Size-selectivity of Supported Silver (Ag) and Gold (Au) Catalysts: From Nanoparticles to Mass-selected Clusters

Atomic O

Mass-selected clusters

CO

Au Au Ag Ag

CO2

Nanoparticles

CO

Atomic O

Dissertation

Zur Erlangung des akademischen Grades Des Doktors der Naturwissenschaften An der Universitt Konstanz Mathematisch-Naturwissenschaftliche Scktion, Fachbereich Physik Vorgelegt von Dong Chan Lim

Tag der mndlichen Prfung: Referent: Prof. Dr. Gerd Gantefr Referent: Prof. Dr. Paul Leiderer

Konstanzer Online-Publikations-System (KOPS) URL: http://www.ub.uni-konstanz.de/kops/volltexte/2007/3142/ URN: http://nbn-resolving.de/urn:nbn:de:bsz:352-opus-31421

Cover picture; Size-selective oxidation and reduction properties of Ag and Au catalysts: from nanoparticles to mass-selected clusters

List of Publications

1. Defect formation of Au thin films on SiO2/Si upon annealing; D. CHAN LIM, I. LOPEZ-SALIDO, R. DIETSCHE and Y. DOK KIM, Philosophical Magazine, 2005, Volume. 85, No. 29, 34773486 2. Ag nanoparticles on highly ordered pyrolytic graphite (HOPG) surfaces studied using STM and XPS; Ignacio Lopez-Salido, Dong Chan Lim and Young Dok Kim, Surface Science, 2005, Volume 588, Issues 1-3, 6-18 3. Size selectivity for CO-oxidation of Ag nanoparticles on highly ordered pyrolytic graphite (HOPG); Dong Chan Lim, Ignacio Lopez-Salido and Young Dok Kim, Surface Science, 2005, Volume 598, Issues 1-3, 96-103 4. Oxidation of Au nanoparticles on HOPG using atomic oxygen; Dong Chan Lim, Ignacio Lopez-Salido, Rainer Dietsche, Moritz Bubek, Young Dok Kim, Surface Science, 2006, Volume 600, Issues 3, 507-513 5. Electronic and geometric properties of Au nanostructures on HOPG studied using XPS and STM; Ignacio Lopez-Salido, Dong Chan Lim*, Rainer Dietsche, Nils Betram, and Young Dok Kim*, Journal of physical chemistry B, 2006, Volume 110, Issue 3, 1128-1136 6. Characterization of Ag nanoparticles on Si wafer prepared using Tollens reagent and acid-etching; Dong Chan Lim, Ignacio Lopez-Salido, Young Dok Kim, Applied surface science, 2006, Volume 253, Issue 2, 959-965

7. Size-selectivity of the oxidation behaviors of Au nanoparticles; Dong Chan Lim, Ignacho Lopez-Salido, Rainer Dietsche, Moritz Bubek, Young Dok Kim, Angewandte Chemie International Edition, 2006, Volume 45, Issue 15, 2413-2415 8. Experimental studies on plasmon resonance of Ag nanoparticles on Highly Ordered Pyrolytic Graphite (HOPG); Ignacio Lopez-Salido, Nils Bertram, Dong Chan Lim, Gerd Gantefr and Young Dok Kim, Bulletin of the Korean Chemical Society, 2006, Vol 24, No 4, 556-562 9. Oxidation and Reduction of Mass-selected Au clusters (Aun, n=2-10) Deposited on SiO2/Si : every atom counts; Dong Chan Lim, Rainer Dietsche, Moritz Bubek, Gerd Gantefr, Young Dok Kim, ChemPhysChem, 2006, Volume 7 (9), 1909-1911 10. Oxidation and reduction of rough and flat Au surfaces; Dong Chan Lim, Young Dok Kim, Applied surface science, 2006, Volume 253, Issue 5, 2984-2987 11. Electronic and chemical properties of supported Au nanoparticles; Dong Chan Lim, Ignacio Lopez-Salido, Rainer Dietsche, Moritz Bubek, Young Dok Kim, Chemical physics, 2006, Volume 330, Issue 3, 441-448 12. Interaction of silver with oxygen on sputtered pyrolytic graphite S.H. Jung, D.C. Lim, J.-H. Boo, S.B. Lee, A.N. Hwang, C.G. Hwang, Y. D. Kim, Applied catalysis A: General, 2007, Volume 320, 152-158

13. Chemistry of mass-selected Au clusters deposited on sputter-damaged HOPG surfaces: the unique properties of Au8 clusters; Dong Chan Lim, Rainer Dietsche, Moritz Bubek, Thorsten Ketterer, Gerd Gantefr, Young Dok Kim, Chemical Physics Letters, 2007, Volume 439, 364368 14. Experimental and theoretical studies on materials consisting of magic Si7 clusters; Felix v. Gynz-Rekowski, Rainer Dietsche, Dong Chan Lim, Nils Bertram, Tim Fischer, Gerd Gantefr, Wolfram Quester, Peter Nielaba, Young Dok Kim, The European Physical Journal D, 2007_accepted

Contents
1 Introduction .....................................................................................................1 1.1 What is catalysis........................................................................................3 1.2 Importance of catalysts and supports........................................................4 1.3 Metal-oxygen interaction...........................................................................5 1.4 Silver and Gold..........................................................................................7 1.5 Motivation and goals.................................................................................9

2 State of the Art................................................................................................14 2.1 Perimeter model......................................................................................15 2.2 Structure of particles (geometric) model................................................19 2.3 Structure of particles (electronic) model................................................21 2.4 Others......................................................................................................27 2.4.1 Adsorption process of oxygen (dissociative or molecular).27 2.4.2 Cluster size effects..28 2.4.3 Moisture effects..30

3. Experimental setup....31 3.1 UHV 1 (for mass-selected Au clusters)......31 3.2 UHV 2 (for Au and Ag nanoparticles)....39 3.3 X-ray Photoelectron Spectroscopy (XPS)...............................................40 3.4 Scanning Tunnelling Microscopy (STM)....54 3.5 Procedure.....57

Results and Discussion

4 Ag nanoparticles on HOPG (Highly Ordered Pyrolytic Graphite)..62 4.1 Preparation of of Ag nanoparticles on HOPG ............................62 4.2 Oxidation properties of Ag nanoparticles................................................70 4.3 Reactivity toward CO of the oxidized Ag nanoparticles.........................88

5 Au nanoparticles on HOPG ......95 5.1 Preparation of of Au nanoparticles on HOPG 95 5.2 Oxidation properties of Au nanoparticles .....101 5.3 Reactivity toward CO of the oxidized Au nanoparticles.......................107

6 Au nanoparticles on SiO2/a-Si.114 6.1 Preparation of of Au nanoparticles on SiO2/a-Si .................................114 6.2 Oxidation properties of Au nanoparticles.on SiO2/a-Si .......................119 6.3 Reactivity toward CO of the oxidized Au nanoparticles .124 6.4 Rough and Flat Au surface.......127 7 Mass-selected Au clusters on HOPG ..141 7.1 Preparation of mass-selected Au clusters on HOPG ............................143 7.2 Oxidation properties of mass-selected Au clusters on HOPG..............155 7.3 Reactivity toward CO of mass-selected Au clusters on HOPG............158 8 Mass-selected Au clusters on SiO2/a-Si ......166 8.1 Preparation of mass-selected Au clusters on SiO2/a-Si.166 8.2 Oxidation properies of mass-selected Au clusters on SiO2/a-Si...........182 8.3 Reactivity toward CO of mass-selected Au clusters on SiO2/a-Si ...188 9 Current results .............................................................................................193

10 Conclusions.203

11 Zusammenfassung..207

A list of figure..212 B References229 Acknowledgement............................................................................................250

1. Introduction
Nanotechnology can been defined as the control and restructuring of a matter below 100 nanometer in size in order to create materials, devices, structures and functional systems [1], i.e. the direct manipulation of matter on the level of atoms and molecules. Restructuring nature at the nanoscale leads to materials with novel properties, which are significantly different to their larger equivalent. Strength, electrical and thermal conductivity, colour, magnetic properties, and chemical reactivity may all vary in extraordinary ways [2]. The novel characteristics of nanomaterials arise from the complex interplay between quantum physics and classical mechanics that occur in the nano realm [3, 4]. As a result, these properties and effects are often highly unpredictable. A basic understanding of the nature of these special properties opens fascinating routes to design devices and potential technological applications. Nanomaterials in the form of nanoscale powders and fibres are already being used in sunscreens, cosmetics, sports equipment, self-cleaning paints and glass, fuel additives, batteries and a range of products [2]. For example, zinc oxide and titanium dioxide particles in the size regime of 40 50 nm are transparent while still retaining the ability to block UV rays, whereas conventional powders are opaque [6, 7]. Researchers at Rutgers University in the US have been developing nanoscale iron and cobalt particles as catalysts for use in the chemical conversion of coal to diesel [5]. Especially novel metal atom clusters (nanoparticles) have attracted much attention because of their potential applications in biosensing, photonics, catalysis, information storage and singleelectron transfer [9 - 16]. Clusters are aggregates of atoms (or molecules) containing between two and a few thousand atoms. They have properties intermediate between those of the isolated monomer (atom or molecule) and the bulk or solid-state material. One of the characteristics of clusters which are responsible for many of their interesting properties is the large numbers of atoms at the surface compared to those in the cluster interior.

-1-

Even in a cluster as big as 105 atoms, almost 10 % of the atoms are at the surface. Clusters can be sorted out four categories [8]; a) Nanoclusters have 2 to 13 atoms. b) Small clusters have 13 to about 100 atoms. c) Large clusters have 100 to 1000 atoms. d) Small nanoparticles and nanocrystals have at least 1000 atoms. Gold clusters form the sites at which proteins or other organic molecules can be immobilized, leading to the intriguing possibility of nanoscale biosensors [13, 17]. In order to develop nanodevices, the electron transfer between nanoclusters and substrate is considered to be of fundamental importance, thus it is also necessary to investigate the nanoclusters-substrate interaction. In this experiment we have focussed our attention on the catalytic aspects of Ag and Au nanoparticels (size-selected clusters) deposited on a surface. The well known activity of Ag as a catalyst for selective oxidation of a number of hydrocarbons (not only ethylene, or methanol, but also styrene and methane [18 - 23]) has been widely studied. In the case of Au, there are a number of curious aspects remaining to catalysis that are attracting academic investigation such as oxidation of carbon monoxide, alkanes, alkenes, hydrochlorination of ethyne (acetylene), production of vinyl acetate monomers, liquid phase oxidation of ethane-1,2-diol and water gas shift reactions [24 - 32], while the observation that gold-based catalysts are active at room temperature and below is driving considerable industrial interest [33 - 39].

-2-

1.1 What is Catalysis?

The term catalysis (from the Greek and , roughly wholly loosening) was used by Berzelius in 1836 [40]. Armstrong proposed the word catalyst in 1885. A catalyst is an active chemical spectator. It takes part in a reaction but is not consumed. A catalyst produces its effect by changing activation barriers as shown in Fig. 1.1. By lowering the height of an activation barrier, a catalyst speeds up a reaction. It does not, however, change the properties of the equilibrated state. It is important that the acceleration of reactions is not the only key factor in catalytic activity. Catalysts can be designed not only to accelerate reactions; the best of them can also perform this selectively. In other words, it is important for catalysts to speed up the right reactions, not simply every reaction. In Fig. 1.1, the activation barrier for the desired product B is decreased more than the barrier for the undesired product C.
homogeneous reaction

Ehom > Ecat Ehom Ecat

Energy

undesired catalytic reaction

A
reactants Hr

B
desired products preferred catalytic reaction

Reaction path

Fig. 1.1 Activation energies and their relationship to an active and selective catalyst (A) Reactant; (B) desired product; (C) undesired product; Ehom, activation barrier for the homogeneous reaction; Ecat, activation barrier with use of catalyst; Hr, change in enthalpy of reactants compared with product [41]

-3-

1.2 Importance of catalysts and support

The chemical industry of the 20th century could not have developed to its present status on the basis of non-catalytic, stoichiometric reactions alone. Approximately 85 - 90 % of the products of chemical industry are made in catalytic processes. Reactions can in general be controlled on the basis of temperature, concentration, pressures and contact time. Catalysts accelerate reactions by orders of magnitude, enabling them to be carried out under the most favorable thermodynamic regime and at much lower temperatures and pressures. In this way efficient catalysts, in combination with optimized reactors and total plant design, are the key factor in reducing both the investment and operation costs of chemical processes [184]. It is also a matter of great interest in industry and at home that the use of toxic and hazardous reagents and solvents can be avoided while formation of waste or undesirable byproducts is minimized. Catalytic routes often satisfy these criteria. For example, the selective oxidation of ethylene to ethylene epoxide, an important intermediate in chemical industry, creates a waste like salt that was traditionally solved by dumping it in a river when non-catalytic synthesis route was used. However, the catalytic route is simple and clean, although it does produce a small amount of CO2. A catalyst offers an alternative, energetically favorable mechanism to the non-catalytic reaction. Summarizing, catalysts are indispensable in production of transportation fuels in the approximately 440 oil refineries all over the world in production of bulk and fine chemicals in all branches of chemical industry and in prevention of pollution by avoiding formation of waste (unwanted byproducts) [5, 184] Another important factor for the catalytic activity is the supporting material of a catalyst. Heterogeneous catalytic reactions occur in systems in which two or more phases are present, for instance, solids and liquids, or gases and solids. Liquid-solid and gas-solid interfaces are of particular interest because the solid surface gives us a place to deposit and immobilize a catalytic substance [41]

-4-

However, surfaces are of particular interest not only because they are where phases meet but also because they give us a place to put catalysts. The surface of a solid is inherently different from the rest of the solid, since the bonding at surface is different than that in the bulk. The chemical activity can be changed depending on the surface properties of the support.

1.3 Metal-oxygen interaction

Interaction of oxygen with metal has been one of the most widely studied subjects in chemistry and physics due to its implication for the corrosion process [195] and heterogeneous catalysis [194]. The metal-oxygen interaction can change as a function of particle size on the nanoscale, which can cause an interesting size-selectivity of a heterogeneously catalyzed reaction [42 - 48]. Among various metals, Au has been attracting particular attention due to the strong size-selectivity for various catalytic reactions [43, 47, 48]. Au-oxide species has been often observed from catalytically active Au nanoparticles on various oxide supports, and oxidized Au has been proposed to be a catalytically active species [26, 95, 207 - 209]. Free Au atom anions can react with atomic oxygen, forming AuO3 species, which can readily react with CO to form CO2 [210]. It was also suggested that Au-oxide (Au (III)) at the interface of metallic Au nanoparticles or negatively charged Au on metal oxide supports are key points toward CO oxidation reactions [26, 205]. In other studies, only metallic Au (0) was suggested to be a catalytically active species [48, 94]. For other catalytically active metal surfaces, the oxidation of Ru (0001) is one of the best studied systems in the literature [49 - 51], in which chemisorbed oxygen, surface oxides, buried oxides, and subsurface oxygen may coexist in the nearsurface region (Fig. 1.2), and this complexity is characteristic of the oxygen chemistry of many transition metal surfaces [42]. The formation of surface oxide layers was suggested to be related to the poisoning of the surface [194], implying that the high catalytic activities generally come from metallic part of the catalysts surface.

-5-

Fig. 1.2 The rich oxygen chemistry of Ru (ruthenium) (0001) [42] There are, however, also examples that oxide layers formed on the surface of metals increase chemical reactivity. RuO2 layers on Ru surfaces were shown to be extraordinarily catalytically active [193], and Ag-oxide was also suggested to be directly involved in heterogeneously catalyzed reactions such as partial oxidation of ethylene [196 - 204]. Oxide species cannot be easily prepared under high-vacuum conditions, yet they can readily form under high-pressure conditions, and therefore, enhanced catalytic activities of oxide surfaces are suggested to be responsible for the so-called pressure gap in surface science [52]. To shed light on the nature of the oxygen species formed on Au nanoparticles under high-pressure reaction conditions, atomic oxygen species were produced on Au bulk crystals and nanoparticles using various methods: on one hand, stronger oxidizing molecules than O2 such as NO2 or ozone were used to deposit atomic oxygen on Au surfaces, on the other hand atomic oxygen was created using hot-filaments, plasma jet equipments or electronbombardment [3, 53 61, 111, 143, 156]. Using these methods, oxygen species could be synthesized, and their nature was characterized to shed light on the elementary steps of the catalytic process under high-pressure conditions.

-6-

1.4 Silver and Gold

Silver
Silver has been known since ancient times. Silver is somewhat rare and expensive, although not as expensive as gold. It is mentioned in Genesis. Slag dumps in Asia Minor and on islands in the Aegean Sea indicate that man learned to separate silver from lead as early as 3000 B.C. Silver is one of the elements which has an alchemical symbol; Its name is originated from the Anglo-Saxon word "siolfur" meaning "silver" and the origin of the symbol Ag comes from the Latin word "argentum" meaning "silver" [62].

Gold
Gold is a very noble metal and belongs to the nine elements (Cu, Ag, Pt, Fe (!), As, Bi, S, C) which occur in nature in a pure, native state. Egyptian inscriptions dating back to 2600 B.C. describe gold. It is mentioned several times in the Old Testament, and it is also one of the elements which have an alchemical symbol; Its name is assumed to originate from the Old English word geolo (yellow), while the chemical symbol, Au, is derived from the Latin word aurum. Pure gold crystallizes in a face-centered cubic lattice (space group Fm-3m) with a lattice constant of 4.0782 [62]. The chemistry of gold is mainly dominated by large relativistic effects. The electron configuration of gold is [Xe]4f145d106s1. The s orbitals are energetically lowered and spatially contracted, the d and f orbitals are energetically raised and expanded. The 5d level is higher, the 6s level lower than a comparison with the congener Ag suggests. This relativistic effects lead to the following remarkable properties of gold [59, 62];

-7-

(a) The reduced 6s-5d separation of 2 eV (as compared to 3.5 eV for the 5s - 4d separation in the silver atom) allows an optical transition in the visible range (adsorption for < 560 nm), which causes the yellow colour of bulk gold. (b) Due to the relativistic effects, the first ionization potential, IP, and the electron affinity, EA, of Au are higher than in the case of Ag (Table 1.1). The EA of Au is the highest of all metals and ranges between the values for oxygen (1.465 eV) and iodine (3.063 eV). This may explain why Au shares several properties with the halogens, whereas Hg behaves partly like a closed-shell element. Gold vapour consists of remarkably stable Au2 molecules with a dissociation energy of 232 kJ/mol (cf. Cl2; 244 kJ/mol) [64]. It is clear from Mulliken's definition of the electronegativity, M, as the average of IP and EA that high values of IP and EA also lead to a high electronegativity. Thus, gold has the highest (Pauling) electronegativity of all metals. Au Ionization potentials (eV) Electron affinity (eV) Heats of atomization (kJ/mol) Melting point (oC) A-A distance in f.c.c. cells (), 25oC 1st 2nd 9.225 20.5 2.039 368 1063 2,8840 Ag 7.576 21.49 1.202 285 961 2,8894

Table 1.1: Ionization potentials and electron affinities of silver and gold [63] (c) The destabilization of the d levels explains the existence of high oxidation states of Au, e.g., +3 (AuCl3), +5 (AuF5), and even +7 (AuF7). The most stable oxidation state is +3, in contrast to +1 in the case of silver.

-8-

1.5 Motivation and goals

As mention above, nanomaterials have been attracting attention of chemists and physicists due to their interesting size-selectivity in catalytic properties. The size-dependent electronic structure, morphology, thermal stability, metal support interaction and chemical reactivity of heterogeneous catalysts of nanoparticles and mass-selected clusters have been studied extensively. Oxidation of nanoclusters is particularly interesting, since under real working conditions of heterogeneous catalysis, the formation of metal oxide is more probable than under high vacuum conditions. Metal oxides can be an either active or passive species for heterogeneously catalyzed reactions [65, 66]. Among various metal particles, interests on Ag and Au nanoparticles and clusters have been drastically increasing due to their high catalytic activities in various reactions well below room temperature [67, 68]. However, we are certainly not on the way to a complete understanding of the real silver and gold catalysts, since their outstanding reactivity is probably caused by a higher complexity, e.g., the influences of particle size, support, binding properties between catalysts and reactant molecules and/or others. Therefore, comparison of real catalysts with our idealised system, particularly deep understanding of the oxidation of nanoclusters, may allow us to distinguish between these influences and additional factors, which are related to the chemical natures of silver and gold. Recently, mononuclear supported Au (III) and Au (I) centers formed from ionic complexes have been shown to catalyze CO oxidation and the water-gas shift reaction under ambient reaction conditions [69, 119, 135]. In the case of CO oxidation, Guzman and colleagues [94] demonstrated that metallic gold clusters typically credited as the working catalysts are not solely responsible for the reactivity of these systems. It is speculated that larger clusters may actually serve as CO reservoirs to active cationic mononuclear centers. Another experiment also shows that the gold catalysts with both, totally the 3+ or metallic oxidation state, where found to exhibit activity [26, 94].

-9-

The catalysts were found to decrease in activity with decreasing Au3+ content up to an intermediate calcination temperature, afterward the activity increased sharply [69]. Size-selected studies have also been performed on supported small gold clusters under ultrahigh vacuum (UHV) conditions (model catalysts), revealing that deposited Au1 exhibits nearly zero catalytic activity, while Au clusters seven or eight atoms in size are quite active depending on the oxide supports [70 - 72]. This apparent discrepancy may be the result of a difference in the oxidation state and/or reaction conditions.

In the present work, we studied oxidation patterns of Ag and Au nanoparticles deposited on sputtered HOPG (Highly Ordered Pyrolytic Graphite) and chemically etched SiO2/Si surfaces with native oxide layer. Model catalysts are prepared by evaporation of bulk metals, yielding deposited nanoparticles with a certain size distribution. Considering that catalytic activities can vary strongly as a function of cluster size, model catalysts prepared in this way may be mixtures of catalytically active and non-active clusters. It cannot provide accurate information on cluster size-dependent changes of catalytic activities. For obtaining more reliable information on cluster size effects on chemical activities, the same experiments were carried out with mass-selected Au clusters with a narrow size distribution. This work can be divided into four categories; 1. Ag and Au nanopaticles on HOPG surface (Chapter 4 and 5) 2. Au nanopaticles on SiO2/Si surfaces (Chapter 6) 3. Size-selected Aun clusters on HOPG (n = 2 10) and SiO2/Si (n = 2 13) surfaces (Chapter 7and 8) 4. Current results; Size-selected Aun (n = 5 8, 12, 13) clusters on SiO2/Si surfaces with NaOHH2O (Chapter 9)

- 10 -

Point 1
Here, we show that uptakes of oxygen of smaller nanoparticles are significantly higher than those of larger particles and bulk-like metal. Furthermore, we provide evidence for formation of different oxygen species depending on Ag particle size, resulting in size-selectivity of the catalytic activity of CO oxidation. Ag2O/AgO, which is active for CO oxidation, can form only on Ag nanoparticles larger than 3 nm in diameter. On the basis of the results of Ag on HOPG and Au on HOPG in which, depending on the size of Au nanostructures, two different oxygen species were also identified and one of them can be attributed to lattice oxygen of Au oxide (active for CO oxidation) and the other comes from subsurface oxygen atoms (non-active for CO oxidation), the nature of catalytically active sites of Ag and Au nanocatalysis for CO can be interpreted by the Mars van Krevelene mechanism (a metaloxide species formed on nanoparticles can directly participate in CO oxidation reactions).

Point 2
We also provide evidence that the oxygen-Au interaction on silica surfaces drastically changes as the particle size becomes smaller than ~ 0.7 nm in particle height and ~ 1.5 nm in lateral size: a characteristic feature for the formation of Au2O3 is only visible for larger Au particles, whereas a different oxidation pattern can be found for smaller particles upon atomic oxygen exposure. The oxidized larger Au nanoparticles can be reduced by CO exposure at room temperature, whereas the reversible oxidation/reduction behavior was not found for smaller particles. Here, we also demonstrate that a rough Au thin film shows some common behaviors with very small Au nanoparticles on HOPG and silica regarding reaction with atomic oxygen, namely the oxygen species is inert (or less reactive) towards CO oxidation under our experimental conditions.

- 11 -

Point 3
It has been shown that gold particles smaller than about 3 nm and/or ~ two atomic layers are suitable for CO oxidation at temperatures as low as -70oC [74]. A large number of experimental and theoretical studies have been devoted for a better understanding of the size effects in heterogeneous catalysis of supported metal nanoparticles. Among those studies, deposition of massselected clusters on various substrates opened new insights into the sizeselectivity of heterogeneous catalysis. Cluster size effects for catalytic activities are quite complicate, since it is difficult to control the cluster size with atomic precision. Mass-selected clusters can be soft-landed onto the oxide support, which can create deposited metal clusters with identical sizes, thus give more precise information on the size dependent changes of the chemical properties of deposited metal clusters. Chemical properties were suggested to change with every additional atom. Here, for deposited Aun clusters with = 2 10 on HOPG surfaces, only Au8 was found to be active towards oxidation/reduction of CO. However, for the deposited Aun clusters with = 2 13 on SiO2/Si surfaces, we have direct evidence that each additional Au atom can drastically change the oxidation pattern of the Au clusters. That is, diverse chemical properties of mass-selected clusters are demonstrated, which cannot be observed from larger Au nanoparticles formed on the same substrate, addressing importance of the ability to control the cluster size on an atom-by-atom basis in order to tune catalytic activity. The even-odd patterns in electronic and chemical properties and cluster stabilities have been theoretically and experimentally explored for free Au clusters in the gas phase; however, such a pattern has been observed for deposited clusters for the first time in the present work [75, 76]. Compared to other Au clusters and larger Au nanoparticles on silica, chemical properties of Au5, Au7, and Au13 were shown to be very unique, also suggesting the importance of metal-support interactions for the chemistry of deposited clusters.

- 12 -

Point 4
In very recent results (size-selected Aun clusters on SiO2/Si surfaces were immersed into NaOHH2O solution), we demonstrate that size-selected Au clusters deposited on chemically etched silica surfaces can be highly stable under air even under resistance to chemical solutions. Furthermore, reverse even-odd patterns in electronic and chemical properties of Au clusters were observed by controlling the charge state of Au clusters which was done under ambient conditions. It may be worth mentioning that electronic effects of Au clusters on silica play a more important role than geometric effects toward CO oxidation reactions. In both, the model and the real catalysis system, it is difficult to prepare a surface decorated with single atoms due to the strong sintering of Au atoms on the surface at room temperature [9 - 11]. Sintering is strongly affected by the cluster-support interactions [77]. However, one can absolutely exclude the possibility of cluster sintering under our experimental conditions, implying that the pressure gap between model and real catalysts can be reduced, the control of matter on an atomic scale under ambient conditions could be possible, and one can also expect applications such as cluster materials in a field of applied science technology. That is, our results can bridge the gap between real systems and idealised systems in heterogeneous catalysis.

- 13 -

2. State of the Art

(The nature of high catalytic activity of supported Au nanoparticles for CO oxidation)

Gold is considered a rather inert metal in its bulk form. However, since the discovery of Haruta [27, 78] and Hutching [31] highly dispersed Au nanoparticles on metal oxide exhibit unexpected catalytic activity, gold catalysts hve been investigated in detail. The catalytic activity of supported gold particles can change depending on structural effects as size effect, shape, thickness as well as the oxidation state of the gold, catalyst pre-treatment, CO adsorption, oxygen adsorption, and support and moisture effects etc. Despite all the work concerning the CO oxidation reaction, mechanism of active sites for CO oxidation is still not completely understood and there is no agreement in literature. Previous works on oxide-supported gold nanoparticle catalysts suggest for a wide range of active sites. One model is that the active sites for the CO oxidation are on the gold nanoparticles and the defect sites of oxide support together (2.1 Perimeter model). Another one is that the catalytic activity is entirely related to the presence of neutral gold atoms on the gold nanoparticles. It also can be devided into two models; Geometric and Electronic models (i) The particles have a high degree of under-coordinatied atoms [46, 85 87] and also a special bonding geometry to other gold atoms that exhibits a more reactive orbital [26] (2.2 Structure of particle (geometric) model). (ii) Quantum size effects that alter the electronic band structure of gold nanoparticles [31] (2.3 Structure of particle (electronic) model).

- 14 -

Another unresolved issue is whether a positive [88, 152] or negative [47] charge transfer between the gold nanoparticles and defect sites of the support is the key point to the catalytic activity of these gold catalysts. Some researcher claims that the active sites are an ensemble consisting of metallic Au and AuOH species [95, 99]. However, it is commonly accepted that hemispherical particles compared to more spherical particles [88] and particles with a large contact area with support [47, 138] are showing higher catalytic activity. Furthermore, Au nanoparticles supported on reducible oxides (TiO2, Fe2O3) are more active towards CO oxidation reaction than Au nanoparticles on irreducible oxides (SiO2, Al2O3) [89, 141, 142].

2.1 Perimeter model

The work done by Haruta [27] showed that high catalytic activity of gold could be associated with hemispherical particles of metallic gold with mean diameters of less than 5 nm attached to reducible oxide supports. The active site for CO oxidation is the perimeter of the hemisphere [26, 47]. The general conclusion of theories related to this hypothesis is that the edge of the Au/support interface plays a key role in the mechanism of CO oxidation [79 82, 100]. This mechanism is the most generally accepted one in gold catalysis, although there are differences in details. Fig. 2.1 shows a schematic representation of CO oxidation pathways for TiO2 supported Au catalysts. Gold surfaces can adsorb CO when the diameter of Au particles is smaller than 10 nm due to an increase in step, edge, and corner sites [152].

- 15 -

Fig. 2.1 CO oxidation reaction pathways of the Au on TiO2 catalyst suggested by Haruta et al. [88] The adsorption of O2 takes place at the perimeter interface of Au and TiO2 are sites for O2 adsorption. The perimeter atoms may be in a special state controlled by electronic interactions with the oxide substrate. The strong contact of Au nanoparticles with the support increases the possibility of O2 adsorption much more than Au and TiO2 themselves [47]. The adsorbed O2 can afterwards move to step, edge, and corner sites of Au nanoparticles. It reacts with CO adsorbed on the Au surface at the perimeter interface. This fact is not due to the strong metal support interaction but due to one of the characteristic features of supported Au catalysts [89, 90]. However, one cannot rule out the importance of metal support interaction because different activities are obtained for different supports, even if the Au particle size remains unchanged [89, 91, 142].

- 16 -

Adsorption sites of oxygen species

Even if this mechanism is generally accepted, there is disagreement amongst the supporters of this model concerning the O2 adsorption sites; whether the O2 is adsorbed on the perimeter itself or the nearby oxide support. For example, theoretical calculations of Hammer et al. [101] and others [112, 115] have shown that molecularly chemisorbed oxygen at the Au/TiO2 and Au/MgO oxide surface and/or interface will readily react with CO to form CO2. Whereas Grisel [25], G.C. Bond [26] and R.J. Behm [92] found collecting and disassociating (or not) of O2 at surface anion vacancies or F centres (oxygen vacancies) for transition metal oxide [115]. R.J Behm et al. proposed that the ability of the oxide support to activate oxygen by a superoxo intermediate is crucial for the catalytic activity of oxide supported Au. Whereas the calculation results of Sanchez et al. [71] have shown that O2 should absorb as O2- onto gold atoms along the interface perimeter of Au8 on MgO. Although the O2 is adsorbed onto the oxide support, the perimeter might still play a special role.

The role of positively charged Au

In the perimeter model of Haruta, Au+ played no role in the activity of Au on TiO2. The highest activity was observed when all Aux+ had been reduced to Au0 by calcination. On the other hand, Bond & Thompson [26], amongst others, have proposed that the gold nanoparticles are attached to the oxide support via an intermediate layer of Aux+ atoms and that the oxidized state of these atoms is essential for the catalytic ability. The Au (III) species stabilize the Au nanoparticles on oxide supports, i.e. Au (III) acts as chemical glue, between metallic Au and metal oxide supports.

- 17 -

Fig. 2.2 A representation of the oxidation of carbon monoxide at the periphery of an active gold particle on metal oxide support [26] Fig. 2.2 shows a mechanism suggested by Bond and Thompson. At the left, a carbon monoxide molecule is chemisorbed on a low coordinated gold atom, and a hydroxyl ion has moved from the support to an Au (III) ion, creating an anion vacancy. At the right, they have reacted to form a carboxylate group, and an oxygen molecule occupies the anion vacancy as O2-. Then this O2- molecule oxidizes the carboxylate group, forming carbon dioxide. The resulting hydroperoxide ion HO2- then oxidizes a further carboxylate species forming another carbon dioxide and restoring two hydroxyl ions to the support surface. This completes the catalytic cycle. No attempt is made to suggest the charges carried by the reacting species. The importance of Au+ species was also supported by the some experiment studies [148 - 155], which have shown that the most active catalysts have the highest fraction of Au+ in metal hydroxide supported systems. If this is correct, then Au0, Aux+, and the metal oxide support all have a role to play and the interaction of these species at the interface between the gold and the support is particularly important.

- 18 -

2.2 Structure of particle (geometric) model

It is possible that gold clusters of equal size and similar electronic configuration could exhibit very different chemical reactivity. In this model, it has been suggested by Freund and other co-workers that, if some specific sites (under-coordinated atoms) on the surface of the material have a particular ability to catalyse a chemical reaction, then the number of these undercoordinated atoms (not introduced from quantum size effects as a result of particle thickness) is important to increase the activity of the catalyst [109, 116 118, 126, 128]. The chemical identity of the catalyst support is less important; it exists solely to pin the catalytic particles and prevent them from sintering. The work of Kung et al. [104] has shown that the actual activity of the gold was due to an exceedingly small proportion of the surface gold atoms, about one in a thousand, implying that the reaction occurs at few special sites only. Such specific sites consist of low coordinated gold atoms at steps or other defects, which have already been proposed to be considerably more electropositive and therefore readily can be ionized to Aux+ [26, 104]. Another theoretical work [73] shows that the roughness of Au is a very important factor controlling oxygen binding. O2 binds well to edges of a planar Au cluster, even though it does not bind to its flat facet. Roughness acts by localizing the HOMO of Au and in this way providing a higher electron density at the site where O2 binds.

Fig. 2.3 Step density of hemispherical Au as a function of particle size [102]

- 19 -

This facilitates electronic charge transfer to the * orbital of O2, which leads to bonding. Furthermore, Nrskov and other workers [102, 112, 127] have calculated that the adsorption strength of CO is very sensitive to the coordination site (kinks and steps) and the number of gold atoms. They suggested that the number of low coordinated atoms and corner sites are important factors for making Au cluster considerably more active than extended Au surface. Fig. 2.3 shows the Step density of hemispherical Au particles as a function of particle size. A maximum in the step density is found as particle size is about 3 nm which corresponds to the onset in the reactivity of small gold particles suggested by Haruta et al. Furthermore the size and shape of the particles are both also important parameters [79]. Fig. 2.4 shows a strong dependence of adsorption energies as a function of the Au coordination number. The electronic structure of Au atoms with low coordination numbers has higher lying d states, which interact strongly with the adsorbate states.

-1 3 4

Energy, eV 0 1

5 6 7 8 9 Coordination Number

10

Fig. 2.4 Binding energy of different oxygen species, versus gas-phase O2, plotted with respect to the coordination number of the Au atoms of the site. Yellow spheres (Au), red (O), and gray (C) [79] Very recent experimental results [145 - 147] for TiO2 supported Au and unsupported Au catalysts, such as nanoporous gold with a spongelike morphology, support the idea that the low coordinated atoms on the metallic gold surface play the most important role in the mechanism of CO oxidation.

- 20 -

2.3 Structure of particle (electronic) model

1. The Langmuir- Hinshelwood mechanism; both gases reactants, O2 and CO, are co-adsorbed onto the Au catalyst, where they subsequently react to form CO2.
O2 (g) CO (g) E2 E1 O2 (ad) CO (g) E3 O (ad) CO (g)

Energy

E4 O (ad) CO (ad)

E5 E6

CO2 (g)

CO2 (ad)

Reaction path
ka,O2

O2 (g)
kd,O2

O2 (ad)
kdiss,O2 krec,O2 ka,CO kd,CO kr

O2 (ad) CO (g)

2O (ad) CO (ad) CO2 (ad) CO2 (g)

CO (ad) + O (ad) CO2 (ad)

kd,CO2

Fig. 2.5 Schematic energy diagram for a heterogeneously catalysed CO oxidation (Langmuir-Hinshelwood reaction) [192] For the oxidation of CO on gold catalysts, carbon monoxide and oxygen come into contact with one another on the surface of the gold nanoparticle. While both gases have a negligible affinity for the bulk gold surface [43, 76], adsorption of oxygen and carbon monoxide on neutral [93] or negativelycharged [76] gold nanoparticles is probable. For example, Whetten et al. showed [76] that O2 bonds readily with Aun- (n is even and n 16), with the oxygen attaching as a superoxide (O2-).

- 21 -

In this model, the catalytic activity occurs on the surface of a cluster, and arises from a special electronic configuration of it. The role of the support is simply to modulate the electronic structure of the cluster. In heterogeneous catalysis, the two main reaction mechanisms which have been proposed are the Langmuir Hinshelwood mechanism (LH) and the EleyRideal mechanism (ER). For the vast majority of surface catalytic reactions, it has been accepted that the LH mechanism is preferred, for instance in catalytic CO oxidation on Pt (111) [105] and CO oxidation of Au particles [71, 109]. In Fig. 2.5, the dissociation of O2, the reaction between the adsorbed species CO(ad) and O(ad), and the desorption of the product, CO2, are the activated steps. No metal shows good performance in both the O2 dissociation and the reaction of chemisorbed atomic oxygen with CO. A noble metal will be slow in cleaving O2, but it will readily deliver O to CO, i.e., E2 >> E5. However one can expect unusual aspect for activation of O-O bond with decreasing particles size or other means of novel metal. 2. The Eley-Rideal (ER) mechanism; only one of the reacting species adsorbs, with the other coming into contact with it by kinetic collision from the gas phase.

Fig. 2.6 Calculated energy diagram for the Eley-Rideal reaction mechanism of CO oxidation on Ru (0001); the transition state geometry is indicated in the inset; large (Ru), medium (O), and small (C) circles [106]

- 22 -

The role of negatively charged Au clusters (Au-)

It was observed that the Au clusters deposited onto oxygen vacancy sites of metal oxide support (MgO) show high catalytic activity toward CO oxidation reaction [71, 93]. According to this model, the excess electron density at the Fcentre (oxygen vacancy) site is transferred to the cluster and forward to the adsorbed O2 to form the superoxide needed for the catalytic reaction. However, one cannot observe catalytic activity of gold deposited onto a stoichiometric TiO2 surface [43]. Such a fully oxidised surface will have very few F-centres and therefore, the gold clusters would not be negatively charged and hence not be able to react with O2 to form superoxide. From these results, one can say that negatively-charged gold clusters (Au-) play an important role for the CO oxidation reaction. Depending on the oxide supports, different negative charge states of Au nanoparticles can be observed [96]. Since more charge is transferred from reducible oxides to Au, the Au deposited on these oxides interacts more strongly with oxygen providing an ideal environment for O2 activation and the oxidation reaction. Even though catalytic activity can be changed depending on the supports, Rousset et al. suggested that the key point for the high catalytic activity is not the negatively charged Au but metallic Au [142].

The role of positively charged Au clusters (Aux+)

A number of mechanisms proposed for explaining the high activity of gold nanoparticles such as quantum size effects of the two-layer Au islands [43, 45, 84], negative charging of the Au clusters [70, 81], abundance of low Au-Au coordination sites [102], and the presence of Au-support perimeter sites [82, 100, 101] suggest that an anionic nature of Au particles was inferred.

- 23 -

On the other hand, Bond and Thompson [26] and others [98, 99], as mentioned before (perimeter model), have suggested that the Au particles are oxidized into Au+ or Au3+ sites at the particles support interface being responsible for the catalytic activity. Recent studies support that cationic Aux+ is responsible for the unique Au/oxide catalytic activity [95 - 97]. Au+-O2 bonds, for example at the Au-support perimeter sites, are suggested to be responsible for the catalytic activity of Au nanoparticles in oxidation reaction [100, 101]. Guzman et al. [94] suggested that the change in catalytic activity were the results of changes in the oxidation state of gold. Fig. 2.7 shows the relationship between the activity of the Au catalysts and the amount of cationic and zerovalent gold in it. Higher concentrations of cationic gold lead to higher catalytic activity. Hammer et al. also identify Au+O adhesion bonds of a mixed covalent and ionic nature under more real goldsupport interactions and reaction conditions.

Fig. 2.7 Correlation between the catalytic activity and the percentage / surface concentration of cationic and zerovalent gold [94]; Higher catalytic reaction rate was observed with increasing the concentration of Au+ species in Au catalysts. Furthermore, the importance of positively charged Au species in various reactions like the catalytic hydrogenation of ethane [134] water-gas shift reaction [119] was reported by Gates et al. and Fu et al. respectively.

- 24 -

Fig. 2.8 shows the reaction path way of the water-gas shift reaction of Ce(La)O2 supported Au.

Fig. 2.8 The rate of the water-gas shift reaction is not affected by leaching of metallic gold nanoparticles by a NaCN solution, indicating that the particles are just spectators in this reaction. The active site involves supported cationic gold (Au+, red spheres; the exact charge is not known) [135]

Quantum size effects

Depending on the structural arrangement of the Au catalyst on the support, catalytic activity can change [84]. In an experimental aspect, Goodman's group referred to quantum size effects in order to explain a maximum in CO oxidation activity and suggested that particle thickness, in this case two atomic layer Au particles, may be the key parameter [43, 45, 83, 109, 137].

Fig. 2.9 Activity for CO oxidation at room temperature as a function of Au coverage above the monolayer TiOx on Mo(112) [45]

- 25 -

According to recent results for CO oxidation by Au catalysts prepared on a TiOx film grown on a Mo(112) surface (Fig. 2.9), Au with a partially bilayered structure shows an increased activity more than an order of magnitude higher with respect to monolayered structure. In these systems well-ordered Au monolayers and bilayers that completely wet the oxide support were used, thus eliminating particle shape and direct support effects [45, 136]. Thicknessdependent catalytic activity of Au particles was also observed by C.T. Campbell et al. [109]. The origins of the exceptionally high catalytic activity for the Au bilayer systems are explained in terms of the electronic charge of Au, the binding strength of the reactants, and limited dimension effects. This was supported by DFT calculations of Rodriguez at al. [85]. For the Au catalyst showing the highest catalytic activity, CO molecule binds directly to the on-top site of Au atom. On the other hand, molecular oxygen adsorption is energetically unstable. The adsorbed CO captures an oxygen molecule to form a CO-O2 intermediate complex (peroxo-type in Fig. 2.10). The formation of peroxo-type complex is in agreement with the experimental and theoretical results [79, 86]. After that, O-O is dissociated giving raise to CO2 and a remaining oxygen atom bound onto the nearby Au atoms, which is highly reactive towards further incoming CO [87]. Such a result can be explained by a larger electron density that around the bottom Au layer which would favor a charge transfer into the antibonding 2 orbital of the peroxo-type fragment which obviously weakens the O-O bond and consequently leads to the oxygen activation [46, 85]

Fig. 2.10 Structure of the peroxo-type CO-O2 complex (left) and final state (right) for CO oxidation on the 4/3 layer Au/TiO-Mo(112) surface [85]

- 26 -

2.4 Others 2.4.1 Adsorption process of oxygen (dissociative or molecular)

Another issue is whether the adsorption of the O2 on gold is eventually dissociative or not. It is well known that molecular oxygen does not dissociatively adsorb on bulk gold at pressures as high as 1400 Torr and temperatures between 300 and 500 K [143]. However, the active oxygen species for catalytically relevant chemistry is atomically adsorbed oxygen. Atomic oxygen has been shown to be highly reactive toward CO oxidation reaction when pre-adsorbed on single crystal gold and Au model catalysts [107 - 111]. These studies show that atomically adsorbed oxygen as the key reactant in the CO oxidation reaction would require a dissociation of O2 on the catalyst surface prior to reaction conclusively. Recent experimental results by Cynthia M. Friend [143] and C.T. Campbell et al. [109] show that the high activity of small Au clusters and ultra thin Au particles (about two atomic layers thickness) on metal oxide supports for CO oxidation originates from enhanced O2 dissociation due to the large number of under-coordinated Au atoms [143]. Some density functional theory calculations [79, 82, 102] also have shown that O2 adsorbs preferentially at the Au-TiO2 interface and readily dissociates. However, another theoretical calculation shows that the stronger the bond between O2- and Aun is the greater the O-O distance in the superoxide [103], but this apparently does not occur to a sufficient extent to dissociate the bond. Additionally, it is reported that the barrier to dissociative chemisorption of oxygen on gold is very high and that molecularly chemisorbed oxygen is not stable on clean Au (111) [73, 90]. It may be the unique characteristic of gold nano-clusters to form superoxide adsorbates [79, 103, 131 - 133] rather than disassociated oxides [76]. This result gives doubt on whether atomic oxygen is a key player in the observed catalytic activity of supported gold clusters.

