An (ultra) high-vacuum compatible sputter source for oxide thin film growth Lukas Mayr, Norbert Köpfle, Andrea Auer, Bernhard Klötzer, Simon Penner* Institute for Physical Chemistry, University of Innsbruck, Innrain 52a, A-6020 Innsbruck *Corresponding Author: simon.penner@uibk.ac.at, 00435125072925 1 Tel: 00435125075056, Fax: ABSTRACT A miniaturised CF-38 mountable sputter source for oxide and metal thin film preparation with enhanced high-vacuum and ultra-high-vacuum compatibility is described. The all home-built sputtering deposition device allows a high flexibility also in oxidic sputter materials, suitable deposition rates for preparation of films in the nm- and the sub-monolayer regime and excellent reliability and enhanced cleanliness for usage in UHV chambers. For a number of technologically important - yet hardly volatile - materials, the described source represents a significant improvement over thermal deposition techniques like electron-beam- or thermal evaporation, as especially the latter are no adequate tool to prepare atomically clean layers of refractory oxide materials. Furthermore, it is superior to commercially available magnetron sputter devices, especially for applications, where highly reproducible sub-monolayer thin film preparation under very clean UHV conditions is required (e.g. for studying phase boundary effects in catalysis). The device in turn offers the usage of a wide selection of evaporation materials and special target preparation procedures also allow the usage of pressed oxide powder targets. To prove the performance of the sputter-source, test preparations with technologically relevant oxide components, comprising ZrO2 and yttriumstabilized ZrO2, have been carried out. A wide range of characterization methods (electron microscopy, X-ray photoelectron spectroscopy, low-energy ion scattering, atomic force microscopy and catalytic testing) was applied to demonstrate the properties of the sputterdeposited thin film systems. 2 I. INTRODUCTION Due to the importance of oxidic (coating) materials in a wide area of technologies ranging from nanotechnology over ceramics to catalysis, much effort has been put into the development of thin film preparation methods, leading to well-defined oxide coatings suitable for reliable structural, spectroscopic and chemical analysis. As in fundamental research, this approach is usually related to the use of well-defined and –structured thin film systems, the need for reproducible and reliable preparation pathways is obvious. For many (oxide) film systems, a set of preparation pathways is already available, in essence comprising thermal deposition and electron-beam evaporation methods. In principle, both methods are highvacuum and ultra-high vacuum compatible. However, while this approach is suitable for a variety of sufficiently volatile metallic and/or oxidic systems, it does not offer access to a variety of also technologically important systems with disadvantageous materials properties. These might include hard-to-evaporate “refractory” materials with high melting points and at the same time low vapor pressures, thus leading to very slow deposition rates, chemically reactive materials or structural and compositional complex materials. A well-known example of the former is the system Zr/ZrO2. Oxidized species of zirconium are known to be an essential constituent of several technologically relevant systems including ceramics and heterogeneous catalysts1,2. Despite this importance, due to its high melting point (2125 K)3 in combination with a low vapor pressure (~10-2 Pa at 2260 K)4 it is difficult to prepare wellstructured, clean layers of defined (especially submonolayer-) thickness using a thermal or electron-beam evaporator. However, clean and well-defined depositing conditions are required for the preparation of meaningful (ultra) high vacuum systems. This is especially true for model catalyst studies, which are essential to understand metal-oxide support interactions and phase boundary phenomena5,6, in order to identify and characterize catalytically active sites and consequently design an even more selective and active catalytic surface or metal3 oxide interface. In this respect, ZrO2 layers are usually prepared by sputtering techniques such as magnetron sputtering7, ion beam sputtering8 or reactive cathodic arc deposition9. These techniques are also applied for Zr-containing alloys10-12. However, all these methods suffer from the serious drawbacks of being on the one hand technologically very sophisticated and demanding, on the other hand, defined “in-situ” model system preparation in e.g. a surface science and/or catalysis-dedicated apparatus13 is at least very difficult. A representative example for a structural and chemical very complex oxidic material are mixed oxides in general, and to once again reference to the ZrO2 system, yttria-stabilized ZrO2 (YSZ). The latter adopts cubic or tetragonal structures with varying yttrium content14. Moreover, it is of significant technological interest and is one of the main constituents of solid oxide fuel cells15. However, while it is possible to prepare YSZ films on large scale by magnetron- or radiofrequency sputtering16, these films are not suitable as model systems because the preparation method is less compatible with (ultra-) high vacuum surface and structural analysis. Furthermore, due to the different vapor pressures and melting points, reproducible preparation by thermal and electron-beam evaporation, with defined yttria