Growth morphology of thin films on metallic and oxide surfaces.

Home

Search

Collections

Journals

About

Contact us

My IOPscience

Growth morphology of thin films on metallic and oxide surfaces

This content has been downloaded from IOPscience. Please scroll down to see the full text. 2014 J. Phys.: Condens. Matter 26 053001 (http://iopscience.iop.org/0953-8984/26/5/053001) View the table of contents for this issue, or go to the journal homepage for more

Download details: IP Address: 130.63.180.147 This content was downloaded on 11/08/2014 at 22:55

Please note that terms and conditions apply.

Journal of Physics: Condensed Matter J. Phys.: Condens. Matter 26 (2014) 053001 (26pp)

doi:10.1088/0953-8984/26/5/053001

Topical Review

Growth morphology of thin films on metallic and oxide surfaces Aleksander Krupski1 Department of Physics, University of Warwick, Coventry CV4 7AL, UK E-mail: [email protected] Received 9 September 2013 Accepted for publication 10 December 2013 Published 17 January 2014

Abstract

In this work we briefly review recent investigations concerning the growth morphology of thin metallic films on the Mo(110) and Ni3 Al(111) surfaces, and Fe and copper phthalocyanine (C32 H16 N8 Cu) on the Al2 O3 /Ni3 Al(111) surface. Comparison of Ag, Au, Sn, and Pb growth on the Mo(110) surface has shown a number of similarities between these adsorption systems, except that surface alloy formation has only been observed in the case of Sn and Au. In the Pb/Mo(110) and Pb/Ni3 Al(111) adsorption systems selective formation of uniform Pb island heights during metal thin film growth has been observed and interpreted in terms of quantum size effects. Furthermore, our studies showed that Al2 O3 on Ni3 Al(111) exhibits a large superstructure in which the unit cell has a commensurate relation with the substrate lattice. In addition, copper phthalocyanine chemisorbed weakly onto an ultra-thin Al2 O3 film on Ni3 Al(111) and showed a poor template effect of the Al2 O3 /Ni3 Al(111) system. In the case of iron cluster growth on Al2 O3 /Ni3 Al(111) the nucleation sites were independent of deposition temperature, yet the cluster shape showed a dependence. In this system, Fe clusters formed a regular hexagonal lattice on the Al2 O3 /Ni3 Al(111). Keywords: aluminum, nickel, aluminum oxide, iron, copper phthalocyanine, molybdenum, lead, silver, gold, tin, STM, STS, LEED, SPA-LEED, AES, ab initio, density functional theory, calculations, growth, thin film growth, cluster growth and nucleation, metal–metal interfaces, electron–solid interactions, low-index single crystal surface, surface alloys (Some figures may appear in colour only in the online journal)

Contents

1. Introduction

2

2. Thin metal films on the Mo(110) and Ni3 Al(111) surfaces

3

2.1. 2.2. 2.3. 2.4.

Mo(110) Ni3 Al(111) Ag/Mo(110) Au/Mo(110)

10

2.6. Pb/Mo(110)

11

2.7. Pb/Ni3 Al(111)

14

3. Molecule and metal growth on the Al2 O3 /Ni3 Al(111) 16

3 4 6 7

3.1. Al2 O3 /Ni3 Al(111)

16

3.2. C32 H16 N8 Cu/Al2 O3 /Ni3 Al(111)

18

3.3. Fe clusters on Al2 O3 /Ni3 Al(111) 4. Conclusions

1

Permanent address: Institute of Experimental Physics, University of Wrocław, plac Maxa Borna 9, 50-204 Wrocław, Poland. 0953-8984/14/053001+26$33.00

2.5. Sn/Mo(110)

1

20 21

Acknowledgments

23

References

23 c 2014 IOP Publishing Ltd

Printed in the UK

J. Phys.: Condens. Matter 26 (2014) 053001

A Krupski

1. Introduction

Al2 O3 /NiAl(110) [93, 94]) and the formation of more complex systems like CuPc/Al2 O3 /Ni3 Al(111) [95] or Fe/Al2 O3 / Ni3 Al(111) [96]. Ni3 Al can be seen as a model system for use as a template for a well-ordered Al2 O3 surface [92], which is suitable for studies with several surface sensitive investigation methods requiring electric conductivity. Aside from the usage as a support material for heterogeneous catalysts [97], Al2 O3 is also of increasing relevance as an insulator for semiconductor devices because of its superior dielectric properties compared to the classical insulator SiO2 [98]. The investigation of organic thin films grown on semiconducting surfaces has already been of interest for many years. The high thermal stability and electronic properties of phthalocyanine (Pc) compounds and their metal complexes (e.g. CuPc) are very interesting candidates for the development of new technical applications in various fields; examples include highly specific gas sensors [99–101], organic light emitting diodes [102], organic field effect transistors [103, 104] and photovoltaic cells [105]. The main focus of our work with the CuPc/Al2 O3 /Ni3 Al(111) system [95] was to determine whether the well-established template effect of the Al2 O3 /Ni3 Al(111) system for metal cluster growth [106, 107] could also be exploited for an ordered arrangement of organic molecules. This ordering is extremely important for the development of novel technologies. Since the first scanning tunneling microscopy (STM) investigations of CuPc on a polycrystalline Ag film by Gimzewski et al [108] many different substrates have been considered for the growth of ordered CuPc films. In the initial work it was not possible to reach submolecular resolution and no surface induced orientation of the molecules was achieved by Lippel et al on Cu(100) [109]. However, the formation of an ordered CuPc monolayer was found later on other metal surfaces like Ag(111) [110] and Au(111) [111]. Some semiconducting substrates like graphite and MoS2 were also observed to exhibit an ordered planar growth of the molecules [112]. The first STM and scanning tunneling spectroscopy (STS) investigations of CuPc on an ultra-thin alumina film on Ni3 Al(110) by Ho and co-workers [113] did not show a significant template effect of the substrate but the authors could distinguish between two different kinds of adsorbate configurations with D4h - and D2h symmetry, respectively. Finally, it is necessary to understand metal–oxide interfaces for progression of many technologies including metal–ceramic sensors, microelectronics, spintronic devices and oxide supported transition metal catalysts [114–119]. In our work concerning Fe/Al2 O3 /Ni3 Al(111) [96] we focus on the creation of small oxide supported iron clusters serving as a model system to investigate the catalytic and magnetic properties of nanosized metallic particles on an insulating substrate. The catalytic properties of Fe clusters on oxides are of particular interest with regard to nitrogen fixation, being of vital importance in biological systems and of the highest relevance in technological processes [120]. Biological nitrogen fixation was the subject of numerous experimental as well as theoretical studies aimed at creating a detailed picture of the catalytic activity of an enzyme and of the reaction pathway [121–123]. A recent publication investigating small Fe clusters on MgO(100) attributes to this model system

Ultra-thin epitaxial film systems exhibit a variety of interesting properties owing to the strong correlation between the electronic structure of the film and its morphology, strain, and defect structure [1, 2]. Knowledge of the interface structure between face-centered cubic (fcc) thin metal films and bodycentered cubic (bcc) metals is important in understanding the initial growth of these adsorption systems. Structural studies of fcc/bcc systems provide a great deal of information on the connection between the geometrical properties of the adsorbed atomic layers and the atomic arrangements of the substrates. Until now, the surface structures and growth modes of various adsorbate metals (e.g. Cu [3], Co [4], Fe [5], Ni [6], Mg [7], Ba [8], K [9], S [10], In [11], Ag [12–17], Au [18, 19], Sn [20, 21], Pb [22–25], etc) on Mo(110) substrates have been investigated by a variety of experimental and theoretical methods in a number of works. Recently, the formation of oxide layers on transition metal (TM) surfaces has also received considerable attention [26–29], including the Mo(110) surface [30, 31]. In addition, Pb layers have also been of interest in recent years because the islands formed during growth are of a special uniform height (often referred to as a ‘magic’ or ‘wedding cake’ island) with flat tops and sharp step edges. This phenomenon has been observed during metal thin film growth in a number of material systems such as Pb/Mo(110) [22], Pb/Ni3 Al(111) [32, 33], Pb/Ni(111) [34–38], Pb/Cu(111) √ √ [39–42], Pb/Si(111) [41, 43–55], In/Si(111)-Pb-α- 3 × 3 [56], Ag/Fe(001) [57], Ag/GaAs(110) [58], Ag/Si(111)7 × 7 [59] and has been interpreted in terms of quantum size effects (QSEs) [60–64]. Quantum size effects were initially observed experimentally by Jaklevic et al [65–67] for metals during measurements of tunneling currents in metal–metal and oxide–metal junctions and later by Jonker et al [68–70] in low-energy electron transmission experiments. Since then, QSE-related oscillations have been found in other transport property measurements of thin films (e.g. superconducting films, the Hall coefficient, and the electric resistivity), all of which depend on the density of states at the Fermi energy. QSE studies indicate that, in the equilibrium distribution of island heights, some heights appear much more frequently than others. ‘Magic’ preferred heights correspond to islands with a quantum well state far from the Fermi energy, while ‘forbidden’ heights appear to be those that have a quantum well state close to the Fermi level. When these uniform height islands are observed the growth conditions for temperature and flux are different for each substrate type, suggesting that kinetic factors play a role. γ 0 -Ni3 Al is of great interest for its attractive applications in the power and aerospace industries as a high-temperature structural material. In a bulk or thin film state this compound exhibits unique physical and mechanical properties such as a high melting point, high hardness, high yield strength, low density and good oxidation and corrosion resistance [71–77]. Nickel aluminate surfaces, including the Ni3 Al(111) and Ni3 Al(001) surfaces, have been investigated experimentally [78–82] and theoretically [81–91], and are often used as substrates in studies of a variety of different surface phenomena such as oxidation (e.g. Al2 O3 /Ni3 Al(111) [92], 2


A Krupski

Figure 1. (a) Schematic view of the bulk unit cell of Mo. The (110) plane is marked. (b) Top view of the Mo(110) surface. (c) LEED pattern

of the clean Mo(110) surface recorded for E = 300 eV at T = 300 K, and normal electron incidence. The (1 × 1) unit cell of the substrate is drawn in yellow. (d) STM image of the clean Mo(110) at T = 300 K (1000 Å × 1000 Å, IT = 0.8 nA, Ubias = 100 mV). The line scan shown in the inset evidences terraces between approximately 100 and 400 Å wide. Part (d) is taken from [23]. Copyright 2009 by the American Physical Society.

catalytic properties that are similar to the active sites in the biological enzyme nitrogenase [124]. The effects of the reduced particle size and the insulating substrate are not only crucial for catalytic activity but are expected to introduce new physical properties. The magnetic properties of the small iron particles supported by oxide systems are of particular interest with respect to the development of high-density magnetic data recording devices [125, 126]. Density functional calculations revealed that small Fen clusters (n ≤ 8) deposited on an inert substrate have magnetic moments greatly exceeding the Fe bulk value [127], which is consistent with the same observation for metal clusters in the gas phase [128]. In this article we review our recent investigations concerning the growth morphology of thin metallic films (Ag, Au, Sn) on Mo(110), Pb on Mo(110) and Ni3 Al(111) surfaces, and Fe and copper phthalocyanine (C32 H16 N8 Cu) on Al2 O3 /Ni3 Al(111). Our studies were performed with the use of well known and widely described experimental methods such as STM/STS [129–136], low-energy electron diffraction (LEED) [137] including spot profile analysis low-energy electron diffraction (SPA-LEED) [138, 139], Auger electron spectroscopy (AES) [140] and theoretical calculations based on the density functional theory (DFT) [141] and plane-wave basis set, as implemented in the Vienna ab initio simulation package (VASP) [142–145]. The presented STM data were processed by freeware image-processing software [146]. The rest of this article is divided into the following sections. Section 2 will give important information on thin metal films on Mo(110) and Ni3 Al(111) surfaces. In section 3

molecule and metal growth on Al2 O3 /Ni3 Al(111) are discussed. The conclusion will consist of a brief summary of the previous sections. 2. Thin metal films on the Mo(110) and Ni3 Al(111) surfaces 2.1. Mo(110)

Mo has a body-centered cubic (bcc) crystal structure with the ¯ with lattice parameter a = 3.15 Å [147, space group I m 3m 148], see figure 1(a). The structural model for the clean Mo(110) surface and the corresponding LEED pattern are shown in figures 1(b) and (c), respectively. Figure 1(d) displays an STM image that was taken on a low-index Mo(110) substrate with terraces between 100 and 400 Å wide separated by monoatomic steps aligned along the direction. The height of the steps on the Mo(110) surface was measured by STM to be 2.3 ± 0.2 Å. Theoretical studies of the Mo(110) surface [23] performed with the use of ab initio calculations based on the density functional theory and the plane-wave basis set, as implemented in the VASP package, indicate that the relaxation process of the Mo(110) surface system is limited solely to its topmost part. The first interlayer distance d1,2 was calculated to be reduced by 5% (i.e., somewhat above 0.1 Å), while the subsequent distance d2,3 increased by 0.5–0.9% (0.01–0.03 Å). Following this, the change in the third interlayer distance d3,4 was seen already as negligibly small. These results were in good agreement with earlier theoretical studies [149–152] as well as LEED measurements [153]. The results presented 3


A Krupski

Figure 2. (a) Schematic view of the bulk unit cell of L12 -ordered Ni3 Al. The (111) and (001) planes are outlined. (b) Top view of the Ni3 Al(111) surface. (c) LEED pattern of the clean p(2 × 2)-Ni3 Al(111) surface recorded for E = 136 eV at T = 200 K, and normal electron incidence. (d) STM image of the clean p(2 × 2)-Ni3 Al(111) surface at T = 23 K (75 Å × 75 Å, IT = 260 pA, Ubias = 50 mV). Tunneling spectra were measured at the points indicated on the STM image. Curves 1–5 were acquired after the feedback loop was opened at 85 mV at a tunneling current of 120 pA. (e) STM image of the Al2 O3 /Ni3 Al(111) surface at T = 23 K (71 Å × 71 Å, IT = 72 pA, Ubias = 3.21 V). Tunneling spectra were measured at the points indicated on the STM image. Curves 1–5 were acquired after the feedback loop was opened at 190 mV at a tunneling current of 72 pA. The corresponding modulation frequency was 5 kHz with an amplitude of 20 mV. The solid red line shows the DOS calculated for the Ni3 Al(111) surface. The (1 × 1) and (2 × 2) unit cells are drawn in blue and red, respectively. Part (c) is adapted from [32]. Copyright 2013 by Elsevier. Parts (d) and (e) are taken from [81]. Copyright 2009 by the American Physical Society.