- 27 -

As mentioned before (perimeter model), some theoretical results suggested that a molecularly chemisorbed oxygen species directly reacts with CO to form CO2 via a peroxolike reaction intermediate, COO2 [100]. Other interesting result reported by Mavrakakis and co-workers [90] is that a strained gold surface results in conditions suitable for molecular chemisorption of oxygen. Steps and defects on the Au surface have been shown to be suitable sites for molecular chemisorption of oxygen. Nrskov et al. also show that molecular adsorption of oxygen on an Au10 cluster is possible and reveal that it is as reactive as atomically adsorbed oxygen for CO oxidation [79]. In investigations of gas phase anionic gold clusters with oxygen, Aun- (n = 2 - 20), oxygen has been observed to interact with gold cluster anions with an odd number of electrons, showing molecularly adsorption of oxygen on these clusters [75, 86, 113, 114]. It is known by some experimental results that O2 does not readily chemisorb on gold single crystal [59, 61, 107] or TiO2 supported Au clusters larger than 1 nm [109] either dissociatively or molecularly. Buddie Mullins et al. present experimental evidence in recent that molecularly chemisorbed oxygen species [129, 130] can participate directly in the CO oxidation of TiO2 supported Au clusters model catalysts in the 2 - 5 nm range [87, 129]. That means that the dissociation of oxygen does not play a key role in CO oxidation. In an alternative but nearly equivalent view, it is the low bond strengths between Au and the reacting species that makes gold such an active catalyst [25, 115].

2.4.2 Cluster size effects

The catalytic reaction studies reported by Haruta [139] and Goodman et al. [43] show a pronounced size effect of the catalytic activity of the TiO2 supported Au clusters for the CO oxidation reaction, with the maximum reactivity at a range of about 3.5 nm exhibiting (Fig 2.11).

- 28 -

0.

2.

4.

b
6. 8. 10.

ClusterDiameter Fig. 2.11 CO oxidation turnover frequencies (TOF) (a) at 300 K [139] and 350 K [43] as a function of the average particle size of the Au cluster on TiO2 support Scott L. Anderson et al. [140] are also showing that the activity is strongly dependent on the deposited cluster size, because electronic and geometric structures are both size-dependent. Au7 cluster on TiO2 in Fig. 2.12 shows the highest catalytic activity toward CO oxidation reaction. According to the results of Heiz and co-workers [71], significant reactivity for clusters as small as Au8 was observed for CO oxidation reaction of Au/MgO system.
0.08 0.10 0.12 0.14 0.16 0.00 0.02 0.04 0.06

CO2/Langmuir CO/Au atom

TiO2 Au1

Au2

Au3 Au4 Au5

Au6 Au7

Fig. 2.12 Size dependence of CO oxidation activity [140] It was claimed by Metiu et al. [73] that the finite size of small clusters leads to better binding because the HOMO is more localized (confinement effect; the finite size of the cluster prevents the delocalization of the HOMO). This effect facilitates charge transfer into the * orbital of O2, which induces the molecule to bind to gold.

- 29 -

2.4.3 Moisture effects

Some experimental results [121 - 125] have shown that the presence of moisture on Au catalyst, unlike for other common catalysts, increases its catalytic activity up two orders of magnitude. Theoretical work done by U. Landman et al. revels that coadsorption of H2O and O2 (Fig. 2.13 b) leads to formation of a complex well bound to the gold cluster. And then CO induced proton transfer resulting in formation of a hydroperoxyl-like group (left) and a hydroxyl (right) in Fig. 2.13 c. Consequently the O-O bond is activated, leading to a weakened peroxo or superoxolike state, and the reaction with CO to form CO2 occurs with a small activation barrier.

Fig. 2.13 Atomic configurations displaying several stages in the simulation of the coadsorption of H2O and O2 on the top facet of a Au8 cluster supported on MgO(100), and the subsequent reaction with gaseous CO to form CO2. Yellow (Au), red (O), white (H), green (Mg), and aquamarine (C) [120]. Recent results reported in literature for active sites (mechanisms) of supported Au catalyst cannot be interpreted with only one mechanism. Depending on the particles size, maybe, different mechanisms can be required for explaining the high catalytic activity of supported Au nanoparticles towards CO oxidation, i.e. two or even more different mechanisms are responsible under model and real catalysis system.

- 30 -

3. Experimental Setup
Nowadays, several routes are used to prepare nanoparticles with narrow size distributions. Most widely used methods for synthesizing nanoparticles are based on chemical reactions in solution, yielding monolayer-protected nanoparticles which self assemble on suited substrates [157 - 159]. However, one cannot completely exclude the possibility of a contamination of these particles, which originates from chemical solution. A method already used in the earliest studies to reduce the contamination to a minimum was thermal evaporation of metals. Further developments of new cluster sources allow direct deposition of pre-formed, size-selected particles [160, 161]. In this experiment, two different Ultrahigh Vacuum (UHV 1 and UHV 2) systems were used for the creation and analysis of the size-selected Aun (n = 2 13) clusters and Au and Ag nanoparticles, respectively. All experiments were performed under Ultrahigh Vacuum (UHV) conditions with base pressures of 2 x 1010 mbar.

3.1 UHV 1 (for mass-selected Au clusters)

The UHV 1 system can be separated into four parts (creation and deposition of the clusters, spectroscopy analysis, preparation chamber for making nanoparticles and oxidation experiments, and portable UHV system for sample transfer between UHV 1 and UHV 2) it is shown in Fig. 3.1. In the main chamber (analysis part), the pressure is maintained at about 2.0 x 10-10 mbar by an ion getter pump, a turbo molecular pump, and a titan sublimation pump.

- 31 -

Part 1; For the creation of the mass-selected Au clusters, a magnetron-sputtersource, originally designed by Prof. Haberlands group in the University of Freiburg, Germany was used [331]. For the mass-separation of the clusters, a magnetic field mass spectrometer was used. All the clusters were soft-landed on the surface, i.e. a kinetic energy of 0.2 eV per atom was used.
Photoelectron spectroscopy
Cylindrical hemispherical anaylser (CHA) Mass separation magnet

Sputter gun Cooling Trap

Ion extraction Skimmer, diff. pumpingstages

X-ray source (Al Ka 1486eV) UHV transfer system

Ion pump

Soft landing

HOPG SiO2/Si CO / O2 Experiment & Nanoparticle evaporation

substrate

Magnetron cluster source

Ag & Au rod W wire

RT_STM

Fig. 3.1 Schematic drawing of the UHV 1 system (upper) and real image of the cluster deposition and XPS analysis chambers (lower)

- 32 -

Part 2; Analysis part is equipped with a Cylindrical Hemispherical Analyzer (CHA: OMICRON, pass energy = 20 eV used in this experiments), an X-ray source with an Al target (K, photon energy = 1486.6 eV), and a He UV lamp for XPS (X-ray Photoelectron Spectroscopy) and UPS (Ultraviolet Photoelectron Spectroscopy) analysis respectively.

Part 3; Ag and Au nanoparticles were also prepared in UHV 1 system part with (a base pressure of 2 x 10-9 mbar) using thermal evaporation method. Ag and Au rods (purity 99.999 % from Alfa Aesar) are wrapped by a W (tungsten) wire (usually 10 cm in length, 0.5 mm in diameter), which was resistively heated. To produce atomic oxygen environments for oxidation experiments, a hot Pt (platinum) wire (usually ~ 10 cm in length, 0.3 mm in diameter) located at the backside of the samples was heated in oxygen atmosphere [53]. Oxygen and CO molecular gas are connected to preparation chamber via leak valves for CO oxidation experiment.

Part 4; A small portable UHV chamber is employed to transfer the sample between UHV 1 and UHV 2. The pressure of this system was kept at about 2.0 x 10-9 mbar by an ion getter pump.

- 33 -

The procedure of the deposition of mass-selected clusters on the surface

The fundamental structure of an experiment for cluster deposition using magnetron sputter source was already described in detail by von GynzRekowski [189] and is therefore only briefly described. Cluster ions are generated using a magnetron sputter source and mass-separated by a magnetic field mass spectrometer. Maximum currents of mass-selected cluster ions of below 1.0 nA (usually 0.25 nA) of Aun (n = 2 13) clusters during deposition onto the surface are achieved. The mass-selected cluster ions are soft-landed on a substrate and characterized by XPS (x-ray photoelectron spectroscopy) in UHV 1 system. A sketch of the experimental setup for cluster deposition is shown in Fig. 3.1. It can be separated into four parts (1.1 Magnetron sputter cluster source, 1.2 ion extraction, 1.3 mass separation, 1.4 soft landing) prior to spectroscopy analysis with XPS, which will be discussed in the following.

1.1 Magnetron sputter cluster source

The clusters were produced in a magnetron sputter source in the gas phase and extracted over differential pumping stages from the feed gases. After three pumping stages the vacuum condition is already under 10-6 mbar. Within this range clusters are accelerated by the guiding voltage (0.8 ~ 1.0 kV). The structure of magnetron sputter source is shown in Fig. 3.2. The magnetron sputter source consists of the sputtering head, with target materials (Au or Ag) in front and water-cooling system and high voltage (~ - 300 V) attached to the reverse side, and a metal cylinder cap, which is connected to the gas in-let system and grounded. Both parts are isolated by a quartz glass cylinder.

- 34 -

The distance between target and metal cylinder is typically 2 4 mm. Ar and He gases were used at the ratio of 1 to 10 for the sputtering and carrier process. The entire structure is installed on a mobile bar system and is mounted in a cylindrical aggregation chamber cooled by liquid nitrogen. At the end of the aggregation chamber the mixture of cluster material and feed gases is focused by a conical nozzle with a diameter of 5 mm (skimmer 0) and extracted in the next vacuum system.

Sputter & Carrier_gas Ar & He Anode

Isolation flange

Cooling water

Cathode ~ -300V Quartz glass_cylinder

Magnet Sputter-target

Skimmer 0 1 mbar

Lens 0 Skimmer 1 10-2 mbar

Skimmer 2 10-4 mbar

Skimmer 3 10-5 mbar 10-6 mbar

5 mm

6 mm

11 mm

16 mm

Magnetron Sputter source

8 mm Root pump 300 l/s

21 mm

52 mm

76 mm

Turbo molecular pump Oil diffusion pump Turbo molecular pump 2 x 500 l/s 2200 l/s 12000 l/s

Fig. 3.2 Structure of the magnetron sputter source (upper) and the arrangement of skimmers in different pumping stages for the extraction of cluster ions from the source into the next vacuum system (lower) [189]

- 35 -

1.2 Ion extraction

The source and the sample are set to ground potential. The cluster ions are accelerated to a kinetic energy of 800 1000 eV, and the ion beam is guided through a metal shield kept at the corresponding potential (+ 800 1000 V for negatively charged Au clusters). A focusing of the created cluster ion beam is adjusted by ion optics (Einzel lens). A first Einzel lens forms a well-defined cluster-ion beam, which is subsequently focussed by a second Einzel lens onto the entrance aperture of a sector magnet.

1.3 Mass separation

Mass separation is achieved by deflection in a magnetic field. The maximum field strength of the deflection magnet is ~ 1 T. For the experiments described here, some settings were adjusted. For example, the entrance aperture of the sector magnet was set to a diameter of 8 mm and the distance between magnet and sample was extended by about 1 m and a further Einzel lens was installed behind the sector magnet for better resolution. Based on this, single charged clusters with a mass up to ~ 2800 amu can be deflected and the mass resolution of clusters is m/m = 22. An example of a mass spectrum of Aun clusters with n = 2 13 is displayed in Fig. 7.1.

1.4 Soft-landing ion optics

After passing the magnet the ion beam is focussed by a third Einzel lens into the UHV chamber. The cluster ion beam, which was guided by HV (high voltage) potential, is decelerated to half of HV and focussed by fifth Einzel lens. In a small gap between the guiding tube of the ion beam and substrate (approximately 10 mm), the cluster ions are decelerated in a homogenous electrostatic field to the applied deposition voltage.

- 36 -

The potential of the sample can be varied by 100 V with respect to ground potential, i.e. a kinetic energy of 0.2 eV per atom was used typically. The spot size of soft-landed clusters onto the substrate is typically below 3 mm in a diameter.

Fig. 3.3 Schematic drawing of the ion optics for soft-landing of the size-selected cluster ions onto the surface [189]

The collision of clusters with surfaces has been studied by many groups [162, 163]. Their studies were focused on cluster implantation [164], fragmentation [165, 166], reflection [162, 163], or soft landing [167]. In the soft-landing regime, the incident kinetic energy is generally insufficient to overcome the adhesion between the surface and the cluster, resulting in collisions that always lead to adhesion. At high kinetic energies, one enters the fragmentation regime where the cluster fragments upon impact into atoms or several large components [165] or the implantation regime where the cluster buries itself in the surface [168]. Only recently an intermediate regime has been identified where clusters were observed to undergo a transition from adhesion to reflection (Fig. 3.4) [169].

- 37 -

Fig. 3.4 Diagram of the interaction between energetic clusters of atoms and metal substrates [163]; ( E kin and
cl

/ N cl is the kinetic energy/atom in the cluster

sub Ecoh is the cohesive energy of the substrate)

Clusters with impact energies of 1.0 eV per atom could be landed nondestructively on the bare substrate, whereas at higher kinetic energies fragmentation and substrate damage were observed. It is worth to mention that, on the basis of previous reports and our experimental results, size-selected clusters created in this experiment are soft-landed on the surface without fragmentation or penetration.

- 38 -

3.2 UHV 2 (for Au and Ag nanoparticles)

The UHV 2 system is equipped with a room temperature STM (Scanning Tunnelling Microscope: OMICRON STM) with Pt / Ir (90 % / 10 %), a Cylindrical Mirror Analyzer (CMA) for Auger Electron Spectroscopy (AES), a He UV Lamp, a Low Energy Electron Diffraction system (LEED), two electron beam evaporators (Tectra) and an Ar ion sputter gun. In the STM chamber the pressure is maintained at about 2 4 x 10-10 mbar by an ion getter pump and a titan sublimation pump. STM chamber and preparation chamber are directly connected to each other (Fig. 3.5).

Fig. 3.5 Three dimensional drawing of the UHV 2 system (upper) and real image of the STM stage (lower) [190]

- 39 -

3.3 X-ray Photoelectron Spectroscopy (XPS)

XPS was developed in the mid 1960s by K. Siegbahn and his research group. K. Siegbahn was awarded the Nobel Prize for Physics in 1981 for his work in XPS. The phenomenon is based on the photoelectric effect outlined by Einstein in 1905 where the concept of the photon was used to describe the ejection of electrons from a surface when photons impinge upon it. X-ray photoemission spectroscopy (XPS), also known as Electron Spectroscopy for Chemical Analysis (ESCA), is among the most frequently used techniques to determine atomic compositions (The peak areas can be used (with appropriate sensitivity factors) to determine the composition) and get information about the types of bonding that occur within various compounds. XPS is not sensitive to hydrogen or helium, but can detect all other elements. XPS must be carried out under UHV conditions.

Fig. 3.6 "Universal curve" of mean free path () versus kinetic energy (KE) (eV) of electrons in different materials [182, 183] ((IMFP is average distance between inelastic collisions ())

- 40 -

A technique becomes surface sensitive if the radiation or particles to be detected can travel no more than a few atomic distances through the solid. Fig. 3.6 shows that the mean free path, , of electrons in elemental solids depends on the kinetic energy in the range 15 - 1000 eV [179]. Optimum surface sensitivity ( = 0.5 nm) is achieved for electrons with kinetic energies in the range 50 250 eV, where almost half of the photoelectrons come from the outermost layer [184].

3.3.1 XPS emission process (Basic principles)

Core electrons have binding energies corresponding to energies of photons that lie in the X-ray region. When a solid absorbs a photon with an energy in excess of the binding energy of an electron, a photoelectron is emitted and the kinetic energy of the photoelectron (Ek) is related to the energy of the photon (h). Core electrons do not participate in bonding, and their energies are characteristic of the atom from which they originate. Ek = h - EB where EB is the electron binding energy h is the Planck constant is the frequency (Hz) of the radiation. Whereas ultraviolet photoemission spectra (UPS) are commonly referenced to the Fermi energy (EF), XPS are conventionally referenced to the vacuum level. The additional amount of energy that an electron needs to overcome the energy difference between the Fermi energy (EF) and the vacuum energy (Ev) is known as work function of the sample (sample). Considering this additional energy, the kinetic energy (Ek) that an electron has after leaving the sample, is given by EK = h - EB - sample (2) (1)

- 41 -

Sample

e-

Spectrometer

h Vacuum level es

Ek Vacuum level espec -es espec EF Fermi level Eb Ebv

sample

spectrometer

Fig. 3.7 Schematic energy diagram of X-ray photoelectron spectra. The measured kinetic energy of the ejected electron is given by E*k [190] Since the spectrometer posses its own work function (spectrometer), Ek is not the kinetic energy, which is measured by the spectrometer, the spectrometer work function must be accounted for when interpreting the measured electron kinetic energies (EK) (Fig. 3.7). The kinetic energy (E*k) measured by the spectrometer is given by E*k = h - EB - spectrometer (3) The work function of spectrometer can be measured with a known bulk sample.

3.3.2 Koopmans Theorem

Koopmans theorem is based on the effective one-electron in the frozen orbital approximation in which the binding is equal to the difference in energies between the atom with n electrons (initial state) and the ion with n-1 electrons (final state) (Fig. 3.8) BE = Efinal(n -1) - Einitial (n)

Fig. 3.8 Emission process based on the Koopmans theorem [191]

- 42 -

If there were no rearrangement of all spectator electrons, the binding energy would be equal the negative of the orbital energy of the initial state of the electron, - k, (BE = - k). However, since the electrons are not frozen, the measured binding eneries and calculated orbital energies are different by 10 30 eV. The final state achieved by removing one electron corresponds to an ionic state in which a hole exists in place of the ejected photoelectron. The remaining electrons in the atom can relax and also the electrons on neighbouring atoms can relax in response to the ionization event. In reality, initial state effects and final state effects affect both the measured binding energy. This means, as a consequence, that the binding energy of a photoelectron contains information about the initial and the final state. The charge potential model elegantly explains the physics behind such binding energy shifts, by means of the formula [184];
i Eb = kqi + j

qj rij

+ Ebref

in which
i Eb is the binding energy of an electron from an atom i

qi is the charge on the atom k is a constant qj is the charge on a neighbouring atom j rij is the distance between atom i and j

Ebref is the suitable energy reference

The first term in this formula indicates that the binding energy goes up with increasing positive charge on the atom from which the photoelectron originates. In ionic solids, the second term counteracts the first, because the charge on a neighboring atom will have the opposite sign. Because of its similarity to the lattice potential in ionic solids, the second term is often referred to as the Madelung sum.

- 43 -

Photoemission process often envisaged as three steps: (i) Absorption and ionization (initial state effects) (ii) Response of atom and creation of photoelectron (final state effects) (iii) Transport of electron to surface and escape (extrinsic losses) All can contribute structure to XPS spectra.

Primary structure in XPS

3.3.3. Inelastic Background

XPS spectra show characteristic "stepped" background. The intensity of background to high binding energy side of a photoemission peak is always greater than on the low energy side due to inelastic processes (extrinsic losses) from deep in bulk (Fig. 3.9). Only electrons close to surface can, on average, escape without energy loss, i.e. electrons deeper in surface loose energy and emerge with reduced kinetic energy (increased binding energy).

Fig. 3.9 Survey XPS spectrum with steplike background [191]

- 44 -

3.3.4 Auger Peaks

The Auger lines like LMM, LMV, and LVV are clearly visible in the XPS spectra, are produced by from excess energy of atom during relaxation (after core hole creation). Since the Auger lines are element specific, they can be used for element identification. The energy of an Auger line is independent of the incident photon energy, while the energy of the photoelectron line varies linearly with the incident photon energy.

3.3.5 Spin-Orbit Splitting (SOS)

Photoemission from p, d, and f electronic states with nonzero orbital angular momentum produces a spin-orbit doublet such as the 4f7/2 and 4f5/2 (Fig. 3.10). The two lines correspond to final states with (j+ = l + s) and (j- = l s). The intensity ratio of the lines is given by the ratio (2j- + 1) / (2j+ + 1), which gives a ratio of line intensities of 3 : 4 for f5/2 to f7/2.

Fig. 3.10 Spin-orbit coupling leads to the splitting of the Au4f photoemission peak into two peaks (l; orbital angular momentum, s; spin angular momentum) [191]

- 45 -

3.3.6 Core Level Chemical Shifts

The exact binding energy for an electron in a given element depends on the chemical environment of that element. Any change in the chemical environment of the element will involve a spatial redistribution of the valence electrons of this atom and the creation of a different potential seen by the core electrons. This results in a change their binding energies (Fig. 3.11 a).

Si 3+ Si 2+ Si 1+

a)

SiO2

Si

Si(100)

b)

Fig. 3.11 a) the chemical shift in binding energy of the Si 2p line for elemental Si and SiOx [185] and b) carbon 1s chemical shifts in ethyl trifluoroacetate. The four carbon lines correspond to the four carbon atoms within the molecule [186] The concept of chemical shifts is based on the idea that the inner electrons feel an alteration in energy due to a charge in the valence shell contribution to the potential based on the outer electron chemical binding. In the simplest picture, valence electrons are drawn either from or toward the nucleus depending on the type of bond. The greater the electonegativity of the surrounding atoms, the more the displacement of electronic charge from the atom and the higher the observed binding energies of the core electrons (Fig. 3.11 b).

- 46 -

Secondary Structure in XPS 3.3.7 X-ray Satellites

In order to observe sharp photoemission lines in XPS, x-ray source should be monochromatic. X-ray monochromators usually make use of quartz crystal diffraction to energy select the beam. Non-monochromatic x-ray sources produces "ghost" peaks in XPS spectrum at lower binding energy beside the two K lines (K2 corresponding to 2p1/2 1s and K1 corresponding to 2p3/2 1s).

3.3.8 Surface Charging

Electrical insulators cannot dissipate charge generated by loss of the electrons in solids by photoionization process. Surface picks up excess positive charge, which gives rise to all peaks shift to higher binding energy (1 ~ 5 eV).

3.3.9 Final State Effects

Final state effects arise during atom relaxation and core-hole creation after photoelectron creation, which affects the energy distribution of the emitted electrons.

1. Relaxation effect The orbitals of the (n-1) remaining electrons adjust to the new filed by a screening effect caused from effective charge, i.e. they may relax into states in lower energies. Energy emitted in this relaxation process is called relaxation energy. This energy has to be carried away by the ejected photoelectron, so that the kinetic energy of the photoelectron increases, i.e. binding energy decreases.

- 47 -

1-1 Intra atomic relaxation; in the case of a free atom, removal of an electron from an atomic orbital creates a positive hole and the remaining electron orbitals relax to minimize the system total energy. 1-2 Extra atomic relaxation; in the case of a large molecular or of a solid the local hole created by the photoemission process acts as a local positive charge, and attracts electrons that tend to screen it. Because additional electronic charge can flow towards the positive hole, the relaxation energy is always larger than in an atom.

2. Multiplet splitting Whereas in atoms and molecules with closed shell configurations the atomic 1s level are spin degenerate (no energy splitting in the spectra), in open shell configurations the core level after photoemission may be split by exchange interaction with the unfilled shell, i.e. removal of a core electron give rise to peaks at different binding energies in the photoelectron spectrum by L-S coupling. The XPS peak intensities of the lines are proportional to the multiplicity.

FeF2
Counts / sec

6.0 eV 4.2 eV

K4FeF(CN)6

Binding Energy (eV)

Fig. 3.12 Multiplet splitting of the Fe 3s lines [191]

- 48 -

3. Multi electron excitation When a core electron is removed by x-ray photon there is a sudden change in effective charge due to the loss of a shielding electron. This sudden change in effective charge gives rise to the possibility for the formation of monopole excited states or doubly ionized states. 3-1 Shake up; if an electron in an outer orbital is transferred to an excited orbital in which only the principal quantum number has been changed, this transition will appear as a discrete satellite photoelectron line with a lower kinetic energy than the main line corresponding to the difference between the ground and excited states of the ion with a core vacancy. 3-2 Shake off; if an electron in an outer orbital is excited to the continuum state, leaving a doubly ionized atom with holes in the core as well as valence shell, a continuous spectrum will appear.

4. Energy loss (plasmon) Photoelectrons may give up some energy to the electron gas situated in a valence band of a conductor before leaving the material, i.e. interaction between photoelectron and other electrons in the surface region gives rise to the collective oscillations (plasmons) with a characteristic frequency. Due to this oscillation, peaks with higher binding energies than the original states can be observed in the XPS spectrum (Fig. 3.13).

Bulk plasmon Surface plasmon

Fig. 3.13 Energy loss lines associated with the 2s line of aluminium [187]

- 49 -

3.3.10 Experimental Aspects of XPS

An XPS spectrometer contains an X-ray source, usually Al K (1486.6 eV) or Mg K (1253.6 eV) and an analyzer. Other X-ray lines can also be sometimes chosen such as Ti K (2040 eV). In this experiments, we used the Al K (1486.6 eV) source. The width of the line for the exciting X-ray radiation K is about 0.85 eV. This value is good compared to the resolution of other Xray anodes [187].

Fig. 3.14 Schematic drawing of a Concentric Hemispherical Analyzer (CHA) in XPS [190] The energy of the photoelectrons leaving the sample can be determined using a CHA (Concentric Hemispherical analyzer; Omicron EA125 U5 used here) (Fig. 3.14). It consists of two concentric hemispheres of mean radius R, mounted with a common centre, to which two different potentials are applied. The outer surface is more negative than the inner, so that a mean equi-potential surface at R0 between the hemispheres is formed. There are two slits located at the entrance and exit of the hemispheres and both are centred on R. A lens system is located before the slit at the entrance. These lenses are used to focus the emitting surface area. In the entrance tube, the electrons are retarded or accelerated to a value called the pass energy, at which they travel through the hemispherical filter.

- 50 -

The lower the pass energies, the smaller the number of electrons reaching the detectors gets, but their energy is determined more precisely (Fig. 3.15). Behind the energy filter is the actual detector, which consists of an electron multiplier or a channeltron, which amplifies the incoming photoelectrons to measurable currents. Our hemispherical analyzer contains up to five channeltrons. The resolution of XPS is determined by the line width of the X-ray source, the broadening due to the analyzer, and the natural line width of the level under study. The line width of the X-ray source is in the order of 1 eV for Al or Mg K sources but can be reduced to about 0.3 eV by using of a monochromator.
HOPG at RT Step 0.05 EP = pass energy

C1s
Ep 60 eV Ep 50 eV Ep 40 eV Ep 30 eV Ep 20 eV Ep 10 eV

Full Width Half Maximum (FWHM) (eV)

1,8

Intensity (arb. units)

1,6

1,4

1,2

1,0

0,8
292 290

Binding Energy (eV)

288

286

284

282

280

10

20

Pass Energy (eV)

30

40

50

60

Fig. 3.15 XPS C1s peaks as a function of the pass energy The broadening due to the analyzer depends on the energy at which the electrons travel through the analyzer and the width of the entrance and exits slits. The analyzer contribution to the line width becomes irrelevant at low pass energies, however, at the cost of intensity. The natural line width is determined by Heisenbergs uncertainty relation E t h / 2 where E is line width. t is the life time of the core-ionized atom. h is Plancks constant

- 51 -

The life time of the core-ionized atom is measured from the moment it emits a photoelectron until it decays by Auger processes or X-ray fluorescence. As the number of decay possibilities is greater for an ion with a core hole in a deep level (e.g. the 3s level) than for an ion with a core hole in a shallow level (e.g. the 3d level), a 3s peak is broader than a 3d peak.

3.3.11 Quantitative analysis (XPS Intensity & Sample Composition)

Almost all photoelectrons used in XPS have kinetic energies in the range of 0.2 to 1.5 keV. According to the inelastic mean free path data in Fig. 3.6, the probing depth of XPS varies between 1.5 nm and 6 nm, depending on the kinetic energy of the photoelectron. When determining concentrations, this effect must be accounted for. Actually, concentrations cannot be calculated without assuming a structure model. For instance, a metal foil with a thin oxide passivation layer on top will give an intense peak of oxygen in the XPS spectrum, whereas the nominal oxygen concentration for the entire foil is negligeable. In the case of a homogeneous concentration through the sample, the following expression holds for the intensity of each element: I = Fx S(Ek) (Ek) n (Ek) cos - I is the intensity of the XPS peak (area) - Fx is the X-ray flux on the sample. - S(Ek) is the spectrometer efficiency for detecting the electron at kinetic energy. Ek (called transmission function) - (Ek) is the cross section for photoemission. - n(z) is the concentration, in number of atoms per unit volume. - (Ek,z) is the mean free path fo the photoelectron at kinetic energy Ek through the material present at depth z. - is the take-off angle, i.e. angle between the direction in which the photoelectron is emitted and the surface normal.

- 52 -

In the case of supported nanoparticles, XPS recognizes how well particles are dispersed over a support. Fig. 3.16 shows two catalysts with the same amount of material in supported particles but with different dispersions. When the particles are small, almost all atoms are at the surface, while the support is covered to a large extent. In this case, XPS measures a high intensity Ip from the particles, but a relatively low intensity Is for the support. Consequently, the ratio Ip / Is is high. For poorly dispersed particles, on the other hand, the ratio Ip / Is is low. Thus XPS intensity ratio Ip / Is reflects the dispersion of a catalyst. One can go a step further and use the Ip / Is ratio for a quantitative estimate of the dispersion.
Low dispersion High dispersion

Particles support
XPS : Ip/Is small Ip/Is large

Fig. 3.16 XPS intensity ration of signals from particles and supports, Ip / Is reflects the dispersion of the particles over the support [184]

- 53 -

3.4 Scanning Tunnelling Microscopy (STM)

The Scanning Tunnelling Microscope (STM) was invented in 1983 by G. Binnig and H. Rohrer [171, 172] and within a few years it emerged to one of the most widely used tools in surface science. Thereafter it was possible to image regular surfaces and surface bound species with atomic scale resolution i.e. one can get information about the topography and the electronic structure of a surface at the atomic scale [173 - 178]. Fig. 3.17 shows the main experimental principle of STM.

Fig. 3.17 Main principle of scanning tunneling microscopy [181] The STM makes use of the quantum mechanical tunneling effect: A sharp metallic tip is positioned a few Angstrom (: usually one or two atomic layers) over a conductive surface and a voltage in the order of ~ 4 V is applied. The potential barrier between tip and surface is larger than the electrons energy, thus forbidding a current flow in the classical picture. Nevertheless, the electronic wave functions of tip and surface, decaying into the junction gap, overlap each other, leading to a finite probability for tunnelling of the electrons. The tunnel current is usually in the order of pico to nano ampere and depends exponentially on the distance between tip and sample, changing about one order of magnitude with 1 change in the tip-sample distance.

- 54 -

A measurement of this current at constant height (Constant height mode), or, of the voltage at constant current by regulating the tip height (z) through a feedback loop (Constant current mode which is typical working mode) can be done while the tip is scanned line wise over a region of the surface. A map of the surface is obtained by assigning to each lateral x / y - position of the scanned region the tip height z (x, y). In this mode the tip follows roughly the corrugation of the surface, creating a nearly topological map.

Fig. 3.18 STM image (topography) of the graphite surface (upper) [180] and scanning tunneling spectroscopy (STS) result of a Bi2Sr2CaCu2O8 sample (lower) [181] Fig 3.18 shows the topography of graphite surface measured by STM with atomic resolution. However, STM image is also affected by electronic effects. More precisely, the tip moves on a surface of constant local density of states (LDOS) close to the Fermi energy. In general it is not simple to distinguish between topologic and electronic effects. In order to achieve high resolution, high stability and precision of the tip movement are required. For the tip positioning, piezoelectric crystals (piezos) are employed. These materials deform in dependence on the applied voltage. For the used piezos at temperatures of about 7 K, the elongation is in the order of 10 / V and the maximum voltage that can be transformed into an elongation is about 100 V.

- 55 -

Therefore the maximal scan area is in the order of 1 m2. The spatial resolution, however, is not limited by the piezos, but by electronic, and mechanic noise as well as STM tip quality. In our STM experiment, in order to reduce the mechanical noise, the STM stage is suspended by four soft springs, which are protected by surrounding columns. The vibrations of the STM stage are intercepted using a non-periodic current damping mechanism. This is achieved by surrounding the STM stage by a ring of copper plates, which come down between magnets. The induced current damps the vibrations (Fig. 3.5 lower) [190]. In addition, one can also measure the local electronic density of the states (LDOS) with STS (scanning tunneling spectroscopy) (Fig. 3. 18 lower). By varying the potential difference between tip and sample, one can measure I (V) spectra as well as the differential conductance dI / dV (V).

- 56 -

3.5 Procedure 3.5.1 For mass-selected Aun (n = 2 - 13) clusters on HOPG and SiO2/Si
In UHV 1 system, mass-selected Au clusters were first deposited on sputtered HOPG and SiO2/Si surfaces with native oxide layer, and then analyzed using XPS (X-ray Photoelectron Spectroscopy). For deposition of Au clusters on a substrate, about ~ 30 minutes were required. Fig. 3.19 shows a scheme of the experimental procedure. Deposition of the size-selected Au clusters on HOPG and SiO2/Si surface (UHV 1)

XPS analysis (UHV 1)

Oxidation experiment with atomic oxygen (UHV 1)

STM analysis (UHV 2)

Fig. 3.19 Scheme of the experimental procedure related to mass-selected Au clusters in UHV 1 and UHV 2 The HOPG surfaces prepared by the scotch-tape pilling method were mounted on a metal sample-holder and inserted into the UHV 1 system (part 2), then outgassed at about 700 K for longer than 12 hours. Subsequently, the HOPG surfaces were sputtered by Ar ions with a kinetic energy of 0.5 kV for about 20 seconds in order to create defect sites, stabilizing Au clusters deposited on the surface. Altering the sputtering time and Ar pressure could control the defect density. The sample current during sputtering was typically ~ 4.0 A, while Ar pressure of 4.0 x 10-6 mbar was used. The purity and defect density was estimated by means of XPS: with increasing sputtering time, the main XPS C1s peak centered at 284.6 eV became broader.

- 57 -

And the peak intensity corresponding to the HOPG surface plasmon resonance (in C1s state at 290.6 eV) decreased. Before loading the Si wafer into the UHV chamber, Si wafers were chemically etched using a mixture of three different acids and water (H3PO4, HNO3, CH3COOH and H2O with a volume ratio of 3:3:23:1). The Si wafers with native silicon oxide layer were immersed into these chemical solutions for 3 min. Si wafers were degreased in ethanol solution for 5 min with the sonicator, and then rinsed in distilled water. The chemical etching was done in order to increase the defect density of the silica surface, reducing the diffusion of the clusters on the surface. After loading the chemically etched Si water into sample holder in UHV 1, size-selected Au clusters created using magnetron sputter source was soft-landed on sputtered HOPG and chemically etched SiO2/Si surfaces. Mass spectra of Au clusters created by magnetron sputtering source was collected during deposing. After XPS analysis of the deposited Au cluster, these samples were moved into the preparation chamber for the oxidation experiment. The analysis chamber is connected via a gate valve to the preparation chamber directly. In the preparation chamber, the backside of the samples was exposed to a hot Ptfilament (~ 1000 K), and at the same time, the chamber was filled with molecular oxygen. Even when the Pt-filament was located on the front side of a sample or a higher current (4.5 A instead of 4.0 A which was decided depending on the length of Pt wire) to resistively heat the Pt-filament was used in the oxygen atmosphere, no Pt-peak could be detected in XPS spectra. Each sample was exposed to 8 x 10-5 mbar of O2 for 30 minutes. This procedure may yield either atomic or thermally excited molecular oxygen. The CO exposure experiments were also conducted in the preparation chamber. After the CO gas exposure (usually 3.0 x 10-5 mbar for 100 sec), the samples were immediately transferred to the XPS analysis chamber, maintaining the pressure below 1 x 109 mbar while the sample is transferred. In addition, deposited clusters on sputtered and/or non-sputtered HOPG surfaces were transferred into UHV 2 chamber for STM analysis using a portable UHV system with a base pressure of ~ 2 x 10-9 mbar.

- 58 -

3.5.2 For Ag and Au nanoparticles on HOPG and SiO2/Si (Type 1)

Ag and Au nanoparticles can be prepared using both UHV 1 and UHV 2 systems. After XPS and STM analysis of prepared nanoparticles in UHV 1 and UHV 2 repectively, each sample was transferred using a portable UHV transfer system with a base pressure below 10-9 mbar between UHV 1 and UHV 2. Before loading the substrates into the UHV chamber, HOPG and SiO2/Si substrates were prepared by the scotch-tape pilling method and chemically etched by chemical solution, respectively. The HOPG sample can be outgassed at about 1100 K.

Evaporation of Au and Ag nanoparticles on HOPG and SiO2/Si surface using two electron beam evaporators (UHV 2)

STM analysis (UHV 2)

XPS analysis (UHV 1)

Oxidation experiment with atomic oxygen (UHV 1)

Fig. 3.20 Scheme of the experimental procedure related to Ag and Au nanoparticles in UHV 1 and UHV 2; Type 1

- 59 -

First, in UHV 2 system, the loaded HOPG substrate was sputtered using Ar ion sputter gun installed in main chamber of UHV 2. In this system, 0.5 kV was used to accelerate Ar ions to the surface and defect density was controlled by altering the sputtering time between 2 and 5 sec. The sample current during sputtering and Ar pressure are typically ~ 1.0 A, and 2.0 x 10-5 mbar, respectively. After Ar ion sputtering, the sample was heated again at about ~ 400 K for 5 min to remove the remaining Ar ions on HOPG surface. Temperature was measured with an optical pyrometer (Impact). Ag and Au nanoparticles were grown by evaporating Ag and Au rods (ChemPur, purity: 99.999%) using electron bombardment heating (evaporator made by TECTRA). The flux of Ag could be kept constant by controlling the emission current (typically 10 mA) between the W filament and the Ag and Au targets. All STM images were taken using constant current mode. After STM measurements, the samples were further characterized in UHV 1: XPS analysis and oxidation experiments (Identical experimental conditions were applied to this like mentioned above for size-selected clusters). Fig. 3.20 shows a scheme of the experimental procedure of type 1.

- 60 -

3.5.3 For Au and Ag nanoparticles (Type 2)

Evaporation of Au and Ag nanoparticles on HOPG and SiO2/Si surface using thermal evaporation (Au & Ag were wrapped with a W wire, which was resistively heated (preparation chamber in UHV 1)

XPS analysis (UHV 1)

STM analysis (UHV 2)

Oxidation experiment with atomic oxygen (UHV 1) Fig. 3.21 Scheme of the experimental procedure related to Ag and Au nanoparticles in UHV 1 and UHV 2; Type 2 On the basis of the STM and XPS results of Ag and Au nanoparticles prepared in UHV 2 system, one can also prepare Ag and Au nanoparticles in UHV 1 system equipped with XPS and the particle size can be determined by only XPS binding energy (Fig. 3.21 Type 2). Before evaporating Ag and Au nanoparticles, identical experimental conditions were applied to prepare the HOPG and Si substrates. And then, Ag and Au nanoparticles were evaporated in the UHV 1 system (preparation chamber) at room temperature. The Ag and Au rods were wrapped in a W wire, which was resistively heated and the substrates mounted onto the sample holder stand opposite to each other (the distance between W wire and the substrates is below 3 cm). In this system, typically 3.5 A was used to heating the W wire and the particle size was controlled by altering the current and evaporation time. It took about half an hour to grow Au films, which start to show yellow color. After preparing the Ag and Au nanoparticles, XPS analysis and oxidation experiment were carried out under same experimental condition like before. Furthermore, these Ag and Au nanoparticles were also transferred into UHV 2 chamber for STM analysis.

- 61 -

Results and Discussion

4. Ag nanoparticles on HOPG (Highly Ordered Pyrolytic Graphite)
4.1 Preparation of Ag nanoparticles on HOPG
To investigate the chemical activity toward atomic oxygen and CO depending on Ag nanoparticle size, in the present chapter, Ag nanoparticles with various particle sizes were grown on point defect HOPG surface which was created by Ar ion sputtering [212]. Then these Ag nanoparticles were exposed to atomic oxygen atmosphere, and their interaction with CO was studied using Scanning Tunneling Microscopy (STM) and X-ray Photoelectron Spectroscopy (XPS) [213, 220] In order to prepare Ag nanoparticles with various particle sizes on HOPG surface, different experimental parameters (Ar-sputtering time and Ag evaporation time) were used. It is well known that particles structure (size, shape, and etc) depends on the structure of support strongly [190]. We prepared four different size Ag nanoparticles on sputtered HOPG (Table 4.1). Figs. 4.1 4.4 show STM images of four samples used for the CO oxidation experiments in the present work. The lateral size of Ag nanoparticles on sputtered HOPG was measured by the width of the half maximum of the particle profile. The particle size determined in this way involves an overestimation due to the limited lateral resolution of STM tips, which is going to be corrected by combined XPS and STM studied. The actual lateral particle size is about 60% of the apparent particle size of the STM images [190].

- 62 -

The particles grow almost three dimensionally on the sputtered HOPG surfaces, which is different from that of previous studies about Au and Ag particles on WSe2 surface in which almost two dimensional grow (Width : Height = 20 : 2) can be observed as the particle height becomes about 2 nm [190, 222, 223]. This result indicates that different growth behaviors of metal particles can occur on various van der Waals surfaces. It is also worth mention that Ag nanoparticles can be prepared on a sputtered HOPG surface with various sizes and relatively narrow size distribution using our experimental condition.

