Characterization of Bimetallic Au/Pd (110) Surfaces

e-Journal of Surface Science and Nanotechnology

4 April 2009

Conference - ICSFS-14 -

e-J. Surf. Sci. Nanotech. Vol. 7 (2009) 448-454

Characterization of Bimetallic Au/Pd(110) Surfaces∗ M. Moors† Institute of Physical and Theoretical Chemistry, University of Bonn, Wegelerstr. 12, D-53115 Bonn, Germany,

T. Kobiela Faculty of Chemistry, Warsaw University of Technology, Noakowskiego 3, 00 664, Warsaw, Poland

M. Kralj Institute of Physics, P.O. Box 304, Zagreb HR-10000, Croatia

T. Pertram, C. Becker, and K. Wandelt Institute of Physical and Theoretical Chemistry, University of Bonn, Wegelerstr. 12, D-53115 Bonn, Germany (Received 30 May 2008; Accepted 8 October 2008; Published 4 April 2009) The growth as well as the compositional, electronic and structural properties of thin Au films deposited on a Pd(110) single crystal have been studied by means of ultraviolet photoelectron spectroscopy (UPS), photoemission of adsorbed xenon (PAX) and scanning tunneling microscopy (STM) as a function of film thickness (ranging from submonolayer amounts up to multilayers) and temperature. Our investigations indicate a Volmer-Weber growth mode of Au on Pd(110) at a deposition temperature of 300 K and below. As former studies already have shown, depositing Au amounts of less than 1 ML results in the formation of unreconstructed pseudomorphic Au islands. Above a critical thickness of 2 ML the formation of a (1 × 2) missing-row reconstruction typical for Au can be observed. This reconstructed Au bilayer is still in registry with Pd(110) and, thus, strained. Furthermore, our experiments show a strong temperature dependence of the surface morphology. Au multilayers prepared at 150 K, which are already quite flat, undergo a weak smoothening of the topmost atomic layer by annealing the sample up to 350 K. At a surface temperature of ∼ 450 K the diffusion of Au atoms into the bulk is initiated at the Au-Pd interface. Annealing a 3 ML thick Au film to 600 K finally results in the formation of an Au-Pd surface alloy, which is stable up to 900 K. A complete loss of Au from the surface can be detected at around 1050 K. Adsorption of CO as a probe molecule on the Au containing surfaces is only possible at low temperatures. [DOI: 10.1380/ejssnt.2009.448] Keywords: Photoelectron spectroscopy; Scanning tunneling microscopy; Gold; Palladium; Alloy

I.

INTRODUCTION

Single crystal metal alloy surfaces have always attracted much attention not only because they provide good model systems for heterogeneous catalysis [1] but also for their vital role in emerging new technologies relevant for environmental and energy-related applications, such as fuel cells [2]. In this study we present the characterization of ultrathin Au films deposited on a Pd(110) single crystal surface and their behavior at different temperatures using UPS, PAX and VT-STM. For a long time Au has been regarded as the least reactive metal for catalytic reactions but during the last years this opinion has changed considerably [3]. Highly dispersed Au deposited on a metal substrate exhibits a surprisingly high catalytic activity for several reactions even at low temperatures (e.g. for CO oxidation) [4]. Recently it has been reported that this feature depends strongly on the preparation conditions and the resulting size of the Au nanostructures [5]. The system Au on Pd(110) has already been studied extensively. In addition to theoretical calculations by Nieminen [6] experimental investigations by Schmitz et al. [7, 8], Kaukasoina et al. [9], Vos et al. [10] and Kralj et

∗ This

paper was presented at the 14th International Conference on Solid Films and Surfaces (ICSFS-14), Trinity College Dublin, Ireland, 29 June - 4 July, 2008. † Corresponding author: [email protected]

al. [11] showed a pronounced temperature dependence of its chemical and morphologic properties. The published results concerning the growth mode on the other hand contradict completely. While the extensive LEED studies of Schmitz et al. [8] led to the conclusion, that the growth of Au on Pd(110) is of Stranski-Krastanov type with a critical coverage of two monolayers, the medium energy ion scattering results of Vos et al. [10] argue for a Frankvan der Merve or layer by layer growth mode. STM and XRD investigations by Kralj et al. [11] finally showed a clear Volmer-Weber or island growth mode.

II.