shows hexagonal arrangements with a lattice constant of about 5.074 Å. This distance corresponds well with the (2 × 2) superstructure of the surface unit cell due to the ordered distribution of Al and Ni atoms [154]. The nearest Ni–Ni distance at the surface is 2.54 Å and cannot be seen in STM images. Nevertheless, as was shown in the work of Jurczyszyn et al [156], this fact is not contradictory to the bulk-like stoichiometric composition of the surface layer. However, instead of a long-range periodicity, the STM image showed relatively small domains with a highly visible (2 × 2) order, see figure 2(d). Our study indicates that such a short-range order of the surface structure is an intrinsic property of the Ni3 Al(111) system. The bias voltage in this experiment was +50 mV, and it is important to stress that the experimental setup requires this voltage to be applied to the sample. This means that at the given voltage tunneling occurs from the tip into unoccupied states of the sample. It followed from our experimental and theoretical studies [81] that the topographies of Ni3 Al(111) STM images were strongly influenced by intra-atomic s–pz interference. This kind of interference increased the efficiency of electron tunneling through s and pz orbitals of surface Al atoms, while reducing considerably the corresponding current contributions flowing through the surface Ni atoms. Earlier studies [156] indicated that this factor might have been responsible for the domination of surface Al atoms in STM images. In particular, it was found that even for an unrelaxed surface structure, where surface Al and Ni atoms

in [23] showed that increase in the slab thickness above seven layers did not considerably change the atomic structure of the surface. Local density of states (LDOS) distributions that were calculated for the three topmost atomic layers of Mo(110) indicated that electronic states in the energy region around the Fermi level were distinctly localized at the surface atomic layer while at lower energies their localization was shifted toward subsurface layers. LDOS calculations performed for the topmost atomic layer showed that the surface electronic structure of Mo(110) is dominated by d states within the whole energy range that was investigated [23]. 2.2. Ni3 Al(111)

Ni3 Al has an A3 B–L12 - or Cu3 Au-type structure with space ¯ group Pm 3m, and the equilibrium lattice parameter a = 3.57 Å [154, 155], see figure 2(a). The structural model for the clean Ni3 Al(111) surface and corresponding LEED pattern are shown in figures 2(b) and (c), respectively. The face-centered cubic (fcc) (111) surface of this perfectly ordered alloy, shown in figure 2(b), has exactly the same stoichiometric composition as the bulk. Each aluminum atom is embedded in the nickel matrix. This results in a 2 × 2 aluminum arrangement in the surface layer. The lattice constant of the surface unit cell is 5.04 Å (double the parameter of the primitive fcc (111) unit cell). Figure 2(d) shows an atomically resolved STM image of the clean (111) surface of Ni3 Al. The image 4


A Krupski

Figure 3. (a) STM image of the clean Ni3 Al(001) surface at T = 4 K (100 Å × 100 Å, IT = 5 nA, Ubias = 85 mV). The FFT shown in the

inset evidences a square lattice arrangement of protrusions (Al atoms) with a nearest neighbor distance of 3.7 Å. (b) Top view of the Ni3 Al(001) surface. (c) LEED pattern of the clean Ni3 Al(001) surface recorded for E = 120 eV, and normal electron incidence. The unit cells are outlined. Parts (a) and (c) are adapted from [82]. Copyright 2008 by Elsevier.

have the same vertical positions (as takes place in the bulk), the topography of the calculated STM image was dominated by Al atoms. More specifically, the surface Al atoms appeared in the simulated STM images of the (2 × 2) superstructure with the same shape and size as they appeared in experimental STM data, while the surface Ni atoms were completely invisible. In the experimentally measured relaxed Ni3 Al(111) surface structure, the domination of Al atoms in STM images was additionally supported by the differences in vertical positions of Al and Ni atoms [81]. Figure 2(e) shows the STS spectra obtained for five different horizontal positions of the tip, marked by arrows in figure 2(d). This set of spectra indicates that, except for the maximum located at the Fermi level, the energies of the rest of the STS features dominating in the considered energy range somewhat depend on the position of the STM tip. Apart from the experimental data, figure 2(e) also presents the LDOS distribution at the surface layer obtained from our present ab initio calculations (solid red line). It should be pointed out here that in the standard STS measurements, the obtained spectra mainly reflected the energy distributions of s and p states of the substrate surface, as d states are not very active during the tunneling process. Therefore, for comparison with experimental results, the LDOS distribution plotted in figure 2(d) takes into account only the contributions connected with s and p states of the surface atoms. This theoretical result represents the averaged LDOS distribution over three Ni atoms and one Al atom from the (2 × 2) surface unit cell. Below the Fermi level, the theoretical distribution indicates the presence of LDOS features located 0.4 and 0.6 eV below E F . The lower one (at −0.6 eV) corresponds well with a distinct STS peak that appears around −0.6 eV in each experimental curve. However, the higher theoretical feature (at −0.4 eV) can only be linked with a relatively weak shoulder, which can be seen in STS spectra 1, 2, and 4 around −0.4 eV. The dependences presented in figure 2(d) also indicate that theoretical results reproduce the measured STS features in the vicinity of the Fermi level well. For unoccupied states, the calculated LDOS distribution shows a distinct feature located around 0.35 eV. Such a feature can be seen in STS spectra 2 and 5, though in spectrum 2 it is somewhat shifted toward higher energies. Much weaker maxima within this energy range can also be observed in curves 1 and 3. However, as mentioned earlier, the experimental

studies of Ni3 Al(111) indicated a lack of the long-range order along this surface (see figure 2(d)), while the theoretical study was performed assuming ideal translation symmetry. This fact needs to be taken into account when comparing the STS spectra and LDOS distributions presented in figure 2(e). Previous STM/STS study [157] showed that for the Al2 O3 /Ni3 Al(111) system, the ordering of the substrate underneath the oxide film is much higher than for the clean Ni3 Al(111) surface. The properties of the Al2 O3 /Ni3 Al(111) system are described in more detail in section 3.1 of this review. The STS data indicated the presence of a large oxide bandgap [118, 157–159] that allowed STM/STS measurements of the Ni3 Al(111) substrate surface through the oxide film. Figure 2(e) shows an STM image of Al2 O3 /Ni3 Al(111) collected with a bias of 3.2 V, i.e., when the topography of the image was dominated by the parameters of the Al2 O3 structure. However, the set of STS spectra presented in figure 2(e) was obtained for the bias range −1.5–1.5 V, which corresponds to the energy gap of the oxide. Since for such voltages the Al2 O3 film appeared invisible in STM/STS measurements [81], a reduction of the tip–sample separation enabled us to interrogate the electronic structure of the Ni3 Al(111) substrate. Figure 2(e) shows a set of STS spectra measured at five different horizontal positions of the STM tip, marked by arrows in figure 2(e). This sequence of dependences shows that, in contrast to the case of clean Ni3 Al(111), see figure 2(d), the STS spectra had very little dependence on the horizontal position of the tip. We may say that for energies within the oxide bandgap, the presence of the oxide film does not modify the surface electronic structure of the substrate much, as compared to the corresponding results for a clean Ni3 Al(111) surface. In contrast to the Ni3 Al(111) surface, the Ni3 Al(001) surface [82] clearly exhibits a longrange character, see figure 3(a). The structural model of the clean Ni3 Al(001) surface and corresponding LEED pattern are presented in figures 3(b) and 2(c), respectively. The topographies from the obtained STM images indicate a surface superstructure with a square lattice unit and a lattice constant equal to (=3.7 Å) the distance between the nearest neighbor surface atoms of the same type (i.e., nearest Al or nearest Ni atoms). This indicates that the STM image is formed by one type of surface atom only, while the atoms of the second type remain invisible. It was shown in [143] that the STM process 5


A Krupski

Figure 4. Ag deposition on the Mo(110) face at T = 300 K. (a) AES(t) plot of Mo MNN and Ag MNN peak heights. (1) AES(t) plot calculated for Frank–van der Merwe growth. STM images at the following coverages. (b) θ ≈ 0.20 ML (4010 Å × 4010 Å, IT = 2.85 nA, Ubias = 0.1 mV). (c) θ ≈ 1.00 ML (1000 Å × 1000 Å, IT = 6.97 nA, Ubias = 20 mV). The arrows in this image point to Ag islands which extend over two terraces of the substrate, and their thickness does not change by 1 ML from terrace to terrace. (d) θ ≈ 7 ML (800 Å × 800 Å, IT = 0.2 nA, Ubias = 2.0 V). The arrows indicate the positions where the Mo substrate is still visible. (e) Post-annealed θ ≈ 1.0 ML at T = 700 K (5047 Å × 5047 Å, IT = 0.2 nA, Ubias = 2.5 V). (f) Line scan corresponding to line A drawn in (e) demonstrating that the Ag nanostripes protrude by 0.39 ± 0.06 Å with respect to a single step of the Mo terrace. The insets in (d) and (e) present differentiated STM images with enhanced contrast. Adapted from [12]. Copyright 2010 by Elsevier.

at the Ni3 Al(001) surface is strongly influenced by the intraatomic s–pz interface. This kind of interference reduces the s and pz current contributions flowing through Ni surface atoms and increases considerably the s–pz contributions connected with surface Al atoms. A detailed analysis has shown that this factor leads to Al atoms being the dominant species in simulated STM images of the Ni3 Al(001) surface [82, 156].

Curve (1) in figure 1 was calculated for FM growth using the formula described above, under the assumption that at the (h S1 ; t1 ) = (0.555; 198) point the first silver layer was completed. It can be seen from a comparison of our calculated and experimental AES(t) plots that two-dimensional growth of the second silver layer should have also been observed. However, one cannot recognize any distinct breaking points by eye after deposition times t2 , t3 or t4 on the experimentally found AES(t) plot. Fitting an exponential to the data provided a closer agreement. Such a shape of an AES(t) plot suggests the Stranski–Krastanov growth mode (3D). Because the scatter of h S1 and t1 values for these AES(t) plots was not large, we will further assume that the initial linear part of the curve corresponds to the formation of the first layer (θ = 1 ML) of silver, where 1 ML of Ag(111) film corresponds to an atomic packing density of 13.80 × 1014 atoms cm−2 . Owing to the method used for the recording of AES data, the dependences of AES signals for both substrate and adsorbate on adsorbate deposition time were represented by as many as 20–30 experimental data points for each monolayer of adsorbate. Thus, profiles of AES kinetics could be determined more precisely than for the kinetics presented in [13] where one monolayer was represented by only 14 points and only by adsorbate AES kinetics. In addition, in our measurements, adsorbate deposition was not interrupted for the recording of Auger peaks. Thus, our AES kinetics analysis was for

2.3. Ag/Mo(110)

It is well know that plotting the AES peak heights of the substrate and of the adsorbate as a function of deposition time (AES(t) plots) enables the determination of the growth mechanism as well as monolayer formation [33, 160–164]. In our studies at room temperature, only one linear region in the AES(t) plots for the 186 eV molybdenum and 351 eV silver peaks could be distinguished, see figure 4(a). It could be deduced from this region in the AES(t) plots that twodimensional growth of the first silver layer was observed. Subsequently, the non-linear shape of the AES(t) plots suggested the occurrence of three-dimensional (3D) growth. If αSA = h S1 / h S0 defines the coefficient of attenuation of the substrate Auger peak due to the presence of a monolayer of adsorbate, then the expected height of the substrate Auger peak for layerby-layer growth after completion of the n = second, third, fourth, . . . layer is given by the equation h Sn = h S0 (αSA )n [33]. 6


A Krupski

der Merwe (FM) growth is expected if γS ≥ γA + γI . In hetero-epitaxial systems, the surface free energy increases with the number of film layers (n) so that the FM condition will no longer be valid at a critical film thickness (n ∗ ) at which point three-dimensional (3D) crystals begin to form (Stranski–Krastanov (SK) mode). The Volmer–Weber (VW) mode, where 3D growth begins immediately, corresponds to n ∗ = 1. The higher surface energy of the Mo substrate allows the Ag film to grow in pseudomorphic registry with the substrate. This was confirmed by Mae [169], using a molecular dynamics aided kinetic Monte Carlo method. Table 1 contains a set of structural parameters and surface energies (γ ) of Mo, Ni3 Al and also Pb, Ag, Au, Sn with respect to Miller plane orientation reproduced from [44, 147, 148, 154, 171–180]. When the sample was post-annealed to 700 K, the morphology of the surface changed, as shown in figures 4(e)–(f). At this temperature, growth was driven by a pure step-flow mechanism. Step-flow growth in the submonolayer and monolayer regime gives rise to a regular Ag nanostripe network attached to Mo(110) step edges, figure 4(e) for 1.0 ML. The corrugation profiles (A) in figure 4(f) highlight the protrusion of silver nanostripes equal to p2 = 0.39 ± 0.06 Å for 1.0 ML with respect to a single step of a Mo terrace. In addition, the width and shape of the top of the nanostripes showed a dependence on coverage. It is suggested that the variations seen in the morphology of Ag films as the temperature changes from room temperature to 700 K were due to the high sensitivity of the surface diffusion coefficient of Ag adatoms to temperature. The mobility of Ag adatoms on Mo(110) was reduced near room temperature and at a temperature approaching 700 K the diffusion mean free path was long enough for adatoms to reach the step edges. The occurrence of mass transport over step edges was not investigated in the present work. The desorption process of silver atoms from Mo(110) at 700 K did not take place, as shown by Bauer [13]. The authors used AES to study the annealing behavior (up to 1050 K) of thick Ag films on the Mo(110) face. Here, the AES results showed two temperature regions in which the Auger signal changed only slightly with temperature. The authors concluded [13] that these two small decreases in the silver AES signal at 700 and 900 K were due to agglomeration of silver atoms. In addition, at a sample temperature of T = 1000 K it was observed that silver desorption from Mo(110) began [13]. There is also no evidence in the literature for alloy formation between Ag and a Mo(110) substrate. Similar behavior for the Fe/Mo(110) adsorption system has been observed [181]. These authors studied the growth of iron on a Mo(110) face by STM. The Fe/Mo(110) system exhibited the formation of large Fe islands at room temperature. At temperatures slightly higher than room temperature the growth proceeded by incorporation of Fe at Mo step edges.

