Strict thermodynamic definitions were used to relate the formation energy of an arbitrary crystalline defect, in an ordered compound, to its energy (as calculated using a perfectly ordered crystal as the reference state) and to the chemical potentials of the components in a uniform alloy at absolute zero. The chemical potentials were, in turn, expressed in terms of the formation energies of constitutional defects in off-stoichiometric alloys. Expressions were derived for the chemical potentials and true energies of vacancies and antisites in a triple-defect compound. Specific calculations were performed for the present B2 compound by using molecular statics and embedded atom methods. It was found that existing embedded atom potentials for this compound were unsuitable because the point defect energies which were obtained were inconsistent with the triple-defect model. But, because this model had been experimentally confirmed, it was necessary to modify the existing embedded atom potentials in order to reconcile the simulation results with experimental data. The modified embedded atom potentials were empirically fitted to self-diffusion data on pure Ni and Al and made NiAl into a triple-defect compound. By using the modified embedded atom potentials, calculations were made of the chemical potentials of Ni and Al, of the true formation energies of point defects, and of the binding energies of their complexes. The resultant data were used to calculate the excess energies of extended defects, such as grain boundaries.

Y.Mishin, D.Farkas: Philosophical Magazine A, 1997, 75[1], 169-85