EXPERIMENTAL

The vacuum system used for the UPS and PAX experiments consisted of a stainless steel vessel, equipped with a CMA-Auger electron spectrometer, a three grid LEED optics, a quadrupole mass spectrometer, a sputter gun and a helium resonance lamp together with a hemispherical electron energy analyser to register He(I) and He(II) excited UV photoelectron spectra. The sample holder enabled a tilt of the sample about an axis parallel to the sample’s surface as well as direct heating to 1200 K and cooling down to 60 K by a closed-cycle two stage He refrigerator. The sample temperature was measured by a NiCr-Ni thermocouple. The Pd(110) surface was cleaned by repeated cycles of sputtering with 3 keV Argon ions at T = 650 K and annealing to T = 1100 K. After annealing to 1100 K the sample was allowed to cool down to 600 K, where the residual carbon was removed in 7.0×10−4 Pa of

c 2009 The Surface Science Society of Japan (http://www.sssj.org/ejssnt) ISSN 1348-0391 °

448

e-Journal of Surface Science and Nanotechnology oxygen, followed by desorption of any remaining oxygen at 1100 K. The deposition of Au on the Pd(110) sample was achieved by evaporation from a Knudsen cell. This cell had been constructed from an Al2 O3 crucible with a diameter of 5 mm. It was filled with an Au wire and closed by a 2-hole ceramic. The Knudsen cell was heated by a tungsten wire (diameter 0.3 mm) wound around the crucible and thermally shielded by a water-cooled jacket. In order to control the deposition time a rotatable shutter was placed in front of the cell opening. The base pressure after bake out reached the 1.0 × 10−8 Pa range, while the working pressure during Au deposition was below the 1.0 × 10−7 Pa. The Au deposition rate was determined by work function measurements and adjusted to 0.5 ML min−1 . For the STM investigations a second stainless steel UHV chamber, containing a home- built VT-STM was used, whose scanning head had been designed according to a design of Stipe at al. [12]. Furthermore, this chamber was equipped with an Auger spectrometer and a quadrupole mass spectrometer. After bake out a final base pressure of 1.0 × 10−8 Pa was established, which also did not exceed 1.0 × 10−7 Pa during Au deposition. The construction enabled an operating temperature range between 150 and 300 K. Sample preparation was possible up to 1100 K. In this chamber the Au deposition was evaporated from a resistively heated tungsten basket containing an Au droplet. The deposition rate, as determined by STM for submonolayer coverages, was adjusted to 0.025 ML min−1 . All STM data were processed using the WSxM [13] freeware image processing software. III. A.

RESULTS AND DISCUSSIONS Deposition of Au on Pd(110)

The deposition of thin Au films up to a thickness of around 3 monolayers (ML) has been investigated in this work. It is expected that such small Au coverages should be affected noticeably by the (1 × 1) lattice of the underlying Pd(110) surface. In Fig. 1(a) the He(I) exited photoemission spectra (hν = 21.2 eV) of Au overlayers deposited on Pd(110) at room temperature are displayed in a differential form. Already at a coverage of 0.25 ML an Au related peak at 5.5 eV is clearly visible. With increasing coverage it shifts towards higher binding energy up to a final value of 6.0 eV at a monolayer coverage. At higher Au coverages the peak energy does not change any more. The same behavior of this characteristic peak stemming from the Au 5d3/2 derived sub band has been observed e.g. by Salmeron et al. [14] and Kobiela et al. [15] on different Pt surfaces and by Citrin et al. [16] on a glass substrate, indicating its independence from the used substrate. At a coverage of 0.5 ML and higher additional Au d-band derived peaks at 2.9, 3.5 and 4.3 eV become visible and gain in intensity with increasing Au amount. These peaks show no energy shifting at all. Furthermore, it can be determined from the integral spectra (Fig. 1(b)) that with increasing Au amount on the surface the electron density near the Fermi edge characteristic for the Pd substrate is continuously decreasing. The peak at 7.8 eV, visible at Au coverages of 0.25 and 0.5 ML can be assigned to

