1997 MRS Fall Meeting & Exhibit

Chairs

Dennis Dimiduk, Air Force Wright Laboratory
Mark Eberhart, Colorado School of Mines
Anthony Giamei, United Technologies Res Ctr
David Srolovitz, Univ of Michigan

Symposium Support 

• United Technologies Research Center

* Invited paper

SESSION P1: POTENTIALS FOR MATERIALS DEVELOPMENT THROUGH MODELING AND SIMULATIONS 
Chair: Christopher F. Woodward 
Monday Morning, December 1, 1997 
Gardner (S)

8:30 AM *P1.1 
SYSTEMS DESIGN OF HIERARCHICALLY-STRUCTURED MATERIALS: ADVANCED STEELS. G. B. Olson, Northwestern University, Dept. Mat. Sci. & Eng., Evanston, IL.

Over the past decade, the multi-institutional Steel Research Group (SRG) has pioneered a systems approach to the integration of processing/structure/property/performance relations in the science-based conceptual design of new materials, using the example of high performance alloy steels. Deriving quantitative property objectives from Ashby property/performance cross-plots, coordinated research integrating materials science, applied mechanics and quantum physics provides mechanistic models for key structure/property and processing/structure relations. Design integration employing the THERMOCALC thermochemical database and software system and DICTRA diffusion code has been successfully applied in Materials Design class team projects delivering promising new alloys for advanced structural, gear and bearing applications. A new Materials Design Initiative is exploring the extension of this approach to ceramics, polymers and biomimetic composites. Research sponsored by ARO, ONR, EPRI, DOE, NASA, AFOSR and NSF, with industry fellowship support.

9:00 AM P1.2 
SURFACE AND INTERFACIAL RELAXATION USING A FULL-POTENTIAL LMTO METHOD. D.L. Price, Dept. of Physics, University of Memphis, Memphis, TN; B.R. Cooper, Dept. of Physics, West Virginia University, Morgantown, WV.

We have added the capability to calculate forces to our full-potential LMTO (Linear Muffin-Tin Orbital) electronic structure method, in a manner similar to that of other recent all-electron force methods. We have applied this new capability to a number of test calculations: We will report here on calculations for TiC and TaC (001) surface layer relaxation. These provide a good case for an initial study, since we have previously1 explored the total energy as a function of surface layer relaxation, and further have performed a detailed investigation of the physical origin of the Hellman-Feynmann forces. The access to forces allows energy minimization in configuration spaces of large dimension, and so allows such calculations as multilayer relaxation, and interfacial relaxation. We shall present results for a few such systems, including multilayer surface relaxation for the carbides, and interfacial relaxation such as for the Ni/Ni3AL interface.

9:15 AM *P1.3 
FIRST-PRINCIPLES CALCULATION OF PHASE STABILITY IN METAL, SEMICONDUCTOR, AND OXIDE ALLOYS. C. Wolverton, National Renewable Energy Laboratory, Golden, CO.

There are currently many research groups performing first-principles electronic structure calculations at zero temperature, studying a wide variety of interesting phenomena. However, a whole class of important effects in alloy systems occurs only at finite temperature: order-disorder transitions, short-range ordered solid solutions, finite solubilities, etc. Further, these phenomena occur on time scales which precludes the efficient use of molecular dynamics calculations. On the other hand, an efficient technique has recently been proposed to extend the ``state-of-the-art'' first-principles LDA calculations into the realm of thermodynamics of alloys: A small set of first-principles total energies of atomically relaxed configurations are mapped onto a generalized (mixed-basis, extended pair, many-body) Ising-like Hamiltonian, followed by statistical mechanical treatment of this Hamiltonian by Monte Carlo simulation. This technique effectively extends the ``state-of-the-art'' in first-principles calculations (i.e., full potential, atomically relaxed, total energy) to 10,000-atom alloy systems at finite temperature. Furthermore, this technique is not restricted to a single class of materials, but can be used to study the thermodynamics of alloys composed of metals, semiconductors, or ceramics. Applications will be shown for a variety of alloy systems (Cu-Au, Cu-Ag, Ni-Au, Al-Mg, GaP-InP, LiO-CoO, etc.): Prediction of unsuspected ground state structures, short-range order in disordered solid solutions, order-disorder transitions, free energies and entropies, and bond length distributions.

10:15 AM *P1.4 
A PHENOMENOLOGICAL THERMODYNAMIC APPROACH TO OBTAIN PHASE DIAGRAMS. S.-L. Chen, CompuTherm, LLC, Madison, WI; Y. A. Chang, Department of Materials Science & Engineering, University of Wisconsin, Madison, WI; W. A. Oates, Institut fur Festkorperforschung, Julich, GERMANY; R. Schmid-Fetzer, Technical University Clausthal, Metallurgical Center, Clausthal-Zellerfeld, GERMANY.

Temperature-composition phase diagrams are important road maps for materials design and processing. They have traditionally been determined experimentally but this approach can be costly and time consuming and is not really practical for obtaining phase diagrams for many multi-component systems of technological importance. In this presentation we will show how a combined computational thermodynamic approach coupled with the determination of key experimental results can be used to obtain the phase diagrams of quite complex systems. For disordered solution phases, we use a phenomenological equation based on the expansion of the regular solution model to represent the Gibbs energies. For highly ordered intermediate phases, we use a generalized bond energy model using the point approximation to account for the configurational entropy. When the ranges of homogeneity of these phases become small, they are treated as line compounds. We will present examples to demonstrate the success of this approach in obtaining phase diagrams for practical applications. In systems which involve intermediate phases which undergo order/disorder transformations, however, the point approximation is unable to correctly describe the topology of the phase diagram. In view of the practical difficulty of using the cluster variation method for multi-component systems, we are exploring the use of a cluster/site approximation (1) to describe the thermodynamic behavior of these phases.

10:45 AM *P1.5 
FIRST-PRINCIPLES METHODS FOR THE DESIGN OF RECHARGEABLE LITHIUM BATTERIES. Gerbrand Ceder, Kadri Aydinol, Anton Van der Ven, Adrian Kohan, Massachusetts Institute of Technology, Department of Materials Science and Engineering, Cambridge, MA.

First-principles computational methods have reached the level where they can be used for the design and development of new materials. Because of their predictive capabilities they can be used to conduct virtual experiments from which the structure-property relations of a material can be easily deduced. We demonstrate the application of first-principles methods to the design of lithium-metal-oxides for use in rechargeable Li-batteries. By computing the Li-intercalation voltage for materials with a systematically varied metal chemistry or crystal structure, we can identify how these factors influence the battery voltage. This has led us to suggest higher-voltage cathode compositions. Recent experiments on these materials confirm the theoretical predictions. To our knowledge, this is one of the first examples where a materials has been completely designed with first-principles methods and also successfully synthesized and tested. In addition, we show calculations on several potential failure modes of lithium-metal-oxide cathodes. In this study, first-principles methods are indispensable as often data is needed that is very difficult to obtain experimentally.

11:15 AM P1.6 
A THEORETICAL SEARCH FOR NEW HIGH-PERFORMANCE MAGNETO-OPTIC MEDIA. R.H. Victora, C.F. Brucker, T.K. Hatwar, Imaging Research and Advanced Development, Eastman Kodak Company, Rochester, NY; and J.M. MacLaren, Physics Department, Tulane University, LA.

Ab initio electronic structure calculations, predicting magnetic anisotropy, Kerr effect, and magnetization, were used to search for potential magneto-optic recording media. The computations used the local spin density approximation in combination with the layer Korringa Kohn-Rostoker and the linearized augmented Slater orbital techniques. Results of the theoretical search suggested that Tb/Bi/FeCo and Tb/Pb/FeCo superlattices would offer a large Kerr rotation and the necessary perpendicular anisotropy. Subsequent experimental fabrication demonstrated that they offer Kerr rotation and figure of merit (reflectivity times Kerr rotation) substantially exceeding that offered by conventional TbFeCo recording materials particularly at blue wavelengths. For example, 4.4Å Tb/2.0Å Bi/5.0Å Fe0.6Co0.4 has figures of merit 0.20, 0.18, and 0.14 at wavelengths of 780, 650, and 430 nm respectively. The 430 nm figure of merit is approximately 60% larger than that of the standard recording alloy. The superlattice displays perpendicular anisotropy and squareness suitable for magneto-optical recording. Dynamic testing with 780 nm and 490 nm light demonstrates that these high figures of merit translate to carriers several dB above those produced by comparably sensitive TbFeCo alloys. Furthermore, very high carrier-to-noise ratios such as 60 dB at long mark lengths (1 MHz) are obtained using 490 nm light and a bandwidth of 30 kHz. The overall results imply that these new materials are suitable for future high density recording.

11:45 AM P1.7 
A QUASICONTINUUM METHODOLOGY FOR SMALL-SCALE ENGINEERING. E. B. Tadmor, E. Kaxiras, N. Bernstein, Harvard University, Division of Applied Science, Cambridge, MA.

Recently a Quasicontinuum method has been introduced for the study of the multiple scale mechanical behavior of materials. The method offers a graded atomistic description of the material within a finite element formalism which reproduces the results of conventional atomistics in regions where the variation in deformation is commensurate with the lattice spacing and reduces to a local hyperelastic continuum theory in the far field. By greatly reducing the number of degrees of freedom that are explicitly treated relative to standard atomistics this methodology makes possible the study of atomic scale phenomena such as the structure and interaction of dislocations and grain boundaries in much larger systems than normally possible. A formulation incorporating the local limit of the Quasicontinuum method could be a useful tool for the design and analysis of small-scale single crystal or polycrystal devices such as microelectromechanical systems (MEMS). These devices are currently being designed using small strain isotropic elasticity theory which fails to do justice to the broad range of possible responses of real crystals. A more realistic material response is captured in the local limit of the quasicontinuum where the hyperelastic response function is obtained directly from atomistic calculations and automatically inherits the periodicity and symmetries of the underlying crystal structure. As a first step, the concept is demonstrated by application of the method to several test cases in Si: (1) The stress-induced -Sn transformation in Si; and (2) Nanoindentation into a Si single crystal.

SESSION P2: INTERMETALLICS - MODELING AND DEVELOPMENT 
Chair: Mark Eberhart 
Monday Afternoon, December 1, 1997 
Gardner (S)

1:30 PM *P2.1 
UNDERSTANDING THE INFLUENCE OF CHEMISTRY ON FLOW BEHAVIOR IN TIAL. C. Woodward1, S.I. Rao1, J.P. Simmons1, S.A. Kajihara1, and D.M. Dimiduk, Materials Directorate, Wright Laboratory, ML/MLLM Wright Patterson AFB OH; 1Materials Research Division, UES Inc, Dayton, OH.

The transition metal intermetallic alloys, such as TiAl and NiAl, are an important class of structural materials due to their low density, high melting temperature and excellent strength retention at high temperature. Alloy designers optimize the structural properties of these materials by varying alloy composition, microstructure and the volume fraction of the existing phases. Traditionally, the role alloying additions play in the deformation of metals has been separated into extrinsic and intrinsic effects. Extrinsic effects are manifested through changes in transformation kinetics, microstructural evolution, and phases present; while the intrinsic effects are reflected principally in dislocation dynamics. While extrinsic effects tend to dominate metallurgical design and development of structural materials, these are presently only amenable to empirically-based modeling and simulation. Conversely, significant progress has been realized in modeling and simulation of intrinsic properties for these systems. The results of several computational studies of the intrinsic effects of alloy chemistry in single crystalline TiAl will be reviewed. First principles and atomistic methods have been used to study dislocation mobility and stability as a function of alloy composition in single crystal -TiAl. Also, first principles calculations have been used to predict the point defect densities and site preferences of solid solutions (Si, Nb, Mo, Ta and W) in TiAl. These calculations produce solute-disocation interaction parameters used in continuum models of solid-solution strengthening. Solute-dislocation interaction strengths have been calculated for these solutes and ordinary screw dislocations in -TiAl using anisotropic elasticity theory. Current results point to two alloy compositions with improved flow behavior relative to binary -TiAl.