Ag on HOPG (STM images) Sputtering (sec) Evaporation (min) Defect density (% of a ML) Ag particles size; Height (nm) Ag particles size; Width (nm)

Fig.4.1 5 8 ~2.5 1 3

Fig.4.2 5 30 ~2.5 3 7

Fig.4.3 2 60 ~1.0 2 5

Fig.4.4 2 90 ~1.0 5 10

Table 4.1 STM results (in fig. 4.1-4.4) of Ag nanoparticles on sputtered HOPG surface. The coverage was estimated with a combined STM and XPS analysis [190]

- 63 -

a)
50 40

b)

Count

30 20 10 0 2 4 6 8 10 12 14 16 18 20

Particle Diameter (nm)

50 40

c)

Count

30 20 10 0 0 1 2 3 4 5 6 7 8 9 10

Particle Height (nm)

Fig. 4.1 a) STM image of Ag nanoparticles on sputtered HOPG (5 sec. sputtering and 8 min. Ag evaporation), b) particle diameter distribution, c) particle height distribution, Tunneling parameters (135.3 nm x 132.3 nm. -3.0 V, 0.7 nA) [212]

- 64 -

a)
80

b)

60

Count

40

20

0 2 4 6 8 10 12 14 16 18 20

Particle Diameter (nm)

160 140 120

c)

Count

100 80 60 40 20 0 1 2 3 4 5 6 7 8 9 10

Particle Height (nm)

Fig. 4.2 a) STM image of Ag nanoparticles on sputtered HOPG (5 sec. sputtering and 30 min. evaporation), b) Particle diameter distribution, c) Particle height distribution, Tunneling parameters (214.5 nm x 214.5 nm. -1.5 V, 1.0 nA) [212]

- 65 -

a)
50 40

b)

Count

30 20 10 0 2
80 70 60 50

Particle Diameter (nm)

10

12

14

16

18

20

c)

Count

40 30 20 10 0 1 2 3 4 5 6 7 8 9 10

Particle Height (nm)

Fig. 4.3 a) STM image of Ag nanoparticles on sputtered HOPG (2 sec. sputtering and 60 minutes Ag evaporation). b) Particle diameter distribution c) Particle height distribution. Tunneling parameters (154.9 nm x 154.9 nm. -2.0 V, 2.4 nA) [212]

- 66 -

a)
60 50 40

b)

Count

30 20 10 0 2 4

Particle Diameter (nm)

10

12

14

16

18

20

120 100 80

c)

Count

60 40 20 0 1 2 3

Particle Height (nm)

10

Fig. 4.4 a) STM image of Ag nanoparticles on sputtered HOPG (2 sec. sputtering and 90 min. evaporation), b) Particle diameter distribution c) Particle height distribution, Tunneling parameters (394.4 nm x 349.8 nm. 1.9V, 2.1 nA) [212]

- 67 -

Fig. 4.5 shows the XPS results of Ag nanoparticles on sputtered HOPG surfaces (Figs. 4.1 - 4.4). XPS has been widely used to characterize electronic structures of metal nanoparticles on various substrates. XPS Ag 3d core level spectra are very sensitive to the particle size. The positive core level shift with decreasing particle size has been reported for Ag on various substrates by many other groups over the last 20 years, and also confirmed here. For the smallest Ag nanoparticles with an average diameter of 2 nm, a positive core level shift of 0.6 eV can be observed with respect to the 3d levels of Ag bulk crystals (368.3 eV) in fig. 4.5. Typically, 0.5 - 2.0 eV chemical shifts have been observed previously for small nanoparticles compared to the core levels of the respective bulk crystals [224 232]. The positive core level shifts for metal nanoparticles were often interpreted in terms of final state effects [224 - 228]. For smaller clusters on less conductive substrates, the positive hole, which is the final state of the photoemission process, can be less efficiently screened, causing a positive core level shift. However, initial state effect can also contribute to the positive chemical shifts, since changes of valence band positions or depletion of valence band electrons with decreasing particle size can affect the core level shifts [232]. Increased lattice strain with decreasing particle size can also contribute to the positive core level shifts by initial state effect [233]. That is, the core level shifts of Ag nanoparticles on HOPG prepared on our experiment conditions are related to the intrinsic initial electronic structure of the Ag particles in addition to the final state effect, and the initial state shifts can be explained by combinations of metal to support charge transfer, quantum-size effect (metalinsulator transition), presences of under-coordinated atoms and possibly also enhance lattice strain in smaller particles.

- 68 -

With decreasing particle size, the width of the Ag 3d state increases in Fig.4.5.b, and this does not necessarily reflect a broadening of the particle size distribution. In STM images (Figs.4.1 - 4.4), particle size distribution becomes rather narrower with decreasing particle size. The broadening of the Ag 3d core levels for smaller metal particles has been also found and attributed to the different life time broadenings of the hole state for different sized particles [234]. Other possibility is that deposition of Ag nanoparticles on different point defect site which induce different electronic affects between Ag nanoparticles and HOPG surface. More reasonable explanation for XPS analysis of Ag nanoparticles on HOPG was discussed in Ph.D. thesis of Ignachio [190] in more detail.
Ag3d 5/2
Intensity (arb. units.)
Ag 2 nm

3/2

a)

Ag 4 nm

Ag 6 nm

pure Ag
362 364 366 368 370 372 374 376 378 380

Binding Energy (eV)

369,2 369,1 369,0 1,5

b)
1,4

Binding Energy (eV)

FWHM (eV)

368,9 368,8 368,7 368,6 368,5 1,2 368,4 368,3 368,2 2 3 4 5 6 7 8 9 10 11 1,1 1,3

Ag nanoparticle size (nm)

Fig. 4.5 Changes of the Ag 3d states (Binding Energy and Full Width of Half Maximum) as a function of particle size on HOPG [212]

- 69 -

4.2 Oxidation properties of Ag nanoparticles on HOPG

Understanding the size-dependent electronic, structural and chemical properties of metal clusters on various substrates is an important aspect of heterogeneous catalysis. As mentioned before in chapter 2 (State of the art), catalytic activity can drastically change as a function of particle size. For Ag, similar size-selectivity in catalysis was reported for propylene partial oxidation and low temperature CO oxidation, the Ag nanoparticles smaller than 5 nm in diameter, whereas for the ethylene epoxidation, only Ag particles larger than about 30 nm can catalyze the reaction [201 - 204]. The reactivity between metal nanoparticles/surface and oxygen species is one of the important factors to explain the nature of high catalytic activity for CO oxidation. In general, increasing exposures of metal surfaces to oxygen yield various oxide formations [42, 235];

Chemisorbed atomic oxygen increasing dissolved oxygen in the bulk + surface oxide increasing increase of the thickness of the oxide layer (generally Ag2O for Ag).
In situ oxidation of Ag yielding Ag2O was achieved by exposing Ag(110) into a dc glow discharge plasma [23]or Ag (111) into a free-radical oxygen source (cold discharge, magnetically confined) [236]. In this study, in order to mimic severe oxidizing conditions of highpressure catalytic reactions under UHV conditions with minor contributions of impurities, Ag nanoparticles on HOPG were exposed to atomic oxygen (see chapter 3). Using Ag nanoparticels grown on HOPG with various particle sizes (chapter 4.1), we provide evidence for formation of different oxygen species depending on Ag particle size, resulting in size-selectivity of the catalytic activity of CO oxidation reaction.

- 70 -

There is no change of XPS C1s peak upon atomic oxygen exposure in this experiment, i.e. one cannot observe that the graphite structure is destroyed by atomic oxygen, due to the formation of CO or CO2, even though this cannot be completely excluded. According to the previous results [196, 202, 213, 220], the electronic and chemical properties of Ag nanoparticles on carbon surface are not much sensitive to the preparation methods of substrates, thus the oxidation behaviors of Ag would not much be affected by different structures of carbon supports.

a)
Intensity (arb. units)

Ag 3d

5/2

b)
Intensity (arb. units)
Ag2O

O1s
surface atomic oxygen dissolved oxygen

0.1 eV

Ag 3d

370

368 Binding Energy (eV)

366

increasing O exposure

5/2

3/2

364

366

368

370

372

374

376

378

526

Binding Energy (eV)

Binding Energy (eV)

528

530

532

534

536

Fig. 4.6 Oxidation of Ag bulk crystals using atomic oxygen; a) Ag 3d core level shifts b) O 1s states. The O 1s state at 529 eV can be attributed to the Ag2O formation. Atomic oxygen exposure is increasing from bottom to up For the Ag polycrystalline thin film on HOPG with a thickness of about 3 - 5 nm estimated by the attenuation of the C 1s peak in XPS, increased exposure of atomic oxygen yield appearance of the O 1s state at 529 eV, concurrently with the binding energy shifts of the Ag 3d core levels toward lower binding energies regime by 0.1 eV (Fig. 4.6). This result indicates the formation of Agoxide.

- 71 -

It is well known that the observed core-level binding energies for most transition metals are shifted towards higher binding energies upon oxidation. The strong electron affinity of oxygen should lead to an electron transfer to the absorbed oxygen atoms. However oxidation of Ag and other elements such as Cd, Cs, Rb and Ba [237, 238] results in negative core level shifts, which is opposite to most of other transition metals. The reason for this unusual shifts originate from their adverse contribution to the negative shift from ground-state charge distribution and from final-state relaxation processed different for metal and oxide binding states due to different core-hole screening abilities. Quantitative analysis reveals that at most first 1 ~ 2 atomic layers (< 0.5 nm) undergo oxidation under our experimental conditions in Fig. 4.6. The formation of dissolved oxygen and surface oxygen species, beside the Ag-oxide formation, can be identified in the O 1s peaks at 530 - 533 eV [235]. This result is in a good agreement with previous studies on Ag surfaces oxidized under high pressures O2 conditions at elevated sample temperatures. Under our experimental conditions, thicker oxide layers could not be obtained; however, previous experimental studies observed formation of Ag2O layers with a thickness of several tens of nm which was identified by a single distinct peak of the O 1s state at 529 eV prepared using free oxygen radical source at a sample temperature of room temperature [239]. Using molecular oxygen, the O 1s state at 529 eV cannot be found in the present work, indicating that Ag-oxide does not form at room temperature using molecular oxygen under high vacuum conditions.

- 72 -

XPS Ag 3d states of the Ag nanoparticles on HOPG

In Figs. 4.7 - 4.9, Ag 3d core level shifts of the Ag nanoparticles (1 to 6 nm in particle diameter; actual size) as a function of atomic oxygen exposure are displayed. In chapter 4-1, positive core level shifts and broadening of the Ag 3d width with decreasing Ag particle size on HOPG was confirmed, indicating that using XPS data, one can estimate the particle size without STM analysis [213]. For smaller Ag nanoparticles, onset of the chemical shift of the Ag 3d states to lower binding energies appears at much lower atomic oxygen exposures compared to those of larger particles: For Ag nanoaprticles with a mean diameter of 1 nm, a 10 minutes exposure of oxygen already leads to the saturation of the oxygen uptake. For larger particles, saturation can only be reached upon prolonged exposure times, implying that the smaller Ag nanoparticles can be more efficiently oxidized compared to the larger Ag nanoparticles. One of the reasons for this result is that the surface with more under-coordinated atoms to volume ration is much higher than that of larger particles. With increasing particle size, uptake of oxygen decreases (Table 4.2), is in line with the Ag 3d level shifts upon atomic oxygen exposures. For Ag nanoparticles with mean particle size of 5 nm and 6 nm in diameters (Fig. 4.8), very high atomic oxygen exposures yield positive chemical shifts, which does not reconcile the Ag-oxide formation. The origin of the positive Ag 3d state is not clear: one possible explanation might be that charging problem may be caused, when large Ag nanoparticles are heavily oxidized, resulting in the positive core level shifts. However, it is important to mention that the Ag 3d / C 1s intensity ratio remains constant during the oxidation experiments, indicating that significant changes of the particle size upon oxidation like sintering can be excluded [213]

- 73 -

Ag nanoparticles (1nm) on HOPG

Ag3d
3/2 5/2

Ag nanoparticle (3nm) on HOPG

Ag3d

Intensity (arb. units)

Intensity (arb. units)

380 376 372 368 364

Binding Energy (eV)

380 376 372 368 364

Binding Energy (eV)

Ag nanoparticels (4nm) on HOPG

Ag3d

Intensity (arb. units)

increasing O exposure

380 376 372 368 364

Binding Energy (eV)

Fig. 4.7 Changes of the Ag 3d core levels of the Ag nanoparticles with a mean size of 1, 3, and 4 nm as a function of atomic oxygen exposure [213]

- 74 -

Ag nanoparticles (5nm) on HOPG Ag nanoparticle (6 nm) on HOPG

Ag3d
Ag3d

Intsnsity (arb. units)

Intesntiy (arb. units)

380

376

Binding Energy (eV)

372

368

364

380

376

Binding Energy (eV)

372

368

364

Fig. 4.8 Changes of the Ag 3d core levels of the Ag nanoparticles with a mean size of 5 nm and 6 nm as a function of atomic oxygen exposure [213]
Ag3d5/2 Binding Energy (eV)
-1,0

-0,8

Ag particle size
1nm 3nm 4nm 5nm 6nm bulk

-0,6

-0,4

-0,2

0,0

20

40

60

80

100

120

140

Oxygen Exposure Time (min)

Fig. 4.9 Changes of the Ag 3d states of Ag nanoparticles with various sizes upon exposing to atomic oxygen [213]; For the last two steps of the atomic oxygen exposure for bulk-like Ag and 5 nm sized particles, a Pt filament current of 3.6 A was used to create atomic oxygen. For other cases, 3.5 A was used. The partial pressure of O2 in the chamber was 8 x 10-5 mbar during oxidation

- 75 -

XPS O1s states of the Ag nanoparticles on HOPG

To shed light on properties of oxygen in the oxides of Ag nanoparticles, O 1s core level shifts were collected (Figs. 4.10 - 4.16). The amount of O on a bare sputtered HOPG detected here is estimated to be below 10 % of a monolayer. Exposure of Ag nanoparticles on HOPG to atomic oxygen atmosphere yields various distinct states in the O 1s core level spectra (Fig. 4.11 - 4.15). The O 1s states between 529 eV 532 eV can be assigned to oxygen species bound to Ag nanoparticles, since these peaks cannot be detected, when Ag is absent on HOPG. The O 1s states at the binding energy regime above 530 eV are partially attributed to the oxygen species in the pure HOPG, since these peaks at 533 eV with a shoulder at 531 eV can be detected in the O1s state, when a pure sputtered HOPG is exposed to atomic oxygen (Fig. 4.10). It is not clear, how oxygen atoms are bound to bare HOPG. However sputtered HOPG is used in this experiment to making point defect sites on HOPG surface. Most likely, defect sites may interact with atomic oxygen, or small amount of water existing in our vacuum system can also adsorb on the HOPG surface or in the subsurface regime.

O 1s

Intensity (arb. units)

524

increasing O exposure

528

532

536

Binding Energy (eV)

Fig. 4.10 O 1s spectra from a pure HOPG sample obtained after exposing to various amounts of atomic oxygen [214]

- 76 -

O 1s spectra of oxidized Ag particles (1nm) on HOPG

O of Ag
Intensity (arb. units)

O of HOPG

increasing O exposure

524

528 532 Binding Energy (eV)

536

Fig. 4.11 O 1s spectra from 1 nm-sized Ag nanoparticles obtained after exposing to various amounts of atomic oxygen [213]

O 1s spectra of oxidized Ag particles (3nm) on HOPG

Intensity (arb. units)

Increasing O exposure

524

528 532 Binding Energy (eV)

536

Fig. 4.12 O 1s spectra from 3 nm-sized Ag nanoparticles on HOPG obtained after exposing to various amounts of atomic oxygen [213]

- 77 -

O 1s spectra of oxidized Ag particles (4nm) on HOPG

Intensity (arb. units)

Increasing O exposure

524

528 532 Binding Energy (eV)

536

Fig. 4.13 O 1s spectra from 4 nm-sized Ag nanoparticles on HOPG obtained after exposing to various amounts of atomic oxygen [213]

O 1s spectra of oxidized Ag particles (5nm) on HOPG

Intensity (arb. units)

Increasing O exposure
Ag2O/AgO

524

528 532 Binding Energy (eV)

536

Fig. 4.14 O 1s spectra from 5 nm-sized Ag nanoparticles on HOPG obtained after exposing to various amounts of atomic oxygen [213]

- 78 -

O 1s spectra of oxidized Ag particles (6nm) on HOPG

Intensity (arb. units)

Increasing O exposure

524

528 532 Binding Energy (eV)

536

Fig. 4.15 O 1s spectra from 6 nm-sized Ag nanoparticles on HOPG obtained after exposing to various amounts of atomic oxygen [213]

The O 1s spectra (O/Ag/HOPG) were summarized in Fig. 4.16 after subtraction of the respective O/HOPG background spectra (Fig. 4.10). For 1 nm and 3 nm-sized Ag nanoparticles, a single peak can be found centered at 531.8 eV and 531 eV, respectively, when atomic oxygen was exposed, whereas for larger particles (> 4 nm), two O 1s states at 529 eV and 531 eV can be observed at the initial stage of the atomic oxygen exposures. The O 1s state at 529 eV can be assigned to the formation of Ag2O based on previous results on Ag bulk crystals, even though AgO formation cannot be completely excluded, since Ag2O and AgO show nearly same binding energies of the O 1s [235, 240, 241]. Moreover, both AgO and Ag2O show negative Ag 3d core level shifts with respect to the metallic Ag [240, 241].

- 79 -

a)1 nm
additional 60 min. 3.5 A

b)3 nm
additional 60 min 3.5 A

c)4 nm
additional 60 min. 3.6~3.7 A

d)5 nm
additional 120 min 3.7~3.8 A

e)6 nm

Intensity (arb. units)

additional 30 min. 3.5 A

additional 30 min. 3.5 A

additional 30 min. 3.5 A

additional 60 min. 3.5~3.7 A

10 min. 3.5 A

10 min. 3.5 A

10 min. 3.5 A

10 min. 3.5 A

Ag2O

528 532 536 528 532 536 528 532 536

528 532 536 528 532 536

Binding energy (eV)

Fig. 4.16 Oxidation of Ag nanoparticles with a mean size of a) 1 nm b) 3 nm c) 4 nm d) 5 nm e) 6 nm. Increasing O exposure from bottom to up; (8 x 10-5 mbar, Pt filament current of 3.5 ~ 3.8 A). Atomic oxygen exposure times and Pt filament currents are given in the figure. For e) a different Pt filament was used, and therefore direct comparison of the experimental conditions is difficult. The corresponding O 1s spectrum of a pure HOPG exposed to atomic oxygen was subtracted from each spectrum. The O 1s state at 531 - 532 eV has been generally assigned to the dissolved oxygen in the bulk or atomically bound oxygen on metallic Ag surfaces in previous studies [202, 203, 235]. One can actually attribute O 1s states at 531 - 532 eV in Fig. 4.16 partially to dissolved oxygen species; however, considering that the appearance of the O 1s state at 531 - 532 eV accompanies a large Ag 3d core level shift in Figs. 4.7 - 4.9, the O 1s state at about 531-532 eV cannot be completely rationalized by taking only the formation of the dissolved/subsurface oxygen species into account.

- 80 -

Note that the formation of dissolved oxygen in metallic Ag does not accompany an Ag 3d core level shift [235]. It is interesting to note that the O 1s peak position of the Ag-oxide of 1 nm sized Ag particles (531.8 eV) is higher than those of larger nanoparticles by about 0.8 eV. This may be related to the different final state screening of the O 1s hole state in a smaller oxidized Ag nanoparticles superposed with some initial state contribution. Also, one should take into account that the error of determining binding energy of the O 1s peak of O/Ag nanoparticels becomes larger with decreasing particle size due to relatively large O/HOPG features overlapping with the O/Ag peaks. For the Ag nanoparticles showing two O 1s states centered at 529 eV and 531 eV at the initial stage of the oxidation (particle size > 4 nm), further exposure of atomic oxygen decreases the intensity of the O 1s state at 529 eV, and concomitantly the O 1s state centered at 531 eV further increases. This result indicates that the O 1s state at 529 eV corresponding to the Ag2O/AgO formation can be further transformed into a new species upon oxidation, identified by the O 1s state at 531 - 532 eV. The O 1s state at 531 - 532 eV has been generally interpreted to be dissolved oxygen in Ag, as mentioned above. Assigning the O 1s state of the oxidized Ag nanoparticles at 531-532 eV to the dissolved oxygen, one has to come up to a conclusion based on our data that further oxidation of Ag2O/AgO results in formation of dissolved oxygen. The formation of Ag-oxide to the dissolved oxygen species upon further atomic oxygen exposure is not likely and thus the peak at 531 - 532 eV is not attributed solely to the dissolved/subsurface oxygen species. Rather, it seems that Ag2O/AgO nanoparticles can transform into another novel oxide species upon further oxidation, which cannot be found on bulk-like Ag, and the O 1s state of the new oxide species is located at the same energy regime as that of the dissolved oxygen by chance.

- 81 -

To reveal the nature of oxygen species on oxidized Ag nanoparticles precisely, quantitative analysis of the XPS spectra was done. For the quantitative analysis of the data, the O 1s spectra in Fig. 4.18, in which O 1s correspond to oxygen on HOPG was subtracted from each O1s spectra of Ag on HOPG systems.

Particle size Oxidation condition O 1s peak center 1 nm 10 min (3.5A) 531,8 0,2 eV 3 nm 100 min (3.5A) 531 0,2 eV 4 nm 100 min (3.5A) 531 0,2 eV 5 nm 190 min (3.5-3.8A) 531 0,2 eV 6 nm 531 0,2 eV bulk 190 min (3.5-3.8A)

Ag : O 1:1.7 0,3 1:1.0 0,2 1:1.0 0,2 1:0,6 0,1 1:1.1 0,1 1:<0,1

Table 4.2 Ag particles on HOPG with various sizes were analyzed with XPS after the final stage of the oxidation experiments in the present work [220] For the determination of Ag : O ratio using XPS, different cross sections of Ag 3d state and O 1s state in the photoemission were taken into account [212]. Using an identical method, Ag : O ratio of 2 : 1 was found for the Ag oxide layers on Ag bulk crystals, indicative of Ag2O formation. For the particle size of 6 nm, a different setup was used for the atomic oxygen exposure, and therefore, the preparation conditions are not comparable with those of other particles. The Pt filament currents used for the atomic oxygen preparation are given in the parentheses. For all cases shown in this table, negative chemical shifts of 0.6 ~ 0.7 eV of the Ad 3d states were found with respect to those of the respective bulk like Ag particles, indicative of Ag-oxide formation. Note that Ag is known to show negative chemical shifts upon oxidation.

- 82 -

As it is shown table 4.2, the Ag : O stoichiometry reaches ~ 2 : 3, when very small Ag nanoparticles with a mean diameter of 1 nm on HOPG are exposure to atomic and/or thermally excited molecular oxygen. Considering that the number of O atoms can be even larger than that of Ag atoms in an oxidized particle, and in this case, the O 1s spectra show a single peak centered at 531 - 532 eV, appearance of the O 1s state at 531 - 532 eV by exposing Ag nanoparticles to atomic oxygen cannot be completely rationalized by dissolved oxygen species alone. The nature of the novel oxide species, which can be only found in these small Ag nanoparticles, is still questionable. Besides Ag2O, AgO is also one of the most stable oxides of Ag, and therefore one may assume that small Ag nanoparticles can form AgO [30]. Actually, prolonged exposure of ozone yields conversion of Ag2O to AgO on Ag crystalline [242]. The Ag : O ratios estimated by the XPS data are actually 1: 1 for the Ag nanoparticles with sizes of 3 and 4 nm; however, formation of pure AgO is not likely. Rather, formation of mixtures of metallic Ag, small amount of Ag2O and AgOx with x > 1 is more probable, because details of our XPS results do not match with the AgO formation. Ag nanoparticles on titania were treated using oxygen plasma, and in this case, AgO was identified by the Ag Auger features at 1132 eV, which was not observed in the present work [243]. For Ag nanoparticles with a diameter of 5 nm, one can argue that AgO or Ag2O are formed upon oxygen exposure, since the relative amount of Ag with respect to O is larger than 1. However, it should be noted that the O 1s peak centered at 531 eV is not in good agreement with the Ag2O/AgO formation. It should be noted that the O 1s binding energies of the oxidized Ag nanoparticles with mean diameters of 5 nm and 1 nm are analogous (531 - 532 eV) and significantly differ from those of Ag2O/AgO. Therefore, one may suggest that the 5 nm sized Ag nanoparticels form a mixed structure, in which a part of the particle corresponds to novel species with a Ag : O stoichiometry of ~ 2 : 3 and the rest of the particle being metallic Ag possibly with some dissolved oxygen species.

- 83 -

As mentioned above, the O 1s core level spectra of AgO and Ag2O are almost identical, i.e. the O 1s state of AgO is located at 528 - 529 eV, also disagreeing with our results of the novel oxide species [240, 241]. Similar results to ours (negative Ag core level shifts in XPS, O 1s at 531 eV) were observed on Ag samples, exposed to air for long time; however, the Ag : O stoichiometry was not well defined [244, 245]. When Ag-oxide prepared under O2 microwave excitation (showing a single O 1s state at 529 eV) was exposed to high O2 pressure at 420 K, the O 1s state at 531 eV increased, indicating that the O 1s state at 531 eV can be related to more oxygen-rich Ag-oxide [246]. Structures of the novel Ag-oxide species found in the present work should be determined using experimental and theoretical methods in the future. It is worth mentioning that using atomic oxygen with a high kinetic energy at a sample temperature of about 500 K (hyperthermal atomic oxygen), not only AgO and Ag2O but also Ag2O3 and Ag3O4 could be observed on Ag (111) and Ag(100) surfaces [247]. One can eventually observe formation of Ag2O3 or Ag3O4 using specific synthesis methods; however, Ag2O3 and Ag3O4 are known to be metastable and volatile at room temperature so that observation of those oxides with higher Ag : O stoichiometries under UHV conditions at room temperature is not expected. One may argue that originally unstable compounds may become stable with decreasing particles, i.e. the thermodynamic stability of a compound may exhibit some size effects. Other possible explanation of our data is the formation of a carbonate compound, Ag2CO3. In Fig 4.17 a), however, the C 1s signals, formation of carbonate species is vague in our XPS, since carbonate formation should give a distinct peak at 287 - 288 eV. The valence band spectra measured in syncrotron radiation by Kim et al. [220] is showing the evidence of Ag-carbonate formation (Ag2CO3 ) upon atomic oxygen exposures (Fig. 4.17 b) and it is in good agreement with our Ag 3d core level shift results depending on the oxygen exposure.

- 84 -

Both the Ag-oxide and the Ag-carbonate formation gives a negative core level shifts of ~ 0.5 eV and therefore one cannot discriminate between oxide and carbonate formation of Ag based on the Ag 3d core level shift data [248].

Difference spectra

a)
Intensity (arb. units)

Intensity (arb. units)

before and after oxidation Ag(6 nm) on HOPG

b)

Carbonate

carbonate

a) O/Ag/HOPG

b) O/HOPG

282

Binding Energy (eV)

285

288

291

294

5 10 15 Binding Energy (eV)

20

Fig. 4.17 a) C 1s specta of Ag (6nm) on HOPG: before (circle) and after (square) oxidation. O 1s shows a strong peak at 531 eV, which we assign to Agoxide, b) The valence band spectrum of the Ag/HOPG sample (a) and the HOPG sample (b) were subtracted from that of O/Ag/HOPG (O exposure time 70 minutes) and that of O/HOPG, respectively [220].

In order to check the reliability of our quantitative analysis of the XPS data, an Ag polycrystalline surface was heavily oxidized using atomic oxygen, and the stoichiometry of the Ag oxide layers was determined. As it is shown in Fig. 4.18, the Ag oxide layers can be identified by the O 1s state at 529 eV, and Ag 3d states shifted to the lower binding energy by 0.7 eV with respect to the non-oxidized Ag. The Ag : O stoichiometry of the oxide layer is estimated to be ~ 2 : 1, in almost agreement of the Ag2O formation. However, one cannot fully exclude the formation of Ag-carbonate in the oxide layer of Ag polycrystalline.

- 85 -

Intensity (count rate)

a)
160000
Ag5/2

Ag 3d metallic Ag
Ag3/2

80000 Ag-oxide 0 366 372 Binding Energy (eV) 378

b)
Intensity (count rate)

O 1s

40000

Ag oxide

20000

0 528 532 Binding Energy (eV) 536

Fig. 4.18 a) Ag 3d and b) O 1s spectra of an Ag polycrystalline sample, which was oxidized using atomic oxygen. Gauss functions were used to fit the experimental spectra, and these were used for the quantitative analysis of the Ag : O stoichiometry of the surface oxide layers. Taking the different cross sections of Ag 3d and O 1s for photoemission into account [212], we come up to the stoichiometry of ~ 2 : 1 for Ag : O. The data acquisition time per data point for the O 1s spectrum was four times longer than that of the Ag 3d spectrum which had to be taken into account for the quantitative analysis.

- 86 -

Previously, it was suggested that existence of the ionic oxygen is necessary for the ethylene epoxidation. The ionic oxygen was suggested to be atomically bound oxygen in Ag (I)-oxide-like structures [196, 200, 202, 203]. In the Ag bulk crystal and larger Ag particles, Ag2O is stable under oxidizing conditions, whereas under similar conditions, Ag2O of smaller nanoparticles can further transform into other Ag-oxides, in which the formal oxidation state of Ag is probably higher than (I), i.e. surfaces of small Ag nanoparticles are easily passivated under oxidizing conditions. This result reconciles the decreased catalytic activities of the Ag nanoparticles smaller than 30 nm for ethylene epoxidation. Similar O 1s spectra to those of ours were observed after exposing Ag nanoparticles to high-pressure O2 (100 pa) at 470 K, suggesting that the experimental results in the present work could be also relevant under real catalytic conditions [202, 203]. Based on our quantitative analysis and further experiment data done by ref. [220], one could conclude that a Ag-compound with a Ag : O stoichiometry of ~ 2 : 3 can form in Ag nanoparticles supported by carbon. More detailed studies using photoemission spectroscopy suggest that at the initial stage of the oxidation, nanoparticles consists of Ag-oxides, Ag-carbonate and metallic parts with some subsurface/dissolved oxygen. At the later stage of oxidation, Agoxide is further converted into Ag-carbonate, thus forming a mixture of Agcarbonate and metallic Ag with subsurface/dissolved oxygen. Considering that the Ag2O/AgO at 529 eV was previously found to be the only active species for the CO oxidation, whereas the species characterized by the O1s state at 531 532 eV is not (discussed in chapter 4.3), the carbonate formation can be suggested to be responsible for the deactivation of Ag catalysts.

- 87 -

4.3 Reactivity toward CO of the oxidized Ag nanoparticles

To shed light on the nature of the oxygen species of oxidized Ag nanoparticles, CO molecule was exposed to these nanoparticles. In Fig. 4.19, the Ag 3d and O 1s core level spectra of partially oxidized Ag nanoparticles with a mean diameter of 4 nm before and after exposing to 1000 L (1 Langmuir = 1 x 10-6 mbar x 1 second) of CO are shown. The O/HOPG background was not subtracted from the O/Ag/HOPG in O1s spectra in order to observe the reactivity the CO with oxygen on HOPG. The oxygen species on pure HOPG does not react with CO. The O 1s states above 530 - 531 eV also do not changes upon CO exposure, whereas the O 1s state at 529 ~-530 eV decreases in intensity after exposing the sample to CO (Fig. 4.19 b). In the Ag 3d state in Fig. 4.19 a, also, it is evident that the Ag nanoparticles are initially oxidized (negative core level shift), then reduced by CO (positive core level shift). Obviously, the O 1s state centered at 529 eV, which is assigned to Ag2O/AgO, can readily react with CO to CO2, whereas other oxygen species (mostly regard as Ag-carbonate) are inert toward CO [220]. The Ag nanoparticles smaller than 3 nm show a single peak at 531 eV in the O 1s spectrum from the very early stage of the atomic oxygen exposure (Fig. 4.16), and CO does not react with oxygen in these Ag nanoparticles (Fig. 4.20). This result was observed for Ag nanoparticles with various oxygen uptakes, indicating that the 3 nm sized particles are not good catalysts for the CO oxidation, independently of amount of oxygen incorporated in Ag. Summarizing our results on CO oxidation, Ag nanoparticles smaller than 3 nm in diameter are most likely not reactive towards CO oxidation under real catalytic conditions when HOPG is used as substrate, whereas larger Ag nanoparticles can become active towards CO oxidation, when not too highly oxidizing conditions are used, in which the active Ag oxide can survive, e.g. when a relatively high CO/O2 ratio is used, Ag nanoparticles larger than 3 nm can become quite active for CO oxidation.

- 88 -

Ag nanoparticles : 4 nm

a)
Intesnity (arb. units)

Ag 3d
A g 3 d 5 /2
C O 1000L O e x p o su re 372 374 376

CO 1000L O exposure

5/2
366 369 372

3/2
375

as prepared Ag 378 381 384

Binding energy (eV)

b)
Intesntiy (arb. units)

O 1s

1) O exposure

2) CO 1000L 526 528 530 532 534 536

Binding Energy (eV)

Fig. 4.19 a) Ag 3d and b) O 1s core level spectra from Ag nanoparticles with a mean size of 4 nm after exposing to atomic oxygen and subsequent reduction by CO at room temperature [213]. The O/HOPG background was not subtracted from the O/Ag/HOPG in O1s spectra.

- 89 -

Ag nanoparticles : 3 nm

a)
Intesntiy (arb. units)

Ag 3d

CO 1000L O exposure CO 1000 L

5/2
366 369 372

3/2
375

O exposure as prepared Ag 378 381 384

Binding Energy (eV)

b)
Intensity (arb. units)

O 1s

CO 1000 L O exposure CO 1000 L O exposure 528 530 532 534 536 538 540

Binding Energy (eV)

Fig. 4.20 a) Ag 3d and b) O 1s core level spectra from Ag nanoparticles with a mean size of 3 nm after exposing to atomic oxygen and subsequent reduction by CO at room temperature [213]. Oxygen peak at 533-534 eV are attributed to the O/HOPG.

- 90 -

It is important to mention that the oxidation and reduction cycle of Ag nanoparticles larger than 4 nm can be repeated. As it is illustrated in Figs. 4.21 and 4.22, the oxides of Ag nanoparticles reduced by CO can be restored using atomic oxygen atmosphere, and in this case the catalytically active Ag-oxide species, which has been consumed by CO, can form again. After the initially oxidized Ag nanoparticles are reduced, subsequent atomic oxygen exposures result in appearance of 529 eV shoulder in the O 1s core level spectrum with a higher intensity than that before the reduction. An additional reduction using 1000 L of CO yields disappearance of the shoulder at 529 eV. As it is shown in Fig. 4.22, the oxidation/reduction cycle can be repeated implying that Ag nanoparticles could probably act as a good catalyst for CO oxidation und CO/O2 steady-state conditions. When a larger amount of atomic oxygen was used to leave a single peak at 531 eV for oxygen in Ag without the shoulder at 529 eV (circle in Fig. 4.22; as mentioned in Fig. 4.16), CO oxidation does not take place, confirming that the Ag2O/AgO is the only oxygen species here, which can react with CO to CO2 in our experimental conditions. In addition to the change of O1s state, concurrently the negative and positive Ag 3d core level shifts can be observed by oxidation and reduction cycle of the Ag nanoparticles in Fig. 4.21. It is interesting to note that after exposing to 1000 L of CO, the Ag2O/AgO is almost completely consumed, suggesting that the CO to CO2 conversion probability is higher than 0.1 % [249]. This value is comparable to that of oxygen-rich Ru surfaces which turned out to be RuO2 layers. Ru has been shown to be less reactive towards CO oxidation under UHV conditions; however, under high-pressure conditions, CO oxidation reactivity of Ru is higher than that of the Pt-group metals, due to the formation of catalytically active oxide species. It is noteworthy that the CO oxidation reactivity of Agoxide nanoparticles is comparable to that of the catalytically active RuO2 layers.

- 91 -

Ag nanoparticles : 6 nm

Ag 3d
Intensity (arb. units)

a)
O exposure CO 1000L O exposure CO 1000L O exposure CO 1000L O exposure CO 1000L O exposure

5/2

3/2

366 369 372 375 378 381 384

Binding Energy (eV)

3/2

Ag 3d

b)
O CO O CO O CO O CO O

373

374

375

376

377

Fig. 4.21 Ag 3d core level spectra from Ag nanoparticles with a mean size of 6 nm after exposing to atomic oxygen and subsequent reduction by CO at room temperature [213]; The reduced particles can be oxidized using atomic oxygen and reduced again using CO, i.e. the oxidation and reduction cycles is reversible.

- 92 -

Ag nanoparticles : 6 nm

O 1s
CO 1000L

Intensity (arb. units)

O exposure O exposure CO 1000L O exposure CO 1000L O exposure CO 1000L O exposure CO1000 L O exposure 528 530 532 534 536 538 540 542

Binding Energy (eV)

Fig. 4.22 O 1s core level spectra from Ag nanoparticles with a mean size of 6 nm after exposing to atomic oxygen and subsequent reduction by CO at room temperature [213]; The reduced particles can be oxidized using atomic oxygen and reduced again using CO, i.e. the oxidation and reduction cycles is reversible.

Most of the Pt group based heterogeneous catalytic reactions are considered to occur via Langmuir-Hinshelwood mechanism. Moreover, a slightly different reaction mechanism (Mars van Krevelene mechanism) has been proposed, in which the lattice oxygen of the oxides can be involved in chemical reactions [193, 250]. Our results indicate that the Mars van Krevelen mechanism can become more important. For CO oxidation on Ag nanoparticles larger than 3 nm, we found evidence that the Mars van Krevelen mechanism operates.

- 93 -

Summarizing, increased chemical activities of Ag nanoparticles towards oxide formation were found in the Ag particle diameter range below 6 nm. Moreover, different Ag-oxides were found for these Ag nanoparticles, which have not been observed for the Ag bulk crystals with similar preparation methods. For the Ag nanoparticles larger than 4 nm, Ag2O/AgO can form at the initial stage of the atomic oxygen treatment, whereas for the larger oxygen exposures, Ag2O/AgO transformations into a different oxygen species most likely Ag-carbonate. However, for smaller Ag particles (< 4 nm), Ag2O/AgO cannot be detected. Among various oxygen species observed in the present work, only oxygen identified by the O 1s state at 529 eV (Ag2O/AgO) can react with CO to CO2.

- 94 -

5. Au nanoparticles on HOPG
5.1 Preparation of Au nanoparticles on HOPG
The interactions of oxygen and Au nanoparticles have been subject of many previous studies due to their strong size-selectivity of toward heterogeneously catalyzed reactions [44 48, 70]. In spite of extensive studies on Au nanocatalysis, as mentioned above (chapter 2), many questions still remain open. A number of mechanisms have been proposed explaining the high activity of gold nanoaprticles toward CO oxidation reactions. These include quantum size effects of two-layer Au islands [45], negative charging of the Au clusters [71] and the abundance of low Au-Au coordination sites [102]. For example, the oxygen-Au interaction (dissociative chemisorption of oxygen as well as molecular adsorption) was suggested to be one of the most important elementary steps under reaction conditions. CO-oxidation on Au nanoparticles was suggested to be mediated by either dissociatively chemisorbed, or molecularly bound oxygen forming carbonates-like species with CO as a reaction intermediate [76, 79, 103, 109, 143]. In general, more attention has been paid to the activation of oxygen molecules on Au than the reaction between CO and O on Au nanoparticles; previous studies have proposed that atomic oxygen attached to Au nanoparticles can readily react with CO to CO2 regardless of the particle size, implying that activation of molecular oxygen is the key of the size-selectivity of heterogeneously catalyzed reactions [111, 112]. No reaction of Au55 clusters with atomic oxygen was suggested, which was attributed to the closed-shell geometry of Au55 [44]. Another important issue is related to the charge state of catalytically active Au; it has been argued that positively, neutral or negatively charged Au is catalytically active [70, 81, 95, 96, 142].

- 95 -

In the present work, we investigated chemical reactivity toward CO oxidation of the oxygen species on Au nanoparticles and nanostructured thin films grown on highly ordered pyrolytic graphite (HOPG) surfaces. Au-oxide species of Au nanoparticles can directly participate in catalytic reactions of CO oxidation. However, different chemical activities toward the CO oxidation reaction were observed for Au nanoparticles with very small particles size (<< 3 nm), which will be discussed in chapter 6. First, three different Au samples with different particle sizes were prepared. Fig. 5.1 shows the XPS Au 4f chemical shifts. One can observe no significant change with in about 0.1 eV with respect to the bulk like Au value, even for particle sizes smaller than 5 nm in a diameter. Previously, ~ 0.4 eV XPS chemical shifts have been observed for Ag nanoparticles with similar size on a HOPG surface. The different behavior as a function of particle size on both systems has been discussed in the Ph. D. thesis of Ignacio [190] in more detail. Summarizing, a different metal to support charge transfer gives rise to different chemical shifts depending on the system.

0.1 eV

Au 4f

Intensity (arb. units)

Au bulk sample C

7/2

5/2

sample B sample A

80

82

84

86

88

90

92

Binding Energy (eV)

Fig. 5.1 XPS Au 4f level shifts as a function of particle size [190]

- 96 -

a)
25 20 15 10 5 0 0
50 40 30

b)

Count

10

12

14

16

Particle Diameter (nm)

c)

Count

20 10 0 0 2 4

Particle Height (nm)

10

12

14

16

Fig. 5.2 a) STM image of Au nanostructures on sputtered HOPG (10 sec. sputtering and 2 min. Au evaporation),300 nm x 300 nm, -5.3 V, 0.08 nA, Sample A [190]

- 97 -

a)
30 25 20

b)

Count

15 10 5 0 0 2 4

Particle Diameter (nm)

10

12

14

16

70 60 50

c)

Count

40 30 20 10 0 0 2 4 6 8 10 12 14 16

Particle Height (nm)

Fig. 5.3 a) STM image of Au nanostructures on sputtered HOPG (5 sec. sputtering and 5 min. Au evaporation), 256 nm x 256 nm, -2 V, 0.08 nA, Sample B [190]

- 98 -

a)
14 12 10

b)

Count

8 6 4 2 0 5 10 15 20 25 30 35 40 45 50

Particle Diameter (nm)

20

c)

15

Count

10

0 2 4

Particle Height (nm)

10

12

14

Fig. 5.4 a) STM image of Au nanostructures on sputtered HOPG (3 sec. sputtering and 15 min. Au evaporation 220 nm x 220 nm, -1.2 V, 1.3 nA, Sample C [190]

- 99 -

Figs. 5.2 - 5.4 show STM images of three samples used for the COoxidation experiments in the present work. As it has been shown for Ag on HOPG in chapter 4, metal atoms nucleate at defect sites on the HOPG surface created by sputtering. Metal particles can be prepared with relatively narrow size distributions, and the particle density and the mean particle size can be varied by using different sputtering conditions and metal coverages. STM images in Fig 5.2 and 5.3 show Au nanoparticles with mean lateral particle sizes of below 5 nm and 8 nm, respectively. The mean particle heights of both samples are ~ 2 nm and ~ 3 nm, respectively. In the STM images, as mentioned before, the particle size is generally overestimated due to the limited resolution of the tip. It was previously shown that the actual lateral particle size is about 60 % of the apparent particle size of the STM images, which is determined by the width of the half maximum of the particle profile. For sample C, more than 90 % of the carbon surface is covered by Au, and hexagonal structures of Au can be observed, implying that the growth of the Au film along the (111) direction perpendicular to the surface is favored. Due to the high coverage of Au in the Fig. 5.4, one may assume that the electronic properties of Au for this sample are nearly bulk-like. Au on HOPG (STM images) Sputtering (sec) Evaporation (min) Emission current (mA) Defect density (% of a ML) Au particles size; height (nm) Au particles size; width (nm) Sample A Fig.5.2 10 2 15 ~5 ~2 ~5 Sample B Fig.5.3 5 5 17 ~2.5 ~2-3 ~8 Sample C Fig.5.4 3 15 22.5 ~1.5 ~3-4 ~23

Table 5.1 STM results of Au nanoparticles on a sputtered HOPG surface. The coverage was estimated with a combined STM and XPS [190]

- 100 -

5.2 Oxidation properties of Au nanoparticles on HOPG

To shed light on the nature of oxygen species formed by exposing various Au nanostructures to atomic oxygen atmosphere, XPS was used. When the nanostructured Au film (Fig. 5.4; sample C) was oxidized, increase of a single peak centered at 529 - 530 eV can be observed in the O 1s state in Fig. 5.5 c. A shoulder at higher binding energies can be observed in the Au 4f states (Fig. 5.7 c; square zone), which could not be observed before the oxidation. At the same time, the main Au 4f peak decreases in intensity upon oxidation. The appearance of the shoulders at higher binding energies in the Au 4f states upon oxidation has been widely interpreted as formation of Au2O3 (Au (III)). Oxidation of metal generally results in the appearance of shoulders at higher binding energies, or shifts of the metal core levels to higher binding energies, which can be understood in terms of strong metal to oxygen charge transfer (initial state effects) in combination with final state effects. It is worth emphasizing that XPS is one of the most widely used experimental methods to identify metal oxide species in metal catalysts [26, 265]. Also for Au nanoparticles, Au (III) has been generally characterized using XPS, allowing us to attribute the shoulder of the Au 4f level to the Au (III) species [44]. The quantitative analysis of the Au (III) species in the Au 4f states and the O 1s signals also suggests a stoichiometry of ~ 2 : 3 for Au : O, in line with Au2O3 formation. For this quantitative analysis, the cross sections of Au4f and O1s in the photoemission process were assumed to be 3 : 10 [183].

