A molecular-dynamics study of the dynamic properties of liquid rubidium. I. Collective correlation functions

Abstract

We present results of a molecular-dynamics study where we calculated the dynamic structure of six states of liquid rubidium. The systems are chosen along the liquid-gas coexistence curve; four of these states have been considered in a recent neutron-scattering experiment by Winter et al. (1987) The interatomic forces are based on an Ashcroft empty-core pseudopotential. This present paper-the first of two contributions-contains results for the collective correlation functions. For the dynamic structure factor S(q, omega ) we find for temperatures up to approximately 1700 K very good quantitative agreement with the experimental results. Discrepancies which we encountered in the high-temperature state (1873 K) may undoubtedly be attributed to the inadequacy of the underlying interaction: near the critical point a metal/nonmetal transition occurs for Rb and the interatomic forces (based on mean electronic density) we used are no longer able to describe the complex changes in the electronic structure correctly. We find that the behaviour of the intermediate scattering function near qp (the position of the main peak in the structure factor S(q)) may be very well understood in terms of a simple memory-function model, involving static properties only. The interpretation of both the longitudinal and transverse current correlation functions using reliable hydrodynamic and memory-function models has revealed that, for intermediate- and high-temperature states, the temperature influence enters only via the static moments of the correlation functions while the parameters of the models (which are determined in a least-squares fit to the correlation functions) turn out to be practically temperature independent. Once more we experience that it is extremely difficult to determine elastic and thermodynamic quantities from the computer data of the correlation functions in a reliable and accurate way: the quality of the results may be very sensitive to the underlying model; results obtained via different routes and interpretations may differ substantially and discrepancies with experiment of 10-20% have to be considered as normal.