content in the deposited film, is not possible. To gain access to both classes of preparation-wise very demanding materials, a novel sputtering device was developed, meeting the following criteria: (a) being fully UHV and HV-compatible with respect to cleanliness of operation, (b) being long-term stable, (c) allowing reproducible atomically clean preparation of (ultra-) thin (oxide) films, (d) allowing highly precise and controllable deposition rates, (e) being particularly flexible in material choice, and, last but not least, (f) being small enough to fit inside a standard-size UHV or HV chamber using a conventional DN 35 or DN 40 flange port. It represents an allover selfdeveloped and -constructed device, differing from conventional ion-beam or magnetron sputter sources with respect to smaller size, higher simplicity, better cleanliness, enhanced 4 UHV compatibility and more appropriate sputter rates for sub-monolayer deposition. It additionally allows overcoming the restrictions of thermal and/or electron-beam evaporation. It permits not only the preparation of thin oxide and/or metal films of almost arbitrary thickness, but is also an adequate tool for reactive sputtering, for example leading to growth of oxide films when sputtering in defined background pressures of oxygen. Besides pure oxidic films, by usage of special sputtering targets, the sputter source is also suitable for preparation of more complex oxide systems, such as mixed oxides with defined composition. To appreciate the full applicability of the so-prepared systems, the resulting films have been subjected to a variety of different surface- and bulk structural, spectroscopic and property characterization, including electron microscopy (TEM), electron diffraction (SAED), X-ray photoelectron spectroscopy (XPS), atomic force microscopy (AFM), low-energy ion scattering (LEIS) and catalytic testing in methanol steam reforming. Particular emphasis has been given to proper analysis of film homogeneity and cleanliness. The sputter source was tested in a high-vacuum system regarding long-term stability of preparation conditions and quality of sputtered material. After transfer to an UHV chamber, surface analysis by XPS, AES and LEIS and, via an attached high-pressure catalytic cell, catalyst testing under technologically relevant conditions13 was carried out. Subsequently, exsitu bulk structural characterization by TEM and AFM was performed. For details on both the HV and UHV chamber we refer to previous publications17-21. The capabilities of the sputter source are exemplarily shown for the demanding preparation of ZrO2 and yttria-stabilized ZrO2 films, but can be deliberately extended to the preparation of similar systems. 5 II. EXPERIMENTAL SETUP A. Technical Description The design is generally based on the concept of a hot-cathode ionization gauge (Bayert Alpert gauge), but with strongly increased ion acceleration voltage and ion target current. Figure 1 highlights an engineering-detailed drawing (Autocad ©) of the sputter source. It is an (at least with respect to the hot parts) all-tantalum construction except for the thoriated tungstenfilament (A), three welding rods (B) and Al2O3 ceramics (Al23 Degussit ceramics from Friatec AG) for electric insulation. A three-way electric feed-through (C), appropriate for a 2.75 inch CF flange, serves as base and mechanical foundation for three long stainless steel welding rods (D, 2x300 mm, 1x250 mm) that are used to reduce the distance to the sample. Using this feed-through, easy and fast mounting to both the HV system and an UHV chamber of approximately 20 cm radius (with respect to the inner wall) was possible. By variation of the length of the stainless steel rods, an adaptation to smaller UHV chamber sizes down to 10 cm radius is feasible. An additional 150 mm linear manipulator between feed-through and flange allows varying the sample distance when mounted to the UHV chamber. To ensure a well-defined geometry at the whole length of the sputter gun, a tantalum sheet is mounted approximately half-way between flange and source, serving as a mechanical stabilizer and thereby ensuring a constant distance of the welding rods (not shown). These rods provide not only the mechanical fixing, but also serve as electrical conductors. All components in the top part of the source are anticipated to reach significant high temperatures during sputtering and are therefore made of tantalum. All connections in this area are fixed by electric spot welding and no bolted assemblies where used to reduce impurity atoms that may act as potential contaminants of the sample. A truncated-cone-formed grid (E), fixed by spot-welding of tantalum wires (Plansee, 0.3 mm and 1 mm), is spot-welded to a 1 mm diameter welding rod end. To improve mechanical 6 stabilization, the point of weld is sheathed with a 0.2 mm thick Ta foil, which is also spot welded to the welding rod and the grid. The grid has two integrated electronic feed-throughs consisting of Al23-ceramic-wrapped tantalum wires (0.3 mm), that are on one end connected to both the second welding rod and the ground, respectively. On the other end they serve as attachment for the spiraled and looped thoriated tungsten-filament (Goodfellow, d = 0.125 mm, total length: 200 mm, total resistivity: ~ 10 Ω, approximate diameter of loops: 2 mm, number of loops: 30, made by twisting the wire 30 times around a 2 mm diameter bar as a template). The fourth welding rod is shorter (F, 250 mm), centered and serves as a