- 101 -

O1s

a)

Intensity (arb. units)

b)

c)

526

528

530

532

534

536

Binding Energy (eV)

Fig. 5.5 O1s XPS spectra of oxidized various Au nanostructures on HOPG. (before subtraction of the O/HOPG signals (red zone) from O/Au/HOPG) a) O1s peak for sample A (~ 5 nm in diameter), b) sample B (~ 8 nm in diameter) and c) sample C (~ 23 nm in diameter)

For the Au nanoparticles with apparent lateral size of ~ 8 nm (sample B), exposures of the sample to atomic oxygen result in increase of three different states in the O 1s spectrum (Fig. 5.5 b). The O 1s peaks centered at 533 - 534 eV can be attributed to the O on/in HOPG (red zone in Fig. 5.5 b), since exposure of the sputtered bare HOPG to atomic oxygen yields a major O 1s peak at 533 - 534 eV and a shoulder at 531 eV (Fig. 4.12) [213, 266]. In Figs. 5.5 and 5.6, the O 1s signals are displayed before and after subtraction of the O/HOPG signals, respectively. Therefore one can say that the O signal in Fig. 5.6 (blue zone and green curve in Fig. 5.5) can be attributed to oxygen species bound to Au.

- 102 -

The O 1s state at 529 - 530 eV can be attributed to Au2O3 based on comparison with the results in Fig. 5.6 c. The oxygen species showing the O 1s state at 531 - 532 eV cannot be found for nanostructured Au films (sample C), implying that decrease of the particle size opens a formation route of a new oxygen species under the same experimental conditions. The nature of the oxygen species identified by the O 1s state at 531 - 532 eV is not clear yet. As it will be shown later, this O species does not react with CO to form CO2, and this oxygen species is not responsible for the Au 4f core level shifts. Since no Au 4f core level shifts can be associated to this oxygen species, this species can be assigned to either chemisorbed oxygen on Au surfaces or subsurface oxygen (or dissolved oxygen). However, we assign the O 1s state at 531 - 532 eV to subsurface oxygen species, since chemisorbed oxygen on Au surfaces should readily react with CO to CO2 according to previous studies [55, 61, 156, 267]. We can not exclude the possibility that there might be additional oxygen states of Au above 533 eV, which can be buried by the O/HOPG signals. Also, it is possible that the O 1s peak at 529 - 530 eV comes from more than two different oxygen species, which we cannot clearly be discriminated, i.e. besides Au-oxide, chemisorbed oxygen, which is electrically quite analogous to the lattice oxygen of Au-oxide, can exit and show an O1s feature at the same binding energy regime [156].

- 103 -

O1s Intensity (arb. units)

a)
5.0 x 10 , 40 min 5.0 x 10 , 10 min 5.0 x 10 , 180 sec 5.0 x 10 , 60 sec as prepared Au
-5 -5 -5 -5

526

528

530

532

534

536

538

Binding Energy (eV)

O1s Intensity (arb. units)

b)

5.0 x 10 , 30 min 5.0 x 10 , 10 min 5.0 x 10 , 180 sec 5.0 x 10 , 60 sec as prepared Au
-5 -5 -5

-5

526

528

530

532

534

536

538

Binding Energy (eV)

O1s Intensity (arb. units)

c)

8.0 x 10 , 60 min 8.0 x 10 , 30 min 8.0 x 10 , 10 min as prepared Au

-5 -5

-5

526

528

530

532

534

536

538

Binding Energy (eV)

Fig. 5.6 Oxidation of various Au nanostructures studied using XPS (O 1s states) [214] a) O1s peak for sample A (~ 5 nm in diameter), b) sample B (~ 8 nm in diameter) and c) sample C (~ 23 nm in diameter)

- 104 -

Au 4f Intensity (arb. units)

a)
increasing O exposure

as prepared Au
82 84 86 88 90 92 94

Binding Energy (eV)

Intensity (arb. units)

Au 4f

b)
increasing O exposure

as prepared Au
82 84 86 88 90 92 94

Binding Energy (eV)

Au 4f Intensity (arb. units)

c)
increasing O exposure

as prepared Au Au bulk
82 84 86 88 90 92 94

Binding Energy (eV)

Fig. 5.7 Oxidation of various Au nanostructures studied using XPS (Au 4f states) [214] a) Au4f peak for sample A (~ 5 nm in diameter), b) sample B (~ 8 nm in diameter) and c) sample C (~ 23 nm in diameter)

- 105 -

In general, metal oxide formation results in a significant change of the metalmetal distance, since direct metalmetal bonds are broken and oxygen atoms are inserted into the metalmetal bonds upon oxidation. The metal oxide formation therefore leads to a different electronic structure with respect to the metallic counterpart, e.g. loss of the plasmon resonance and chemical shifts of the metal core levels can be observed. Metal oxides can show a band gap since there is no overlap of the metalmetal bond anymore, even though there are exceptional cases such as RuO2, which is metallic [193]. In contrast to the oxide formation, subsurface species or dissolved oxygen species are incorporated within the lattice of metallic substrates without significant change of the metalmetal bonds, even though small changes of the metalmetal distance due to the relaxation or reconstruction cannot be excluded. The metallic characteristics of substrates can be preserved upon subsurface or dissolved oxygen formation. For example, no chemical shift of Ag is observed, when chemisorbed or dissolved oxygen exists, whereas oxidation of Ag results in significant Ag 3d shifts [235]. For even smaller Au nanoparticles (sample A), two different oxygen species are found, similar to the results of larger particles (Fig. 5.6). For the Au 4f levels (Fig. 5.7) also, shoulders appear at higher binding energies upon oxidation. We cannot obtain particles completely oxidized, implying that the relatively thin oxide layers prevent further oxidation of the deeper layers of Au nanoparticles.

- 106 -

5.3 Reactivity toward CO of the oxidized Au nanoparticles

To study chemical reactivity of different oxygen species formed by exposing various Au nanostructures to atomic oxygen, CO was exposed to the oxidized Au/HOPG. The nanostructured Au film (sample C) oxidized by atomic oxygen shows a single peak of the O 1s state centered at 529 eV (Fig. 5.8 c), which is rapidly decreased in intensity with increasing exposure of CO. The shoulders of the Au 4f level associated to the Au (III) species disappear concomitantly (Fig. 5.9 c), implying that oxygen species bound to Au (III) can actively participate in catalytic reactions such as a CO oxidation reaction. This result suggests that the Au (III) species of Au catalysts not only plays a role as chemical glue [26], but oxygen atoms in the Au-oxide lattice can directly participate in chemical reactions. Our results rather suggest that once Au-oxide forms under real catalytic conditions, they can readily react with CO [214]. We have shown that two different oxygen species can form for Au nanoparticles smaller than ~ 10 nm oxidized by atomic oxygen in Fig. 5.6 (samples A and B). The results in Fig. 5.8 clearly show that the oxygen (at 529 - 530 eV) associated with Au (III) readily reacts with CO, whereas the other oxygen species (at 531 - 532 eV), which we suggested to be subsurface oxygen, is inert. In Fig 5.9, Au 4f level also shows the disappearance of the Au (III) species upon exposing to CO. Previously, various oxygen species on Au single crystal surfaces were characterized using thermal desorption spectroscopy (TPD), ultraviolet photoelectron spectroscopy (UPS) and work function measurements, and it was concluded that the chemisorbed oxygen and Auoxide can react with CO, whereas dissolved oxygen is inert towards CO oxidation. Our results here are in line with the previous results [55, 156].

- 107 -

O1s Intensity (arb. units)

a)

O exposure CO 1,000 L CO 3,000 L

as prepared Au

526

528

530

532

534

536

Binding Energy (eV)

O1s Intensity (arb. units)

b)

O exposure CO 1,000 L CO 3,000 L

as prepared Au

526

528

530

532

534

536

Binding Energy (eV) O1s

c)

Intensity (arb. units)

O exposure CO 1,000 L as prepared Au

528 530 532 534 536

526

Binding Energy (eV)

Fig. 5.8 Reduction of oxidized Au nanostructures (O 1s states) [214] a) O1s peak for sample A (~ 5 nm in diameter), b) sample B (~ 8 nm in diameter) and c) sample C (~ 23 nm in diameter)

- 108 -

Au 4f Intensity (arb. units)

a)

CO 3,000 L CO 1,000 L O exposure as prepared Au

82 84 86 88 90 92 94

Binding Energy (eV)

Au 4f Intensity (arb. units)

b)
CO 3,000 L CO 1,000 L O exposure as prepared Au

82

84

86

88

90

92

94

Binding Energy (eV)

Intensity (arb. units)

Au 4f

c)

CO 1,000 L O exposure as prepared Au

82 84 86 88 90 92 94

Binding Energy (eV)

Fig. 5.9 Reduction of oxidized Au nanostructures (Au 4f states) [214] a) Au4f peak for sample A (~ 5 nm in diameter), b) sample B (~ 8 nm in diameter) and c) sample C (~ 23 nm in diameter)

- 109 -

According to the Langmuir-Hinshelwood mechanism (or Mars van Krevelene mechanism), CO should adsorb on the surface before reacting with oxygen [chapter 2]. The adsorption sites of CO on oxidized Au nanoparticles are not clear yet. It is possible that the surface of an Au particle is completely converted into stoichiometric Au oxides, and CO adsorbs on coordinatively unsaturated Au atoms of the Au-oxide surface. However, we cannot exclude the possibility that the fully oxidized Au nanoparticle surface consists of mixtures of metallic and oxidized Au, and CO adsorbs on metallic Au, reacting with lattice oxygen of the Au oxide, i.e. the reaction takes place at the boundary of Au and Au-oxide. To shed light on this problem, CO adsorption on stoichiometric Au oxide surfaces would be required experimentally or theoretically. Considering that several thousands of Langmuirs (Langmuir (L), 1L = 1 x 10-6 Torr x 1sec) of CO are required to completely consume the oxygen species of Au-oxides, the CO to CO2 conversion probability can be estimated to be on the same order of magnitude to the reactivity of RuO2 layers, which is shown to be very active for CO oxidation at room temperature. Many different reaction mechanism reported in the literature (CO + dioxygen species on metallic Au, CO + O, reactions on the periphery of Au/oxide supports; discussed in chapter 2 in detail) can be responsible for the enhanced catalytic reactivity of Au nanoparticles, and we suggest that the direct participation of the lattice oxygen of Au-oxides for catalytic reactions, which is known to be Mars van Krevelene mechanism, can also play a significant role for the enhanced catalytic activity of Au nanoparticles. Considering that many transition metal oxides such as AgO2, RuO2, and Au2O3 can readily react with CO to form CO2 with a CO to CO2 conversion probability of the order of ~ 0.1 % or even higher, and smaller nanoparticles can more efficiently form metal oxides than bulk metals under real catalytic conditions, we suggest that the Mars van Krevelene mechanism (consumption of lattice oxygen of oxides during reactions) may generally play an important role for catalytic reactions of metal nanoparticles [212, 214].

- 110 -

Au4f

a)

O exposure

Intensity (arb. units)

as prepared Au

b)

O exposure CO 3,000 L

c)

CO 3,000 L O2 exposure
82 84 86 88 90 92 94

Binding Energy (eV)

Fig. 5.10 Au 4f XPS core level spectra: a) After the Au nanostructure was oxidized by atomic O, b) after reduction by CO and c) after reduced Au nanostructurse by CO were oxidized by molecular O (~ 400000 L) again Inserts; each spectras were compared after fitting the base line

- 111 -

To make the CO oxidation of Au-oxide completely catalytic, the reduced Au-oxide should be able to be efficiently re-oxidized. It has been suggested that Au-oxide is the catalytically active species of Au-nanocatalysts. To study whether the oxidation/reduction cycle of Au-oxide can take place, the partially reduced Au-oxide surfaces were exposed to molecular oxygen (~ 400000 L) at room temperature (Fig. 5.10). In Fig. 5.10 c, we could not observe any restoration of the reduced oxide by molecular oxygen. When the reduced oxide surfaces by CO were exposed to oxygen at a sample temperature of 400 K, the O 1s intensity further decreased in Fig. 5.11. This result is not surprising, considering that the decomposition temperature of Au-oxide is known to be only slightly higher than 400 K [212]. This result implies that the oxidation/reduction cycle of Au nanoparticles on HOPG cannot take place under our experimental condition, which was suggested to take place for other oxide catalysts [190].

O1s Intensity (arb. units)

after CO after heating at 400 K

526

528

530

532

534

536

Binding Energy (eV)

Fig. 5.11 O1s XPS core level spectra: before (black curve) and after (red curve) heating the reduced using CO Au nanostructure

- 112 -

We have previously shown that oxidation of Ag nanoparticles smaller than ~ 10 nm by an atomic oxygen atmosphere can result in the formation of Ag-oxide, which is completely inert towards CO oxidation [212]. Au nanoparticles, in contrast, do not show any deactivation of the particles towards CO oxidation, even though severe oxidation conditions were used for the oxidation of Au. This might be also a reason that Au nanoparticles show different chemical activities under high-pressure conditions compared to other metal nanoparticles such as Ag. Summarizing, we have shown that two different oxygen species can be formed on Au nanoparticles smaller than ~ 10 nm in diameter oxidized by atomic oxygen. The oxygen species at 529 - 530 eV associated with Au (III) readily reacts with CO, whereas the other oxygen species at 531 - 532 eV, which we suggested to be subsurface oxygen, is inert. However, only one oxygen species at 529 - 530 eV which is active for CO to form CO2 was observed for nanostructured Au films, implying that decrease of the particle size opens a formation route of a new oxygen species under the same experimental conditions. We have direct evidence that lattice oxygen of Auoxide can react with CO to form CO2. This result implies that under real catalytic conditions, Au-oxide species of Au nanoparticles can directly participate in catalytic reactions. Together with our previous results on the reactivity of Ag nanoparticles and the high catalytic activity of RuO2 reported in the literature, we suggest that participation of lattice oxygen of oxide species formed on the surface of metal nanoparticles can be also relevant for many other metal catalysts.

- 113 -

6. Au nanoparticles on SiO2/Si
6.1 Preparation of Au nanoparticles on SiO2/Si
As already mentioned before, the catalytic activity of supported Au catalysts for low temperature CO oxidation depends on the particle size, the type of support, etc. In system for catalysis with Au, many oxides except for Al2O3, SiO2 and activated carbon can be used as a support to induce catalytic activity at temperature below 300 K [89, 141, 142]. Since Al2O3, SiO2, and WO3 are used as a support, gold exhibits poor activity. And also, it has been accepted that semi-conductive metal oxide supports such as TiO2, Fe2O3, and NiO provide more stable Au catalysts than insulating metal oxides [251]. In the present work, Au nanoparticles with different sizes on SiO2/Si with a native oxide layer were oxidized using atomic oxygen/excited molecular oxygen environments to investigate size-dependent chemical activity [216]. In addition, measurements of the oxidation pattern of Au films with flat and rough Au surfaces on silica upon atomic oxygen exposures were carried out [214, 218]. In order to prepare Au nanoparticles with different sizes on a silica surface, different amounts of Au were deposited at a sample temperature of room temperature [190]. Fig. 6.1 shows the XPS spectra of the Au 4f states collected from Au nanoparticles on SiO2/Si [190]. For the spectra a) to g), the Au 4f peak intensity with respect to the Si feature at 89.7 eV decreases, i.e. Au coverage becomes smaller from a) to g). Positive shifts in the Au 4f core level can also be observed with decreasing Au coverage: the Au 4f peaks of the sample with the lowest Au coverage (below 10 % of a monolayer) show a positive core level shift of 0.8 eV with respect to the bulk values.

- 114 -

Au 4f7/2 Binding Energy (eV)

Intensity (arb. units)

Si
g) f) e) d) c) b) a)
92 90

Au4f

a)

85,0 84,8
H: ~ 0.7 nm W < 1.5 nm

b)

84,6 84,4 84,2 H: ~ 4 nm

W < 10 nm

/1.8 /5

5/2
88 86

7/2
84 82

/5 /20
80

84,0 83,8

H: ~ 1 nm W < 3 nm

bulk Au

a)

b)

c)

d)

e)

f)

g)

Binding Energy (eV)

Au coverage decreasing

Fig. 6.1 a) XPS spectra of Au particles grown on SiO2/Si, b) binding energy (Au4f7/2) shifts [190]; g) 10 % of a monolayer

Positive core level shifts with decreasing particle size have been observed for different systems consisting of metal nanoparticles on support materials with relatively low conductivities, such as Au and Ag nanoparticles on oxides surfaces [227, 230 233, 268]. However, much less core level shifts were found for similar Au coverage compared to the previous results like Au/SiO2/Mo(110). This result can be understood by taking different thickness of the silica layers (different charge transfer between support and Au nanoparticles) into account. More systematic analysis for the XPS core level shifts of Au on Silica was done in the thesis of Ignacio Lopez [190]. The gradual shifts of the Au 4f states to the higher binding energy region with decreasing Au coverage confirm that the particle size decreases as the Au coverage decreases from a) to g) in Fig. 6.1 b). The particle sizes of three samples (samples a, b and e) estimated by scanning tunneling microscopy (STM) studies are also given in Figs. 6.2 - 6.4, confirming decreasing particle size with decreasing Au coverage and that Au nanoparticles show three dimensional growth with increasing Au coverage. Furthermore, one can observe that the high stability of Au nanoparticles on silica surfaces due to the high defect density of the support enables preparation of narrow particle size distributions without significant particle sintering.

- 115 -

a)
32 28 24 20

b)

Count

16 12 8 4 0 0 2 4 6 8 10 12

Particle Diameter (nm)

25 20

c)

Count

15 10 5 0 0 2

Particle Height (nm)

Fig. 6.2 a) STM image of Au nanoparticles on SiO2/Si (corresponding to the XPS spectra a) in Fig. 6.1), b) Particle diameter distribution c) Particle height distribution; Tunneling parameters (193 nm x 193 nm, 2 V, 0.1 nA) [190, 216]

- 116 -

a)
40 35 30

b)

Count

25 20 15 10 5 0 0 2 4 6 8 10

Particle Diameter (nm)

25 20

c)

Count

15 10 5 0 0 1 2 3 4 5 6 7 8 9

Particle Height (nm)

Fig. 6.3 a) STM image of Au nanoparticles on SiO2/Si (corresponding to the XPS spectra b) in Fig. 6.1), b) Particle diameter distribution c) Particle height distribution; Tunneling parameters (124 nm x 124 nm, 4.7 V, 0.1 nA) [190, 216]

- 117 -

a)
40 35 30 25

b)

Count

20 15 10 5 0 0 1 2 3 4 5 6 7 8 9 10

Particle Diameter (nm)

25 20 15 10 5 0 0 1 2

c)

Count

Particle Height (nm)

Fig. 6.4 a) STM image of Au nanoparticles on SiO2/Si (corresponding to the XPS spectra e) in Fig. 6.1), b) Particle diameter distribution c) Particle height distribution; Tunneling parameters (123 nm x 123 nm, 4.7 V, 0.1 nA) [190, 216]

- 118 -

6.2 Oxidation properties of Au nanoparticles on SiO2/Si

All of the Au nanostructures prepared in Fig 6.1 were exposed to an atomic oxygen atmosphere in order to investigate the origin of the oxygen species formed on Au nanostructures. For most of the samples in Fig. 6.5, additional peaks at the higher binding energy region (yellow bar) with respect to the metallic Au 4f state (Au (0)) appear in the XPS spectra upon treatment of most of the samples examined with atomic oxygen. This result can be attributed to the formation of Au2O3 (Au (III)) [44]. When the particle size decreases from 4 nm to ~ 0.7 nm in height (samples a) - d) in Fig. 6.5), the relative intensity of the Au2O3 peak compared to the respective metallic Au peak increases, which can be understood by increase of the surface/volume ratio with decreasing particle size, i.e. surface layers of nanoparticles are oxidized, forming core (metallic) / shell (oxidic)-type particles. In contrast, when the mean height of the particles is lower than about ~ 0.7 nm (samples e) - g) in Fig. 6.5), the Au2O3 peak becomes less pronounced with decreasing particle size. The sample with the smallest mean particle size does not show Au2O3 peaks in the XPS spectrum, whereas the Au 4f peaks are shifted to the lower binding energy by approximately ~ 0.3 eV after reaction under atomic oxygen conditions (Fig. 6.6 a). It should be noted that the accuracy of the determination of the binding energies is better than 0.1 eV. All of the Au nanoparticles studied here show changes upon exposure to atomic oxygen, which implies that none of the Au nanoparticles studied here are completely inert towards oxygen uptake, which has been previously suggested to be the case for an Au cluster consisting of 55 atoms [44]. We cannot synthesize a mono-disperse surface consisting of only the magic Au55 clusters by using our particle preparation method. In the work [44] of Boyen et al. mentioned above, all Au particles except Au55 were shown to be able to form Au2O3, which disagrees with our data.

- 119 -

The difference in the results can most likely be rationalized by taking into account the different oxidation conditions used by Boyen et al., whereas we have used atomic oxygen created by a hot Pt filament, which may provide a milder oxidizing environment than the oxygen plasma.

g) f) e)
Intensity (arb. units)

Au4f7/2

Si

g) f)
H: ~ 0.7 nm W < 1.5 nm Oxidation Atomic O

e) d) c)
Au(III) Au(o)

d) c) b) a)
7/2
80 82 84 86

H: ~ 1 nm W < 3 nm
H : ~ 4 nm W < 10 nm
90 92

b) a)
80

5/2
88

Binding Energy (eV)

Binding Energy (eV)

82

84

86

88

90

92

Fig. 6.5 Left panel; XPS spectra of the Au 4f states for the Au particles grown on SiO2/Si are displayed [216]. From a) to g), the Au coverage is reduced, i.e. the average particle size decreases from a) to g). For a), b), and e), the average particle heights (H) and widths (W) estimated on the basis of the STM data in Figs. 6.2 - 6.4 are approximately given. Right panel; XPS spectra of the Au 4f states of the samples in the left panel after exposing them to the atomic oxygen atmosphere. The shoulders at the higher binding energy region correspond to the formation of Au2O3. (marked by the yellow bar)

- 120 -

It is worth emphasizing that the sample with the smallest mean Au nanoparticles size in Fig. 6.5 g) shows no Au2O3 peak, yet these particles are not inert to the reaction with atomic oxygen. Our XPS results show that the AuO interaction becomes completely different as the particle size becomes smaller than ~ 0.7 nm in height. Previously, the non-appearance of the Au2O3 peak in XPS upon oxidation has been attributed to the inertness towards oxygen uptake, yet according to our results, conclusions about the inertness towards uptake of oxygen based on the XPS data should be scrutinized, since this can only indicate formation of different types of Au-O bonds [44]. In chapter 5, we have shown that Au nanoparticles on HOPG as large as 3 ~ 10 nm in diameter form two different oxygen species upon exposing to atomic oxygen environments, which have been attributed to the Au2O3, and subsurface oxygen species [214]. Subsurface oxygen species were identified by the appearance of the O 1s state at about 531 eV and a low reactivity towards CO oxidation at room temperature. Formation of the subsurface oxygen species does not cause strong positive shifts of Au 4f states, which is the case for the Au-oxide formation. For the Au nanoparticles larger than ~ 20 nm in diameter, only Au2O3 formation can be found without subsurface species. To shed light onto the oxidation behaviors of the Au nanoparticles on HOPG even smaller than those studied in the previous chapter 5, Au nanoparticles were prepared on a heavily sputtered HOPG surface (Fig. 6.6 c). Due to very weak metal-support interactions, which can cause tip-induced movements of the nanoparticles on the surface, no reasonable STM images could be obtained for the Au nanoparticles smaller than 3 nm in diameter on HOPG. Though, XPS data clearly show that the particle size of the sample in Fig. 6.6.c should be much smaller than 3 nm in diameter: when the particle size is larger than 3 nm, shifts of Au 4f level with respect to the bulk value are at most 0.1 eV (Fig. 5.1), whereas a Au 4f core level shift of 0.3 eV could be observed for the sample in Fig. 6.6 c).

- 121 -

Intensity (arb. units)

as prepared sample g) after atomic oxygen

a)

0.3 eV

Au4f7/2

80

82

Binding Energy (eV)

84

86

88

90

92

Intensity (arb. units)

as prepared sample c) after atomic oxygen

b)
5/2

Au4f
7/2

Au(III) Au(o)

80

82

Binding Energy (eV)

0.3 eV

84

86

88

90

92

Intensity (arb. units)

as prepared: Au on HOPG after atomic oxygen

c)

Au4f

80

82

Binding Energy (eV)

84

86

88

90

92

Fig. 6.6 a), b): XPS Au 4f spectra (samples g) and c) in Fig 6.1), c): XPS Au4f spectra of the Au nanoparticles on HOPG (<< 3 nm in diameter) were taken before (black) and after (red) atomic oxygen exposure [216]

- 122 -

When this sample was treated with atomic oxygen conditions, no Au2O3 shoulder could be observed at the higher binding energy region, whereas the Au 4f peaks only shifted to the lower binding energies, similar to the case of the small Au nanoparticles on SiO2/Si in Fig. 6.6 a).

Intensity (arb. units)

O1s

Bare HOPG exposed to O Au/HOPG exposed to O Au/HOPG exposed to O + CO (3000L)

526

528

Binding Energy (eV)

530

532

534

536

538

Fig. 6.7 XPS O 1s spectra of the Au nanoparticles on HOPG exposed to atomic oxygen and subsequently to CO [216]

When the Au nanoparticles were supported on Si wafers with native oxide layers, the O 1s signal from the substrates (SiO2) is much larger than that of the oxygen atoms bound to the Au nanoparticles, thus preventing a clear discrimination of the O/Au signals by XPS. In contrast, this problem does not arise with HOPG substrates because of their low oxygen uptake. Fig. 6.7 shows the O1s spectra of the Au nanoparticles (<< 3 nm) on HOPG exposed to atomic oxygen, and subsequently to CO. For comparison, the O 1s spectrum of a bare sputtered HOPG surface exposed to the similar amount of atomic oxygen is shown. When these Au nanoparticles on HOPG were exposed to atomic oxygen atmosphere, an additional state at about 531 eV could be observed, which can be attributed to the oxygen species bound to Au nanoparicles.

- 123 -

6.3 Reactivity toward CO of the oxidized Au nanoparticles

When the Au nanoparticles showing the Au2O3 shoulders after oxidation were exposed to 3000 L of CO at room temperature, the reduction of Au2O3 could be observed (Fig. 6.8 b). The Au (III) species are not completely removed by 3000 L of CO, most likely because of the fact that not only the topmost Au layers are oxidized, but deeper layers are also oxidized, which cannot be reduced by CO efficiently. No change could be observed in the Au 4f level (Fig. 6.8 a) when the small Au nanoparticles on SiO2/Si that showed only a negative shift in the Au 4f state without appearance of the Au2O3 peak upon atomic oxygen treatment were exposed to CO, thus demonstrating that different oxygen species with various characteristics in XPS show dissimilar chemical properties. However, for the Au nanoparticles (<< 3 nm) on HOPG, the O 1s signal decreases in intensity by about 30 % upon exposing to 3000 L of CO (Fig. 6.7). Since the Au-oxide species formed on Au particles larger than 20 nm (Fig. 5.7 c) can be nearly completely removed by 1000 L of CO at room temperature (Au2O3 species react with CO to CO2 with a mean CO/CO2 conversion probability of about 0.1 %), the atomic oxygen species on smaller Au nanoparticles are much less reactive for the CO oxidation at room temperature [213, 214]. Based on the data in Fig. 6.7 c), the average CO/CO2 conversion probability for the oxygen species bound to very small Au particles is estimated to be at most 0.01 %. This result also confirms that the negative shifts in the Au 4f level of smaller Au particles upon atomic oxygen treatment are not simply due to the structural change of Au nanoparticles (e.g. sintering) without formation of an Au-O bond, but from the adsorption/absorption of oxygen on Au nanoparticles. It is worth mentioning that the formation of oxygen species that are inactive for the oxidation of CO has been suggested for Au/TiO2 systems, which implies that the unique behavior of the small Au nanoparticles seems to be independent of the substrate [111].

- 124 -

Intensity (arb. units)

after atomic oxygen after CO 3000 L

Sample g)

a)
Si

Au4f7/2

80

82

Binding Energy (eV)

84

86

88

90

92

Intensity (arb. units)

after atomic oxygen after CO Au4f Sample c)

7/2

b)
5/2

Au(III)

Au(o)

80

82

Binding Energy (eV)

84

86

88

90

92

Intensity (arb. units)

after atomic oxygen after CO

Au on HOPG

c)

80

82

Binding Energy (eV)

84

86

88

90

92

Fig. 6.8 a), b): XPS Au4f spectra (samples g) and c) in Fig 6.1) and c): XPS Au4f spectra of the Au nanoparticles on HOPG (<< 3 nm in diameter) exposed to an atomic oxygen atmosphere (red) and further exposed to 3000 L of CO (blue). No change can be seen after exposure to CO for a) and c) [216]

- 125 -

It has been previously suggested that the activation of oxygen molecules on Au nanoparticles is the key to the increased catalytic activities of Au: once oxygen molecules are dissociated on Au nanopartocles, they can readily react with CO to form CO2 at below 80 K [111]. In the present work, in contrast, we found oxygen species formed by deposition of atomic oxygen on Au nanoparticles that hardly react with CO at room temperature. As already mentioned, the mean CO/CO2 conversion probability of this oxygen species is at most 0.01 %, which is at least one order of magnitude smaller than those of Au oxide or oxygen atoms chemisorbed on top of bulk-like Au surfaces. This less reactive oxygen species for the oxygen of CO is tentatively assigned to either the subsurface oxygen or surface oxygen species, which are not active for CO oxidation (either they occupy the same adsorption sites as CO, thus preventing CO chemisorption on Au nanoparticles, or the Au-O interaction is too strong). Ag nanoparticles smaller than ~ 4 nm in diameter show much different oxidation behaviors than the larger Ag nanoparticles in chapter 4, and it is likely that such different oxidation behaviors of small nanoparticles compared to the larger ones can be relevant for many other transition metals. Based on the that fact the oxidation pattern of very small (< 0.7 nm in particle height which roughly corresponds to 2 - 3 atomic layers) and lager Au nanoparticles are dissimilar, One can provide the following new insights into the mechanisms of Au nanocatalysis; 1) It has been suggested that dissociatively chemisorbed oxygen can readily react with CO to CO2 on small Au nanoparticles; however, we demonstrated that catalytically inactive atomically bound oxygen species can exist on small Au nanoparticles. For obtaining a complete understanding of the Au-nanocatalysis, this inactive O species should be taken into account. 2) Au nanoparticles smaller than 2 - 3 atomic layers are generally suggested to be catalytically active [43]. Since those small Au particles are not able to form Au (III), the active Au nanoparticles should be not in the oxidic (III), but metallic (0) state under real catalytic conditions [217].

- 126 -

6.4 Rough and Flat Au surface: Preparation &Reactivity toward O and CO

As mentioned before, to shed light on the origin of the dissimilar chemical properties of differently sized Au nanoparticles, relatively flat and very defective Au films grown on Si with native oxide layers were characterized using XPS and STM, then exposed to molecular/atomic oxygen and CO.

Preparation of rough and flat Au surface

Au/Si systems have been widely studied in the past, and it is known that Au can diffuse into the Si bulk [269 - 271]. In some previous studies, a strong Au-Si bonding has been found [272 - 274]. Moreover, the strong Au-Si bonding can induce Au-silicide formation. Here, we show that annealing causes formation of defects on Au thin films, which is most likely driven by strong SiAu interactions. This result is in line with those from recent studies of Palmer et al, in which formation of cracks in the micrometer-scale was observed using Scanning Electron Microscopy (SEM) [275]. In ref. [275], Cr wetting layers were used, whereas Au was directly deposited on a Si wafer in the present work. Defect (crack) formation of Au films upon annealing was confirmed using STM with a higher resolution. In additions, XPS could shed light on surface structures of Au films after heat treatments. Inter-diffusion of Au into Si, segregation of Si onto Au and/or decomposition of Au films are generally expected to result in the increase of the Si signal and decrease of the Au signal upon annealing; however, we found an increase of the Au signals upon annealing at lower temperature range (< 200 C), suggesting an increased surface area of Au upon annealing.

- 127 -

In Fig. 6.9, XPS survey spectra reveal that Au films are free from contaminations except C and sometimes Sn. Taking different cross sections of various elements in the photoionisation process into account, concentrations of C and Sn in the Au films are estimated to be < 4 % and < 1% respectively. Sn disappears after annealing at 350 C. One can suggest that Sn does not have influence on our key results shown in the present work, since samples without Sn impurities show essentially the same behaviors upon annealing at elevated temperatures.

Au4s

Au4p1/2Au4p3/2
O

Au4d3/2 Au4d5/2

Au4f5/2 Au4f7/2 Si Si

Intensity (arb. units)

Heating 500 C

Si

Heating 350 C

Heating 200 C Sn3d3/2 Sn3d5/2 Au/SiO2/Si C1s

800

600

400

200

Binding Energy (eV)

Fig. 6.9 XPS survey spectra of Au thin films before and after annealing at various temperatures. Annealing time was 30 minutes for each experiment [211]

- 128 -

A u4f Intensity (arb. units)

500 o C 350 o C

a)

200 o C

RT
80 82

7/2
84 86

5/2
88 90

Binding Energy (eV )

O 1s

b)

Intensity (arb. units)

500 C 350 C 200 C RT

526
o o

B in d in g E n e rg y (e V )
S i2 p

528

530

532

534

536

538

c)

Intensity (arb. units)

500 C 350 C 200 C RT

96
o o

B in d in g E n e rg y (e V )

98

100

102

104

106

108

Fig. 6.10 XPS spectra of Au thin films before and after annealing at various temperatures (narrow scan). a) Au 4f, b) O 1s, c) Si 2p levels. Spectra are displayed after background subtraction [211]

- 129 -

Fig. 6.10 shows the XPS spectra (narrow scan) of an Au film grown on a Si wafer before and after annealing at various temperatures. Before annealing, two sharp peaks at 84.0 eV and 87.7 eV corresponding to Au 4f7/2 and Au 4f5/2 can be discriminated with a Full-Width-Half Maxima of 1.1 eV and 1.2 eV, respectively (Fig. 6.10 a), which are identical to those of pure Au samples. No Si and O signal can be detected (Fig. 6.10 b and c) in this case, implying that the Au film thickness was much larger than the mean free path of the photoelectrons emitted from the Si 2p level, i.e. the Au film thickness was estimated to be larger than 5 nm. After annealing, significant changes in the XPS spectrum can be observed. The results of the XPS spectra were summarized in Table 6.1. After heating to 200 C, the Au 4f peaks undergo broadening, accompanying shifts of the peak positions to higher binding energies by about 0.2 eV. The absolute intensities of the Au peaks increased significantly after annealing at 200 C. In this experiment, the sample position was not changed during annealing, and in this case, in general, we have a good reproducibility of XPS spectra in terms of absolute intensity. Shifts of the Au 4f peaks to higher binding energies can be interpreted in various ways. One may argue that Au became Au oxide, or other compounds, in which Au is (partially) positively charged, which can rationalize the shift to higher binding energies in terms of an initial state effect. However, it is unlikely that annealing under UHV conditions can lead to the oxidation of Au. Also, there are no O and Si signals detected after annealing at 200 C in Fig. 6.10 b and c, even though Au 4f peaks have shifted to higher binding energy significantly under the same conditions. Another way to explain these results is that the Au film became much more defective after annealing. On defect sites such as steps and edges, under-coordination numbers of Au atoms are higher with respect to those of the terrace atoms. According to final state effects, the positive holes created by the photoemission process can be less efficiently screened (less efficient extra-atomic relaxation), when the number of directly neighboring atoms is reduced, leading to the core level shifts to higher binding energies.

- 130 -

Also, in terms of initial state effect, under-coordinated atoms may have different electronic structures, leading to the experimentally found positive core level shift. Similar positive core level shifts with increasing number of defect (step, edge and corner) sites (or decreasing particle size) have been found previously [224, 226, 276 - 279]. It is worth emphasizing that no Si and O signal can be found upon annealing at 200 C, and therefore the role of the Sisupport on the Au 4f core level shifts can be ruled out, i.e. the Au 4f level shifts should be interpreted exclusively in terms of morphological change of the Au surface. Our observation that the Au 4f signal increased in intensity by about a factor of 1.4 after annealing at 200 C also indicates increased surface area and roughness after annealing.

RT Au4f7/2 (Binding Energy eV) Au4f7/2 (Relative Intensity) Au4f7/2 O 1s (Relative Intensity) Si 2p (Binding Energy eV) Si 2p (Relative Intensity) (Full-WidthHalf-Maximum eV) 84.0 1.0 1.05

200oC 84.20 1.4 1.06

350oC 84.45 0.8 1.29 1.0 99.74 1.0

500oC 84.54 0.6 1.34 1.7 99.94 1.8

x x x

x x x

Table 6.1 Analysis of the XPS results of Fig. 6.10. A Gaussian function was used for fitting each spectrum.

- 131 -

After annealing the Au films at above 350 C, signals from the underlying Si substrate can be observed in the XPS spectrum (Fig. 6.10 b and c), indicative of segregation of Si onto the surface, or diffusion of Au into Si bulk. At the same time, the Au 4f states are further shifted to higher binding energies. In this case, the positive Au 4f level shifts can be rationalized by considering both increased under-coordinated Au atoms and effects of the SiO2/Si substrate. The final state of the Au 4f level of Au/SiO2 in the photoemission process can be less efficiently screened than in a pure Au sample, due to the lower conductivity of SiO2, contributing to the positive core level shift. Formation of Au-silicide may also come into play. After annealing at 500 C, Si and O signals further increase, whereas the Au signal is reduced, indicative of further inter-diffusion of Au and Si at the interface. Note that no evaporation of Au is expected at 500 C. Fig. 6.10 shows a small positive shift of the main Si 2p peak upon annealing at 500 C compared to the case after annealing at 350 C, which may result from an increased number of Au-Si bonding (or Au-silicide formation) upon annealing at a higher temperature. In the Si 2p signal, existence of SiO2 gives a peak at about 103 eV, and the thickness of the silica layer is estimated to be about 1 nm, i.e. 1 nm oxide layers are thin enough to allow diffusion of Au into Si [280].

- 132 -

a)

b)

c)
Fig. 6.11 STM pictures of Au thin films before (a: 0.2 V, 0.6 nA, 238 nm x 238 nm), and after annealing at 200 C (b: 2V, 1.7 nA, 297 nm x 297 nm) and 500 C (c: 4.2 V, 0.5 nA, 250 nm x 250 nm). For a) and b) the same height contrast (0 ~ 12 nm) was used for display, whereas for c) 0 ~ 19 nm was used [211]

- 133 -

To study the influence of annealing on the surface structure of Au on Si, STM investigations were carried out (Fig. 6.11). Before annealing (Fig. 6.11 a), the surface consists of grains as large as 20 - 100 nm with relatively smooth surface structures. After annealing at 200 C (Fig. 6.11 b), STM image shows much enhanced surface roughness, as the XPS data suggested in Fig. 6.10. It is worth emphasizing that only Au could be detected for this surface without Si and O in XPS. After annealing at 500 C (Fig. 6.11 c), very rough surface structures can be observed using STM. As mentioned above [272 - 274], the inter-diffusion of Au into the Si substrate and/or Au-silicide formation at the Au/Si interface may be responsible for the formation of defects upon annealing the Au films. Another possible explanation of the defect formation upon annealing is a different thermal expansion coefficient of Au from those of Si or SiO2. It is important to mention that the cracking of an Au film upon annealing was observed in sub-micrometer scale using Scanning Electron Microscope (SEM) and Atomic Force Microscopy (AFM) previously; however, the defect formation in the nanometer scale has not been found previously [275]. After the ex-situ STM measurements of the sample annealed at 500 C, XPS spectra were collected again to study the influence of the surface structure by exposing it to air. As it is shown in Fig. 6.12, the thickness of SiO2 layers was significantly increased after exposing to air, which is evident by the increase of the Si 2p state at about 103 eV and increase of the O 1s peak. C 1s peak also increases in intensity, accompanying a positive shift of the C 1s state. The C 1s state after exposing to air agree very well with that of a graphite sample (peak position and also the tale at higher energies, which can be attributed to the plasmon peak), indicating that graphite layers are formed on the sample under air, which most likely comes from decomposition of hydrocarbons on the surface. The shifts of Si, O and Au peaks upon exposing to air can also be explained by additional graphite layers, since additional graphite layers on the surface can more efficiently screen the final state of the photoemission in Au/SiO2/Si, causing the negative shifts of the peaks.

- 134 -

Intensity (arb. units)

after STM measurements

before

Intensity (arb. units)

C 1s

a)

O 1s

b)

280

282

284

286

288

290

292

294

526

528

530

532

534

536

538

Binding Energy (eV)

Si 2p

Binding Energy (eV)

c)
Intensity (arb. units)

Au 4f

d)

Intesnsity (arb. units)

96

98

100

102

104

106

108

80

82

84

86

88

90

92

Binding Energy (eV)

Binding Energy (eV)

Fig. 6.12 XPS spectra of Au thin films on Si heated at 500 C before (black) and after ex-situ STM experiments (red). a) C 1s, b) O 1s, c) Si 2p, and d) Au 4f. Spectra are displayed after background subtraction [211]

Similar effects were also observed in our recent studies on tungsten oxide layers deposited on Ag nanoparticles on HOPG [213]. It is also important to point out that the Au/Si ratio did not change significantly upon exposing to air, and therefore, the Au surface structure did no change significantly. Note that roughening or flattening of the Au surface should yield a significant change of the Au/Si intensity ratio, which was not found here.

- 135 -

Oxidation property of rough and flat Au surface

We have shown that the number of under-coordinated Au atoms is significantly increased upon annealing. This kind of Au surface has never been thoroughly characterized using spectroscopic tools before. As mentioned before, chemical reactivity of the under-coordinated Au atoms is an important issue due to its relevance in Au catalysis [43, 47, 53, 79, 112, 281]. Various theoretical studies have suggested dissociative and non-dissociative chemisorption of O2 on under-coordinated Au atoms of a metallic Au surface. Previously, O2 chemisorption of a relatively smooth Au surface has been studied; however, O2 chemisorption on extremely rough surfaces like the one prepared in the present work has not been investigated [53, 281]. To shed some light on the role of the under-coordinated Au atoms on defect sites of metallic Au surfaces for O2 chemisorption, an Au film annealed at 200oC was exposed to 400000 L of O2 at room temperature. Before exposing to O2, an O 1s peak at 533 eV corresponding to SiO2 of the substrate was discriminated (Fig. 6.13 b). The Au 4f states are shifted to higher binding energy by about 0.3 eV compared to those before annealing. In Fig. 6.10, no O 1s signal could be found upon annealing at 200 C, whereas here a small amount of SiO2 can be observed, most likely due to a different initial Au film thickness before annealing. Our quantitative analysis of the O 1s and Au 4f signals suggests that at least ~ 97 % of the surface should be covered by Au (the absolute intensity of the O 1s peak in Fig. 6.12 is only about 1 % of that of the Au 4f peak. Si 2p intensity is even smaller), and therefore, the positive core level shift after annealing at 200 C can only be explained by increased number of defect sites upon annealing, whereas SiO2/Si substrate plays a minor role for the positive Au 4f level shift.