2:00 PM P2.2 
THE YIELD STRENGTH OF FULLY LAMELLAR Ti-Al. P.M. Hazzledine. S.J. Rao and Y.Q. Sun, Materials Directorate, Wright Laboratory, WL/MLLM, Wright-Patterson Air Force Base, OH.

The yield strength of lamellar Ti-AI alloys is known to be controlled by the Hall-Petch effect . In the theory, the leading member of a pile up of dislocations is forced across an interface by the applied stress magnified by the following members of the pile up. The three parameters which control the yield strength are the intrinsic strength of the material, the strength of the interfacial barrier and the length of the pile up. To determine these parameters requires calculation and simulation on a variety of length scales. The intrinsic strength of the material involves elastic calculation of the coherency stresses and atomistic simulations of kink and jog motion along glissile dislocations. The strength of the interfacial barrier has several components which may be estimated classically but which are best determined by atomistic simulation. The properties of a pile up may be estimated from the continuum theory but a more accurate calculation involves computer simulation of pile ups of discrete dislocations in a material which has three distinctly different grain sizes namely the lamellar thickness, the domain size and the regular grain size. In this paper the results of atomistic simulations, dislocation simulations and continuum theory are combined, within the framework of the Hall-Petch theory, to calculate the yield stress of fully lamellar and polysynthetically twinned Ti-Al.

2:15 PM P2.3 
FLUCTUATIONS OF THE MICROCRACK DENSITY IN AN GAMMA BASED TITANIUM ALUMINIDE. Michael J. Pfuff, Bettina U. Wittkowsky, Rui He, Barbara Wiesand, GKSS Research Centre and SFB 371, Institute for Materials Research, Geesthacht, GERMANY.

Microcracking events have been observed in many intermetallic alloys during straining at room temperature. However, there seems to be no quantitative understanding of the frequency of such events and of the relationships between microcracking and the macroscopic failure of those materials. The failure behavior of a TiAl alloy with a globular, near gamma microstructure has been investigated. Densities of microcracks formed during monotonic deformation have been measured at the surface of rectangular tensile specimens. Due to the heterogeneity of the microstructure the microcrack density exhibits large local fluctuations within the surface of one specimen, as well as on corresponding areas on different specimens of a sample of nominally identical specimens. The experimental results are analyzed in terms of a mesoscale lattice of beams model, which is based on the idealizing assumption that the heterogeneity of the material might be expressed by a statistical distribution of local rupture stress. The importance of the microcrack density fluctuations on the macroscopic failure behavior, such as the absolute value of the rupture strength and its scatter is considered.

3:00 PM P2.4 
AB-INITIO CALCULATIONS OF VACANCY FORMATION ENERGIES INCLUDING LATTICE RELAXATION IN Fe3Al AND FeAl. L.S.Muratov, B.R.Cooper, West Virginia University, Dept of Physics, Morgantown, WV; J.M.Wills, Los Alamos National Laboratory, Los Alamos, NM.

Transition-metal aluminides are of considerable interest as structural materials. Their properties, however, are extremely sensitive to the type and the concentration of point defects. Ab-initio quantum mechanical total-energy calculations based on full-potential linear combination of muffin-tin orbitals (LMTO) provide a convenient method of studying not only formation energies for various defects but also other relevant characteristics such as changes in density of states, and charge redistribution around defects. Augmented with Hellmann-Feymann forces, LMTO allows calculations of relaxation geometries and relaxation energies. We have performed such calculations for iron monovacancy formation in two iron aluminides - FeAl with (B2) and Fe3Al with (DO3) crystallographic structures. We found that the iron vacancy formation energy for FeAl is close to 1 eV. For Fe3Al, two types of iron are characterized by two distinguishable local environments. One type has 8 atoms of iron as nearest neighbors, while the other one has 4 aluminum atoms and 4 iron atoms. Electronic densities of states sharply reflect those features. Our calculated vacancy formation energy for the iron site of the first type is 3.0 eV, while for the second it is 1.8 eV. Relaxations around vacancies in both cases are noticeably high; displacements of atoms reached 15% of interatomic distances, and energies of relaxation lie in the 0.8-1.0 eV range. For both materials we obtain indications of a favoring of vacancy clustering. Comparison between 16 and 32 atom unit cell simulations suggested that the binding interaction between vacancies exceeds 0.2 eV and depends not only on the distance between two vacancies but also on their relative spatial configuration. The dependence of the vacancy formation energy on the size of unit cells suggests that increasing the unit cell (up to 64 atoms) might improve the accuracy of the calculations. Preliminary results for such a unit cell will be presented.

3:15 PM *P2.5 
EFFECT OF HYDROGEN ON THE GRAIN BOUNDARY IN NI3AL: A FIRST PRINCIPLES STUDY. Nicholas Kioussis, Gang Lu, California State Univ., Northridge, Northridge, CA*; Mikael Ciftan, US Army Research Office, Research Triangle Park, NC; * Supported by US Army Research Office under contract No. DAAG55-97-1-0093.

The atomic and the electronic structure of the 5 (210) [001] tilt grain boundary in Ni3Al and the effect of H on the cohesion of the grain boundary have been investigated using the full potential linearized-augmented plane-wave method. The strain field normal to the boundary plane and the excess grain boundary volume are calculated and compared with the results from the EAM calculations. The interlayer strain normal to the boundary plane oscillates with increasing distance from the grain boundary plane . The bonding charge distributions suggest that bonding in the boundary region is significantly different from that in the bulk. The grain boundary energy and the Griffth cohesive energy with and without the H impurity are calculated and compared with the EAM results.

3:45 PM *P2.6 
PREDICTING BRITTLE VS. DUCTILE BEHAVIOUR OF COMPLEX CRYSTALS FROM FIRST PRINCIPLES: THE CASE OF MoSi2. Umesh V. Waghmare, Efthimios Kaxiras, Dept of Physics, Harvard Univ, Cambridge, MA; V. Bulatov, Dept of Mech Engineering, MIT, Cambridge, MA; M. S. Duesbery, Fairfax Material Res, Springfield, VA.

We investigate the possibility of enhancing the ductility of complex crystals by sustitutional alloying, through the changes that this introduces to the relevant surface and unstable stacking fault energies. We obtain accurate values for these quantities using the ab initio pseudopotential total energy method based on a conjugate gradient algorithm. As a representative example, we consider MoSi2, which is a technologically important but brittle material. Effects of V, Nb, Tc substitution for Mo, and Mg, Al, Ge, P substitution for Si are investigated. Our results reveal a simple relationship between the surface and stacking fault energies and the percentage of substitution. Incorporating these results in the Rice model for dislocation nucleation and the Griffith criterion for brittle failure, we predict the effects of substitutional alloying on ductility of MoSi2 for all the substitutional elements considered. Self consistent electronic charge densities are examined to understand the changes in bonding responsible for the changes in ductility.

4:15 PM P2.7 
MEASUREMENT OF FAULT ENERGIES IN Ni3Ge-Fe3Ge INTERMETALLIC ALLOYS. Mukul Kumar, T. John Balk, and Kevin J. Hemker, Johns Hopkins University, Dept. of Mechanical Engineering, Baltimore, MD.

A combination of transmission electron microscopy (TEM) with image simulations has facilitated a highly quantitative measure of superdislocation dissociations. The experimental observations have been corrected for image shifts within the framework of anisotropic elasticity by comparison with simulated images. This allows experimental quantification of planar fault energies, thus providing a benchmark for first principles and atomistic simulations and fundamental insight towards alloy modeling and design. Such measurements of superdislocation dissociations on the order of 2-10 nm have been recorded for the pseudo-binary Ni3Ge-Fe3Ge alloy system. These detailed measurements of fault widths obtained by weak-beam TEM observations of deformation structures will be presented and discussed as a function of alloy composition. The transition from anomalous to normal yielding behavior in these alloys will also be discussed in terms of planar fault energies calculated using the above measurements.

4:30 PM *P2.8 
DISLOCATIONS AND FRACTURE PROPERTIES OF INTERMETALLICS FROM ELECTRONIC STRUCTURE CALCULATIONS. O.N. Mryasov, A.J. Freeman, Northwestern Univ, Dept of Phys and Astron, Evanston, IL; Yu.N. Gornostyrev, Inst of Metal Phys, Ekaterinburg, RUSSIA.

The mechanisms of the ductile/brittle behaviour and anomalous mechanical response of intermetallic aluminides still far from being well understood and remain a challenge to theoretical explanation. This problems seems to be extremely difficult since accurate, microscopic descriptions of the directional covalent chemical bonding are required for modeling the mesoscale phenomena (dislocations). We present results of fundamental, comparative studies of the dislocations and fracture properties for ordered NiAl and FeAl, TiAl and CuAu, and the fcc metals Ir and Au on the basis of the ab-initio determination of the cleavage and generalized stacking fault (GSF or -surface) energetics needed for further Peierls-Nabarro (PN) model and Rice-Thomson (RT) criteria theoretical analyses. We proposed a general and physically transparent scheme for analyzing the dislocation structure and mobility based on solution of the PN model with a two component displacement field and generalized restoring force law determined from first-principles total energy calculations. We used this method to determine the structure and Peierls stress of the single dislocation for L10 TiAl, CuAu, fcc Ir and Au. The approach allows one to establish an explicit relation between quantities which can be accurately determined using microscopic quantum mechanical methods (-surface) and dislocation properties (processes approaching mesoscopic scales). In particular, for the first time the dislocation splitting scheme was consistently determined within the Peierls model concept and found to be substantially different from that usually assumed; as we demonstrate, this result is also very important in the context of analyzing fracture poperties within RT criteria. The proposed scheme allows also one to solve the inverse problem ie., to determine -surface energetics features for a given dislocation structure. In light of these findings, we discuss the importance of the -surfaceology and further improvements of PN model for analyzing dislocation and fracture properties.

SESSION P3: MICROSTRUCTURE MODELING FOR APPLICATIONS I 
Chair: Gregory B. Olson 
Tuesday Morning, December 2, 1997 
Gardner (S)

8:30 AM *P3.1 
MODELING THE ANISOTROPIC PROPERTIES OF MATERIALS. Anthony D. Rollett, Carnegie Mellon University, Pittsburgh, PA.

The successful modeling of properties for materials development requires multiple length scales to be considered. We take iron-based alloys for magnetic applications in rotating machinery as an example of the application of materials modeling. Properties such as permeability and core loss are critical to the use of soft magnetic materials. Modeling the development of these materials , i.e. optimization of properties, requires an understanding of properties such as magnetocrystalline anisotropy, magnetoelastic anisotropy (magnetostriction), grain boundary energy, boundary mobility, in addition to characterization of microstructure. In principle, a wide range of length scales should be considered from electronic structure to the structural analysis of electric equipment. To illustrate the relevance of modeling to materials development, we analyze key processing steps at the mesostructural level for an electrical steel, and for a high temperature iron-cobalt alloy. By so doing, we can show that microstructure, texture and magnetic properties are intimately linked. We analyze specific cases of the modeling of elastic and magnetic properties such as anisotropic modulus and torque curves, based on orientation image maps and X-ray texture determination.

9:00 AM P3.2 
SELECTIVITY OF GOSS GRAINS DURING ABNORMAL GRAIN GROWTH IN Fe-Si ELECTRICAL STEELS. Narayanan Rajmohan, Jerzy A. Szpunar, Department of Mining and Metallurgical Engineering, McGill University, Montreal, Quebec, CANADA; Yasuyuki Hayakawa, Technical Research Laboratories, Kawasaki Steel Corporation, Okayama, JAPAN.

Abnormal growth of Goss grain is studied in detail by various computer models which use computer specimens with imposed orientation distribution function (ODF) of primary recrystallized Fe-3%Si steel. The computer models take into account the anisotropy of GB energy and mobility. From various computer experiments conducted, the importance of initial matrix texture, percentage of mobile boundaries around a growing grain, the mobility differences between GBs and the time of release of various grain boundaries during annealing is realized. Among the various Goss nucleii (which are deviated from the real Goss orientation by few degrees) in the initial primary recrystallized matrix, selection of a particular Goss grain is discussed in detail using the results obtained.

