Molecular dynamics simulation of microstructure transitions in a large-scale liquid metal Al system during rapid cooling processes

We perform a molecular dynamics simulation of microstructure transitions in a large-scale system consisting of 400000 atoms of liquid metal A1 by the Clare supercomputer. A cluster-type index method is proposed to describe the structures of various short-range-order clusters in the liquid system. It is demonstrated that the icosahedron cluster (12 0 12 0) plays the most important role in the microstructure transition and that some larger clusters (containing more than 130 atoms) are formed in the system during the rapid cooling processes. It is obvious that the larger clusters are formed by means of combining some middle clusters, and that the middle clusters are formed with several smaller clusters. However, the larger clusters are not formed to be the multi-shell configuration as shown in the mass spectrum analysis of the cluster configurations of A1 obtained by gaseous deposition, ionic spray methods and so on. This result can be used to explain the essential distinction between the cluster configurations of A1 formed in two different ways.