Two procedures were developed to fit interatomic potentials of the embedded-atom method (EAM) form and applied to determine a potential which describes crystalline and liquid iron. While both procedures use perfect crystal and crystal defect data, the first procedure also employs the first-principles forces in a model liquid and the second procedure uses experimental liquid structure factor data. These additional types of information were incorporated to ensure more reasonable descriptions of atomic interactions at small separations than is provided using standard approaches, such as fitting to the universal binding energy relation. The new potentials (provided herein) are, on average, in better agreement with the experimental or first-principles lattice parameter, elastic constants, point-defect energies, bcc–fcc transformation energy, liquid density, liquid structure factor, melting temperature and other properties than other existing EAM iron potentials.

The recently published semi-empirical potentials of Finnis and Sinclair for the metals V, Nb, Ta, Mo and W appear to give unphysical results for properties involving small interatomic separation. This is remedied by adding to the potentials cores fitted to electron gas calculations on dimers. The adjusted potentials are shown to predict a more realistic pressure-volume relationship. Interstitial formation energies are calculated for various configurations, using quenched molecular dynamics and static relaxation. Some preliminary results on interstitial migration are presented.

The behaviour of copper atoms in dilute solution in α-iron is important for the microstructural changes that occur in ferritic pressure vessel steels under fastneutron irradiation. To investigate the properties of atomic defects that control this behaviour, a set of many-body interatomic potentials has been developed for the Fe—Cu system. The procedures employed, including modifications to ensure suitability for simulating atomic collisions at high energy, are described. The effect of copper on the lattice parameter of iron in the new model is in good agreement with experiment. The phonon properties of the pure crystals and, in particular, the influence of the instability of the metastable, bcc phase of copper that precipitates during irradiation are discussed. The properties of point defects have been investigated. It is found that the vacancy has lower formation and migration energy in bcc copper than in α-iron, and the self-interstitial atom has very low formation energy in this phase of copper. The threshold displacement energy in iron has been computed as a function of recoil orientation for both iron-and copper-atom recoils. The differences between the energy for the two species are small.

Using the approach of Finnis and Sinclair, N-body potentials for copper, silver, gold and nickel have been constructed. The total energy is regarded as consisting of a pair-potential part and a many body cohesive part. Both these parts are functions of the atomic separations only and are represented by cubic splines, fitted to various bulk properties. For the noble metals, the pair-potentials were fitted at short range to pressure-volume relationships calculated by Christensen and Heine so that interactions at separations smaller than that of the first-nearest neighbours can be treated in this scheme. Using these potentials, point defects, surfaces (including the surface reconstructions) and grain boundaries have been studied and satisfactory agreement with available experimental data has been found.

A general problem in particle-based modeling is one of showing the various configurations which emerge from simulations. A particular problem is to determine the local coordination, distinguishing fcc, hcp, bcc, and other relatively close-packed structures. Here we describe an approach to this problem, which attempts to optimize the differentiation between different structure. The method is then applied to differentiate dynamically stabilized phases from mechanically stable ones in bcc crystals, and to identify dislocation defects in random hexagonal close-packed colloidal suspensions, in each case extracting much more information from the raw data than had previously been possible.