9:45 AM *P3.3 
INDUSTRIAL APPLICATION OF MICROSTRUCTURAL MODELS FOR AIRCRAFT ENGINES. Daniel Y. Wei, Edward S. Russell, Daniel G. Backman, GE Aircraft Engines, Lynn, MA.

Increasingly, the aerospace industry is using microstructural models to improve materials and processes for practical applications. This effort has been motivated by the need to accelerate the development process, reduce cost, and refine the quality of aerospace products. Although considerable fundamental research has been directed to mathematically describe the evolution of microstructure, additional effort is needed to apply these relationships to complex industrial alloys and to integrate individual models. This paper will address these issues in the context of a recently developed model that describes gamma-prime precipitation during continuous cooling of multicomponent superalloy engine parts. The model which simulates precipitate nucleation, growth, and coarsening will be described and results will be compared to experimental data. Other examples of industrial microstructural modeling will also be presented.

10:15 AM P3.4 
FROM MACROSCOPIC TRANSFER MODELING OF SILICON CARBIDE SUBLIMATION PROCESS TO MICROSCOPIC MODELING TRENDS. Michel Pons, Elisabeth Blanquet, Claude Bernard, Laboratoire de Thermodynamique et Physicochimie Metallurgiques, UMR CNRS/INPG/UJF 5614-ENSEEG, Institut National Polytechnigue de Grenoble, Saint-Martin, FRANCE; Micha Anikin, Karim Chourou, Jean-Marc Dedulle, Roland Madar, Laboratoire de Materiaux et de Genie Physique, UMR CNSR/INPG 5628-ENSPG, Institut National Polytechnique de Grenoble, Saint Martin D'Heres, FRANCE.

The deposition of single SiC crystals has been processed inside a sealed enclosure at temperatures above 2300 K and pressures lower than 5x103 Pa by the modified Lely method. The first purpose of this work is to present different optimized macroscopic models, thermodynamics, heat and mass transfers used in the simulation of the growth of such crystals. Thermodynamic modeling has been used to determine the most important reactive species involved in equilibrium conditions. Induction heating modeling has allowed the calculation of the actual temperatures inside the reactor which are not well known because of the difficulty associated with their measurement. Finally, mass transport modeling provided the calculated deposition rate. It was found that the calculated growth rates were close to the experimental ones which may indicate a good representation of the actual phenomena involved in the crucible. Moreover, detect formation, which is the primary obstacle to the production of large area devices, is explained by local sublimation of the growing ingot and related to the temperature gradients existing within the crystal and graphite holder. By analyzing the role of localized thermal gradients in the formation of macro-defects, we have demonstrated that a simple link permit to obtain semi-quantitative microscopic trends from a macroscopic approach. The proposed modelling route could allow in the future a precise control of the cavity configuration to decrease or to modify the density of macro-defects. From a macroscopic point of view, these models and associated thermodynamic and thermal data are to be linked with plastic deformation models to better quantify the defect appearance.

10:30 AM *P3.5 
MICROMECHANICS BASED DESIGN MODELS OF SIZE AND GEOMETRY DEPENDENT STRENGTH AND TOUGHNESS IN METAL-INTERMETALLIC MICROLAMINATES. John Heathcote, Dept of Materials, G. Robert Odette and Glenn E. Lucas, Dept of Mechanical and Environmental Engineering, University of California, Santa Barbara, Santa Barbara, CA.

Experimental and computational techniques were combined to model the effect of various microscale processes on the stress(S)-displacement (U) function, S(U), of constrained micron-scale ductile layers. The s(u) function increases up to a peak Sp(Up), followed by a declining to 0 at a maximum displacement Umax. A large scale bridging model was used to self-consistently calculate S(U) for intermetallic/metallic microlaminates from mechanical tests. Experimental observations were used to guide the development of a finite element model, which was used to simulate the individual and combined effects of the various microscale fracture processes and constituent parameters on S(U). The model treats crack blunting as a function of displacement of the crack faces (U) in the adjoining brittle layers. Crack blunting, or geometrically mediated necking, in the ductile layer results in large-scale geometry changes that control the overall S(U) function. Microscale processes included residual stresses, tunnel crack growth, fracture path selection (e.g., slanted and offset cracks), internal inclusion debonding/microvoid damage and effective layer debonding by brittle matrix splitting. The constituent parameters included ductile layer strength, strain hardening exponents and intrinsic brittle matrix toughness. The effect of loading rate, fracture mode and statistical distributions in key parameters were also modeled. High metal layer strength and strain hardening increase both peak Sp(Up) values and Umax. Crack processes such as slanted, offset cracks or splitting cracks in the intermetallic layers reduce Sp(Up) but increase Umax. Damage in the metal layer, in the form of microvoids or debonding inclusions, reduces both the Sp(Up) and the failure extension. The relation between these basic processes and design criteria are discussed in a companion presentation.

SESSION P4: IN-ROOM POSTER SESSION 
Tuesday Morning, December 2, 1997 
11:00 A.M. 
Gardner (S)

P4.1 
A NOVEL PARALLEL-ROTATION ALGORITHM FOR ATOMISTIC MONTE CARLO SIMULATION OF POLYMER MELTS AND GLASSES. Serge Santos, Ueli W. Suter, Departement of Materials Science, Institute of Polymers, Eidgenoessische Technische Hochschule (ETH), Zurich, SWITZERLAND; Matthias Mueller, Juerg Nievergelt, Department of Computer Science, Institute of Theoretical Computer Science, Eidgenoessische Technische Hochschule (ETH), Zurich, SWITZERLAND.

We develop and test a new elementary Monte Carlo move for use in the off-lattice simulation of polymer systems. This novel Parallel-Rotation algorithm (ParRot) permits moving very efficiently torsion angles that are deeply inside long chains in melts. The Parallel-Rotation move is extremely simple and is also demonstrated to be computationally very efficient and appropriate for Monte Carlo simulation. The ParRot move does not affect the orientation of those parts of the chain outside the moving unit. The move consists of a concerted rotation around four adjacent skeletal bonds. No assumption is made concerning the backbone geometry other than that bond lengths and bond angles are held constant during the elementary move. Properly weighted sampling techniques are needed for ensuring microscopic reversibility because the new move involves a correlated change in four degrees of freedom along the chain backbone. The ParRot move is supplemented with the classical Metropolis Monte Carlo and Reptation techniques in an isothermal-isobaric Monte Carlo simulation of melts of short and long chains. The present composite algorithms are able to reproduce quite efficiently equilibrium thermodynamic properties such as the density. Comparisons are made with the capabilities of the Continuum-Configurational-Bias Monte Carlo to move the torsion angles in the middle of the chains. We demonstrate that ParRot constitutes a highly promising Monte Carlo move for the treatment of long polymer chains in the off-lattice simulation of realistic models of polymer melts and glasses.

P4.2 
INTERFACIAL PHONONS AND LONG-RANGE STRUCTURAL CORRELATIONS IN NANOPHASE CERAMICS VIA MULTISCALE PARALLEL MOLECULAR DYNAMICS*. Kenji Tsuruta, Andrey Omeltchenko, Kajiv K. Kalia, and Priya Vashishta, Concurrent Computing Laboratory for Materials Simulation, Dept. of Physics & Astronomy, Dept. of Computer Science, Louisiana State University, Baton Rouge, LA.

We investigate multiple scale phenomena in nanophase silicon nitride (Si3N4) using mutiresolution molecular dynamics on parallel machines. In a consolidated nanophase system the structure and the density-of-states in interfacial regions are similar to bulk amorphous Si3N4. The specific heat of the nanophase system is larger than that of the single crystal. This is due to an enhancement of low-energy interfacial phonon modes. The calculated static structure factor of the nanophase system reveals multiple length-scale correlations: Large peaks at small q regions (0.1) of the structure factor shed light on the intercluster structural correlations in the consolidated system. These results will be compared with small-angle neutron scattering experiments and a theoretical model for fractal objects [1].

P4.3 
EFFECT OF ULTRAFINE MICROSTRUCTURES ON MECHANICAL FAILURE IN NANOPHASE SILICON CARBIDE: MULTISCALE PARALLEL MOLECULAR DYNAMICS SIMULATIONS. Alok Chatterjee, Andrey Omeltchenko, Kenji Tsuruta, Rajiv K. Kalia, and Priya Vashishta, Concurrent Computing Laboratory for Materials Simulation, Dept. of Physics & Astronomy, Dept. of Computer Science, Louisiana State University, Baton Rouge, LA.

Using multiresolution algorithms we perform multi-million atom molecular dynamics simulations on parallel architectures to investigate structural correlations and mechanical failure in nanocluster-assembled SiC. The simulations are based on an empirical bond-order potential [1]. We investigate sintering, the structure of interfacial regions, and the influence of grain size and porosity on mechanical properties of nanophase SiC. Fracture toughness, crack-front morphology, crack-tip speed, and the effect of strain rate on dynamic fracture are determined in crystalline and nanophase SiC. These results are compared with corresponding results in other nanophase ceramics.

P4.5 
STRUCTURE AND MECHANICAL BEHAVIOR OF NANOCLUSTER-ASSEMBLED SILICA. Timothy J. Campbell*, Kenji Tsuruta*, Rajiv K. Kalia*, Shuji Ogata**, Aiichio Nakano*, and Priya Vashishta*, *Concurrent Computing Laboratory for Materials Simulation, Department of Physics & Astronomy, Department of Computer Science, Louisiana State University, Baton Rouge, LA; **Dept. of Applied Sciences, Yamaguchi Univ., Ube, JAPAN.

Million atom molecular-dynamics (MD) simulations are performed to investigate the structure and mechanical behavior of nanocluster-assembled SiO2. The MD simulations are implemented using multiscale techniques and an effective interatomic potential which includes two-body and three-body interactions1. The solids are assembled from clusters of amorphous Si02 with radii of 40 (10,905 atoms) and sintered at various pressures to obtain consolidated systems with mass densities ranging from 1 to 2 g/cc. Structure, pore morphology, and mechanical behavior at various mass densities will be presented.

P4.6 
PARALLEL MOLECULAR DYNAMICS SIMULATIONS OF DYNAMIC, FRACTURE IN NANOWIRES*. Phillip Walsh, Kenji Tsuruta, Rajiv Kalia, and Priya Vashishta, Concurrent Computing Laboratory for Materials Simulations, Dept. of Physics & Astronomy, Dept. of Computer Science, Louisiana State University, Baton Rouge, LA.

Million atom molecular dynamics simulations are performed on parallel architectures to study dynamic fracture in SiSe2 and Si3N4 nanowires. Interaction potentials for these systems include charge transfer, steric repulsion, and electronic polarizability effects via two-body terms. Covalent effects are included through bond stretching and bond bending three-body potentials. Results for mechanical properties of nanowires, critical strain for fracture, crack velocities, and local stress distributions near and far from the crack tip will be presented.

P4.7 
NUMERICAL SIMULATIONS OF THE STRESS REPONSE OF POROUS ISOTROPIC AND ANISOTROPIC MATERIALS UNDER TRANSIENT, DYNAMIC LOADING. J.B. Aidun, R.M. Brannon, V. Tikare, Sandia National Laboratories, Albuquerque, NM.

We will present numerical simulations that will give the transient stress response of isotropic and anisotropic materials with 10% porosity. The Potts Monte Carlo method is used to generate a self-scaling microstructure with porosity. Finite element techniques are used to determine the transient dynamic stresses. The microstructure is meshed to preserve the grain and pore structure. Materials parameters such as Young's modulus and Poisson's ratio of typical ceramics is assumed for the grains and pores are assumed to have no load bearing capability. The stress response of such a material to dynamic loading will be presented, with emphasis on local stress variations due to microstructural features.

