Efficient statistical mapping of energy surfaces of nanoclusters and molecules

Abstract

A statistical technique to efficiently map out the energy surfaces of nanoclusters and molecules is described. Global energy minimizations are performed to reach of the catchment basins of the lowest energy stationary points. Saddle points are located by using a large value of the iterative energy change as the stopping criterion of a final local relaxation. Minima are derived from saddle points by simply tightening the stopping criterion and continuing the relaxation. A statistical approximation to the widths of the paths in phase space between saddle points and minima is obtained. Application is made to argon clusters of 7 and 38 atoms.