Meetings & Events

fall 1997 logo1997 MRS Fall Meeting & Exhibit

December 1 - 5, 1997 | Boston
Meeting Chairs:
 Harry A. Atwater, Peter F. Green, Dean W. Face, A. Lindsay Greer 

Symposium S—Microscopic Simulation of Interfacial Phenomena in Solids and Liquids


Paul Bristowe, Cambridge Univ
Simon Phillpot, Argonne National Laboratory
John Smith, GM NAO R&D Center
David Stroud, Ohio State Univ

Symposium Support 

  • Argonne National Laboratory 
  • General Motors Research & Development

* Invited paper

Chairs: Peter S. Pershan and Stuart A. Rice 
Monday Morning, December 1, 1997 
Commonwealth (S)

9:00 AM *S1.1 
X-RAY SCATTERING MEASUREMENTS OF THE STRUCTURE OF LIQUID SURFACES. P.S. Pershan, Harvard Univ., Physics Dept. and the Div. of Eng. and Applied Sciences, Cambridge, MA.

We will first discuss the angular dependence of x-ray reflectivity measurements from the surface of liquid metals. A reflectivity peak at a wavevector close to the maximum in the liquid structure factor establishes the existence of predicted surface induced atomic layers for three metals studied, Hg[l], In[2] and Ga[3]. Although the temperature dependence of the amplitude peak for Ga is accounted for by a Debye Waller like effect due to thermal surface waves[3] the same is not true for Hg[4]. An 5 thick oxide layer on the surface of liquid Ga suppresses thermal capillary waves[3]. We will also discuss application of the grazing incidence diffraction technique to characterize amorphous positional order parallel to the surface for Langmuir monolayers(LM) on the surface of water. For typical surfactants, like fatty acids, diffraction peaks are only observed for crystalline LM; however, peaks associated with a two-dimensional amorphous phase are observed from LM of a fullerene denvative[5]. We will discuss prospects for characterizing phase transitions from both this amorphous phase, and a similar anisotropic one formed from monolayers of rod like -helix segments of the polypeptide poly--benzyl-L-glutamate[5] *AII of the results on liquid metals result from a long standing collaboration with Ben Ocko (BNL), Moshe Deutsch (Bar Ilan, Israel), our students and postdoctoral associates.

9:30 AM *S1.2 
THEORETICAL STUDIES OF THE STRUCTURES OF THE LIQUID-VAPOR INTERFACES OF METALS AND BINARY ALLOYS. Stuart A. Rice, Department of Chemistry and The James Franck Institute, The University of Chicago, Chicago, IL.

The results of self consistent quantum Monte Carlo simulations of the liquid-vapor interfaces of metals and binary alloys will be described. Comparisons of the predicted and experimentally determined structures of those interfaces will be made for all the cases for which data are available. The extent to which the theory suggests qualitative features which are similar in the liquid-vapor interfacial structures of all metals, e.g., an approximate theorem of corresponding states, will be discussed.

10:30 AM *S1.3 
MONTE CARLO AND MOLECULAR DYNAMICS SIMULATIONS OF LIQUID SEMICONDUCTOR SURFACES. Zhiqiang Wang, Lucent Technologies, Inc. Holmdel, NJ; Wenbin Yu, Eidea Labs. Inc. Arlington, VA; David Stroud, Ohio State University, Columbus, OH.

We have numerically studied the surface tension and surface profiles of several liquid semiconductors, including Si, Ge, GaAs, CdTe, and their alloys, as a function of temperature and concentration. Two kinds of simulations have been carried out: direct free-energy calculations using Monte Carlo methods, and force summations using molecular dynamics. We use empirical two- and three-body interatomic interactions based on the form originally proposed by Stillinger and Weber for Si, in conjunction with simulation cell sizes ranging from 216 to as large as 8000 atoms and several novel numerical techniques including a direct calculation of the surface entropy. In the case of alloys, we find a striking segregation of the low-surface-tension component to the surface even when the alloy components are miscible at all concentrations. In most cases, the calculated results are in good agreement with available experimental data. () Work supported by NASA under grant NAG 3-1437.

11:00 AM S1.4 

The density functional theory (DFT) has been used by many authors for analyses of inhomogeneous fluids at solid surfaces and in pores. Conventional versions of the DFT imply minimization of the grand thermodynamic potential with respect to the fluid density within fixed solid boundaries at the given temperature and chemical potential. At certain conditions, multiple equilibrium states and associated hysteresis are observed. There exist many experimental evidences of such hysteretic behavior. Some of the prominent examples are the capillary condensation phenomena in nanopores and the adsorption-desorption transitions in mono- and multilayers. In order to study the hysteresis phenomena in greater detail, we have developed two new versions of the DFT, which allow us to trace not only the stable and the metastable states along the hysteresis loop, but also the unstable states inside the hysteresis loop. The obtained isotherms are typically S-shaped (like the van-der-Waals isotherms) with the unstable states inside the spinodal. These unstable states cannot be realized in a physical or a numerical experiment unless fluctuations, which drive the system to one of the metastable or stable states, are completely suppressed. In the developed DFT, the fluctuations are suppressed by imposing severe restrictions on the total amount of fluid molecules within the solid boundaries or on the exchange of molecules between the coexisting phases. While the conventional DFT corresponds to the grand canonical ensemble, the new versions of DFT correspond to the canonical ensemble and the Gibbs ensemble respectively. The theory is applied for studies of multiple states of inhomogeneous fluids at solid surfaces and in pores.

11:15 AM S1.5 
MOLECULAR DYNAMICS SIMULATIONS OF Cu3Sn LIQUID Sn INTERFACES. J. Aguilar, R. Ravelo, Univ. of Texas-El Paso, Dept of Physics and Materials Research Inst. El Paso, TX; M. Baskes, Sandia National Laboratories.

We present computer simulations of the Cu3Sn-liquid Sn interface. We will discuss Molecular Dynamics results of the structure and thickness of the interface as well as calculations of the surface tension and excess surface free energies as a function of temperature. The atomic interactios are calculated using the modified embedded atom method (MEAM). This modification of the EAM includes the angular dependence of the electron density which describes the bond bending forces necessary to model covalent materials. This material is based upon work supported by DOE contract DE-AC04-94AL85000 and CONACYT MX.

11:30 AM S1.6 

It is known that the states of fluids confined in pores or by surfaces, can undergo abrupt changes, reminiscent of the gas-liquid bulk phase transitions. In such transitions the densities of fluids change rapidly with pressure, and often the isotherms exhibit permanent hysteresis loops. Such a behavior is usually linked to the existence of meta- and unstable states. Since a real physical equilibrium system cannot exist in the unstable states, it spontaneously passes into a stable state, minimizing the corresponding thermodynamic potential. In such a case, the gaps on pvt or other thermodynamic diagrams are observed. However, if one artificially constrains the system, so that it cannot undergo this transition, it may remain in the internal equilibrium for an infinitely long time. Such constrained equilibrium states are usually very hard to achieve in the real world. However, as our study shows, they can be realized in numerical experiment, such as the Monte Carlo simulations. The Monte Carlo simulation methods of description of the matter are considered to be exact in principle for given molecular interaction potentials. These methods are widely used for studies of inhomogeneous fluids, confined in the narrow pores of solid materials. We use the Canonical Ensemble Monte Carlo (CMC), Gibbs Ensemble Monte Carlo (GMC), and Grand Canonical Ensemble Monte Carlo (GCMC) simulations to model behavior of the model Lennard-Jones fluid confined in the pores. The potential interaction and other parameters of the model used in the numeric experiment are chosen to represent nitrogen within the carbon slit pores. This system has been extensively studied previously by using Monte Carlo and Molecular Dynamics techniques, however no special attention has been paid to the region where both the metastable and unstable states could be found. Our studies have showed for the first time that considerably large regions of the unstable states of confined fluids can be found in the pores of width above certain critical size. On the other hand, in sufficiently narrow pores these states are not observed at all. When presented on the ''density versus pressure'' plane, the unstable states form a continuous S-shaped curve connecting the meta-stable states of the confined fluid with low and high density. GCMC simulations model metastable states (unconstrained) and, as a consequence, produce hysteresis loops in the metastable regions. GMC simulations model partially constrained states; by the appropriate choice of the parameters of the model (number of particles, size of the simulation cell), it is possible to constrain the system and observe unstable internally equilibrium states and the van der Waals-type loops connecting the metastable states of the confined fluid. CMC simulations model constrained states and produce unstable internally equilibrium states and van der Waals-type loops connecting the metastable states.

Chairs: Efthimios Kaxiras and David Vanderbilt 
Monday Afternoon, December 1, 1997 
Commonwealth (S)

1:30 PM *S2.1 
ATOMIC AND ELECTRONIC STRUCTURE OF VACANCIES ON THE GALLIUM ARSENIDE SURFACE. James R. Chelikowsky and Hanchul Kim, Department of Chemical Engineering and Materials Science, Minnesota Supercomputer Institute, University of Minnesota, Minneapolis, MN.

Relatively less progress has been made in the microscopic understanding of single atom vacancies on semiconductor surfaces in contrast to adatoms or bulk vacancies. However, surface vacancies occur abundantly in nature, and they play an increasingly important role in modern technologies which involve thin film growth, or layer-by-layer etching of surfaces. An accurate understanding of the physical and chemical properties of surface vacancies is a challenging problem in materials theory owing to the computational complexities of this system. We will present results of a study for the structural and the electronic properties of As vacancies on the GaAs (110) surface using ab initio pseudopotentials and a plane wave basis. Our minimum energy surface geometry corresponds to a symmetric inward relaxation of the neighboring surface Ga atoms. This predicted geometry is in contradiction with previous work based on tight binding calculations that predicted an outward relaxation (Lengel et al., Phys. Rev. Lett. 72, 836 (1994)). However, we will demonstrate that our geometry yields theoretical scanning tunneling microscopy images which accurately replicate the fundamental features of the experimental images: the reduced image at the vacancy site for the occupied states and the enhanced image of the adjacent Ga sites for the empty states. We also find that the single positive charge state is stable in p-type materials which is in good agreement with experiment.

2:00 PM *S2.2 
THEORETICAL STUDIES OF ATOMIC STRUCTURE AND DYNAMICS ON SEMICONDUCTOR SURFACES AND INTERFACES. Efthimios Kaxiras, Dept of Physics and Division of Engineering and Applied Sciences, Harvard Univ, Cambridge, MA.

The combined use of ab initio calculations based on density functional theory, and simulations employing Monte Carlo techniques or phenomenological continuum theories, allow comprehensive studies of complex phenomena such as epitaxial growth, chemical modification of substrates, and manipulation of the mechanical properties of solids. We discuss examples of such studies including: atomic and cluster diffusion, island nucleation and the early stages of epitaxial growth on semiconductor surfaces; the exploitation of surface strain and mobility to enhance the solubility of impurities; the enhancement of ductility by selective microalloying at interfaces; and adsorption and desorption processes that lead to passivation of semiconductor surfaces. In all these problems the input of quantum mechanical calculations, which afford high accuracy and chemical specificity, is crucial in determining key parameters such as activation energies and the relative stability of competing structures. The combination of these results with simple simulations or phenomenological models provides a powerful means for sheding light into complicated phenomena and making predictions for technologically relevant applications. This work is supported by the Office of Naval Research, and by the Materials Research Science and Engineering Center of Harvard University; it was carried out in collaboration with Kyeongjae Cho, Pantelis Kelires, Normand Modine and Umesh Waghmare.

2:30 PM *S2.3 

Using first principles molecular dynamics, we have studied the reconstructions and thermal properties of the (001) surfaces of cubic SiC. We show that C-terminated surfaces can have different reconstructions, depending on preparation conditions and thermal treatments [1]. Furthermore we analyze how the reconstructions of the Si-terminated surfaces depend upon stress and the presence of defects. Our findings allow us to interpret recent experiments about (001) SiC surfaces, in particular STM images [2].

3:15 PM *S2.4 
STOICHIOMETRY AND STRUCTURE OF POLAR GROUP-III NITRIDE SEMICONDUCTOR SURFACES. Jurgen Fritsch, Otto F. Sankey, Kevin E. Schmidt, John B. Page, Arizona State University, Dept. of physics And Astronomy, Tempe, AZ.