P4.8 
LOCAL STABILITY OF HIGHER-ENERGY PHASES IN NiAl: A COMPARATIVE STUDY OF FIRST-PRINCIPLES CALCULATIONS AND FINNIS-SINCLAIR POTENTIALS. L.G. Wang, Institute of Physics of Materials, Academy of Sciences of the Czech Republic, Brno, CZECH REPUBLIC; V. Paidar, Institute of Physics, Academy of Sciences of the Czech Republic, Prague, CZECH REPUBLIC; M. Sob, Institute of Physics of Materials, Academy of Sciences of the Czech Republic, Brno, CZECH REPUBLIC, and Department of Materials Science and Engineering, University of Pennsylvania, Philadelphia, PA; V. Vitek, Department of Materials Science and Engineering, University of Pennsylvania, Philadelphia, PA.

The variation of the energy of NiAl deformed following the tetragonal (Bain's) and trigonal deformation paths is evaluated using the full-potential linearized augmented plane waves (FLAPW) method and central-force many-body potentials of the Finnis-Sinclair type. Comparison of these two approaches reveals features that are sensitive to the details of the electronic structure, which is fully included in FLAPW. Configurations corresponding to energy minima are particularly important since they represent metastable structures that may play a role in extended defects, such as interfaces. Some such minima correspond to symmetry-dictated extrema and are generic for a given atomic ordering. However, other minima are not imposed by symmetry and may depend sensitively on interatomic forces and/or details of the electronic structure. Understanding of these features aids to assess the extent of utility of simple potentials in atomistic studies.

P4.9 
CLASSICAL MOLECULAR DYNAMICS WITHOUT PAIR POTENTIALS FOR COMPLEX MATERIAL. S. T. Pantelides and J. S. McCarley, Department of Physics and Astronomy, Vanderbilt University, Nashville, TN.

Classical molecular dynamics expresses the total energy of a system of atoms in terms of pair potentials plus higher-order terms. Construction of these potentials is an arduous task and becomes virtually impossible for complex materials involving several atomic species. We describe a new approach where the final form of the total energy is derived from density functional pseudopotential theory via systematic approximations. The net result is an expression that depends only on atomic coordinates. The method is equally applicable to multicomponent systems, where dynamical effective charges arise naturally and track charge transfer. The method has been implemented for simple metals where it is found to be entirely predictive without empirical adjustment. For Si and compound semiconductors, an empirical correction is needed, but the fit is linear and thus very robust and can be easily reoptimized for custom applications. For transition metals and first-row elements, additional empirical adjustments are needed. The method is also applicable to magnetic materials, where spin-up and spin-down electron densities are introduced. Initial applications to aluminum grain boundaries will be discussed. Work supported in part by ONR Grant N00014-96-1-1042.

P4.10 
EVOLUTION OF VOIDS IN A SINGLE CRYSTAL. John Y. Shu, Chemistry and Materials Science Department, Lawrence Livermore National Laboratory, Livermore, CA.

Multiscale modeling of material deformation is becoming increasingly viable as computer power explodes. Considerable amount of research has been focused on numeric techniques to bridge macroscale polycrystal and mesoscale single crystal mechanics simulations. However the classical single crystal plasticity theory does not contain any length scales in the constitutive laws, therefore resulting simulations predict the same stress and strain no matter what the size of the sample is, cm or micron. Accumulating experimental evidences indicate the inadequacy of the classical theory at micron scale. Therefore it is necessary to have a modified continuum plasticity theory which can predict deformation of a single crystal with size from micron to cm and larger. In this talk, an elastic-viscoplastic strain gradient crystal plasticity theory is used to simulate the evolution of voids in a single crystal to demonstrate its applicability across broader length scale spectrum. Also studied is the softening of the crystal associated with the presence of the voids. A finite element code (GRACY2D) with elements specially designed for the strain gradient theory is used to conduct the simulations. Various void sizes but the same volume fraction are considered. A dramatic decrease in the void growth rate is predicted when its size is reduced to the microscopic material length scale. This indicates that 'small' voids are less susceptible to growth than 'large' voids. Computations show that 'small' voids cause a smaller reduction than 'large' voids in the average applied stress of that of a void-free crystal, ie, a smaller softening effect. As the void size increases, the gradient theory predictions gradually approach the classical theory predictions which are size-independent.

P4.11 
FIRST PRINCIPLES LINEAR-SCALING CALCULATIONS IN NANOSTRUCTURED MATERIALS. Daniel Sánchez-Portal, Emilio Artacho, José M. Soler, Dept. Fisica de la Materia Condensada, Universidad Autónoma de Madrid, Madrid, SPAIN; Pablo Ordejón, Departamento de Fisica, Universidad de Oviedo, Oviedo, SPAIN.

Selfconsistent density-functional calculations are much more reliable than conventional tight-binding approaches to describe atoms in very different chemical enviroments and bonding situations. Besides, LCAO basis sets allows the use of the linear-scaling techniques initially developed within the tight-binding scheme. With these techniques the computer time and memory scale linearly with the size of the system studied. We have implemented a method based in norm-conserving pseudopotentials and numerical atomic basis functions, constructed from solutions of the atomic pseudopotential. We use the Mauri-Ordejón-Kim energy functional, which does not require the explicit orthogonalization of the wave-functions but converges to the same solution as the Kohn-Sham functional. Forces and stresses are also calculated efficiently and accurately, allowing relaxation of structures and molecular dynamics simulations. With a minimal basis set, we are able to study very large systems, similar to those treated with tight-binding methods, but fully selfconsistently. By systematically increasing the basis with multiple-zeta and polarized functions, we can get fully converged results with a precision similar to that of plane-wave methods. Applications to vibrational and elastic properties of carbon nanotubes are presented. We also present calculations for several systems including minerals and biological systems.

P4.12 
MULTI-SCALE CONTINUUM MODELING OF THE THERMOMECHANICAL BEHAVIOR OF HIGH SPEED STEELS. H.J. Bohm, A.F. Plankensteiner, F.G. Rammerstorfer, Institute of Light Weight Structures and Aerospace Engineering, Vienna University of Technology, Vienna, AUSTRIA.

The thermomechanical behavior of high speed steels (HSSs) is studied by continuum mechanical models that account for the length scales of the primary carbides (microscale) and of carbidic stringers (mesoscale) to describe the response at the macroscale. On the one hand, two-dimensional models are generated by assigning either matrix or inclusion properties to hexagonal subcells tiling a unit cell. This way generic phase arrangements can be investigated via a finite element based approach, layer-structured HSSs being treated as graded metal matrix composites. On the other hand, unit cell models are obtained by discretizing metallographic sections of HSSs in terms of mesoscale particle-rich clusters and stringers embedded in an inclusion-poor matrix. The thermomechanical properties of these zones, in turn, are described by an elastoplastic micromechanically based material model that allows the phase averaged microfields to be accessed directly. These two approaches are used to evaluate phase arrangements in HSS in terms of their vulnerability to the initiation of damage in the carbidic particles, in the matrix and at the interface.

SESSION P5: MICROSTRUCTURE MODELING FOR APPLICATIONS II 
Chair: David J. Srolovitz 
Tuesday Afternoon, December 2, 1997 
Gardner (S)

1:30 PM P5.1 
QUASICONTINUUM DYNAMICS. Vijay B. Shenoy, Rob Phillips, Division of Engineering, Brown University, Providence, RI; M. Ortiz, Graduate Aeronautical Laboratories, Caltech, Pasadena, CA.

The recently proposed quasicontinuum method is an efficient approach for modeling of multiple scale phenomena in crystalline solids. Extension of this method to include dynamic and thermal effects are explored, where phenomena involve multiple temporal scales in addition to multiple spatial scales. The method is based on selecting a set of reduced degrees of freedom which are the positions of a subset of atoms (called representative atoms) that make up the solid. The position of any other atom is then obtained by finite element interpolation based on a mesh constructed with the representative atoms as nodes. This procedure effects the bridging of the spatial scales. The temporal scales are bridged using the method of subcycling where the equations of motion of the representative atoms in the far-field region are integrated using a larger time step than that of the atoms in the nonlinear fully refined regions. The atoms in the fully refined regions are explicitly treated using Newton's laws of motion. In the far-field region a thermodynamic formulation is applied to treat the thermal processes using a local harmonic approximation for the free energy and the local temperature. Heat and linear momentum are transferred from the fully refined region to the far-field region using constrained dynamics at the interfa ce between the two regions. Applications of the method are discussed.

1:45 PM P5.2 
MODELING AND SIMULATION OF CERAMIC GAS SENSORS ACROSS ALL LENGTH SCALES. Bruce R. Patton,The Ohio State University, Dept of Physics, Columbus, OH; Yunzhi Wang, The Ohio State University, Dept of Materials Science and Engineering, Columbus, OH.

Electrical transport in ceramic materials depends critically on structural properties and heterogeneities like grain boundaries, pores, second phases and interfacial impurity segregation. At the same time the transport properties are of key importance in materials used in devices such as gas or temperature sensors. We propose an integrated computational approach which unites the structural and transport modeling of electronic ceramic materials across a wide range of length scales from microscopic through mesoscopic to macroscopic. The gas sensing ceramic composites are taken as a model system. At the microscopic level, we study the gas absorption and reaction processes at the surface of the grains through Monte Carlo modeling together with calculations of the percolation conductivity of the composite. These results form the input to field kinetic modeling of the growth of granular sintered ceramic regions. For the first time, the conductivity of the sample is calculated concurrently as the structure evolves. Finally, coarse graining of the mesoscalar field kinetic results leads to an integrated phenomenological model that describes the gas reactions and percolation conduction at the macroscopic level. The contrast between different proposed reaction mechanisms is clearly shown.

2:00 PM P5.3 
THEORY AND SIMULATION OF SHEAR FLOW-INDUCED MICROSTRUCTURE IN MESOPHASE PITCHES. Arvinder P. Singh and Alejandro D. Rey, Department of Chemical Engineering, McGill University, Montreal, Quebec, CANADA.

Mesophase pitches are an important class of low cost precursor materials that are used to manufacture high performance mesophase pitch-based carbon fibers which possess excellent mechanical and thermal transport properties. These fibers are used to produce a new generation of composite materials that are revolutionizing the aircraft, electronics and automotive industries. The superior properties of mesophase carbon fibers depend on the texture that evolves during the spinning process, and which is a function of the operating conditions, geometry, and material properties. The development of microstructure during the fiber formation process of mesophase pitch-based carbon fibers is critical to optimize their mechanical and thermal transport properties. Numerous experimental studies have been conducted worldwide to explore the flow behavior of these precursor materials. The need to understand the fundamental principles that govern the evolution and control of fiber microstructures during the spinning of these materials has been widely recognized. Since melt-spinning involves shearing of precursors inside the spinneret, a fundamental understanding of shear flow-induced microstructure is highly desirable. In this work a constitutive equation (CE) of mesophase pitches is formulated by taking into account the full microstructural characteristics. The CE is subjected to simple shear flows and the microstructural responses are computed, characterized, and explained. Advanced mathematical tools such as bifurcation methods have been employed to capture the various microstructure modes and their corresponding stabilities as function of process conditions. The bifurcational analysis reveals that the present CE predicts the simultaneous presence of stable planar (steady and periodic) and non-planar orientation modes, and transition among the planar periodic and steady states with increasing shear rate. Consistencies among the simulated microstructure features and experimentally observed fiber microstructures are presented.

2:15 PM P5.4 
FORMATION AND SELF-ORGANIZATION OF NANOCLUSTERS BY ION BEAM SYNTHESIS: A COMBINED ATOMISTIC AND CONTINUUM DESCRIPTION. Matthias Strobel, Karl-Heinz Heinig, Wolfhard Moeller, Forschungszentrum Rossendorf, Dresden, GERMANY.

