文摘
The simulation of segregation in multicomponent alloy surfaces is challenging with atomistic approaches because of the need to model a very large number of possible configurations with a high degree of accuracy. Density functional theory (DFT) is too expensive to use directly, and atomistic potentials are often a compromise between accuracy and computational speed. In this work we develop a neural network (NN) atomistic potential capable of predicting accurate energies for any configuration of a AuPd(111) slab. The fully trained neural network spanning all configurations and lattice constants of a AuPd binary alloy is trained from only 3914 DFT calculations. Using this NN, segregation profiles are created spanning bulk compositions between 10 and 90% Au, and at temperatures ranging from 700 to 1000 K using Monte Carlo simulations. These profiles are then fit to the Langmuir–McLean formulation of the Gibbs-isotherm with a model for the enthalpy of segregation. The simulation results are in excellent agreement with available experimental LEIS data for the composition of the top layer. Site distributions were computed and compared to random distributions, indicating the presence of some short-range ordering favoring the formation of Au–Pd surface bonds.