mn30 4 and C~04, it is believed that the absence of occupied tetrabedral sites at the surface is a more general property of spinels.

'Tbe contents of tbis chapter bas been based on the following: J.-P. Jacobs, A. Maltha, J.G.H. Reintjes, J. Drimal, V. Ponec and H.H. Brongersma, J. CataL 147 (1994) 294-300; S.M. Meijers, J.-P. Jacobs, T.P. Pruys van der Hoeven, J.J. W.M. Rosink, H.H. Brongersma and V. Ponec, J. Catal. (suhmitted).

79

Chopter 4

1. INTRODUCTION

An important part of the products of the chemical industry is fonned by catalytic oxidation reactions. These reactions are not only of commercial interest, but also of interest from the point of view of fundamental research in catalysis. In genera!, catalytic reactions on metals can be described in tenns of reactions running in an adsorbed layer, which means that the resetion is limited to the surface. In heterogeneaus oxidation reactions, in which catalytically active oxides are used, the situation is different. In almost all cases, lattice oxygen is used by the reaction. Lattice oxygen then appears in the oxidation products, while the oxygen vacancy in the catalyst is replenisbed by gaseous molecular oxygen. This mechanism is well known astheMars-Van Krevelen mechanism [1]. Here the activity and the selectivity of the oxidation reaction depend, inter alia, on the metal-oxygen bond strength (the valency of the metal ion), the coordination of the metal in the lattice, the stability and properties of defects, the morphology of the material.

In actdition to selective oxidation reactions, oxides can also be used to catalyze selective reduction reactions, that is, the selective removal of oxygen atoms from organic molecules, such as nitro-compounds and carboxylic acids (de-oxygenation). For these reactions a common reducing agent {H2, CO, CIJ..) can be used or a part of a larger molecule (auto-reduction). In this chapter the selective (auto-) reduction of nitrobenzene to nitrosobenzene is studied on oxides with a spinel structure.

When MnO:z is used as a catalyst it will be reduced flrst to a-Mn30 4, which will then act as the catalyst in the reduction. This catalyst bas been shown to be very active in the selective (auto-) reduction of nitrobenzene to nitrosobenzene [2,3,4]. The Mn30 4

catalyst has the hausmannite structure which is a tetragonally distorted spinel. The structure and properties of spinels have a wide interest. Spinel ferrites are

used as magnetic materials in a wide variety of applications in the electronics industry, ranging from transfonner materials to recording media. Another example is the protective oxygen layer fonned on some alloys to prevent corrosion. For example, on NiCr alloys, a protective spinel-structured oxide layer is formed. In catalysis, spinel structures can be found in various systems, for example, 'Y-Al20 3, a very common support material, supported oxide catalysts [5,6, 7], and the catalytically active spinels discussed in this

investigation. Spinels are compounds of the general fonnula (N+)tetr[Bz3+]""'04 , in which the

formal valency and the cation coordination are indicated. A principal and interesting question to ask is, which valency and which

coordination are responsible for the catalytic activity and selectivity? Since the oxides of

80

Sutface of Cata/ytically-Active Spinels

Fig. 1 The unit cell ofthe normal spinel structure: (AZ+)~arfBl+ro4•

several metals are virtually inactive in the reduction of nitrobenzene (Al, Zn), it is possible to study the activity of the different sites separately by the substitution of the active transition element (Mn, Co) by inactive Al, or Zn ions [2,3,4]. From the present research, it has been concluded that the ions in the tetrabedrally coordinated A-sites, i.e. Mn2+ or Co2+ are either inactive or contribute only little to the overall activity in the nitrobenzene reduction. Similar conclusions have been made earlier with regard to simple mcidation reactions [8,9,10].

The fact that tetrabedral sites are not active could originate from stronger metal­oxygen honds due to the lower valency and coordination number. Another possible explanation is that the tetrabedral sites are not accessible to the reactants. Therefore, the sutface structure of spinels has to be considered in more detail. The unitcellof "normal"

81

Chapter 4

spinel, see Fig. 1, consistsof 32 cubic-close-packed oxygen anions. In this unit cell, 8 of the 64 tetrabedral interstices are filled with divalent metal cations, and 16 of the 32 octahedral interstices are filled with trivalent metal cations. In the literature, usually only the low-index planes are taken into consideration when discussing the surfaces of spinels [10,11,12]. Following the suggestion by Knözinger and Ratnasamy [11], and using their notation, one can distinguish 6 different low index-surface planes, which are shown in Fig. 2. From the figure, it follows that the A(lll), C(llO), E(lOO) and F(lOO) planes

a b

d

'/he /ow-index planes of a normal spinel structure: a) A(lll), b) B(lll), c) C(llO), d) D(llO), e) E(lOO), J) F(JOO) (notation according to Ref. {11]). '/he open spheres represent the oxygen anions, the solid spheres the octahedrally coordinated cations and the cross-hatched spheres the tetrahedrally coordinated cations in the normal spinel structure.

82

Sulface of Catalytically-Active Spinels

have both tetrabedral and octahedral sites on the surface while B(lll) and 0(110) only expose octahedrally-coordinated cations. Thus, it is possible, in principle, to construct a

spinel surface wbere the tetrabedral sites are not exposed. Theoretical predictions for the ~04 spinels have been made already by

Ziólkowski and Barbaux [10]. According to their calculations, the A(lll) or D(llO)

planes (using the notation as described above [11]) are prefeered in the surface, but these predictions do not form a final conclusion. Experimentally, an attempt to elucidate the

surface structure bas been made by Shelef and co-workers [8, 13] using LEIS. They support the idea that the tetrabedrally coordinated cations are not accessible but they stress the preliminary nature of their results due to the rather low sensitivity in their

experiment. Various studies on spinels using XPS have been reported [14,15,16]. G.C.

Allen et al. [14] concluded from XPS studies that the surface composition of a number of mixed transition-metal oxide spinels reflects the bulk stoichiometry. But, as also pointed out by Vepi'ek et al. [15] who performed both LEIS and XPS on CuMn20 4, one sbould

keep in mind that the information depth of XPS arnounts to several atomie layers of the solid. This is in contrast to LEIS, where it is limited to the topmost atomie layer.

Beaufils and Barbaux have concluded from surface neutron differentlal diffraction (SOO) studies that the surfaces of the ideal normal spinels MgA120 4 [17] and C0)04

[18] consist of a mixture of the (111) and (110) plane. Furthermore, because the caleulated distances matebed only the octahedral Co3+ atoms, the exposed faces are, according to the authors, limited to 0(110) and B(lll) planes, which contain only octahedrally coordinated cations. Por the distorted spinel structure of 'Y-Al20 3 [17], it is suggested that 80% of the exposed faces have the (110) direction. However, Angevaare et

al. [3] found an indication (by CO adsorption and IR spectroscopy) that both the divalent

and trivalent sites interact with CO. Obviously, there are new experiments required to solve the problem. These have

been performed with LEIS because of its extreme sensitivity to the topmost atomie layer. The surface concentrations obtained by LEIS will be discussed below in relation to the most relevant catalytic data of the spinel surface. However, when spinel catalysts are studied, some complications can arise due to the redistribution of the cations arnong the

octahedral and tetrahedral sublattice. The spinels can exhibit so-called "inversion", or altematively "oxidative transfer" [4,19]. Then the cations, which in the ideal case of a normal spinel are only in one sublattice (tetrahedral or octahedral), can migrate to the other sites. The two phenomena mentioned distinguish themselves from the fact that in the case of "inversion" the valency of the cations that undergo the transfer keeptheir valency.

83

Chopter 4

In the case of "oxidative transfer" the cations that are transferred change their oxidation state from 2 + to 3 + when going from tetrabedral to octahedral coordination, or vice

versa. Therefore, in addition to the nonnal spinels, which show no "oxidative transfer", some other spinels were synthesized and included in this study. Another reason for doing so was that some spinels with "oxidative transfer" are especially important in catalysis.

To study the influence of the composition and structure of spinel surfaces in detail,

it is clear that spinels which exhibit "oxidative transfer" are undesired. This would only

disturb the results. Therefore also a series of Co1 +"Al2_"04 (0 s x s 2) was prepared. This series was expected not to suffer from the complications due to "oxidative transfer".

2. EXPERIMENTAL

2.1. Catalyst preparation

The simpte spinels were synthesized from hydroxides, obtained at room temp­erature by precipitation of the corresponding metal nitrates with ammonia at pH =9 and room temperature. Then the hydroxides were decomposed in air at 400 K. CoA120 4 was

calcined in air at 1273 K for 20 h. MnA120 4 was calcined in air at 873 K for 20 h after the decomposition of the nitrates, foliowed by a heat treatment at a flow of 5% H2/N2 at 1273 K for 48 h to form the required structure. The Zn1.xMnxA120 4 catalysts were prepared by the same coprecipitation, but then subsequently calcined at 1173 K for 20 h and calcined at 1423 K for 20 h.

A series of Co1+"A12.x04 catalysts was prepared with 0 s x s 2. For each compound a stoichiometrie mixture of Co- and Al-nitrates and citric acid was dissolved in water and heated at 363 Kat 15 mbar for 24 h. The resulting purple solid (mainly cobalt aluminum citrate) was ground and heated in a flow of oxygen with a temperature increase of 25 K/h until 823 K was reached to accomplish a slow decomposition of the citrate mixture. At 823 K the gas flow was stopped and the sample was kept at that temperature for 24 h and then allowed to cool to room temperature in about 4 h. The obtained spinels were calcined during 24 h in air at 1073 K [20]. The decomposition of the citrate during the formation of the spinel consumes oxygen. As a result the degree of "oxidative transfer" can be suppressed [19]. The Co~04 prepared from Co-citrate will be denoted as Co30 4 • Additional, a COj04 catalyst, denoted as COj04(E), was prepared from Co(OHh, obtained by precipitation of CQz(N03) 2 in ethanol at 243 K, while gaseous NH3 was led through.

84

Sulface of Catalytically-Active Spinels

2.2. Catalyst dlaracterization

Catalytic activity measurements were carried out in an open flow system with a

fixed bed reactor. The reaction conditions were as follows: the reaction temperature was

573 K at 1 bar, the carrier gas was He, containing nitrosobenzene at a partial pressure of

70 Pa, the flow rate was 25 ml/min. The reaction was monitored using a gas

chromatograph.

The specific surface areas of the catalysts was determined using nitrogen adsorption (BET method). XRD was performed to ensure that the spinel structure was

formed. The bulk composition of the catalysts was determined by quantitative XRD,

neutron diffraction and AAS. For forther details about the catalytic activity measurement

and the bulk characterization, see Refs. [4,21].

The surface compositions of the catalysts were studied using LEIS. With LEIS,

there is also a smooth background as a result of ions that suffer many collisions at greater

depth. In catalytic investigations, the selective sensitivity for the outermost atomie layer is

essential because when the analysis depth is two or more layers, as in XPS experiments, no exact information can be obtained for the catalytically active top layer and both

tetrabedral and octahedral sites will be detected at the same time. Recent investigations

have proven the value of the technique even for the study of rough insulating surfaces,

which catalysts mostly are. Reviews of LEIS in catalysis research (and its latest

developments) can be found in Refs. [22,23] and chapter 2 of this thesis.

The experiments were performed with the LEIS apparatus NODUS, of which the

basic design bas been described elsewhere [24]. In this setup it is possible to

compensate for surface charging by flooding the surface with thermal electroos from all

sides. The base pressure in the UHV system is 1•10·9 mbar, which increases during

operation to 1•1 0"8 mbar. This increase is due to the inert gas of the ion beam, and does

not affect the measurements.

The samples used for the LEIS experiments consisled of powders pressed into

pellets. FortheLEIS analysis a 4He+ ion beam of 3 keV was used. To retard destroelion

a light ion (4He) is used and the ion flux on the sample is kept as low as possible (less

than 1013 ion/cm2s), while only the interval of interest is measured. Time-dependent

measurements show that the influence of damage by the beam on the present data can be

neglected. Forthermore from earlier results [25] it was found that the influence of

roughness on LEIS results is not as important an effect as was previously reported in

lirerature [26]. Provided the scattering conditions are kept constant during the

experiments, LEIS signals of the different powder catalysts, having specific surface areas

85

Chapter4

ranging from 0.7 to 30 m2/g, can be compared in an absolute sense.

Compound LEIS-signals (10' cts) Yield of~NO

(10"10 mol 0 Al Mn Co Zn m-2s-I)

1) -y-AI:P3 2.1 3.9 < 0.1

2) MnMn20 4 (a-Mn304) 1.4 24.6 5.5

3) Mn.Ak04 3.2 3.3 12.4 2.3

4) ZnMn204 2.0 23.9 1.1 5.3

5) CoQb04 (Co304) 1.4 14.8 1.6

6) Co_Ak04 2.2 4.3 1.9 < 0.1

7) ZnCQz04 2.0 14.1 3.0 1.5

8) ZnAl:P4 2.0 4.2 <0.1 < 0.1

9) 20%-MnO/ZnO 1.3 5.8 9.8

10) ZnO 1.7 15.8 < 0.1

Table 1 Results of the activity measurements in the selective reduction of nitro- to nitroso-benzene, and of the LEIS experiments. The underlined characters indicate octahedral site preference.

3. RFSULTS

3.1. Introduetion

This study is divided into three parts. First, various simple spinels are discussed.

Second, a series of mixed Zn1_xMnxAl20 4 spinels is discussed, where x ranges from zero

to one. XRD showed, as discussed in more detail by Maltha et al. [4] and Meijers et al. [21], that all catalysts studied indeed had the spinel structure. Some "oxidative transfer"

was observed for manganese. This will be discussed in more detail elsewhere [19].

Finally, a series of normal Co1+xAl2_,.04 spinels was prepared to study the influence of the

86

Fig. 3

Suiface of Catalytically-Active Spinels

2000 -U) -0 ->. 1500 := U) c: (J,) -.5 1000 en w -' 500

0

(a)

ûi 1500 t5 -z­(i.i ~ 1000 -.5 en ~ 500

(b)

---- ZnMnp4

··········· MnA~04 MJ\04

0

1000 1500

Al

Mn I\ I

Zn

[\ j j '

"'···~· ........... ~.,r .. l '\ I I

2000 2500 Final energy (eV)

ZnCop4 ··········· CoAI

20

4

Co30 4

1000 1500 2000 Final energy (eV)

Co

Zn

2500

LEIS spectrafrom simple spinels (3keV, He+)

a) Mn-spinels b) Co-spinels

87

Chapter4

composition on the catalytic activity in more detail.

Some typical LEIS spectra of the simpte spinels are shown in Figs. 3a and 3b. The

different elements can be clearly distinguished wben no Zn is present. For the Zn­

containing spinels, this is unfortunately not the case. Therefore, the individual peak areas wbicb repcesent the surface sensitive information are extracted using a deconvolution of the peaks after background subtraction. The peak sbape and the background curve are based on the semi-empirical model suggested by Nelson [27] and Young et al. [28] and was adapted for our purpose [29]. Due to the overJap of the Mn-Zn and Co-Zn

signals and the fact that the Zn signal in the spinel is rather small, the error in the Zn

signal is rather large (up to 50% ). In most of the spinels no contamination was found by

LEIS. This subject will be revisited for tbe Co1+xAlz..x04 series.

