PHYSICAL REVIEW B VOLUME 57, NUMBER 12 15 MARCH 1998-II Growth and structure of ultrathin FeO films on Pt 111... studied by STM and LEED M. Ritter, W. Ranke, and W. Weiss* Fritz-Haber-Institut der Max-Planck-Gesellschaft, Faradayweg 4-6, 14195 Berlin, Germany Received 28 July 1997 The growth of iron-oxide films on Pt 111 prepared by iron deposition and subsequent oxidation was studied by scanning tunneling microscopy STM and high-resolution low-energy electron diffraction LEED . Despite a 10% lattice mismatch to the substrate, an epitaxial growth of well-ordered films is observed. The oxide starts to grow layer by layer in a 111 orientation of the metastable cubic FeO structure up to a thickness of about 2.2 monolayers ML . The completion of the second and third FeO layer depends on the precise oxidation temperature, and at coverages of approximately 2 ML three-dimensional Fe3O4(111) islands start to grow. The FeO 111 layers consist of hexagonal close-packed iron-oxygen bilayers that are laterally expanded when compared to bulk FeO and slightly rotated against the platinum substrate. They all exhibit oxygen-terminated unreconstructed (1 1) surface structures. With increasing coverage several structural film changes occur, and four coincidence structures with slightly different lateral lattice constants and rotation misfit angles against the platinum substrate are formed. In the submonolayer regime an FeO 111 bilayer with a lattice constant of 3.11 Å and rotated by 1.3° against the platinum substrate is observed. Upon completion of the first layer the film gets compressed leading to a lattice constant of 3.09 Å and a rotation misfit angle of 0.6°. Between 1.5 and 2 ML a coincidence structure rotated by 30° against the platinum substrate forms, and at 2 ML a nonrotated coincidence structure with a lattice constant of 3.15 Å evolves. All these coincidence structures exhibit large periodicities between approximately 22 and 38 Å that are visible in the STM images up to the third FeO layer surface. The LEED patterns exhibit characteristic multiple scattering satellite spots. The different coincidence structures reflect lowest-total-energy arrangements, balancing the contributions of substrate-overlayer interface energies and elastic energies within the strained oxide overlayer for each coverage. S0163-1829 98 02311-X I. INTRODUCTION epitaxy.14,15 They found that the selective growth of these oxide phases critically depends on the growth rate deter- The preparation of thin metal-oxide films is becoming an mined by the iron and oxygen fluxes and the substrate tem- important technique in material and surface science. Single perature. High growth rates and low partial pressures are crystalline films allow us to study ordered oxide surfaces required for Fe3O4 while low growth rates and high oxygen without using single-crystal samples, which sometimes are pressures are needed for -Fe2O3. The different iron-oxide not available or may cause electrostatic charging problems phases can transform into each other depending on the am- when applying electron spectroscopy techniques or scanning bient conditions such as temperature and oxygen partial pres- tunneling microscopy. The properties of clean metal-oxide sure. Their stability ranges in thermodynamic equilibrium surfaces and the adsorption of gases thereon is of great in- with the oxygen gas phase are given by the iron-oxygen terest in catalysis research, since not much is known about phase diagram.16 The substrate temperatures and oxygen par- the atomic-scale surface chemistry on metal oxide catalysts tial pressures used in molecular-beam epitaxy growth of yet.1,2 Magnetic oxidic multilayers of Fe3O4 combined with single phased iron oxide films mostly differ from the equi- other oxides are used to study magnetic coupling across non- librium stability ranges of these phases, indicating that the magnetic barriers and between antiferromagnetic layers.3 kinetics of the iron-oxide formation is determining the oxide These properties are also important for the development of phase that forms during epitaxial growth. magnetic-field sensors and of high-density magnetic record- The growth mode of iron-oxide films on metal-oxide sub- ing media.4 Iron oxide is also utilized as catalyst material for strates depends on the lattice mismatch between the oxygen a number of different important chemical processes,5 in par- sublattices and therefore on the particular metal-oxide sub- ticular, the technical dehydrogenation of ethylbenzene to strate used and its orientation. Fe3O4 grows layer by layer styrene.6,7 onto 100 -oriented metal-oxide substrates with small lattice Well-ordered oxide films can be prepared by oxidizing the mismatches, as was observed in the molecular-beam experi- surface region of the corresponding metal single crystals, ments by Kim, Gao, and Chambers for Fe3O4 on which was done, for example, with several transition metals MgO 100 14 by Lind et al. for Fe3O4 and NiO/Fe3O4 super- by Freund and co-workers.8 Heteroepitaxial growth can be lattices with sharp interfaces on MgO 100 substrates,17 as achieved by repeatedly depositing the metal and oxidizing it well as by Gaines et al., who grew smooth Fe3O4(100) films afterwards,9­11 by molecular-beam epitaxy12 or by reactive about 500 Å thick onto MgO 100 and studied them with vapor deposition.13 Kim, Gao, and Chambers were able to STM afterwards.18 Nonstoichiometric Fe3O4 films have been grow single crystalline and pure phased Fe3O4 and -Fe2O3 produced by evaporating iron in a controlled NO2 flux,19 and films 100­1000 Å thick onto MgO and Al2O3 substrates with a columnar growth of single crystalline Fe3O4(111) particles different orientations by plasma-assisted molecular-beam on -Al2O3(0001) was observed.20 Kim, Gao, and Cham- 0163-1829/98/57 12 /7240 12 /$15.00 57 7240 © 1998 The American Physical Society 57 GROWTH AND STRUCTURE OF ULTRATHIN FeO FILMS . . . 7241 bers also observed an initial Fe This paper is organized as follows: In Sec. II the experi- 3O4(111) island growth and a subsequent island coalescence on Al mental procedures are explained, in Sec. III A the FeO coin- 2O3(0001) substrates, as well as faceted surfaces on Fe cidence structures that we observe on Pt 111 are explained 3O4(110) films grown onto MgO 110 .15 together with the LEED patterns they form. The growth of Not much is known about the details of epitaxial metal- the first and second FeO layers is presented in Secs. III B and oxide growth, especially in the inital growth stage. In het- III C, the growth of the third FeO layer together with the eroepitaxy the growth mode on lattice mismatched substrates initial growth of Fe3O4 islands is presented in Sec. III D. The is always determined by substrate-interface energies, over- LEED beam intensity evolution during the iron-oxide film layer surface energies, and elastic energy in the strained growth is presented in Sec. III E, and the experimental re- overlayer that can be reduced