Papers

AVS1997 Session SS2-WeA: Diffraction Techniques and Ordered Overlayers

Wednesday, October 22, 1997 2:00 PM in Room A1/2-A

Wednesday Afternoon

Start	Invited?	Item
2:00 PM		SS2-WeA-1 Accurate Cluster-Based Methods of Modeling Photoelectron Diffraction Data: Comparisons and Applications to Adsorption and Epitaxial Growth Y. Chen (Lawrence Berkeley Nat'l Lab); H. Wu (Univ. of Hong Kong); A. Chassé (Martin Luther Univ., Germany); F.J. García de Abajo, D.A. Shirley, Z. Hussain (Lawrence Berkeley Nat'l Lab); A.P. Kaduwela (California Air Resources Board); R.X. Ynzunza, C.S. Fadley, M.A. Van Hove (Lawrence Berkeley Nat'l Lab) Photoelectron diffraction (PD) is a widely used probe of surface structures, and its most accurate modeling requires multiple scattering theory. The advent of third- generation synchrotron radiation sources has much increased the rates of acquisition of such data, which can be obtained with unprecedented resolutions in angle and energy, with electron spin detection and with variable photon polarization. Thus, theory is challenged to be accurate and general enough, as well as fast enough, to describe in a convenient way all the observed effects. The multiple scattering cluster method has been applied to PD simulations previously, using the convenient and accurate separable Green's function approach of Rehr and Albers in 2nd order¹^,²]. A more recent implementation, using the backward summing method³, has improved the computation speed of this by over an order of magnitude, in particular by allowing quick rotations of incident and exit angles in any combination. These approaches will be described and compared to calculations from more exact, albeit slower, cluster methods (e.g., ⁴). Their application to experimental data recently obtained at the Advanced Light Source will also be presented, such as for photoelectron diffraction data obtained from adsorbate-covered metal surfaces, including metallic epitaxial layers. In addition, extensions to model the effects of photon polarization, magnetism and electron spin will be described, allowing the simulation of various forms of circular and magnetic dichroism as well as relativistic effects (spin flip). Work supported by U.S. DOE. ¹J.J. Rehr and R.C. Albers, Phys. Rev. B41, 8139 (1990). ²A.P. Kaduwela et al., J. Electron Spectrosc. 57, 223 (1991). ³H.S. Wu, Y. Chen, D.A. Shirley and Z. Hussain, in preparation. ⁴P. Rennert and A. Chassé, Exp. Techn. Phys. 35, 27 (1987); O. Speder, P. Rennert and A. Chassé, Surf. Sci. 331-333, 1385(1995).
2:20 PM		SS2-WeA-2 Interaction of Se with Si(111): A Photoelectron Diffraction Study S. Meng, B.R. Schroeder, M. Leskovar, M.A. Olmstead (University of Washington) The structure and composition of a silicon surface after exposure to individual constituent elements are crucial in understanding the initial formation of compound heteroepitaxial films. The initial stages of film growth often involve saturation or reaction of substrate dangling bonds. For example, the growth of GaAs on Si(111) begins with As saturation of the substrate dangling bonds, whereas the initial growth of ZnSe/Si(100) involves Se reaction with surface Si, forming SiSe₂. In this study we have looked at Se/Si(111) as the initial step in GaSe or ZnSe growth on Si(111). We have used X-ray photoelectron spectroscopy (XPS), component-resolved X-ray photoelectron diffraction (CR-XPD) and low energy electron diffraction (LEED) to study the growth of Se on clean Si (111) 7x7 at various substrate temperatures. Our LEED observations indicate that the adsorption of Se on Si(111) surfaces does not lead to an ordered termination of the 7x7 surface. We find that the growth of Se reaches a saturation coverage during exposure of Si samples at temperatures above 380 C and our XPS data are consistent with a monolayer or submonolayer coverage of Se. Deposition of Se at room temperature followed by annealing at higher temperatures leads to similar results. In either case our preliminary CR-XPD studies show no interdiffusion of Se at the interface and no evidence of SiSe₂ formation.