- 136 -

Au4f

a)

Intensity (arb. units)

200 C-annealed after O2 before O2 before annealing

80 82 84 86 88 90 92 94

Binding Energy (eV)

O1s

b)

Intensity (arb. units)

200 C-annealed after O2 before O2 before annealing

526 528 530 532 534 536 538 540

Binding Energy (eV)

Si2p

c)
200 C-annealed after O2 before O2
o

Intensity (arb. units)

96

98

100

102

104

106

108

110

Binding Energy (eV)

Fig. 6.13 XPS spectra of the defective Au thin films before and after exposing to 400000 L of O2 at room temperature; a) Au 4f, b) O 1s, and c) Si 2p. Spectra are displayed after background subtraction [211]

- 137 -

Upon O2 exposure, no changes can be observed either in the O 1s or in the Au 4f states, indicating that O2 chemisorption does not take place under these conditions (Fig. 6.13). Since the presence of the under-coordinated Au atoms at defect sites on this Au film was clearly discriminated using XPS, chemisorption of O2 on these defect sites should be able to be identified using XPS. From these results, we estimate the sticking probability of O2 on defect sites of Au to be about less than 1x10-6. It is worth mentioning that Si was not further oxidized upon this large amount of oxygen exposure, since most likely the Si wafer was covered by SiO2 layers which prevent Si from oxygen adsorption. As mentioned before, O2 chemisorption over under-cooridinated Au atoms has been in debate so far. Previous Density Functional Theory (DFT)calculations from Hu et al. predicted that O2 molecularly chemisorbs on the step sites of Au surfaces, whereas Norskov et al. suggested that O2 should efficiently dissociate on undercoordinaterd Au atoms at room temperature [79, 112]. Based on our experimental data, the theoretical prediction of Norskov et al. could not be confirmed. To shed light on the origin of dissimilar chemical properties differently sized Au nanoparticles, these relatively flat (before annealing) and rough Au surface structures annealed at 200oC were exposed to atomic oxygen environment at room temperature (Fig. 6.14). When a flat Au surface was exposed to atomic oxygen, appearance of the shoulders at higher binding energies can be found in the Au 4f levels (Fig. 6.14 a). In contrast, no oxidation could be observed in the Au 4f level when the rough Au surface was exposed to the same amount of atomic oxygen (Fig. 6.14 b). The rough surface shows a similar oxidation pattern to those of very small Au nanoparticles. Note that for very small Au nanoparticles supported by oxide surfaces, the O 1s signals of the supporting materials overwhelm, making an identification of Au-O species in the O 1s spectra difficult. When a flat Au surface is oxidized, the O 1s state at 529 eV becomes visible (Fig. 6.14 c), which can be assigned to the chemisorbed oxygen or/and Au2O3 [214]. In addition to the O1s peak at 529 eV attributed to the Au2O3 formation, an additional state at 533 eV can be found.

- 138 -

Intensity (arb. units)

Intensity (arb. units)

a)

Au4f

after O exposure

c)

O1s

Flat Au surface 526 528 530 532 534 536

Flat Au surface 82

80

Binding Energy (eV)

Au4f

84

86

88

90

92

Binding Energy (eV)

Intensity (arb. units)

after O exposure

Intensity (arb. units)

b)

d)

O1s

Rough Au surface

80

82

Binding Energy (eV)

84

Rough Au surface 86 88 90 92

526

528

Binding Energy (eV)

530

532

534

536

Fig. 6.14 XPS Au 4f (a and b) and O 1s (c and d) spectra of the as grown flat Au thin film on SiO2/Si and the rough Au surface annealed at 200oC on SiO2/Si; before (black) and after (red) exposing to atomic oxygen at room temperature [218]

When the extremely rough Au surface is oxidized (Fig. 6.14 d), the 533 eV peak in the O 1s state becomes much more pronounced than the state at 529 eV, i.e. oxidation of Au is much less pronounced for rough Au surfaces, in line with the Au 4f level data. The structures of the oxygen species, which is responsible for the O 1s state at 533 eV, is still unclear, even though similar O 1s states have been generally attributed to the subsurface oxygen species [214, 235]. If this assignment is right, one may argue that the defect sites on rough surfaces can facilitate dissolution of oxygen atoms in the bulk, rather forming surface oxide.

- 139 -

Reactivity toward CO of the oxidized rough and flat Au surface

When these Au structures reacted with atomic oxygen are exposed to CO, only the O 1s state at 529 eV attributed to the Au2O3 species, can be titrated by CO, whereas the O 533 eV peak does not change upon CO exposure. This result implies that the O 1s peaks at 533 eV can either be attributed to subsurface oxygen, since previous studies have shown that all on-surfaces oxygen on Au surfaces can efficiently be removed by CO at room temperature, or surface oxygen species, which is not reactive due to the occupation of the same adsorption site as CO, or a too strong Au-O bond [216, 267].

Intensity (arb. units)

atomic oxygen

Intensity (arb. units)

a)

O1s

Flat Au surface

b)

Rough Au surface

CO 3000L atomic oxygen 526 528 530 532 534 536 538

CO 3000 L 526 528

Binding Energy (eV)

530

532

534

536

Binding Energy (eV)

Fig. 6.15 a) Relatively flat and b) rough Au surface structures on SiO2/Si exposed to the atomic oxygen atmosphere are reduced by CO. The O 1s spectra are taken before and after CO exposures [218]

The rough Au thin film and very small Au nanoparticles on HOPG and silica have some common behaviors regarding the reaction with atomic oxygen, namely that the oxygen species are inert (or less reactive) towards CO oxidation. This result confirms that the increase of the number of under-coordinated atoms of smaller nanoparticles should be responsible for different oxidation patterns.

- 140 -

7. Mass-selected Au clusters on HOPG

A large number of experimental and theoretical studies including our previous works have been devoted for better understanding of the interaction of metal clusters-supports and the size effects of the as-deposited, isolated clusters themselves in heterogeneous catalysis. As mention before, oxide-supported small gold nanoparticles show the highest reactivity at sizes of about 2 to 3 nm, whereas unsupported gold nanoparticles are weakly active at best [102, 328]. The choice of support, the method of preparation and the pretreatment before use are very important factors that control the reactivity [26, 88, 89]. In previous chapters (chapter 4, 5 and 6), we also have shown the size-selectivity for CO oxidation of Ag and Au nanoparticles deposited on HOPG and SiO2/Si, which is due to different oxide species formed on nanoparticles. Based on these results, we have claimed that the unique behavior of the very small Au nanoparticles seems to be independent of the substrate [216, 217]. In order to gain better understanding for the metal-support interaction, we use mass-selected Au clusters deposited on HOPG (chapter 7) and SiO2/Si (chapter 8). Deposition of mass-selected clusters opens a new insight into the size-selectivity of chemical properties of clusters consisting of less than ~ 20 atoms and metal-support interactions. For example, Au clusters consisting of 8 atoms on MgO were shown to be able to catalyze CO oxidation efficiently [70, 81]. In combination with theoretical calculations, it was suggested that charge transfers from the defect sites of the MgO surface to the Au clusters play an important role for the enhanced catalytic activities [47, 48, 67]. On TiO2, Au7 clusters were found to exhibit a higher catalytic activity towards CO oxidation than other clusters [72].

- 141 -

In fact, Haruta suggests that Au clusters (smaller than ~ 2 nm) have some different chemical properties than that of Au nanoparticles (larger than 2 nm) [309].

Au clusters
Non-metallic Quantum Size Effect 3D Structure No Surface Plasmon Absorption

Au nanoparticles
Metallic Edge and Corners Contact Interfaces Surface Plasmon Absorption

Table 7.1 The difference between Au clusters and Au nanoparticles suggested by Haruta [309]

For Au nanopaticles larger than 2 nm in particle size, metallic properties, edges and corner and interfaces are the active sites, whereas, for decreasing particles size (Au clusters), non-metallic properties and quantum size effects are the key points for the high catalytic activity toward CO oxidation. It has been shown that catalytic activity can be also changed depending on the geometric structure of Au13 clusters [310]. Au13 clusters supported on Mg(OH)2, being a structure of isomers having Icosahedral and Cubo-Octahedral structures with a ratio of 58 : 42 show a much higher CO oxidation activity than with a ratio of 7 : 93. In our work (chapters 7 - 9), we also found direct evidence of the difference between Au clusters and Au nanoparticles. In this chapter, oxidation and reduction of Au clusters deposited on sputtered HOPG were studied. Size selectivity for the oxidation of Au clusters was observed on sputtered HOPG. Au8 was only found to be active towards the oxidation and reduction of CO.

- 142 -

7.1 Preparation of Mass-selected Au clusters on HOPG

For the creation of mass-selected Au clusters, a magnetron-sputter source was used. A magnetic field mass spectrometer was used for the mass separation of the clusters. Mass-selected Aun (n = 2 - 9) clusters were soft-landed on the sputtered HOPG surface, that is, a kinetic energy of 0.2 eV per atom was used. 1x 1013 clusters were deposited on the surface, estimated by measuring the sample current during deposition. (see chapter 3)

Au3

Intensity (arb. units)

Au4 Au2 Au1 Au5 Au7 Au 8 Au9 Au6 Au10 Au11 Au12 Au

13

Number of Au atoms

10

12

14

Fig. 7.1 Mass-spectrum of Au cluster anions created by the magnetron-sputtersource

Fig. 7.1 shows a mass spectrum of the Au clusters before deposition. Due to the limited resolution of the mass spectrum, the purity of the clusters can hardly be determined in Fig. 7.1; however, it has been previously shown that Au cluster anions with very high purities can be relatively easily prepared [75]. One of the possible major contaminations for metal clusters is oxygen molecules. Based on the mass spectrum in Fig. 7.1, contamination of Au clusters by oxygen molecules can be ruled out: Among the anionic Au clusters, Au2- is known to be most reactive towards O2 chemisorption, but the O2 chemisorption on Au2- is not detectable in Fig. 7.1 [75, 76].

- 143 -

Since no information is available for the sticking probability of Au clusters on sputtered HOPG surfaces, the Au coverage cannot be precisely determined. Our STM (Figs. 7.2 - 7.4) and XPS in Fig. 7.6 results are suggestive of less than ~ 10 % of a monolayer equivalent of Au coverage for Au7. However, Au clusters are weakly adsorbed on a perfect surface of HOPG and show high lateral diffusion implying spontaneous nucleation and formation of dendritic structures of Au clusters [300 - 302]. It is well known from STM studies of evaporated Au and Ag that metal particles sinter upon reaching the surface to form larger particles [295 - 299]. Generally, in the case of mass-selected clusters, low landing energies (a few eV/atom) or high kinetic energies (~ 100 eV/atom) were used to deposit cluster onto the surface leading to diffusion / aggregation and fragmentation / implantation /substrate damage, respectively [293, 294]. To avoid fragmentation and sintering of deposited Au clusters on HOPG, in this experiment, the HOPG surface was sputtered using an Ar-ion sputter gun to increase the point defect site density before depositing clusters, and then Au clusters were soft-landed (0.2 eV / atom) on these surfaces. Another effective method for preventing diffusion of clusters after deposition is that clusters impinge on the substrate with high kinetic energy [17, 283]. Defects on the surface introduce pinning centers for metal clusters that can promote cluster and film growth, and one monolayer deep nanometer-sized pits have been used to trap Ag and Au clusters on HOPG [222, 284, 285]. In our previous results [190], it was also shown that particle size and growth behaviour were influenced by the defect density. Various defect sites on carbon supports like HOPG and CNTs were observed in previous experimental results such as pentagon-heptagon pairs, mono-vacancies and multivacancies, adatoms, and interstitials [286 290, 323 - 325]. Recently, Ar+ ion-induced defect formations on HOPG surface were imaged with STM [322].

- 144 -

Two distinctive types of defects can exist, as depicted in Fig. 7.2. One type (a) has as an almost circular dome-like structure showing the undisturbed lateral atomic arrangement of the (0001) HOPG surface, while the other (b) exhibits a distorted atomic structure. The much more abundant distorted type of defect like b) appears unaffected by annealing. The origin of this type of defect is associated with a distortion of the lattice atoms due to the formation of vacancies. One can also expect this kind of defects on used HOPG, even though Ar+ ions with much higher kinetic energy were used here.

b) a)

Fig. 7.2 STM image (35 x 35 , It; 0.44 nA and Vt; -9.8 mV) of two types of defects on the HOPG surface; a) dome-like defect and b) a distorted atomic structure defect) [322]. (Low-energetic Ar+ ions; 20 s, 150 eV acceleration voltage)

Moreover, theoretical results reported recently [282] show that the surface vacancy (single carbon atom vacancy) on HOPG can form chemical bonds with an Au atom and Au6 clusters as the three nearby carbons align their dangling bonds toward the gold clusters. However, for a perfect surface, Au6 adsorption is very week.

- 145 -

a)

b)

Fig. 7.3 STM images of Au7 deposited on a) sputter-damaged HOPG (150 nm x 150 nm, 1.61 v, 0.7 nA), b) non-sputtered HOPG. (12.9 nm x 12.9 nm, 1.67 v, 0.3 nA) [221]

- 146 -

a)

b)
Fig. 7.4 STM images of Au7 deposited on non-sputtered HOPG a) 2D STM image b) 3D image (50 nm x 50 nm, 1.67 v, 0.7 nA)

- 147 -

3 1 2

b)
4
0,8

a)
No.1 No.2
0,6

c)

nm

0,4

0,2

0,0 0,8

No.3

No.4

0,6

nm

0,4

0,2

0,0
0 5 10 15 20 25 30 0 5 10 15 20 25 30

nm

nm

Fig. 7.5 STM images of Au7 deposited on non-sputtered HOPG a) 2D STM image b) 3D image (50 nm x 50 nm, 1.67 v, 0.7 nA) c) Line profiles of each clusters on a)

- 148 -

Our STM results in Figs. 7.3 - 7.5 show that deposited Au7 clusters stay as individual clusters with almost no sintering. As it is shown in Fig. 7.3 a, the STM image of Au7 clusters deposited on sputtered HOPG shows many particles distributed on the surface with a very narrow particles size distribution. One cannot determine the particles size (width) on this STM image precisely due to the relatively higher roughness of the sputtered-HOPG surface and the particletip convolution. One may suggest that the cluster diameter in Figs. 7.3 - 7.5 should be smaller than ~ 2 nm and the average particle height is about 0.3 nm, corresponding to two atomic layers and a 3 dimensional Au cluster structure. These results are in line with those of Buratto et al., in which Aun+ clusters (n = 1 - 8) deposited with low kinetic energy on a single crystal rutile TiO2 (110)(1x1) surface at room temperature [291, 292]. They showed that Au dimers, trimers, and tetramers lie flat on the surface, while Au5, Au6, Au7 and Au8 exhibit two-layer, three-dimensional structures (Fig. 7.6).
layers

4,5 4,0

4 3 2 1

Average Height ( )

3,5 3,0 2,5 2,0 1,5 1,0 0,5 Au1 Au2 Au3 Au4 Au5 Au6 Au7 Au8

Size-selected Clusters

Fig. 7.6 The average cluster heights of Aun (n=1-8) on the TiO2 (110)-(1x1) surface. The dashed lines indicate heights expected for various gold layers in the cluster [291]

- 149 -

However, both STM results (Au clusters on HOPG and on TiO2 surfaces) are not in line with the experimental and theoretical results reported in ref [303 - 305], which claimed that the 2D to 3D transition occurs at least for Au12 and Au14 clusters for isolated Au clusters, respectively. That means the interaction between clusters and support influences the properties of the cluster. Therefore, one can expect different chemical activities from the gas phase, depending on the surface. It is also worth mentioning that one can observe many more particles (almost 5 times more) surviving as individual on sputtered HOPG compared to non-sputtered. This can be seen in these STM images on sputtered HOPG (Fig. 7.3, 150 nm x 150 nm) and non-sputtered (Fig. 7.4 a, 50 nm x 50 nm). This result is in line with our previous STM results, in which we have shown that particle density increases with increasing defect density of HOPG [190]. In the STM image of Fig. 7.3 a, larger particles can be found in the near of vacant areas, which we suggest to be due to the agglomeration of the clusters induced by the STM tip. On non-sputtered HOPG (Fig. 7.3 b - 7.5), very small clusters can be identified with a very low density, suggesting that small clusters can be pinned on defect sites (dark areas in STM images), which were created by natural processes during preparing the HOPG sample, even though they are not stable on perfect surfaces at room temperature. The Au/HOPG systems are difficult to study with STM due to the weak metal-support interactions. Some previous studies [313, 314] suggested that asymmetric STM images can be derived from both mechanic and electronic effects. Adatoms and dimers induce surface polarization, as well as surface deformation. These results indicate that the properties of the surface, deposited particles and the STM tip will interact with each other, which affects the STM image both mechanically and electronically [311, 312].

- 150 -

The XPS Au 4f core level spectra of Aun (n = 2 - 9) clusters deposited on a sputtered HOPG surface were collected in Fig. 7.7. The size dependent changes of the Au 4f7/2 binding energies are also summarized in Fig. 7.7 b, in which almost bulk-like binding energies of the Au 4f levels were observed. Previously, Au 4f spectra of Au and Ag nanoparticles formed on the sputtered HOPG surface and silica by evaporating Au atoms on the surface were thoroughly studied. In contrast to the case of Au/SiO2 and Ag/HOPG, no significant change of the binding energy of the Au 4f state was observed as a function of particle size.

Au 4f
Intensity (arb. units)

a)
Au9 Au8 Au7 Au6 Au5 Au4 Au3 Au2
90 92 94

7/2
80 82 84 86

5/2
88

Binding Energy (eV)

84.6

b)

Binding Energy( eV)

84.4 84.2 84.0 83.8 1 2 3

Au bulk

Number of Au atoms

10

Fig. 7.7 Au 4f states of the deposited Au clusters on sputter-damaged HOPG; a) Au 4f states of Aun with n = 2 - 9 and b) Summary of the Au4f energies as a function of cluster size [221]
7/2

binding

- 151 -

A positive core level shift with decreasing particle size can be observed in general. For Au/SiO2, the positive core level shift was mostly attributed to final state effects, i.e. the hole state, which is the final state of the photoionization process, is less efficiently screened for smaller particles on poorly conducting substrates. For Ag nanoparticles on HOPG, in contrast, Ag to HOPG charge transfer forming partially positively charged Ag, was suggested to be related to the positive core level shift with decreasing particle size. In contrast, for Au nanoparticles on HOPG, existence of partially negatively charged Au was proposed, and thus, the initial and final state effects compensate each other, showing no significant core level shifts as a function of particle size [215]. According to ref. [315 - 321], smaller clusters are positively charged in a photoemission process due to insufficient contact with a carbon matrix, while for larger charged clusters, increasing contact area reduce coulomb charging. The size-dependent difference in the environment around Au nanoclusters leads to the deviation of the electronic structure compared to that of isolated clusters. However, for mass-selected Au clusters deposited on the sputtered HOPG in our experiment, as aforementioned, the Au 4f levels are analogous to the bulk in terms of binding energy and are in line with our previous results (Au nanoparticles on sputtered HOPG). It may be expected that more active binding sites (defect sites) induced by Ar ion sputtering give rise to partial charge transfer from these defect (or trap) sites to deposited Au clusters, resulting in Au-(defects or traps)+ formation. Another possibility is depth-dependent effects induced by Ar+ ion sputtering with relatively high kinetic energy (0.5 keV used here). Usually, implanted atoms as well as defects distribute nonuniformly along the depth, resulting in depth-dependent nucleation and growth of nanoclusters [321, 329, 330]. Furthermore, one cannot observed the gradual change of the Au 4f7/2 binding energies in depending on their number of atoms in a cluster, which may caused by different binding sites formed on sputtered HOPG. As mentioned before, various defect sites can exist on a HOPG surface.

- 152 -

However, it is difficult to control the amount and types of defects on the surface, which is a barrier for systematic experimental studies on the interaction between metal clusters and support.

Au2 Au4f
7/2 5/2

Au3
(a) (b)

Intensity (arb. units)

Au4

Au9

80

82

84

86

88

90

92

94 80

82

84

86

88

90

92

94

Binding Energy (eV)

Au2 Intensity; (b) / (a) 0.23

Au3 0.19

Au4 0.12

Au9 0.10

Fig. 7.8 Au 4f states of the deposited Aun (n = 2, 3, 4 and 9) clusters on sputterdamaged HOPG. Each spectrum was fitted by 4 Gaussian functions. The table shows the relative intensity ratio [(b) / (a)] of each spectrum.

Generally, peak asymmetry (Doniach-Sunjic line shape) can be seen in XPS spectra of metals caused by small energy electron excitations near Fermi energy level (EF) of metal. The degree of asymmetry is proportional to the electron density of states (DOS) at EF. However, XPS spectra in this experiment exhibit a different line shape (asymmetry) from that of bulk like Au.

- 153 -

The line shape can be decomposed by two double spectra, using a Gaussian curve fitting method. One is located around the value of bulk-like gold (a), the other is located in the high binding energy regime (b). One of the possible interpretations of those XPS data appearing at higher binding energy (b) is that small clusters can be readily trapped and/or pinned at defect sites on the carbon surface, leading to much stronger cluster-surface interaction. The relative intensity ratio (Fig 7.8) is in line with this assumption, because the intensity ratio [(b) / (a)] decreases with increasing cluster size, implying smaller clusters can be more efficiently trapped and/or pinned at the defects sites of HOPG under our experimental conditions. According to the previous works on Au nanoparticels on HOPG with variation of argon-ion sputtering times, the occurrence of (b) is directly related to the carbon surface defect density [306]. Gold carbide formation was also observed for ionized clusters in the gas phase [307], though being a very unstable species, and carbon atoms on defect sites of HOPG, as mention before, can form bonds with Au clusters directly. Core level spectra of Pt clusters deposited on HOPG also show a high binding energy component, which is attributed to Pt atoms trapped at defects [308]. Another possibility is the contribution of smaller clusters resulting from cluster fragmentation (creating a fraction of smaller clusters) and/or the existence of different isomers on the surface. However, one may rule out those possibilities because Au2 clusters do not have isomers and the clusters were soft-landed onto the surface with a very low kinetic energy (0.2 eV/atoms) to prevent fragmentation. Moreover, doublet intensity ratio [(b) / (a)] changes depend on the cluster size continuously. It is worth mentioning that Au8 was only found to be active towards oxidation and reduction (discussed later) even though the doublet intensity ratio [(b) / (a)] shows almost identical values for Au4 and Au9 clusters. Compared the doublet intensity ratio [(b) / (a)] (minimum value; 0.95) in literature in which Au nanoparticles were prepared on different carbon surfaces (pristine HOPG, Ar-ion irradiated HOPG and amorphous carbon) [306], the doublet intensity ratio [(b) / (a)] in here is also very small as it can be excluded

- 154 -

7.2 Oxidation properties of Mass-selected Au clusters on HOPG

Aun clusters (n = 2 -9) on HOPG were exposed to atomic/molecular oxygen environments and the XPS Au 4f spectra were collected. When Au 4f spectra of bare Aun/HOPG surfaces in Fig. 7.7 are compared to the results of the respective Aun/HOPG sample acquired after atomic/excited molecular oxygen exposure in Figs. 7.9 7.12, the following can be observed: for all clusters with one exception, only minor changes are observed after exposure to atomic/molecular oxygen. Only Au8 clusters are strongly oxidized, evidenced by additional peaks at higher binding energies. Type i) Au2 and Au3; The Au 4f peaks become slightly narrower after oxygen treatment. The FWHM (Full-Width-Half-Maximum) of Au2 and Au3 clusters is 0.07 eV and 0.08 eV, respectively.

A u2 Intensity (arb. units)

A u4f

A u3

A u4f

5 /2

7 /2

5 /2

7 /2

92

90

88

B in d in g E n e r g y ( e V )

86

84

82

80

92

90

88

86

84

82

80

Fig. 7.9 Au 4f spectra of the deposited Au2 and Au3 clusters after atomic/excited molecular oxygen exposure; before (black-bottom) and after atomic/excited molecular oxygen exposure (red-upper) [221]

- 155 -

Type ii) Au4; Atomic oxygen exposure induces a slight Au 4f peaks shifts to the lower binding energy range (- 0.17 eV).
Au4 Intensit (arb. units) Au4f
0.17 eV

5/2

7/2

92

90

88

86

84

82

80

Binding Energy (eV)

Fig. 7.10 Au 4f spectra of the deposited Au4 clusters after atomic/excited molecular oxygen exposure; before (black-bottom) and after atomic/excited molecular oxygen exposure (red-upper) [221]

Type iii) Aun with n = 5 - 7, 9; No change in the Au 4f level can be found upon the oxygen treatment.
Au5

Au4f

Au6

Au4f

Intsnsity (arb. units)

5/2

7/2

5/2

7/2

Au7

Au4f

Au9

Au4f

5/2
92 90 88 86

7/2
84 82 80 92 90

5/2
88 86

7/2
84 82 80

Binding Energy (eV)

Fig. 7.11 Au 4f spectra of the deposited Aun with n = 5 - 7 and 9 clusters after atomic/excited molecular oxygen exposure; before (black-bottom) and after atomic/excited molecular oxygen exposure (red-upper) [221]

- 156 -

Type iv) Au8; Au8 clusters are strongly oxidized. Au (III) peaks are observed in the high binding energy regime (green circles in Fig. 7.12).

Intensity (arb. units)

Au8

Au4f
Au (0)

5/2
Au (III)
92

7/2

Binding Energy (eV)

90

88

86

84

82

80

Fig. 7.12 Au 4f spectra of the deposited Au8 clusters after atomic/excited molecular oxygen exposure; before (black-bottom) and after atomic/excited molecular oxygen exposure (red-upper) [221]

However, it is not clear, whether the narrowing or the negative shift of the Au 4f states of some clusters are related to the reaction between Au and O, or if it is an influence of the C-O bond formation. As mentioned in a previous chapter (chapter 4), note that bare sputtered HOPG surfaces can also take some oxygen under the same conditions. Due to the change of the substrate upon oxygen exposure, the Au 4f state may change. Indeed, the line-width for the nanoclusters increases with a decrease in size, since the atoms in the nanoclusters have various coordination numbers. However, one can ignore the change of line-width, since the FWHM (Full-Width-Half-Maximum) of Au2 and Au3 clusters before and after oxygen treatment is very small (0.07 eV and 0.08 eV, respectively). As it is shown in Fig. 7.12, for all of the clusters (Aun, n = 5-7, 9) studied here except n=8, no indication for the oxidation of Au could be identified. Only for Au8, appearance of the Au (III) states at higher binding energies (marked with green circles in Fig. 7.12) is obvious.

- 157 -

7.3 Reactivity toward CO of Mass-selected Au clusters on HOPG exposed to atomic oxygen

When the deposited Au clusters are exposed to atomic/excited molecular oxygen and subsequently reacted with 3000 L of CO, some changes of the Au 4f states can be found only for Au4 and Au8 in Fig 7.13. For Au4, the Au 4f peaks shift to the higher binding energy side, recovering the original Au 4f binding energies state. The distinct Au (III) states appearing after oxygen treatment of deposited Au8 clusters are nearly completely removed by CO exposure. For all other clusters, no change of the Au 4f peaks can be found upon CO exposure.
Au2 Au4f
Au3 Au4

Intensity (arb. units)

5/2 7/2

Au5

Au6

Au7

Au8

Au8

Au9

92

90

88

86

84

82

80 92

90

88

86

84

82

92 80

90

88

86

84

82

80

Binding Energy (eV)

Fig. 7.13 Au 4f spectra of the deposited Au clusters with n = (2 - 9) after atomic/excited molecular oxygen exposure; after O (red-bottom) and after CO (blue-upper)

- 158 -

The O 1s spectra collected from the samples of Fig. 7.13 are displayed in Fig. 7.14. The interpretation of the O 1s spectra is rather complicated: the sputtered HOPG can take some oxygen, forming C-O and C=O species. These two different oxygen species bound to carbon can show the O 1s states at 531 eV and 533 eV, respectively. In addition, various Au-O species may exist, which are electronically and also chemically different. It has been shown in chapter 5 that oxygen treatment of Au nanoparticles on sputtered HOPG surfaces results in formation of at least two electronically and chemically different oxygen species bound to Au. Although the assignment of the O1s spectra is complicated, the data in Fig. 7.14 clearly shows that the chemical properties of Aun/HOPG surfaces are very sensitive to the size of the clusters deposited. Only for n=8 a pronounced change indicating oxidation of Au is clearly observed. The results in Fig. 7.14 are summarized as below.
A u2
O 1s

A u3

Au4

Intensity(arb. units)

A u5

A u6

Au7

/2
A u8 A u9

/2
538 536 534

B in d in g E n erg y (eV )

532

530

5 28

5 26 5 36 38

5 34 5 32

530 528 526

Fig. 7.14 O 1s spectra of the deposited Au clusters with n = (2 - 9) after atomic/excited molecular oxygen exposure (line with circles), and after subsequent CO exposure (solid) [221]

- 159 -

Type i) For Au3, Au7, and Au9; A broad peak centered at ~ 533 eV can be observed in the O 1s spectra after exposing the surfaces to atomic/excited molecular oxygen environments. For these clusters, a subsequent CO exposure does not result in any change of the O 1s and Au 4f spectra. Since the O 1s state at 533 eV is much more pronounced than in the case of the oxygen-treated bare sputtered HOPG surface, the 533 eV peak for these samples should be related to the oxygen species attached to Au, which is inert towards CO oxidation. The nature of this inactive oxygen species is not clear. Oxygen species can be inert towards CO oxidation either due to the fact that oxygen occupies the same adsorption site as CO, inhibiting the Langmuir-Hinshelwood type CO oxidation, or the binding energy of oxygen is too high. Formation of carbonate species by the reaction between Au, oxygen and carbon from the substrate may be also one possible explanation for the existence of inert oxygen species of Au. Since this oxygen species does not lead to a core level shift, the nature of the Au-O should be covalent rather than highly ionic.
A u3
A u7 Au 9

Intensity (arb. units)

O1s

O1s

O1s

a)

/2
538 536 534 532 530 528

/2
526 538
536 534 532 530 528 526

538 526

536

534

532

530

528

Binding Energy (eV)

Intensity (arb. units)

Au 3

O1s

Au 7

O1s

Au 9

O1s

b)

/2
538 536 534 532 530 528

/2
526 538
536 534 532 530 528 526

538 526

536

534

532

530

528

Binding Energy (eV)

Fig. 7.15 O 1s spectra of the deposited Aun clusters with n = (3, 7, 9) after atomic/excited molecular oxygen exposure (a), and after subsequent CO exposure (b)

- 160 -

Type ii) For Au2 and Au6; A peak centered at 533 eV with a shoulder at 531 eV can be found in the O 1s spectra. No CO oxidation takes place in this case. The O 1s spectra of these samples are very close to that of the bare substrate exposed to atomic/excited molecular oxygen, even though the peak at 533 eV is more pronounced than in the case of the pristine sample.
Intensity (arb. units)

Au2
O1s

A u6

a)

O1s

538

536

534

532

530

528

536 5 28

536

534

532

530

528

526

B in d in g E n erg y (eV )
Intensity (arb. units)

A u2
O1s

Au6

b)

O1s

538

536

534

532

530

528

536 528

536

534

532

530

528

526

B in d in g E n e rg y (eV )

Fig. 7.16 O 1s spectra of the deposited Aun clusters with n = (2, 6) after atomic/excited molecular oxygen exposure (a), and after subsequent CO exposure (b)

Intensity (arb. units)

Intensity (arb. units)

A u4
O 1s

a)

A u4
O 1s

b)

538

536

534

532

530

528

526

B in d in g E n e r g y ( e V )

538

B in d i n g E n e r g y ( e V )

536

534

532

530

528

526

Fig. 7.17 O 1s spectra of the deposited Au4 clusters after atomic/excited molecular oxygen exposure (a), and after subsequent CO exposure (b)

- 161 -

Type iii) For Au4; Formation of Au-O species upon oxygen treatment and removal of the oxygen species at 531 eV by CO can be observed in the O 1s spectrum, even though a large amount of oxygen remains unreacted. The Au 4f spectra in Fig. 7.13 also show negative and positive shifts after O and CO treatments, in line with the O 1s data in Fig. 7.17.

Type iv) For Au5; The shoulder at 531 eV becomes larger than for other cases mentioned above. Here again, no indication for CO-oxidation can be observed.

Intensity (arb. units)

O 1s

Intensity (arb. units)

O 1s

Intensity (arb. units)

A u5

a)

A u5

b)

A u5 vs. A u6

c)

Au5 Au6
538 536 534 532 530 528 526

538

536

534

532

530

528

526

B in d in g E n e rg y (e V )

538

536

534

532

530

528

526

B in d in g E n e rg y (e V )

B in d in g E n e rg y (e V )

Fig. 7.18 O 1s spectra of the deposited Au5 clusters after atomic/excited molecular oxygen exposure (a) and after subsequent CO exposure (b) (c); O 1s peaks intensity at 531 eV are compared.

- 162 -

Type v) For Au8; A shoulder at 530 eV is more pronounced, which is heavily reduced by CO. The O 1s peak at 530 eV can be assigned to oxygen atoms of Au (III). After exposing the oxidized Au8 clusters to CO, the O 1s spectrum becomes identical to that of the bare substrate.
A u8

Intensity (arb. units)

O 1s

Intensity (arb. units)

a)

A u8

b)

O1s

538

536

534

532

530

528

526

538

B in d in g E n e r g y ( e V )

B in d in g E n e r g y ( e V )

536

534

532

530

528

526

Fig. 7.19 O 1s spectra of the deposited Au8 clusters after atomic/excited molecular oxygen exposure (a), and after subsequent CO exposure (b)

We will show that Au5, Au7, and Au13 clusters on silica are inert towards oxidation in chapter 8, whereas other Au clusters smaller than Au13 were found to be able to be oxidized and reduced under similar O/CO conditions. On the other hand, Au8 on HOPG was found to be the only cluster, showing a significant activity for oxidation and reduction. We have shown that the chemical properties of deposited clusters are very sensitive to the substrate. Since Au nanoparticles on HOPG larger than 2 - 3 nm can form Au-oxide under similar oxidation conditions used in the present work and which can be reduced by CO [214, 216], one may argue that the Au8 clusters on HOPG sinter significantly forming larger particles rather than existing as individual clusters. However, our O 1s spectra unambiguously show that the chemical properties of the deposited mass-selected clusters are very unique and much different from the larger Au nanoparticles: when larger Au nanoparticles are oxidized, two different oxygen species form, one of them is active towards CO oxidation, and the other one is not (Fig. 7.20 a).

- 163 -

In contrast, Au8 forms exclusively active oxygen species as shown in Fig. 7.20 b. Accumulation of catalytically inactive oxygen species can poison catalysts, but Au8 on carbon surfaces may not experience this problem, which is different from other Au clusters and nanoparticles. This result shows that optimization of the catalytic activity can be achieved by control of cluster size on an atom-byatom basis.

Au nanoparticles Intensity (arb. units)

O1s

a)

after atomic O after CO

526

528

530

532

534

536

Binding Energy (eV)

Au8 Intensity (arb. units)

O1s

b)

after atomic O

after CO
526 528 530 532 534 536

/4

538

Binding Energy (eV)

Fig. 7.20 O 1s spectra of Au nanoparticles with a laterial size of 2 - 3 nm (a) and deposited Au8 clusters (b) on HOPG after atomic/excited molecular oxygen exposure (red line), and after subsequent CO exposure (blue line)

- 164 -

Besides Au8, other clusters studied here also show different chemical properties from those of larger Au nanoparticles. Au4 clusters show negative and positive shifts of the Au 4f states without Au (III) species formation upon O/CO exposure, which has never been observed from larger Au nanoparticles on HOPG. Note that the relatively small Au nanoparticles show a negative core level shift upon oxygen treatment, yet no positive shift upon CO exposure (Chapters 5 and 6). Furthermore, the O 1s spectra of Aun with n = 2, 3, 6, 7, 9 showing a pronounced feature centered at 533 eV, are much dissimilar to the respective data from larger Au nanoparticles (Fig. 7.20 a). For Au5, the O 1s state centered at 530-531 eV is not reduced by CO, whereas for larger Au nanoparticles, the O 1s state at the same binding energy range is easily removed by CO at room temperature. All these results demonstrate unique chemical properties of the mass-selected deposited Au clusters, and indicate that the mass-selected clusters are not sintering but rather surviving as individual clusters on the surface after deposition on sputtered HOPG surfaces. In conclusion, we found that the oxidation and reduction pattern of massselected Aun clusters with n = 2 -9 deposited on sputtered HOPG are much different from those of Au nanoparticles deposited on sputtered HOPG and silica (discussed in chapter 8), suggesting that experiments using mass-selected clusters can shed light onto metal-support interactions as well as size-dependent changes of chemical properties in more detail. Only Au8 clusters on sputtered HOPG show chemical activity toward the oxidation and reduction by atomic O and CO, respectively.

- 165 -

8. Mass-selected Au clusters on SiO2/a-Si

8.1 Preparation of Mass-selected Au clusters on SiO2/a-Si

In the present chapter, we study oxidation patterns of mass-selected Aun clusters (n = 2 - 13) deposited on silica surfaces. In the previous chapter, only Au8 was found to be active towards oxidation and reduction. However, we demonstrate that each additional Au atom can drastically change the oxidation pattern of the Au clusters on SiO2/Si. In particular, Au5, Au7 and Au13 are not oxidized at all under our oxidation condition, suggesting an extraordinarily high stability of these clusters on silica [217]. For the creation of mass-selected Au clusters, the same magnetronsputter source was used. All the clusters were soft-landed on the amorphous silicon surfaces with native oxide layer, that is, a kinetic energy of 0.2 eV per atom was used. Before the Si wafers were put into UHV, they were chemically etched using a mixture of three different acids and water (see chapter 3). A typical mass spectrum of the Au clusters is shown in Fig. 7.1. As mentioned before, one can rule out contamination of the Au clusters by oxygen molecules during deposition. In XPS spectra (Fig. 8.1), one cannot detect any kind of contamination except carbon (C) and oxygen (O) that were already on the silica surface before deposition.

- 166 -

O1s

Intensity (arb. units)

Si2p

Si2s

C1s Au4f

100

200

300

400

500

600

Binding Energy (eV)

Fig. 8.1 XPS survey spectra of Au clusters deposited on SiO2/Si surface

For each sample, 2 x 1012 Au clusters were deposited on Si wafer surfaces (Fig. 8.2) corresponding to Au a coverage of about 5 % of a monolayer for Au13. The corresponding Au 4f XPS spectra in Fig. 8.2 are presented as acquired for the different Aun cluster sizes (n = 2 - 13) after deposition on SiO2/Si. Since the same number of clusters was deposited for all samples, Au 4f peak intensity should increase with increasing cluster size. Fig. 8.3 shows the change of the intensity ratio of the Au 4f7/2 peak with respect to the Si satellite peak at 89.7 eV as a function of cluster size. For all clusters, an almost linear increase of the Au 4f7/2 peak intensity as a function of cluster size can be found. However, one can see some deviations from the linear tendency, particularly Au9 and Au10 clusters. This may be due to slightly different sample position, angle or condition of silica surface (defect density and/or defect sites), and so on. However, it is important to mention that all the XPS data shown in the present work are highly reproducible.

- 167 -

Intensity (arb. units)

Au4f7/2

a)
Au 13 Au 12 Au 11 Au 10 Au 9 Au 8 Au 7 Au 6 Au5 Au4 Au 3 Au 2

Si
92 88 84

Binding Energy (eV)

80

Fig. 8.2 Au4f XPS spectra of various Au clusters on SiO2/Si; typically Au4f7/2 at 84.0 eV, Au4f5/2 at 87.6 and 2 Si satellite peaks at 87.7 eV and 89.7 eV

1,0

Au 4f 7/2/Si Sattelite

0,8 0,6 0,4 0,2

2 4 6 8 10 12 14

Number of atoms in a cluster

Fig. 8.3 Au 4f7/2 intensity ratio corresponding to the Au 4f XPS spectra in Fig. 8.2

- 168 -

In Fig. 8.2, one cannot observe the Au4f5/2 peak clearly due to the Si satellite peak with a relatively high intensity [216]. To avoid this problem, we subtract the silicon background from the raw spectrum (Fig. 8.2) by fitting gauss functions (Fig. 8.4). Each spectrum was fitted by 7 gauss functions (2 for Au (0), 2 for Au (III) and 3 for Si satellites). During the fitting procedure, the three functions describing the Si satellite peaks (green curves) were only marginally varied from spectrum to spectrum (Fig. 8.4). Fig. 8.5 shows the XPS Au 4f level spectra of Aun clusters (n = 2 - 13) on SiO2/Si after subtraction the Si background. One can clearly observe the splitting of Au 4f7/2 and Au4f5/2 peaks.

Au5 : pristine sample Intensity (arb. units)

Au 4f

Si 5/2

7/2

80

82

Binding Energy (eV)

84

86

88

90

92

Fig. 8.4 Raw XPS spectra of Au5 clusters on SiO2/Si fitted by 7 gauss functions [217]

- 169 -

Au4f

Au2

Au 3

Au4

Au5

7/2

5/2

Intensity (arb. units)

Au6

Au7

Au8

Au9

Au10

Au11

Au12

Au 13

Binding Energy (80-93 eV)

Fig. 8.5 Au 4f level spectra of deposited Au clusters after subtraction the silicon background

- 170 -

The native SiO2 films of a thickness of about 1 nm on Si wafers could be identified in the Si 2p spectra (Fig. 8.6). The dashed lines at 99.6 eV and 103.6 eV in the Fig. 8.6 indicate the peak position of Si 2p levels corresponding to Si (0) and Si (IV), respectively. Oxide thickness can be measured by various techniques (Table 8.1). Technique TEM (Transmission Electron Microscopy) XPS (X-ray Photoelectron Spectroscopy) AES (Auger Electron Spectroscopy) SIMS (Secondary Ion Mass Spectrometry) RBS (Rutherford Backscattering Spectroscopy) Ellipsometry Advantages Accurate without standards Precise to
0.1 nm

Disadvantages Expensive; very small area analyzed Less precise for thick films; accurate over wide range Not accurate, further calibration necessary Further calibration required

Optimum range

Useful range

>1 nm

>1 nm

For film < 3nm Inexpensive , readily available Qualitative film thickness Accurate without standards Inexpensive , fast

<10 nm

20 nm

<4 nm

20 nm

10 nm

Not precise for films 1 nm Not accurate for films < 2 nm

1 nm

1 nm

>10 nm

>2 nm

Table 8.1 Analytical techniques to measure the oxide thickness [343]

- 171 -

In XPS, the binding energies of the oxide and the metallic species are usually separated by a few electron volts. Thus, when the oxide is thin (< 10 nm), it is possible to distinguish the contribution from both oxide and metal photoelectrons. In order to measure the SiO2 thickness in our experiment, equation was used the following [343];

1 I 2 tox = SiO 2 sin ln{( )( SiOexp ) + 1} I Si

: the angle between the sample surface plane and the electron analyzer

exp

SiO2: the attenuation length of the Si 2p photoelectrons in SiO2 (2.7 2 nm) : the Si 2p intensity from infinitely thick SiO2 and Si (0.83) ISiO2 exp / Isiexp : the ratio of intensities of the unknown film

Si2p

Intensity (arb. units)

Si(IV)
106 104 102

Si(0)
100 98 96

Binding Energy (eV)

Fig. 8.6 XPS Si2p spectra after exposing the samples to the oxygen atmosphere with a hot Pt-filament, no change of the Si-signal can be found. The peak at ~103 - 104 eV corresponds to the Si (IV) (oxide layer), whereas the peak at ~ 99 eV comes from Si (0)

- 172 -

As already mentioned before, Au7 on rutile TiO2 [72] and Au8 on MgO film [71] are the smallest Au cluster with a measurable reaction rate for CO oxidation. Comparing the activity of supported Au clusters on different supports, it has been demonstrated that Au clusters on reducible oxides, e.g. TiO2, Fe2O3, are more active than Au cluster with almost identical particles size on irreducible oxides e.g. SiO2, Al2O3 under similar conditions [89, 141, 142]. Irreducible oxides are characterized by higher metal oxygen bond strength and a larger band gap compared to reducible oxides. Charge transfer from oxide support to Au cluster can be considered one of the key points to enhance catalytic activity for CO oxidation. Because of the different oxidation states of supported Au clusters depending on the form of the oxide support, the catalytic activity of Au clusters can change. Since more charge is transferred from reducible oxide to Au, the Au deposited on these oxides interacts more strongly with oxygen than Au on irreducible oxide support, providing an ideal environment for O2 activation and for the oxidation reaction. The capacity of Au atoms adsorbed on reducible oxide to be readily oxidized and reduced at mild conditions might explain the unique low-temperature oxidation activity of Au on reducible oxides. One can say the strong metal-support interaction affects the catalytic activity. There is, however, a pressure gap between model catalysts and real catalysis systems. Usually, in experimental and theoretical works, well-defined oxide supports were used for model catalysts, whereas this is not the case in real catalysis. Deposition of transition metal clusters and particles on silicon dioxide surface has been investigated for catalytic applications [332, 333]. These studies are dealing with amorphous silica surface due to the difficulties in growing and preparing well-defined crystalline surfaces of this material. Only recently the synthesis of a new crystalline phase of SiO2 in form of a thin film has been reported by the group of Freund et al. [334 - 336]. In general, the adsorption of metal atoms onto defect free silica films is very weak like for HOPG surface. Due to the small diffusion barriers, metal atoms can diffuse almost freely on the surface at room temperature and even below forming larger particles.