by dislocation defects.21 Mc- sults are discussed in Sec. IV. Kee and co-workers have demonstrated the crucial role of interfacial energy minimization at the first atomic layers for II. EXPERIMENT the heteroepitaxial growth mode of metal oxides.22 They found that ion size and electrostatics at the interface deter- The experiments were performed in an UHV chamber de- mine the growth mode for BaTiO3(100) on MgO 100 . An scribed in detail in reference.34 It is equipped with a com- ordered FeO 111 monolayer film was grown onto Pt 111 mercial STM head Burleigh Instruments , a backview and Pt 100 surfaces for the first time by Vurens et al.23 LEED optics and a cylindrical mirror analyzer Auger spec- Later this monolayer structure was further characterized with trometer Omicron . The base pressure of the system is 5 low-energy electron diffraction LEED Refs. 9 and 24 and 10 11 mbar. All STM measurements were performed in scanning tunneling microscopy STM .25 Galloway and co- the constant current mode using tunneling currents between workers proposed a model for this monolayer film, which 0.2 and 1.0 nA and bias voltages between 0.3 and 1.3 V. consists of an FeO 111 bilayer with an expanded lateral Tungsten tips were sharpened ex situ by electrochemical lattice constant if compared to bulk FeO and rotated by 0.6° etching in NaOH. The high-resolution LEED measurements against the platinum substrate.25 Photoelectron diffraction were performed in a seperate chamber equipped with a measurements reveiled an oxygen-terminated surface for this Henzler-type spot profile analysis LEED system film.26 This was substantiated later by STM image calcula- SPA-LEED 35 and a double pass cylindrical mirror analyzer tions applying electron-scattering quantum-chemistry for photoelectron spectroscopy. This chamber also has a base theory.27 Galloway and co-workers also performed some pressure of 5 10 11 mbar. STM measurements on iron-oxide films several layers thick, The sample preparation was performed in both chambers where they observed coexisting -Fe2O3 and Fe3O4 in the same way. The Pt 111 surface is cleaned by repeated islands.28 cycles of argon sputtering and annealing to 1300 K until it Here we present a detailed study of the initial growth exhibited a sharp (1 1) LEED pattern and no AES con- stage of iron-oxide films on Pt 111 combining LEED and tamination signals anymore. Iron is deposited onto this plati- STM. The films were prepared by repeated deposition of iron num surface at room temperature by thermal evaporation and subsequent oxidation. The role of the interface structure from an iron wire wrapped around a resistively heated tung- and of the iron-oxide phase thermodynamics for the epitaxial sten wire. After the deposition the iron is oxidized for 2 min growth is adressed. STM allows characterization of the at temperatures between 870 and 1000 K in 10 6 mbar oxy- atomic surface structures and of the film morphologies up to gen partial pressure. This produces a well-ordered first oxide the m range. With high-resolution LEED we characterize layer as discussed in the following sections. To further in- the whole sample surface and can determine average lattice crease the film thickness this procedure is repeated. Up to 1 constants with high precision. This makes a detailed investi- ML coverage the oxidation temperature was always T gation on the epitaxial film growth possible. 1000 K, above 1 ML coverage oxidation temperatures be- Several STM studies were performed on surfaces of tween T 870 and 920 K were applied. In the STM chamber mostly natural iron-oxide single crystals prepared by ion the FeO film thickness FeO was determined by the STM bombardment and annealing. Tarrach et al. studied measurements. With increasing FeO film coverage FeO dif- Fe3O4(001) surfaces,29 Jansen, Brabers, and van Kempen ferent film structures exhibiting characteristic LEED patterns Fe3O4(110) surfaces.30 Lennie et al. observed two different are formed. In the SPA-LEED chamber the FeO coverage terminations on Fe3O4(111) surfaces seperated by steps and FeO was determined with the help of these LEED patterns exposing iron and oxygen atoms in the topmost layer.31 They for coverages above 1 ML. In the submonolayer regime the also observed different coexisting oxide phases arranged in coverage was controlled by valence-band photoemission of ordered patches on the surface that they call biphase struc- adsorbed ethylbenzene molecules. Since at room temperature tures, namely, FeO 111 and Fe2O3(0001) phases on ethylbenzene only adsorbs on Pt 111 and not on the FeO- -Fe2O3(0001) crystals32 as well as FeO 111 and covered parts of the surface,36 the adsorbate signal could be Fe3O4(111) phases on Fe3O4(111) crystals.33 All these stud- used to titrate the submonolayer coverage of the oxide over- ies show that iron-oxide surface structures critically depend layer and to determine the iron evaporation rate. From this on the preparation conditions and that stoichiometric evaporation rate the effective overlayer thickness EFF cor- -Fe2O3 surfaces can not be prepared under vacuum condi- responding to the total amount of iron deposited onto the tions. A comparison between iron oxide surface structures surface is estimated. EFF deviates from the FeO film thick- formed on single-crystal samples and on epitaxially grown ness FeO for coverages above 1.5 ML. One reason for this films may provide a deeper insight into the formation and deviation is the growth of Fe3O4 islands starting at FeO cov- energetics of metal-oxide surface structures. erages of approximately 2 ML. A second reason is a possible 7242 M. RITTER, W. RANKE, AND W. WEISS 57 FIG. 2. Model of an FeO 111 bilayer on Pt 111 . The overlayer has a lattice constant of 3.11 Å and is rotated by 1.3° against the 110 direction, forming a (8 2 1 10) coincidence structure with the coincidence site 1 and the large unit cell indicated. Sites 2­4 FIG. 1. 55 55 Å2 STM image of a submonolayer FeO film indicate coincidence sites of structures 2 ­ 4 as discussed in the grown onto Pt 111 . An atomic periodicity of 3.1 Å is modulated by text. a large 25 Å periodicity creating a moire´ superstructure. The direc- tion of this superstructure indicated by the marked atoms is rotated site on the platinum surface after going eight platinum lattice by about 11° against the small FeO 111 -(1 1) unit cell that is spacings along the 110 direction and two platinum lat- also indicated. UT 0.9 V, IT 0.3 nA. tice spacing along the 101 direction. This site is labeled 1 in Fig. 2 and was chosen arbitrarily as a top site on a diffusion of iron into the platinum substrate. In the following platinum surface atom. The coincidence overlayer structure the oxide overlayer coverage is always given in terms of the has the large unit cell indicated in Fig. 2, which is 25.4 Å in real FeO coverage FeO if not stated otherwise. size and rotated by 10.9° with respect to the (1 1) unit cell of the Pt 111 surface and by 12.2° with