Volume 7 (2009)

small CO impurities, which were adsorbed during the Au deposition process from the residual gas. At higher Au coverages this peak is no longer visible any more because CO is not able to adsorb on a closed Au surface at room temperature [17]. Figure 1(c) shows the variation of the work function values as a function of the Au coverage on the surface calculated from the afore discussed He(I) exited UPS spectra. After a short increase at the beginning of the Au deposition, which could be due to the mentioned CO contamination, we observe an exponential decrease of the work function (φ) down to a final value of 5.36 eV, which matches perfectly the φ value obtained on an Au(110) single crystal surface [18]. This suggests that evaporating Au under the used conditions results in quite flat and well-ordered films. Furthermore, the exponential decrease of φ with increasing Au coverage points to a Volmer-Weber or island growth mode of Au on Pd(110). This growth mode is also verified by our STM measurements, which have been performed with Au coverages of 0.3, 0.6, 1.0 and 2.5 ML, since the second layer starts to grow before the first one is completed. As displayed in Fig. 2 the Au islands are orientated along the [1¯10] direction of the Pd substrate. They are not atomically compact, but are rather rugged on the nanometer scale, especially at the lower Au coverages. Although the islands at 2.5 ML Au coverage become more compact the Au film is far from being flat. At the highest Au coverage some nanometer sized regions of (1 × 2) and, very rarely, also of (1 × 3) periodicity can be detected, while at the lower coverages no ordered structures have been observed at all. Both reconstructions, i.e. (1 × 2) and (1 × 3), had been reported several times in the literature for bulk Au(110) [19–22]. This indicates that at least for multilayer coverages the effect of the Pd substrate is not strong enough to prevent a reconstruction of the Au film. B.

Au multilayer films on Pd(110) at different temperatures

Au is known to form ordered alloys with Pd already at moderate temperatures [23]. Since the first LEED studies by Schmitz et al. [8] it was assumed that annealing thin Au films to around 600 K results in a stoichiometric Au-Pd surface alloy. To study this interesting effect we prepared a 3 ML thick Au film on a clean Pd(110) surface at 150 K and stepwise annealed it for 5 minutes, respectively, up to a final temperature of 1050 K. In Fig. 3(a) the He(II) exited photoemission spectra (hν = 40.8 eV) are displayed. Up to an annealing temperature of 350 K no significant change in the spectrum shape or intensity can be observed. At 450 K a first increase of the Pd related intensity at 1.1 eV is detectable. Further annealing to 650 K results in additional intensity near the Fermi edge. At the same time the characteristic Au peak at 6.0 eV starts to decrease in intensity and shifts to lower energy down to a final value of 5.3 eV at 950 K, indicating Au diffusion from the surface into the Pd bulk. After annealing to 1050 K no Au related intensity is detected any more, proving that the Au film has been completely dissolved in the Pd bulk. In order to reach a more quantitative information on

http://www.sssj.org/ejssnt (J-Stage: http://www.jstage.jst.go.jp/browse/ejssnt/)

449

Volume 7 (2009)

Moors, et al.

(a)

(b) He(I)-UPS spectra at 330 K He(I)-UPS spectra at 330 K

3.0 ML Au

2.0 ML Au 1.0 ML Au

3.0 ML Au

Intensity [a.u.]

2.0 ML Au

Intensity [a.u.]

0.8 ML Au

0.5 ML Au

0.3 ML Au

1.0 ML Au 0.8 ML Au

clean Pd 0.5 ML Au 0.3 ML Au clean Pd

0

2

4 6 Energy [eV]

8

10

0

2

4 6 Energy [eV]

8

10

(c)

Work function [eV]

5.9 5.8 5.7 5.6 5.5 5.4 5.3 0.0

0.5

1.0

1.5 2.0 Au coverage [ML]

2.5

3.0

FIG. 1: Summary of UPS spectra (He(I) radiation: hν = 21.2 eV) taken from various Au films deposited on Pd(110) at 300 K shown in (a) differential and (b) integral form and (c) work function variation as a function of the Au amount on the surface.