2:40 PM		SS2-WeA-3 Atomic Structural Model of Ga:Si(112): Theory and Experiment A.A. Baski (Virginia Commonwealth University); S.C. Erwin, L.J. Whitman (Naval Research Laboratory) The Ga:Si(112) system is a striking example of how the presence of an adsorbate can dramatically influence the morphology of a surface. Our previous scanning tunneling microscopy (STM) studies of the clean Si(112) surface revealed a quasi-periodic sawtooth morphology consisting of ~10 nm wide nanofacets ¹. These nanofacets can be eliminated, however, by the adsorption of less than an atomic layer of Ga, which restores the surface to a planar (112) orientation with a 6x1 (or 5x1) reconstruction ². We now report a comprehensive theoretical and experimental study of Ga:Si(112) which supports a vacancy model first proposed by Jung et al. ³. The vacancy model incorporates Ga atoms in three-fold bonding sites along the step edges of the unreconstructed Si(112) surface, with a Ga vacancy occurring every four to five atoms along the step. At each metal vacancy, the adjacent Si atoms at the step edge dimerize to further lower the density of dangling bonds. According to first-principles, electronic-structure calculations, it is the Si step edge atoms -- not the adsorbed Ga atoms -- that are observed in filled-state STM images. These calculations also indicate that the energetically preferred ground state for a generalized nx1 vacancy model does in fact occur for n equals five to six, as is experimentally observed. At this n value, the strain energy associated with the Ga-Si bonds (compressive) and the dimerized Si-Si bonds (tensile) also shows a minimum, indicating that strain plays an important role in this system. ¹A.A. Baski and L.J. Whitman, Phys. Rev. Lett. 74, 956 (1995). ²A.A. Baski and L.J. Whitman, J. Vac. Sci. Technol. B 14, 992 (1996). ³T.M. Jung, S.M. Prokes, and R. Kaplan, J. Vac. Sci. Technol. A 12, 1838 (1994).
3:00 PM		SS2-WeA-4 Alkali Metal Adsorption on Cleaved Si(111): A LEED and Photoemission Study of New Surface Phases E.J. Nelson (Stanford Synchrotron Radiation Laboratory); T. Kendelewicz, P. Liu (Stanford University); P. Pianetta (Stanford Synchrotron Radiation Laboratory) Alkali metal (AM) adsorption on semiconductors is notable for the large decrease in work function, the ordered formation of a single AM monolayer (1 ML) at saturation, and the simple electronic structure of AM's, making the AM/semiconductor interface a model for other metal/semiconductor interfaces. Previous work on room temperature (RT) adsorption of AM's on the cleaved Si(111)2x1 surface found the surface converts to a 1x1 LEED pattern at 0.5 ML Na coverage, converts to a √3x√3-R30 pattern at 1 ML Cs coverage, and maintains its 2x1 pattern at 1 ML K coverage.^1,2,3 Our results confirm the Na adsorption results; however, for 1 ML saturation, the LEED patterns are 1x1 with a weak higher-order star-shaped pattern for Cs, and 1x1 for K. Perhaps more interesting are the six surface phases seen for the Cs- and K-dosed surfaces as AM coverage is increased from 0 to 1 ML in small RT doses. The LEED pattern progressions are analogous for Cs and K, with the √3x√3-R30 pattern for Cs corresponding to the 3x1 pattern for K at about 0.5 ML coverage. Na on cleaved Si(111) has a less complex RT surface phase diagram, with an additional 2x2 LEED pattern at 0.25 ML. Annealing the 1 ML AM-dosed cleaved Si(111) surface produces 3x1 LEED for Na and K, with AM coverage reduced to about 1/3 ML. The annealed Na/Si(111)3x1 and K/Si(111)3x1 and the RT-dosed K/Si(111)3x1 and Cs/Si(111)√3x√3-R30 surfaces are passivated to oxygen, lasting days in UHV without degradation of LEED patterns or a noticeable O 1s core-level peak. This paper provides LEED photographs of new AM/Si(111) surface phases. It sets limits on AM coverages for these phases using photoemission and secondary electron cutoff results. Changes in photoemission spectra with increasing AM coverage are investigated. These new LEED results show more complex interaction between AM's and the cleaved Si(111) surface than previously thought and hopefully will encourage further exploration of AM/semiconductor interfaces. ¹B. Reihl, S.L. Sorenson, R. Dudde, and K.O. Magnusson, Surf. Sci. 269-270, 1005 (1992). ²K.O. Magnusson and B. Reihl, Phys. Rev. B 39, 10456 (1989). ³B. Reihl and K.O. Magnusson, Phys. Rev. B 42, 11839 (1990).
3:20 PM		SS2-WeA-5 Magnetic Surface-Alloy Structure Determination by Photoelectron Holography, Surface X-ray Diffraction, and Quantitative Photoelectron Diffraction S. Banerjee, X. Chen (University of Wisconsin, Milwaukee); J. Denlinger (University of Michigan); S. Ravy, Y. Garreau, M. Sauvage (University of Paris-South, France); B.P. Tonner, D.K. Saldin (University of Wisconsin, Milwaukee) The system c(2x2) Mn/Ni(001) is unusual both structurally and magnetically. The magnetic properties of this two-dimensional alloy show that it is ferromagnetic, with parallel alignment of the Mn and Ni moments. However, the bulk binary alloy MnNi is an antiferromagnet, suggesting that the details of the structure of the surface phase drive a radically different magnetic alignment. We have performed three different types of surface structure determination on this system, for both single layer and multi-layer films. These measurements include a no-assumptions photoelectron hologram, multiple scattering photoelectron diffraction, and surface x-ray scattering experiments.
