LAUSR

Empirical potential is vital to the classic atomic simulation, especially for the study of phase transitions, as well as the solid-interface. In this paper, we attempt to set up a uniform procedure for the validation among different potentials before the formal simulation study of phase transitions of metals. Two main steps are involved: (1) the prediction of the structures of both solid and liquid phases and their mutual transitions, i.e. melting and crystallization; (2) the prediction of vital thermodynamic (the equilibrium melting point at ambient pressure) and dynamic properties (the degrees of superheating and undercooling). We applied this procedure to the testing of seven published embedded-atom potentials (MKBA (Mendelev et al 2008 Philos. Mag. 88 1723), MFMP (Mishin et al 1999 Phys. Rev. B 59 3393), MDSL (Sturgeon and Laird 2000 Phys. Rev. B 62 14720), ZM (Zope and Mishin 2003 Phys. Rev. B 68 024102), LEA (Liu et al 2004 Model. Simul. Mater. Sci. Eng. 12 665), WKG (Winey et al 2009 Model. Simul. Mater. Sci. Eng. 17 055004) and ZJW (Zhou et al 2004 Phys. Rev. B 69 144113)) for the description of the solid–liquid transition of Al. All the predictions of structure, melting point and superheating/undercooling degrees were compared with the experiments or theoretical calculations. Then, two of them, MKBA and MDSL, were proven suitable for the study of the solid–liquid transition of Al while the residuals were unqualified. However, potential MKBA is more accurate to predict the structures of solid and liquid, while MDSL works a little better in the thermodynamic and dynamic predictions of solid–liquid transitions.