the Au amount on the surface an integration of the Au 5d band related intensity of the Pd background subtracted He(II) exited UPS spectra was used (see Fig. 3(b)). Annealing the surface to 450 K leads to a strong decrease of the formerly stable peak intensity. This strong decrease of the Au signal proceeds up to 650 K. The loss of more than 80% of the Au related signal from the near surface region in the temperature range between 450 and 650 K is a clear indication for the Au-Pd alloying process. Above this temperature the residual peak intensity decreases again slowly up to 950 K, where it vanishes completely. The data suggest that an Au-Pd surface alloy is present in the temperature range from at least ∼ 450 to ∼ 850 K. Due to the finite information depth of He(II) excited UPS spectra it is certainly not possible to determine exactly when Pd atoms finally reach the outermost atomic layer. But we suggest that the first occurrence of the Pd peak at 1.1 eV visible already at 450 K in Fig. 3(a) is 450

an effect caused by the onset of the alloying process at the Au-Pd interface while the main rise of Pd related intensity at 650 K in combination with the behavior of the Au peak at 6.0 eV indicates that the alloying process has reached the surface. In order to verify this assumption we decided to use a very surface sensitive variant of UPS: the photoemission of adsorbed xenon atoms (PAX). The power of the PAX method as a technique for surface studies is based on the invariance of the binding energies of the Xe 5p1/2 photoelectrons with respect to the vacuum level which is due to the weak interaction of xenon with surfaces and its rather large atomic diameter (∼ 4.5 ˚ A) [24]. This results in a high sensitivity of PAX to local differences in the work function of a heterogeneous surface. In contrast to regular UPS experiments the PAX spectra are only affected by the topmost substrate layer. The occupation of different sites on heterogeneous surfaces leads to a superposition of the Xe 5p spectra, shifted by the corresponding differ-


e-Journal of Surface Science and Nanotechnology (a)

(b)

Volume 7 (2009) (a) UPS He(II) spectra 3 ML Au on Pd(110)

1050 K 950 K

850 K

Intensity [a.u.]

(c)

750 K

(d)

650 K 550 K

450 K 350 K

250 K 150 K

0

4 Energy [eV]

6

8

(b) 100 80 Au amount [%]

FIG. 2: STM images of Au-Pd(110) surfaces with a total Au coverage of (a) 0.3 ML (120.6 nm×120.6 nm, Ubias = 0.29 V, IT = 487 pA), (b) 0.6 ML (100.2 nm×100.2 nm, Ubias = 0.25 V, IT = 520 pA), (c) 1.0 ML (102.0 nm×102.0 nm, Ubias = 2.89 V, IT = 718 pA) and (d) 2.5 ML (124.8 nm×124.8 nm, Ubias = 2.00 V, IT = 712 pA) deposited at 300 K. The Au islands are orientated along the [1¯ 10] direction of the Pd substrate.

2

60 40 20

ences in the “local work function” [25]. This makes the PAX method an interesting alternative for the morphological characterization of multicomponent systems like bimetallic surfaces [26–29] and alloys [30–35]. Performing He(I) exited UPS spectra on a homogeneous surface covered with xenon atoms results usually in two additional Xe related peaks which can be assigned to the 5p3/2 and the 5p1/2 electronic final states of Xe. Due to the fact that the Xe 5p3/2 peak at lower binding energy is composed of two peaks with different magnetic quantum numbers [36], it is decidedly broader than the Xe 5p1/2 peak, so that we will focus in our discussion only on the 5p1/2 peak. Figure 4 shows the background subtracted PAX spectra of a freshly at 150 K prepared and stepwise annealed 3 ML thick Au film on Pd(110) with a total coverage of 5 L Xe. The Xe adsorption on the respective surface was performed always below 80 K. Up to an annealing temperature of 350 K we can observe a continuous shifting of the Xe 5p1/2 peak towards lower binding energy. This can be explained by a smoothening of the Au film, which was also observed with STM [11], due the increased mobility of the Au atoms with increasing temperature. Above 350 K the peak position remains stable at 6.9 eV but at 650 K the rise of a second 5p1/2 peak at 7.1 eV can be observed. We assume that this new state can be assigned to Xe atoms adsorbed on an Au-Pd surface alloy. The intensity of this peak reaches a maximum at 750 K indicating a maximal concentration of Au-Pd alloy in the topmost layer after annealing to

0 200

300

400

500 600 700 Temperatue [K]

800

900

1000

FIG. 3: (a) Summary of UPS spectra (He(II) radiation: hν = 40.8 eV) of a 3 ML thick Au film deposited on Pd(110) at 150 K and annealed up to the given temperatures for 5 min. in each case and (b) relative intensity of Au 5d band related peaks as a function of annealing temperature. The values can be interpreted as an indicator for the relative Au amount on the surface. Markings indicate the development of the Pd related intensity (box) and of the characteristic Au peak (line).

