Advances in all-electron local density functional theory approaches to complex materials structure and properties made possible by the implementation of new computational/theoretical algorithms on supercomputers are exemplified in our full potential linearized augmented plane wave (FLAPW) method. In this total energy self-consistent approach, high numerical stability and precision (to 10 in the total energy) have been demonstrated in a study of the relaxation and reconstruction of transition metal surfaces. Here we demonstrate the predictive power of this method for describing the structural, magnetic and electronic properties of several systems (surfaces, overlayers, sandwiches, and superlattices).

This paper gives a brief account of recent calculations of equilibrium configurations, formation energies, and migration energies of intrinsic lattice defects (vacancies, self-interstitials) and complexes of dopant impurities (phosphorus, aluminum) with these defects. The results have been used to provide a comprehensive interpretation of low- and high-temperature diffusion data.

The electronic properties of liquid silicon were computed by coupling molecular dynamics and tight binding methods. By employing the Stillinger-Weber potential, atomic configurations of liquid Si at 1740°C were generated by molecular dynamics. Tight binding parameters chosen to fit fcc,bcc, simple cubic and diamond cubic band structures of silicon, were then used to obtain the electronic properties of the system. All states within 10eV of the Fermi level are found to be delocalized, the density of states spectrum similar (but much broadened) to that of diamond cubic silicon and the optical conductivity is found to be almost featureless with no Drude behavior.

We perform self-consistent density functional calculations on semiconductor heterojunctions, using ab initio nonlocal pseudopotentials, and derive valence band discontinuities for many different lattice matched interfaces. Spin-orbit effects are included a posteriori. A comparison is made with previous calculations, which used empirical pseudopotentials, and with other heterojunction theories. We find reasonable agreement with reported experimental values, and derive some important conclusions about the nature of the lineups.

In the past few years, experimental developments in electron energy loss spectroscopy have allowed study of the dispersion relations of surface phonons, on clean and adsorbate covered surfaces. Also, theoretical analyses of the angle and energy variation of the excitation cross sections have been developed. These have guided the choice of scattering geometry, and also have assisted directly in the interpretation of the data. We review the recent work, with emphasis on the theoretical side.

The sputtering of In and Ga atoms from a “liquid” target composedof gallium coveredby a surface monolayer of indium by incident 5 keV Ar+ ions was simulated using the multiple interaction molecular dynamics technique. Yields, energy distributions, and angular distributions of sputtered atoms were obtained at a temperature above the melting point for the eutectic alloy. Similar information was obtained for a pure gallium and a pure indium target. Our results for layer yield ratios and angular distributions are in good qualitative agreement with Dumke's experimental data for the Ar+, In-Ga system. Absolute yields, however, were found to be sensitive to the detailed nature of the two-body potentials used to describe the atom-atom interactions.

A tight-binding theory is used to study vibrational excitations of the π-bonded chain model of the Si(111)2×l surface. Some aspects of the surface phonon spectrum are discussed. We study the charge fluctuations driven by the vibrational excitations and the surface conductivity associated with the phonons. We find that a longitudinal optical phonon on the surface chains dominates the surface conductivity.

It is generally the case that simple direct interpretation of high resolution electron microscope images is not possible due to the phase contrast imaging modes necessary to achieve atomic level spatial resolution. Therefore, an extensive number of computer programs have been developed to perform electron diffraction and image computations. Both single scattering or dynamical scattering processes can be simulated as well as any form of imaging mode currently available on most modern high performance transmission electron microscopes. Since one is interested in imperfections rather than perfect crystal structures, a large number of sampling points in real and reciprocal space are required. Often, large atom position arrays must be sampled requiring large mainframe computer memories and fast CPU's. High quality displays are also required for realistic image representations and even faster computational methods via television rate digital frame store devices. This paper will be centered about a number of materials areas requiring high resolution electron microscopy computer simulation from atomic structure models. These areas include: organometallic molecules, point defects, surface structure and reconstructions, amorphous thin films, quasi-crystals, semiconductor interfaces and grain boundary structure in metals.

We describe computer-based atomistic simulation studies of defect behaviour in oxides. In illustrating the value of the technique, contact with experiment is discussed through comparison with measured energies of ion migration and other energies. Since the success of these methods is dependent on the extent to which the interatomic potentials used provide a realistic model of the material, consideration is given to transferability of potentials, especially to tenary compounds where the widely adopted practice of fitting potential parameters to available crystal data is not feasible. Our principal examples will include ionic conductivity in gadolinium zirconate, electronic defect properties of BaTiO3 and grossly nonstoichiometric rutile.

The Embedded Atom Method (EAM) is combined with Monte Carlo simulation techniques to determine the equilibrium segregation at internal defects and surfaces. This approach has been applied in the Ni-Cu alloy system to the calculation of the surface composition profiles and the segregation at an edge dislocation. The surface composition profile of these alloys as a function of distance from the surface is found to vary non-monotonically with the top atomic layer strongly enriched in Cu and the near surface atomic layers enriched in Ni. The compositional variation in the core region of an edge dislocation shows enrichment of Ni in the compressed regions of the partial dislocation core and Cu enrichment in the expanded regions. In addition, the composition changes abruptly at the slip plane of the dislocation.

