The static dipole polarizability for the ground state of C(3Π ), N(4∑+), O(3Π ) and F(2∑+ ) have been calculated using ab initio molecular orbital techniques. The polarizabilities are obtained from unrestricted Hartree-Fock and Many-Body Perturbation Theory wave functions with and without spin projection, which are computed by finite-field differentiation. The effect of spin contamination is shown to be important on the calculation of static dipole polarizabilities for the ground state of C, N, O and F.

The temperature dependence of the in-plane C11 C22. C12 and C66 modes between 80 and 260 K of superconducting crystals of Bi2Sr2Ca1Cu208 have been obtained via the measurements of ultrasonic-velocities. The anisotropic elasticity in the a-b plane of single crystal Bi2 Sr2Ca1Cu2O8 is manifest. The shear modulus of sound propagation along the [110] with the polarization has been also calculated and shows an overall trend of softening over a wide temperature range above Tc. The shear modulus C6 6 shows three obvious softening minima around 240–250 K, 150 K and 100 K.

The possibility of passivating the Si(100) surface by adsorption of Group-VI atoms (S and Se) is investigated through first-principles calculations. The structure of the ideal (1×1) configuration with the Si surface dangling bonds saturated by full monolayer coverage is examined in detail. The Group-VI adsorbates form covalent bonds to the substrate with bond-lengths very close to the sums of the covalent radii. The bond-angles are larger than in bulk configurations of the Group-VI elements. The ideal (1×1) configuration gives rise to a surface electronic state with large dispersion spanning the entire band-gap of Si. Deviations from this configuration by in-phase or out-of-phase tilting of the adsorbate atoms result in energy costs which can give qualitative information on the relative strength of adsorbate-adsorbate and adsorbate-substrate interactions.

The atomic origins of the intrinsic mechanical and thermal properties of the interfaces between niobium and its carbides and silicides are explored using both the layer Korringa-Kohn-Rostoker (LKKR) technique and model cluster calculations. The relative stability of an idealised [100] interface is predicted from charge transfer arguments.

The electronic structure of the interface between α-Al2O3(0001) and Nb layers has been calculated using the empirical tight-binding method and the slab model. It has been shown that a direct chemical bond of both covalent and ionic characters can be established at the interface between the surface O atoms of α-Al2O3 and the Nb atoms, which is consistent with the recent photoelectron spectroscopy study. General trends of the electronic structure and chemical bond at the interfaces between α-Al2O3.(0001) and a series of 4d transition metals have been examined using the same theoretical method. In all cases, there exist similar interactions between the surface O atoms and the metal atoms, and it has been observed that the occupancy in the portion of the antibonding peaks of the local density of states at the interface increases as the atomic number of the transition metal increases.

A unique set of physico-chemical properties of the implantation phases (transition metal carbides and nitrides) and their wide-scale application in production processes motivated an intensive research in the electronic properties of the phases [1, 2]. A great attention attaches recently to the study of specific features of the electronic surface states of refractory compounds, the role of the electronic states being decisive in the formation of adsorption, catalytic, emission and other characteristics of the compounds. Available calculations of the energy spectrum of the high-symmetry faces of some 3d-5d-metal carbides and nitrides performed with the use of the band theory methods [4, 5] made it possible to establish principal features peculiar to the formation of surface states for ideal systems.

The energies of carbon defects in silicon have been calculated, using an empirical classical potential, and used to infer defect properties and reactions. The results accord well with a variety of experimental data, including defect structures, activation energies for defect motion, and coupling to strain.

The field of geological materials is rapidly opening up to computer modelling and ice poses stimulating new problems. The normal pressure and 0°C form of ice is ice Ih in which the oxygen sublattice is wurtzite. Imagining first that the protons occupy O–O “bond-centre” sites, it is generally believed that in ice Ih these protons are displaced along the bond direction in a correlated fashion such that each oxygen has two close protons (at about 1Å ) and two more distant ones (at about 1.7Å ). The correlated “disorder” of protons causes frustration of certain atomic movements, such as water molecule reorientation (as measured by dielectric relaxation times) and dislocation glide. Frustration can be relieved (and dislocation glide facilitated) by proton order defects, such as ionic defects, like an oxygen surrounded by only one close proton (OH−) or three (H3O+), or Bjerrum defects, like an O–O link with no intervening proton (L defect) or two protons (D defect).

