Papers

AVS1997 Session SSPD-ThE: Postdeadline Discovery Session

Friday, October 24, 1997 8:10 PM in Crystal Room (Fairmont Hotel)

Friday Evening

Start	Invited?	Item
8:10 PM		SSPD-ThE-1 Is the di-σ Bonding Model Correct for Acetylene Adsorbed on the Si(100)-2x1 Surface G.J. Lapeyre, Y. Yang, S.H. Xu (Montana State University); E. Rotenberg (ALS/LBNL); C. Chen (Montana State University); J.T. Yates, Jr. (University of Pittsburgh)
8:30 PM		SSPD-ThE-2 Atomic Landslides on Surfaces: A Novel Effective Channel for Interlayer Mass Transport M. Giesen, G. Schulze Icking-Konert, H. Ibach (Forschungszentrum Jülich, Germany) The morphology of epitaxially grown thin films on solid surfaces is frequently controlled by kinetics rather than equilibrium thermodynamics. A delicate interplay between nucleation, diffusion and the mass exchange between terraces of different height can cause a surprising complexity of morphological features. The mass exchange between terraces is often hindered by an additional activation barrier for the diffusion of adatoms across step edges (Schwoebel-Ehrlich (SE) barrier). Using scanning tunneling microscopy, we have investigated the decay of multilayer islands of copper on a Cu (111) surface. We report on the observation of a novel and very effective mechanism of interlayer mass transport which bypasses the SE barrier for the diffusion of atoms over step edges. The channel for a rapid mass transport opens when a two-dimensional island engaged in a random walk on a surface touches the boundary of a descending step. The decay rate of the island then increases by about two orders of magnitude. Even entire mounds can disappear in a very short time due to ledge contact events caused by equilibrium fluctuations of step edges. Similar to a landslide flowing down a mountain side, an avalanche of atoms flow down the descending step edge. Such an event is initiated and controlled by the decay of the lowest layer island in a mound. The ledge contact decay channel has significant consequences for the stability of mounds. A normal diffusion limited decay of a mound would proceed via a sequential decay of the islands from top to the bottom, the decay of each layer being slowed down by the SE barrier. Due to the ledge contact mechanism, the entire mound disappears with the decay rate of the lowest layer island. For mounds consisting of a large number of layers, the time for a total decay is therefore reduced by orders of magnitude. This rapid decay should have significant consequences also for the long time limit scaling behavior in epitaxial growth.
8:50 PM		SSPD-ThE-3 Holography Guided LEED Structure Analysis of SiC(111)-(3x3) K. Reuter, J. Schardt, J. Bernhardt, M. Franke, H. Wedler, U. Starke, K. Heinz (University of Erlangen-Nürnberg, Germany) Recent reports on homoepitaxial growth of SiC by gas source molecular beam epitaxy (MBE) under silicon rich conditions have drawn much interest on the SiC(111)-(3x3) structure which is observed during growth. The atomic arrangement in this structure may be crucial for the step flow growth occuring. For this system a conventional low energy electron diffraction (LEED) analysis is bound to fail due to the excessive number of atoms in the (3x3) unit cell -- the complexity being simply too large for a trial and error based method. We have used a combined effort of LEED holography and scanning tunneling microscopy (STM) to guide the quantitative LEED structure analysis. In STM images a single protrusion per unit cell is observed. The holographic inversion of the acquired LEED intensity resulted in a clear, unambiguous 3D image showing atoms down to the fourth layer in the local surrounding of the topmost adatom seen with the STM. These findings enabled the refining structure analysis in a now drastically reduced parameter space revealing the exact atomic geometry of the complete surface. Apart from its per se importance for SiC materials research, the success of the analysis shows that LEED holography can be applied to ordered superstructures, even though originally developed for the case of diffuse adsorption. Further, it finally proves the long claimed capability of LEED holography to provide substantial direct information on complex surfaces and to act as a guiding precursor for a subsequent quantitative LEED structure analysis.