3.2. Simpl~spinel surfaces

A wbole series of different spinels bas been investigated in which tbe Co and Mn

cations in the different sites were substituted: (a) by Zn in the tetrabedral interslices and

(b) by Al in octahedral interstices. The reactivity obtained in tbe selective reduction of

nitro- to nitrosobenzene togetber witb tbe LEIS signals, wbich represent tbe surface

composition, are summarized in Table 1. From tbe table, it can be seen that in the Zn-containing spinels only very little Zn

can be detected by LEIS. This confirms tbe LEIS results of Shelef et al. [13], discussed

in tbe introduction. The LEIS signals and, therefore, surface concentration of tbe catalytically-active Mn, are the same for O!-Mn30 4 and ZnMn20 4• A similar effect can be

notleed wben C~04 and ZnC~04 are compared. This not only holds for tbe surface composition but also for tbe catalytic activity (see Table 1). Replacing tbe cations in tbe

tetrabedral sites by Zn, which is not active in the reaction, has no effect on tbe

performance of the catalyst. The fact that no Zn is detected by LEIS on the surface of the

spinels cannot be attributed to a lack of sensitivity to Zn. For ZnO, and for a 20 wt% MnO/ZnO mixture, Zn is easily detected. However, for the ZnA120 4 spinel, no Zn is

detected on tbe surface, just as in tbe otber simpte oxide spinels. If the octahedral sites are occupied by Al (CoAl20 4, MnA120 4) the Co and Mn

LEIS signal and tbe catalytic activity decrease. For the CoA120 4, only a very small

amount of Co is detected and the catalytic activity is also found to decrease dramatically.

The analogous decrease in tbe Mn spinel is not so pronounced. About half of tbe Mn

content is still found on tbe surface, while tbe activity, compared to O!-Mn30 4, is also

about half. This difference between the Co and Mn compounds can be attributed to the

88

Suifoce of Catalytically-Active Spinels

"oxidative transfer" of Mn from tetra- to octahedral sites [4]. This transfer is much

stronger in the Mn-conta.ining spinels than in the Co-conta.ining ones. The fact that no Zn is detected on the surface of the spinels supports the idea that

practically no tetrabedral sites are exposed on the surface. lt is also striking that the

activity correlates nicely with the surface composition as measured by LEIS. Thus all the

Mn or Co detected by LEIS is also active in the catalytic reaction. As pointed out in the

introduction, this is in good agreement with earlier results. It seems correct to conclude

that the surface of spinel oxide powders contains mainly octahedral sites. This fact limits the possible exposed faces to B(ll1) or D(llO), as shown in Fig. 2.

In the second part of the investigation, a series of Zn1.xMnxAl20 4 spinels was

studied. The results of the activity measurements, as a function of the Mn content, are

shown in fig. 4a. The corresponding LEIS results can be found in Table 2 and the LEIS

signal of the catalytically-active Mn is shown in Fig. 4b. One notices immediately an

enrichment of Mn in the surface of the catalyst. This is also reflected by its catalytic activity. The surface concentration of Zn is very small. This is in full agreement with the previous section. Since zinc ions do oot move to the octahedral sites, it is obvious that no Zn should be detected when the surface consists only of occupied octahedral sites.

LEIS signals (104 cts)

x in

Zn1.xMnxA120 4 0 Al Mn Zn

0 2.0 4.2 0.0 < 0.1

0.2 2.3 2.8 5.5 0.3

0.3 2.5 3.9 6.4 0.9

0.5 2.4 3.1 7.8 0.3

0.7 2.3 2.8 8.8 0.6

1.0 2.3 3.3 12.4 0.0

Table 2 LEIS signals ofZn1ftn.Al:P4 spinels.

89

Clulpter 4

4 (i) ... "' § 3

0 ... E

0

";"0 2 ..,.. ->- :.t. ... • 1::::

~ :& ctl 0

0.0 0.5 1.0 (a) x in Zn1.xMn,.AI:P4

15 -(J)

ö "'o 10 ..,.. -a;

c: Cl 5 'f.i)

t (IJ jjj -1

0 0.0 0.5 1.0

(b) x in Zn1.,Mn,AI20 4

0.2

!-c: 6 0 0.1 ë Q) -c: 0 0

..lè ::; 0.0 .0 0.0 0.5 1.0

(c) x in Zn1 .• Mn,.A120 4

Fig. 4 Results for Zn1.;xMn,.Alz()4 spinels a) activity in the selective reduction of nitrobenzene to nitrosobenzene b) LEIS signals of Mn (see table 2) c) •oxidative transfer" of Mn2+ to Mn!+ into octahedral sites [19]

90

Suiface of Catalytically-Active Spinels

60 OCo

-50 vAl :::J *0 -e 40 m - Co30iC) ~ 30

0

c 0 Cl) 0 -c

20 0 en w ...J 10

0 0.0 0.5 1.0 1.5 2.0

x in Co1+xAI2_x04

Fig. s LEIS signals of Co, Al and 0 as ajUnetion ofx in Coi+xAlz...04 (3keV 4He+).

Quantitative XRD indicated that in the bulk no "oxidative transfer" could be assigned [21]. Therefore the general formula can be asscribed as (Ccf-+)-­

[AP+ 2.xCo3+ JociQ4• By increasing x, Al is gradually substituted by Co in the octahedrally

coordinated sites. In Fig. 5 the LEIS signals of Co, Al and 0 are shown as a function of

x in Coi+xA12.x04• The gradual substitution of Al by Co is reflected by a decrease and

increase of their surface peak: areas, respectively. All catalysts showed no significant

contaminations except C0:304(C), which showed Na and K/Ca on the surface. The LEIS

intensity of Co decreased accordingly. Although the Na concentration was below the

detection limit by AAS [21], no Na free C0:304(C) could be obtained. For C0:30iE) LEIS

showed no contamination on the surface. The catalytic activity follows the trend observed

by LEIS as it did in the the case of simple spinel structures. In Fig. 6 the surface

91

Chapter 4

composition as determined by LEIS and the occupation of the octahedral sites as determined by XRD are plotted against the catalytic activity. The results confirm the

observations of the previous sections, namely that it is the c<f+ in the octahedral sites

which is responsible for the catalytic activity. The catalytic activity of C<>.J04(E) foliowed

the trend of the series. However, C<>.J04(C) showed an unexpected low activity. This

could be assigned to the alkali surface contaminations detected by LEIS.

4. DISCUSSION

This investigation has been performed to support and rationalize the information

obtained by determining the catalytic activity in the reduction of nitrobenzene to

nitrosobenzene (de-oxygenation). The spinel Co30 4 is an active catalyst for the above mentioned reaction. The

activity is not decreased when Co2+ ions in the tetrabedral positions of the spinel are

replaced by redox-inactive Zn2+ ions. The spinel ZnC~04 shows almost the same area1 activity as the pure Co-spinel, Co30 4• However, the activity of another fully substituted spinel CoA120 4, is practically zero. Two possible explanations which can account for this

effect will be considered. One is that tetrabedrally coordinated co2+ is not active when it

appears on the surface. Alternatively, the surface of spinels is formed by such

crystallographic planes that the Co2+ is not directly exposed to the molecules from the gas

phase which collide with the surface, or it is, at least, screened by neighboring oxygen

atoms. The surface-analytical technique LEIS, with its selective sensitivity for the

outermost atomie layer is extremely well-suited to distinguish between the two proposed explanations. It appears that in spinels (A2+)tetrlB23+]""'04 hardly any A2+ cations are

exposed to direct collisions of the prohing ions. LEIS results thus confirm the claims of

Beaufils and Barbaux [17,18] and of Shelef and Yao [8,13]. Also, the theoretica!

predictions of Ziólkowski and Barbaux [10] seem to be confirmed. The catalytic activity of ZnMn20 4 is almost the same as that of Mn30 4, in full

analogy with the Co-spinels. However, the activity of MnA120 4 although being lower than

that of Mn30 4 and ZnMn20 4, it is still much higher than that of CoA120 4• Two possible explanations for this phenomenon can be found in literature. First, there could be an inversion of Mn2+ into octahedral positions with an accompanying move of Af+ in the

tetrabedral position [30,31]. The second explanation can be derived from the paper

of Eggert and Riedel [32], namely, there is an "oxidative transfer" of Mn2+ in the tetra-

92

Sulface of Catalytically-Active Spinels

- 3.0 ..... . (fJ

~

E 0 E 2.0 0 ..... 'o ....--CD 1ü 1.0 1...

c:::: • 0 D :;::1 Co30 4(C) 0 <0 CD 1...

0.0 0 20 60 80 100

Surface coverage of CoOx by LEIS(%)

Fig. 6 Reaction rate in the selective reduction of nitrobenzene to nitrosobenzene as a function of the sulface concentration of CoOx (determined by LEIS in %).

bedrally coordinated sites to Mn3+ in octahedral positions, accompanied by exclusion of

A120, from the spinel. A detailed theoretical analysis of the spinel formation (Ziólkowski

et al. [19]) revealed that in MnA120 4 and in the spinels of nomina! composition

Zn1-xMnxA120 4, an "oxidative transfer" takes place indeed, whereby most likely alsopart of the Mn2+ in the tetrabedral sites beoomes oxidized to (Mn3+]''"1• Keeping in mind what

bas been established with fully substituted spinels, namely that mainly (or even solely)

[Mn3+]001 ions are detected by LEIS in the outermost atomie layer, we can conclude, when

inspeeling Fig. 3a, that the transfer of Mfi2+ to octahedral sites made Mn visible for LEIS

detection. These experiments thus self--consistently support the conclusions conceming

"oxidative transfer", as they have been made from the XRD measurements and theoretical

calculations (Ziólkowski et al. [19]).

The next question is whether the extent of the presence of MnH in the surface of

93

Chapter 4

Zn1.xMnxA120 4 spinels corresponds with the extent of the "oxidative transfer" in the bulk.

When studying the two relevant curves in Fig. 4 we are inclined to conclude that the Mn­

surface concentration is too high to set the two values equal. The cause for this

enrichment can be explained as follows. The "oxidative transfer" within the outermost atomie layer of the spinel unit cell

could be higher at the surface. Alternitavely a Zn depletion could account for the fact. This depletion can be caused by surface segregation or by evaporation of Zn during the

spinel preparation. Due to either of these two effects, the surface layer could tend to form a Mn30 4-like spinel and the catalytie activity would be higher than expected on grounds of

the average bulk composition. However, according to Overbury et al. [33] surface

segregation would cause a strong enrichment of ZnO.

XPS results on these mixed oxides performed by Maltha et al. [4] suggest a depletion of Zn in the surface region while the Mn follows the trend of the bulk

composition in Zn1.xMnxA120 4• However, it should be kept in mind that the mean free path for the Mn(2p1n) photoelectrons is about 10 atomie layers, while for the Zn(2P312) it

is limited to only 5 atomie layers. This indicates that the depletion of Zn and the

enriehment of Mn that are observed in LEIS are indeed limited to the outermost surface region [4].

Although for the establishment of the spinel defect structure this is a marginal effect, surface enrichment by Mn should have an important effect on the catalytic activity. The results presenled in Fig. 5 show that this is indeed the case. Hence, the catalytic

activity measurements, also reflect the above mentioned surface enrichment.

It is important to realize the difference in the surface composition and catalytic

activity of the Co-spinels on the one hand and that of the Mn-spinels on the other. A

comparison of entties 5, 6 and 7 in Table 1 shows that a relative smalt amount Co can be detected on the surface of CoA120 4 and the activity, in compliance with it, is very low.

However, the MnA120 4 spinel shows relatively more surface Mn and a higher activity. A

similar surface enrichment of Mn ooropared to Co in other mixed oxides is also reported by Yang et al. [9].

When the mixed Co-Al-oxide series is analyzed in more detail a linear increase of

the surface concentration is observed as a function of the Co eoncentration in the

octahedral sites. The relatively smalt amount of Co that is detected with CoAl20 4 can

possibly be assigned to exposure to other than the proposed B(lll) and 0(110) faces.

which will be, for example, visible at the edges of the smalt particles. Alternatively, an

enhanced "oxidative transfer" at the surface as in the Zn-Mn-oxide case would also account for this effect. However, the catalytie activity was only found significant when

94

Sulface of Oltalytically-Active Spinels

already about 40 % of the surface was covered by cobalt oxide. Fig. 7 shows that the

observed catalytic behavior cannot be described by a simple linear relationship. Therefore

it was suggested that not a mononuclear Co-species did not determine the catalytic activty behavior but rather a more complex active site was necessa.ry. However, the limited data

prohibited further speculation on this subject.

The influence of surface contamination was nicely illustrated by the combination

of LEIS and the catalytic activity measurements. Since according to Fig. 7 the decrease in

activity of CG.304(C) was accompanied by a proportional decrease in Co detected by

LEIS, it could be concluded that the Co cations shielded by Na or K/Ca to detection by LEIS are similarly blocked for the reactants.

5. CONCLUSIONS

The results demonstrate the advantage of the use of the surface-analysis technique

of LEIS in the analysis of catalytically active materials. By the use of absolute LEIS

signals in the comparison of the different catalysts, the catalytic data on various spinels

could be rationalized and additional support could be obtained for a model of the "oxidative transfer" of tetrabedral Mn2+ ions to the octahedral positions. The data for the mixed-oxide spinels indicate a strong surface enrichment in Mn, which was not found for

the Co-containing mixed oxide spinels, where no "oxidative transfer" was expected. It can

furthermore be concluded that the strong correlation between the different techniques

suggests that no chemically induced segregation occurs in these oxides. This should be

taken into account when discussing reaction roodels for selective reduction.

A model could be proposed where only the highest coordinated cations were found

at the surface which is, in the case of the spinels, the octahedrally coordinated cation. For another well-known mixed oxide structure, the perovskite, similar observations were

made. Like the spinel, the perovskite bas an cubic oxygen lattice with cations in two

different oxygen coordination, namely a octahedral and twelvefold site. At the surface of

SmCo03 and a series of Sm1_xSrxCo03, which is a candidate for a fast oxygen transport

material for the use in fuel cells or solid state membranes, a strong enrichment of the

twelvefold coordinated cations (Sm,Sr) was detected by LEIS [34]. Although the

present results pertain to Mn30 4 and C0]04, results on other spinels such as F~04 and ZnA120 4 seem to suggest that the absence of occupied tetrabedral sites at the surface is a more genhal property of spinels.

95

Chapter4

References

[1] P. Mars and D.W. van Krevelen, Chem. Eng. Sci. 3 (1954) 41. [2] T.L.F. Favre, P.J. Seijsener, P.J. Kooyman, A. Maltha, A.P. Zuur, and V. Ponec, Catal. Lett. 1

(1988) 461. [3] P.A.J.M. Angevaare, J.R.S. Aarden, J.R. Linn, A.P. Zuur, and V. Ponec, J. Electron Spectrosc.

Relat. Pheoom. 54 (1990) 795. [4] A. Maltha, H.F. Kist, B. Brunet, J. Zióllrowski, H. Onishi, Y. Iwasawa and V. Ponec, J. Catid.

