Available online at www.sciencedirect.com
Surface Science 602 (2008) 499–505 www.elsevier.com/locate/susc
Silver nanostructures on a c(4 · 2)-NiOx/Pd(1 0 0) monolayer S. Agnoli a, M. Sambi a, G.A. Rizzi a, G. Parteder b, S. Surnev b, F.P. Netzer b, G. Granozzi a,* b
a Dipartimento di Scienze Chimiche, INSTM and INFM Research Unit, Universita` di Padova, Padova, Italy Institut fu¨r Physik, Oberfla¨chen-und Grenzfla¨chenphysik, Karl-Franzens Universita¨t Graz, A-8010 Graz, Austria
Received 25 September 2007; accepted for publication 31 October 2007 Available online 7 November 2007
Abstract The growth, morphology and epitaxial relationship of Ag nanostructures deposited onto the c(4 · 2)-NiOx/Pd(1 0 0) surface have been investigated by photoemission (both core and valence levels), scanning tunneling microscopy and angle-scanned photoelectron diffraction. Small Ag nanoparticles are obtained on the terraces, whereas the tendency of Ag to decorate the step edges can lead to the formation of extended nanowire-like features. The Ag nanoparticles adopt a fcc structure epitaxially related to the substrate according to the following relationship: Ag(1 0 0)[0 0 1]//Pd(1 0 0)[0 0 1], while by means of STM we have found that the shape of the islands is typically rectangular with the edges running along the [0 1 1] directions. According to the reported data, the NiOx ML is locally disrupted in the vicinity of the Ag nanoparticles so that the c(4 · 2)-NiOx/Pd(1 0 0) surface cannot be considered as an efficient template to stabilize metal nanoparticles against Ostwald ripening phenomena. 2007 Elsevier B.V. All rights reserved. Keywords: Ag nanoclusters; NiO/Pd(1 0 0) monolayer; STM; Photoemission; Photoelectron diffraction
1. Introduction Metal nanostructures show a large variety of interesting properties finding applications in many different fields, such as microelectronics and photonics [1], energetics [2], magnetism [3] and catalysis [4]. These properties are in general critically dependent on the structure of the nanoparticles at the atomic level. Therefore, many efforts have been undertaken to develop efficient synthetic routes to accurately control the size and the morphology of metal nanostructures [5]. Nanofabrication, i.e. the control of the shape, dimensions and order of such nanostructures, can be achieved with the wellknown sequential top-down approach, e.g. by nanolithographic methods, or by atomic manipulation via Scanning Tunneling Microscopy (STM) [6]. Alternatively, size selected cluster deposition methods can be used, even if clus*
Corresponding author. Tel.: +39 049 8275158; fax: +39 049 8275161. E-mail address:
[email protected] (G. Granozzi).
0039-6028/$ - see front matter 2007 Elsevier B.V. All rights reserved. doi:10.1016/j.susc.2007.10.047
ter ordering is difficult to achieve [7]. However, the current frontier in nanofabrication is associated with the formation of nanostructures by a fast parallel self-assembly process. One of the most promising techniques is the use of a suitable template surface (e.g. an oxide ultrathin film with an ordered array of defects which act as preferential nucleation centres), which is able to orient the subsequent growth of metal particles leading to the desired surface nanostructures [8–10]. Ag based nanostructures, in particular, play an important role in nanotechnology since this material is very suitable for the development of basic device elements, such as electrical devices with quantum conductance [11] or plasmon based optical elements [12]. Moreover, Ag nanoparticles (NP) and nanowires are also widely used as catalysts [13] and are at the forefront of the new single molecule sensors [14,15]. For these reasons, the study of surface supported Ag NPs is widely addressed a topic in the literature: a large amount of papers considering widely different substrates, ranging from semiconductors [16,17] to
500
S. Agnoli et al. / Surface Science 602 (2008) 499–505
oxides [18–20], from flat to vicinal surfaces [21,22], have been published. Also, Ag on NiO is of relevance because Ag is used as a promoter in NiO based catalysts [23,24] or as an additive to tailor the optical properties in electrochromic devices [25]. In recent years, the reversed catalyst model system i.e. the growth of NiO on Ag(1 0 0), has been thoroughly studied and the NiO/Ag(1 0 0) interface has assumed the status of a reference system for a well-matched ultrathin film [26– 32]. We have reported previously the structure and morphology of NiO/Pd(1 0 0) ultrathin films, with particular attention towards novel structures and properties stemming from the large lattice mismatch [33–37]. In the monolayer (ML) coverage range a new defective NiOx c(4 · 2) structure (x = 1.33) has been thoroughly investigated by both static and dynamic low energy electron diffraction (LEED) [36], STM [37], and photoemission spectroscopy (both valence and core levels) [38]. In the following we present a morphological and chemical characterization of the Ag/ NiOx/Pd(1 0 0) system, prepared by Ag UHV evaporation on the c(4 · 2) NiOx/Pd(1 0 0) ML film, for different Ag coverages. The choice of the investigated system was dictated by the goal of testing if the defective c(4 · 2) ML is sufficiently stable to act as an effective template for the growth an ordered array of Ag NPs, as demonstrated in other cases [8–10]. 