3:40 PM		SS2-WeA-6 Photoelectron Diffraction as a Contrast Mechanism and a Crystallographically-Sensitive Tool in Electron Microscopy M. Zharnikov, M. Neuber, M. Grunze (Universität Heidelberg, Germany) The applicability of photoelectron diffraction as a contrast mechanism in electron microscopy was investigated with a Ni polycrystal as a test system. Because of the anisotropy of both outgoing photoelectron wave and consequent electron-atom scattering some crystallographic contrasts can be observed at a particular kinetic energy of photoelectrons if several microcrystallites with the different crystallographic orientation are exposed to X-rays. Kinetic energies of several hundreds of electron volts, at which the zero-order interference or forward focusing dominates over other scattering events, are however preferable because of the pronounced strength of forward scattering and direct correlation between photoelectron intensity and the orientation of the interatomic axes in the sample. Well-resolved images of individual microcrystallites on the surface of Ni polycrystal have been obtained by using the Ni 2p_3/2 photoelectrons with a kinetic energy of 635 eV and a widely varying direction of emission. Even some crystallites, which were not distinguished by an optical microscope, have been easily discriminated by the photoelectron diffraction contrast. This contrast amounted almost 50 % of the whole intensity scale and could be observed directly during acquisition of the images. A crystallographic information on the orientation of the microcrystallites constituting the polycrystal was obtained: Some crystallites with the low-index surfaces were identified, some microcrystallites with the same orientation were recognized, and the symmetry of the surfaces of the individual crystallites was controlled. An approximate identification of the crystallographic orientation of several individual microcrystallites in the investigated Ni polycrystal has been achieved.
4:00 PM		SS2-WeA-7 Genetic Algorithms and Parallel Computing for the Global Search in Complex Surface Structure Determination by LEED M.A. Van Hove, R. Döll, S. Sachs, G. Stone (Lawrence Berkeley National Laboratory) The main remaining limitation in surface crystallography is the difficulty of finding the globally-best model of complex structures. "Direct methods" have only succeeded for some simple structures. For example, without prior knowledge, it is very difficult to determine in what way a surface is reconstructed, whether an adatom resides as an overlayer or substitutionally or interstitially, and whether an adsorbed molecule is intact or decomposed. Many methods exist for global searches, including simulated annealing, neural networks, maximum entropy and genetic algorithms. In prior work¹, genetic algorithms were found to be very promising in low-energy electron diffraction (LEED). They simulate the natural evolution of living organisms by means of crossover and mutation of chromosomes, generating individuals of greater "fitness". A particular surface structure can be encoded as one instance of a chromosome (e.g. as a bit-string), and its fitness is quantified in LEED through its R-factor (which is the degree of misfit between theory and experiment). Calculating the R-factor is time-consuming for complex structures due to multiple scattering in LEED. This problem is ideal for parallel computing. We therefore explore and optimize the behavior of genetic algorithms for LEED on a Cluster of Multi-Processor Systems, which is a promising architecture for next-generation supercomputers. In addition, we explore the acceleration of genetic algorithms by combining them with automated tensor LEED, which provides particularly rapid local optimization through steepest descent. Work supported by U.S. DOE. ¹R. Döll and M.A. Van Hove, Surf. Sci. 355, L393 (1996).
4:20 PM		SS2-WeA-8 Adsorption Structures on Co(0001) J. Lahtinen, T. Vaara, J. Vaari (Helsinki University of Technology, Finland); P. Kaukasoina, M. Lindroos (Tampere University of Technology, Finland) Low energy energy diffraction (LEED) has been utilized to characterize the clean Co(0001) and the (2x2)-K and (√3x√3)R30-CO adsorption structures on Co(0001). Results from all systems are presented. The measured intensity versus energy curves of the diffraction spots are compared to the theoretical curves calculated using the van Hove/Tong program. The LEED patterns were recorded at 160 K in order to minimize thermally induced vibrations. The experimental setup is designed to minimize the electron beam induced damage. This is done by deflecting the incident beam away from the surface by applying a bias voltage to the sample both when the beam energy is being varied and between individual video frame recordings. The analysis suggests that the K-atoms take the on-top site above the first layer cobalt atoms inducing large rumpling of 0.2Å in the outermost Co layer. The CO molecules also adopt the on-top site above the first layer cobalt atoms with their molecular axis perpendicular to the surface and attached to the surface with their C-end.