1.., (1994) 256. [S] G.C. van Leerdam, J.-P. Jacobs and H.H. Brongersma, Surf. Sci. 268 (1992) 45. [6] J.-P. Jacobs, G.C. van Leerdam and H.H. Brongersma, in Flllllkunental Aspects of Heterogenrous

Calalysis Stilllied by Partiele Beams, p. 399 , Eds. H.H. Brongersma and R.A. van Santen, NATO-ASI Series B 265, Plenum press, New York, 1991.

[7] B.R. Strohmeier and D.M. Hercules, 1. Catal. 841 (1984) 266. [8] H.C. Yao and M. Shelef, J. Phys. Chem. 78 (1974) 2490. [9] B.L. Yang, S.F. Chan, W.S. Chang and Y.Z. Chen, 1. Catal. 130 (1991) 52. [10] J. Ziólkowski and Y. Barbaux, J. Mol. Catal. 67 (1991) 199. [11] H. Knözinger and P. Ratnasamy, Catal. Rev.-Sci. Eng. 17 (1979) 31. [12] B.C. Lippens and J.J. Steggerda, in Physical and Chemical Aspect of Adsorbents and Catalysts, p.

171, Ed. B.G. Linsen, Academie Press, New York, 1970. [13] M. Shelef, M.A.Z. Wheeler and H.C. Yao, Surf. Sci. 47 (1975) 697. [14] G.C. Allen, S.J. Harris, J.A. Jutson and J.M. Dyke, Appl. Surf. Sci. 37 (1989) 111. [IS] S.Veprek, D.L. Cocke, S. Kehl and H.R. Oswald, J. Catal. 100 (1986) 250. [16] V.A.M. Brabers, F.M. van Setten and P.S.A. Knapen, J. Solid State Chem. 49 (1983) 93. [17] J.P. Beaufils and Y. Barbaux, J. Chim. Phys. 78 (1981) 347 (in French). [18] J.P. Beaufils and Y. Barbaux, J. Appl. Cryst. IS (1982) 301. [19] J. Ziólkowski, A.Maltha, H.F. Kist, E.J. Grootendorst and V. Ponec, a paper in.preparation [20] J. Zióllrowski, K. Krupa and K. Mocala, J. Solid State Chem. 48 (1983) 376. [21] S.M. Meijers, J.-P. Jaoobs, T.P. Pruys van der Hoeven, J.J.W.M. Rosink, H.H. Brongersma and

V. Ponec, J. Catal. (submitted). [22] H.H. Brongersma and G.C. van Leerdam, in Fundamental Aspects of Heterogeneous Caralysis

Stilllied by Partiele Beams, p. 283, Eds. H.H. Brongersma and R.A. van Santen, NATO-ASI Series B 265, Plenum Press, New York, 1991.

[23] H. Niehus, W. Heiland and E. Taglauer, Surf. Sci. Rep. 17 (1993) 213. [24] H.H. Brongersma, N. Ha.wwindus, J.M. van Nieuwland, A.M.M. Otten, and A.J. Smets, Rev.

Sci Instrum. .., (1978) 707. [25] J.-P. Jaoobs, S. Reijne, R.J.M. Elfrink, S.N. Mikbailov and H.H. Brongersma, J. Vac. Sci.

Technol. AU (1994) 2308-2313. [26] R. Margraf, H. Knözinger and E. Taglauer, Surf. Sci 211/212 (1989) 1083. [27] G.C. Nelson, J. Vac. Sci. Tecbnol. A4 (1986) 1567. [28] V.Y. Young, G.B. Hoflund and A.C. Miller, Surf. Sci 235 (1990) 60. [29] J. Drimal and V. Koksa, unpublished [30] S. Greenwald, S.J. Piekart and F.H. Grannis, J. Chem. Phys. 22 (1954) 1597. [31] F.C.M. Driessen, Inorg. Chim. Acta 1 (1967) 193. [32] A. Eggert and E. Riedel, Z. Naturforsch. 46b (1991) 653. [33] S.H. Overbury, P.A. Bertrand and G.A. Somorjai, Chem. Rev. 75 (1975) 547. [34] I.C. Fullarton, J.-P. Jaoobs, H.E. van Benthem, J.A. Kilner, H.H. Brongersma, P.J. Scanlon, and

B.C.H. Steele, Proc. of 1st Euroconf. on Solid State Ionics, Zak:ynthos, Greece, September 11-18, 1994 (in press).

96

The Growth Mechanism of Nickel Oxide in the Preparation of Ni/ Al20 3 Catalysts studied by

LEIS, XPS and Catalytic Activity*

Abstract

A series of Ni/ Al20 3 catalysts prepared from the vapor phase by ALE have been studied. A model is proposed for the growth mechanism of nickel oxide on alumina, from sequences of treatments with Ni(acac)2 and air. In this study activity measurements were

combined with surface analysis by LEIS and XPS. During the fust preparation sequence ( < 5 wt% Ni) atomically dispersed nickel oxide is obtained on the alumina support. This nickel oxide is catalytically inactive, but offers the nuclei for the growth of nickel oxide which can be reduced and shows catalytic activity. The highest utilization of nickel in the hydragenation of toluene was obtained when the nickel oxide nuclei were covered by another layer of nickel oxide.

'The contentsof this chapter bas previously appeared as: J.-P. Jacobs, L.P. Lintifors, J.G.H. Reintjes, 0. Jylhli and H.H. Brongersma, Catal. Lett. 25 (1994) 315-324.

97

Chapter 5

1. INTRODUCTION

Supported niekel catalysts are widely used in a number of industrial processes sueh as hydrogenation. The aetivity of the supported catalysts is strongly dependent on the preparation metbod used and on the choice of reagents and support. In order to keep up with the inereasing demands regarding the performance of catalysts,- a thorough onderstanding of the mechanisms involved in the catalyst preparation is important. In this study, the preparation of Ni/ Al20 3 catalysts from vapor phase by ALE, was studied using LEIS and XPS. The catalysts were prepared by ehemisorbing niekelacetylacetonate, Ni(acach, onto the alumina support using atomie layer epitaxy (ALE).

The characteristic feature of ALE is the controlled build-up of surface struetures [1]. This is achieved by a sequential introduetion of different reagents in saturating gas­

solid reactions. The active species or their precursors, introduced from vapor phase, saturate the surface by chemisorption. Thus, the toading of the active species is determined by the ability of the support to bind the reactant, not by dosing of the reactant [2]. The Ni/ Al20 3 catalysts were prepared in the oxidic form by cbemisorbing nickel

acetylacetonate, Ni(acac)2, onto alumina foliowed by the removal of ligand residues with air treatment [3]. The nickel content was regulated by the number of reaction eycles.

In LEIS a beam of mono-energetic ions is directed towards the surface to be analyzed. The information on the surface of a specimen is obtained from the energy distribution of the ions that are scattered back by the surface. Conservalion of energy and momenturn during the ooilision with a surface atom ensures that for given experimental conditions (incident energy, type of ion and scattering angle) the scattered ion energy only depends on the mass of the surface atom from which it was scattered. An energy spectrum of the scattered ions is thus equivalent to a mass spectrum of the surface atoms. It bas been demonstrated earlier that when inert gas ions, such as He+ or Ne+ are used to

probe the surface, the information oornes exclusively from ions scattered by the outermost atomie layer of the solid [4].

Recent investigations proved the value of the technique even when studying rough insulating surfaces, such as catalysts in general [5,6]. A review of LEIS in catalysis researchandrecent developments bas been reported elsewhere [7,8].

The activity of the catalysts was evaluated in gas phase hydrogenation of toluene

in a laboratory scale flxed bed reactor after in situ rednetion of the accessible nickel species. The aim of this study was to suggest a model of the structure of the catalyst and to establish the effect of alumina pretreatment, prior to catalyst preparation, on the structure and activity of the catalysts. It is well known that the interaction between metal

98

Growth Mechanism of Ni/AlP3 Catalysts

oxide supports and active species is crucial for the catalytic performance. Several studies dealing with the interaction between nickel and alumina have been reported [9,10,11,12]. The interaction can be varled e.g. by pretreating the support and

thus influencing the number, strength and nature of the bonding sites between the active species and the support. The dispersion of the active species and their resistance towards agglomeration as well as the actlvities of the catalysts are strongly dependent on this

interaction.

2. EXPERIMENTAL

2.1 Catalyst preparafton

The alumina used as support (AKZO 000-1.5E) was crushed and sieved to a size of 0.15-0.35 mm. It was then heat treated in air at varlous temperatures to vary the fJI'Y phase ratio. The preparation of the Ni/ Al20 3 catalysts by ALE has been discussed in more detail elsewhere [3]. The interaction between Ni(acac)z and alumina was restricted to che­misorption by the proper choice of reaction temperature. Surface saturation was ensured by using an excess of reactant. The chemical nature of the nickel species was changed by removing the ligands with air at elevated temperatures at the end of each sequence.

The catalyst preparation was studied using an alumina, having a 91-y phase ratio of - 0.25 as support (catalysts Ni-1, Ni-2, Ni-4", Ni-6, Ni-8 and Ni-10 in Table 2.1). The nickel content was varled from 3 to 21 wt% by changing the number of preparation sequences from one to ten.

In order to study the effect of alumina pretreatment on the activity, three catalysts containing - 10 wt% nickel were prepared on the alumina having fJ/'Y phase ratiosof -0, 0.25 and 0.30 (catalysts Ni-4., Ni-4" and Ni-4. respectively, in Table 1).

2.2 Catalyst cbaracterization

The nickel content was determined using XRF. Possible crystalline nickel species and alumina phase ratios were determined by means of XRD. The specitic surface area, SA, pore volume, PV, and average pore diameter dP, of the alumina and the catalysts were determined using nitrogen adsorption and condensation [3]. The values of the SA, PV and dP were between 125 and 155 m2/g, 0.3 and 0.4 cm3/g, and 100 and 115 A, respectively, for all catalysts. The corresponding values for the pure fJ and 'Y alumina

99

Chapter5

were 120 and 190 m2/g, 0.45 and 0.49 cm3/g, and 150 and 100 Á, respectively. The surface compositions of the catalysts were studied using LEIS and XPS. The

LEIS characterization was performed with the LEIS apparatus NODUS, of which the basic design bas been described elsewhere [13]. In its setup it is nowadays possible to compensate for surface charging, when the ion beam is directed onto the target, by flooding the surface by thermal electroos from all sides. The base pressure of the UHV system is 1•1 o-9 mbar and increases during operation to 1•1 0"8 mbar. This is due to the inert gas of the ion beam, which does not affect the measurements. Prior to the LEIS measurements the catalysts were treated in 20 mbar oxygen at - 250°C for about 15 min in order to remove contamination by water and hydrocarbons. Since the scattering conditions were kept constant during the experiments, the LEIS signals of the different catalysts can be compared in an absolute sense. A typical LEIS spectrum of the studied catalysts is shown in Fig. 1. The signals ,for the various elements are the peak heights with respect to the background on the high energy side of the peak. Instead of peak

heights one can also use peak areas. The numbers are then different, but it does not effect the conclusions.

The catalysts as well as the pure alumina were analyzed by XPS. The photoelectron spectra were measured with an X-probe model lOl spectrometer manufactured by Surface Science Instruments (VG Fisons, Mountain View, CA) equipped with a monochromated Al Ka X-ray source. A flood gun was used to minimize sample charging. The background pressure during data acquisition was lower than lQ-8 mbar. Por the analysis the Ni 2P312 signal with its corresponding satellite were used. High resolution spectra of the elements of interest and wide scan survey spectra were run for each sample. The relative surface composition in atomie percent was calculated from high resolution spectra for aluminum, niekel and oxygen. The atomie surface composition was caleulated from the Ni 2P3f2, Al 2p and 0 ls level peak intensities using atomie sensitivity factors provided by the instrument manufaeturer. (Note: allthough this method is common practise it should be noted that for quantitative measurements the use of mathematical models introduced by Kerkhof and Moulijn [14] and Kuipers et al. [15], which take

the morphology of the supported catalyst into account, should be preferred. This would take the XPS results of Fig. 5 eloser to the LEIS measurements. The model for the growth mechanism will however be unaffected.)

The activity of the catalysts was studied in the hydrogenation of toluene to methylcyclohexane. The catalysts were tested, in a flow of gases, in a differential fixed bed vertical tubular reactor following a procedure simular to that described elsewhere [16]. The catalysts were reduced in situ in flowing hydrogen at 500°C (0.5 mol/h) at

100

Growth Mechanism of M/Al,P3 Catalysts

atrnospheric pressure prior to testing. The reactor was operated at atrnospheric pressure,

T,-175"C at a rnolar ratio of hydrogen to toluene, nmfflroLUPNB of -3.1. Methylcyclo­bexane (MECH) and unreacted toluene along with traces of rnethylcyclohexene were

analyzed by the on-line GC.

Catalyst, No. of prep.seq. Ni Activity Activity (wt%) ~/g}ó/h) (~/fk"/h)

Ni-1,1 3.0 5 <1 Ni-2,2 5.9 60 4

Ni-4o,4 10.8 205 22

Ni-6,6 14.0 140 20 Ni-8,8 17.2 130 22 Ni-10,10 20.7 135 28

Table 1 Number of preparadon sequences, corresponding M-loadings anti hydrogenation activity of the catalysts.

3. RESULTS AND DISCUSSION

3.1. The Diekei growth mecbanism in the preparation

3.1a Activity during the hydrogenation of toluene In Table 1 the reaction rates calculated in grarns of produced MECH per gram of

nickel/catalyst and hour are shown. The results describe the steady-state performance of the catalysts. The selectivity to MECH was always above 95 %. Por a more detailed discussion the reader is referred to Ref. [3].

3.1b Sutface characterlzation LEIS was used for the study of the nickel oxide growth. The surface cornposition

was studied as a function of the nurnber of preparadon sequences (nickelloading) ranging frorn one to ten (catalysts Ni-1 to Ni-10 in Table 2.1) along with the pure alurnina which was used as support in the preparadon of the catalysts. Each sample was rneasured twice, except for catalysts Ni-6 and Ni-40 , which were rneasured four times. The LEIS

101

Chapter 5

intensities were reproducible, with a deviation of less tban 5 % between measurements of the same sample. The average LEIS signal intensities are shown in Fig. 2 as a function of niclrel content.

As expected, the nickel signal increases and the aluminum signa} decreases with

increasing nickel loadings. Tbis indicates, that nickel is shielding the aluminum.

However, as can be seen from Fig. 2, the LEIS nickel signal does not increase linearly

with the nickel loading: starting from the second preparation sequence, i.e. at nickel

loadings above 5 wt%, the rate of increase of the LEIS Ni-signal decreases. Thus, more

and more of .tfle nickel oxide appears on already deposited nickel oxide instead of on

uncovered alumina. The theoretical mooolayer of NiO on a support having a specific surface ara of about 130 rrr/g, i.e. similar to that of the alumina used in the catalyst preparation, was cak:ulated. Both the crystal ionic radü of the elements and unit cell

dimensions for NiO, were used. The monolayer coverage occurs at about 13 wt% Ni/Al203.