2. Experimental The experiments have been performed in two different UHV systems. The first one, used for the photoemission experiments (both core, XPS, and valence band, UPS, measurements), is a two-chamber UHV system (a modified VG ESCALAB MK II, Vacuum Generators, Hastings, England), equipped with a four grids rear view LEED, two electron beam evaporators with an integrated flux monitor, a quadrupole mass spectrometer, a twin (Mg/Al) anode Xray source, a discharge lamp for noble gas ionization (VUV HIS 13 Omicron), a sputter gun, and a hemispherical electrostatic analyzer ending with a five channeltrons detector. The sample is mounted on a two-axis goniometer capable of sample rotations in polar angle h (defined with respect to the surface normal) and in azimuthal angle /, (defined with respect to the [0 0 1] direction of the substrate surface), thereby allowing us to collect angle-scanned X-ray photoelectron diffraction (XPD) data. Angular accuracy is always better than ±1 in both directions. The angular acceptance of the analyzer can be varied between 1.5 and 8 (the latter used for UPS experiments). The binding energy (BE) calibration was performed using the Fermi Edge (EF) and 4f peaks of a gold sample. The STM experiments were performed in Graz, in a UHV system equipped with a variable-temperature STM (Oxford Instruments), LEED, an Auger electron spectrometer (AES), and facilities for crystal cleaning and metal evaporation (Omicron EMF3 triple electron beam evapo-
rator). The STM images were recorded in constant current mode at RT, with electrochemically etched W tips, which have been cleaned in situ by electron bombardment. Typical tunneling conditions employed for imaging were 1.5 V and 0.1 nA. For all the experiments the Pd crystal was cleaned by repeated cycles of argon ion sputtering (E = 2 keV), and annealing in 1 · 106 mbar O2 (uncorrected ion gauge reading) and a final flash to T = 970 K. The cleanliness of the substrate surface prior to the experiments was always checked by using XPS or AES and the surface order was probed by means of LEED. In order to grow the c(4 · 2)-NiOx structure the standard recipe was followed, described in details in Ref. [34]. Silver was evaporated using an e-beam evaporator under UHV conditions at room temperature (RT). The evaporation rate of both Ni and Ag was determined by angle resolved XPS (ARXPS) in the photoemission studies or by a quartz microbalance calibration in the STM experiments. The deposition rates employed for Ni and Ag were 0.5 MLE/min and 0.2 MLE/min, respectively. One monolayer equivalent (MLE) is referred to the atom density of the Pd(1 0 0) surface and corresponds to 1.3 · 1015 metal atoms/cm2. 3. Results and discussion In the present study, we have investigated the chemical, morphological and structural evolution of the Ag/NiOx(1 ML)/Pd(1 0 0) system for increasing amounts of Ag. In each experiment, the initial surface was prepared by reactively depositing 0.75 MLE NiO ðP O2 ¼ 106 mbarÞ on Pd(1 0 0) at RT and successively annealing the sample at 250 C in 5 · 107 mbar O2. Under these conditions a flat continuous layer of c(4 · 2)-NiOx is formed with very few vacancy islands. For a detailed description of this surface we refer to our previous work [34–37]. 3.1. XPS results In Fig. 1 we report the photoemission data of the Ag/ NiOx/Pd(1 0 0) system as a function of the Ag coverage. All core level spectra were taken at grazing incidence (h = 70 from the surface normal) with an Al Ka nonmonochromatized source. Fig. 1a reports the XPS Ni 2p. For a detailed interpretation of the Ni 2p photoemission spectrum of the c(4 · 2) phase we refer the reader to a previous publication [38]. Here it is sufficient to note that as soon as Ag is deposited, a shoulder appears at the higher BE side of the main peak, and the peak maximum progressively shifts towards higher BE while increasing the Ag dose (at 2.5 MLE DEBE = 0.8 eV). At 2.5 MLE a single peak is observable, whose full width at half maximum (FWHM) is slightly decreased (by 0.2 eV) with respect to the c(4 · 2) ML. The structure and the intensity of the satellite, on the other hand, do not undergo significant changes after the Ag deposition. Fig. 1b shows the XPS spectrum of the
S. Agnoli et al. / Surface Science 602 (2008) 499–505
501
Fig. 1. Photoemission data of the Ag/NiOx/Pd(1 0 0) system as a function of the Ag coverage: (a) Ni 2p XPS region (Al Ka source), (b) Ag 3d region (Al Ka source) and (inset) Ag MNN Auger peak for the 0.3 MLE deposit, (c) VB region (He-II source).