continuous growth of the deposited layer. Figures 4(b)–(d) show STM images of the Mo(110) surface taken with different Ag coverages in order to illustrate the development of the morphology of the Ag layers deposited on Mo at room temperature. Figure 4(b) shows a typical STM image corresponding to a submonolayer coverage of θ = 0.2 ML. The bright features represent one monolayer high silver islands. The observed Ag islands on the substrate had regular shapes, sharp edges, and smooth tops. At this point, an analysis of the STM measurements indicated that, for coverage less than 1 ML, the growth of Ag was mediated by a two-dimensional step-flow mechanism. This means that, during the initial stage of growth, the first Ag film nucleates and creates islands at Mo step edges. That is, the growth proceeds by incorporation of Ag adatoms at ascending step edges and the film advances across the atomic terraces along the mis-cut direction. The estimated average size of the islands is about 180 ± 20 nm2 . Figure 4(c) shows an STM image corresponding to monolayer coverage θ = 1.0 ML, as deduced from the AES(t) plot in figure 4(a). However, the first two-dimensional silver layer was not complete as was expected from the AES(t) plots. At this coverage, the two-dimensional step-flow mechanism was continued. The silver islands of the first layer were then larger than in the submonolayer regime. In addition, between these silver islands, additional small Ag islands without flat tops appeared. Also, the second silver layer started to grow before completion of the first silver layer, figure 4(c). In this image the first and second layers are marked by the numbers I and II, respectively. The arrows in this image point to Ag islands which extended over two terraces of the substrate, and their thickness did not change by 1 ML from terrace to terrace. This island shape type has been termed a ‘nano-wedge island’ [165]. Closer inspection of figure 4(c) shows that, in the monolayer coverage range, the decoration of the substrate steps by Ag could be distinguished by the presence of a fractional step of p1 = 0.86 ± 0.6 Å height at the Ag–Mo boundary. This fractional step may have been due to a difference in either the local density of states (Mo(110) [166, 167], Ag [168]) or the atomic radii of the two metals (ar,Mo = 1.40 Å, ar,Ag = 1.44 Å). At coverages greater than θ = 7 ML, see figure 4(d), further formation of three-dimensional (3D) islands with sharp edges and mostly smooth tops was observed. The silver islands probably minimize their surface free energy by having flat tops. The arrows in figure 4(d) indicate the areas where the Mo substrate was still visible. These STM results agreed well with our AES studies and confirmed a 3D growth mode, a result that contradicted the literature [13–17], where layer-by-layer growth at room temperature up to 3 ML was reported. In addition, the results presented here are in good agreement with a simulated growth mode [169], where a 3D growth mode was shown with island growth commencing from the second Ag layer. A thermodynamic description of growth is based on the specific surface free energy γ of a particular crystal facet, defined as the reversible work per unit area required for the creation of the surface [170]. Different growth modes may be distinguished according to a balance between the surface free energy of the substrate (γS ) and the film (γA ) and the interface energy (γI ). Namely, layer-by-layer or Frank–van

2.4. Au/Mo(110)

Figure 5 shows STM images of the Mo(110) surface with varying Au coverage in order to illustrate the morphology of the Au layers deposited on Mo at room temperature. Figure 5(a) shows a typical STM image corresponding to a 7

8

fcc

fcc

bct

Au

Sn

fcc

Pb

Ag

fcc

Ni3 Al

1.40 [147]

1.44 [147]

1.45 [147]

1.75 [147]

Ni: 1.25 [147] Al: 1.43 [147]

Atomic Structure radiation a (Å) r bcc 1.36 [147]

Element, alloy Mo

a = 5.82 [148] c = 3.17 [148]

4.08 [148]

4.09 [148]

4.95 [148]

3.57 [154]

Lattice constant a (Å) 3.15 [148]

−45.9

−22.8

−23.0

Mo: −36.4 Ni3 Al: −27.8

(Substrate)

(Substrate)

A (%) f = aSa−a A

(001) (110) (100)

(111) (110) (100)

(111) (110) (100)

(111) (110) (100)

(111) (110) (100)

Orientation (hkl) (111) (110) (100)

0.387 [171] 0.509 [171] 0.716 [171]

1.283 [171] 1.700 [171] 1.627 [171]

1.172 [171] 1.238 [171] 1.200 [171]

0.321 [171] 0.445 [171] 0.377 [171]

1.887 [178] 2.290 [178] 2.108 [178]

3.740 [171] 3.454 [171] 3.837 [171]

0.709 [172]

1.506 [172]

1.250 [172]

0.593 [172]

1.755 [177] 2.060 [177] 1.910 [177]

3.000 [172]

0.675 [173]

1.500 [173]

1.246 [173]

0.600 [173]

2.907 [173]

0.588 [44] 0.625 [44]

1.480 [44] 1.850 [44] 1.690 [44]

1.140 [44] 1.420 [44] 1.290 [44]

2.840 ([44]) 3.040 [44] 2.120 [44]

Surface energy γ (J m−2 )

0.799 [175] 0.999 [175] 1.038 [175]

2.900 [174]

1.520 [176] 1.800 [176] 1.940 [176]

1.200 [176] 1.400 [176] 1.510 [176]

0.550 [176] 0.640 [176] 0.700 [176]

4.620 [176] 3.200 [176] 3.180 [176]

Table 1. Structural parameters and surface energies (γ ) of Mo, Ni3 Al, Pb, Ag, Au, and Sn with respect to Miller plane orientation reproduced from [44, 147, 148, 154, 171–178].

J. Phys.: Condens. Matter 26 (2014) 053001 A Krupski


A Krupski

Figure 5. STM images of Au deposited on the Mo(110) face at T = 300 K. (a) θ = 0.4 ML (2923 Å × 2923 Å, IT = 0.5 nA, Ubias = 1.0 V). (b) θ ≈ 1.0 ML (2480 Å × 2480 Å, IT = 0.5 nA, Ubias = 1.16 V). (c) θ ≈ 1.2 ML (2494 Å × 2494 Å, IT = 0.5 nA, Ubias = 1.0 V). (d) θ ≈ 2.0 ML (2500 Å × 2500 Å, IT = 0.5 nA, Ubias = 1.16 V). (e) Post-annealed θ ≈ 1.0 ML at T = 800 K (5000 Å × 5000 Å, IT = 0.5 nA, Ubias = 1.4 V). (f) Post-annealed θ ≈ 1.2 ML at T = 800 K (2410 Å × 2410 Å, IT = 0.5 nA, Ubias = 1.37 V). The insets in (a)–(c), (e) and (f) present differentiated STM images with enhanced contrast. Adapted from [18]. Copyright 2011 by Elsevier.

submonolayer coverage of θ ≈ 0.4 ML. The bright features represent gold islands. These were one monolayer high islands. The Mo substrate was still visible between the gold islands. An analysis of the STM measurements indicated that for coverage less than 1 ML the distribution of the gold islands was not random on the Mo terraces. Namely, from figure 5(a) it seemed that the gold islands grew preferentially along the ¯ and [111] ¯ directions. The estimated average diameter of [001] such gold islands was approximately 12 ± 5 nm. Figure 5(b) shows an STM image corresponding to a monolayer coverage θ ≈ 1.0 ML, as deduced from the preceding STM images and deposition time. The first two-dimensional disordered gold layer was completed in the manner predicted from the previous AES, LEED, and TDS experimental [19, 182, 183] and theoretical [184] studies. In addition, it was not possible to identify further positions of the gold islands in the first layer because of the observed per-coalescence effect [185–192] between θ = 0.55 and 0.65 ML. In figure 5(b), the arrows indicate where the second two-dimensional gold layer started to grow after completion of the first gold layer. Closer inspection of figure 5(b) when the coverage increased to 1.2 ML showed that the decoration of substrate steps by one atomic height Au islands could be distinguished from other randomly distributed gold islands on the first disordered gold layer. For a coverage θ ≈ 2 ML, see figure 5(d), a second completed but disordered gold layer was visible, as expected from the literature [182]. The higher surface energy of the Mo substrate (see table 1) allowed the Au film to grow in pseudomorphic registry with the substrate. This

was originally postulated by Gruzza and Gillet [182] using a simulation of a random adsorption process based on the following assumptions. (a) Every atom impinging on a free site is adsorbed instantaneously. If the site is occupied the atom joins either the first neighbor site or a second layer position. (b) Steric effects can induce displacements of 5% around an adsorption site. (c) An adsorption site is inhibited when it is surrounded by at least four adatoms. Based on these assumptions and taking into account the fact that the distance between neighboring sites in the (110) molybdenum plane (2.73 Å) is lower than the atomic diameter of the gold atoms (2.88 Å), the distribution of gold atoms in the saturated first layer shows that the ordered zones are no greater than 20 Å. When the sample was post-annealed to 800 K, the morphology of the surface changed, as shown in figures 5(e) and (f). At this temperature the rearrangement of an existing gold film was observed. These STM results showed similarity to previous AES studies concerning the possibility of a 2D growth mode [183] between 773 and 973 K. However, one should remember that post-annealing to a certain temperature (in our case 800 K) and growth at the same temperature, 800 K, cannot be compared directly because of differing kinetic pathways. This rearrangement that occurred between the monolayer and two layer regimes gave rise to a non-regular well separated Au island network, see figures 5(e) and (f) for 1.0 and 1.2 ML, respectively. The 1.0 and 1.2 ML coverages discussed here were the initial coverages before the post-annealing procedure. The fact that a complete monolayer annealed at 800 K led to island formation is indicative of gold alloying into the molybdenum substrate. This agrees well with what is known 9


A Krupski

Figure 6. Sn deposition on the Mo(110) face at T = 300 K. (a) AES(t) plot of Mo MNN and Sn MNN peak heights. (1) AES(t) plot

calculated for Frank–van der Merwe growth. STM images at the following coverages. (b) θ ≈ 0.70 ML (2350 Å × 2350 Å, IT = 0.33 nA, Ubias = 987 mV). (c) θ ≈ 1.0 ML (2300 Å × 2300 Å, IT = 0.2 nA, Ubias = 1.0 V). (d) θ ≈ 2.00 ML (2500 Å × 2500 Å, IT = 0.73 nA, Ubias = 2.1 V). (e) Post-annealed θ ≈ 0.7 ML at T = 800 K (1360 Å × 1360 Å, IT = 0.21 nA, Ubias = 790 mV). (f) Post-annealed θ ≈ 1.0 ML at T = 800 K (1210 Å × 1210 Å, IT = 0.2 nA, Ubias = 1.0 V). The insets in (b), (e) and (f) present differentiated STM images with enhanced contrast. Adapted from [20]. Copyright 2011 by Elsevier.

points for the tin signal was large because this signal was low because it was taken with the use of the retarding field analyser (RFA), and the ratio of noise to signal was large. It was apparent from this part of the AES(t) plot that two-dimensional growth of the first and second layers was observed. Curve (1) in figure 6(a) was calculated for Frank–van der Merwe (FM) growth under the supposition that at the (h S1 ; t1 ) point the first tin layer was completed. From comparison of our calculated and experimental AES(t) plots it could be seen that two-dimensional growth of the second and third tin layers is also observed. However, it was impossible to distinguish any distinct breaking point after deposition time t3 on the experimentally found AES(t) plot. The shape of the AES(t) plot suggested the Stranski–Krastanov growth mode after formation of the second layer. Further, we will assume that the first linear region of the curve corresponds to the formation of the first layer (θ = 1 ML) of tin, where 1 ML of Sn(111) film corresponds to an atomic packing density of 5.48 × 1014 atoms cm−2 . Figures 6(b)–(f) show STM images of the Mo(110) surface with varying Sn coverage in order to illustrate the morphology of Sn layers deposited on Mo at room temperature. The insets of figures 6(b), (e), and (f) present corresponding differentiated STM images with enhanced contrast. Figure 6(b) shows a typical STM image corresponding to a submonolayer coverage of θ ≈ 0.7 ML. The bright features show one monolayer high tin islands. Such behavior of tin in the initial stage of growth on the Mo(110) surface confirmed experimental results obtained during our

from the literature [182]; 800 K is below the desorption temperature of gold from the Mo(110) face. In this publication, the authors used thermal desorption spectroscopy to study the annealing behavior (up to 1550 K) of thick Au films on Mo(110) surfaces. The TDS results have shown that the desorption process begins at about 1080 K. Further inspection of the data showed desorption of a full gold multilayer at T = 1180 K and of a single monolayer at T = 1300 K. The analysis of post-annealed figures showed that the height of the gold islands corresponded to one-atom high islands, mainly 2.3 Å. From the presented STM results here it is clear the two-dimensional gold islands observed after the post-annealing procedure grew on top of the Au–Mo(110) surface alloy. The rearrangement of Au films with change in temperature from room temperature to 800 K was most likely due to the high sensitivity of the surface diffusion coefficient of Au adatoms to temperature. The mobility of Au adatoms on Mo(110) was reduced near room temperature and at a temperature approaching 800 K the diffusion mean free path was long enough for adatoms to reach the step edges. The occurrence of mass transport over step edges was not observed in this work. 2.5. Sn/Mo(110)

In case of this adsorption system at room temperature two linear regions of the AES(t) plot for 186 eV molybdenum can be distinguished, see figure 6(a). The scatter of experimental 10


A Krupski

(high temperature and pressure is necessary to stabilize such a compound), yet on the surface these size restrictions are much less dominant. To a certain extent a somewhat larger size of the adatom can even be favorable for incorporation because in this case the surface can achieve a larger packing density, which is favorable. The rearrangement of Sn films with change in temperature from room temperature to 800 K was probably due to the high sensitivity of the surface diffusion coefficient of Sn adatoms to temperature. The mobility of Sn adatoms on Mo(110) was reduced near room temperature and increased with temperature until at temperatures approaching 800 K the diffusion mean free path was long enough for adatoms to reach the step edges. The occurrence of mass transport over step edges was not observed in this work. Another important parameter, which determines the structure of epitaxial films, is the lattice misfit f [196–199] of the film to the substrate and can be accommodated by elastic strain or by the formation of dislocations. The misfit is given by f = (aS − −aA )/aA , where aS and aA are the lattice parameters of the substrate and adsorbate, respectively. Since the film’s elastic energy is proportional to f 2 , and the dislocation energy is proportional to f, a critical misfit exists below which pseudomorphic growth can be expected. In addition, strain energy contributes to the surface energy of the deposited layer. As a result, if the strain energy becomes large it could be energetically more favorable to form three-dimensional islands on top of the initially formed film (SK growth mode). Comparison of misfit for different adsorbates is presented in table 1.