4

""":" ::I 3 ..c 'I-

a:s -~ "(i) 2 c Q) -.5

Cf) 1 w

...J

0 1000 1500 2000 2500

Final energy (eV)

Fig. 1. Typical LEIS spectrum of an alumina supported nickel catalyst (Ni-4~ rHe+, primary energy of 3 ke V)

102

Growth Mechanism of Ni!AlP3 Catalysts

It appears from Fig. 2 that, at a niekel loading corresponding to the theoretica! monolayer coverage, only half of the niekel is visible with LEIS. Thus, one can assume that niekel is present in clusters of NiO. If one assumes "atomie" dispersion of niekel during the first preparation sequence, it is possible to approxirnate the number of Ni species detectable by LEIS. This exercise can also be perforrned for the XPS results, whieh are presented in Fig. 3. One should keep in rnind that LEIS probes only the topmost atomie layer while in XPS several layers are seen. LEIS is, therefore, mueh more sensitive for spreading behavior than XPS. Combining these results and the aetivity measurements, as shown in Fig. 4, a model for the niekel growth in the preparation of Ni/A120 4 catalysts via ALE using Ni(acach as a starting compound is proposed.

Fig. 2.

200

-::J 150 -e co -~ ·u; 100 c Q) ...... c

• ~--

---------· •

AO! j •Al! L ___ !_~_U

en w ...J

- &_- - • .c:--_ - -it - - _..._ 50 _/ _ _.,_- - ...

//t

0 .------'------'------------'-----L__-_j

0 10 20 1\li content (wt%)

LEIS intensities for various nickel loadings.

3.Ic Model for the growth mechanism

During the first reaetion sequence, when condensation of the reaetants is prevented, the bulky acae ligands guarantee the distance between niekel. The theoretica! monolayer coverage of the surface complexes of Ni(acach on the alumina used in these

103

Clulpter 5

experiments corresponds well with the nickel content obtained by XRF (3.0 wt%). This highly dispersed catalyst was found to be catalytically inactive in the hydrogenation of toluene (fable 2.1, Fig. 4). This seems reasonable, since the Ni-acac and NiO wi1l interact with the strongest binding sites of the alunrlna support and this makes the rednetion of Ni2+ to Ni0 difficult [17,18]. The dissociation of H2 is an important step in hydragenation reactions and it requires metallic (Ni~ surface [17,18,19].

Fig. 3.

;g 60 1a -en a.. >< 40 ~ c: 0

:;::: "(i)

g_ 20 E 8

.. ···• ........ .·

........

"-Ü

•Al : .... ~ .. r-.J.L{?><) ...

0 .,_ _ __,_ __ j___~-----0 10 20

Ni content (wt%)

XPS results in atomie percentfor various nickelloadings, where Ni+Al+O = 100 at-%.

After 2-4 reaction sequences with Ni(acac)2 and air, (5-10 wt% Ni), the nickel signal obtained by XPS still increases while the LEIS signal increases less steeply. This suggests that the well dispersed nickel oxide, strongly bound to the alumina during the first sequence, induces further growth of nickel oxide clusters and is thus screened from observation by LEIS. The catalytic activity increases dramatically (Fig. 4) during cycles 2-4. At higher Ni-loadings the weakly bonded nickel oxide is easily reduced producing nickel which can interact with the gaseous reactants. LEIS can detect the formation of clusters after the first ALE sequence. After the second sequence, there is about twice as much Ni deposited on the surface, as shown by XRF and the doubled XPS signal,

104

Growth Mechanism of Ni/Al/)3 Catalysts

whereas tbe LEIS signal increases only by - 50% (Fig. 4). This indicates that the nickel

species from the second sequence shield the nickel nuclei obtained during the primary

reaction in the fust sequence. At higher nickel loadings, -10-20 wt% Ni, both the hvdrogenation activity and

the surface composition measured by LEIS do notchange much. This indicates that the

ratio of the number of nickel species accessible for the hydrogenation and the total

number of nickel atoms bas decreased by a factor of 2 when tbe nickel content bas increased from 10 to 20 wt%. The nickel oxide clusters are growing, covering tbe precursor from previous sequences. At 20 wt% (10 sequences) about one quarter of the

deposited nickel is detected by LEIS. We suggest that the clusters have grown from an

average thick.ness of two NiO layers after four preparanon sequences to an average

tbick.ness of four NiO layers after ten preparation sequences. This proposal is supported

by tbe XRD results: When tbe catalysts containing less than 14 wt% nickel were studied

by means of X-ray diffraction, the nickel oxide was found to be XRD-amorphous. For the

catalysts containing more than 14 wt% nickel, peaks in the X-ray diffractograms were

observed in the region of 43-44° 28 corresponding to NiO. However, the intensity was very low. Even for the catalyst containing more than 20 wt% Ni, the crystallite size of

the nickel oxide species was estimated to be below 40 A, using a profile fitting program and the Scherrer formula. It seems likely, by combining the results obtained from LEIS,

XPS and XRD measurements, that the nickel oxide clusters are smalt, even at nickel

loadings of - 20 wt%.

Catalyst, fJI'Y Al20 3 Activity Activity LEIS AllO LEIS

Ni-loading phase (gMECH/ gN/h) (gMECH/~h) signal Ni (wt%) ratio ratio• signal•

Ni-4., 10.8 -0 120 13 2.8 68 Ni-4.,, 10.8 0.25 205 22 2.7 76 Ni-40 , 10.1 0.30 230 23 2.5 83

• For the LEIS signals the peak heights in the energy spectra are taken.

Table 2 Hydrogenation activity and LEIS signals of catalysts prepared on alumina having various phase ratios.

105

Chapter 5

At even higher loadings the nickeValumina ratio observed by LEIS bas been reported to increase [20]. This should increase the total activity of the catalyst. In this

investigation, however, there was not enough data to confirm this observation. Although

the LEIS results of similar studies [20,21] in general show good agreement with our

results, one should keep in mind that the catalysts have been prepared in a different way in this work and that the growth mechanism is dependent on the nietrel compound,

alumina and the preparation method/conditions used.

Fig. 4.

-""

15

·e 10 0

(f)

.~ 5 z

o activity ~LEIS

+XPS • •

0

...... e·····o-···· o·······~········~ oa--e~~--~--~----~--~

0 10 20 Ni content (wt%)

100 )> s

80 c:· ~ -(Q

60 s::: m (')

ia-40 ~ ;;r -20

0

Results of the LEIS, XPS and activity measurements as a function of Ni loading on the catalyst.

3.2. Effect of alumina pretreatment on activity and surface compostion

In order to study the effect of alumina pretreatment on the catalyst activity, three

catalysts (Ni-4.,Ni-4.,,Ni-4c) were tested in the hydragenation of toluene. The surface of

the fresh catalysts was studied using LEIS. The only variabie in the preparation of the catalysts was the pretreatment of the alumina. The alumina phases were investigated by

106

Growth Mechanism of Ni/Alz(J3 Catalysts

XRD. The XRD pattems of the alumina as received (pure -y-alumina), after heating at

950°C for 6 hours (pure 8-alumina), and mixtures of these alumiDas were studied and

used as reference samples. It was then possible to determine the alumina phase ratios in

the catalysts. In table 2 the Al/0 ratio of the aluminas and the Ni-signals of the

corresponding catalysts observed by LEIS are shown along with the hydrogenation activities. There appears to be a correlation between the hydrogenation activity and the

phase ratio of the alumina; with increasing 91-y phase ratio, both the activity and the LEIS Ni-signal increase. Nickel oxide seems to be better dispersed on the alumina when the fraction of the 9 phase is increased from 0 to 0.3. The available bonding sites and

especially the reducibility of nickel seems to be higher on the mixture of aluminas, containing less OH-groups than the pure -y-alumina.

4. CONCLUSIONS

The growth mechanism of nickel oxide was studied in the sequential prepatation of

Ni/A120 3 catalysts via ALE by bringing Ni(acach vapor into contact with an alumina support and removing the acac ligands with air. Activity measurements were combined

with LEIS and XPS surface analysis in order to develop a model for the nickel growth

process. Nickel species brought into contact with alumina during the first preparation

sequence bonds with the heterogeneaus alumina surface. These nickel oxide species,

serving as nucleation centers, are highly dispersed but remain after reduction unreduced

and catalytically inactive. They interact too strongly with the alumina. During the fust

prepatation sequence, the highest possible dispersion of nickel oxide species is ensured by

the steric bindrance of the acac ligands in Ni(acach.

During the 2-4 sequences (5-10 wt%), nickel oxide species grow on the nucleation

centers but also to some extent on the unoccupied surface on the support. This can be seen in the slopes of the LEIS signals and the catalytic activity as a function of the number the prepatation sequences (Fig. 4). During 6-10 sequences (10-20 wt% Ni), both

the LEIS signal of the Ni-oxide and the activity of the metallic nickel remain relatively

constant (Figs. 2 and 4), because nickel oxide is growing preferentially on nickel oxide. The highest activity in the hydrogenation of toluene, i.e. the maximum nickel utilization,

was obtained at a Ni-loading of about 10 wt% nickel.

107

Chapter 5

References

[1] T. Suntola, Mater.Sci.Rep. 4 (1989) 261. [2] M. Lindblad and L.P. L.indfurs, in New Frontiers in Catalysis, p. 1763, Proc. lOth Int. Congr. on

Catalysi.s, Eds. L. Ouczi, F. Solymosi and P. Tétényi, Akadmiai Kiado, Budapest, 1993. [3] M. Lindblad, L.P. Lindfors and T. Suntola, Catal. I..ett. 27 (1994) 323. [4] H.H. Brongersmaand P.M. Mul, Surf.Sci. 35 (1973) 303. [5] O.C. van Leerdam, J.-P. Jacobs and H.H. Brongersma, Surf. Sci ~ (1992) 45. [6] M. Skoglundh, L.O. Löwendahl, P.O. Menon, B. Stenbom, J.-P. Jacobs, 0. van Kessel and H.H.

Brongersma, Catal. Lett. 13 (1992) 27. [7] H.H. Brongersma, and O.C. van Leerdam, in Fundamental Aspects of Heterogeneous Catalysis

Studied by Partiele Beams, p. 283, Eds. H.H. Brongersma and R.A. van Santen, NATO ASI Series B 2(;5, Plenum, New York, 1991.

[8] H.H. Brongersma, R.H. Bergmans, L.O.C. Buijs, J.-P. Jacobs, A.C. Kruseman, C.A. Severijns, and R.O. van Welzenis, Nucl. Instr. Meth. 868 (1992) 207.

[9] S. Narayanan, K. Uma, J. Chem. Soc. Faraday Trans. I 81 (1985) 2733. [10] S. Narayanan, 0. Sreekanth, J. Chem. Soc. Faraday Trans. I SS (1989) 3785. [11] C.H. Bartholomew and R.J. Farrauto, J. Catal. 45 (1976) 41. [12] S.L. Cben, H.L. Zhang, J. Hu, C. Contescu and J.A. Schwarz, Appl. Catal. 73 (1991) 289. [13] H.H. Brongersma, N. Hazewindus, J.M. van Nieuwland, A.M.M. Otten, and A.J. Smets, Rev.

Sci. Instrum. 49 (1978) 707. [14] F.P.J.M. Kerkhof and J.A. Moulijn, J. Phys. Chem. 83 (1979) 1612. [15] H.P.C.E. Kuipers, H.C.E. van Leuven and W.M. Visser, Surf. Interface Anal. 8 (1986) 235. [16] L.P. Lindfors, T Salmi and S. Smeds, Chem. Eng. Sci. ~ (1993) 3813. [17] C.H. Bartholomew and R.B. Pannell, J. Catal. 65 (1980) 390. [18] C.H. Bartholomew, in Hydrogen Effects in Catalysis: Fundamental and Practical Applications, p.

139, Eds. Z. Paál and P.O. Menon, Marcel Dekker Inc., New York, 1988. [19] R.Z.C. van Meerten and J.W.E. Coenen, J. Catal. ~ (1977) 13. [20] M. Wu and D.M. Hercules, J. Phys. Chem. 83 (1979) 2003. [21] S. Kastzelan, J. Grimblotand J.P. Bonnelle, J. Phys. Chem. 91 (1987) 1503.

108

Surface Analysis and Activity of Chromia/alumina Catalysts prepared by Atomie Layer Epitaxy*

Abstract

A series of Cr0j'Y-Al20 3 catalysts was prepared by atomie layer epitaxy (ALE) from the gas

phase using Cr(acac)3 as a precursor. The toading was varled from 1.3 to 8.8 wt% Cr by the

sequential saturative reactions of Cr(acac)3 vapor and air. Loadings as low as 0.16 wt%

were achieved by blocking the alumina surface partially by acetylacetone or

dipivaloylmethane before introducing Cr(acac)3• lt was shown that the catalytically-active

material was evenly and homogeneously distributed throughout the catalyst particles. LEIS

showed a dispersed layer of CrOx up to 7.4 wt% of Cr. XPS and UV-VIS spectrophotometry

showed that already at the lowest loadings (0.16 wt% Cr0xf'Y-A~03) both er>+ and ct+ species were present. This was attributed to a stabilization of cel+ on the alumina support

in the ALE process. Catalytic activity in the dehydrogenation of i-butane to i-butene

increased linearly with toading up to a monolayer coverage, which pointed to a mononuclear

er>+ species as the catalytically active center.

"The content ofthis chapter bas been submitted for publicationas A. Kytökivi, J.-P. Jacobs, A. Halwli, J. Meriläinen and H.H. Brongersma, J. Calal.

109

Chapter 6

1. INTRODUCTION

Chromia/alumina catalysts possess high activity and selectivity in the dehydrogenation

of light alkanes such as propane and butane. The system bas been already studied intensively

[1] andrecent studies have shown there is still a considerable interest in the system [2-19].

In these studies the catalysts have usually been prepared by impregnation using Cr03 or

Cr(N03) 3 as a precursor whereas the present work describes the preparation of

chromia/alumina catalysts by atomie layer epitaxy (ALE) using sequentia} saturative reactions

of Cr(acac)3 vapor and treatment by air. ALE is a gas-phase technique, which application has been expanded, in addition to

thin film growth [20,21], to the preparation of catalysts only recently [22]. It is a new

approach which shows great promise in, the preparation of tailored for the task and

well-defined catalysts. ALE distinguishes itself from the chemical vapor deposition techniques

by the use of saturative gas solid reactions to form the required surface. These saturative and

self-terminating reactions ensure a uniform distribution of the chemisorbed species throughout

the porous catalyst support. The surface of the support and the reagent used determine the

density of the surface species during each reaction step. Thus, the concentration of the

desired component is not dependend on the dose of the reactant. 1t is only necessary to

ensure that the dose exceeds the amount required for surface saturation.

The use of Cr(acac)3 vapor in the preparation of catalysts bas been demonstrated

earlier by Köhler et al. [16, 17]. They, however, did not utilize saturative reactions but

controlled the concentration of chromium by regulating the contact time between Cr(acac)3

and oxide supports in a fluidized bed. Haukka et al. [23,24] used a single saturative reaction

of Cr(acac)3 to introduce chromium, but on a silica support.

The aim of this paper is to study the growth process of chromia on 'Y-alumina during

the preparation from vapor phase by ALE and correlate it to the catalytic behavior of the

selective dehydrogenation of i-butane to i-butene. Therefore a series of Cr-loadings was

prepared and macroscopie homogencity of the samples was ensured by SEMIEDS, XRD and

BET measurements. A more detailed investigation of the surface formed was performed by

LEIS and XPS. In this way the growth process and the nature of the Cr-species on the

alumina support during the ALE preparation could be monitored. The findings on the

CrOJ'Y-Al2~ will be compared and contrasred to earlier workon NiO/'Y-AlP3 [25,26] and

CrOJSi02 [23,24] prepared by ALE. Finally, the series of Cr0J'Y-Al20 3 catalysts was tested

in the dehydrogenation of i-butane to i-butene.

