Integrated CALPHAD and machine learning approach for the design of high-strength Sn-In-based low-temperature solders

The demand for reliable low-temperature solders is increasing with the development of flexible electronics. Eutectic Sn–In alloys are promising candidates due to their low melting temperature (118 °C); however, their relatively low mechanical strength limits broader applications. In this work, a data-driven framework with Gaussian kernel ridge regression (GKRR) integrating CALPHAD approach is developed for design of Sn–In–X alloys (X = Cu, Ag, Zn, Bi, Sb, Al, Ni). The GKRR model demonstrates good predictive capability, with a five-fold cross-validation R 2 of 0.78 and RMSE of 6.95 HV, and a testing R 2 of 0.91 with RMSE of 3.97 HV. Based on the integrated strategy, a new alloy, Sn–47In–1.59Ag–0.27Cu, was designed and experimentally validated. The designed alloy exhibits a Vickers hardness of 13.44 HV, approximately twice that of eutectic Sn–52In, and an ultimate tensile strength of 26.6 ± 0.8 MPa, representing an approximately threefold improvement while maintaining a reasonable elongation of 21.9 ± 1.4%. Microstructural characterization and thermodynamic analysis confirmed the formation of β-In 3 Sn, γ-InSn 4 , and (Ag, Cu)(In, Sn) 2 phases, with a melting point of 130.56 °C. These results demonstrate that the integration of machine learning and CALPHAD method provides an effective approach for accelerating the design of low-temperature Sn-In solders