AES measurements and previous studies [24]. Analysis of the STM measurements indicated that, for a coverage less than 1 ML, the distribution of the tin islands was random on the Mo terraces. This can be seen in figure 6(b) where it appears that the tin islands do not exhibit preferential growth directions. Figure 6(c) shows an STM image corresponding to a complete monolayer, i.e. θ ≈ 1.0 ML, as deduced from our AES experiment. The first two-dimensional tin layer is disordered. For a coverage θ ≈ 2 ML, see figure 6(d), a second complete but disordered tin layer was visible, as expected from the literature [24] and our AES studies. The higher surface energy of the Mo substrate (and the similar atomic size of Sn, see table 1) allowed the Sn film to grow in pseudomorphic registry with the substrate. When the sample was post-annealed to 800 K, the morphology of the surface changed, as shown in figures 6(e) and (f). At this temperature the rearrangement of an existing tin film was observed. These STM results were consistent with previous studies concerning the possibility of a 2D growth mode [24] between 790 and 1020 K. However, again, one should remember that post-annealing a sample grown at room temperature to the growth temperature (in our case 800 K) is not a process that can be compared directly to a growth at that same temperature because of different kinetic pathways. This rearrangement that occurred between the monolayer and bilayer regimes gave rise to a non-regular well separated Sn island network, see figures 6(e) and 6(f) for 0.7 and 1.0 ML, respectively. The 0.7 and 1.0 ML used in this description are the initial coverages measured before any post-annealing procedure. The average size of the islands was estimated to be about 10 ± 3 nm. The corrugation changed from 0.4 to 2.4 Å after post-annealing. The fact that a complete monolayer annealed at 800 K led to island formation suggested that tin alloyed into the molybdenum substrate. It is probable that some of the tin atoms dissolved into the bulk sample to form a Sn–Mo alloy. This is expected from the literature [24], where it has been shown that 800 K is below the desorption temperature of tin from the Mo(110) face. The authors here studied the annealing behavior (up to 1550 K) of thick Sn films on the Mo(110) surface using TDS. It was shown that the desorption process starts at about 1000 K. Desorption of a tin multilayer was shown to occur at T = 1180 K and a monolayer at about T = 1500 K. Analysis of figure 6 showed that the height of the tin islands largely corresponded to single atomic height islands. It can be deduced from the STM presented here that the two-dimensional tin islands that were observed after the post-annealing procedure most likely grew on the Sn–Mo(110) surface alloy. It should be noted that the incorporation of tin into the Mo(110) surface that is proposed here is unusual, because firstly neither bulk alloys nor stable intermetallic compounds are known for this system, and secondly in other metal systems where a low surface energy metal is adsorbed on a high surface energy substrate, the first monolayer does not mix with the substrate. This has only been observed previously when a substrate is soluble in the related film as for Pd on Mo(110) [193], or when the adsorbate and substrate material form intermetallic compounds such as Pb on Au(100) [194, 195]. On the other hand, tin is too large for molybdenum to form stable bulk compounds

2.6. Pb/Mo(110)

Figure 7 shows STM images from the Mo(110) surface with different Pb coverages in order to illustrate the morphology of the Pb layers deposited on Mo at room temperature. A typical STM image for 0.4 ML Pb deposition can be seen in figure 7(a). Pb islands with an irregular shape are shown as bright features and were deduced to be monolayer height islands. As the Pb coverage approached 1 ML, the Pb wet the Mo(110) surface almost completely, as can be seen in figure 7(b). It is not easy to determine whether an adsorbate wets a surface or not by STM. However, almost perfect wetting occurred due to the high specific surface free energy of the Mo(110) surface compared to that of the Pb(111) surface. As the total specific surface free energy is always minimized by nature, a covered Mo(110) surface was favored. As can be seen from the inset of figure 7(b), the first Pb layer on the Mo(110) surface was not free from dislocation defects. The pseudomorphic growth of the first two-dimensional Pb layer is thermodynamically driven by the lower surface free energy of Pb, and is the dominant process so that the Pb layer wets the Mo surface despite the elastic energy required as a result of the large lattice mismatch. Detailed inspection of figure 7(b) showed that the second lead layer starts to grow before complete formation of the first layer. The arrows in figure 7(b) indicate where the Pb adatoms of the second layer started to nucleate [200], which occurred preferentially at dislocation defects of the first Pb layer. As can be seen, the density of the dislocation defects was higher at monatomic 11


A Krupski

Figure 7. STM images of Pb deposited on the Mo(110) face at T = 300 K. (a) θ = 0.4 ML (1140 Å × 1140 Å, IT = 0.2 nA, Ubias = 1 V). (b) θ = 1.0 ML (1951 Å × 1951 Å, IT = 0.3 nA, Ubias = 800 mV). The inset presents a zoom (242 Å × 242 Å) in the area indicated by a

square. The arrows indicate the places where the Pb atoms of the second layer start to nucleate at dislocations of the first Pb layer. (c) θ = 1.2 ML (10 000 Å × 10000 Å, IT = 0.2 nA, Ubias = 1 V). (d) Height profile along the line B from the image in (c) demonstrating that the height of the islands corresponds to the height of a double Pb step height. (e) θ = 2.5 ML (8563 Å × 8563 Å, IT = 0.2 nA, Ubias = 1.0 V). The height of the Pb islands corresponds to 2 and 4 atomic layers of Pb(111). (f) θ = 5 ML (10 000 Å×10 000 Å, IT = 0.2 nA, Ubias = 1 V). The island height distribution (number of Pb islands with a given height) obtained from this image has shown the strongly preferred heights of the Pb islands corresponding to 2, 4 and 6 atomic layers of Pb(111) [22]. Copyright 2009 by the American Physical Society.

Figure 8. STM images of Pb deposited (θPb ≈ 5.0 ML) on the Mo(110) face at T = 300 K and post-annealed at 1050 K. (a) (20 000 Å ×

20 000 Å, IT = 0.21 nA, Ubias = 559 mV) showing the observed ring morphology (indicated by white arrows) on the 5 atomic layers of Pb(111) islands. (b) (6100 Å × 6100 Å, IT = 1.18 nA, Ubias = 298 mV). The height of the islands corresponds to 2, 5 and 7 atomic layers of Pb(111) [201]. (c) The height profile along line C from the image in (b) demonstrating that the heights of the rings correspond to 2 atomic layers of Pb(111). Part (a) is adapted from [20]. Copyright 2011 by Elsevier.

ledges. At higher coverages (1 ML < θ ≤ 5 ML) the growth became three-dimensional (figures 7(c)–(f)). These results contradict the literature data [25], where layer-by-layer growth at room temperature was reported. Figures 7(c) and (e) show STM images after 1.2 and 2.5 ML of Pb were deposited on the Mo(110) surface at RT. Height profiles that were taken across the Pb islands (figure 7(d)) show that the Pb crystallites had a thickness of about 6 Å. The interlayer spacing along the [111] direction of Pb was d111 = 2.86 Å and therefore the islands

had a thickness corresponding to the stacking of two atomic layers. The heights of the Pb islands were measured from the wetting layer. At coverages below 2 ML, Pb islands were observed with an average diameter of 800 Å and a standard deviation of 70 Å. As the Pb coverage was increased up to 2.5 ML (figure 7(e)), more than 75% of the islands were two layers in height. Four-layer high islands were also observed; however, they represented less than 25% of the total population of the Pb islands. In figure 7(e), the heights of the Pb islands 12


A Krupski

Figure 9. Pb deposition on the Mo(110) face at T = 300 K. STM images (differentiated in the insets) of two kinds of Pb superstructure. (a) 1 (80 Å × 80 Å, IT = 1.7 nA, Ubias = 149 mV). (b) 2 (50 Å × 50 Å, IT = 3.3 nA, Ubias = 76 mV). The white parallelograms indicate unit cells of the respective Pb structures with geometrical parameters measured to be equal to a = 4.3 Å, b = 5.2 Å, and α = 84◦ for 1 , and a = 4.1 Å, b = 3.8 Å, and α = 92◦ for 2 . Parts (c) and (d) present the overlap of the geometrical models of the superstructures 1 and 2 with their filtered STM images, respectively. The parameters of the simulated unit cells are a = 4.1 Å, b = 5.2 Å, α = 84◦ and a = 4.1 Å, b = 3.9 Å, α = 90◦ for 1 and 2 , respectively. The red dashed lines denote the supercells used in simulations [23]. Copyright 2009 by the American Physical Society.

layer (equal to 9.4 × 1014 atoms cm−2 ). Taking into account the experimental results, structural models of the Pb/Mo(110) surface have been constructed reproducing the topography of the obtained STM images. The lateral geometrical properties of this model (distances between the nearest lead adatoms along the directions aE and bE and the angle between them) are almost the same as those deduced from the STM measurements (cf the captions of figures 9(a) and (c)). Structural relaxation showed that such a model would be stable, i.e., the lateral positions of all lead adatoms in the relaxed structure were the same as in the starting configuration while PbS adatoms protruded 0.08 Å above PbL . The superstructure ε2 is, in turn, modeled by an adlayer with PbL , PbS , and PbT atoms (cf figure 9(d)), corresponding to the coverage of 4/9 ML. The lateral parameters of this model (i.e., aE , bE and α) reproduce the STM data well (cf captions of figures 9(b) and (d)). The results of structural simulations show PbT adatoms located 0.09 Å above PbS and 0.17 Å above PbL while the lateral positions of all lead atoms remain fixed during the relaxation. The supercells used for simulation of both superstructures are different, and, therefore, a direct comparison of their total energies is not possible. Thus, to determine the energy relationship between them, the average adsorption energy (E b ) per one Pb adatom has been calculated, defined as E b = − N1 (E Pb/Mo(110) − E Mo(110) − N E Pb ), where E Mo(110) is the energy of a clean Mo(110) slab, E Pb/Mo(110) is the energy of an adsorbed slab, E Pb is the energy of a free Pb atom, and N is the number of adsorbate atoms in the cell. To this end, E b was averaged over

measured in atomic layers are indicated. The observed Pb islands on the wetting layer had regular shapes, sharp edges and smooth tops, and tended to grow up with higher coverages suggesting that such islands are stable. It should be pointed out that above θ > 1 ML, one-layer thick Pb islands were never observed. At coverages greater than 3 ML (figure 7(f)), further formation of islands with sharp edges and smooth tops was observed. The island height distribution (number of Pb islands with a given height), however, showed strong peaks at relative heights corresponding to the number (N = 2, 4, and 6) of Pb atomic layers. During post-annealing to 1050 K, a change of the height of the lead magic islands was observed. Figure 8 shows STM images obtained after the post-deposition annealing (1050 K, 5 s) of Pb films (θ = 5.0 ML) deposited on Mo(110) at room temperature [20, 201]. The height profile (C), taken across the Pb islands (figure 8(c)), shows that only islands containing 2, 5, and 7 atomic layers of Pb were observed above the wetting layer. Close inspection of the STM topography image in figure 7(b) revealed the presence of two well-ordered lead superstructures denoted ε1 and ε2 [23]. Their geometrical parameters are described in the captions of figures 9(a) and (b). The structures ε1 and ε2 were observed simultaneously in different areas of the Mo(110) substrate. STM images indicated in both cases that a long-range order in the surface system was present. The densities of Pb atoms determined for the ε1 and ε2 structures were 4.5 × 1014 and 6.4 × 1014 atoms cm−2 , respectively, which correspond to 48% and 68% of the density of the close-packed (111) Pb 13


A Krupski

Figure 10. Pb deposition on the Ni3 Al(111) face at T = 300 K. (a) AES(t) plot of Ni MVV and Pb NVV peak heights. (1) AES(t) plot calculated for Frank–van der Merwe growth. (2) AES(t) plot calculated for the growth of three atomic layer thickness flat islands on the first lead layer. STM images at the following coverages. (b) θ = 1.00 ML (2070 Å × 2070 Å, IT = 122 pA, Ubias = 2.0 V). The inset presents a zoom (392 Å × 392 Å) in the area marked by a square. The arrows indicate holes in the wetting layer. (c) θ = 3.5 ML (1520 Å × 1520 Å, IT = 100 pA, Ubias = 3.2 V). The height of the Pb islands corresponds to 3, 5, 7 and 11 atomic layers of Pb(111). The hexagonal shape of the Pb islands is marked. (d) θ = 5.5 ML (4400 Å × 4400 Å, IT = 146 pA, Ubias = 7.1 V). The Pb islands have regular hexagonal shapes I, II and III with side lengths of 25, 30 and 45 nm, respectively. The different orientation of the Pb islands with respect to each other is marked. (e) Post-annealed θ = 5.5 ML at T = 400 K (6600 Å × 6600 Å, IT = 130 pA, Ubias = 7.6 V). The height of the Pb islands corresponds to two atomic layers of Pb(111) measured from the wetting layer. The inset presents differentiated STM images with enhanced contrast. (f) The height profile along line D from the image in (e). Adapted from [33]. Copyright 2013 by Elsevier.

not fit the experimental data (see curve (1) in figure 10(a)). A much better fit was obtained for the plots for growth up to 2.5 ML when a growth of flat Pb islands composed of three atomic layers (curve (2) in figure 10(a)) was assumed in the calculations. Another assumption that was made was that the first linear region of the curve corresponded to the formation of the first layer (θ = 1.0 ML) of lead. Figures 10(b)–(d) show STM images of Ni3 Al(111) surfaces with various Pb coverages and illustrate the evolution of growth morphology of the Pb layer deposited on Ni3 Al(111) at room temperature. Figure 10(b) shows a typical STM image corresponding to a monolayer coverage θ = 1.0 ML where Pb was seen to wet the Ni3 Al(111) surface almost completely. At this coverage only LEED spots associated with the p(4 × 4) structure were observed [32]. The density of Pb atoms in the p(4 × 4) structure is known to be 10.42 × 1014 atom cm−2 , which corresponds to 110% of the density of the close-packed (111) Pb layer. The almost perfect wetting in this system was due to the high specific surface free energy of the Ni3 Al(111) surface as compared with that of the Pb(111) surface. Since the total specific surface free energy is always required to be minimized, a covered Ni3 Al(111) surface was favored. As seen in the inset of figure 10(b), the first Pb layer on the Ni3 Al(111) surface was not free from dislocation defects. The arrows in

two or four adatoms from the supercell corresponding to the superstructures ε1 and ε2 , respectively (cf figures 9(c) and (d)). The obtained averaged adsorption energies are equal to 4.28 eV for ε1 and 4.29 for ε2 . It should be noted that the energetic properties of the lead superstructures considered in this section are consistent with experimental results presented in [24], indicating that the adsorption energy of lead adatoms increases with coverage. Indeed, the adsorption energies calculated for dense superstructures corresponding to a coverage of 2/3 ML were above 4.4 eV, and were noticeably higher than the values (below 4.3 eV) obtained for the more dilute superstructures given in figures 9(c) and (d). 2.7. Pb/Ni3 Al(111)

In the growth of lead on Ni3 Al(111) at room temperature, one linear component of the AES(t) plots for both the 61 eV nickel and 94 eV lead peaks was observed, see figure 10(a). This component of the AES(t) plots indicates a two-dimensional (2D) growth of the first lead layer. After the first monolayer, the quasi-linear shape of the second region of the AES(t) suggested that a two-dimensional growth continued that was not of the standard Frank–van der Merwe (FM) type. In this region the calculated AES(t) plot for a substrate with FM growth did 14


A Krupski

Table 2. This table summarizes the observed layer thicknesses (N) of magic height Pb islands for different adsorption systems. Layer thicknesses are measured with respect to the wetting layer; (u) indicates unstable Pb islands. Reproduced from [20, 22, 32–36, 40, 43–48, 50, 52, 61, 201].