respect III. RESULTS to the FeO(111)-(1 1) unit cell. It also can be described by ( 84 84)R10.9° or by (8 2 ) superstructure cells using A. FeO coincidence structures 1 10 the Wood or matrix notation, respectively. The angle Figure 1 displays a constant current 55 55 Å2 atomic 12.2° and the lattice constant of 3.10 Å obtained from this resolution STM image of a FeO film less than 1 ML thick, model agree reasonably with the moire´ angle of 11° observed grown on Pt 111 . It exhibits a hexagonal surface structure by STM in Fig. 1 and the 3.11 Å lattice constant obtained with an atomic periodicity of 3.11 Å as determined precisely from the LEED measurement in Fig. 4 b . In Table I these from the high-resolution LEED intensity scan of such a film experimentally observed lattice constants and rotation misfit shown in Fig. 4 b . This atomic periodicity is modulated by angles are listed together with the theoretical values expected a larger periodicity of about 25 Å, which creates the moire´ from coincidence structure 1 . superstructure in the STM image. The large hexagonal unit The iron-oxygen bilayer model in Fig. 2 was proposed cell of this moire´ structure can be defined by the brightest previously by Galloway, Benitez, and Salmeron,25 who ob- atomic features in the STM image and is rotated by about served a very similar constant height STM image on an FeO 11° with respect to the small (1 1) surface unit cell on the monolayer grown onto Pt 111 . They observed an atomic oxide film. This is indicated by the marked atoms with equal periodicity of 3.09 Å and a 26 Å moire´ superstructure rotated brightness in Fig. 1, which do not line up with the atom rows by 5° 1°. This was explained by an iron-oxygen bilayer on the FeO 111 surface. with a 3.09 Å lattice constant and rotated by 0.6° against the We propose the model shown in Fig. 2 for this submono- platinum substrate, so that the coincidence site 2 in Fig. 2 is layer film. It consists of a laterally expanded oxygen- reached after going nine platinum lattice spacings along the terminated FeO 111 bilayer on top of the Pt 111 surface, 110 direction and one platinum lattice spacing along the where the iron atoms are seperated by 3.10 Å and form rows 101 direction. Galloway, Benitez, and Salmeron ob- that are rotated by 1.3° with respect to the underlying plati- served this structure in an atomic resolution STM image.25 num atom rows along the 110 and 1­10 directions. For We observe this compressed structure 2 after completion of comparison the interatomic distance within the 111 planes the first monolayer as described in Sec. III B. Galloway and of the cubic sodium chloride FeO bulk structure is 3.04 Å. In co-workers applied electron-scattering quantum-chemistry this rigid model the rotational mismatch of 1.3° between theory to calculate the contrast in this STM image.27 They the FeO bilayer and the platinum substrate and the lateral showed that the image contrast is not directly related to the FeO lattice constant of 3.10 Å lead to an iron coincidence surface topography and that for Pt tips the maxima occur 57 GROWTH AND STRUCTURE OF ULTRATHIN FeO FILMS . . . 7243 TABLE I. FeO/Pt 111 unit cells and their orientations from SPA-LEED measurements. The structures are numbered according to their appearance with increasing FeO coverage. EFF is the coverage according to the total amount of evaporated Fe and FeO is the real FeO coverage as deduced from the STM measure- ments. The difference exists in form of Fe3O4 islands. The underlined numbers are the experimentally observed values, the values printed italic are those expected from the corresponding models. The FeO rows are rotated by with respect to the Pt atomic rows and the superstructure unit cell vectors are rotated by with respect to the substrate. is the ``moire´ angle'' between the overlayer atomic rows and the connecting line of the moire´ maxima. No. EFF ML FeO ML LEED structure aFeO Å 1 1 1 8 2 ( 84 84)R10.9° 3.11 1­1.5° 2 10 3.102 1.3° 10.9° 12.2° 2 1 1 9 1 ( 91 91)R5.2° 3.09 small 1 10 3.093 0.6° 5.2° 5.8° 3 2 1.8 8 8 (8) 8))R30° 38.0a 30° 8 16 38.38a 30° 4 2.5 2 8 0 8 8 3.15 small 0° 0° 0 8 3.166 0° aThis is the length of the superstructure unit cell vector and not aFeO which is not known since we have no model for this structure. over oxygen positions. Photoelectron diffraction measure- cesses between the platinum substrate and the oxide over- ments also reveiled an oxygen-terminated surface for this layer as well as in terms of diffraction at the large superstruc- FeO bilayer.26 Based on these findings we interpret the ture unit cell of a buckled overlayer into fractional order atomic resolution STM images we observe on FeO 111 spots. Both descriptions lead to the same satellite spot posi- films also as oxygen-terminated surface structures. tions that are given by linear combinations of platinum sub- All FeO 111 films from submonolayer up to 2.2 ML strate and oxide overlayer surface reciprocal lattice vectors thickness and beyond exhibit similar hexagonal LEED pat- g(hk)Pt g(hk)FeO , where h and k denote the indices of the terns which are consistent with the model in Fig. 2. A sche- integer diffraction spots.37 From the weak satellite spot in- matic representation of these LEED patterns is shown in Fig. tensities, if compared to the substrate and overlayer integer 3. The first-order platinum substrate spots are still visible and spot intensities, it can be concluded that multiple scattering occur at the same positions as on the clean surface crosses . is the dominating mechanism creating the satellite spots. They correspond to the Pt(111)-(1 1) surface unit cell This indicates a small buckling in the oxide overlayer, which with a lattice constant of 2.77 Å. The oxide film also forms a is in line with STM image simulations that revealed the hexagonal LEED pattern that is superimposed to the hexago- atomic corrugations in the STM images to be mainly due to the local electronic surface structure and not to the surface nal platinum LEED pattern. The first-order FeO spots are topography.27 located closer to the specular beam because of the larger In the multiple scattering picture satellite spot 1 near the FeO(111)-(1 1) surface unit cell with a lattice constant of 00 beam in Fig. 3, for example, is created by double scat- about 3.1 Å large dots . The formation of the satellite dif- tering described by the scattering vector sum g(10) fraction spots around the 00 and FeO 10 beams small Pt g ( 10) dots can be discussed in terms of multiple scattering pro- FeO . Spot 2 near the 10 beams is created by double scattering described by g(01)Pt g(1 1)FeO and spot 3 by g( 11)Pt g(01)FeO . Spot 4 is created by double scattering g(0 1)Pt g(11)FeO , spot 5 by g( 10)Pt g(20)FeO and spot 6 by g( 11)Pt g(2 1)FeO . We also observe very weak spots due to triple scattering events, which, however, are hardly visible in the gray-scale intensity plots shown in Fig. 4. For a nonrotated FeO bilayer perfectly