- 173 -

Defects on crystalline silica films have been studied and three major defects are expected to be present; extended defects (steps and kinks), line defects (antiphase domain boundaries), and point defects (oxygen vacancies) [326, 339 - 341]. Defect-poor and defect-rich silica films can be prepared by changing the preparation method. For example, silica films with a high density of defects can be prepared using a sample annealing temperature of 1100 K [71, 344 - 346]. In the results of Goodmans group, it was suggested that the most probable defects are oxygen vacancies, since silanol groups are unlikely given the high annealing temperature [344] However, non-bridging oxygen could form under these conditions. While on low-defect films the Au clusters nucleate along the line defects, on the high-defect film the clusters nucleate primarily at point defects on the terrace sites. The results show that the cluster density correlates with the number of point defects [344]. In this experimental work, we used chemically etched amorphous Si wafers with a native oxide layer of thickness about 1 nm as a support which has a lot of defect sites and some molecular species related carbon, oxygen and H2O. It is therefor more closely to a real catalyst system. One can expect some advantages; preventing the charging problem, no sintering of Au clusters on the surface after deposition and high catalytic activity. It is known, for example, that Au clusters deposited on defect-poor MgO surface show less catalytic activity than on defect-rich MgO surfaces and that moisture on Au/oxide influences the catalytic activity [347]. The presence of moisture on gold catalysts can increase the catalytic activity by up to 2 orders of magnitude [121 - 124].

- 174 -

In order to investigate the influence of defect sites created in our experiment on the binding property and chemical activity of Au clusters towards atomic O, the chemically etched amorphous Si wafers loaded on sample holder in UHV 1 chamber were heated before depositing Au7 cluster at two different temperature conditions: 250oC and 500oC.
Au7 Intensity (arb. units)

0.1 eV

Au4f

b) 500 C
7/2

5/2

a) 250 C

82

84

86

88

90

92

Binding energy (eV)

Fig. 8.7 The Au4f XPS spectra of the Au7 cluster deposited on pre-heated SiO2/Si at different heating temperature conditions; a) at 250oC and b) at 500oC

Au7 Intensity (arb. units)

Au4f

after atomic O

Au (III) Au (0)

b) 500 C

7/2

5/2

a) 250 C

82

Binding energy (eV)

84

86

88

90

92

Fig. 8.8 The Au4f XPS spectra of the Au7 cluster deposited on pre-heated SiO2/Si at different heating temperatures; after exposing to oxygen with a hot Pt-filament; a) at 250oC and b) at 500oC

- 175 -

The binding energy (BE) of the Au7 clusters deposited on SiO2/Si preheated at 500oC (Fig. 8.7.a) shifts to lower binding energy regime compared to BE of Au7 clusters deposited on SiO2/Si pre-heated at 250oC (Fig. 8.7.b). One reasonable explanation is that the defect density decreases (SiO2/Si surface are smoothed) with increasing heating temperature, giving rise to an increase in the mobility of deposited Au clusters on the surface, which then form larger particles. Usually, the BE shifts to higher binding energies with decreasing particle size. This possibility can be verified by exposing these samples to atomic oxygen environments. As mentioned in chapter 6, Au nanoparticles larger than 0.7 nm in particle height on SiO2/Si can be readily oxidized. The oxidation properties of the Aun clusters (n = 2 - 13) on SiO2/Si will be discussed in a moment. Au7 is not oxidized under our experimental condition. However, this oxidation behavior is changed depending on the pre-heating conditions of SiO2/Si: only Au7 clusters deposited on SiO2/Si pre-heated at 500oC oxidized (Fig. 8.8.b), evidenced by additional peaks at higher binding energies. The Au particle size could be larger than 0.7 nm in height after being deposited on SiO2/Si pre-heated at 500oC suggesting cluster sintering on these surfaces. On a basis of these results, it is important to mention that defects sites created by chemical etching method in our experiment are very importants, preventing the cluster from sintering on the SiO2/Si surface, i.e. each Au clusters survives as an individual on these surfaces.

- 176 -

The Au coverage could be determined by measuring the total sample current during deposition of Au clusters. As the cluster size (n) increases from 2 to 13, the binding energies of the Au 4f states vary between ~ 84.0 eV and 84.5 eV. Fig. 8.9 plots the shift of the core level energies of the Au clusters with respect to that of the bulk material for Au. In our repeated experiments, the binding energies of the deposited Au clusters are reproducible within a deviation of 0.2 eV.
84,8 84,4
Au bulk

Binding Energy / eV

84,0 83,6 0 2 4 6 8 10 12 14

Number of Au atoms
Fig. 8.9 XPS Au 4f7/2 core level shifts as a function of number of Au atoms in a Au cluster on SiO2/Si [217; until Au10]

The Au 4f binding energy for Au/silica systems generally shifts to high binding energy regime with decreasing particle size mostly due to final state effects. Positive core level shifts larger than 1.0 eV with respect to the bulk values could be found for the Au clusters consisting of less than ~ 13 atoms and small nanoparticles [217, 342]. In Fig. 8.10, the Au 4f binding energy shifts were summarized as measured on ensembles of isolated gold nanoparticles or clusters prepared by different techniques [342]. The Au 4f binding energy shifts were interpreted using 1/d behavior justifying the modelling of the nanoparticles as extremely small spherical capacitors (inset of Fig. 8.10). These small particles are positively charged with one elemental charge during the escape of the photoelectron, thereby reducing its kinetic energy due to coulomb interaction.

- 177 -

Fig. 8.10 Au 4f binding energy shifts of isolated gold nanoparticles or clusters. The solid line represents 1/d behaviour [342] - Full squares (): micellar method, SiO2/Si substrates - Open squares (): ligand-free Au55 clusters on SiO2/Si - Full circles (): size-selected clusters on SiO2/Si [337] - Open circles (): size-selected clusters on amorphous carbon [338]

On the other hand, the Au 4f binding energies of the deposited Aun clusters (n = 2 - 13) on SiO2/Si in the present work are close to the bulk like value, even though cluster size is very small. That is not in line with the previous XPS results reported in ref. [217, 342] (Figs. 8.10 and 8.11). The Pt clusters particularly show larger shifts and more variations with cluster size than the Au clusters (Fig. 8.11). It was suggested that the deposited Au clusters (independent of size) are somewhat mobile on silica and thus form larger clusters depending on cluster size [337]. However, particle sintering is not likely in our experiment, since the chemical properties of the deposited clusters change significantly with each additional atom, as it will be shown later.

- 178 -

Moreover, it should also be mentioned that the positive Au4f level shifts are significantly larger (> 0.6 eV) than those of the deposited Au clusters found here, when Au atoms are evaporated onto the same substrate, resulting in a similar Au coverage as those of the mass-selected deposited Au clusters [217].
0.0

Our data points

Energy Difference to Bulk BE (eV)
-0.5 -1.5 -1

Au clusters Pt clusters
0 1 2 3 4 5 6 7 NUMBER of ATOMS 8

Fig. 8.11 A plot of the 4f7/2 binding energy shifts of the Au (filled squares), Pt (open squares) clusters in ref. [337 - 338], and our data (green block) relative to the bulk values for each metal cluster as a function of cluster size. Au and Pt clusters were deposited onto Si (100) wafers with natural oxide layer

The Au 4f core level shifts of Au clusters obtained in our experiment cannot be explained by using only the former interpretation method mentioned in Ref. [337, 338]. One possible explanation for the Au4f level shifts is that deposition of the Au clusters can induce formation of specific point defects on silica surfaces, which are particularly electron-rich. On those defect sites created by deposition, Au clusters may be pinned and partially negatively charged [72]. Another possibility is the bonding formation like negatively charged Au clusters-positively charge defect sites (Aunx-(defects)y+), induced by strong metal support interaction between Au clusters and defect sites created by chemical etching method.

-2

- 179 -

As already mentioned before, defect formation on silica is important for the nucleation of Au clusters on the surface. Pacchioni et.al revealed in detail ([326, 339, 341]) that several different point defect sites on the surface of amorphous silicon oxide can exist such as; Si dangling bond(Si), 2) nonbridging oxygen(Si-O), 3) silanolate(Si-O-), 4) peroxyl radical(Si-OO), 5) oxygen vacancy(Si-Si), 6) oxygen divacancy(Si-Si-Si), 7) twocoordinated silicon(=Si:), 8) peroxo group (Si-O-O-Si), 9) hydride group (Si-H), and etc. Defect sites like Si-O and Si-O- efficiently trap Au clusters, preventing diffusion and aggregation. Because the Si wafer was etched in our case, one can expect even more complicated defect sites with defferent sizes, depths and oxidation states.

Si dangling bond(Si)

Silanolate(Si-O-) Nonbridging oxygen(Si-O)

Oxygen vacancy(Si-Si)

Fig. 8.12 The geometric structure of the Si-Au complex formed on Si dangling bonds (Si), Si-O-Au complex formed on nonbridging oxygen(Si-O) and silanolate(Si-O-) defects, and Si-Au-Si complex formed on oxygen vacancy(Si-Si) [339]

Interaction between Si atoms in mono- or di-oxygen vacancies or Si dangling bonds and Au may result in charge transfer due to a higher electronegativity of Au compared to Si. According to the definition of Paulings electronegativity: Electronegativity is the power of an atom in a molecule to attract electrons to itself , the value is 2.54 and 1.90, respectively.

- 180 -

In Fig. 8.12, for example, the Si dangling bond (Si) induces a strong bond between the Au and Si atom due to the coupling of the unpaired electron in the sp3 hybrid orbital of the Si atom and the Au 6s electron. This kind of interaction also suggests a high thermal stability of gold clusters on silica surfaces. The Au atom can also form a direct covalent bond with nonbridging oxygen (Si-O) by coupling the 6s electron with the unpaired O2p electron. In particular, it is worth mentioning that the soft-landing of Au clusters can readily form negatively charged Aun-O-Si species on silica surfaces. Extra electron on silanolate (Si-O-) point defect site contributes to the O-Au molecular orbital. This could result in a charge transfer to the adsorbed Au atom and form a SiO-Au- complex. Silanolate defect site plays an important role in the optical properties of Au atoms and clusters deposited on an amorphous silica surface [326, 340]. Another charge transfer process from support to Au clusters is that Au serves as the missing oxygen atom, taking its place in the bridge position of silicon bond and attracting charge from the Si neighbors because of its high electron affinity. The unpaired electron remains localized on the Au atom. Based on these previous experimental and theoretical works; excess electrons in these kinds of defect sites could be localized on Au clusters. The Au 4f core level shifts which were closer to the Au bulk value than expected could be rationalized. However, one cannot observe the gradual change of the Au4f7/2 binding energies depending on the number of atoms in the cluster. One appropriate explanation for our results is that; depending on the Au cluster size, different types of Au-slilica complexes can be present, which give rise to different charge states of Au clusters and therefore the degree of binding energy shifts can be varied.

- 181 -

8.2 Oxidation properties of Mass-selected Au clusters on SiO2/a-Si

The deposited mass-selected Aun clusters (n = 2 - 13) were exposed to oxygen atmosphere with a hot Pt filament to shed light onto the size-dependent change of the oxidation pattern. In this way one can produce atomic oxygen and/or thermally excited molecular oxygen species. Most model catalyst studies done with Au/TiO2 system under UHV conditions have used atomic oxygen, because O2 has a low sticking probability on TiO2 and Au/TiO2 [109, 111], whereas atomic O was found to bind to nanometer size Au particles on Au/TiO2. The changes of the Au 4f state of the deposited Au clusters on SiO2/Si upon oxidation are displayed in Figs. 8.13 - 8.15. First, we focus on the Au clusters with n = 2 - 4, 6, 8 - 12. After exposing these clusters to the oxygen atmosphere, the appearance of additional states at the higher binding energy sides can be found. Concomitantly, the original Au 4f peaks decrease in intensity. Fig. 8.14 shows the XPS spectra after subtracting the Si satellite background from Fig. 8.13. One can clearly observe the splitting of Au 4f levels (Au4f7/2 and Au4f5/2) and the additional peaks at the high binding energy regime. The additional peaks appearing after oxygen exposure are characteristic features of oxidized Au; Au (III) formation. The Au (III) peaks in the core level spectra can be also observed, when Au bulk crystal or the Au nanoparticles larger than ~ 1.5 nm in diameter are oxidized under similar conditions [217]. In contrast to the other Au clusters already mentioned, Au5, Au7, and Au13 do not show any change of the Au 4f state upon oxygen exposure.

- 182 -

Au2
Si Au4f7/2

Au3

Au4

Au5

Intensity (arb. units)

Au6
Au(o)

Au7

Au8

Au9

Au(III)

Au10

Au11

Au12

Au13

Binding Energy (80-90 eV)

Fig. 8.13 Raw Spectra of the Au 4f levels of deposited clusters before (black) and after (red) exposing to oxygen [217; until Au10]

- 183 -

Au4f

Au 2

Au 3

Au 4

Au 5

Intensity (arb. units)

Au 6

Au 7

Au 8

Au9

Au 10

Au 11

Au 12
Au(III)

Au 13
Au(0)

7/2 5/2

Binding Energy (80-93 eV)

Fig. 8.14 XPS spectra of the deposited Au clusters consisting of 2 - 13 atoms after exposing to the atomic oxygen atmosphere. Each spectrum was originally fitted with 7 Gauss functions. Three of them corresponding to the Si satellite features are removed, and only the Au peaks are shown [217; until Au10]

- 184 -

In Fig. 8.15, a strong even-odd alternation behavior for Aun clusters with n = 4 8 and for Au12, Au13 cluster can be observed regarding the appearance of the Au-oxide peak. However, the spectra of Au9 and Au11 represent a clear exception in the even-odd pattern.

Au(III) / Au(0) ratio

1.2 0.8 0.4 0.0 2 4 6 8 10 12 Number of Au atoms in a cluster 14

Fig. 8.15 Relative intensities of the oxidic and metallic peaks in Fig. 8.14; Au 4f
7/2

state was used for this quantitative analysis [217; until Au10]

One cannot explain the origin of even-odd oxidation pattern for Aun clusters with n = 4 8 accurately. The chemical stability of clusters might be influenced by a high density of defects of the supporting material. As mentioned before, defects in the support are known to modify the reactivity of Au clusters. That is, the even-odd oxidation pattern and the corresponding stabilities are caused by cluster size-induced modifications in the electronic structure. On the other hand, on the basis of previous results of Boyen et al. [44] (they suggested that magic Au55 cluster are completely inert toward oxygen uptake), one may also suggest that the nature of higher stability toward oxidation of the Au13 cluster is due to the closed-shell model (magic number of Au atoms).

- 185 -

If this model is the most important factor for the high stability of Au cluster toward oxidation, one can also expect this kind of stability for larger magic number clusters like Au147. However, one cannot apply this model to cluster sizes smaller than 13 in this experiment. Upon deposition of the clusters, originally flat Au clusters may experience structural transformation [305]. During this process, fragmentation of the clusters cannot be completely excluded even in the case of the soft-landing (less than 0.2 eV per atom). For Aun clusters (n = 2 - 4, 6, 8 - 12), in fact, one cannot completely rule out cluster fragmentation based on our XPS data. It is notable that those clusters are only partially oxidized. For the clusters smaller than 4 atoms, it is expected that atomic or thermally excited molecular oxygen should completely oxidize the clusters. Formation of the partially oxidized state of Au (such as Au (I) or Au (II)) is not likely. One may suggest that either different nucleation sites which induce different electronic states of the clusters yields the different chemical properties of the clusters of the same size, or some of the clusters fragment or sinter. The even-odd pattern in electronic and chemical properties, and cluster stabilities have been theoretically and experimentally explored for free Au clusters in the gas phase; however, such a pattern has been observed for deposited clusters for the first time in the present work [75, 76]. The higher stabilities of the odd-numbered clusters do not reconcile with simple electronic models (such as one-electron shell model) of the neutral free clusters, since higher cluster stabilities (or chemical inertness) due to the closed-shell configuration should be observed for even-numbered clusters; however, the interactions between Au and defect sites of silica may change the electronic and geometric structures of Au clusters significantly. This kind of explanation is also in agreement with the Au 4f core level shift data mentioned above.

- 186 -

However, one can exclude the possibilities of cluster fragmentation or sintering as well as geometric effects (closed shell model; magic number of Au clusters) in our experiment, since this even-odd oxidation pattern of deposited Aun clusters (n = 5 8, 12, 13) inverts after reacting with NaOH solution regarded as a promoter to change of electronic property (which will be discussed in next chapter). Consequently, Au clusters on SiO2/Si are present as individual clusters and the nature of different chemical properties of Aun clusters is due to modified electronic states. Compared to Au clusters and larger Au nanoparticles on silica surfaces which were prepared under the same experimental conditions, chemical properties of Au5, Au7 and Au13 were shown to be very unique. Au5, Au7 and Au13 are suggested to be highly stable and do not experience fragmentation or sintering, since those processes should result in formation of Au-O bond, which was not found here. To shed light onto the origin of the even-odd alternation of the oxidation behavior, further studies, in particular those using theoretical approaches are required.

- 187 -

8.3 Reactivity toward CO of Mass-selected Au clusters on SiO2/Si exposed to atomic oxygen

To get information about the chemical activities of the oxygen species formed on deposited mass-selected Au clusters on silica surfaces, the oxidized Au clusters were exposed to 3000 Langmuir of CO at room temperature. Figs. 8.16 and 8.17 (before and after subtracting the Si satellite backgrounds) show the XPS core levels of Au 4f after exposing CO. The intensities of the peaks corresponding to Au (III) decreased: most of the oxidized Au clusters can efficiently be reduced by CO. Considering that several thousands of Langmuir of CO are required to completely consume the oxygen species of Au-oxides, the CO to CO2 conversion probability (~ 0.1 %) can be estimated to be in the same order of magnitude than the reactivity of RuO2 layers, which are shown to be very active for CO oxidation at room temperature [214].
Au2
Si Au4f7/2

Au3

Au4

Au5

Intensity (arb. units)

Au6
Au(o)

Au7

Au8

Au9

Au(III)

Au10

Au11

Au12

Au13

Binding Energy (80-90 eV)

Fig. 8.16 Raw spectra of the Au 4f levels of deposited clusters before (red) and after (blue) exposing to 3000 L of CO [217; until Au10]

- 188 -

Au4f

Au2

Au3

Au4

Au5

Intensity (arb. units)

Au6

Au7

Au8

Au9

Au10

Au11

Au12

Au13

Binding Energy (80-93 eV)

Fig. 8.17 The XPS spectra of the deposited Au clusters first exposed to oxygen atmosphere with a hot Pt-filament and then exposed to 3000 L of CO at room temperature (after subtracting the Si satellite backgrounds in Fig. 8.16) [217; until Au10]

It is important to mention that the reduced Au clusters can be oxidized using atomic oxygen and reduced using CO again, i.e. the oxidation and reduction cycles are reversible. Figs. 8.18 and 8.19 (before and after subtracting the Si satellite background, respectively) show the XPS spectra for the oxidation/reduction cycle of Au6 clusters on SiO2/Si.

- 189 -

Au6

Intensity (arb. units)

Au(0) Au(III)

CO_5,000 L Atomic O CO_5,000 L Atomic O CO_3,000 L Atomic O Au6 on SiO2/Si bare Si wafer
76 78 80 82

Au4f7/2

Si

Binding Energy (eV) Fig. 8.18 Au6 clusters on SiO2/Si undergo reversible oxidation and reduction

84

86

88

90

92

cycle by reacting with atomic oxygen and CO atmosphere

CO 5,000 L

g) f) e) d) c)
Au(0) Au(III)

atomic O

Intensity (arb. units)

CO 5,000 L

atomic O

CO 3,000 L

atomic O

b) a)
5/2

Au6
7/2

Au4f

80

82

Binding Energy (eV)

84

86

88

90

92

94

Fig. 8.19 Au6 clusters on SiO2/Si undergo reversible oxidation and reduction cycle by reacting with atomic oxygen and CO atmosphere (after subtracting the Si satellite backgrounds in Fig. 8.18)

- 190 -

After the initially oxidized Au6 clusters are reduced using 3000 L of CO (Fig. 8.19 a-c), subsequent atomic oxygen exposures result in appearance of Au (III) species again at the higher binding energy regime with a relatively lower intensity (Fig. 8.19 d) than that before the reduction. An additional reduction using 5000 L of CO yields disappearance Au (III) peaks (Fig. 8.19 e), coincidentally the peak intensity of the Au (0) recovers the former intensity. However, the Au (III) species are not completely removed by CO, when being repeated oxidation and reduction cycle, because of the fact that not only the topmost Au layers but also deeper layers are oxidized. Larger amount of CO is required for removing the Au-O species completely (Fig. 8.19).
1,6

f) b) d)

Au(III) / Au(0) ratio

1,2

0,8

0,4

0,0

a)

c)

e)

g) Au(0)

Oxidation & Reduction cycle

Fig. 8.20 Relative intensity ratio of the oxidic and metallic peaks of Au6 clusters on SiO2/Si in Fig. 8.19

Another possibility is the formation of strong Au-O bond. When we compare the relative intensity (Au (III) / Au (0)) of Au clusters from every oxidation and reduction process (Fig. 8.20), one can observe that more Auoxygen species (Au (III)) are still remaining on the Au clusters at the end of the cycle. The remaining oxygen species on Au clusters can prevent the CO adsorption onto the Au cluster, occupying the same binding sites as CO, giving rise to a decrease in the CO to CO2 conversion probability of Au clusters on SiO2/Si.

- 191 -

Summarizing, on the basis of these results, one can confirm that the appearance of the additional peaks in the Au 4f level by oxygen exposures should come from the formation of Au-O bonds. The oxygen species of Au oxide participate in the oxidation of CO directly. Much different size selectivity of the oxidation of Au was observed on the silica compared to the results from sputtered HOPG (only Au8 was found to be active towards the oxidation and reduction), suggesting importance of metal-support interaction for chemistry of deposited clusters. As already mentioned before, Au8 is the smallest catalytically active cluster on MgO and Au3 and Au7 clusters deposited on TiO2 have shown significantly high catalytic activities towards CO oxidation. The sizes of catalytically active Au clusters on titania turned out be different from those on MgO, pointing towards importance of metal-support interactions. Diverse chemical properties of mass-selected clusters, which cannot be observed for larger Au nanoparticles formed on the same substrate, are demonstrated; Au5, Au7, and Au13 clusters on SiO2/Si do not oxidize at all, addressing importance to the ability to control the cluster size on an atom-byatom basis to tune catalytic activity. Thus, Au5, Au7, and Au13 clusters on SiO2/Si may play a significant role in the design of nanoscaled devices where chemical inertness is of crucial importance.

- 192 -

9. Current results
To shed light on size selectivity in heterogeneous catalysis, extensive theoretical and experimental studies on both gas phase and deposited clusters have been devoted. We have also shown the change of chemical activities toward oxidation/reduction as a function of particle size for the deposited Ag and Au nanoparticles in previous chapters. Beside these results, even-odd alternation pattern in oxidation properties, which was not observed for Au nanoparticles in a similar particle size regime, has been observed for massselected Au clusters deposited on silica surface, i.e. the chemical properties can be changed with every additional atom. It is important to mention that the changes in the electronic structure of mass-selected Aun clusters on silica dominate the chemical properties instead of geometric structure. However, one can not fully exclude geometric effects, because Au13 cluster shows unexpected chemical behaviour while Au9 and Au11 represent a clear exception to the evenodd pattern. In order to verifying the electronic effects on the change of chemical properties of Au clusters on silica, mass-selected Aun clusters with n = 5 8, 12, 13 deposited on chemically etched SiO2/Si were immersed into diluted sodium hydroxide solution. And then oxidation experiment by using atomic oxygen on these clusters was carried out determine the influence of sodium hydroxide on chemical properties of Au clusters in UHV conditions. Electronic structures of Au clusters can be strongly modified by alkali metal atoms, due to the charge transfer between alkali metal and gold. Alkali metals easily donate electrons to neighbours, and thus alkali metal doping on supporting materials or metal clusters can give rise to the formation of partially negatively charged clusters, which can significantly change chemical properties of deposited metal clusters. Opposite even-odd patterns in oxidation property for deposited Au clusters on SiO2/Si are observed after immersing into NaOHH2O solution. That is, the change of chemical activities of deposited Au clusters on silica can be interpreted by using not the geometric effects but the electronic effects rather.

- 193 -

Au5
Intensity (arb. units)

Au (0)

Au4f
5/2
+ 0.08 eV

7/2

atomic O

after water

pristine Au5
80 82 84 86 88 90 92 94

Binding Energy (eV)

Au (0)

Au8
Intensity (arb. units)
x2

Au4f
Au (III)

atomic O
+ 0.08 eV

after water
7/2 5/2

pristine Au8

80

82

Binding Energy (eV)

84

86

88

90

92

94

Fig. 9.1 XPS Au 4f core level spectra; comparative studies on oxidation properties of Au5 and Au8 clusters before and after exposing to distilled water. Si satellite back ground peaks were subtracted.

Before proceeding these experiments, the stability of deposited massselected Au clusters on silica under our experimental condition has been explored. The stability of the nanoclusters is crucial for all applications, i.e. an understanding of the sintering process for nanoclusters is also important for long lifetime performance of novel devices and materials based on nanoclusters [170].

- 194 -

The deposited mass-selected Au clusters on silica was exposed to air, and then oxidation experiment by using atomic oxygen on these clusters with identical oxidation conditions like before was carried under UHV conditions. Moreover, the same experiment was done using distilled water. Oxidation property of these Au clusters is not changed before and after exposing to air and distilled water. That is, the Au5 clusters on silica, which did not oxidize before exposed to air and water, still do not oxidize and the Au8 clusters on silica, which was oxidized before exposed to air and water, are oxidized (Fig. 9.1), implying deposited Au clusters under our experimental conditions are quite stable under ambient conditions. All experimental data are not shown in this thesis, and will be discussed in the Ph. D thesis of Rainer Dietsche. Based on these results, next experiment was carried out;
Deposition of the size-selected Aun clusters (n=5-8, 12, 13) on SiO2/Si surface (UHV 1)

XPS analysis (UHV 1)

Air transfer
Oxidation experiment with atomic oxygen (UHV 1) Immerse into NaOHH2O

Fig. 9.2 Scheme of the experimental procedure

In UHV 1 system, size-selected Aun clusters (n = 5-8, 12, 13) were first deposited on chemically etched SiO2/Si surfaces with native oxide layer, and then analyzed using XPS (X-ray Photoelectron Spectroscopy). After this, these samples were moved out of UHV 1 chamber, immersed into diluted sodium hydroxide solution (0.04 M NaOHH2O, except Au12 for which 0.1 M NaOHH2O was used) for 5 min.

- 195 -

Before loading these samples into the UHV 1 chamber, these samples were dried in air. After XPS analysis, these samples were moved into preparation chamber for the oxidation experiment. Fig. 9.3 shows the Au4f core level shifts of Aun (n = 5-8, 12, 13) clusters before and after immersing into NaOHH2O solution. A negative core level shift of 0.2 - 0.5 eV can be observed after immersing into NaOHH2O with respect to the 4f level of Au clusters before immersing into NaOHH2O solution. The negative chemical shifts, i.e. electron rich states are formed on the surface of Au clusters, can be interpreted by charge transfer between Au clusters and/or silica surface and NaOHH2O.
Au clusters on SiO2/Si After reacting with NaOH

Au 5

- 0.31 eV
7/2 5/2

Au 6

- 0.22 eV

Intensity (arb. units)

Au 7

- 0.20 eV

Au 8

- 0.50 eV

Au 12

- 0.39 eV

Au 13

- 0.35 eV

80

82

84

86

88

90

92

94 80

82

84

86

88

90

92

94

Binding Energy (eV)

Fig. 9.3 Au4f core level spectra of Aun (n = 5-8, 12-13) clusters before (blue) and after (red) reacting with NaOHH2O solution. Si satellite back ground peaks were subtracted.

- 196 -

One can expect two possibilities; one is charge transfer from NaOH to Au clusters, i.e. Au clusters can make an interaction directly with Na or OH species. These species can exist in form of ion, salt, and/or sodium clusters. Based on the XPS results of Au 4f (negative shift) and Na 1s core level spectra (positive shift, not shown here), one can suggest that sodium atoms (ion and/or clusters) donate electrons to Au clusters. To unravel the origin of Na species, the same experiment was carried out using chemically etched bare SiO2/Si sample without Au clusters. One can also observe Na 1s species in XPS spectra of these bare systems. However in the case of clusters, two sodium species were observed different from the bare systems. One is located at an almost similar binding energy regime to that of the bare sample, and another one is ~ 0.2 V at the higher binding energy regime with respect to that of the bare systems implying this sodium species reacts with Au clusters directly. However, one cannot fully rule out the formations of different types of sodium species such as Na4SiO4, Na2SiO3, Na2SiO3, and Na6Si2O7 under NaOHH2O solution, since sodium can react rapidly with other species to form sodium silicate. It is worth mentioning that higher oxidation conditions were used for the oxidation of Au clustersNaOH samples (typically 4.5 A and 8 x 10-5 mbar of O2 for 60 minutes), whereas typically 4.0 A was used to heating the Pt wire and Au clusters samples were exposed to 8 x 10-5 mbar of O2 for 30 minutes in oxidation experiment. That means some molecular species like Na2CO3, which can be readily formed, are surrounding the Au clusters or making a bond. However, these species do not affect the chemical properties of Au clusters. One can also expect a different type of reaction between Au cluster and NaOH like Au-OH formation. However, one cannot explain the negative chemical shift of Au4f states with this possibility. According to the previous results [264], the formation of Au-OH bond leads to charge transfer from Au clusters OH species resulting in positively charged Au clusters.

- 197 -

The other possibility for the negative chemical shifts of Au 4f states in Fig. 9.3 is a charge transfer from NaOH to silica surface, which gives rise to changes of the local work function of the surface, type of defect sites, and/or increasing the number of electron rich defect sites, forming negatively charged Au clusters. The support can influence the electronic properties of the deposited particles through specific interactions with surface sites, and in some cases enhanced catalytic reactivity has been observed. Some defects such as low coordinated ions and point defects clearly play a crucial role in stabilizing metal atoms at the metal oxide interface and ultimately influencing the catalysis of the supported metal particles [252 - 254]. Extensive studies have been devoted to better understanding the interaction between low ionization energy metals (essentially alkali metals) and oxides supports. For instance Me-MgO systems (Me = Li, Na, K, Rb, Mg) [255 - 263] have been used as promoters to modify electronic and chemical properties of metal supports. It can be expected that the degree of charge transfer which influences the binding energy of Au 4f states can be different depending on the amount of sodium and hydroxide species. However, gradual binding energy shifts of the Au 4f states to lower binding energy depending on the cluster size were not observed due to the degree of chemical shifts of Au12 cluster is not so much with respect to other clusters while it was immersed into more highly concentrated NaOHH2O solution than others by about factor two. That is, the degree of chemical shift is not related to the Au cluster size and the concentration of NaOHH2O solution. On the basis of these results and suggested possibilities, it is worth to mention that sodium species can affect electronic properties of Au clusters as well as silica surface, i.e. a fraction of the sodium species reacts with Au clusters corresponding to the amount of binding sites of Au clusters while the remaining sodium species can be adsorbed on silica surface.

- 198 -

However, one cannot explain the origin of chemical shifts shown in Fig. 9.3 precisely. To gain accurate information on the interaction between Au clusters and Na and/or OH species, more experimental and theoretical results are required.

Au 5

Au (III)

Au 6
7/2
5/2

Au (0)

Intensity (arb. units)

Au 7

Au 8

Au 12

Au 13

80

82

84

86

88

90

92

9480

82

84

86

88

90

92

94

Binding Energy (eV)

Fig. 9.4 Oxidation pattern of mass-selected Aun (n = 5-8, 12-13)NaOH cluster after exposing to atomic oxygen atmosphere. Si satellite back ground peaks were subtracted.

In order to ascertain whether Na and/or OH species can affect the chemical properties of Aun clusters with n = 5 - 8, 12, 13 on silica surface after immersing into NaOHH2O solution, these clusters were exposed to atomic and/or thermally excited molecular oxygen environments and the Au 4f spectra were collected. For the Au5, Au7, and Au13 clusters, after exposing these clusters to the oxygen atmosphere, the appearance of additional states at the higher binding energy sides, Au (III), can be found (Fig. 9.4). Simultaneously the original Au 4f peaks decrease in intensity.

- 199 -

In contrast to these clusters, Au6, Au8, and Au12 clusters do not show any change of the Au 4f state upon the oxygen exposures. As mentioned in the previous chapter, the strong even-odd patterns in oxidation property have been observed for deposited Aun clusters with n = 4 8, 12, 13 on SiO2/Si for the first time in the present work. Opposite even-odd patterns in oxidation property for deposited Aun clusters with n = 5 8, 12, 13 on SiO2/Si are observed after immersing into NaOHH2O solution which is remarkably. Fig. 9.5 shows comparison of the even-odd oxidation patterns as a function of number of Au atoms in a cluster for both systems.

1,6

Au(III)/Au(0) ratio

/2

After reacting with atomic O Au clusters on SiO2/Si Au clusters on SiO2/Si_Na

1,2

0,8

0,4

0,0

Number of Au atoms in a cluster

12

13

Fig. 9.5 Relative intensities of the oxidic and metallic peaks in Fig. 9.4 were compared with that of Au clusters before immersing into NaOHH2O solution.

Based on these results, one can claim that the electron density of Au clusters was changed by the charge transfer between Au clusters and NaOH and/or silica surface modified by sodium species, and thus, chemical properties of the deposited Au clusters are also changed, i.e. the change of chemical activities of deposited Au clusters on silica is essentially due to electronic effects rather than geometric effects. One can also probably exclude the possibility of size-dependent cluster sintering effects under our experimental conditions.

- 200 -

- 0.6 eV
1,0

after oxidation

Au4f
7/2 5/2
Pristine

Au4f
Au (III)

Intensity (arb. units)

0,8

0,6

a)
NaOHH2O

c)
Au (0)

0,4

0,2

b)
0,0 80 82 84 86 88 90 92 94 80 82 84 86 88 90

d)
92 94

Binding Energy (eV)

0,4

- 0.4 eV

after oxidation
Pristine

Intensity (arb. units)

Au4f

Au4f
7/2 5/2

a)
0,2

c)

NaOHH2O

Au (0)

b)
0,0

d)

80

82

84

86

88

90

92

94 80

82

84

86

88

90

92

94

Binding Energy (eV)

Fig. 9.6 Au4f core level spectra of Au nanoparticles with ~ 2.5 nm in diameter (upper) and Au nanoparticles with almost smallest particle size (lower); Oxidation patterns of these Au nanoparticles before and after immersing into NaOHH2O solution were exhibited.

For comparison, Au nanoparticles with ~ 2.5 nm in diameter (upper in Fig. 9.6) and Au nanoparticles with almost smallest particle size (lower in Fig. 9.6) were prepared. After immersing into NaOHH2O solution, oxidation experiment was carried out with these samples. XPS results for these experiments are shown in Fig. 9.6. Similar chemical shifts, negative core level shifts of 0.6 eV and 0.4 eV, can be observed with respect to the pristine 4f level of the Au nanoparticles, after immersing into NaOHH2O solution. One can also expect charge transfer between Au nanoparticles and NaOH and/or silica surface modified by sodium species.

- 201 -

In contrast to the results of Au clusters, however, one cannot observe an inversion in the oxidation properties. It is worth to mention that one can rule out the possibility of cluster sintering under our experimental conditions. Diverse chemical properties of mass-selected clusters are demonstrated, which cannot be observed for Au nanoparticles formed on the same substrate, emphasizing the importance of the ability to control the cluster size on an atom-by-atom basis in order to tune chemical property. In previous reports, chemical activities of Au clusters depend strongly on the support material, for example Au clusters on TiO2, and Mg(OH)2 exhibit much higher catalytic activities towards CO oxidation in comparison to those on SiO2. However, on the basis of our results, one can expect that the chemically etched silica surface can also be one of the reliable candidates as a support material, since Au clusters are very stable on these surfaces even under ambient conditions and their chemical properties can be controlled simply by modifying the electric states using alkali metals. However, the nature of chemical shifts of Au 4f state and the opposite oxidation property of Au clusters after immersing into NaOHH2O solution is still indistinct. To reveal these, more systematic theoretical and experimental efforts are required; 1) Using of different analysis tools such Cs ion scattering, SIMS (Secondary Ion Mass Spectrometer), and etc. 2) Using ammonium hydroxide solution instead of sodium hydroxide solution to verify whether the OH- species affects the electronic state of Au clusters or not. 3) Using the different type of alkali metals, due to the difference in electronegativity of these alkali metals 4) High-pressure CO-oxidation reaction with these samples. Most important, one can claim that the mass-selected Au clusters on silica are quite stable even under ambient conditions and their chemical property can be altered on an atom-by-atom basis.

- 202 -

10 Conclusions
The size-dependent electronic structure, morphology, thermal stability, metal support interaction and chemical reactivity of heterogeneous catalysts consisting of nanoparticles and mass-selected clusters have been studied extensively. Oxidation of nanoclusters is particularly interesting, since under real working conditions of heterogeneous catalysis, the formation of metal oxides is more probable than under high vacuum conditions. Metal oxides can be an either active or passive species for heterogeneously catalyzed reactions. However, we are certainly not on the way to a complete understanding of the real silver and gold catalysts, since their outstanding reactivity is probably caused by a higher complexity, e.g., the influences of particle size, support, binding properties between catalysts and reactant molecules and/or others. Therefore, deep understanding of the oxidation of nanoclusters, may allow us to distinguish between these influences and additional factors, which are related to the chemical natures of silver and gold. In the present work, we studied oxidation patterns of Ag and Au nanoparticles deposited on sputtered HOPG (Highly Ordered Pyrolytic Graphite) and chemically etched SiO2/Si surfaces with native oxide layer. And the same experiments were carried out with mass-selected Au clusters for obtaining more reliable information on cluster size effects on chemical activities. These results were characterized using Scanning Tunneling Microscopy (STM) and X-ray Photoelectron Spectroscopy (XPS). The conclusions can be summarized as follows;

- 203 -

Chapter 4; Ag nanoparticles on HOPG

Ag nanoparticles with various sizes and relatively narrow size distribution on sputtered HOPG were prepared. We found that the CO oxidation reactivity on Ag can be very size sensitive, which is due to different oxide species formed on Ag nanoparticles with various sizes. Increased chemical activities of Ag nanoparticles towards oxide formation were found in the Ag particle diameter range below 6 nm. Moreover, different Ag-oxides were found for these Ag nanoparticles, which have not been observed for the Ag bulk crystals with similar preparation methods. For the Ag nanoparticles larger than 4 nm, Ag2O/AgO can form at the initial stage of the atomic oxygen treatment, whereas for the larger oxygen exposures, Ag2O/AgO transforms into a different oxygen species most likely Ag-carbonate. However, for smaller Ag particles (< 4 nm), Ag2O/AgO cannot be detected. Among various oxygen species observed in the present work, only oxygen identified by the O 1s state at 529 eV (Ag2O/AgO) can react with CO to CO2.

Chapter 5; Au nanoparticles on HOPG

Au nanostructures with different mean size were prepared on HOPG and oxidized using atomic oxygen under ultrahigh-vacuum conditions. Depending on the size of Au nanostructures smaller than ~ 10 nm in diameter, two different oxygen species were identified in oxidized Au nanostructures, one of which can be attributed to lattice oxygen of Au oxide (Au (III)) at 529 530 eV and the other comes from subsurface oxygen atoms at 531 532 eV. Oxygen atoms on Au-oxide lattice can directly react with CO to form CO2, whereas the subsurface oxygen species do not participate in catalytic reaction. On nanostructured Au thin films, in contrast, only a single oxygen species at 529 530 eV can be observed, which is attributed to the Au-oxide species, can readily react with CO to form CO2.

- 204 -

Chapter 6; Au nanoparticles on SiO2/Si

Au nanoparticles with different sizes supported by silica surfaces were oxidized using atomic oxygen environments. We found evidence that the oxidation/reduction of Au nanoparticles on silica smaller than ~ 0.7 nm in height is much different than those of larger particles; upon atomic oxygen exposure, a characteristic feature for the formation of Au2O3 is only visible for larger particles, whereas a different oxidation pattern can be found for the smaller particles (negative shift). Almost identical oxidation/reduction was observed for the very small Au nanoparticles on HOPG surfaces. Furthermore, the oxidized larger Au nanoparticles on silica can be reduced by CO exposure at room temperature, whereas the reversible red/ox-behavior was not found for the smaller particles (~ 0.7 nm in height). We also found that the oxidation pattern of a flat Au surface is analogous to those of the larger Au nanoparticles, whereas a rough surface shows a similar oxidation pattern as those of the smaller particles. The rough Au thin film, and very small Au nanoparticles on HOPG and silica have some common behaviors regarding reaction with atomic oxygen, namely oxygen species inert (or less reactive) towards the CO oxidation. This result confirms that the increase of the number of undercoordinated atoms of smaller nanoparticles should be responsible for different oxidation patterns.

Chapter 7; Mass-selected Au clusters on HOPG

We found that the oxidation and reduction pattern of mass-selected Aun clusters with n = 2 -9 deposited on sputtered HOPG are much different from those of Au nanoparticles deposited on sputtered HOPG and silica, suggesting that experiments using mass-selected clusters can shed light onto metal-support interactions as well as size-dependent changes of chemical properties in more detail. Only Au8 clusters on sputtered HOPG show chemical activity toward the oxidation and reduction by atomic O and CO, respectively.

- 205 -

Chapter 8; Mass-selected Au clusters on SiO2/a-Si

Much different size selectivity of the oxidation of mass-selected Au clusters was observed on the silica compared to the results from sputtered HOPG, suggesting importance of metal-support interaction for chemistry of deposited clusters. Diverse chemical properties of mass-selected clusters, which cannot be observed from larger Au nanoparticles formed on the same substrate, are demonstrated; Au5, Au7 and Au13 clusters on SiO2/Si do not oxidize at all. These properties of Au clusters on silica do not change before and after exposing to air and distilled water, implying deposited Au clusters on silica are quite stable under ambient conditions. Furthermore, reverse even-odd oxidation patterns of Au clusters after immersing into NaOHH2O solution were observed. It may be worth mentioning that electronic effects of Au clusters on silica play an important role toward unexpected chemical properties. However, one cannot fully rule out the geometric effects. In the present study, we demonstrate that depending on the particle size, different oxygen species can form, which could have implications in the size dependence of heterogeneous catalysis. The nature of catalytically active sites of Ag and Au nanocatalysts for CO oxidation reaction in this experiment can be interpreted by Mar van Krevelene mechanism (consumption of lattice oxygen of oxides during reactions). Compared to Au nanoparticles, chemical properties of mass-selected Au clusters were shown to be very unique, addressing importance of the ability to control the cluster size on an atom-by-atom basis in order to tune catalytic activity. Most important, the mass-selected Au clusters on silica are quite stable even under ambient conditions and their chemical properties can be altered on an atom-by-atom basis. This kind of information can be used to find a suitable particle size for catalytic reactions.