this temperature. At 850 K even a third Xe species can be detected at a 5p1/2 peak energy of 7.3 eV. With further increasing annealing temperature this becomes the only remaining species on the surface, suggesting that it is due to Xe atoms adsorbed on pure Pd. Thus, the PAX measurements indicate the existence of an Au-Pd surface alloy within a temperature range between 600 and 900 K. The alloyed Au-Pd surface has also been investigated by STM. Figure 5 shows a 2.5 ML thick Au film, which has been annealed to 600 K. Obviously the surface is much smoother after this treatment. The anisotropy of the Au islands is still present, but their average size has increased significantly. Also the size of the areas, in which atomic resolution could be achieved, differs in magnitude compared to the surface prepared at room temperature. Al-


451

Volume 7 (2009)

Moors, et al. (a)

(b)

1050 K 950 K

850 K

Intensity [a.u.]

750 K 650 K 550 K 450 K 350 K

250 K

5p3/2

5

5p1/2

6 7 Energy [eV]

150 K

8

FIG. 4: Summary of UPS spectra (He(I) radiation: hν = 21.2 eV) of Xe-covered Au films deposited on Pd(110) at 150 K and annealed up to the given temperatures for 5 min. in each case. All spectra except the Xe-covered clean Pd(110) are background subtracted. Xe exposures are all 5 L. In all cases 5 L Xe were adsorbed. Lines indicate the different Xe species.

though STM is of course not very suited for quantitative assessment the Au amount on the surface seems to be reduced, too. Former STM studies with lower Au coverage [11] have shown a decrease of the alloying temperature with decreasing Au amount on the surface, supporting our assumption that the Au diffusion is initiated at the Au-Pd interface already at much lower temperatures.

C.

CO adsorption

The change of the chemical properties of a surface is an important effect that can be induced by surface alloy formation in multicomponent systems. In contrast to Pd, Au is known to be inactive for CO adsorption at moderate temperatures [37, 38]. This leads to the question if alloying a Pd surface with Au results in a changed reactivity towards CO, what may be an advantage for many Pd catalyzed reactions where CO acts as a catalytic poison. To answer this question we dosed 10 L of CO on differently prepared Au-Pd surfaces, which were produced by annealing an Au film on Pd(110) to the temperatures indicated in Fig. 6. There the He(II) exited photoemission spectra after CO exposure at temperatures of 80 K (Fig. 6(a)) and at room temperature (Fig. 6(b)) are shown. For the low adsorption temperature weak CO related peaks at 10.2 eV, 12.1 eV and 13.5 eV can be found on the freshly prepared Au film and on the surfaces, which have been 452

FIG. 5: STM images of an Au coverage of 2.5 ML deposited on Pd(110) at 300 K after 5 min. of annealing to 600 K ((a) 148.9 nm—By 148.9 nm, Ubias = 1.99 V, IT = 962 pA and (b) 18.4 nm×18.4 nm, Ubias = 1.99 V, IT = 953 pA). The Au islands still exhibit a strong anisotropy.

annealed up to 550 K. These peaks are in perfect agreement with the overlapping 5σ/1π and 4σ orbitals and a satellite peak, respectively, which had been observed in CO adsorption studies of Duckers et al. on a (1 × 2) reconstructed Au(110) single crystal surface [39]. With the onset of alloy formation in the topmost atomic layer we observe a continuous shifting of the 5σ/1π orbital peak towards lower binding energy up to an annealing temperature of 850 K. The 4σ orbital peak and the satellite peak remain stable. In agreement with our PAX measurements also the rise of an additional peak at 8.2 eV is observed at this temperature, which can be clearly assigned to the 5σ/1π peak of CO adsorbed on pure Pd. At 950 K and above this peak and its associated 4σ peak at 11.2 eV are the only remaining CO peaks, evidencing once again that no Au is left in the topmost atomic layer. Dosing CO at room temperature changes the adsorption properties of the investigated surfaces considerably. As expected CO does not adsorb on the closed Au film which covers the majority of the surface up to 550 K. But even after annealing up to 750 K no CO adsorption is detectable, although due to the alloying process Pd atoms should be available on the surface. The Au-Pd surface alloy seems to be rather unreactive for CO adsorption at room temperature. After annealing to 850 K first CO related peaks at 7.8 and 10.8 eV become visible, which rise in intensity with further increasing annealing temperature. These peaks can be assigned to CO molecules adsorbed on bare Pd patches. Our PAX measurements have shown that such patches are formed during the progressing Au diffusion into the Pd bulk at high annealing temperatures. Interestingly, the CO binding energy on the Pd patches is decreased by 0.4 eV when dosing the molecules at room temperature and not at 80 K, which indicates a different bonding type at the two investigated dosing temperatures. Such a temperature dependence has also been observed in the LEED studies of CO adsorbed on clean Pd(110) by He et al. [40]. They observed a reversible phase transition from the at higher CO doses dominant (2 × 1) CO phase to a (4 × 2) CO phase by changing the exposition temperature from 130 to 330 K and explained it with a possible displacement of Pd surface atoms.