aligned to the platinum substrate we expect all diffraction beams to be sharp, neglecting step induced spot broadening at the corre- sponding out-of-phase electron energies.38 If domains with FeO bilayers rotated by different angles coexist on the platinum surface, a characteristic broadening or splitting of some LEED spots independent of the electron energy scat- tering vector component perpendicular to the surface but dependent on the parallel component of the scattering vector FIG. 3. Schematic LEED pattern of FeO 111 films on Pt 111 . is expected. The largest spot broadening is observed for sub- Crosses indicate platinum integer spots, large dots FeO integer monolayer films. This can be seen in the high-resolution spots, and small dots double scattering satellite spots. LEED intensity plot in Fig. 4 b . The gray-scale plots display 7244 M. RITTER, W. RANKE, AND W. WEISS 57 FIG. 4. LEED intensity line scans between the 00 and 10 beams left side and LEED pattern gray-scale plots of the region around the 00 and 10 beams for clean Pt 111 and for epitaxial FeO films on Pt 111 . All curves and patterns are scaled in the same way expanded line scans: 10 . The coverage EFF corresponds to the total amount of deposited Fe, the coverage FeO corresponds to the real FeO coverage as deduced from the STM measurements. Above 1.5 ML coverage both values differ because of the formation of Fe3O4 islands. Fe3O4 spots are marked by arrows. The shift of the FeO 10 spot between 0.4 and 1.2 ML and its splitting at FeO 2.2 ML is emphasized by lines. The dashed lines in the lowest coverage LEED pattern on the right side indicate the 10 scattering vector directions expected for two FeO domains rotated by against the platinum substrate lattice. the 00 and 10 beams with their surrounding double scat- All these spots around the FeO 10 beam are elongated per- tering satellite spots. Only the 00 and the Pt 10 beams are pendicular to the direction connecting them with the Pt 10 round shaped, the FeO 10 beam is elongated perpendicular spot, a consequence of the involved overlayer scattering vec- to the direction connecting the 00 and 10 beams. This is tor directions. due to the coexistence of domains with different rotational We observe four different coincidence structures with in- mismatches between the FeO bilayer and the platinum sub- creasing FeO coverage FeO , which will be presented in the strate as discussed above. The rotation angle can occur in following sections. We numbered them 1­4 according to the both directions and is of the order of 1° leading to domains sequence of their appearance with increasing FeO coverage. rotated by 1°. This leads to a splitting or an elongation They are listed in Table I together with their superstructure of the FeO beams depending on the resolution of the LEED unit cells in matrix and Wood notations, their lateral lattice system. The spot splitting is determined by the angle 2 and constants aFeO obtained from the rigid coincidence models the scattering vector length parallel to the surface as indi- and observed experimentally aFeO is the interatomic dis- cated in the gray-scale plot in Fig. 4 b . If several rotation tance within the iron and oxygen 111 planes . The rotation angles are present a spot elongation perpendicular to the par- misfit angles to the platinum substrate obtained from the allel scattering vector is expected. This is analogous to models and observed experimentally, the misfit angles be- LEED beam broadening on mosaic crystal surfaces, which tween the superstructure cell and the Pt 111 -(1 1) cell increases with increasing scattering vector perpendicular to as well as the moire´ angle between the superstructure cell, the surface.39 and the FeO 111 -(1 1) cell , which is observed in The full width at half-maximum of the 00 beam does not the STM images, are also listed. change considerably with increasing FeO coverage, because no scattering vector component parallel to the surface is in- volved in this spot. The satellite spots around the FeO 10 B. First layer structures beam are elongated according to the length of the involved As discussed in the previous section, for submonolayer overlayer scattering vectors parallel to the surface. Spots 2 coverages we observe the LEED pattern shown in Fig. 4 b and 3 are least elongated because they are created by double and the STM image shown in Fig. 1. From the broadening of scattering involving the shortest overlayer scattering vector the FeO 10 beam we derive rotational misfit angles rang- g(10)FeO , whereas spots 4 and 6 are more elongated as they ing between 0 and about 2°. The lattice constant obtained are created by double scattering with the longer overlayer from the LEED line scan and the moire´ angle between scattering vector g(11)FeO . Spot 5 is most elongated because the small and large unit cells on the oxide overlayer obtained the longest overlayer scattering vector g(20)FeO is involved. from the STM image are listed in Table I and agree well with 57 GROWTH AND STRUCTURE OF ULTRATHIN FeO FILMS . . . 7245 will be explained in the next paragraph. The 3° direction change indicates the existence of superstructures with other rotation misfit angles. In the lower part of the image indi- cated by the number 3 several honeycomb superstructure rows end at locations where small uncovered platinum areas are present. Two honeycomb rows formerly separated by a row in between them move together at these points. They are inclined by small angles with respect to each other. The hon- eycomb row ending points are located along a tilt grain boundary where slightly inclined superstructure rows meet. This tilt grain boundary in an epitaxial monolayer film is the two-dimensional analogy to the well-known tilt grain bound- aries in three-dimensional crystals.40 The different rotational misfits that we observe at the domain and grain boundaries also contribute in small part to the FeO LEED beam broad- ening that we observe on these films. At coverages above 1 ML the FeO 10 beam position moves away from the specular beam as indicated in the line scan in Fig. 4 c . This reveals a new lattice constant of 3.09 Å of the FeO layer, slightly smaller than the submonolayer lattice constant of 3.11 Å. This compression is reproducibly observed upon completion of the first FeO monolayer. Al- though we could not obtain an atomic resolution STM image FIG. 5. 