Ag 3d photoemission line. It can be clearly seen that the BE maximum of the Ag 3d5/2 component shifts progressively towards higher BE as a function of coverage (DEBE = 0.3 eV), while the FWHM remains constant. At 2.5 MLE the BE matches that of an Ag(1 0 0) bulk-like sample. In order to check the oxidation state of Ag on the surface, we have measured the Ag MNN peak (see the inset in Fig. 1b in the case of a 0.3 MLE deposit) to evaluate the Auger parameter. It turns out that this it changes from 720.1 eV in the case of the 0.15 MLE to 720.0 eV in the case of the 2.5 MLE, i.e. a difference which is not significant within the experimental error, while both values are compatible with a pure metallic state. The XPS region containing the Pd 3p3/2 and O 1s peaks is rather uninformative, due to the large overlap between the two peaks. Anyway, no significant change in the band envelope as a function of the Ag coverage is observed and the peak positions are those already reported for the substrate (Pd 3p3/2 component at 531.9 eV, O 1s component at 529.3 eV in NiOx/Pd(1 0 0)) [38]. Fig. 1c shows the He II excited VB spectra taken at normal emission (Cpoint) as a function of the Ag coverage. The most relevant trend is the progressive formation of two distinct peaks in the region between 3 eV and 7 eV which are clearly due to the progressive formation of the Ag 4d band. Just below EF (region between 0 and 3 eV), the changes are less distinct, but one can see that the features peculiar of the c(4 · 2)-NiOx/Pd(1 0 0) system [38] become broader and less intense because of the attenuating effect of the Ag overlayer and of the overlap with its 5sp band. 3.2. STM results The morphological changes resulting from the Ag deposition have been investigated by STM. After the deposition
of 0.15 MLE Ag on the c(4 · 2) NiOx/Pd(1 0 0) phase (Fig. 2a), several irregular Ag NPs can be seen on the surface. In particular, they preferentially decorate the (zigzag reconstructed) step edges caused by the NiOx layer. The ˚ and NPs have a size dispersion ranging between 20 A ˚ 50 A, and can be single (apparent height with respect to ˚ ) or double layered (3.5 A ˚ ). The inset the NiOx ML 1.8 A of Fig. 2a shows a high resolution image of an Ag NP on the c(4 · 2) superstructure: as it can be seen, the oxide layer is partially disrupted in the close proximity to the Ag NP. Increasing the coverage to 0.3 MLE (Fig. 2b) leads to a stronger decoration of the step edges, which straighten out and become decorated by long nanowire-like features ˚ . Also, islands running continuously for several hundred A localized on the terraces become more frequent and tend to assume better developed rectangular shapes. In addition, it is possible to distinguish the formation of a second layer on larger islands. The mean size of the islands on the terraces ˚ and the island edges increases to approximately 80–100 A are in general oriented along the {0 1 1} directions of the substrate (see Fig. 2c). In Fig. 2d–e two STM images corresponding to 0.6 MLE Ag are shown. In this case, the formation of extended smooth nanowires (up to several ˚ long) along the edges is evident. These features thousand A ˚ wide and mostly only one layer high, but some are 30–40 A two layer high sections can be identified as well. The islands on the terraces show almost the same mean size as observed in the case of the previous Ag dose (0.3 MLE), the increase of the metal coverage leads only to an increase of the fraction of two layer high NPs. At the Ag coverage of 0.6 MLE the LEED pattern of the c(4 · 2)-NiOx superstructure is still visible, although the diffraction spots are very diffuse and the background intensity is high. On the other hand, no superstructure, originating from the NPs is detectable.