connector for the plate-shaped, approximately 10 x 10 mm sized sputter target (G). It may consist of either a piece of metal foil, spot-welded to an approximately 35 mm long rod of the same material for (reactive) metal sputtering, or of a pelletized oxide plate pressed onto a small Ta spiral, attached to the end of a Ta support rod. Since this connector is at the same voltage as the target and therefore potentially sputtered too, it has to be centered behind the target and consequently shielded. For high-vacuum preparation, an additional Ta shielding (H, see front and side view in the bottom panel) is used to reduce tungsten contaminations from the filament. These contaminations turned out to be a relevant issue when using a, compared to a UHV chamber, high background pressure (see E. “Purity”). Upon mounting the gun, there are three electrical connectors accessible from the outside plus one for ground connection. All necessary voltages and currents can be applied separately to grid, target and filament according to the electric connection scheme in Figure 1. The filament is connected to ground on one side, the low voltage AC heating current for the filament to another connector. Second and third isolated connectors supply negative high voltage with respect to ground to the sputter target and (with respect to the grounded filament) positive acceleration voltage for the electrons to the grid. As power supplies a Tectra-e-flux electron 7 beam evaporator power supply unit and a Fluke Model 410B High Voltage power supply with polarity-invertible stabilized high voltage up to 10 kV are in use. As deposition templates, ultra-clean Cu foils for methanol model catalysis studies, enabling surface analysis in the UHV chamber, and NaCl(001) single crystals for the preparation of epitaxial thin film samples for structural analysis of epitaxially grown nanoparticles by TEM, were chosen. Both templates were mounted side-by-side in the HV chamber during sputtering, thereby ensuring identical sample preparation. 8 FIG 1: Technical-detailed drawing of the sputter source consisting of the following parts: thoriated-W-Filament (A), electric connectors (B), electric feed troughs (C), four long welding rods (D), tantalum-grid (E), target holder (F), target connector (G), optional shielding (H). 9 B. Functionality The operation principle of the sputter gun is schematically depicted in Figure 2. The topmost part, consisting of the thoriated W-Filament and the tantalum grid, is responsible for argon ion production. Emitted electrons from the filament, which is operated at 2 A and approximately 17 V, are accelerated to the tantalum grid, which is positively charged. While the highest effective cross section for ionizing Argon atoms to Ar+ is at about 70 eV, the applied high voltage to the grid is empirically optimized to +300 V to partly overcome the strongly repulsive negative electric field between target and grid. Due to the fact that the anode is a grid, electrons can pass through it and are then forced by the opposite-charged electric field at the backside of the grid backwards, leading to strongly curved trajectories. The so-enlarged trajectory of the electrons from cathode to anode increases the probability of collision with argon atoms and therefore the number of generated argon ions. As a result, an emission current, collected at the grid, of at least 150 mA can be measured. The ionized argon atoms are subsequently accelerated with an applied target-voltage of also empirically optimized 2000 V to the sputter target. The simple, but effective geometry ensures a favoured sputter direction toward the substrate. In case oxidic powder sputtering targets are used, the respective oxide powders are pressed into round thin pellets of approximately 1 cm diameter and 1 mm thickness, supported on sufficiently flexible 0.3 mm diameter Ta spiral wire supports to partly overcome their low electronic conductivity. The Ta wire is then used to connect the pellet to the target pin. During operation, a background pressure of 10-4 mbar argon was in use, resulting in a sputter drain current, collected at the target, of approximately 50-100 µA. 10 FIG 2: Functionality scheme of the sputter procedure. Emitted electrons from the filament are accelerated to the grid and ionize on their preferable long parabolic pathway Ar atoms to Ar+ions. These generated ions are then affected by -2000V target voltage and sputter therefore the target surface under an angle that causes a preferred direction of sputtered material towards the centered hole in the grid. C. Quantitative Analysis The time-resolved recording of the sputter-current is also suitable for quantitative sputter operation, as shown in Figure 3, and the integrated value in µA∙s was then set proportional to real film thickness measured by a XPS sputter profile. These calibration was done a couple of times and the results accommodated with the XPS over layer model22, which allows to calculate the theoretical thickness of an overlayer on a substrate directly from the XPS intensities. An exemplary simulation for a ZrO2 overlayer is described in Section III. The device allows the deposition of nm-thick films in reasonable time. However, its significant advantage is the possibility to set the sputter rate to a constant, reliable and reproducible low value of approximately 0.05 nm/min by varying the Ar pressure in the low 10-5 mbar regime and the filament emission current and to hence deposit material in the sub-monolayer regime (e.g. being