- 206 -

11 Zusammenfassung

Die grenabhngige elektronische Struktur, Morphologie, Stabilitt, SubstratMetall Wechselwirkung bestehend und chemische aus Reaktivitt von heterogenen und Katalysatoren metallischen Nanopartikeln

massenselektierten Clustern wurden ausfhrlich untersucht. Die Oxidation von Nanoclustern ist besonders interessant, da die Bildung von Metalloxiden unter realen Arbeitsbedingungen der heterogenen Katalyse wahrscheinlicher ist als unter Hochvakuumbedingungen. Metalloxide knnen entweder eine aktive oder passive Spezies fr heterogen katalysierte Reaktionen sein. Jedoch sind wir sicher nicht auf dem Weg zu einem vollstndigen Verstndnis der realen Silberund Goldkatalysatoren, auf da deren auergewhnliche wie z. B. Reaktivitten Einflsse der wahrscheinlich komplexere Faktoren,

Teilchengre, Substrat, Bindungseigenschaften zwischen Katalysatoren und Eduktmoleklen etc., zurckzufhren sind. Daher kann ein besseres Verstndnis der Oxidation von Nanoclustern uns erlauben, zwischen diesen Einflssen und zustzlichen Faktoren zu unterscheiden, die mit den chemischen Eigenschaften von Silber und Gold zusammenhngen. In der vorliegenden Arbeit wurde das Oxidationsverhalten von Ag und Au Nanopartikeln untersucht, die auf gesputtertem HOPG (Highly Ordered
Pyrolytic Graphite) und auf chemisch getzten SiO2/Si Oberflchen mit nativer

Oxidschicht deponiert wurden. Zudem wurden die gleichen Experimente mit massenselektierten Au Clustern ausgefhrt, um zuverlssigere Informationen ber den Einfluss der Clustergre auf die chemische Aktivitt zu erhalten. Die Ergebnisse wurden mittels STM (Scanning Tunneling Microscopy) und XPS (X-ray Photoelectron Spectroscopy) gewonnen. Die Schlussfolgerungen knnen wie folgt zusammengefasst werden:

- 207 -

Kapitel 4: Ag Nanopartikel auf HOPG

Ag Nanopartikel mit verschiedenen Gren und verhltnismig schmaler Grenverteilung wurden auf gesputtertem HOPG hergestellt. Wir fanden, dass die Reaktivitt fr die CO-Oxidation von Ag sehr grenempfindlich sein kann, was an den unterschiedlichen Oxidsorten liegt, die auf Ag Nanopartikeln mit verschiedenen Gren gebildet werden. Erhhte chemische Aktivitten der Ag Nanopartikel bei der Oxidbildung wurden fr einen Grenbereich der Ag Partikel von unter 6 nm Durchmesser gefunden. Zudem wurden fr diese Ag Nanopartikel unterschiedliche Ag-Oxide gefunden, die im Fall von Ag Festkrperkristallen bei hnlichen Vorbereitungsmethoden nicht beobachtet wurden. Fr Ag Nanopartikel mit einer Gre von mehr als 4 nm kann sich zu Beginn der Behandlung mit atomarem Sauerstoff Ag2O/AgO bilden, whrend sich bei lngerem Sauerstoffangebot Ag2O/AgO in eine andere Oxidart, hchstwahrscheinlich Ag Karbonat, umwandelt. Fr kleinere Ag Partikel (< 4 nm) konnte jedoch Ag2O/AgO nicht festgestellt werden. Von den verschiedenen Sauerstoffarten, die in der vorliegenden Arbeit beobachtet wurden, kann nur die Sorte Sauerstoff mit CO zu CO2 reagieren, die sich durch den O 1s Zustand bei 529 eV (Ag2O/AgO) auszeichnet.

Kapitel 5: Au Nanopartikel auf HOPG

Au Nanopartikel mit unterschiedlichen Durchschnittsgren wurden auf HOPG hergestellt und mit atomarem Sauerstoff unter Ultrahochvakuumbedingungen oxidiert. Fr Au Nanopartikel mit einem Durchmesser von weniger als ~10 nm konnten im Fall der oxidierten Au Nanopartikel in Abhngigkeit von der Gre zwei unterschiedliche Sauerstoffarten identifiziert werden: Die eine Art bei 529-530 eV kann dem Gittersauerstoff des Au Oxides (Au (III)) zugeordnet werden und die andere bei 531-532 eV kommt von Sauerstoffatomen unter der Oberflche. Sauerstoffatome auf dem Au-Oxid Gitter knnen direkt mit CO reagieren um CO2 zu bilden, whrend die unter der Oberflche liegende

- 208 -

Sauerstoffart nicht an der katalytischen Reaktion teilnimmt. Dagegen kann auf nanostrukturierten dnnen Au-Filmen nur eine einzelne Sauerstoffsorte bei 529530 eV beobachtet werden, die Au-Oxid zugeordnet wird und bereitwillig mit CO zu CO2 reagiert.

Kapitel 6: Au Nanopartikel auf SiO2/Si

Au

Nanopartikel

mit

unterschiedlichen

Gren

auf

einer

Siliziumoxidoberflche wurden mit atomarem Sauerstoff oxidiert. Wir fanden Belege, dass die Oxidation/Reduktion von Au Nanopartikeln auf Siliziumoxid mit einer Hhe kleiner als ~ 0,7 nm anders als die der greren Partikel abluft: Nach dem Anbieten von atomarem Sauerstoff ist die Bildung von Au2O3 nur fr grere Partikel sichtbar, whrend bei kleineren Partikeln ein anderes Oxidationsverhalten gefunden werden kann (negative Verschiebung). Ein fast identisches Oxidations-/Reduktionsverhalten wurde fr die sehr kleinen Au Nanopartikel auf HOPG Oberflchen beobachtet. Darber hinaus knnen die oxidierten greren Au Nanopartikel auf Siliziumoxid durch Anbieten von CO bei Raumtemperatur reduziert werden, whrend das reversible Red/OxVerhalten nicht fr die kleineren Partikel (~ 0,7 nm Hhe) gefunden werden konnte. Wir fanden auch heraus, dass das Oxidationsverhalten einer flachen Au Oberflche anolog zu dem der greren Partikel ist, whrend das Oxidationsverhalten einer rauen Oberflche dem der kleineren Partikel hnelt. So haben raue Au Dnnfilme und sehr kleine Au Nanopartikel auf HOPG und Siliziumoxid einige gemeinsame Verhaltensweisen bei der Reaktion mit atomarem Sauerstoff, nmlich Sauerstoffspezies, die inert (oder weniger reaktiv) gegenber der CO Oxidation sind. Dieses Resultat besttigt, dass Zunahme der Zahl von unterkoordinierten Atomen bei kleineren Nanopartikeln fr das unterschiedliche Oxidationsverhalten verantwortlich sein soll.

- 209 -

Kapitel 7: Massenselektierte Au Cluster auf HOPG

Wir fanden, dass das Oxidations- und Reduktionsverhalten massenselektierter Aun Cluster mit n = 2 - 9, die auf gesputtertem HOPG deponiert wurden, sehr verschieden ist von dem von Au Nanopartikeln auf gesputterten HOPG und Siliziumoxid. Dies zeigt, dass Experimente mit massenselektierten Clustern detaillierteren Aufschluss ber Metall-Substrat Wechselwirkungen als auch grenabhngige nderungen der chemischen Eigenschaften geben knnen. Nur Au8 Cluster zeigen auf gesputtertem HOPG eine chemische Aktivitt gegenber Oxidation durch atomaren Sauerstoff und Reduktion durch CO.

Kapitel 8: Massenselektierte Au Cluster auf SiO2/a-Si

Verglichen mit den Resultaten fr gesputtertes HOPG wurde im Falle von Siliziumoxid eine deutlich andere Grenabhngigkeit des Oxidationsverhaltens von massenselektierten Au Clustern beobachtet. Dies unterstreicht die Bedeutung der Metall-Substrat Chemie deponierter Cluster. Verschiedene Wechselwirkung fr die chemische Eigenschaften

massenselektierter Cluster werden aufgezeigt, die bei greren Nanopartikeln auf demselben Substrat nicht beobachtet werden knnen: Au5, Au7 und Au13 Cluster oxidieren auf SiO2/Si berhaupt nicht. Diese Eigenschaften der Au Cluster auf Siliziumoxid verndern sich nicht vor und nach dem Aussetzen an Luft und destilliertem Wasser. Dies impliziert, dass die auf Siliziumoxid deponierten Au Cluster unter den Bedingungen unseres Experimentes unter Raumbedingungen ziemlich stabil sind. Des Weiteren wurde ein umgekehrtes gerade-ungerade Oxidationsverhalten beobachtet, nachdem die Au Cluster in NaOHH2O Lsung getaucht wurden. Es drfte erwhnenswert sein, dass elektronische Effekte der Au Cluster auf Siliziumoxid eine wichtige Rolle fr die unerwarteten chemischen Eigenschaften spielen. Jedoch knnen die geometrischen Effekte nicht vllig vernachlssigt werden.

- 210 -

In der vorliegenden Arbeit zeigen wir, dass sich in Abhngigkeit von der Partikelgre unterschiedliche Sauerstoffarten bilden knnen, die Implikationen fr die Grenabhngigkeit der heterogenen Katalyse haben knnten. Die Art der katalytisch aktiven Pltze von Ag und Au Nanokatalysatoren fr die CO Oxidation in diesem Experiment kann durch den Mar von Krevelene Mechanismus verstanden werden (Verbrauch des Gittersauerstoffs der Oxide whrend der Reaktion). Verglichen mit Au Nanopartikeln sind die chemischen Eigenschaften massenselektierter Au Cluster sehr einzigartig. Dies unterstreicht die Mglichkeit mittels einer atomgenauen Kontrolle der Clustergre die katalytische Aktivitt steuern zu knnen. Das Wichtigste ist, dass die massenselektierten Au Cluster auf Siliziumoxid sogar unter Raumbedingungen ziemlich stabil sind und dass ihre chemischen Eigenschaften auf einer atomaren Skala gendert werden knnen. Diese Information kann weiter verwendet werden, um eine fr katalytische Reaktionen geeignete Partikelgre zu finden.

- 211 -

A List of Figures

Fig. 1.1

Activation energies and their relationship to an active and selective

catalyst (A) Reactant; (B) desired product; (C) undesired product; Ehom, activation barrier for the homogeneous reaction; Ecat, activation barrier with use of catalyst; Hr, change in enthalpy of reactants compared with product [41].......................................................................................................................3 Fig. 1.2 The rich oxygen chemistry of Ru (ruthenium) (0001) [42]..................6 Fig. 2.1 CO oxidation reaction pathways of the Au on TiO2 catalyst suggested by Haruta et al. [88]...........................................................................................16 Fig. 2.2 A representation of the oxidation of carbon monoxide at the periphery of an active gold particle on metal oxide support [26]......................................18 Fig. 2.3 Step density of hemispherical Au as a function of particle size [102] ............................................................................................................................19 Fig. 2.4 Binding energy of different oxygen species, versus gas-phase O2, plotted with respect to the coordination number of the Au atoms of the site. Yellow spheres (Au), red (O), and gray (C) [79]...............................................20 Fig. 2.5 Schematic energy diagram for a heterogeneously catalysed CO oxidation (Langmuir-Hinshelwood reaction) [192]...........................................21 Fig. 2.6 Calculated energy diagram for the Eley-Rideal reaction mechanism of CO oxidation on Ru (0001); the transition state geometry is indicated in the inset; large (Ru), medium (O), and small (C) circles [106] ..............................22

- 212 -

Fig. 2.7 Correlation between the catalytic activity and the percentage / surface concentration of cationic and zerovalent gold [94]...........................................24 Fig. 2.8 The rate of the water-gas shift reaction is not affected by leaching of metallic gold nanoparticles by a NaCN solution, indicating that the particles are just spectators in this reaction. The active site involves supported cationic gold (Au+, red spheres; the exact charge is not known) [135].........................25 Fig. 2.9 Activity for CO oxidation at room temperature as a function of Au coverage above the monolayer TiOx on Mo(112) [45]......................................25 Fig. 2.10 Structure of the peroxo-type CO-O2 complex (left) and final state (right) for CO oxidation on the 4/3 layer Au/TiO-Mo(112) surface [85]..........26 Fig. 2.11 CO oxidation turnover frequencies (TOF) (a) at 300 K [139] and 350 K [43] as a function of the average particle size of the Au cluster on TiO2 support................................................................................................................29 Fig. 2.12 Size dependence of CO oxidation activity [94]..................................29 Fig. 2.13 Atomic configurations displaying several stages in the simulation of the coadsorption of H2O and O2 on the top facet of a Au8 cluster supported on MgO(100), and the subsequent reaction with gaseous CO to form CO2. Yellow (Au), red (O), white (H), green (Mg), and aquamarine (C) [120].....................30 Fig. 3.1 Schematic drawing of the UHV 1 system (upper) and real image of the cluster deposition and XPS analysis chambers (lower).....................................32 Fig. 3.2 Structure of the magnetron sputter source (upper) and the arrangement of skimmers in different pumping stages for the extraction of cluster ions from the source into the next vacuum system (lower) [189].......................................35

- 213 -

Fig. 3.3 Schematic drawing of the ion optics for soft-landing of the size-selected cluster ions onto the surface [189].....................................................................37 Fig. 3.4 Diagram of the interaction between energetic clusters of atoms and metal substrates [163]....................................................................................38 Fig. 3.5 Three dimensional drawing of the UHV 2 system (upper) and real image of the STM stage (lower) [190]...............................................................39 Fig. 3.6 "Universal curve" of mean free path () versus kinetic energy (KE) (eV) of electrons in different materials [182, 183] ((IMFP is average distance between inelastic collisions ()).........................................................................40 Fig. 3.7 Schematic energy diagram of X-ray photoelectron spectra. The measured kinetic energy of the ejected electron is given by E*k [190]..............42 Fig. 3.8 Emission process based on the Koopmans theorem [191].................42 Fig. 3.9 Survey XPS spectrum with steplike background [191]........................44 Fig. 3.10 Spin-orbit coupling leads to the splitting of the Au4f photoemission peak into two peaks (l;orbital angular momentum, s;spin angular momentum) [191]...................................................................................................................45 Fig. 3.11 a) the chemical shift in binding energy of the Si 2p line for elemental Si and SiOx [185] and b) carbon 1s chemical shifts in ethyl trifluoroacetate. The four carbon lines correspond to the four carbon atoms within the molecule [186]...................................................................................................................46 Fig. 3.12 Multiplet splitting of the Fe 3s lines [191].........................................48

- 214 -

Fig. 3.13 Energy loss lines associated with the 2s line of aluminium [187] ............................................................................................................................49 Fig. 3.14 Schematic drawing of a Concentric Hemispherical Analyzer (CHA) in XPS [188]...........................................................................................................50 Fig. 3.15 XPS C1s peaks as a function of the pass energy................................51 Fig. 3.16 XPS intensity ration of signals from particles and supports, Ip / Is reflects the dispersion of the particles over the support [184]..........................53 Fig. 3.17 Main principle of scanning tunneling microscopy [181]...................54 Fig. 3.18 STM image (topography) of the graphite surface (upper) [180] and scanning tunneling spectroscopy (STS) result of a Bi2Sr2CaCu2O8 sample (lower) [181]......................................................................................................55 Fig. 3.19 Scheme of the experimental procedure related to mass-selected Au clusters in UHV 1 and UHV 2............................................................................57 Fig. 3.20 Scheme of the experimental procedure related to Ag and Au nanoparticles in UHV 1 and UHV 2; Type 1.....................................................59 Fig. 3.21 Scheme of the experimental procedure related to Ag and Au nanoparticles in UHV 1 and UHV 2; Type 2.....................................................61 Fig.4.1 a) STM image of Ag nanoparticles on sputtered HOPG (5 sec. sputtering and 8 min. Ag evaporation), b) particle diameter distribution, c) particle height distribution, (135.3 nm x 132.3 nm. -3.0 V, 0.7 nA) [212]...................................................................................................................64

- 215 -

Fig. 4.2 a) STM image of Ag nanoparticles on sputtered HOPG (5 sec. sputtering and 30 min. evaporation), b) Particle diameter distribution, c) Particle height distribution, (214.5 nm x 214.5 nm. -1.5 V, 1.0 nA) [212]...................................................................................................................65 Fig. 4.3 a) STM image of Ag nanoparticles on sputtered HOPG (2 sec. sputtering and 60 minutes Ag evaporation). b) Particle diameter distribution c) Particle height distribution. (154.9 nm x 154.9 nm. -2.0 V, 2.4 nA) [212]...................................................................................................................66 Fig. 4.4 a) STM image of Ag nanoparticles on sputtered HOPG (2 sec. sputtering and 90 min. evaporation), b) Particle diameter distribution c) Particle height distribution, (394.4 nm x 349.8 nm. -1.9V, 2.1 nA) [212]...................................................................................................................67 Fig. 4.5 Changes of the Ag 3d states (Binding Energy and Full Width of Half Maximum) as a function of particle size on HOPG [212] ................................69 Fig. 4.6 Oxidation of Ag bulk crystals using atomic oxygen; a) Ag 3d core level shifts b) O 1s states. The O 1s state at 529 eV can be attributed to the Ag2O formation. Atomic oxygen exposure is increasing from bottom to up................71 Fig. 4.7 Changes of the Ag 3d core levels of the Ag nanoparticles with a mean size of 1, 3, and 4 nm as a function of atomic oxygen exposure [213]..............74 Fig. 4.8 Changes of the Ag 3d core levels of the Ag nanoparticles with a mean size of 5 nm and 6 nm as a function of atomic oxygen exposure [213]..............75 Fig. 4.9 Changes of the Ag 3d states of Ag nanoparticles with various sizes upon exposing to atomic oxygen [213] .............................................................75

- 216 -

Fig. 4.10 O 1s spectra from a pure HOPG sample obtained after exposing to various amounts of atomic oxygen [214]...........................................................76 Fig. 4.11 O 1s spectra from 1 nm-sized Ag nanoparticles obtained after exposing to various amounts of atomic oxygen [213]........................................77 Fig. 4.12 O 1s spectra from 3 nm-sized Ag nanoparticles on HOPG obtained after exposing to various amounts of atomic oxygen [213]...............................77 Fig. 4.13 O 1s spectra from 4 nm-sized Ag nanoparticles on HOPG obtained after exposing to various amounts of atomic oxygen [213]...............................78 Fig. 4.14 O 1s spectra from 5 nm-sized Ag nanoparticles on HOPG obtained after exposing to various amounts of atomic oxygen [213]...............................78 Fig. 4.15 O 1s spectra from 6 nm-sized Ag nanoparticles on HOPG obtained after exposing to various amounts of atomic oxygen [213]...............................79 Fig. 4.16 Oxidation of Ag nanoparticles with a mean size of a) 1 nm b) 3 nm c) 4 nm d) 5 nm e) 6 nm..........................................................80

Fig. 4.17 a) C 1s specta of Ag (6nm) on HOPG: before (circle) and after (square) oxidation. O 1s shows a strong peak at 531 eV, which we assign to Agoxide, b) The valence band spectrum of the Ag/HOPG sample (a) and the HOPG sample (b) were subtracted from that of O/Ag/HOPG (O exposure time 70 minutes) and that of O/HOPG, respectively [220] .......................................85 Fig. 4.18 a) Ag 3d and b) O 1s spectra of an Ag polycrystalline sample, which was oxidized using atomic oxygen......................................................................86

- 217 -

Fig. 4.19 a) Ag 3d and b) O 1s core level spectra from Ag nanoparticles with a mean size of 4 nm after exposing to atomic oxygen and subsequent reduction by CO at room temperature [213]. The O/HOPG background was not subtracted from the O/Ag/HOPG in O1s spectra.................................................................89 Fig. 4.20 a) Ag 3d and b) O 1s core level spectra from Ag nanoparticles with a mean size of 3 nm after exposing to atomic oxygen and subsequent reduction by CO at room temperature [213]. Oxygen peak at 533-534 eV are attributed to the O/HOPG.......................................................................................................90 Fig. 4.21 Ag 3d core level spectra from Ag nanoparticles with a mean size of 6 nm after exposing to atomic oxygen and subsequent reduction by CO at room temperature [213]. The reduced particles can be oxidized using atomic oxygen and reduced again using CO, i.e. the oxidation and reduction cycles is reversible. The O/HOPG background was not subtracted from the O/Ag/HOPG in O1s spectra.....................................................................................................92 Fig. 4.22 O 1s core level spectra from Ag nanoparticles with a mean size of 6 nm after exposing to atomic oxygen and subsequent reduction by CO at room temperature [213]..............................................................................................93 Fig. 5.1 XPS Au 4f level shifts as a function of particle size [190]...................96 Fig. 5.2 a) STM image of Au nanostructures on sputtered HOPG (10 sec. sputtering and 2 min. Au evaporation),300 nm x 300 nm, -5.3 V, 0.08 nA, Sample A [190]...................................................................................................97 Fig. 5.3 a) STM image of Au nanostructures on sputtered HOPG (5 sec. sputtering and 5 min. Au evaporation), 256 nm x 256 nm, -2 V, 0.08 nA, Sample B [190]...............................................................................................................98

- 218 -

Fig. 5.4 a) STM image of Au nanostructures on sputtered HOPG (3 sec. sputtering and 15 min. Au evaporation 220 nm x 220 nm, -1.2 V, 1.3 nA, Sample C [190]..................................................................................................99 Fig. 5.5 O1s XPS spectra of oxidized various Au nanostructures on HOPG. (before subtraction of the O/HOPG signals (red zone) from O/Au/HOPG) a) O1s peak for sample A (~ 5 nm in diameter), b) sample B (~ 8 nm in diameter) and c) sample C (~ 23 nm in diameter)...........................................102 Fig. 5.6 Oxidation of various Au nanostructures studied using XPS (O 1s states) [214] a) O1s peak for sample A (~ 5 nm in diameter), b) sample B (~ 8 nm in diameter) and c) sample C (~ 23 nm in diameter).................................104 Fig. 5.7 Oxidation of various Au nanostructures studied using XPS (Au 4f states) [214] a) Au4f peak for sample A (~ 5 nm in diameter), b) sample B (~ 8 nm in diameter) and c) sample C (~ 23 nm in diameter).................................105 Fig. 5.8 Reduction of oxidized Au nanostructures (O 1s states) [214] a) O1s peak for sample A (~ 5 nm in diameter), b) sample B (~ 8 nm in diameter) and c) sample C (~ 23 nm in diameter)...........................................108 Fig. 5.9 Reduction of oxidized Au nanostructures (Au 4f states) [214] a) Au4f peak for sample A (~ 5 nm in diameter), b) sample B (~ 8 nm in diameter) and c) sample C (~ 23 nm in diameter)...........................................109 Fig. 5.10 Au 4f XPS core level spectra: a) After the Au nanostructure was oxidized by atomic O, b) after reduction by CO and c) after reduced Au nanostructurse by CO were oxidized by molecular O (~ 400000 L) again ..........................................................................................................................111 Fig. 5.11 O1s XPS core level spectra: before (black curve) and after (red curve) heating the reduced using CO Au nanostructure..................................112

- 219 -

Fig. 6.1 a) XPS spectra of Au particles grown on SiO2/Si, b) binding energy (Au4f7/2) shifts [190] ........................................................................................115 Fig. 6.2 a) STM image of Au nanoparticles on SiO2/Si (corresponding to the XPS spectra a) in Fig. 6.1), b) Particle diameter distribution c) Particle height distribution; Tunneling parameters (193 nm x 193 nm, 2 V, 0.1 nA) [190, 216] ......................................................................................................................116 Fig. 6.3 a) STM image of Au nanoparticles on SiO2/Si (corresponding to the XPS spectra b) in Fig. 6.1), b) Particle diameter distribution c) Particle height distribution; Tunneling parameters (124 nm x 124 nm, 4.7 V, 0.1 nA) ..........................................................................................................................117 Fig. 6.4 a) STM image of Au nanoparticles on SiO2/Si (corresponding to the XPS spectra e) in Fig. 6.1), b) Particle diameter distribution c) Particle height distribution; Tunneling parameters (123 nm x 123 nm, 4.7 V, 0.1 nA) ..........................................................................................................................118 Fig. 6.5 Left panel; XPS spectra of the Au 4f states for the Au particles grown on SiO2/Si are displayed [216]. From a) to g), the Au coverage is reduced, i.e. the average particle size decreases from a) to g). For a), b), and e), the average particle heights (H) and widths (W) estimated on the basis of the STM data in Figs. 6.2 - 6.4 are approximately given. Right panel; XPS spectra of the Au 4f states of the samples in the left panel after exposing them to the atomic oxygen atmosphere. The shoulders at the higher binding energy region correspond to the formation of Au2O3. (marked by the yellow bar) .......................................120 Fig. 6.6 a), b): XPS Au 4f spectra (samples g) and c) in Fig 6.1), c): XPS Au4f spectra of the Au nanoparticles on HOPG (<< 3 nm in diameter) were taken before (black) and after (red) atomic oxygen exposure [216] ........................122 Fig. 6.7 XPS O 1s spectra of the Au nanoparticles on HOPG exposed to atomic oxygen and subsequently to CO [216] ............................................................123

- 220 -

Fig. 6.8 a), b): XPS Au4f spectra (samples g) and c) in Fig 6.1) and c): XPS Au4f spectra of the Au nanoparticles on HOPG (<< 3 nm in diameter) exposed to an atomic oxygen atmosphere (red) and further exposed to 3000 L of CO (blue) [216] ..........................................................................................125 Fig. 6.9 XPS survey spectra of Au thin films before and after annealing at various temperatures. Annealing time was 30 minutes for each experiment [211] ................................................................................................................128 Fig. 6.10 XPS spectra of Au thin films before and after annealing at various temperatures (narrow scan). a) Au 4f, b) O 1s, c) Si 2p levels. Spectra are displayed after background subtraction [211] ................................................129 Fig. 6.11 STM pictures of Au thin films before (a: 0.2 V, 0.6 nA, 238 nm x 238 nm), and after annealing at 200 C (b: 2V, 1.7 nA, 297 nm x 297 nm) and 500 C (c: 4.2 V, 0.5 nA, 250 nm x 250 nm). For a) and b) the same height contrast (0 ~ 12 nm) was used for display, whereas for c) 0 ~ 19 nm was used [211] ..........................................................................................................................133 Fig. 6.12 XPS spectra of Au thin films on Si heated at 500 C before (black) and after ex-situ STM experiments (red). a) C 1s, b) O 1s, c) Si 2p, and d) Au 4f. Spectra are displayed after background subtraction [211] .......................135 Fig. 6.13 XPS spectra of the defective Au thin films before and after exposing to 400000 L of O2 at room temperature; a) Au 4f, b) O 1s, and c) Si 2p. Spectra are displayed after background subtraction [211] .........................................137 Fig. 6.14 XPS Au 4f (a and b) and O 1s (c and d) spectra of the as grown flat Au thin film on SiO2/Si and the rough Au surface annealed at 200oC on SiO2/Si; before (black) and after (red) exposing to atomic oxygen at room temperature [218] ................................................................................................................139

- 221 -

Fig. 6.15 a) Relatively flat and b) rough Au surface structures on SiO2/Si exposed to the atomic oxygen atmosphere are reduced by CO. The O 1s spectra are taken before and after CO exposures [218] ..............................................140 Fig. 7.1 Mass-spectrum of Au cluster anions created by the magnetron-sputtersource...............................................................................................................143 Fig. 7.2 STM image (35 x 35 , It; 0.44 nA and Vt; -9.8 mV) of two types of defects on the HOPG surface; a) dome-like defect and b) a distorted atomic structure defect) [322]. (Low-energetic Ar+ ions; 20 s, 150 eV acceleration voltage) ............................................................................................................145 Fig. 7.3 STM images of Au7 deposited on a) sputter-damaged HOPG (150 nm x 150 nm, 1.61 v, 0.7 nA), b) non-sputtered HOPG. (12.9 nm x 12.9 nm, 1.67 v, 0.3 nA) [221] ...................................................................................................146 Fig. 7.4 STM images of Au7 deposited on non-sputtered HOPG a) 2D STM image b) 3D image (50 nm x 50 nm, 1.67 v, 0.7 nA) .......................................147 Fig. 7.5 STM images of Au7 deposited on non-sputtered HOPG a) 2D STM image b) 3D image (50 nm x 50 nm, 1.67 v, 0.7 nA) c) Line profiles of each clusters on a) ...................................................................................................148 Fig. 7.6 The average cluster heights of Aun (n=1-8) on the TiO2 (110)-(1x1) surface. The dashed lines indicate heights expected for various gold layers in the cluster [291] ..............................................................................................149 Fig. 7.7 Au 4f states of the deposited Au clusters on sputter-damaged HOPG; a) Au 4f states of Aun with n=2-9 and b) Summary of the Au4f
7/2

binding

energies as a function of cluster size [221] .....................................................151

- 222 -

Fig. 7.8 Au 4f states of the deposited Aun (n = 2, 3, 4 and 9) clusters on sputterdamaged HOPG. Each spectrum was fitted by 4 Gaussian functions. The table shows the relative intensity ratio [(b) / (a)] of each spectrum. .......................153 Fig. 7.9 Au 4f spectra of the deposited Au2 and Au3 clusters after atomic/excited molecular oxygen exposure; before (black-bottom) and after atomic/excited molecular oxygen exposure (red-upper) [221] ...............................................155 Fig. 7.10 Au 4f spectra of the deposited Au4 clusters after atomic/excited molecular oxygen exposure; before (black-bottom) and after atomic/excited molecular oxygen exposure (red-upper) [221] ...............................................156 Fig. 7.11 Au 4f spectra of the deposited Aun with n = 5 - 7 and 9 clusters after atomic/excited molecular oxygen exposure; before (black-bottom) and after atomic/excited molecular oxygen exposure (red-upper) [221] .......................156 Fig. 7.12 Au 4f spectra of the deposited Au8 clusters after atomic/excited molecular oxygen exposure; before (black-bottom) and after atomic/excited molecular oxygen exposure (red-upper) [221] ...............................................157 Fig. 7.13 Au 4f spectra of the deposited Au clusters with n = (2 - 9) after atomic/excited molecular oxygen exposure; after O (red-bottom) and after CO (blue-upper) .....................................................................................................158 Fig. 7.14 O 1s spectra of the deposited Au clusters with n = (2 - 9) after atomic/excited molecular oxygen exposure (line with circles), and after subsequent CO exposure (solid) [221] ............................................................159 Fig. 7.15 O 1s spectra of the deposited Aun clusters with n = (3, 7, 9) after atomic/excited molecular oxygen exposure (a), and after subsequent CO exposure (b) .....................................................................................................160

- 223 -

Fig. 7.16 O 1s spectra of the deposited Aun clusters with n = (2, 6) after atomic/excited molecular oxygen exposure (a), and after subsequent CO exposure (b) .....................................................................................................161 Fig. 7.17 O 1s spectra of the deposited Au4 clusters after atomic/excited molecular oxygen exposure (a), and after subsequent CO exposure (b).........161 Fig. 7.18 O 1s spectra of the deposited Au5 clusters after atomic/excited molecular oxygen exposure (a) and after subsequent CO exposure (b) (c); O 1s peaks intensity at 531 eV are compared. .........................................162 Fig. 7.19 O 1s spectra of the deposited Au8 clusters after atomic/excited molecular oxygen exposure (a), and after subsequent CO exposure (b)..163 Fig. 7.20 O 1s spectra of Au nanoparticles with a laterial size of 2 - 3 nm (a) and deposited Au8 clusters (b) on HOPG after atomic/excited molecular oxygen exposure (red line), and after subsequent CO exposure (blue line).................164 Fig. 8.1 XPS survey spectra of Au clusters deposited on SiO2/Si surface....167 Fig. 8.2 Au4f XPS spectra of various Au clusters on SiO2/Si; typically Au4f7/2 at 84.0 eV, Au4f5/2 at 87.6 and 2 Si satellite peaks at 87.7 eV and 89.7 eV .....168 Fig. 8.3 Au 4f7/2 intensity ratio corresponding to the Au 4f XPS spectra in Fig. 8.2 168 Fig. 8.4 Raw XPS spectra of Au5 clusters on SiO2/Si fitted by 7 gauss functions [217] ................................................................................................................169 Fig. 8.5 Au 4f level spectra of deposited Au clusters after subtraction of the silicon background...........................................................................................170

- 224 -

Fig. 8.6 6 XPS Si2p spectra after exposing the samples to the oxygen atmosphere with a hot Pt-filament, no change of the Si-signal can be found. The peak at ~103 - 104 eV corresponds to the Si (IV) (oxide layer), whereas the peak at ~ 99 eV comes from Si (0) ...................................................................172 Fig. 8.7 The Au4f XPS spectra of the Au7 cluster deposited on pre-heated SiO2/Si at different heating temperature conditions; a) at 250oC and b) at 500oC................................................................................................................175 Fig. 8.8 The Au4f XPS spectra of the Au7 cluster deposited on pre-heated SiO2/Si at different heating temperature; after exposing to oxygen with a hot Ptfilament; a) at 250oC and b) at 500oC.............................................................175 Fig. 8.9 XPS Au 4f7/2 core level shifts as a function of number of Au atoms in a Au cluster on SiO2/Si [217; until Au10] ...........................................................177 Fig. 8.10 Au 4f binding energy shifts isolated gold nanoparticles or clusters. The solid line represents 1/d behaviour [342] ................................................178 Fig. 8.11 A plot of the 4f7/2 binding energy shifts of the Au (filled squares), Pt (open squares) clusters in ref. [337 - 338], and our data (green block) relative to the bulk values for each metal cluster as a function of cluster size. Au and Pt clusters were deposited onto Si (100) wafers with natural oxide layer ..........................................................................................................................179 Fig. 8.12 The geometric structure of the Si-Au complex formed on Si dangling bonds (Si), Si-O-Au complex formed on nonbridging oxygen(Si-O) and silanolate(Si-O-) defects, and Si-Au-Si complex formed on oxygen vacancy(Si-Si) [339] ...................................................................................180 Fig. 8.13 Raw Spectra of the Au 4f levels of deposited clusters before (black) and after (red) exposing to oxygen [217; until Au10] ......................................183

- 225 -

Fig. 8.14 XPS spectra of the deposited Au clusters consisting of 2 - 13 atoms after exposing to the atomic oxygen atmosphere. Each spectrum was originally fitted with 7 Gauss functions. Three of them corresponding to the Si satellite features are removed, and only the Au peaks are shown [217; until Au10] ..........................................................................................................................184 Fig. 8.15 Relative intensities of the oxidic and metallic peaks in Fig. 8.14; Au 4f
7/2

state was used for this quantitative analysis [217; until Au10] ................185

Fig. 8.16 Raw spectra of the Au 4f levels of deposited clusters before (red) and after (blue) exposing to 3000 L of CO [217; until Au10] .................................188 Fig. 8.17 The XPS spectra of the deposited Au clusters first exposed to oxygen atmosphere with a hot Pt-filament and then exposed to 3000 L of CO at room temperature (after subtracting the Si satellite backgrounds in Fig. 8.16) [217; until Au10] ........................................................................................................189 Fig. 8.18 Au6 clusters on SiO2/Si undergo reversible oxidation and reduction cycle by reacting with atomic oxygen and CO atmosphere.............................190 Fig. 8.19 Au6 clusters on SiO2/Si undergo reversible oxidation and reduction cycle by reacting with atomic oxygen and CO atmosphere (after subtracting the Si satellite backgrounds in Fig. 8.18) ..............................................................190 Fig. 8.20 Relative intensity ratio of the oxidic and metallic peaks of Au6 clusters on SiO2/Si in Fig. 8.19......................................................................................191 Fig. 9.1 XPS Au 4f core level spectra; comparative studies on oxidation properties of Au5 and Au8 clusters before and after exposing to distilled water. Si satellite back ground peaks were subtracted. .............................................194

- 226 -

Fig. 9.2 Scheme of the experimental procedure...............................................195 Fig. 9.3 Au4f core level spectra of Aun (n=5-8, 12-13) clusters before (blue) and after (red) reacting with NaOHH2O solution. Si satellite back ground peaks were subtracted. ....................................................................................196 Fig. 9.4 Oxidation pattern of mass-selected Aun (n=5-8, 12-13)NaOH cluster after exposing to atomic oxygen atmosphere. Si satellite back ground peaks were subtracted. ..............................................................................................199 Fig. 9.5 Relative intensities of the oxidic and metallic peaks in Fig. 9.4 were compared with that of Au clusters before immersing into NaOHH2O solution. ..........................................................................................................................200 Fig. 9.6 Au4f core level spectra of Au nanoparticles with ~ 2.5 nm in diameter (upper) and Au nanoparticles with almost smallest particle size (lower); Oxidation patterns of these Au nanoparticles before and after immersing into NaOHH2O solution were exhibited. ...............................................................201

Table 1.1 Ionization potentials and electron affinities of silver and gold [63] ..............................................................................................................................8 Table 4.1 STM results (in fig. 4.1-4.4) of Ag nanoparticles on sputtered HOPG surface. The coverage was estimated with a combined STM and XPS analysis [190] ..................................................................................................................63 Table 4.2 Ag particles on HOPG with various sizes were analyzed with XPS after the final stage of the oxidation experiments in the present work [220] ............................................................................................................................82

- 227 -

Table 5.1 STM results of Au nanoparticles on a sputtered HOPG surface. The coverage was estimated with a combined STM and XPS [190] ......................100 Table 6.1 Analysis of the XPS results of Fig. 6.10. A Gaussian function was used for fitting each spectrum. ........................................................................131 Table 7.1 The difference between Au clusters and Au nanoparticles suggested by Haruta [309] ...............................................................................................142 Table 8.1 Analytical techniques to measure the oxide thickness [343] ..........171

- 228 -

B References
[1] Center for responsible nanotechnology (http://www.crnano.org/whatis.htm) [2] Nanotechnology: Assessing the Environmental Risks for Australia, Earth Policy Centre (2006) [3] V.A. Bondzie, S.C. Parker, C.T. Campbell, J. Vac. Sci. Technol., A 17, 1717 (1999) [4] Nanotechnology: Assessing the Environmental Risks for Australia [5] Clean Diesel from Coal MIT Technology Review (2006) [6] Safety of sunscreens containing nanoparticles of zinc oxide or titanium dioxide, TGA (2006) [7] http://en.wikipedia.org/wiki/Nanotechnology [8] Atom cluster Te McGraw-Hill Encyclopedia of Science & Technology, 9th. Edition, (2002) [9] C. A. Mirkin, R. L. Letsinger, R. C. Mucic, and J. J. Storhoff, Nature, 382, 607 (1996) [10] M. P. A. Viegers and J. M. Trooster, Phys. Rev. B 15, 72 (1977) [11] T. Sato, H. Ahmed, D. Brown, and B. F. G. Johnson, J. Appl. Phys. 82, 696 (1996) [12] D. M-K. Lam, B. W. Rossiter, Sci. Am., 265, 80 (1991) [13] Leung C, Xirouchaki C, Berovic N and Palmer, Adv. Mater., 16, 223 (2004) [14] Liang Z J, Susha A and Caruso F. Chem. Mater., 15 3176 (2003) [15] C. B. Murray, S. Sun, H. Doyle, T. Betley, Mater. Res. Soc. Bull., 26, 985 (2001) [16] P. V. Kamat et.al., J. Phys. Chem. B, 106, 7729 (2002) [17] R.E. Palmer, S. Pratontep, and H.-G. Boyen, Nat. Mater., 2, 44(2003) [18] C. T. Campbell and M. T. Paffett, Surf. Sci. 139, 396 (1984)

- 229 -

[19] R. B. Grant and R. M. Lambert, J. Catal. 92, 364 (1985) [20] R. A. van Santen and C. P. M. de Groot, J. Catal., 98, 530 (1986) [21] R. A. van Santen and H. P. C. E. Kuipers, Adv. Catal., 35, 265 (1987) [22] M. A. Barteu and R. J. Madix, in The Chemical Physics of Solid Surfaces and Heterogeneous Catalysis, edited by D. A. King and D. P. Woodruff (Elsevier, Amsterdam, 1982), vol 4, Chap. 4. [23] S. Hawker, C. Mukoid, J. P. S. Badyal, and R. M. Lambert, Surf. Sci., 219, L615 (1989) [24] D. Thompson, Gold Bulletin, 32(1), 12-19 (1999) [25] R. Grisel, K.J. Westrate, A. Gluhoi, and B.E. Nieuwenhuys: Gold Bulletin, 35(2), 39-45 (2002) [26] D.T. Thompson and G.C. Bond: Gold Bulletin, 33(3), 41-51 (2000) [27] M. Haruta, T. Kobayashi, H. Sano, and M. Yamada: Chemistry Letters, 405-408 (1987) [28] D.A.H. Cunningham, W. Vogel, R.M. Torres- Sanchez, K. Tanaka, and M. Haruta: J. Catalysis, 183, 24-31 (1999) [29] D. Andreeva: Gold Bulletin, 35(3), 82-88 (2002) [30] K. Tanaka, T. Akita, D.A.H. Cunningham, S. Tsobota, et al.: Structural analyses and characterization of gold catalysts, Report of the Osaka National Research Institute, No.393, Osaka, Japan, pp.11-35 (1999) [31] G.J. Hutchings: J. Catalysis, 96, 292- 295 (1985) [32] L. Prati and G. Martra: Gold Bulletin, 32(3), 96-101 (1999) [33] Hughes, M. D. et al. Nature 437, 11321135 (2005) [34] Bond, G. C. & Thompson, D. T. Catal. Rev. Sci. Eng., 41, 319388 (1999) [35] Prati, L. & Rossi, M. J. Catal., 176, 552560 (1998) [36] Hayashi, T., Tanaka, K. & Haruta, M. J. Catal., 178, 566575 (1998) [37] Xu, Y.-J. et al. Catal. Lett., 101, 175179 (2005)

- 230 -

[38] Haruta. M. Date.M, Applied Catalysis A. General, 222, 1, 427-437 (2001) [39] Jacob I Kleiman, Protection of Space Materials from the Space Enviroment. Editor Springer, (2001) [40] K.J. Laidler, The World of Physical Chemistry (oxford university press, oxford, 1993) [41] Surface Science, Foundations of Catalysis and Nanoscience, Kurt W. Kolasinski, JOHN WILEY & SONS, LTD, (2001) [42] H. Over and A. P. Seitsonen, science, 297, 20 (2002) [43] M. Valden, X. Lai, D.W. Goodman, Science, 281, 1647-1650 (1998) [44] H.-G Boyen, G. Kstle, F. Weigl, B. Koslowski, C. Dietrich, P. Ziemann, J.P. Spatz, S. Riethmller, C. Hartmann, M. Mller, G. Schmid, M.G. Garnier, P. Oelhafen, Science, 297, 1533-1536 (2002) [45] M.S. Chen, D.W. Goodman, Science, 306, 352-355 (2004) [46] B. Yoon, H. Hkkinen, U. Landman, J. Phys. Chem. A, 107, 40664071 (2003) [47] M. Haruta, S. Tsubota, T. Kobayashi, H. Kageyama, M.J. Genet, B. Delmon, J. Catal., 144, 175-192 (1993) [48] R. Meyer, C. Lemire, Sh. K. Shaikhutdinov, H.-J. Freund, Gold Bulletin, 37, 72-124 (2004) [49] M. Rodorova et al., phys. Rev. Lett., 89, 096103 (2002) [50] Y.D. Kim, H.Over, G.Krabbes, G.Ertl, Top. Catal., 14, 95 (2001) [51] H. Over et al., Surf. Sci., 515, 143, (2002) [52] K. Christmannm Introduction to Surface Physical Chemistry, K. (Steinkopff-Verlag Darmstadt, Springer-Verlag, New York, 1999). [53] N.D.S. Canning, D. Outka, R.J. Madix, Surf. Sci., 141, 240 (1984) [54] D.A. Outka and R.J. Madix, Surf. Sci., 179, 351 (1986) [55] J.M. Gottfried, N. Elghobashi, S.L.M. Schroeder, K. Christmann, Surf. Sci., 523, 89 (2003)