Pb/substrate Ni3 Al(111)

Layer thickness (N) 3 3, 5, 7, 11 2

Temperature (K) 300 300 After annealing to 400

Methods AES, LEED, DEPES STM

Reference [32] [33]

Mo(110)

2, 4, 6, 7, 9 2, 5, 7 3, 5, 7, 9 2, 3 4(u), 5, 6, 7 5 5 5, 7 7 7 3, 5, 6, 7 4(u), 5 4, 5, 6, 7 4, 5, 6, 7 5, 6(u), 7, 8(u) 9, 10(u), 11 5 5, 6(u), 7, 8(u), 9, 10(u), 11 6, 8, 10–11, 15, 17, 20

300 After annealing to 1050 422 ≤ T ≤ 474 180 < T < 270 240 210 T ≤ 155 155 < T < 175 T > 175 185 180 40 240 200 200 180 200 300 ≤ T ≤ 400

STM STM LEEM, µLEED AES STM STM

[22] [20, 201] [34, 35] [36] [52] [43]

Ni(111) Si(111)

Ge(111) Cu(111)

figure 10(b) indicate areas where holes with an average depth of dh = 1.02 ± 0.35 Å in the wetting layer were observed. The observed two-dimensional growth of the first Pb wetting layer was thermodynamically driven by the lower surface free energy of Pb compared to Ni3 Al, see table 1. This wetting was still favorable despite the elastic energy resulting from the large lattice mismatch between Pb and Ni3 Al, ∼39% [32]. A similar wetting layer growth with dislocations was observed for Pb adsorption on Mo(110) [22] as described above in section 2.5. The presence of compressed wetting layers with a density higher than the metallic Pb(111) density by approximately 5% has been identified as the key factor that provides unusually fast and non-classical kinetics in the uniform height islands [34, 53]. It is not unusual to expect that the p(4 × 4) phase plays the same role and superfast diffusion could have been present in this system as well. At coverages of θ > 1 ML, three-dimensional growth was observed. Figure 10(c) shows STM images taken after 3.5 ML of Pb was deposited on the Ni3 Al(111) surface at room temperature. The Pb islands have a hexagonal shape, flat top and thickness of approximately 9, 14, 20, or 32 Å. The interlayer spacing along the [111] direction of Pb was measured to be d111 = 2.86 Å and therefore the islands had a thickness corresponding to 3, 5, 7 or 11 atomic layers, respectively. As Pb coverage was increased to 5.5 ML, three types of regular hexagonal islands with side lengths of 25, 30 and 45 nm were measured (figure 10(d)). Smaller irregularly shaped Pb islands were also observed between the main regular hexagonal islands. Furthermore, three different adsorption configurations of Pb were distinguishable, rotated by α = 10◦ ± 1◦ and β = 30◦ ± 6◦ with respect to each other. It should be noted that the different orientation of these hexagonal Pb islands was not correlated with their size. The presence of these multilayer lead islands agreed

LEED STM

[45]

SPA-LEED STM STM

[44] [46] [47]

STM/STS STM X-ray STM AES

[48] [50] [61] [50] [40]

with the island growth deduced from AES analysis at room temperature [32]. During post-deposition annealing (T = 400 K, 60 s) of Pb films (θ = 5.5 ML), a change of the height of the lead magic islands was observed, as seen in figures 10(e) and (f). Only islands containing two atomic layers of Pb were observed above the wetting layer. LEED patterns only corresponding to a weakly ordered p(4 × 4)-Pb structure were observed after post-deposition annealing [32]. These STM results confirmed AES studies at T = 400 K concerning a 3D growth mode at the same height above the wetting layer. Table 2 shows the addition of the Pb–Mo(110) and Pb–Ni3 Al(111) adsorption systems to the known group of systems where uniform island height selection during metal thin film growth has already been observed and interpreted in terms of quantum size effects (QSEs) [60–65]. Typically, these electronic effects have been observed on semiconducting [41, 43–59] or metal substrates [22, 32–42, 58]. For example, STM/STS observations of Pb islands on the Cu(111) surface face [40] indicated that the equilibrium distribution of island heights showed some heights appearing much more frequently than others. In all these systems a confining barrier restricted the electrons to be within the film because of misaligned electronic bands in the metal film and the substrate so that no states were available to move from the film to the substrate. The magic preferred heights correspond to islands with a quantum well state far from the Fermi energy, while the forbidden heights appear to be those that have a quantum well state close to the Fermi level. Another typical example of uniform island growth is Pb deposited on Si(111) at a temperature between 170 and 250 K and typically results in the formation of Pb islands with N = 4 (unstable), 5, 6, 7 and 9 atomic layer heights above the wetting layer. Nevertheless, the occurrence of magic heights is related to the increased stability associated 15


A Krupski

Figure 11. (a) SPA-LEED image of Al2 O3 /Ni3 Al(111) taken at an electron energy of 110 eV. (b) Simulation of the LEED image. The units cells are p(2 × 2) aluminum sub-lattice of the substrate (green), network structure (blue), and dot structure (red). Adapted from [92]. Copyright 2005 by Elsevier.

These two rings could be related to two superstructures, the ‘network’ and the ‘dot’ structures. The fact, that we observed 12 spots was related to the superposition of two hexagonal rotational domains of the alumina film, which have already been observed by STM [207]. The angle between domains was 24◦ which was in good agreement with the STM results. The larger of the two rings could be attributed to the diffraction from the ‘network’ structure. The size of the unit cell of this structure (blue lines in figure 11(b)) could be calculated when using the known size of the substrate lattice. The value of bnet (SPA-LEED) was 24 Å. Similarly, the inner circle could be attributed to diffraction from the ‘dot’ structure. In this case the value for the unit cell size bdot (SPA-LEED) was equal to 41.6 Å. Furthermore, it was determined that the angle of rotation between the network structure and the substrate was 18◦ as calculated from the SPA-LEED image. Based on these findings it was possible to simulate the LEED pattern using the substrate lattice and unit cells of the ‘network’ and ‘dot’ structures, the blue and red circles in figure 11(b), respectively. The strong dependence of the corrugation on the tunneling voltage can be seen in the STM images of figure 12. At bias voltages around Ubias = 3.2 V (see figure 12(a)) a hexagonal arrangement of hollows was observed where every hollow was surrounded by a hexagonal ring of bright dots. This structure possessed an apparent lattice constant of 24 Å and will be referred to as ‘network structure’ in the following text. However, closer inspection of this structure indicated that not all hollows had the same depth. Some hollows were up to 40 pm shallower than the others. This fact is clearly visible in the cross section shown in figure 12(d). The shallower hollows formed a regular arrangement, which formed the basis of a hexagonal unit cell with a lattice vector of 41.6 Å. This unit cell of the alumina film was more easily observed in the STM images, figures 12(b) and (c), due to a contrast reversal, which was found at bias voltages of 2.0 and 4.2 V, and will be referred to as ‘dot structure’ in the following text. Recent AFM measurements have shown that, indeed, the surface was atomically flat and that the ‘dot structure’ corresponded to the actual surface unit cell [208, 209]. Scanning tunneling spectroscopy (STS) was used to study the electronic properties of the system Al2 O3 /Ni3 Al(111) [157]. In figure 13(a) STS

with islands of specific thickness. Ayuela et al [202] studied quantum size effects of Pb overlayers using density functional theory (DFT) calculations and analytical models. This work demonstrated that the high stability of Pb islands on metallic or semiconductor substrates at even higher coverages was supported by an extra second quantum beat pattern in the energetics of the metal film as a function of the number of atomic layers. It seemed that this pattern was triggered by the butterfly-like shape of the Fermi surface of lead in the [111] direction and supported the detection of stable magic islands of greater heights than measured until now. The most stable magic islands from this DFT study had heights corresponding to the number (N = 2, 4, 6, 7, 9, 11, 13, 14, 16, 18, 20 ¨ . . . ) of lead atomic layers. Recently, Ozer et al [203] and Jia et al [204] have shown that, similarly to the case of pure Pb, Pb1−x Bix alloy films on Si(111) 7 × 7 exhibit a reentrant bilayer-by-bilayer growth mode but with different quantum beating patterns. The observed stable thickness sequence was 4, 6*7, 9, 11, 13, 15, 17, 19*20. . . , ML, where the asterisks indicate the locations of the even–odd crossover. At the higher concentration of Pb80 Bi20 , the alloys did not exhibit quantum growth but simply followed classical layer-by-layer growth. 3. Molecule and metal growth on the Al2 O3 /Ni3 Al(111) 3.1. Al2 O3 /Ni3 Al(111)

It is known that controlled oxidation of the clean Ni3 Al(111) surface results in the formation of a closed double layer alumina film with a distinctive long-range order [156, 157, 205, 206]. Recent SPA-LEED and STM investigations [92] showed that under the preparation conditions used, the Al2 O3 film forms a long-range ordered hexagonal superstructure on the nanometer scale. Figure 11(a) shows an overview SPA-LEED image of the alumina film on Ni3 Al(111) at an electron energy of 110 eV. Apart from the aluminum p(2 × 2) superstructure of the substrate (figure 11(b)), a very large number of additional superstructure spots originating from the Al2 O3 film could be seen. Around the (0, 0) spot two rings consisting of twelve superstructure spots each (figure 11(a)) could be observed. 16


A Krupski

Figure 12. LT-STM images of Al2 O3 /Ni3 Al(111) measured at T = 23 K and different bias voltages. The unit cell of the hexagonal alumina

superstructure is drawn in red. The image parameters are (a) the ‘network structure’ at Ubias = 3.2 V (278 Å × 278 Å, IT = 122 pA), (b) the ‘dot structure’ at Ubias = 2.0 V (278 Å × 278 Å, IT = 105 pA), (c) the ‘dot structure’ with inverted contrast at Ubias = 4.2 V (123 Å × 123 Å, IT = 103 pA). In (d) two cross sections taken at the positions indicated in (a) and (b) are shown. Parts (a), (b) and (d) are adapted from [92]. Copyright 2005 by Elsevier. Part (c) is adapted from [95]. Copyright 2008 by Elsevier.

Figure 13. LT-STS measurements of Al2 O3 /Ni3 Al(111) at T = 23 K [157]. (a) Tunneling spectra measured at the points indicated on the STM image (50 Å × 50 Å, IT = 100 pA) in the inset. Curves (A) and (B) were acquired after the feedback loop was opened at 3.2 V. The corresponding modulation frequency was 10 kHz with an amplitude of 40 mV. (b) Fourier filtered STM image (53 Å × 53 Å, IT = 95 pA, Ubias = 3.2 V). (c) Tunneling spectra (A)–(F) acquired at the points indicated in (b).

spectra are shown together with the corresponding STM image in which the positions are shown where the spectra were acquired. Curves (A) and (B) were taken at a deep hollow site and on the hexagon surrounding that site, respectively. In the range −4 to −6 V the main ascent of the density of states of the oxide valence band was visible. At the large tip–sample separation used for these spectra, no electronic states in the range −4–2.7 V were detected. Above a value of 2.7 V an increase of the density of states was seen, which corresponded to the lower edge of the oxide conduction band. From this it was possible to derive a value of 6.7 eV for the bandgap of the Al2 O3 film. The bandgap of an Al2 O3 film prepared

on NiAl(110) has also been quoted before at 6.7 eV [210]. The comparison of curves (A) and (B) in figure 13(a) shows that the bandgap measured by STS was a function of the tip position. In figure 13(b) an STM image is shown for which a set of tunneling spectra were measured at different locations (A)–(F) and the corresponding STS spectra are displayed in figure 13(c). The center of this STM image (A) shows a shallow hollow that was a coincidence point of the commensurate Al2 O3 structure [92]. The spectra were taken over a bias voltage range of 1.5–3.3 V. This corresponded to an energy range in the vicinity of the oxide conduction band edge where electronic states originating from the oxide could be detected. 17


A Krupski

coverage of θCuPc = 0.03 ± 0.02 ML, figure 14. Individual molecules were imaged as four-lobe protrusions; this was in good accordance with the calculated LUMO [220]. Small differences like the obviously missing electron density at the outer carbon atoms of the benzene rings were most likely due to interactions with the substrate. The surprisingly large side length of the adsorbed CuPc in the STM images of ∼20 Å compared to the expected value of 12 Å (see figure 14(a)) was a frequently observed phenomenon, which could be explained by a small overlap of the orbitals of tip and molecule even at greater distances [221]. This resulted in a small tunneling current, even when the tip was not exactly positioned above the molecule. At this coverage of θCuPc = 0.03 ± 0.02 ML local dI /dV curves of individual CuPc molecules were recorded. Since the CuPc molecules and their arrangement on the surface were quite sensitive to the alternating high electric field of the tip during the STS measurements some precautions had to be taken. Only spectra that did not influence the position and the shape of the molecules were considered. To ensure this, STM images of the individual molecule were taken before and after the spectroscopic measurements. Interestingly, at different positions of the tip above the molecule (figure 14(c)) the local dI /dV curves shown in figure 14(d) looked quite similar. They exhibit only one maximum centered around 1.2 eV, which could be clearly identified as the energy of the degenerate LUMO. This exhibited a marked difference from former investigations of CuPc adsorbed on Al2 O3 /NiAl(110) by Qiu et al [113]. These authors observed two distinct molecular symmetries (D4h and D2h ). The ones with D2h -symmetry featured two different maxima in the local dI /dV curves, which were detected on two pairs of oppositely located lobes. This was explained by the lifting of the degeneracy of the LUMO by reason of a Jahn–Teller effect. However, we could only detect molecules with D4h -symmetry and degenerate LUMO on Al2 O3 /Ni3 Al(111). This could have been due to small differences in the interaction of CuPc with Al2 O3 /N3 Al(111) as compared to Al2 O3 /NiAl(110). In fact, we found only one molecular CuPc species, which adopted two molecular orientations rotated by 30◦ with respect to each other. Figure 15(a) shows a larger STM image of the oxidized Ni3 Al(111) surface with a CuPc coverage of θCuPc = 0.03 ± 0.02 ML. At this coverage the molecules were always adsorbed with the molecular plane parallel to the substrate surface. At this low coverage only individual molecules were found, which apparently did not form an ordered long-range structure. Closer inspection of the images showed that the metal center of the CuPc was always positioned above one of the bright spots of the hexagons surrounding the hollows of the ‘network structure’. However, the ‘dot structure’ also seemed to affect the position of the molecules, even though they were never centered directly on a coincidence point [92] in this structure. A significant affinity to surface defects, e.g. terrace edges or imperfections in the oxide layer, was not, however, observed at low temperature and this was probably due to the low mobility of the CuPc. Furthermore, a distinct orientation of the CuPc molecules with respect to the substrate could be observed at low coverage. As demonstrated in figure 15(a), only two different molecular adsorption configurations of

Spectrum (A) was measured directly at the coincidence point. It exhibited one maximum at approximately 2.4 V, which was in agreement with previous studies [211]. Spectra (A)–(F) were measured at an increasingly larger distance from the coincidence point. Going from spectrum (B) to spectrum (E) the maximum at 2.4 V decreased and was no longer visible in spectrum (E). In addition, an electronic state around 3.0 V appeared. In spectra (B) and (C) the maximum around 2.4 V could be distinguished as well as the one around 3.0 V. These results clearly showed that in this energy range there existed two electronic states with different origins at different locations within the oxide supercell. The electronic state at 3.0 V was present everywhere on the surface except at the coincidence points. Spectrum (F) was measured in the center of a deeper hollow of the structure and was nearly identical to spectrum (E), which corresponded to a completely different point on the alumina film. Previous studies on the Al2 O3 /Ni3 Al(111) system showed a pronounced template effect for growth of a number of metals. While the ‘network structure’ was shown to have a direct influence on the growth of V [212] and Mn [213] clusters, a template effect of the ‘dot structure’ for Cu [212], Ag [214], Au [214] and Pd [107] clusters has been demonstrated. The next logical step was the investigation of less strongly interacting adsorbates such as organic molecules, and this will be discussed in section 3.2. 3.2. C32 H16 N8 Cu/Al2 O3 /Ni3 Al(111)

Thin films formed from organic molecules including copper phthalocyanine are rapidly gaining attention in different technological fields [215]. Copper phthalocyanine is an organic molecule with a molecular weight of 576 g mol−1 and a side length of 12 Å. It possesses D4h -symmetry and is completely planar in the gas phase [216]. The extensive aromatic π -electron system and the chelating properties of the phthalocyanine core result in stable complexes with copper and many other metal ions. In the ground state copper is in the oxidation state +II. Pure CuPc is known to crystallize in several modifications [217]. The α- and β-modifications are the most stable structures. Both are composed of stacked layers of planar Pc molecules, which show different stacking sequences and layer spacings of 0.34 nm [216] and 0.33 nm [218], respectively. The stacked arrangement of the molecules enables strong interactions between their p-electron systems (‘π -stacking’). This results in a high thermal stability and low vapor pressure. DFT calculations have shown that the HOMO is strongly affected by the dx 2 −y 2 -orbital of the copper ion [219]. This results in a relatively high energy due to the electronic splitting in the quadratic planar ligand field. Because of the d9 -configuration of the copper ion the HOMO is only half filled and may theoretically take part in the tunneling process. However, due to the missing z-component of the dx 2 −y 2 -orbital this seems to be improbable, so the STM depiction should be dominated by the LUMO, which has eg-symmetry. The large energy gap of more than 1.5 eV between the LUMO and the higher unoccupied orbitals should prevent their participation in the tunneling process. This assumption was proven by our STM investigations [95] at a 18


A Krupski

Figure 14. Copper phthalocyanine (C32 H16 N8 Cu). (a) Chemical structure. (b) Calculated shape of the LUMO [109]. (c) STM image

measured at T = 23 K of a single CuPc molecule adsorbed on Al2 O3 /Ni3 Al(111) (33 Å × 33 Å, IT = 100 pA, Ubias = 1.2 V). (d) dI /dV tunneling spectra corresponding to the numbered positions in (c) (AC modulation voltage of 60 mV at 1 kHz). Adapted from [95]. Copyright 2008 by Elsevier.