e-Journal of Surface Science and Nanotechnology

Volume 7 (2009)

(a)

(b) 1050 K

950 K

1050 K

850 K

950 K

750 K

Intensity [a.u.]

850 K

550 K

450 K

Intensity [a.u.]

650 K 750 K

650 K

550 K

350 K

450 K

250 K

150 K

UPS He(II) spectra 8 10 L CO at 80 K

10 12 14 Energy [eV]

350 K

UPS He(II) spectra 8 10 L CO at 300 K

10 12 Energy [eV]

14

FIG. 6: Summary of UPS spectra (He(II) radiation: hν = 40.8 eV) of CO-covered Au films with a thickness of 3 ML deposited on Pd(110) at 150 K and annealed up to the given temperatures for 5 min. in each case. The CO was adsorbed at (a) 80 K and (b) 300 K, respectively; in all cases 10 L. Lines indicate the different CO orbital peaks.

IV.

CONCLUSIONS

In this work we investigated the deposition and annealing behavior of Au films on Pd(110). Work function measurements and STM images clearly indicate a VolmerWeber or island growth mode during deposition at room temperature or below. Au coverages below 1 ML form unreconstructed pseudomorphic Au islands, which adopt the structure of the unreconstructed Pd(110)-(1 × 1) surface underneath and are orientated along the [1¯ 10] direction of the substrate. Above a critical thickness of 2 ML the formation of a (1 × 2) missing-row reconstruction typical for Au has been found. This reconstructed Au bilayer is quite flat and still pseudomorphic with the Pd(110) substrate. The electronic and morphologic properties of Au multilayer films are strongly temperature dependent. Annealing a film to room temperature, which has been prepared at 150 K, results in an ordering of the topmost atomic layer. At 450 K an alloy formation is starting at

[1] R. M. Lambert in Chemisorption and Reactivity on Supported Clusters and Thin Films, ed. R. M. Lambert and

the Au-Pd interface. The anisotropy of the Au islands on the surface is still present, but their average size has increased significantly. Further annealing leads to the formation of an Au-Pd surface alloy, which is stable up to ∼ 850 K. A maximum of the Au-Pd alloy concentration in the topmost layer can be detected after annealing to 750 K for 5 minutes. At higher annealing temperatures than 900 K the remaining Au amount is quickly decreasing at the surface. The surface alloy exhibits only a low reactivity towards CO adsorption at room temperature, which is comparable to that of a closed Au(110) surface and much lower than observed on pure Pd. Acknowledgments

T. Kobiela and M. Kralj express their appreciation to the Alexander von Humboldt foundation for financing their research stays in Germany.

G. Pacchioni (Kluwer, Dordrecht, 1997), pp. 1-26.


453

Volume 7 (2009)

Moors, et al.

