Effect of concentration-dependent diffusion coefficient on dendrite growth in directional solidification

Abstract

Solute diffusion is an important process that determines the dendrite growth during solidification. The theoretical model generally simplifies the solute diffusion coefficient in liquid phase into a constant. Nevertheless, the composition of the boundary layer changes greatly in the solidification process, the diffusion coefficient will no longer be a constant and is dependent on concentration. In this paper, the quantitative phase field model is used to simulate the effect of concentration-dependent diffusion coefficient on dendrite growth in directional solidification. In the model, the concentration-dependent diffusion process is investigated by coupling the concentration-dependent diffusion coefficient in the liquid solute diffusion equation. A series of simulation results confirms that the concentration-dependent diffusion process has a significant effect on the dendrite growth. The results show that the increase of the coupling intensity of solute concentration will enhance the diffusion of solute in the mushy zone between primary dendrites to the dendrite tip, resulting in the increase of solute enrichment at the dendrite tip, thereby increasing the tip undercooling. The variation of diffusion coefficient in liquid phase has little effect on the tip radius of dendrite, and the simulation results are in good agreement with those from the theoretical model. Moreover, the amplitude of dendritic side branches decreases with the increase of solute diffusion coefficient. In the study of dendrite arrays, it is found that the concentration-dependent diffusion coefficient increases the primary spacing and reduce the tip position. The results of this study indicate that for a system with a concentration-dependent coefficient significantly, the effect of concentration-dependent diffusion on tip undercooling and side branches should be considered in the quantitative and experimental verification of the existing model.