Advancing breakthrough design of superior strength-ductility synergy of lead-free solder alloys via interpretable machine learning and iterative optimization

The miniaturization, high-density integration, and multifunctionalization of electronic devices impose increasingly stringent performance requirements on solder alloys employed in microelectronic packaging. In this study, an interpretable machine-learning-based design strategy is developed by integrating element importance ranking, a small experimental dataset, and iterative optimization, with the objective of simultaneously enhancing the strength and ductility of SnAgCu (SAC)-based solder alloys. After four iterative optimization cycles, a novel Sn-Ag-Cu-Zr-Al-Ti (SACZAT) solder alloy is successfully designed, exhibiting an ultimate tensile strength of 48.26 ± 1.64 MPa and an elongation of 83.42 ± 8.94%, corresponding to improvements of 9.98% in strength and 42.02% in ductility, respectively, relative to the conventional Sn3Ag0.5Cu alloy. Microstructure analyses combined with first‑principles calculations reveal that the exceptional ductility of the SACZAT alloy originates from accelerated dynamic recrystallization (DRX) of β-Sn during deformation and β-Sn grain refinement. Meanwhile, the enhanced strength is attributed to the precipitation strengthening from newly formed intermetallic compounds and grain-refinement strengthening of the β-Sn. This study provides a machine learning framework for designing high-performance lead-free solder alloys with tunable mechanical properties by adjusting feature weights, thereby offering materials support for the development of high-reliability packaging devices.