Ostwald ripening controls the evolution and self-organization of precipitates in bulk phases. Computational studies of Ostwald ripening firstly performed by Voorhees and Glicksman [1] have been recently extended to spatially inhomogeneous systems (layers) [2], but for these models the initial conditions have been chosen to some extent arbitrarily. However, since coarsening starts usually with precipitate distributions formed by preceding nucleation and growth processes, a detailed knowledge of the depth and size distribution is essential to predict the (spatial) evolution of the nanoclusters. Therefore, this paper aims at a complete and consistent description of nucleation, growth and ripening using two computer simulation methods: 
(i) On the atomic scale a kinetic Monte Carlo (MC) program allows a unified investigation of the nucleation, growth and coarsening of nanoclusters. (ii) Within the framework of a local mean-field approach the study of Ostwald ripening of nanoclusters on the mesoscopic scale is performed by a stepwise numerical integration of the reaction-diffusion equations. 
Depth and size distributions obtained by the MC method are used as initial conditions for ripening simulations with method (ii). Predicted pattern formation will be compared with the experimentally observed multilayer formation. Furthermore, the evolution and micro-structure of nanocluster ensembles during ion implantation and/or during subsequent annealing have been simulated and the dependence on process parameters will be discussed.

3:00 PM P5.5 
NANOSCALE SINTERING PHENOMENA. P. C. Clapp, P. Zeng, S. Zajac and J. A. Rifkin, Center for Materials Simulation, Institute of Material Science, Univ of CT, Storrs, CT.

We are using Molecular Dynamics techniques with Embedded Atom Method potentials to study sintering, surface diffusion and grain boundary mobility in nanoparticle arrays. Simulations several hundred degrees below the melting point of pure Cu and Au 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 (e. g. Ashby sintering diagrams) will have to be heavily revised, if not abandoned, before accurate predictions of nanoscale sintering kinetics will be possible. Computer movies will be shown to illustrate the competing sintering processes at the nanoscale.

3:15 PM P5.6 
COMPUTATIONAL EVALUATION ACROSS LAMELLAE AND GRAIN STRUCTURAL SCALES IN SPHERULITE DEVELOPMENT OF SIME-CRYSTALLINE POLYMER FILMS. Tao Huang, Tomohiro Tsuji, M.R. Kamal, and A.D.Rey, McGill Univ, Dept of Chemical Engineering, Montreal, CANADA.

An envelope evolution equation was developed to predict and analyze nucleation-growth processes and grain patterns of semi-crystalline polymer spherulitic structures. Computational modeling with this equation was performed and evaluated using experimental data. Simulation results and visualization of the spherulitic textures indicate the origin of the Maltese cross pattern that appears under polarized light. A free energy density equation was derived to analyze the lamellar spatial orientation. Dynamic comparison between computational and experimental results was done by quantitatively matching the simulated growth patterns to the experimental images. Domain-spatial correlation was used to validate the computational results. This study provides a physical model for prediction and characterization of the development of semi-crystalline polymer microstructure across mesoscopic length and macroscopic grain length scales.

3:30 PM P5.7 
MICROSTRUCTURAL EVOLUTION IN THE PRESENCE OF VACANCIES. Sonali Mukherjee, Bernard R. Cooper, West Virginia Univ., Dept. of Physics, Morgantown, WV.

Using phenomenological Monte-Carlo (MC) simulations in 3-dimensions, we have explored microstructural evolution at atomic scales in binary alloys in the presence of vacancy-mediated diffusion. The vacancy trajectory and consequently the coarsening mechanism through which the microstructure evolves is determined by the the interplay of interfacial energy and thermal energy. Reduction of interfacial energy makes the vacancies favor a position near domain interfaces; whereas thermal energy tends to randomize the vacancy trajectory. The competition between thermal energy and interfacial energy leads to crossover between coarsening via cluster-cluster aggregation (CCA) and Ostwald ripening (OR) with increasing temperature. At lower temperatures relative to Tc (the temperature which marks the onset of phase separation), vacancy-interface affinity leads to evaporated atoms recondensing onto the same domain. This leads to domain shape deformation without domain breaking and results in the movement of the domain center-of-mass. Consequently, the mobile domains collide with other domains (clusters) and aggregate via (CCA). With increasing temperature, the thermal energy increases the probability of atoms evaporating from a domain to diffuse into the matrix along with the vacancy. The evaporated atoms eventually get attached to other domains. This process of evaporation, diffusion and condensation of atoms onto other domains results in an increasing contribution of coarsening via OR. Consequently, CCA gradually gives way to OR as the temperature is increased towards Tc. Currently, we are also investigating the material-specific diffusion barriers entering the coarsening process using local density approximation (LDA) based energy calculations.

3:45 PM P5.8 
COMPETITION BETWEEN GRAIN GROWTH AND GRAIN BOUNDARY SEGREGATION IN BINARY ALLOYS: A MONTE-CARLO APPROACH. J.M.Liu, Z.G.Liu, National Laboratory of Solid State Microstructures, Nanjing University, Nanjing, CHINA.

The kinetics of grain growth in polycrystalline materials represents a very attractive field for recent research activities in microstructural evolution modelings. However, for binary alloys the grain boundary segregation operates as a prominent process controlling the microstructure. We propose a microscopic thermodynamic model where the grain growth is approached by the Q-state Potts model with non-conserved order parameter, and the component diffusion is simulated with the “spin-exchanging” Ising kinetics. A Monte-Carlo algorithm of the competition between the grain growth and the grain boundary segregation is developed. Our simulations reveals significant segregation of the solute species on the grain boundaries for alloys above the co-existence curve, and also second phase precipitation on the boundaries as the alloys are quenched below the co-existence curve. Both the Lifshitz-Slyozov-Wagner law of the microstructural coarsening and scaling law of the grain growth are broken due to the competition between the two processes .

SESSION P6: IN-ROOM POSTER SESSION 
Tuesday Afternoon, December 2, 1997 
4:00 P.M. 
Gardner (S)

P6.1 
BIAXIAL CREEP OF TEXTURED RECRYSTALLIZED CP-TITANIUM TUBING AND CODF-PLASTICITY MODEL. B.V. Tanikella, St. Gobain/Norton Company, Northboro R&D Center, Northboro, MA and K. Linga Murty, North Carolina State University, Raleigh, NC.

Biaxial creep behavior of thin-walled tubing of recrystallized cp-titanium alloy was investigated using internal pressurization superimposed with axial load. Equibiaxial (hoop stress = axial stress) creep tests were performed in the temperature region from 673 to 723 K which revealed creep anisotropy in terms of higher (by about 4) axial creep-rates. Both hoop and axial creep-rate data revealed an activation energy of 46±1 kCal/mole which is identified with that for self-diffusion. Tests under varied stress ratios (hoop/axial) from 0 to 2 were performed at 723 K from which the creep locus was constructed at a constant energy dissipation with power-law dependencies of the component (hoop and axial) strain-rates under varied stress ratios. The experimental results were fit to the modified Hill1s equation with anisotropy parameters, R and P which are also the contractile strain (rate) ratios. The crystallographic texture of the tubing was characterized through inverse and direct pole figures from which the crystallite orientation distribution functions (CODF) were derived. The CODF was combined with plasticity model based on power-law stress dependence of the strain-rate and dominance of basal, prism and pyramidal slip systems were considered. In contrast to the recrystallized Zircaloy, experimental results on cp-Ti deviated from the prism-model predictions. These deviations are tentatively attributed to the relatively high Oxygen-equivalent content which is known to result in reduced CRSS for basal slip. This work is supported by the National Science Foundation.

P6.2 
ISLAND-CORNER BARRIER EFFECTS IN TWO-DIMENSIONAL PATTERN FORMATION AT SURFACES*. Zhenyu Zhang, Tianjiao Zhang, Solid State Division, Oak Ridge National Lab, Oak Ridge, TN; Max G. Lagally, Depts of Materials Science and Engineering and Physics, Univ of Wisconsin, Madison, WI.

Recent scanning tunneling microscope studies of epitaxial growth in both metal and semiconductor systems have revealed the formation of fractal-like islands at submonolayer coverages. When the growth temperature is sufficiently high, only islands of compact shapes are formed. Using kinetic Monte Carlo simulations, we show that the ultimate atomistic process controlling the fractal-to-compact island shape transformation is atom motion across island corners. As long as the diffusion barrier preventing atoms on one edge of an island from reaching another edge is still effective, the island is bound to be fractal-like, a conclusion true on both triangular and square lattices. Possible exotic atomistic processes for effective corner crossing are discussed for different systems of interest, using semi-quantitative total-energy calculations.

P6.3 
DYNAMIC MODELING OF SOLIDIFICATION. Nicolay Bodyagin, Alexey Ufimcev, Arcady Aivakov, Moscow The Inst of Electronic Engineering, Moscow, RUSSIA; Sergey Vikhrov, Ryazan Radioengineering Academy, Ryazan, RUSSIA.

Abstract is non-reproducible.

P6.4 
CELLULAR AUTOMATA MODEL FOR THREE-DIMENSIONAL SIMULATION OF MULTIPLE CRYSTALLIZATION. Serguei Mourachov, Mauricio Cardoso Couto, Laboratorio de Materiais Avanados, Universidade Estadual do Norte Fluminense, Campos-RJ, BRAZIL.

The mass crystallization phenomena occur in various natural and technological processes. In the present work model of the mass crystallization from a solution based on three-dimensional cellular automata was created. The model describes processes of diffusion, nucleation, surface diffusion and interface kinetic phenomena on a mesoscopic level. Temperature, local fluctuation of crystallization parameters and anisotropy of the superficial processes were taken into account. Adequate simulation results describing dependence of solubility from temperature, dependence of critical radius of a crystal nucleus on concentration and temperature, and also kinetic curves for various growth system parameters were obtained. These results allow to speak about applicability of the described model for mass crystallization phenomena simulation.

P6.5 
COMPUTER SIMULATION OF MICROSTRUCTURAL EVOLUTION IN COHERENT TWO-PHASE ALLOYS UNDER APPLIED STRESSES. D.Y. Li and L.Q. Chen, The Pennsylvania State University, Dept. of Materials Science & Engineering, University Park, PA.

A coherent precipitate phase usually has a number of variants which are oriented in different crystallographic directions. Morphology of a precipitate variant is dependent on the elastic energy which is a function of the shape, size, and the lattice mismatch between the precipitate and the matrix. In a multi-variant system, the morphology of a precipitate variant is also influenced by the elastic interaction among variants. An anisotropic distribution of the precipitate variants may result in anisotropic properties of the two-phase material. The variant distribution can be controlled by applying stress during aging, under which, the growth of differently oriented variants becomes selective. A simulation study was conducted to investigate the morphological evolution and the selective variant growth under applied stresses. Several coherent two-phase alloy systems were chosen for case studies, using a diffuse-interface kinetic model. The simulation demonstrates that the morphology of a coherent precipitate is primarily determined by the coherent elastic strain energy; and that an applied stress may affect nucleation and growth behavior of the precipitate phase, thus leading to an anisotropic distribution of differently oriented variants of the precipitate phase.

P6.6 
A STOCHASTIC CELLULAR AUTOMATON TECHNIQUE FOR SIMULATING RECRYSTALLIZATION AND GRAIN GROWTH PHENOMENA AT THE MESOSCALE. Dierk R. Raabe, Institut fur Metallkunde und Metallphysik, Aachen, GERMANY.

A stochastic cellular automaton technique is presented which allows to simulate recovery, recrystallization, and grain growth phenomena at the mesoscale. The local transformation rules for the interface movement are formulated in terms of a discretized phenomenological differential rate equation. This mesoscopic rate equation is derived on the basis of a symmetric, linearized atomic rate equation for diffusion through the interface. The relevant state variables, i.e. the dislocation density and the crystal lattice orientation, are simultaneously updated in each time step. The simulation algorithm which allows to discretely predict kinetics, microstructure, and texture development on a real time scale will be discussed in detail. Various simulations which incorporate experimentally obtained grain boundary mobility data will be presented.

P6.7 
SIMULATION OF ZENER PINNING EFFECT WITH GROWING SECOND PHASE PARTICLES USING PHASE FIELD FORMULATIONS. Danan Fan, S.P. Chen, Los Alamos National Laboratory, Los Alamos, NM; Long-Qing Chen, The Pennsylvania State University, Dept. of Materials Science and Engineering, State College, PA.