The polarity of compound semiconductors substantially influences the atomic structure and the stoichiometry of surfaces that are parallel to pure cation or anion planes. Well-known examples are the (001) and (111) surfaces of GaAs and other III-V compounds. Recently, wide band gap semiconductors like GaN and AlN have become subject of intense investigations. These materials are grown on polar substrates like the hexagonal surfaces of sapphire and SiC or the cubic surfaces of GaAs and Si. We employ a multicenter tightbinding-like formalism based on the local-density approximation and the pseudopotential method to examine the stoichiometry and the atomic and electronic structure of the anion and cation terminated (0001) surfaces of wurtzite-phase GaN and AlN. All stable surface configurations differ in atomic composition and periodicity from the ideal bulk-like termination. Vacancy structures are found to be the most stable configurations for the anion and cation terminated surfaces. For metal rich growth conditions our calculations favor the adsorption of metal atoms on the cation terminated surface while the adsorption of hydrogen stabilizes the surfaces in the presence of hydrogen. Surface chemical reactions relevant for the growth of thin nitride films like the adsorption of hydrogen and nitrogen from decomposed ammonia are discussed. The validity of the autocompensation principle is verified by a close inspection of all relevant surface electronic states.

3:45 PM *S2.5 
THEORY OF BaTiO3 SURFACES AND FERROELECTRIC DOMAIN WALLS. David Vanderbilt and J. Padilla, Dept. of Physics and Astronomy, Rutgers University, Piscataway, NJ.

Earlier work has shown that ab-initio studies are capable of providing accurate information on the bulk structural phase transitions and dielectric properties of perovskite ferroelectrics such as BaTiO3. Here, we focus on related studies of surfaces and interface properties of BaTiO3. First, ab-initio ultrasoft-pseudopotential calculations of (100) surfaces, terminated either on a BaO plane or on a TiO2 plane, will be presented. Both of these surfaces are non-polar, and therefore relatively low in energy. We consider surfaces of both the cubic and the tetragonal ferroelectric phase. In the latter case, the polarization is taken parallel to the surface; otherwise strong depolarization fields destroy the ferroelectricity. We present energies, structural relaxations, and electronic states for fully-relaxed surface geometries. The presence of the surface is found to have only a weak effect upon the ferroelectricity in the underlying layers, in spite of the fact that the surface relaxation energy is much larger than the characteristic ferroelectric distortion energy. The surfaces do not give rise to deep gap states in our calculations, in agreement with experiment. Second, we present results on 180 ferroelectric domain walls in tetragonal BaTiO3. This work is based on Monte Carlo simulations applied to an effective Hamiltonian that is extracted from ab-initio calculations. The ferroelectric order parameter is found to reverse abruptly on the scale of 1-2 lattice constants, and shows no tendency to rotate in the domain wall. We compute the domain-wall energy, free energy, and thickness, and study its spatial fluctuations.

4:15 PM S2.6 
AB-INITIO THEORY OF INITIAL OXIDATION OF SILICON (001) SURFACES. N. A. Modine, Gil Zumbach, Efthimios Kaxiras, Department of Physics, Harvard University, Cambridge, MA.

The oxidation of the Si(001) surface is an important process on both technological and theoretical grounds. Experimental studies have not provided a clear picture of even the relevant atomic structures during the initial stages of oxidation, while previous theoretical studies of these processes have yielded contradictory results. Using careful first principles total-energy calculations based on density functional theory, we study several mechanisms of incorporating a sub-monolayer coverage of oxygen into the characteristic dimer reconstruction of the Si(001) surface. Our recently developed Adaptive Coordinate Real-space Electronic Structure (ACRES) method allows us to obtain results that are adequately converged with respect to the numerous computational parameters associated with this difficult system. We compare our results with previous theoretical work and propose a physically motivated two step pathway for the initial incorporation of an oxygen atom into the dimerized surface. Based on our results, we can explain what formerly appeared to be puzzling Ultraviolet Photoelectron Spectroscopy measurements which indicated that each initial oxygen atom saturates two dangling surface bonds.

4:30 PM S2.7 

We have investigated the energies of a number of phases in the Zr-O system using the ab initio plane wave approach with ultra-soft pseudopotentials. In a first step, we have studied the four structures of zirconia compound which are observed with increasing temperature and pressure. We show that gradient corrections are necessary to reproduce the experimental structural sequence. The mechanism of the cubic-tetragonal distorsion is discussed. In a second step, we have studied the oxygen adsorption in Zr bulk and on the Zr(0001) surface using a supercell technique. We show that the heat of absorption displays similar values for both bcc and hcp zirconium structures and the energetically most favorable occupation site for oxygen are the octahedral sites. For the adsorption of oxygen on Zr(0001), we found that the energetically most favorable sites for oxygen are subsurface octahedral sites. The adsorption is found to be strongly exothermic and the oxygen atoms at subsurface adsorption sites are energetically more favorable that those in bulk oxides. The initial oxidation process of Zr surface is discussed in the framework of our ab initio results.

4:45 PM S2.8 
DIFFUSION PROCESSES AND PRE-EXPONENTIAL FACTORS IN HOMO-EPITAXIAL GROWTH ON Ag(100) *. Ulrike Kurpick, Talat S. Rahman, Department of Physics, Kansas State University, Manhattan, KS.

We explore theoretically the microscopic factors controlling adatom self diffusion on Ag(100). We calculate the local thermodynamic properties of the system, in the harmonic/quasi-harmonic approximation of lattice dynamics, using force constant matrices based on interatomic potentials from the embedded atom method. Through usage of these thermodynamic properties, within the framework of transition state theory, we provide a recipe [1] for the calculations of diffusion coefficients for both hopping and exchange processes. We find vibrational entropy contributions to be central to the calculation of the pre-exponential factor, as it impacts the latter significantly. The vibrational internal energy, on the other hand, contributes to the softening of the activation energy barriers, as a function of surface temperature. By examining the pre-exponential factors and activation barriers for hopping and exchange mechanisms for adatom self diffusion on flat (100) terraces, and on those with a <100>, or a <110> step-edge, we show for the first time that for self diffusion on Ag(100) the prevalent path for interlayer transport is via exchange over the <100> step-edge with a negative Schwoebel barrier. We also find the relative magnitudes of these activation barriers and pre-exponential factors to change with temperature. As a result, exchange processes become more probable for self diffusion on Ag(100) at higher temperatures, while hopping is favored at lower temperatures. We examine the implications of these results in the light of experimental data for growth on Ag(100). Additionally, we compare the results with that for homo-epitaxial growth on Cu(100).

Chairs: Robin W. Grimes and John H. Harding 
Tuesday Morning, December 2, 1997 
Commonwealth (S)

8:30 AM *S3.1 
COMPUTER SIMULATION OF INTERFACES IN CERAMICS. J. H. Harding, University College London, Dept. of Physics and Astronomy, London, UNITED KINGDOM; S. C. Parker and D. J. Harris, Univ. of Bath, School of Chemistry, Bath, UNITED KINGDOM.

The methods of computer simulation of grain boundaries and hetero-interfaces have long been available. What has been lacking has been the computer power to implement these methods effectively. Thus simulation of these interfaces has often been confined to highly symmetric (often twin) boundaries in oxides of the rock-salt structure using simple classical potentials. These calculations can now often be done with LDA methods and we discuss some comparisons with classical models. However, it is not obvious that such calculations are reasonable models for general boundaries. We will review developments in the field in two areas; the simulation of hetero-interfaces (both metal/ceramic and ceramic/ceramic) and the study of general grain boundaries. In both cases, we shall discuss the validity of extrapolation from the symmetric case.

9:00 AM S3.2 
AB-INITIO STUDY OF ELECTRONIC AND GEOMETRIC STRUCTURES OF METAL/CERAMIC HETEROPHASE BOUNDARIES. Sibylle Köstlmeier, Christian Elsässer, Bernd Meyer, Max-Planck-Institut für Metallforschung, Stuttgart, GERMANY; Michael W. Finnis, Queen's University, Belfast, UNITED KINGDOM.

The atomic structures at heterophase boundaries between cubic spinel MgAl2O4 and cubic metals Al and Ag have been determined ab initio via local density-functional calculations using a mixed-basis pseudopotential method. An analysis of the corresponding electronic structures yields insights into the chemical bonding at the interface between a metallic and an ionic material. The theoretical results are used for interpretations of both high-resolution and analytical transmission-electron-microscopy experiments, and they provide data for further developments of empirical potential models for atomistic simulations of heterophase interfaces.

9:15 AM S3.3 
ATOMIC AND ELECTRONIC STRUCTURE OF A POLAR CERAMIC/METAL INTERFACE: {222}MgO/Cu. R. Benedek, D.A. Shashkov, D.N. Seidman, Northwestern University, Evanston, IL; D.A. Muller, J. Silcox, Cornell University, Ithaca, NY; M.F. Chisholm, Oak Ridge National Laboratory; L.H. Yang, Lawrence Livermore National Laboratory, Livermore, CA.

Atomic and electronic properties of the polar ceramic/metal interface {222}MgO/Cu are simulated using ab initio local density functional theory (LDFT) and classical molecular statics. The simulations are compared with experimental observations by atom-probe field-ion microscopy, Z-contrast scanning transmission electron microscopy, and spatially resolved electron energy loss (EELS) spectroscopy. LDFT calculations with the plane wave pseudopotential method are employed to design an interatomic potential suitable for classical simulations of interface atomic structure, taking account of the relatively large misfit (0.15). The proposed interface potential (assuming a flat terminating MgO layer) consists of a one-body bonding contribution, parametrized in terms of the Universal Binding Energy function, and a two-body repulsive Born-Mayer contribution. Most of the work addresses the O-terminated {222} interface, the only MgO/Cu termination observed experimentally in internally oxidized specimens, and which has the largest calculated (coherent interface) adhesive energy. Molecular statics simulations of the interface atomic structure exhibit a network of partial dislocations along 1/2<110> directions with Burgers vectors 1/6<211>, and zero standoff. Layer densities of electronic states show bulklike behavior except for the terminating O layer, which contains a localized metal-induced gap state 1 eV above the MgO valence-band edge, 0.5 eV below the Fermi level. The spectral density above the Fermi level associated with the gap state may be responsible for the observed shift in the oxygen K-edge in the interface EELS spectrum. Work supported by the U. S. Department of Energy.

9:30 AM S3.4 
AB INITIO CALCULATIONS OF THE SiC/Al INTERFACE. John Hoekstra, Masanori Kohyama, Osaka National Research Institute, Department of Materials Physics, Ikeda, Osaka, JAPAN.

The SiC/Al interface has been studied using the ab initio pseudopotential method, with the conjugate-gradient technique proposed by Bylander-Kleinman-Lee, and Troullier-Martins soft pseudopotentials. Ionic and electronic structure at the interface, local density of states, Schottky-barrier heights, and bond adhesion between the two materials were determined for both the silicon terminated and carbon terminated surfaces. Results show a distinct difference between the Al-Si and the Al-C interactions effecting all aspects of the chemical bond, as well as bond adhesion.

10:15 AM S3.5 
ATOMISTIC SIMULATIONS OF THE WORK OF ADHESION AT METAL OXIDE INTERFACES. Tatsuya Ohira, Yukihiko Inoue, Advanced Technology Research Center, Mitsubishi Heavy Industries, Ltd., Yokohama, JAPAN.

We report some results of the work of adhesion at technologically important metal oxide interfaces predicted by atomistic simulations with interatomic potentials, the Modified Embedded Atom Method (MEAM). In this paper we study magnetite scale adhesion on a tube made from some different type materials, and treat metal oxide interfaces between magnetite scales (Fe3O4) and stainless steel tube surface oxides (NiFe2O4) or chromium coating tube surface oxides (Cr2O3, FeCr2O4). The MEAM potentials for iron, nickel, chromium, and oxygen have been previously developed. But Fe-O, Cr-O, Ni-O, Fe-Ni, and Fe-Cr pair interaction parameters have not been presented. In this work we have defined new parameters for those pair interaction based on experimental information such as lattice constants, cohesive energies, bulk moduli of those metal oxides, and have calculated the work of adhesion at Fe3O4 (110) / NiFe2O4(110), Cr2O3(110), Cr2O3(100), FeCr2O4(110) interfaces where a cluster-surface interface model is adopted. The work of adhesion was obtained from a minimum value in a potential energy curve vs. a separation at interfaces. As the results of the calculation, it was found that the work of adhesion to stainless steel tube surface is larger than that of adhesion to chromium coating tube surface. In addition, we found there is a potential barrier, which prevents a scale from approaching a tube surface, in a potential energy curve vs. a separation between scales and chromium coating surfaces. We further discuss the effect of surface directions at interfaces on the work of adhesion.

10:30 AM S3.6 
A METHOD TO PREDICT ADHESION ENERGY OF CERAMIC: METAL WETTING BASED ON CHEMICAL THERMODYNAMICS CONSIDERATION. Xiaodong Wang, Shanghai Institute of Metallurgy, Academia Sinica, Shanghai, CHINA; Jianxin Wu, Dept. of Mechanical and Aerospace Engineering, University of California, Los Angeles, CA.