110

Surface Analysis and Activity of Chromia/alumina Catalysts

2. EXP~AL

2.1 Catalyst preparation

The support material was 'Y-alumina (000-1.5E) from AKZO-NOBEL. This material

was crushed and a fraction with apartiele size of0.15-0.35 mm or 0.7-1.0 mm was used as

a support. The values of specific surface area and pore volume for uncalcined material were 195 m2/g and 0.49 g/cm3, respectively. Cr(acac)3, i.e. Cr(C5û:!H7) 3, 99% pure was purchased

from Riedel-de Haën. Synthetic air was used for the removal of the ligands. Acetylacetone

(C5H80 2, Hacac) from Merck and dipivaloylmethane (C11H200 2, Hthd) from Fluka were used

as surface blocking agents.

The support material (4-10 g) was preheated for 16 h in a muffle fumace at 600"C.

The heat pretteatment was continued at 200"C for 3 h in a fixed-bed flow-type reactor [26]

(Microchemistry Ltd.) under a pressure of 50.100 mbar in a nitrogen flow. The solid

Cr(acac)3 was vaporized at 190"C, and was transported in nitrogen downwards through the solid support bed. The reaction of Cr(acac)3 with the support was performed at 200'C. A

nitrogen purge foliowed this reaction sequence for at least two hours at the reaction

temperature. To remove the ligand residues from the samples, air treatment was started at

200"C and continued at 600'C for 3-4 h. The high preheat temperature and air treatment temperature used in the preparation were chosen to correspond to the conditions in the

dehydrogenation reaction. To increase the Cr concentration of the samples the Cr(acac)3,

nitrogen and air sequences were repeated between one and seven times. To decrease the Cr concentration Hacac or Hthd was bound to the surface at 200"C foliowed by nitrogen purge at 200"C before introducing Cr(acac)3 into the reactor. Hacac and Hthd were vaporized at

60"C and SO"C, respectively. The vaporized amount of Cr(acac)3, Hacac or Hthd was always

kept sufficiently high to ensure surface saturation.

The Hacac-blocked sample and the samples with a toading of one and seven ALE

cycles were also calcined at 500"C for 16 h in air to check the influence of the calcination step on the crl+ lef'+ distribution.

2.2 Catalyst characterization

The carbon content after Cr(acac)3 adsorption was determined using a Leco CR12

carbon analyzer. The Cr concentration of the samples was measured by instromental neutron

activation analysis (INAA). cf'+ content of the catalysts, after ligand removal, was

determined in basic solution as chromate by UV-VIS-spectrophotometry [27]. Extra heating

111

Chopter 6

of the samples in a muffle fumace in air at 600"C for an hour preceded the dissolution of the

ct+ species. The number of Cr atoms per nor (Cr/nm2) and wt% of Cr metal will be used

to indicate the Cr concentrations of the catalyst, where an average specific surface area of

190 m2/g was taken for the computation.

Specific surface areas, according to the BET metbod by nitrogen adsorption, and pore

volumes were determined using Micromentics ASAP 2400 equipment. X -ray diffraction

(XRD) measurements were made with a Siemens Diffrac 500 diffractometer using Cu Ka

radiation. For the scanning electron microscopy (SEM) measurements the sample particles

were embedded in epoxy resin and cleaved to expose surfaces inside the particles.

Measurements were carried out with a Jeol JSM 840A scanning electron microscope

combined with a PGT IMIX m energy dispersive X-ray (EDS) and image analyzer.

X-ray photoelectron (XPS) spectra were measured with an X-probe model 101

spectrometer (Surface Science lnstruments,, VG Fisons) equipped with a monochromatized

Al Ka X-ray source. The samples were inertly transferred to the equipment after cooling the

samples in flowing nitrogen. During the XPS analysis Cf+ reduction by the impinging X­

rays was minimized by a short analysis time. All binding energies were referenced to the Al

2p signal for A120 3 at 74.5 eV.

Low-energy ion scattering (LEIS) measurements were performed using the LEIS­

machine, NODUS. Fora description of its basic design, as it is used nowadays, and a review

of the applications of the LEIS technique to oxides, the reader is referred to Ref. [28]. The

primary 4He+ ions are generated in a Leybold IQE 12/38 souree and are directed

perpendicularly onto the target. Ions scatteredover 142° are energy selected by a modified

cylindrical mirror analyzer (CMA). Charging of insulating materials can be effectively

compensated for by flooding from all sides with low-energy electrons. The base pressure is

in the low l Q-9 mbar, which would increase during the operation of the ion beam to the high

10·9 mbar. This is due to the inert gas of the ion beam, which will not influence the analysis

results. All catalysts powders were pressed into wafers in an ambient atmosphere and prior

to analysis chemically cleaned by a calcination step at 200'C in 30 mbar of oxygen in a

connecting chamber.

2.3 Activity Measurements

The activity of the catalysts was studied in dehydrogenation of i-butane at atmospheric

pressure at 54QOC in a fixed bed flow reactor using a catalyst charge of 200 mg. i-Butane,

diluted with nitrogen (1:10), was used as a feed. Prior to testing, the catalysts were first calcined in-situ with synthetic air at 600"C for 16 h, and then reduced with diluted hydrogen

112

Suiface Analysis and Activity of Chromialalumina Catalysts

(1:10) at 540"C for 0.5 h. The product distribution was monitored with an on-line FT-IR

analyser. Spectra were measured for 1 s at a time interval of 8 s. The multicomponent

metbod used in the analysis of measured IR spectra has been described in detail elsewhere

[29,30]. i-Butene along with some cracking and isomerization products were detected in the

product flows. The activity was expressed as conversion of i-butane from the time period of

30-60 s time on stream. This means the results describe the initial activity.

3. RESULTS

The chromium concentrations and UV-VIS-spectrophotometry results of the catalysts as a function of number of the ALE reaction cycles.

113

Chapter 6

3.1 Catalyst preparation

The first ALE sequence of Cr(aca.c)3 led toa sample containing 1.3 wt% of Cr. This

corresponded toa saturation density of 0.8 Cr/nm2• The carbon concentration of the sample

was also determined before ligand removal to study the binding of Cr(aca.c)3 to the surface.

The value of 3.4 wt% C, giving a C/Cr ratio of 11, implied that one ligand was released in

the surface reaction teaving two ligands bound to the chromium center. To form the final

oxide surface the ligands were removed by heating the sample in air at 600"C. During this

treatment the color of the sample cbanged from green to yellow indicating that most of cii+ oxidized to er>+. To study the catalysts at even lower Cr loadings with an optimal dispersion

a saturative sequence ofHthd or Hacac preceded the Cr(aca.c)3 sequence. In this way part of

the surface was effectively blocked for Cr(aca.c)3 resulting in samples of 0.16 and 0.32 wt%

Cr with Haca.c and Hthd, respectively (see Table 1). The use ofblocking agents for this kind

of atomie engineering with ALE bas been earlier described by Haukka et al. [23].

..... .~ . I I I

0 100 200 300 position (~J.m)

Fig. 2. EDS line scan through a partiele afterseven Cr(acac)1 and air cycles.

By repeating the ALE sequences of Cr(acac)3 at 200'C and air at 600"C the Cr toading

of the catalysts could be increased from 1.3 wt% to 8.8 wt% after seven ALE cycles (Table

1 ). As Fig. 1 shows the toading increases linearly with the number of the cycles, which was

114

Surface Allalysis and Activity of Chromia/alumina Catalysts

also the case for NiO/'Y-alumina system prepared by ALE using Ni(acach and air [26]. This

indicated that about the same amount of chromium was bound to the surface after each

Cr(acac)3 sequence. The color of the samples changed from yellow todark brown with

increasing chromium concentration. The average surface density of the highest loadings, 7.4 wt% and 8.8 wt%, was 4.5 and 5.4 Cr/nm\ respectively. These values exceeded the

monolayer coverage of impregnated chromia/'Y-alumina catalysts of 3. 7 Cr/nl:lt as reported

by Wachs et al. [7]. The valenee state distribution of cr6+ /Ct+ as a function of loading was measured by

measuring the cr6+ content by UV-VIS-spectrophotometry, where it was assumed that only

ct+ and cr6+ were present in the oxidized catalysts. The results are shown in Fig. 1 and

Table 1. ct+ could be assigned in all samples including the blocked ones with lowest

loadings. The amount of cr6+ in the samples increased during the fust cycles but seemed to

level off in the two samples with the highest Cr concentration. Since Wachs et al. [7] stated that Cr can be fully oxidized to cr6+, a calcination at 500"C for 16 h was applied to the

catalysts with a toading of 0.16 wt%, 1.3 wt% and 8.8 wt%. However, UV-VIS

spectrophotometry did not show any effect on the cr6+ content by the calcination.

Numberof Cr ct+ ct+ Cf surface pore Cr(acac)3 and (wt%) UV-VIS UV-VIS XPS area volume

air cycles (wt%) (%) (%) m2/g cm3/g

support 195 0.49 11 Hacac-blocked 0.16 0.13 81 11 Hthd-blocked 0.32 0.26 81 1 1.3 1.0 77 66 200 0.49 2 2.5 1.6 64 66

3 3.7 2.2 59 63 195 0.47 4 5.0 2.4 48 45 5 6.5 2.6 40 50 6 7.4 2.7 36 40 185 0.42 7 8.8 2.7 31 38 185 0.42

Table 1. Cr and cr6+ concentradons ofthe CrOJ'Y-Al/)3 catalysts.

115

Fig. 3.

5

--e 4 ctS -~ ·c;; 3 c (I) -c 'ï Cf) 2 w ...J

1

1000

Chapter 6

1500 2000 Final energy (eV)

LEIS spectra ofthe Cr0,/"{-Alz(J3 catalysts. (3 keV 4He+)

8.8%

7.4%

6.5%

3.7%

2.5%

1.3%

0.32%

2500

Although all the catalysts had a homogeneons appearance with respect to color, they were studied by SEMIEOS to ensure penetration of chromium into the heart of the support particles. No agglomerates could be distinguished throughout the support using SEM in a backscattered imaging mode. Furthermore, EDS line scans measured through the cleaved surfaces confirmed that chromium was evenly distributed through the particles. An example of the Cr distribution measured by EDS is shown in Fig. 2 for the sample of the highest Cr loading (8.8 wt% ).

The specific BET surface areas and pore volumes are listed in Table 1. The results showed that the surface area of alumina could be maintained since the decrease in the valnes

116

Surface Analysis and Activity of Chromialalumina Catalysts

corresponded to the weight increase of the catalysts calculated as C~. XRD measurements did not show any evidence of formation of an a-Cr20 3 or other chromium phase. It can be

concluded that the ALE preparation resulted in catalysts where the catalytically active material was distributed homogeneously through the catalyst and no large agglomerates of

chromia were formed.

1.00 0

0

0 0.75 -as 0 Al/0 .... as c.: 0.50 0)

Cl) • Cr/0 en • w _J 0.25

0.00 0 2 4 6 8 10

wt.% Cr Fig. 4. Cr/0 and AllO LEIS-signa/ ratios ofthe CrOxl-y-Al/)3 catalysts.

3.2 Surface characterization

To obtain a more detailed picture of the chromium dispersion besides the fact that the

distribution is homogenous, surface analysis by XPS and LEIS was needed. LEIS is the more surface sensitive technique, prohing only the topmost atomie layer [28], while XPS provides chemical information in addition to the surface composition. The surface analysis relles on

117

Chapter 6

the assumption that the external surface is a good representation of the internal surface. XPS

measurements after grinding of the 7.4 wt% sample showed, within experimental error, no

deviation from the unground sample, confirming the good homogeneity of the catalysts. The relative abondance of Cr'+ and ct+ in the samples was determined by

deconvoluting the Cr 2P3n peak. The amount of ct+ was in agreement with the UV-VIS-spectrophotometry measurements of samples, within the experimental error (Table

1). The binding energy values measured by XPS of 517.2 ± 0.2 eV for cr-3+ and 579.8 ± 0.3 eV for er>+, respectively, corresponded well with the values reported by Grünert et al. [31] and by Rahman et al. [15] for chromia/alumina catalysts.

30

/ '* •

20 • -c: 0 U) ..... <D > c:

10 0 • 0

0 0 2 4 6 8 10

wt.% Cr

Fig. s. Initia/ activity of the chromialalumina catalysts in dehydrogenation of i-butane.

The results of the LEIS experiments are shown in Figs. 3 and 4. For the analysis a

light ion (4He+) was used and the ion flux on the sample was kept as low as possible by

118

Suiface Analysis and Activity of Chromia/alumina Catalysts

measuring only the interval of interest. Time-dependent measurements showed that the influence of danlage by the ion beam on the presenled data could be neglected. Tbe spectra were calibrated to the oxygen signals where it is assumed that the oxygen signal was constant for the loadings studied. Tbis bas been proven valid for a number of systems in earlier studies [19,25,32]. Tbe ratios were determined in three different ways using peak heights, peak areas after linear background subtraction and peak areas using a semi-empirical model fust introduced by Nelson [33] and adapted for our purposes [34]. All three methods showed the same trend and the average was taken for Fig. 4. The results show a linear increase in the Cr signal up toa loading of about 7.5 wt% Cr after which the decrease in the rate of increase indicated multilayer growth. The LEIS results show that, using ALE, a monolayer coverage was reached at about 4.5 Cr/nm2

• The catalysts show a higher dispersion than the catalysts prepared by impregnation techniques, where a maximum coverage of 3. 7 Cr/nrrf was achieved for full dispersion [7].

3.3 Activity in dehydrogenation of i-butane

The initial activity of the catalysts, is shown in Fig. 5. Our results describe a linear increase in activity with chromium loading. The linear increase from the low loadings, where isolated CrOx species are assumed, to monolayer coverage, suggest that mononuclear Cr species acts as an active site

In the interpretation of the results it should be noted that the reported activity includes also cracking and isomerization activity, which occurred in addition to dehydrogenation. Both alumina and chromium oxide surface can be involved in these side reactions. On pure alumina a conversion of 7% was found. The selectivity to i-butene was also measured. On chromium oxide catalysts the selectivity to i-butene increased with Cr toading from 43% to 82%.

4. DISCUSSION

4.1 Dispersion of cbromia/alumina prepared by ALE

ALE is based on the use of saturating self-terminating gas solid reactions. In the case of Cr(acac)3 on ')'-alumina the saturation level under the used reaction conditions was found to be 0.8 Cr/nm2

• On the basis of the carbon content measured aftera single sequence of Cr(acac)3 it seemed that one of the acac ligands is released when the molecule binds to the

119

Chapter6

surface. These results coincide well with the earlier ALE work with ,S-diketonates. In agreement with our experiments on alumina a saturation density of Cr(acac)3 on silica of

0.8/nm2 was found, while also in this case one of the ligands of the Cr(acac1 was released

[23,24]. In the ALEprocessof Ni(acac1 on alumina a saturative level of 2.0 Ni/nm2 was

reported. But in that case the acac/Ni ratio was one. Therefore, more Ni can be deposited

per ALE cycle [26].