inevitable for studying phase boundary effects in catalysis6). A sub-monolayer of 11 any material can therefore be deposited at the time scale of minutes, allowing for a precise time measurement with a negligible standard error. Deposition rates of other preparation techniques, such as thermal evaporation, but also of commercial magnetron sputter sources with considerably higher deposition rates (e.g. around 20 nm/min corresponding to roughly 100 ML/min), are way too high for reproducible sub-monolayer deposition. To prepare half a monolayer, the deposition time would have to be controlled within fractions of a second (or the distance between gun and sample accordingly unrealistically increased). For preparation of thicker films, higher deposition rates would in principle be advantageous, but the gain in increasing the deposition rate has to be balanced against the corresponding higher amount of impurities in the sputtering gas (estimation see below in section E. Purity). 3 400x10 Sputter Current /µA 50 300 40 200 30 Sputter Current integrated Sputter Current 20 100 integrated Sputter Current /µAs 60 10 0 0 0 2000 4000 Time /s 6000 8000 FIG 3: Quantitative analysis of deposited films was realized by integration of the recorded sputter current being collected at the target and calibrated via a sputter depth profile. Results were subsequently correlated with XPS film thickness simulations22. 12 D. Parametrization The maximum argon pressure and the target high-voltage that could be used is limited by uncontrolled spark discharge in the area of the electronic feed-throughs or even in the target area. Since a high voltage of -2000 V is obligatory to transfer enough Ar+ kinetic energy to the surface, the maximum argon pressure should not exceed the low 10-4 mbar range. Increasing the sputter current, measured at the target’s high voltage, could also be reached by increasing the electron emission to the tantalum grid. The used voltage feed provided a maximum of 250 mA, but usually, an electron emission at the grid of 150-180 mA was used. This limitation also reduced sample heating by the sputter source filament. For nm-thick films, the sample distance must be kept as small as possible (at the same time avoiding significant sample heating by the filament) to compensate the - in comparison to other evaporation techniques - lower sputtering rates. For these purposes, the following optimal parameters were used: operating the sputter source at a distance of 30 mm causes a sample temperature of approximately 500 K and the deposition of approx. 1 monolayer (ML) of ZrO2 onto a polycrystalline copper foil in less than 10 minutes. E. Purity To ensure that no impurity atoms are evaporated, all prepared films were without exception checked by XPS. Besides the common impurities caused by the sample transportation through air to the UHV chamber, especially carbon, at first unacceptable tungsten contamination levels between 0.5 and 10 atomic percent caused by filament oxidation or oxidative sputtering, turned out to be a serious issue, especially when operating the source under wateror oxygen-rich HV conditions. Using UHV conditions, and additionally seeding 10% hydrogen in the Ar sputter gas, led to effective suppression of tungsten contamination and consequently, the most prominent W 4f7/2 XPS peak was hardly distinguishable from the background noise. A remaining, very low tungsten contamination (less than 0.1 atomic %) 13 cannot be excluded. To eliminate tungsten also from sample prepared under HV conditions, a specially designed tantalum shielding, positioned in line-of-sight between filament and substrate (see Figure 1), was installed. This effectively removed tungsten from the sample. However, traces of tantalum (<0.8 atom %) are unavoidable. Figure 4 shows representative examples of the W 4f7/2 peak under high-vacuum conditions before and after installation of the tantalum shielding, clearly showing that use of the tantalum aperture effectively reduces the tungsten impurity level to below the detection level of XPS. FIG 4: W 4f XPS spectra of the deposited YSZ film under HV conditions (background pressure 10-6 mbar) with and without shielding, respectively. The corresponding spectrum taken under UHV conditions is also shown. The formation mechanism of the tungsten and tantalum contaminations is less clear. Two different mechanisms can be in principle thought of, consequently giving rise to enhanced impurity levels. On the one hand, corresponding thermal evaporation studies showed, that 14 oxidizing atmospheres (caused by a low O2 background pressure), inevitably lead to formation of volatile (sub-) stoichiometric tungsten and tantalum oxides if tungsten and tantalum crucibles are used as evaporation sources. Consequently, heavily contaminated samples result. In turn, a small oxygen partial pressure in the argon sputtering gas may cause similar problems. On the other hand, all components located between grid and sample (essentially the filament or the optional aperture to shield the filament) are in fact susceptible to unwanted sputtering by reversely accelerated argon ions. However, these argon ions are produced between filament and grid and are therefore not affected by the -2000 V electric field from the target penetrating through it. Consequently, they are accelerated by the potential difference between +300 V at the grid and approximately +17 V filament voltage. Because their kinetic energy is low and their sputter yield consequently too, as well as