The Zener pinning effect with growing second phase particles in Al2O3-ZrO2 composite systems were studied in 2-D using phase field formulations. In these systems, all second phase particles are distributed at grain corners and boundaries. The second phase particles grow continuously and the motion of grain boundaries of the matrix phase is pinned by second phase particles which coarsen through long-range diffusion. It is found that the mean size of matrix phase (D) depends linearly on the mean size of second phase particles (r) for all volume fractions of second phase from 10 to 40%, which agrees well with experimental results. It is shown that D/r is proportional to volume fraction of second phase (f) as f-1/2 for volume fraction less than 30%, which agree with the Hillert and Srolovitz's prediction for 2-D systems. It is also found that D/r is not proportional to f-1/3 and f-1 in 2-D simulations with growing second phase particles.

P6.8 
COMPUTER SIMULATION OF COMPOUND FORMATION: EFFECT OF INITIAL CONFIGURATIONS. Satoshi Kitaoka, Hideaki Matsubara, Fine Ceramics Research Association, Synergy Ceramics Laboratory, Nagoya, JAPAN.

Microstructural evolution of materials often accompanies compound formation in systems with different phases or compositions. This process would be useful to obtain complex microstructures with synergistic functions. The reactivity of the raw materials strongly depends on initial configurations such as a grain size and dispersibility, etc. In this study, the effect of initial configurations of reactants on the microstructural evolution during compound formation due to additive reaction has been simulated mainly by the Monte Carlo method using a two-dimensional triangular lattice. In the case of a reaction with low driving force, decreasing the grain size of the reactants effectively enhances the reactivity in spite of the constant reactant interface length. The reaction site such as the junction between a grain boundary in the reactant and the reactant interface was also controlled by the grain boundary energy of the reactants. The control of the initial configurations makes it possible to design novel structures during the compound formation simulation. This research was carried out under the Synergy Ceramics Project of the ISTF program promoted by AIST, MITI, Japan. Under this Program, part of the work has been supported by NEDO.

P6.9 
CONTINUUM FIELD KINETIC MODEL AND SIMULATION OF PRECIPITATION OF Ll2 ORDERED INTERMETALLICS FROM FCC SOLID SOLUTION. Dipanwita Banerjee, Yunzhi Wang, Department of Materials Science and Engineering, The Ohio Sate University, Columbus, OH; Armen G. Khachaturyan, Department of Ceramic Engineering, Rutgers University, Piscataway, NJ.

A phenomenological field kinetic model of coherent precipitation of L12 ordered intermetallics from a disordered fcc solid solution will be presented. It explicitly takes into account both, the lattice misfit strain and the four types of antiphase domains resulted from the L12 ordering. A coarse-grained free energy functional of concentration and long-range order parameters is formulated based on the concentration wave representation of the L12 ordering and its relation to the atomistic model will be discussed. Precipitation kinetics and microstructural development of phase in Ni-based superalloys are investigated by computer simulations based on the model. The results reveal new features introduced by the L12 ordering which changes the coarsening mechanisms and kinetics and affect greatly the morphology of the mesoscopic microstructure formed by the ordered precipitates.

P6.10 
THREE-DIMENSIONAL SIMULATION OF DIAMOND SINGLE-CRYSTAL GROWTH BY TEMPERATURE-GRADIENT METHOD. Serguei Mourachov, Vladimir Prokofievich Poliakov, Laboratorio de Materiais Avanados, Universidade Estadual do Norte Fluminense, Campos-RJ, BRAZIL.

For modeling the process of diamond growth by the method of a temperature gradient was used model based on three-dimensional probabilistic cellular automata. The model describes diffusion of carbon in a metal-solvent, dependence of solubility of carbon on temperature and kinetic phenomena on a crystal surface. The modeling was carried out for system configurations really used for the perfect diamond single-crystal growth. As parameters of model the real values were used: coefficient of diffusion of carbon in metal - solvent, dependence of solubility of carbon on temperature, geometry of the growth system, temperature and temperature gradient in the high pressure chamber. The results obtained during modeling have quantitative conformity with real processes for dependence of diamond crystal weight on time and geometrical ratio of the crystal sizes.

P6.11 
SEGREGATION EFFECTS AT ANTIPHASE BOUNDARIES. Marcel Sluiter, Hai-Ping Wang, Yoshiyuki Kawazoe, Institute for Materials Research, Tohoku University, Sendai, JAPAN.

Macroscopic deformation behavior is intricately linked to the microscopic interaction of dislocations and defects. In L12 type ordered structures, such as Al3Li and Ni3Al, dislocation motion at the edges of (111) Anti-Phase Boundaries (APB) can be both enhanced and reduced by segregation effects. In this paper, the effect of impurity elements and of off-stoichiometric defects is studied within the context of a generalized Ising model. Approximate solutions to the Ising model are obtained with the Cluster Variation method of Kikuchi. It will be shown that defects can be classified according to features of the multi-component phase diagram.

SESSION P7: MAGNETIC MATERIALS - MODELING AND DEVELOPMENT 
Chair: David M. Wood 
Wednesday Morning, December 3, 1997 
Gardner (S)

8:30 AM *P7.1 
MODELING OF SPIN-DEPENDENT TRANSPORT IN MAGNETIC MULTILAYERS. W. H. Butler, X.-G. Zhang, T. C. Schulthess, D. M. C. Nicholson, Oak Ridge National Laboratory, Oak Ridge, TN; J. M. MacLaren, Tulane University, New Orleans, LA.

In 1988 it was discovered that when a magnetic field is applied to a magnetically inhomogeneous conductor and brings the moments in different regions into alignment, there may be an associated drop in the electrical resistance. This effect is known as the giant magnetoresistance (GMR) effect and arises because electron transport in magnetic materials depends on the relative alignment of the electron spin and the magnetic moments of the atoms. The GMR effect can be used to make sensitive and inexpensive sensors for magnetic fields that are extremely small. These have potential technological applications, e.g. as read sensors in disk drives, as non-volatile random access memory devices, and as position and motion sensors. We have calculated the electronic and magnetic structure of magnetic multilayers and have used the results to model the transport properties of GMR devices. We have used both a fully quantum mechanical transport model based on the Kubo formula and a semi-classical model which utilizes the Boltzmann equation. Both models presently incorporate scattering rates which can either be determined from single films or can be used as fitting parameters. Several interesting phenomena have emerged from the calculations including the possibility of increasing the GMR through a wave guide effect.

9:00 AM *P7.2 
MICROMAGNETIC BEHAVIOR IN GMR MULTILAYERS DEVICES. Jian-Gang Zhu and Youfeng Zheng, Department of Electrical and Computer Engineering, Carnegie Mellon University, Pittsburgh, PA.

Magnetoresistance in GMR multilayers depends on relative magnetization orientation between adjacent layers. In GMR multilayer devices, including spin valve device, domain configurations can be very complex due to intra-layer and inter-layer magnetic interactions. Complicated domain configurations and magnetization switching processes can cause performance degradation and instability. This paper concerns micromagnetic behavior in spin valve, pseudo spin valve and multilayer devices with various geometries. Magnetization processes and switching properties are characterized via dynamic micromagnetic modeling. The model combines the classic micromagnetic theory with the Landau-Lifshitz-Gilbert equations. This dynamic approach allows us to study transient states of magnetization processes, including formation of domain walls and nucleation of new domains. In addition to the effect of various geometic structures, correlations with device microstructure and film intrinsic magnetic properties have been systematically investigated.

10:00 AM *P7.3 
MAGNETO-OPTICAL PROPERTIES OF MAGNETIC MATERIALS. J. van Ek and J.M. MacLaren, Tulane University, New Orleans, LA.

Although the theory for the magneto-optical Kerr effect (MOKE) in polar geometry was developed decades ago, only recently various groups started applying theoretical/computational methods to more complex magnetic materials. In this presentation an overview of the modeling of the MOKE will be presented. Applications to magnetic 3d-metals, binary compounds, L21 and C1b ternary Heusler alloys, as well as magnetic multilayers will be discussed. With results of electronic structure calculations at hand, detailed analysis of the interband contribution to the MOKE can be made. As an example, it will be shown that the MOKE in half-metallic ferromagnetic PtMnSb is not rooted in the semiconducting spin channel, but rather that it originates from the metallic spin channel. Fast band structure codes facilitate the search for materials with interesting magneto-optical properties. In particular for magnetic multilayers, theoretical guidance before the actual fabrication of the multilayers has become available.

10:30 AM *P7.4 
ELECTRONIC STRUCTURE AND PROPERTIES OF PERMANENT-MAGNET MATERIALS. S. S. Jaswal and R. F. Sabiryanov, University of Nebraska, Dept of Physics and Astronomy, Lincoln, NE.

The applications of permanent magnets range all the way from simple household devices to the most sophisticated miniature instruments in science and technology. This fairly large and everincreasing market is a source of impetus for research in this field which led to the discovery of a remarkable magnet, Nd2Fe14B, in 1984. The research in this area is aimed at finding M(Fe,Co)-rich systems with a large uniaxial magnetic anisotropy but minimum reduction in the magnetization and Curie temperature. Since rare earth (R) have large anisotropies, the systems that are normally investigated in this field are the basic units RmMn and their modifications, where . We have performed ab-initio self-consistent spin-polarized electronic-structure calculations to understand the magnetic properties of Nd2Fe14B, RM12 and R2M17 and their modifications with results in very good overall agreement with the experimental data.1-4 These results will be reviewed in this paper.

11:00 AM *P7.5 
MODELING MAGNETIC TAPE PROCESSING. P. B. Visscher, Dept of Physics and Astronomy, Univ of Alabama, Tuscaloosa, AL; Yuksel Gunal, Dept of Chemistry, Duke University, Durham, NC.

Magnetic tape is made by drying a thin layer of a colloidal suspension of magnetized particles, which has been coated onto polyethylene tape. The recording quality depends critically on the degree of particle orientation and short- range order: clustering causes noise in playback. However, the structure is very difficult to probe directly: the opacity of the particles precludes optical microscopy, scattering, and diffusing wave spectroscopy. Most experimental information comes from indirect probes of magnetic susceptibility or rheological response, and computer simulations are a useful tool for relating these measurements to particle properties and motions. We have done simulations of realistic systems of acicular (cigar-shaped) particles, incorporating steric and magnetostatic interactions, as well as Stokesian drag and Brownian forces. The most time-consuming thing is the calculation of the magnetostatic interactions, because it is an order(N2) problem (N is the number of particles). For systems of point particles, fast- multipole methods have been introduced, which reduce this to order(N). We have developed a simplified fast- multipole method for systems of extended particles of arbitrary shape, using an object-oriented hierarchical approach. One long-range objective of research on nano-scale magnetic systems is the synthesis of ordered arrays of particles that will allow very high-density (one bit per particle) recording. We are exploring possible mechanisms for such ordering, and have observed in simulations the formation of a layered structure reminiscent of a smectic liquid crystal, similar to the Schiller layers observed experimentally in aqueous suspensions of colloidal FeOOH. An animated visualization of colloid ordering and magnetic response will be shown. *Supported by the NSF MRSEC program, Award No. DMR-9400399.

11:30 AM *P7.6 
DOMAIN STRUCTURE FORMATION AND EVOLUTION IN FERROELECTRIC MATERIALS: A COMPUTER SIMULATION STUDY. Hong-Liang Hu, Long-Qing Chen, The Pennsylvania State University, Dept of Materials Science and Engineering, University Park, PA.

Computer simulation studies of ferroelectric domain formation and evolution are performed both in two dimensions (2D) and in three dimensions (3D), using a computer simulation model based on time dependent Ginsburg-Landau equations. The effects of various interactions, namely, the long range elastic interaction, the long range dipole-dipole interaction, the local domain wall energy, and the depolarization energy, are studied. The influence of defects and electric field on the transition process is given an in depth consideration.