Evaluation of the adhesion energy of ceramic:metal interfaces at liquid or solid state is very important for material processing and appliction. Here a previous method is refined to predict the adhesion energy of ceramic:liquid metal wetting based on chemical thermodynamics consideration. By treating the interface as a type of locally ordered atomic structure, the first layer of metal atoms are matched with the contacting ceramic species, which gives a proportinal constant relating the adhesion energy to the formation energy of the ceramic materials, especially in the case of oxides. The advantage of this method is that the proportional constant is calcualted from the atomic or ionic radii of the species concerned, unlike in other methods where the constant is obtained from regression of the measured adhesion energy against the formation energy of the oxide. Application to SiO2, ZrO2, and Al2O3 systems yields results in very good agreement with experimental measurement. The reasonable results suggest that the valent electronic structure of the ceramic surface ions, regardless of covalent or ionic, might change to nondirectional when in contact of a huge number of metal atoms.

10:45 AM S3.7 

The growth of thin heterovalent interlayers at semiconductor interfaces can be used to engineer band discontinuities in semiconductor heterostructures [1]. For any practical device application of such engineered heterojunctions, it is indispensable to assess whether the interlayer will produce localized states that may degrade the transport properties. Using the ab-initio pseudopotential approach, we have investigated the existence of localized states and resonances in abrupt GaAs/Si/GaAs (100)- and (110)-oriented heterostructures incorporating a 2 monolayer (ML) thick Si interlayer, as well as in fully developed GaAs/Si (110) heterojunctions. In the (100)-oriented heterostructure, we find both valence- and conduction-band-related near-band-edge states localized at the Si/GaAs interface. In the (110) systems, interface states occur deeper in the valence band, the highest valence-related interface states being about 1 eV below the band maximum. Furthermore, localized states near the conduction band minimum are observed in the (110)-oriented heterostructure incorporating 2 ML of Si, but not in the GaAs/Si (110) heterojunction. Using their characteristic bonding properties and atomic character, we were able to follow the evolution of localized states and resonances from the fully developed binary junction to the ternary GaAs/Si/GaAs (110) systems incorporating 2 and 1 ML of Si. This approach allowed us to show also the link between the interface states of the (110) and (100) systems. In order to interpret our results, we developed a Koster-Slater-type of model for the interface-state problem. Using this model, we can predict the conditions for the existence of localized interface states and resonances in terms of simple bulk-band-structure and interface-bonding parameters.

11:00 AM *S3.8 
ADVANTAGES OF USING THE COMBINATION OF ATOMISTIC SIMULATION AND HREM TO RESOLVE STRUCTURES IN DEFECTIVE ELECTROCERAMICS. Robin W. Grimes, Michael A. McCoy, Dept. of Materials, Imperial College, London, UNITED KINGDOM; William E. Lee, Dept. of Engineering Materials, Sheffield University, Sheffield, UNITED KIDGDOM.

Atomistic level computer simulation and high resolution electron microscopy (HREM) are natural partners for studying complex ceramic materials. Experimental HREM images are interpreted through image simulation calculations which require, as a starting point, a structure which can be provided by the atomistic simulation. Simulation of complex materials often requires the use of approximate parameterized methods. Confidence in model parameters is provided by a successful match between the simulated and the experimental image. Once the parameterization is supported, the simulation can be used to model structures associated with point defects which do not exhibit a periodic arrangement through the lattice. For example, the segregation of defects to boundaries or the partition of point defects between component structures can be investigated. Such data can then be used to interpret the influence of point defects on interface boundary contrast. In this presentation results from recent studies of the In2O3(ZnO)m system and the isovalent doped SrO/SrTiO3 system will be used to illustrate the general methodology outlined above.

11:30 AM S3.9 
EQUILIBRIUM STRUCTURE AND SCHOTTKY BARRIERS AT ERAS/GAAS INTERFACES. A. G. Petukhov, B. Hemmelman, Physics Department, South Dakota School of Mines and Technology, Rapid City, SD; W.R.L. Lambrecht, Department of Physics, Case Western Reserve University, Cleveland, OH.

ErAs/GaAs heterostructures are of great interest because of their already demonstrated use in novel electronic devices such as spin-dependent resonant tunneling diodes and metal-base transistors and the observation of giant magnetoresistance in ErAs quantum dots in GaAs. ErAs is a magnetic semimetallic compound with rocksalt structure that is closely lattice matched to GaAs. Recent experimental studies on epitaxially grown films of ErAs on GaAs substrates have shown that the Schottky barrier height strongly depends on the direction of growth. We used full-potential linear muffin-tin orbital calculations to study the structural relaxation, bonding, and Schottky barrier formation at the ErAs/GaAs (001) interface. The two structures investigated contain a common As sublattice but the GaAs is either terminated in a Ga layer or in an As layer. We find that the former has lower energy and exhibits an outward relaxation of the interplanar distance at the interface. The Fermi level at the interface is pinned by interface states which have GaAs surface dangling bond character. The electronic structure at the interface resembles closely that of the ideal surfaces and is indicative of weak bonding. Magnetic field effects on the Schottky barrier will also be discussed.

11:45 AM S3.10 
SURFACES, INTERFACES, AND FRACTURE OF GALLIUM ARSENIDE: A VARIABLE-CHARGE MOLECULAR-DYNAMICS STUDY. Aiichiro Nakano, Rajiv K. Kalia, Priya Vashishta, Concurrent Computing Laboratory for Materials Simulations, Louisiana State University, Baton Rouge, LA; Anupam Madhukar, Photonic Materials and Device Laboratory, University of Southern California, Los Angeles, CA.

A variable-charge interatomic potential model is developed for molecular dynamics (MD) simulations of GaAs surfaces and interfaces. Large-scale MD simulations are performed on parallel computers to investigate: i) stress distribution in stepped surfaces and mesa structures; ii) structural and mechanical properties of twist-bonded layers; and iii) fracture energy of various surfaces. Work supported by NSF, DOE, AFOSR, USC-LSU Multidisciplinary University Research Initiative, ARO, PRF, and Louisiana LEQSF.

Chairs: Michael William Finnis and Arthur J. Freeman 
Tuesday Afternoon, December 2, 1997 
Commonwealth (S)

1:30 PM *S4.1 
AB INITIO AND EMPIRICAL MODELLING OF A (104) MIRROR TWIN GRAIN BOUNDARY IN -Al2O3. Yin-Min Huang, Martin Exner and Mark Wilson, Max-Planck-Institut für Metallforschung, Stuttgart, GERMANY; Michael W. Finnis, Atomistic Simulation Group, School of Mathematics and Physics, The Queen's University of Belfast, Belfast, UNITED KINGDOM.

The equilibrium structure of the (104) mirror twin grain boundary in -Al2O3 (sapphire) has been calculated ab initio, using pseudopotentials with a basis of plane waves and a supercell containing 80 atoms. The results were compared with those obtained using classical interatomic potentials based on the ionic model, the shell model and the recently derived compressible ion model (CIM-DQ), which includes dipole and quadrupole polarisability of the anions. Two variants of this boundary were considered, corresponding to Al- and O-terminations, both of which have been observed experimentally. The electronic density and density of states was obtained from the ab initio calculations. There is a slight narrowing of the band gap at the grain boundary and no gap states appear. The CIM-DQ model is significantly more successful than the shell model. The charge density map reveals that the oxygen atoms at the boundary are strongly polarised.

2:00 PM S4.2 
DEFECTS IN ALUMINUM: A DENSITY FUNCTIONAL . R. Ramprasad and S. R. Atlas, Department of Physics and Astronomy and Center for Advanced Studies, University of New Mexico, Albququerque, NM.

The chemistry of point defects such as vacancies and substitutional impurities plays a crucial role in determining diffusion-dominated materials properties, particularly in the vicinity of internal interfaces such as grain boundaries. To investigate these effects, we have performed a series of self-consistent plane wave pseudopotential calculations for point defects and grain boundaries in Al. The formation energy of an isolated vacancy, the dilute heat of solution for a Cu impurity atom in bulk Al, and the and grain boundary formation energies have been determined at the local density approximation (LDA) level of theory, using large supercells and dense Brillouin zone meshes. Defects cause local changes in the steric field of the material as well as in the electronic structure, each having interesting consequences. We address the former issue by allowing lattice relaxations and determining the stress concentrations at the defect sites, and the latter by local density of states analysis.

2:15 PM S4.3 

Atomistic simulations of grain-boundary structures in body-centered cubic transition metals have revealed that angle-dependent contributions to interatomic interactions are essential (see, e.g., [1,2]). Unfortunately, the results of presently available empirical many-body potentials are not yet always sufficiently accurate or unique for quantitative theoretical predictions of grain-boundary structures, which are consistent with experimental observations, e.g., by high-resolution transmission electron microscopy. Ab-initio electronic-structure calculations based on the local density-functional theory offer the possibility to determine accurately the microscopic structures of special, high-symmetry grain boundaries, which can be further used as data bases for the improvement of empirical many-body potentials. Results of such ab-initio calculations, using a mixed-basis pseudopotential method and grain-boundary supercells, are presented for selected symmetrical tilt grain boundaries in Niobium and Molybdenum.

2:30 PM S4.4 
SIMULATION OF SYMMETRIC TILT GRAIN BOUNDARIES IN SILICON. James R. Morris, David Ring, Kai-Ming Ho, Cai-Zhuang Wang, Ames Laboratory, Ames, IA; Chong-Long Fu, Oak Ridge National Laboratory, Oak Ridge, TN.

Using a combination of classical, tight-binding and first-principles calculations, we have examined a number of different structures of tilt grain boundaries in Si, including the {510} grain boundary. This latter boundary has been observed in both Si and Ge using high-resolution electron microscopy. There are a number of possible structures, and observations in Ge suggest that the real boundaries may be mixtures of these. Our study examines the structure and energies of the competing geometries.

2:45 PM S4.5 
POTENTIAL ENERGY CALCULATION OF TWO STRUCTURES OF THE = 11<011> TILT GRAIN BOUNDARY IN SILICON AND GERMANIUM. Alain Hairie, Jun Chen, Benoît Lebouvier, Gérard Nouet and Eric Paumier, Laboratoire d'Etudes et de Recherches sur les Matériaux, UPRESA CNRS, Institut des Sciences de la Matière et du Rayonnement, Caen, FRANCE.

The relative stability at zero temperature of the two possible structures (A and B) of the - 11<011> tilt grain boundary is studied in silicon and germanium. With empirical potentials (Keating one, Baraff, Kane and Schluter one, Stillinger and Weber one) the A structure is obtained as the stable one. The results obtained with tight-binding methods are more varied. The method of Goodwin, Skinner and Pettifor applied to silicon only, and the method of Mercer and Chou confirm the A structure as the more stable in silicon. However, the latter merhod leads to an opposite result for germanium. This difference between silicon and germanium is, for the first time, in agreement with experimental observations by high resolution electron microscopy: at low temperature, the A structure has been observed in silicon and the B one in germanium. Nevertheless, the difference between silicon and germanium which are usually considered as very near each others, seems to be an interesting feature of the method of Mercer and Chou.

3:30 PM *S4.6 
MAGNETIC EFFECTS AT SURFACES AND INTERFACES (INCLUDING GRAIN BOUNDARIES). A.J. Freeman, Northwestern University, Department of Physics, Evanston, IL; R.Q. Wu, California State University, Department of Physics, Northridge, CA.

The last two decades have witnessed an explosive growth in the science and technology of thin film and interface magnetism as a result of several revolutionary developments which have advanced the study of artificial magnetic materials (surfaces, interfaces and superlattices or multilayers). As is well-known, first-principles electronic structure studies based on local spin density functional theory and performed on extremely complex simulations of ever increasingly realistic systems, play a very important role in explaining and predicting surface and interface magnetism [1]. This has led to solving even more challenging problems like the embrittlement of the Fe grain boundary [2]. Now, a major issue for first-principles theory is the treatment of the weak spin-orbit coupling (SOC) in magnetic transition metals and their alloys and its subsequent effects. A major breakthrough in eliminating the numerical randomness for the determination of the magneto-crystalline anisotropy was made with the state-tracking and torque approaches [3]. This now enables us to treat magnetostriction and its inverse effect, strain-induced magnetic anisotropy in transition metal bulk, thin films and alloys [4]. With the same SOC origin, the magneto-optical Kerr effects (MOKE) and x-ray magnetic circular dichroism (XMCD) are now directly calculated and compared with experiments. In all this work, and more recently, on the first-principles calculations of giant magneto-resistance in multilayers, extensive first-principles calculations and model analyses provide simple physical insights and guidelines to search for new magnetic recording and sensor materials. Finally, we describe our recent first-principles efforts to understand the role played by impurities (including hydrogen) and transition metal additions in the intergranular cohesion/embrittlement of the Fe grain boundary.