because of the disadvantageous geometry, the amount of sputtered tungsten or tantalum onto the deposition template is very low. Considering the fact, that low background pressures, like operating the source under UHV conditions, and additionally reducing atmospheres suppress tungsten below the detection limit of the XPS signal even without an aperture, it is very likely that thermal evaporation of sub-oxides plays a more important role. Anyhow, a combination of both mechanisms is supposable including the idea of a mechanism where sputtered Zr, deposited on the filament, forms zirconium oxides and is therefore facilitating the formation of tungsten oxides. Additionally, UHV experiments using a thinner/hotter W-filament (0.1 mm instead of 0.125 mm) for better suppression of WOx formation, which is rather likely at lower surface temperatures of the cathode, were performed. Subsequent XPS analysis (W 4f and W 4d) demonstrated clearly, that no detectable W impurities were present any more, even when the background pressure was set as high as the low 10-7 mbar range and no H2 was added. It is conclusive, that the reason is the thinner filament, consequently being hotter to achieve the 15 same electron emission level. This temperature is now obviously higher than the upper limit of the stability range of volatile W-oxides, which are consequently not formed anymore, even at higher O2 partial pressures. The so prepared films are therefore free of any impurities as can be deduced from the most recent XPS experiments (see corresponding spectrum in Figure 4). Regarding UHV and background gas purity, we note that the device contains no temperaturesensitive parts, thus, there are no restrictions concerning bake-out temperature. In consequence, a proper UHV degassing procedure at high temperatures up to 1000 K can be performed. Since this is not the case for commercial magnetron sputter devices, where temperature-sensitive permanent magnets are in use, much lower partial pressures of unwanted contaminations in the Ar sputtergas during operation are a further advantage. The usual operational Ar pressure for the newly designed sputter-source is, depending on the desired sputter rates, between 10-5 mbar and 10-4 mbar. A 0.5 vpm O2 contaminated Ar therefore causes an O2 partial pressure in the 10-11 mbar range and, considering a sticking coefficient of 1, full surface coverage within more than 10 hours. Commercial magnetron sputter devices operate between 10-3 and 10-2 mbar, resulting in a two orders of magnitude higher partial pressure of background contaminations during sputtering using Ar with the same purity level. This accordingly causes lower times for surface coverage and unwanted surface- and also gas-phase reactions. It also makes it more difficult to rapidly reach UHV conditions with a pressure about 10-10 mbar after sputtering, which is mandatory for surface science studies. Therefore, the further developed sputter source features a much higher UHV surface science application suitability than commercial sputter devices. F. Sample characterization The HV and UHV chambers used for preparation and surface analysis are in detail described elsewhere and only the most important features are reported. The HV chamber exhibits a base pressure of 10-6 mbar and basically consists of a Duran glass cross, to which all the necessary 16 equipment for sputtering can be attached via dedicated flanges. The deposition templates are mounted on a specially-designed copper holder, which can be heated to 623 K. The UHV chamber (base pressure low 10-10 mbar range) is specially-equipped and designed for surface analysis in combination with catalytic testing under technically relevant conditions (up to 1 bar)13. Characterization of the samples is performed using a Thermo electron Alpha 110 XPS/Auger/LEIS spectrometer and a standard Mg/Al anode X-ray gun (XR 50, SPECS), an Omicron ISE 100 ion gun to provide the focused 1 keV He+ ions for LEIS, an electron beam heater, an ion sputter gun and a mass spectrometer (Balzers). The described sputter source is by standard mounted on a 2.75 inch CF flange. AFM analysis is performed under ambient conditions using a Veeco Dimension 3100 atomic force microscope in tapping mode and phosphorous-doped Si cantilevers with a spring constant of 20 N/m and a resonance frequency of 256 kHz. TEM measurements are conducted on a ZEISS EM 10C electron microscope operated at 100 kV. Catalytic measurements were performed in the high-pressure catalytic cell attached to the UHV chamber in batch mode. Detection of the reaction products and intermediates with high sensitivity is possible by discontinuous injection into the gas-chromatography-mass spectrometry (GC-MS) setup (HP G1800A) or by direct online MS analysis of the reaction mixture via a capillary leak into the GC/MC detector. Sample transfer is ensured by means of a magnetically coupled transfer rod from the UHV holder to a quartz glass sample holder inside the all-glass reaction cell. III. APPLICATIONS OF THE HOME-BUILT SPUTTER SOURCE: EXPERIMENTAL RESULTS To fully appreciate the capabilities of the sputter source for the preparation of both technologically relevant, but at the same time preparation-wise very demanding systems, the 17 ZrO2 and yttria-stabilized ZrO2 systems were chosen. As outlined in the introductory chapter, preparation of ZrO2 films, in case of simple thermal evaporation, usually requires the use of tungsten or tantalum crucibles and a defined oxygen partial pressure to ensure