1000 1000 Å2 STM image of a 0.9 ML FeO film on 8 2 of this compressed FeO film we propose structure 2 listed Pt 111 . The honeycomb moire´ pattern of the ( 1 10) coincidence in Table I, since this structure was observed by Galloway, superstructure is visible. The defects indicated are explained in the Benitez and Salmeron in an atomic resolution STM image.25 text. UT 0.5 V, IT 0.2 nA. It has a lattice constant of 3.09 Å and is rotated by the values expected from a (8 2 0.6° against the platinum substrate. The coincidence site 1 10) coincidence structure. Model structure 1 involves rotation angles of 1.3°, labeled 2 in Fig. 2 is reached after going nine platinum lat- which according to the resolution of our LEED system tice spacings along the 110 direction and one platinum 9 1 would create a splitting of the FeO 10 beam or at least an lattice spacing along the 101 direction. The ( 1 10) su- intensity profile with a central minimum indicating a spot perstructure cell is rotated by 5.2° against the Pt 111 - splitting. Instead, we always observe intensity profiles with a (1 1) unit cell. Against the FeO 111 -(1 1) unit cell it is central maximum. Because the atom rows in Fig. 1 do not rotated by the moire´ angle 5.8°. A smaller rotational form straight lines but wiggled lines, additional rotational mismatch of this film compared to the submonolayer struc- mismatches between the first layer FeO 111 -(1 1) unit ture 1 is also evident from the less elongated FeO 10 beam cells and the platinum substrate are created that range be- and its surrounding satellite spots. This can be seen in inten- tween 7°. This explains the central maximum in the sity line scans of the FeO 10 beam along the elongation FeO 10 spots. direction, which are not shown here. The existence of different rotation misfit angles is also evident from the 1000 1000 Å2 images of an 0.9-ML-thick C. Second layer structures FeO film shown in Fig. 5, where several structural defects Figure 6 displays 1500 1500 Å2 a and 2000 2000 can be seen. The 25 Å coincidence superstructure creates a Å2 b STM images of FeO films 1.2 and 1.6 ML thick, honeycomb moire´ pattern clearly visible in the STM image. respectively. The second FeO layer in Fig. 6 a has grown in Between the FeO-covered regions and the dark bare plati- hexagonally shaped islands with step edges running along num areas we measure a step height of 2 Å, which cannot be the main crystallographic directions on the FeO 111 sur- interpreted as the real topographic height difference because face, namely, the 110 and 1 10 directions. The is- of the different electronic surface structures of Pt 111 and lands are randomly distributed on the surface. The step FeO 111 . At the upper left corner a monoatomic platinum height measured between the first and second layer is 2.5 Å, step seperating two FeO covered terraces is visible, where which corresponds to the distance between consecutive iron- we measure the real platinum step height of 2.3 Å. Two oxygen 111 bilayers in the bulk FeO structure. At 1.2 ML domain boundaries meeting at the upper left can be seen, at coverage only the compressed monolayer structure 2 exists, which lateral shifts between the honeycomb coincidence su- which is deduced from the LEED pattern in Fig. 4 c reveal- perstructure cells occur as indicated for example by arrow 1. ing the 3.09 Å periodicity and the absence of LEED spots Presumably former separated FeO island have grown to- related to any other structures. The honeycomb moire´ pattern gether at these boundaries. Above the domain boundary of the (9 1 1 10) coincidence structure is also visible in Fig. marked by arrow 1 a well ordered superstructure without 6 a . visible defects has formed. At the position marked by arrow On the 1.6-ML-thick FeO film shown in Fig. 6 b the 2 the superstructure cell direction changes by 3° and arrow 4 second layer exhibits hexagonal shaped holes exposing the indicates a 7° direction change of the rows formed by the first FeO layer surface. Again all steps between the first and superstructure cells. The 7° direction change can be ex- second layer are about 2.5 Å high and run along the main plained by the coexistence of structures 1 and 2 , the latter crystallographic directions on the FeO 111 surface. The ex- 7246 M. RITTER, W. RANKE, AND W. WEISS 57 FIG. 6. a 1500 1500 Å2 STM image of an 1.2 ML FeO film; UT 0.5 V, IT 0.1 nA. b 2000 2000 Å2 STM image of an 1.6 ML FeO film; UT 1.0 V, IT 0.2 nA. posed first FeO layer forms the compressed monolayer struc- developed best exhibiting the highest LEED spot intensities. ture 2 as deduced from the 3.09 Å lattice constant observed With further increasing coverage FeO it gets replaced by a in the LEED pattern in Fig. 4 d and from the moire´ super- new FeO coincidence structure. In the LEED pattern of a 2.2 structure observed by STM on the 1-ML-thick regions. On ML thick FeO film in Fig. 4 e no (8) 8))R30° spots the second layer surface of this film a new coincidence struc- are visible anymore. In addition to the FeO 10 beam corre- ture is observed. This is deduced from STM measurements sponding to the 3.09 Å periodicity of structure 2 , a second and from the appearance of additional LEED spots that are FeO 10 beam located closer to the specular beam and cor- not related to structure 2 . The gray scale plot of an 1.7 ML responding to a larger lattice constant of 3.15 Å has evolved, film in Fig. 4 d shows these spots around the FeO 10 beam. which can be seen in the line scan on the left side. This In the corresponding line scan the FeO 10 spot position is agrees with STM observations showing that the triangle unchanged and corresponds to the 3.09 Å lattice constant, the structure 3 completely disappears upon completion of the left shoulder of the FeO 10 beam is due to a fractional order second FeO layer. spot of structure 3 . Figure 7 b shows a 90 90 Å STM The coincidence structure 4 has a lattice constant of 3.15 image of this new structure. Triangles with a side length of Å as deduced from the LEED measurements. Figure 8 shows about 35 Å occur with a periodicity of about 38 Å along an atomic resolution 70 70 Å STM image of this structure directions rotated by 30° to the 110 directions on the measured on an almost completed second layer surface. It Pt 111 surface. Fig. 7 a displays the LEED pattern around exhibits an unreconstructed FeO 111 -(1 1) surface struc- the 00 beam of a 1.8-ML-thick film. In addition to the ture that forms a moire´ superstructure with a periodicity of satellite spots of structure 2 new spots appear at positions about 22 Å. This distance corresponds to seven lattice spac- corresponding to a (8) 8))R30° superstructure, referred ings on the FeO 111 surface and is smaller than the moire´ to the Pt 111 -(1 1) unit cell. In the matrix notation struc- superstructure period observed on structures 1 and 2 . The ture 3 is given by a (8 8 8 16) unit cell as listed in Table I. moire´ superstructure now is not rotated anymore against the Since we obtained no atomic resolution STM images of this small FeO 111 -(1 1) unit cell. We propose a nonrotated structure we do not propose a model for it. In such a model (8 8) coincidence structure on the platinum surface as a the coincidence site labeled 3 in Fig. 2 must be reached after model. The coincidence site 4 in Fig. 2 is reached after going going eight platinum spacings along the 110 and eight platinum lattice spacings along the 110 direction. 