We present results of ab initio total energy calculations on the formation of these defects.

We report the results of ab-initio molecular dynamics simulations for small clusters of Si and C atoms. Ground-state geometries and vibrational spectra are presented. We also describe a Bayesian spectral estimation technique which we have found to be useful in analyzing molecular dynamics trajectories.

In order to calculate thermodynamic properties of disordered alloys, it is necessary to extract certain parameters (namely effective cluster interactions) from electronic band structure models. Several of these methods exist, thus necessitating a comparison of their accuracy and convergence. Direct configurational averaging is performed in both a canonical (concentration-dependent) and grand canonical (concentration- independent) scheme. These results are compared with those derived from the Connolly-Williams Method and “exact” results obtained on a simple, tight-binding, d-band model.

A dynamical simulation of curved grain boundaries composed of pyramidal-shaped ledges has shown that the boundaries can move by local conservative shuffles of atoms or groups of atoms such that one adjoining crystal grows at the expense of the other. In the model system studied, the shuffles often take the form of correlated rotational displacements about the axis normal to the boundary. The simulations provide support for the atomic mechanism proposed by Babcock and Balluffi to explain their observation of grain boundary migration without the participation of SGBDs.

The atomic and electronic structure of the {122} Σ=9 grain boundary in cubic SiC has been calculated for the first time using the self-consistent tight-binding (SCTB) method. An atomic model consisting of zigzag arrangement of 5-membered and 7-membered rings similar to that in the same boundary in Si or Ge has been constructed from a HREM image, although Si-Si and C-C wrong bonds are repeated alternately at the interface in this model. We have also performed calculations of the same boundary in Si using the SCTB method for comparison, and have obtained the results similar to those previously obtained by other theoretical schemes. The calculated boundary energy in SiC has shown that the present atomic model can exist stably as compared with the two surfaces, and the calculated boundary electronic structure in SiC has no deep states in the gap as well as that in Si. However, it has been found that the the increase in the electrostatic energy caused by the wrong bonds is a large part of the present boundary energy in SiC differently from that in Si, and it has been shown that the wrong bonds introduce the wrong-bond localised states at the band edges and within the valence band.

In a recent publication [1], we presented results of electronic-structure calculations on Σ5 grain boundaries in Cu. These calculations were performed using the real-space multiplescattering theory (RSMST), a first-principles method which does not rely on translational invariance and Bloch's theorem [2, 3], and hence avoids the problems associated with treating interfaces in a periodic repeating slab geometry. A brief summary of this method follows.

The RSMST is based on the concept of semi-infinite periodicity (SIP), defined as the regular repetition along a given direction of a scattering unit (atom, planes of atoms, etc.), or a set of such units. Systems with SIP possess the property of removal invariance, which states that the scattering properties (scattering matrices) of any such system remain invariant when an integral number of scattering units is removed from, or added to, the free end of the system. Using this property in conjunction with multiple-scattering theory, one can determine the electronic Green function, and hence all one-particle quantities such as the density of states (DOS), directly in real space.

The essence of the method consists in a prescription for the proper renormalization of the scattering properties of the boundary sites of a cluster of atoms. Unrenormalized or “bare” sites in the interior of the cluster describe the region of interest, such as a grain boundary, while the renormalized boundary sites represent the infinite medium surrounding the cluster. This “dressing up” of the cluster is done independently for each part of a system that is characterized by its own SIP, so that grain boundaries between essentially arbitrary crystal structures can be treated.

The calculations described previously [1] used given atomistic configurations and electronic one-particle potentials and computed the local DOS at atoms of interest within the grain boundaries. We are currently extending our method to incorporate charge self-consistency and total-energy capabilities. In connection with this, we have implemented four different methods of expressing mathematically the invariance properties of semi-infinite systems, and are evaluating them for optimal rates of convergence and computational efficiency. Calculations on Cu (100) monolayers will be used to illustrate these methods.

The interaction between screw dislocations and [1 1 0] symmetric tilt boundaries is investigated by atomistic calculations. In order to study the differences between fcc and ordered alloys and to study the effect of increasing ordering tendency, many-body potentials representing Cu, Cu3Au and Ni3Al were used. For the ordered alloys, the different possible ordering configurations of the boundaries that were studied are discussed.