a full ZrO2 stoichiometry in the deposited films. As this in turn causes considerable impurity levels in the resulting films, clean ZrO2 films are hard to obtain. Mixed oxide film preparation is also a very demanding procedure, in order to control the composition, structure and morphology of the samples. We exemplify the capabilities of our sputter source for the reproducible preparation of yttria-stabilized ZrO2 films. A. Deposition of ZrO2 films As sputter target a 0.25 cm2 metallic zirconium foil (Alfa Aeser, Purity 99.5%) was used. To remove oxidic passivation layers, a preliminary lead time of at least 10 minutes was used for all sputter procedures. Reproducible preparation of ZrO2 films is possible by simple sputtering the Zr foil in argon at the background pressure in the UHV apparatus of about 2∙10-9 mbar, which is equal to the pressure in the UHV chamber before the sputter source was turned on and after switching off. The argon bottle that was used for sputtering (Linde argon 5.0) contains 1 ppm O2, which obviously is enough to cause full oxidation of surface-near zirconium in the deposited films. The formation of the latter is also aided by the use of a Cu foil (GoodFellow, 99.95%), with a relatively high amount of bulk-dissolved oxygen. Zirconium tends to extract this dissolved oxygen from the Cu bulk by additionally forming zirconia (ZrO2) on the surface. Figure 5 shows Zr 3d spectra of the deposited film under changing sputter atmosphere. Because of the high formation energy of ZrO2 it is very easy to perform successful reactive sputtering. Complete oxidation (top-most spectrum), yielding peaks at 183.3 and 185.0 eV binding energy, can already be obtained by adding very low amounts of O2. In fact, O2 impurities in argon 5.0 are already sufficient. Peaks at lower binding energy due to metallic Zr are clearly absent. Reduction of the O2 content in Ar by 18 using an oxygen trap, and further improving background pressure conditions, leads to incomplete oxidation (middle spectra) with a metallic Zr 3d doublet at lower binding energies. However, still partial oxidation is observed when using O2-cleaned Ar (Supelco Supelpure®O Oxygen/Moisture Trap – less than 2 ppb O2) caused by residual O-containing gas phase species in the UHV chamber (lowest spectra). A low-energy ion scattering spectrum of a thin ZrO2 film with a theoretical thickness of approx. 0.7 nm (~2ML) is shown in Figure 6. Considering the relative sensitivity factors of Cu and Zr in ZrO2, that have been determined by measuring both the pure Cu foil and the fully surface-oxidized Zr foil, a ratio of Cu:Zr of 1:6.7 can be determined. Thus, ZrO2 is apparently not entirely covering the Cu surface, but forms three-dimensional clusters. Figure 7 shows a representative three-dimensional AFM image of a thicker ZrO2 film deposited on the Cu foil. On the 500x500 nm2 image, a very regular array of rounded grains is observable. The film itself is very flat, as evidenced by the roughness analysis, yielding an rms roughness of approximately 2 nm. 19 increasing O2 content in sputter atmosphere Coun Counts /a.u. 188 184 180 Binding Energy /eV FIG 5: Zr 3d XPS spectra collected under increasingly oxidizing conditions for reactive sputtering. Fully oxidized zirconium can already be obtained by adding more than 1 ppm of O2 to the sputtering atmosphere (top-most spectrum). 20 4000 Counts /a.u. 3000 Cu Zr 2000 1000 0 780 800 820 840 860 Kinetic Energy /eV 880 900 FIG 6: He+-ion scattering spectrum of a Cu-ZrO2 surface. The higher mass of the zirconium leads to a higher kinetic energy of scattered ions. The measurement clearly demonstrates that Cu and Zr atoms are present at the surface. FIG 7: Three-dimensional AFM image of the ZrO2 film grown on a Cu foil at 300 K in false colors. The film has an overall dimension of approximately 15 nm in the z-direction, indicating that the film is very flat. 21 To avoid extensive sample heating by the filament of the sputter source, a distance of 30 mm between the hot filament at the top of the sputter source and the sample was set. Using a sputter current of approximately 50 µA, 1 ML ZrO2 is fully deposited in approximately 10 minutes and full coverage of the surface by ZrO2, resulting in complete suppression of the LEIS and AES signals of the Cu template, was achieved after sputtering for approximately 25 minutes. Subsequent high-pressure catalytic test reactions point out the catalytic relevance of the prepared Cu-ZrO2 surfaces (for details of the model catalysis experimental procedures, see6). Exemplified methanol steam reforming (MSR) reaction results on clean, submonolayer ZrO2covered and fully ZrO2-covered Cu foil are compared in Figure 8. Starting conditions included a reaction mixture of 12 mbar methanol, 24 mbar water, with He added to a total pressure of 1 bar. In comparison to a pure polycrystalline Cu surface, which produces a majority of formaldehyde by partial dehydrogenation of methanol (Fig. 8a), the prepared submonolayer oxide-metal system, which allows for a considerable amount of metal-oxide phase boundary sites (Fig. 8b), shows a bi-functional promotional effect. On the one hand, still an effective activation of methanol toward formaldehyde caused by the remaining Cu surface fraction is evident (0.020 mbar/min at maximum), but on the other hand activation of water and thus, strongly promoted total oxidation of formaldehyde toward CO2 (rate maximum: 0.025 mbar/min, presumably caused by the special phase boundary sites located at the interface of Cu and ZrO2) is observed. From Fig. 8c it is evident that a fully Cu-blocking ZrO2 film completely suppresses the reactivity, because neither Cu- nor phase boundary sites are available. Full dehydrogenation to CO on any model surface is hardly measurable. The further optimization of the phase boundary for promotion of total oxidation is currently under investigation. 