The use of sequential reactions led to a series of samples in which the concentration

increased linearly, indicating that the saturation level is the same for every cycle. This linear

behavior is not a self evident property of the ALE method, since the surface has changed

after every ALE cycle, but a property of the reagent/support combination in question.

Apparently in the Cr(acac)3/alumina system the steric bindrance by the acac groups, and not

the number of bonding sites, determined the saturation level. The work of Haukka et al. (20)

shows that the adsorption of Cr(acac1 on silica occurs selectively through reaction with OH

groups. This would suggest that OH groups behave as bonding sites on alumina too. But in

addition, the adsorption of Cr(acac)3 to coordinatively unsaturated aluminum and oxygen

atoms, i.e. to Lewis sites, existing on dehydroxylated alumina cannot be ruled out. When the

bonding would be determined solely by the OH groups, the formation of new OH groups

during air treatment after the Cr(acac)3 sequences would be required.

XRD, SEMIEDS and BET measurements showed that the chromium oxide species

was homogeneously distributed throughout the catalyst particles. The linear increase of the

LEIS signals from the lowest loadings (0.16 wt% and 0.32 wt%), where a monodisperse

dispersion was guaranteed by the use of btocking agents, to a toading of 7.4 wt% after 6

ALE cycles, indicated a dispersion of the Cr species until a monolayer coverage at about 4.5

Cr/nm2 was reached. The highest toading showed no further increase of the LEIS signal

pointing to cluster growth. XPS confirmed the homogenons deposition of the Cr species and

XRD gives no evidence of the formation of any chromia phases. The results showed that due

to the controlled deposition of Cr on the alumina surface in the ALE process a higher

coverage can be reached at full dispersion than by conventional impregnation techniques

where, aftera surface concentration of about 4 Cr/nm2 agglomeration has been reported [3-7]

The LEIS measurements showed that the ALE growth of CrOx on alumina differs

from that of NiO [25] although for both systems ,S-diketonates were used as a precursor for

the deposition of the catalytically active species [26]. For both cases it was observed that

during the fust ALE preparation cycle an atomically dispersed species was formed on the

alumina support. However, the growth ofNi(acac)2 on nickel oxide started during the second

ALE cycle [25], while Cr(acac)2 preferred bonding to the alumina up to six reaction cycles.

120

Suifoce AMlysis and Activity of ChromialalumiM Catalysts

4.2 Oxidation state of Cr in chromia/alumina prepared by ALE

Until now the ALE prepared catalysts have shown a similar behavior as the

conventionally impregnated catalysts, namely a monolayer spreading. One exception was that,

due to the ALE process a higher monolayer coverage could be reached. When the growth

mechanism was considered in the light of oxidation states of chromium a completely different

picture emerged ooropared to the conventional impregnated catalysts.

1t is generally accepted that on calcined catalysts at low loadings tbe chromium is

anchored on tbe alumina surface as cti+ (e.g. Refs. [3,9,12,19,35]). At higher loadings the

opinions about the valenee state distribution of the chromia are divided. For example,

Vuurman et al. [2-4] showed that up toa monolayer (4 Cr/nm~ all the chromium oxide

species are stabilized in oxidation state 6+ on the alumina support at moderate calcination

temperatures ( < 500"C). The work of Kozlowski et al. [36] showed that 60% of Cr was

found in oxidation state 6+ at a monolayer coverage (3.9 Cr/nm~ by chemical analysis but

due to the domination of Cr20 3 features in both EXAFS and XANES spectra a more detailed

study was not possible. Also Rahman et al. [15] reported cfl+ at the same loadings. They

pointed to calcination-induced rednetion of er>+ due to a relative high calcination

temperatures of 600"C as an explanation for this behavior. Above monolayer coverage the

picture is consistent. The occurrence of ct+ is related to the formation of a-Cr20 3• In

addition, highly dispersed, isolated species containing cfl+ were reported by Fouad et al.

[12]

Differently than with the just mentioned procedures, ALE catalysts showed the

presence of cfl+ already at low loadings of the samples, both according to UV-VIS-spectrophotometry and XPS measurements. Furthermore, although the amount of

cf+ increased with increasing loading, even at a monolayer coverage only 35 % of the Cr

was found as ct+. The formation of cfl+ in our case may be connected to the high stage

of dehydroxylation of alumina, since in literature the role of hydroxyl groups in stahilizing

the monolayer and different chromate structures bas been emphasized. In the work of Turek

et al. [6] and Vuurman et al. [2-4] it was shown that the increase in the chromium

concentration up to the monolayer coverage can be seen as a consumption of the alumina OH

groups. Wachs et al. [7] more generally proposed that monolayer coverage is typical for each

transition metal/support pair, one of the determining factors being the surface hydroxyl

chemistry of the support. Weckhuyzen et al. [10] showed by the study of Cr on SiQz,

SiQz•Al20 3 and Al20 3 that the Cr dispersion on the different supportscan be explained by

the amount and the characteristics of the hydroxyl groups. The more OH-groups available

the better the dispersion and the acidic Cr03 has a tendency to react more easily with the

121

Chapter 6

basic OH-groups on the alumina, again resulting in a better dispersion. Also, Fouad et al. [11] stated that the high surface OH density of alumina prevents the polymerization of surface CrOx species to a-Cr20 3 bulk phase and thus ensures the stability of dispersed high valenee species.

In addition to studies on valenee state distribution after the fust calcination, the chromia/alumina system has been studied intensively after reduction/recalcination processes. Weckhuysen et al. [8-10] studied chromialalumina catalysts of 0.2-0.8 wt.% of Cr showing only chromate species on the surface after calcination at 72fi'C and 55fi'C. Upon rednetion with co these cts+ species were reduced to ct+ but further reduction was mainly prevented by a stahilizing effect of the alumina. It was proposed from 27 Al MAS NMR results that part of the ct+ is incorporated in the vacant octahedral sites of the spinel surface. Since the ionic radü and charge of ct+ and Al3+ are similar, ct+ diffusion into vacant octahedral Al3+ sites in the alumina is facilitated [9]. Altemativel.y it was suggested [10] that due to the structural similarity of the alumina and chromia an epitaxial growth of chromia on the -y-alumina spinel structure could be envisaged, which then would have a stahilizing effect on the Cr-species. A similar model for an "epitaxial" growth of V20 5 on -y-Al20 3 has also been proposed [32,37]. Upon recalcination at 550"C, Weckhuysen et al. [9,10] observed that most of the ct+ could be reoxidized to cts+ but the isolated or dispersed ct+ resisted oxidation, tbis effect was also assigned to the stabilization of the Cr-species by the alumina support. The work of De Rossi et al. [13] supports tbis observation. Upon recalcination they found that a small fraction of ct+ produced in the reductions could not be reoxidized on alumina whereas on silica and zirconia systems the Cr was oxidized reversibly. They suggested that the ct+ species were less exposed on the surface of -y-A~03•

Our results showed that after the ALE process a recalcination at 500"C in air for 16 h did not change the ct+ tets+ ratio indicating high stability of the ct+ species even without any rednetion treatment. The high stability of ct+ in the ALE-prepared catalysts could originate from the ALE process, since no calcination-induced rednetion of cts+ in the air-treatment at 600"C could be expected to this extend, i.e. it is unlikely that the ct+ species adsorbed during the Cr(acac)3 sequence went through oxidation state 6+ back to oxidation state 3 + during the ALE preparation cycle.

If we now consider our results in the light of the lirerature then the following growth mechanism can be proposed. The ALE process on the alumina ensured a high dispersion over the support and stahilized ct+ already at low loadings of Cr. When already during the fust cycle ct+ would be incorporated into the spinel structure, following the work of Weckhuysen et al. [9], all Cr would still be visible by LEIS. Jacobs et al. [38] have shown that the surfaces of spinels, including -y-Al2~, expose the octahedral sites only. This would

122

Surface Analysis and Activity of Chromia/alumina Catalysts

greatly facilitate the incorporation of CrH into the octahedral sites on the surface or

altematively, when an epitaxial growth would be suggested, the placement of CrH in close

vicinity of octahedrally coordinated All+. The following ALE sequences to increase the Cr toading on the surface ensure high dispersion and in combination with an air-treatment step

after every Cr(acac)3 sequence, the er+ bas the possibility to be stabilized, resulting in a

relatively high fraction of Cr in the oxidation state of 3 +. Due to strong chromia-alumina

interaction this stabilization made the formation of er+ irreversible.

4.3 Catalytic activity of chromia/almnina catalysts prepared by ALE.

The catalytic activity measurements in this investigation supported the conclusions of

DeRossiet al. [13] and Gorriz et al. [18]. Gorriz et al. [18] showed from a combined use

of XPS, microgravimetrie and activity data in the catalytic dehydrogenation of propane, that

the catalytic activity is related to ei'+. They furthermore reported an increase of the catalytic

activity parallel with the surface Cr content as measured by XPS. After a monolayer

coverage was reached (- 4 Cr/nm2), further increase ofthe Cr-loading was accompanied by

a much lower increase in the dehydrogenation activity. However, since the number of data

points is limited, it cannot be concluded that the activity of the Cr species in the second layer

is lower. De Rossi et al. [13] studied the propane dehydrogenation on chromialalumina

catalysts. Chromia/-y-alumina catalysts with a toading up to toading of 1.00 Cr/nrnl showed

a linear increase in the catalytic dehydrogenation activity of propane in accordance with our

findings. According to De Rossi et al. [13] the linear correlation observed between

dehydrogenation activity and chromium concentranon suggested mononuclear chromium

species as the active site. This was consistent with our results which showed that the

catalysts, where a dispersion was guaranteed by the use of blocking agents, and the catalysts

with higher loadings showed the same trend. As also Wachs [7] pointed out, when special

configurations of the active sites, such as pairs, would be important, one would expect the

turnover frequency of the dehydrogenation per Cr atom to change with loading.

In recent lirerature the dehydrogenation activity was related to ct+ species. De Rossi

[13] suggested that the active er+ with coordinative vacancies originated from cf+ species

by reduction, whereas Gorritzet al. [18] stated that the amount of cfi+ can be related to the

initial activity, while the conversion stability before deactivation is related to the dispersion

of the chromium oxide phase. They furthermore stated that due to a preteduction of the

catalysts the participation of cfi+ in the dehydrogenation of propane could be excluded. In

the present work, the LEIS measurements showed that all the chromium sites were available

for the molecules in the catalytic reaction up to the monolayer coverage. Thus, if cfi+ were

123

Chapter 6

the active precursor, one would expect to find a correlation between the amount of er'+, determined by UV-VIS and XPS, and the catalytic activity. However, the results showed that the catalytic activity followed the total amount of Cr on the surface instead of the ct+ content. This could suggest that cf+ is not exclusively the precursor for the active crl+ species. However, according to Wachs et al. [7] only a small part of the Cr-species may be active.

The catalytic behavior of the catalyst above monolayer coverage could not be assessed. Although only one data point above monolayer coverage (8.8 wt%) was studied, it was clear that since the number of mononuclear sites on the surface could no longer increase, a further increase in catalytic activity should be related to multilayered Cr-species. However, the limited dataforthese high coverages prohibit further speculation. Therefore, experiments at even higher loadings are in progress to shed a light on this phenomenon. Furthermore, catalysts prepared by ALE and impregnation techniques will be compared to study the influence of the preparation in more detail.

5. CONCLUSIONS.

The chemisorption of Cr(acac)3 on -y-Al20 3 in the ALE process proved independent of the toading of Cr by preceding ALE sequences. A saturative coverage of 0.8 Cr/mrr was found. It is believed that the steric bindrance of the acac groups is the determining factor, since the number of available binding sites is large.

The ALE preparation metbod resulted in a homogeneously distributed Cr coverage tbrooghout the catalyst particles even after seven ALE sequences. BET surface and pore volume measurements were in agreement with the weight increase of Cr.

Surface characterization by LEIS showed a monolayer dispersion up toa 4.5 Cr/nm2•

It was shown that for the catalysts prepared by ALE a higher dispersion could be reached than using impregnation methods [7]. The dispersion of Cr on the -y-A~03 support was very different from the growth mechanism of Ni on Al20 3 in the preparation from vapor phase by ALE. In the Ni case the formation of clusters was observed by LEIS [25] already during the second ALE sequence.

XPS and UV-VIS spectrophotometry showed that already from the fust ALE sequence on, a significant amount of the Cr-species was in the ct+ oxidation state. 1t is believed that due high dispersion on the dehydrated -y-Al20 3 surface in the ALE process, the Cr could be more easily stabilized than in wet-chemical methods where the stabilization will only occur aftera few oxidationlreduction cycles [9,10,13]. This stabilization could be as octahedrally

124

Surface Allillysis and Activity of Chromia/alumina Catalysts

coordinated crl+ diffused in the vacant AlH positions of the defect spinel structure of the

support or altematively as an stabilization by an epitaxial growth of Cr-species on the "(­

Al203 spinel [9, 10]. The catalytic activity increased linearly with the Cr-loading. This pointed to a

mononuclear Cr-center which determines the catalytic activity. The present results showed

that this can be extended to monolayer coverage. Experiments, where by the use of Hacac and Hthd blocking agents optimal dispersion of the Cr-oxide species was ensured, were in

agreement with these findings.

References

[1] C.P. Poole Jr., and D.S. Maclver, Adv. Catal. 17 (1967) 223 and references there in. [2] M.A. Vuurman, D.I. Stufkens, A. Oskam, J.A. Moulijn and F. Kapteijn, J. Mol. Catal. 60 (1990)

83. [3] M.A. Vuurman, F.D. Hardcastie and I.E. Wachs, J. Mol. Catal. 84 (1993) 193. [4] M.A. Vuurman, I.E. Wachs, D.I. Stufkeus and A. Oskam, J. Mol. Catal. 80 (1993) 209. [5] D.S. Kim and I.E. Wachs, J. Catal. 142 (1993) 166. [6] A.M. Turek, I.E. Wachs and E. DeCanio, J. Phys. Chem. 96 (1992) 5000. (7] I.E. Wachs, G. Deo, M.A. Vuurman, H. Hu, D.S. Kim and J.-M. Jehng, J. Mol. Catal. 82 (1993)

443. [8] B.M. Weckhuysen, L.M. De Ridder and R.A. Schoonheydt, J. Phys. Chem. 97 (1993) 4756. [9] B.M. Weckhuysen, A.A. Verberckmoes, A.L Buttiens and R.A. Schoonheydt, J. Phys. Chem. 98

(1994) 579. [10] B.M. Weckhuysen, L.M. De Ridder, P.J. Grobet and R.A. Schoonheydt, J. Phys. Chem. 99 (1995)

320. [11] N.E. Fouad, H. Knözinger, M.I. Zaki and S.A.A. Mansour, Z. Phys. Chem. 171 (1991) 75. [12] N.E. Fouad, H. Knözinger and M.I. Zaki, Z. Phys. Chem. 186 (1994) 231. [13] S. De Rossi, G. Ferraris, S. Fremiotti, E. Garrone, G. Ghiotti, M.C. Campa and V. Indovina, J.