SESSION P8: SEMICONDUCTING MATERIALS: MODELING AND DEVELOPMENT 
Chair: Anthony Giamei 
Wednesday Afternoon, December 3, 1997 
Gardner (S)

1:30 PM *P8.1 
SEMICONDUCTOR MATERIALS PROPERTIES FROM FIRST PRINCIPLES: A STATUS REPORT. David M. Wood, Department of Physics, Colorado School of Mines, Golden, CO.

`First principles' methods, based on density functional theory using as input atomic numbers and weights and (if needed) the crystal structure, have rationalized or predicted structural and electronic properties of crystalline elemental, binary, and ternary semiconductors, superlattices, quantum wells, etc., for 20 years. For a new material, experiment requires (at least!) information about thermodynamics (stability, phase transitions, phase coexistence), electronic structure (bands andthe nature and strengths of transitions between them), diagnostics (the effects of ordering on optical properties and Raman spectra, etc.), and kinetics (time-dependent phenomena). Progress has occurred on allfronts. Recent theoretical extensions permit computation of (i) the full spectrum and thermodynamic contributions of lattice vibrations, hence the complete (P,T) phase diagram for a crystalline compound, to an accuracy of a few percent; (ii) the (x,P,T) phase diagrams of bulk or arbitrarily strained (or surface reconstruction-driven) substitutionally disordered alloys such as Ga1-xInxP, using generalized multi-atom, multi-neighbor Ising models and computational statistical mechanics methods such as the cluster variation method; (iii) direct quantum mechanical calculation of electronic properties of systems with 103-106 atoms; (iv) some time-dependent phenomena, via `first-principles molecular dynamics,' applicable even to topologically disordered materials. Capabilities and limitations of these methods as currently applied to semiconductors will be reviewed, with examples of how growth has been impacted or clarified. Hybrid approaches whose inputs are taken from (and whose scope of validity is assessed by) the first-principles methods offer great promise for device applications.

2:00 PM *P8.2 
III/V EPITAXY: LINKS BETWEEN FUNDAMENTAL UNDERSTANDING AND DEVICE APPLICATION. Jeff Y Tsao, Sandia National Laboratories, Albuquerque, NM.

III/V compound semiconductors are now in wide use in optoelectronic, sensor and high-frequency electronic devices. Many of these devices rely on epitaxial growth of complex laterally or vertically tailored heterostructures. In this talk we will discuss, with a historical perspective, some of the ways in which fundamental understanding and modeling of the epitaxial growth process has led (or followed) progress in materials development for device applications. We will draw upon examples from the four III/V materials families (nitrides, phosphides, arsenides, antimonides) and from a variety of device types. We will also discuss ways in which improved fundamental understanding might accelerate further development of device applications.

3:00 PM P8.3 
FINITE ELEMENT ANALYSIS OF QUANTUM MECHANICAL TRANSPORT IN A STRAINED SEMICONDUCTOR NANOSTRUCTURE. H. T. Johnson, L. B. Freund, A. Zaslavsky, Brown University, Division of Engineering, Providence, RI.

Semiconductor nanostructures consisting of lattice mismatched materials are characterized by both large nonuniform mechanical strains and novel quantum effects; the coupling of these features presents a unique challenge in modeling the structures. A technique is presented for calculating the wavefunction for a single charge carrier in a heterostructure by means of the finite element method. The effects of both the nonuniform strain as well as the material nonuniformity are incorporated explicitly into the FEM analysis of the quantum mechanical problem. By making use of the fact that the steady-state Schrodinger equation has the same form as the equation governing acoustic wave propagation, a conventional structural mechanics finite element program can be used to solve the full, three dimensional quantum mechanics problem. In this convenient analogy the carrier wavefunction corresponds to the acoustic pressure, while the energy is analogous to the acoustic wave frequency. Using this technique, it is possible to solve for the wavefunction in the presence of an arbitrary, spatially varying potential field. As an example, this finite element technique is used to analyze a columnar structure based on a SiGe quantum well grown between Si barriers which is being studied experimentally at Brown University. Carriers tunnel into and out of quantized states in the well at resonant energies. It is shown that for structures of small enough lateral dimension, the nonuniformity of the relaxed lattice mismatch strain contributes to quantization in the lateral direction, which results in fully quantized zero-dimensional states. The method is shown to be a reliable and convenient tool for analyzing the effects of strain on the performance of semiconductor heterostructures used in quantum devices.

3:15 PM *P8.4 
THE ROLE OF MODELING IN THE DEVELOPMENT OF GAS SOURCE MBE GROWTH OF InTlV ALLOYS. M. J. Antonell, C. R. Abernathy, M. Berding*, A. Sher* and M. Van Schilfgaarde*, Department of Materials Science and Engineering, University of Florida, Gainesville, FL; *SRI Int., Menlo Park, CA.

InTlV alloys have been proposed as potential IR materials based upon calculations done at SRI using an LDA, linear muffin-tin orbital method. In addition to the desirable bandgaps, the lattice constants are expected to differ from those of the respective InV substrate by less than 2 percent. This talk will describe the interplay of theory and experiment used in an effort to develop a synthetic route to realize these materials. Experimentally we have found that attempts to synthesize TlP and TlAs using elemental Tl in gas-source molecular beam epitaxy produce only metallic Tl, with no evidence of the formation of a TlV phase. Introduction of In to form the ternary alloy similarly produces a two phase mixture of InV and Tl under most conditions. The theoretical modeling has been quite useful in explaining these experimental results and in evaluating potential methods of overcoming the kinetic obstacles to synthesis. For example, thermodynamic data obtained using the SRI model suggest that one difficulty in forming TlP and TlAs is the high group V vapor pressures expected above these compounds, pointing toward the use of low growth temperatures as a solution. Another complication identified both experimentally and theoretically is the difficulty in hybridization of the Tl bonding orbitals. Chemically induced hybridization may overcome this problem, and is supported by results of modeling of the growth surface.

3:45 PM *P8.5 
SIMULATION OF CHEMICAL VAPOR DEPOSTION ACROSS DIFFERENT LENGTH SCALES. K. F. Jensen, I. Lengyel, T. Mihopoulos, S. Rodgers, H. Simka, and R. Venkataramani, Department of Chemical Engineering, Massachusetts Institute of Technology, Cambridge, MA.

Chemical vapor deposition (CVD) of thin films involves reactive gas flow combined with surface processes including adsorption, diffusion, nucleation, and growth. The complex coupling of transport phenomena with gas-phase and surface chemical kinetics on different length scales implies that more than one type of modeling approach must be applied to predict the performance of a particular CVD process. It is necessary to predict macroscopic (growth rate, film uniformity, and film composition), mesoscopic (surface morphology), and microscopic (adatom diffusion and reaction) qauntities to realize particular thin film characteristics. A methodology is presented for linking the different length scale models for the CVD process. Two- and three-dimensional finite element (FEM) simulations are used to solve the governing macroscopic conservation equations describing fluid flow, heat and mass transfer with chemical kinetics in CVD reactor enclosures. These computations predict the type and concentration of growth and impurity precursors arriving at the growth front. Three-dimensional kinetic Monte Carlo (MC) simulations of growth front evolution provide understanding of surface morphology evolution and impurity incorporation mechanisms. Ab initio molecular orbital and density functional theory quantum chemistry computations, combined with transition state calculations, are used to determine thermochemical and kinetic data for reaction pathways needed in the different levels of physical models. Each of the length scale-specific simulations is validated through comparison with experimental results from Si and compound semiconductor applications. The linked model is shown to provide new insight into macroscopic and microscopic experimental observations that cannot be accurately represented by a single length scale simulation approach.

4:15 PM P8.6 
MODEL-BASED DESIGN OF DIRECTED VAPOR DEPOSITION. J. F. Groves, H. N. G. Wadley, Univ of Virginia, Dept of Materials Science and Engineering, Charlottesville, VA.

Recently, Directed Vapor Deposition (DVD) has been developed as a new physical vapor deposition (PVD) technology for the creation of engineered films. The primary DVD system configuration employs electron beam evaporation in combination with a supersonic gas jet to focus vapor onto a substrate for film creation. The original DVD system was configured using traditional engineering design strategies employing empirical and theoretical relations to ensure that a functional system was developed. While knowledge of desired material synthesis capabilities (e.g. ability to vary adatom energy) also guided the design, no general material synthesis design tool was employed to construct the system. Subsequently a Monte Carlo model of vapor transport in the DVD system has been created which allows the energy and position of individual vapor atoms to be tracked from vapor source to deposition substrate. For those atoms which land on the substrate, the model records the energy, angle, location, and efficiency of deposition. Since these process parameters fundamentally affect growing film microstructure, the vapor transport model now makes possible ìempiricalî process optimization and helps suggest different DVD system configurations for various applications. The validity of the vapor transport modelís results are verified through comparison with experimental data from DVD and sputtering film synthesis. The model is used to illustrate how different configurations maximize DVD system performance (e.g. deposition efficiency, adatom energy, adatom deposition angle) for flat substrate and continuous fiber substrate coating.

4:30 PM P8.7 
CALCULATION OF COMPOSITIONAL DISTRIBUTION FOR MOVPE GROWN QUATERNARY ALLOYS. Jun Xu, Xiaojin Cheng, I. Physikalisches Institut IA, Aachen, GERMANY.

Metall Organic Vapour Phase Epitaxy (MOVPE) is an important methode to produce a large scale of semiconductor thin films. But the non-uniformity of the compositional distribution in alloys on the sample surface is a general problem in this system, especially, in a horiziontal reactor. In this work we consider the MOVPE growth process as a chain reaction. According to the theory of chemical reaction kinetics and Vergard´s law a calculation model is developed to analyze the dependence of the compositional distribution of quaternary alloys on the position of the sample surface. The results of model calculation are corresponding to the experimental results. According to this model some suggestions are proposed to improve the compositional uniformity on the sample surface for MOVPE grown quaternary alloys.

4:45 PM P8.8 
NUMERICAL SIMULATION OF MICRODEFECT DISTRIBUTIONS IN SILICON CRYSTALS GROWN FROM THE MELT. Talid Sinno and Robert A. Brown, Department of Chemical Engineering, Massachusetts Institute of Technology, Cambridge, MA.

The control of microdefect sizes and distribution in single crystal silicon grown from the melt is becoming critical to the performance of microelectronic devices grown on wafers made of this material. Numerical simulation, performed by synthesizing atomistic, mesoscale and continuum modelling, is an extremely valuable tool for understanding the physical processes leading to microdefect formation and for prediction of the sizes and distributions of these defects as a function of operating conditions for crystal growth systems. We focus on the microdefects formed by the aggregation of native point defects - self-interstitials and vacancies - which are receiving much experimental attention. It is demonstrated that analysis of continuum-scale models based solely on the transport and recombination of native point defects can quantitatively delineate the regions where vacancy and self-interstitial aggregates are dominant and which are separated by the oxidation-induced stacking-fault ring or OSF-ring. The size distributions and densities of these aggregates are predicted by coupling the model of IV dynamics to discrete rate equations (for small clusters) and Fokker-Planck equations which describe the growth or dissolution of noninteracting clusters. Thermodynamic parameters in the diffusion-limited aggregation models are based on results of atomistic simulations using the Stillinger-Weber interatomic potential for small clusters (2-20 point defects) and on phenomenological descriptions for larger clusters (10 - 108 point defects). The combined point-microdefect dynamics model is solved by a finite-element/finite-difference numerical method that preserves the positivity of the microdefect concentrations. Simulations for typical temperature fields and growth rates for Czochralski crystal growth predict a critical temperature range in the cooling crystal for rapid cluster growth and a maximum in the cluster size distribution for large clusters of vacancies; both features agree with experiments on the microdefects seen within the OSF-Ring. The sensitivity of this maximum in the cluster size and of the entire size distribution to operating conditions is explored.

SESSION P9: MODELING DISLOCATIONS AND DISLOCATION DYNAMICS 
Chair: Dennis M. Dimiduk 
Thursday Morning, December 4, 1997 
Gardner (S)

8:30 AM P9.1 
PEIERLS STRESSES FOR FCC METALS FROM FIRST PRINCIPLES.. J. Hartford, B. von Sydow, G. Wahnstrom and B.I. Lundqvist, Chalmers University of Technology and Goteborg University, Dept of Applied Physics, Goteborg, SWEDEN.