502
S. Agnoli et al. / Surface Science 602 (2008) 499–505
˚ · 1000 A ˚ , V = 0.6 V; I = 0.2 nA), the inset Fig. 2. Evolution of the Ag/NiOx/Pd(1 0 0) surface for increasing Ag coverages: (a) 0.15 MLE Ag, (1000 A ˚ · 75 A ˚ , V = 1.25 V; I = 0.2 nA); (b) 0.3 MLE Ag shows an high resolution image of a silver cluster surrounded by the c(4 · 2) structure, (75 A ˚ · 1000 A ˚ , V = 1.5 V; I = 0.1 nA); (c) 0.3 MLE Ag (250 A ˚ · 250 A ˚ , V = 1.5 V; I = 0.1 nA); (d) 0.6 MLE Ag (1000 A ˚ · 1000 A ˚ , V = 1.6 V; (1000 A ˚ · 250 A ˚ , V = 0.5 V; I = 0.4 nA); (f) 1 MLE Ag (1000 A ˚ · 1000 A ˚ ,V = 1.5 V; I = 0.1 nA). I = 0.1 nA); (e) 0.6 MLE Ag (250 A
Fig. 2f shows the surface after deposition of 1 MLE Ag metal on the NiOx/Pd(1 0 0) substrate. The surface is covered by an almost continuous ML of Ag metal scattered on top by smaller islands and NPs. The long metal nanowire-like features, which decorated the edges before, are now embedded in the main Ag layer, which contains troughs along the main crystallographic directions due to uncovered oxide regions (see e.g. the ellipses in Fig. 2f). The LEED pat-
tern observed on this surface is a simple (1 · 1), with a high background and diffuse spots. No atomic resolution was possible on the NPs or on the Ag overlayer. 3.3. XPD results In order to obtain crystallographic information on the Ag NPs and to detect any possible epitaxial ordering, we
S. Agnoli et al. / Surface Science 602 (2008) 499–505
503
Fig. 3. 2p XPD plots of the Ag 3d intensity (Ekin 1119 eV) for the Ag/NiOx/Pd(1 0 0) system as a function of Ag coverage.(a) 0.3 MLE, (b) 0.8 MLE, (c) 1.5 MLE. The raw intensity has been normalized with respect to 1/cos h typical of ultrathin films. The colour scale ranges from black (minimum) to yellow (maximum). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
have undertaken an angle-scanned XPD investigation [39,40]. In Fig. 3 we report 2p plots of the intensity of the Ag 3d photoelectrons (Ekin 1119 eV) for different coverages. In this type of presentation, the centre of the plot corresponds to the surface normal, the radial section displays a polar scan, the circular section an azimuthal scan and the photoelectron intensity is given by the corresponding value of the colour scale. Exploiting the observed symmetry, the whole set of XPD data has been fourfold averaged; the polar angle ranges from h = 0 to h = 66. As the photoelectron kinetic energy is rather high, the interpretation of the Ag 3d XPD pattern can be carried out by a simple kinematical single-scattering model, where the strongest intensity enhancements are interpreted as derived from a forward-focusing effect [40]. Starting with the lowest coverage (Fig. 3), strong intensity maxima can be observed at h = 45 along the [0 0 1] direction and at 34 and 55 along the [0 1 1] azimuth. This pattern remains unchanged and becomes even more evident for the highest investigated coverage (1.5 MLE). These findings clearly indicate that the Ag nanostructures (either NPs or ultrathin film) grow azimuthally ordered, adopting the epitaxial (1 0 0)[0 0 1]//(1 0 0)[0 0 1] relationship with respect to the
Pd(1 0 0) substrate. The presence of the intensity maximum at 45 along the [0 0 1] direction, which is observed already for an Ag dose of 0.3 MLE, can be explained assuming that the Ag NPs are at least two layers high, hence indicating a 3D growth mode. Similarly, the strong intensity maximum clearly observed along the surface normal at a coverage of 1.5 MLE is only compatible with the existence of a third layer. In Fig. 4 we compare the XPD patterns of the Ag 3d and Ni 2p photoemission peaks for a coverage of 0.8 MLE. Both plots present the typical fingerprint of a fcc (1 0 0) oriented structure. However in the case of Ni, the pattern is extremely weak. If the c(4 · 2)-NiOx structure remained intact after the Ag deposition and provided a simple epitaxial relationship between Ag and the underlying NiOx ML, one would expect a strong modulation of the Ni signal, since Ag is a strong scatterer. One has to consider that the c(4 · 2)-NiOx structure is essentially similar to a NiO(1 0 0) surface, hence the epitaxial relationship of Ag(1 0 0) clusters could be derived from the reverse system NiO(1 0 0)/Ag(1 0 0) [30,32], i.e. with the metal Ag atoms sitting on top of the c(4 · 2) oxygen atoms. This means that Ni and Ag atoms would follow an fcc stacking. Therefore,