4:00 PM *S4.7 
IMPURITY AND MAGNETIC EFFECTS AT INTERFACES. Eric R. Roddick, Dept of Materials Science and Engineering, University of Michigan, Ann Arbor, MI.

Strong bonds can form at interfaces between metals, intermetallics, and ceramics. We have found that these bond strengths can depend sensitively on the presence of interfacial impurities as well as on magnetic effects. In turn, local magnetic properties can be significantly different at surfaces and interfaces as compared to their bulk values. Analysis of adhesive energetics and forces has been simplified by the discovery of a universal behavior. Modelling these interfaces has been complicated by the fact that a single interface can contain several different elements interacting together via all the bond types - ionic, covalent, and metallic. This means that fully self-consistent first principles computations must be done.Results of such computations will be reviewed for a number of interfaces and free surfaces.

4:30 PM S4.8 
MONTE CARLO SIMULATION OF MAGNETIZATION REVERSAL IN FE SESQUILAYERS ON W(110). M. Kolesik, SCRI, Florida State Univ, Tallahassee, FL;  M. A. Novotny, SCRI, Florida State Univ, Tallahassee, FL and Dept of Electrical Engineering, Florida A&M Univ-Florida State Univ, Tallahassee, FL; Per Arne Rikvold, CMRT, Dept of Physics, and SCRI, Florida State Univ, Tallahassee, FL.

Iron sesquilayers on W(110) are ultrathin films with coverages between one and two atomic monolayers. When grown at room temperature, they consist of a nearly ideal monolayer with compact islands of the second monolayer on top.1 They exhibit unusual mechanical and magnetic properties, including a pronounced coercivity maximum near a coverage of 1.5 atomic monolayers.2 On lattices which faithfully reproduce the morphology of the real films, a kinetic Ising model is utilized to simulate the domain-wall motion.3 Simulations reveal that the interface dynamics is dominated by the second-layer islands, which act as pinning centers. Based on nucleation-theoretical arguments, a functional form for the domain-wall velocity vs. field is proposed. Together with the simulation data, this form is used to calculate the coercivity of the film. The simulated dependencies of the coercivity on the film coverage, as well as on the temperature and the frequency of the applied field, are very similar to those measured in experiments. Unlike previous micromagnetic models, the presented approach provides insight into the dynamics of the domain-wall motion and clearly reveals the role of thermal fluctuations.

Chairs: Paul D. Bristowe and Simon R. Phillpot 
Tuesday Evening, December 2, 1997 
8:00 P.M. 
Grand Ballroom (S)

MICROSTRUCTURAL CHARACTERIZATION OF FERROELECTRIC Pb(Sc1/2Ta1/2)O3 CERAMICS BY FRACTAL GEOMETRY. Z. X. Xiong, K. Z. Baba-Kishi and F. G. Shin, Department of Applied Physics and Materials Research Center, The Hong Kong Polytechnic University, HONG KONG.

Following Mandelbrot's fractal theory, the irregular characteristics of the microstructural features of ferroelectric Pb(Sc1/2Ta1/2)O3 ceramics, including grain boundaries and dislocation networks, were investigated. The microstructural features were imaged by electron microscopy. The fractal analyses were carried out manually and by image processing techniques, which show the value of the fractal dimension, D, varies according to the regularity of the microstructure. The value of D close to unity is an indication of increasing degree of microstructural regularity, which is in agreement with the simulated results. The temperature dependence of the dielectric constant of the ceramics was also measured. It was found that the fractal dimension, D, can be correlated with the microstructure, and consequently with the dielectric property of the ceramic.

MICROSCOPIC INVESTIGATION OF SURFACE REGION IN FERROELECTRIC OXIDES. Sergey Prosandeyev, Alexander Maslennikov, Physics Department, Rostov State University, Rostov on Don, RUSSIA.

A Green-function computation has been carried out to give a microscopic description of the polarization behavior in the surface region of ferroelectric oxides. The Green-functions were constructed on the basis of the APW-calculations with the help of the tight-binding method. To compute the polarization an expression for calculating the Lorentz matrix for a polarized slab was derived. The influence of the polaronic effect on the surface region width is also uder discussion.

THE EFFECT OF THE INTERFACE ON THE OVERALL DC CONDUCTIVITY OF MULTI-PHASE COMPOSITES. Robert Lipton, Department of Mathematical Sciences, Worcester Polytechnic Institute, Worcester, MA.

We consider two different models of electrical transport across phase interfaces: the first model represents an electrical contact resistance, the second model allows for conduction along the interface. We apply these models to show how the particle size distribution influences the overall conductivity of the composite. Rigorous rules of thumb are presented for the design of multi-phase composites with optimal conductivity properties.

SIMULATIONS OF THE INTERPHASE BOUNDARY STABILITY IN NON-CRYSTALLINE MATERIALS. Tatiana Ischenko, Sergey Demishev, General Physics Inst RAS, Submillimetre Spectroscopy Dept, Moscow, RUSSIA.

We report our results on the theoretical investigations of the interphase boundary stability in a mixture of stable and metastable phases. It is well known that under definite conditions the process of the phase transition and interphase boundary movement may be strongly non-equilibrium. For example, athermal fast process may be induced be laser radiation in amorphous films and then rapid laser driven recrystallisation occurs. The microscopic mechanism of these phenomena is analysed in detail. According to suggested approach the non-equilibrium phase transition goes through intermediate excited states. Excited states are described as high-energy vibrations of disordered network localised at a scale L 1 nm corresponding to the medium range order scale and having a lifetime 0.1-1 ns. Localisation of the phonon modes leads to a stronger coupling with electrons and consequently work as an extra source of excitation of the electronic subsystem and as an origin of an anomalous diffusion regime, which results in an enhancement of the concentrational diffusivity. The localised phonon states store the energy on the interphase boundary for the non-equilibrium regime and control the heat transfer process, that leads to the renormalization of the thermal diffusivity. Hence the excited states become the necessary condition for the appearance of the self- sustaining movement of the interphase boundary in amorphous materials. The simple computer models is developed based on the mentioned ideas. The predicted threshold characteristics of the interphase boundary movement are found to be consistent with the existing experimental data. The possibility of stochastic regime of the interphase boundary movement is discovered: when the life-time of the excited states exceeds some critical value, the interphase boundary acquires the beam-like or fractal-like structure. Finally, the possible applications of the model to the other problems of amorphous films are discussed.

THEORETICAL ANALYSIS OF THE LOCAL ELECTRONIC STRUCTURE AT METAL/CERAMIC INTERFACES. Sibylle Köstlmeier, Christian Elsässer, Bernd Meyer, Max-Planck-Institut für Metallforschung, Stuttgart, GERMANY.

Local electronic structures in a set of ionic materials (MgO, Al2O3, MgAl2O4) are analyzed in terms of site- and angular-momentum-projected densities of states, which are obtained from self-consistent ab-initio band structure calculations (c.f. [1]). The occupied valence states can be analyzed by a projection on an atomic orbital basis (c.f. [2]) in order to characterize the chemical bonding types. The unoccupied conduction states can be compared to energy-loss near-edge spectra (ELNES), which are measured by analytical electron microscopy with high spatial resolution. These concepts are applied to the analysis of the chemical bonding at an atomically sharp coherent heterophase interface Al/MgAl2O4.

TOTAL ENERGY DIFFERENCES BETWEEN SILICON CARBIDE POLYTYPES AND THEIR IMPLICATIONS FOR CRYSTAL GROWTH. Sukit Limpijumnong and Walter R. L. Lambrecht, Dept of Physics, Case Western Reserve University, Cleveland, OH.

Silicon carbide polytypes are periodic arrangements of twin boundaries in an otherwise cubic or hexagonal lattice. Their energy differences are of the order of meV/atom. We have carried out well converged calculations of the energy differences for the 3C, 6H, 4H, 2H, 9R and 15R polytypes using the full-potential linear muffin-tin orbital method and the generalized gradient approximation to the exchange-correlation density functional. From these results, we extract parameters for a generalized anisotropic next nearest neighbor Ising model proposed by Heine et al. in the late 80's and written as , in which (anti-)parallel spins indicate (hexagonal) cubic stacking. We find that the energy of formation of a twin, 2(J1+2J2)<0, with J1>0 indicating that many twins in an otherwise cubic lattice are favorable. This explains why polytypes with consecutive bands of 2 and 3 paralallel spins have the lowest energies even though the results are not close to the ``multi-phase degeneracy point'' J1=-2J2, and why polytypes of high hexagonality are unfavorable. Unlike other recent calculations, we find J1>|J2|, which has consequences for stabilization of cubic material in epitaxial growth. We find that -J1/J3 is 4 or 9 depending on whether or not the K term is included and that the inclusion of the latter significantly improves the range of applicability of the model. We discuss the relevance of these parameters in epitaxial growth taking into account the island sizes for which the energies become comparable to kT at the growth temperature and the large estimated value of lateral versus interlayer spin interactions. Supported by NSF-95-29376.

CONTROLLED WETTING ON HETEROGENEOUS INTERFACES. Mariela Araujo and Y. Carolina Araujo, Reservoir Department, INTEVEP, Caracas, VENEZUELA.

We present an experimental and numerical study of the influence of surface wettability defects on the behavior of a liquid front advancing on a disordered substrate, a modified Hele-Shaw cell. A liquid-gas interface moves in a very thin rectangular cell of a characteristic thickness. For the experiments, defects are made by deposition of silane and ink droplets on one of the glass surfaces of the cell. Their strength is determined by using atomic force microscopy and the characterization also include measurements of local advancing and receding contact angles. Fluid is injected at constant rate from one of the cell's edges. The concentration of each defect type allows to control wetting properties of the surface. The advance of the front is registered as a function of time and the roughness features of the interface are calculated. In the simulation, the experimental conditions are closely reproduced by using a discrete two dimensional lattice model in which local rules for the motion of the interface are included to mimic the cooperative effect of neighbouring sites. The evolution of the interface roughness, its amplitude and scaling properties are studied as a function of the density of defects. The classical and interface exponents are measured and compared with values reported in the literature.

DIFFUSION OF Si ADATOMS ON H-TERMINATED Si(001) SURFACES. Takahisa Ohno, Jun Nara, Taizo Sasaki, National Research Institute for Metals, Ibaraki, JAPAN.

Hydrogen termination has attracted considerable interest because of the possibility of changing the growth mode of epitaxial semiconductor films. Adsorption of hydrogen can alter the surface properties such as the surface energy, the adatom diffusion, and the nucleation processes, and result in the modification of the morphology of the epitaxial films. Recently, Copel and Tromp have reported that the Si homoepitaxial growth is disrupted on the hydrogenated Si(001) surface, and that H atoms segregate from the Si(001) surface [1]. On the other hand, Ogitsu et al. [2] claimed from their ab initio calculations that H atoms cannot segregate to the surface. The role of surface H atoms in the Si homoepitaxy has not been understood satisfactorily at the microscopic level. In this paper, we theoretically investigate the Si adsorption on the monohydride terminated Si(001)-(2x1) surface by using first-principles total-energy calculation techniques. We find that the Si adatom spontaneously segregates one H atom from a surface Si dimer during adsorption, and further captures the remaining H atom of the same Si dimer during surface migration, leading to the most stable adsorption geometry. The migration of the Si adatom is assisted by the mobility of H atoms, being reduced compared with that on the bare Si surface. The reduction of the Si diffusion may have disruptive effects on the Si homoepitaxy, together with the stable Si addimer structures being not at epitaxial growth sites.


A theoretical atomic process model of the interfacial oxidation reaction on a silicon substrate is proposed based on first-principles calculation. The model is based on a plausible assumption that the remaining stress in the oxidized region is kept minimum without breaking the grown Si-O-Si network when the oxidation proceeds. As the natural consequence, Si atoms are emitted from the interface as the oxidation proceeds to release the stress due to substitution of Si-Si bonds by Si-O-Si bonds. This emission is consistent with well-known experiments of oxidation induced stacking faults and oxidation enhanced diffusion. This model reveals that oxidation proceeds atomically with the following processes: (1) O atoms cut into interfacial Si-Si bonds. (2) Stress due to the formation of Si-O-Si bonds accumulates in the neighboring Si-Si bonds. (3) Si atoms with highly strained Si-Si bonds are emitted from the interface to release the accumulated stress. (4) Remaining dangling bonds due to Si-Si bond breaking are terminated by O atoms by forming Si-O-Si bonds. If the strain is completely released only by the emission, it is estimated that approximately one Si atom is emitted from the interface when six Si atoms are oxidized. Moreover, these processes can depend on the shape of the system and Fermi energy position as well as local strain. This model presents a microscopic approach to understand the silicon oxidation process, whereas widely accepted Deal-Grove model presents a macroscopic one.

