Calculation of Viscosity of Liquid Nickel by Molecular Dynamics Methods

Computer models of casting processes are becoming useful tools in the foundry industry. These models are used to simulate the heat and fluid flows that control the microstructural development of castings. The accuracy and usefulness of these computer models depend critically upon the availability of reliable thermophysical properties of metallic melts. For instance, the simulation of the solidification of metallic alloys includes the numerical solution to the conservation equation for momentum. The viscosity of liquid metals is therefore required in the solution to the momentum equation. Unfortunately, viscosity data for metallic alloys are sparse. Even the viscosity data for pure liquid metals reported in the literature are quite scattered. As an illustrative example, the viscosity of pure Ni, as measured by various investigators may vary by as much as 50% (1). This lack of reliable experimental data and advances over the past decade in assigning credible interatomic potential functions for transition elements are compelling arguments for attempting to predict viscosity liquid transition metals with atomistic computer simulation methods. In the present contribution, we concern ourselves with the application of the method of molecular dynamics (MD) and the Embedded Atom Method (EAM) to calculating the viscosity of a liquid transition metal, namely Ni. For high temperature liquid metals, on account of the very large discrepancies in experimental data, it has been recommended that values of viscosity calculated from theoretical models might be as appropriate as the experimental value (1). To date such theoretical and semiempirical predictions of viscosity have relied on (a) input from experimental pair distribution functions for representing the liquid structure and/or (b) pair potentials for modeling the atomic interactions. The choice of a pair potential cannot be justified for transition elements (2). Apart from first-principle quantum calculations, the EAM many-body interatomic potentials represent at this time one of the most realistic ways of modeling liquid transition metals. We show, here, that MD simulations in conjunction with the EAM derived potentials provide a reliable method of predicting viscosity.