Fig. 4. XPD pattern of the Ni 2p (Ekin 634 eV) and Ag 3d (Ekin 1119 eV) peaks taken for the Ag/NiOx/Pd(1 0 0) system at a 0.8 MLE Ag coverage. The raw intensity has been normalized with respect to 1/cos h typical of ultrathin films. The colour scale ranges from black (minimum) to yellow (maximum). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
504
S. Agnoli et al. / Surface Science 602 (2008) 499–505
if the Ag atoms give rise to strong scattering events at h = 45 along the [0 0 1] direction, which is typical of a two layer high fcc structure, in the case of Ni we would expect a strong maximum along the surface normal due to a three layers high fcc structure. The absence of such a peak in the Ni pattern excludes the possibility that the c(4 · 2)NiO phase remains unaltered after the Ag deposition. On the other hand, the faint XPD pattern indicates that NiOx clusters are formed with an fcc structure characterized by some residual order. One possible explanation is that some Ni atoms are diluted in the growing Ag film similarly to what has been reported in the case of the deposition of Au on NiO [41]. However, this would lead to the formation of a metal-like component in the Ni 2p XPS spectrum, which is not in tune with our findings (see Fig. 1). On the contrary, we observe a shift towards higher BE, which would better fit with the hypothesis of a transition of the c(4 · 2)-NiOx 2D monolayer to 3D NiO clusters, as already suggested in Ref. [38]. 3.4. General discussion Both STM and XPD data clearly evidence that the Ag growth follows a Stranski–Krastanov scheme, which is in agreement with the dewetting behaviour of Ag on bulk NiO [42]. The XPD data provide unequivocal evidence for an epitaxial relationship of the Ag nanostructures with respect to the Pd(1 0 0) substrate. This could be related to the disruption of the c(4 · 2)-NiOx template as a consequence of the Ag deposition. The most interesting morphological feature of the Ag nanostructures is the strong tendency to decorate the upper substrate step edges to form long features with a nanowirelike appearance. If we look at the details of the wetting
behaviour of the c(4 · 2)-NiOx layer we can see that the upper rim of the step edges is often not covered by the NiOx layer: this may be seen in the STM picture of Fig. 5 where a detail of the step edge is shown [34]. If we consider that this is the area which is decorated during the first stages of the Ag deposition, we can argue that the nanowire-like features are characterized by a direct Pd–Ag interaction. The presence of a Pd–Ag interface would also explain why at very low coverages, when most of the Ag is concentrated at step edges, the BE of the Ag 3d is shifted to lower values (see Fig. 1) with respect to the BE typical of bulk Ag. Actually it has been demonstrated both theoretically and experimentally that Pd–Ag alloys and bimetallic NPs show the same behaviour [43], which originates from the change in the atomic charge due to the hybridization between VB electrons [44]. On the contrary, in the case of Ag NPs on oxides, the observed BE shift is always towards higher BE because of strain and initial state effects [45–47]. 