22 623 K 296 K linear Ramp isothermal Reaction -3 25x10 a) CO2 Formaldehyde CO 20 15 10 Formation Rate / mbar/min 5 0 -3 25x10 b) 20 15 10 5 0 -3 25x10 c) 20 15 10 5 0 0 10 20 30 40 Time /min 50 60 FIG 8: Exemplified catalysis result of a methanol steam reforming reaction on a Cu-ZrO2 bifunctional surface (b). Starting conditions: 12 mbar methanol, 24 mbar water, He added to 1 bar total pressure. The other viewgraphs show a comparison to selectivity patterns on clean, undoped Cu (a) and an inactive, fully surface-covering ZrO2 film (c). B. Deposition of Yttria-stabilized ZrO2 (YSZ) films A pellet of YSZ powder (Sigma-Aldrich, 8 mol% Y, calcined overnight at 1173 K in air to remove traces of water), pressed onto a spirally formed 0.1 mm thick tantalum wire to ensure 23 target conductivity as well as mechanical support, was used as the sputtering target. The YSZ thin film was sputter-deposited under HV conditions (base pressure 10-6 mbar) onto a polycrystalline Cu foil for XPS purity and AFM analysis and on a NaCl single crystal for TEM (and also AFM) characterization. To ensure film purity, the Ta shielding, as mentioned above (Figure 1, H), was in use. To directly correlate XPS, AFM and TEM analysis, NaCl films, freshly deposited onto the Cu foil, were also used as growth templates. This procedure is necessary, since TEM-suitable films are exclusively prepared by growth on NaCl(001) growth templates. Using a mixed-oxide target for sputtering might in principle result in preferential sputtering of individual target components. However, preferential sputtering of one component leads to an enrichment of the non-preferential sputtered components at the target surface and therefore to an increasing sputter rate of this component. As a consequence, all atoms are sputtered in the correct ratio over a wide period of sputter time, exactly as they appear in the target. Variations of the original atomic compositions are consequently expected only at the beginning and at the end of the sputtering process, where an enrichment of the non-preferential sputtered component might appear. Thus, the yttrium content of the sputter-deposited film is very close to the yttrium content in the powder used for target preparation. Figure 9 shows the XPS overview spectrum as well as detailed spectra of the O 1s, Zr 3d and Y 3d regions of the prepared film. Besides the ubiquitous carbon contamination (C 1s at 284.3 eV) from transportation through air and traces of oxidized tantalum (<0.8 %), whose origin has already been discussed, no impurity atoms, except those of the underlying Cu foil, can be detected. The ratio of yttrium (Y 3d5/2 at 356.7 eV) to zirconium (Zr 3d at 182.9 eV) to O (O 1s at 530.9 eV) can be determined from XPS as 7:31:62, considering that O 1s also includes contributions from oxidized Cu species. A quantitative analysis of the film thickness by XPS using the XPS Thickness Solver Tool22 yields a value of 2 nm. The thickness of the YSZ layer 24 was calculated using the SRD 82 NIST database23 for estimating the three required electron attenuation lengths (EAL): Cu 2p photoelectron in Cu (5.7 Å), Cu 2p photoelectron in YSZ (6.6 Å) and Zr 3d Photoelectron in YSZ (16.81 Å). Asymmetric Parameters were taken from the ELETTRA-Database24,25. Y 3d Zr 3d Counts /a.u. Counts /a.u. O 1s 540 530 184 180 164 160 156 Cu 3p Cu 2p Cu 3d O KLL Cu 3s Ta 4f C 1s Cu Auger 1000 FIG 9: 800 600 400 Binding Energy /eV 200 0 Overview X-ray photoelectron spectra of the YSZ film grown on the freshly deposited NaCl(001) film, using a Cu foil as substrate material. The most prominent peaks are marked and the O 1s, Zr 3d and Y 3d regions shown as insets. For structural analysis, the corresponding YSZ film grown in-parallel on the NaCl(001) template was detached by dissolution in water and mounted on gold grids for electron microscopy. Figure 10 in turn highlights a representative overview TEM image of this film. The film exhibits a very low-contrast structure without pronounced structural or 25 morphological details. Accordingly, the selected area electron diffraction pattern only shows diffuse halos indicating an amorphous structure. FIG 10: Overview TEM image of the YSZ film grown on NaCl(001) at 373 K substrate temperature. The SAED pattern shown as inset confirms the amorphous structure of the film. IV. SUMMARY The specific advantages of the newly-designed sputter-device in comparison with other stateof-the-art sputter techniques and evaporation methods can be summed up as follows: Full UHV compatibility 1. A small 2.75 inch CF flange allows fast application in most surface-science UHV chambers without further adaption. 2. No bake-out limitations are present, i.e. full degassing at the usual bake-out temperatures without mandatory removal of parts like magnets is possible. 