Figure 15. LT-STM images measured at T = 23 K of CuPc adsorbed on Al2 O3 /Ni3 Al(111). (a) θ = 0.03 ± 0.02 ML (130 Å × 130 Å,

IT = 107 pA, Ubias = 1.6 V). The different orientation of the single molecules with respect to each other is marked. (b) θ = 1.0 ± 0.2 ML (174 Å × 174 Å, IT = 110 pA, Ubias = 3.2 V). (c) Height profile along the marked line in (b). (d) θ = 0.03 ± 0.02 ML (138 Å × 138 Å, IT = 110 pA, Ubias = 3.2 V), CuPc on the ‘striped phase’ of Al2 O3 [222]. Parts (a)–(c) are taken from [95]. Copyright 2008 by Elsevier.

19


A Krupski

30◦

CuPc were distinguishable, rotated by with respect to each other. This suggested that a distinct adsorption geometry was preferred in order to maximize the adsorption energy as a result of the interaction of the square shaped molecule and the hexagonal alumina film. The intermolecular interactions between the aromatic π -systems of the molecules became ever more important with increasing CuPc coverage. Figure 15(b) shows that the template effect of the substrate was not strong enough to compensate for these interactions, even at a moderate coverage of θCuPc = 1.0 ± 0.2 ML. At 140 K the growth of the second molecular layer started before the monolayer was complete, resulting in the formation of randomly orientated molecular clusters [95]. Here, the molecules were always oriented with the molecular plane parallel to the substrate. Interestingly, even in the second layer the CuPc could be imaged with submolecular resolution. In an earlier work by Lippel et al [109] it was reported that this was not possible on Cu(100), Si(111) and Au(111). The measured height difference between molecules in the first and second layers of 0.35 nm (figure 15(c)) was comparable to the layer distance of 0.34 nm in the α-modification of pure CuPc, indicating a similar intermolecular interaction as in the bulk of this compound. The layer distance, under normal thermodynamic conditions, in the stable β-modification was 0.33 nm. Figure 15(d) shows an STM image of the CuPc molecules adsorbed on the ‘striped phase’ of Al2 O3 film [222]. In order to characterize the interaction of the molecules and the surface in greater detail the surface was annealed. After annealing to 300 K the formation of CuPc agglomerates could be observed, indicating a high lateral mobility of CuPc. Single molecules could not be observed. After annealing the sample to 350 K the CuPc was desorbed completely. According to this, it could be concluded that CuPc only weakly chemisorbs on to the oxidized Ni3 Al(111) surface and that the intermolecular forces are much stronger than the molecule–surface bond [95].

with the observation of three-dimensional (3D) particles at 300 K. In figures 16(a) and (c) the iron clusters grown at a low deposition temperature (130–160 K) are depicted. The surface structure grown to low iron coverage (θ = 0.07 ± 0.02 ML) is shown in figures 16(a), (d) and (e). The iron clusters appear as bright spots on top of the long-range modulated alumina film. As evidenced in the upper part in figure 16(a) taken at Ubias = 3.5 V and in figure 16(d) taken at 3.2 V, the iron clusters nucleated first in the hollows corresponding to points of the ‘network structure’ of the alumina film. A similar behavior was observed for V and Mn [106], also on alumina. The noble metals Ag, Cu, Au and Pd, however, nucleate preferentially on the ‘dot structure’ of Al2 O3 /Ni3 Al(111) with clusters being 41.6 Å apart from each other [106, 107]. The apparent height of the iron clusters grown between 130 and 160 K was 2.6 ± 0.3 Å with respect to the oxide surface, as inferred from the line profile shown in figure 16(b). It is concluded from STM that the islands are one atomic layer in height. Cluster growth is a non-equilibrium process and the final macroscopic state of a system depends on the different activation energies for all microscopic kinetic processes that are involved. At 130 K, the growth is limited to 2D, since a diffusing Fe atom laterally approaching a 2D island will attach itself to the step but cannot climb up to reach the island’s top surface. However, an Fe atom landing on top of an Fe island can overcome the Ehrlich–Schwoebel barrier and will therefore move down-step. At an Fe coverage of θ = 1.0 ± 0.2 ML the nucleation sites of the ‘network structure’ were all filled with Fe clusters which partly coalesced (figure 16(c)). From the Fourier transform (FFT) shown in the inset, a long-range hexagonal order of the iron clusters that corresponded to a distance of 24 Å between the clusters could be observed. Here, the iron clusters were still two-dimensional as could be deduced from their apparent height of 2.6 ± 0.3 Å. In another experiment (not shown) iron clusters grown at 150 K were annealed for 20 min at 350 K and STM images were then taken at 150 K. Again, the apparent cluster height was the same as that attributed to 2D islands. Thus, the shape of the iron clusters was determined by the sample temperature during deposition and remained unchanged during annealing to room temperature or even slightly above. The information gained here on the stability was an important prerequisite to subsequently running catalytic reactions or investigating the temperature dependence of magnetic properties, such as the zero field susceptibility [232]. The fact that the growth and annealing temperature have quite different influences on the cluster shape has been previously observed in the case of Pd/MgO(100) [115]. This phenomenon was caused by the generally higher energy barrier required to transform an entire island into its equilibrium shape as compared to that required to put the atoms, being added one by one during growth, onto thermodynamically favorable sites, and being lower coordinated. Therefore islands may well remain 2D upon annealing to temperatures higher than the growth temperature where they are 3D. Figure 17 shows the bias voltage dependence on the height and diameter of the iron clusters [222]. The observed increase in cluster height at around 1.5 V could be attributed to empty states of the Fe clusters [233] whereas empty states of the Al2 O3 film contributed around 2.4 V [157].

3.3. Fe clusters on Al2 O3 /Ni3 Al(111)

Several recent review articles have described the growth, structure and electronic properties of various metallic overlayers on different oxide surfaces [115, 223–228]. In this section, we concentrate our review on recent results concerning the behavior of Fe on Al2 O3 . The iron clusters grew and gradually covered the entire Al2 O3 /Ni3 Al(111) surface with increasing coverage [96]. However, the growth of a continuous wetting layer of Fe has not been observed. This result was in agreement with the thermodynamic criterion for the wetting of alumina [229–231]. From energetic considerations wetting should not occur on the alumina surface if E adh < 2γv/m , where E adh is the adhesion energy and γv/m is the surface free energy at the vacuum metal interface for liquid metal. The definition of E adh is the energy needed to separate the metal and the oxide at their interface resulting in a bare oxide surface. E adh for iron on alumina was measured to be 1205 mJ m−2 at 1853 K [231]. E adh is mainly energetic and therefore nearly temperature independent. γv/m could be extrapolated to temperatures below the phase transition entailing an increasing γv/m by measuring γv/m at 1853 K to be 1787 mJ m−2 [229– 231]. The thermodynamic criterion (no wetting) was consistent 20


A Krupski

Figure 16. STM images of Fe clusters on Al2 O3 /Ni3 Al(111). (a) Deposited at 160 K (θ = 0.07 ± 0.02 ML, 198 Å × 98 Å, IT = 106 pA, Ubias = 3.5 and 4.2 V in upper and lower parts, respectively; TSTM = 23 K). The unit cell of the hexagonal alumina superstructure is drawn in red. (b) Corresponding line profiles along the blue and green lines in (a). (c) Deposited at 130 K (θ = 1.0 ± 0.2 ML, 681 Å × 681 Å, IT = 78 pA, Ubias = 0.7 V, TSTM = 300 K). The FFT shown in the inset evidences a hexagonal arrangement of the iron clusters with a nearest neighbor distance of 24 Å. (d) Fourier filtered STM image (62 Å × 62 Å, IT = 151 pA, Ubias = 3.2 V). (e) Fourier filtered STM image (62 Å × 62 Å, IT = 151 pA, Ubias = 0.19 V), The (2 × 2) unit cell of the substrate is drawn in yellow [222]. Parts (a)–(c) are adapted from [96]. Copyright 2006 by Elsevier.

4. Conclusions

800 K, rearrangement of existing Au and Sn films occurred and Au–Mo and Sn–Mo surface alloys were formed. Finally, for Pb films on Mo(110) and Ni3 Al(111) surfaces, two-dimensional growth of the first Pb monolayer (wetting layer) took place below 1 ML. For θ > 1 ML, three-dimensional growth of the Pb islands was observed on the wetting layer with strongly preferred atomic scale ‘magic heights’ and flat tops. Here, at coverages between 1 and 2 ML, STM showed the growth of two-atom high (N = 2) lead islands. Furthermore, at higher coverages, the island height distribution showed peaks at relative heights corresponding to N = (2, 4, 6, 7 and 9) and N = (3, 5, 7 and 11) atomic Pb layers for Pb/Mo(110) and Pb/Ni3 Al(111), respectively. As the Pb/Mo(110) and Pb/Ni3 Al(111) samples were post-annealed to 1050 K and 400 K, respectively, a change of the height of the lead ‘magic islands’ was observed. The island height distribution exhibited peaks in the relative heights corresponding to N = (2, 5 and 7) and N = (2) atomic Pb layers above the wetting layer for the Pb/Mo(110) and Pb/Ni3 Al(111) respectively. In the second section of this review the properties of alumina films and the growth morphology of copper phthalocyanine and Fe on the Al2 O3 /Ni3 Al(111) surface were focused on. The structural relation between an ultra-thin Al2 O3 film and a Ni3 Al(111) substrate has been investigated. It has been found that only the so-called ‘dot structure’ is a real superstructure of the √ √ Al2 O3 film. The unit cell size of this ( 67 × 67)R47.784◦ superstructure (‘dot structure’), bdot , has been determined with

Recent studies on the growth morphology of thin films on metallic and oxide surfaces have been summarized. The investigation leads to the following new results. In the first section of the review the growth behavior of ultra-thin Ag, Au, Sn and Pb films on Mo(110) was discussed as well as, in addition to this, Pb on the Ni3 Al(111) surface. For Ag on the Mo(110) surface, an analysis of measurements indicated that three-dimensional growth of a Ag film was observed. For submonolayer coverage, the growth of Ag was mediated by a two-dimensional step-flow mechanism. During the initial stage of this growth, the first Ag layer nucleated and created islands at Mo step edges. In the monolayer coverage range, the decoration of substrate steps by Ag could be distinguished by the presence of a fractional step height at the Ag–Mo boundary. As the sample was post-annealed to 700 K, the morphology of the surface changed. The step-flow growth in this case gave rise to a regular Ag nanostripe network attached to Mo(110) step edges. In the case of Au and Sn layer-by-layer growth for the first two layers was observed. For submonolayer coverage, the gold and tin created one-atom high islands on the Mo terraces. Gold nucleated preferentially ¯ and [111] ¯ directions; however, tin did not along the [001] exhibit a preferential nucleation direction. In the case of gold and tin no ordered regions were observed in the completed first and second layers. As the samples were post-annealed to 21


A Krupski

Figure 17. (a) Set of STM images of Fe clusters deposited at 160 K on the Al2 O3 /Ni3 Al(111) surface recorded as a function of bias voltage (θ = 0.07 ± 0.02 ML, 480 Å × 480 Å, IT = 102 pA, TSTM = 23 K). Parts (b) and (c) present the bias voltage dependence of the height and diameter of the Fe clusters marked in the STM images, respectively. The increase in cluster height at ∼1.5 eV can be attributed to empty states of the Fe clusters whereas empty states of the Al2 O3 film contribute around ∼2.4 eV [222].

high precision to be 41.6 Å. SPA-LEED as well as LT-STM measurements clearly showed the commensurability of the Al2 O3 film and the Ni3 Al(111) surface. This revealed that the structure of the alumina film is determined by both the intrinsic structure of the film itself and the structure of the substrate. In STM the appearance of the copper phthalocyanine as four-lobe protrusions was in good accordance with the calculated shape of the lowest unoccupied molecular orbital (LUMO). The weakly chemisorbed CuPc molecules showed a high lateral mobility above 250 K. At 350 K, thermal desorption was complete. At very low coverages, below 0.1 ML, a weak template effect of the alumina surface was observed. Only two different molecular adsorption orientations of CuPc were distinguishable, which were rotated by 30◦ with respect to each other. With increasing CuPc coverage the strong intermolecular π -electron interactions compensated the weak template effect of the surface resulting in the growth of 3D molecular clusters without any visible ordering. The distance between the molecules in the first and second layers was 3.5 Å, which is similar to that in the α-modification of bulk CuPc. By the use of STS the degenerated LUMO of the adsorbed

CuPc molecules has been identified at an energy of 1.2 eV above E F . Finally, the results presented here for the growth morphology of iron on Al2 O3 /Ni3 Al(111) indicated that the nucleation site of iron clusters was largely independent of deposition temperature, while the cluster shape was 2D in the range 130–160 K and 3D at 300 K. At coverages close to one monolayer the Fe clusters formed a regular hexagonal lattice with 24 Å spacing, while at low coverages they occupied only some of the available nucleation sites. In comparison with Mn- and V-cluster growth on alumina films, the observed perfection of the order of the cluster arrays increases from Mn- to Fe- to V-cluster arrays [106]. Iron clusters grown between 130 K and 160 K were stable against coarsening and shape changes upon prolonged annealing to temperatures up to 350 K. Future theoretical and experimental work is required firstly to further study the electronic structure and stability of these uniform Pb ‘magic islands’ on Mo(110) and Ni3 Al(111) surfaces, as this type of growth could have impact on the engineering of stable materials and devices with nanometer-scale dimensions, and secondly to confirm Sn–Mo and Au–Mo alloy formation. 22