[2] J. Zhang, K. Saski, E. Sutter, and R. R. Adzic, Science 315, 220 (2007). [3] B. Hammer and J. K. Nørskov, Nature 376, 238 (1995). [4] R. Meyer, C. Lemire, S. K. Shaikhutdinov, and H.-J. Freund, Gold Bulletin 37, 72 (2004). [5] S. Ichikawa, T. Akita, M. Okumura, M. Koyhyama, and K. Tanaka, JEOL News 38, 6 (2003). [6] J. A. Nieminen, Phys. Rev. Lett. 74, 3856 (1995). [7] P. J. Schmitz, H. C. Kang, W.-Y. Leung, and P. A. Thiel, Surf. Sci. 248, 287 (1991). [8] P. J. Schmitz, W.-Y-Leung, H. C. Kang, and P. A. Thiel, Phys. Rev. B 43, 8834 (1991). [9] P. Kaukasoina, M. Lindroos, O. L. Warren, and P. A. Thiel, Surf. Sci. 318, 243 (1994). [10] M. Vos and I. V. Mitchell, Phys. Rev. B 45, 9398 (1992). [11] M. Kralj, A. Bailly, M.-C. Saint-Lager, S. Degen, A. Krupski, C. Becker, P. Dolle, M. De Santis, and K. Wandelt, Surf. Sci. 600, 2614 (2006). [12] B. C. Stipe, M. A. Rezaei, and W. Ho, Rev. Sci. Instrum. 70, 137 (1999). [13] I. Horcas, R. Fernandez, J. M. Gomez-Rodriguez, J. Colchero, J. Gomez-Herrero, and A.M. Baro, Rev. Sci. Instrum. 78, 013705 (2007). [14] M. Salmeron, S. Ferrer, M. Jazzar, and G. A. Somorjai, Phys. Rev. B 28, 6758 (1983). [15] T. Kobiela, M. Moors, W. Linhart, I. Cebula, A. Krupski, C. Becker, and K. Wandelt, in preparation. [16] P. H. Citrin and G. K. Wertheim, Phys. Rev. Lett. 41, 1425 (1978). [17] A. Sandell, P. Bennich, A. Nilsson, B. Hernnas, O. Bjorneholm, and N. Martensson, Surf. Sci. 310, 16 (1994). [18] J. M. Gottfried, K. J. Schmidt, S. L. M. Schroeder, and K.Christmann, Surf. Sci. 525, 197 (2003). [19] D. G. Fedak and N. A. Gjostein, Acta Metall. 15, 827 (1967). [20] W. Moritz and D. Wolf, Surf. Sci. 88, L29 (1979). [21] G. Binnig, H. Rohrer, Ch. Gerber, and E. Weibel, Surf. Sci. 131, L379 (1983).

454

[22] A. Held, J. L. Jordan-Sweet, P. M. Horn, A. Mak, and R. J. Birgeneau, Solid State Commun. 72, 37 (1989). [23] R. P. Elliot and F. A. Shunk, Bull. Alloy Phase Diagrams 2, 482 (1982). [24] K. Wandelt, Appl. Surf. Sci. 111, 1 (1997). [25] A. Jablonski and K. Wandelt, Surf. Interface Anal. 17, 611 (1991). [26] A. Jablonski, S. Eder, and K. Wandelt, Appl. Surf. Sci. 22/23, 309 (1985). [27] K. Wandelt, K. Markert, P. Dolle, A. Jablonski, and J. W. Niemantsverdriet, Suf. Interface Anal. 12, 15 (1988). [28] M. Alnot, V. Gorodetskii, A. Cassuto, and J. J. Ehrhardt, Surf. Sci. 162, 886 (1985). [29] J. A. Barnard, J. J. Ehrhardt, H. Azzouizi, and M. Alnot, Surf. Sci. 211/212, 740 (1989). [30] G. Sch¨ onhense, A. Eyers, U. Friess, F. Schafers, and U. Heinzmann, Phys. Rev. Lett. 54, 547 (1985). [31] J. K¨ uppers, K. Wandelt, and G. Ertl, Phys. Rev. Lett. 43, 928(1979). [32] G. R. Castro, H. Isern, U. Schneider, M. St¨ ocker, and K. Wandelt, Vacuum 41, 393 (1990). [33] J. Hulse, J. K¨ uppers, K. Wandelt, and G. Ertl, Appl. Surf. Sci. 6, 453 (1980). [34] T. Berghaus, C. Lunau, H. Naddermayer, and V. Rogge, Surf. Sci. 182, 13 (1987). [35] D. Fargues, J. J. Ehrhardt, M. Abon, and J. C. Bertolini, Surf. Sci. 194, 149 (1988). [36] B. J. Waclawski and J. F. Herbst, Phys. Rev. Lett. 35, 1594 (1975). [37] M. Mavrikakis, P. Stoltze, and J. Nørskov, Catal. Lett. 64, 10 (2000). [38] S. R. Bahn, N. Lopez, J. Nørskov, and K. W, Jacobsen, Phys. Rev. B 66, 081405(R) (2002). [39] K. D¨ uckers and H.P. Bonzel, Surf. Sci. 213, 25 (1989). [40] J.-W. He and P. R. Norton, J. Chem. Phys. 89, 1170 (1988).