4. Conclusions The growth of a metal on oxides is an extremely important topic for the development of tailored nanostructures as well as to get some insight into heterogeneous catalysis, where metallic NPs grown on ultrathin oxide films are used as model systems. We have reported on the growth morphology of Ag nanostructures (nanoparticles, nanowire-like, nanolayers) deposited onto the c(4 · 2)-NiOx/ Pd(1 0 0) surface. Small clusters are obtained on the terraces, whereas the tendency of Ag to decorate the step edges can lead to the formation of extended nanowire-like features, reinforcing the general idea of using a stepped substrate as an efficient template for the growth of an array of metal nanowires [48]. XPD measurements have evidenced that the NPs adopt a fcc structure epitaxially related to the substrate according to the following relationship: Ag(1 0 0)[0 0 1]//Pd(1 0 0)[0 0 1], while by means of STM we have found that the shape of the islands is typically rectangular with the edges running along the [0 1 1] directions. The chemical situation following the Ag deposition resulted to be quite complex: the NiOx ML is locally disrupted in the vicinity of the Ag NPs and gets destroyed as a function of Ag coverage and a rather disordered NiOx phase is formed. As a consequence, the c(4 · 2)-NiOx/ Pd(1 0 0) surface cannot be considered as an efficient template to stabilize metal nanoparticles against Ostwald ripening phenomena. On the other hand, chemical modifications of Ag are observed in XPS. In particular, it may be speculated that a Ag–Pd bimetallic alloy is formed which may be interesting for heterogeneous catalysis. Acknowledgements
Fig. 5. A zoom-in STM image of the step edge of the NiOx/Pd(1 0 0) ˚ · 100 A ˚ , V = 2.1 V; I = 0.1 nA). surface (100 A
This work has been funded by European Community through the STRP project (Growth and Supra-organization
S. Agnoli et al. / Surface Science 602 (2008) 499–505
of Transition and Noble Metal Nanoclusters) and by the Italian Ministry of Instruction, University and Research (MIUR) through the fund ‘‘Programs of national relevance’’ (PRIN-2003, PRIN-2005). The STM experiments in Graz are also supported by the Austrian Science Funds. References [1] (a) G.D. Scholes, G. Rumbles, Nature Mater. 5 (2006) 683; (b) E. Ozbay, Science 311 (2006) 189. [2] M. Za¨ch, C. Ha¨gglund, D. Chakarov, B. Kasemo, Current Opin. Solid State Mater. Sci. 10 (2006) 132. [3] Y. Shiratsuchi, M. Yamamoto, S.D. Bader, Prog. Surf. Sci. 82 (2007) 121. [4] (a) A.T. Bell, Science 299 (2003) 1688; (b) C.T. Campbell, A.W. Grant, D.E. Starr, S.C. Parker, V.A. Bondzie, Top. Catal. 14 (2001) 43; (c) A.K. Santra, D.W. Goodman, J. Phys.: Cond. Matt. 15 (2003) R31; (d) B. Hammer, J. Nørskov, Nature 376 (1995) 238. [5] C. Burda, X. Chen, R. Narayanan, M.A. El-Sayed, Chem. Rev. 105 (2005) 1025. [6] R. Garcia, R.V. Martinez, J. Martinez, Chem. Soc. Rev. 35 (2006) 29. [7] C. Binns, Surf. Sci. Rep. 44 (2001) 1. [8] S. Degen, C. Becker, K. Wandelt, Faraday Discuss. 125 (2004) 343. [9] G. Hamm, C. Becker, C.R. Henry, Nanotechnology 17 (2006) 1943. [10] F. Sedona, S. Agnoli, M. Fanetti, I. Kholmanov, E. Cavaliere, L. Gavioli, G. Granozzi, J. Phys. Chem. C 111 (2007) 8024. [11] J. Zhao, C. Buia, J. Han, J. Ping Lu, Nanotechnology 14 (2003) 501. [12] H. Ditlbacher, A. Hohenau, D. Wagner, U. Kreibig, M. Rogers, F. Hofer, F.R. Aussenegg, J.R. Krenn, Phys. Rev. Lett. 95 (2005) 257403. [13] Y. Chen, C. Wang, H. Liu, J. Qiu, X. Bao, Chem. Commun. (2005) 5298. [14] S. Nie, S.R. Emor, Science 275 (1997) 1102. [15] A.D. McFarland, R.P. Van Duyne, Nano Lett. 3 (2003) 1057. [16] J.H.G. Owen, K. Miki, Surf. Sci. 600 (2006) 2943. [17] R.A. Zhachuk, S.A. Teys, A.E. Dolbak, B.Z. Olshanetsky, Surf. Sci. 565 (2004) 37. [18] X. Tong, L. Benz, A. Kolmakov, S. Chre´tien, H. Metiu, S.K. Buratto, Surf. Sci. 575 (2005) 60. [19] N. Nilius, M. Kulawik, H.-P. Rust, H.-J. Freund, Surf. Sci. 572 (2004) 347. [20] A.K. Santra, B.K. Min, D.W. Goodman, Surf. Sci. 515 (2002) L475. [21] K.R. Roos, K.L. Roos, M. Horn-von Hoegen, F.-J. Meyer zu Heringdorf, J. Phys.: Condens. Matter 17 (2005) S1407. [22] E. Fort, C. Ricolleau, J. Sau-Pueyo, Nano Lett. 3 (2003) 65.