- 231 -

[56] B. Koslowski, H.-G. Boyen, C. Wilderotter, G. Kstle, P. Ziemann, R. Wahrenberg, P. Oelhafen, Surf. Sci., 475, 1 (2001) [57] H. Ron and I. Rubinstein, Langmuir 10, 4566 (1994) [58] J. Wang and B.E. Koel, Surface Sci., 436, 15 (1999) [59] D.H. Parker and B.E. Koel, J. Vac. Sci. Technol., A 8, 2585 (1990) [60] N. Saliba, D. H. Parker, B. E. Koel, Surf. Sci., 410, 270 (1998) [61] J.M. Gottfried, K.J. Schmidt, S.L.M. Schroeder, K. Christmann, Surf. Sci., 511, 65 (2002) [62] www.webelements.com [63] N. Bartlett, Gold Bull, 31, 22 (1998) [64] Holleman-Wiberg, Lehrbuch der Anorganischen Chemie, 91.-100. Aufl., de Gruyter, Berlin (1985) [65] H. Over, A.P. Seitsonen, Science, 297, 2003-2004 (2002) [66] J. Amann, D. Crihan, M. Knapp, E. Lundgren, E. Lffler, M. Muhler, V. Narkhede, H. Over, M. Schmid, A.P. Seitsonen, P. Varga, Angew. Chem., 117, 939-942 (2005) [67] A. Cho, Science, 299, 1684-1685 (2003) [68] R. Meyer, C. Lemire, Sh. K. Shaikhutdinov, H. J. Freund, Gold Bulletin, 37, 72-124 (2004) [69] Stephen J. Roberts, Gary Pattrick, Elma van der Lingen : Project AuTEK, Mintek, Private Bag X3015, Randburg, 2125, South Africa [70] B. Yoon, H. Hkkinen, U. Landman, A.S. Wrz, J. Antonietti, S. Abbet, J. Judai, U. Heiz, Science, 307, 403-407 (2005) [71] A. Sanchez, S. Abbet, U. Heiz, W.-D. Schneider, H. Hkkinen, R.N. Barnett, U. Landman, J. Phys. Chem. A, 103, 9573-9578 (1999) [72] S. Lee, C. Fan, T. Wu, S.L. Anderson, J. Am. Chem. Soc., 126, 5682-5683 (2004) Mills, G.; Gordon, M. S.; Metiu, H., J. Chem. Phys., 118, 4198 (2003) [74] M. Haruta, N. Yamada, T. Kobayashi, S. Iijima, J. Catal., 115, 301 (1989)

- 232 -

[75] D. Stolcic, M. Fischer, G. Gantefr, Y.D. Kim, Q Sun, P. Jena, J. Am. Chem. Soc., 125, 2848-2849 (2003) [76] B.E. Salisbury, W.T. Wallace, R.L. Whetten. Chem. Phys., 262, 131-141 (2000) [77] Laurent D.Menard, Fengting Xu, Ralph G.Nuzzo and Judith C. Yang, Journal of catalysis, 243,1,64-73 (2006) [78] M. Haruta, H.Sano, and T. Kobayasi: Method for manufacture of catalyst composite having gold or mixture of gold with catalytic metal oxide deposited on carrier, US Patent 4698324 (1987) [79] Lopez, N.; Nrskov, J. K. J. Am. Chem. Soc., 124, 11262 (2002) [80] Molina. L. M.; Hammer, B. Appl. Catal., A, 291, 21 (2005) [81] Remediakis, I. N.; Lopez, N.; Nrskov, J. K. Angew. Chem., Int. Ed., 44, 1824 (2005) [82] Liu, Z.-P.; Gong, X.-Q.; Kohanoff, J.; Sanchez, C.; Hu, P. Phys. ReV. Lett., 91, 266102 (2003) [83] V. Valden, S. Pak, X. Lai, D.W. Goodman, Catal. Lett., 56, 7 (1998) [84] Chen, M. S.; Cai, Y.; Yan, Z.; Goodman, D. W., J. Am. Chem. Soc., 128, 6341 (2006) [85] Norge Cruz Hernandez, Javier Fdez. Sanz,, and Jose A. Rodriguez. J. Am. Chem. Soc., 128, 15600 (2006) [86] Huber, H.; McIntosh, D.; Ozin, G. A. Inorg. Chem., 16, 975 (1997) [87] Stiehl, J. D.; Gong, J.; Ojifinni, R. A.; Kim, T. S.; McClure, S. M.; Mullins, C. B. J. Phys. Chem. B, 110, 20337 (2006) [88] M. Haruta, CATTECH, 6, 102 (2002) [89] M. Haruta, J. New materials for electrochemical systems 7, 163-172 (2004) [90] Y. Xu. M. Mavrikakis, J. Phys. Chem. B, 107, 9298 (2003) [91] A. Wolf and F. Schth, Applied Catal. A, General, 226, 1-13 (2002)

- 233 -

[92] M.M. Schubert, S. Hackenberg, A.C. van Veen, M.Muhler, V.Plzak, and R.J. Behm, J. Catalysis, 197, 113-122 (2001) [93] U. Heiz, A. Sanchez, S. Abbet, W.D. Schneider, Chemical Physics, 262, 189-200 (2000) [94] J. Guzman and B.C. Gates, J. Am. Chem. Soc., 126, 2672 (2004) [95] L. Fu, N.Q. Wu, J.H. Yang, F.Qu, D.L. Jonson, M.C. Kung, H.H. Jung, and V.P. Dravid, J. Phys. Chem. B, 109, 3704 (2005) [96] S. Laursen and S. Linic, PRL 97, 026101 (2006) [97] J.G. wang and B. Hammer, PRL 97, 136107 (2006) [98] J. Guzman and B.C. Gates, J. Phys. Chem. B 106, 7659 (2002) [99] C.K. Costello, M.C. Kung, H.S. Oh, Y. Wang, and H.H. Kung, Appl. Catal., A 232, 159 (2002) [100] Molina. L. M et al., Phys. Rev. Lett., 90, 206102 (2003) [101] Molina. L. M et al., J. Chem. Phys., 120, 7673 (2004) [102] M. Mavrikakis et al., Catal. Lett., 64, 101 (2000) [103] G. Mills, M.S. Gordon, and H. Metiu, Chemical Physics Letters, 359, 493-499 (2002) [104] H.S. Oh, J.H. Yang, C.H. Costello, Y.M. Wang, S.R. Bare, H.H. Jung, and M.C. Kung: J. of Catalysis, 210, 375-386 (2002) [105] R. J. Baxter and P. Hu, journal of chemical physics, 116, 11, 4379-4381 (2002) [106] C. Stampfl and M. Scheffler, Surface Science, 433-435, 119-126 (1999) [107] Outka, D. A.; Madix, R. J. Surf. Sci., 1987, 179, 351 () [108] Gottfried, J. M.; Christmann, K., Surf. Sci., 566, 1112 (2004) [109] Bondzie, V. A.; Parker, S. C.; Campbell, C. T., Catal. Lett., 63, 143 (1999) [110] Stiehl, J. D.; Kim, T. S.; Reeves, C. T.; Meyer, R. J.; Mullins, C. B. J. Phys. Chem. B, 108, 7917 (2004)

- 234 -

[111] Kim, T. S.; Stiehl, J. D.; Reeves, C. T.; Meyer, R. J., Mullins, C. B. J. Am. Chem. Soc., 125, 2018 (2003) [112] Liu, Z.-P.; Hu, P.; Alavi, A. J. Am. Chem. Soc., 124, 14770 (2002) [113] Kim, Y. D.; Fischer, M.; Ganterfor, G. Chem. Phys. Lett., 377, 170 (2003) [114] Socaciu, L. D.; Hagen, J.; Bernhardt, T. M.; Wste, L.; Heiz, U.; Hkkinen, H.; Landman, U. J. Am. Chem. Soc., 125, 10437 (2003) [115] J.-D. Grunwaldt, A. Baiker, J. Phys. Chem. B, 103, 1002 (1999) [116] C. Lemire, R. Meyer, S. Shaikhutdinov, H. J. Freund, Angew. Chem. Int. Ed., 43, 118 (2004) [117] N. Lopez, T. V. W. Janssens, B. S. Clausen, Y. Xu, M. Mavrikakis, T. Bligaard, and J. K. Nrskov, J. Catal., 223, 232 (2004) [118] Rodolfo Zanella, Suzanne Giorgio, Chae-Ho Shin, Claude R. Henry and Catherine Louis, J. Catal., 222, 357 (2004) [119] Q. Fu, H. Saltsburg, M. Flytzani-Stephanopoulos, Science, 301,935 (2003) [120] Angelo Bongiorno and Uzi Landman, PRL, 95, 106102 (2005) [121] M. Haruta, in Catalytic Science and Technology, edited by S. Yoshida, N. Takezawa, and T. Ono et al., (Kodansha, Tokyo, 1991), Vol. 1, p. 331. [122] M. Date and M. Haruta, J. Catal., 201, 221 (2001) [123] M. Date et al., Catal Today, 72, 89 (2002) [124] M. Date et al., Angew. Chem. Int. Ed., 43, 2129 (2004) [125] H. H. Kung et al., J. Catal., 216, 425 (2003). [126] C.Lemire, R. Meyer, Sh.K. Shaikhutdinov, H.-J. Freund, surface science, 552, 27-34 (2004) [127] S.R. Bahn, N. Lopez, J.K. Nrskov, K.W. Jacobsen, Phys. Rev. B 66, 081405 (2002)

- 235 -

[128] S. Shaikhutdinov, R. Meyer, M. Naschitzki, M. Baumer, H.-J. Freund, Catal. Lett., 86, 211 (2003) [129] James D. Stiehl, Tae S. Kim, Sean M. McClure, and C. Buddie Mullins, J. Am. Chem. Soc., 126, 13574-13575 (2004) [130] James D. Stiehl, Tae S. Kim, Sean M. McClure, and C. Buddie Mullins, J. Am. Chem. Soc., 126, 1606-1607 (2004) [131] Okomura, M., Kitagawa, Y., Haruta, M., Yamagishi, K. Chem.Phys.Lett, 346, 163 (2001) [132] Okomura, M., Coronado, J.M., Soria,J., Haruta, M., Conesa, J.C., J. Catal., 203,168 (2001) [133] Liu, H., Kozlov, A.I., Kozlova,A.P.,Shido,T.,Asakura,K.,Iwasawa, Y., J.Catal.,185,252 (1999) [134] J. Guzman, B.C. Gates, Angew. Chem. Int. Ed. 42, 690, (2003) [135] Robert J. Davis, science, 301 926-927 (2003) [136] Campbell, C. T., Science, 306, 234 (2004) [137] Goodman, D. W., J. Catal., 216, 213 (2003) [138] J. J. Pietron, R. M. Stroud, and D. R. Rolison, Nano Lett., 2, 545 (2002) [139] G.R. Bamwenda, S.Tsubota, T.Nakamura, M.Haruta, Catal. Lett., 44, 83, (1997) [140] Sungsik Lee, Chaoyang Fan, Tianpin Wu, and Scott L. Anderson, J. Am. Chem. Soc., 126, 5682-5683 (2004) [141] S.D. Lin, M. Bollinger, and M.A. Vannice, Catal. Lett., 17, 245 (1993) [142] S. Arrii, F. Morfin, A.J. Renouprez, and J.L. Rousset., J. Am. Chem. Soc., 126, 1199 (2004) [143] Xingyi Deng, Byung Koun Min, Amado Guloy, and Cynthia M. Friend, J. Am. Chem. Soc., 127, 9267-9270 (2005) [144] Sault, A.G., Madix, R.J., Campbell, C.T., Surf. Sci., 169, 347 (1986) [145] Norbert Weiher, Angela M. Beesley, Nikolaos Tsapatsaris, Laurent Delannoy, Catherine Louis, Jeroen A. van Bokhoven, and Sven L. M. Schroeder, J. Am. Chem. Soc., 129(8); 2240-2241 (2007)

- 236 -

[146] Volkmar Zielasek, Birte Jrgens, Christian Schulz, Jrgen Biener, Monika M. Biener, Alex V. Hamza, and Marcus Bumer, Angew. Chem. Int. Ed., 45, 8241 8244 (2006) [147] Caixia Xu, Jixin Su, Xiaohong Xu, Pengpeng Liu, Hongjuan Zhao, Fang Tian, and Yi Ding, J. AM. CHEM. SOC., 129, 42-43 (2007) [148] F. E. Wagner, S. Galvagno, C. Milone, A. M. Visco, L. Stievano, S. Calogero, J. Chem. Soc. Faraday Trans., 93, 3403 (1997) [149] Y. Kobayashi, S. Nasu, S. Tsubota, M. Haruta, Hyperfine Interactions, 126, 95 (2000) [150] H. Kageyama, N. Kamijo, T. Kobayashi, M. Haruta, Physica B,158, 183 (1989) [151] Z. Hao, L. An, H. Wang, T. Hu, React. Kinet. Catal. Lett., 70, 153 (2000) [152] F.Boccuzzi, A. Chiorino, M. Manzoli, P. Lu, T. Akita, S. Ichikawa, M. Haruta, J. Cata., 202, 256 (2001) [153] S. Tsubota, D. A. H. Cunningham, Y. Bando, M. Haruta, Stud. Surf. Sci. Catal., 91, 227 (1995) [154] D. Horvth, L. Toth, L. Guczi, Catal. Lett., 67, 117 (2000) [155] S. Minic, S. Scir, C. Crisafulli, A. M. Visco, S. Galvagno, Catal. Lett., 47, 273 (1997) [156] Gottfried, J. M.; Schmidt, K. J.; Schroeder, S. L. M.; Christmann, K. Surf. Sci., 525, 197 (2003) [157] X.M. Lin, H.M. Jaeger, C. Sorensen, K. Klabunde, J. Phys. Chem. B, 105, 3353 (2001) [158] Jana, N.R.; Gearheart, L.; Murphy, C.J. J.Phys.Chem.B, 105, 4065 (2001) [159] Walker, C.H.; St. John, J.V., Wisian-Neilson, P. J. Am. Chem. Soc., 123, 3846 (2001) [160] M.-H. Schaffner, J.-F. Jeanneret, F. Patthey, W.-D. Schneider, Appl. Phys. 31, 3177 (1998) [161] A. Perez, L. Bardotti, B. Prevel, P. Jensen, M. Treilleux, P. Melinon, J. Gierak, G. Faini, D. Mailly, New J. Phys. 4, 76 (2002)

- 237 -

[162] W. Harbich, in Metal Clusters at Surfaces, edited by H. H. Meiwes-Broer (Springer, Berlin, 2000), 108150 [163] H. Hsieh, R. S. Averback, H. Sellers, and C. P. Flynn, Phys. Rev. B, 45, 4417 (1992) [164] Y. Yamamura and T. Muramoto, Radiat. Eff. 130131, 225 (1994) [165] M. Chatelet, A. De Martino, J. Pettersson, F. Pradere, and H. Vach, Chem. Phys. Lett., 196, 563 (1992) [166] J. B. C. Pettersson and N. Markovic, Chem. Phys. Lett., 201, 421 (1993) [167] H. P. Cheng and U. Landman, J. Phys. Chem., 98, 3527 (1994) [168] D. J. Kenny, R. E. Palmer, C. F. Sanz-Navarro, and R. Smith, J. Phys. Condens. Matter, 14, L185 (2002) [169] J. G. Partridge, S. A. Brown, A. D. F. Dunbar, M. Kaufmann, S. Scott, M. Schulze, R. Reichel, C. Seigert, and R. Blaikie, Nanotechnology 15, 1382 (2004) [170] Yu Chen, Richard E. Palmer, and Jess P. Wilcoxon, Langmuir, 22, 28512855 (2006) [171] G. Binning, H. Rohrer, Ch. Gerber, E. Weibel, Phys. Rev. Lett., 49, 57-61 (1982) [172] G. Binnig, H. Rohrer, C. Gerber, and E. Weibel., Applied Physics Letters, 40, 178, 180 (1982) [173] M. C. Desjonqures and D. Spanjaard, Concepts in Surface Physics, 2nd Edition, Springer, Berlin, (1996) [174] J. Venerables, Introduction to Surface and Thin Film processes, Cambridge University press, (2001) [175] R. Wiesendanger, "Scanning Probe Microscopy and Spectroscopy: Methods and Applications", Cambridge University Press, (1998) [176] D. Bonnell, "Scanning Probe Microscopy and Spectroscopy: Theory, Techniques, and Applications", 2nd, Wiley-VCH, New York, (2001) [177] H.-J. Guentherodt, R. Wiesendanger, "Scanning Tunneling Microscopy", Vol. I, II, and III, Springer, (1993, 1995, 1996)

- 238 -

[178] http://www.almaden.ibm.com/vis/stm/ [179] G.A.Somorjai, Chemistry in Two Dimensions, Surfaces, Cornell University Press, Ithaca (1981) [180] Kugler et al., Rev. Sci. Instrum., 71, 1475 (2000) [181] Hoogenboom et al., Physica C, 332, 440 (2000) [182] T.E. Madey and J.T.Yates, J. Vac. Sci. Techn., 8, 525 (1971) [183] G. Ertl, K. Kppers, Low Energy Electrons and Surface Chemistry, VCH Verlagsgesellschaft, Weinheim, (1985) [184] J.W. Niemantsverdriet, Spectroscopy in Catalysis, VCH (1993) [185] F.J. Himpsel et al., Phys. Rev. B, 38, 6084 (1988) [186] L. C. Feldman and J. W. Mayer, Fundamentals of surface and thin film analysis, Elsevier Science Publishing, (1986) [187] H. Ibach, Topics in Current Physics 4: Electron spectroscopy for Surface Analysis. Springer Verlag, Berlin, Heidelberg, New York (1997) [188] Omicron, EA 125 Energy Analyzer, Technical Reference Manual (1997) [189] Sauerstoffadsorption an freien und deponierten Clustern, GynzRekowski, Felix von, Universitt konstanz, Dissertation (2005) [190] Electronic and Geometric Properties of Silver and Gold Nanoparticles, Universitt konstanz, Ignacio Lopez Salido, Dissertaion (2007) [191] www.cem.msu.edu/~cem924sg/Topic09.pdf [192] CO oxidation over Gold, Freien Universitt Berlin, Jrg Michael Gottfried, Dissertation (2003) [193] Over, H.; Kim, Y.D.; Seitsonen, A.P.; Wendt, S.; Lundgren, E.; Schmid, M.; Varga, P.; Morgante, A.; Ertl, G. Science, 287,1474-1476 (2000) [194] Goodman, D.W.; Peden, C.H.F. J. Phys. Chem., 90, 4839-4843 (1986) [195] V.E. Henrich and P.A. Cox, The Surface Science of Metal Oxide (Cambridge Univ. Press. Cambridge (1996)

- 239 -

[196] Bukhtiyarov, V.I.; Hvecker, M.; Kaichev, V.V.; Knop-Gericke, A.; Mayer, R.W.; Schgl, R. Phys. Rev. B, 67, 235422-1-12 (2003) [197] Li, W.-X.; Stampfl, C.; Scheffler, M. Phys. Rev. Lett., 90, 256102-1-4 (2003) [198] Li, W.-X.; Stampfl, C.; Scheffler, M. Phys. Rev. B, 67, 045408-1-16 (2003) [199] van Santen, R.A.; de Groot, C.P.M. J. Catal., 98, 530-539 (1986) [200] Bocquet, M.; Michaelides, A.; Loffreda, D., Sautet, P.; Alavi, A., King. D.A., J. Am . Chem. Soc., 125, 5620-5621(2003) [201] de Oliveira, A.L.; Wolf, A.; Schth, F. Catal. Lett., 73, 157-160 (2001) [202] Bukhtiyarov, V.I.; Carley, A.F.; Dollar, L.A.; Roberts, M.W. Surf. Sci. Lett., 381, L605-L608 (1997) [203] Bukhtiyarov, V.I.; Kaichev, V.V. J. Mol. Catal. A, 158, 167-172 (2000) [204] Verykios, X.E.; Stein, F.P.; Couglin, R.W. J. Catal., 66, 368-382 (1980) [205] L. Guczi, D. Horvath, Z. Paszti, L. Toth, Z.E. Horvath, A. Karacs, G. Peto, J. Phys. Chem. B 104, 3183 (2000) [206] Nordan & Holman, Industrial Biotechnology, Vol 1 No 3, p 146 (2005) [207] R.M. Finch, N.A. Hodge, G.J. Hutschings, A. Meagher, Q.A. Pankhurst, M.R.H. Siddiqui, F.E. Wagner, R. Whyman, Phys. Chem. Chem. Phys., 1, 485 (1999) [208] A.M. Visco, F. Neri, G. Neri, A. Donato, C. Milone, S. Galvagno, Phys. Chem. Chem. Phys., 1, 2869 (1999) [209] E.D. Park, J.S. Lee, J. Catal., 186, 1, (1999) [210] M.L. Kimble, A.W. Castleman, Jr., R. Mitric, C. Burgel, V. BonacicKoutecky, J. Am. Chem. Soc., 126, 2526 (2004) [211] Defect formation of Au thin films on SiO2/Si upon annealing; D. CHAN LIM, I. LOPEZ-SALIDO, R. DIETSCHE and Y. DOK KIM, Philosophical Magazine, 85, 29, 3477348 (2005)

- 240 -

[212] Ag nanoparticles on highly ordered pyrolytic graphite (HOPG) surfaces studied using STM and XPS; Ignacio Lopez-Salido, Dong Chan Lim and Young Dok Kim, Surface Science, 588, 1-3, 6-18 (2005) [213] Size selectivity for CO-oxidation of Ag nanoparticles on highly ordered pyrolytic graphite (HOPG); Dong Chan Lim, Ignacio Lopez-Salido and Young Dok Kim, Surface Science, 598, 1-3, 96-103 (2005) [214] Oxidation of Au nanoparticles on HOPG using atomic oxygen; Dong Chan Lim, Ignacio Lopez-Salido, Rainer Dietsche, Moritz Bubek, Young Dok Kim, Surface Science, 600, 3, 507-513 (2006) [215] Electronic and geometric properties of Au nanostructures on HOPG studied using XPS and STM; Ignacio Lopez-Salido, Dong Chan Lim*, Rainer Dietsche, Nils Betram, and Young Dok Kim*, Journal of physical chemistry B, 110, 3, 1128-1136 (2006) [216] Size-selectivity of the oxidation behaviors of Au nanoparticles; Dong Chan Lim, Ignacho Lopez-Salido, Rainer Dietsche, Moritz Bubek, Young Dok Kim, Angewandte Chemie International Edition, 45, 15, 2413-2415 (2006) [217] Oxidation and Reduction of Mass-selected Au clusters (Aun, n=2-10) Deposited on SiO2/Si : every atom counts; Dong Chan Lim, Rainer Dietsche, Moritz Bubek, Gerd Gantefr, Young Dok Kim, ChemPhysChem, 7 (9), 19091911 (2006) [218] Oxidation and reduction of rough and flat Au surfaces; Dong Chan Lim, Young Dok Kim, Applied surface science, 253, 5, 2984-2987 (2006) [219] Electronic and chemical properties of supported Au nanoparticles; Dong Chan Lim, Ignacio Lopez-Salido, Rainer Dietsche, Moritz Bubek, Young Dok Kim, Chemical physics, 330, 3, 441-448 (2006) [220] Interaction of Ag on sputtered HOPG with atomic oxygen: an example of strong metal-support interaction?; S.H. Jung, D.C. Lim, J.-H. Boo, S.B. Lee, A.N. Hwang, C.G. Hwang, Y. D. Kim_accepted Applied catalysis A: General, (2007) [221] Chemistry of mass-selected Au clusters deposited on sputter-damaged HOPG surfaces: the unique properties of Au8 clusters; Dong Chan Lim, Rainer Dietsche, Moritz Bubek, Thorsten Ketterer, Gerd Gantefr, Young Dok Kim, Chemical Physics Letters, 2007_accepted [222] H. Hvel, Th. Becker, A. Bettac, B. Reihl, M. Tschudy, E.J. Williams, J. Appl. Phys., 81, 154 (1997)

- 241 -

[223] A. Rettenberger, P. Bruker, M. Metzler, F. Mugele, Th.W. Matthes, M. Bhmisch, J. Boneberg, K. Friemelt, P. Leiderer, Surf. Sci. 402404 (1998) 409. Also Ph.D. Thesis J. Zimmermann, University of Konstanz (2004) [224] K. Luo, T.P.St. Clair, X. Lai, D.W. Goodman, J. Phys. Chem. B, 104, 3050 (2000) [225] K. Luo, D.Y. Kim, D.W. Goodman, J. Mol. Catal. A, 167, 191 (2001) [226] J.A. Rodriguez, M. Kuhn, J. Hrbek, J. Phys. Chem., 100, 18240 (1996) [227] W. Eberhardt, P. Fayet, D.M. Cox, Z. Fu, A. Kaldor, R. Sherwood, D. Sondericker, Phys. Rev. Lett., 64, 780 (1990) [228] A. Tanaka, Y. Takeda, M. Imamura, S. Sato, Phys. Rev. B, 68, 195415 (2003) [229] J. Radnik, C. Mohr, P. Claus, Phys. Chem. Chem. Phys., 5, 172 (2003) [230] P. Zhang, T.K. Sham, Phys. Rev. Lett., 90, 245502 (2003) [231] G.K. Wetheim, S.B. DiCenzo, S.E. Youngquist, Phys. Rev. Lett., 51, 2310 (1983) [232] V. Vijayakrishnan, A. Chainani, D.D. Sarma, C.N.R. Rao, J. Phys. Chem., 96, 8679 (1992) [233] B. Richter, H. Kuhlenbeck, J.-J. Freund, P. Bagus, Phys. Rev. Lett., 93, 026805 (2004) [234] A. Sandell, J. Libuda, P.A. Bru hlwiler, S. Andersson, A.J. Maxwell, M. Bumer, N. Martensson, H.J. Freund, J. Vac. Sci. Technol. A, 14 1546 (1996) [235] X. Bao, M. Muhler, Th. Schedel-Niedrig, R. Scho gl, Phys. Rev. B, 54, 2249 (1996) [236] A.J. Nagy, G. Mestl, and R. Schloelgl, J. Catal., 188, 58 (1999) [237] J. S. Hammond, S. W. Gaarenstroom, Chem. Phys. Lett., 60, 459 (1979) [238] J. F. Moulder, W. F. Stickle, P. E. Sobol, and K. D. Bomben, Handbook of X-ray Photoelectron Spectroscopy, edited by J. Chastain (Perkin-Elmer, Eden Prairie, 1992) [239] L.H. Tjeng, M.B.J. Meinders, J. van Elp, J. Gihjsen, and G.A. Sawatzky, Phys. Rev. B, 41, 3190 (1990)

- 242 -

[240] G.B. Hoflund, J.F. Weaver, and W.S. Epling, Surf. Sci. Spec., 3,163 (1995) [241] G.B. Hoflund, J.F. Weaver, and W.S. Epling, Surf. Sci. Spec., 3, 157 (1995) [242] G.I.N. Waterhouse, G.A. Bowmaker, and J.B. Metson, Appl. Surf. Sci., 183,191 (2001) [243] J. He, I. Ichinose, S. Fujikawa, T. Kunitake, and A. Nakao, A. Chem. Comm., 17, 1910 (2000) [244] http://www.xpsdata.com/HB1_demo.pdf [245] A. Majid, F. Bensebaa, P. LEcuyer, G. Pleizier, and Y. Deslandes, Rev. Adv. Mater. Sci., 4, 25 (2003) [246] A.I. Boronin, S.V. Koscheev, K.T. Murzakhmetov, V.I. Avdeev, and G.M. Zhidomirov, Appl. Surf. Sci., 165, 9 (2000) [247] L. Li, and C. Yang, Materials at High Temperatures, 20, 601 (2000) [248] Handbook of X-ray Photoelectron Spectroscopy, J.F. Moulder, W.F. Stickle, P.E. Sobol, K.D. Bomben, Physical Electronics, Inc. USA (1995) [249] A. Bttcher, H. Niehus, S. Schwegmann, H. Over and G. Ertl, J. Phys. Chem., 101, 11185 (1997) [250] P. Mars and D.W. van Krevelen, Chem. Eng. Sci., 3, 41 (1954) [251] M, Haruta, Gold Bulletin, 37, 1-2 (2004) [252] Abbet, S.; Sanchez, A.; Heiz, U.; Schneider, W.-D.; Ferrari, A. M.; Pacchioni, G.; Rosch, N. J. Am. Chem. Soc., 122, 3453 (2000) [253] Ferrari, A. M.; Giordano, L.; Rosch, N.; Heiz, U.; Abbet, S.; Sanchez, A.; Pacchioni, G. J. Phys. Chem. B, 104, 10612 (2000) [254] Heiz, U.; Sanchez, A.; Abbet, S.; Schneider, W.-D. J. Am. Chem. Soc., 121, 3214 (1999) [255] Kasai, P. H.; Bishop, R. J., Jr. J. Phys Chem., 77, 2308 (1973) [256] Edwards, P. P.; Harrison, M. R.; Klinowski, J.; Ramadas, S.; Thomas, J. M.; Johnson, D. C. J. Chem. Soc. Chem. Commun., 982 (1984)

- 243 -

[257] Edwards, P. P.; Woodall, L. J.; Anderson, P. A.; Armstrong, A. R.; Slaski, M. Chem. Soc. ReV., 305 (1993) [258] Zecchina, A.; Scarano, D.; Marchese, L.; Coluccia, S.; Giamello, E. Surf. Sci., 194, 513 (1988) [259] Giamello, E.; Ferrero, A.; Coluccia, S.; Zecchina, A. J. Phys. Chem., 95, 9385 (1991) [260] Giamello, E.; Murphy, D. M.; Ravera, L.; Coluccia, S.; Zecchina, A. J. Chem. Soc., Faraday Trans., 90, 3167 (1994) [261] Murphy, D. M.; Giamello, E. J. Phys. Chem., 98, 7929 (1994) [262] Murphy, D. M.; Giamello, E.; Zecchina, A. J. Phys. Chem., 97, 1739 (1993) [263] Murphy, D. M.; Giamello, E. J. Phys. Chem., 99, 15180 (1995) [264] William T. Wallace, Richard B. Wyrwas, Robert L. Whetten, Roland Mitric, and Vlasta Bonacic-Koutecky, J. AM. CHEM. SOC., 125, 8408-8414 (2003) [265] A. Krozer, M. Rodahl, J. Vac. Sci. Technol. A, 15, 1704 (1997) [266]Y.-J. Zhu, A. Schnieders, J.D. Alexander, T.P. Beebe Jr., Langmuir, 18, 5728 (2002) [267] J.M. Gottfried, K.J. Schmidt, S.L.M. Schroeder, K. Christmann, Surf. Sci., 525, 184 (2003) [268] Luo, K.; Lai, X.; Yi, C.-W.; Davis, K.A.; Gath, K.K.; Goodman, D.W. J. Phys. Chem. B, 109, 4064-4068 (2005) [269] L. Braicovich, C.M. garner, P.R. Skeath, C.Y. Su, P.W. Chye, I. Lindau, and W.E. Spicer, Phys.Rev. B, 20 5131 (1979) [270] C. Grupp, and A. Taleb-Ibrahimi, Phys. Rev. B, 57 6258 (1998) [271] J.S. Custer, M.O. Thompson, D.C. Jacobson, and J.M. Poate, Phys. Rev. B, 44 8744 (1991) [272] R.S. Bauer, R.Z. Bachrach, and L.J. Brillson, Appl. Phys. Lett., 37 1006 (1980) [273] F.H. Baumann, and W. Schrter, Phys. Rev. B, 43 6510 (1991)

- 244 -

[274] E. Landree, D. Grozea, C. Collazo-Davila, and L.D. Marks, Phys. Rev. B, 557910 (1997) [275] J.L Plaza, S. Jacke, Y. Chen, and R.E. Palmer, Phil. Mag., 83 1137 (2003) [276] R. L. Whetten, D. M. Cox, D. J. Trevor, A. Kaldor, Phys. Rev. Lett., 541494 (1985) [277] H. P. Steinrck, F. Pesty, L. Zhang, T. E. Madey, Phys. Rev. B, 51 2427 (1995) [278] L-. Zhang, R. Prsaud, T. E. Madey, Phys. Rev. B, 56 10549 (1997) [279] T. P. St. Clair, D. W. Goodman, Top. Catal., 13 5 (2000) [280] D.A. Cole, J.R. Shallenberger, S.W. Novak, R.L. Moore,M.J. Edgell, S.P. Smith, C.J. Hitzman, J.F. Kirchhoff, E. Principe, W. Nieveen, F.K. Huang, S. Biswas, R.J. Bleiler, and K. Jones, J. Vac. Sci. Tech. B, 18 404 (2000) [281] D.D: Eley, and P.B. Moore, Surf. Sci., 52 21 (1975) [282] Jaakko Akola and Hannu Hkkinen, PRB, 74, 165404 (2006) [283] M. Di Vece, S. Palomba, and R. E. Palmer, Phys. Rev. B, 72, 073407 (2005) [284] H. Hvel, Appl. Phys. A: Mater. Sci. Process. 72, 295 (2001) [285] T. Irawan, I. Barke, and H. Hvel, Appl. Phys. A: Mater. Sci. Process. 80, 929 (2005) [286] A. Hashimoto, K. Suenaga, A. Gloter, K. Urita, and S. Iijima, Nature, 430, 870 (2004) [287] Nordlund, K., Keinonen, J. & Mattila, T. Formation of ion irradiation induced small-scale defects on graphite surfaces. Phys. Rev. Lett., 77, 699702 (1996) [288] Krasheninnikov, A. V., Nordlund, K., Sirvio, M., Salonen, E. & Keinonen, J. Phys. Rev. B, 63, 245405 (2001) [289] Kelly, K. F. & Halas, N. J., Surf. Sci. 416, L1085L1089 (1998) [290] Ouyang, M., Huang, J.-L., Cheung, C. L. & Lieber, C. M., Science 291, 97100 (2001)

- 245 -

[291] Xiao Tong, Lauren Benz, Paul Kemper, Horia Metiu, Michael T. Bowers, and Steven K. Buratto, J. AM. CHEM. SOC., 127, 13516-13518 (2005) [292] Lauren Benz, Xiao Tong, Paul Kemper, Horia Metiu, Michael T. Bowers, and Steven K. Buratto JPCB, 110, 663-666 (2006) [293] Bromann, K.; Felix, C.; Brune, H.; Harbich, W.; Monot, R.; Buttet, J.; Kern, K., Science, 274, 956 (1996) [294] Bromann, K.; Brune, H.; Felix, C.; Harbich, W.; Monot, R.; Buttet, J.; Kern, K. Surf. Sci., 377, 1051 (1997) [295] Lai, X.; St Clair, T. P.; Valden, M.; Goodman, D. W., Surf. Sci., 59, 25 (1998) [296] Lai, X. F.; Goodman, D. W., J. Mol. Catal. A, 162, 33 (2000) [297] Parker, S. C.; Grant, A. W.; Bondzie, V. A.; Campbell, C. T. Surf. Sci., 441, 1 (1999) [298] Wahlstrom, E.; Lopez, N.; Schaub, R.; Thostrup, P.; Ronnau, A.; Africh, C. Laegsgaard, E.; Norskov, J. K.; Besenbacher, F. Phys. ReV. Lett., 90, 026101 (2003) [299] Tong, X.; Benz, L.; Kolmakov, A.; Chretien, S.; Metiu, H.; Buratto, S. K. Surf. Sci., 575, 60 (2005) [300] R. Anton and I. Schneidereit, Phys. Rev. B, 58, 13874 (1998) [301] R. Anton and P. Kreutzer, Phys. Rev. B, 61, 16077 (2000) [302] N. Vandamme, E. Janssens, F. Vanhouette, P. Lievens, and C. Van Haesendonck, J. Phys.: Condens. Matter, 15, S2983 (2003) [303] Hakkinen, H.; Yoon, B.; Landman, U.; Li, X.; Zhai, H.-J.; Wang, L.-S. J. Phys. Chem. A., 107(32); 6168-6175 (2003) [304] Gilb, S.; Weis, P., Furche, F.; Ahlrichs, R.; Kappes, M. M. J. Chem. Phys., 116, 4094 (2002) [305] Furche, F.; Ahlrichs, R.; Weis, P., Jacob, C.; Gilb, S.; Bierweiler, T.; Kappes, M. M. J. Chem. Phys., 117, 6982 (2002) [306] Michael Bttner, Peter Oelhafen, Surface Science, 600, 11701177 (2006)

- 246 -

[307] J.K. Gibson, J. Vac. Sci. Technol. A, 16, 53 (1998) [308] K. Fauth, M. Heissler, D. Batchelor, G. Schuth, Surf. Sci., 529, 397 (2003) [309] Plenary talk, 13th International Symposium on Small Particles and Inorganic Clusters, Sweden (2006) [310] D. A. H. Cunningham, W. Vogel, H. Kageyama, S. Tsubota and M. Haruta, Journal of Catalysis,177, 1, 10 (1998) [311] Guan Ming Wang,1 Joseph J. BelBruno,1 Steven D. Kenny,2 and Roger Smith2, PHYSICAL REVIEW B, 69, 195412 (2004) [312] Guan Ming Wang, Joseph J.BelBruno, Steven D.Kenny, Roger Smith, Surface Science, 576,107115 (2005) [313] F. Atamny, O. Spillecke, and R. Schlgl, Phys. Chem. Chem. Phys., 1, 4113 (1999) [314] E. Ganz, K. Sattler, and J. Clarke, Surf. Sci., 219, 33 (1989) [315] G. K. Wertheim, S. B. DiCenzo, and S. E. Youngquist, Phys. Rev. Lett., 51, 2310 (1983) [316] M. Quinten, I. Sander, P. Steiner, U. Kreibig, K. Fauth, and G. Schmid, Z. Phys. D: At., Mol. Clusters, 20, 377 (1991) [317] T. Ohgi and D. Fujita, Phys. Rev. B, 66, 115410 (2002) [318] A. Howard, D. N. S. Clark, C. E. J. Mitchell, R. G. Egdell, and V. R. Dhanak, Surf. Sci., 518, 210 (2002) [319] J. N. O'Shea, M. A. Phillips, M. D. R. Taylor, P. Moriarty, M. Brust, and V. R. Dhanak, Appl. Phys. Lett., 81, 5039 (2002) [320] M. Imamura and A. Tanaka, Phys. Rev. B 73, 125409 (2006) [321] K. Takahiro, S. Oizumi, A. Terai, and K. Kawatsura, B. Tsuchiya and S. Nagata, S. Yamamoto, H. Naramoto and K. Narumi, M. Sasase, Journal of Applied Physics, 100, 084325 (2006) [322] Kibsgaard, J.; Lauritsen, J. V.; Lagsgaard, E.; Clausen, B. S.; Topsoe, H.; Besenbacher, F., J. Am. Chem. Soc., 128(42), 13950-13958 (2006)

- 247 -

[323] Marton, D.; Bu, H.; Boyd, K. J.; Todorov, S. S.; Albayati, A. H.; Rabalais, J. W. Surf. Sci., 326, L489-L493, 34 (1995) [324] Hahn, J. R.; Kang, H. Phys. Rev. B, 60, 6007-6017 (1999) [325] Marton, D.; Boyd, K. J.; Lytle, T.; Rabalais, J. W. Phys. Rev. B, 48, 6757-6766 (1993) [326] J.M. Antonietti, M. Michalski, U. Heiz, H. Jones, K.H. Lim, N. Rosch, A. Del Vitto, G. Pacchioni, Phys. Rev. Lett., 94, 213402 (2005) [327] G. de la Puente, J.J. Pis, J.A. Menedez, P. Grange, J. Anal. Appl. Pyrol., 43, 125-138 (1997) [328] M. Haruta, Chem. Record, 2, 75 (2003) [329] J.C. Pivin1, M. Sendova-Vassileva, M. Nikolaeva, D. DimovaMalinovska, A. Martucci, Appl. Phys., A 75, 401410 (2002) [330] 13th International School on Quantum Electronics: Laser Physics and Applications. Edited by Atanasov, Peter A.; Gateva, Sanka V.; Avramov, Lachezar A.; Serafetinides, Alexander A. Proceedings of the SPIE, Volume 5830, pp. 31-39 (2005) [331] H. Haberland, M. Mall, M. Moseler, Y. Quiang, T. Reiners, Y. Thurner, J. Vac. Sci. Technol. A, 12, 2925-2930 (1994) [332] Grinter, P.L.J. et al.. Rev.Sci. Eng., 39, 77 (1997) [333] Basset, J. M., Lefebvre, F., Santini,C. Coord. Chem. Rev., 178, 1703 (1998) [334] Schroeder, T., Adelt, M., Richter, B., Naschitzki, M., Baeumaer, M., Freund, H.-J. Surf. Rev. Lett., 7, 7 (2000) [335] Schroeder, T., Giorgi, J. B., Baeumaer, M., Freund, H.-J. Phys. Rev. B, 66, 165422 (2002) [336] Todorova, T. K., Sierka, M., Sauer, J., Kaya, S., Weissenrieder, J., Lu, J.L., Gao, H.-J., Shaikhutdinov, S., H.-J. Phys. Rev. B, 73, 165414 (2006) [337] D.M. Cox, W, Eberhardt, P.Fayet, Z.Fu, B. Kessler, R.D. Sherwood, D. Sondericker, and A. Kaldor, Z.Phys. D 20, 385 (1991) [338] S.B. DiCenzo, S.D. Berry, and E.H. Hartford, Jr., Phyts. Rev. B, 38, 8465 (1988)

- 248 -

[339] Umberto Martinez, Livia Giordano, and Gianfranco Pacchioni, J. Phys. Chem. B, 110 (34), 17015-17023 (2006) [340] Del Vitto, A.; Pacchioni, G.; Lim, K. H.; Rosch, N.; Antonietti, J.-M.; Michalski, M.; Heiz, U.; Jones, H., J. Phys. Chem. B, 109, 19876 (2005) [341] Gianfranco Pacchioni and Gianluigi Ierano. Phyts. Rev. B, 57, 818 (1998) [342] H.-G. Boyen, A. Ethirajan, G. Kastle, F. Weigl, P. Ziemann, G. Schmid, M.G. Garnier, M. Buttner, P. Oelhafen, Phys. Rev. Lett., 94, 016804 (2005) [343] Shallencerger et. al, J VAC SCI TECHNOL B, 18 (1), 440-444 (2000) [344] B.K. Min, W.T. Wallace, A.K. Santra, and D.W. Goodman, J. Phys. Chem. B, 108, 16339-16343 (2004) [345] Zhen Yan, Sivadinarayana Chinta, Ahmed A. Mohamed, John P. Fackler, Jr., and D. Wayne Goodman, Am. Chem. Soc., 127 (6), 1604 -1605 (2005) [346 ]Chen, Y., Gonzalez, R.; Schow, O. E.; Summers, G. P. Phys. Rev. B, 27, 1276-1282 (1983) [347] Matthias Arenz, Uzi Landman, Ueli Heiz, ChemPhysChem, 7, 9, 18711879 (2006)

- 249 -

Acknowledgment

There are many people who supported me and I would like to express my gratitude to all those who gave me the possibility to complete this thesis, especially to those whom I cannot mention explicitly in the following. First of all, I would like to thank my supervisor, Prof. Dr. Gerd Gantefr and Prof. Dr. Young Dok Kim for attracting me to studying and for their supports and continuous readiness for discussion. I also would like to express my gratitude to Prof. Dr. Leiderer for his effort to be the second referee of my dissertation. Among my colleagues (and former colleagues), I am especially indebted to Ignacio Lopez and Rainer Dietsche, who are my good colleagues in the practical work and life in German. I have furthermore to thank Marco Niemietz, Tim Fischer, Wilko Westhuser, Jrn Cordes, Matthias Gtz, Tobias Mangler, Karsten F. Vetter, Dr. Kiichirou Koyasu and former colleagues (Thorsten Ketterer, Moritz Bubek, Dr. Nils Bertram, Markus Engelke, Dr. Felix v.Gynz-Rekowski) for all their help, support, interest and valuable hints. I also want to thank the staff of the Fachbereich Physik in Universitt Konstanz. Especially, I would like to give my special thanks to my family whose patient love enabled me to complete this work.

- 250 -