Catal. 148 (1994) 36. [14] D. Cordischi, V. Indovina and M. Occhiuzzi, Appl. Surf. Sci. 55 (1992) 223. [15] A. Rahman, M.H. Mohamed, M. Ahmed and A.M. Aitani, Appl. Catal. A121 (1995) 203. [16] S. Köhler, PhD. Thesis, Ruhr-University Bochum, Germany, 1994. [17] S. Köhler, M. Reiche, C.Frobel and M. Baems, in Preparation of Catalysts VI: Scientific Bases for

the Preparation of Heterogeneous Catalysts 3, p. 231, Ed. G. Poncelet, Elsevier Science B.V., Amsterdam, 1995.

[18] O.F. Gorriz, V. Cortés Corberán and I.L.G. Fierro, Ind. Chem. Eng. Res. 31 (1992) 2670. [19] S.J. Scierka, M. Houalla, A. Proctor and O.M. Hercules, J. Phys. Chem. 99 (1995) 1537. [20] T. Suntola, Mater. Sci. Rep. 4 (1989) 261. [21] T. Suntola, in Handhook of Crystal Growth 3, p. 601, Ed. D.T.J Hurle, Elsevier Science B.V.,

Amsterdam, 1994. [22] E.-L. Lakomaa, Appl. Surf. Sci. 75 (1994) 185. [23] S. Haukka, E.-L. Lakomaa and T. Suntola, Appl. Surf. Sci. 75 (1994) 220. [24] S. Haukka, E.-L. Lakomaa and T. Suntola, Appl. Surf. Sci. 82/83 (1994) 548. [25] J.-P. Iacobs, J.G.H. Reinges, H.H. Brongersma, L.P. Lindfors and 0. Jylhä, Catal. LeU. 25 (1994)

315. [26] M. Lindblad, L.P. Lindfors and T. Suntola, Catal. Lett. 27 (1994) 323. [27] S. Haukka and A. Saastamoinen, Analyst 117 (1992) 1381.

125

Chapter 6

[28] H.H. Brongersma, P.A.C. Groenen, J.-P. Jaoobs in Science ofCeramic 1ntofaces 11, p. 113, &IJ. Nowotny, Mat. Sci. Monographs 81, Elsevier Science B.V., Amsterdam, 1994.

[29) P. Saarinen and J. Kauppinen, Appl. Spectrosc. 45 (1991) 953. [30) A. Halruli, A. Kytökivi, E.-L. Lakomaa, and 0. Krause, Allal. Chem. accepted for publication. [31] W. Griinert, E.S. Shpiro, R. Feldhaus, K. Anders, G.V. AntoshinandKh.M. MÎJlachev, J. Catal.lOO

(1986) 138. [32] J.-P. Jaoobs, G.C. van Leerdam and H.H. Brongersma, in Fundamental Aspects of Heterogeneaus

Catalysis Studied by Partiele Beams, p. 399, Eds. H.H. Brongersmaand R.A. van Santen, NATO ASI series B 265, Plenum Press, New York, 1991.

[33] G.C. Nelson, J. Vac. Sci. Teclm. A4 (1986) 1567. [34] J. Dri:mal and V. Koksa, in preparation. [35] K. Jagannatban, A. Srinivasan and C.N.R. Rao, J. Catal. 69 (1981) 418. [36] R. Kozlowski, Bul. Polish Acad. Sci. Chem. 35 (1986) 365. (37] A.W. Stobbe-Kreemers, G.C. van Leerdam, J.-P., Jacobs, H.H. Brongersma andJ.J.F. Schotten, J.

Catal. 152 (1995) 130. [38] J.-P. Jaoobs, A. Maltba, J.G.H Reinljes, I. Drimal, V. Ponec, and H.H. Brongersma, J. Catal. 147,

(1994) 294.

126

AAS Hacac AES ALE CMA CVD EARISS EDS ERISS ESCA HDS HSA ICISS IR

ISS LEED LEIS MBE MECH MOCVD NMR PV SA SDD SEM SIMS SMSI Hthd TOF-SARS UHV UV-VIS XPS XRD XRF

=

=

=

=

= =

=

=

=

=

= =

=

=

=

= = =

=

=

=

=

=

=

=

Acronyms

Atomie Adsorption Speetrometry Acetylacetone (C5H80z} Auger Electron Speetroscopy Atomie Layer Epitaxy Cylindrical Mirror Analyzer Chemical Vapor Deposition Energy and Angular Resolved Ion Scattering Speetroscopy Energy Dispersive Spectroscopy Energy Resolved Ion Scattering Spectroscopy Electron Spectroscopy for Chemical Analysis Hydro-DeSulferization Hemi-Spherical Analyzer Impact Collision Ion Scattering Spectroscopy Infra-Red Spectroscopy Ion-Scattering Spectroscopy Low-Energy Electron Diffraction Low-Energy Ion Scattering Molecular Beam Epitaxy Methylcyclohexane Metal Organic Chemical Vapor Deposition Nuclear Magnetic Resonance Pore Volume Specific surface Area Surface neutron Differentlal Diffraction Scanning Electron Microscopy Secondary Ion Mass Speetrometry Strong Metal Support Interaction Dipivaloylmethane (CuH200:J Time-Of-Flight Scattering And Recoil Speetrometry ffitra High Vacuum ffitra-Violet VlSible speetrop hotometry X-ray Photo-electron Spectroscopy X-Ray Diffraction X-Ray Fluorescence

127

Samenvatting

De toepassing van een katalysator, een materiaal dat een chemische reactie selectief versnelt zonder zelf verbruikt te worden, is van groot belang in de chemische

industrie. Men spreekt over heterogene katalyse wanneer katalysatoren in de vaste fase

worden gebruikt met reactanten in de vloeibare of gasfase. In de heterogene katalyse zorgt de interactie tussen de reactanten en het oppervlak van de katalysator ervoor dat bepaalde bindingen in de moleculen gemakkelijker verbroken of gevormd kunnen worden. Hierdoor kunnen chemische reacties verlopen onder mildere omstandigheden, lagere druk

en lagere temperatuur, en kunnen gewenste produkten selectief worden gevormd. Dit

heeft een aantal grote voordelen. Er worden minder strenge of minder extreme eisen aan de reactoren gesteld, het energieverbruik is lager en er zullen minder (ongewenste)

bijprodukten ontstaan. Deze voordelen leveren kostenbesparingen op en maken het proces minder belastend voor het milieu.

Uit het voorafgaande blijkt al dat het oppervlak van de katalysator een belangrijke

rol speelt en dat derhalve een gegronde kennis hiervan essentieel is voor het begrijpen van

de katalytische werking. De oppervlakte analysetechniek, lage-energie ionen verstrooiing (LEIS of low-energy ion scattering), wordt sinds kort toegepast voor het katalyse

onderzoek. Bij deze techniek richt men een bundel mono-energetische edelgas-ionen (He+, Ne+) op het oppervlak en meet men de energieverdeling van de verstrooide ionen.

In eerste benadering kan men het botsingsproces met een (elastische) biljartballenbotsing

beschrijven, waaruit volgt dat de energieverdeling in feite de massaverdeling van de atomen aan het oppervlak weergeeft. Door alleen de verstrooide edelgasionen (en niet de

geneutraliseerde deeltjes) te meten bereikt men bovendien een extreme oppervlakte­

gevoeligheid voor de buitenste atoomlaag. Zo kan zelfs voor ruwe isolerende

oppervlakken, denk bijvoorbeeld aan oxidische poeders, de chemische samenstelling van de buitenste atoomlaag worden bepaald. Het is nu juist deze laag die met de gasmoleculen

in de chemische reactie in contact komt en derhalve het katalytisch gedrag voor een groot gedeelte bepaalt.

In dit proefschrift wordt voor de eerste keer de oppervlaktesamenstelling van katalytisch actieve oxides, kwantitatief bepaald met LEIS, vergeleken met katalytische activiteit. Dit geeft nieuwe informatie over de katalytische werking van het oppervlak:. Na

een korte inleiding in hoofstuk 1 worden in hoofdstuk 2 de basisprincipes van LEIS uitgelegd. Hier wordt beschreven hoe LEIS aan ruwe isolerende materialen kan worden verricht en dit wordt toegelicht met toepassingen aan oxidische materialen met

voorbeelden uit het verrichte onderzoek en de literatuur. In hoofdstuk 3 wordt de

kwantitatieve bepaling van de oppervlaktesamenstelling uit de LEIS signalen behandeld.

Voor spinel oxides (stoichiometrie: AB:zOJ is het verband gelegd tussen de oppervlaktestructuur en de katalytische activiteit in de selectieve reductie van nitrobenzeen

128

Samenvatting

naar nitrosobenzeen. De spinelstructuur heeft een kubisch zuurstofrooster met de metaal

kationen in een tetraëdrische, viervoudige, zuurstofomgeving (atomen A) of in een

octaëdrische, zesvoudige, zuurstofomringing (atomen B). Uit katalytische

activiteitsmetingen, blijkt dat, voor Co-oxides en Mn-oxides, alleen kationen in de octaëdrische omgeving het katalytisch gedrag bepalen. Met LEIS kan worden aangetoond dat dit niet toegeschreven moet worden aan een inactiviteit van de kationen in de

tetraëdrische omgeving, maar dat de tetraëders simpelweg bijna niet aan het oppervlak

voorkomen. Daarom zijn het alleen de octaëders die het katalytische gedrag bepalen.

Onderzoek aan andere spinellen toonde verder aan dat dit een algemene eigenschap is van

de spinelstructuur. Hieruit blijkt dat zelfs voor hoog poreuze poeders het oppervlak goed

gedefinieerd kan zijn.

Tenslotte wordt het groeimechanisme tijdens de bereiding van NiO en Cr03 op "Y­

Al203 katalysatoren in hoofdstukken 5 en 6 bepaald. In het hier beschreven onderzoek zijn

de katalysatoren bereid met een nieuwe techniek: Atomie Layer Epitaxy (ALE). ALE is gebaseerd op de zelfpassiverende chemisorptie van een metaalcomplex (precursor voor het

katalytische actieve materiaal) vanuit de gasfase. Door de keuze van de precursor, het

aantal cycli, de voorbehandeling van de drager en de reactiecondities, kan de belading en

de dispersie van de actieve plaats gecontroleerd worden. Deze techniek zou in de toekomst een goed alternatief kunnen zijn voor huidige nat-chemische methodes voor de

preparatie van hoogwaardige heterogene katalysatoren. Zowel voor NiO/"Y-Al20 3 als CrûJ"Y-Al20 3 zijn actieve katalysatoren bereid waarbij de gecontroleerde depositie vanuit de gasfase met ALE garant stond voor een homogene verdeling van het actieve materiaal. Uit de correlatie van oppervlaktesamenstelling, bepaald met LEIS, en de katalytische

activiteit kan een model van de actieve plaats(en) worden bepaald. Een gedetailleerd

model voor de depositie van zowel NiO als Cr03 tijdens de bereiding van de katalysator

is opgesteld.

Concluderend is LEIS een oppervlakte-analysetechniek waarmee zelfs voor ruwe

isolerende materialen de samenstelling van de buitenste atoomlaag kan worden bepaald.

Doordat het juist de eerste atoomlaag is die voor een groot deel de katalytische

eigenschappen van de vaste stof bepaalt, is de combinatie van oppervlaktekarakterisering met LEIS samen met onder andere katalytische activiteitsmetingen zeer waardevol

gebleken.

129

Summary

The use of a catalyst, a material that selectively accelerates a chemical reaction, without being consumed itself, is of major importance in chemical industry. One speaks about heterogeneaus catalysis when the catalyst in the solid state is used with reaetauts in the liquid or gas phase. In heterogeneaus catalysis, due to the interaction of the reaetauts and the surface of the catalyst, specific bondsin the moleculescan be more easily formed or broken. In this way the chemical reactions can proceed under milder conditions, e.g. lower pressure and temperature, and with a high selectivity. This results in some major benefits. Because of the milder operating conditions, less restraints have to be put on the design of the reactors and less energy is consumed. Due to the selectivity, less (undesirable) side products will be produced. Furthermore these benefits result in reduced costs and less environmental pollution.

From the above it is clear that the surface of the catalyst plays a key role and a thorough knowledge of this surface is therefore essential for the understauding of catalytic action. Recently the surface analytical technique of low-energy ion scattering (LEIS) is applied in catalytic research. Here a mono-energetic beam of inert-gas ions (He+, Ne+) is directed onto the surface to be studied and the energy distribution of the scattered ions is measured. To a frrst approximation the scattering process can be described in terms of (elastic) billiard-ball collisions. From bere one can derive that the energy distribution gives, in principle, the atomie mass distribution of the surface. Moreover, by measuring the scattered inert-gas ions only (and not the neutralized particles), one reaches extreme surface sensitivity to the topmost atomie layer. Even for rough insulating surfaces, for example oxide catalysts, the composition of the outermost atomie layer can be determined. This is the very layer which is in contact with the gas molecules and therefore toa large extend responsible for the catalytic behavior.

In this thesis the surface composition quantitatively determined by LEIS is, for the first time, combined with catalytic activity measurements. This gives new additional information about the catalytic action of the surface. After a short introduetion in chapter 1, the basic principles of LEIS are explained in chapter 2. Here it is described how LEIS can be performed quantitatively on rough insulating surfaces and this is illustrated by examples from the current research and from the literature. In chapter 3 the quantification of the surface composition from LEIS signals is treated.

For oxide spinels (stoichiometry: A~04) the relation between the surface composition and the catalytic activity of the selective rednetion of nitrobenzene to nitrosobenzene is rationalized in chapter 4. The spinel structure consists of a cubic oxygen lattice where the metal cations are bonded in a tetrahedral, fourfold, (atoms A) or in a octahedral, sixfold, (atoms B) oxygen coordination. Catalytic activity measurements showed that, for Co- and Mn-oxides, the catalytic behavior is only influenced by the

130

Summary

octahedral coordinated sites. LEIS showed that this can not be attributed to an inactivity of the cations in the tetrabedral sites but that the tetrabedral coordinated cations are hardly present on the surface. Research on other spinel structured rnaterials showed this to be a general property of the spinel structure. It can be concluded that even for highly porous powders the surface structure can be well-defmed.

Finally the growth mechanism of NiO and Cr03 on -y-Alz03 is determined in chapters 5 and 6. Using LEIS the preparanon of a catalyst, in our case Cr and Ni on alumina, is monitored. In this investigation the catalysts were prepared by a new technique: Atomie Layer Epitaxy (ALE). ALE is based on self-passivating gas-solid interaction where a precursor of the catalytic active material is deposited from the gas phase. By the choice of the precursor, the number of ALE-sequences, the pretreatment of the support and the reaction conditions, the toading and dispersion of the active material can be controlled. This technique could be a good alternative for the present wet-chemical methods for the preparanon of high-technology catalysts. Both the Ni0/-y-Al20 3 and Cr03/-y-Al20 3 catalysts showed high activity, where the controlled deposition from the gas phase using ALE garanteed a homogeneons distrubution of the catalytic active rnaterial. The correlation between the surface composition, determined by LEIS, the catalytic activity allowed fora model of the active site(s). A detailed model for the deposition of the Ni-oxide and Cr-oxide species in the preparation by ALE is developed.

In conclusion, it is shown that LEIS can be used to determine the composition and structure of outermost atomie layer even of rough insulating surfaces. Because this is the very layer which determines, to a large extend, the catalytic behavior, the combination of surface analysis by LEIS and, amongs others, catalytic activity measurements bas proven to be very valuable.