505
[23] A.M. Turky, Appl. Catal. A 247 (2003) 83. [24] J.T. Richardson, B. Turk, M. Lei, K. Forster, Appl. Catal. A 83 (1992) 87. [25] F.F. Ferreira, M.C.A. Fantini, J. Phys. D: Appl. Phys. 36 (2003) 2386. [26] K. Marre, H. Neddermeyer, Surf. Sci. 287–288 (1993) 995. [27] T. Bertrams, H. Neddermeyer, J. Vac. Sci. Technol. B 14 (1996) 1141. [28] M. Caffio, B. Cortigiani, G. Rovida, J. Phys. Chem. B 108 (2004) 9919. [29] J. Wollschlger, D. Erdos, H. Goldbach, R. Hopken, K.M. Schroeder, Thin Solid Films 400 (2001) 1. [30] C. Lamberti, E. Groppo, C. Prestipino, S. Casassa, A.M. Ferrari, C. Pisani, Phys. Rev. Lett. 91 (2003) 46101. [31] M. Caffio, B. Cortigiani, A. Atrei, G. Rovida, J. Phys. Condens. Matter 18 (2006) 2379. [32] E. Groppo, C. Prestipino, C. Lamberti, R. Carboni, F. Boscherini, P. Luches, S. Valeri, S. D’Addato, Phys. Rev. B 70 (2004) 165408. [33] T. Orzali, S. Agnoli, M. Sambi, G. Granozzi, Surf. Sci. 569 (2004) 105. [34] J. Schoiswohl, S. Agnoli, B. Xu, S. Surnev, M. Sambi, M.G. Ramsey, G. Granozzi, F.P. Netzer, Surf. Sci. 599 (2005) 1. [35] J. Schoiswohl, J. Zheng, S. Surnev, M.G. Ramsey, G. Granozzi, S. Agnoli, F.P. Netzer, Surf. Sci. 600 (2006) 1099. [36] S. Agnoli, M. Sambi, G. Granozzi, A. Altrei, M. Caffio, G. Rovida, Surf. Sci. 576 (2005) 1. [37] S. Agnoli, M. Sambi, G. Granozzi, J. Schoiswohl, S. Surnev, F.P. Netzer, M. Ferrero, A.M. Ferrari, C. Pisani, J. Phys. Chem. B 109 (2005) 17197. [38] S. Agnoli, A. Barolo, P. Finetti, F. Sedona, M. Sambi, G. Granozzi, J. Phys. Chem. C 111 (2007) 3736. [39] C.S. Fadley, Prog. Surf. Sci. 16 (1984) 275. [40] G. Granozzi, M. Sambi, Adv. Mater. 8 (1996) 315. [41] S. Benedetti, P. Torelli, P. Luches, A. Rota, S. Valeri, Surf. Sci. 600 (2006) 4251. [42] H. De Monestrol, L. Schmirgeld-Mignot, P.J.A. Molina`s-mata, S. Poissonnet, G. Martin, Acta Mater. 49 (2001) 1655. [43] N.A. Khan, A. Uhl, S. Shaikhutdinov, H.-J. Freund, Surf. Sci. 600 (2006) 1849. [44] I.A. Abrikosov, W. Olovsson, B. Johansson, Phys. Rev. Lett. 87 (2001) 176403. [45] P.S. Bagus, A. Wieckowski, H.-J. Freund, Chem. Phys. Lett. 420 (2006) 42. [46] K. Luo, X. Lai, C.-W. Yi, K.A. Davis, K.K. Gath, D.W. Goodman, J. Phys. Chem. B 109 (2005) 4064. [47] I. Lopez-Salido, D.C. Lim, Y.D. Kim, Surf. Sci. 588 (2005) 6. [48] J. Schoiswohl, F. Mittendorfer, S. Surnev, M.G. Ramsey, J. Andersen, F.P. Netzer, Surf. Sci. 600 (2006) L274.