THE PINNING PATHS OF AN ELASTIC INTERFACE. Hernan A. Makse, Sergey Buldyrev, Heiko Leschhorn, and H. Eugene Stanley, Center for Polymer Studies and Dept. of Physics, Boston University, Boston, Massachusetts.

We introduce a model describing the paths that pin an elastic interface moving in a disordered medium. We find that the scaling properties of these ``elastic pinning paths'' (EPP) are different from paths embedded on a directed percolation cluster, which are known to pin the interface of the ``directed percolation depinning'' class of surface growth models. The EPP are characterized by a roughness exponent , intermediate between that of the free inertial process () and the diode-resistor problem on a Cayley tree (). We also calculate numerically the mean cluster size and the cluster size distribution for the EPP.

ETCHING PROCESS OF Al OXIDE ON Si SURFACE WITH HF TREATMENT. T. Hoshino, Faculty of Pharm. Sci., Chiba Univ., JAPAN; Y. Nishioka, Texas Instruments, Tukuba R&D Center.

For the purpose of development of the cleaning technique of semiconductor surfaces, ab initio theoretical calculations have been performed to obtain atomic-scale understanding of the reaction mechanism of HF etching of AlO metal oxide on Si surfaces. The following two-step process has been revealed to be the most probable reaction path. HF + Al-O-Si4H9 H-O-Si4H9 + Al-F (1) HF + H-O-Si4H9 F-Si4H9 + H2O (2) Both of the reactions are exothermic by 0.3 and 1.1eV, respectively, and also show a chemical complex by the adsorption of HF molecule. The energy barrier in each reaction is 0.8 and 0.9eV. Another etching reaction which proceeds with one-step process HF + Al-O-Si4H9 F-Si4H9 + Al-O-H (3) requires a larger activation energy. A part of this work was performed under the management of ASET in the MITI's R&D Program supported by NEDO.

MD-SIMULATION OF INTERFACE EVOLUTION IN INDENTATION, ATHERMIC SLIDING AND FRICTION OF METALLIC NANOPARTICLES. Vladimir Pokropivny, Valery Skorokhod, Aleksy Pokropivny, Institute for Problems of Material Science, National Academy of Science, Kiev, UKRAINE.

The processes of nanoindentation, stretch and shock of bcc-Fe-asperity with (100) and (114) bcc-Fe-surface are computer simulated at atomic level by molecular dynamics technique. Atomistic mechanisms of contact creation during push indentation and contact break during following pull stretch are investigated. Correlation between structure transformation and variation of adhesive bonds number,viscosity of contact, energy and force of adhesion is analyzed. Crowdion mechanism of plastic deformation under indentation is confirmed. Adhesive model for SD-effect is developed. Athermic shear of special tilt boundary is simulated. Atomistic mechanism is studied of shear energy to transform in hypersonic phonon vibrations. Effect of "phonon skidding" is advanced. During sliding the epitaxial mechanism of accompanied boundary migration is observed as well as the formation of number of metastable states in all of which a grain is stuck and braked. Mechanism and elementary act of cohesion friction and wear of atomic-sharp W-roughness over bcc-Fe-(114) -surface were simulated. Correlation of calculated energy and force of adhesion, viscosity, force of internal and external friction with transformations of atomic structure was investigated during friction in constant height mode. In constant force mode both the mechanisms of seizure, exhibited a plastic deformation of slider under great load,and the atomic-force microscope images of surface roughness under small load were examined. Force criteria for wear, seizure and fracture of asperity during friction were suggested.

SPIN-POLARISED DENSITY OF STATES AND ELECTRON TUNNELING FROM THE Co/Al2O3 INTERFACE. D. Nguyen-Manh, E.Yu. Tsymbal, D.G. Pettifor, Oxford University, Department of Materials, Oxford, UNITED KINGDOM; C. Arcangeli, R. Tank and O.K. Andersen, Max-Plank-Institut fur Festkorperforschung, Stuttgart, GERMANY.

In order to elucidate the mechanism of tunneling magnetoresistance in the cobalt-alumina system, spin-polarised electronic structure calculations of Co/Al2O3 interface have been performed self-consistently with and without gradient correction using a new LMTO technique. The system was represented by a supercell consisting of 3 monolayers of hcp Co (0001) and 7 monolayers of Al2O3 (0001) terminated by Al layers. The position of the Co atoms with respect to the Al atoms was chosen to minimise the lattice mismatch between Al2O3 and Co. Since the results of the calculations are very sensitive to the distance between the Co and Al planes, we have minimised the total energy with respect to this distance. It is found that the minimum of energy is reached at the distance of about 2.3A. Our calculations show that at the Fermi energy a strong bonding between the 3d-electrons of Co with the sp-electrons of Al at the interface can have an important influence on spin polarisation of the local density of states (LDOS) of inner Al and O layers. Since the Fermi energy is situated inside the minority-spin d-band of Co but above the majority-spin d-band, the sp-d bonding results in a smaller LDOS of the minority-spin electrons of interfacial Al layers in comparison to the LDOS of the majority-spin electrons at the Fermi energy. This asymmetry in the LDOS extends to the inner Al2O3 layers implying a positive spin polarisation of the tunneling density of states. This result is coherent with experimental observations on tunnelling from cobalt through alumina where positive values of the spin polarisation of the tunnelling current were measured.

H2S ADSORPTION ON A CHARGED CU SURFACE IN AN ELECTROLYTE: EFFECT OF IONIC STRENGTH. W. D. Wilson, Sandia National Laboratories, Livermore, CA; C. M. Schaldach, U.C. Berkeley, Berkeley, CA.

We have calculated the minimum energy configurations of a neutral H2S molecule impinging upon a charged Cu <100> surface in the presence of an electrolyte. Partial charges for the atoms in the molecule are obtained from quantum chemical calculations. A molecular surface is constructed surrounding the H2S molecule and a discretized Cu surface (slab) forming boundary elements. The linearized Poisson-Boltzmann equation is then solved for the induced surface charges and surface potentials using a Green's function method. Real surface charge is applied to the surface elements reflecting the pH of the solution. Repulsive interactions between the atoms of the molecule and those of the surface are obtained using a rigid-ion Hartree-Fock method. The minimum energy configuration of the H2S molecule is determined as a function of its distance and orientation relative to the surface. Binding energies of the molecule to the surface are determined as a function of the real surface charge imposed and also the ionic strength of the solution. It is found that small surface charges can completely (180ƒ) reorient the H2S molecule and that the ionic strength of the solution can completely screen binding effects of even large surface charges.


A homogeneous two-dimensional (2D) interacting electron system in a uniform perpendicular magnetic field is studied by a molecular dynamics simulation method. The calculations were done for 256 particles in a rectangular cell with periodic boundary conditions, for a range of values of the applied magnetic field from 0 to 10 T. A predictor-corrector method involving up to five time derivatives of the position was used to integrate Newton's Equations of motion for the interacting system. In this simulation, we include the effect of magnetic field classically through the Lorentz force. Both the Coulomb interaction and the magnetic field are included directly in the electron dynamics to study their combined effect on the transport properties of the 2D systems. We will present the results for the pair correlation function, the static structure factor, the mean square displacement and the density correlation function, in an external magnetic field Our simulation results show that the magnetic field has a significant effect on the dynamics of the system.

EPITAXIALGROWTH MECHANISM OF ZnO ULTRAVIOLET-LASER-EMITTING MATERIAL AS INVESTIGATED BY MOLECULAR DYNAMICS. Momoji Kubo, Yasunori Oumi, Ryuji Miura, Hiromitsu Takaba, Abhijit Chatterjee, Akira Miyamoto, Graduate School of Eng., Tohoku Univ., Sendai, JAPAN; Masashi Kawasaki, Interdisciplinary Graduate School of Engineering, Tokyo Inst. Technol., Yokohama, JAPAN; Mamoru Yoshimoto, Materials and Structures Lab., Tokyo Inst. Technol., Yokohama, JAPAN; Hideomi Koinuma, Materials and Structures Lab., Tokyo Inst. Technol. and CREST-JST, Yokohama, JAPAN.

Recently Koinuma et al. fabricated ZnO quantum dots on alpha-Al2O3(0001) and discovered that the above material emits ultraviolet laser. In order to control the atomic structure of the ZnO quantum dots, the crystal growth mechanism of the quantum dots was studied. In the present study, we developed a novel molecular dynamics code MOMODY to simulate the epitaxial growth process of metal oxide surfaces and applied it to the homoepitaxial growth process of ZnO(0001). Totally 10 ZnO molecules were deposited on flat ZnO(0001) one by one with definite velocities at 700 K. All deposited ZnO molecules were faced repulsion from the surface. It indicates that the crystal growth does not easily occur on the terrace of the surface. Afterwards, 10 ZnO molecules were deposited on the ZnO(0001) with a step. The crystal growth suddenly occurred at the step of the surface, which is a totally different phenomenon with that occurring on the flat ZnO(0001). Such crystal growth mechanism was explained in terms of the stability of the deposited ZnO molecules on the terrace and step of the surfaces The deposited ZnO molecules on the step were much more stabilized in comparison to that on the terrace; this justifies the fact that the ZnO molecules on the step did not evaporate from the surface. The above epitaxial growth process of the ZnO(OOOl ) was much different from that of MgO(001) 1). It was observed that such difference was due to the different surface structures of ZnO(0001) and Mg0(001).

THEORETICAL STUDY ON THE OXIDATION OF SI(III) SURFACE: DENSITY FUNCTIONAL APROACH. Akira Endou, Ryo Yamauchi, Momoji Kubo, Abhijit Chatterjee, Akira Miyamoto, Graduate School of Eng., Tohoku Univ., Sendai,JAPAN; András Stirling, Institute of Isotopes, Hungarian Academy of Sciences, Budapest, HUNGARY; Ewa Broclawik, Institute of Catalysis and Surface Chemistry, Polish Academy of Sciences, Cracow, POLAND; Kazutaka G. Nakamura, Masahiro Kitajima, National Research Institute for Metals, Tsukuba, JAPAN.

In the oxidation process of Si, formation and desorption of SiO takes place at the surface temperature of more than 900K. Nakamura et al. [1] have reported the rotatinoal-vibrational distribution state of the desorbing SiO molecule. In this study, we investigated the formation and desorption process of SiO using density functional (DF) calculations. We used the Si4H9 cluster model as a representative of Si (111) surface. To describe the oxidation state of silicon surface, the adsorption state of the oxygen atom lying at the top of the Si surface was optimized. Further the energy profile for SiO was obtained by optimizing SiO molecule in terms of the angle between Si-O bond axis and a plane which includes 3 other silicon atoms considered as bulk. It was found that the bridge adsorption states (Si-O-Si) was 16.7 kcal/mol (NLSD level) more stable than the adsopton state for the position of oxygen at top of the surface. As for the desorption step of SiO, two models from both of the adsorption states were studied. Energy profiles shows smooth desorption paths. The activation energies for both desorption models were 79.6 and 96.3 kcal/mol (NLSD level), respectively. They showed a qualitative agreement with the experimental result.

MECHANISM FOR BREAKAGE OF Si-O NETWORKS OF SiO2 FILMS IN HF SOLUTIONS. Tomoki Oku, Ryo Hattori and Kazuhiko Sato, Mitsubishi Electric Corporation, Optoelectronic and Microwave Devices Laboratory, Hyogo-Prefecture, JAPAN.

Abstract not available.

8:30 AM *S6.1 
QUASICONTINUUM MODELS OF INTERFACIAL DEFORMATION. Rob Phillips, V. Shenoy, Brown University, Div. of Engineering, Providence, RI; E. Tadmor, R. Miller, Harvard University, Cambridge, MA; M. Ortiz, California Institute of Technology, Pasadena, CA.

Mixed atomistic and continuum methods offer the possibility of explicit simulation of interfacial processes such as the interaction of dislocations and cracks with grain boundaries. This talk will review recent progress in using the quasicontinuum method to examine interfacial structure and deformation. In particular, interaction of dislocations with grain boundaries will be examined within the setting of a nanoindentation simulation. The second key example will be that of cracks interacting with a grain boundary with special reference to the stress induced motion of a grain boundary.