We outline a new method for the calculation of free energies from a cumulant analysis of the internal-energy distribution function obtained from a Monte-Carlo simulation. The new methodology is validated by comparing the free energy calculated for a perfect crystal from this approach with the free energy calculated by a temperature integration scheme. By comparing the two methods for an inhomogeneous system containing grain boundaries, we investigate the effects of the local-strain distribution on the free energy; for the superlattice of twist grain boundaries studied here, the effects are found to be small.

A self-consistent non-orthogonal semi-empirical tight binding Hamiltonian is proposed for GaAs, or any sp system, which is simple, reliable, transferable, accurate and fast to evaluate. Matrix elements are functions of charges, distances between atoms and simple cosines of angles between s and p-electron densities and interatomic vectors which maintain the simplicity of Slater-Koster parameterizations. The tight binding scheme is fit against a large data base of local density functional derived total energies for systems of differing coordination and geometry. The Hamiltonian fulfills the correct Virial constraint, invokes the physically correct relationship between overlap and kinetic energy matrix elements and defines charges via Mulliken or Löwdin schemes. Such Hamiltonians will allow the reliable simulation of statistical mechanically interesting systems of order hundred or more atoms over physically useful periods of time of order tens to hundreds of thousands of time steps within not unreasonable supercomputer budgets.

The microscopic structure of a tilt and a twist boundary in germanium are explored using a state-of-the-art total energy calculation. The structure of the tilt boundary (Σ = 5 (310)) is found to be simple as it exhibits a well defined minimum energy structure, consistent with previous experimental and theoretical results on this and other tilt boundaries. The structure of the twist boundary (Σ = 5 (100)), however, is found to be very complex. The boundary bonds are distorted and weak, and their weakness makes the twist boundary exhibit a wealth of local energy minima. The different types of energy minima for the twist boundaries are identified and studied in some detail. We summarize the consequence of the orientation of the two crystal grains upon the microscopic structure of the boundaries, and speculate about future applications.

This paper reports two applications of computer simulations based on the embedded atom method to the structure of grain boundaries. First, the equilibrium segregation at symmetric twist grain boundaries in Ni-Cu alloys is computed. These simulations include both compositional rearrangements and atomic displacements. It is found that Cu segregates to the boundary and that the change in composition is confined to a few atomic layers near the boundary. The degree of segregation was found to be greater for Ni-rich alloys and for large angle boundaries. The variation of the concentration within the planes parallel to the boundary for a low angle twist boundary is studied and the variations are correlated with the positions of the boundary dislocations. The second application addresses the possibility of structural disorder in the Σ5(001) twist boundary in Au. These results show that defects corresponding to local motion of the boundary normal to the interface are low in energy. The energetics of these defects are then mapped onto the body-centered-solid-on-solid model of surface roughening. This model predicts that a roughening transition will occur in this grain boundary at 100 K or 500 K depending on the choice of interatomic interactions. The effect of this roughening transition on the x-ray scattering from the boundary is also computed.

Non-equilibrium molecular dynamics is used to study the indentation of an ideal metallic surface by a hard spherical ball. The problem is symmetric about the axis of the ball and we choose to study the deformation in two dimensions. We use the simple embedding function F(ρ) = 2eρln ρ along with the Lennard-Jones two body potential to model a material resembling copper. We present a calculation of the st'ress field (from the internal atomistic description) for both elastic and plastic deformations. The material deforms plastically through the creation of edge dislocations and the flow of material around the sides of the indenter. The creation of dislocations is reversible—they return to the surface (anneal out) when the indenter is removed. The plastic flow leads to a permanent indentation in the surface.

The embedded atom type potentials and static relaxation method combined with steepest gradient computational technique have been used to simulate the grain boundary cohesive energies, the distribution of electron density and stress field in the grain boundary region, and other related problems of [100], [110] and [111] symmetric tilt grain boundaries in Ni3 Al with different grain boundary geometrical index and composition. Their relations with the segregation of boron, behaviors of the grain boundary, and especially the stoichiometrical effect of B induced ductility have also been studied and discussed.