Both silica (SiO2 ) and water are ubiquitous in our environment and our technologies. They come together in two important processes: the oxidation of silicon by steam, which is a process step in integrated circuit fabrication, and the hydrolytic weakening of quartz, which helps us understand plate tectonics. In the former process water molecules must diffuse through an amorphous silica layer and in the latter they must diffuse through quartz to reach dislocations and bubbles to promote plasticity.

We have taken a molecular fragment of SiO2 based on an a quartz c channel, with a hydrogenated external surface and with an internal water molecule. Ab initio, local density, total energy calculations, relaxing the water molecule and the atoms lining the channel, have been applied to discover the insertion energy of water in quartz. In addition, this quantity has been examined in a c axis dislocation core, in order to evaluate the ease of pipe diffusion in the case of hydrolytic weakening and to better simulate amorphous SiO2 in the case of silicon oxidation.

Silicon is an ideal model for strong covalent solids because it is an elemental semiconductor amenable to many well-defined structural, mechanical and electronic experiments. It was shown some 25 years ago that electronically active dopants enhance dislocation mobility in silicon, but recently Sumino and co-workers have shown that phosphorus also strongly pins dislocations at low stresses. Almost paradoxically the same dislocations move with enhanced mobility once they are in motion at higher stresses.

We have performed local density functional pseudopotential calculations on a molecular fragment taken from the core of a 90° partial dislocation of silicon. The surface dangling bonds were hydrogenated and phosphorus was substituted for silicon in various positions. Our results suggest that there is a strong valence effect in operation that causes phosphorus to bind very strongly to three fold coordinated sites in the core, simultaneously losing electrical activity. It is this binding that inhibits dislocation motion, because motion would require the coordination of phosphorus to change from three-fold to the more energetic fourfold.

The stresses, displacement gradients, and Eshelby's F and M integrals are obtained for two crack orientations in an EAM atomic model of aluminum. For a sharp crack, the stresses are shown to agree quite well with the linear elastic prediction, and F is essentially path independent and also in good agreement with the linear elastic prediction. When dislocation emission and blunting ensues, the path independence of F disappears. In addition, for circular contours with origin at the crack tip, the M-integral is linear in contour radius with slope equal to twice the surface energy and zero intercept for a sharp crack, but acquires a nonzero intercept as blunting occurs. The shift in intercept is related to the movement of singularities away from the origin.

An atomistic simulation of H-Pd(100) provided a phase diagram for the c2×2 H overlayer phase. The Embedded Atom Method (EAM) calculated energy of each configuration of atoms and the Metropolis Monte Carlo algorithm equilibrated the structure and generated configurations from which to sample the structure factor for the H overlayer. The procedure provided the expectation of the square of the structure factor modulus, < |S2| >, as a function of temperature at three coverages. The inflection point of the < |S2| > versus T curve estimated the critical temperature for disordering, Tc,, for one value of coverage, θ. The plot of Tc versus θ, the phase boundary for the c2×2 phase, lay about 125 K below the experimentally determined boundary. A comparison of the energies of ordered and disordered phases showed ΔE = 0.016 eV per hydrogen atom. Equating this unrealistically small energy difference to thermal kinetic energy (3/2)kBTc at the critical temperature implies Tc ≈ 100 K. Obtaining – |S2| > values relatively free of noise at such low temperatures required large numbers of Monte Carlo steps. The c2×2 phase is the experimentally determined stable low temperature phase, and was assumed to be the lowest-energy phase possible in this simulation. The very small ΔE indicates that some other ordered phase may be more stable than c2×2 in the EAM model.

We propose a simple interatomic potential model for zirconium which successfully reproduces the phonon dispersion curves in both the α- and the β-phases as well as the temperature-induced phase transformation. Two interesting phenomena that help to understand the anomalous diffusion in the high-temperature β-phase are observed: the spontaneous formation of Frenkel-pairs and the highly correlated walk of the vacancy.

Surface and near-surface processes have been studied during low energy Xe ion bombardment of Si (001) and fcc surfaces using molecular dynamics simulations. Defect production is enhanced near the surface of smooth Si (001) surfaces with respect to the bulk in the energy range 20–150 eV, but is not confined exclusively to the surface layer. The extent and qualitative nature of bombardment-induced dissociation of small fcc islands on an otherwise smooth fcc (001) surface is found to depend strongly on island cohesive energy.