Peierls barriers and stresses for edge dislocations in fcc metals (Al, Ni and Pd) are calculated from generalised stacking fault energy curves, . The generalised stacking fault energies are computed with a pseudo-potential plane-wave implementation of density-functional theory. We compare results for the classical Peierls-Nabarro model with a new discrete model (V.V. Bulatov and E. Kaxiras, PRL 78, 4221 (1997)), which link with dislocation mobility. In the case of Pd we also compare the model results for the Peierls stress with a full-scale molecular dynamics simulation using a model potential. In the simulations we have also studied the temperature dependence of the Peierls stress.

8:45 AM P9.2 
KINETIC MONTE CARLO SIMULATION OF DISLOCATION DYNAMICS. Karin Lin, Dept of Physics, Univ of California, Berkeley, CA, and Division of Materials Science, Lawrence Berkeley National Laboratory, Berkeley CA; D. C. Chrzan, Dept of Materials Science and Mineral Engineering, Univ of California, Berkeley, CA, and Division of Materials Science, Lawrence Berkeley National Laboratory, Berkeley, CA.

A kinetic Monte Carlo simulation is used to study the dynamics of a single two-dimensional dislocation moving under an applied stress. The dislocation is assumed to consist of pure screw and edge segments in the slip plane of a finite medium, and motion is controlled by the processes of double-kink nucleation and lateral kink motion. Stress fields of the dislocation segments are calculated within isotropic elasticity theory, and surface effects are taken into account rigorously. Applications of this model to predictions of internal friction and dislocation velocity as a function of stress are discussed. This work is supported by the US DOE, Office of Basic Energy Sciences, under contract DE-AC03-76SF00098.

9:00 AM P9.3 
EFFECT OF PRESSURE ON DISLOCATION DYNAMICS. Sriram Swaminarayan and Richard A. Lesar, Los Alamos National Laboratory, Center for Materials Science, Los Alamos, NM.

The mechanical properties of metallic materials depend strongly on the microstructure in general, and the dislocation substructure in particular. The dislocation substructure, in turn, is determined by the interactions between the different dislocations in the material and the resistance to dislocation motion offered by the discrete nature of the metallic lattice, i.e. the Peierls stress. Both these phenomena are strongly affected by the pressure under which the deformation is taking place. Although the effect of pressure on the core structure of specific dislocations has been studied extensively, the effect of an applied pressure on the annihilation of dislocations has not received much attention. We present the results of a series of atomistic simulations of the effect of pressure on the annihilation of two screw dislocations. We will present data on the change in the core structure, the Peierls stress and the geometry of annihilation. We also compare the energetics results for the interactions between dislocations with linear elastic theory and the applicability of these results to larger scale simulations of dislocation dynamics.

9:15 AM P9.4 
SHORT RANGE DISLOCATION REACTIONS IN BCC METALS: MOLECULAR DYNAMICS AND DISLOCATION DYNAMICS STUDIES. Hanchen Huang and Tomas Diaz de la Rubia, Lawrence Livermore National Laboratory, Livermore, CA; Nasr Ghoniem, Mechanical and Aerospace Engineering Department, UCLA, Los Angeles, CA; Moono Rhee, Hussein Zbib, and John Hirth, School of Mechanical and Materials Engineering, Washington State University, WA.

As a part of an effort to develop a three dimensional dislocation dynamics model for BCC metals, short range interactions of dislocations are investigated at the atomistic level (via molecular dynamics) and the obtained reaction mechanisms and critical parameters are incorporated into the dislocation dynamics model. Two straight dislocations are placed close to each other in a molecular dynamics simulation cell to obtain the reaction mechanisms of the two dislocations without applied external stresses. At the same time, critical parameters, such as largest separation of two dislocation before they react are also obtained. With these physically based reaction mechanisms and critical parameters, dislocation dynamics simulations are carried out to investigate the mechanical response of the material as a function of strain rate and temperature.

10:00 AM P9.5 
ENERGETICS OF DISLOCATION JUNCTIONS. Vasily V. Bulatov, MIT, Cambridge, MA.

Energetics of dislocation junction formation and dissociation are examined in detailed atomistic calculations using semi-empirical interatomic potentials. A cross-over from the linear elastic regime (1/r) of dislocation-dislocation interaction to strongly non-linear interaction at close encounters (core-core overlap) is the primary focus of this work. Based on the atomistic simulation results, an energy balance equation is developed for the driving force of junction formation, which includes both elastic interaction energy and core contributions from each participating dislocation. The magnitude of the energy gain (or cost) of junction formation is an essential atomic plug-in for meso-scale dislocation dynamics simulations, as the means for understanding the nature of microstructural evolution in large dislocation ensembles.

10:15 AM P9.6 
3D MIXED ATOMISTIC/CONTINUUM SIMULATION OF DISLOCATION INTERACTIONS. David E. Rodney and Rob Phillips, Division of Engineering, Brown University, Providence, RI.

Continuum modeling of plastic flow and other mechanical properties requires such phenomenological inputs as the hardening matrix or the annihilation distance between neighboring dislocations. The present work aims to compute such quantities on the basis of 3-dimensional quasicontinuum simulations of dislocation junction configurations with special reference to the case of FCC metals. The resulting geometries and junction strengths depend on features such as the atomic level geometries and the dislocation line directions, showing a complexity which is generally not taken into account in micromechanical models. Simulation of nanoindentation in 3 dimensions is also reported in conjunction with experimental studies, giving insights into such processes as dislocation emission and the interaction of dislocations beneath the indentor.

10:30 AM P9.7 
DISLOCATION INTERSECTION PROCESS INVESTIGATED WITH LARGE-SCALE MOLECULAR DYNAMICS. Shujia Zhou, Dean Preston, and Peter Lomdahl, Los Alamos National Laboratory, Los Alamos, NM.

Deformation of metals and intermetallics is closely related to dislocations and their interactions. One of important questions is how a dislocation moves through a dislocation forest, which is one of the important mechanisms restricting the dislocation motion and contributing to work hardening. In this meeting, we will report our study on dislocation intersection process both in copper and aluminum single crystals with 3D molecular dynamics simulations with up to 3.5 million atoms at very low temperature: Two perpendicular dislocations, one pure screw dislocation and one 60 degree mixed dislocation, have been introduced into the computational system with elastic dislocation theory. These two dislocations immediately dissociate into four partials. Under compressive strain, the 90 degree partial of the 60 degree extended dislocation moves down and passes the extended screw dislocation by LOOPING. Then the 30 degree partial CUTS through the extended screw dislocation and move away together with the 90 degree partial. This LOOPING mechanism is true for various dislocation lengthes and loading rates. This intersection process is strikingly different from previous assumption. With those findings, interaction function in the process of dislocation intersection is derived with good agreement with the values measured from MD simulations. This physical information is essential to macro/meso-level deformation modellings.

10:45 AM *P9.8 
DISLOCATION NUCLEATION IN COPPER: EXPERIMENT AND ACROSS-SCALE SIMULATIONS. R. Phillips,W.W. Gerberich, D. Kramer and E.B. Tadmor.

In a large number of crystalline systems such as Al, Au, Cu, Fe, Mo, Ni, NiAl, Al2O3, MgO, Ti and W, we and others have shown unique and discrete yield events associated with dislocation punching as induced by nanoindentation.(1-3) Unsolved problems include an unequivocal dislocation nucleation criterion and an arrest model associated with the tip coming to equilibrium with the multiple dislocations formed following the first. Initial results suggest that multiple length scales may be involved in this process. For example, on the elasticity side, continuum theory may be fully applicable of predicting nucleation with the correct treatment of the Peierls potential. At the elastic-perfectly plastic extreme, continuum mechanics may predict the extent of plasticity. However, it seems very unlikely that continuum scale models can predict the enormous pressures that can be locally supported after the first plasticity appears. This is commonly referred to as the indentation size effect. It is logical then that quasicontinuum models which involve multiple length scales be applied to such problems. As has been initially outlined elsewhere,(4) an adaptive mesh routine capable of following singularities allows atomistics and finite element solutions to be used simultaneously in such a dislocation nucleation and arrest problem. We have decided to reexamine Cu with a more sensitive instrument which has accuracy in the nN regime. By evaluating <100.

11:15 AM P9.9 
ON THE CONSIDERATION OF CLIMB IN DISCRETE 3D DISLOCATION DYNAMICS. Dierk R. Raabe, Institut fur Metallkunde und Metallphysik, Aachen, GERMANY.

A concept is outlined to incorporate dislocation climb in 3D space and time discretized simulations of dislocation dynamics. Each dislocation line consists of a sequence of interconnected piecewise straight segments which are embedded in a homogeneous linear elastic medium. The dynamics are described by solving Newton's equation of motion for each portion of dislocation. Non-conservative dislocation motion is introduced by considering the osmotic force that arises from emitting or adsorbing point defects at the climbing segment. The osmotic force on each segment depends on the local point defect concentration. The mechanical structure evolution law must thus be complemented with a chemical structure evolution equation. Corresponding formulations are derived using the continuity equation and Fick's first law of diffusion.

11:30 AM P9.10 
COOPERATIVE DISLOCATION GENERATION UNDER APPLIED LOADS VIA KOSTERLITZ-THOULESS MECHANISM: A MONTE CARLO STUDY. M. Ling, M. Khantha and V. Vitek, Department of Materials Science and Engineering, University of Pennsylvia, Philadelphia, PA.

The Kosterlitz-Thouless (K-T) mechanism of cooperative thermally-driven dislocation generation takes place in two-dimensional (2D) crystals just below the melting temperature in the absence of applied loads. It has been recently proposed that a similar cooperative dissociation of dislocation dipoles can also occur in loaded solids at temperatures well below the melting temperature (Phys. Rev. Lett. 73, 684 (1994)). This stress assisted, thermally-driven K-T instability can cause a sudden onset of extensive plastic deformation due to a dramatic increase of the density of glissile dislocations. A Monte Carlo simulation of this instability is carried out on a 2D lattice. The Hamiltonian of this system represents a collection of interacting dislocation dipoles composed of point dislocations subject to a shear stress. The simulation is carried out on a square lattice with periodic boundary conditions. The effective shear modulus which includes the contribution from the plastic strain associated with the dislocation dipoles is calculated as a function of temperature for different shear stresses. The results show that dislocation dipoles dissociate collectively above a critical temperature in the presence of a stress field and this temperature is well below the value obtained in the zero-stress K-T limit. The critical temperatures obtained from simulation are compared with theoretical predictions.

11:45 AM P9.11 
COMPUTER MODELING OF SHEAR DEFORMATION AT HIGH STRAIN RATES USING FAST-MOVING DISLOCATIONS. A. Roos, J.Th.M. De Hosson, Department of Applied Physics, University of Groningen, Groningen, NETHERLANDS; H.H.M. Cleveringa, E. van der Giessen, Delft University of Technology, Delft, NETHERLANDS.

In this paper, shear deformation at high strain rates is modeled using the framework of discrete dislocation plasticity. The total stress state is split into the stress due to the dislocations in an infinite medium, and the complementary stress state due to the linear elastic finite matrix. Sources and obstacles are put into the material. Rules are implemented for the creation and annihilation of dislocations. The dislocations form pile-ups against obstacles that are present in the material. At the tip of such a pile-up, the stress may be high enough to initiate a crack. When the velocity of a dislocation approaches that of the local shear wave speed, the stress- and displacement fields undergo a Lorentz-contraction in the direction of motion. Furthermore, other mechanisms of dislocation damping come into play. The addition of the fields of the moving dislocations may alter the stress field at the tip of a pile-up in such a way that crack opening is achieved on the plane of the shear band, as is observed experimentally. In this work, the high-velocity stress and displacement fields are used, as well as the Gillis-Gilman-Taylor drag-relation. The results are compared with calculations using the static fields and a linear drag relation.