26 3. An in comparison to other sputter sources low operational Ar pressure of 10-4 to 10-5 mbar reduces the partial pressure of impurities and therefore minimizes unwanted side reactions during sputtering. This also significantly improves the purity by allowing a fast pump down to UHV conditions after sputtering. Applications for surface science studies 1. Highly precise calibration of the Ar-ion flux onto the target via the target drain current allows to control and to adjust the sputter rate (by varying Ar pressure and applied voltages). Once the voltages and currents are set to the desired values, the sputter source runs stable ad thus, the obtained results are highly reliable and reproducible. 2. Usage of sufficiently low sputter rates for reliable and reproducible preparation of sub-monolayer-thick films is possible. 3. Atomically clean layers can be easily obtained. 4. No magnets are present inside the UHV chamber, which is mandatory to perform high-resolution electron spectroscopy in the same system. Flexibility 1. Possibility of varying the kinetic energy of the Ar ions and to adapt it to the specific sputter target needs. 2. No water cooling and no water-cooled feed-throughs are required. V. CONCLUSION We successfully demonstrated the development and capabilities of an all-purpose sputter source compatible for high-vacuum and ultra-high vacuum usage. It offers the tremendous 27 advantages of providing rather easy access to preparation of atomically clean model systems of technologically relevant systems, which are otherwise hard to gain access to, neither with commercial sputter devices nor with thermal or electron-beam evaporators. Although the capabilities of the sputter source could be demonstrated only for selected examples, the application possibilities can be deliberately extended to other metallic, oxidic, intermetallic or mixed-oxide systems. A particular advantage over comparable systems is the in-parallel preparation of model samples for detailed surface and bulk structural and spectroscopic analysis, which can be also combined with property or catalytic characterization. This is in turn especially favorable for the establishment of structure-activity/selectivity/property relationships, which provides key access to the full understanding of the materials. In terms of model catalysis experiments dedicated to the role of the oxide-metal phase boundary, especially the controlled deposition of sub-monolayer oxide film coverages is important, which has been clearly shown to be possible with our setup. VI. ACKNOWLEDGMENTS The authors thank the FWF (Austrian Science Foundation) for financial support under project F4503-N16. We are also indebted to R. Pramsoler for the assistance in preparation of the Autocad drawings. 28 References [1] S. H. Liu, G. K. Chuah, S. Jaenicke, J. Mol. Catal. A: Chem. 2004, 220, 267-274. [2] H. Purnama, F. Girgsdies, T. Ressler, J. H. Schattka, R. A. Caruso, R. Schomacker, R. Schlogl, Catal. Lett. 2004, 94, 61-68. [3] I. Barin, Thermodynamical Data of Pure Substances, VCH, Weinheim, 1993. [4] D. A. Glocker, S. I. Shah, Handbook of Thin Film Process Technology, Vol. 2, Institute of Physics Publishing, 1995. [5] C. Rameshan, C. Weilach, W. Stadlmayr, S. Penner, H. Lorenz, M. Hävecker, R. Blume, T. Rocha, D. Teschner, A. Knop-Gericke, R. Schlögl, D. Zemlyanov, N. Memmel, G. Rupprechter, B. Klötzer, J. Catal. 2010, 276, 101-113. [6] C. Rameshan, W. Stadlmayr, S. Penner, H. Lorenz, N. Memmel, M. Hävecker, R. Blume, D. Teschner, T. Rocha, D. Zemlyanov, A. Knop-Gericke, R. Schlögl, B. Klötzer, Angew. Chem. 2012, 124, 3057-3061. [7] H. H. Zhang, C. Y. Ma, Q. Y. Zhang, Vacuum 2009, 83, 1311-1316. [8] S.-W. Yeh, T.-Y. Hsieh, H.-L. Huang, D. Gan, P. Shen, Mater. Sci. Eng., A 2007, 452–453, 313-320. [9] J. Čyvienė, M. Laurikaitis, J. Dudonis, Mater. Sci. Eng., B 2005, 118, 238-241. [10] E. Juzeliūnas, A. Grigucevičien, K. Leinartas, R. Jušknas, Electrochem. Commun. 2004, 6, 678-682. [11] S. Misumi, A. Okada, Y. Okamoto, M. Inoue, Procedia CIRP 2013, 6, 487-492. [12] X. Y. Li, E. Akiyama, H. Habazaki, A. Kawashima, K. Asami, K. Hashimoto, Corros. Sci. 1997, 39, 1365-1380. [13] W. Reichl, G. Rosina, G. Rupprechter, C. Zimmermann, K. Hayek, Rev. Sci. Instrum. 2000, 71, 1495-1499. [14] H. Ding, A. V. Virkar, F. Liu, Solid State Ionics 2012, 215, 16-23. 29 [15] J. Maček, B. Novosel, M. Marinšek, J. Eur. Ceram. Soc. 2007, 27, 487-491. [16] F. Kuhl, M. Piechotka, D. Reppin, T. Henning, J. Janek, P. J. Klar, MRS Online Proc. Libr. 2013, 1495, 2013.2017/2011-2013.2017/2016. [17] C. Rameshan, H. Lorenz, L. Mayr, S. Penner, D. Zemlyanov, R. Arrigo, M. Haevecker, R. Blume, A. Knop-Gericke, R. Schlögl, B. Klötzer, J. Catal. 2012, 295, 186-194 [18] C. Rameshan, W. Stadlmayr, S. Penner, H. Lorenz, L. Mayr, M. Hävecker, R. Blume, T. Rocha, D. Teschner, A. Knop-Gericke, R. Schlögl, D. Zemlyanov, N. Memmel, B. Klötzer, J. Catal. 2012, 290, 126-137. [19] H. Lorenz, S. Penner, W. Jochum, C. Rameshan, B. Klötzer, Appl. Catal., A 2009, 358, 203-210. [20] H. Lorenz, S. Turner, O. I. Lebedev, G. Van Tendeloo, B. Klotzer, C. Rameshan, K. Pfaller, S. Penner, Appl. Catal., A 2010, 374, 180-188. [21] S. Penner, H. Lorenz, W. Jochum, M. Stöger-Pollach, D. Wang, C. Rameshan, B. Klötzer, Appl. Catal., A 2009, 358, 193-202. [22] K. Smith, D. A. Saenz, D. Zemlyanov, A. A Voevodin 2012, "XPS Thickness Solver," https://nanohub.org/resources/xpsts. (DOI: 10.4231/D3N29P603) [23] C. J. Powell, A. Jablonski, NIST Electron Effective-Attenuation-Length Database, Vol. Version 1.3, National Instititue of Standards and Technology, Gaithersburg, 2011. [24] J. J. Yeh, I. Lindau, Atomic Data and Nuclear Data Tables, Vol. 32, 1985. [25] J. J. Yeh, Atomic Calculation of Photoionization Cross-Sections and Asymmetry Parameters, Gordon and Breach Science Publishers, Langhorne, PE (USA), 1993. 30