101 directions each. This triangle structure 3 always In a rigid model a FeO 111 -(8 8) coincidence structure is forms on the second layer surface at coverages between 1.5 obtained with a lattice constant of 3.16 Å, in good agreement and 2 ML. with the experimentally observed lattice constant of 3.15 Å. On the left-hand side of the image in Fig. 6 b a feature The disappearance of the rotational mismatch is also evident can be seen that looks like a third layer island. But the sur- from the round-shaped diffraction spots in the gray-scale face of this island also exhibits the triangle structure 3 in LEED intensity plot in Fig. 4 e . The spots indicated by the the STM image. It is located on an almost circular region arrows are due to Fe3O4(111) islands that start to grow at about 250 Å in diameter that exhibits the compressed mono- this stage. The growth of these islands is discussed layer structure 2 and which is located 2.2 Å above the elsewhere.41 These islands are the reason for the deviation of lowest regions where the first layer is exposed. The same the FeO film thickness FeO and the effective film thickness step height of 2.2 Å, which corresponds to the step height on corresponding to the total amount of deposited iron EFF . Pt 111 2.26 Å , is measured between the high island and the other 2-ML-thick regions of the film. Therefore, this re- gion is the first and second FeO layer on top of a one-layer- D. Third layer structures and Fe3O4 111... island growth high platinum island with a diameter of 250 Å. The completion of the second and third FeO layer and the As discussed above the (8) 8))R30° triangle struc- characteristics of the Fe3O4(111) island growth starting at ture 3 starts to evolve on the second FeO layer surface at FeO coverages around 2 ML critically depend on the film coverages around 1.5 ML. At coverages around 1.8 ML it is oxidation temperature, which will be discussed more detailed 57 GROWTH AND STRUCTURE OF ULTRATHIN FeO FILMS . . . 7247 FIG. 8. 70 70 Å2 STM image of the second FeO layer surface exhibiting an unreconstructed, oxygen-terminated FeO 111 -(1 1) surface structure. The 3.15 Å unit cell is indicated. The moire´ superstructure of the nonrotated (8 8) coincidence structure is vis- ible and its 22 Å periodicity is indicated by the marked atoms. UT 0.3 V, IT 1.0 nA. again is 2.5 Å corresponding to the distance between iron- oxygen 111 bilayers in bulk FeO. On the 2-ML-thick film structures 2 and 4 coexist as deduced from the LEED pattern in Fig. 4 e . On the third FeO layer surface a moire´ superstructure indicating the existence of the coincidence structure 4 is observed. On the film prepared at T 920 K FIG. 7. a Gray-scale plot of the 00 LEED beam and its environment for FeO 1.8 ML, where the (8) 8))R30° struc- ture 3 is developed best. The fractional order spots of the (8) 8))R30° superstructure and the double scattering satellite spots of the coexisting structure 2 are visible. The latter are elongated due to the existence of two rotation domains. b 90 90 Å2 STM image of the (8) 8))R30° structure 3 . UT 1.3 V, IT 1.0 nA. in a forthcoming paper.41 This oxidation temperature depen- dence is demonstrated in Figs. 9 and 10, which show large- area STM images of two films prepared at T 870 and 920 K, respectively. The film prepared at T 870 K in Fig. 9 exhibits a closed second FeO layer and small third FeO layer islands. At this temperature FeO grows layer by layer up to a FIG. 9. 4400 4400 Å2 STM image of an FeO film prepared at thickness of about 2.5 ML. Fe3O4(111) islands start to grow T 870 K, FeO 2.2 ML thick. The second layer is completely upon completion of the second FeO layer, and further iron closed and small third FeO layer islands can be seen. The LEED deposition and oxidation results in increasing Fe3O4 island pattern of such a film is shown in Fig. 4 e . UT 1.0 V, IT sizes. The step height between the second and third layer 0.1 nA. 7248 M. RITTER, W. RANKE, AND W. WEISS 57 FIG. 10. 4400 4400 Å2 STM image of an FeO film prepared at FIG. 11. 90 90 Å2 STM image of the third FeO layer surface. T 920 K, FeO 1.6 ML thick. A third FeO layer island has It exhibits a similar unreconstructed and oxygen terminated formed. UT 1.0 V, IT 0.9 nA. FeO 111 -(1 1) surface structure with the moire´ pattern of the (8 8) coincidence superstructure 4 as observed on the second in Fig. 10 the second layer is not completed and looks like layer surface in Fig. 8. On the lower right a defect region attributed the 1.6 ML film shown in Fig. 6 b . Only the (8) to surface oxygen vacancies can be seen. U 8))R30° structure 3 is observed on the second layer T 0.3 V, IT 1.0 nA. surface of this film. A third layer FeO island has formed, the intensities of the corresponding LEED spots. The integral which is about 1200 Å 500 Å in size and located in the beam intensities were obtained from LEED patterns as upper part of Fig. 10. The darkest areas in the image are shown in Fig. 4 and are displayed in Fig. 12. Initially, the located one Pt 111 interlayer spacing deeper due to plati- FeO 10 intensity increases and the Pt 10 intensity de- num steps underneath the oxide film. These steps are not creases. Structure 1 is formed at submonolayer coverages. visible because the FeO coverage there changes from 1 to 2 For the chosen primary electron energy of Ep 90 eV the ML as also observed on the platinum island shown in Fig. 00 beam of Pt 111 is very weak. Therefore the 00 beam 6 b . intensity increase is almost entirely caused by the FeO over- Since FeO always grows layer by layer we interpret the layer formation and follows exactly the increase of the third layer island in Fig. 10 as the initial growth stage of an FeO 10 intensity. At 1 ML coverage the transformation Fe3O4(111) island. This interpretation is further evidenced from the submonolayer structure 1 into the compressed by a different surface structure on this island. Usually third structure 2 takes place. This must change the dynamic FeO layer FeO islands like those in Fig. 9. always exhibit the form factors determining the absolute beam intensities, as moire´ superstructure corresponding to structure 4 . In con- above 1 ML coverage the FeO 10 and 00 beams decrease trast to that two regions with different contrasts are visible in different ways and their intensity ratio changes. At cover- on the island surface in Fig. 10. In high-resolution scans the ages around FeO 1.2 ML only the compressed structure brighter regions show cluster like features without long- 2 exists. Between FeO 1.5 and 2.0 ML the (8) range order, whereas the darker regions are well ordered. 