In this study, the orientation-dependent solidification behavior of CMSX-4 single-crystal superalloy welds was systematically investigated using phase-field simulations. A multiphase-field model implemented in MICRESS, coupled with CALPHAD-based thermodynamic and mobility databases, was employed to simulate weld solidification under gas tungsten arc welding (GTAW) conditions. The numerical model was validated by comparing the simulated primary dendrite arm spacing (PDAS) with experimental observations, showing good agreement. Initial crystallographic misorientations of 0° and 15° were introduced to represent epitaxial single-crystal and high-angle polycrystalline solidification, respectively. The results reveal that, although the mushy zone range remains nearly identical for both orientations, the morphology and continuity of residual liquid at the terminal stage of solidification differ significantly. For the epitaxially grown condition (0°), residual liquid is distributed in a discontinuous droplet-like form due to early dendrite arm coalescence. In contrast, high-angle misorientation (15°) delays dendrite coalescence, leading to the formation of continuous liquid films enriched with solute elements such as Hf. This continuous liquid film provides an effective pathway for crack initiation and propagation, resulting in a substantially higher BTR. These findings demonstrate that the solidification cracking susceptibility of single-crystal superalloy welds is governed primarily by the morphology of residual liquid during the final stage of solidification rather than by the mushy zone range itself. The present study provides a mechanistic framework for understanding BTR variation in single-crystal superalloy welds and offers critical insights for controlling epitaxial growth and improving weldability in repair welding and additive manufacturing applications.
Building similarity graph...
Analyzing shared references across papers
Loading...
Hee-Eun Kim
Eun-Joon Chun
Journal of Welding and Joining
Building similarity graph...
Analyzing shared references across papers
Loading...
Kim et al. (Fri,) studied this question.
synapsesocial.com/papers/69a286600a974eb0d3c014fd — DOI: https://doi.org/10.5781/jwj.2026.44.1.7