A Krupski

Acknowledgments

[16] Gotoh Y, Horii A, Kawanowa H, Kamei M, Yumoto H and Gonda T 1995 J. Cryst. Growth 146 198 [17] Horii A, Kawanowa H, Kamei M, Yumoto H, Gonda T and Gotoh Y 1996 Cryst. Res. Technol. 31 875 [18] Krupski A 2011 Surf. Sci. 605 424 [19] Rodriguez J A and Kuhn M 1995 Surf. Sci. 330 L657 [20] Krupski A 2011 Surf. Sci. 605 1291 [21] Maehara Y, Kimura T, Kawanowa H and Gotom Y 2005 J. Cryst. Growth 275 1619 [22] Krupski A 2009 Phys. Rev. B 80 0354241 [23] Ko´smider K, Krupski A, Jelinek P and Jurczyszyn L 2009 Phys. Rev. B 80 1154241 [24] Tikhov M and Bauer E 1988 Surf. Sci. 203 423 [25] Jo S, Gotoh Y, Niski T, Mori D and Gonda T 2000 Surf. Sci. 454–456 729 [26] Gustafson J, Mikkelsen A, Borg M, Lundgren E, Kohler L, Kresse G, Schmid M, Varga P, Yuhara J and Torrelles X 2004 Phys. Rev. Lett. 92 126102 [27] Gustafson J, Mikkelsen A, Borg M, Andersen J N, Lundgren E, Klein C, Hofer W, Schmid M, Varga P and Kohler L 2005 Phys. Rev. B 71 115442 [28] Campbell C T 2006 Phys. Rev. Lett. 96 066106 [29] Todorova M, Li W X, Ganduglia-Pirovano M V, Stampfl C, Reuter K and Scheffler M 2002 Phys. Rev. Lett. 89 096103 [30] Radican K, Berdunov N, Manai G and Shvets I V 2007 Phys. Rev. B 75 1554341 [31] Okada A, Yoshimura M and Ueda K 2007 Surf. Sci. 601 1333 [32] Mi´sków K and Krupski A 2013 Appl. Surf. Sci. 273 554 [33] Mi´skòw K, Krupski A and Wandelt K 2014 Vacuum 101 71 [34] Bollmann T R J, Gastel R, Zandvliet H J W and Poelsema B 2011 New J. Phys. 13 103025 [35] Bollmann T R J, Gastel R, Zandvliet H J W and Poelsema B 2011 Phys. Rev. Lett. 107 136103 [36] Krupski A and Mróz S 2002 Phys. Rev. B 66 035410 [37] Krupski A 2007 Ann. Chim. Mater. 32 395 [38] Krupski A 2006 Phys. Status Solidi b 243 467 [39] Hinch B J, Koziol C, Toennies J P and Zhang G 1989 Europhys. Lett. 10 341 [40] Otero R, Vázquez de Parga A L and Miranda R 2002 Phys. Rev. B 66 115401 [41] Slomski B, Landolt G, Muff S, Meier F, Osterwalder J and Dil J H 2013 arXiv:1306.0245 [cond-mat.mtrl-sci] [42] Dil H J, Kim J W, Gokhale S, Tallarida M and Horn K 2004 Phys. Rev. B 70 045405 [43] Kuntova Z, Hupalo M, Chvoj Z and Tringides M C 2006 Surf. Sci. 600 4765 [44] Budde K, Abram E, Yeh V and Tringides M C 2000 Phys. Rev. B 61 10602 [45] Menzel A, Kammler M, Conrad E H, Yeh V, Hupalo M and Tringides M C 2003 Phys. Rev. B 67 165314 [46] Chan T L, Wang C Z, Hupalo M, Tringides M C and Ho K M 2006 Phys. Rev. Lett. 96 226102 [47] Binz S M, Hupalo M and Tringides M C 2008 Phys. Rev. B 78 193407 [48] Su W B, Chang S H, Jian W B, Chang C S, Chen L J and Tsong T T 2001 Phys. Rev. Lett. 86 5116 [49] Kim J, Qin S, Yao W, Niu Q, Chou M Y and Shih Ch 2010 Proc. Natl Acad. Sci. 107 12761 ¨ [50] Ozer M M, Jia Y, Wu B, Zhang Z and Weitering H H 2005 Phys. Rev. B 72 113409 [51] Kesaria M, Kumar M, Govind and Shivaprasad S M 2009 Appl. Surf. Sci. 256 576 [52] Hupalo M and Tringides M C 2007 Phys. Rev. B 75 235443

The projects described in this review were financially supported by the Ministerium für Wissenschaft und Forschung des Landes Nordrhein-Westfalen (MWF-NRW) and the Deutsche Forschungsgemeinschaft (DFG) through SFB 624 and the priority program ‘Organic Field Effect Transistors’, all of which are gratefully acknowledged. The work of AK was supported by the University of Wroclaw under grant nos 2016/W/IFD/2006, 2016/W/IFD/2007, 2016/W/IFD/2008, 2342/W/IFD/10, 2016/W/IFD/10, 1010/S/IFD/10 and 1010/S/ IFD/11. The author also gratefully acknowledges financial support gained through a fellowship from the Alexander von Humboldt and Hertie Foundations (III 2003–XII 2004, I–III 2011). The work reviewed here would have been impossible without the contribution of my colleagues. I am particularly grateful to Klaus Wandelt for many insightful discussions and his continuing support, which made part of this work possible. I am, furthermore, greatly indebted to the past members of his research group at Bonn University who have contributed partly to the results reviewed here, in particular Stefan Degen, Marko Moors, Marko Kralj and Conrad Becker for their contributions to studies of aluminum oxide. I would like to thank all the people who have contributed with experimental, theoretical and technical assistance work for metal growth on Mo(110), in particular Leszek Jurczyszyn, Krzysztof Ko´smider, Barbara Pieczyrak, Adam Paruszewski, Wojciech Linhart, Dominika Grodzińska, Izabela Cebula, Zbigniew Jankowski, Mirosław KarpickiAntczak, Jakub Cichos, Sławomir Chomiak and his coworkers (Jerzy Chmiel, Ryszard Wiacek and Jerzy Kurowski). Finally, I would like also to thank Katarzyna Mi´sków for her contribution to the studies of Pb on Ni3 Al(111) and S R C McMitchell for his help and very useful discussions during preparation of this topical review. References [1] Himpsel F, Ortega J, Mankey G and Willis R 1998 Adv. Phys. 47 511 [2] Larsen J and Chorkendorff I 1999 Surf. Sci. Rep. 35 163 [3] Kawanowa H, Suzuki J, Mahara Y, Gotom Y and Souda R 2005 J. Cryst. Growth 275 1615 [4] Mikkelsen A, Ouattara L and Lundgren E 2004 Surf. Sci. 557 109 [5] Murphy S, Osing J and Shvets I V 2003 Surf. Sci. 547 139 [6] Murphy S, Usov V and Shvets I V 2007 Surf. Sci. 607 5576 [7] Tamori T, Kazanowa H and Gotom Y 2007 Surf. Sci. 601 4412 [8] Jo S and Gotom Y 2000 Surf. Sci. 464 145 [9] Maehara Y, Kazanowa H and Gotom Y 2006 Surf. Sci. 600 3575 [10] Zhou Y G, Zu X T, Nie J L and Ciao H Y 2008 Chem. Phys. 353 109 [11] Katoh A, Miwa H, Maehara Y, Kazanowa H and Gotom Y 2004 Surf. Sci. 566–568 181 [12] Krupski A 2010 Surf. Sci. 604 1179 [13] Bauer E and Poppa H 1984 Thin Solid Films 121 159 [14] Gotoh Y and Yanokura E 1990 J. Cryst. Growth 99 588 [15] Gotoh Y and Yanokura E 1990 Surf. Sci. 269/270 707 23


A Krupski

[90] Boucetta S, Chihi T, Ghebouli B and Fatmi M 2010 Mater. Sci.-Poland 28 347 [91] Oramus P, Masssobrio C, Kozłowski M, Kozubski R, Pierron-Bohnes V, Cadeville M C and Pfeiler W 2003 Comput. Mater. Sci. 27 186 [92] Degen S, Krupski A, Kralj M, Langner A, Becker C, Sokolowski M and Wandelt K 2005 Surf. Sci. 576 L57 [93] Kresse G, Schmid M, Napetschnig E, Shishkin M, Koehler L and Varga P 2005 Science 308 1440 [94] Stierle A, Renner F, Streitel R, Dosch H, Drube W and Cowie B C 2004 Science 303 1652 [95] Moors M, Krupski A, Degen S, Kralj M, Becker C and Wandelt K 2008 Appl. Surf. Sci. 254 4251 [96] Lehnert A, Krupski A, Degen S, Franke K, Decker S, Rusponi S, Kralj M, Becker C, Brune H and Wandelt K 2006 Surf. Sci. 600 1804 [97] Thomas J M and Thomas W J 1997 Principles and Practice of Heterogeneous Catalysis (Weinheim: VCH) [98] Wilk D, Wallace R M and Anthony J M 2001 J. Appl. Phys. 89 5243 [99] Tada H, Fujii Y and Matsushige K 1998 Mol. Cryst. Liq. Cryst. 316 371 [100] Newton M I, Starke T K H, McHale G and Willis M R 2000 Thin Solid Films 360 10 [101] Tada H, Tanimura Y, Fujii Y and Matsushige K 1999 Mol. Cryst. Liq. Cryst. 327 283 [102] Friend R H et al 1997 Nature 397 121 [103] Clemens W, Fix W, Ficker J, Knobloch A and Ullmann A 2004 J. Mater. Res. 19 1946 [104] Horowitz G 2004 J. Mater. Res. 19 1924 [105] Kahn A, Koch N and Gao W 2003 J. Polym. Sci. B 41 2529 [106] Becker C, Rosenhahn A, Wiltner A, Bergmann K, Schneider J, Pervan P, Milun M, Kralj M and Wandelt K 2002 New J. Phys. 4 75 [107] Degen S, Becker C and Wandelt K 2004 Faraday Discuss. 125 343 [108] Gimzewski J K, Stoll E and Schlittler R R 1987 Surf. Sci. 181 267 [109] Lippel P H, Wilson R J, Miller M D, Wöll C and Chiang S 1989 Phys. Rev. Lett. 62 171 [110] Grand J Y, Kunstmann T, Hoffmann D, Haas A, Dietsche M, Seifritz J and Möller R 1996 Surf. Sci. 366 403 [111] Stöhr M, Wagner T, Gabriel M, Weyers B and Möller R 2001 Adv. Funct. Mater. 11 175 [112] Ludwig C, Stohmaier R, Petersen J, Gompf B and Eisenmenger W 1994 J. Vac. Sci. Technol. B 12 1963 [113] Qiu X H, Nazin G V and Ho W 2004 Phys. Rev. Lett. 92 206102 [114] Campbell C T 1997 Surf. Sci. Rep. 27 1 [115] Henry C R 1998 Surf. Sci. Rep. 31 231 [116] Büumer M and Freund H J 1999 Prog. Surf. Sci. 61 127 [117] Chambers S A 2000 Surf. Sci. Rep. 39 105 [118] Santra A K and Goodman D W 2002 J. Phys.: Condens. Matter 15 R31 [119] Wolf S A, Awschalom D D, Buhrman R A, Daughton J M, Molnár S, Roukes M L, Chtchelkanova A Y and Treger D M 2001 Science 294 1488 [120] Ertl G and Freund H J 1999 Phys. Today 52 32 [121] Howard J B and Rees D C 1996 Chem. Rev. 96 2965 [122] Burgess B K and Lowe D J 1996 Chem. Rev. 96 2983 [123] Hinnemann B and Nørskov J K 2004 J. Am. Chem. Soc. 126 3920

[53] Li M, Wang C Z, Evans J W, Hupalo M, Tringides M C and Ho K M 2009 Phys. Rev. B 79 113404 [54] Man K L, Tringides M C, Loy M M T and Altman M S 2013 Phys. Rev. Lett. 110 036104 [55] Kim J, Zhang Ch, Kim J, Gao H, Chou M Y and Shih Ch K 2013 Phys. Rev. B 87 245432 [56] Chen J, Hupalo M, Ji M, Wang C Z, Ho K M and Tringides M C 2008 Phys. Rev. B 77 233302 [57] Evans D A, Alonso M, Cimino R and Horn K 1993 Phys. Rev. Lett. 70 3483 [58] Luh D A, Miller T, Paggel J J, Chou M Y and Chiang T C 2001 Science 292 1131 ¨ [59] Unal B, Belianinov A, Thiel P A and Tringides M C 2010 Phys. Rev. B 81 085411 [60] Wu B and Zhang Z 2008 Phys. Rev. B 77 035410 [61] Hong H, Wei C M, Chou M Y, Wu Z, Basile L, Chen H, Holt M and Chiang T C 2003 Phys. Rev. Lett. 90 076104 [62] Wei C M and Chou M Y 2002 Phys. Rev. B 66 233408 [63] Zhang Z, Niu Q and Shih C K 1998 Phys. Rev. Lett. 80 5381 [64] Jia Y, Wu B, Weitering H H and Zhang Z 2006 Phys. Rev. B 74 035433 [65] Materzanini G, Saalfrank P and Lindan P J D 2001 Phys. Rev. B 63 235405 [66] Jaklevic R C, Lambe J, Mikkor M and Vassell W C 1971 Phys. Rev. Lett. 26 88 [67] Jaklevic R C and Lambe J 1975 Phys. Rev. B 12 4146 [68] Jonker B T, Bartelt N C and Park R L 1983 Surf. Sci. 127 183 [69] Jonker B T and Park R L 1984 Solid State Commun. 51 871 [70] Iwasaki H, Jonker B T and Park R L 1985 Phys. Rev. B 38 5272 [71] Sheng L Y, Zhang W, Guo J T, Wang Z S, Ovcharenko V E, Zhou L Z and Ye H Q 2009 Intermetallics 17 572 [72] Stoloff N S, Liu C T and Deevi S C 2000 Intermetallics 8 1313 [73] Morsi K 2001 Mater. Sci. Eng. A 299 1 [74] Ozdemir O, Zeytin S and Bindal C 2012 Tribol. Int. 53 22 [75] Scheppe F, Sahm P R, Hermann W, Paul U and Preuhs J 2002 Mater. Sci. Eng. A 329–331 596 [76] Yeh C L and Sung W Y 2004 J. Alloys Compounds 384 181 [77] Davis J R 2000 Specialty Handbook: Nickel, Cobalt, and Their Alloys (Materials Part, OH: ASM International) p 92 [78] Sondericker D, Jona F and Marcus P M 1986 Phys. Rev. B 33 900 [79] Sondericker D, Jona F and Marcus P M 1986 Phys. Rev. B 34 6775 [80] Bikonda O, Castro G R, Torrelles X, Wendler F and Moritz W 2005 Phys. Rev. B 72 195430 [81] Jurczyszyn L, Krupski A, Degen S, Pieczyrak B, Kralj M, Becker C and Wandelt K 2007 Phys. Rev. B 76 04510101 [82] Kurnosikov O, Jurczyszyn L, Pieczyrak B and Krupski A 2008 Surf. Sci. 602 2994 [83] Min B I, Freeman A J and Jansen H J F 1988 Phys. Rev. B 37 6757 [84] Lekka Ch E, Papageorgiou D G and Evangelakis G A 2003 Surf. Sci. 541 182 [85] Han Y, Unal B and Evans J W 2012 Phys. Rev. Lett. 108 216101 [86] Ruban A V 2002 Phys. Rev. B 65 174201 [87] Gopal P and Srinivasan S G 2012 Phys. Rev. B 86 014112 [88] Saadi S, Hinnemann B, Appel Ch C, Helveg S, Abild-Pedersen F and Norskov J K 2011 Surf. Sci. 605 582 [89] Fatmi M, Ghebouli M A, Ghebouli B, Chihi T, Boucetta S and Heiba Z K 2011 Rom. J. Phys. 56 935 24