3))R30° structure 3 develops, above 2.0 ML it disap- The step height between the second FeO layers and the or- pears again and is replaced by the nonrotated structure 4 . dered surface regions on the island again is 2.5 Å. An atomic The compressed structure 2 still coexists but its intensity resolution 90 90 Å2 STM image of the ordered region is decreases as structures 3 and 4 develop. Along with the shown in Fig. 11, which exhibits an unreconstructed formation of structures 3 and 4 also Fe FeO(111) (1 1) surface. Here the film forms the same 3O4 derived spots appear not shown in Fig. 12 , in agreement with the obser- nonrotated (8 8) coincidence structure 4 with a periodic- vation of Fe ity of about seven lattice spacings on FeO 111 as observed 3O4 islands by STM. The formation of structures 3 and 4 at coverages around on the second layer surface of a 2 ML film shown in Fig. 8. EFF 2 ML is accompanied The superstructure corrugation on the third layer surface is a by an increase of the Pt 10 spot intensity, which finally even little weaker. Several randomly distributed defects are ob- exceeds the intensity on the clean platinum surface. The rea- served. The missing corrugation maxima are in registry with son for this must be an increased scattering of structures 3 the corrugation maxima on the defect-free surface areas, and and 4 into the Pt 10 spot position as the oxide overlayer therefore we interpret them as oxygen vacancies. thickness increases. E. Leed beam intensities IV. DISCUSSION The subsequent formation of the coincidence structures In this work we prepared all oxide films by oxidizing iron 1 ­ 4 described in the previous sections is also reflected in at temperatures between 870 and 1000 K in 10 6 mbar oxy- 57 GROWTH AND STRUCTURE OF ULTRATHIN FeO FILMS . . . 7249 and electrostatics at the interface play a dominant role for the growth mode of BaTiO3(100) on MgO 100 .22 Similar ef- fects might determine the initial stage of iron-oxide growth on Pt 111 . The large number of successively formed and often coexisting FeO coincidence structures on Pt 111 with slightly different lattice constants and rotation misfit angles seems confusing. They must reflect lowest total-energy ar- rangements, balancing the contributions of substrate- overlayer interface energies and strain energies within the oxide layer for each coverage. These lowest-energy configu- rations are always coincidence structures. This is clearly evi- denced by the observation of three FeO structures with dif- ferent lattice constants, structures 1 , 2 , and 4 , each having a rotation misfit angle against the platinum substrate that fits only to one particular coincidence structure. The rotated structure 1 is the coincidence structure with the lowest substrate-overlayer interface energy, because at sub- monolayer coverages the elastic energy within the laterally expanded oxide overlayer lattice is still small and overbal- anced by the energy gain due to the formation of the lowest- energy interface structures. This situation changes upon completion of the first monolayer, when the compressed structure 2 is formed and the FeO lattice clicks into new coincidence sites on the substrate surface. Now the oxide FIG. 12. LEED beam intensities at Ep 90 eV as a function of the FeO overlayer coverage given in the amount of deposited iron overlayer has reduced its strain energy by the lattice constant compression for the price of a less favorable coincidence EFF lower scale and in terms of the real FeO coverage FeO upper scale . The curves are scaled arbitrarily in order to make their interface structure, which continues to exist above 2 ML relative changes visible. The curves for FeO 10 squares and 00 coverage. Between 1.5 and 2 ML coverage the (8) circles beams are rescaled to each other at small coverages. The 8))R30° structure 3 forms which obviously is a transi- structures observed in the different coverage ranges numbering as tion structure that transforms further into the expanded unro- in Table I are indicated. tated (8 8) structure 4 upon completion of the second layer. Structure 2 first coexists with structure 3 and then gen partial pressure for 2 min. If solid bulk iron oxide gets with 4 . Structure 4 is also observed on the third layer into thermodynamic equilibrium with the oxygen gas phase surface and is the most expanded one. The reason for its Fe3O4 magnetite would coexist with -Fe2O3 hematite under stability is not clear, as one would expect thicker FeO films these conditions.16 Recent calculations reveiled an equilib- to adopt the properties of bulk FeO. Perhaps the Fe3O4 is- rium molar ratio of 25%:75% between Fe3O4 and -Fe2O3 at lands have an influence on the stability of structure 4 10 6 mbar oxygen partial pressure and T 1000 K.42 The within their environment. FeO wustite phase is thermodynamically stable only at tem- All observed coincidence structures have expanded lattice peratures above 840 K. At an oxygen partial pressure of constants 3.09­3.15 Å when compared to the bulk value of 10 6 mbar FeO is stable only at a temperature of about 1700 FeO 3.04 Å . A reduced interplanar Fe-O spacing of 0.65 Å K, but not under the preparation conditions we applied.43 As compared to the bulk spacing of 1.25 Å was obtained on the Fe3O4 and -Fe2O3 are the stable phases at our preparation FeO 111 monolayer film by photoelectron diffraction, conditions, the interaction with the platinum surface must which was explained by the lateral expansion of the FeO stabilize the initial formation of metastable FeO films. On lattice.26 However, a 50% reduction of the interlayer spacing lattice mismatched substrates epitaxial growth is determined seems very large to be caused by a lateral expansion of 1.6%. by the interface energy between the substrate and the first We cannot determine the topographic height difference be- strained layer and by the energy of the islands in the case of tween the Pt 111 substrate and the first FeO layer surface island formation. The latter is given by the sum of the island because of their different electronic surface structures. How- surface energies and the island-substrate interface energy ever, we observe the FeO bulk step height of 2.5 Å between that may contain dislocation defects.21 A stable interface the first and second and between the second and third FeO structure is formed between an expanded first FeO 111 layer layers, which all are laterally expanded, too. An angle depen- and the Pt 111 surface. FeO continues to grow layer by dent x-ray absorption study also reveiled the FeO bulk inter- layer, leading to an increasing elastic energy within the layer distance between the iron and oxygen planes in strained overlayer, until around 2 ML thickness Fe3O4(111) FeO 111 films grown on Pt 111 .44 islands start to grow. Both, the elastic energy increase and All FeO 111 films form oxygen terminated unrecon- the thermodynamic stability of Fe3O4 are the driving forces structed (1 1) surface structures based on the STM calcu- for the island formation around 2 ML FeO coverage. lations performed by Galloway and co-workers for the FeO The crucial role of interfacial energy minimization at the monolayer film with structure 2 . On a single-crystal sample first atomic layers for the heteroepitaxial growth of oxides an unreconstructed polar surface termination would be un- was demonstrated by McKee et al., who found that ion size stable, because its surface dipole leads to a diverging