9:10 PM		SSPD-ThE-4 Strain Field of Diluted-Dimer Rows X.R. Qin (University of Wisconsin, Madison); B.S. Swartzentruber (Sandia National Laboratories); M.G. Lagally (University of Wisconsin, Madison) We have made high-resolution scanning tunneling microscopy studies of diluted ad-dimer rows on the Si(001)-2x1 surface. Using bias-dependent measurements we show that when Ge or Si ad-dimers form diluted rows there is a clear reconstruction of the substrate dimers along the longitudinal sides of the ad-dimer row. In the most surface state sensitive imaging mode this reconstruction appears as "wing"-like structures adjacent to diluted ad-dimer rows. The strain field reflected by the wing structure involves a gradual transition in the adjacent substrate from pinned symmetric dimers to the common buckled ones. The effect of the tensile stress in the "wings" changes the appearance of the dimers from that of a normal dimer for a distance at least three dimers away from the ad-dimer location. These results provide a more realistic structural model for the diluted ad-dimer rows than has been considered in recent state-of-the-art theoretical calculations, by including changes in the vicinity of the rows. The anisotropic shape of the wing structure further provides quite naturally an atomistic picture of the growth-shape anisotropy that is so obvious in Si-on-Si and Ge-on-Si growth. In analysing the experimental results, we revisit conventional interpretations of empty state images of a clean Si(001)-2x1 surface, delineate its inadequacy with a more complete set of bias-dependent data, and separate the contribution from the surface states of dimer atoms from that of the back bonds. ¹Supported by NSF and by Sandia National Laboratories.
9:30 PM		SSPD-ThE-5 Theoretical Studies of Silicon Oxidation B.B. Stefanov, K. Raghavachari (Bell Laboratories, Lucent Technologies)
9:50 PM		SSPD-ThE-6 Thin Film vs. Bulk-Truncated Al₂O₃(0001): Metal Overlayers and Interfaces D.R. Jennison (Sandia National Laboratories); C. Verdozzi (Sandia National Laboratories (also at IRCSS, Univ. of Liverpool, UK)); P.A. Schultz, M.P. Sears (Sandia National Laboratories) Using first-principles LDA, we explore structural and adsorbate properties of α-Al₂O₃, and compare the bulk-truncated Al-terminated surface with an O-terminated thin film on a metallic (Al) substrate. a) Bulk Truncated: Significant surface relaxations extend to the fifth atomic layer, making previous slab calculations too thin to contain them. Ag and Pt epitaxial overlayers bind electrostatically and both prefer atop-Al sites, which bind twice as strongly than the next site, atop-O; there is negligible charge transfer. Even weakly bound Ag produces large (0.3 Å) relaxations in the oxide surface; Pt is much more strongly bound. Thus thick slabs are needed to study this rather open oxide. b) Thin Film: Noting experimental evidence that the film grown on NiAl(110) resembles γ-Al₂O₃, is O-terminated, and is three O-layers (~5Å) thick [1], we have modeled such a film. Again, large relaxations are noted. We compare metal binding here with the bulk truncated surface. c) The Al-Al₂O₃ Interface: Several models were tried. The preferred structure begins with O(1x1)/Al(111), overlain by two layers of O₃Al₂, but reversing the fcc order to prevent the lowest Al in the oxide from being above the top Al in the metal. A low energy closely packed interface results, which stabilizes the charged top surface, with fcc ordering within both the (γ-like) oxide and the metal. The Al and O layering through the interface is then A-B-C-B-A. [1] J. Libuda, et al., Surf. Sci. 318 (1994) 61. a) also at IRCSS, Univ. of Liverpool, Liverpool L69 3BX, UK * Sandia is a multiprogram laboratory operated by Sandia Corporation, a Lockheed Martin Company, for the U.S. DOE under Contract DE-AC04-94AL85000.
10:10 PM		SSPD-ThE-7 Is Feibelman Correct?: Structure of the Clean Ru(0001) Surface D.M. Zehner (Oak Ridge National Laboratory); J.P. Pierce (University of Tennessee and ORNL); C.F. Walters (University of Pennsylvania and ORNL) The interlayer spacing in the selvedge of a clean crystal may be modified by the presence of adsorbates. For Ru(0001), the first interlayer spacing has been experimentally determined to be contracted by 2 percent; however, results of calculations yield a contraction of 4 percent. To explain this difference, Feibelman has recently suggested that measurements performed on a supposedly "clean" sample were in fact performed on a hydrogen contaminated one, and that the amount of hydrogen present was sufficient to remove half of the interlayer contraction¹. We have used elastic recoil detection to measure both the hydrogen coverage on a "clean" surface, and the rate of adsorption from the background gas. Measurements performed subsequent to the standard cleaning procedure show that there is no hydrogen present on the sample. Additionally, the measured sticking coefficient is so small that only a few percent of a monolayer will adsorb in the time typically required to obtain the data used to determine the interlayer spacing. Thus, Feibelman's postulate is nullified, the difference between the experimentally and theoretically determined values of the interlayer contraction is not due to hydrogen contamination, and apparently ... A Ph.D. Is Not Enough! .. one also needs data. This work was supported by the Division of Materials Sciences, US DOE, under contract DE- AC05-96-OR22464 with LMER Corp. ¹P. J. Feibelman, et. al., Surf. Sci. 302 (1994) 81