Quantitative prediction of rapid solidification by integrated atomistic and phase-field modeling

Abstract

Systematic integration of atomistic simulations with phase-field modeling is presented for quantitative predictions of cellular growth and solute trapping during solidification of alloys for solidification velocities relevant to additive manufacturing. For parametrization of the phase-field model, molecular dynamics simulations are utilized as an alternative to complex experiments to obtain the anisotropic crystal-melt interface free energy, kinetic coefficient, and diffusive interface velocity. The accuracy of this integrated model is tested for rapid solidification of Ti-3.4at.%Ni alloy. The predicted solute trapping of the proposed phase-field model is comparable with the continuous growth model for solidification velocities of additive manufacturing. The predicted primary dendritic arm spacing is weakly dependent on the diffuse interface width enabling simulations in larger length scales. The concentration profile and partition coefficient obtained from both two-and three-dimensional phase-field simulations are comparable to the results of Kurz-Fisher's analytical and continuous growth models, respectively. Unlike other computational models for rapid solidification, the proposed model enables predictions completely based on computations without fitting to experiments.