surface 7250 M. RITTER, W. RANKE, AND W. WEISS 57 free energy.45 Polar surfaces can be stabilized by reconstruc- submonolayer coverages this first layer consists of an oxygen tions, adatoms or vacancies that reduce the surface charge terminated FeO 111 bilayer that is laterally expanded if thereby lowering the surface energy.46 A (2 2) LEED pat- compared to bulk FeO and rotated by 1.3° against the plati- tern was observed on thin FeO 111 films prepared by ox- num substrate. This leads to a (8 2 1 10) coincidence structure idion of an Fe 110 single crystal surface, and an octopolar with respect to the Pt 111 surface. With increasing coverage reconstruction was proposed for this surface.47 The unrecon- different coincidence structures with different FeO lattice structed polar surfaces of the ultrathin FeO films on Pt 111 constants and rotation misfit angles are formed. They reflect must be stabilized by an image dipole in the platinum sub- lowest total energy arrangements balancing the correspond- strate underneath, which compensates the dipole in the oxide ing interface and elastic energies within the strained oxide overlayer. On the third FeO layer surface in Fig. 10 we ob- overlayer for each coverage. Around 2 ML FeO coverage the serve oxygen vacancies, which reduce the polar FeO 111 thermodynamic stable Fe3O4 phase starts to grow in its bulk surface energy on the somewhat thicker FeO film. This sur- structure, forming 111 oriented three-dimensional islands face energy lowering might be the driving force for the de- on top of the lattice mismatched platinum substrate or on top fect formation, as iron cation vacancies and not oxygen va- of the expanded first FeO layer. All films exhibit oxygen cancies are the predominant defects occurring in bulk terminated unreconstructed FeO 111 -(1 1) surface struc- Fe1 xO wustite, which for that reason exhibits large devia- tures. These polar surfaces are stabilized by an image dipole tions from stoichiometry.48 in the platinum substrate. On the third FeO layer surface oxygen vacancies are formed that stabilize the polar surface V. SUMMARY by reducing the surface charge. The initial growth of FeO on Pt 111 was studied by scan- ning tunneling microscopy and high resolution low energy ACKNOWLEDGMENT electron diffraction. The FeO oxide grows layer by layer We thank Manfred Swoboda for his excellent technical forming a very stable first layer on the Pt 111 surface. At assistance as well as Robert Schlo¨gl for helpful discussions. *Author to whom correspondence should be addressed. Fax: 49 Lieuwma, R. M. Wolf, A. Reinders, R. M. Jungblut, P. A. A. 30 84134401. Electronic address: weiss­w@fhi-berlin.mpg.de van der Heijden, J. T. W. M. van Eemeren, J. van de Stegge, and 1 Gabor A. Somorjai, Introduction to Surface Chemistry and Ca- W. J. M. de Jonge, Surf. Sci. 373, 85 1997 . talysis John Wiley & Sons, New York, 1994 . 19 F. C. Voogt, T. Hibma, G. L. Zhang, M. Hoefman, and L. Niesen, 2 H. H. Kung, Transition Metal Oxides: Surface Chemistry and Surf. Sci. 331, 1508 1995 . 20 Catalysis Elsevier, New York, 1991 . T. Fujii, M. Takano, R. Katano, Y. Bando, and Y Isozumi, J. 3 P. J. van der Zaag, R. M. Wolf, A. R. Ball, C. Bordel, L. F. Appl. Phys. 666, 3168 1989 . 21 E. Pehlke, N. Moll, A. Kley, and M. Scheffler, Appl. Phys. A to Feiner, and R. M. Jungblut, J. Magn. Magn. Mater. 148, 346 be published . 1995 . 22 R. A. McKee, F. J. Walker, E. D. Specht, G. E. Jellison, L. R. 4 J. C. Mallinson, The Foundations of Magnetic Recording, 2nd ed. Boatner, and J. H. Harding, Phys. Rev. Lett. 72, 2741 1994 . Academic, New York, 1993 . 23 G. H. Vurens, V. Maurice, M. Salmeron, and G. A. Somorjai, 5 J. W. Geus, Appl. Catal. 25, 313 1986 . Surf. Sci. 268, 170 1992 . 6 M. Muhler, R. Schlo¨gl, and G. Ertl, J. Catal. 138, 413 1992 . 24 M. Ritter, H. Over, and W. Weiss, Surf. Sci. 371, 245 1997 . 7 D. Zscherpel, W. Weiss, and R. Schlo¨gl, Surf. Sci. 382, 326 25 H. C. Galloway, J. J. Benitez, and M. Salmeron, Surf. Sci. 298, 1997 . 127 1993 . 8 H.-J. Freund, Angew. Chem. 109/5, 444 1997 . 26 C. S. Fadley, M. A. Van Hove, Z. Hussain, and A. P. Kaduwela, 9 W. Weiss and G. A. Somorjai, J. Vac. Sci. Technol. A 11, 2138 J. Electron Spectrosc. Relat. Phenom. 75, 273 1995 . 1993 . 27 H. C. Galloway, P. Sautet, and M. Salmeron, Phys. Rev. B 54, 10 K. Takeuchi, S. S. Perry, M. Salmeron, and G. A. Somorjai, Surf. R11 145 1996 . Sci. 323, 30 1995 . 28 H. C. Galloway, J. J. Benitez, and M. Salmeron, J. Vac. Sci. 11 A. B. Boffa, H. C. Galloway, P. W. Jacobs, J. J. Benitez, J. D. Technol. A 12, 2302 1994 . Batteas, M. Salmeron, A. T. Bell, and G. A. Somorjai, Surf. Sci. 29 G. Tarrach, D. Bu¨rgler, T. Schaub, R. Wiesendanger, and H.-J. 326, 80 1995 . Gu¨ntherodt, Surf. Sci. 285, 1 1993 . 12 B. A. Parkinson and F. S. Ohuchi, K. Ueno, and A. Koma, Appl. 30 R. Jansen, V. A. M. Brabers, and H. van Kempen, Surf. Sci. 328, Phys. Lett. 58, 472 1991 . 237 1995 . 13 T. Fujii, M. Takano, R. Katano, Y. Bando, and Y. Isozumi, J. 31 A. R. Lennie, N. G. Condon, F. M. Leibsle, P. W. Murray, G. Appl. Phys. 68, 1735 1990 . Thornton, and D. J. Vaughan, Phys. Rev. B 53, 10 244 1996 . 14 Y. J. Kim, Y. Gao, and S. A. Chambers, Surf. Sci. 371, 358 32 N. G. Condon, F. M. Leibsle, A. R. Lennie, P. W. Murray, D. J. 1997 . Vaughan, and G. Thornton, Phys. Rev. Lett. 75, 1961 1995 . 15 Y. Gao, Y. J. Kim, G. Bai, and S. A. Chambers, J. Vac. Sci. 33 N. G. Condon, F. M. Leibsle, T. Parker, A. R. Lennie, D. J. Technol. A 15, 332 1997 . Vaughan, and G. Thornton, Phys. Rev. B 55, 15 885 1997 . 16 A. Muan, Am. J. Sci. 256, 171 1958 . 34 W. Weiss, M. Ritter, D. Zscherpel, M. Swoboda, and R. Schlo¨gl, 17 D. M. Lind, S. D. Berry, G. Chern, H. Mathias, and L. R. Tes- J. Vac. Sci. Technol. to be published . tardi, Phys. Rev. B 45, 1838 1992 . 35 U. Scheithauer, G. Meyer, and M. Henzler, Surf. Sci. 178, 441 18 J. M. Gaines, P. J. H. Bloemen, J. T. Kohlhepp, C. W. T. Bulle- 1986 . 57 GROWTH AND STRUCTURE OF ULTRATHIN FeO FILMS . . . 7251 36 W. Ranke, D. Zscherpel, W. Weiss and R. Schlo¨gl unpublished . 43 L. S. Darken and R. W. Gurry, J. Am. Chem. Soc. 67, 1398 37 M. A. Van Hove, W. H. Weinberg, and C.-M. Chan, Low-Energy 1945 . Electron Diffraction Springer-Verlag, Berlin, 1986 . 44 Th. Schedel-Niedrig, W. Weiss, and R. Schlo¨gl, Phys. Rev. B 52, 38 W. Weiss, D. Schmeisser, and W. Go¨pel, Surf. Sci. 207, 401 17 449 1995 . 1989 . 45 P. W. Tasker, J. Phys. C 12, 4977 1979 . 39 M. Henzler and W. Go¨pel, Oberfla¨chenphysik des 46 A. Barbieri, W. Weiss, M. A. Van Hove, and G. A. Somorjai, Festko¨rpers Teubner Studienbu¨cher, Stuttgart, 1994 . Surf. Sci. 302, 259 1994 . 40 W. Kleber, Einfu¨hrung in die Kristallographie VEB Verlag 47 D. Cappus, M. Haßel, E. Neuhaus, M. Heber, F. Rohr, and H.-J. Technik, Berlin, 1979 . Freund, Surf. Sci. 337, 268 1995 . 41 W. Weiss, M. Ritter, and R. Schlo¨gl unpublished . 48 W. Schweika, A. Hoser, M. Martin, and A. E. Carlsson, Phys. 42 W. Weiss, Surf. Sci. 377, 943 1997 . Rev. B 51, 15 771 1995 .