Predictions of melting, crystallization, and local atomic arrangements of aluminum clusters using a reactive force field

A parametrized reactive force field model for aluminum ReaxFF(Al) has been developed based on density functional theory (DFT) data. A comparison has been made between DFT and ReaxFF(Al) outputs to ascertain whether ReaxFF(Al) is properly parametrized and to check if the output of the latter has correlation with DFT results. Further checks include comparing the equations of state of condensed phases of Al as calculated from DFT and ReaxFF(Al). There is a good match between the two results, again showing that ReaxFF(Al) is correctly parametrized as per the DFT input. Simulated annealing has been performed on aluminum clusters Al(n) using ReaxFF(Al) to find the stable isomers of the clusters. A plot of stability function versus cluster size shows the existence of highly stable clusters (magic clusters). Quantum mechanically these magic clusters arise due to the complete filling of the orbital shells. However, since force fields do not care about electrons but work on the assumption of validity of Born-Oppenheimer approximation, the magic clusters are therefore correlated with high structural symmetry. There is a rapid decline in surface energy contribution due to the triangulated nature of the surface atoms leading to higher coordination number. The bulk binding energy is computed to be 76.8 kcal/mol. This gives confidence in the suitability of ReaxFF for studying and understanding the underlying dynamics in aluminum clusters. In the quantification of the growth of cluster it is seen that as the size of the clusters increase there is preference for the coexistence of fcc/hcp orders at the expense of simple icosahedral ordering, although there is some contribution from distorted icosahedral ordering. It is found that even for aluminum clusters with 512 atoms distorted icosahedral ordering exists. For clusters with N>/=256 atoms fcc ordering dominates, which implies that at this point we are already on the threshold of bulklike bonding.