Empirical potential derivation for ionic materials

A program has been developed for the derivation of empirical interatomic potentials, with particular regard to ionic materials and the use of shell models, incorporating two new methods of fitting. Concurrent fitting of multiple structures is found to enhance greatly the reliability of the derived potentials and can lead to a physically sensible O-O potential without the use of constraints. Inclusion of gas-phase cluster information appears to be beneficial even within an ionic model. By combining free-energy minimization with empirical fitting based on displacements, rather than gradients, it is now possible to determine interatomic potentials with correct treatment of thermal effects and the zero-point energy.