9:00 AM *S6.2 

At the continuum level, a crack on an elastically mismatched interface gives rise to a singular mixing of modes at the tip. We show that this singularity never causes crack closure, but that it can lead to dislocation shear breakdown and dislocation emission. We discuss the problem of cleavage and emission in the face of this added complexity for the interface crack, and propose a diagramatic criterion for crack behavior which includes all the varieties of crack behavior: cleavage, emission, and branching. The criteria for these various types of behavior is couched in terms of the external load and the form of the force laws in the material. Localized chemical effects at the crack tip are also included when chemical size effects are not important. The results are demonstrated for a very simple generic force law and lattice geometry.

10:00 AM *S6.3 
INTERFACIAL PROPERTIES AND MECHANICAL BEHAVIOR OF TITANIUM ALUMINIDES. M. H. Yoo and C. L. Fu, Oak Ridge National Laboratory, Metals & Ceramics Div, Oak Ridge, TN.

The role of various interfaces in deformation and fracture behavior of two-phase TiAl-Ti3Al alloys is analyzed on the basis of the specific interfacial and surface energies determined from ab initio calculations. The propensity of twinning observed in these alloys is consistent with the calculated low energies of coherent true-twin and pseudo-twin boundaries. The strong plastic anisotropy reported in TiAl polysynthetically twinned (PST) crystals is attributed partly to the localized slip along lamellar interfaces, thus lowering the yield stress for soft orientations. Interfacial fracture (ideal) energies can be estimated from the calculated interfacial and cleavage energies and the energy released by interfacial dislocations. The mode mixity plays an important role in the crack-tip plasticity by ordinary slip and true-twinning, leading to translamellar and interfacial fracture. Available experimental data on environmental embrittlement are discussed in view of the effects of hydrogen on cohesive energies and electronic structure.

10:30 AM S6.4 
MIXED ATOMISTIC/CONTINUUM SIMULATION OF THE INTERACTIONS BETWEEN GRAIN BOUNDARIES AND BRITTLE CRACKS. Ron Miller, Rob Phillips, Vijay Shenoy, and David Rodney, Brown University, Division of Engineering; Ellad Tadmor, Harvard University, Department of Physics; Michael Ortiz, Caltech, Graduate Aeronautical Laboratories.

Grain boundaries (GBs) play an important role in the fracture of polycrystalline solids. For instance, whether a material fails by intergranular or transgranular fracture depends on the toughness of the GBs relative to the bulk crystals. As well, it has been observed that the macroscopic toughness of a material depends on the grain size - and hence on the number of GBs present in the sample. Any modeling effort which aims to describe the complex interactions between cracks and GBs must be able to span the wide range of length scales inherent in the problem. On the one hand, realistic simulation of the loading of a crack requires a model with lengths on the micron scale. On the other hand, Angstrom scale resolution is critical in describing the atomistic details of the GB structure and of the advancing crack tip. Using the recently developed Quasicontinuum Method of Tadmor, Ortiz and Phillips, we model cracks approaching a number of different tilt boundaries in fcc nickel. The simulation technique is a mixed atomistic (embedded atom method) and continuum formulation, which allows for realistic boundary conditions and system sizes while at the same time capturing important atomic scale effects near the crack tip. First, we demonstrate that the quasicontinuum method will correctly model the details of atomic level GB structure, with results that are virtually indistinguishable from full lattice statics simulations. We then simulate the GB/crack interaction by loading a pre-existing crack in a bicrystal. The simulations elucidate a number of different deformation mechanisms, depending on the GB being studied. GB migration due to the stress field around the crack tip, dislocation emission from the GB, GB sliding, and intergranular fracture are all found to be important deformation mechanisms.

10:45 AM S6.5 
ATOMISTIC SIMULATION OF 3 (111) GRAIN BOUNDARY FRACTURE IN TUNGSTEN CONTAINING VARIOUS IMPURITIES. M. Grujicic, H. Zhao, Clemson Univ., Dept. of Mechanical Engineering, Clemson, SC; Genrich L. Krasko, U.S. Army Research Laboratory, Materials Division, Aberdeen Proving Ground, MD.

The effect of various impurities and micro-alloying elements (B, C, N, C, O, Al, Si, S, and P) on the intrinsic resistance of the 3 (111) grain boundary (GB) in W, has been investigated using Molecular Dynamics simulation. The Finnis-Sinclair many-body potential for W. and the so-called ``environment-sensitive embedding energies''1 for the impurities were used. The fracture resistance of the GB has been characterized by computing, in each case, the ideal work of GB separation, the Mode I stress intensity factor and the Eshelby's F1 conservation integral at the onset of crack propagation. The results obtained suggest that pure W is relatively resistant towards GB decohesion, this resistance is further enhanced by the presence of B. C and N. On the other hand, O. Al and Si have a relatively minor effect on the cohesion strength of the GB. In sharp contrast, S and P greatly reduce this strength thus enhancing the W brittleness. These results have been correlated with the effect of the impurity atoms on the material evolution at the crack tip.

11:00 AM *S6.6 
COMPARISON OF ATOMISTIC AND CONTINUUM-MECHANICS MODELLING OF GRAIN-BOUNDARY FRACTURE. F. Cleria,b, D. Wolfa and S.R. Phillpota, aMaterials Science Division, Argonne National Laboratory, Argonne, IL; bDivisione Materiali Avanzati, ENEA, Centro icerche Casaccia, Roma, ITALY.

Recent results of atomic-level simulations of dislocation emission from a crack tip in a perfect fcc crystal suggest a strategy to improve the macroscopic-scale modelling of such processes. By extracting stress displacement relations to be used as boundary conditions into the Peierls Nabarro model, the full atomic-scale complexity of the dislocation nucleation process can be accounted for. Moreover, other important atomic-scale features such as the surface step contribution (the so called Œ¹ledge effect¹¹) are naturally included. The new predictive capability of which continuum models are endowed by such atomic-level approach is tested in a realistic case, by successfully predicting the direction-dependent fracture behavior of symmetric-tilt grain-boundaries in fcc metals starting from single-crystal atomistic inputs. As a further check, direct atomistic simulations of a microcrack in the interface plane of a symmetric-tilt grain boundary in fcc Cu are performed. Direction dependent brittle-ductile response is observed, in agreement with both the experimental observations and the continuum-model predictions. However, detailed analysis of the atomistic simulations reveals the importance of several effects, which are currently neglected even in the atomistically-modified Peierls-Nabarro model, such as the shielding of the stress field by the emitted dislocations and the role of the excess interfacial stress in modifying the critical loads for ductile and brittle fracture. Further improvements of continuum models are thus suggested.

11:30 AM S6.7 
COMPUTER SIMULATION OF SLIP IN TITANIUM CARBIDE. R.M. Harris and P.D. Bristowe, Cambridge University, Department of Materials Science and Metallurgy, Cambridge, UNITED KINGDOM.

At least two modes of slip are observed in TiCx and are a function of temperature and composition. For nearly stoichiometric material (x > 0.75), {110} slip is observed at low temperatures while above the brittle-to-ductile transition temperature (800C), {111} slip is preferred. In this paper we have investigated the energetics of {110} and {111} slip in stoichiometric TiC using an orthogonal tight binding model to calculate interatomic forces and crystal binding energies. Five different slip systems have been considered including three on {111} which involve the dissociation of perfect lattice dislocations into partials. By calculating the appropriate cross-section through the -surface, the critical resolved shear stress for each system has been determined. The likely slip system active at low temperatures is confirmed to be 1/2[10](110) and it is suggested that the origin of the brittle-to-ductile transition in this material is a result of a change in slip system permitted by increased carbon vacancy diffusion.

11:45 AM S6.8 
MOLECULAR DYNAMICS STUDY OF Si/Si3N4 INTERFACE AND Si/Si/Si3N4 MESA*. Martina Bachlechner, Andrey Omeltchenko, Kenji Tsuruta, Aiichiro Nakano, Rajiv K, Kalia, and Priya Vashishta, Concurrent Computing Laboratory for Materials Simulations, Dept of Physics and Astronomy, Dept of Computer Science, Louisisana State Univ, Baton Rouge, LA; Ingvar Ebbsjo, Studsvik Neutron Research Lab, Univ of Uppsala, SWEDEN; Anupam Madhukar, Dept of Materials Science and Engineering, Univ of Southern California, CA.

The interface structure, stress distributions, crack propagation and fracture in a Si3N4 film on Si substrate are studied using molecular dynamics simulations on parallel computers. Bulk Si is described by Stillinger-Weber potential and Si3N4 is represented by a combination of two- and three-body interactions, which include charge transfer, polarizability and covalent effects. At the interface, the charge transfer is taken from LCAO electronic structure calculations1. MD simulations are performed to study structural correlations in the intefacial region and stress distribution in a Si/Si/Si3N4 mesa. Results for crack propagation and fracture will also be presented.

Chairs: Jeffrey M. Rickman and Dieter Wolf 
Wednesday Afternoon, December 3, 1997 
Commonwealth (S)

1:30 PM *S7.1 
MOLECULAR-DYNAMICS SIMULATION OF GRAIN-BOUNDARY MIGRATION. B. Schoenfelder1,2, D. Wolf1, S.R. Phillpot1 and M. Furtkamp(1,2(1,2%%(1,2,^1Materials Science Division, Argonne National Laboratory,^2^-9s

typically accessible by molecular-dynamics simulation. For this model high-angle grain boundary we demonstrate that (a) the drift velocity is, indeed, proportional to the applied driving force thus enabling us to determine the boundary mobility, (b) the activation energy for grain-boundary migration is distinctly lower than that for grain-boundary self-diffusion or even self-diffusion in the melt and (c) in agreement with earlier simulations the migration mechanism involves the collective reshuffling during local disordering (Œ¹melting¹¹) of small groups of atoms and subsequent resolidification onto the other crystal.

2:00 PM S7.2 

We employ a molecular dynamics simulation technique to observe constant curvature, steady-state grain boundary migration using a U-shaped half-loop grain geometry. Both special, near special and general grain boundaries are simulated and their migration kinetics compared. Examination of the atomic motions reveal several basic features of the grain boundary migration mechanism. The migration mechanism can be described as thermally activated single atom hops across the grain boundary, followed by correlated atomic shuffles. This model is applied to understand the misorientation dependence of boundary migration. Experimental and simulation evidence suggest that grain boundary migration may be described by an Arrhenius form. Several experiments show that when the activation energy increases (e.g., due to misorientation change), the pre-exponential factor also tends to rise. This has become known as the 'compensation effect.' We perform a series of simulations on several grain boundaries to verify the existence of this effect and identify the temperature at which the mobilities of all of the grain boundaries are equal (i.e., the compensation temperature Tc). The observed compensation behavior is rationalized in terms of the boundary migration mechanism.

2:15 PM S7.3 
AB INITIO CALCULATIONS OF THE STRENGTH OF GRAIN BOUNDARIES IN SiC. Masanori Kohyama, Dept. of Material Phys., Osaka National Research Institute, AIST, Ikeda, Osaka, JAPAN.

It is of great importance to study the structure and properties of grain boundaries in SiC, because grain boundaries often dominate mechanical and electronic properties of SiC ceramics. Recently, we have investigated the microscopic structure and various properties of coincidence tilt boundaries in cubic SiC by using the first-principles molecular dynamics method [1]. In this paper, we present our recent results on the mechanical properties obtained by the ``tensile test'' using ab initio calculations. We deal with the {122} = 9 boundary. 64-atom supercells of two kinds of polar and one kind of non-polar interfaces are constructed by inverting the polarity of the grains. Conjugate gradient techniques [2] and optimized pseudopotentials [3] are used. Stable fourfold-coordinated configurations contain either C-C or Si-Si wrong bonds for the polar interfaces, and both kinds of wrong bonds for the non-polar interface. The behavior of such wrong bonds and other normal bonds under stresses has been examined. We have found that the wrong bonds have significant effects on the strength and mechanical properties of grain boundaries in SiC.

2:30 PM S7.4 
FRICTION BETWEEN BARE AND HYDROGENATED SILICON SURFACES. Mark O. Robbins, Johns Hopkins Univ, Dept Physics and Astronomy, Baltimore, MD; Raymond D. Mountain, National Institute of Standards and Technology, Gaithersburg, MD.