We directly simulate the dynamics of carbon and silicon clusters to study the relevance of the probability distribution of coordinates to experimentally measured phenomena. We believe that the high temperature conformations are not necessarily related to the ground state minimum, and that care should be exercised when comparing ground state T = 0 calculations to experiments performed at high temperature.

Recently it has been proposed that as two metal surfaces are gradually brought toward contact, atomic layers can collapse or avalanche together when the interfacial spacing falls below a critical distance. This avalanche can occur regardless of the stiffness of external supports and is accompanied by a discontinuous drop in the adhesive binding energy. Here we re-examine this phenomenon using a simple zero-temperature Monte-Carlo simulation using the Equivalent Crystal Method. In particular, we investigate the effect of loss of registry between the two contacting surfaces in order to delimit the circumstances under which avalanche can occur. We find that the avalanche is suppressed when the two surfaces are sufficiently far out of registry, at least when only a few layers near the surface are allowed to relax.

A modified version of the multiple-interaction code SPUT2 has been used to simulate impacts of 63-atom Cu clusters on six–layer Cu targets. Simulations were carried out with cutoff times of 100 and 500 fs for an incident cluster energy of 63 keV (1 keV/atom). Significant enhancements were observed in the maximum potential and kinetic energies achieved in the early phase of the collision cascade. Some hard collisions yielded atoms with potential energies as high as 925 eV (in the CM frame). This is almost twice the energy allowed in an isolated two-body collision. The number of hard collisions per time-step vs potential energy is well-fitted with a decaying exponential, allowing extrapolation to higher energies. These results together with similar results for Al clusters impacting Au targets suggest that non-linear collisional effects cannot explain the high D–D fusion rates seen in Beuhler, Friedlander, and Friedman's recent experiment [l].

The migration of a (100) Θ = 43.6° (Σ29) twist grain boundary is observed during the course of a molecular-dynamics (MD) simulation. The migration is investigated by calculating changes in the planar structure factor Sp(k), the orientation of interparticle bonds in the migrating plane, and the motion of the center of mass of planes near the grain boundary. It is found that (a) migration is accompanied by sliding, and (b) each migration step consists of a series of local bond rearrangements. It appears that, as a precursor to migration, Frenkel-like defects are produced at the grain boundary.

The effect of structure and geometry on grain boundary self-diffusion is investigated. Using static structures found by relaxation techniques (conjugate gradients) as starting points for a molecular dynamic simulation of a bicrystal model, diffusion coefficients and activation energies are calculated for (100) fcc Al and bcc α-Fe tilt boundaries. These quantities are derived by monitoring the mean-squared displacement of atoms in the grain boundary region as the simulation progresses and as the temperature of the simulated solid is changed. The angular, directional, and structural dependence of the simulated diffusion are discussed and compared to experimental measures and theoretical predictions. The implementation of the molecular dynamics algorithm on a parallel supercomputer is also briefly discussed to illustrate the performance benefits these computers make possible.

The energies and configurations of interstitials and vacancies in the ordered compound CuTi were calculated using atomistic simulation. Vacancies created by the removal of either a Cu or Ti atom resulted in a vacant Cu site, with an antisite defect in the latter case. The vacancy at the Cu site was found to be very mobile within two adjacent (001) Cu planes, resulting in two dimensional migration. Interstitials created by inserting either a Cu or Ti atom had complicated configurations containing one or more antisite defects.

A Cu(110) surface can exhibit missing-row type reconstructions if potassium or oxygen atoms are chemisorbed on the surface. In the case of potassium every second of the closepacked rows in the top layer of the copper surface is removed. The reconstruction induced by oxygen on the same surface is of a very different kind where the missing rows are formed by picking out every second atom in the close-packed rows of the copper surface. The energetics and the driving force behind the two different reconstructions are discussed and compared on the basis of effective medium calculations.

A systematic geometrical procedure for predicting favored boundaries in the structural unit model is presented. The method is applicable to both symmetric and asymmetric tilt boundaries. The predictions are confirmed by modeling the structures of tilt boundaries belonging to low symmetry ( and [221]) axes in f.c.c. and b.c.c. structures. The results confirm the applicability of the structural unit model for relatively high-index tilt axes.