131

Publications

LEIS and oxides (ad chapter 2)

• J.-P. Jacobs, L.C.A. van den Oetelaar, H.H. Brongersma, M.A. Valenzuela, P. Bosch, Análisis de Supeljicies de Catalizadores Heterogéneos: dispersión de iones de baja energfa, La Revista de la Sociedad Quimica de Mexico (submitted).

• J.-P. Jacobs, L.C.A. van den Oetelaar and H.H. Brongersma, 1he Outennost Atomie Layer of Oxides, Proc. Symp. Surf. Sci. Kitzsteinhorn!Kaprun, Austria, April 23-29, 1995, p. 116-120,

Eds. P. Vargaand F. Amayer, HTU-Druck, Vienna (1995).

• H.H. Brongersma, P. Groenen and J.-P. Iacobs, Application of Low Energy Ion Scattering to

Oxidic Surfaces, in Science of Ceramic Interfaces ll, p. 113-182, Ed. J. Nowotny, Material Science Monographs 81 Elsevier Science B.V., Amsterdam, (1994).

• H.H. Brongersma and J.-P. Jacobs, Applications of Low-Energy Ion Scattering to Studies of Growth, Appl. Surf. Sci 75 (1994) 133-138.

Quantif"acation of LEIS (ad chapter 3)

• J.-P. Jacobs, S. Reijne, R.J.M. Elfrink, S.N. Mikhailov, and H.H. Brongersma, Quantitative Analysis of Metal Surfaces and Their Oxides Using Low-Energy Ion Scattering, I. Vac. Sci. Technol. Al2 (1994) 2308-2313.

• S.N. Mikhailov, R.J.M. Elfrink, J.-P. Jacobs, L.C.A. van den Oetelaar, P.J. Scanion and H.H. Brongersma, Quantification in Low-Energy Ion Scattering: Elemental Sensitivity Factors and Charge-Exchange Processes, Nucl. Instr. Meth. B93 (1994) 145-155.

• I.-P. Jacobs, L.C.A. van den Oetelaar, M.I. Mietus, H.H. Brongersma, V.N. Semenov and V.G. Glebovsky, Korr!ftJ:eCTBeHHHtî aHaJI!f3 rroBepxHOCT!f crrrraBoB Nb/Ta MeTonoM

pacceSIH!fSI HOHOB H!f3KOtî :>HeprHH, Poverkhnost 5 (1993) 15-21.

• L.C.A. van den Oetelaar, I.-P. Iacobs, M.I. Mietus, H.H. Brongersma, V.N. Semenov and V.G. Glebovsky, Quantitative Surface Analysis of NbxTa1.x Alloys by Low-Energy Ion Scattering, Appl. Surf. Sci. 70nl (1993) 79.

Spinel surfaces (ad chapter 4)

• S.M. Meijers, J.-P. Jacobs, T.P. Pruys van der Hoeven, J.J.W.M. Rosink, H.H. Brongersma and V. Ponec, Col+.Al2.P4 Catalysts in the Selective Reduction of Nitroben:zene to Nitrosoben:zene, J. Catal. (submitted).

• M.A. Valenzuela, I.-P. Jacobs, B. Zapata, S. Reijne, P.Bosch, H.H. Brongersma, 1he lnfluence of the Preparation Method on the Surface Structure of ZnAlzD.,, J. Appl. Catal. (in preparation).

• J.-P. Iacobs, A. Maltha, J.G.H. Reinljes, J. Drimal, V. Ponec and H.H. Brongersma, 1he Surface of Catalytically-Active Spinels, J. Catal. 147 (1994) 294-300.

132

Supported eatalysts (ad cbapter 5 aud 6)

• A. Kytökivi, J.-P. Jacobs, A. Hakuli, J. Meriläinen and H.H. Brongersma, Surface Analysis and Activity of Chromia/Alumina Cotalysts prepared by Atomie Layer Epitaxy, J. Catal. (submitted).

• A.W. Stobbe-Kreemers, G.C. van Leerdam, J.-P. Jacobs, H.H. Brongersma and J.J.F. Schotten, Characterization of 'Y-Alumina-Supported Vanadium Oxide Monolayers, J. Catal. 152 (1995) 130-136.

• J.-P. Jacobs, L.P. Lindfors, J.G.H. Reintjes, 0. Jylhä and H.H. Brongersma, The Growth Mechanism of Nickel in the Preparation of Ni/A/1)3 Catalysts studled by LEIS, XPS, and

Catalytic Activity, Catal. Lett. 25 (1994) 315-324. • M. Skoglundh, L.A. Löwendahl, P.G. Menon, B. Stenbom, J.-P. Jacobs, 0. van Kessel and

H.H. Brongersma, Characterization of a Pt-Pd Combustion Catalysts on an AJumina Washcoat, withand without prior Hydrothermal Treatment, Catal. Lett. 13 (1992) 27-38.

• G.C. van Leerdam, J.-P. Jacobs and H.H. Brongersma, A LEIS Study of the Structure of a Molybdenum on 'Y-Alumina Cotalyst, Surf. Sci. 268 (1992) 45-56.

• J.-P. Jacobs, G.C. van Leerdam and H.H. Brongersma, Study of the Surface Structure of V/)51-y-Al/)3 Catalysts by LEIS, in Fundamental Aspectsof Heterogeneous Catalysis studled by Partiele Beams, p. 399-403, Eds. H.H. Brongersma and R.A. van Santen, NATO ASI series B 265, Plenum Press, New York (1991).

Others

• W.C.A.N. Ceelen, J.-P. Jacobs, H.H. Brongersma, E.G.F. Sengers and F.J.J.G. Janssen, Cesium DiJfusion in Sodium Borosilicate Glass studled by Low-Energy Ion Scattering, Surf. Interface Anal. (accepted).

• I.C. Fullarton, J.-P. Jacobs, H.E. van Benthem, J.A. Kilner, H.H. Brongersma, P.J. Scanlon, and B.C.H. Steele, Study of Oxygen Transport in Acceptor Doped Sanulrium Cobalt Oxide, Proceedings of 1st Euroconference on Solid State Ionics, Zakynthos, Greece, September ll-18, 1994 (in press).

• S. Boudin, J.-L. Vignes, G. Lorang, M. Da Cunha Belo, G. Blondiaux, S.N. Mikhailov, J.-P. Jacobs and H.H. Brongersma, Analytica/ and electrochemical study of passive films formed on nickel-chromium alloys: injluence of the chromium bulk concentration, Surf. Interface Anal. 22 (1994) 462-466.

• W.C.A.N. Ceelen, J.-P. Jacobs, H.H. Brongersma, E.G.F. Sengers and F.J.J.G. Janssen, OPSLAG VAN HOOGRADIOACllEF AFVAL: een studie naar de diffusie van cesium in borosilicaatglas m.b.v. lage-energie ionen verstrooiing, Nevacblad, 4 (1992) 91-97.

• R.H. Bergmans, P. Brinkgreve, H.H. Brongersma, J.-P. Jacobs, C.A. Severijns, en R.G. van Welzenis, Oppervlakte-onderzoek met EARISS, NTvN 14 (1992) 225-228.

• H.H. Brongersma, R.H. Bergmans, L.G.C. Buijs, J.-P. Jacobs, A.C. Kruseman, C.A. Severijns and R.G. van Welzenis, Developments in Low-Energy Ion Scattering from Surjaces, Nucl. Instr. Meth. B68 (1992) 207.

133

134

Epilogue/N a woord

Finally the pleasant task remains to thank all who contributed to this thesis. 1t

would have been impossible for me to do all the research described in this thesis without

the help, encouragement and discussions with so many. In particular I would like to

mention:

In de eerste plaats wil ik mijn eerste promotor Hidde Brongersma bedanken voor

zijn enthousiasme waarmee hij mij in de toegepaste wetenschap en de ionenverstrooiing

heeft ingeleid. Hidde, ook bedankt voor de vrijheid en het vertrouwen dat je me gaf om

de resultaten in binnen- en buitenland uit te dragen. Ik wil Loon van den Oetelaar en Rob

Bergmans bedanken voor de prettige samenwerking op het gebied van de katalyse en

ionenverstooiing en hun vriendschap die ik altijd ontzettend heb gewaardeerd. Paul van

Dijk, bedankt voor je enthousiaste plannen, de tripjes met je vehikel(s), en het vele

plezier. I am grateful to Des O'Mahony for proofreading my thesis and for much more.

De goede sfeer in de groep Fysica van Oppervlakken en Grenslagen heeft er zeker

toe bijgedragen dat ik altijd met veel plezier zal terugzien op mijn promotietijd. De

gezonde (al dan niet) wetenschappelijke discussies onder het genot van (al dan niet)

alcoholhoudende drankjes met Will Ceelen, Amoud Dernier van der Gon, Kees Flipse, Ronald van den Oetelaar, Alan Partridge, en Rob van Welzenis hebben me altijd veel

goed gedaan. Gerard Wijers, Wijnand Dijkstra, Hans Dalderop en de technici van de

CTD ben ik zeer erkentelijk dat al onze "briljante" ideeën ook verwezenlijkt konden

worden. Zonder de vele studenten was het heel wat saaier op het lab geweest. Ik wil

Hans van Bentum, Robert Elfrink, Mark Mietus, Stef Reijne, John Reintjes en Hans

Rosink danken voor hun bijdrage tijdens hun afstudeerperiode.

Mijn tweede promotor Vladimir Ponec wil ik bedanken voor de vele discussies

over spineloppervlakken, tevens dank aan Sylvia Meijers en Annemarieke Maltha voor de

leuke samenwerking op het gebied van de spinellen.

The cooperation of Microchemistry Ltd. and Neste Oy from Finland is gratefully

acknowledged. I would like to thank Lars Peter Lindfors, Arla Kytökivi, Olli Jylhä and

Eeva-Liisa Lakomaa for their contribution and for valuable discussions conceming atomie

layer epitaxy. The enthusiasm and expertise of the foreign guests Anantharaman, Jiri

Drimal, Ian Fullarton, Vladim Glebovsky, John Kilner, Vit Koksa, Sergei Mikhailov, Pat

Scanion and Valery Semenov has been of great help to me during the last few years.

Tambien quiero agradecer a Pedro Bosch, Luis Javier Alvarez y Miguel Angel Valenzuela

por tener la oportunidad de trabajar con ellos en Mexico.

Ook wil ik mijn familie hartelijk bedanken voor de nooit aflatende steun die ik heb

mogen ontvangen. Quiero agradecer la familia Benitez Hemandez por su hospitalidad, la

cual me ha servido mucho. Finalmente quiero agradecer a Gabriela por su amor y

paciencia. Me hace mucho ilusión compartir mi vida contigo.

135

Curriculum Vitae

Jean-Paul Jacobs is geboren op 24 januari 1967 te Heerlen. Na het behalen van het gymnasium-~ diploma aan bet Collegium Marianum te Venlo is hij in 1985 begonnen met zijn studie Technische Natuurkunde aan de Technische Universiteit Eindhoven. Tijdens

zijn afstudeerproject bestudeerde hij onder begeleiding van dr. G.C. van Leerdam en

prof. dr. H.H. Brongersma de oppervlaktestructuur van vanadia/alumina katalysatoren

met lage-energie ionenverstrooiing. Zijn ingenieursdiploma behaalde hij op 20 februari

1991. Op 1 april 1991 trad hij als onderzoeker in opleiding in dienst bij de Stichting

Scheikundig Onderzoek in Nederland (SON), financieel ondersteund door de Nederlandse Organisatie voor Wetenschappelijk Onderzoek (NWO). Onder leiding van prof. dr. H.H. Brongersma verrichtte hij zijn promotie-onderzoek binnen het Schuit Katalyse Instituut in de vakgroep Vaste Stof Fysica aan de Faculteit der Technische Natuurkunde van de

Technische Universiteit Eindhoven. De belangrijkste resultaten van dit onderzoek zijn

beschreven in dit proefschrift.

136

Stellingen

behorende bij het proefschrift

Catalytically-Active Oxides

studied by

Low-Energy Ion Scattering

door

Jean-Paul Jacobs

26 september 1995

1) De beweóng, dat een lage inelastische achtergrond in LEIS spectra karakteóstiek is voor verstrooiing aan metallische of electósch

geleidende oppervlakken, is onjuist.

J.E. Drawdy, G.B. Hoflund, MR. Davidson, B.T. Upchurch and DR. Schryer, Surf. Interface Anal. 19 (1992) 559

2) Het feit dat AFM topografische beelden, opgenomen in de contact­

mode, geen defecten in het kóstalrooster van de vaste stof

oppervlak vertonen, doet vermoeden dat deze techniek het

oppervlak te rooskienóg afschildert.

G. Binnig, Ch. Gerber, E. Stolt, T R. Albrecht and CF. Quate, Europhys. Lett. 3 ( 1987) 1281

3) Het verdient aanbeveling om voor de bepaling van de oppervlaktecompositie met LEIS na te gaan of matóx effecten kunnen worden uitgesloten.

Dit proefschrift

4) De katalytische activiteit van oxidische spinellen wordt alleen bepaald door de kationen in de octaedósche omgeving aangezien

alleen zij zich bijna exclusief aan het oppervlak bevinden.

Dit proefschrift

5) Het presenteren van "spin coating" op vlakke modeldragers onder

de noemer van de fabócage van hoge technologie katalysatoren voor de nabije toekomst getuigt niet van een groot realiteitsbesef.

F.H. Ribeiro and GA. Somorjai, Receuil des Travaux Chimiques des Pays-Bas 113 (1994) 419

6) Het feit dat gesputterde deeltjes na laag-energetisch ionen bombardement op een rhodium cluster niet noodzakerlijkerwijs van het oppervlak van deze cluster afkomstig hoeven te zijn, compliceert de oppervlakte analyse van kleine clusters met behulp van statische secundaire ionen massa spectrometrie (SSIMS).

NM. Reed and J.C. Vickerman, in: Fundamental Aspects of Heterogenous Catalysis studied by Partiele Beams, p. 357, Eds. HR. Brongersmaand RA. van Santen, NATO AS/ series B 265, Plenum Press, New York (1991); W.K. den Otter, HR.

Brongersma and H. Feil, Surf. Sci. 306 (1994) 215

7) De gewoonte om op internationale conferenties de karakterisering van de structuur van een éénkristal oppervlak te presenteren zonder algemene modelvorming zou zoveel mogelijk beperkt moeten worden.

Lord Rutherford, In science there is physics and stamp-collecting

8) De wetenschappelijke strijd tussen fysici en chemici wordt niet alleen gevoed door een zekere arrogantie van de fysicus maar is voor een groot gedeelte ook te wijten aan valse bescheidenheid van de chemicus.

9) Degene die door speculatie een munteenheid weet te devalueren en daarmee miljoenen verdient, zou gerechtelijk vervolgd moeten worden aangezien hij moet inzien hoeveel leed hij onder de bevolking van het desbetreffende land aanricht.

10) Dat de mens van tegenwoordig de gevoelens en ideeën van de griekse helden in de IDias van Homerus nog steeds herkent als de zijne, geeft aan dat iedere poging om de menselijke natuur te veranderen een illusie is gebleken.