A Krupski

[124] Sljivancanin Z and Pasquarello A 2005 Phys. Rev. B 71 081403 [125] Plumer M L and Weller J 2001 The Physics of Ultra-High-Density Magnetic Recording (Berlin: Springer) [126] Bader S D 2002 Surf. Sci. 500 172 [127] Sljivancanin Z and Pasquarello A 2003 Phys. Rev. Lett. 90 247202 [128] Billas I M L, Ch telain A and Heer W A 1994 Science 265 1682 [129] Meyer E, Hug H J and Bennewitz R 2004 Scanning Probe Microscopy. The Lab on a Tip (Berlin: Springer) [130] Besenbacher F 1996 Rep. Prog. Phys. 59 1737 [131] Hofer W A, Foster A S and Shluger A L 2003 Rev. Mod. Phys. 75 1287 [132] Binning G and Rohrer H 1999 Rev. Mod. Phys. 71 324 [133] Binning G and Rohrer H 1982 Helv. Phys. Acta 55 726 [134] Binning G, Rohrer H, Gerber Ch and Weibel E 1983 Phys. Rev. Lett. 50 120 [135] Tersoff J and Hamann D R 1983 Phys. Rev. Lett. 50 1998 [136] Tersoff J and Hamann D R 1985 Phys. Rev. B 31 805 [137] Van Hove M A, Weinberg W H and Ch M 1986 Low-Energy Electron Diffraction Experiment, Theory and Surface Structure Determination (Springer Series in Surface Sciences vol 6) [138] Scheithauer U, Meyer G and Henzler M 1986 Surf. Sci. 178 441 [139] Henzler M 1985 Surf. Sci. 152/153 963 [140] Hofmann S 2013 Auger- and X-ray Photoelectron Spectroscopy in Materials Science: A User-Oriented Guide (Berlin: Springer) [141] Sholl D S and Steckel J A 2009 Density Functional Theory. A Practical Introduction (Hoboken, NJ: Wiley) [142] Kresse G and Hafner J 1993 Phys. Rev. B 47 558 [143] Kresse G and Furthmüller J 1996 Comput. Mater. Sci. 6 15 [144] Kresse G and Furthmüller J 1996 Phys. Rev. B 54 11169 [145] Kresse G and Joubert D 1999 Phys. Rev. B 59 1758 [146] Horcas I, Fernández R, Gómez-Rodr´ıguez J M, Colchero J, Gómez-Herrero J and Baro A M 2007 Rev. Sci. Instrum. 78 013705 [147] Kittel Ch 1996 Introduction to Solid State Physics 7th edn (New York: Wiley) [148] Martienssen W and Warlimont H (ed) 2005 Springer Handbook of Condensed Matter and Materials Data. Part 2. The Elements (Berlin: Springer) [149] Kohler B, Ruggerone P and Scheffler M 1997 Phys. Rev. B 56 13503 [150] Methfessel M, Hennig D and Scheffler M 1992 Phys. Rev. B 46 4816 [151] Che J G, Chan C T, Jian W-E and Leung T C 1998 Phys. Rev. B 57 1875 [152] Kohler B, Ruggerone P, Wilke S and Scheffler M 1995 Phys. Rev. Lett. 74 1387 [153] Arnold M, Sologub S, Hupfauer G, Bayer P, Frie W, Hammer L and Heinz K 1997 Surf. Rev. Lett. 4 1291 [154] Sondericker D, Jona F and Marcus P M 1986 Phys. Rev. B 34 6770 [155] Wang C and Wang Ch 2008 Surf. Sci. 602 2604 [156] Jurczyszyn L, Rosenhahn A, Schneider J, Becker C and Wandelt K 2003 Phys. Rev. B 68 115425 [157] Krupski A, Degen S, Kralj M, Becker C and Wandelt K 2004 unpublished results

[158] Schintke S, Messerli S, Pivetta M, Patthey F, Libioulle L, Stengel M, Vita A and Schneider W D 2001 Phys. Rev. Lett. 87 276801 [159] Rosenhahn A, Schneider J, Kandler J, Becker C and Wandelt K 1999 Surf. Sci. 433–435 705 [160] Gallon T E 1969 Surf. Sci. 17 486 [161] Jackson D C, Gallon T E and Chambers A 1973 Surf. Sci. 36 381 [162] Seah M P 1972 Surf. Sci. 32 703 [163] Biberian J B and Somorjai G A 1979 Appl. Surf. Sci. 2 352 [164] Ossicini S, Memeo R and Ciccacci F 1985 J. Vac. Sci. Technol. A 3 387 [165] Altfeder I B, Matveev K A and Chen D M 1997 Phys. Rev. Lett. 78 2815 [166] Zhou Y G, Zu X T, Nie J L and Xiao H Y 2008 Chem. Phys. 351 19 [167] Sancho M P L 1984 Solid State Commun. 50 629 [168] Wang Y, Wang W, Fan K N and Deng J 2001 Surf. Sci. 490 125 [169] Mae K 2001 Surf. Sci. 482–485 860 [170] Argile C and Rhead G E 1989 Surf. Sci. Rep. 10 277 [171] Vitos L, Ruban A V, Skriver H L and Kollar J 1998 Surf. Sci. 411 186 [172] Tyson W R and Miller W A 1977 Surf. Sci. 62 267 [173] Boer F R, Boom R, Mattens W C M, Miedma A R and Niessen A K 1988 Cohesion in Metals (Amsterdam: North-Holland) [174] Baskes M I 1992 Phys. Rev. B 46 2727 [175] Sellers M S, Schultz A J, Basaran C and Kofke D A 2010 Appl. Surf. Sci. 256 4402 [176] Jiang Q, Lu H M and Zhao M 2004 J. Phys.: Condens. Matter 16 521 [177] Chen S P, Srolovitz D J and Voter A F 1989 J. Mater. Res. 4 62 [178] Kuznetsov V M, Kadyrov R I and Rudenskii G E 1998 J. Mater. Sci. Technol. 14 320 [179] Kim Y S, Lee S M and Song J Y 2010 Appl. Surf. Sci. 256 3603 [180] Mehl M J and Papaconstantopoulos D A 1996 Phys. Rev. B 54 4519 [181] Malzbender J, Przybylski M, Giergiel J and Kirschner J 1998 Surf. Sci. 414 187 [182] Gruzza B and Gillet E 1980 Thin Solid Films 68 345 [183] Pavlovska A, Paunov M and Bauer E 1985 Thin Solid Films 126 129 [184] Payne S H, Kreuzer H J, Pavlovska A and Bauer E 1996 Surf. Sci. 345 L1 [185] Liu S, Bonig L and Metiu H 1997 Surf. Sci. Lett. 392 L56 [186] Yoon B, Akulin V M, Cahuzac Ph, Carlier F, Frutos M, Masson A, Mory C, Colliex C C and Brechignac C 1999 Surf. Sci. 443 76 [187] Bales G S and Chrzan D C 1994 Phys. Rev. B 50 6057 [188] Tomellini M and Fanfoni M 1996 Surf. Sci. 349 L191 [189] Perez M, Dumont M and Acevedo-Reyes A 2008 Acta Mater. 56 2119 [190] Zuo J K, Wendelken J F, Durr H and Liu C L 1994 Phys. Rev. Lett. 72 3064 [191] Chen L H, Chen C Y and Lee Y L 1999 Surf. Sci. 429 150 [192] Bartelt M C and Evans J W 1993 Surf. Sci. 298 421 [193] Park C, Bauer E and Popa H 1985 Surf. Sci. 154 371 [194] Green A K, Prigge S and Bauer E 1978 Thin Solid Films 52 163 25


A Krupski

[195] Biberian J 1978 Surf. Sci. 74 437 [196] Frank F C and Van der Merwe J H 1949 Proc. R. Soc. A 198 216 [197] Van der Merwe J H 1963 Appl. Phys. 34 123 [198] Matthews J W and Blakeslee A E 1974 J. Cryst. Growth 27 118 [199] Matthews J W 1975 J. Vac. Sci. Technol. 12 126 [200] Brune H 1998 Surf. Sci. Rep. 31 121 [201] Krupski A 2006 unpublished results [202] Ayuela A, Ogando E and Zabala N 2007 Phys. Rev. B 75 153403 ¨ [203] Ozer M M, Jia Y, Zhang Z, Thompson J R and Weitering H H 2007 Science 316 1594 [204] Jia Y, Wang S Y, Chen W G, sun Q, Weitering H H and Zhang Z 2010 Phys. Rev. B 81 245425 [205] Bardi U, Atrei A and Rovida G 1990 Surf. Sci. 239 L511 [206] Bardi U, Atrei A and Rovida G 1992 Surf. Sci. 268 87 [207] Rosenhahn A, Schneider J, Becker C and Wandelt K 2000 J. Vac. Sci. Technol. A 18 1923 [208] Gritschneder S, Becker C, Wandelt K and Reichling M 2007 J. Am. Chem. Soc. 129 4925 [209] Hamm G, Barth C, Becker C, Wandelt K and Henry C R 2006 Phys. Rev. Lett. 97 126106 [210] Andersson S, Brühwiler P, Sandell A, Frank M, Libuda J, Giertz A, Brena B, Maxwell A, Bäumer M and Freund H 1999 Surf. Sci. 442 L964 [211] Maroutian T, Degen S, Becker C, Wandelt K and Berndt R 2003 Phys. Rev. B 68 155414 [212] Wiltner A, Rosenhahn A, Schneider J, Becker C, Pervan P, Milun M, Kralj M and Wandelt K 2001 Thin Solid Films 400 71 [213] Becker C, Bergmann K, Rosenhahn A, Schneider J and Wandelt K 2000 Surf. Sci. 486 443

[214] Rosenhahn A 2000 PhD Thesis University of Bonn [215] Hlawacek G and Teichert Ch 2013 J. Phys.: Condens. Matter 25 1432002 [216] Li D, Peng Z, Deng L, Shen Y and Zhou Y 2005 Vib. Spec. 39 191 [217] Hoshino A, Takenake Y and Miyaji H 2003 Acta Crystallogr. B 59 393 [218] Brown C J 1968 J. Chem. Soc. A 2488 [219] Liao M S and Scheiner S 2001 J. Chem. Phys. 114 9780 [220] Sharp J H and Abkowitz M 1972 J. Phys. Chem. 77 477 [221] Sautet P and Joachim C 1992 Surf. Sci. 271 387 [222] Krupski A, Moors M, Degen S, Kralj M, Becker C and Wandelt K 2005 unpublished results [223] Persaud R and Madey T E 1997 The Chemical Physics of Solid Surfaces and Heterogeneous Catalysts ed D A King and D P Woodruff (Amsterdam: Elsevier) p 407 [224] Henry C R 1997 Cryst. Res. Technol. 33 1119 [225] Cambell C T Surf. Sci. Rep. 27 1 [226] Baumer M and Freund H J 1999 Prog. Surf. Sci. 61 127 [227] Cosandey F and Madey T E 2001 Surf. Rev. Lett. 8 73 [228] Gavioli L, Cavaliere E, Agnoli S, Barcaro G, Fortunelli A and Granozzi G 2011 Prog. Surf. Sci. 86 59 [229] Adamson A W 1990 Physical Chemistry of Surfaces (New York: Wiley) [230] Chatain D, Coudurier L and Eustathopoulos N 1988 Rev. Phys. Appl. 23 1055 [231] Chatain D, Rivollet I and Eustathopoulos N 1986 J. Chim. Phys. 83 561 [232] Rusponi S, Cren T, Weiss N, Epple M, Buluschek P, Claude L and Brune H 2003 Nature Mater. 2 546 [233] Castro M 1997 Int. J. Quantum Chem. 64 223

26

Growth of Epitaxial Oxide Thin Films on Graphene.

Tailored surfaces of perovskite oxide substrates for conducted growth of thin films.

Growth and composition of nanostructured and nanoporous cerium oxide thin films on a graphite foil.

Heteroepitaxy of Cerium Oxide Thin Films on Cu(111).

Epitaxial growth of highly-crystalline spinel ferrite thin films on perovskite substrates for all-oxide devices.

Influence of texture coefficient on surface morphology and sensing properties of W-doped nanocrystalline tin oxide thin films.

Formation of nanostructured metallic glass thin films upon sputtering.

Metallic glass thin films for potential biomedical applications.

Plasmonic sensing using metallic nano-sculptured thin films.

Cell growth on different types of ultrananocrystalline diamond thin films.

Lateral solid-phase epitaxy of oxide thin films on glass substrate seeded with oxide nanosheets.

Formation of carbonated hydroxyapatite films on metallic surfaces using dihexadecyl phosphate-LB film as template.

Low Temperature Reactive Sputtering of Thin Aluminum Nitride Films on Metallic Nanocomposites.

Atomically defined templates for epitaxial growth of complex oxide thin films.

KCl ultra-thin films with polar and non-polar surfaces grown on Si(111)7 × 7.

Ordered fragmentation of oxide thin films at submicron scale.

Transparent p-type epitaxial thin films of nickel oxide.

Capturing by self-assembled block copolymer thin films: transfer printing of metal nanostructures on textured surfaces.

Scanning tunnelling spectroscopy of superconductivity on surfaces of LiTi2O4(111) thin films.

The structures of hexadecylamine films adsorbed on iron-oxide surfaces in dodecane and hexadecane.

Room temperature self-assembled growth of vertically aligned columnar copper oxide nanocomposite thin films on unmatched substrates.

Vertical-interface-manipulated conduction behavior in nanocomposite oxide thin films.

Growth, structure and stability of sputter-deposited MoS2 thin films.

Multiferroic iron oxide thin films at room temperature.