We present the results of computer simulations of body centered cubic (bcc)/melt interfaces, with particular emphasis on the “width” of the interface. Both static and dynamic properties of single crystal/liquid interfaces are examined. The implications for crystal growth near equilibrium are discussed. The results of these computer “experiments” are compared with an extended density functional theory of the solid-liquid interface.

The ab-initio calculation of phase diagrams for metals and alloys using electronic and thermodynamic perturbation theories is discussed.

Ion dynamics simulations are presented to model the system SiO2 over wide ranges of temperature and density as it passes from liquid into glassy states. Essential features of the diffusive mechanism by means of which the system configuration space is explored are examined in relation to density and structure. The presence of a water-like density maximum in the liquid state of this system is established by using the negative pressure regime to stretch the system and thereby sharpen the phenomenon.

A molecular dynamics technique has been used to simulate the melting of a 2-dimensional diatomic crystal and to quench the liquid phase to a solid phase. We demonstrate that a 2-dimensional dense amorphous structure can be obtained and that a 2-dimensional glass transition does exist. Furthermore, atomic vibrations in the liquid can be separated from motion produced by diffusion. The relaxation time during which atoms have a vibratory motion but do not diffuse, diverges to infinity near the observed glass transition. Because of the 2-dimensionality. we are able to display the microscopic processes associated with the glass transition.

A dispersion equation is obtained for plane wave propagation in a discrete random medium. The effect of multiple scattering between the discrete inhomogeneities, statistical correlation in the position of the scatterers, details of the geometry, size and properties of the inhomogeneity via the T-matrix are considered. The resulting effective wavenumber for the average or composite medium depends on the above parameters and the frequency. The complex effective wavenumber in turn can be related to the effective properties of the composite material. The formalism is extremely well suited for numerical computation and can thus yield results suited for engineering applications. The other advantage is that the generality of the T-matrix description makes it convenient for describing acoustic, electromagnetic and elastodynamic problems.

Through the development of a total energy pseudopotential approach, it has become possible to compute structural, electronic, and vibrational properties of solids using only the atomic numbers and atomic masses of the constituent atoms as input. The method has been applied to semiconductors, insulators, and metals; and agreement with experiment for most properties is usually within a few percent.

Applications include the determination of lattice constants for specific structural phases, properties of structural phase transitions, cohesive energies, bulk moduli, lattice vibrational spectra, electron-lattice interaction parameters, and electronic and superconducting properties. A recent example is the prediction of superconductivity in highly condensed hexagonal silicon which was subsequently found experimentally.

Comments are made on total energy band calculations as tools for exploring properties of solids; the importance of fixed spin moment calculations is noted. Use of energy - magnetisation curves to locate magnetic phases is described. Detailed results for fcc and bcc Co and Ni and phase diagrams on the magnetisation - volume plane exhibit two new phases for each metal and show that ferromagnetic fcc Co and bcc Ni break down at small volumes and make first order transitions to nonmagnetic phases in a metamagnetic volume range.

The scope of molecular dynamics problems designed to simulate materials properties is described, focussing on the limits computation imposes on space and time scales, as well as the limits theoretical understanding imposes on our knowledge of interatomic forces. Five strategies for improving the efficiency of the simulations are described. Shock-induced solid-solid phase transformations are discussed to illustrate these ideas.

Over the past two decades, grain boundary structures have been investigated, with increasing frequency, using computer simulation techniques. The accuracy of the potentials used to describe the interaction between atoms is essential to the success of any computer modelling and practically all these studies have been made assuming that the atoms exert pair-wise forces on each other. This paper discusses the applicability and limitations of this assumption and the validity and applicability of the simulation results obtained with pair-potentials. First, it is shown that certain structural features of grain boundaries are independent of the potential used. Such results form, for example, the basis of the structural unit model which relates the structures of boundaries corresponding to different misorientations. Secondly, the multiplicity of boundary structures and its physical implications are discussed. In particular, we report on the complexities of the structure of Σ=5 [001] twist boundaries and on alternative structures of [111] twist boundaries near the Σ=3 twin orientation, both found by computer modelling. These results, which are of general nature, are compared with experimental observations by X-ray diffraction and transmission electron microscopy and a very good agreement is found. However, exact positions of atoms in calculated structures depend on the potential used. Hence, to study properties sensitively dependent on these positions requires to go beyond the pair-potentials.

The authors have previously introduced a method, based on Monte Carlo techniques, for simulation of crystal growth processes in a continuous space. We have applied the method, initially used to simulate growth of two-dimensional Lennard-Jones systems, to treat growth of silicon in three dimensions. The interaction model for silicon is taken to be the recently introduced Stillinger-Weber (S-W) potential, which is a two- and threebody classical potential. Although the early stages of growth seem to be well modelled by the S-W potential, growth of even a single monolayer of epitaxial (111) silicon does not seem to be possible. Modifications to the S-W potential were considered, and found to be unacceptable physically. More accurate treatment of non-ideal atomic configuration energies is necessary to arrive at physically realistic growth simulations.