Comparison of embedded-atom models and first-principles calculations for Al phase equilibrium

We have performed total energy calculations on Al in the face-centered cubic (fcc), body-centered cubic (bcc), ideal hexagonal close packed (hcp), and simple cubic (sc) crystal structures over a range of unit cell volumes. We employed density functional theory (DFT) in the local density approximation (LDA) and the generalized gradient approximation (GGA) as well as two different forms of the embedded-atom method (EAM) empirical atomistic potential with the aim of evaluating the predictive range of the EAM relative to first-principles methods. All four calculations correctly give the fcc structure as the preferred one at zero pressure, with the DFT results in good agreement with the experimental equation of state and the model potentials in exact agreement (by construction). The hcp structure is found to be fairly close in energy to the fcc structure in all cases, and the sc structure is always found to be energetically very unfavorable. However, for the energetics of the bcc phase there is a serious disagreement between first-principles and atomistic calculations: The bulk modulus of bcc Al is much lower as predicted by the model potentials, with the result that its energy approaches that of the fcc phase as the volume is reduced. For the potential of Mishin et al. an fcc → bcc phase transition is predicted at a pressure of 27.4 GPa, in disagreement with the experimental fact that fcc Al is stable to at least 220 GPa. Our DFT results are consistent with earlier DFT calculations and with experiment in predicting no transition over the density or pressure range considered.