We have performed molecular dynamics simulations of the friction between bare and hydrogen terminated silicon surfaces using forces calculated from a tight-binding Hamiltonian. We first fixed the atomic positions and examined the normal and lateral forces as two (111) or (100) surfaces were brought together. If one or both surfaces was terminated with hydrogen, there was a small (0.1 to 0.2 GPa) adhesive pressure. At larger pressures the lateral forces scaled linearly with the applied load. The slope gives the static friction coefficient .When atoms on the surfaces were allowed to relax, the value of dropped roughly in half. For two hydrogenated surfaces .Removing one hydrogen layer doubles .Moreover, the remaining layer is much less stable. Increasing load, speed or temperature leads to rupture of the layer and direct silicon bonding across the surface. The lateral forces needed to shear the resulting junctions are comparable to the ideal shear strength of silicon. The sliding or kinetic friction was negligible except at high velocity (> 200 m/s) or pressure (> 10 GPa). The dependence of the results on the mechanical properties of the device that moves the surfaces will be discussed, and results for small AFM-like tips will be presented.

2:45 PM S7.5 
MONTE CARLO MODELING OF THE KIRKENDALL EFFECT. Michael V. Yarmolenko, Institute of Engineering and Technology, Dept of Physics, Cherkasy, UKRAINE.

A computer program was worked out to model the Kirkendall effect during mutual diffusion between two solid substances by Monte Carlo calculations. It takes into account the random vacancy jumps in a cubic lattice, appearance and disappearance of vacancies due to climbing of dislocations to be parallel to the initial interface, vacancy quasi equilibrium in all points of each substance, different frequency of exchange of vacancy and the atoms of different kinds, and external stress gradient. We solved a random-walk vacancy problem with appropriate boundary conditions. The main assumption is pB/pA=const>1. Here PA (PB) is the probability that an A-atom (a B-atom) will jump into any given neighboring vacant site. Monte Carlo modeling shows that a) the equilibrium vacancy concentration in substance B is greater than that in substance A; b) the net vacancy flux is directed into substance B; c) the shift of an inert marker is directed m the same direction and is proportional to the square root of time of diffusion; and d) the time rate of change of the shift decreases exponentially with distance increase from the initial interface. The results of computer modeling agree with the real experimental data, and we can conclude that the modeling describes correctly this very interesting phenomenon.

3:30 PM *S7.6 
EXTENDED DEFECTS IN MATERIALS: MICROSTRUCTURE AND KINETICS. J.M. Rickman, Lehigh University, Department of Materials Science and Engineering, Bethlehem, PA.

The modeling of defect structures and the impact of extended defects, such as grain boundaries, on transport phenomena and phase formation are discussed. In particular, we examine the impact of heterogeneous boundary nucleation on the microstructure of product grains generated during a phase transformation in order to relate defect and product grain length scales. In addition, we discuss spatial and temporal correlations associated with grain boundary diffusion in terms of a lattice gas model. Finally, we consider both discrete and spatially coarse-grained models of overdamped dislocation motion which properly incorporate long-ranged, Peach Koehler interactions in order to describe the resulting dislocation microstructures.

4:00 PM S7.7 
MICROSTRUCTURE EVOLUTION IN THE BCC-HCP PHASE TRANSITION OF ZIRCONIUM. G.J.Ackland and Udomsilp Pinsook, Department of Physics, University of Edinburgh, Edinburgh, UNITED KINGDOM.

We present computer simulations of the martensitic pahase transformation between bcc and hcp phases of zirconium, described by a many body potential. We show that the low symmetry (hcp) phase is characterised by laminates separated by twin boundaries, and that stacking faults and plastic deformation occur in such a way as to rotate the hcp crystalites to a low energy twinning direction. Implications for efficiency of shape memory effect alloys wil be discussed.

4:15 PM S7.8 
PRESSURE INDUCED PHASE TRANSFORMATIONS IN SILICA. Tahir Cagin, Ersan Demiralp, William A. Goddard III, Materials and Process Simulation Center, California Institute of Technology, Pasadena, CA.

We have recently developed an interatomic potential for SiO2 crystalline polymorphs, glasses and melt. This interaction potential uses a 2-body morse and electrostatic interactions between ions. We have chosen the Morse form for stability conserns under extreme pressures and temperatures. Moreover the charges on the ions are made configuration dependent and were determined from charge equilibration method of Rappe and Goddard as the configuration changes in molecular dynamics studies. We have applied these potentials to study pressure induced structural phase transformations in silica under various pressure loading conditions. The particular transformation studied are alpha-quartz to stishovite, coesite to stishovite and fused glass to stishovite-like dense dominantly six-coordinated glassy phases. The simulations are performed under constant loading rates ranging 0.1 GPa/ps to 2 GPa/ps and at temperatures 300, 500, 700 and 900 K. All the transfromations are conducted using constant-stress constant temperature molecular dynamics with thermostatting supplied by Nose-Hoover formalism. We observe transformations to occur reconstructively in crystal-crystal transformations whereas it occurs in a smooth displacive manner from glass to stishovite-like phase confirming the conjecture of Stolper and Ahrens. To elucidate the shock loading experiments we studied the dependence of the transition pressure is on the loading rate and the temperature. To assess the hysterisis effect we also studied the unloading behavior of each transformation.

4:30 PM S7.9 
ANISOTROPIC SOLUTE SEGREGATION AT THE MIGRATION OF ANTIPHASE BOUNDARIES IN A B2 ORDERED PHASE. Qiang Wang, Long-Qing Chen, The Penn State Univ, Dept of Materials Science & Engineering, University Park, PA.

The anisotropy of solute segregation at and migration of antiphase boundaries (APB's) in a B2 ordered single-phase were investigated using computer simulations based on microscopic master equations. Both stoichiometric and nonstoichiometric compositions of the B2 phase were considered. For the particular case of a cylindrical APB along the z-direction, it is found that (1) for the stoichiometric composition the maximum segregation in magnitude occurs along the [1,1,0]-direction and essentially zero along the [1,,0]-direction, and (2) for the nonstoichiometric compositions the maximum segregation occurs at directions slightly different from the [1,1,0]-direction and the segregation is nonzero along the [1,,0]-direction. Particularly, the local compositional profile across APBs, the relationship between size of antiphase domain (APD) and time, and the mobility and width of APB, were studied. Despite the strong segregation anisotropy in stoichiometric alloy, it is demonstrated that the decrease in the radius of the cylindrical antiphase domain in single phase (B2) region is linearly proportional to t1/2 where t is time.

Chairs: Jonathan C. Boettger and Pawel Keblinski 
Thursday Morning, December 4, 1997 
Commonwealth (S)

8:30 AM *S8.1 
THICKNESS DEPENDENCIES IN THE CALCULATED PROPERTIES OF ULTRA-THIN FILMS. Jonathan C. Boettger, Theoretical Division, Los Alamos National Laboratory, Los Alamos, NM.

Ultra-thin film (UTF) electronic-structure calculations are an important tool for investigating the properties of surfaces. In this approach, the surface of interest is modelled by a two-dimensionally periodic UTF that is only a few atomic distances thick. For this approximation to be useful, the model UTF must be thick enough that the surfaces are decoupled and the interior is bulk-like, yet thin enough that a high-precision electronic-structure calculation is affordable. These opposing conditions can be met only if the surface properties of interest converge rapidly as the thickness of the UTF is increased. In this talk, UTF electronic-structure calculations for Al(111) films ranging from one to twelve atoms thick will be used to illustrate some of the difficulties which can arise when one attempts to extract surface properties from UTF calculations.

9:00 AM S8.2 
LATERAL-SIZE-DEPENDENT INTERFACIAL STRAIN-FIELD IN THIN-FILM MICROSTRUCTURES. Qun Shen and Stefan Kycia, Cornell High Energy Synchrotron Source (CHESS), Cornell University, Ithaca, NY.

Interface and surface induced strain fields in thin-film microstructures play an important role in determining the physical properties of a thin-film device. In this paper we present new advances in x-ray diffraction analysis and theoretical simulations of the interfacial strain fields in crystalline semiconductor nanostructures. One of such advances is the observation of diffuse interference fringes and asymmetric patterns due to the strain-varying regions near the boundaries of the nanostructure. From these additional diffraction features, both the longitudinal and the transverse strain gradients are determined, in addition to the average strain. The strain field is found depend strongly on the lateral size of the nanostructures. These experimental results are compared with the simulations of elastic deformations, and can be used to interpret other physical properties in nanostructures such as quantum-confinement potentials, band-gap energies, and surface morphologies.

9:15 AM S8.3 
ATOMISTIC STUDY ON NANO-STRESS-STRAIN CURVES OF -Fe. Shenyang Hu, Matthias Ludwig, Liam Farrissey, Siegfried Schmauder, Staatliche Materialpruefungsanstalt, (MPA), University of Stuttgart, GERMANY.

The atomistic processes and stress-strain-curves during uniaxial tensile deformation of a single -Fe nanocrystal have been studied. Periodic boundary conditions are imposed along one direction perpendicular to the tensile axis to model plane strain conditions. The effects of the model sizes, boundary conditions, crystal orientations and loading rates on the stress-strain curves are systematically simulated. Various deformation evidences such as dislocation movement, dislocation piling up and twinning are clearly observed. The deformation and fracture characteristics of -Fe and their dependencies on the boundary conditions are investigated.

10:00 AM *S8.4 
MOLECULAR-DYANMICS SIMULATION OF GRAIN-BOUNDARY DIFFUSION CREEP. P. Keblinskia,b, D. Wolfa, and H. Gleiterb, aMaterials Science Division, Argonne National Laboratory, Argonne, IL; bForschungszentrum Karlsruhe, Karlsruhe, GERMANY.

Molecular-dynamics (MD) simulations were used to study high temperature grain-boundary diffusion creep of a model polycrystalline silicon microstructure. Our fully dense model microstructures, with a grain size of up to 7.5 nm, were grown by MD simulations of a melt into which small, randomly oriented crystalline seeds were inserted1. In order to prevent grain growth and enable steady-state diffusion creep to be observed on a time scale accessible to MD simulations (of typically 10-9s), our input microstructures were tailored to (i) have a uniform grain shape and a uniform grain size of nm dimensions and (ii) contain only high-energy grain boundaries which are known to be highly disordered even at low temperature2 and exhibit rather fast, liquid-like self-diffusion at a high temperature. Our simulations reveal that under relatively high tensile stresses, these microstructures indeed, exhibit steady-state diffusion creep that is homogenous ( i.e., involving no grain sliding), with a strain rate that agrees quantitatively with that given by the Coble-creep formula.

10:30 AM S8.5 
SIC:GROWTH AND SINTERING. P. C. Clapp, S-H Wang and J. A. Rifkin, Center for Materials Simulation, Institute of Materials Science, Univ. of CT, Storrs, CT.

Using Molecular Dynamics simulation methods combined with Tersoff three body potentials for SiC, C and Si, studies have been made of vapor phase growth of beta SiC from a single crystal seed, and the sintering processes among nanoparticles of beta SiC. Preliminary results will be presented and the vapor phase growth, as well as the sintering sequences, will be illustrated with computer movies.

10:45 AM S8.6 

Molecular dynamics computer simulations of large nanophase samples of pure Ni have been performed on a massively parallel computer. Samples containing up to 50 grains in the range of 2 to 8 nm have been constructed by filling a volume with a polycrystal nucleated from different seeds chosen stochastically. In some samples orientations are chosen fuller randomly, resulting in samples with mainly high energy grain boundaries, in others a restriction on the orientation is imposed in order to obtain samples with a lot of low energy grain boundaries. All samples are relaxed to stable configurations with final density close to bulk values. The relaxed samples are then annealed at higher temperatures in order to change the grain boundary structure to lower energy configurations. The structure of the grain boundaries is studied by means of pair distribution functions, coordination number, energy contour plots, and simulated electron diffraction patterns. The relaxed samples are then loaded and sequentially unloaded with uniaxial stress and the elastic and plastic deformations are studied at different deformation temperatures. The structure of the grain boundaries is changed by annealing the sample at higher temperatures and the influence of grain boundary structure on the deformation mechanism is discussed.

11:00 AM S8.7 
MELTING, FREEZING, DIFFUSION AND COALESCENCE OF GOLD NANOCLUSTERS. Laurent J. Lewis, Département de physique et GMC, Université de Montréal, Succ. Centre-Ville, Montréal, Québec, CANADA; and Pierre Deltour, Pablo Jensen, and Jean-Louis Barrat, Département de physique des matériaux, Université Claude-Bernard Lyon-I, Villeurbanne, FRANCE.

