Dynamics of Lennard-Jones clusters: A characterization of the activation-relaxation technique

Abstract

The potential energy surface of Lennard-Jones clusters is investigated using the activation-relaxation technique (ART). This method defines events in the configurational energy landscape as a two-step process: (a) a configuration is first activated from a local minimum to a nearby saddle-point and (b) is then relaxed to a new minimum. Although ART has been applied with success to a wide range of materials such as a-Si, a-SiO2 and binary Lennard-Jones glasses, questions remain regarding the biases of the technique. We address some of these questions in a detailed study of ART-generated events in Lennard-Jones clusters, a system for which much is already known. In particular, we study the distribution of saddle-points, the pathways between configurations, and the reversibility of paths. We find that ART can identify all trajectories with a first-order saddle point leaving a given minimum, is fully reversible, and samples events following the Boltzmann weight at the saddle point.