We present a detailed molecular-dynamics study of the melting, freezing, diffusion, and coalescence of gold nanoclusters within the framework of the embedded-atom method. We find cluster melting to first affect the surface, then to proceed inwards. The curve for the melting temperature vs cluster size is found to agree reasonably well with predictions of phenomenological models based on macroscopic concepts. We observe a large hysterisis of the transition upon quenching, consistent with recent experiments on lead. In contrast, we find macroscopic sintering theories to be unable to describe the coalescing behaviour of two small clusters, a failure we attribute to the fact that the nanocrystals are faceted; this has important consequences for the morphology of cluster-assembled materials. Finally, we investigate the diffusion of a 250-atom gold cluster on nickel surfaces.

11:15 AM S8.8 
TEST OF SINTERING LAWS AT THE NANOSCALE. P. Zeng, P. C. Clapp and J. A. Rifkin, Center for Materials Simulation, Institute of Material Science, Univ. of CT, Storrs, CT.

Molecular Dynamics techniques with Embedded Atom Method potentials have been used to study sintering in arrays of pure Cu nanofibers. The sintering studies on multiparticle arrays several hundred degrees below the melting point of pure Cu show unexpectedly large contributions from plastic deformation processes, mechanical rotations, amorphization and highly driven diffusion effects. These results strongly indicate that the standard sintering theories developed for micron scale powders do not apply at the nanoscale. A detailed test of Herring's scaling law has also been performed, and it is found to fail over the entire range of sintering sizes at the nanoscale. Reasons for this failure will be offered. Computer movies will be displayed to illustrate the dynamics of the competing sintering processes.

11:30 AM S8.9 
MICROSTRUCTURAL SIZE AND DIELECTRIC SUSCEPTIBILITY EFFECTS ON THE DIELECTRIC RESPONSE OF INHOMOGENEOUS MEDIA. Kim F. Ferris, Larry A. Chick and Gregory J. Exarhos, Pacific Northwest National Laboratory, Materials and Chemical Sciences Center, Richland, WA; and Steven M. Risser, Texas A&M University-Commerce, Department of Physics, Commerce, TX.

In modeling the dielectric properties of an inhomogeneous material, the treatment of the electric field interactions differentiate the usual modeling formalisms (such as the Maxwell-Garnett and Bruggeman effective medium methods) and their accuracy. Using the LOCALF method to analyze the electric field response, we have developed models for Al2O3 and TiO2 dielectric films of varying microstructure and void content. Surface and bulk defects differentially affect the local electric fields in dielectric media, resulting in marked variations in their dielectric properties. While both cubic and columnar voids result in bimodal distributions of electric field enhancements, columnar voids skew the distribution towards greater effective dielectric constants and enhancements factors. The extent of these effects is dependent on the dielectric constant of the bulk component. In this paper, we will make comparisons for the currently accepted models of dielectric materials with our direct determination method, to illustrate the differences imposed by material microstructures inherent to fabrication processes. Further, we show interdependence of the dielectric response on the characteristic size of the microstructure, and the dielectric constant of the material itself. This work was supported by the U.S. Department of Energy, Office of Basic Energy Sciences, Materials Sciences Division under contract DE-AC06-76RLO 1830.

11:45 AM S8.10 

Besides the contribution from the thermally activated flips of the polar regions, the contribution from the surface of frozen polar regions to the dielectric behavior of relaxors is introduced and discussed. Based on it, a formula is proposed to describe the temperature and frequency dependence of the dielectric constant. The formula is strictly certified with the measured data of some relaxors in a wide temperature range at frequencies from 100 Hz to 100 kHz. It is indicated that the dielectric response of the surface motion of the frozen polar regions is resonance. The difference between the relaxation and resonance processes is given, and a relationship between them is proposed. The frozen polar region is nano-order cluster. The difference between the frozen polar regions and the general ferroelectric nanocrystal is discussed.

Chairs: Anne M. Chaka and John Ferrante 
Thursday Afternoon, December 4, 1997 
Commonwealth (S)

1:30 PM *S9.1 
COPPER CORROSION MECHANISMS OF POLYSULFIDES. Anne M. Chaka, The Lubrizol Corporation, Research Division, Wickliffe, OH; John Harris, Molecular Simulations, Inc., San Diego, CA.

Polysulfide lubricant additives are effective antiwear agents which protect ferrous metal components, but also cause corrosion of copper-based alloys such as bronze and brass. This is a serious limitation as ferrous and non-ferrous metals are commonly used to fashion different parts of the same mechanical system. In commercial organopolysulfides of the type R-(S)n-R, the corrosive behavior of polysulfides dramatically increases when n 4. Currently little is known about the corrosion mechanisms involved which can explain this difference in reactivity. Three hypotheses proposed to explain this behavior are examined at the atomistic level using local and nonlocal density functional theory (DFT), as well as post-Hartree-Fock calculations at the many-body perturbation and coupled-cluster theory level. In addition we compare the description of the potential energy surface obtained using traditional Kohn-Sham DFT methods with a relatively new density functional program, Fast-Structure, based on the non-self-consistent Harris functional with trial densities constructed from spherically symmetric site-densities.

2:00 PM *S9.2 
ACTIVE SITE STRUCTURES IN ZEOLITE-SUPPORTED LEAN NOx CATALYSTS. Richard J. Blint, Physics and Physical Chemistry Department, Research and Development Center, General Motors Corporation, Warren, MI.

Abstract not available.

2:30 PM S9.3 
ALLOYING ELEMENTS AT GRAIN BOUNDARIES IN IRON. X. Chen, D.E. Ellis, G.B. Olson, Northwestern Univ, Evanston, IL.

The first principles Embedded Cluster Density Functional (ECDF) approach is used to study electronic effects of alloying elements X (X=Pd, Mo, Mn, Cr) at grain boundaries(GB) and corresponding free surfaces(FS) in iron. A modified Molecular Dynamics/Monte Carlo (MD/MC) scheme is used to estimate relaxed atomic positions defining the selected GB and FS, as input to the ECDF procedure. The effects of X-Fe interactions are studied both in the dilute and saturated(monolayer) limits, by analysis of bonding interactions, charge and spin densities, and spectroscopic properties inferred from the electronic states. Calculated cohesive energies are used to assess alloying effects on the Griffith work of fracture.

2:45 PM S9.4 
ENERGETICS OF SEGREGATION OF Cu AND Si IMPURITIES TO AL GRAIN BOUNDARIES. Stephen M. Foiles, Andrew A. Quong, Sandia National Laboratories, Livermore, CA; Geoffrey H. Campbell, Lawrence Livermore National Laboratories, Livermore, CA.

Cu and Si are common alloying additions to Al interconnect lines and their segregation to grain boundaries is known to affect the stability of these lines. In this paper, the energetics of segregation of these impurities to a 11 symmetric tilt boundary is computed using ab initio electronic structure calculations. The results are compared to experimental observations of the same boundary. Further, the results are interpreted in terms of size mismatch and elastic modulus mismatch interactions. These ideas provide a qualitative understanding of the results and so a means to extend these results to more general boundaries.

3:30 PM *S9.5 
BFS METHOD SURVEY OF SURFACE AND INTERFACIAL PROPERTIES OF MULTICOMPONENT METALLIC SYSTEMS. John Ferrante, NASA Lewis Research Center, Cleveland, OH and Department of Physics, Cleveland State University, Cleveland, OH; Guillermo Bozzolo, Ohio Aerospace Institute, Cleveland, OH.

The calculation of alloy properties has proven to be a difficult topic, particularly for multicomponent systems. Although first-principles methods have high accuracy, the application to complex systems has been limited to simple cases, because of the excessive computer time required for calculations. Semiempirical techniques have been applied to binary systems but extensions to multicomponent systems have been limited. Recently, a new semiempirical technique has been developed for calculating the configurational energetics of multicomponent systems, the BFS (Bozzolo-Ferrante-Smith) method. After successfully predicting bulk heats of formation and surface segregation profiles for binary systems, the method has been applied to more complex problems, including three- and four-element alloys and their surfaces, the analysis of multiple phase alloys, their interfaces, segregation of alloying additions to interfaces, etc. This paper will be devoted to presenting a summary of the results of such studies for alloys of aeronautical interest as well as presenting experimental verification of some of the theoretical predictions.

4:00 PM S9.6 
MONTE CARLO SIMULATION OF SOLUTE-ATOM SEGREGATION AT GRAIN BOUNDARIES IN SINGLE-PHASE BINARY FCC ALLOYS*. J.D. Rittner and D.N. Seidman, Dept of Materials Science and Engineering, Northwestern University, Evanston, IL.

Monte Carlo techniques (Metropolis algorithm and overlapping distributions MC) have been used extensively to explore systematically the eight-dimensional grain boundary (GB) phase space for both twist and tilt boundaries in dilute single-phase binary f.c.c. alloys; the atomic interactions are described by EAM potentials. We use lattice statics calculations initially to determine the lowest energy GB structures and then MC simulation to calculate the distribution of solute atoms and the Gibbsian interfacial excess at elevated temperatures for each GB structure studied. ODMC simulation is used to calculate binding free energies of solute atoms at specific sites in a GB. The effects of both the five macroscopic and three microscopic degrees of freedom (DOFs) are studied and it is demonstrated that the Gibbsian excess is a complicated function of both macroscopic and microscopic DOFs. Also a GB's atomic structure determines the partition of segregating solute atoms between the cores of dislocations and in the elastic stress fields of GB dislocations. It is concluded that none of the geometric criteria suggested in the literature is capable of predicting the propensity for GB segregation, much less the magnitude.newline *This research is supported by the National Science Foundation-Division of Materials Research.

4:15 PM S9.7 
INFLUENCE OF ALUMINUM GRAIN BOUNDARY MISORIENTATION ON EMBRITTLEMENT BY LIQUID GALLIUM. R.C. Hugo and R.G. Hoagland, Washington St. Univ, School of Mechanical and Materials Engineering, Pullman, WA.

When liquid gallium wets the surface of solid aluminum, the gallium rapidly penetrates the aluminum grain boundaries. Previous studies performed in-situ in the transmission electron microscope have shown that different boundaries are penetrated at rates that can differ by two orders of magnitude. In this study, we correlate thermodynamic properties calculated from atomistic grain boundary models with experimentally determined penetration speeds. From misorientation data obtained in TEM experiments, we construct grain boundary models based on Embedded Atom Method potentials. Relaxation proceeds via molecular statics. In addition calculating the grain boundary energy, we compute the excess volume using a Voronoi tesselation. We find that boundaries with higher energy and excess volume are penetrated more rapidly; however, the variation in these thermodynamic properties is much smaller than the variation in penetration speeds. Other factors that may explain the large variation in penetration speed, such as channels of high excess volume, are discussed. This research is supported by the US Dept of Energy under grant DE-FG06-87ER45287.

4:30 PM S9.8 
AN ATOMISTIC STUDY OF Ti SEGREGATION TO LAMELLAR INTERFACES IN TI-RICH TiAl. K. Ito, R. Siegl, V. Vitek, Department of Materials Science and Engineering, University of Pennsylvania, Philadelphia, PA.

The most interesting titanium-aluminum alloys based on L1o TiAl ( phase) are two-phase lamellar structures. The corresponding interfaces are all parallel to (111) planes in the phase and understanding of their structure is an essential precursor for fundamental comprehension of properties of these materials. Lamellar TiAl is, in general, Ti rich and, therefore segregation of Ti to the lamellar interfaces mayoccur. This may influence considerably their structure and properties and induce structural changes. In this paper we present recent results of molecular statics and Monte Carlo studies of Ti segregation to several types of / interfaces. These calculations suggest that a thin layer of Ti3Al, one or two (111) layers thick, is formed at pseudotwins and 120 rotational faults. On the other hand no segregation takes place to ordered twins, and twins with APB transform upon segregation into pseudotwins and 120 rotational faults. This research was supported by the U.S. Department of Energy, Office of Basic Energy Sciences, Grant No. DE FG02-87ER46295 and JSPS Postdoctoral Fellowship for Research Abroad (KI).

4:45 PM S9.9 
IMPURITY INDUCED STRUCTURAL TRANSFORMATION OF AN MGO GRAIN BOUNDARY. Y. Yan, M.F. Chisholm, S.J Pennycook, Oak Ridge National Laboratory, Oak Ridge, TN; A. Maiti, Molecular Simulations Inc, Boston, MA; and S.T. Pantelides, Vanderbilt University, Nashville and Oak Ridge National Laboratory, Oak Ridge, TN.

A combination of first-principles density-functional simulations and atomic-resolution Z-contrast images provides evidence that the segregation of Ca impurities in MgO grain boundaries induces a structural transformation. The presence of Ca in these new boundary structures is independently confirmed and quantified by electron energy loss spectroscopy. The calculations establish that, if the Ca impurities are removed, the observed structure boundary structure is unstable and will spontaneously revert to the ground state configuration of the pure grain boundary.